solidwork stress project

profilestrength
term.zip

TermProject-CantileverBeams.pdf

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 1/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

�� TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

CANTILEVER BEAMS

PART I: AXIAL TENSION

Date: 03-11-2018

The main goal of this term project is to compare solutions obtained by the 1-D Mechanics-of-Materials (MoM) approach, which are approximate yet more practical, with the 3-D finite element (FE) solutions. Whenever the exact and rigorous Elasticity solutions are available, they should also be included for comparison.

All FE simulations must be performed using SolidWorks 2016 version. Students are strongly encouraged to use the computer stations in ME CAD Room (Steinman Room ST-213). Once executed successfully, be sure to save all key results (e.g., stresses and displacements in ASCII/text, MS Excel and/or graphic formats), which are necessary for comparison with the theoretical MoM/Elasticity predictions. When presenting results graphically, students are suggested to use the deformed geometry. If possible, also superimpose them the undeformed meshes. Otherwise, juxtaposing the deformed and undeformed geometries for easy comparison. It should be noted that SolidWorks, just like almost all commercial general-purpose FE software, allows users to adjust scale factors to control Deformation Shape and Stress Fringes for better viewing.

In the term project, each student is assigned with a different set of load values, although all students will work on the same geometries and boundary conditions of the 3-D bars. All student are expected to add his/her own proper loading and boundary conditions, material properties and other information/data, if so required by SolidWorks and abstract all needed results (graphical, texted and/or tabulated) for the report.

In order to compare the SolidWorks FE simulations with theoretical MoM/Elasticity results numerically and/or graphically, the analytical solutions should first be formulated, derived and/or simply cited with proper references. When an analytical solution is adopted from a know source, the equation(s) needed for comparison with the FE results should only be included with concise explanation in the main text of the report and the formula can be simply, yet clearly, referred to the textbook or class notes (e.g., Eq. so and so in Page so and so, Section so and so, etc.). If a equation for comparison is not available for lifting off directly from the textbook or class notes, the student should derive it by him/herself. Details of the formulation/derivation should be given in the Appendix of the report.

PART I: AXIAL TENSION

I.1 Cantilever Beams of Circular Solid Sections (Required for all students)

As shown in Fig I.1, a prismatic cantilever beam of length L and circular solid cross-section of radius 0R is fixed at one end (A) while loaded axially at the other (B). The x-axis is through the centroidal axis (C.A.) of the axial bar, which is made of a linearly elastic material of Young’s modulus E and Poisson’s ratio Q. The bar is loaded axially at the free end by (a) a uniform tension 0V and (b) a concentrated axial force

2 0 0 0 0P A RV S V through the centroid of the end cross-section. In this term project, each student is assigned

with a different 0V value (thus, of course a different 0P ) although all students will work on the same geometries, boundary conditions and material properties. Refer to the file: Term-Project Case Assignment (2018-02-27) for the load values you need to work on and Table I.1 for the material and dimensional data.

yousef ha
yousef ha
yousef ha
yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 2/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

z

y

R0

y

x

L

V0

A BC

2x L

z

y

R0

P0

L

y

x P0BC

2x L

A

(a) uniform axial tension (b) concentrated axial force through C.A.

FIGURE I.1 An axially loaded circular cylindrical bar.

TABLE I.1 Material and dimensional data of bars of circular cross-section.

E Q 0R L 1R 2R

> @GPa > @- > @m > @m > @m > @m

2 0.25 0.2 06R 00.5R 01.5R

The afore-mentioned bar geometry has been prepared in a SolidWorks Assembly file: Assembly 2_1.SLDASM, which represents a quarter of the circular bar in the 1st octant � �0, 0, 0x y zt t t of the Cartesian x-y-z coordinate system. Each student is expected to add his/her proper loading and boundary conditions, material properties and other information/data, if so required by SolidWorks.

For Case (a): Uniform Axial Tension, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Roller/Slider option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x , the

horizontal � �0z and vertical � �0y symmetric/central planes.

x A uniform tension of 0V value should be applied on the free end face � �x L through the Pressure option of SolidWorks “External Loads”.

x An element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes (approximate 202,000 elements) through the Create Mesh option of SolidWorks “Mesh”.

Similarly, for Case (b): Concentrated Tensile Force, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Fixed Geometry option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x whereas, again, the Roller/Slider option should be applied in SolidWorks “Fixtures” to the horizontal � �0z and vertical � �0y symmetric/central planes.

x A concentrated tensile axial force of 0P value should be applied at the center of on the free-end

� � � �, , ,0,0 mx y z L . x Once again, an element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes

(approximate 202,000 elements).

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 3/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

Report: (minimum items to be included) 1. Summarize the MoM solutions of the problem, including the two field variables: stress tensor

� �, 1,2,3ij i jW and displacement vector � �1,2,3iu i . 2. For each FE case, show the mesh with loading and restraint B.C.’s. Be sure to include the coordinate system

used in the FE simulation. 3. Compare the MoM solutions with the two FEM results for the following cases:

a) Plot on the same graph the axial extensions u in [mm] from both FE simulations and MoM prediction along the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

b) Plot on the same graph the transverse contraction v in [mm] from both FE simulations and MoM

prediction along the vertical diameter of the mid-section 0;0 ; 0 2 Lx y R z§ · d d ¨ ¸

© ¹ . In a separate plot,

demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

c) Display 3 normal stress contour plots for � �, ,x y zV V V , respectively, on the mid-span cross-section

2 Lx§ · ¨ ¸

© ¹ on the same page from the Axial-Tension Case. Repeat the same for the Concentrated-Force

Case. Compare the results with the MoM predictions. Hint: The MATLAB command meshgrid can be used to defined the contour plot area whereas, as the

name implies, the command contour is for contour plots.

d) Display 3 shear stress contour plots for � �, ,xy yx zxW W W , respectively, on the mid-span cross-section 2 Lx§ · ¨ ¸

© ¹

on the same page from the Axial-Tension Case. Repeat the same for the Concentrated-Force Case. Compare the results with the MoM predictions.

e) Consider the area: 0 0;0 ; 0 2 2 R RL x L y z§ ·� d d d d ¨ ¸

© ¹ , which is a vicinity within the x-y plane around the

loading point, i.e., � � � �, , ,0,0x y z L . Display 3 xV stress contour plots in the area from the Concentrated-Force Case FE result and two Elasticity solutions based on the 2-D Flamant and 3-D Boussinesq normal-force solutions:

A. 2-D HALF-PLANE NORMAL FORCE (FLAMANT SOLUTION)

� � � �

� � � �

� � � �

3 22 0 0 0

2 2 22 2 22 2 2

2 2 2

0

x y xy

z xz yz

P L x P L x y P L x y

L x y L x y L x y V V W

S S S

V W W

� � � �

ª º ª º ª º� � � � � �¬ ¼ ¬ ¼ ¬ ¼

/ (I.1)

where the out of x-y plane line-load 0P has a unit of force

out-of-plane thickness ª º « » ¬ ¼

. For this term project,

simply choose 0 0P P .

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 4/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

B. 3-D HALF-SPACE NORMAL FORCE (BOUSSINESQ SOLUTION)

� �

� � � � � � � �

� � � � � � � �

� �

� �

3 0

5

2 2 220

3 2

2 2 220

3 2

2 0

5

2 0

5

0 3 5

3 2

3 21 2 2

3 21 2 2

3 2

3 2

3 2

x

y

z

xy

xz

yz

P L x R

y L x y R L xP R R L x L x R R R L x R L x

z L x z R L xP R R L x L x R R R L x R L x

P y L x R

P z L x R

P yz x R R

V S

QV S

QV S

W S

W S

W S

­ ½ª º� � ��° ° � � � � � �® ¾« »� � � �° °¬ ¼¯ ¿ ­ ½ª º� � ��° ° � � � � � �® ¾« »� � � �° °¬ ¼¯ ¿

� �2

21 2 R L x R L x

Q

­ ° ° ° ° ° ° ° °° ® ° ° ° ° ° ° ° � �ª º° � �« »° � �¬ ¼¯

(I.2)

where � �2 2 2R L x y z � � � .

f) Repeat e) for yV , xyW , maxV , minV , vMV and TrW (called Stress Intensity in SolidWorks), respectively.

g) Consider the area: 0 0 00 ; ; 0

2 2 R Rx y R z§ ·d d d d ¨ ¸

© ¹ , which is a vicinity within the x-y plane around the

bottom edge point of the fixed end, i.e., � � � �0, , 0, ,0x y z R . Display the stress contour plots of xV , yV ,

xyW , maxV , minV , vMV and TrW in the area from the Concentrated-Force Case FE result. h) (Option) Compare the FE results in 7) with available Elasticity solutions, e.g., Sinclair, GB “Stress

singularities in classical elasticity - I. Removal, Interpretation and Analysis and II. Asymptotic identification,” Applied Mechanics Review, Vol. 57, No. 4, pp. 251-297 and No. 5, pp. 385-439, 2004.

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 5/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

I.2 Truncated Conical Cantilever Beam (Required for all students)

Figure I.2 shows a truncated conical bar of length L and end radii � �1 2,R R is fixed as a roller/slider boundary at

the restrained end (A) while loaded axially by a uniform tension: 2 0

1 02 1

R R

V V at the free end (B). In the figure

the x-axis is through the centroidal axis (C.A.) of the bar, which is made of a linearly elastic material of Young’s modulus E and Poisson’s ratio Q.

R1

R2

z

y L

x

y

V1

2x L

CA B

FIGURE I.2 A truncated conical bar loaded by uniform axial tension.

Refer to the file: Term-Project Case Assignment (2018-02-27) for the load values you need to work on and Table I.1 for the material and dimensional data. Finally, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Roller/Slider option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x , the

horizontal � �0z and vertical � �0y symmetric/central planes.

x A uniform tension of 1V value should be applied on the free end face � �x L through the Pressure option of SolidWorks “External Loads”.

x An element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes (approximate 215,000 elements) through the Create Mesh option of SolidWorks “Mesh”.

x

Report: (minimum items to be included) 1. Show the mesh with loading and restraint B.C.’s. Be sure to include the coordinate system used in the FE

simulation. 2. Find the MoM solution of the axial normal stress � �xV distribution.

Hint: Force balance. 3. Find the MoM solution of the displacement field � �1,2,3iu i .

Note: The governing eq of the axial displacement � �1u u of this axially loaded bar with variable cross-section is

� � � �, 0

du x tdE A x dx dx

ª º « »

¬ ¼ (I.3)

4. Due to its axi-symmetrical nature, the conventional MoM approach concludes this axially loaded circular object will not have transverse shear stresses � �,xy xzW W , hence � �,xr xTW W , where � �, ,r xT constitutes a cylindrical coordinate system. Inspecting the equilibrium equations in cylindrical coordinate system, Eq (1.8-C3), to prove this conclusion is wrong. Indeed, an Advanced MoM approach taking into account the tapered-beam effect, does prove the existence of transverse shears. The results are summarized in the file:

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 6/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

Transverse Shear Stresses in Tapered Beams (2017-12-06)

5. Compare the above MoM/Advanced MoM solutions with the FEM results for the following cases: a) Plot on the same graph the axial extensions u in [mm] from the FE simulation and MoM prediction along

the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

b) Plot on the same graph the transverse contraction v in [mm] from the FE simulation and MoM prediction

along the vertical diameter of the mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate plot,

demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

c) Plot on the same graph the axial normal stress xV in [MPa] from the FE simulation and MoM prediction along the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

d) Plot on the same graph the transverse shear stress xyW in [MPa] from the FE simulation and MoM

prediction along the vertical diameter of the mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate

plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

e) Plot the transverse normal stress yV in [MPa] from the FE simulation along the vertical diameter of the

mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate plot, demonstrate how the data are abstracted

using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

yousef ha

values.png

stress.zip

stress/.DS_Store

__MACOSX/stress/._.DS_Store

stress/Ch 1 Analysis of Stress-Part 1 (2018-01-25).pdf

SECTION 1.1 INTRODUCTION PAGE 1/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

CHAPTER 1: ANALYSIS OF STRESS

1.1 INTRODUCTION (Lecture)

Mechanics of Materials vs Theory of Elasticity

  elementary

theory for approximate practical solutions~ technical

~ exact and rigorous solutions

yet           

Mechanics of Material

Theory of Elast

Solid M s

e M

ch

icity

oM anics

~ body & surface forces

~ normal & shear stresses

~ vibration & wave propagation

~ humidity & temper

external force

quasi - static

internal force

hygrothermal loadin

dynamic impact

l

l

loa

g

o

oading

adi

d

ng

ing

  

ature effects

       

Def: body force: an external force acts throughout the entire body V of a solid. It has a unit of force per unit

volume. Examples of body forces include gravitational-weight force, inertial force, magnetic force, etc.

Def: surface force: an external force, acts over the entire or part of the surface S of a solid. It has a unit of

force per unit area. Examples of surface forces include pressure and aerodynamic lift/drag, etc.

(a) body force:

cantilever beam under its own weight

(b) surface force:

aerodynamic lift and drag over an airfoil

FIGURE 1.1-A1 Examples of body and surface forces.

SECTION 1.1 INTRODUCTION PAGE 2/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Classification of Structures: Geometry & Loading (Self-Study)

Def: structure: A collection of bodies arranged and supported so that it can resist and transmit loads. They

can be classified into following groups, based upon a combination of geometric configurations and

loading characteristics.

A. 1-D Structures (or Bars): 1-D straight or curved structural member possessing one dimension

significantly greater than the other two.

rod (or tie member or tensile bar): a straight bar loaded in tension along the longitudinal axis.

cable (or string): a flexible tie with zero or negligible flexural rigidity and can sustain only axial tensile

forces.

column (or compressive bar): a straight bar loaded in compression along the longitudinal axis. (Note: Slender

columns are susceptible to failure in buckling.)

torsional bar (or shaft): a straight bar loaded by twisting torques about the longitudinal axis.

beam: a straight bar possessing one dimension significantly greater than the other two, bent flexurally in

directions normal to the longitudinal axis.

beam on elastic foundation: a loaded beam resting on an elastic foundation.

beam-column: a beam loaded simultaneously by bending and compression. (Note: Slender beam-columns

are susceptible to failure in buckling.)

beam-tie (or tension-beam): a beam loaded simultaneously by bending and tension.

curved beam: a curved beam subject to bending, twisting, shear and axial loads.

arch: a curved beam supported at its ends and loaded primarily in direct compression.

ring: a closed curved beam.

truss: a structure consisting of two or more axial bars joined by frictional hinges and with each member

loaded by an axial force only.

frame: a structure made of two or more bars, which are rigidly attached and under bending, shear and axial

loads.

B. 2-D Structures: a 2-D flat or curved structural member possessing two dimensions significantly large in

comparison with the third.

panel: a 2-D flat structural member subject to in-plane loads, which act in directions tangent to the mid-

surface.

shear panel: a panel loaded only by in-plane shears.

membrane: a flexible panel with zero or negligible flexural rigidity and can resist only in-plane tensions.

balloon: a curved membrane.

plate: a 2-D flat structural member subject to out-of-plane loads, which act in directions perpendicular to

the mid-surface.

shell: a curved plate, which can be loaded simultaneously by the in-plane stretching, compression and shear

as well as out-of-plane bending and twisting.

stiffened panel, plate or shell: a panel, plate or shell reinforced with bars.

SECTION 1.1 INTRODUCTION PAGE 3/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Rod (or Tie Member or Axially-Tensile Bar)

Column (or Compressive Bar)

Torsional Bar (or Shaft)

Beam

Beam on Elastic Foundation

Beam-Column

Beam-Tie (or Tension Beam)

Plate (Out-of-Plane) Panel (In-Plane)

Shell (or Curved Plate)

Frame

Truss

Curved Beam

Ring

Arch

SECTION 1.1 INTRODUCTION PAGE 4/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

(Axial-)Bar Tension ( 0C member) Closed-Section Torsion (

0C member)

EA

L  x,u

y,v

P 0

L

GK t

T 0

y,v

x,

end elongation:   0P L u L

EA    end angle of twist:   0 0

24t

T L T SL L or

GK G t   

normal stress distribution

   x

x

P x x

A  

shear stress distributions:

   

, T x r

x r J

  &    

  ,

2

T x x s

t s  

normal force vs axial-load intensity relation:

   xdP x

s x dx

 

twisting torque vs torque intensity relation:

   t

dT x m x

dx  

normal force vs axial displacement relation:

   

x

du x P x EA

dx 

twisting torque vs angle of twist relation:

   

t

d x T x GK

dx

 

axial rigidity: EA torsional rigidity: tGK

EA

x,u

y,v

P x

dx

P x +dP

x

s(x)

GK t

R

x,u

y,v

T

m t (x)

dx

T+dT

ds d

D.E.:  

  2

2

d u x EA s x

dx   D.E.:

   

2

2t t

d x GK m x

dx

  

B.C.:  

  0

0 : 0 0 @

: x

x u

x L P L P

  

  B.C.:

 

  0

0 : 0 0 @

:

x

x L T L T

  

 

SECTION 1.1 INTRODUCTION PAGE 5/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Beam Bending ( 1C member) (Thin-Walled) Open-Section Torsion (

1C member)

EI z

L

x,u

y,v

V 0

x,

y,v

B 0

T 0

EI 

& GK t

L

end deflection:  

3

0

3 z

V L v L

EI    end angle of twist:   0 0

24t

T L T SL L or

GK G t   

bending normal/shear stress distributions:

   

     

  , ; ,

y zz

x xy

z z

V x Q yM x y x y x y

I I b y    

warping normal/shear stress distributions:

     

     

  , ; ,

B x s T x Q s x s x s

I I t s

  

 

 

    

bending moment/shear force vs load intensity

relations:    

  2

2

yz dV xd M x

p x dx dx

  

bimoment/twisting torque vs torque intensity

relations:    

  2

2 t

d B x dT x m x

dx dx

    

bending moment/shear force vs deflection

   

   2 3

2 3 ;z z y z

d v x d v x M x EI V x EI

dx dx   

Bimoment/twisting torque vs angle of twist:

   

     2 3

2 3 ; t

d x d x d x B x EI T x GK EI

dx dx dx    

      

bending (or flexural) rigidity: zEI warping rigidity: EI & torsional rigidity: tGK

M z  dM

z

V y  dV

y

dx

p(x)

M z

V y

x,u

y,v

EI z

D.E.:  

  4

4z

d v x EI p x

dx  D.E.:

     

4 2

4 2t t

d x d x EI GK m x

dx dx 

   

B.C.:    

    0

0 : 0 0 & 0 0 @

: 0 &z y

x v

x L M L V L V

   

   B.C.:

 

  0

0 : 0 0 @

:

x

x L T L T

  

 

SECTION 1.1 INTRODUCTION PAGE 6/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Questions: 1. How many types of structural members used in this modern high-speed railway station?

2. In addition to Structural Engineering (or Stress Analysis), what other disciplines are needed for

the design, manufacture, test and operation of this modern technological product?

SECTION 1.1 INTRODUCTION PAGE 7/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Case Histories of Error & Judgment in Stress Analysis (Option)

Reference: H. Petroski, Design Paradigms: Case Histories of Error and Judgment in Engineering, Cambridge

University Press, NY, 1994.

Ancient Italian marble column in storage: top, with

modified support; bottom, as originally supported.

Galileo’s illustration of two failure modes.

Mid-air explosion of space shuttle Challenger during

launching in a chilly winter morning (January 28, 1986). It

was caused by failure of an O-ring, which was designed to seal

the shuttle booster rocket (SBR’s).

O-ring designs for Titan III and space shuttle

booster rockets.

SECTION 1.1 INTRODUCTION PAGE 8/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

New joint design and other changes due to the Challenger accident.

SECTION 1.1 INTRODUCTION PAGE 9/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

(a) Sagging

(b) Hogging

William Fairbairn’s illustration of ship loadings caused by wave motion; top, sagging as supported on two wave

crests; bottom, hogging as supported on a single wave crest, 1865.

A failed Liberty ship, circa 1940.

SECTION 1.1 INTRODUCTION PAGE 10/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

The Tacoma Narrows Bridge (Tacoma, Washington, USA) in its fatal torsional oscillation mode and collapsing,

1940. The resonant twisting motion was caused by fluid-induced vibration due to aerodynamic Kármán

vortices.

The Bronx-Whitestone Bridge, as built in 1939 (left) and as modified in 1946, employing the original stiffening

girders as the bottom chord of an unattractive stiffening truss (right).

SECTION 1.1 INTRODUCTION PAGE 11/11

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Architectural rendition of suspended walkways in the Kansas City Hyatt Regency Hotel.

Connection detail of upper suspended walkway in the Kansas City

Hyatt Regency Hotel, which failed in 1981; left, as built; right, as

originally designed.

Failed walkway connection.

SECTION 1.2 SCOPE OF TREATMENT PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.2 SCOPE OF TREATMENT (Self-Study)

Principal Topics of Mechanics of Solids (Mom)

1. Analysis of the stresses and deformations within a body subject to a prescribed system of forces. This is

accomplished by solving the governing equations that describe the stress and strain fields (theoretical stress

analysis). It is often advantageous, where the shape of the structure or conditions of loading preclude a

theoretical solution or where verification is required, to apply the laboratory techniques of experimental

stress analysis.

2. Determination by theoretical analysis or by experiment of the limiting values of load that a structural

element can sustain without suffering damage, failure, or compromise of function.

3. Determination of the body shape and selection of the materials that are most efficient for resisting a

prescribed system of forces under specified conditions of operation, such as temperature, humidity,

vibration, ambient pressure, etc. This is the design function.

SECTION 1.3 ANALYSIS & DESIGN PAGE 1/1

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.3 ANALYSIS & DESIGN (Self-Study)

    

energy msolution meth

num

force moment e

erical finite

e

e

th

le

quili

ment

briu

meth

m met

od

h

od

d

o

o

ds

Basic Principles of Mechanics/Materials-Based Engineering Analysis

1. Equilibrium Conditions. The equations of equilibrium of forces/moments must be satisfied throughout

the member.

2. Material Behaviors/Constitutive Relations. The stress-strain or force-deformation relations (for

example, Hooke’s law) must apply to the material behavior of which the member is constructed.

3. Geometry of Deformation/Compatibility Conditions. The compatibility conditions of deformations must

be satisfied: that is, each deformed portion of the member must fit together with adjacent portions. (Note:

For mathematical strictness, the matter of compatibility should always be complied in Theory of Elasticity;

however, it may not always be broached in Mechanics of Materials analysis.)

4. Boundary and Initial Conditions. The stress and deformation obtained through the use of the above three

principles must conform to the initial conditions: the initial values of displacements (and velocities for

dynamic problems) of the member as well as satisfy the boundary conditions: conditions of loading

imposed at the boundaries of the member.

Rational Procedure in Mechanics/Materials-Based Engineering Design

1. Evaluate the most likely modes of failure of the member. Failure criteria that predict the various modes of

failure under anticipated conditions of service are discussed in Ch. 4.

2. Determine the expressions relating applied loading to such effects as stress, strain, and deformation.

Often, the member under consideration and conditions of loading are so significant or so amenable to

solution as to have been the subject of prior analysis. For these situations, textbooks, handbooks, journal

articles, and technical papers are good sources of information. Where the situation is unique, a

mathematical derivation specific to the case at hand is required.

3. Determine the maximum usable value of stress, strain, or energy. This value is obtained either by reference

to compilations of material properties or by experimental means such as simple tension test and is used in

connection with the relationship derived in Step 2.

4. Select a design factor of safety. This is to account for uncertainties in a number of aspects of the design,

including those related to the actual service loads, material properties, or environmental factors. An

important area of uncertainty is connected with the assumptions made in the analysis of stress and

deformation. Also, we are not likely to have a secure knowledge of the stresses that that may be introduced

during machining, assembly, and shipment of the element.

(design) factor of safety:  

maximum usable stress

allowable working stress n

or  (1.1)

  , ~ brittle materials

, ~ ductile materia

ls

    yd

maximum usable stre ultimate tensile stregth str

yield strength str ss

es

e

s

ss u

σ

σ

SECTION 1.4 CONDITIONS OF EQUILIBRIUM PAGE 1/1

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.4 CONDITIONS OF EQUILIBRIUM (Self-Study)

Equilibrium Equations of Statics

for 3D problems: 0 0 0

0 0 0

x y z

x y z

F F F

M M M

    

  

     

~ 6 eqs (1.2)

for planar problems:  0 0 0x y zF F M     ~ 3 eqs (1.3)

alternatively,

0 0 0x Z ZA B F M M     provided that line AB is not  to x-axis; (1.4a)

or 0 0 0Z Z ZA B C M M M     where points A, B, and C are not collinear. (1.4b)

Static Determinate vs Indeterminate Systems

A structure is statically determinate when all forces on its members can be found by using only the conditions

of equilibrium. If there are more unknowns than available equilibrium equations of statics, the problem is

called statically indeterminate. The degree of static indeterminacy is equal to the difference between the

number of unknown forces and the number of relevant equilibrium conditions. Any reaction that is in excess

of those that can be obtained by statics alone is termed redundant. The number of redundants is therefore the

same as the degree of indeterminacy.

  • 1.1 Introduction (2018-01-25)
  • 1.2 Scope of Treatment (2018-01-25)
  • 1.3 Analysis & Design (2018-01-25)
  • 1.4 Conditions of Equilibrium (2018-01-25)

__MACOSX/stress/._Ch 1 Analysis of Stress-Part 1 (2018-01-25).pdf

stress/Ch 1 Analysis of Stress-Part 2 (2018-01-25).pdf

SECTION 1.5 STRESS TENSOR: DEFINITION & STRESS COMPONENTS PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.5 STRESS TENSOR: DEFINITION & STRESS COMPONENTS (Lecture)

FIGURE 1.5 (a) Prismatic bar in uniaxial tension; (b) stress distribution across the cross-section.

1-D normal stress: x

P

A   (1.10)

FIGURE 1.1 Method of sections & free-body diagram: (a) sectioning of a loaded body;

(b) free body with external and internal forces; (c) enlarged area A with components of the force F .

0 0 0 lim lim lim

y yx x z z x xy xz

A A A

F dFF dF F dF

A dA A dA A dA   

     

       

   (1.5)

FIGURE 1.2 Element subjected to three-dimensional stress. All stresses have positive sense.

stress tensor:   xx xy xz x xy xz

ij yx yy yz yx y yz

zx zy zz zx zy z

     

      

     

       

             

τ ~ 2 nd

-rank tensor (1.6)

Note: The notations: ij vs ij (or similarly,  τ vs  σ ) are used interchangeably in this course.

SECTION 1.5 STRESS TENSOR: DEFINITION & STRESS COMPONENTS PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Sign convention of stress component: For a stress component ij , the subscript i represents the positive

surface normal whereas subscript j points to the positive force direction.

Tensorial-Indicial Notation (see Sec 1.17)

Range convention: When a lowercase alphabetic subscript is unrepeated, it takes on all values indicated.

(Einstein) summation convention: Unless stated otherwise, when a lowercase alphabetic/Greek subscript

appears twice in the same term, then summation over the range (e.g., from 1 to 3 for a 3-D problem) of that

subscript is implied, making the use of the summation symbol  unnecessary.

Note: It should be apparent that ii jj kk     , and therefore the repeated subscripts or indices are

sometimes called dummy subscripts. Unspecified indices that are not repeated are called free or distinct

subscripts.

Special Stress States

a. Triaxial Stress. An element subjected to only stresses and acting in mutually perpendicular directions is

said to be in a state of triaxial stress. Such a state of stress can be written as:

  1

2

3

0 0

0 0

0 0

x xy xz

ij yx y yz

zx zy z

   

    

   

       

           

τ (a)

Note: The absence of shear stresses indicates that the preceding stresses are the principal stresses for the

element.

a1. Spherical or Dilatational or Hydrostatic Stress. A special triaxial-stress case occurring if all principal

stresses are equal: 1 2 3    . Equal triaxial tension/compression is also called hydrostatic

tension/compression. An example of hydrostatic compression is found in liquid under hydrostatic

pressure:

 

0 0

0 0

0 0

x xy xz

ij yx y yz

zx zy z

p

p

p

  

   

  

       

            

τ where p h gh   (b)

FIGURE 1.3 Examples of special stress states: (a) Element in plane stress;

(b) two-dimensional presentation of plane stress; (c) element in pure shear.

SECTION 1.5 STRESS TENSOR: DEFINITION & STRESS COMPONENTS PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

b. Two-Dimensional or Plane Stress. Only the x and y faces of the element are subjected to stress, and all

the stresses act parallel to the x and y axes, as shown in Fig 1.3a & b.

 

0

0

0 0 0

x xy xz x xy

ij yx y yz xy y

zx zy z

    

     

  

       

           

τ or simply,   x xy

ij

xy y

  

 

     

  τ (1.8)

Note: The relation: ij ji  or

xy yx

yz zy

zx xz

 

 

 

 

  

, i.e., the stress tensor is symmetric will be proven in Sec 1.8.

b1. Biaxial Stress. A special plane-stress case occurs if only two normal stresses are present:

  1

2

0

0

x xy

ij

xy y

   

  

         

   τ (c)

Note: The absence of shear stresses implies the principal stresses of a biaxial-stress state are:

 1 2 0  .

c. 3-D Pure Shear. The element is subjected to shear stresses only:

  1 2

1 3

2 3

0

0

0

x xy xz

ij yx y yz

zx zy z

    

     

    

       

           

τ (d)

c1. (2-D) Pure Shear. The element is subjected to plane shear stresses only (Fig. 1.3c):

  0

0

0 0

0 0

0 0 0

x xy xz

ij yx y yz

zx zy z

   

    

  

       

           

τ or simply,   0

0

0

0 ij

 

     

  τ (e)

Note: A typical 2-D pure shear occurs over the cross

sections and on longitudinal planes of a circular

shaft subjected to torsion.

d. Uniaxial Stress. When normal stresses act along one

direction only, the one-dimensional state of stress is

referred to as a uniaxial (or simple) tension/compression:

  0

0 0 0

0 0

0 0 0

x xy xz

ij yx y yz

zx zy z

  

    

  

       

           

τ (f)

FIGURE 1.5-A1 Sample special stress states.

SECTION 1.6 INTERNAL FORCE/MOMENT-RESULTANTS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.6 INTERNAL FORCE/MOMENT-RESULTANTS (Lecture)

FIGURE 1.4 Positive forces and moments on a cut section of a body

and components of the force dF on an infinitesimal area dA .

internal force & moment resultants:  

x y xy z xz

xz xy y x z x

P dA V dA V dA

T y z dA M zdA M ydA

  

   

    

    

  

   (1.9)

1. The axial force P or N tends to lengthen or shorten the member.

2. The shear forces yV and zV tend to shear one part of the member relative to the adjacent part and are often

designated by the letter V.

3. The torque or twisting moment T is responsible for twisting the member.

4. The bending moments yM and zM cause the member to bend and are often identified by the letter M.

SECTION 1.6 INTERNAL FORCE/MOMENT-RESULTANTS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Table 1.1 Commonly used elementary (MoM) formulae for stress a

a Detailed derivations and limitations of the use of these formulae are described in Secs 1.6, 5.7, 6.2 & 13.13.

SECTION 1.7 STRESSES ON INCLINED SECTIONS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.7 STRESSES ON INCLINED SECTIONS (Self-Study)

Axially Loaded Members

FIGURE 1.6 (a) Prismatic bar in tension; (b, c) side views of a part cut from the bar.

axially loaded member:

2cos cos

sin sin cos

x x

x

x y x

x

P

A

P

A

   

    

 

  

       

(1.11)

max

max

when 0 180

1 when 45 135

2 2

x

x

P or

A

P or

A

  

  

     

         

(1.12)

SECTION 1.7 STRESSES ON INCLINED SECTIONS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Example 1.1 State of Stress in a Tensile Bar (Self-Study)

Compute the stresses on the inclined plane with 35   for a prismatic bar of a cross-sectional area 800 mm 2 ,

subjected to a tensile load of 60 kN (Fig 1.6a). Then determine the state of stress for 35   by calculating the

stresses on an adjoining face of a stress element. Sketch the stress configuration.

Solution The normal stress on a cross section is:

   

3

6

60 10 75 MPa

800 10 x

P

A 

   

Introducing this value in Eqs (1.11) and using θ = 35°, we have:

 

  

22cos 75 cos35 50.33 MPa

sin cos 75 sin 35 cos35 35.24 MPa

x x

x y x

  

   

 

     

       

The normal and shearing stresses acting on the adjoining y′ face are 24.67 MPa and 35.24 MPa, respectively, as

calculated from Eq (1.11) by substituting the angle 90 125     . The values of  ,x x y    are the same on

opposite sides of the element. On the basis of the established sign convention for stress, the required sketch is

shown in Fig 1.8.

FIGURE 1.7 Variation of stress at a point with the inclined section in the bar shown in Fig 1.6a.

FIGURE 1.8 Stress element for 35   .

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 1/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.8 EQUILIBRIUM EQUATIONS ~ VARIATION OF STRESS WITHIN A BODY (Lecture)

(a) 3-D stress components at a material point

(b) X-Y planar view with body force components within a differential element

FIGURE 1.9 Differential element with stresses and body forces in Cartesian coordinates.

Differential Scheme

With respect to the center of the differential element, taking moment balance about the z-direction:

,center 0zM       0 2 2 2 2

xy xy xy yx yx yx

dx dx dy dy d dydz dydz d dxdz dxdz          

Apply the chair rule for partial differentiation, we get:

xy xy

xyd dx dy x y

  

   

 

xy dz

z

 

yx

yxd dx x

 

 

yx yx dy dz

y z

    

 

     

 0 2 2 2 2

xy yx

xy xy yx yx

dx dx dy dy dx dydz dydz dy dxdz dxdz

x y

     

             

    

or     2 21 1 1 1 1 1

0 2 2 2 2 2 2

xy yx

xy xy yx yxdxdydz dx dydz dxdydz dxdydz dx dy dz dxdydz x y

     

             

    

Ignoring the higher-order terms, we obtain: xy yx 

Similarly, taking moment balance about y-direction  zx xz 

and taking moment balance about x-direction  yz zy 

(1.7)

Thus, equality of shear stresses  ij ji   2 nd

-rank symmetric tensor

Paradox: Breakdown of symmetry in stress tensor: Pure Shear vs Simple Shear (to be explained in lecture).

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 2/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Now taking force balance in x-direction:

0xF         0x x x yx yx yx zx xz zx xd dydz dydz d dxdz dxdz d dxdy dxdy F dxdydz                 

Similarly, apply the chair rule for partial differentiation, we get:

x x

xdx dydz dydz x

 

      

yx yx

yxdy dxdz dxdz y

 

    

  zx zx

zxdz dxdy dxdy z

 

    

  0xF dxdydz 

 0 yxx zx

xF dxdydz x y z

        

    , or

0 yxx zx

xF x y z

      

    0

xyx xz xF

x y z

      

  

Similarly, force balance in y-direction: 0 xy y zy

yF x y z

        

    0

xy y yz

yF x y z

        

  

and force balance in z-direction: 0 yzxz z

zF x y z

      

  

(1.14)

In tensor-index notation: , 0 where , , , ij

i ij j i

j

F F i j x y z x

 

     

 (1.15)

In vector-matrix notation:   0  τ F (1.8-A1)

For 2-D case: , 0 where , ,F F x y x



   

   

     

 or

0

0

xyx x

xy y

y

F x y

F x y



 

   

       

  

(1.13)

Notes: a. In Eq (1.15), i is a free index whereas j is a dummy index; similarly,  is free while  is dummy in

Eq (1.13).

b. In Eq (1.14), there are 6 unknown stress components: , , , , ,x y z xy xz yz      , but only 3 equations;

similarly, 3 unknown stresses: , ,x y xy   with 2 equations only. Hence, stress analysis problems are

in general internally statically indeterminate.

c. zero body force: 0x y zF F F   , then:

0

0

0

xyx xz

xy y yz

yzxz z

x y z

x y z

x y z

 

  

 

    

      

     

     

  

or , 0 where , , , ij

ij j

j

i j x y z x

 

   

 (1.8-A2)

indicating that the sum of the three stress derivatives is zero.

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 3/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Example 1.2 The Body Forces in a Structure (Self-Study)

The stress field within an elastic structural member is expressed as follows:

3 2 3 2 2 3

2 3 2

1 2 4

2

5 2 0

x y z

xy xz yz

x y x y y z

z y xz x y

  

  

      

    

(d)

Determine the body force distribution required for equilibrium.

Solution Substitution of the given stresses into Eq (1.14) yields:

     

     

     

2 2

3 2

3 4 3 0

0 0 0

2 0 3 0

xyx xz

x x

xy y yz

y y

yzxz z z z

x y xzF F x y z

F y F x y z

F z xy z F x y z

 

  

 

         

      

         

            

  

The body force distribution, as obtained from these expressions, is therefore

2 2

2 3

3 4 3

2 3

x

y

z

x y xzF

yF

F xy z z

    

      

(e)

The state of stress and body force at any specific point within the member may be obtained by substituting the

specific values of x, y, and z into Eqs (d) and (e), respectively.

Integral Scheme (Option)

Figure 1.8-B1 Body and surface forces acting on an arbitrary portion of a continuum.

conservation of linear momentum (force balance principle): 0n

i i S V T dS FdV   (1.8-B1)

n

i ji jT n  0ji j i S V

n dS FdV    (1.8-B2)

divergence theorem: ,ji j ji j S V

n dS dV     , 0ji j i V

F dV   (1.8-B3)

zero-value theorem  equilibrium eqs in tensor-index notation: , 0ji j iF   V (1.8-B4)

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 4/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

vector-matrix notation:   0  τ F (1.8-B5)

scalar-component notation:

0

0

0

yxx zx x

xy y zy

y

yzxz z z

F x y z

F x y z

F x y z

 

  

 

     

      

      

      

  

(1.8-B6)

conservation of angular momentum (moment balance principle): 0n

ijk j k ijk j k S V

x T dS x F dV    (1.8-B7)

n

k lk lT n  0ijk j lk l ijk j k S V

x n dS x F dV    

Gauss divergence theorem:   ,ijk j lk l ijk j lk lS V

x n dS x dV        ,

0ijk j lk ijk j klV x x F dV    

 

expand and simplify the integral  0ijk jk V

dV  

zero-value theorem  0ijk jk  

 symmetric stress tensor: ij ji  

xy yx

yz zy

zx xz

 

 

 

 

  

(1.8-B8)

 equilibrium eqs in tensor-index notation: , 0ij j iF   (1.8-B9)

scalar-component notation:

0

0

0

xyx xz x

xy y yz

y

yzxz z z

F x y z

F x y z

F x y z

 

  

 

     

      

      

      

  

(1.8-B10)

Note: The stress tensor is symmetric only if there is no body moment or force doublet, which exist in strong

electromagnetic fields. See M.H. Sadd, Elasticity, 3rd ed., Ch 15: Micromechanics Applications for

Mindlin’s micropolar & stress-couple theories.

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 5/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Curvilinear Coordinate Systems (Option)

Figure 1.8-C1 Cylindrical and spherical coordinate systems.

Cylindrical coordinate system:

 

radial coordinate :

polar coordinate :

axial longitudinal coordinate :or

r

z

    

cylindrical vs Cartesian coordinates:

2 2

1

cos

sin tan

x r r x y

y y r

x

z z z z

  

     

  

  

(1.8-C1)

stress components in cylindrical coordinate system:   r r rz

ij r z

rz z z

  

  

   

  

   

      

τ (1.8-C2)

equilibrium eqs:

  1 1

0

1 2 0

1 1 0

rr rz r r

r z r

zrz z rz z

F r r z r

F r r z r

F r r z r

 

    

   

   

  

          

   

       

         

(1.8-C3)

SECTION 1.8 EQUILIBRIUM EQUATIONS PAGE 6/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Figure 1.8-C2 Stress components in cylindrical and spherical coordinate systems.

Spherical coordinate system:  

 

radial coordinate :

azimuthal angle coordinate :

polar angle coordinate :

R

    

spherical vs Cartesian coordinates:

2 2 2

1

2 2 2

1

cos sin

sin sin cos

cos tan

x R R x y z

z y R

x y z

y z R

x

 

  

 

      

      

  

(1.8-C4)

stress components in cylindrical coordinate system:   R R R

ij R

R

 

  

  

  

   

  

   

       

τ (1.8-C5)

equilibrium eqs:

 

 

 

1 1 1 2 cot 0

sin

1 1 1 cot 3 0

sin

1 1 1 2 cot 3 0

sin

R RR R R R

R

R

R R

F R R R R

F R R R R

F R R R R

    

  

   

    

      

  

      

  

    

  

         

      

              

        

(1.8-C6)

  • 1.5 Stress Tensor (2018-01-25)
  • 1.6 Internal Force-Moment Resultants (2018-01-25)
  • 1.7 Stresses on Inclined Sections (2018-01-25)
  • 1.8 Equilibrium Equations (2018-01-25)

__MACOSX/stress/._Ch 1 Analysis of Stress-Part 2 (2018-01-25).pdf

stress/Ch 1 Analysis of Stress-Part 3 (2018-01-25).pdf

SECTION 1.9 PLANE-STRESS (2-D) TRANSFORMATION PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.9 PLANE-STRESS (TWO-DIMENSIONAL) TRANSFORMATION (Lecture)

FIGURE 1.10 Plane-stress example:

a thin plate subject to in-plane loads.

FIGURE 3.1 Plane- strain example:

a long cylindrical body with closed ends.

FIGURE 1.11 Elements in plane stress.

From the wedge element ABO shown in Fig 1.11b, we have: cos sin

AO BO AB

   

Assume unit thickness and by force balance in x- & y-directions, respectively, in the wedge element, we get:

cos sin

cos sin

x x xy x x y

y xy y xy y

p AB AO BO AB AB

p AB AO BO AB AB

     

     

       

   

 cos sin

cos sin

x x xy

y xy y

p

p

   

   

  

  (1.16)

where xp and yp , called tractions, are the components of the stress resultant acting on the AB plane in the x-

and y-directions, respectively. The normal and shear stresses in the -x y  coordinate system are obtained by

projecting xp and yp in the -x and -y directions, then summing the results, respectively:

cos sin

cos sin

x x y

x y y x

p p

p p

  

  

 

  

  

   

2 2

2 2

cos sin 2 sin cos

cos sin sin cos

x x y xy

x y xy y x

       

       

 

    

   

(1.17a,b)

Note that the normal stress y  acting on the y face of an inclined element (Fig 1.11c) may readily be

obtained by substituting 2  for θ in the expression for x  . In so doing, we have:

2 2sin cos 2 sin cosy x y xy           (1.17c)

SECTION 1.9 PLANE-STRESS (2-D) TRANSFORMATION PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

trigonometric identities:

2

2

1 cos 2 cos

2

1 cos 2 sin

2

sin 2 sin cos

2

 

 

  

 

 

  

 

 transformation equations for plane stress:

   

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

cos sin sin cos sin 2 cos 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

x y x y

x x y xy xy

x y

x y xy y x xy

x y x y

y x y xy xy

              

            

              

 

       

 

       

       

(1.18)

Note: constantx y x y        ~ stress invariant (see Sec 1.13)

By interchanging symbols:    , , , ,x y x y    , we also get:

   

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

cos sin sin cos sin 2 cos 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

x y x y

x x y x y x y

x y

xy x y x y x y

x y x y

y x y x y x y

              

            

              

   

     

 

     

   

     

       

      

       

       

(1.9-A1)

Figure 1.9-A1 Two-dimensional orthogonal (or rotational) transformation.

SECTION 1.9 PLANE-STRESS (2-D) TRANSFORMATION PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Stress Trajectory - Polar Representation of State of Plane Stress (Option)

FIGURE 1.12 Stress trajectories: polar representations of x  and x y   (in MPa) vs .

Cartesian Representation of State of Plane Stress (Option)

FIGURE 1.13 Graph of normal stress x  and shear stress x y   within angle 0 180   .

SECTION 1.10 2-D PRINCIPAL STRESSES AND MAXIMUM SHEAR STRESS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.10 TWO-DIMENSIONAL PRINCIPAL STRESSES AND MAXIMUM SHEAR STRESS (Lecture)

From Eq (1.18a), in order to obtain the maximum or minimum x  , we have

0xd

d

     cos 2 sin 2 sin 2 2 cos 2 0

2 2

x y x y

xy x y xy

d

d

           

          

  (a)

 2 sin 2

tan 2 cos 2

xy

x y

  

    

  principal (stress) direction: 1

21 tan

2

xy

p

x y

 

 

 

(1.19)

  

 

2 2

2 2

2 sin 2

4

cos 2

4

xy

p

x y xy

x y

p

x y xy

 

  

  

  

 

    

  

 

(1.10-A1)

Eq (1.10-A1)  Eqs (1.18a,c), we obtain

 

   

 

   

2 2

2 2 2 2

2 2

2 2 2 2

21 cos 2 sin 2

2 2 2 2 4 4

21 cos 2 sin 2

2 2 2 2 4 4

p

p

x yx y x y x y xy

x p xy p

x y xy x y xy

x yx y x y x y xy

y p xy p

x y xy x y xy

 

 

           

     

           

     

 

 

           

         

           

 principal stresses:

2

2

max,min 1,2 2 2

x y x y

xy

      

       

  (1.20)

Notes: a. Since  tan 2 tan 2p p    , hence there are two principal directions:  ,p p   correspond to

 1 max 2 min,     , respectively. Furthermore, p p   .

b. It is necessary to substitute one of the p values into Eq (1.18a) to determine which of the principal

directions corresponds to the maximum principal stress 1 .

c. Shear stress vanishes on a principal plane: 0 p p

xy x y         

  .

Similarly, use Eq (1.18b) to obtain the maximum shear stress, we have:

0 x yd

d

     sin 2 cos 2 cos 2 2 sin 2 0

2

x y

xy x y xy

d

d

         

           

 sin 2

tan 2 2 cos 2

x y

xy

   

 

     maximum shear direction: 11

tan 2 2

y x

s

xy

  

 

 (1.21)

SECTION 1.10 2-D PRINCIPAL STRESSES AND MAXIMUM SHEAR STRESS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

  

 

2 2

2 2

sin 2

4

2 cos 2

4

x y

s

x y xy

xy

s

x y xy

  

  

 

  

  

       

 

(1.10-A2)

Eq (1.10-A2)  Eq (1.18b), we obtain:

 

   

2 2

2 2 2 2

21 sin 2 cos 2

2 2 4 4

p

x yx y xy

x y s xy s

x y xy x y xy

 

       

       

     

   

 maximum shear stresses:

2

2max min1 2 max

2 2 2

x y

xy

     

         

  (1.22)

Notes: a. Since  tan 2 tan 2s s    , hence there are also two maximum shear directions:  ,s s  

correspond to  max max,   , respectively, with s s   .

b. It is necessary to substitute one of the s values into Eq (1.18b) to determine which of the

maximum shear directions corresponds to the “” maximum shear stress max .

Substituting Eq (1.10-A2) into Eqs (1.18a,c):

cos 2 sin 2 2 2 2

cos 2 sin 2 2 2 2

s

s

x y x y x y

x s xy s

x y x y x y

y s xy s

 

 

         

         

 

 

      

       



max min1 2 ave

2 2 2

x y      

       (1.23)

The above results are illustrated in Fig 1.14. Note that the diagonal of a stress element toward which the shear

stresses act is called the shear diagonal. The shear diagonal of the element on which the maximum

 2

 1

 1

 2

 max

' ave

' ave

' ave

 max'

ave 45 45

45

p  

p  

 x

 y

 xy

FIGURE 1.14 Planes of principal stresses, maximum shear stresses and shear diagonal.

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 1/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESSES (Lecture)

Procedure for Drawing Mohr’s Circle

1. Establish a rectangular - coordinate system. Both stress scales must be identical.

2. Locate the center C of the circle on the horizontal -axis a distance   1

2 x y  from the origin.

3. Locate point A by coordinates  ,x xy  . These stresses may correspond to any face of an element such as

in Fig 1.15a. Nevertheless, it is usual to specify the stresses on the positive x face.

4. Draw a circle with center at C and of radius equal to CA.

5. Draw line AB through C; thus, point B will have coordinates  ,x xy  .

FIGURE 1.15 (a) Stress element; (b) Mohr’s circle of stress; (c) interpretation of positive shear stresses.

Notes: a. The angles on the circle are measured in the same direction as θ is measured in the stress element

(Fig 1.15a). However, an angle of 2θ on the circle corresponds to an angle of θ on the stress

element.

b. The state of stress associated with the original x and y planes corresponds to points A and B on the

circle, respectively.

c. Points lying on any diameter, such as A′ and B′, define states of stress w.r.t. x′-y′ coordinates rotated

relative to the original x-y coordinates through an angle θ, see Eq (1.18).

   

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

cos sin sin cos sin 2 cos 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

x y x y

x x y xy xy

x y

x y xy y x xy

x y x y

y x y xy xy

              

            

              

 

       

 

       

       

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 2/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

d. Points 1A and

1B on the circle locate the principal stresses  1 2,  . Their magnitudes and

directions are defined by Eqs (1.19) and (1.20):

2

2

1,2 max,min

1

2 2

21 tan

2

x y x y

xy

xy

p

x y

      

 

 

              

 

e. Points D and E represent the maximum shear stresses. Their magnitudes and directions are defined

by Eqs (1.21) and (1.22):

2

2 max min1 2 max

1

2 2 2

1 tan

2 2

x y

xy

x y

s

xy

      

  

               

      

  

f. The radius of the circle is: 2 2CA CF AF  where 2

x y

xy

CF

AF

 

 

  

, thus, the radius equals the

magnitude of the maximum shear stress max .

g. Mohr’s circle shows that the planes of maximum shear are always located at 45° from planes of

principal stress, as already indicated in Fig. 1.14 (Sec. 1.10).

0 

0 

0 

0 

0 

0 

0 

0 

FIGURE 1.11-A1 Two versions of pure shear.

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESSES PAGE 3/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

SPECIAL 2-D MOHR’S CIRCLES

Case Schematic Stress state Mohr’s circle

Uniaxial

tension

0 

x

y

0

0 0

0

x xy

yx y

 

  

          

0 0

2 p

0

2

0

0

center @ ,0 2

radius 2

      

   

01

2

0

max

0

2

90p

 

 

 

  

 

 

Pure shear

0 

x

y

0 

0

0

0

0

x xy

yx y

  

  

          

0 

0 

0 

2 p

 

0

center @ 0,0

radius 

 



01

02

0max

45p

 

 

 

 

  

   

x

y

0 

0 

0

0

0

0

x xy

yx y

  

  

          

0

 0

 

0 

2 p

 

0

center @ 0,0

radius 

 



01

02

0max

90p

 

 

 

 

  

   

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESSES PAGE 4/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Case Schematic Stress state Mohr’s circle

Biaxial

tension x

y

0 

0 

0

0

0

0

x xy

yx y

  

  

          

0 

degenerate circle

 0center @ ,0

radius 0

 



01 2

0max

~ undefinedp

  

 

  

  

Cylindrical

pressure

vessel x

y pr

t

2

pr

t

0 2

0

x xy

yx y

pr

t

pr

t

 

 

    

         

(surface only)

3 center @ ,0

4

radius 4

pr

t

pr

t

      

   

pr

t 2

p 

2

pr

t

4

pr

t

1

2

max

2

4

90p

pr

t

pr

t

pr

t

 

    

 

 

Spherical

pressure

vessel

x

y 2

pr

t

2

pr

t

0 2

0 2

x xy

yx y

pr

t

pr

t

 

 

    

         

(surface only)

pr

t

degenerate circle

center @ ,0 2

radius 0

pr

t

     

   

01 2

0max

~ undefinedp

  

 

  

  

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 5/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Example 1.3 Principal Stresses in a Member (Self-Study)

At a point in the structural member, the stresses are

80 MPa

40 MPa

30 MPa

x

y

xy

  

  

as represented in Fig. 1.16a. Employ Mohr’s

circle to determine:

a. the magnitude and orientation of the principal stresses and

b. the magnitude and orientation of the maximum shear stresses and associated normal stresses.

In each case, show the results on a properly oriented element and represent the stress tensor in matrix form.

FIGURE 1.16 (a) Element in plane stress; (b) Mohr’s circle of stress;

(c) principal stresses; (d) maximum shear stress.

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 6/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

EXAMPLE 1.4 Stresses in a Frame (Self-Study)

The biaxial stress state acting on an element of a loaded frame is shown in Fig 1.17a

28 MPa

14 MPa

0 MPa

x

y

xy

  

   

. Apply

Mohr’s circle graphically to determine the normal and shear stresses acting on a plane defined by 30   .

Check the results using Eq (1.18).

FIGURE 1.17. Example 1.4. (a) Element in biaxial stresses; (b) Mohr’s circle of stress;

(c) stress element for 30   .

Solution Mohr’s circle of Fig 1.17b describes the state of stress given in Fig 1.17a. Points 1A and 1B represent

the stress components on the x and y faces, respectively. The center and radius of the circle are,

 

  22

2

28 14 center: 7 MPa

2 2

28 14 radius: 0 21 MPa

2 2

x y

x y

xy

OC

CA CB

 

  

     

 

                  

, respectively.

Corresponding to the 30° plane within the element, it is necessary to rotate through 60° counterclockwise on the

circle to locate point A′. A 240° counterclockwise rotation locates point B′. Referring to the circle, we get

cos 2 7 21cos60 17.5 MPa

sin 2 21sin 60 18.19 MPa

cos 2 7 21cos60 3.5 MPa

x

x y

y

OC CA

CA

OC CB

 

 

 

 

       

       

      

Figure 1.17c indicates the orientation of the stresses. The results can be checked by applying Eq (1.18), using

the initial data as

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 7/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

cos 2 2 2

x y x y

x xy

      

    

   28 14 28 14 sin 2 cos60 17.5 MPa

2 2

sin 2 2

x y

x y xy

     

       

   

 28 14 cos 2 sin 60 18.19 MPa

2

cos 2 2 2

x y x y

y xy

      

      

    

   28 14 28 14 sin 2 cos60 3.5 MPa

2 2 

          

     

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 8/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

EXAMPLE 1.5 Cylindrical Vessel Under Combined Loads (Lecture)

A thin-walled cylindrical pressure vessel of 250-mm diameter and 5-mm wall thickness is rigidly attached to a

wall, forming a cantilever with the following loads being applied: internal pressure 1.2 MPap  , torque

3 kN-mT  , and direct force 20 kNP  (Fig. 1.18a).

a. Find the principal stresses and directions at point A of the cylindrical wall.

b. Determine the maximum shear stresses and the associated normal stresses at point A. Show the results on a

properly oriented element.

FIGURE 1.18. Example 1.5. Combined stresses in a thin-walled cylindrical pressure vessel:

(a) side view; (b) free body of a segment; (c) and (d) element A (viewed from top).

Solution The internal force resultants on a transverse section through point A are found from the equilibrium

conditions of the free-body diagram of Fig 1.18b. They are 20 kNV P  ,   20 0.4 8 kN-mM P AP  

and 3 kN-mT  . In Fig 1.18c, the combined axial, tangential, and shear stresses are shown acting on a small

element at point A. These stresses are (Tables 1.1 and C.1):

axial stress (bending):

 

 

3 3

33 3 3

250 10 8 10

2 32.6 MPa

250 10 5 10

2

b

Mr Mr

I r t 

 

 

    

      

   

shear stress (torque):

 

 

3 3

33 3 3

250 10 3 10

2 6.112 MPa

2 250 10 2 5 10

2

t

Tr Tr

J r t 

 

 

    

      

   

axial stress (internal pressure):

 

 

3 6

3

250 10 1.2 10

2 15 MPa

2 2 5 10 a

pr

t 

    

    

SECTION 1.11 MOHR’S CIRCLE FOR TWO-DIMENSIONAL STRESS PAGE 9/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

tangential stress (internal pressure):

 

 

3 6

3

250 10 1.2 10

2 2 30 MPa

5 10 a

pr

t  

    

     

We thus have:

 

32.6 15 47.6 MPa

30 MPa

6.112 MPa : shear-stress Wh siy gn convent" "? ion

x b a

y

xy t Hint

  

 

 

      

 

 

    

Note that for element A, 0Q  (Why? Hint: 1 st moment of the shaded area w.r.t. N.A.)

 direct shear stress: 0xz d

VQ

Ib    

a. The principal stresses are from Eq. (1.20):

  2 2

22

max,min 1,2

49.51 MPa47.6 30 47.6 30 47.6 30 6.112

28.09 MPa2 2 2 2

x y

xy

    

                  

   

To find the principal directions, we use Eq (1.19):

 1 1 2 17.42 6.1121 1

tan tan 72.62 2 47.6 30

xy

p

x y

 

 

       

           

To differentiate the maximum & minimum principal directions, we choose 17.4p    and use Eq (1.18a):

   

cos 2 sin 2 2 2

47.6 30 47.6 30 cos 2 17.4 sin 2 17.4 49.51 MPa

2 2

x y x y

x xy

xy

       

    

                 

 17.4

72.6

p

p

       

b. The maximum shear stresses are from Eq (1.22):

  2 2

22

max

47.6 30 6.112 10..71 MPa

2 2

x y

xy

   

               

  

To locate the maximum shear planes, we use Eq (1.21):

  1 1

27.61 1 47.6 30 tan tan

117.62 2 2 2 6.112

x y

s

xy

  

      

          

Choosing 27.6s   and applying Eq (1.18b), then

      47.6 30

sin 2 cos 2 sin 2 27.6 6.112 cos 2 27.6 10.71MPa 2 2

x y

x y s xy s

      

              

Hence, 27.6

27.6

s

s

      

Equation (1.23) yields the average (or mean) stress, which is also the normal stresses associated with max at

point A: ave

47.6 30 38.8 MPa

2 2

x y   

      

These stresses are shown in their proper directions in Fig 1.18d.

  • 1.9 Plane-Stress (2-D) Transformation (2018-01-25)
  • 1.10 2-D Principal Stresses-Maximum Shear Stress (2018-01-25)
  • 1.11 2-D Mohr’s Circle (2018-01-25)
    • 1.11 2-D Mohr’s Circle-Part 1 (2018-01-25)
    • 1.11 2-D Mohr’s Circle-Part 2 (2018-01-25)
    • 1.11 2-D Mohr’s Circle-Part 3 (2018-01-25)

__MACOSX/stress/._Ch 1 Analysis of Stress-Part 3 (2018-01-25).pdf

stress/Ch 1 Analysis of Stress-Part 4 (2018-01-25).pdf

SECTION 1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION PAGE 1/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION (Lecture)

Def: direction cosines: The direction cosines of a vector: ˆ ˆ ˆ x y zV V V  V i j k w.r.t. the x-y-z coordinates are:

 

 

 

cos cos ,

cos cos ,

cos cos ,

x

y

z

V l x

V m y

V n z

   

  

          

V V

V V

V V

, where 2 2 2 V x y zV V V   (1.24)

FIGURE 1.19 Stress components on a tetrahedron.

If V is a unit normal vector n̂ to a plane ABC (Fig 1.19), i.e., ˆ 1n , then

ˆ ˆ ˆ ˆ ˆ ˆˆ x y zn n n l m n     n i j k i j k (1.12-A1)

and 2 2 2 1l m n   (1.25)

 

 

 

ˆ ˆ ˆ ˆ ˆˆ

ˆ ˆ ˆ ˆ ˆˆ

ˆ ˆ ˆ ˆ ˆˆ

QAB ABC ABC ABC

QAC ABC ABC ABC

QBC ABC ABC ABC

l m n l

l m n m

l m n n

                              

n i i j k i

n j i j k j

n k i j k k

(a)

Def: stress traction vector (or surface traction): ˆ ˆ ˆ x y zp p p  p i j k ~ unit:

2

N Pa

m

   

  (1.12-A2)

force balance on the plane ABC:

 

 

 

0 :

0 :

0 :

x x ABC x QAB xy QAC xz QBC x xy xz ABC

y y ABC xy QAB y QAC yz QBC xy y yz ABC

z z ABC xz QAB yz QAC z QBC xz yz z ABC

F p l m n

F p l m n

F p l m n

     

     

     

              

            

           

SECTION 1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION PAGE 2/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

thus, in scalar-component form:

x x xy xz

y xy y yz

z xz yz z

p l m n

p l m n

p l m n

  

  

  

    

      

(1.26)

in tensor-index form: i ji j ij jp n n    , 1,2,3i j  (1.12-A3)

in vector-matrix form:  ˆp n τ (1.12-A4)

Special Case: Surface Tractions in Plane Problems

In 2-D (x-y plane), 0n  , then Eq (1.26) reduces to: x x xy

y xy y

p l m

p l m

 

 

  

  (1.12-A5)

Consider now a Cartesian x′-y′-z′ coordinate system with the same origin Q as the afore-mentioned x-y-z

coordinates. If the unit normal vector n̂ is placed to coincide with the x′-axis, then a set of direction cosines:

       1 1 1, , cos , ,cos , ,cos ,l m n x x x y x z     

will be formed. By the same token, one can place n̂ to coincide with the y′- and z′-axes, respectively, to form

another two sets of direction cosines:

       2 2 2, , cos , ,cos , ,cos ,l m n y x y y y z      &        3 3 3, , cos , ,cos , ,cos ,l m n z x z y z z     

Mnemonics: The relations among the two Cartesian coordinate systems and the direction cosines are listed in

Table 1.2.

TABLE 1.2 Notation for Direction Cosines.

Note that the sets:

       

       

       

1 2 3

1 2 3

1 2 3

, , cos , ,cos , ,cos ,

, , cos , ,cos , ,cos ,

, , cos , ,cos , ,cos ,

l l l x x x y x z

m m m y x y y y z

n n n z x z y z z

       

       

      

are the respective direction cosines of

the Cartesian x-, y- and z-axes w.r.t. the Cartesian x′-y′-z′ coordinate system. By placing the unit normal vector

n̂ to coincide with the x′-, y′- and z′-axes, respectively, Eq (1.26) becomes:

1 1 1

1 1 1

1 1 1

x x xy xz

y xy y yz

z xz yz z

p l m n

p l m n

p l m n

  

  

  

    

      

2 2 2

2 2 2

2 2 2

x x xy xz

y xy y yz

z xz yz z

p l m n

p l m n

p l m n

  

  

  

    

      

3 3 3

3 3 3

3 3 3

x x xy xz

y xy y yz

z xz yz z

p l m n

p l m n

p l m n

  

  

  

    

      

(1.12-A6)

Following the same logic, we can project the components of the above-mentioned stress traction vector ˆ ˆ ˆ

x y zp p p p  i j k in the x′-, y′- and z′-axes, respectively, then take force balances along the x′-, y′- or z′-

direction to obtain:

SECTION 1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION PAGE 3/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1 1 1

2 2 2

3 3 3

x x y z

x y x y z

x z x y z

p l p m p n

p l p m p n

p l p m p n

 

 

    

      

1 1 1

2 2 2

3 3 3

x y x y z

y x y z

y z x y z

p l p m p n

p l p m p n

p l p m p n

 

 

    

      

1 1 1

2 2 2

3 3 3

x z x y z

y z x y z

x x y z

p l p m p n

p l p m p n

p l p m p n

 

 

    

      

(1.27)

Substitute Eq (1.12-A6) into Eq (1.27), we get:

       

     

2 2 2

1 1 1 1 1 1 1 1 1

1 2 1 2 1 2 1 2 2 1 2 1 1 2 1 2 2 1

1 3 1 3 1 3 1 3 3 1 1 3 3 1 1 3 3 1

2 2 2

2 2 2

2

2

x x y z xy yz xz

x y x y z xy yz xz

x z x y z xy yz xz

y x y z

l m n l m m n l n

l l m m n n l m l m m n m n l n l n

l l m m n n l m l m m n m n l n l n

l m n

      

      

      

    

 

 

     

        

        

     

       

2 2 2 2 2 2

2 2 2

3 3 3 3 3 3 3 3 3

2 3 2 3 2 3 2 3 3 2 2 3 3 2 2 3 3 2

2

xy yz xz

z x y z xy yz xz

y z x y z xy yz xz

l m m n l n

l m n l m m n l n

l l m m n n l m l m m n m n l n l n

 

      

      

 

     

  

               

(1.28)

Note: Owing to the symmetry of the stress tensor: ij ji  , only 6 of the 9 stress components thus developed

are unique.

The direction cosine components in Table 1.2 can be collected to form:

direction cosine (or orthogonal rotation) matrix:   11 12 13 1 1 1

21 22 23 2 2 2

31 32 33 3 3 3

ij

l l l l m n

l l l l l m n

l l l l m n

       

             

l (1.12-A7)

Note: The direction cosine matrix is NOT symmetric: ij jil l

The scalar-component form of Eq (1.28) can then be expressed in the short-hand forms:

tensor-index notation: rs ir js ij

rs ri sj ij

l l

l l

 

 

  

 (1.29)

or vector-matrix notation:      

      

T

T

   



τ l τ l

τ l τ l (1.12-A8)

Finally, it can be proven the components of the direction cosine (or orthogonal rotation) matrix have the

following properties:

2 2 2

1 1 1

2 2 2

2 2 2

2 2 2

3 3 3

1

1

1

l m n

l m n

l m n

    

      

and

1 2 1 2 1 2

2 3 2 3 2 3

1 3 1 3 1 3

0

0

0

l l m m n n

l l m m n n

l l m m n n

   

      

     1

l l T

 (1.30)

Similarly,

2 2 2

1 2 3

2 2 2

1 2 3

2 2 2

1 2 3

1

1

1

l l l

m m m

n n n

    

      

and

1 1 2 2 3 3

1 1 2 2 3 3

1 1 2 2 3 3

0

0

0

l m l m l m

m n m n m n

n l n l n l

   

      

(1.12-A9)

SECTION 1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION PAGE 4/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Special Case: In-plane (or 2-D) stress tensor transformation

 

cos sin 0

sin cos 0

0 0 1

ijl

 

 

   

       

l and  

cos sin 0

sin cos 0

0 0 1

T

ji l

 

 

   

      

l (1.12-A10)

Figure 1.12-A1 Two-dimensional (orthogonal) rotational transformation.

then in component notation:

 

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

sin cos sin cos cos sin sin 2 cos 2 2

c

x y x y

x x y xy xy

x y x y

y x y xy xy

y x

x y x y xy xy

z z

x z zx

              

              

              

 

 

 

 

       

       

       

 os sin

cos sin

yz

y z yz zx

  

    

            

 

(1.12-A11)

or  

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

sin cos sin cos cos sin sin 2 cos 2 2

x y x y

x x y xy xy

x y x y

y x y xy xy

x y

xy x y xy xy

              

              

             

             

             

          

cos sin

cos sin

z z

yz yz zx

zx zx yz

 

   

    

        

     

  

(1.12-A12)

Mnemonics: The equations in each pair of the relations in the 2-D Special Case are interchangeable by

switching the primed and unprimed and by replacing  with .

SECTION 1.12 THREE-DIMENSIONAL STRESS TRANSFORMATION PAGE 5/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Example: 2-D Cartesian vs. polar coordinates

If the 2-D primed coordinate system  ,x y  is the polar coordinate system  ,r  , then

 

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

sin cos sin cos cos sin sin 2 cos 2 2

x y x y

r x y xy xy

x y x y

x y xy xy

y x

r x y xy xy

              

              

              

       

  

      

       

(1.12-A13)

or

 

2 2

2 2

2 2

cos sin 2 sin cos cos 2 sin 2 2 2

sin cos 2 sin cos cos 2 sin 2 2 2

sin cos sin cos cos sin sin 2 cos 2 2

r r x r r r

r r y r r r

r xy r r r

    

    

   

              

              

              

       

  

      

      

(1.12-A14)

which are the same as Eqs (1.18) and (1.9-A1) in Sec 1.9.

Large Deformation/Finite Elasticity Theory (Option)

Following the principles of Small Deformation/Infinitesimal Elasticity Theory, the previous definitions for the

stress tensor and traction vector do not make a distinction between the deformed and undeformed (or

reference) configurations of the body since such a distinction only leads to small modifications that are

considered higher-order effects and are normally neglected. However, for Large Deformation/Finite Elasticity

Theory, sizeable differences exist between these configurations and several stress tensors have been defined.

For instance, both directions of the force and surface normal in the true (Cauchy) stress tensor is defined

based on the deformed configuration. Other examples of stress tensors defined in Finite Elasticity are:

Kirchhoff stress tensor, the 1 st & 2

nd Piola-Kirchhoff stress tensors and Biot stress tensor. In addition,

there are also several incremental stress updates, also called stress rates: Jaumann, Green-Naghdi, Oldroyd,

Trusdell, convective, etc. It should be noted that since the physical meaning of the integral of stress with

respect to strain is the strain energy, the selection of proper pair of stress and strain tensors should obey the

rule of stress objectivity. That is, the superimposition of rigid-body motion on the deformed configuration

should not alter the stress state and should produce no extra strain energy.

SECTION 1.13 THREE-DIMENSIONAL PRINCIPAL & MAXIMUM SHEARING STRESSES PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.13 THREE-DIMENSIONAL PRINCIPAL & MAXIMUM SHEARING STRESSES (Lecture)

Let ij j p jn n  or   ˆ ˆτ n np or

x xy xz

pyx y yz

zx zy z

l l

m m

n n

  

   

  

           

              

(1.13-A1)

 stress tensor eigenvalue problem:

0

0

0

px xy xz

pyx y yz

pzx zy z

l

m

n

   

   

   

           

               

(1.31)

non-trivial solution, i.e.,

0

0

0

l

m

n

       

           

, we must have:

     3 2

1 2 3det det det 0

px xy xz

ij p ij p p p p pyx y yz

pzx zy z

I I I

   

          

   

   

                

Iτ (1.32 & 33)

here  

         1 2 3

1 2 3

eigenvalues: , , the extreme stationary values of stresses

ˆ ˆ ˆ ˆeigenvector: , ,

p

p

or      



th

th

principal stresses

principal stress directions

n n

p

p n n

In the cubic equation, Eq (1.33),  1 2 3, ,I I I are the three fundamental stress invariants:

Def : 1 st stress invariant: 1 1 2 3x y zI            (1.34a)

Def: 2 nd

stress invariant:

 

 

 

2

1 2 2 3 3

2

1

2 2

1

2

x xy y yz z zx

ii jj ij ji

yx y zy z xz x

x y y z z x xy yx yz zy zx

xy yz x

xz

z

I      

         

           

    

 

    

  

  

 

  

 (1.34b)

Def: 3 rd

stress invariant:

 

   

 

3

1

2 2 2

2 3

1 1 1 det det

3 6

2

2

x xy xz

ij yx y yz ij jk ki ij ji kk ii jj kk

zx zy z

x y z xy yz zx yx zy xz x yz zy y zx xz z xy

x yz y xz z xyxy y

x

z

y

zx

I

  

            

  

      

        

          

  

       

     

 

 

σ

(1.34c)

Def: (3-D) maximum shearing stress: 1 2 2 3 3 1

max max 0 2 2 2

      

       

  (1.13-A2)

SECTION 1.13 THREE-DIMENSIONAL PRINCIPAL & MAXIMUM SHEARING STRESSES PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Special Case: Plane-Stress Problems

plane-stress condition: 0z xz zx yz zy           3

3 3 0n   &      3 3 3

1 2 ˆ ˆˆ n n n i j (1.13-A3)

Eq (1.31) becomes

 

 

1

2

0

0

p

px xy

p pyx y

n

n

  

  

          

      

    2 2det 0 px xy

p p xyx y x y pxy y

         

  

        

  (1.13-A4)

1 1

2

3

2

2 1 2

0

x y

x xy

x y xy

yx y

I

I

I

   

      

 

     

   

  

Note: Typo in textbook (1.35)

2 2

2 11 1,2 2

2 2

21 2 1

max 2

: 2 2 2 4

: 2 2 4

x y x y

xy

x y

xy

I I I

I I

     

    

                   

        

principal stresses

maximum shearing stress

(1.20 & 22)

   

   

1

1

21 : tan two values

2

1 : tan two values

2 2

xy

p

x y

y x

s

xy

 

 

  

 

 

   

principal stress directions

maximum shearing stress directions

(1.19 & 21)

FIGURE 1.13-A1 The principal directions p and the maximum shear directions s .

SECTION 1.13 THREE-DIMENSIONAL PRINCIPAL & MAXIMUM SHEARING STRESSES PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Eigendecomposition & Diagonalization (Self-Study)

Def : eigen matrix:        1 2 3ˆ ˆ ˆn n n n   

(1.13-A5)

then       1 n n n

T or

  (1.13-A6)

&      1

2

3

0 0

0 0

0 0

n τ n

         

(1.13-A7)

Example 1.7 Three-Dimensional Stress in a Machine Component (Self-Study)

The stress tensor at a point in a machine element with respect to a Cartesian coordinate system is given by the

following array:

 

50 10 0

10 20 40 MPa

0 40 30

x xy xz

yx y yz

zx zy z

  

  

  

       

          

τ (f)

Determine the state of stress and  1 2 3, ,I I I for an x′, y′, z′ coordinate system defined by rotating x, y through an

angle of θ  45° counterclockwise about the z-axis (Fig 1.21a).

FIGURE 1.21 Direction cosines for 45   .

SECTION 1.14 NORMAL AND SHEAR STRESSES ON AN OBLIQUE PLANE PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.14 NORMAL AND SHEAR STRESSES ON AN OBLIQUE PLANE (Self-Study)

Normal & Shear Tractions

Figure 1.14-A1 Comparison of general and principal stress states

p

(a) (b)

FIGURE 1.22 (a) Element in triaxial stress state; (b) traction vector decomposition.

in a general coordinate system (x-y-z or ˆ ˆ ˆ- -i j k ):  τ x xy xz

ij xy y yz

zx zy z

  

   

  

   

       

~ 6 stress components

in principal coordinate system (      1 2 3ˆ ˆ ˆ- -n n n ):  

1

2

3

0 0

0 0

0 0

ij

 

   

      

 ~ 3 stress components

Note: For the principal coordinate system, all shearing stresses vanish and thus the state includes only normal

stresses, which are the principal stresses themselves. That is to say, under transformation to principal

axes, the matrix form of the stress tensor will reduce to a diagonal form with the principal stresses as its

diagonal components.

Consider Fig 1.22, an element in a triaxial stress state:    

   

1 2 3

0 0 0

x y z

xy xz yz

     

  

  



, the traction vector p

on a surface with a unit normal vector ˆ ˆ ˆˆ l m n  n i j k , from Eq (1.26), becomes:

   1 2 3x y z p p p l m n   (a)

SECTION 1.14 NORMAL AND SHEAR STRESSES ON AN OBLIQUE PLANE PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

With the aid of Fig 1.22b, p can also be decomposed into a normal traction vector σ and a shear traction

vector τ as:

 p σ τ with Eq (1.12-A4)  

2 2

ˆ

Pythagorean theorem

    

 

normal traction :

shear traction :

σ p n

τ p σ (1.14-A1)

Eq (a) into Eq (1.14-A1a)  2 2 2 2 2 2 2 2 2

1 2 3p l m n         (1.36)

Eq (1.28a)  2 2 2 2x y z xyl m n      = yzlm  xzmn  ln

 2 2 2

1 2 3l m n      (1.37)

Alternatively, Eqs (1.14-A1a) & (1.12-A3)  2 2 2

1 2 3i i ij jp n n l m n       =

Substitute Eq (1.37) into Eq (1.36), we get:

    2

2 2 2 2 2 2 2 2 2 2 2 2

1 2 3 1 2 3p l m n l m n               (1.38)

Since Eq (1.25): 2 2 2 1l m n          1 2

2 2 22 2 2 2 2 2

1 2 2 3 3 1l m m n n l              

(1.39)

Note: Eq (1.39) indicates that if the principal stresses are all equal: 1 2 3    , the shear stress  vanishes,

regardless of the choices of the direction cosines:  , ,l m n .

Finally, for a general stress state (Fig 1.14-A1a), we have:

 

     

2 2 2

1 2 2 2 2

2

2x y z xy yz xz

x xy xz xy y yz xy yz z

l m n lm mn ln

l m n l m n l m n

      

          

       

             

=

(1.40 & 41)

Lamé’s Stress Ellipsoid (Option)

FIGURE 1.23 Stress ellipsoid.

22 2

1 2 3

1yx z pp p

  

           

      (1.39)

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 1/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES (Lecture)

Spherical Stress Tensor

Def: mean (or average or hydrostatic) stress:

1 2 3 1

3 3 3

x y z

m

I     

      (1.14-S1)

Def: spherical stress tensor:

0 0

0 0

0 0

m

ij m ij m

m

p

  

   

      

(1.14-S2)

where Kronecker delta ij is defined as:

Def:  

   

1 0 0 1 if

1 00 0 if

0 0 1

unit identity matrixIij

i j no sum or

i j 

     

       

(1.14-S3)

fundamental spherical stress invariants

Def: 1 st spherical stress invariant:

1 1 2 3 13x y z mI p p p p p p I        (1.14-S4a)

Def: 2 nd

spherical stress invariant:

2

2 2 1

1 2 2 3 3 1

0 00

0 00

3 3

x zy

y xz

x y y z z x m

p pp I

p pp

I p p p p p p p p p p p p 

  

       

(1.14-S4b)

Def: 3 rd

spherical stress invariant:

3 3 1

3 1 2 3

0 0

det 0 0 27

0 0

x

ij y x y z m

z

p I

I p p p p p p p p

p

        (1.14-S4c)

Note: As presented in next topic: Chapter 2: Strain & Material Properties, the 2 nd

spherical stress

invariant 2

21 2 3

3 m

I I   is directly proportional to the dilatational component of strain energy.

Deviatoric Stress Tensor

Def: deviatoric stress tensor:

x m xy xz

ij ij ij ij m ij yx y m yz

zx zy z m

s p

   

       

   

   

         

(1.14-S5)

Notes: a. total stress = spherical stress + deviatoric stress: ij ij ijp s   (1.14-S6)

b. The spherical stress ij m ijp   is an isotropic stress tensor. That is, its components are the same

and equal to the mean stress m in all coordinate systems and the principal spherical stress

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 2/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

directions are arbitrary.

c. Since the spherical stress ij m ijp   is an isotropic stress tensor. That is, its components are the

same in all coordinate systems and principal directions are arbitrary. Thus, the principal directions

of the deviatoric stress tensor ij ij ij ij m ijs p       are the same as those of the stress tensor ij

itself.

fundamental deviatoric stress invariants

Def: 1 st deviatoric stress invariant:

     1 3x y z x m y m z m x y z mJ s s s                      

From Eq (1.14-S1)  1 0J  (1.14-S7a)

Def: 2 nd

deviatoric stress invariant:

       

     

2 2 2 2 2 2

2

2 2 2 2

1 2 2 3 3 1 1 2

1 6

6

1 1

6 3

x y y z z x xy yz zxJ

I I

        

     

            

          

(1.14-S7b)

Def: 3 rd

deviatoric stress invariant:     3

3 1 2 3 1 1 2 3

2 1

27 3 m m mJ I I I I            (1.14-S7c)

(a) (b)

FIGURE 1.24 (a) An octahedron containing 8 octahedral stress planes;

(b) On an octahedral plane: l m n  i.e., 1cos 54.74       .

Octahedral Stresses

Def: principal stress space: A 3-D stress space (Note: not a physical space) with the three principal stresses

 1 2 3 , ,   (that is, their principal directions:

 1 n̂ ,

 2 n̂ ,

 3 n̂ ) as the coordinate axes.

Def: In the principal stress space, a plane whose normal vector makes equal angles with each of the principal

axes (i.e. having direction cosines equal to 1

cos54.74 3

  ) is called an octahedral stress plane. There

are a total of eight octahedral stress planes, as shown in Fig 1.24. The shear and normal components of

the stress tensor on these planes are called octahedral shear stress oct and octahedral normal stress

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 3/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

oct , respectively:

      2 2 2

oct 1 2 2 3 3 1 2

1 2 3 1 oct

1 2

3 3

3 3 m

J

I

      

    

       

 

     

(1.43 & 44)

Von Mises & Tresca Stresses

Def: von Mises (or effective or equivalent) stress:

       

     

2 2 2

vM

2 2 2

1 2 2 3 3 1 oct 2

1 6

2

1 3 3

2 2

x y y z z x xy yx yz zy zx xz

J

            

      

        

       

(1.14-S8)

Def: Tresca stress:  Tr max 1 2 2 3 3 12 max , ,             (1.14-S9)

Notes: a. It can be proven that: Tr vM

max oct

 

 

 

 (1.14-S10)

b. vM (or equivalently, oct and 2J ) and Tr (or equivalently, max ) play significant roles in Failure

by Yielding (Ch 4).

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 4/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Example 1.6 Three-Dimensional Stress in a Hub (Lecture)

A steel shaft is to be force fitted into a fixed-ended cast-iron hub. The shaft is subjected to a bending moment

M, a torque T, and a vertical force P. Suppose that at a point Q in the hub, the stress field is, as shown below,

represented by the matrix:

19 4.7 6.45

4.7 4.6 11.8 MPa

6.45 11.8 8.3

x xy xz

ij yx y yz

zx zy z

  

   

  

        

           

a. Determine the principal stresses  1 2 3, ,   and their corresponding principal directions       1 2 3ˆ ˆ ˆ, ,n n n .

b. Prove eigendecomposition and diagonalization numerically by performing the matrix multiplication:

             1 2 3 1 2 3ˆ ˆ ˆ ˆ ˆ ˆn n n n n n T

       

.

c. Obtain the associated spherical and deviatoric stress tensors.

d. Find the invariants of the original, spherical and deviatoric stress tensors, respectively.

e. Find the maximum shear, von Mises, Tresca and octahedral normal and shear stresses.

f. Plot the 3-D Mohr’s circle and indicate the locations of the principal stresses  1 2 3, ,   and the maximum

shear max .

FIGURE 1.20 (a) Hub-shaft assembly. (b) Element in three-dimensional stress.

Solution

a. Principal stresses and directions.

For the stated stress tensor, the stress tensor eigenvalue problem, Eq (1.13-A1), becomes:

19 4.7 6.45

4.7 4.6 11.8

6.45 11.8 8.3

x xy xz

pyx y yz

zx zy z

l l l

m m m

n n n

  

   

  

                   

                          

The eigenvalues (i.e., principal stresses) and the corresponding eigenvectors (i.e., principal directions) can be

obtained using the Matlab command: eig. The results are:

1

2

3

11.6178

9.0015 MPa

25.3163

       

           

&      

1 2 3

1 2 3

1 2 3

1 2 3

0.0266 0.6209 0.7834

ˆ ˆ ˆ 0.8638 0.3802 0.3306

0.5031 0.6855 0.5262

n n n

l l l

m m m

n n n

             

           

(a1)

Verification. Substituting the above direction-cosine matrix into the stress tensor eigenvalue problem,

Eq (1.31), the following relations should be satisfied.

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 5/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

For 1 11.6178 MPa  :

 

 

 

1 1 1

1 1 1

1 1 1

19 11.6178 4.7 6.45 0

4.7 4.6 11.6178 11.8 0

6.45 11.8 8.3 11.6178 0

l m n

l m n

l m n

                

  1

1 1 1 ˆ ˆ ˆ ˆ ˆ ˆˆ 0.0266 0.8638 0.5031l m n      n i j k i j k Note: 2 2 2

1 1 1 0l m n   (a2)

For 2 9.0015 MPa   :

 

 

 

2 2 2

2 2 2

2 2 2

19 9.0015 4.7 6.45 0

4.7 4.6 9.0015 11.8 0

6.45 11.8 8.3 9.0015 0

l m n

l m n

l m n

                

  2

2 2 2 ˆ ˆ ˆ ˆ ˆ ˆˆ 0.6209 0.3802 0.6855l m n     n i j k i j k Note: 2 2 2

2 2 2 0l m n   (a3)

For 3 25.3163 MPa   :

 

 

 

3 3 3

3 3 3

3 3 3

19 25.3163 4.7 6.45 0

4.7 4.6 25.3163 11.8 0

6.45 11.8 8.3 25.3163 0

l m n

l m n

l m n

                

  3

3 3 3 ˆ ˆ ˆ ˆ ˆ ˆˆ 0.7834 0.3306 0.5262l m n      n i j k i j k Note: 2 2 2

3 3 3 0l m n   (a4)

b. Eigendecomposition and diagonalization.

             1 2 3 1 2 3

0.0266 0.8638 0.5031 19 4.7 6.45 0.0266 0.6209 0.7834

ˆ ˆ ˆ ˆ ˆ ˆ 0.6209 0.3802 0.6855 4.7 4.6 11.8 0.8638 0.3802 0.3306

0.7834 0.3306 0.5262 6.45 11.8 8.3 0.5031 0.6855 0.5262

n n n n n n T

                    

                

       

               1

1 2 3 1 2 3

2

3

11.6178 0 0 0 0

ˆ ˆ ˆ ˆ ˆ ˆ 0 9.0015 0 MPa 0 0

0 0 25.3163 0 0

n n n n n n T

             

              

(b1)

c. Spherical and deviatoric stress tensors.

From Eq (1.14-S2), we have:

mean (or average or hydrostatic) stress: 19 4.6 8.3

7.5667 MPa 3 3

x y z

m

   

         (c1)

The spherical stress tensor can be obtained using Eq (1.14-S2):

0 0 7.5667 0 0

0 0 0 7.5667 0 MPa

0 0 0 0 7.5667

m

ij m

m

p

       

             

(c2)

The deviatoric stress tensor can be obtained using Eq (1.14-S5):

11.4333 4.7 6.45

4.7 12.1667 11.8 MPa

6.45 11.8 0.7333

x m xy xz

ij yx y m yz

zx zy z m

s

   

   

   

        

             

(c3)

d. Invariants.

The invariants of the stress tensor can be obtained from Eqs (1.34a,b,c):

1 1 2 3 22.7 MPaI        (d1)

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 6/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

  2

2 1 2 2 3 3 1 170.8 MPaI           (d2)

  3

3 1 2 3 2,647.5 MPaI     (d3)

The invariants of the spherical stress tensor can be obtained from Eqs (1.14-S4a,b,c):

1 1 22.7 MPaI I   (d4)

  2

21 2 171.7633 MPa

3

I I    (d5)

  3

31 3 433.2253 MPa

27

I I    (d6)

The invariants of the deviatoric stress tensor can be obtained from Eqs (1.14-S7a,b,c):

1 0J  (d7)

        2 2 2 2

2 1 2 2 3 3 1

1 342.5758 MPa

6 J             

  (d8)

      3

3 1 2 3 488.5897 MPam m mJ           (d9)

e. Maximum shear, von Mises, Tresca & octahedral stresses.

The maximum shear stress can be obtained from Eq (1.13-A2):

1 2 2 3 3 1

max max , , 18.467 MPa 2 2 2

      

       

  (e1)

The octahedral stresses can be obtained from Eqs (1.43 & 44):

oct 2

oct

2 octahedral shear stress: 15.1124 MPa

3

octahedral normal stress: 7.5667 MPam

J

 

  

    

(e2)

The von Mises (or effective or equivalent) stress can be obtained from Eq (1.14-S8):

vM oct

3 32.0582 MPa

2    (e3)

The Tresca stress can be obtained from Eq (1.14-S8):

Tr max2 36.9341MPa    (e4)

SECTION 1.14a SPHERICAL, DEVIATORIC, OCTAHEDRAL, VON MISES & TRESCA STRESSES PAGE 7/7

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

f. 3-D Mohr’s circles.

 max

=1.8.467 MPa

 1 =11.6178 MPa

2 =9.0015 MPa

3 =23.3163 MPa

SECTION 1.15 MOHR’S CIRCLES IN THREE DIMENSIONS PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

1.15 MOHR’S CIRCLES IN THREE DIMENSIONS (Option)

FIGURE 1.25 Triaxial state of stress: (a) wedge; (b) planes of maximum shear stress.

FIGURE 1.26 (a–c) Views of elements in triaxial stresses on different principal axes;

(d) Mohr’s circles for three-dimensional stress.

SECTION 1.15 MOHR’S CIRCLES IN THREE DIMENSIONS PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

Equations of Three Mohr’s Circles for Stress:

direction cosines: 2 2 2 1l m n  

normal traction: 2 2 2

1 2 3l m n     

shear traction:       2 2 22 2 2 2 2 2 2 2 2 2 2 2 2 2

1 2 3 1 2 2 3 3 1l m n l m m n l n                   

(a)

Solve Eq (a) 

  

  

  

  

  

  

2

2 32

1 2 1 3

2

3 12

2 3 2 1

2

1 22

3 1 3 2

0

0

0

l

m

n



  

   

   

   

  

   

 

     

     

   

         

(Note: typos in textbook) (1.46)

Without loss in generality, assume 1 2 3    , we have:

Mohr’s circles of stresses:

  

  

  

2

2 3

2

3 1

2

1 2

0

0

0

   

   

   

     

        

(Note: typos in textbook) (b)

     

     

     

2 2 22 1 1 2 3 23 2 32 4max

2 2 22 1 1 1 3 13 1 32 4max

2 2 22 1 1 1 2 12 1 22 4max

     

     

     

         

         

        

(Note: typos in textbook) (1.47)

absolute maximum shearing stress:     1 3 max 13 max 2a

   

   (1.45)

SECTION 1.15 MOHR’S CIRCLES IN THREE DIMENSIONS PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 1 – ANALYSIS OF STRESS

EXAMPLE 1.8 Analysis of Three-Dimensional Stresses in a Member

The state of stress on an element of a structure is illustrated in Fig. 1.27a. Using Mohr’s circle, determine

a. the principal stresses,

b. the maximum shearing stresses; Show results on a properly oriented element; Also,

c. apply the equations developed in Sec 1.14 to calculate the octahedral stresses.

FIGURE 1.27 Example 1.8. (a) Element in three-dimensional stress;

(b) Mohr’s circles of stress; (c) stress element for 26.56p  .

  • 1.12 3-D Stress Transformation (2018-01-25)
  • 1.13 3-D Principal & Maximum Shear Stresses (2018-01-25)
  • 1.14 Normal-Shear Stresses on Oblique Plane (2018-01-25)
  • 1.14a Spherical-Deviatoric-Octahedral-vonMises-Tresca Stresses (2018-01-25)
  • 1.15 3-D Mohr’s Circles (2018-01-25)

__MACOSX/stress/._Ch 1 Analysis of Stress-Part 4 (2018-01-25).pdf

stress/Ch 2 Strain & Material Properties-Part 1 (2018-01-25).pdf

SECTION 2.2 DEFORMATION PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

CHAPTER 2: STRAIN & MATERIAL PROPERTIES

2.2 Deformation (Lecture)

FIGURE 2.1 Two-dimensional deformation: planar displacement and strain in a body.

Displacement Gradient Tensor (Option)

FIGURE 2.2-A1 General deformation between two neighboring points

As shown in figure above, the displacement vectors representing the movements of points Po and P in the

undeformed state to points oP and Pare o u and u , respectively. From vector algebra, we have:

        o o o o o

P P P P P P P P PP or  o u + r r + u

 relative position vector:      o r r r u u (2.2-A1)

Since P and Po are neighboring points, we can use a Taylor series expansion around point Po to express the

components of u as:

SECTION 2.2 DEFORMATION PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

o

x y z

o

x y z

o

x y z

u u u u u r r r

x y z

v v v v v r r r

x y z

w w w w w r r r

x y z

       

       

      

       

  

(2.2-A2)

where  , ,u v w and  , ,x y zr r r are the Cartesian components of the displacement and position vectors, u and

r , respectively. Hence, Eq (2.2-A1) becomes:

in scalar- component notation:

o

x x x x y z

o

y y y x y z

o

z z z x y z

u u u r r r u u r r r

x y z

v v v r r r v v r r r

x y z

w w w r r r w w r r r

x y z

           

       

          

           

  

or in tensor-index notation: ,i i j jr u r 

or in vector-matrix notation:  r u r

(2.2-A3)

Here, the displacement gradient tensor: ,i j

u u u

x y z

v v v u

x y z

w w w

x y z

              

                    

u (2.2-A4)

Notes: a. The higher-order terms of the Taylor series expansion have been dropped since the components of

u and r are small. This approximation is only good for the so-called Small Displacement (or

Deformation or Strain) Theory (or Infinitesimal Elasticity Theory). On the contrary, if u is large,

Large Displacement (or Deformation) or Finite Strain Theory, which is also called Finite Elasticity

Theory, should be used instead.

b. Principle of Superposition. The small displacement assumption leads to one of the basic

fundamentals of solid mechanics, called the principle of superposition. This principle is valid

whenever the quantity (stress or displacement) to be determined is a linear function of the loads

that produce it. For the foregoing condition to exist, the material must also be linearly elastic (see

Sec 2.9). In such situations, the total quantity owing to the combined loads acting simultaneously

on a member may be obtained by determining separately the quantity attributable to each load and

combining the individual results. Clearly, superposition cannot be applied to (post-yielding) plastic

deformations. The main motivation for superposition is the replacement of a complex load

configuration by two or more simpler loads.

SECTION 2.3 STRAIN DEFINED PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.3 STRAIN DEFINED (Lecture)

FIGURE 2.2 Normal strain in a prismatic bar: (a) undeformed state; (b) deformed state.

(1-D) uniaxial (normal) strain: 0 0

0 0

L L

L L

 

   (2.2)

1-D normal strain: 0

limx x

u du

x dx 

 

  

 (2.1)

Plane (2-D) Strains

FIGURE 2.3 Strain components x , y and xy in the x-y plane.

SECTION 2.3 STRAIN DEFINED PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

FIGURE 2.4 Deformations of an element: (a) normal strain; (b) shearing strain.

From Fig 2.4a, normal strains:

x

y

u dx u dx u dx

xA B AB

dxAB

v dy v dy v dy

yA D AD

dyAD

                  

 

                

From Fig 2.4b, (engineering) shearing strain: xy x y

uv dydx yx

dx dy   



   

 2-D (in-plane)

 

normal strains:

engineering shearing strain:

x

y

xy

u

x

v

y

u v

y x

    

     

      

(2.3)

Three-Dimensional Strains

strain tensor:  

1 1 2 2

1 1 2 2

1 1 2 2

x xy xz

ij yx y yz

zx zy z

  

   

  

   

       

 ~ 2 nd

-rank tensor (2.7)

in scalar-component notation:

engineering strains:

x y z

xy yz zx

u v w

x y z

v u w v u w

x y y z z x

  

  

        

            

      

(2.4)

in tensor-index notation:  , ,

1 1

2 2

ji ij i j j i

j i

uu u u

x x 

         

 ij ji  ~ symmetric (2.5)

SECTION 2.3 STRAIN DEFINED PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

in vector-matrix notation:     1

2

T      

u u (2.3-A1)

Notes: a. The strain-displacement relations, e.g., (2.4 & 2.5) are also called the kinematic relations.

b. The strain components defined above are based on the Small Deformation Theory (or Infinitesimal

Elasticity Theory); that is, the displacement vector u is infinitesimally small. When u is large, the

Large Deformation Theory (or Finite Elasticity Theory) should be used. In that case, there are four

strain measures:

Def: engineering strain: 0

0

E

L L

L 

  where

 

  0 original initial length

instantaneous current length

L or

L or

 



Def: true (or natural or logarithmic) strain: 0

0ln ln L

L L

dL L L

L    

Def: Green-Lagrangian strain: 2 2

0

2

02 G

L L

L 

  based on the Lagrangian (material) description

Def: Almansi-Euler strain: 2 2

0

22 A

L L

L 

  based on the Euler (spatial) description

For small deformation, i.e.,  0 0L L L L or L   , we have: 0

E L G A

L

L    

    

Example 2.1 Plane Strains in a Plate (Self-Study)

A 0.8m0.6m rectangle ABCD is drawn on a thin plate prior to loading. Subsequent to loading, the deformed

geometry is shown by the dashed lines in Fig 2.5. Determine the components of plane strain at point A.

FIGURE 2.5 Deformation of a thin plate.

SECTION 2.4 SAINT-VENANT STRAIN COMPATIBILITY PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.4 SAINT-VENANT STRAIN COMPATIBILITY (Lecture)

6 kinematic (strain-displacement) relations:

x y z

xy yx yz zy zx xz

u v w

x y z

v u w v u w

x y y z z x

  

     

        

               

      

(2.4)

or  , ,

1 1

2 2

ji ij i j j i

j i

uu u u

x x 

         

, where , , ,i j x y z (2.5)

implies

   

 

continuous, single-valued continuous, single-valued

continuous, single-valued

displacement field : strain field :

strain field : dis

i ij

ij

u 





differentiation

integration

3 6

6   may not be

continuous, single-valued

placement field : iu 3

Def: strain compatibility (or continuity or integrability) equations: The additional relations that the strain

tensor ij , which has 6 components, must satisfy to ensure a continuous, single-valued displacement

field iu , which has only 3 components.

Figure 2.4-A1 Physical interpretation of strain compatibility.

SECTION 2.4 SAINT-VENANT STRAIN COMPATIBILITY PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Integrate Eq (2.5) twice w.r.t. kx , and

lx   , , ,

1

2 ij kl i jkl j iklu u  

Through simple interchange of subscripts 

 

 

 

, , ,

, , ,

, , ,

1

2

1

2

1

2

kl ij k lij l kij

ik jl i kjl k ijl

jl ik j lik l jik

u u

u u

u u

  

 

   

  

Assume the displacement field iu is continuous; then the order of differentiation on iu is immaterial. That is:

, ,i jkl i kjlu u , , ,j ikl j liku u , , ,k lij k ijlu u , , ,l kij l jiku u  Saint-Venant strain compatibility equations:

  , , , ,tensor-index notation: 0

vector-matrix notation: 0

ij kl kl ij ik jl jl ik       

   ε (2.4-A1)

Although (2.4-A1) would lead to 81 individual equations, most are either identities or repetitions, and only 6

are meaningful:

2 22 2

2 2

2 2 22

2 2

2 22 2

2 2

2

2

2

y xy yz xyx x xz

y yz y xy yzxzz

xy yzx xz xzz z

y x x y y z x x y z

z y y z z x y y z x

x z z x x y z z x y

     

    

    

               

                 

                  

             

          

   

(2.12)

Furthermore, these 6 equations are not independent. Only 3 of them are independent:

4 3

2 2

4 3

2 2

4 3

2 2

2

2

2

yz xyx xz

y xy yzxz

xy yz xzz

y z x y z x y z

z x x y z y z x

x y x y z z x y

  

  

  

          

             

              

          

         

(2.4-A2)

For 2-D cases, the 6 compatibility equations reduce to 1:

2 22

2 2

y xyx

y x x y

     

    (2.11)

Finally, the compatibility equations (2.12) are necessary and sufficient conditions that the strain components

ij give continuous, single-valued displacements iu for a simply-connected domain. For a multiply-connected

domain, however, these conditions are necessary but generally not sufficient.

SECTION 2.4 SAINT-VENANT STRAIN COMPATIBILITY PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Figure 2.4-A2 Continuity of displacements.

Figure 2.4-A3 Examples of domain connectivity.

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 1/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.5 STATE OF STRAIN AT A POINT (Lecture)

FIGURE 2.2 Normal strain in a prismatic bar: (a) undeformed state; (b) deformed state.

FIGURE 2.3 Strain components x , y and xy in the x-y plane.

FIGURE 2.4 Deformations of an element: (a) normal strain; (b) shearing strain.

Transformation of 2-D Strain:

     

2 2

2 2

2 2

cos sin sin cos cos 2 sin 2 2 2 2

2 sin cos cos sin sin 2 cos 2

sin cos sin cos cos 2 sin 2 2 2 2

x y x y xy

x x y xy

x y y x xy x y xy

x y x y xy

y x y xy

              

            

              

 

       

        

        



(2.13 & 14)

Note: constantx y x y        ~ strain invariant

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 2/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

By interchanging symbols:    , , , ,x y x y    , we also get:

     

2 2

2 2

2 2

cos sin sin cos cos 2 sin 2 2 2 2

2 sin cos cos sin sin 2 cos 2

sin cos sin cos cos 2 sin 2 2 2 2

x y x y x y

x x y x y

xy x y x y x y x y

x y x y x y

y x y x y

              

            

              

     

   

       

     

   

       

      

       

     

(2.5-A1)

Figure 2.5-A1 Two-dimensional orthogonal (or rotational) transformation.

Transformation of 3-D Strain:

       

       

2 2 2

1 1 1 1 1 1 1 1 1

1 2 1 2 1 2 1 2 2 1 1 2 2 1 2 1 1 2

1 3 1 3 1 3 1 3 3 1 1 3 3 1 3 1 1 3

2 2 2

2 2 2

2

2

x x y z xy yz xz

x y x y z xy yz xz

x z x y z xy yz xz

y x y z

l m n l m m n l n

l l m m n n l m l m m n m n l n l n

l l m m n n l m l m m n m n l n l n

l m n

      

      

      

    

 

 

     

        

        

   

       

2 2 2 2 2 2

2 2 2

3 3 3 3 3 3 3 3 3

2 3 2 3 2 3 2 3 3 2 2 3 3 2 2 3 3 22

xy yz xz

z x y z xy yz xz

y z x y z xy yz xz

l m m n l n

l m n l m m n l n

l l m m n n l m l m m n m n l n l n

 

      

      

 

     

                    

(2.18)

Note: Owing to the symmetry of the strain tensor: ij ji  , only 6 of the 9 stress components thus developed

are unique.

As shown in Sec. 1.12, the direction cosine components in Table 1.2 can be collected to form:

direction cosine (or orthogonal rotation) matrix:   11 12 13 1 1 1

21 22 23 2 2 2

31 32 33 3 3 3

ij

l l l l m n

l l l l l m n

l l l l m n

       

             

l (1.12-A6)

Note: Again, the direction cosine matrix is NOT symmetric: ij jil l

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 3/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

The scalar-component form of Eq (2.18) can then be expressed in the short-hand forms:

tensor-index notation: rs ir js ij

rs ri sj ij

l l

l l

 

 

  

 (2.19)

or vector-matrix notation:      

      

T

T

   



l l

l l

 

  (2.5-A2)

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 4/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Example 2.2 Three-Dimensional Strain in a Block (Lecture)

A 2m1.5m1m parallelepiped is deformed by movement of corner point A (2, 1.5, 1) to A′ (1.9985, 1.4988,

1.0009), as shown by the dashed lines in Fig 2.7. During this deformation, point O remains fixed. Calculate the

following quantities at point A:

a. the strain components w.r.t. the x-y-z coordinates;

b. the normal strain in the direction of line AB ; and

c. the normal strain in the direction of line AC and the shearing strain for perpendicular lines AB and AC .

d. Also find the third line AD so that it is perpendicular to both AB and AC . Calculate the normal and shear

strains associated with this direction.

FIGURE 2.7 Deformation of a parallelepiped.

Solution The components of displacement of point A are given by:

1.9985 2 0.0015 m 1.5 mm

1.4988 1.5 0.0012 m 1.2 mm

1.0009 1 0.0009 m 0.9 mm

A

A

A

u

v

w

      

          

(d)

a. Inverse Method (Sec. 3.7): Assume a displacement field:

     1 2 3, , , , , ,u x y z c xyz v x y z c xyz w x y z c xyz   , where 1c , 2c and 3c are constants. (A.1)

Note: The assumed displacement field satisfies the fixed constraint at the origin, point O.

Eq (d)  Eq (A.1), we have:

   

   

   

1 2

Pt

2 2

Pt

3 2

Pt

0.0015 1 500μ

2 1.5 1 m

0.0012 1 400μ

2 1.5 1 m

0.0009 1 300μ

2 1.5 1 m

A

A

A

u c

xyz

v c

xyz

w c

xyz

     

 

    

       

(A.2)

Here 6μ 10 . Applying Eq (2.4), we have

1 2 3

2 1 3 2 1 3

x y z

xy yz xz

u v w c yz c xz c xy

x y z

v u w v u w c yz c xz c xz c xy c xy c yz

x y y z z x

  

  

           

                  

      

(f)

By introducing Eq (f) into Eq (2.12), we have

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 5/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2

2

x

y



2

2

y

x

 

2

xy

x y

   

2

2 yzx

y z x x

    

   

2

2

xyxz

y

y z

z



    

   

2

2

z

y

 

2

yz

y z

 

 

2

2 y yz xz

x z y x y

      

    

2

2

xy

z

z

x

     

2

2

x

z

 

2

xz

x z

 

 

2

2 yz xyxzz

x y z x y z

       

     

       

       

22 2

1

2 2 2

2

2 22

3

2 2

2 2

2 2

xyx xz

y yz xy

yz xzz

c y z x y x z

c x z x y y z

c x y x z y z

 

  

 

     

         

        

      

     

 Saint-Venant’s compatibility conditions are satisfied; thus, the strain field obtained is therefore possible.

The strain components can be calculated as follows

           

           

           

1 2 1

2 3 2

3 1 3

500μ 1.5 1 750με 400μ 1.5 1 500μ 2 1 1600με

400μ 2 1 800με 300μ 2 1 400μ 2 1.5 600με

300μ 2 1.5 900με 500μ 2 1.5 300μ 1.5 1 1050με

x xy

y yz

z xz

c yz c yz c xz

c xz c xz c xy

c xy c xy c yz

 

 

 

             

                      

Note: The strain unit “” is dimensionless.

The approach above allows the calculations of the full displacement and strain fields. However, since the

displacement at point A is known, the calculations can be simplified to find its strains alternatively

, ,

Pt Pt

, ,

Pt Pt

,

1.5 mm 1.2 mm 1.5 mm 750με 1600με

2 m 2 m 1.5 m

1.2 mm 0.9 mm 1.2 mm 800με 600με

1.5 m 1.5 m 1.0 m

A A A x A xy A

A A

A A A y A yz A

A A

z A

u v uu v u

x x x y x y

v w vv w v

y y y z y z

w

 

 

                           

                          

  

,

Pt Pt

0.9 mm 1.5 mm 0.9 mm 900με 1050με

1 m 1 m 2 m

A A A xz A

A A

w u wu w

z z z x z x 

       

                     

which are identical to the results obtained by the full-field approach.

b. Let x′-axis be placed along the line ˆ ˆ ˆ ˆ ˆ ˆ2 1.5 0x x xAB a b c        i j k i j k

       2 2 22 2 2 2 1.5 0 2.5 mx x xAB a b c           . From Sec 1.12, the direction cosines of AB are:

 

 

 

1

1

1

2 cos , 0.8

2.5

1.5 cos , 0.6

2.5

0 cos , 0

2.5

x

x

x

a l x x

AB

b m x y

AB

c n x z

AB

 

         

             

(B.1)

Applying Eq (2.18a), we thus have:

2 2 2

1 1 1x x y zl m n       1 1 1 1xy yzl m m n   1 1xzl n

          2 2

750 0.8 800 0.6 1600 0.8 0.6 1536με           

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 6/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

c. Let the y′-axis be placed along the line ˆ ˆ ˆ ˆ ˆ ˆ0 0 1y y yAC a b c       i j k i j k

       2 2 22 2 2 0 0 1 1my y yAC a b c          and the direction cosines of AC are:

     2 2 2

0 0 1 cos , 0 cos , 0 cos , 1

1 1 1

y yy a b c

l x m y n zy y y AC AC AC

                  (C.1)

Applying Eqs (2.18d,b), we have: 2

2y x l   2

2y m 2

2 2z xyn l   2m 2yz m 2 2xzn l    2

2 900 0.1 900μεn   

1 22x y xl l    1 2ym m 1z n 2 1 2xyn l m 2l 1 1 2 2yzm m n m  1n  2xz l 1n         

1 2

600 0.6 1 1050 0.8 1 1200με

l n

         

Note: Since         2 0 1.5 0 0 1 0x y x y x yAB AC a a b b c c                AB AC

Question: Why y z   ?

d. Let the z′-axis be placed along the line ˆ ˆ ˆ z z zAD a b c    i j k . Since AB AD & AC AD , we have

2 1.5 0

0

x z x z x z z z

y z y z y z z

AB AD a a b b c c a b

AC AD a a b b c c c

       

      

        

     

1.5

2

0

z

z

z

a

b

c

  

 

 choose ˆ ˆ ˆ1.5 2 0AD   i j k

       2 2 22 2 2 1.5 2 0 2.5 mz z zAD a b c          and the direction cosines of AD are:

 

 

 

3

3

3

1.5 cos , 0.6

2.5

2 cos , 0.8

2.5

0 cos , 0

2.5

z

z

z

a l z x

AD

b m z y

AD

c n z z

AD

         

             

(D.1)

Applying Eqs (2.18e,f,c), we have: 2 2 2

3 3 3z x y zl m n       3 3 3 3xy yzl m m n   3 3xzl n

          2 2

750 0.6 800 0.8 1050 0.6 0.8 278με         

22y z x l    3 2yl m 3 2 3zm n n  2xy l 3 3 2m l m  2yz m 3n 3 2 2xzm n l  3n         

3 2

600 0.8 1 1050 0.6 1 150με

l n

       

1 3 1 3 12x z x y zl l m m n        3n   1 3 3 1 1 3xy yzl m l m m n    3 1m n  3 1xz l n 1 3l n                 2 750 0.8 0.6 800 0.6 0.8 1600 0.8 0.8 0.6 0.6 496με                    

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 7/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Example 2.3 State of Plane Strain in a Plate (Self-Study)

The state of strain at a point on a thin plate is given by 510μεx  , 120μεy  and 260μεxy  . Using Mohr’s

circle of strain to determine:

a. the state of strain associated with axes x′, y′, which make an angle θ  30° with the axes x, y (Fig 2.8a);

b. the principal strains and directions of the principal axes; and

c. the maximum shear strains and associated normal strains; and

d. display the given data and the results obtained on properly oriented elements of unit dimensions.

FIGURE 2.8 (a) Axes rotated for θ  30°; (b) Mohr’s circle of strain.

FIGURE 2.9 (a) Element with edges of unit lengths in plane strain; (b) element at θ  30°;

(c) principal strains; and (d) maximum shearing strains.

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 8/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

 EXAMPLE: KINEMATIC (u-) RELATIONS

BIAXIAL NORMAL STRAINS

A proposed strain field for a two-dimensional elasticity problem of a rectangular panel stretched by uniform

edge loadings is:

0

0

x xy

ij

xy y

A

B

  

 

        

  

where A and B are constants. Assume the problem depends only on x and y.

a. Verify if the strain field is compatible.

b. Integrate the 2-D strain-displacement relations, Eq (2.3), to determine the displacement components u and v.

c. Identify all rigid-body (RB) motion terms:  0 0 , ,

z u v  , which are the RB translations along the x- and

y-directions and the RB rotation about the z-axis, respectively.

d. If the panel is restrained so that there is no RB motion, represent graphically the deformed and undeformed

shapes of the panel.

Sol:

a. 2-D compatibility condition, Eq (2.11):

2 22

2 2

y xyx

y x x y

     

    

2

2

A

y

2

2

B

x

  

2 0

x y

   

 satisfied

b. Eq (2.3):

   

   

integrate

1

integrate

1

,

,

0

x

y

xy

u u x y x u yA A

x

v B v x y By v x

y

u v

y x

       

       

       

(b1)

Substitute Eqs (b1-1 & 2) into Eq (b1-3)     1 1 0x u y By v xA y x

            

    1 1

0 du y dv x

dy dx  

    1 1

constant du y dv x

a dy dx

    (b2)

Integrate Eq (b2)   

  1

1

u y ay b

v x ax c

   

  (b3)

Combine Eqs (b1-1,2 & b3)  2-D displacement field:  

 

,

,

u x y x ay bA

v x y By ax c

   

   (b4)

c. Physically, the integration constants  , ,a b c represent the rigid-body motion:  0 0 , ,

z u v . Thus,

 

  0

0

,

,

z

z

u x y x y uA

v x y By x v

   

   (c1)

SECTION 2.5 STATE OF STRAIN AT A POINT PAGE 9/9

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

d. Assume a panel of unit size (11), if the panel is restrained at the center so that there is no RB motion, i.e.,

0 0 0

z u v    , then

 

 

,

,

u x y xA

v x y By

 

 and the deformed and undeformed shapes are represented graphically

below, where   Poisson’s ratio. These figures illustrate the possibilities of generating the same strain tensor

by uniaxial or biaxial tension.

Vertical Uniaxial Tension Biaxial Tension Horizontal Uniaxial Tension

A B B A

B

A

A

B

AB

  • 2.2 Deformation (2018-01-25)
  • 2.3 Strain Defined (2018-01-25)
  • 2.4 Saint-Venant Strain Compatibility (2018-01-25)
  • 2.5 State of Strain at a Point (2018-01-25)

__MACOSX/stress/._Ch 2 Strain & Material Properties-Part 1 (2018-01-25).pdf

stress/Ch 2 Strain & Material Properties-Part 2 (2018-01-25).pdf

SECTION 2.9 HOOKE’S LAW AND POISSON’S RATIO PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.9 HOOKE’S LAW AND POISSON’S RATIO (Lecture)

Hooke’s law: x xE  , where E  Young’s modulus or modulus of elasticity (2.26)

Hooke’s law in shear: xy xyG  , where G  shear modulus or modulus of rigidity (2.27)

Poisson’s ratio: lateral strain

or axial strain

y z

x x

  

       (2.28)

Note:  2 1

E G

 

 (2.35)

FIGURE 2.15 Lateral contraction of an element in tension.

Volume Change

original volume:    0V dx dy dz

final volume:            1 1 1 1 1 1f x y z x x xV dx dy dz dx dy dz                                

         2 2 3

0 01 1 1 1 1 2 2f x x x x x xV V V                    

ignore the higher order terms:  2

x and 3

x    01 1 2f xV V     

volume change:    0 0 0 01 1 2 1 2f x xV V V V V V             

dilatation or dilation (unit volume change):    

0

1 2 1 2

3

x x x

V e

V E K

    

      (2.29)

where K  bulk modulus of elasticity. Note:  3 1 2

E K

 

 (2.39)

SECTION 2.9 HOOKE’S LAW AND POISSON’S RATIO PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Example 2.2 Deformation of a Tension Bar (Self-Study)

An aluminum alloy bar of circular cross-sectional area A and length L is subjected to an axial tensile force P

(Fig 2.16). The modulus of elasticity and Poisson’s ratio of the material are E and v, respectively. Determine for

the bar:

a. the axial deformation ;

b. the change in diameter d;

c. the change in volume ΔV;

d. the strain energy stored in the bar after tension U (see Sec 2.14). Also

e. evaluate the numerical values of the quantities obtained in (a) through (d) for the case in which P  60 kN,

d  25 mm, L  3 m, E  70 GPa and   0.3.

FIGURE 2.16 A bar under tensile forces.

Solution

2

4 axial stress:

Hooke's law:

axial strain:

P P

A d

E

L

 

 

 

  

 

   

SECTION 2.9 HOOKE’S LAW AND POISSON’S RATIO PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 1/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.10 GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS (Lecture)

2.11 HOOKE’S LAW FOR ORTHOTROPIC MATERIALS (Lecture)

Theorem: Neumann principle: Symmetry in material microgeometry corresponds to identical symmetry in

the constitutive response.

FIGURE 2.10-A1 Material microstructures.

Generalized Hooke’s Law for Linearly-Elastic Materials Under Isothermal Condition

tensor-index notation: ij mnij mn

ij mnij mn

c

s

 

 

 

 (2.33)

where  

  mnij

mnij

c

s

 



4th -order elasticity tensor

4th -order compliance tensor

Since 9 stress components & 9 strain components  81 elasticity (or compliance) constants

Because ij ji  & mn mn  (i.e., symmetric)  mnij nmij mnji nmji

mnij nmij mnji nmji

c c c c

s s s s

   

  

 36 elasticity (or compliance) constants

Def: 0

0

0 0

strain energy density:

complementary energy density:

ij

ij

ij ij

ij ij

U d

U d

 

 

      

 

 

 

0

0

ij

ij

ij

ij

ij

ij

U

U

 

 

  

 

 



(2.54)

linearly elastic   0 0

1 1

2 2 ij ij x x y y z z xy xy yz yz zx zxU U                      (2.51)

 mnij ijmn

mnij ijmn

c c

s s

 

  21 elasticity (or compliance) constants (2.10-A1)

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 2/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 3/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

 scalar-component notation:

11 12 13 14 15 16

12 22 23 24 25 26

13 23 33 34 35 36

14 24 34 44 45 46

15 25 35 45 55 56

16 26 36 46 56 66

x x

y y

z z

xy xy

yz yz

xz xz

c c c c c c

c c c c c c

c c c c c c

c c c c c c

c c c c c c

c c c c c c

 

 

 

 

 

 

                          

                           

  

(2.32)

or

11 12 13 14 15 16

12 22 23 24 25 26

13 23 33 34 35 36

14 24 34 44 45 46

15 25 35 45 55 56

16 26 36 46 56 66

x x

y y

z z

xy xy

yz yz

xz xz

s s s s s s

s s s s s s

s s s s s s

s s s s s s

s s s s s s

s s s s s s

 

 

 

 

 

 

                          

                           

  

(2.10-A2)

in tensor-index/vector-matrix notations: i ij j

i ij j

c

s

 

 

 

 or

    

    

  



c

s

 

  ~ Voigt contraction (2.10-A3)

Notes: a.     1

s c

b. ij ji

ij ji

c c

s s

 

 ~ symmetric (2.56)

c. The material is also called triclinic (or general anisotropic).

Monoclinic Material (Single Plane of Symmetry): 13 elasticity (or compliance) constants

Figure 2.10-A2 Plane of symmetry for a monoclinic material.

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 4/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

11 12 13 16

12 22 23 26

13 23 33 36

44 45

45 55

16 26 36 66

0 0

0 0

0 0

0 0 0 0

0 0 0 0

0 0

x x

y y

z z

xy xy

yz yz

xz xz

c c c c

c c c c

c c c c

c c

c c

c c c c

 

 

 

 

 

 

                          

                              

(2.10-A4)

Orthotropic (or Orthorhomic) Material (3-Perpendicular Planes of Symmetry):

9 elasticity (or compliance) constants

Figure 2.10-A3 Three planes of symmetry for an orthotropic material.

11 12 13

12 22 23

13 23 33

44

55

66

0 0 0

0 0 0

0 0 0

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0

x x

y y

z z

xy xy

yz yz

xz xz

c c c

c c c

c c c

c

c

c

 

 

 

 

 

 

                          

                              

(2.40 & 41)

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 5/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

or

11 12 13

12 22 23

13 23 33

44

55

66

1 0 0 0

0 0 0 0 0 0

0 0 0

0 0 0

0 0 0 0 0

0 0 0 0 0

0 0 0

1

0 0

zy

y

yx zx

x y z

xy

x zx x

y y xz

z z x

xy xy

yz yz

xz xz

E E E

E Es s s

s s s

s s s E

s

s

E

s

 

 

  

 

 

 

 

                              

                              

0 0 0

1 0 0 0 0 0

1 0 0 0 0 0

1 0 0 0

1

0 0

x

yyz

zy

xy

yzxy

xz

yz

z

xz

E

G

E

G

G

                                          

               

(2.42)

Due to symmetry of the matrix  s  ij ji

i jE E

   or

xy yx

x yE E

   ,

yz zy

y zE E

   , xz zx

x zE E

   (2.43)

Notes: a. Eq (2.42) in textbook contains typos.

b. In the above,

, ,x y zE E E  orthotropic moduli of elasticity in the three directions of material symmetry

, ,xy yz xzG G G  shear moduli in the three orthogonal planes of material symmetry

 where , , , , j

ij

i

i j x y z i j 

 

     Poisson’s ratios between  ,i j directions (2.10-A5)

c. For isotropic materials, Poisson’s ratio is limited to 1 0.5   to ensure Young’s, shear and bulk

moduli  , ,E G K are all positive. For anisotropic materials, that constraint, however, does not

apply. Instead, in order to preserve the ve-definiteness of strain energy density, the Poisson’s

ratios need to satisfy the following relations:

1 0

1 0

1 0

xy yx

xz zx

yz zy

 

 

 

     

  

(2.10-A6)

and 1 2 0xy yx yz zy zx xz yx zy xz             (2.10-A7)

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 6/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Non-Rectilinear Orthotropic Material:

Circumferentially orthotropic (wood annulus) Spirally orthotropic (filament winding)

Notes: a. Examples of special anisotropic materials: (appear in crystallography, biology, etc.)

b. Extension-shear coupling in triclinc (general anisotropic) and monoclinic material, but not in

orthotropic, transversely isotropic and isotropic materials.

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 7/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Mechanical behaviors of various materials under tension or shear

Isotropic Orthotropic Anisotropic

Tension

Shear

Transversely Isotropic (or Hexagonal) Material (Axis of Symmetry)

5 elasticity (or compliance) constants

 

11 12 13 16

12 11 13 26

13 13 33 36

44

44

11 12

0 0

0 0

0 0

0 0 0 0 0

0 0 0 0 0

1 0 0 0 0 0

2

x x

y y

z z

xy xy

yz yz

xz xz

c c c c

c c c c

c c c c

c

c

c c

 

 

 

 

 

 

                                                   

(2.10-A5)

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 8/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Figure 2.10-A4 Axis of symmetry for a transversely isotropic material.

Tetragonal, Trigonal & Cubic Materials

Tetragonal Materials

7 elasticity (or compliance) constants 6 elasticity (or compliance) constants

 

 

11 12 13 16

12 11 13 16

13 13 33

44

55

16 16 11 12

0 0

0 0

0 0 0

0 0 0 0 0

0 0 0 0 0

1 0 0 0

2

ij

c c c c

c c c c

c c c

c c

c

c c c c

   

                   

c  

 

11 12 13

12 11 13

13 13 33

44

55

11 12

0 0 0

0 0 0

0 0 0

0 0 0 0 0

0 0 0 0 0

1 0 0 0 0 0

2

ij

c c c

c c c

c c c

c c

c

c c

                     

c

Trigonal Material Cubic Material

7 elasticity (or compliance) constants

6 elasticity (or compliance) constants

 

 

11 12 13 14 25

12 11 13 14 25

13 13 33

14 14 44 25

25 25 55 14

25 14 11 12

0

0

0 0 0

0 0

0 0

1 0 0 0

2

ij

c c c c c

c c c c c

c c c

c c c c c

c c c c

C c c c

   

                  

c  

11 12 12

12 11 12

12 12 11

44

44

44

0 0 0

0 0 0

0 0 0

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0

ij

c c c

c c c

c c c c

c

c

c

       

            

c

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 9/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Isotropic Material (Complete Symmetry): 2 elasticity (or compliance) constants,  ,E 

1 0 0 0

1 0 0 0

1 0 0 0

1 0 0 0 0 0

1 0 0 0 0 0

1 0 0 0 0 0

x x

y y

z z

xy xy

yz yz

xz xz

E E E

E E E

E E E

G

G

G

 

   

    

 

 

 

    

        

                  

                           

       

or

 

 

 

1

1

1

xy

x x y z xy

yz

y y x z yz

xz z z x y xz

E G

E G

E G

      

      

      

       

 

        

       

(2.34)

Note:  2 1

E G

 

 = shear modulus (2.35)

2 0 0 0

2 0 0 0

2 0 0 0

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0

x x

y y

z z

xy xy

yz yz

xz xz

G

G

G

G

G

G

   

  

   

 

 

 

         

             

                   

        

or

2

2

2

x x xy xy

y y yz yz

z z xz xz

e G G

e G G

e G G

    

    

    

   

      

(2.36)

where change in volume

original volume dilatation volume strain  is defined as

  0

1 2 x y z x y z

V e

V E

      

         (2.37)

and Lamé’s constants:   

 

1 1 2

2 1

E

E G

 

 

   

    

(2.38)

Also bulk modulus of elasticity:  

mean stress

volume strain 3 1 2

m p E K

e e

     

 (2.39)

SECTION 2.10

SECTION 2.11

GENERALIZED HOOKE’S LAW-LINEAR ELASTIC MATERIALS

HOOKE’S LAW FOR ORTHOTROPIC MATERIALS PAGE 10/10

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

Note:  

 

: 0.25 0.8

e.g., rubber : 0.5 , 3

e.g., cork : 0 0 2 3

G E

E G K

E E G K

 

 

 

               

Poisson's solid

incompressible solid

solids with zero Poisson's ratio

SECTION 2.12 MEASUREMENT OF STRAIN: STRAIN ROSETTE PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.12 MEASUREMENT OF STRAIN: STRAIN ROSETTE (Lecture)

Figure 2.20 (a) Strain gage (courtesy of Micro-Measurements Division, Vishay Intertechnology, Inc.) and

(b) schematic representation of a strain rosette.

2 2

2 2

2 2

cos sin sin cos

cos sin sin cos

cos sin sin cos

a x a y a xy a a

b x b y b xy b b

c x c y c xy c c

       

       

       

    

      

(2.44)

Table 2.2 Strain rosette equations

SECTION 2.12 MEASUREMENT OF STRAIN: STRAIN ROSETTE PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

SECTION 2.12 MEASUREMENT OF STRAIN: STRAIN ROSETTE PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

  • 2.9 Hooke’s Law & Poisson’s Ratio (2018-01-25)
  • 2.10 Generalized Hooke’s Law-Linear Elastic Materials (2018-01-25)
  • 2.12 Measurement of Strain-Strain Rosette (2018-01-25)

__MACOSX/stress/._Ch 2 Strain & Material Properties-Part 2 (2018-01-25).pdf

stress/Ch 2 Strain & Material Properties-Part 3 (2018-01-25).pdf

SECTION 2.13 STRAIN ENERGY PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.13 STRAIN ENERGY (Lecture)

Strain Energy Density for Normal and Shear Stresses

FIGURE 2.21 (a) Deformation/displacement under uniaxial stress; (b) work done by uniaxial stress or

strain energy stored due to uniaxial deformation.

strain energy stored  net work done:

0 0 0 0

x x x x

x x x x

u u u dU dW d u dx dydz dudydz d dxdydz d dV

x x x

   

           

                  

   

Since x

u

x 

  

  0

x

x xdU d dV 

  

Def: strain energy density: strain energy per unit volume: o

dU U

dV 

Def: (total) strain energy: o o V

U U dV U dxdydz  

Def: complementary energy density: complementary energy per unit volume: *

*

o

dU U

dV 

Def: (total) complementary energy: * * *

o o V

U U dV U dxdydz  

(2.58)

 uniaxial tension: 0

*

0

x

x

o x x

o x x

U d

U d

 

 

      

 Note: *

o o x xU U    (2.48 & 49)

linearly elastic: x xE   * 2 21 1 1

2 2 2 o o x x x xU U E

E        (2.13-A1)

pure shear: 0

*

0

xy

xy

o xy xy

o xy xy

U d

U d

 

 

      

 Note:

*

o o xy xyU U    (2.13-A2)

linearly elastic: xy xyG   * 2 21 1 1

2 2 2 o o xy xy xy xyU U G

G        (2.13-A3)

SECTION 2.13 STRAIN ENERGY PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

FIGURE 2.22 Deformation due to pure shear.

Strain Energy Density for 3-Dimensional Stresses (Linearly-Elastic):

 

     

   

*

2 2 2 2 2 2

2 2 2 2 2 2 2

1

2

1 1

2 2

1 2

2

o o x x y y z z xy xy yz yz zx zx

x y z x y y z x z xy yz zx

x y z xy yz xz

U U

E E G

e G G

           

            

      

      

        

         

(2.51-53)

SECTION 2.14 STRAIN ENERGY IN COMMON STRUCTURAL MEMBERS PAGE 1/2

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.14 STRAIN ENERGY IN COMMON STRUCTURAL MEMBERS (Lecture)

Strain Energy for Axially Loaded Bars

constantx

P

A   

linearly-elastic non-prismatic bar: 2 2

02 2

L x

o V V

P U U dV dV dx

E EA

      (2.58)

prismatic  2

2

P L U

EA  Note: EA  axial rigidity (2.59)

FIGURE 2.23 Nonprismatic bar with varying axial loading.

Strain Energy of Circular Bars in Torsion

constant T

J

   

linearly-elastic non-prismatic shaft:   2 2 2

2

20 02 2 2

L L

o V V

T T U U dV dV dA dx dx

G GJ GJ

         (2.61)

prismatic and constant twisting torque  2

2

T L U

GJ  Note: GJ  torsional rigidity (2.62)

Strain Energy for Beams in Bending:

constantz x

z

M y

I    

linearly-elastic variable-cross sectional beam    2 2

2

202 2

L x z

o V V

z

M U U dV dV y dA dx

E EI

       (2.14-A1)

constant cross section and constant bending moment  2

20 2

L z

z

M U dx

EI   (2.63)

Note: EI  bending/flexural rigidity

SECTION 2.14 STRAIN ENERGY IN COMMON STRUCTURAL MEMBERS PAGE 2/2

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

SECTION 2.15 COMPONENTS OF STRAIN ENERGY PAGE 1/2

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

2.15 COMPONENTS OF STRAIN ENERGY (Lecture)

dilatational (or mean or spherical) stress tensor:

0 0

0 0

0 0

m

ij m

m

p

   

     

(1.14-S2)

distortional (or deviatoric) stress tensor:

x m xy xz

ij yx y m yz

zx zy z m

s

   

   

   

   

     

(1.14-S5)

total stress = dilatational stress + distortional stress: ij ij ijp s   (1.14-S6)

FIGURE 2.25 Resolution of (a) state of stress into (b) dilatational stresses and (c) distortional stresses.

total strain energy density = dilatational energy density + distortional energy density:

o ov odU U U  (2.15-A1)

linearly elastic       2 2 2 2 2 21 1

2 2 2 o x y z x y y z x z xy yz zxU

E E G

                     (2.52)

then the (elastic) dilatational energy density:

    2 2

221 1 2 3

1 1

2 18 18 18

m ov x y z

I U

K K K K

              (2.64)

and the (elastic) distortional energy density:

     

       

2 2 2 2oct 2

1 2 2 3 3 1

2 2 2 2 2 2

3 1

4 122

1 6

12

od

x y y z z x xy yz zx

J U

G GG

G

      

        

          

            

(2.65)

Here

 

       

1

2 2 2

oct 2

1 :

3 3

1 2 : 6

3 3

m x y z

x y y z z x xy yx yz zy zx xz

I

J

   

            

    

              octahedra

mean stress

l shear stress

Note: : change only in volume, not in shape

: change only in shape, not in volume

  di

dilatational

stortional

SECTION 2.15 COMPONENTS OF STRAIN ENERGY PAGE 2/2

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 2 – STRAIN & MATERIAL PROPERTIES

  • 2.13 Strain Energy (2018-01-25)
  • 2.14 Strain Energy in Common Structural Members (2018-01-25)
  • 2.15 Components of Strain Energy (2018-01-25)

__MACOSX/stress/._Ch 2 Strain & Material Properties-Part 3 (2018-01-25).pdf

stress/Ch 3 Problems in Elasticity A-Part 1 (2018-02-09).pdf

SECTION 3.2

SECTION 2.16

FUNDAMENTAL PRINCIPLES OF ANALYSIS

SAINT-VENANT’S PRINCIPLE PAGE 1/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

CHAPTER 3: PROBLEMS IN ELASTICITY

PART A - FORMULATION AND METHODS OF SOLUTION

3.2 FUNDAMENTAL PRINCIPLES OF ANALYSIS (Lecture)

Basic Principles of Mechanics/Materials-Based Engineering Analysis

1. Equilibrium Conditions. The equations of equilibrium of forces/moments must be satisfied throughout

the member.

2. Material Behaviors/Constitutive Relations. The stress-strain or force-deformation relations (for

example, Hooke’s law) must apply to the material behavior of which the member is constructed.

3. Geometry of Deformation/Compatibility Conditions. The compatibility conditions of deformations must

be satisfied: that is, each deformed portion of the member must fit together with adjacent portions. (Note:

For mathematical strictness, the matter of compatibility should always be complied in Theory of Elasticity;

however, it may not always be broached in Mechanics of Materials analysis.)

4. Boundary and Initial Conditions. The stress and deformation obtained through the use of the above three

principles must conform to the initial conditions: the initial values of displacements (and velocities for

dynamic problems) of the member as well as satisfy the boundary conditions: conditions of loading

imposed at the boundaries of the member.

   

 

 

 

~~

or

     

    

 



Plane - stress problems

Plane - strain pr

Cartesian coordin

Torsion problems

Plane problems

Axisymmetric problems

ates Two - dimensional problems

Polar coordinaoblems tes

Thr

x, y

x, y r,z

r,

r,θ

z

         ee - dimensional problems

Three-Dimensional Problems

 

 

 

, , , ,

, , , ,

,

,

,

,

x y z xy x

x y z xy xz yz

z yz

u v w

     

     

     

15 unkn 6 stresse

6 st

o

rains

3 displacements

s swn

equilibrium equations:

0

0

0

xyx xz x

xy y yz

y

yzxz z z

F x y z

F x y z

F x y z

 

  

 

     

      

      

      

  

~ 3 (1.14)

displacement-strain/kinematic relations:

x y z

xy yz zx

u v w

x y z

v u w v u w

x y y z z x

  

  

        

            

      

~ 6 (2.4)

SECTION 3.2

SECTION 2.16

FUNDAMENTAL PRINCIPLES OF ANALYSIS

SAINT-VENANT’S PRINCIPLE PAGE 2/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

stress-strain/constitutive relations or generalized Hooke’s law:

 

 

 

1

1

1

xy

x x y z xy

yz

y y x z yz

xz z z x y xz

E G

E G

E G

      

      

      

       

 

        

       

or

2

2

2

x x xy xy

y y yz yz

z z xz xz

e G G

e G G

e G G

    

    

    

   

      

~ 6 (2.34 & 36)

strain compatibility equations:

2 22 2

2 2

2 2 22

2 2

2 22 2

2 2

2

2

2

y xy yz xyx x xz

y yz y xy yzxzz

xy yzx xz xzz z

y x x y y z x x y z

z y y z z x y y z x

x z z x x y z z x y

     

    

    

               

                 

                  

             

          

   

~ only 3 independent (2.12)

Boundary Conditions in Terms of Surface Forces (or Tractions) (see Sec 1.16)

x x xy xz

y xy y yz

z xz yz z

p l m n

p l m n

p l m n

  

  

  

    

      

(1.26 or 48)

FIGURE 1.19 Stress components on a tetrahedron.

SECTION 3.2

SECTION 2.16

FUNDAMENTAL PRINCIPLES OF ANALYSIS

SAINT-VENANT’S PRINCIPLE PAGE 3/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

2.16 SAINT-VENANT’S PRINCIPLE (Lecture)

Saint-Venant’s principle: If an actual distribution of forces is replaced by a statically equivalent system, the

distribution of stress and strain throughout the body is altered only near the regions of load application.

Figure 2.26 Stress distribution due to a concentrated load in a rectangular elastic plate, confirming the

Saint-Venant’s principle. Note that the average stress avg is constant at all cross-sections.

Saint-Venant’s principle: avg A

P dA A   at any cross section (2.16-A1)

FIGURE 2.16-A1 Statically equivalent loadings for regions far away from the loadings.

Apply Saint-Venant’s principle  weak B.C.:

0

0

0

x x xA

y xy xA

z x xA

R dA

R dA

M ydA

    

  

  

(2.16-A2)

Figure 2.27 Cantilever beam illustrating use of Saint-Venant’s principle: (a) actual support;

(b) statically equivalent loading obtained by replacing the clamped end with its reactions.

SECTION 3.2

SECTION 2.16

FUNDAMENTAL PRINCIPLES OF ANALYSIS

SAINT-VENANT’S PRINCIPLE PAGE 4/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

SECTION 3.2

SECTION 2.16

FUNDAMENTAL PRINCIPLES OF ANALYSIS

SAINT-VENANT’S PRINCIPLE PAGE 5/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

SECTION 3.3 PLANE STRAIN PROBLEMS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.3 PLANE STRAIN PROBLEMS (Lecture)

FIGURE 3.1 Plane strain in a long cylindrical body.

Uplane-strain condition in x-y plane: z-independence: 0 z

 

displacement field: (2 non-zero displacement components)

       , , , , 0u u x y u z v v x y v z w     (3.3-A1)

strain-displacement (or kinematic) relations: (3 non-zero strain components)

Eq (3.3-A1) into Eq (2.4) 

0 0 0

x y xy

z xz yz

u v u v

x y y x

w u w w v

z z x y z

  

  

           

             

     

(3.1 & 2)

stress-strain (or constitutive) relations: (4 non-zero stress components, only 3 independent)

Eq (3.2) into Eq (2.36) 

     

 

2

2 0

0

x x y x z x y x y

y x y y xz

xy xy yz

G

G

G

           

     

  

         

    

 

(3.3 & 4)

or 2 21 1

1 1

xy

x x y y y x xy E E G

          

 

            

     (3.5)

equilibrium equations: (2 equations)

Eq (1.14) 

0

0

xyx x

xy y

y

F x y

F x y



 

   

       

  

(3.6)

SECTION 3.3 PLANE STRAIN PROBLEMS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

traction boundary conditions: (2 equations)

Eq (1.26 or 48)  x x xy

y xy y

p l m

p l m

 

 

  

  (3.7)

FIGURE 3.2 Surface forces (or tractions).

compatibility equation: (1 equation)

Substitute Eqs (3.1 & 2) into Eq (2.12), we get

Saint-Venant strain compatibility equation:

2 22

2 2

y xyx

y x x y

     

    (3.8)

Eq (3.5) into Eq (3.8)      22 2

2 2 1 1 2 xy

x y y x y x x y

      

                

(a)

From Eq (3.6), we have 2 22

2 2 2

y yxy x x FF

y yx y x x

                    

(b)

Substitute Eq (b) into Eq (a), we obtain the Beltrami-Michell stress compatibility equation:

    2 2

2

2 2

1

1

yx yxx y

FF

x y x y   

              

       (3.9)

where the 2-D Laplacian operator in Cartesian coordinates: 2 2

2

2 2x y

    

  (3.3-A2)

+

Special Case: zero-body forces  0x yF F 

The Beltrami-Michell stress compatibility equation, Eq (3.9), becomes

    2 2

2

2 2 0yxx y

x y   

         

   ~ 2-D Laplace equation (3.3-A3)

Generalized Plane-Strain Condition: constant 0z  

~ e.g., rotating long cylinders or pressurized hollow thick cylinders with closed ends.

SECTION 3.4 PLANE STRESS PROBLEMS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.4 PLANE STRESS PROBLEMS (Lecture)

FIGURE 3.3 Thin plate under plane stress.

Uplane-strain conditionU in x-y plane: z-independence: 0 z

 

stress field: (3 non-zero stress components)

           , , , , ,

0

x x x y y y xy xy xy

z xz yz

x y z x y z x y z        

  

      

   (a)

stress-strain (or constitutive) relations: (4 non-zero strain components, only 3 independent)

Eq (a) into Eq (2.34) 

     

 

1

1

1 0

0

x x y z x y x y

y y x xz

xy

xy yz

E E

E

G

         

   

  

         

 

    

  

(3.10 & 11)

or    2 21 1 x x y y y x xy xy

E E G       

      

  (3.4-A1)

strain-displacement (or kinematic) relations: (4 non-zero relations)

Eq (2.4) 

0 0

x y z

xy xz yz

u v w

x y z

u v u w v w

y x z x z y

  

  

        

              

      

(3.4-A2)

equilibrium equations: (2 equations)

Eq (1.14) 

0

0

xyx x

xy y

y

F x y

F x y



 

   

       

  

~ same as the plane-strain (pl-) case (3.6)

SECTION 3.4 PLANE STRESS PROBLEMS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

compatibility equation: (1 equation)

Substitute Eq (3.4-A2) into Eq (2.12), we get

Saint-Venant strain compatibility equation:

2 22

2 2

y xyx

y x x y

     

    ~ same as the pl- case (3.8)

Eqs (3.10 & 11) into Eq (3.8)        22 2

2 2 2 1

xy

x y y x y x x y

     

      

    (a)

From Eq (3.6), we have 2 22

2 2 2

y yxy x x FF

y yx y x x

                    

(b)

Substitute Eq (b) into Eq (a), we obtain the Beltrami-Michell stress compatibility equation:

      2 2

2

2 2 1

yx yxx y

FF

x y x y    

               

       (3.12)

+

Special Case: zero-body forces  0x yF F 

The Beltrami-Michell stress compatibility equation, Eq (3.12), becomes

    2 2

2

2 2 0yxx y

x y   

         

   ~ 2-D Laplace equation~ same as the pl- case (3.3-A3)

Generalized Plane-Strain Condition:   0z z  yet   1

0 2

h

z h

z dz h

 



Stress–Strain Relations for Orthotropic Materials (Self-Study)

0

0

yx y yzx xz x z x y

y xy x

y xz

xy

xy yz

xy

E E E E

E E

G

       

    

  

     

 

    

  

(3.10 & 11)

or     1 1

x x yx y y y xy x xy xy xy

xy yx xy yx

E E G         

        

  (3.15)

SECTION 3.5 COMPARISON OF TWO-DIMENSIONAL ISOTROPIC PROBLEMS PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.5 COMPARISON OF TWO-DIMENSIONAL ISOTROPIC PROBLEMS (Lecture)

Table 3.1 Conversion between plane stress and plane strain solutions (Enhanced)

Material Property To convert from pl- solution

to pl- solution

To convert from pl- solution

to pl- solution

Young’s Modulus

E

Replace E in pl- solution by

21

E



Replace E in pl- solution by

 

  2

1 2

1 E

Poisson’s Ratio

Replace  in pl- solution by

1



Replace  in pl- solution by

1



Shear Modulus

G no change no change

Lamé’s Constant

Replace  in pl- solution by

1 2

1 3

 

Replace  in pl- solution by

1 2

1

 

Bulk Modulus

K

Replace K in pl- solution by

 

  

1 2

1 3 1 K

 

 

Replace K in pl- solution by 2

2

1 4

1 K

Coefficient of Linear

Thermal Expansion

Replace  in pl- solution by

 1 

Replace  in pl- solution by

1

1 2

 

Notes: a. Shear modulus G remains the same between the pl- vs pl- conversion.

b. When 0  , pl- case  pl- case.

  • 3.2 Fundamental Principles of Analysis (2018-02-09)
  • 3.3 Plane Strain Problems (2018-02-09)
  • 3.4 Plane Stress Problems (2018-02-09)
  • 3.5 Comparison of 2-D Isotropic Problems (2018-02-09)

__MACOSX/stress/._Ch 3 Problems in Elasticity A-Part 1 (2018-02-09).pdf

stress/Ch 3 Problems in Elasticity A-Part 2 (2018-03-02)(1).pdf

SECTION 3.6 AIRY’S STRESS FUNCTION PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.6 AIRY’S STRESS FUNCTION (Lecture)

Assume a potential function  ,V x y exists and relates to the body forces xF and yF such that:

 

 

,

,

x

y

V x y F

x

V x y F

y

  

 

    

(3.6-A1)

Note: Many body forces (e.g., gravity loading, electromagnetic field, etc.) found in applications fall into such

a category.

plane (or 2-D) equilibrium eqs, Eq (3.6) 

 

 

0

0

xyx

yxy

V

x y

V

x y





   

  

     

The equations will be satisfied identically if

   

   

 

2

2

2

2

2

, ,

, ,

,

x

y

xy

x y x yV

y

x y x yV

x

x y

x y

    

   

  

    

 

(3.6-A2)

Specially, no body force 0x yF F V   :

 

 

 

2

2

2

2

2

,

,

,

x

y

xy

x y

y

x y

x

x y

x y

   

   

 

    

 

(3.16)

Here  ,x y  is called the Airy’s stress function.

Substitute Eqs (3.6-A1) & (3.6-A2) into plane (or 2-D) compatibility eqs: Eqs (3.9) & (3.12), we have:

     

     

       

       

2 22 2

2 2 2

2 22 2

2 22 2

2 2 2

2 22 2

pl- :

pl-

, ,, , 1 1 2 , ,

1 1

, ,, , 2 1: 1, ,

V x y V x yx y x y x y x yV V

x yx y

V x y V x yx y x y x y x yV V

x yx y

 

 

                     

         

                         

  

SECTION 3.6 AIRY’S STRESS FUNCTION PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

 

4 4 4 4 2 2 2

4 2 2 4

4 4 4 4 2 2 2

4 2 2 4

pl- :

pl- 2 1:

1 2 2

1 V

x x y y

V x x y y

               

    

                  

  

(3.6-A3)

where the 2-D biharmonic operator in Cartesian coordinates is defined as:

2 2 2 2 4 4 4 4 2 2

2 2 2 2 4 2 42 2

x y x y x x y x

                   

         (3.6-A4)

Special Case: zero-body forces  0x yF F    2 , 0x yV 

 4 4 4

4 2 2

4 2 2 4 2 0

x x y y

              

    ~ 2-D biharmonic equation (3.17)

Note: For the case of zero body forces, the equation governing Airy stress function, Eq (3.17), is the same for

both pl- and pl- problems; that is, the equation is independent of elastic constants.

SECTION 3.7 SOLUTION OF ELASTICITY PROBLEMS PAGE 1/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.7 SOLUTION OF ELASTICITY PROBLEMS (Lecture)

Direct Method: This method seeks to determine the solution by direct integration of the field equations

(Sec. 3-2) governing the stress, strain and displacement components. Boundary conditions are to be satisfied

exactly. This method normally encounters significant mathematical difficulties, thus limiting its application

to problems with simple geometry.

Inverse Method: For this technique, particular displacements or stresses are selected that satisfy the basic

field equations. A search is then conducted to identify a specific problem that would be solved by this

solution field. This amounts to determine appropriate problem geometry, boundary conditions, and body

forces that would enable the solution to satisfy all conditions on the problem. Using this scheme it is

sometimes difficult to construct solutions to a specific problem of practical interest.

Semi-Inverse Method: In this scheme part of the displacement and/or stress field is specified, and the other

remaining portion is determined by the fundamental field equations (normally using direct integration) and

the boundary conditions. It is often the case that constructing appropriate displacement and/or stress solution

fields can be guided by approximate strength of materials theory. The usefulness of this approach is greatly

enhanced by employing Saint-Venant’s principle, whereby a complicated boundary condition can be

replaced by a simpler statically equivalent distribution.

Analytical Solution Techniques a. Power-series (or polynomial) method

b. Separation of variables method using Fourier series/integrals

c. Integral transform methods: Laplace, Fourier, Mellin, Hankel, Legendre, Tchebycheff, Kontorovich-

Lebedev, Mahler-Fock, etc.

d. Complex variable methods: Muskhelishvili, Lekhnitskii, Westergaard, etc.

Polynomial Solutions

no body force 0x yF F   2-D biharmonic eq: 4 4 4

4 2 2

4 2 2 4 2 0

x x y y

              

    (3.17)

separation of variables  assume Airy’s stress function:

  0 0

2 2 3 2 2 3

00 10 01 20 11 02 30 21 12 03

4 3 2 2 3 4

40 31 22 13 04

5 4 3 2 2 3 4 5

50 41 32 23 14 05

6 5 4 2 3 3 2 4 5

60 51 42 33 24 15

, m n

mn

m n

x y A x y

A A x A y A x A xy A y A x A x y A xy A y

A x A x y A x y A xy A y

A x A x y A x y A x y A xy A y

A x A x y A x y A x y A x y A xy

 

 

 

         

    

     

     



6

06A y

(3.7-A1)

Substitute Eq (3.7-A1) into Eq (3.17):

           4 2 2 4

4 0 2 2 0 4

1 2 3 2 1 1 1 2 3 0m n m n m n

mn mn mn

m n m n m n

m m m m A x y m m n n A x y n n n n A x y      

   

     

            

Collect like powers of x and y:

              2 2

2, 2 2, 2

2 2

2 1 1 2 1 1 2 1 1 0m n

m n mn m n

m n

m m m m A m m n n A n n n n A x y  

 

   

 

            

              2, 2 2, 22 1 1 2 1 1 2 1 1 0m n mn m nm m m m A m m n n A n n n n A              (3.7-A2)

SECTION 3.7 SOLUTION OF ELASTICITY PROBLEMS PAGE 2/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

4

2 2

3 2 2 3

3 2 2 3 4

5 4 2 3 3 2 4 5

6 5 4 2 3 3 2 4 5 6

1

x x y x y xy y

x x y x y x y xy y

x x y x y x y x

x x

y

y y

x x y xy y

xy

y

y

x

FIGURE 3.7-A1 The Pascal triangle.

For a complete polynomial of degree 6 

40 22 04

50 32 14

23 41 05

60 42 24

51 33 15

42 24 06

3 3 0

5 0

5 0

15 2 0

5 3 5 0

2 15 0

A A A

A A A

A A A

A A A

A A A

A A A

   

        

       

  

(3.7-A3)

no body force 0x yF F  :      2 2 2

2 2

, , , xyx y

x y x y x y

y x x y  

        

    (3.16)

Substitute Eq (3.7-A1) into Eq (3.16), we obtain:

2 2 3 2 2 3

02 12 03 22 13 04 32 23 14 05

4 3 2 2 3 4

42 33 24 15 06

2 2 6 2 6 12 2 6 12 20

2 6 12 20 30

x A A x A y A x A xy A y A x A x y A xy A y

A x A x y A x y A xy A y

          

     (3.7-A4a)

2 2 3 2 2 3

20 30 21 40 31 22 50 41 32 23

4 3 2 2 3 4

60 51 42 33 24

2 6 2 12 6 2 20 12 6 2

30 20 12 6 2

y A A x A y A x A xy A y A x A x y A xy A y

A x A x y A x y A xy A y

          

     (3.7-A4b)

2 2 3 2 2 3

11 21 12 31 22 13 41 32 23 14

4 3 2 2 3 4

51 42 33 24 15

2 2 3 4 3 4 6 6 4

5 8 9 8 5

xy A A x A y A x A xy A y A x A x y A xy A y

A x A x y A x y A xy A y

           

     (3.7-A4c)

Constant and Linear Polynomial:   00 10 01 ,x y A A x A y   

Apparently, the constant and linear polynomial of the Airy stress function satisfy Eq (3.17), the biharmonic

eq: 4 4 4

4 2 2

4 2 2 4 2 0

x x y y

              

    , identically and generate zero stress.

Quadratic Polynomial:   2 2

20 11 02 ,x y A x A xy A y    (3.20)

The quadratic polynomial satisfies the biharmonic eq identically and generate the following stress

components:

SECTION 3.7 SOLUTION OF ELASTICITY PROBLEMS PAGE 3/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

02 20 112 2 xyx y

A A A     (3.7-A5)

which correspond to a constant stress state: biaxial tension  pure shear.

Cubic Polynomial:   3 2 2 3

30 21 12 03 ,x y A x A x y A xy A y     (3.21)

The cubic polynomial satisfies the biharmonic eq identically and generate the following stress components:

12 03 30 21 21 122 6 6 2 2 2 xyx y

A x A y A x A y A x A y        (3.7-A6)

Special Case:

1. 03 0A  , 30 21 12 0A A A    036x A y  , 0y xy    pure bending

2. 30 0A  , 21 12 03 0A A A    306y A x  , 0x xy    pure bending

FIGURE 3.4 Stress fields of (a) Eq. (3.20) and (b) Eq. (3.21).

4 th

-Order Polynomial:   4 3 2 2 3 4

40 31 22 13 04 ,x y A x A x y A x y A xy A y      (3.22)

Since  4

404

, 24

x y A

x

  

 ,

 4

222 2

, 4

x y A

x y

  

  &

 4

044

, 24

x y A

y

  

 , then in order for the 4

th -order

polynomial to satisfy the biharmonic eq, we must have:

4 4 4

40 22 044 2 42 2 24 8 24 0A A A

x x y y

          

    40 22 043 3 0A A A   (3.7-A3a)

thus, the coefficients are no longer independent. The associated stress components are:

2 2

22 13 04

2 2

40 31 22

2 2

31 22 13

2 6 12

12 6 2

3 4 3

x

y

xy

A x A xy A y

A x A xy A y

A x A xy A y

    

       

 2 2

22 13 40 22

2 2

40 31 22

2 2

31 22 13

2 6 12 4

12 6 2

3 4 3

x

y

xy

A x A xy A A y

A x A xy A y

A x A xy A y

     

       

(3.7-A7)

Special Case: 40 0A  , 31 22 13 04 0A A A A     2

4012x A y   , 2

4012y A x  , 0xy 

 biaxial parabolic tension/compression

5 th

-Order Polynomial:   5 4 3 2 2 3 4 5

50 41 32 23 14 05 ,x y A x A x y A x y A x y A xy A y       (3.23)

Since  4

50 414

, 120 24

x y A x A y

x

   

 ,

 4

32 232 2

, 12 12

x y A x A y

x y

   

  &

 4

14 054

, 24 120

x y A x A y

y

   

 , then

SECTION 3.7 SOLUTION OF ELASTICITY PROBLEMS PAGE 4/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

in order for the 5 th

-order polynomial to satisfy the biharmonic eq, we must have:

      4 4 4

50 41 32 23 14 054 2 42 2 120 24 2 12 12 24 120 0A x A y A x A y A x A y

x x y y

              

  

    50 32 14 23 41 0524 5 24 5 0A A A x A A A y       50 32 14

23 41 05

5 0

5 0

A A A

A A A

   

   (3.7-A3b,c)

The associated stress components are:

3 2 2 3

32 23 14 05

3 2 2 3

50 41 32 23

3 2 2 3

41 32 23 14

2 6 12 20

20 12 6 2

4 6 6 4

x

y

xy

A x A x y A xy A y

A x A x y A xy A y

A x A x y A xy A y

     

         

 

 

   

3 2 2 3

32 23 50 32 05

3 2 2 3

50 05 23 32 23

3 2 2 3

05 23 32 23 50 32

2 6 60 12 20

20 60 12 6 2

20 4 6 6 20 4

x

y

xy

A x A x y A A xy A y

A x A A x y A xy A y

A A x A x y A xy A A y

      

           

(3.7-A8)

6 th

-Order Polynomial:   6 5 4 2 3 3 2 4 5 6

60 51 42 33 24 15 06 ,x y A x A x y A x y A x y A x y A xy A y       

Since  4

2 2

60 51 424

, 360 120 24

x y A x A xy A y

x

    

 ,

 4

2 2

42 33 242 2

, 24 36 24

x y A x A xy A y

x y

    

  &

 4

2 2

24 15 064

, 24 120 360

x y A x A xy A y

y

    

 , then in order for the 6

th -order polynomial to satisfy the

biharmonic eq, we must have:

   

 

4 4 4 2 2 2 2

60 51 42 42 33 244 2 42

2 2

24 15 06

2 360 120 24 2 24 36 24

24 120 360 0

A x A xy A y A x A xy A y x x y y

A x A xy A y

            

  

   

      2 2

60 42 24 51 33 15 42 24 0624 15 2 24 5 3 5 24 2 15 0A A A x A A A xy A A A y        

60 42 24

51 33 15

42 24 06

15 2 0

5 3 5 0

2 15 0

A A A

A A A

A A A

   

      

(3.7-A3d-f)

The associated stress components are:

4 3 2 2 3 4

42 33 24 15 06

4 3 2 2 3 4

51 42 33 24

4 3 2 2 3 4

51 42 33 24 15

2 6 12 20 30

30 20 12 6 2

5 8 9 8 5

x

y

xy

A x A x y A x y A xy A y

x A x y A x y A xy A y

A x A x y A x y A xy A y

      

           

     

     

     

4 3 2 2 3 4

06 60 51 15 60 06 15 06

4 3 2 2 3 4

51 06 60 51 15 60 06

4 3 2 2 3 4

51 60 06 51 15 06 60 15

10 20 10 10 60 120 20 30

30 20 60 120 10 10 10 20

5 80 40 15 15 80 40 5

x

y

xy

A A x A A x y A A x y A xy A y

x A x y A A x y A A xy A A y

A x A A x y A A x y A A xy A y

         

                 

(3.7-A9)

SECTION 3.8 THERMAL STRESSES PAGE 1/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

3.8 THERMAL STRESSES (Lecture)

Def: thermal strains: normal:

shear: 0

t

t

T 

 

 (3.25)

where T is the temperature change. That is free expansion or contraction, i.e., equi-dimensional volume

change only, no distortion (or change in shape).

temperature change: T

L

l

r

R

B

H

coefficient of linear thermal expansion: 

h

b

Figure 3.8-A1 Free thermal expansion of a block with a circular hole due to temperature change T.

As shown in Fig 3.8-A1, when subject to a temperature rise T, a block of dimensions: lbh with a circular

hole of radius r will be expanded to a larger block of dimensions: LBH with a bigger circular hole of radius

R. Assume the block has a coefficient of linear thermal expansion . The dimensions after free thermal

expansion are:

       

       

1 1 1 1

1 1 1 1

t t

t t

L T B b b T

H h h T R r r T

   

   

       

        (3.8-A1)

Equations of Thermoelasticity

1-D Case:

total strain elastic strain thermal strain  or normal strain:

shear strain:

el t el

el t

T    

  

   

  el

 

 (3.8-A2)

 Hooke’s law:  normal:

shear:

el t

el t

E E E E T

G G

     

   

    

   G

 

    

(3.8-A3a)

or

normal:

shear:

el t

el t

T E

    

  

   

  G

     

(3.8-A3b)

SECTION 3.8 THERMAL STRESSES PAGE 2/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

Notes: (1) Thermal stresses may develop in a statically indeterminate structure or result from nonuniform

temperature change.

(2) In general, thermal stresses are self-equilibrating so long as there are no reaction forces on the

boundary. This implies a design issue: residual stresses may be caused by thermal stresses

resulting from thermal treatment and/or manufacturing processes.

3-D Case (Duhamel-Neumann thermoelastic constitutive laws):

total strain elastic strain thermal strain  or , ,ij ij el t ij el T        (3.8-A4)

 

 

 

1

1

1

xy

x x y z xy

yz

y y x z yz

xz z z x y xz

T E G

T E G

T E G

       

       

       

        

 

         

        

(3.8-A5)

pl- Case:

2 21 1

1 1

xy

x x y y y x xyT T E E G

            

 

              

     (3.8-A6)

       

 

2 3 2 2 3 2x x y x y x y y

xy xy z x y

G T G T

G E T

               

      

           

   

(3.8-A7)

pl- Case:

   

 

1 1 x x y y y x

xy

xy z x y

T T E E

T G E

       

      

      

       

(3.26a)

    2 21 1 1 1 x x y y y x xy xy

E E T E E T G

         

          

    (3.26b)

Compatibility Equation:

no body force      2 2

2

2 2 0yxx y

E T E T x y

       

           

(3.27)

Combine with Eq (3.16), we obtain:

4 4 4 4 2 2 2 2 2

4 2 2 4 2 0E T E T E T

x x y y   

                   

     (3.28)

Def: thermal force:

thermal moment:

t A

T A

N E TdA

M E TydA

   



 (3.8-A8)

SECTION 3.8 THERMAL STRESSES PAGE 3/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

 EXAMPLE: 1-D THERMO-MECHANICAL STRESSES

THERMO-MECHANICAL STRESS IN AN AXIAL BAR (Lecture)

As shown in the figure below, a bar of length L, cross-sectional area A, Young’s modulus E and coefficient of

linear thermal expansion  is fixed at one end A and restrained at the other end B by a spring of spring constant

k. Assume the structure is stress free initially and the spring is insensitive to temperature change.

a. Thermo-mechanical stress, elongation & reactions. If the axial bar is undergone simultaneously a

temperature change T and subject to an axial force F at the bar-spring interface B, determine the

thermo-mechanically induced stress  in the bar, elongation Bu at the interface B and the reactions at the two

rigid walls,  ,A CN N , respectively.

b. Special cases. Assume 0F  , find  , , ,B A Cu N N two special cases:

:

: 0

rigid restraint

free restraint

i k

ii k

 



EA

L

B

A

k temperature change: T

Cx,u

y,v

F

Sol:

a. Thermo-mechanical stress, elongation & reactions.

Taking a free-body diagram at the bar-spring interface B, as shown below, we have

interface 0

B F   C AN N F  (a1)

N A

B

F

N C

CA

k EAL

axial bar,elastic axial bar,elastic

axial bar,thermal axial bar,thermal

spring

A

C

N L u L L

E EA

u L TL

N u

k

 

 

   

  

    

(a2)

compatibility condition: axial bar axial bar,elastic axial bar,thermal springu u u u   (a3)

Substitute Eq (a2) into Eq (a3)  CA NN L

TL EA k

   (a4)

SECTION 3.8 THERMAL STRESSES PAGE 4/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

Combine Eqs (a1) and (a4)   

A

EA F k TL N

EA kL

 

 ~ reaction force @ wall A (a5)

Substitute Eq (a5) into Eq (a1)   

C

kL F EA T N

EA kL

  

 ~ reaction force @ wall C (a6)

 thermal stress:  

A E F k TLN

A EA kL

 

  

 (a7)

elongation @ the bar-spring interface B:  

axial bar springB

F EA T L u u u

EA kL

   

 (a8)

b. Special cases.

i) rigid restraint: 0F  & k   E T   , 0Bu  & A CN N EA T   (b1)

ii) free restraint: 0F k   0  , Bu TL & 0A CN N  (b2)

SECTION 3.8 THERMAL STRESSES PAGE 5/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

SECTION 3.8 THERMAL STRESSES PAGE 6/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 3 – PROBLEMS IN ELASTICITY

  • 3.6 Airy's Stress Function (2018-02-09)
  • 3.7 Solution of Elasticity Problems (2018-02-09)
  • 3.8 Thermal Stresses (2018-03-02)

__MACOSX/stress/._Ch 3 Problems in Elasticity A-Part 2 (2018-03-02)(1).pdf

stress/Ch 4 Failure Criteria-Part 1 (2018-01-25).pdf

SECTION 2.6 ENGINEERING MATERIALS PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

CHAPTER 4: FAILURE CRITERIA

2.6 ENGINEERING MATERIALS (Self-Study)

Table 2.1 Typical Engineering Materials

Common Engineering Materials:  Metals

 Ceramics

 Polymers

 Composites

Advanced/Modern Materials:  Smart materials (e.g., shape memory alloys, piezoelectric materials, magnetostrictive materials)

 Nanomaterials (e.g., fullerenes, carbon nanotubes, graphene)

 Metamaterials (e.g., auxetics - tunable/negative Poisson’s ratios)

SECTION 2.7 STRESS–STRAIN DIAGRAMS PAGE 1/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

2.7 STRESS-STRAIN DIAGRAMS (Self-Study)

The Complete Stress-Strain Curve

Figure 2.7-A1 Typical uniaxial stress-strain curves for three structural metals.

E

1

E

1

1

4

2

3

hysteresis

fracture

necking

(strain softening)

strain (work)

hardening non-linearly

elastic

linearly

elastic

1

2

3

4 tensile strength (ultimate stress)

yield strength (lower yield point)

elastic limit (upper yield point)

proportional limit

Lüders (slip, shear)

bands

 plastic

 elastic

plasticelastic

Figure 2.7-A2 The complete stress-strain curve.

SECTION 2.7 STRESS–STRAIN DIAGRAMS PAGE 2/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Ductile Materials in Tension

FIGURE 2.10 (a) Stress–strain diagram of a typical ductile material;

(b) determination of yield strength by the 0.2% offset method.

FIGURE 2.11 Geometry change in a typical round specimen of ductile material in tension:

(a) necking; (b) fractured.

0

0

0

0

percent elongation 100%

percent reduction in area 100%

f

f

L L

L

A A

A

  

 

    

(2.23)

True Stress and True Strain

 

  0

0

0 0

0

engineering nominal strain:

true logarithmic strain: ln

o

L

L

L L L or

L L

L L or

dL L

   

     

(2.11-A1)

0

0 0

ln ln L LL

L L 

      ln 1 o   (2.24)

Within the plastic range , material is incompressible  volume constancy: 0V AL AL  (a)

0 0 0

0 0 0 0

L LP P L L

A A L L L   

      true stress:  0 1 o    (2.25)

SECTION 2.7 STRESS–STRAIN DIAGRAMS PAGE 3/3

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

FIGURE 2.12 Stress–strain curves for a plain low-carbon 1005 steel (0.05% wt C) in tension.

Brittle Materials in Tension

FIGURE 2.13 Cast iron in tension: (a) Stress–strain diagram; (b) fractured specimen.

SECTION 2.8 ELASTIC VS. PLASTIC BEHAVIOR PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

2.8 ELASTIC VS PLASTIC BEHAVIOR (Self-Study)

Figure 2.14 Stress–strain diagrams showing (a) elastic behavior; (b) partially elastic behavior.

SECTION 4.3 FAILURE BY YIELDING PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.3 FAILURE BY YIELDING (Self-Study)

Creep & Relaxation: viscoelastic behaviors (also called time-dependent elastoplastic behaviors)

Figure 4.1 Typical creep curves for a bar in tension.

SECTION 4.4 FAILURE BY FRACTURE PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.4 FAILURE BY FRACTURE (Self-Study)

Types of Fracture in Tension

Figure 4.2 Necking of a bar in tension: (a) the distribution of the axial stresses; (b) stress elements in the plane

of the minimum cross section.

Progressive Fracture: Fatigue

Figure 4.3 Typical S–N diagram for steel.

  • 2.6 Engineering Materials (2018-01-25)
  • 2.7 Stress–Strain Diagrams (2018-01-25)
  • 2.8 Elastic vs Plastic Behavior (2018-01-25)
  • 4.3 Failure by Yielding (2018-01-25)
  • 4.4 Failure by Fracture (2018-01-25)

__MACOSX/stress/._Ch 4 Failure Criteria-Part 1 (2018-01-25).pdf

stress/Ch 4 Failure Criteria-Part 2 (2018-02-28).pdf

SECTION 4.6 MAXIMUM SHEAR STRESS THEORY (TRESCA YIELDING CRITERION) PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.6 MAXIMUM SHEAR STRESS THEORY-TRESCA YIELDING CRITERION (Lecture)

Maximum Shear Stress Theory (Tresca Criterion)

Yielding occurs when the maximum shear stress reaches the value of the yield strength under simple shear;

that is:

then

2 3 yp yp 2 1 3yp

1 2 3 yp

3 2 yp yp 2 1 3

3 1 yp yp 3 2 1yp

2 3 1 yp

1 3 yp yp 3 2 1

1 2 yp yp 1 3 2yp

3 1 2 yp

2 1 yp yp

2 if1

2 if2 2

2 if1

2 if2 2

2 if1

22 2

          

      

          

      

          

   

          

    

          

    

         

   1 3 2if   

       

  

  

(4.1 & 2)

where Def: yield strength in shear: yp

yp 2

   (4.6-A1)

Note: In 3-D, the maximum shearing stressing theory/Tresca criterion is a regular hexagon formed by 6 linear

equations.

Plane-stress case:  0 z xz yz

      3

0 

1 2 yp 1 2 1 2 yp 2 1

1 yp 1 2 1 yp 1 2

2 yp 2 1 2 yp 2 1

if 0 if 0

if 0 if 0

if 0 if 0

        

       

       

          

              

~ irregular hexagon (4.3)

 yp

 yp

 yp

 yp

 1

 2

yp1 2    

yp2 1    

Figure 4.4 Yield criterion based on the maximum shearing stress theory (Tresca criterion)

for plane-stress cases.

Notes: a. A limitation of the maximum shear stressing criterion is the requirement that yield strengths in

tension and compression be equal.

b. The theory is in fair agreement with experiments and is used to a considerable extent by designers.

c. However, it suffers from a major setback; that is, it is necessary to know in advance which are the

maximum and minimum principal stresses.

SECTION 4.7

SECTION 4.8

MAXIMUM DISTORTION ENERGY THEORY (VON MISES YIELDING CRITERION)

OCTAHEDRAL SHEARING STRESS THEORY PAGE 1/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.7 MAXIMUM DISTORTION ENERGY THEORY (VON MISES YIELDING CRITERION) (Lecture)

4.8 OCTAHEDRAL SHEAR STRESS THEORY (Lecture)

Maximum Distortion Energy Theory (von Mises Criterion)

Octahedral Shear Stress Theory

Yielding begins when the distortion energy density odU equals the distortion energy density at yield ,ypodU in

simple tension. Here:

     

       

2 2 2 2oct 2

1 2 2 3 3 1

2 2 2 2 2 2

3 1

4 122

1 6

12

od

x y y z z x xy yz zx

J U

G GG

G

      

        

          

            

(2.65)

For yielding in uniaxial tension: 1 yp

2 3 0

 

 

 

  

2

,yp yp

1

6 odU

G 

Thus, ,ypod odU U 

     

     

2 2 2 2 2 2 2

yp

2 2 2 2

1 2 2 3 3 1 yp

oct yp yp

2

2 yp

vM yp

6 6 6 2

2

2 0.47

3

1

3

x y y z z x xy yz xz

or

J

         

      

  

 

                   

  

 



(4.4 & 6)

Plane-stress case:  0 z xz yz

      3

0  

2 2 2

1 1 2 2 yp

2 2 2 2

yp3x x y y xy

    

     

    

    ~ ellipse (4.5)

Notes: a. Hencky showed that the statement in Eq (4.6): oct yp yp

2 0.47

3 or   was equivalent to

assuming that yielding will take place when the octahedral shear stress oct reaches the octahedral

shear stress at yield in simple tension oct,yp yp0.47  , which has a difference of about 5.72% from

the maximum shear stress at yield in simple tension yp

max,yp 2

   .

b. By the same token, the statement: 2

2 yp

1

3 J  implies that yielding will occur when the 2

nd invariant

2J of the stress deviatoric tensor ijs equals a critical value at yield in simple tension 2

2,yp yp

1

3 J  .

c. Similarly, the statement: vM yp  implies that yielding will occur when the equivalent (or

effective or effective or generalized or von Mises) stress vM reaches a critical value at yield in

simple tension vM,yp yp  .

SECTION 4.7

SECTION 4.8

MAXIMUM DISTORTION ENERGY THEORY (VON MISES YIELDING CRITERION)

OCTAHEDRAL SHEARING STRESS THEORY PAGE 2/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

 yp

 yp

 yp

 yp

2 2 2

1 1 2 2 yp       

 1

 2

Figure 4.5(a) The distortion energy theory (von Mises-Hencky criterion) for plane-stress cases.

SECTION 4.7

SECTION 4.8

MAXIMUM DISTORTION ENERGY THEORY (VON MISES YIELDING CRITERION)

OCTAHEDRAL SHEARING STRESS THEORY PAGE 3/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

SECTION 4.7

SECTION 4.8

MAXIMUM DISTORTION ENERGY THEORY (VON MISES YIELDING CRITERION)

OCTAHEDRAL SHEARING STRESS THEORY PAGE 4/4

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

SECTION 4.9 COMPARISON OF THE YIELDING THEORIES PAGE 1/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.9 COMPARISON OF THE YIELDING THEORIES (Lecture)

Table 4.1 Shear stress and strain energy at the start of yielding

Figure 4.9-A1 Plane-stress Tresca vs von Mises yield loci on the 1 2-  plane when

3 0  .

Question: From Design Perspective, which yielding criterion is more conservative?

SECTION 4.9 COMPARISON OF THE YIELDING THEORIES PAGE 2/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

 EXAMPLE: SIMPLE YIELD CRITERIA

PURE SHEAR

A rectangular panel is loaded to yield by an in-plane pure shear. That is, ypxy

  . Correlate the yield strength

in shear yp with the yield strength in uniaxial tension yp on von Mises and Tresca yield criteria, respectively.

Sol:

 yp

 yp

 yp

 yp

 yp

 yp

 yp

 yp

1 yp  

2 yp   

max yp  

pure-shear stress tensor: 

yp

yp

0 0

0 0

0 0 0

x xy xz

ij yx y yz

zx zy z

   

    

  

       

          

(A1)

plane-stress Mohr’s circle 

2

2

yp1

2

2

yp2

3

2 2

2 2

0

x y x y

xy

x y x y

xy

      

      

                            



(A2)

As shown in the left figure, the pure-shear stress state is equivalent to:

bi-axial stress tensor: 

yp

yp

0 0

0 0

0 0 0

x xy xz

ij yx y yz

zx zy z

   

    

  

       

           

(A3)

which will have the same set of principal stresses in Eq (A2).

Now, substitute Eq (A2) into Eq (4.5), von Mises criterion:       2 2 2 2

1 2 2 3 3 1 yp2            , we have:

SECTION 4.9 COMPARISON OF THE YIELDING THEORIES PAGE 3/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

      2 2 2

2

yp yp yp yp yp0 0 2                  

 2 2

yp yp6 2   yp

yp,von Mises yp0.577 3

   (A4)

i.e., based on von Mises criterion, the yield strength in shear yp,von Mises is 1

3 times yield strength in tension

yp .

Similarly, substitute Eq (A2) into Eq (4.3), Tresca criterion: yp

max 1 2

1

2 2

      , we have:

  yp

yp yp

1

2 2

     

yp

yp,Tresca 2

   (A5)

i.e., based on Tresca criterion, the yield strength in shear yp,Tresca is 1

2 times yield strength in tension yp .

The relative different between von Mises and Tresca yield criteria is maximum when the material is in a

pure-shear stress state can be estimated as:

yp yp

yp,von Mises yp,Tresca

ypyp,Tresca

23 %max difference

2

 

 



 

   %max difference 15.47% (A6)

which, as shown in the right figure, represents the maximum relative difference between the von Mises and

Tresca yield criteria under plane-stress condition.

pure shear

 1

 2

 yp

 yp

 yp

 yp

SECTION 4.9 COMPARISON OF THE YIELDING THEORIES PAGE 4/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

 EXAMPLE: THIN-WALLED TUBES

TAYLOR-QUINNEY EXPERIMENT

As shown in the figure, a closed-end thin-walled vessel with a radius R and a thickness t is subject to a

combined loading of an axial force F and a torque T. Define:

2

2

2

F

Rt

T

R t

 

 

 

   

a. Mohr’s circle. Construct the Mohr’s circle under the stated condition.

b. Yield loci. Determine the yield loci based on the von Mises and Tresca criteria. Plot these loci in the same

graph with the normalized Y

 and

Y

 as the coordinates.

c. Implication in engineering design. Comment on possible implication in engineering design.

Sol:

a. Mohr’s circle.

From Mechanics of Materials, we have

2

2

0

2

x

y

xy

F

Rt

T

R t

  

  

  

 

    

(B1)

Then, the Mohr’s circle can be constructed as

2 2

2 2

center: ,0 ,0 2 2

radius: 2 2

x y

x y

xy

  

    

            

    

        

(B2)

2

SECTION 4.9 COMPARISON OF THE YIELDING THEORIES PAGE 5/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

b. Yield loci.

 principal stresses:

2 2

2 2

1

2 2

2 2

2

3

0 2 2 2 2

0 2 2 2 2

0

x y x y

xy

x y x y

xy

       

       

                                         



 plane-stress (B3)

von Mises criterion: 2 2 2

1 1 2 2 yp        22 2 yp3    ~ ellipse (B4)

Tresca criterion: 1 2 yp     2

2

yp2 4

    

2 22

yp4    ~ ellipse (B5)

c. Implications in engineering design.

The graph of yield loci implies that: The experimental data are closer to the prediction made by the

von Mises criterion. However, the engineering design based on the Tresca criterion is more conservative than

the one using the von Mises criterion since the Tresca yield loci of former are fully within the von Mises

yield locus.

  • 4.6 Maximum Shear Stress Theory-Tresca Yielding Criterion (2018-02-08)
  • 4.7 Maximum Distortional Theory-von Mises Yielding Criterion (2018-02-08)
  • 4.9 Comparison of the Yielding Theories (2018-02-08)

__MACOSX/stress/._Ch 4 Failure Criteria-Part 2 (2018-02-28).pdf

stress/Ch 4 Failure Criteria-Part 3 (2018-02-08).pdf

SECTION 4.10 MAXIMUM PRINCIPAL STRESS THEORY-RANKINE FRACTURE CRITERION PAGE 1/1

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.10 MAXIMUM PRINCIPAL STRESS THEORY-RANKINE FRACTURE CRITERION (Lecture)

Fracture occurs when one of the principal stresses becomes equal to the ultimate strength in tension u

 or in

compression u

  , i.e.,

1 2 3or or or u u

      (4.10)

Plane-stress case:  0 z xz yz

      3

0   1 2or or

u u      (4.11)

u  

u  

u 

u 

 1

 2

Figure 4.8 Fracture criterion based on the maximum principal stress theory (Rankine criterion)

for plane-stress case.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 1/5

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.15 FAILURE CRITERIA FOR METAL FATIGUE (Option)

Figure 4.14 Typical stress–time variation in fatigue: an alternating sinusoidal stress a

superimposed on a constant stress m .

 

 

max min

max min

1 mean stress:

2

1 alternating range stress:

2

m

mor

  

  

  

    

(4.20)

Single Loading

Table 4.4 a Failure criteria for fatigue

Modified

Goodman

z

Yield Line

elliptic

Gerber

 u

 yp

 yp

 cr

 m

 a

Soderberg

 

 

cr

2

2 2

c

yp

c

cr

r

r

r

c

Modified Goo

elliptic: 1

least co

Gerbe

Soderberg: 1

most conserva

nservativ

dman: 1

SAE:

ti

r

e

1

ve

: 1

a m

u

a m

u

f

a m

u

a m

a m

 

 

 

 

 

 

      

      

        

       

 

FIGURE 4.15-A1 Comparison of various fatigue failure criteria.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 2/5

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 3/5

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Example 4.7 Fatigue Load of Tension-Bending Bar (Option)

A long plate of width 2w is subjected to a tensile force P in longitudinal direction with a safety factor of n (see

case A, Table 4.2). Determine the thickness t required (a) to resist yielding, (b) to prevent a central crack from

growing to a length of 2a. Given: w  50 mm, P  50 kN, n  3 and a  10 mm. Assumption: The plate will be

made of Ti-6AI-6V alloy. A square prismatic bar of sides 0.05 m is subjected to an axial thrust (tension)

90 kNmF  (Fig 4.15). The fatigue strength for completely reversed stress at 10 6 cycles is 210 MPa and the

static tensile yield strength is 280 MPa. Apply the Soderberg criterion to determine the limiting value of

completely reversed axial load aF that can be superimposed to

mF at the midpoint of a side of the cross section

without causing fatigue failure at 10 6 cycles.

FIGURE 4.15 Bar subjected to constant axial tension mF and alternating eccentric

aF loads.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 4/5

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Combined Loading

       

       

2 2 2 2 2 2 2

2 2 2 2 2 2 2

6 6 6 2

6 6 6 2

xa ya ya za za xa xya yza xza ea

xm ym ym zm zm xm xym yzm xzm em

         

         

                     

(4.21)

or

     

     

2 2 2 2

1 2 2 3 3 1

2 2 2 2

1 2 2 3 3 1

2

2

a a a a a a ea

m m m m m m em

      

      

       

     

(4.22)

Example 4.8 Fatigue Pressure of a Cylindrical Tank (Option)

Consider a thin-walled cylindrical tank of radius r  120 mm and thickness t  5 mm, subject to an internal

pressure varying from a value of 4p to p. Employ the octahedral shear theory together with the Soderberg

criterion to compute the value of p producing failure after 10 8 cycles. The material tensile yield strength is

300 MPa and the fatigue strength is cr 250 MPa  at 10

8 cycles.

Solution

plane stress: 3 3 0a m   

2 2

1 1 2 2

2 2

1 1 2 2

a a a a ea

m m m m em

    

    

    

  

(4.23)

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 5/5

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Fatigue Life

Def : fatigue life:

1 b

cr cr f

f

N N 

     

 

where    

ln

ln

f e

f e

b N N

   (4.24 & 25)

Table 4.5 Fracture stress f (fracture cycles fN ) and fatigue strength e (fatigue life

eN ) for steels

Example 4.9 Fatigue Life of an Assembly (Option)

A rotating hub and shaft assembly is subjected to bending moment, axial thrust, bidirectional torque, and a

uniform shrink fit pressure so that the following stress levels (in MPa) occur at an outer critical point of the

shaft:

700 14 0 660 7 0

14 350 0 , 7 350 0

0 0 350 0 0 350

        

              

These matrices represent the maximum and minimum stress components, respectively. Determine the fatigue

life, using the maximum energy of distortion theory of failure together with (a) the SAE fatigue criterion and (b)

the Gerber criterion. The material properties are 2400 MPau  MPa and K  1.

  • 4.10 Maximum Principal Stress Theory-Rankine Fracture Criterion (2018-02-08)
  • 4.15 Failure Criteria for Metal Fatigue (2017-09-24)

__MACOSX/stress/._Ch 4 Failure Criteria-Part 3 (2018-02-08).pdf

stress/Ch 4 Failure Criteria-Part 4 (2018-02-08).pdf

SECTION 4.13 FRACTURE MECHANICS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.13 FRACTURE MECHANICS (Option)

Mode I Mode II Mode III

(a) (b) (c)

FIGURE 4.12 Crack deformation/ propagation types: (a) mode I, opening, tensile;

(b) mode II, sliding, in-plane shear; (c) mode III, tearing, out-of-plane/anti-plane.

mode I fracture (opening or tensile mode): The crack surfaces move directly apart. It is the most common

fracture mode with a stable crack extension.

mode II fracture (sliding or in-plane shear mode): The crack surfaces slide over one another in a direction

perpendicular to the leading edge of the crack with a crack extension tending to be unstable.

mode III fracture (tearing or out-of-plane/anti-plane shear mode): The crack surfaces move relative to one

another and parallel to the leading edge of the crack with a crack extension also tending to be unstable.

Question: Why Mode I crack extension tends to be stable whereas Mode II & III are not?

Stress-Intensity Factors

Def : stress-intensity factor: K a  or K a  ~ unit: MPa m   

or ksi in   

(4.16)

where applied normal stress full half crack length

geometry factor, depends on , full half member width

a or

a w w or

  

  (see Table 4.2)

SECTION 4.13 FRACTURE MECHANICS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Table 4.2 Geometry-loading factors  for sample crack problems

SECTION 4.14 FRACTURE TOUGHNESS PAGE 1/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.14 FRACTURE TOUGHNESS (Option)

Table 4.3 Yield strength yp and fatigue toughness cK for some materials

fracture-mechanics based factor of safety: c cK K n

K a    (4.17 & 18)

ASTM Standard: , 2.5 c

yp

K a t

     

 

Note: For plane-strain fracture toughness testing (4.19)

SECTION 4.14 FRACTURE TOUGHNESS PAGE 2/2

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Example 4.5 Aluminum Bracket with an Edge Crack (Option)

A 2024-T851 aluminum alloy frame with an edge crack supports a concentrated load (Fig. 4.13a). Determine

the magnitude of the fracture load P based on a safety factor of n = 1.5 for crack length of a  4 mm. The

dimensions are w  50 mm, d  125 mm and t  25 mm.

Solution

2

6 a b

P M

wt tw     (a)

FIGURE 4.13 Aluminum bracket with an edge crack under a concentrated load.

Example 4.6 Titanium Panel with a Central Crack (Option)

A long plate of width 2w is subjected to a tensile force P in longitudinal direction with a safety factor of n (see

case A, Table 4.2). Determine the thickness t required (a) to resist yielding, (b) to prevent a central crack from

growing to a length of 2a. Given: w  50 mm, P  50 kN, n  3 and a  10 mm. Assumption: The plate will be

made of Ti-6AI-6V alloy.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 1/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

4.15 FAILURE CRITERIA FOR METAL FATIGUE (Option)

Figure 4.14 Typical stress–time variation in fatigue: an alternating sinusoidal stress a

superimposed on a constant stress m .

 

 

max min

max min

1 mean stress:

2

1 alternating range stress:

2

m

mor

  

  

  

    

(4.20)

Single Loading

Table 4.4 a Failure criteria for fatigue

Modified

Goodman

z

Yield Line

elliptic

Gerber

 u

 yp

 yp

 cr

 m

 a

Soderberg

 

 

cr

2

2 2

c

yp

c

cr

r

r

r

c

Modified Goo

elliptic: 1

least co

Gerbe

Soderberg: 1

most conserva

nservativ

dman: 1

SAE:

ti

r

e

1

ve

: 1

a m

u

a m

u

f

a m

u

a m

a m

 

 

 

 

 

 

      

      

        

       

 

FIGURE 4.15-A1 Comparison of various fatigue failure criteria.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 2/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 3/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Example 4.7 Fatigue Load of Tension-Bending Bar (Option)

A long plate of width 2w is subjected to a tensile force P in longitudinal direction with a safety factor of n (see

case A, Table 4.2). Determine the thickness t required (a) to resist yielding, (b) to prevent a central crack from

growing to a length of 2a. Given: w  50 mm, P  50 kN, n  3 and a  10 mm. Assumption: The plate will be

made of Ti-6AI-6V alloy. A square prismatic bar of sides 0.05 m is subjected to an axial thrust (tension)

90 kNmF  (Fig 4.15). The fatigue strength for completely reversed stress at 10 6 cycles is 210 MPa and the

static tensile yield strength is 280 MPa. Apply the Soderberg criterion to determine the limiting value of

completely reversed axial load aF that can be superimposed to

mF at the midpoint of a side of the cross section

without causing fatigue failure at 10 6 cycles.

FIGURE 4.15 Bar subjected to constant axial tension mF and alternating eccentric

aF loads.

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 4/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Combined Loading

       

       

2 2 2 2 2 2 2

2 2 2 2 2 2 2

6 6 6 2

6 6 6 2

xa ya ya za za xa xya yza xza ea

xm ym ym zm zm xm xym yzm xzm em

         

         

                     

(4.21)

or

     

     

2 2 2 2

1 2 2 3 3 1

2 2 2 2

1 2 2 3 3 1

2

2

a a a a a a ea

m m m m m m em

      

      

       

     

(4.22)

Example 4.8 Fatigue Pressure of a Cylindrical Tank (Option)

Consider a thin-walled cylindrical tank of radius r  120 mm and thickness t  5 mm, subject to an internal

pressure varying from a value of 4p to p. Employ the octahedral shear theory together with the Soderberg

criterion to compute the value of p producing failure after 10 8 cycles. The material tensile yield strength is

300 MPa and the fatigue strength is cr 250 MPa  at 10

8 cycles.

Solution

plane stress: 3 3 0a m   

2 2

1 1 2 2

2 2

1 1 2 2

a a a a ea

m m m m em

    

    

    

  

(4.23)

SECTION 4.15 FAILURE CRITERIA FOR METAL FATIGUE PAGE 5/5

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 4 – FAILURE CRITERIA

Fatigue Life

Def : fatigue life:

1 b

cr cr f

f

N N 

     

 

where    

ln

ln

f e

f e

b N N

   (4.24 & 25)

Table 4.5 Fracture stress f (fracture cycles fN ) and fatigue strength e (fatigue life

eN ) for steels

Example 4.9 Fatigue Life of an Assembly (Option)

A rotating hub and shaft assembly is subjected to bending moment, axial thrust, bidirectional torque, and a

uniform shrink fit pressure so that the following stress levels (in MPa) occur at an outer critical point of the

shaft:

700 14 0 660 7 0

14 350 0 , 7 350 0

0 0 350 0 0 350

        

              

These matrices represent the maximum and minimum stress components, respectively. Determine the fatigue

life, using the maximum energy of distortion theory of failure together with (a) the SAE fatigue criterion and (b)

the Gerber criterion. The material properties are 2400 MPau  MPa and K  1.

  • 4.13 Fracture Mechanics (2018-02-08)
  • 4.14 Fracture Toughness (2018-02-08)
  • 4.15 Failure Criteria for Metal Fatigue (2018-02-08)

__MACOSX/stress/._Ch 4 Failure Criteria-Part 4 (2018-02-08).pdf

stress/Stress Singularities in Classical Elasticity I.pdf

ical ay der rom ec- ciate en nts

di-

set

Stress singularities in classical elasticity–I: Removal, interpretation, and analysis

GB Sinclair Department of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803-6413

This review article has two parts, published in separate issues of this journal, which consider the stress singularities that occur in linear elastostatics. In the present Part I, after a brief re- view of the singularities that attend concentrated loads, attention is focused on the singulari- ties that occur away from such loading, and primarily on 2D configurations. A number of ex- amples of these singularities are given in the Introduction. For all of these examples, it is absolutely essential that the presence of singularities at least be recognized if the stress fields are to be used in attempts to ensure structural integrity. Given an appreciation of a stress sin- gularity’s occurrence, there are two options open to the stress analyst if the stress analysis is to actually be used. First, to try and improve the modeling so that the singularity is removed and physically sensible stresses result. Second, to try and interpret singularities that persist in a physically meaningful way. Section 2 of the paper reviews avenues available for the re- moval of stress singularities. At this time, further research is needed to effect the removal of all singularities. Section 3 of the paper reviews possible interpretations of singularities. At this time, interpretations using the singularity coefficient, or stress intensity factor, would appear to be the best available. To implement an approach using stress intensity factors in a general context, two types of companion analysis are usually required: analytical asymptotics to char- acterize local singular fields; and numerical analysis to capture participation in global configu- rations. Section 4 of the paper reviews both types of analysis. At this time, methods for both are fairly well developed. Studies in the literature which actually effect asymptotic analyses of specific singular configurations will be considered in Part II of this review article. The present Part I has 182 references.@DOI: 10.1115/1.1762503#

1 INTRODUCTION

1.1 Objective and scope

Stress singularities are not of the real world. Nonetheless, they can be a real fact of a stress analysis. Then it is essential to take them into account if the analysis is to be of any real use. The primary objective of this review is to assist in this regard. That is, in the first instance, to aid in the all-important task of recognition of a singularity’s presence, then, in the second instance, to aid in removal or interpretation.

Throughout this review we takestress singularitiesas in- volving stresses which, in themselves, are unbounded. Spe- cifically, we are concerned with when such singularities can occur in the linear elastic regime. This is a key regime since elastic response physically precedes plastic flow, so that in- troducing plasticity does not remove the singular character in any true sense.1 To keep the scope of the article within rea- sonable limits, we further restrict attention to materials which are homogeneous, or piecewise so, and isotropic. We

also focus on loading which is quasi-static. For such class elasticity fields, two classes of singular configurations m be distinguished: those wherein singularities occur un concentrated loads, and those wherein they occur away f any concentrated loading. For either, it is important to r ognize the presence of stress singularities and to appre their nature. In what follows we give examples of both, th turn our attention to the latter because it typically prese greater difficulties to the stress analyst.

1.2 Examples of stress singularities under concentrated loads

Concentrated loading configurations induce singularities rectly by applying finite stress resultants~eg, forces, mo- ments! over regions with vanishingly small areas~eg, points, lines!. As such they may be termedsingular loads: Table 1 exhibits the singular character of the stresses for a basic of such loads.

Appl Mech Rev vol 57, no 4, July 2004 25

Transmitted by Editorial Advisory Board Member R. C. Benson

1We expand on this point in Section 2.1.

© 2004 American Society of Mechanical Engineers1

252 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Table 1. Basic singular loads of classical elasticity

Load type 3D stress state at load„r\0…

2D stress state at load„r\0…

Isolated force ord(r 22) ord(r 21) Doublet state ord(r 23) ord(r 22)

i

r u r

u r

t

n l

n

u

d

n on.

ee m

-

Fig. 1 Some limiting configurations for doublet states:a) concen- trated moment,b) force doublet without a moment,c) center of compression

se ity

o a the s

t of be-

ich by ave s, d if se- ite at

ite re- erg it is lly

blet y r-

nal b- rec- as in

In Table 1,r is the distance from the point of applicatio of a singular load, and we have employed the ord notat For a functionf (r ), here this has

f ~r !5ord~r 2g! as r→0 (1.1)

if

r g f ~r !5cÞ0 as r→0 (1.2)

whereg and c are constants. The traditional large orderO notation, in contrast, admits the possibility thatc50. Pro- vided nonzero loads are being applied,c cannot be every- where zero for the stresses in Table 1.

Examples of solutions for isolated force problems in th dimensions are: the point load in the infinite elastic medi by Kelvin ~Thomson@1#!, the normal point load on the su face of an elastic half-space of Boussinesq@2#, the tangential point load on a half-space surface of Cerutti@3#, and point loads within a half-space in Mindlin@4#. A convenient com- pendium of these closed-form solutions may be found Poulos and Davis@5#, Section 2.1. Inspection of these sol tions demonstrates compliance with the order of singula for point loads given in Table 1. Analogous solutions ex for isolated force problems in two dimensions, namely: line load in an infinite elastic medium in Michell@6#, the normal line load on the surface of an elastic half-space Flamant@7#, the tangential line load on a half-space surfa in Boussinesq@8#, and line loads within a half-space i Melan @9#.2 These may be found ibid, Section 2.2, and a demonstrate compliance with their singular order given Table 1.

Examples of doublet states are indicated in Fig. 1. T first of these~Fig. 1a! illustrates a means of obtaining concentrated momentM . This moment is produced by tak ing the limit asd→0 whered is the horizontal separation o two vertical forces of magnitudeF5M /d. The second ar- rangement~Fig. 1b! is a dual of the first and realizes n resultant force or moment in the limit asd→0, yet does have a nontrivial stress field ifF is ord(d21): As a consequence it requires a generalization of the usual notion of a load terming it a ‘‘singular load.’’ The third arrangement~Fig. 1c! is a center of compression produced by superposing the ond in an angular array: It, too, represents a load in a ge alized sense. A precise definition of doublet states in gen is given in Sternberg and Eubanks@11#. Some closed-form solutions for doublet states in three dimensions may be fo in: Love @12# Article 132, Sternberg and Eubanks@11#, Tur- teltaub and Sternberg@13#, Chowdhury@14#, and Chen@15#. Closed-form solutions for doublet states in two dimensio are available in Love@12# Article 152, and Timoshenko an

g se-

in - ity ist he

of ce

so in

he a - f

o

, in

sec- er-

eral

nd

ns

Goodier@16# Articles 36 and 42. The stresses in all of the solutions comply with their respective orders of singular given in Table 1.

The nature of the singularities displayed in Table 1 is, t degree, that expected. For a point force, integration of tractions acting on the surface of a small sphere of radiur centered on the point of application produces a produc stresses withr 2: Hence the stresses can be expected to have like r 22 if a finite force is to result in the limit asr →0. Similarly for a line load, one anticipates stresses wh behave liker 21. And the doublet states, being derivable differentiation of corresponding isolated loads, then beh as r 23 and r 22 in three dimensions and two dimension respectively. However, some care needs to be exercise these expectations are to be realized in the limit by a quence of finite stress fields acting over regions of fin extent—a limiting process for producing singular loads th is physically appealing. Sternberg and Eubanks@11# gives a clear account of the sort of restrictions required on the fin stress distributions used in the limiting process: These strictions have since been refined in Turteltaub and Sternb @13#. In essence, Sternberg and Eubanks establish that insufficient to simply have the distributed fields be statica equivalent to the end stress resultant sought~as Kelvin origi- nally proposed for his problem!. If one merely makes this requirement, then it is possible, for example, to add a dou state of the kind in Fig. 1b to a point load problem, thereb changing the dominant singularity of the latter without alte ing the force exerted. One means of avoiding this additio field for the point force example is to require all the distri uted stresses in the underlying limiting sequence be unidi tional; alternative restrictions for the point load, as well effective requirements for other singular loads, are given Sternberg and Eubanks@11# and Turteltaub and Sternber @13#. Provided proper attention is paid to the generating

2An error in one of the formulas given in Melan@9# is corrected in Kurshin@10#.

e

e

t

r a i a

r

e e n z

a t

r

s

s

t

r

s

he als to-

rack

en- ith

e

ach are - i- ot,

ht arp t is

s as at, st e- lar

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 253

quence of the distributed loads acting on successiv smaller regions, all of the singular loads included in Tabl have unique stress fields with singularities as indica therein.

In practice, concentrated loads usually serve as Gre functions in stress analysis. That is, they are superpose achieve a desired regular distribution of applied loads. Of this superposition is undertaken via numerical analysis demonstration of their use in this way occurs in integ equation approaches, such as the boundary integral equ method which currently enjoys fairly wide application elastic stress analysis. In this role, it is of value to underst the singular nature of the concentrated loads involved in der to design efficient quadrature schemes for their nume integration. However, these integrations typically result finite stresses. Then, one is not faced with the challeng drawing physical inferences, with respect to structural int rity, from nonphysical singular fields. On other occasio though, singular loads can be used to model highly locali loading, such as under a knife edge in the three-point-b specimen of fracture mechanics~eg, at pointP1 in Fig. 2a!. In this instance, if a line load is introduced, it is merely one of a set of three which effect an applied moment for crack. As such, it is not the feature of greatest interest, cally, with respect to potential failure—the crack tip is (P2 in Fig. 2a!.3 Again, one is not faced with interpreting loca fields at singular loads. On the other hand, one must atte this task for the crack, with its classical, inverse-square-ro stress singularity. Indeed, in general this is the case for second class of singular configurations recognized here. cordingly we focus on stress singularities which occur aw from any concentrated loading throughout the remainde this review.

1.3 Examples of other stress singularities

Some illustrative examples of this class of singularity a depicted schematically in Fig. 2. The corresponding order stress singularity present are set out in Table 2.

The first example~Fig. 2a! is the aforementioned cracke elastic plate under three-point bending, with its attenda inverse-square-root, stress singularity reflecting the stres tensification at the crack tip~ie, at P2). For the case of a crack in a large elastic plate under transverse tension, su singularity can be extracted from the corresponding solu for the elliptical hole on passing to the limit as the ho becomes a mathematically sharp slit. The fields required take this limit were first provided in Kolossoff@17# ~see also Kolossoff @18#!, and subsequently derived in Inglis@19#. That the same singularity results for crack tips in gene and for the crack tip in the three-point-bend specimen of F 2a in particular, can be discerned from Williams’ semin paper@20#. In this paper, the asymptotic character of elas stresses in angular plates or wedges under extension i vealed: Letting the angle of the ‘‘free-free’’ wedge go to 2p in Williams @20# recovers singular stresses as in Table 2.

eous r a

ely 1

ted

n’s d to en, . A al tion

n nd or- ical in of

g- s, ed

end

s he lo-

l mpt ot, the Ac- ay of

re of

d nt, in-

ch a ion le

to

al, ig. al tic

re-

As a modification to the first example, we consider t plate now to be comprised of two distinct elastic materi instead of a single one. The two are perfectly bonded gether on an interface extending straight ahead of the c ~indicated by the dashed line in Fig. 2a!. Adding the further discontinuity of an abrupt change in material properties r ders the crack-tip stress singularity more nonphysical, w the inverse square root having multipliers, cos(h ln r) and sin(h ln r), which oscillate an infinite number of times in th limit r→0 whenhÞ0. Hereinh is a material constant given by

h5 1

2p ln

m11k1m2

m21k2m1 (1.3)

where m is the shear modulus,k5324n or (32n)/(1 1n) for plane strain or plane stress,n being Poisson’s ratio, and the subscripts distinguish the different materials on e side of the interface crack. Observe that if the materials taken to be one and the same,h50 and there is no oscilla tory multiplier, as in our original example. Otherwise, typ cally interface cracks have oscillatory, inverse-square-ro stress singularities, as first shown in Williams@21#.

A related pair of examples concerns a tire, under lig load, where it meets a relatively stiff pavement at the sh edge of a pothole~a section through such an arrangemen sketched in Fig. 2b, whereinP3 is the point of interest!. If the pavement is icy, and thereby lubricated, the situation i if the tire were an elastic half-space being indented by a fl frictionless, rigid strip. The solution to this problem was fir given in Sadowsky@22#, and exemplifies the inverse-squar root stress singularity listed in Table 2. That the singu character here is the same as for the crack in a homogen material can be argued as follows. First we note that, fo

odels

Table 2. Some elastic stress singularities away from singular loads

Singular point, Fig. 2 „rÄ0…

Local configuration description

Singular stresses at point „r\0…

P2 Crack tip in three-point- bend specimen

ord(r 21/2)

P2 Interface crack tip in bend specimen

ord(r 21/2 cos(h ln r)) & ord(r 21/2 sin(h ln r)), see Eq.~1.3! for h

P3 Tire at pothole edge under icy conditions

ord(r 21/2)

P3 Adhering nylon tire at pothole edge

ord(r 21/2 cos(h ln r)) & ord(r 21/2 sin(h ln r)), see Eq.~1.4! for h

P4 Edge of piston ring pressed into cylinder wall

ord(r 20.23)

P5 Reentrant corner in stress-free keyway

ord(Tr21/3) ord(Fr 20.46) & ord(Fr 20.09)

P6 Edge of adhering rubber tire on pavement

ord(r 20.41)

P7 Circumference of an epoxy-steel interface

ord(r 21/3)

P8 Edge of a rough heavy block on an elastic slab

ord(ln r)

P9 Edge of a smooth steel chisel on a wooden block

ord(ln r)

P10 Submodel node with displacement shape functions as boundary conditions

ord(ln r)

3If instead the stresses under the knife edge were of greatest concern, better m than a line load are available, as we demonstrate subsequently.

d ion

254 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Fig. 2 Some singular configurations:a) three-point-bend test piece of fracture mechanics,b) section through a tire on a relatively rigi pavement,c) section through a piston with a ring pressed into a cylinder wall,d) section of a shaft with a stress-free keyway under tors and lateral loading,e) adhesive butt joint under tension,f ! rough heavy block sticking to an elastic base,g) steel chisel just starting to indent a wooden slab,h) displacement shape functions as submodel boundary conditions

d c

r o d

l

e

o

o

s

i

i l

o

h t r

ingu- rner are

are dary ex-

lly ne

tely in

e

of f - t . n

er- eets

ain ely r- il-

n’s e

r is r

of ty. r a

onti-

h a ear

to -

y uity hy two

-up if tant ear las-

tart- of

he

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 255

2D elastic half-space, the constant displacement due to rigid strip can be recovered by a rigid body translatio Hence, we need only consider homogeneous conditions der the strip, together with stress-free conditions outside strip. But these conditions under the strip are the same symmetry conditions. Thus the half-space can be reflecte itself to produce a full space with a pair of stress-free cra outside of where the strip punch acts.

For the second of our examples concerning the arran ment in Fig. 2b at P3 , we consider the pavement to be d and the tire to stick to it perfectly. Now the in-plane situati for the section of Fig. 2b is as if the tire were being indente by a flat, adhering, rigid strip. The solution to this proble was first furnished in Abramov@23#, and contains the inverse-square-root stress singularity, with its oscillato multipliers, listed in Table 2. This is the same singularity for the interface crack, except that nowh is given by Eq. ~1.3! with m2→` therein. That is,

h5 1

2p ln k (1.4)

Recall thatk5324n for the plane strain state applicab here: So as to avoidh50, Table 2 specifies a nylon tire (n 50.4) rather than rubber (n50.5) for this case of adhesiv contact. Asymptotically, the configuration can be treated ing the ‘‘clamped-free’’ conditions for a wedge of anglep in Williams @20#, if one sets ‘‘s’’ 5n in Eq. ~17! therein so as to correspond to a state of plane strain. The same singula results.

A further contact example is that of a lubricated pist ring pressed into a cylinder wall as indicated in Fig. 2c. This configuration is axisymmetric rather than being as previ examples which entail states of plane strain. However, first argued in Zak@24#, a plane strain analysis still applie Then, if the ring is taken to be relatively rigid compared the cylinder, the same inverse-square-root singularity res as for an indentation with a flat, frictionless, rigid strip~Fig. 2c at P4). Alternatively, if the more realistic assumption made that the ring is comprised of the same material as cylinder, the weaker singularity of Table 2 results. This s gularity can be identified by solving the pertinent eigenva equation in Dempsey and Sinclair@25#. It is weaker because now the deformation of the ring is being included.

For the example of a stress-free keyway in a shaft un torqueT and transverse loadF ~Fig. 2d!, multiple singulari- ties are present~Table 2!. For the torque, the singularity ac tive at the 90° reentrant corner~ie, atP5) is weaker than if a crack is subjected to torsion, having an exponent of 1/3 co pared to 1/2. This singularity was first identified in Thoms and Tait@26#, Section 710. For the transverse load, two s gularities typically participate. The stronger one is associa with loading which is symmetric about the bisector of t angle at the reentrant corner, the weaker with antisymme Both are weaker than the singularity at a crack, a reent corner of zero angle in effect. The two singularities for th right-angled reentrant corner are included in Brahtz@27#. Al- ternatively, they may be obtained using the ‘‘free-free’’ co ditions in Williams@20#, on taking a wedge angle of 3p/2. In

the n. un- the as on ks

ge- y n

m

ry as

e

us-

rity

n

us as . to ults

s the n- ue

der

-

m- n

in- ted e ric. ant is

n-

general, as the angle at a reentrant corner increases, s larity strength reduces. Eventually, when a stress-free co opens all the way up to a half-space, singular stresses removed.

The disappearance of stress singularities once corners no longer reentrant need not be the case when the boun conditions are mixed, as is demonstrated in our next ample. This concerns the tire again~Fig. 2b!, but now where it meets the pavement at its outside edge~ie, at P6). If the tire adheres perfectly to the relatively rigid pavement, loca this configuration becomes a right-angled corner in pla strain with one face being free of stress, the other comple fixed. The singularity in this instance is characterized Knein @28#. Alternatively, it may be obtained using th ‘‘clamped-free’’ conditions in Williams@20# for a wedge angle of onlyp/2, provided these are adapted to a state plane strain. For rubber (n50.5), the stress singularity o Table 2 results~there is a minor round-off error in the singu larity exponent in Knein@28#!. While this is weaker than tha of a crack, it is nonetheless quite comparable in strength

A similar situation occurs for the butt joint under tensio of Fig. 2e. Herein the points of interest are where the int face between the epoxy adhesive and steel adherend m the outside free surface~eg, P7). As for the piston ring, this configuration is axisymmetric but nonetheless plane str analysis still applies. Again then, since steel is relativ rigid compared to epoxy, a ‘‘clamped-free’’ right-angled co ner in plane strain is appropriate and can be treated via W liams @20#. Taking 3/8 as a reasonable estimate of Poisso ratio for epoxy, this gives the singularity of Table 2. Th reduction in strength here from that of the rubber corne due to the lower value ofn. Indeed, there is no singularity fo such corners whenn50.

Our last three examples give rise to the weakest type stress singularity in elasticity, the logarithmic singulari The first example concerns a heavy rough block, unde lateral force, sticking to a horizontal elastic surface~Fig. 2f!. If one assumes that the normal stresses produce a disc nuity in the surface shear~ie, at P8 , as indicated in the close-up!, then a log singularity in the stresses occurs, wit coefficient that is proportional to the magnitude of the sh stress discontinuity. This result is given in Kolossoff@18#. Alternatively, it can be constructed using auxiliary fields those in Williams@20#. These fields may be found in Demp sey and Sinclair@29#. While the normal stress discontinuit produces no stress singularity, any shear stress discontin on an elastic half-plane does. To see an indication of w this is so, consider the shear stress components on the little square elements outlined by broken lines in the close of Fig. 2f. The left one is in force and moment equilibrium it has no shears on its boundaries. The right one, cons shears. Where they meet, there is an incompatibility in sh stress which cannot be accommodated by any regular e ticity fields known to date.

The second example concerns a piece of wood, just s ing to be indented but not yet cut, by a sharp chisel made relatively rigid steel~Fig. 2g!. Assuming the contact to be frictionless and ignoring any anisotropy in the wood, t

l

e o

l

n

t

y

n n

s o e

u a

u

r

u

ake ap- uc- fi- ak

n, ir- cir-

ust aly- ch

nd e it ge,

ical the ible a-

ity. en t ex- lin- dis- are

he that las- p-

of em tion the ce elf. ld the h a l in si-

en ider

on-

all tip

the en

256 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

log-singularity stress field induced at the cutting edge~ie, at P9) may be found in Sneddon@30#, Section 48.4. Again, alternatively it can be constructed using the auxiliary fie in Dempsey and Sinclair@29#. This log singularity features a coefficient which depends on the chisel tip angle and is p portional to the elastic moduli of the wood. It also has displacement field which is more physically applicable initial knife-edge loading than that of a line load, being fr of unbounded vertical displacement and overlapping h zontal ones.

The third example concerns the use of displacement sh functions as boundary conditions in submodeling in fin element analysis~as suggested in ABAQUS@31#, and AN- SYS @32#!. Along a smooth submodel boundary, spurious singularities can be introduced. An example involving fo node elements is shown in Fig. 2h. Therein a log singularity occurs at the node atP10 whenever there is a discontinuity i the derivatives of either of the boundary displacementsu andv on y50. That is, whenever the constants are such c1

2Þc1 1 or c2

2Þc2 1 . Fields are given in Sinclair and Epp

@33#.

1.4 What to do about stress singularities

The foregoing serves to demonstrate some of the variet singular configurations and stress singularities possible classical elasticity. The natural question which then arise what is to be done about these and like configurations attempting to ensure structural reliability? In the first i stance, it is vital that the stress analyst at least recog when a stress singularity is present.4

That there is a singularity present is not always imme ately obvious. This is especially so in the stress analysi actual engineering components, since frequently the c plexity of such configurations necessitates numerical tr ment, often via finite element analysis~FEA!. Under these circumstances, one does not have available analytical s tions whereby singular character is detectable simply by servation. Nevertheless, it remains essential that the pres of any singular stress field be appreciated.

Consider the alternative. A scenario such as follows then quite possible. On Monday, you complete a first FEA a component subjected to cyclic loading. The maxim stresses found are a factor of two less than the endur stress of the component’s material. You conclude that component has indefinite life, or at least long life. O Wednesday, you check your FEA with a refined grid. T peak stresses are now comparable to the endurance. Yo in somewhat of a quandary as to how much life the p really has. Hence, on Friday you complete a further FEA a still more refined mesh. Now you get stresses that a factor of two greater than the endurance level. The com nent’s life now is, apparently, distinctly limited. Life for yo

ear n a

s-

ds

ro- a

to e ri-

ape ite

og ur

, hat s

of in

s is in - ize

di- of m- at-

olu- ob- ence

is of m nce the n

he are

art on e a po-

is somewhat disconcerting: Such a workweek does not m for a great weekend. More importantly, such a structural praisal has nothing to do with the component’s actual str tural reliability: In the presence of a singularity, any suf ciently refined numerical analysis predicts failure when pe stresses are compared against some finite stress criterio respective of what is physically happening. Under such cumstances, the participation of the singular stresses m first be recognized if any real use is to be made of the an sis. The main aim of this review is to aid in achieving su recognition.

That said, we next turn our attention to the important a challenging task of interpreting singular stress fields onc is apparent that they are active. In taking up this challen we begin by considering the simplifications made in class elasticity since we expect singularities to be a product of modeling in the theory, infinite stresses not being poss physically. Three such simplifying assumptions or lineariz tions can be identified in the classical theory of elastic The first linearization has that the relationship betwe stresses and strains is linear; that is, the stresses do no ceed the limits of elastic material response. The second earization has that the strains depend linearly on the placement gradients; that is, the displacement gradients small. The third linearization has that all loads act on t undeformed shape throughout the entire loading process; is, the deflections are small. The singular stress fields of c sical elasticity are in violation of all three of these assum tions. Yet they do comply with all of the field equations elasticity, as may be established by simply substituting th into these equations. This seemingly paradoxical situa results from the fact that, once an assumption is made in theory of elasticity and equations so simplified, complian with the assumption becomes unpoliced by the theory its This allows singular stress fields to comply with the fie equations of classical elasticity, but remain in defiance of underlying and unpoliced assumptions of elasticity. Suc situation requires some care if one is to be successfu interpreting these fundamentally wayward fields in a phy cally meaningful fashion.

To demonstrate the difficulty of interpreting results wh they lie outside of admissible responses in a theory, cons the following beam example taken from Frisch-Fay@34#. On page one of his monograph, Frisch-Fay considers a horiz tal cantilever beam of length 2.54 m~100 inches!, with a bending stiffness of 2.87 Nm2 (1000 lbf in2), subjected to a vertical concentrated end load of 4.45 N~1 lbf!. Treating this beam within the context of classical beam theory for sm deflections, Frisch-Fay obtains a prediction of a vertical deflection of 8.47 m~333 inches!, or more than three times the beam’s original length. This result suggests strains of order of 300% and the possibility of gross yielding and ev ductile rupture. Subsequently, on page 39 of Frisch-Fay@34#, the same beam is analyzed within the context of nonlin beam theory for large deflections. This analysis results i vertical deflection of 2.06 m~81 inches!, together with a horizontal deflection of 1.42 m~56 inches!, and stress and strain fields that can now comply with the underlying a

icro- fully

4We have not included, in the examples of Table 2, the yet stronger, ord(r 21), singu- larity occurring at dislocations of the Volterra type~see, eg, Love@12#, Appendix to Chapters VIII, IX, or Timoshenko and Goodier@16#, Articles 34, 117!. These fields are used as Green’s functions, and by some theoreticians in an attempt to model m structure. We omit discussion of them primarily because we expect users to be cognizant of the singular character present.

a

n a

i

t h b

e n i

e

v g

l

e

o

u

i

en

lso tip

was

as ress

are han we the ut

mall all rns

we g re- like ble we

ress

al em of l to the ter- is te In- nse ethe- ion x-

w ls for

on- the

ion

si- n- the ow re- , it em-

rely itu-

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 257

sumptions of the theory. The beam’s length is essenti unchanged, in contrast to the earlier and potentially qu misleading prediction. It follows that, in this instance, all o can reasonably directly conclude from the prediction of cl sical, small-deflection, beam theory is that the deflection large, too large to be quantitatively predicted by the theo In essence, the same situation holds with respect to the gular stress fields of classical elasticity. Typically, they a correct qualitatively in implying large stresses, yet quant tively they cannot be relied upon for the magnitude of the stresses. Other less direct interpretations must be mad order to quantify the implications of stress singularities.

The preceding example also illustrates a possible stra for dealing with stress singularities: namely, improving t modeling in the underlying theory so that physically sensi outcomes are predicted. This is what nonlinear beam the did in the example, albeit at the expense of turning a lin theory into a less tractable nonlinear one. Arguably, eve the expense of requiring greater analytical effort, such provements in the physical modeling represent the ultim of ‘‘interpretations’’ of singular stress fields. Accordingly, w consider various means that might effect such improvem next, in Section 2. Currently, not all configurations are am nable to complete amelioration of their singular stress fie via the various means identified. Hence, in Section 3 review interpretations that can be made when singular havior persists. Then we return to our primary intent of he ing a stress analyst appreciate when a stress singularity occur, and what its singular character can be. We begin activity in Section 4 with a description of some method both analytical and numerical, for determining the nature a participation of stress singularities. We then close Part I this review with some concluding remarks. Part II will fo low with a review of contributions in the literature that ha actually carried out characterizations of possible local sin lar stresses for a variety of elastic configurations. Through both parts, there are portions of the text that are tutoria nature. Because a significant amount of today’s stress an sis is carried out in industry and, in the main, by engine with bachelor’s degrees, a serious effort has been mad write these tutorial portions so that they can be underst by such stress analysts.

2 RIDDING CONFIGURATIONS OF NONPHYSICAL STRESS SINGULARITIES

2.1 Possible avenues for removing singularities

In some instances, removing singularities is straightforwa For example, the logarithmic singularities induced by the of displacement shape functions as boundary conditions submodels in finite element analysis~eg, P10 of Table 2 and Fig. 2h!. These can be removed simply by fitting nodal d placements in the global FEA preceding the submodel w curves that are once continuously differentiable, then us intervening values in submodel boundary conditions~eg, by fitting a cubic spline as in Kondo and Sinclair@35#!. In es- sence, all that is required here is an appreciation of the troduction of singularities by a poor choice of boundary co

lly ite e s- is

ry. sin- re ta- se e in

egy e le ory ar at

m- ate e nts e- lds we be- lp- can

this s, nd of

l- e u-

out in

aly- rs

e to od

rd. se on

s- ith ing

in- n-

ditions which have extraneous discontinuities. Th removing such discontinuities removes the singularities.

In other instances, smoothing discontinuities might a appear to remove singularities. For example, rounding the of a crack~as atP2 of Table 2 and Fig. 2a!, or rounding the corner in a keyway~as atP5 of Table 2 and Fig. 2d!, does produce finite stresses. This tactic for the latter example suggested in Thomson and Tait@26# circa 1867, so such an approach is definitely not new. However, in instances such these two wherein the stress singularities reflect real st concentrations, such smoothing is questionable.

How so? For the example of the crack, certainly there no singular stresses with any root radius that is greater t zero. However, we know that for a root radius that is zero get the physical absurdity of infinite stresses. This raises question of just how physically relevant are the finite b extremely large stresses that can result from extremely s root radii. Moreover, crack tips can have extremely sm root radii, so the question is not moot. And similar conce apply to rounding of the keyway corner.

The real removal of stress singularities requires that can be confident that the unbounded stresses are bein placed by physically sensible ones. For the crack and configurations, this really means we want finite sensi stresses when root radii actually go to zero. Only then can be reasonably confident of the physical relevance of st fields for root radii near but not zero.

At first thought, opportunities for achieving the remov of singularities when root radii are zero would appear to st from relaxing the constraints implicit in the linearizations classical elasticity listed earlier. Perhaps the most natura consider in this regard is relaxing the assumption that stresses remain below their elastic limits and, thereby, en taining the possibility of plastic flow. Such a relaxation quite often implied in the literature to be the appropria recourse to take when singularities occur in elasticity. deed, if one insists upon perfectly plastic material respo after elastic, unbounded stresses can be removed. Non less, introducing plasticity does not really effect a resolut of the difficulties with elastic stress singularities, as we e plain next.

To begin, introducing plasticity begs the question of ho to remove singularities for configurations involving materia that are not ductile. Leaving this omission aside, even ductile materials it is not really appropriate. To see this, c sider what happens physically as loading progresses. At outset, loads are small. In fact, for any actual configurat comprised of a material withsY.0, sY being the yield stress, there exists a sufficiently light loading so that, phy cally, no yielding whatsoever is produced. Yet, if the co figuration of interest has an elastic stress singularity, theory predicts yielding for any nonzero load, no matter h small. Given the physical inappropriateness of the initial sponse of plastic fields derived from singular elastic ones is not reasonable to assume that these fields correct th selves as plastic flow increases. Accordingly, one cannot on these fields to accurately capture the physics of the s ation.

o. lure , a uch hat n a

se, qs.

n

hav- m-

w nse

ion the

l-

258 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Fig. 3 Tensile crack in a hardening material

u l o

c

a a

i r

e n

ght lly, nt it els. ear

ll. re- ay

rge ri- ove er of rst

wles

rsis- he

be

the ity

e-

re

or

is

tes er- fter lar, ly wn

- f ile e

on nd ve

u-

t is ial tress

ct -

plasticity irrespective of load level, provided it is not zer Thus, if this value or a lesser stress is chosen as a fai criterion, failure is always predicted: If, on the other hand higher stress value is taken, failure is never predicted. S stress-based predictions have no reliable correlation to w is physically happening. Furthermore, one cannot rely upo strain criterion for this special type of material respon since the strains remain singular—see the second of E ~2.2!, from Hutchinson@37# ~the same result may be found i Cherepanov@36# and Rice and Rosengren@38#!.

Other geometries share the persistence of singular be ior when treated via deformation theory, though not all co ply with the second of~2.2!: see Chao and Yang@40#, Rudge and Tiernan@41#, Zhang and Joseph@42#, and references therein. It follows that introducing yielding and plastic flo when a singularity is already present in the elastic respo does not remove the singularity in any real sense.

Alternatively, one could consider relaxing the assumpt of small displacement gradients and, thereby, entertaining possibility of large strains. Again, though, initially it is a ways possible to have actual load levels which are li enough so that only small strains are induced physica rather than large. This raises questions as to how importa is to include a large strain representation at such load lev Observe, though, that in contrast to plasticity, the nonlin contributions attending large strain representationsare presentat low load levels, even if they are relatively sma Consequently, absent analysis, it is not clear how much laxing the small displacement gradient assumption m remove/alleviate singular stresses.

Turning to analysis then, the general finding is that la strain treatments do typically improve the physical approp ateness of singular fields, and even on occasion rem them, but nonetheless result in the persistence of a numb singularities. For the crack, results of this nature were fi indicated in Wong and Shield@43#, then established for more general circumstances in the two successive papers, Kno and Sternberg@44,45#. Geubelle and Knauss@46# provides a recent large strain treatment of cracks demonstrating pe tence of singular behavior, together with a review of t area.5 There and elsewhere, ther 21 behavior asr→0 of the crack-tip stress-strain product is found to continue to present~cf, the second of Eqs.~2.2!!. On the other hand, a large strain treatment of the interface crack can remove nonphysical oscillatory multiplier of the stress singular noted in the Introduction forP2 of Table 2 and Fig. 2a ~see Geubelle and Knauss@47# and references therein!. It can also remove the oscillatory nature of the singularity for the adh sive flat punch noted forP3 of Table 2 and Fig. 2b ~Knowles and Sternberg@48#!. Furthermore, it does remove the enti singularity for the butt joint noted forP7 of Table 2 and Fig. 2e ~Ru @49#!. However, it does not remove singularities f other bimaterial wedges~Ru @49#!, nor for reentrant corners ~Duva @50#!. In sum, while introducing large strain analys

is to not

There is a further impediment to the use of such estima of elasto-plastic response in structural integrity consid ations. If the material being considered hardens at all a yielding, the stresses can be expected to remain sing though with the strength of their singular behavior typica being abated. That this is so for the case of a crack is sh in Cherepanov@36#, Hutchinson@37#, and Rice and Rosen gren @38#, within the context of total deformation theory o plasticity. By way of specific example, we consider a tens crack tip ~Fig. 3! in a material which hardens in accordan with the model put forward in Ramberg and Osgood@39#. A law for uniaxial tensile stresss t versus tensile strain« t for such a model is

« t

«Y 5

s t

sY 1

1

500S s t

sY D n«

(2.1)

whereinsY continues as the yield stress and«Y5sY /E is the corresponding strain, withE being Young’s modulus and n« the strain hardening exponent. Then, from Hutchins @37# using the coordinates of Fig. 3, the normal stress strain ahead of the crack within deformation theory beh in accordance with

sy5O~x21/(n«11)!, sy«y5O~x21!, as x→01

(2.2)

on y50. For n«51, the classical inverse-square-root sing larity of elasticity is recovered. For 1,n«,`, the stress singularity is weaker but nonetheless persists. Hence futile to compare such stresses directly with finite mate values such as the ultimate stress or the endurance s ~recall the previous discussion in the Introduction!.

For the special and physically atypical case of perf plasticity post yield (n«→`), the crack-tip stresses are co strained to be finite but still cannot be compared in a me ingful way with material values. This is because the stres always locally attain the limiting value set by the perfe

an- ses ct

5Some of these references term themselves ‘‘finite strain’’ treatments. This term underscore the contrast with the infinitesimal strains of classical elasticity: It does imply bounded strains at the crack tip.

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 259

Fig. 4 Genesic Griffith crack configuration

t

m

n

t

a

s

n h

a p

al act e- e

lar. lar

e ot

ysis ef-

t the

ng en- ions on- sses

lly

ral, ns, re- the rst v f

ing re.

er- D

tal

f y

o a

- on. eir p-

as on-

ap- e

n jor

to

nt re-

ess g at

ll is-

e r-

penetration or overlapping of material outside of the origin contact region. If, instead, the two are allowed to cont further without interpenetration, the power singularity is r duced to a log singularity which is similar to that for th chisel indentation configuration (P9 of Table 2 and Fig. 2g!. While this is a weaker singularity, nevertheless it is singu Thus, this last relaxation also fails to really remove singu character once it exists in a classical elasticity solution.

All told, none of the foregoing relaxations fully remov stress singularities when they exist in classical elasticity. N to say that elasto-plastic/large strain/large deflection anal may not be appropriate on occasion once the removal is fected, but that by themselves such analyses do not effec removal. Needed is a different approach.

What other options are there for improving the modeli so that stress singularities are replaced with physically s sible stresses? The answer lies in the boundary condit enforced, both as direct requirements and as auxiliary c straints. We consider some problems where singular stre are alleviated via this approach next.

2.2 Canceling crack-tip singularities: Barenblatt’s ap- proach

We begin our consideration of the effects of more physica appropriate boundary conditions withcrackedconfigurations because of their central role in solid mechanics in gene and fracture mechanics in particular. For such configuratio it is possible to negate singularities produced by loading mote from the crack with those due to tractions acting on crack flanks. Barenblatt credits Khristianovitch as being fi to notice this in his paper with Zheltov in 1955. In Zhelto and Khristianovitch@54#, a large rock stratum comprised o an oil bearing shale is considered with a view to determin when a pressurized flaw within the stratum might fractu The stratum is under all-round pressurep0 while the faces of the flaw near its tips are subjected to a relatively high int nal pressure ofpi ~Fig. 5!. The configuration is treated as 2 and elastic. Then, if the extent of the regions over whichpi

acts,Da, is taken to be an appropriate fraction of the to

Fig. 5 Pressurized crack configuration

improves the physical appropriateness of singular fields degree, this relaxation fails to fully remove them.

The remaining option for relaxation within the simplifica tions of classical elasticity is the small deflection assumpti That is, removing the assumption that the loads act in th entirety on the undeformed state. Relaxation of this assu tion can be performed by applying loads incrementally deformation proceeds: In some sense, one may interpret linear beam theory as an implementation of such an proach. Griffith was first to do this for a crack in an infini plate under all-round tensions0 ~Griffith @51#!. He formed his crack of length 2a as the limit as the semiminor axis,b, of an elliptical hole goes to zero~Fig. 4!. For classical elas- ticity, the maximum stress for the elliptical configuratio used in the limiting process occurs at the ends of the m axis. This peak valuesmax, is given by~from Inglis @19#!

smax5KTs0 , KT52a/b (2.3)

In ~2.3!, KT is the stress concentration factor. On passing the limit of a crack (b→0), KT blows up reflecting the stres singularity so generated. In Griffith’s incremental treatme wherein loading is gradually applied, the corresponding sult is

KT5 E

s0 lnS cosh

2s0

E 1

a

b sinh

2s0

E D (2.4)

whereE remains Young’s modulus and a state of plane str is assumed.6 An analogous treatment for the ellipse tendi to a crack under uniaxial tension yields similar results. T is, stresses are ord(lnb) as b→0 instead of ord(b21) as in Eq. ~2.3!. Nevertheless, they are still singular.

A further example of the effects of relaxing the sm deflection assumption may be obtained on revisiting the ton ring configuration in the Introduction (P4 of Table 2 and Fig. 2c!. Once the ring is allowed to deform along with th cylinder wall, the power singularity of Table 2 leads to inte

6Mansfield@52# derives the same result as Eq.~2.4!. The actual theory used in all o these incremental elasticity analyses is an approximate rate-of-deformation theor~see Truesdell@53#!.

to be

e

s line

e

260 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Fig. 6 Barenblatt’s crack tip

. c

d

e

u

e

h

sity

tip

n-

d of

ing s a his e of d a

s- s- tri-

a-

ith nd

or-

ir by

im- at nks. be

gu- lar

in the is

ne al of

sentative of the applied far-field tensile traction parallel they-axis: These tractions need not be uniform but are to symmetric abouty50. Next, define the coefficient of th singularity due tos0 , the stress intensity factorK I , in ac- cordance with usual practice. To wit,

K I5 lim x→02

A2pxsy u y50

(2.6)

whereinsy is the normal stress component in they-direction induced bys0 alone. Now apply a pair of tensile line load to the crack flanks which are equal and opposite. These loads tend to close the crack. If their strength isF per unit thickness and they act on the crack flanks at a distancx from the crack tip, the associated stress intensity factor,K I8 , is negative and given by~Tada, Paris, and Irwin@55#, p 3.6!

K I852FA 2

px (2.7)

Introducing cohesive stressessc5sc(x) by replacingF by scdx, then integrating, gives the negative stress inten induced by the closing tractions: Equating the result toK I , the factor due to the far-field loadings0 , then renders the configuration singularity free. That is, there is no crack- singular stress field if

K I5A2

p E 0

dc scdx

Ax (2.8)

wheredc is the extent of the cohesive zone. Under this co dition, the crack opening profile forms a cusp~see close-up of Fig. 6!, with a crack opening displacement,n on y50 (x.0), of the form

n u y50 x.0

5 11k

6m sc u

x50

F S x3

L D 1/2

1O~x5/2!G as x→01

(2.9)

whereinm andk are as previously~Eq. ~1.3! et seq!, andL is a normalizing length. The companion tensile stress ahea the crack tip is given by

sy u y50 x,0

5sc u x50

F12S uxu L D 1/2

1O~x3/2!G as x→02

(2.10)

Clearly the crack-tip stress of Eq.~2.10! is free of singulari- ties.

In addition to introducing cohesive stresses and assum the region over which they act is small, Barenblatt make further ad hoc assumption regarding their distribution. T second assumption has that the maximum possible valu the right-hand side of Eq.~2.8! at failure does not depen upon the applied loadings0 , and is always the same for given material. He terms the right-hand side of Eq.~2.8! at failure a material’s ‘‘modulus of cohesion’’ to reflect his a sumption that it is a material property. This simplifying a sumption obviates the need to determine explicitly the dis

,

flaw width, 2a, the compressive stress singularity due top0

is cancelled by the tensile stress singularity due topi . More precisely, if

Da

2a 5sin2S pp0

4pi D ~pi.p0! (2.5)

then there is no singularity for the configuration of Fig. 57

Subsequently, Barenblatt appreciated the fuller impli tions of Zheltov and Khristianovitch@54# ~Barenblatt@56#!: An extensive account of his resulting research, together w a comprehensive bibliography of related work, may be fou in Barenblatt@57#. To extend the applicability of Zheltov an Khristianovitch’s model, Barenblatt introduces cohesive n mal stresses to replace the applied pressurepi . Essentially, he argues as follows:

i! that the heights of cracks are small relative to th lengths so that they can be approximated mathematically-sharp slits.

ii ! that under such circumstances, the immediate prox ity of the crack flanks at the crack tip ensures th intermolecular cohesive stresses act between the fla

iii ! that the distribution of such cohesive stresses can adjusted so that the corresponding compressive sin lar stress field completely negates any tensile sing stress field due to far-field loading.

Barenblatt assumes that the extent of the near-tip zon which cohesive stresses are applied is small relative to overall crack length. Indeed, in a first implementation of ideas for a specific crack configuration in Barenblatt@57#, he considers a semi-infinite crack with a finite cohesive zo ~Fig. 6!: Hence, in effect, his cohesive zone is infinitesim compared to the crack length. Even so, the cancellation singularities can be effected, as shown next.

First, take rectangular Cartesian coordinatesx, y, with origin O at the crack tip, as in Fig. 6. Then lets0 be repre-

7The result in Eq.~2.5! follows directly from the singularity coefficients given in Tada Paris, and Irwin@55# on pp 5.1, 5.13.

a

b t

c s h

i

o e

.

d l e

e

e . a n r

s

a

rgu- im- also d- cs, ny dis- he-

with dis- en

sed w one d is ep,

n re- pa- . In a-

a- rst can s or lk tiff-

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 261

bution of cohesive stresses within the cohesive zone, no easy task at the time of Barenblatt@56,57#. However, it has two serious drawbacks.

First, an immediate consequence of the assumption is the stress intensity factor at fracture due to any applied lo ing such ass0 is also a material property, because it equ in magnitude that due to cohesive stresses. Essentially, th fore, the stress intensity factor due to applied loading comes the key parameter controlling fracture. This is same fracture criterionas used in models wheresingularities are present~see subsequent Sections 3.1 and 3.3!.8 Hence, while Barenblatt does indeed cancel singularities for cra by introducing the concept of cohesive crack-flank stres the manner in which he does so leads to an approach w is equivalent to that practiced when singularities are active far as fracture goes. Accordingly, Barenblatt’s approach c not realize any practical improvement in fracture predict as a result of negating crack-tip stress singularities.

Second, the assumption is not realistic. To explain, c sider its analogue in elasticity in general. Taking the str intensity factor at failure resulting from cohesive stresses a material property would be akin to taking local stress sultants in elasticity as material properties. This is not so elasticity, it is the elastic moduli that are the material pro erties. While local stress resultants can depend on the va of such material properties, they can also depend on loa and geometry and so are not material properties themse With cohesive stresses, then, it is the cohesive str separation laws that are material properties, not the st intensity factors that can attend these laws.

It is possible to extend Barenblatt’s approach and can singularities with cohesive stresses in other otherwise sin lar configurations, albeit with the same drawbacks. For ample, the keyway configuration in the Introduction (P5 of Table 2 and Fig. 2d!. It is not clear, though, how it could b extended to all of the other examples in the Introduction

In sum to date then, modifying the field equations of el ticity would not seem to offer any real means of removi stress singularities. On the other hand, what we learn f Barenblatt@57# is that incorporating cohesive stresses in boundary conditions can remove singular behavior. Cohe stresses have also been used in this way to render mode dislocations free of singularities: Such models have been forward in Peierls@60# and other papers~see Hirth and Lothe @61#, Chapter 8!. It would therefore appear that cohesiv stresses might play a major role in the alleviation of sing larities. Moreover, cohesive stresses are fundamental to s mechanics, being the underlying source of constitutive re tions. In contrast, it is not obvious that there is any fund mental justification for making assumptions regarding th distributions. Consequently, we next look to consider an proach for including cohesive stress action without such sumptions.

eter-

the

h l

t an

that ad- ls ere- e-

he

ks es, ich as

an- on

n- ss as

re- In

p- lues ing ves. ss-

ress

cel gu- x-

s- g

om to ive ls of put

e u- olid la- a- eir p-

as-

2.3 Removing singularities via boundary conditions: Introducing cohesive stresses

In the approach adopted here, we endorse Barenblatt’s a ment that when surfaces come into extremely close prox ity with one another, cohesive stresses have to act. We follow Barenblatt in taking this interatomic action to be mo eled with boundary conditions in continuum mechani thereby facilitating analysis. We do not, though, accept a of Barenblatt’s assumptions concerning cohesive stress tributions. Rather, we introduce cohesive stresses via co sive stress-separation laws and let these laws interact the configuration of interest to determine cohesive stress tributions. Initially, we treat cracks with the approach, th we treat other singular configurations.

To begin, the nature of the cohesive stresses to be u merits further discussion. Acohesive stress-separation la for the normal stress at a single point on the surface of elastic half-space as it is being removed from a secon sketched in Fig. 7. The initial response there exhibits a ste nearly linear, increase in cohesive stresssc with separations above the equilibrium valuese . Thus, ass first exceedsse ,

sc5ke~s2se! (2.11)

where ke is the separation stiffness near equilibrium~the dashed line slope in Fig. 7!. After reaching an ultimate value sU , sc gradually decays to zero ass becomes large. The overall character of the cohesive stress versus separatio sponse is similar to that for the attractive force versus se ration response for an isolated pair of atoms or molecules fact, physically it is the result of an integration or combin tion of such force-separation responses.

Carrying out such integrations via first-principle calcul tions is a challenging analytical task. However, for the fi part of the curve—the linear stress-separation law—we simply estimate the response of the accumulation of atom molecules directly. To do this, we obtain and fit the bu response in experiments so as to back out cohesive law s nesses. For example, to use a uniaxial tension test to d mine ke of Eq. ~2.11!, reconsider Fig. 2e with the epoxy replaced by linear springs that are supposed to replicate

is arities

Fig. 7 Schematic of cohesive stress-separation law

8Willis @58# and Goodier@59# provide alternative arguments that, as a result of assumptions, Barenblatt’s approach reduces to the same as for cracks with singu present.

d

h

a e c l s

n s e

e

o

m

t

e

p- , it m- th-

ws. he tric

3, ese

ack rting :

l the

uce

s the

on-

rp

s

262 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

initial cohesive law for steel. Then matching the respon from the springs with that for a solid steel bar gives

ke5E/se (2.12)

For Fig. 2e, E would be Young’s modulus for steel: In gen eral,E is Young’s modulus for whatever material is involve Hence for uniaxial tension, our initial cohesive law is E ~2.11! with ke as in Eq.~2.12!. A consequence of this mean of estimatingke is that the initial cohesive law passes wh might be termed a ‘‘patch test’’ and is consistent with t surrounding continuum.

The foregoing is a possible, even if somewhat crude, p cedure for estimatingke in the elastic regime since materi defects in bulk specimens do not have a marked influenc response in this regime. At higher stresses, though, we not employ such an inverse approach because materia fects do produce significant effects. For present purpo however, the remainder of the curve is not critical since are primarily concerned with elastic response. According we adopt the highly idealized assumption of a perfe defect-free, brittle material. Then there do exist estima from solid-state physics of a material’s ultimate stress~see, eg, Cherepanov@62#, p 36, which givessU'E/10). We can also set the area under the curve—the work of adhesion twice the surface energy, another material property for wh estimates can be obtained~ibid!. Regarding the decay rate ass→`, we can just directly integrate that associated w pair-wise atomic or molecular forces, ignoring other intera tions. For example, the potential of Lennard-Jones@63# for van der Waals’ forces at large separations has them deca as 1/s7 as s→`: Direct integration then givessc decaying as 1/s3 as s→` ~see, eg, Israelachvili@64#, Section 10.2!. Such a derivation does not properly account for interact and shielding effects, but suffices here.

The choice ofke so that it is consistent with theke im- plicit in the elastic constitutive relations of the surroundi continuum offers some attributes in elastic stress analysi demonstration thereof follows on reconsidering the probl of a circular hole in an elastic plate under all-round far-fie tension~Fig. 4 with a5b therein!. The classical solution to this problem is given in Lame´ @65#, Article 80. It features a KT52 for the hoop stress at the hole’s edge~see Eqs.~2.3! with a5b). However, in this solution, if one sits at the edg of the hole then takes the limit as the hole disappears, obtains the physically anomalous result of the persistenc this stress concentration even when the plate becomes w without a hole. What is needed to remove this anomal result is the recognition thatcohesive tractions must acton the hole surface as it closes. When the hole is very small, associated cohesive stress-separation law takes the for Eq. ~2.11!. Then, provided the stiffness therein is taken so to be consistent with the elastic constitutive relations of surrounding continuum, a state of uniform biaxial tension recovered throughout the plate when the hole disapp ~Sinclair and Meda@66#!.9

ca-

ith c

se

- .

q. s at e

ro- l on an- de- es,

we ly, ct, tes

to ich s ith c-

ying

ion

g . A m

ld

e one

of hole us

the of

as he is ars

While the foregoing describes a greatly simplified a proach for determining cohesive stress-separation laws suffices for the discussion that follows here. We next co pare various treatments of the symmetrically loaded ma ematically sharp crack with and without such cohesive la

The traditional conditions on the crack plane for t stress-free mathematically sharp crack under symme ~Mode I! loading are:

sy5txy50, for x,0

v50, txy50, for x.0 (2.13)

where thex and y rectangular coordinates are as in Fig. sy andtxy are normal and shear stress components in th coordinates, andv is the displacement in they-direction. In contrast, recognizing that for the mathematically sharp cr cohesive stresses must act as Barenblatt did, then inse them via Eq.~2.11!, the conditions on the crack plane are

sy5ke~n12n2!, txy50, for x,0

n50, txy50, for x.0 (2.14)

wheren1 is the displacement of the upper crack flank,v2

that of the lower~ie, n65n at y56se/2). Settingke50 in Eqs. ~2.14! give Eqs.~2.13!. In effect, therefore, traditiona conditions overlook the cohesive interaction between flanks that physically must occur.

What now becomes apparent, once we start to introd interatomic considerations, is that whenx,0, the boundary conditions hold aty5se/2 for the upper crack flank andy 52se/2 for the lower. That is, through the centers of atom comprising the bottom surface of the upper half-space, top of the lower. To be consistent then, we should view c ditions ahead of the crack tip as applying at the same lo tions for their respective half-spaces. As a result, Eqs.~2.13! and~2.14! must have a cohesive law ahead of the crack w

le. If ohe-

Fig. 8 Sketches of atomic or molecular ‘‘springs’’ at a sha crack-tip for various boundary conditions:a) classical stress-free conditions,b) Barenblatt’s cohesive stress conditions,c) consistent cohesive stress conditions,d) alternate cohesive stress condition

9This assumes there is no activation energy or other impediment to closing the ho there were, the closing cohesive law would have to be modified. Nonetheless, a sive law would still have to act as the hole closes.

e

t

h

l

a

a i

g t d

t

n r

r

a

ct at

are the er- not tion as with um ow ack

ati- th s. he

rst are law is,

re- e- de-

ree

tress

ive

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 263

an infinite stiffness forn15n250 there, as required by th second of Eqs.~2.13! and ~2.14!. This situation is indicated schematically in Fig. 8a andb, wherein circles represent a oms and springs with stiffnessk represent cohesive laws.

A more consistent introduction of cohesive laws for t mathematically sharp crack under symmetric loading tak on the crack plane,

sy5ke~n12n2!, txy50, for all x (2.15)

Then the same cohesive law acts throughout, consistent the same material comprising the half-spaces both in fr and in back of the crack tip~Fig. 8c!.

An alternative crack-tip configuration sometimes imp mented in the literature just inserts the cohesive law ahea the tip while maintaining stress-free crack flanks. The con tions on the crack plane for this type of crack tip are:

sy5txy50, for x,0

sy5ke~n12n2!, txy50, for x.0 (2.16)

Now, in effect, the cohesive law in back of the crack tip h zero stiffness~Fig. 8d!. For this choice to be physically jus tifiable, appropriate arguments from solid state physics n to be made. Presumably such arguments reflect a histor the crack flanks which, at one time, had them at significan greater separations than for the mathematically sharp cr

For Eqs.~2.13! and ~2.14!, with their effectively infinite stiffnesses, singularities result. This is shown asymptotic in Williams @20# for the traditional conditions, and in Sincla @67# for Barenblatt’s conditions.10 Indeed, for Eqs.~2.14! a singularity is necessary if cancellation of singularities is be effected as in Barenblatt@57#. For Eqs.~2.15! and~2.16!, with their absence of infinite stiffnesses, no singularities sult ~Sinclair @67#!. For Eqs.~2.15!, no singularity is clearly the result to be expected, there being no discontinuity either boundary directions or conditions.

Thus, the presence of effectively an infinite stiffness in cohesive law is what is the underlying source of the sin larity for the mathematically sharp crack under symme loading. The situation is akin to contact/impact in rigid bo dynamics. There, rigid bodies with their infinite stiffness lead to infinite contact forces. Once deformation is admit and finite stiffnesses introduced, finite contact forces res Likewise with only finite stiffnesses in cohesive laws, fini rather than singular stresses result for the crack.

At this time, the use of cohesive/adhesive laws in bou ary conditions in solid mechanics has seen quite widesp use. Sinclair@68# provides a recent bibliography: Most of th references therein cancel singularities after Barenblatt@57#, but some introduce cohesive/ adhesive laws ahead of c tips instead. An early example of the latter type of impleme tation is Cribb and Tomkins@69#. A fairly recent review of a number of contributions of this ilk is furnished in Needlem @70#. An implementation of Eqs.~2.15! when ke is backed out from constitutive relations is summarized in Sincla Meda and Smallwood@71#.

ack.r

-

e es,

with ont

e- d of di-

as - eed y of tly ck.

lly r

to

re-

in

a u-

ric y

es ed ult. te

d- ead e

ack n-

n

ir,

In Sinclair et al @71#, cohesive laws are taken to a throughout the length of the mathematically sharp crack the outset before any external loading is applied~Fig. 9a!. That is, no assumption is made that cohesive stresses confined to a small region near the crack tip, but rather cohesive law itself interacts with the configuration to det mine cohesive stress distributions. Not surprisingly, this only leads to finite stresses but also to a stress concentra factor of unity for the mathematically sharp crack. This h to be the case when the cohesive law is made consistent the surrounding continuum because then the continu never knows the mathematically sharp crack is present. H is it, then, that the real stress concentration occurring at cr tips can be reflected by this type of modeling?

The answer lies in treating cracks that are not mathem cally sharp. One way of doing this is to proceed as in Griffi @51# and form cracks via elliptical holes in elastic plate Then two types of configuration can be distinguished. T first has stress-free crack surfaces~Fig. 9b!. This occurs when the root radiusr 0 of the elliptical hole is sufficiently large. Here, by sufficiently large is meant such that the fi pair of atoms or molecules on opposite flanks at the tip separated by a sufficient distance so that the cohesive itself sets the surface tractions for this pair to zero. That this distance is a sufficient number of multiplesm of the equilibrium separationse so that the law of Fig. 7 hassc

50 effectively ~see close-up in Fig. 9b!. The second has cohesive stresses acting near its tip~Fig. 9c!. This occurs when the root radius decreases from the minimum value quired for stress-free flanks. Ultimately this configuration b comes the mathematically sharp crack as the root radius creases still further.

Resulting stress concentration factors for stress-f cracks coincide with classicalKT ~ie, as on the right-hand side of Eqs.~2.3! plus one for transverse tension alone!. Thereafter, as root radii are decreased so that cohesive s starts to act,KT fall below classical values. Ultimately for a small but nonzero root radius, the crack closes and aKT of unity results, the same as for the mathematically sharp crein.

Fig. 9 Crack flank configurations when introducing cohes stresses:a) mathematically sharp crack,b) stress-free crack,c) intervening crack

10Here, by Barenblatt’s conditions we mean Eqs.~2.14!: While Barenblatt@57# does not explicitly give these conditions, they are nonetheless implicit in the approach the

f

t

o

i o

i

f

s

e

d

c

s n

e ’

b

o t

e

h

i

an cal m-

in are , a-

und- n a ly, ion

- ing oad ch on-

of -

the ain

e

i-

us

ies

in the es, act we

n, but

264 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Other erstwhile singular configurations can be rende singularity-free by similar means. Viewed from a cohesiv adhesive stress perspective, it is possible to identify ef tively infinite stiffnesses in almost all of the examples stress singularities given in the Introduction. Hence their s gular nature. When cohesive/adhesive laws without infin stiffnesses are introduced, these examples are rid of s singularities.

To explain further, traditional contact and clamped con tions are really simplifications of cohesive or adhesive c ditions. When exchanged for the latter, only two princip types of boundary conditions remain in planar elastic cohesive/adhesive conditions and stress-free conditi When these last two types of boundary conditions act on in-plane geometry which entails a vertex anglef of p or less, no power singularities are possible~Sinclair @67#!. When effectively infinite stiffnesses in cohesive/adhes laws are removed on lines of symmetry or antisymmetry, examples in the Introduction all havef<p, hence no power singularities. Moreover, there are no log singularities these two types of boundary conditions provided there are step discontinuities in shear tractions whenf5p. This last requirement, in particular, means that one cannot implem the shear counterpart of Fig. 8d for antisymmetric~Mode II! loading of a crack if one is to avoid log singularities. It al means any shear tractions in contact problems with sh edges must go to zero continuously, even if very rapid outside of contact regions if one is to avoid log singulariti Further explanation is given in Sinclair@72#.

While the introduction of cohesive/adhesive laws with nite stiffnesses and no shear jumps shows promise of rid elasticity of most if not all stress singularities, the impleme tation of this approach in toto faces some stiff challeng These principally stem from the determination of the app priate cohesive/adhesive law. For example, consider the of brittle fracture, arguably the simplest physical respon once the limit of elastic behavior is reached. For real ma rials that behave in a brittle fashion, there is a question a what ultimate stress governs fracture in the presence of finite but highly concentrated stresses. It is not likely to be high a strength as the material’s theoretical ultimate str sU'E/10. Nor is it likely to be as low as the material ultimate stress as determined using standard tension t su'E/1000. In the short term, an estimate of the applica intervening value for a limited range of sizes might be ma via direct calibration with test results. In the long term, th question is likely to require modeling of the material’s m crostructure itself. In addition to such modeling issues c fronting the full implementation of boundary conditions wi cohesive/adhesive laws, companion analysis is now non ear, even in the elastic regime. And this analysis must b sufficient refinement to accurately capture the local stres involved, with their high gradients. For the present, the fore, we can expect to continue to face the longstanding c lenges represented by singularity analysis and interpreta even for configurations that could be freed of singularit with cohesive/adhesive laws.

red e/ ec- of in- ite ress

di- n- al ty: ns. an

ve the

or no

ent

o arp ly, s.

fi- ing n- es. ro- ase se te-

to ow as ss,

s ests, le

de is i- n-

h lin- of

ses re- al-

tion es

2.4 Removing singularities via boundary conditions: Enforcing inequality constraints

Alternative modifications to boundary conditions which c remove stress singularities may be found in additional lo inequalities that are physically required. We begin by de onstrating the way in which this occurs for somesimple, frictionless, contact problems.

The different types of frictionless contact entertained this regard may be distinguished by whether or not they conforming. Here, by ‘‘conforming’’ is meant contact which from no load to full load, has the indentor and indented m terial share a common tangent as the contact region’s bo ary is approached from outside. An example is a roller o relatively flat surface, as occurs in roller bearings. Initial before any loading, the contact region for this configurat consists of a line through the contact pointC ~Fig. 10a!. Subsequently, under loading,C splits into C and C8 as the contact region spreads~Fig. 10b!. Throughout, contact is conforming atC ~or C8) in the above sense. A further ex ample is the closely conforming contact of a journal bear which tends to produce a larger contact region under l ~Fig. 10c!. In contrast is a sharp-edged indentor or flat pun contacting a horizontal surface. This is an example of n conforming contact at bothC andC8 ~Fig. 10d!.

In addition to assuming contact in the configurations Fig. 10 is frictionless or perfectly lubricated, we further sim plify the exposition by taking the indentors~shown vertically hatched! to be rigid. We also assume that they are long in out-of-plane direction so that the 2D state of plane str applies. Then traditional local boundary conditions atC in Fig. 10b, in terms of ther ,u coordinates of Fig. 11, take th form

su5t ru50 on u5p

uu5u0 , t ru50, on u50 (2.17)

for r .0. The first of Eqs.~2.17! are the stress-free cond tions external to the contact region. The second reflectslocal indentation by an amountu05u0(r ) without any friction within the contact region. The local fields for Eqs.~2.17! admit to being supplemented by their fully homogeneo counterparts, namely those for Eqs.~2.17! with u050. Then we recover the classical boundary conditions for a crack~cf, Eqs. ~2.13!!, so that inverse-square-root stress singularit are possible.

To remove the possibility of stress singularities, we adjo physically sensible constraints. These insist that within contact region there can be no tensile contact stress11

while without there can be no interpenetration or cont between the indentor and the indented material. Thus require

su<0 on u50

uu,R02AR0 22r 2 on u5p (2.18)

11Actually, adhesive stresses can supply tensile stresses within the contact regio for most interfaces these stresses are negligible. Johnson@73#, Section 5.5, has an interesting discussion of such effects.

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 265

for r .0, whereR0 is the radius of the indenting roller. Given compliance with these added restrictions, singular response is no longer possible.

To see this, consider what happens otherwise. There are two cases.

i! Singular stresses participate with a positive stress in- tensity factor ofK I(K I.0).

ii ! Singular stresses participate with a negative stress in- tensity factor of2K I .

Under i , the singular stress field must dominate all others as C is approached from within the contact region, so that the contact stresses must become tensile~indicated onu50 in Fig. 11!. This is in violation of the first of Eqs.~2.18!. Under i i , the displacement of the indented material just outsideC is vertically upwards and consequently interpenetrates the in- dentor~indicated onu5p in Fig. 11!. This is in violation of the second of Eqs.~2.18!. Hence, the classical singular fields associated with a crack cannot participate in the conforming contact configuration of Fig. 10b if the inequality constraints of Eqs.~2.18! are enforced.

Fig. 10 Contact configurations:a) unloaded roller bearing,b) loaded roller bearing,c) journal bearing under load,d) piston ring pressing against a cylinder wall~deformation not indicated!

Fig. 11 Local contact configuration atC in Fig. 10b: coordinates and consequences of singularities

e

s

a i

n s

t

t

v

i f

n e r i

c u

c

i t

e

ting e

h ely n-

n- s e of

f h

qs. o ve d

zian

i-

ng ce- lly

east s- si- and

er-

act on- his o en- e

in,

olved

266 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

The question that now arises is can we, in actuality, force the inequality constraints of Eqs.~2.18! and so remove singularities? The fact that Eqs.~2.18! would seem to be physically sensible and therefore desirable does not nece ily mean they can be enforced within classical elasticity. A ter all, singularities in general are nonphysical so it would physically sensible and desirable if we could simply legisl them out of elastic solutions. Unfortunately, such legislat typically leads to the posing of a problem that has no so tion, the local regular elastic fields being incomplete witho their singular counterparts.

For conforming contact, however, we have an additio degree of freedom of which we can take advantage. Thi the extent of the contact region~ie, the length betweenC and C8 that 2l denotes in Fig. 10b!. By suitably adjusting this extent, the inverse-square-root stress singularity can be moved. Then, since there are no other singular fields wi elasticity satisfying the local boundary conditions Eq ~2.17!, or their homogeneous counterparts, the configura is rendered singularity free.

Implicitly, this adjustment of contact extent so as to r move stress singularities is what Hertz did when he sol contact problems of the genre of the roller of Fig. 10a andb ~Hertz @74#!. His solutions all feature contact stresses wh are nonsingular and, indeed, go to zero at the edges o contact region. For example, for the roller of Fig. 10b, the Hertzian contact stress is

sy52 2F

p l 2 Al 22x2 on y50 (2.19)

for 2 l<x< l , whereF is the force per unit length in the out-of-plane direction, andx and y are now as in Fig. 10b with origin O in the middle of the contact region.

The same situation obtains for frictionless conformi contact by rigid indentors in general. Namely, that the ext of the contact region can be adjusted so that only comp sive tractions occur within it and there is no interpenetrat outside of it. Given compliance with these constrain stresses are nonsingular. An example of more extensive forming contact than that of the roller on the half-space der Hertzian assumptions is furnished in Steuermann@75#. Therein closed-form expressions for contact stresses s they behave as ord(r 1/2) as r→0 at the edges of the conta region, the same behavior as in Eqs.~2.19!. As a further example, the closely conforming contact of Fig. 10c is treated in Persson@76#, and demonstrates that stress sing larities can also be removed in this instance.

The same situation does not obtain for nonconform contact. Herein the sharp edges present can set the limi the contact region so that the contact extent is not availa to be adjusted to remove singular behavior. This is the c for the indentor of Fig. 10d. Such configurations require th introduction of appropriate cohesive/adhesive laws to ren them singularity free~as in Section 2.3!.

We now admittwo extensionsto the limited class of con- tact problems considered heretofore. First, we entertain introduction of friction effects. To obtain a bound on thes effects to complement that of frictionless conditions, we c

n-

sar- f- be te

on lu- ut

al is

re- hin s. ion

e- ed

ch the

g nt es- on ts, on- n-

how t

u-

ng s of ble ase e der

the

an

assume that there is no slipping whatsoever. The resul stick conditions within the contact region, in terms of th coordinates of Fig. 11, take the form

uu5u0 , ur50, on u50 (2.20)

for r .0. In Eqs.~2.20!, ur is the radial displacement whic is set to zero by virtue of the indented material complet sticking to the rigid indentor. Again the homogeneous cou terpart of Eqs.~2.20!, taken together with the stress-free co dition onu5p in Eqs.~2.17!, admits the possibility of stres singularities. These are the same as for the adhering tir the Introduction (P3 of Table 2 and Fig. 2b! which has sin- gularities of ord (r 21/2cos(h ln r)) and ord (r 21/2sin(h ln r)). Hence we can anticipate the same response asr→0 at C in Fig. 11. These two singularities occur in combination intwo distinct local fieldswhich can participate independently o each other~except for incompressible plane strain for whic h50 and there is but one local singular field—see E ~1.4!!. Thus adjusting theoneparameter we have available t us, the contact extent, is generally not sufficient to remo both of them. Accordingly, now it can be impossible to fin elastic solutions in compliance with Eqs.~2.18!, and singular stresses can occur. For example, returning to the Hert contact of the roller of Fig. 10b but now with stick condi- tions as in Eqs.~2.20!, the normal contact stress becomes

sy52 2F

p l 2 FAl 22x2 cosS h lnS l 2x

l 1xD D 1

2h lx

Al 22x2 sinS h lnS l 2x

l 1xD D G (2.21)

on y50, for 2 l ,x, l .12 The shear contact stress is sim larly singular.

To alleviate the singular response of direct conformi contact with no slip, one can allow some lateral displa ment. This can be done by applying the load incrementa so that surface material outside the contact region is at l allowed to move laterally prior to coming into contact. Mo sakovskii@78# describes the implementation of such a phy cally more realistic approach. Results are nonsingular comply with the constraints of Eqs.~2.18!. Indeed, for the normal contact stress,sy is as in Eqs.~2.21! but with h 50, so that the Hertzian contact stress of Eqs.~2.19! is re- covered. Similar results obtain for the axisymmetric count part ~see Goodman@79# and Mossakovskii@80#!.13 For both configurations, though, in the limit as the edge of the cont region is approached from within, the ratio of the shear c tact stress divided by the normal approaches infinity. T implies that an infinite coefficient of friction is needed for n slip once contact is made. This in turn suggests that we tertain the possibility of slip in the outer portions of th contact region itself.

12The derivation of Eq.~2.21! is straightforward using complex potential methods as eg, Gladwell@77#, Chapter 4. 13Spence subsequently showed via dimensional analysis that the stress fields inv are self similar, thereby enabling direct implementation rather than incremental~Spence @81#!.

i h

i

o

e

- i e

o

s

o s

o

c o

t a u

the fine as 3

ar e-

- the

ergy en- its

is he

the d to

ck ute ion

act und- r- n is nd the

gy ed

and ase me

sical ergy er- a- ond

in ity. ing

ss -

ic

ten his the le to e ans-

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 267

For such slip under a rigid indentor up to the contact lim at C in Fig. 11, the boundary conditions take the form

uu5u0 , t ru5 f su , on u50 (2.22)

for r .0. In Eqs.~2.22!, f has the magnitude of the coeffi cient of friction. The sign off is taken to be such thatt ru

opposes any slipping displacementur . Consequently for compressive normal contact stresses, sgnf52sgnur , where sgn is the signum function. The conditions of Eqs.~2.22! with the first of Eqs.~2.17! prescribe local boundary cond tions for a slip-to-free transition: When taken abutting t displacement requirements in Eqs.~2.20! if ut ruu,u f suu, they prescribe local boundary conditions for a slip-to-st transition. For both transition configurations, it is possible show only a single singularity exists. Accordingly, by appr priately adjusting the positions of these two transitions, b singularities can be removed. To capture the physics be the loading needs to continue to be applied incrementally~or effectively so via similarity arguments!. Such an analysis may be found in Spence@82# and produces singularity-fre stresses.

As a second extension to the class of contact proble considered, we admitdeformation of the indentor. Results remain essentially the same. For conforming contact with friction, or with friction but allowing for slip, physically rea sonable inequalities can be complied with by adjust boundary region extents and configurations rendered fre singularities. Dundurs and Comninou@83# furnish asymptotic arguments that obeying such inequality c straints removes singular behavior, while there are a num of examples showing that one can actually adjust extent do this ~eg, Johnson@73#!.

In sum, when sufficient degrees of freedom are availa to enable compliance with the pertinent inequalities, str singularities can be removed from conforming contact pr lems. The resulting nonsingular stresses may be loo termed Hertzian, and have been found to be generally s ported by experiments~Johnson@73#, Chapter 4!. In these circumstances, therefore, the stress analyst should make ery effort to comply with the inequality conditions.

3 TRYING TO MAKE PHYSICAL SENSE OF PERSISTENT STRESS SINGULARITIES

3.1 Interpreting crack-tip singularities: The energy release rate hypotheses

We now turn to configurations in elasticity for which th foregoing strategies, while removing singular stresses, d with an approach that is yet not mature. Principal amon these in their practical importance are those involving cra as treated within classical elasticity, so we initially focus trying to interpret the crack-tip singularities.

Griffith was first to appreciate that it is futile to attempt directly interpret the implications for fracture of singul crack-tip stresses. He also appreciated that, while sing nonetheless these stresses are integrable; therefore the be integrated to arrive at a bounded quantity which may physically interpreted. In essence, the particular integra

it

-

- e

ck to o- th

tter,

ms

out

ng of

n- ber to

ble ess b- ely up-

ev-

e so

gst ks n

o r lar,

y can be ted

quantity taken in Griffith@51# is theenergy release rateac- companying crack extensionG, and Griffith hypothesized that G controls brittle fracture at cracks.

The basic elements of the argument which establish energy release rate for incipient crack propagation and de its role in a brittle fracture criterion may be described follows. To fix ideas, reconsider the tensile crack of Fig. but now with the surrounding material being strictly line elastic all the way to fast fracture. Such a material is som times termed ‘‘perfectly brittle.’’ By symmetry in Fig. 3, ten sile crack extension can be expected to occur along x-axis where the maximum tensile stresses occur. The en available to drive this extension comes from the strain ergy of the material surrounding the crack tip: Unlike contributing stress and strain fields, the strain energy bounded by virtue of being an integral of these fields. If t rate such energy releases at the newly formed surfaces in extension exceeds the rate at which it needs to be supplie form them, brittle fracture is hypothesized to occur.

One way in which such an energy argument for cra extension can be implemented is as follows. First, comp the drop in strain energy accompanying a crack extens within some region surrounding the crack tip. Next, subtr the energy transported away as work terms across the bo ary of this region not including the newly formed crack su faces. Thus, the energy released on the crack extensio obtained. Dividing this energy by the extension length, a taking the limit as this length goes to zero, then gives energy release rate for crack propagation.

Alternatively, one can simply compute the strain ener released as work terms on the boundary of the newly form crack surfaces, then divide by the crack extension length take the limit as it goes to zero to obtain the energy rele rate. Both approaches, properly carried out, give the sa result. Both are true energy balances in the sense of clas physics. Both have the strain energy as the potential en source, since this is the ability of an elastic system to p form work by virtue of its deformed state. Given this equiv lence, we choose to focus further discussion on the sec approach here because it is relatively direct.14

In describing such an energy argument, we follow Irw @84# because the analysis therein is elegant in its simplic Hence we consider a crack tip under symmetric load which produces a small extensionda aligned with the origi- nal crack~Fig. 12!. Prior to the extension, the tensile stre ahead of the crack tipsy and the crack opening displace ment back of itv can be identified using an asymptot analysis as in Williams@20#. Locally this results in

sy5 K I

A2px 1O~x1/2! as x→0~x.0!

14It is unfortunate that the variational statement of equilibrium in elasticity is of termed the ‘‘theorem of minimum potential energy’’ and the functional involved in t theorem the ‘‘potential energy.’’ This has led some to confuse this functional with true elastic potential energy, the strain energy. By serendipity, though, it is possib make this mistake and defineG as a derivative of this functional and still obtain th correct energy release rate~essentially this happens because there is no energy tr ported across parts of the boundary where displacements are held fixed!.

268 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Fig. 12 Tensile stress ahead of a crack and displacements ac panying a small extension under symmetric~Mode I! loading

l

t c

e h

com-

Fig. 13 Modes of deformation at a crack tip:a) Mode I, b) Mode II, c) Mode III

I,

ion

rate

a of-

so- m ividu- ful ce,

v5 11k

2m K IA2x

2p 1O~ uxu3/2! as x→0~x,0! (3.1)

for y50, whereinK I continues as the symmetric, or Mode stress intensity factor. The displacement accompanying extensiondv therefore is

dv5 11k

2m K I@11O~da!#Ada2x

2p

1O~~da2x!3/2! as x→da~x,da! (3.2)

on y50. In Eqs.~3.2!, the term in square brackets accoun for the perturbation inK I resulting from the extension. Now the strain energy released on the extension must equa work needed to heal it and restore the crack to its unexten state. For an infinitesimal element of the upper flank in extension, this healing work is one half force times displa ment, or (1/2)(sy dx)(dv). Adding up all such contribu- tions for both the upper and lower flanks of the crack ext sion gives the total work needed to heal it. Dividing by t extension lengthda, then taking the limit asda→0, gives the energy releaserate for crack propagationG. That is,

G5 lim da→0

1

da E0

da

sy dv dx (3.3)

To evaluateG, we introducesy of Eq. ~3.1! anddv of Eq. ~3.2! into Eq. ~3.3! to obtain

G5 11k

4pm K I

2 lim da→0

F 1

da E0

daAda2x

x dx1o~da!G (3.4)

The integral in Eq.~3.4! is readily performed by takingx 5da sin2 t, thereby giving

the

ts

the ded he e-

n- e

GI5 11k

8m K I

2 (3.5)

In Eq. ~3.5! we have added the subscript I toG to distinguish it as being associated with Mode I or tensile crack extens ~with deformation as sketched in Fig. 13a!.

It is also possible to determine the energy release associated with Mode II or shear crack extension~Fig. 13b!. The corresponding energy release rateGII is

GII5 lim da→0

1

da E0

da

txy du dx (3.6)

Proceeding analogously to the derivation of Eq.~3.5! leads to

GII5 11k

8m K II

2 (3.7)

whereK II is the stress intensity factor in Mode II. There is further mode of crack propagation associated with out- plane shear, Mode III~Fig. 13c!. For this mode, a similar derivation gives

GIII 5 K III

2

2m (3.8)

whereK III is the stress intensity factor in Mode III. Each of the foregoing modes of crack extension is as

ciated with a different way of separating material to for new surfaces. Consequently, each must be assessed ind ally in a given application in order to enable meaning comparisons with corresponding critical values. In practi

e

o

c

a

s

l

s a

r

a n s

W

on- ay

h

ave i-

l glo- b- r ex-

r In

rate een lier. in

Ac- ses n in

ter-

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 269

brittle fracture typically occurs in tension rather than shear. This means that under general loading, with both s metric and antisymmetric contributions, crack extension m well not occur aligned with the originating crack~as in Fig. 12!, but rather along a ray emanating from the crack which maximizes the energy release rate in Mode I. Even we still need to be able to distinguish amongst the differ contributions to the total energy release rate so as to ob the maximumGI and determine the ray on which it acts f a given loading.

Two observations can be made on the foregoing res for energy release rates. First, only the singular stresses their displacements contribute to these rates. Second, ac ing the hypothesis thatGI , GII , andGIII control brittle frac- ture in their various modes is completely equivalent to cepting the corresponding stress intensity factors,K I , K II , andK III , in this role.

The literature has a number of other developments of elastic energy argument for brittle fracture which are con tent with the preceding.15 Several of these express the ener release rate with path-independent integrals which enc the crack tip: in chronological order, Eshelby@86#, Sanders @87#, Cherepanov@36#, and Rice@88#. The isolation of the contributions from different modes of crack propagation i little more awkward to effect with these integrals. This m in part account for current practice preferring to expre brittle fracture criteria in terms of stress intensity facto rather than energy release rates.

Originally, Griffith hypothesized that brittle fracture oc curs when the energy release rate equals the surface en of the solid being fractured. Later, Irwin@89# and Orowan @90# independently argued that the energy ‘‘sink’’ for fractu could also include some plastic dissipation, provided the tent of any accompanying yield region is limited to the im mediate neighborhood of the crack tip. This extension Griffith’s original hypothesis realized the practical benefit enabling the approach to be applied to metals. Aside fr these hypotheses as to acceptable energy sinks for frac there is a further basic hypothesis underlying either proach. This has that the integral of something which is physically appropriate—namely singular crack-tip stresse can yet furnish something which is—namely the energy lease rates accompanying crack extension. Thus, the en arguments of classical fracture mechanics contain two potheses: one for the energy sources for fracture, the o for the energy sinks. Each one needs to be complied with the approach to be successful. The extent to which they in fact, can be judged by the degree of agreement of physical evidence with predictions based on the pair. review some physical data with this issue in mind Section 3.4.

3.2 Energy release rates for interface cracks

The ease with which the singular fields for a crack can integrated to provide energy release rates suggests tryin

-

in ym- ay

tip so, nt

tain r

ults and ept-

c-

the is-

gy ose

a y ss rs

- ergy

e ex- - of

of om ture, p- ot —

re- ergy hy- ther for are, the

e in

be g to

extend this means of singularity interpretation to other c figurations. Perhaps the most natural to consider in this w is theinterface crackconfiguration wherein material on eac side of the crack plane can have different elastic moduli~Fig. 14!. Now, as noted in the Introduction, Williams@21# shows that the inverse-square-root singularity of a crack can h oscillatory multipliers. For example, in terms of the cylindr cal polar coordinates of Fig. 14,

su5O~r 21/2cos~h ln r !!1O~r 21/2sin~h ln r !!

as r→0 (3.9)

on u50, whereh is as in Eq.~1.3!. That these possible loca singular stresses do in fact participate in the response of bal configurations is confirmed by solutions to such pro lems, as in England@91#. Undertaking an analysis as earlie for companion energy release rates associated with crack tension along the interface then gives, for Mode I,

GI5c1 lim da→0

@c8 cos~2h ln da!1c9 sin~2h ln da!#

(3.10)

where c, c8, and c9 are constants (generallyÞ0). Clearly the limit in Eq. ~3.10! does not exist. A like result holds fo GII , and accordingly neither is a well-defined quantity. combination, however, the terms that are undefined inGI and GII can be shown to cancel, so that a total energy release does exist. Nonetheless, the inability to distinguish betw modes is unsatisfactory for the reasons indicated ear Moreover, the situation is not improved if crack extension directions other than along the interface is entertained. cordingly, the generally more nonphysical singular stres of interface cracks would seem to require special attentio order to effect satisfactory physical interpretations.

This need has occasioned a series of models for the in face crack to be put forward since Williams@21#: the contact zone model of Comninou@92#, the crack-opening-angle model of Sinclair@93#, the intervening-layer models of At

be in

Fig. 14 An interface crack configuration

15There are also some articles which are not consistent and may be shown to error—see Keating and Sinclair@85# for a review.

270 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Fig. 15 Crack-tip models for the interface crack:a) contact zone model,b) crack opening angle model,c) intervening layer model with constant moduli,d) intervening layer model with continuously varying moduli

o

g

d

ials the

kinson @94#, and the perturbed moduli model of He an Hutchinson@95# and Suo and Hutchinson@96#. We review these models next.

There is a further unsatisfactory aspect of elastic soluti for interface cracks based on the model of Williams@21#. As pointed out in England@91# and Malyshev and Salganik@97#, the crack-flank displacements also oscillate and in so do interpenetrate one another. This interference between crack flanks lead Malyshev and Salganik to suggest introd ing a contact zone for the crack flanks immediately conti ous to the crack tip. Such acontact zone modelwas first pursued in Comninou@92#, and subsequently has seen qu extensive investigation—see references in Comninou@98#. The basic elements of such a model are as follows.

In terms of the cylindrical polar coordinates of Fig. 15a, three types of conditions near the original crack tip atO are prescribed in the contact zone model for the interface cra the matching conditions for perfect bonding ahead ofO,

d

ns

ing the uc- u-

ite

ck:

suu u501

5suu u502

, t ruu u501

5 t ruu u502

ur u u501

5ur u u502

, uuu u501

5 uuu u502

(3.11)

for r .0; the frictionless contact conditions behindO,

suu u5p

5suu u52p

, uuu u5p

5uuu u52p

t ru50 on u56p (3.12)

for 0,r , l , where l is the extent of the contact zone; an the stress-free conditions once contact ceases atO8,

su5t ru50 on u56p (3.13)

for r . l . Equations~3.11!–~3.13!, when taken together with the planar elastic field equations for the respective mater and boundary conditions describing loading remote from

r

s

m

a

s s

i e b

o

e e

a

h

f a l ons tip, he

o-

at

in-

tact ack up- del, part ad, del

tion e

astic he

i- ne ied. olu- en not os-

k as ig-

ed i ho-

of the

r let- e

yer

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 271

crack, can constitute a complete problem statement in e ticity. In addition, though, we would like to adjoin a con straint which prohibits interpenetration once conta ceases—a primary motivation for the model in the fi place—as well as a constraint which only admits compr sive stresses within the contact zone. Can we do this? answer is yes for two reasons. First, we can adjust the ex of the contact regionl so as to remove the one singular fie possible atO8, thereby ensuring no interpenetration: This the same adjustment as used to the same end in friction conforming contact in Section 2.4. Second, the local field O ~elucidated in the Appendix of Comninou@92#! can feature a compressive inverse-square-root singularity in the nor stress within the contact zone~viz, in su on u5p in Fig. 15a!. Once present, this singular stress means that, no m how hard we pull the overall configuration apart with remo tensile loading, the finite biaxial tensile stresses so induce O can never completely negate the infinite compressive n mal stress there. Hence there can always exist a region, a possibly a small one, in which contact stresses are comp sive.

The question that now arises is when is such a local gular stress field excited in global problems? Somewhat prisingly, Comninou@92# shows that the closing of the crac tips present in the contact zone models can occur in glo configurations when the loading over the crack flanks is p dominantly in the opening mode~as in Fig. 14!. In Comni- nou @92#, when an interface crack is under uniform tens loading at infinity, fields for the contact zone model are d termined which have compressive stresses within the con zone. These fields are also free of any interference betw the crack flanks. Furthermore, the solution obtained once contact zone model is adopted may be shown to be un ~see Comninou@98#!. Looking ahead of the crack tip in th model (O in Fig. 15a!, we find the shear stress alone to singular~see the Appendix in Comninou@92#!. This has to be the case since a normal singular stress there, if tensile, w separate the crack flanks in the contact zone, while if co pressive, would cause them to overlap one another. The sociated energy release rates for crack propagation along interface are~Comninou@98#!

GI50, GII5 m̂1m̂2

4m1m2~m̂11m̂2! K II

2

m̂15m11k1m2 , m̂25m21k2m1 (3.14)

The Mode I energy release rate is zero by virtue of th being no singular normal stress on the bonded interfac the contact zone model.

In an attempt to complement the contact zone model w one which does permit crack propagation along the interf in an opening mode accompanied by a positive energy lease rate, the following simple tactic is suggested in Sinc @93#. While the singular character of a crack can be increa by the introduction of an abrupt material discontinuity on t crack plane, it can be reduced by opening the angle s tended at the crack tip prior to loading~Fig. 15b!. The two effects can be adjusted so as to offset one another and

las- - ct st es- The tent ld is less at

al

tter te d at or- lbeit res-

in- ur-

k bal re-

ile e- tact een the que

e

uld m- as- the

re in

ith ce re-

lair sed e

ub-

re-

cover the nonoscillatory inverse-square-root singularity o crack in a single material. Thiscrack-opening-angle mode shares with the contact zone model the matching conditi for a perfectly bonded interface ahead of the crack namely Eqs.~3.11!, and has stress-free conditions on t crack flanks which are taken to subtend an angle ofF in the unloaded state. That is, in terms of the cylindrical polar c ordinates of Fig. 15b,

su50, t ru50, on u5p2F,2p (3.15)

for r .0. For given material moduli, the value ofF that removes the oscillatory multiplier of the stress singularity O may be found in Sinclair@93#. For this angle, the crack flanks open without interference under tensile loading as dicated in Fig. 15b.

The approach may be viewed as the dual of the con zone model. In the contact zone model, contact of the cr flanks is anticipated and boundary conditions thereon dated to reflect this event. In the crack-opening-angle mo crack opening is anticipated and the crack flanks angled a to promote this event. For a specific applied tensile lo assuming crack opening via the crack-opening-angle mo can be shown to lead to a unique solution~the proof follows along the lines of Knowles and Pucik@99#!. Both the normal and shear stress on the bonded interface in such a solu are singular~provided FÞ0). Companion energy releas rates are given by

H GI

GII J 5 H c

c8J @m1 2~11k2!21m2

2~11k1!2#

m1m2~m̂11m̂2! K2 (3.16)

where m̂1 and m̂2 are as in Eqs.~3.14!, c and c8 are now dimensionless constants whose values depend on el moduli, andK is the one stress intensity factor present in t model.

For a given interface crack configuration with predom nantly tensile loading as in Fig. 14, both the contact zo model and the crack-opening-angle model can be appl Once a decision is made as to which model to use, the s tion for that model is unique and free of interference betwe crack flanks. So which one should we use? The answer is obvious, but quite possibly neither. This is because, in cho ing one or the other, themodeleris making the decision as to the relative contributions of Mode I and Mode II to crac propagation along the interface. In effect this decision h that, for a broad spectrum of remote loadings having a s nificant tensile component, the ratio ofGI to GII is to be in one or the other of the but two fixed proportions prescrib by Eqs. ~3.14! and ~3.16!. If it happens that this a prior selection is physically appropriate, the use of the model c sen may be justifiable. If not, then not. In general, themodel should make the decision as to the relative participation Modes I and II, and this decision should be sensitive to specific loading being applied.

Atkinson @94# furnishes a pair of alternative models fo the interface crack tip. These each have the attribute of ting loading interact with the model itself to set relativ mode participation. They both feature an interface la which contains a stress-free crack. In the first~Fig. 15c!, the

s

x

a

t o

e

e o

i n

p

a i

u s

n

a

i

t

i e

ms

lute

for tio

in his ral-

the

han the

si-

is rate een-

d by lar

lace- g

272 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

intervening layer is taken to be homogeneous with ela moduli which are intermediate to the parent moduli (0<c <1 in Fig. 15c!. In the second~Fig. 15d!, the intervening layer is taken to have varying moduli which effect a contin ous transition from those for one of the parent materials the other~a linear variation is shown by way of simple e ample!. For both of theseintervening-layer models, the inverse-square-root singularities of a crack in a single m rial are recovered together with their associated energy lease rates, Eqs.~3.5! and ~3.7!. Therefore, response as mode of crack propagation is essentially the same as f crack in a single material. That is, in accordance with one the following scenarios.

i! If any tensile Mode I fields are excited by the appli loading, no interference or contact occurs betwe crack flanks and Mode I propagation along the int face is possible. Mode II contributions to propagati may also be present under these circumstances.

ii ! If any compressive Mode I fields are excited, interpe etration of the crack flanks is predicted and must alleviated by admitting contact between them. Cra propagation along the interface can only occur in Mo II under these circumstances.

iii ! If Mode I fields are not excited at all, Mode II is obv ously the only possibility for propagation along the i terface. Such propagation may occur with or witho contact between the crack flanks, depending on the ticipation of other regular crack-tip fields.

Consequently here, under tension, there would not appe be any reason for the modeler to fix the relative participat of modes prior to applying actual loading.

For the models of Atkinson@94# to be physically appro- priate, the heights of the layers (2h andh in Figs. 15c andd! need to be physically reasoned. On the atomic/molec level, one can envisage a small region in which the cohe laws acting within material 1 switch to adhesive laws b tween materials 1 and 2, then to cohesive laws within ma rial 2. The height of this transition region can be expected be of the order of several atomic/molecular diameters. C stitutive laws can therefore also be expected to vary ove similar size scale. Thus the incorporation of intervening la ers into the global analysis of crack configurations is without significant analytical challenges.

Nonetheless, the intervening-layer models of Atkins @94# are conceptually valuable and support the mode of cr propagation being dependent on applied loading in much same way as for the crack in a homogeneous plate. It follo that these models do not in general support the use of e the contact zone model or the crack-opening-angle mo Rather they lend support adopting the strategy for trea interface cracks first put forward in He and Hutchinson@95#, and subsequently amplified in Suo and Hutchinson@96#. This strategy simply setsh of Eqs.~1.3!, ~3.9!, and~3.10! to zero by suitably adjusting material moduli. Such aperturbed moduli modelhas no oscillatory character and accompany crack-flank interference, and has decoupled energy rel rates,GI andGII , as in Eqs.~3.5! and ~3.7!.

tic

u- to -

te- re-

o r a of

d en r- n

n- be ck de

- - ut ar-

r to on

lar ive e- te- to

on- r a y- ot

on ck the ws ther del. ing

ng ase

Further support for the perturbed moduli model ste from the fact that, generally,h is small (uhu,7/40, see Eq. ~1.3!!. Hence setting it to zero does not change the abso value of the singularity exponent much (,6%), nordoes it necessitate dramatic changes in elastic moduli. Indeed, plane strain, it is always possible to maintain the actual ra of shear moduli sought in an application and geth to be zero by adjusting a Poisson’s ratio while still maintaining it with the physical range of zero to one half. More precisely in t regard, one can proceed as follows. Without loss of gene ity, number the materials so thatm1<m2 . Then replace the actualk1 by k̂1 where

k̂15~k221! m1

m2 11 (3.17)

This replacement value by itself ensuresh50 for plane strain. For plane stress, though, some modifications to true ratio of the shear moduli are needed to renderh50 when one shear modulus differs from the other by more t a factor of three. Even so, the strategy would seem to be most effective way of treating interface cracks within clas cal elasticity at this time.

3.3 Interpreting other singularities: The K-controlled annulus hypothesis

In attempting to interpret other singular configurations, it natural at the outset to try and extend energy release arguments. As a demonstration, we consider the sharp r trant corner under symmetric loading~Fig. 16!.

The stresses directly ahead of the corner are dominate the singular field there. Thus, in terms of the rectangu Cartesian coordinates of Fig. 16,

Fig. 16 Tensile stress ahead of a reentrant corner and disp ments accompanying a small extension under symmetric loadin

all . a

i- he en-

for ne

cir-

to

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 273

Fig. 17 K-controlled annulus at a crack tip

a

a i

E

pri- -tip t le, e

ck

of ar one ion r - r,

cular

eld u- l ic

e

than

al ur

tle si- ile

on

at-

rner ith

us-

th ti- rial nd on as

ame

q. r’’

th- ned ase

ilar s.

are in f

linear elastic fields for cracks to instances in which sm scale yielding is admitted. These elements are as follows

In the immediate vicinity of a crack tip there exists region wherein theK fields are just not physically appropr ate by virtue of the singularity they contain. To keep t development simple, enclose this region within a circle c tered on the crack tip (R0 in Fig. 17!. If the radius of this region in small enough, immediately outside it stresses theK fields dominate those for all other elastic fields. As o moves further away, the stresses of theK fields become of comparable magnitude to the others present. Under these cumstances, the region exterior toR0 may be subdivided into an annular regionRa in which theK fields are valid and dominant, and a still further removed exterior regionRe in which theK fields, while valid, have comparable stresses the regular fields~Fig. 17!. Within the annulusRa , then, the K fields can be regarded as prescribing physically appro ate traction boundary conditions for the innermost crack regionR0 . In this sense,K can be expected to control wha happens at the crack tip. If the material is perfectly britt this meansK controls brittle fracture at the crack tip. If th material is ductile but any yielding is confined to withinR0 , K may be still be viewed as controlling fracture at the cra tip.

Before examining the implications of aK-controlled an- nulus interpretation further, some additional clarification the notion of theK stress fields dominating the other regul stresses present in the annulus is helpful. At the outset, might be tempted to adopt the obvious but stringent criter that the tractions fromK fields dominate those from regula fields atall points on a circular arc within the annulus. Un fortunately, this is typically not possible. To explain furthe the stress components that can act as tractions on a cir arc withinRa ares r andt ru ~Fig. 17!. If the crack is under symmetric loading, theK I stress field hass r50 at u56p ~see, eg, Tada et al@55#, p. 1.4b!. Thus, the magnitude of the associated traction cannot dominate that of any regular fi with s rÞ0 atu56p, and there are a number of such reg lar fields for Mode I cracks~these stem from polynomia solutions!. Alternatively, if the crack is under antisymmetr loading, the K II stress field hass r5t ru50 at u5

62 sin211/) ~ibid!. Again there are regular fields whos tractions are not dominated. Accordingly, whiles r and t ru

may act as controlling tractions withinRa , the criterion for them to do so needs to be based on something other their own values point by point throughoutRa .

To develop an alternative criterion, we take the princip physical phenomenon we are trying to capture with o K-controlled annulus interpretation for a crack to be brit fracture on radial rays emanating from the crack tip. Phy cally, brittle fracture is predominantly caused by tens stresses. It follows that we can expect the hoop stress radial rayssu to control this event. Of course, withinR0 this stress component goes to infinity in traditional elastic tre ments. However, withinRa it does not. Hence, in principle we can check if the maximumsu for theK fields inRa is an

sy5 K2g

A2p xg 1o~1! as x→0~x>0! (3.18)

on y50, whereing is the singularity exponent andK2g its associated stress intensity factor. In the event that the co angleF tends to zero and the corner becomes a crack w g51/2, we recover the first of Eqs.~3.1!. Other values of the singularity exponent for other corners can be determined ing Williams @20#. These values show thatg,1/2 for F .0 ~eg, forF590°, g50.46).

Now we entertain a small crack-like extension of leng da. We take this extension to be in the form of a mathem cally sharp small crack when undeformed so that no mate is removed. Its deformed shape is indicated in Fig. 16 the companion displacement assumes the same express previously, namely as in Eq.~3.2!. Furthermore, using the same energy argument as in Section 3.1 results in the s expression for the energy release rate, namely Eq.~3.3!. On introducing Eqs.~3.18! and Eq.~3.2! into Eq.~3.3!, and mak- ing the change of variable as for~3.4!, we obtain

GI5 11k

4pm K2gB~12g,3/2! lim

da→0 K Ida1/22g (3.19)

whereB is the beta function. What becomes apparent in ~3.19! is that the crack is a somewhat fortunate ‘‘corne since it is the only one which has a finiteGI ~becauseg 51/2 andK IÞ0 asda→0). All other corners have zeroGI

~becauseg,1/2 andK I is bounded asda→0). This conclu- sion can also be reached using any of the valid pa independent integrals for the energy release rate mentio in Section 3.1. Accordingly, we cannot use an energy rele rate interpretation for reentrant corners withF.0. More- over, energy release rate arguments break down in a sim way for a wide variety of other singular configuration Needed, therefore, is an alternative interpretation.

One alternative interpretation argues for aK-controlled annulus. The essential elements of such an argument given in Irwin @100#; a more extensive discussion is given Rice @101# in the context of extending the applicability o

e

s

e

e

t t

n i

t t

t

e

o

n i

n c

i

h

eir ons

rc use

not is ot d-

e utt

the e is a

he ic

ne

mon is a

he- her po-

y be ex- ngths in- to ns, less

er ex- ng t

ture

ets

tion

274 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

order of magnitude greater than the maximumsu from any regular fields. If this is the case, it would seem reasonabl view theK fields as dominant.

The foregoing is but one fairly simple possibility for a sessing the dominance ofK fields in the annulus; others ex ist. For Mode I configurations, the resultingK-controlled an- nulus interpretation is the same as accepting the en release rateGI as controlling fracture. For mixed-mode situ ations, it is essentially equivalent to the fracture criteri proposed in Erdogan and Sih@102# and supported by som experimental evidence therein. Thus while this choice is unique, it does not appear unreasonable and it does focus ensuing discussion. Moreover, it is not expected other reasonable alternatives would significantly alter conclusions drawn from this discussion.

One attribute of theK-controlled annulus interpretatio that we have adopted is that it can be applied to singular other than just those at a crack. Reconsider our earlier ample of a reentrant corner under symmetric tensile load ~Fig. 16!. Even for a corner subtending an angle of 90°, singularity is almost as strong as for a crack. Hence, if K-controlled annulus argument is successful for a crack can reasonably be expected to be capable of extensio reentrant corners subtending angles up to 90° and subje to tensile loading. For other singularities which are y weaker, the dominance of the associated stresses is con to a smaller neighborhood of the singular point, but so may be the region wherein stresses are not physically ap priate. Consequently, it would not seem unreasonable to tertain the possibility of aK-controlled annulus for thes situations as well.16

While successful in extending the range of singular c figurations that can be interpreted over that with just ene release rate arguments, there are some shortcomings i K-controlled annulus approach for these other singularit We demonstrate this next with some examples.

Returning to the special case of the 90° reentrant cor typically mixed-mode loading can be expected in practi That is, loading which is neither purely symmetric nor pure antisymmetric about the bisector of the corner angle. Un these circumstances, there are twodifferent typesof stress singularities that can be present~as forF in Fig. 2d andP5

of Table 2!. Thus, in terms of the cylindrical polar coord nates of Fig. 16,

su5 Ks

A2pr 0.46 1

Ka

A2pr 0.09 as r→0 (3.20)

on u50. In Eq.~3.20!, Ks andKa are the generalized stres intensity factors associated with symmetric and antisymm ric loading, respectively. Given the different orders of t stress singularities associated withKs andKa in Eq. ~3.20!, it is not clear that they share a common annulus. Further, e if they do, we cannot always tell if one of the two fields

d it ger ite

c d

to

- -

rgy -

on

not help hat he

ties ex- ing he he , it n to cted et fined oo pro- en-

n- rgy

the es.

er, e. ly der

-

s et- e

ven is

dominant with respect to the other, or how to combine th effects. This is so even if we continue to base our decisi in this regard onsu in the annulus despitesu now having distinctly different dependencies onr . The reason is that, if Ka happens to be bigger thanKs , we cannot tell which has the largersu without a knowledge of the radius of the a whereon we are making the comparison. However, beca we do not know the true physical stress field, we are really able to specify the location of the annulus with th arc. So here aK-controlled annulus interpretation does n readily permit predictions of what happens under mixe mode loading.

A further example of a potential shortcoming in th K-controlled annulus approach is that of the epoxy-steel b joint ~as for P7 of Table 2 and Fig. 2e!. Herein there is but one stress intensity factor for both the normal stress and shear on the interface. Hence, any fracture on the interfac constrained by traditional elastic modeling to occur with fixed ratio of tensile contribution to shear, irrespective of t composition of the far-field loading. This sort of unrealist limitation makes it unlikely that this singleK can be reliably used to predict brittle fracture for widely differing loads.

On the other hand, for a single type of configuration, o may be able to use stress intensity factors torank different adhesives’ strengths. When the adhesives share a com Poisson’s ratio and the adherend is relatively rigid, there common singularity exponent and the value ofK at fracture can be expected to reflect the relative strength of the ad sives for the particular test configuration used. For ot adhesive-adherend interfaces wherein the singularity ex nents are not identical but are close to one another, it ma possible to perturb elastic moduli so that the singularity ponents become the same, and then rank adhesive stre for the specific configuration of concern. However, when s gularity exponents differ to the point that it is not judged be reasonable to coalesce them via moduli modificatio comparisons of adhesive strengths can be expected to be consistent.

To explain further, suppose that, for a butt joint und tension with a single adherend, glue 1 has a singularity ponentg1 and a stress intensity factor at brittle fracture alo the interface ofK1 , while glue 2 has a singularity exponen g2 and a stress intensity factor at fracture ofK2 . Thus, ifs f

a

and s f b are the normal stresses on the interface at frac

for glue 1 and 2, respectively,

s f a5

K1

A2pr g1 1o~1!, s f

b5 K2

A2pr g2 1o~1!, as r→0

(3.21)

wherer is now the distance from where the interface me the outside of the specimen (P7 in Fig. 2e!. Two different situations can now be envisaged. First, we have the situa where we can name the two glues so that

g1.g2 and K1.K2 (3.22)

Then the local stresses are uniformly higher for glue 1 an is reasonable to conclude that glue 1 is probably stron than glue 2 for the given butt joint in tension, and qu

rack uli utch-

16K-controlled annulus interpretations have also been advanced for the interface ~Rice @103#!. They particularly merit being considered when the perturbed mo approach is not applicable. Such can be the case in anisotropic configurations; H inson and Suo@104# reviews such instances.

i

h

p

o

d i

n e

- a

h e p

f

t

s

h

ns n

- for a be- ar a ing ven p-

cur

ex-

ear usly n, ly rt- - p-

ion

iew nd

o is

rm- is ary-

ing

hat

ith

ar t it a

he ot to

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 275

possibly so for like butt joints. Second, we have the situat wherein, regardless of how we name the two glues, we h a jumbled result as in

g1.g2 but K1,K2 (3.23)

or vice versa. Then the local stresses vary as to which g has higher values and it is difficult to decide which is t locally stronger glue just for the given butt joint, let alon more generally. As with the earlier reentrant corner exam we would need to know the actual location of a comm K-controlled annulus to make a judgment, assuming a c mon annulus exists.

As a final example of a shortcoming in theK-controlled annulus approach, we consider a relatively rigid chisel denting a block~as for P9 of Table 2 and Fig. 2g!. The singularity coefficient, or generalizedK, in this instance de- pends upon the angle of the chisel tip and the elastic mo of the block. However, it is independent of the load appl to the chisel. Accordingly, even if aK-controlled annulus existed for such a configuration, theK involved could not be used to estimate loads to fracture.

These examples illustrate the sort of difficulties that c be encountered in employing the more generally applica interpretation of aK-controlled annular region to stress si gularities. They underscore that care needs to be exercis order to appreciate the limitations of the approach and us consistently.

The hypothetical nature ofK-controlled annulus argu ments also bears comment. The basic hypothesis has th one moves away from a singularity, the singular stresses come physically applicable while they still dominate t other stresses present. So what is ‘‘physically applicabl Certainly the singular fields would seem to have to com with all three of the underlying and unpoliced assumptions linear elasticity to be assured of attaining this quality. This not difficult to check in specific instances. For example, the Griffith crack~Fig. 4 with b→0), compliance with the assumptions of elasticity on the crack plane ahead of crack tip is tantamount to insisting that the stresses there at or below yield levels. Using coordinates as in Fig. 3, normal stress on this plane is~see, eg, Tada et al@55#, pp 1.20, 5.1!,

sy

s0 5

x1a

Ax~x12a! on y50 (3.24)

for x.0. Determining the locationxY when this stress equal the yield stress gives

xY

a 5F12S s0

sY D 2G21/2

21 (3.25)

The corresponding contribution of theK field, the singular stresssy

s , is then given by~ibid, pp 1.3, 5.1!

sy s

sY 5

s0

sY A a

2xY (3.26)

Provided far-field loading is maintained below 50% of t yield stress, Eqs.~3.25! and ~3.26! have theK field as con-

on ave

lue e e le,

on m-

in-

uli ed

an ble - d in e it

t, as be- e ?’’ ly of is or

the are

he

e

tributing 90% or more of the normal stress atxY . Conse- quently, here we can readily find a location at which theK field dominates while complying with the assumptio within elasticity theory. Similarly, other configurations ca and should be checked for such compliance.

Unfortunately, while complying with the underlying as sumptions of elasticity may be a necessary requirement the K fields to be physically applicable, it is generally not sufficient one. This is because the physical discrepancy tween the singular stresses and reality that must occur ne singular point may result in the singular stresses continu to deviate significantly from the true physical stresses e when they are in accord with all three linearizing assum tions. We demonstrate how this type of deviation can oc in the simpler context of beam theory next.

To this end, we reconsider the earlier cantilever beam ample taken from Frisch-Fay@34# ~Section 1.4!. This ex- ample compares tip deflections from nonlinear and lin beam theory, and demonstrates that linear theory is serio in error. At the other end of the beam where it is built i though, both theories give the same ‘‘deflection,’’ name zero. Moreover, we can identify a length of the beam, sta ing where it is built in, within which it is reasonable to re gard linear theory as being in compliance with the assum tion that recovers it from nonlinear theory. This assumpt has

F11S dv dxD

2G3/2

'1 (3.27)

where v is now the beam deflection andx the coordinate along the beam’s length. Suppose now we adopt the v that anything up to a 10% difference between the right-ha and left-hand sides of Eq.~3.27! can be regarded as the tw sides being in fair agreement. It follows that linear theory in fair agreement with its underlying assumption if

dv dx

< 1

4 (3.28)

Solving the linear beam problem, withx measured from the built-in end and for the specifications of Frisch-Fay@34#, we find Eq.~3.28! is met for 0<x<64 mm~2 1/2 inches!. Thus, in this range linear beam theory can be regarded as confo ing with its underlying assumption. However, also in th range, linear beam theory has the bending moment as v ing from 11.3 to 11.0 N-m~100 to 97.5 lbf-in!. In fact, non- linear beam theory gives the bending moment as vary from 5.0 to 4.7 N-m~44 to 41.5 lbf-in!. This is a discrepancy of more than a factor of two. What is happening here is t the physically inaccurate predictions of linear theory forx .64 mm are continuing to pollute predictions whenx ,64 mm, even though linear theory is in fair agreement w its underlying assumption in this range.

The same sort of pollution is a possibility for singul stress fields. Not to say that it has to happen, just tha might. Accordingly, we simply cannot know whether K-controlled annulus really exists without knowing what t true physical stress field is, something we typically do n know. Consequently we need to resort to indirect means

e

c

d g o

h t

a

t t n i

e

a

o

u

n

t

o

also

o- lts yer - ns, ent t re-

ngle of

95% e

in

rted at ing yer - ures ci- cat- ted ting

ari-

u-

be-

well ee hat

m- o -

of

se lts

276 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

infer the existence or otherwise of aK-controlled annulus. Arguably the best of these is to examine the physical acc with predictions made viaK. We look to examine some physical evidence in this light next.

3.4 How well doK interpretations work?

As remarked in Section 3.1, the two interpretations of str singularities discussed here are mutually consistent inasm as both identify stress intensity factors as the parameters trolling brittle fracture. The energy release rate hypothe does this for cracks; theK-controlled annulus for cracks an other singularities. Too, both interpretations share an inte tion of singular stresses. The energy release rate appr does this directly as in Eqs.~3.3! and~3.6!; theK-controlled annulus indirectly by, in effect, considering control on t boundary of a region including the singularity. Hencefor therefore, we refer to both as simplyK interpretations.17

By far the greatest practical application ofK interpreta- tions is to configurations entailing cracks. The attend technology is termed linear elastic fracture mechan ~LEFM!. Currently, LEFM plays a central role in attempts try and ensure the structural reliability of components in presence of cracks. Accordingly we focus our assessme how well K interpretations work on their performance with LEFM.

Linear elastic fracture mechanics leads the field of so mechanics when it comes to explicitly recognizing the pr ence of singularities and attempting to interpret them in physically meaningful way. In implementing this activity, has made considerable progress in the last 50 years. At time, the analytical tools for determining stress intensity f tors are well in hand. For most configurations, sufficien accurate determinations of stress intensity factors for pra cal purposes can be made either by drawing directly on c pendia ofK ’s ~Tada et al@55#, Rooke and Cartwright@105#, Sih @106#, and Murakami et al@107–109#!, or by applying suitably-adapted numerical methods that have been de oped. On the testing side, procedures have been well tho out so as to limit plasticity effects and, thereby, enhance applicability of LEFM. For fracture under monotonic load ing, the design of these procedures was led by Srawley Brown @110#. In essence,@110# takes advantage of the con straint inhibiting plastic flow that is produced by increasi thickness. This constraining effect enables restriction of estimate of the yield region extent to being within abou percent % of the crack length. This in turn allows applicati of the approach to metals, and@110# is now the basis of a standard for the determination of plane-strainfracture tough- nessfor metallic materials,K Ic . That is, the test procedure t be followed to ensure limited plasticity when obtainingK Ic , the critical value ofK I at which fracture commences for given metallic material. This standard of the American So ety for Testing and Materials, ASTM E399, has furth evolved since Srawley and Brown@110# to the point that it is

this u

ord

ss uch on-

sis

ra- ach

e h,

nt ics o he t of n

lid s- a

it this c-

tly cti- m-

vel- ght

the - and - g an 2

on

a ci- er

difficult to imagine any further significant refinements.18

Present-day testing practice that applies this standard is typically reliable.

Evidence of the reproducibility achieved by testing lab ratories in applying ASTM E399 is available in the resu found from round-robin testing programs described in He and McCabe@112#, McCabe@113#, and Underwood and Ken dall @114# for bend specimens, compact tension specime and C-shaped specimens, respectively. All told, 17 differ laboratories took part in these programs. From these tes sults, scatter can be estimated inK Ic determination.

In undertaking this assessment, we use results for a si specimen type with a fixed nominal size and comprised the same material. As a measure of scatter we take the confidence limits of the normal distribution divided by th mean and expressed as a percentage. That is,6100 (1.96sK Ic

/K̄ Ic), sK Ic being the sample standard deviation

K Ic , andK̄ Ic being the mean value ofK Ic . We compute these scatter measures using only results which have no repo violations of ASTM E399 whatsoever. That is, results th are free of any designation indicating concerns regard compliance with the standard in Tables 3, 2, and 3 of He and McCabe@112#, McCabe@113#, and Underwood and Ken dall @114#, respectively. We then average the scatter meas so found for the different materials tested with a given spe men type to obtain an overall representative measure of s ter for that particular type of specimen. Results are presen in Table 3. Given the considerable demands placed on tes by ASTM E399, the reproducibility ofK Ic evident in this table is a tribute to the effort and care expended by the v ous laboratories taking part.

In all, the implementation of aK interpretation for crack- tip singularities is a credit to the fracture mechanics comm nity. Hence, in considering how well suchK interpretations work, the practice of them can generally be regarded as ing well done and reliable.

As, arguably, the most basic means of assessing how theK interpretation of LEFM works we can check the degr to which fracture toughness is truly a material property. T is, a property for a given material which remainsconstantfor different configurations which are acceptable within the li its of applicability of the theory. One way of doing this is t consider values ofK Ic for the same material found by differ ent laboratories over the years.

In undertaking this survey we draw upon compendia sources of such data assembled in Hudson and Seward@115- 117#. Materials selected for inclusion in the survey are tho with greater numbers of different sources furnishing resu so as to gauge the presence of any variability better. To

Table 3. Scatter inK Ic testing

Specimen type Intralaboratory Interlaboratory

Bend 67% 611% Compact tension 64% 6 7% C-shaped 65% 610%

17It is also possible to regard Barenblatt’s approach as an ‘‘interpretation’’ of sing stress fields, although it removes them. Viewed in this way, it too identifies K as parameter controlling fracture.

lar the 18A recent version of this standard may be found in ASTM@111#.

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 277

Table 4. Variability in K Ic

Material Sample size Variability

Steels 4340:T1 14 650% 4340:T2 32 662% 4340:T3 39 641% 4340:T4 29 641% Aluminum alloys 6061-T651 11 625% 7075-T6 35 651% 7079-T6 13 644% Titanium alloys Ti-6A1-4V:P1 23 639% Ti-6A1-4V:P2 29 662% Ti-6A1-4V:P3 21 649% Ti-6A1-4V:P4 39 661% Ti-6A1-4V:P5 17 643%

e

f

h t t

i

c n

e p t t t

d n

n r a

Fig. 18 Size dependence of fracture toughness

els

lly of ick- ic ure lute ze ial’s xt. gh- ma- ing

ure in.

on-

c- heir th e is ly. ral r, rom een

atio

of

peci- lied,

end, the following materials are chosen: four 4340 ste distinguished by tempering temperatures~T1–T4!, three alu- minum alloys, and five Ti-6A1-4V titanium alloys separat by processing~P1–P5!. Details of the different tempering temperatures and processing may be found in Hoysan Sinclair @118#. Values ofK Ic are included only if contributors claimed them to be in compliance with ASTM E399. Resu are summarized in Table 4.

Results in Table 4 are drawn from over 50 different lab ratories. When these testing laboratories repeated tests o same material made with the same sized specimens o same kind, just the average value is kept in the survey. the other hand, if one laboratory tested different materials the same material but with different types of specimens, t companion fracture toughness results from that labora are viewed as being independent and included as mul values. The total number of independent measurement K Ic for a given material is designated the sample size Table 4. Also in Table 4, variability for a single material represented as previously~ie, 6100 (1.96sK Ic

/K̄ Ic)). Evident in Table 4 is that there is considerable variation

fracture toughness values. On average over all materials sidered, the variability is647%, or a factor of 2.8 betwee the lowest value and the highest. By way of comparison, yield strength of the same materials varies on average 611%, or a factor of 1.2 between the lowest and the high

Clearly, one needs to exercise care in obtaining a re sentative toughness for a particular material and using i predict fracture in an attempt to guard against this even the material is one in Table 4, then taking the low end of spread should probably furnish a conservative estimate~cor- responding actual numerical values can be found in@118#!. If the material is not one in Table 4, and no information readily available as to its distribution ofK Ic , then dividing an isolated value by a factor of three should probably furn a conservative estimate.

The foregoing raises the issue of the identity of the sou of the substantial variations in fracture toughness reporte Table 4, especially since these variations are significa larger than the scatter indicated in Table 3. Current prac for measuring fracture toughness applies predomina bending loads to cracks in test pieces; for example, th point-bend, compact, disk-shaped compact, and arc-sh

d

and

lts

o- f the the On , or en

ory iple s of in

s

in on-

the by st. re- to

. If he

is

ish

rce in

tly tice tly

ee- ped

specimens.19 Furthermore, test pieces are quite geometrica similar, with cracks penetrating about half the widths specimens and being about equal in extent to their th nesses~see @111#!. Therefore, loading type and geometr proportions can be expected to cause little of the fract toughness variations reported in Table 4. In contrast, abso size is not completely dictated in ASTM E399. Thus si effects are possible sources of discrepancy in a mater fracture toughness. We consider this possibility further ne

To assess whether size has any effect on fracture tou ness, we need results from tests performed on a single terial using a single type of specimen with size alone be altered. Sinclair and Chambers@119# collects data of this genre—specifics of the restrictions enforced to try to ens testing varied solely as to size scale are described there

Focusing on plane-strain brittle response~yield region, if any, of extent less than about 2% of the crack length!, Fig. 18 presents results from 43 different papers which together c tain over 800 distinct tests. In the figure,K Ic

a is the fracture toughness determined via the smaller specimen,K Ic

b via the larger, with G being the scale factor between the two~the inserted compact specimens with aG'2 are merely intended to be illustrative!. While designated asK Ic to reflect being in the plane-strain brittle regime, only about 30% of the fra ture toughness values used in Fig. 4 are claimed by t contributors to be valid in the sense of complying wi ASTM E399. For some, being nonmetals, such complianc not appropriate. For others, it is hard to tell complete However, all were checked for compliance with the cent restriction of ASTM E399, namely limited yielding. Furthe separating those claimed as in accord with the standard f the remainder did not reveal any major differences betw the two sets of results in terms of the ratiosK Ic

a/K Ic b for

different G. For fracture toughness to be size independent, the r

K Ic a/K Ic

b should be unity for all scale factorsG ~the solid line in Fig. 18!. Evident in Fig. 18 are clear demonstrations

19While the compact specimen was formerly referred to as the compact tension s men, this designation was simply to reflect the means by which loading is app namely by pulling. The peak nominal stress ahead of the crack is largely ('90%) due to bending rather than tension.

t

l

o

h

h

e

ens- re

gth. of

hat ata. That is ed on a rior act f a ical real- by ant

th n to inal

ess inal

ize ra- gly day. the ar

es and ral the and stic

on s-

tan-

- e- am- tic

e

278 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

fracture toughnessbeing size dependent. Given that the range of variation ofK Ic

a/K Ic b spans more than a factor o

four, such size dependence could be a source of at least s of the variability found in fracture toughness.

One should not be that surprised that, even when frac is appropriately brittle in nature, the fracture toughness o given material can vary by about a factor of three, qu possibly due to fracture toughness being size depend Fracture toughness’ role as a material constant control brittle fracture is the outcome of the hypothesis~or hypoth- eses! that led to theK interpretation for singularities. Being hypothetical, this role may or may not be fulfilled in practic The physical evidence, in fact, shows that the hypothe underlying LEFM are not complied with, or at least, n closely so.

Nonetheless, LEFM can take credit for predicting tren in the fracture of cracked components. By way of examp we reconsider the data in Fig. 18. For the specimens volved in generating this data, the stress intensity factors in general be expressed by

K15s0Apa f~a/w! (3.29)

wheres0 is an applied stress,a continues as crack length and f (a/w) is a function of geometry withw being some other dimension of the specimen, such as the overall widt the crack plane. At fracture, therefore, for two complete scaled specimens,

K Ic a5s f

aApa f~a/w!, K Ic b5s f

bApGa f~Ga/Gw! (3.30)

wherein the subscriptf is now put on the applied stresses denote values at fracture. If fracture toughness is size in pendent, Eqs.~3.30! have

K Ic a/K Ic

b51, s f a/s f

b5AG (3.31)

That is, the strength or stress at fracture decreases with creasing size. If, on the other hand, the strength is size in pendent, Eqs.~3.30! have

s f b/s f

b51, K Ic a/K Ic

b51/AG (3.32)

The dashed line in Fig. 18 plotsK Ic 1 /K Ic

2 of Eqs.~3.32!. Evi- dent in Fig. 18 is that nearly all the data lie above this das line. This means that nearly all the data comply with t trendpredicted by fracture mechanics of decreasing stren of cracked components with increasing size. Unfortunat all the data do not agree well with theprecise reduction predicted by fracture mechanics, that of Eqs.~3.31!. Typi- cally, this is the case for other predictions of fracture m chanics: Qualitatively they are correct, yet quantitative there is room for considerable improvement.20

3.5 How well do other ‘‘interpretations’’ work?

Given the potential for greater predictive capability, it natural at this point to seek alternatives toK interpretations

t

f ome

ure f a ite ent. ing

e. ses t

ds le, in- can

,

at ly

to de-

in- de-

ed he gth ly,

e- ly

is

of singular stresses. To this end, one could consider disp ing with any 2D or 3D stress analysis and predicting failu by merely comparingnominal net-section stressahead of singularities with some suitable measure of material stren Certainly this is a procedure which appeals in its ease implementation. However, what Fig. 18 demonstrates is t such an approach does not agree well with the physical d This is because it predicts strength size independence. is, it predicts Eqs.~3.32! and the dashed line in Fig. 18. Th prediction is almost an outlier of the physically measur responses. In contrast, at least fracture predictions based K interpretation capture the trend in the data. This supe predictive performance can be attributed, in part, to the f that at least aK interpretation recognizes the presence o stress concentration, albeit with a somewhat nonphys measure. Nominal net-section stress does not. Since, in ity, fracture can be expected to be significantly influenced stress concentrations, this recognition realizes a signific real advantage forK interpretations over ones made wi nominal net-section stress. That said, there is no reaso preclude the use of a fracture criterion based on nom net-section stress in an adjunct role. Indeed, ifK ever failed to predict brittle fracture when the nominal net-section str exceeded the ultimate stress of a brittle material, a nom net-section stress criterion should be enforced.

What about other alternatives that do attempt to recogn the influence of stress concentrations in singular configu tions? Over the years, a number of these have been wittin or otherwise suggested, and they continue to be used to All, in essence, draw on field quantities near but not at singularity of concern to infer failure right at the singul point. As a result they may be termednearby fracture crite- ria.

In implementing nearby fracture criteria, two choic need to be made: what to monitor as governing fracture, where to monitor it. With respect to the first option, seve possibilities have been entertained in the literature over years. Among these are measures reflecting stresses strains at the singularity. Stresses are usually used in ela analyses and typically in complex configurations~eg, for failure in composites as in Chamis@121#; and for biomedical applications as in Valliappan, Kjellberg, and Svenss @122#!. Strains are normally preferred when significant pla tic flow accompanies fracture~eg, Belie and Reddy@123#, Kim and Hsu@124#, and Chen@125#!. Other quantities em- ployed are the crack opening displacement of Wells@126# and the crack opening angle of Andersson@127#, the former having gained sufficient acceptance to have a British S dard@128# and an ASTM Test Procedure@129# to govern its measurement.21 While the majority of these quantities se lected in the role of governing fracture are for plastic r sponse, given that elastic precedes plastic it is fair to ex ine them all with respect to performance in the elas regime.

All of the nearby fracture criteria concomitant with th

20The focus here of the assessment ofK interpretations is on monotonic loading rathe than cyclic. This is because this is the simpler situation and accordingly where could expectK interpretations to perform best. Some evidence that this is in fac may be found in Sinclair and Pieri@120#.

r one so

21Crack opening displacement was independently introduced in Cottrell@130# to effect a somewhat different objective, that of classifying brittle fracture—see Burdekin@131# for a review of its role in fracture mechanics.

t s y c

a c

n a t

u

s s f

d u t r c i

r

e

o

i- is-

ss, ing

s can

or c- in

igns nd

aid an —

ss n

bal or

is At a

ar tic

ing ity. ari- first

ms a- of

ms ri-

to ns.

e

ons lo- rm

ile

as ex-

the

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 279

above measures must make their comparisons at loca which are removedfrom the actual singular point. This i because the elastic stresses and strains at a singularit infinite while, for the case of the tensile crack, the displa ment and opening angle at the tip are fixed independen load level~being zero andp, respectively!.

Indeed, this quality of being removed, even if only by short distance, is a principal motivation for selecting t foregoing quantities in the first place, namely having a qu tity which is responsive to loading yet ‘‘measurable.’’ Su measurement can be effected either via FEA or some o analytical method for nearby stresses and strains, or eve direct physical means for crack opening displacement angle. Nevertheless, the decision to withdraw from the ac singular point of concern should not imply a retreat from t original objective of predicting failure at the singular poin It is therefore a logic requirement of proposers/users nearby fracture criteria to clearly identify what is their co responding fracture criterionat the singularity. Despite infer- ences in various papers to the contrary, the complexity of configuration being considered does not obviate one fr this responsibility.

Turning to a consideration of possible companion fract criteria at the singularity, it is almost embarrassingly obvio to make the following comments at the outset, but neces nonetheless given suggestions made in the literature. It i exercise in futility to attempt to use nearby quantities to in elastic stresses and strains at the singularity because the unbounded there. Thus their true infinite values are use for comparison with any corresponding finite limiting one The fact thatestimatesof stress and strain at a singularity ca be finite reflects the limitations of the procedure used to the estimation, not physics. One simply cannot rely up errors in analysis to make a nonphysical field in a mo physically appropriate. Ultimately, with a sufficiently acc rate analysis, a large enough value of stress or strain a singular point must result so as to exceed any finite co sponding limit imposed, irrespective of load level. Su comparisons are therefore meaningless. Moreover, the s tion is not improved in any real sense by introducing plas flow ~recall the discussion in Section 2.1!.

Given the need for a bounded quantity at the singular and one which reflects load levels, there would not seem be any significantly different alternative toK. In fact, K is the explicit choice made in the elastic regime by crite based upon crack opening displacement or angle. It follo that nearby fracture criteria cannot be expected to realize real improvement in predictive capability over that offer by K interpretations.

Indeed, nearby fracture criteria can typically be expec to be even less reliable. The reason for this is that attemp to determineK from nearby quantities can be, in itself, a unreliable undertaking. A demonstration of this possibility given in Sinclair @132#. Therein, two artificial applications are constructed which each have known closed-form s tions. Thereafter, nearby stress, strain, and crack opening placement and angle are used to inferK, with all specifics of corresponding estimation procedures being set a priori.

ions

are e- t of

a he n- h

ther by

nd ual he t. of

r-

the om

re us ary an

er y are less s. n do on el - the re- h tua- tic

ity, to

ia ws any d

ted ting n is

lu- dis-

All

four estimateKÞ0 whenK50 in the first application, and all four find K50 whenKÞ0 in the second. Such dramat cally erroneous determinations can be attributed to the m chievous intent of the author of this article. Neverthele they do indicate the potential of further discrepancies be introduced by the use of nearby fracture criteria.

At this time then,K interpretations of elastic singularitie represent the best available. Properly implemented, they provide qualitative predictions to guide in designing f structural integrity. Quantitatively, they may provide satisfa tory predictions, but they cannot be relied upon to do so general. Hence, ultimately, one can expect that most des based onK interpretations are going to require specific a rigorous testing.

4 ANALYZING STRESS SINGULARITIES

4.1 Asymptotic identification: Classical analysis

Asymptotic characterization of elastic singularities can the stress analyst in two ways. In the first instance, it c alert the analyst to the possibility of singular stresses ‘‘possibility’’ because whether or not local singular stre fields in fact participate in a particular global configuratio usually depends on the actual far-field loading in that glo configuration. If this possibility is realized, it is essential f it to be appreciated if any useful information whatsoever to be gleaned from such a physically limited model. present, the best use of such a stress analysis is viaK interpretation. This requires a definition of an appropriateK, which in turn requires the identification of the local singul field present. This is the second way in which asympto characterization can be of assistance.

Several methods are available for analytically undertak the asymptotic analysis of stress singularities in elastic One is the use of potentials together with separation of v ables. This approach appeals in its directness. It was used in Knein@28# to identify the singularity in a single elastic configuration. Since, it has seen use in Willia @20,133# and Kitover@134# to establish the eigenvalue equ tions governing singularity exponents for a wide range configurations. These equations are solved in Willia @20,133# so as to explicitly identify possible stress singula ties.

Alternatively, complex variables may be introduced yield an approach which is compact in its representatio This method of analysis was first employed in Huth@135# to treat the same class of problems as in Williams@20#, then shown in Williams@136# to lead to the same results as th earlier separation of variables.

These two methods were applied to elastic configurati having locally homogeneous boundary conditions. When cal boundary conditions are inhomogeneous, transfo methods are natural to consider. Brahtz@27# does this for an angular plate comprised of a single elastic material, wh Bogy @137#, following Tranter@138#, uses the Mellin trans- form for a bimaterial plate, and the Mellin transform h seen extensive use since. It is, though, quite possible to plore the effects of inhomogeneous conditions using

e

n

s - h

h

i

ir-

ing

dis-

r -

to nt

280 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

original separation-of-variables method or complex va ables. Essentially, all three methods enable the same id fication of possible stress singularities to be made. T choice of which one to use is really a matter of perso preference for the analyst undertaking the asymptotics: choose separation of variables here and describe its b elements next.

By way of a simple illustration, we reconsider the elas plate with a stress-free reentrant corner, or notch, and jected to tensile loading~Fig. 16!. The symmetry of the con figuration enables attention to be confined to the upper of the plate which, in terms of the polar coordinatesr andu of Fig. 16, is contained in 0,u,f/2. Heref is the entire angle subtended at the corner within the plate (0,f <2p). What we seek is the local character of the stres s r , su , andt ru , and displacementsur anduu in the corner of the plate~viz, as r→0). These fields are to satisfy th equations of elasticity together with the following tradition boundary conditions: the stress-free conditions on the p edge,

su5t ru50 on u5f/2 (4.1)

for 0,r ,`; and the symmetry conditions ahead of t notch,

uu50, t ru50, on u50 (4.2)

for 0,r ,`. To construct appropriate forms for the solutions to t

field equations of elasticity for complying with the cond tions Eqs.~4.1! and ~4.2!, we follow Williams @20# and let the stresses be generated by an Airy stress functionx in accordance with

s r5 1

r

]x

]r 1

1

r 2

]2x

]u2

su5 ]2x

]r 2 , t ru52 ]

]r S 1

r

]x

]u D . (4.3)

Such stresses satisfy the equilibrium requirements: Prov x is biharmonic, they are also compatible so that compan displacements exist. These may be determined with the of an auxiliary harmonic function which is given in Sectio 2.45, Coker and Filon@139# ~see also Williams@20#!.

The determination of a biharmonicx for Eqs.~4.3! can be further reduced to the determination of two harmonic fun tions C andĈ, sincex admits to being represented by

x5C1r 2Ĉ (4.4)

Separation of variables directly furnishes a candidateC as

C5r l11@c1 cos~l11!u1c2 sin~l11!u# (4.5)

whereinl, c1 , andc2 are constants~the choice ofl11 as an exponent rather than justl follows Williams @20#, and results in somewhat simpler equations forl later!. In select- ing Ĉ, it is essential that the resultingx involve a single power ofr . This is so that each of the boundary conditions Eqs. ~4.1! and ~4.2! holding for 0,r ,` leads to but one condition on the constants inC and Ĉ. Then we have a

ri- nti- he al

We asic

tic ub-

alf

ses

e al late

e

he i-

ded ion aid n

c-

in

chance of complying with these boundary conditions by v tue of having four constants—two each forC and Ĉ—for four equations. Accordingly we take

Ĉ5r l21@c3 cos~l21!u1c4 sin~l21!u# (4.6)

wherec3 andc4 are the further constants. Hence, substitut Eqs. ~4.5! and ~4.6! into Eq. ~4.4! realizes ax with four constants sharing a commonr l11 multiplier. Processing this Airy stress function, using Eqs.~4.3! and the Coker and Filon relations, furnishes corresponding stresses and placements. Thus we have, as ourbasic separable fields,

s r52lr l21@c1 cos~l11!u1c2 sin~l11!u

1~l23!~c3 cos~l21!u1c4 sin~l21!u!#

su5lr l21@c1 cos~l11!u1c2 sin~l11!u

1~l11!~c3 cos~l21!u1c4 sin~l21!u!#

t ru5lr l21@c1 sin~l11!u2c2 cos~l11!u

1~l21!~c3 sin~l21!u2c4 cos~l21!u!# (4.7)

ur5 2r l

2m @c1 cos~l11!u1c2 sin~l11!u

1~l2k!~c3 cos~l21!u1c4 sin~l21!u!#

uu5 r l

2m @c1 sin~l11!u2c2 cos~l11!u

1~l1k!~c3 sin~l21!u2c4 cos~l21!u!#

In Eqs.~4.7!, c1 andc2 of Eq. ~4.5! have been exchanged fo c1 /(l11) andc2 /(l11) so as to slightly simplify expres sions, andm continues as the shear modulus,k as the func- tion of Poisson’s ratio given in Eq.~1.3! et seq.

Now applying the symmetry conditions Eqs.~4.2! gives

c25c450 (4.8)

Applying the outstanding stress-free conditions Eqs.~4.1! to the remaining terms then yields the 232 system of equations

r l21Ac50

A5S l cos~l11!f/2 l~l11!cos~l21!f/2

l sin~l11!f/2 l~l21!sin~l21!f/2D c5S c1

c3 D (4.9)

for 0,r ,`. For this homogeneous system of equations have a nontrivial solution, the determinant of the coefficie matrix must be zero. That is

D50 (4.10)

whereD is the determinant ofA. Hence we obtain theeigen- value equationfor our example as

l2~sinlf1l sinf!50 (4.11)

o

R

d

i

I

i

e t- ly etric m-

ex oci-

n

-

m-

of

oots.

on- ive rack

ary ne er- me ary e ext.

r- ty,

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 281

The eigenvalue Eq.~4.11! is for the eigenvalues, or charac teristic values, of the boundary value problem described Eqs.~4.1!, ~4.2!, and surrounding text.

Given ther -dependence of the stresses in Eqs.~4.7!, if we focus on integrable singular stresseswe can confine our search for roots of Eq.~4.11! to the range

0,l<1 (4.12)

Within this range, a root for each of two special cases immediate:

l51/2 for f52p

l51 for f5p (4.13)

The first of these gives the familiar inverse-square-root s gularity of a tensile crack or a flat lubricated rigid punch a half-space~see Table 2, pointsP2 and P3). For p,f ,2p, there is one real root forl satisfying Eq.~4.11!: For 0,f,p, there are no real roots. Actual values ofl for f in the first of these ranges need to be found numerically. sults so obtained can be fairly readily fitted to within 0.5 by

l50.512.425f̂316.3f̂5

f̂512 f

2p , p<f<2p (4.14)

The fit of ~4.14! recovers thel’s of Eqs.~4.13! and connects the two for otherf.

For anyf.p, the associated singulareigenfunctioncan be assembled as follows. First, substitute the correspon eigenvalue,l from Eqs.~4.14!, into the fields of Eqs.~4.7!. Next, setc25c450 therein in accordance with Eqs.~4.8!. Last, determine the relationship betweenc1 andc3 from Eqs. ~4.1! whenl equals the eigenvalue, and substitute this re tionship into the fields. By way of example, for the spec case of a crack (f5p), these steps givel51/2 and c3

52c1 . Then, on exchangingc1 for K I/2A2p so as to recover the stress intensity factor, the stresses from Eqs.~4.7! are:

H s r

su J 5

K I

4A2pr F H5

3J cos u

2 H 2

1J cos 3u

2 G

t ru5 K I

4A2pr Fsin

u

2 1sin

3u

2 G (4.15)

Displacements follow similarly from Eqs.~4.7!. The stress field of Eqs.~4.15! is one form of the now classical, Mode singular eigenfunction for crack-tip stresses, originally ide tified in Williams @140# and Irwin @84#.

Another form of stress singularity can be directly iden fied via the same approach. This type of singularity ste from complex roots to the eigenvalue equation, a possib appreciated in Williams@20,133#, and further amplified in

- in

is

in- n

e- %

ing

la- al

, n-

ti- ms lity

Williams @21#. For our illustrative example, the eigenvalu Eq. ~4.11! is simply comprised of algebraic and trigonome ric functions ofl. For other configurations, this essential remains the case. Thus, since these functions are symm in the complex domain, complex eigenvalues occur as co plex conjugates. That is, as

l5j6 ih (4.16)

with 0,j<1 as the counterpart of Eq.~4.12!. For such com- plex eigenvalues, the definition of equality in the compl domain assures that the real and imaginary parts of ass ated eigenfunctionsare each individually eigenfunctions i themselves. For example, thec1 contribution tos r of Eqs. ~4.7!, whenl is as in Eq.~4.16!, becomes the two expres sions

s r52c1r j21@cos~j11!u coshhu~j cos~h ln r !

2h sin~h ln r !!1sin~j11!u sinhhu~j sin~h ln r !

1h cos~h ln r !!#

s r52c18r j21@cos~j11!u coshhu~j sin~h ln r !

1h cos~h ln r !!2sin~j11!u sinhhu~j cos~h ln r !

2h sin~h ln r !!# (4.17)

wherec18 is generally a distinct constant fromc1 . Evident in Eqs. ~4.17! is the oscillatory nature that accompanies co plex eigenvalues.

Returning to our illustrative example, the determination eigenvalues of the form Eq.~4.16! proceeds routinely on separating the eigenvalue Eq.~4.11! into real and imaginary parts. Forp,f,2p, given 0,j<1, graphical arguments can then be used to establish that there are no complex r There are complex eigenvalues whenj.1, though these do not give rise to singular stresses. There do exist other c figurations, though, for which complex eigenvalues do g rise to singular stresses. Examples include the interface c and the adhering rigid indentor~as for pointsP2 and P3 in Table 2 and Fig. 2a andb!.

The foregoing analysis can be applied to other bound conditions for in-plane loading, as well as to out-of-pla shear, bending within classical theory, and to composite v sions of all of these configurations. However, there are so further stress singularities and different types of bound conditions for which it is not immediately applicable. W look to its adaptation to accommodate these situations n

4.2 Asymptotic identification: Further developments

An additional form of stress singularity results from ente taining logarithmic character. To investigate this possibili we need to augment the fields of Eqs.~4.7! with ones con- taining lnr. To this end, observe that Eqs.~4.7! satisfy the plane field equations of elasticity for anyl. In fact, these fields are continuously differentiable functions ofl. Hence, because theirr -dependence is of the formr l215e(l21)ln r,

n n

-

a

nts

q. e ing

n-

ure -

rity ang

nd

r ial

t s. . It

re-

lar e

to ad re-

rith- is

282 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

they can be differentiated with respect tol to generate the sought-after fields containing lnr. This is the approach adopted in Dempsey and Sinclair@29#. For the basic fields of Eqs.~4.7! it leads to, as ourauxiliary fields,

s r52r l21@~11l ln r !~c1 cos~l11!u1c2 sin~l11!u!

1~2l231l~l23!ln r !~c3 cos~l21!u

1c4 sin~l21!u!2lu~c1 sin~l11!u2c2 cos~l11!u

1~l23!~c3 sin~l21!u2c4 cos~l21!u!!#

su5r l21@~11l ln r !~c1 cos~l11!u1c2 sin~l11!u!

1~2l111l~l11!ln r !~c3 cos~l21!u

1c4 sin~l21!u!2lu~c1 sin~l11!u2c2 cos~l11!u

1~l11!~c3 sin~l21!u2c4 cos~l21!u!!#

t ru5r l21@~11l ln r !~c1 sin~l11!u2c2 cos~l11!u!

1~2l211l~l21!ln r !~c3 sin~l21!u

2c4 cos~l21!u!1lu~c1 cos~l11!u1c2 sin~l11!

3u1~l21!~c3 cos~l21!u1c4 sin~l21!u!!#

(4.18)

ur5 2r l

2m @~c1 cos~l11!u1c2 sin~l11!u!ln r

1~11~l2k!ln r !~c3 cos~l21!u1c4 sin~l21!u!

2u~c1 sin~l11!u2c2 cos~l11!u

1~l2k!~c3 sin~l21!u2c4 cos~l21!u!!#

uu5 r l

2m @~c1 sin~l11!u2c2 cos~l11!u!ln r

1~11~l1k!ln r !~c3 sin~l21!u2c4 cos~l21!u!

1u~c1 cos~l11!u1c2 sin~l11!u

1~l1k!~c3 cos~l21!u1c4 sin~l21!u!!#

Together, the stresses and displacements of Eqs.~4.18! con- tinue to satisfy the field equations of elasticity because th equations are independent ofl ~that such is the case may b verified by direct substitution!. What now becomes appare is that the stresses of Eqs.~4.18! can also be singular whe l51, the upper limit admitted in Eqs.~4.12!, since then they can go to infinity as lnr whenr→0. Forl.1, though, they remain bounded whenr→0.22

The fields found vial-differentiation can be supple mented by the originating fields of Eqs.~4.7!. When com- bined in this way, the constants need no longer be the s so we now distinguish the constants in Eqs.~4.7! asc8. In- troducing the combination into the boundary conditions our example, the earlier system Eq.~4.9! now involves dif- ferent functions ofr , with

h

ese e t

me

in

r l21 ln rAc1r l21F]A

]l c1Ac8G50 (4.19)

for 0,r ,`. In Eq. ~4.19!, ]A/]l is the matrix with ele- ments obtained by differentiating all corresponding eleme of A of Eqs. ~4.9! with respect tol. For our example, Eq. ~4.19! is a 232 system. In general, it can be annA3nA

system,nA being the order ofA. In either case, the first term in Eq. ~4.19! recovers our original determinant condition, E ~4.10!, for a nontrivial c. The second term requires som analysis to establish the necessary conditions for maintain a nontrivial c—essentially these conditions result from e suring a consistent or solvable system forc8.

Under Eq.~4.10!, D50 and the rank ofA must be less than its ordernA . If the rank ofA is nA21, then necessary conditions for a nontrivialc are

D5 ]D

]l 50 (4.20)

That is, the eigenvalue is a repeated root. Equations~4.20! are effectively the conditions that are noted to hold for a p logarithmic singularity (l51) with inhomogeneous bound ary conditions in Bogy@141#: sansD50, Eqs.~4.20! for l 51 are stated as the condition for a pure log singula under homogeneous boundary conditions in Bogy and W @142#. Equations~4.20! are shown to be necessary forcÞ0 for 0,l<1 under homogeneous boundary conditions, a when the rankr A5nA21, in Dempsey and Sinclair@29#.

If, instead, the rankr A is further reduced tonA22, nA

23,.., the conditions in Eqs.~4.20! are not enough. Unde these circumstances, necessary conditions for a nontrivc are

D5 ]D

]l 5

]2D

]l2 5••

]nA2r AD

]lnA2r A 50 (4.21)

This result is established in Appendix 1,@29#, for either pure logarithmic singularities (l51), or logarithmic intensifica- tion of power singularities (l,1, see Eqs.~4.18!!: It in- cludes the previous result Eqs.~4.20!. We note, however, tha cÞ0 and the existence of local fields of the form of Eq ~4.18! does not necessarily mean local logarithmic terms is possible for a nontrivialc to exist yet the coefficient of lnr terms be zero. This occurs forl51,c15c25c350, andc4

Þ0—see Eqs.~4.18!. Returning to our illustrative example, there are no

peated roots to Eq.~4.11! within the range of Eq.~4.12!. Therefore, there is no logarithmic character in the singu stresses because Eqs.~4.20! are necessary condition for th same. On the other hand, if instead of confining attention symmetric loading in our reentrant corner example we h admitted antisymmetric as well, we would have had a peated root for the case of a crack~viz, for f52p and l 51/2 with a multiplicity of 2!. Accordingly Eqs. ~4.20! would have been satisfied. Nonetheless, there is no loga mic intensification of the singularity in this instance. This because the rank of the now 434 coefficient matrix drops to two for l51/2, and Eqs.~4.20! are not sufficient under suc circumstances. Furthermore, Eqs.~4.21! are not satisfied so

ry

22One might be inclined to try to employ the classical Michell solution for auxilia fields since it does contain log terms. However, in its usual form~eg, Art 43, Timosh- enko and Goodier@16#!, this solution does not contain all the terms in Eqs.~4.18!.

-

l

s

w - s

n h a n o

qs. ar

he

- this

s at the cti-

to t-

ld,

n

em

e . s e ith

q.

r

.

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 283

that no logarithmic intensification results. There are, thou other instances in which Eqs.~4.21! are satisfied and loga rithmic singular character is possible: Quite a number these are identified in Part II of this review.

In summary, then, thesingular stresses that are possib with homogeneous boundary conditionsfollow as below, in what might reasonably be regarded as order of decrea singular character. For any stress components, asr→0,

s5O~r j21 cos~h ln r !!1O~r j21 sin~h ln r !!

when D50 for complex l5j1 ih~0,j,1!

s5O~r l21 ln r !1O~r l21! when D50, ]nD

]ln 50

for n51,..,nA2r A and real l~0,l,1!

s5O~r l21! when D50 for real l~0,l,1! (4.22)

s5O~ ln r ! when D50, ]nD

]ln 50,

for n51,..,nA2r A and l51,

with c1 21c2

21c3 2Þ0

in the auxiliary stress field of Eqs.~4.18!

s5O ~cos~h ln r !!1O~sin~h ln r !!

when D50 for complex l511 ih

Herein,D is the determinant of the coefficient matrixA re- sulting from applying boundary conditions,nA is the order of this matrix, andr A its rank whenl is an eigenvalue. For the single material plate in extension, at mostnA54. For bima- terial plates, both boundary and interface conditions are volved in assemblingA: Then typicallynA58. And so on.

In the last of Eqs.~4.22! we have included, as a type o ‘‘singularity,’’ stresses which in fact are bounded forr 50. These same stresses, though, are undefined forr→0. Conse- quently, to a degree, they share with actual singular stre the futility of trying to use them directly in stress-streng comparisons atr 50.

In addition to the types of singularity in Eqs.~4.22!, it is theoretically possible to have a combination of the first t types in Eqs.~4.22! whenl is a complex root of the appro priate multiplicity,nA2r A . The actual occurrence of this la sort of singularity with homogeneous boundary conditions yet to be noted in the literature. It is also theoretically po sible to have log-squared singularities. Again, the actual currence of this last sort of singularity with homogeneo boundary conditions is yet to be noted in the literature. It c occur, however, with inhomogeneous boundary condition

For inhomogeneous boundary conditions, any respo can include that for the corresponding homogeneous co tions. Further, stress singularities typically stem from the mogeneous boundary conditions, especially if we require applied inhomogeneous conditions to be sufficiently conti ous. Even so, for some inhomogeneous boundary conditi logarithmic singularities can be induced. We illustrate how treat this sort of response by reconsidering our symme notch example.

gh,

of

e

sing

in-

f

ses th

o

t is s- oc- us an s. nse di- o- ny u- ns, to tric

Suppose in this example, the stress-free conditions of E ~4.1! are replaced with conditions applying a uniform she tractionq. That is, with

su50, t ru5q, on u5f/2 (4.23)

for 0,r ,`. Now introducing into Eqs.~4.23! the symmet- ric part of the basic stress field, Eqs.~4.7! with Eqs. ~4.8!, yields the 232 system

r l21 Ac5q (4.24)

for 0,r ,`. Here A and c are as in Eqs.~4.9!, and the vectorq is given byq5(0,q). For Eqs.~4.24! to hold for all r , we set l51. Then solving forc yields c15q cscf and c3

52(q/2)cotf. Hence, for example, the shear stress is

t ru5q sin 2u

sinf (4.25)

for 0<u<f/2. Clearlyt ru of Eqs.~4.25! complies with the shear boundary condition in Eqs.~4.23!. However, what is also clear is that there is a problem with the solution if t vertex angle is such that sinf50. That is, if f5p, 2p. Thent ru is everywhere infinitethroughout the plate. Further more, the other stresses and even the displacements in solution are everywhere infinite. This sort of ‘‘singularity’’ i no longer trying to reflect a physical stress concentration the plate vertex. Rather, it represents a total breakdown in solution procedure adopted, something which must be re fied before any physical interpretation is attempted.

The reason for the breakdown is that the fields used arrive at Eqs.~4.25! are incomplete. To overcome this shor coming we follow Dempsey@143# and supplement them with those of Eqs.~4.18! with Eqs.~4.8! applied. If we continue to usec8 to distinguish the constants in the original stress fie our system for solution becomes

r l21 ln r Ac1r l21S ]A

]l c1Ac8D5q (4.26)

for 0,r ,`. We setl51 again so that the second term o the left-hand side of Eq.~4.26! becomes independent ofr like the right-hand side. Now, though, we still have a syst which depends onr by virtue of the lnr term. The vector coefficient of this lnr term must therefore be zero. For th problem vertex angles,f5p and 2p, this can be arranged This is because the determinant ofA is zero for these angle whenl51 ~see Eqs.~4.9!!: Indeed, in some sense it is th determinant ofA being zero that causes the problem w these angles in the first place by prohibiting a solution to E ~4.24!. Consequently, we merely need to makec a solution of Ac50 for D50. Then it in concert withc8 enables a solution of Eq. ~4.26!. For example, forf5p, a solution isc1

52c3524c38522q/p,c1850. The corresponding shea stress becomes

t ru5 22q

p @~11 ln r !sin 2u1u cos 2u# (4.27)

for 0<u<p/2. Clearlyt ru of Eq. ~4.27! complies with the shear boundary condition in Eqs.~4.23!. Clearly, also,t ru of Eq. ~4.27! is logarithmically singular at the plate vertex

- c n

l

i

t r m

.

y, qs.

ble. ess

s- d

r.

nd-

284 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Given symmetry for our plate of vertex anglep, it realizes a half-space with a jump in surface shear traction from2q to q. Accordingly, the log singularity present is akin to that P8 of Table 2 and Fig. 2f. This sort of singularity is admis sible in elasticity and does reflect the physical stress con tration that occurs at a step discontinuity in shear tractio

While the analysis leading to Eq.~4.27! and like expres- sions for the other stresses does solve the plate loaded uniform shear whenf5p, it does not provide a reasonab transition from stresses like that of Eq.~4.25! as f passes throughp. In fact, from Eq.~4.25! it would appear that, for f near but not equal top, t ru can be made arbitrarily large

To furnish a more sensible transition, Ting@144# supple- ments the solution of Eq.~4.25! with its homogeneous coun terpart ~ie, the stresses forq50). This leaves compliance with Eqs.~4.23! unaltered. By suitably adjusting the partic pation of these additional stresses, a reasonable trans from t ru of Eq. ~4.25! through t ru of Eq. ~4.27! can be effected asf passes throughp. Such transitions are obtaine for the other stresses and for further configurations in T @144#. Since they recover results found via Eqs.~4.18!, the approach in Ting@144# can be used just by itself.

Either via Dempsey@143# plus Ting@144#, or just by Ting @144#, a number of configurations that would otherwise ha breakdowns in their analysis can be treated. Typically, leads to logarithmic stress singularities when constant t tions are applied; ‘‘typically’’ because occasionally syste like Eq. ~4.24! with D50 are still consistent because th augmented matrix also drops in rank. Analogous results h for linear displacements.23

Observe that, for these log singularities withinhomoge- neousboundary conditions, the requirements in the last Eqs. ~4.22! do not apply. All that is required isD50 when l51. Indeed, ifD has a repeated root atl51, further fields other than just those of Eqs.~4.18! are typically needed These fields stem from further differentiation with respect l. As noted in Dempsey and Sinclair@29#, this leads to ln2 r terms. For the auxiliary field of Eqs.~4.18!, it gives thefur- ther auxiliary stress field

s r52r l21@~l ln2 r 12 ln r 2lu2!~ ĉ1 cos~l11!u

1 ĉ2 sin~l11!u1~l23!~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!!22u~11l ln r !~ ĉ1 sin~l11!u

2 ĉ2 cos~l11!u1~l23!~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!!12~11l ln r !~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!22lu~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!#

su5r l21@~l ln2 r 12 ln r 2lu2!~ ĉ1 cos~l11!u

1 ĉ2 sin~l11!u1~l11!~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!!22u~11l ln r !~ ĉ1 sin~l11!u

lates rities nder

at

en- .

by a e

.

-

i- ition

d ng

ve his ac- s

e old

of

to

2 ĉ2 cos~l11!u1~l11!~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!!12~11l ln r !~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!22lu~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!# (4.28)

t ru5r l21@~l ln2 r 12 ln r 2lu2!~ ĉ1 sin~l11!u

2 ĉ2 cos~l11!u1~l21!~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!!12u~11l ln r !~ ĉ1 cos~l11!u

1 ĉ2 sin~l11!u1~l21!~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!!12~11l ln r !~ ĉ3 sin~l21!u

2 ĉ4 cos~l21!u!12lu~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!#

whereinĉi and i 51, 2, 3, 4, are further constants. Similarl expressions can be obtained for displacements. With E ~4.28!, log-squared stress singularities may be possi ‘‘May’’ because there are constants for this additional str field which remove all ln2 r terms whenl51, yet do not remove the field in its entirety. These constants areĉ15 ĉ2

5 ĉ350, ĉ4Þ0 in Eqs.~4.28!. In summary, then, thesingular stresses that can be po

sible with uniform tractions/linear displacements applie follow as below, in order of decreasing singular characte24

For any stress components, asr→0,

s5ord~ ln2 r !1ord~ ln r ! when D50, ]nD

]ln 50,

for n51,..,nA2r A, ĉ1 21 ĉ2

21 ĉ3 2Þ0

in the further auxiliary field of~4.28!

s5ord~ ln r ! when D50, ]nD

]ln 50,

for n51,..,nA2r A, ĉ15 ĉ25 ĉ350 (4.29)

in the auxiliary field of Eqs.~4.28!

s5ord~ ln r ! when D50, ]nD

]ln Þ0,

for n5nA2r A, c1 21c2

21c3 2Þ0

in the auxiliary stress field of Eqs.~4.18!

provided throughout~4.29!

l51, r AÞr A8 (4.30)

wherer A8 is the rank of the augmented matrix formed byA and the forcing vector attending the inhomogeneous bou ary conditions. The conditions in Eqs.~4.29! and ~4.30! can be inferred from Dempsey and Sinclair@29# and Dempsey

e Ting

24Some singularities possible with other inhomogeneous boundary conditions for p in extension are discussed in Part II. In large part, these are self evident. Singula can also be induced with other inhomogeneous boundary conditions for plates u bending: These are also discussed in Part II.

23There are other singular configurations wherein supplementary fields like Eqs.~4.18! are needed to make the analysis complete. These involve concentrated loads. Se @145# for a review.

h

f

c i h

o

b

,

r have

lds law

e the

old ive

of

ols lar al- ess s to bly ent me

i e ysis as- in- set not ress at ry not ly-

ty in de- can

um hal- ted cus

r- y ith

e is resent

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 285

@143#. When Eqs.~4.29! and ~4.30!, or their analogues for other configurations, are satisfied, log singularities do res Examples are given in Part II of this review. When the singularities occur, their participation is controlled by t local applied loading rather than far-field conditions. Wh such local loading is nonzero, they must occur. Hence, use of the ord notation in Eqs.~4.29!. As noted earlier, these singularities can occur in concert with the singularities corresponding homogeneous boundary conditions.

There is an additional type of boundary condition whi requires further consideration. These are conditions wh while homogeneous in themselves, promote equations w are not homogeneous in theirr -dependence when the field of Eqs.~4.7! are introduced into them. For example, suppo normal cohesive stresses are applied ahead of the notc our reentrant corner configuration of Fig. 16. Thus,

su5keuu on u50 (4.31)

for 0,r ,` is exchanged for the first of Eqs.~4.2!, whereke

continues as the cohesive law stiffness. Substituting E ~4.7! into Eqs.~4.31! then gives

lr l21@c11~l11! c3#1 ker

l

2m @c21~l1k! c4#50

(4.32)

for 0,r ,`. Since~4.32! holds for all r and now involves two distinct powers ofr , it is effectively two boundary con- ditions. Taken together with the stress-free conditions of E ~4.1! and the zero-shear condition of Eqs.~4.2!, Eq. ~4.32! realizes five equations in the but four unknowns,c1–c4 . These equations cannot be made consistent for anylÞ0. Therefore, no nontrivial solution exists which is simply the form of Eqs.~4.7!.

To overcome this difficulty, we form fields as series replacingl of Eqs. ~4.7! by ln5l1n, with corresponding constants obtained on extendingci ( i 51,2,3,4) toci 1n , then summing onn. This series approach is the one adopted Sinclair @146# to handle heat conduction problems with co vective cooling: It is also the one used in Ting@147# to handle elastic plates with curved boundaries. With series the fields, the cohesive condition of Eq.~4.31! becomes

suu l05l

1 ( n51

`

[suu ln5l1n

2keuuu ] 5 ln5l1n21

0 (4.33)

for 0,r ,`. The order of the terms in Eq.~4.33! is

O~r l21!1 ( n51

`

O~r l1n21!50 as r→0 (4.34)

The lowest order terms in Eq.~4.34! areO(r l21): Setting to zero the determinant of the coefficient matrix for these ter enables one to determine an eigenvaluel, and corresponding eigenfunction constantsci ( i 5124). For this eigenvalue the next terms areO(r l) and serve to relatec5–c8 to c1–c4 , and the terms thereafter relatec9–c12 to c5–c8 , whence c1–c4 , and so forth~see Sinclair@67# for details!. Thus, each complete eigenfunction itself becomes a series~which can be

ult. se e

en the

or

h ch, ich

s se h in

qs.

qs.

f

y

in n-

for

ms

shown to converge forr ,1). There is one such series fo each eigenvalue. Hence, for a series of eigenvalues, we a series of series for eigenfunctions.

The preceding means of constructing asymptotic fie has, as a direct consequence, that the part of the cohesive that is active in determining eigenvalues is the stress~see Eqs. ~4.7!, ~4.33!, and ~4.34!!. That is, cohesive laws hav exactly the same effect on singular character as setting stress contained within them to zero. Similar outcomes h for other boundary conditions which, on a first analysis, g rise to equations that are inhomogeneous inr . An example is the out-of-plane bending of plates within the context sixth-order theory.

The foregoing summarizes some of the analytical to that can be fairly readily applied to asymptotic singu analysis within classical elasticity. As mentioned earlier, ternatives exist. Faced with a specific problem, the str analyst could entertain using any of these approache check for the possibility of singular stresses. It is proba easier, though, to try to draw on the literature for a pertin analysis. To assist in this activity, there already exist so reviews: for cracks, those in Atkinson@148# and Hwang, Yu, and Yang@149#; for some bimaterial plates, that in Murakam @150#. For other configurations, hopefully Part II of th present review can help. In the event that no such anal can readily be found, one could perform the necessary ymptotics oneself. Ultimately, this may be necessary for terpretation. However, it may be more efficient at the out to carry out a global analysis. This is because, while it is likely it is nonetheless possible that any associated st singularities do not actually participate in the problem hand if it involves local homogeneous bounda conditions.25 Under these circumstances, singularities do have to be asymptotically identified, and just a global ana sis suffices. We turn our attention to this activity next.

4.3 Numerical analysis: Detection of singularities

In order to detect the actual presence of a stress singulari a configuration being numerically analyzed, we need to sign a sequence of successively refined analyses which reasonably be relied on to produce diverging maxim stresses, thereby revealing the singularity. The most c lenging singularities to unearth in this way can be expec to be the weakest, namely log singularities. Hence, we fo attention on this type of singularity initially.

To develop a scheme for detecting logarithmic dive gence, we follow Sinclair@151# and consider an analog with the numerical summation of series. For a series w individual memberssn , we form the partial sumSn̂ in accor- dance with

Sn̂5 ( n51

sn (4.35)

25‘‘Not likely’’ because usually the eigenfunctions remaining once the singular on removed have zero stresses at the singular point, and are therefore unable to rep nonzero stresses there.

s p

s r t

i

o

u

n

is- yses xi-

rm sys- n a

ent s are

ted

ic The n

mal , we

ine we t of

os- o- re,

rity ical erse

rant

re, r the al gu-

286 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

For such partial sums, it is sometimes implied in texts numerical analysis that convergence can be examined considering the sequence

Sn̂ ,S2n̂ ,S3n̂ , . . . (4.36)

With this sequence, the series sum is deemed to have verged if successive differences between sums decrea magnitude with the last difference being less than some scribed tolerancee. That is, the convergence criteria for th sequence of Eq.~4.36! are

uS2n̂2Sn̂u.uS3n̂2S2n̂u, uS3n̂2S2n̂u,e (4.37)

Alternatively, a more stringent test for convergence is ba upon a sequence in which the number of members in pa sums is successively doubled. Then the convergence cri become

uS2n̂2Sn̂u.uS4n̂2S2n̂u, uS4n̂2S2n̂u,e (4.38)

Now consider the application of the foregoing criteria the particular instance of summing a harmonic progress namely,

Sn̂5 ( n51

n̂ 1

n (4.39)

Using an area estimate for Eq.~4.39! gives

Sn̂'E 1/2

n̂11/2 dn

n 5 ln~2n̂11! (4.40)

While Eq. ~4.40! is just an approximation, it does disclos the logarithmic divergence of the sum in Eq.~4.39!. Conse- quently, Eq.~4.39! represents a good series to test the c vergence criteria of Eq.~4.37! and Eqs.~4.38! to see if they can detect divergence.

Using Eq.~4.40! and the first sequence of partial sums Eq. ~4.36!, we have, forn̂ large,

Sn̂' ln n̂1 ln 2

S2n̂' ln n̂12 ln 2, S2n̂2Sn̂' ln 22 ln 150.69 (4.41)

S3n̂' ln n̂1 ln 21 ln 3, S3n̂2S2n̂' ln 32 ln 250.41

Continuing, successive differences equal the difference tween the natural logarithms of two successive integers. A result, the convergence criteria of Eqs.~4.37! can be met, since differences are decreasing in magnitude and event can be made smaller than any prescribede. The first conver- gence criteria of Eqs.~4.37! thereforefail to detect that the series is diverging.

On the other hand, the sequence with doubling has

S2n̂' ln n̂12 ln 2, S2n̂2Sn̂' ln 2

S4n̂' ln n̂13 ln 2, S4n̂2S2n̂' ln 2 (4.42)

Thus, the first convergence criterion of Eqs.~4.38! is not complied with, revealing the lack of convergence of th logarithmically divergent series. The second converge criteria of Eqs.~4.38! thereforepasswith respect to diver- gence detection. This suggests adopting the analogue of ~4.38! when undertaking numerical stress analysis.

on by

con- e in re- e

ed tial eria

to on,

e

n-

in

be- s a

ally

is ce

Eqs.

Accordingly, we successively systematically halve d cretization intervals for a sequence of at least three anal and examine whether the magnitude of differences in ma mum stress values is decreasing. Initially, we favor unifo discretization throughout as a means of readily ensuring tematic refinement. We next demonstrate this approach o set of four sample configurations, each having a differ stress singularity present: The last three of these analyse taken from Sinclair@151#.

The first configuration analyzed features a crack subjec to remote tensile loading~as in Fig. 14 but withm15m2 , n15n2 , andF replaced by a uniform tractions0). So as to limit the extent of discretization, we in fact take a period array of such cracks sharing a common crack plane. cracks all have length 2a and a center-to-center separatio from their nearest neighbor of 4a. An exact solution for such a configuration is given in Westergaard@152# and shows the presence of inverse-square-root singularities in the nor stresses ahead of the cracks. To analyze the configuration use an integral equation derived via periodic Flamant l loads. In the numerical analysis of this integral equation, discretize the unknown as a piecewise constant on a se intervals of equal length. Given an appreciation of the p sible singular character of this unknown, numerical alg rithms of superior efficiency can readily be devised. He though, we are proceeding as if we hadno such awareness and asking the numerics themselves to reveal any singula present. Results from applying our unsophisticated numer analysis are presented in Table 5 for the maximum transv normal stress ahead of a crack (smax), normalized by s0(snom).

The second configuration treated entails a 90° reent corner under tension~as in Fig. 16 withf53p/2). The spe- cific finite elastic plate chosen results from taking a squa with uniform traction s0 applied to its upper and lowe edges, and cutting out a 90° corner on one side so that vertex of the corner is right at the center of the origin square. For such a configuration, the strongest of the sin larities for P5 in Table 2 and Fig. 2d can be anticipated

Table 5. Numerical divergence in the presence of stress singularities

Configuration —analysis

No. of intervals or elements

smax

snom

Percentage change

Periodic crack under tension —integral equation

32 4.49 — 64 6.22 39

128 8.71 55

Reentrant corner under tension —FEA with three node triangles

48 2.75 — 192 3.99 45 768 5.63 60

3072 7.83 80

Epoxy-steel butt joint under tension —FEA with four node quadrilaterals

10 1.22 — 40 1.49 22

160 1.94 37 640 2.51 47

2560 3.23 59

Surface step shear —FEA with four node quadrilaterals

16 1.12 — 64 1.35 21

256 1.59 21

ear

er- by a col-

per-

arity

ge is to For ses for ap- is,

ty, ting al.

: s

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 287

Fig. 19 Finite element grids for reentrant corner under tensiona) initial grid with 48 elements,b) first refinement with 192 element

p

e

t b

t

e o

s

of

alv- gu- eri- er ni- the of

ws.

l if

re- is

rity h. in- riori wl-

naly- also t in uc-

, it is

e ad tor.

ten-

to tion er of

he e

in y ble ly

y-

ect of

s of

ut-

ses ing are

sur-

an stic ni-

d ad-

rilaterals. Results are included in Table 5 withsmax being the peak principal stress at the point of application of the sh step andsnom beingt0 .

To examine convergence in Table 5, we take the diff ence insmax/snom for successive analyses normalized the ratio’s value for thefirst analysis and expressed as percentage. This process leads to the results under the umn headed ‘‘percentage change’’ in Table 5. For these centage changes,no decreasewith successively refined analysis reveals divergence and the presence of a singul ~cf the first of Eqs.~4.38!!.

For the first two configurations treated, the percenta change increases in a similar and pronounced way: This be expected since they have very similar singularities. the third, with its somewhat weaker singularity, the increa are less marked but nonetheless clearly evident. In fact, these first three, stress increases with grid refinement proach that expected from the singularity present. That stresses increasing by a factor of 2g whereg is 1/2, 0.46, and 1/3, respectively. For the last with its logarithmic singulari the percentage changes remain constant, thus still indica this singularity’s presence even if only with a weak sign Hence singularities are numerically unearthed for all four our sample configurations.

The results in Table 5 support the use of systematic h ing of discretization lengths to reveal the presence of sin larities. However, they are merely a set of numerical exp ments for which the approach works—for oth configurations/results it may not. For example, the mag tudes of the singular shear stress on the interface in epoxy-steel butt joint, as found on the same sequence meshes as used for the normal stress there, are as follo

tmax/s0 : 0.29, 0.40, 0.50, 0.61, 0.74

% change: 2, 38, 34, 38, 45 (4.43)

The results in Eq.~4.43! are consistent with a numerica analysis which is converging on the first three grids, even only slowly so. The later grids, though, start to diverge, vealing the singularity present. What is happening here that, in addition to aO(r 21/3) singularity, the shear on the interface can have other regular contributions asr→0. The participation of these regular terms here hides the singula from the coarser grids. Ultimately it has to show thoug Nonetheless, the possibility of regular fields concealing s gular ones to a degree underscores the value of an a p appreciation of potential singular stresses, since such kno edge tends to make one check a more extensive set of a ses for their actual realization. Such an appreciation may enable the region of grid refinement to be confined to tha the neighborhood of the potential singularity, thereby red ing computational effort.

4.4 Numerical analysis: Resolution of singular fields

Once a stress singularity’s presence has been detected necessary toquantify its participation if one is to effect aK interpretation. At this point, asymptotic identification of th nature of the singularity is no longer optional, but inste essential in order to properly define a stress intensity fac

because this is the singularity associated with transverse sile loading. That is, we expect stresses ofO(r 20.46) as r →0, wherer here is the distance from the corner. Turning the analysis of the configuration, symmetry enables atten to be confined to the upper half of the plate. For this up half, we continue to proceed as if we had no appreciation the possibility of a stress singularity and simply employ t finite element method with uniform grids comprised of thr node triangles. The first grid has 48 such elements~Fig. 19a!, the second is formed by halving element sides to resul 192 elements~Fig. 19b!, and subsequent grids are formed further halving element sides. Results are included in Ta 5, wherein smax is the transverse normal stress direc ahead of the corner andsnom is s0 .

The third configuration considered is that of an epox steel butt joint under tension~as in Fig. 2e!. The steel is taken to be rigid and only the epoxy analyzed. The asp ratio of the epoxy layer is set as 10:1. The Poisson’s ratio the epoxy is taken to be approximately 3/8 so that stresse O(r 21/3) asr→0 can be expected,r here being the distanc from points where the epoxy-steel interface meets the side surface of the specimen~as for P7 of Table 2 and Fig. 2e!. Finite element analysis is again unsophisticated and u a sequence of uniform meshes, with the elements compri the meshes now being four node quadrilaterals. Results included in Table 5, whereinsmax is the maximum normal stress on the interface where it terminates at the outer face, andsnom is the nominal stress at such a location.

The fourth and final configuration treated concerns abrupt application of a shear traction to the edge of an ela plate~as in Fig. 2f!. The jump in the shear stress has mag tudet0 . The singularity anticipated is logarithmic~as atP8

in Table 2 and Fig. 2f!. Finite element analysis is performe on a sequence of uniform grids comprised of four node qu

r

h

d

r c t h

t e e p

l i i

t

e o .

f

t

a

al

- me ve the p- s a

ed,

en- n is ce

a its or

the er

g ibly lim- al

to ro- e on- om- ith

by re-

ch- ide e

cali- three tes

, in

ntly t of er- er- id,

not The

r- rior

288 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

With a definition ofK in hand, one can proceed with nume cal assessment using either a boundary integral equation proach or a finite element analysis.

A good review of the early research on the application finite element methods to singular elasticity problems given in Gallagher@153#. This paper was part of a first sym posium on numerical methods in fracture mechanics~Lux- moore and Owen@154#!. Developments since, for bot boundary integrals and finite elements, are reflected in s sequent symposia@155–158#. In what follows, we concen- trate on finite element analysis since it enjoys more wi spread use today.

There are two issues facing the stress analyst when tempting the numerical analysis of a singular problem. Fi to resolve the singular fields themselves sufficiently ac rately numerically. Second, to extract from the numerics associated stress intensity factor without diminishing t level of accuracy. In this section we focus on the first act ity.

With respect to resolution, a number of finite eleme methods have been developed. These may be loosely ca rized as belonging to one of the following three class methods which add special elements, methods which us cal grid gradation, and methods which use superposition cedures. Special elements attempt to improve resolution introducing appropriately singular representations into the ements immediately contiguous to the singular point. G gradation attempts the same goal by suitably increasing number of regular elements in the vicinity of the singu point. And superposition procedures attempt it by super posing analytical singular fields throughout the entire reg of interest, then letting the regular fields in standard eleme effectively correct boundary values so that they comply w the prescribed conditions sought. In terms of implemen tion, special elements typically take the least amount of fort on the part of the stress analyst. This is especially when the singular fields are introduced simply by moving mid-side nodes of isoparametric elements. This techniqu developed for cracks in Henshell and Shaw@159# and Bar- soum @160#. The approach is generalized to apply to oth singularities in Wait @161#.26 Given the relative ease o implementation, if such techniques can provide suffici resolution in return for reasonable levels of computati they would seem to be the method of choice at this time

Before describing an assessment of the resolution of particular method, it is appropriate to outline the elements an evaluation protocolthat needs to be adhered to in ord for an appraisal of any numerical method to be meaning

i! The method needs to be completely prescribed w respect to how it is to be implementedprior to any testing. Under these circumstances, there is no mix of the calibrating of any adjustable parameters in method with true testing of the same.

ii ! The problems employed must have no ambiguity as what are their correct answers so that there is no

ode nd a

i- ap-

of is

-

ub-

e-

at- st, u- he is

iv-

nt ego- s: lo- ro- by

el- rid the

ar m- on nts ith ta- ef- so he e is

er f nt n,

this of

er ul.

ith

ing he

to m-

biguity as to the errors incurred in their numeric analysis. This is probably best achieved viatest prob- lems with known and analytical solutions. Alterna tively, if two or more independent analyses of the sa problem employing different numerical methods ha converged to exactly the same answer to more than number of significant figures being used in the a praisal, the problem can reasonably be viewed a benchmarkone and used.

iii ! An extensive set of such problems should be analyz with each member of the set being markedlydifferent from all the others. Ideally, the set should be repres tative of all the decidedly distinct types of problem o which application of the subject numerical method envisaged. Then it is reasonable to infer performan in general practice from the numerical experiments.

iv! The evaluation should include a check onconvergence. In the case of FEA, this should be undertaken on sequence of grids, with each grid being formed from predecessor by refinement which is systematic, nearly so. In this way one can obtain an estimate of computational level likely to be required should furth accuracy be needed.

Any evaluation that falls significantly short of complyin with the above should be viewed as preliminary, and poss encouraging further appraisal, but nevertheless seriously ited in its ability to justify the general use of the numeric method under consideration. Preliminary evaluations are be expected by the initial developers of novel numerical p cedures: Their contributions principally lie in conceiving th new approach in the first place, then explaining and dem strating its use. However, subsequent evaluations and c parisons with other methods should, in essence, comply w the foregoing protocol.

Returning to the evaluation of special elements formed displacing mid-side nodes, the originating papers are p liminary in this regard. Henshell and Shaw@159# treats some six problems that, as reported anyway, are not strictly ben mark problems in accordance with the protocol. Setting as this limitation, for only three of the problems is the sam grid used—the other three, therefore, can be viewed as brating the respective meshes used to a degree. For the with the common grid, one problem in some sense calibra the approach while the other two are quite similar. Hence effect, there is one trial problem in Henshell and Shaw@159#. For this trial problem, stress intensity factors are appare determined to within about 1–2% using a modest amoun computational effort. Henshell and Shaw also do a conv gence check on one of their problems: This exhibits div gence in computingK between a coarse and a medium gr but convergence from the medium to a fine. Barsoum@160# analyzes only one 2D elastic configuration which again is established as a benchmark problem within the paper. paper uses several different meshes~which together do not constitute a convergence check!, and suggests that quarte point elements formed from six node triangles have supe accuracy to corresponding elements formed from eight n quadrilaterals. Barsoum also considers a thermoelastic a

, and

26Further references on the use of such elements may be found in Lim, Johnston Choi @162#.

n f

r

-

h

t a i

t

s

l i n n

i

i e

w

a

ngu- ial

ges t, six nge- m- t

the

in to

ch in rec-

of ac-

ing

ned el- em- ed

SH, ra- nts le-

ents t of led as lso

sis y m a are

tip

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 289

3D application. In sum, the main thrust of Henshell a Shaw @159# and Barsoum@160# is the development o quarter-point elements for the FEA of cracks, and demons tions of its potential. This is well done in the two pape Thorough testing of the approach, on the other hand, is attempted.

Meda and Sinclair@163# provides an appraisal of the ap proach which adheres to the previous protocol in large p Therein, a series of crack problems are analyzed using quarter-point elements of@159,160#. The problems seek to simulate some of the variety of physical effects encounte in applications. The first configuration considered to this e involves a periodic array of cracks, all of length 2a with center-to-center separations of 2B, and under far-field trans verse tension. By considering different spacings (a/B51/4, 1/2, and 3/4!, crack interaction effects can be studied. T second configuration represents a round compact ten specimen: Herein, loading is predominantly in bending stead of tension. The third configuration simulates cra opening loading: This is basically the same configuration the second except that loading is displacement control The fourth configuration reflects crack arrest by placing crack tip in close proximity to a stiffener. The fifth and fin configuration is a slanted crack under remote tension: Th a mixed-mode situation. Together, the five configurations alize a total of seven different problems.

All of the problems are test problems in that they ha known closed-form solutions. For the first problems with t different crack spacings, these solutions can be taken dire from Westergaard@152#. Notice, though, that Westergaard treatment applies a uniform transverse tension at infin Since FEA requires that we treat a plate of finite height, solution in@152# must be evaluated at the height chosen a the stresses found used to apply tractions there. While s tractions are nearly the same as those for simple ten when the height is greater than the crack spacing, they differ a little. The inclusion of such differences is essentia one is to formulate a problem with a true exact solution us @152#. For problems entailing the next three configuratio exact solutions are constructed by the superposition of fi sets of eigenfunctions~identified using Williams@20#!. The resulting sums maintain stress-free crack flanks and sym try conditions ahead of the crack. They do not replicate boundary conditions elsewhere that perhaps one would m naturally apply, but do reflect the character of the load sought. In any event, whatever conditions they do realize the remainder of the boundary are taken to be, in fact, exact conditions thereon. Thus, these sums themselves the exact solutions to the problems so posed. The last p lem solution is obtained by combining the solution for crack under uniform tension at infinity~see, eg, Tada et a @55#, pp 1.20, 5.1!, with the corresponding solution for un form shear of Irwin@164#. As for the first problems, thes solutions must be evaluated at finite stations in order to m a precise statement of the problem.

For the analysis of this series of test problems, t quarter-point elements are available: one obtained from e node quadrilaterals, the other from six node triangles. Tri

d

tra- s: not

- art. the

red nd

e sion in- ck as

led. he l

s is re-

ve he ctly ’s ity. he nd uch ion do if ng s, ite

me- the ost

ng on the are

rob- a

l -

ake

o ight n-

gular elements are chosen over quadrilateral since the si lar fields are then present within the element on all rad rays originating at its vertex, rather than just along its ed as is the case for quadrilateral elements. To be consisten node triangles are used as host elements. The local arra ment of the quarter-point elements follows ANSYS reco mendations~Chapter 3,@32#!. It features a fan of congruen isosceles triangles spreading out from the crack tip~Fig. 20!. Each triangle subtends an angle ofp/6 at the tip and has an altitude which is about one eighth of the crack lengtha The remainder of the mesh is generated automatically using command AMESH~Chapter 9,@32#!, since this is a conve- nient means of doing so, and one likely to be employed practice. This procedure is adhered to for all problems generate their baseline grids. These grids are taken as su the first instance because they are essentially the grids ommended by ANSYS@32#. Furthermore, in practice, like grids should probably result in no more than an order magnitude greater number of degrees of freedom, and cordingly be computationally tractable~the maximum num- ber of degrees of freedom for the baseline grids used be 2533!.

To examine convergence, baseline grids are coarse and refined by approximately quadrupling and quartering ement areas, respectively. The grid refinement is not syst atic, though it is nearly so. This is because of the continu use of the convenient automatic mesh generator, AME which is not completely systematic in its element configu tions. It is also because of the different types of eleme involved: That is, because the number of quarter-point e ments remains constant while the number of host elem changes. Around the crack tip, though, the arrangemen Fig. 20 is preserved with element altitudes being doub and halved, so that locally grid refinement may be viewed being systematic. These pairs of additional grids are a used in the FEA of all seven test problems.

In evaluating the resolution of the finite element analy in Meda and Sinclair@163#, we focus on the stress intensit factors computed via it since these are the key results fro practical perspective. However, in making this choice, we

Fig. 20 Local arrangement of quarter-point elements at a crack ~following ANSYS recommendations!

18 34 ut , sh,

the te is

hell a N- of ls

de ro-

ith .

290 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Table 6. Resolution ofK fields using quarter-point elements

Test problem description

Number of elements used in coarse, medium and fine grids

Absolute percentage error in K

Periodic crack under tension: a/B51/4, 1/2, 3/4

174 1 620 0

1959 0

Round compact tension specimen

97 1 328 0

1207 0

Specimen with crack opening loading

97 37 328 4

1207 5

Crack arrest at a stiffener

97 4 328 2

1207 1

Mode I for slanted crack under tension Mode II

216 5 488 3

2000 1 216 21 488 10

2000 4

s

e

d

r l

y

r

i

t

t

c

M

een first tion al inten-

s - ost

en lts

l

w e e his an eri-

ped cted is na

rn si- t- stin- and ee s ge th-

o- r e to d

ror

of, e- - this sis

re- d in ons

of ory. d- he

l- ay

- h ach

e sing

eda

en- nd ely.

terms of the previous rating scale, accuracy is excellent in problems, good in 8, and satisfactory in 1. In 33 of the problems, values ofK computed are converging througho the three-grid sequence employed. In the other problemK values are diverging from the coarse to the medium me but converging from the medium to the fine. Based on number of elements used, the average convergence ra about halfway between linear and quadratic.

In sum, therefore, the quarter-point elements of Hens and Shaw@159# and Barsoum@160#, when arranged around crack tip in accordance with the recommendations of A SYS @32#, would seem to offer good levels of resolution the singular fields involved in return for quite modest leve of computation. In general, then, the shifting of mid-si nodes on isoparametric elements would appear likely to p vide a means of numerically analyzing singular fields w more than adequate accuracy, and doing so fairly readily

4.5 Numerical analysis: Extraction ofK ’s

With respect to extraction, a variety of procedures have b suggested over the years. Probably the most obvious approach is to attempt to take advantage of an apprecia of the asymptotic character of a singularity to fit loc stresses or displacements and, thereby, estimate stress sity factors. An early development of suchlocal fitting meth- ods is Chan, Tuba, and Wilson@166#. Alternatively, for cracks, one can obtainK by computing energy release rate ~as in ~3.5!, ~3.8!, and ~3.9!!. Several distinct implementa tions of this approach have been put forward. The m widely practiced is via theJ integral of Rice@88#. Others include the stiffness derivative technique of Parks@167#, and the virtual crack extension method of Rybicki and Kannin @168#. The virtual crack extension method uses local resu to estimateG, henceK: accordingly it qualifies as a loca fitting method. TheJ integral, on the other hand, is apath- independent integraland consequently does not have to dra on local fields. Parks@167#, in an Appendix, shows that th stiffness derivative technique is an area-analogue of thJ integral, so it also does not need to rely on local fields. T is a positive attribute since fields close to any singularity c be expected to be the least accurately determined via num cal analysis. A further set of procedures for extractingK which share this attribute are based on specially develo path-independent integrals. These integrals are constru by an adroit invoking of Betti’s reciprocal theorem: Th leads to integrands that are akin to those in Somiglia integrals.27 The method of construction has its origin in Ste @169#. The integrals that result are devoid of the direct phy cal interpretation ofJ, but are computationally more adap able. For cracks, as a consequence, they can readily di guish between different modes, as in Stern, Becker, Dunham@170#. They can also be adapted to the fixed-fr corner, as in Stern and Soni@171#, and the interface crack, a in Hong and Stern@172#. Others have since taken advanta of the ideas underlying the construction of such pa

unfortunately combining an appraisal of singular field re lution with one ofK extraction capability. Provided the latte is consistently reasonably accurate, we should still be abl infer the effective levels of singular field resolution obtain ~we examine the issue ofK extraction in the next section!.

With respect to the accuracy sought, we view 0–1% er as excellent, 11 – 5% as good, 51 – 10% as satisfactory, an greater than 10% as unsatisfactory. In justification there given the likely level of agreement between physical sponse and predictions made viaK, we can expect an exce lent analysis, and even just a good analysis, to leave agreement largely unaltered, while a satisfactory anal probably would not impair it significantly.

The number of elements actually used and the co sponding errors in stress intensity factors are summarize Table 6~the same errors are obtained for all three separat in the periodic crack problem!. On the baseline~medium! grid, the average absolute error is 2.4%, with four results excellent accuracy, three with good, and one just satisfac With the exception of the problem with crack opening loa ing where results have yet to converge in going from medium to the fine mesh, all results are converging.

A further evaluation of the resolution of quarter-point e ements which largely adheres to the protocol given here m be found in Cooper et al@165#. This features more displace ment controlled/Mode II loadings, the two situations whi would appear to promote the greatest errors for the appro ~Table 6!. All told, 34 test problems are constructed in@165#, with 18 being Mode I, 16 Mode II, and half for each mod having prescribed displacements. They are analyzed u the same elements and mesh generation scheme as in and Sinclair@163# ~ie, following ANSYS @32# recommenda- tions coupled with easy-to-implement automatic mesh g eration!. This results in a baseline grid of 276 elements, a coarse and fine grids of 57 and 916 elements, respectiv For the baseline grid, the average absolute error inK for all 34 test problems is 1.6%, while the maximum is 5.9%.

In

27A statement of Betti’s reciprocal theorem may be found in Art 121, Love@12#: Somigliana integrals Art 169, ibid.

n od

een on- the

ress

of

ons e of loy

the can s a

le- ills

her. hey lier ted at ro- r of

the

ive .2% ed th- ting

uite nd one

as are

s

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 291

independent integrals to develop them for further singular configurations. Examples include the stress-free reentrant corner in Carpenter@173# and in Sinclair et al@174#, the butt joint in Okajima and Sinclair@175#, and the bimaterial reen- trant corner in Carpenter and Byers@176#. The path- independent line integral of Okajima and Sinclair@175# can also be found in Banks-Sills@177#, together with an equiva- lent area integral.

At the outset in evaluating these competing techniques, local fitting methods appeal in their ease of implementation and adaptability to different singularities. However, once corresponding path-independent integrals have been devel- oped and made available as algorithms within standard codes, implementation typically requires little if any extra effort on the part of the stress analyst. Further, using Stern’s approach, path-independent integrals are quite adaptable. So the initial advantages of local fitting methods can be ex- pected to be of no great consequence in practice as the de- velopment of path-independent integrals continues. There is, though, an inherent deficiency in local fitting methods that is of concern in practice.

This deficiency stems from the fact that local fitting meth- ods fit quantities near but not at the singularity. They must avoid the singular point because stresses there are un- bounded and therefore not fitting, while displacements are zero leaving nothing to fit. Given such necessary backing off from the singular point, other regular fields can participate in any fit. This participation cannot generally be either com- pletely accounted for by any local fitting method, or com- pletely eliminated. As a result, local fitting methods have the potential to be unreliable in their accuracy. That is not to say they cannot furnish accurate, or even occasionally extremely accurate, estimates ofK: Just that they can also provide un- satisfactory estimates.

One might think that all that is required to overcome such deficiencies is to develop a better local fitting method. Logi- cally, though, this cannot be done in any complete sense. To explain further, any fit must match afinite number of quan- tities. Hence, since singular configurations can have aninfi- nite number of regular eigenfunctions participate in addition to their singular ones, there always exists the possibility of some being left unaccounted. Indeed, the existence of such unfitted eigenfunctions at a crack tip is what is exploited in Sinclair @132# to cause the complete inaccuracy of several local fitting methods~ie, to have them estimateK50 when

KÞ0, and vice versa!.28 Moreover, the app be adapted to ensure the downfall of any once the specifics of how it is to be impl decided. However, these types of demon trived test problems, so that there is an degree such difficulties are actually enco this issue, we draw on evaluations in the

The originators of local fitting metho extractingK from numerical analyses—he and Rybicki and Kanninen@168#—underst preliminary evaluations that showcase p rather than establish accuracy levels for true test or benchmark problems. Further their methods in conjunction with quarter- selected approach for resolving singular be said of the originators of path-indepe means ofK extraction—here Parks@167# @170#.29 Turning to evaluations that do us ments, two limited assessments are ava and Sherman@178#, Pang@179#, and Pang

In Banks-Sills and Sherman@178#, thr problems are analyzed, two being quite s These problems are not true test prob qualified in@178# as benchmarks in the se in the evaluation protocol. However, they in the latter role with a less stringent d adopted here. The best crack-flank displ cedure considered results in an appare 0.9%, with excellent accuracy for two pro other one. The path-independent integra as evaluated either directly or by the technique, results in an apparent average with excellent accuracy on all three prob evaluation, therefore, would seem to independent integrals are more accura methods.

In Pang@179# and Pang and Leggat@18 set of crack problems is analyzed. The s distinct configurations, eight different twenty-seven different analyses/stress in

28Essentially, this is the same characteristic of l employed to produce the erroneous results for nearb described in Section 3.5. 29Rice @88# introduces theJ integral to a different end not attempt any evaluation of it as a numerical tool.

Table 7. Comparison of someK-extraction methods using Pang and Leggat†180‡

Trial problems involved

Measures of apparent absolute error

J via stiffness derivative

Crack-flank displacement fit

Virtual crack extension

Stress fit ahead of crack

No. 1 with 8 grids

Average error ~%!

0.5 1.2 2.3 3.8

Accuracy distribution

8e 5e, 3g 8g 6g, 2s

Nos. 2–8 Average error ~%!

2.8 3.0 — 4.7

Accuracy distribution

6e, 10g 3s

4e, 12g 3s

— 1e, 11g 6s, 1u

Key: e...excellent~0–1%!, g...good (11 – 5%), s...satisfactory (51 – 10%), u...unsatisfactory (.10%)

roach therein ca local fitting meth

emented have b stration are for c open issue as to untered. To add literature.

ds as a means re Chan et al@166# andably perform otential applicati

a diverse rang , they do not emp point elements, fields. The same ndent integrals a and Stern et al e quarter-point e ilable in Banks-S and Leggat@180#. ee planar crack imilar to each ot lems, nor are t nse defined ear could be accep efinition than th acement fitting p nt average erro blems, good on l used, theJ integral stiffness derivat

error of about 0 lems. This limit

indicate that pa te than local fit

0#, an extensive et entails five q geometries, a

tensity factors. N

ocal crack-tip fields y fracture criteria that

and accordingly doe

re er- o: are er felt

ni-

ac- tu-

292 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

Table 8. Evaluation of two K-extraction methods

Test problems involved

Absolute error measures for baseline grid

H via direct integration

Crack-flank displacement fit

7 ~with 8 K ’s), from Meda & Sinclair @163#

Average error ~%!

2.4 10.6

Accuracy distribution

4e, 3g 1s

4e, 1g 1s, 2u

34, from Cooper et al @165#

Average error ~%!

1.6 9.3

Accuracy distribution

18e, 15g 1s

7e, 10g 7s, 10u

Key: e...excellent~0–1%!, g...good (11 – 5%), s...satisfactory (51 – 10%), u...unsatisfactory (.10%)

r

u

s n ,

s

e

p

e

i

o

s

s lgo- ad is

in

em- are

lies

ele-

n

ude

-

o a ns:

ex- s a der- to rals f a is u- ed tion

tion to ac-

sity oth

an

s, in- is- ath-

e. s all

are ing -

lent

me cel- ny- l- s th- ac-

f re to are te- n - nre

t al s are

Re- zed

e- ry

ber ~12!. Half of the latter instances entail errors that a greater than 20%. Moreover, while performance can gen ally be improved by grid refinement, this is not always s The displacement fitting procedure yields results which not converging on going from a baseline grid to a yet fin grid on seven occasions. What is making its presence here is the inherent unreliability of local fitting methods.

The H integrals of Stern, on the other hand, almost u formly provide good to excellent estimates ofK on baseline grids. Moreover, the two instances of merely satisfactory curacy converge to at least good on grid refinement. Ac ally, what is being displayed here is not the accuracy ofH integrals in extractingK: They essentially do this exactly, a can be established by feeding the integrands in their a rithms exact values of the field quantities called for inste of finite element estimates. Thus, what is being shown really the accuracy of the FEA determination of the fields these singular test problems.30 This accuracy is more than adequate provided fields in the quarter-point elements th selves, as well as those immediately contiguous to them, avoided. This can readily be done by taking a path which outside these elements when computing anH integral: Where precisely does not matter as long as these inner ments are not on it.

In Meda and Sinclair@163#, there is a further compariso of an H integral withJ. SinceJ by itself cannot distinguish between different modes, the comparison does not incl the mixed-mode test problem.31 On a common subset of five test problems analyzed on the baseline grid,J averages 1.0% absolute error inK, while H averages 0.4%. Hence, if any thing, this limited evaluation would indicate thatH is slightly more accurate thanJ.

The foregoing discussion focuses on 2D analysis. T degree, similar capabilities are available in three dimensio see Banks-Sills@181# and Meda et al@182#

In all, therefore, path-independent integrals can be pected to be more reliable than local fitting methods a means of extracting stress intensity factors. There are un lying reasons to think this might be so, and evaluations date demonstrate that it is. Hence, path-independent integ are to be preferred in practice. For cracks, the choice o specific integral and the way in which it is computed largely a matter of availability/convenience. For other sing lar configurations, it may well be that integrals develop along the lines of Stern and his coworkers are the only op for obtaining a corresponding path-independent integral.

When path-independent integrals are used in conjunc with isoparametric elements with mid-side nodes shifted reflect the stress singularity present, more than adequate curacy in the resolution and the extraction of stress inten factors can be obtained in return for reasonable levels of b implementation and computation. Other approaches~see, eg, @154–158#!, in concert with path-independent integrals, c

y

able

are strictly qualified as true test or benchmark problem though they could be viewed as the latter with a less st gent definition. Local fitting methods considered include d placement and stress fits, and virtual crack extension. P independent integrals considered include theJ integral as calculated directly or via the stiffness derivative techniq Results for the best displacement and stress fits, as well a those available from virtual crack extension and for theJ integral computed via the stiffness derivative technique, summarized in Table 7. The apparent order of decrea accuracy is:J integral, displacement fit, virtual crack exte sion, and stress fit. When theJ integral is computed directly apparently the average absolute error is 1.4% with excel accuracy in six instances, good in three: For theJ integral computed via the stiffness derivative technique on the sa set, apparently the average absolute error is 1.7% with ex lent accuracy in three instances, good in six. Hence, if a thing, this limited evaluation would indicate that direct ca culation ofJ is slightly more accurate than via the stiffne derivative. Irrespective of the means of computation, pa independent integrals would definitely appear to be more curate than local fitting methods in Pang and Leggat@180#.

In the earlier cited papers, Meda and Sinclair@163# and Cooper et al@165#, there are also contained assessments oK extraction methods. These assessments do basically adh the evaluation protocol. The two methods so evaluated crack-flank displacement fitting and path-independent in grals developed a` la Stern. For the first, its specifics are give in ANSYS @32#: Reasons for believing the particular a proach prescribed therein is amongst the best of its ge available are given in Cooper et al@165#. For the second, the precise forms of the integrals used are given in Sinclair e @174#, wherein they are dubbedH integrals. Both approache are applied to all the test problems for planar cracks that set out in@163,165#. Implementation is in concert with th same arrangement of quarter-point elements as earlier. sults for the errors incurred on baseline grids are summar in Table 8.

Evident in Table 8 is that the displacement fitting proc dure on baseline grids typically leads to barely satisfact estimates ofK. What is more disconcerting is the scatter performance. While this local fitting method produces e mates of excellent accuracy for a number of problems~11!, it also furnishes unsatisfactory estimates on a comparable n

in ti-

um-

30It follows that our previous use ofK to assess FEA resolution is not polluted b extraction error becauseH integrals are employed in this appraisal. 31The J integral can be supplemented by a further path-independent integral to en the participation of different modes to be distinguished: see Ch 5, Cherepanov@62#.

a

w

t

a e

r

g

a .

i

m

o -

In d-

s

to- en-

- s

ned nded east m- nt els lar- in-

ty’s

nt,

city ity

nd las-

, it her

ac- o ca- rs. s is

of ia

ly sid-

ter

at-

with

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 293

offer comparable and even superior accuracy for the s computation levels, but can take more effort on the part the stress analyst.

5 CONCLUDING REMARKS

In classical elasticity, stress singularities can occur un point loads, line loads, and so on. They can also occur a from any such concentrated loading. Then typically they flect, albeit crudely, physical stress concentrations. In role, these singular stresses direct attention towherefailure is likely to occur, but are useless in themselves for predict when it occurs. It is this latter type of singularity that is o concern in attempting to ensure structural integrity. Acco ingly, this type of singularity is the focus here, as well as Part II of this review.

When stress singularities occur away from concentra loads, they do so in concert with discontinuities. These d continuities can be in boundary directions, or in bound conditions, or in elastic moduli. While such discontinuiti do not have to have associated stress singularities, often do. Discontinuity singularitiesare thus far from rare in elas ticity ~Part II of this review amplifies their occurrence fu ther!.

At the outset in dealing with discontinuity singularities, is essential that theirparticipation be recognized. Otherwise one risks making stress-strength comparisons in their p ence, an exercise in futility. Given recognition of a discon nuity singularity, the engineer has three options in seekin ensure structural integrity.

i! To rely primarily on testing and forego analysis, oth than perhaps nominal~1D! stress analysis.

ii ! To proceed with classical stress analysis~2D or 3D!, then try to interpret the stress singularity.

iii ! To improve the modeling so that the singularity is r placed with physically sensible stresses that can compared with strengths.

For the all-important first step of identifying the presence a stress singularity in elasticity, two types of analysis available: analytical asymptotics and numerical methods

With respect toasymptotics, three principal approache exist for 2D analysis: via the Airy stress function, v Kolossoff-Muskhelishvili complex potentials, and via th Mellin transform. These approaches are well developed this time. Properly implemented, all three identify the sa stress singularities: Hence, the choice of which to use largely a matter of personal preference. In two dimensio the various elastic stress singularities actually identified date with these approaches may be summarized as foll For any stress components, as the singular point is ap proached, elasticity can have:

s5O~r 2g cos~h ln r !!1O~r 2g sin~h ln r !!

s5O~r 2g ln r !1O~r 2g!

s5O~r 2g!

s5ord~ ln2 r !1ord~ ln r ! (5.1)

me of

der ay

re- his

ing f rd- in

ted is- ry s

they - -

it

res- ti- to

er

e- be

of re

s a e at e is

ns, to ws.

s5ord~ ln r !

s5O~ ln r !

s5O~cos~h ln r !!1O~sin~h ln r !!

as r→0, whereing is the singularity exponent (0,g,1), and h is the imaginary part of the eigenvalue involved. Eqs.~5.1!, O is associated with locally homogeneous boun ary conditions, ord with locally inhomogeneous~ord being defined in Section 1.2!. Corresponding stress fields for plate in extension may be found in Section 4.1, Eqs.~4.7! and ~4.17! (g512l,12j, respectively!, and in Section 4.2, Eqs. ~4.18! and ~4.28! (g512l). Further corresponding stress fields for other configurations are given in Part II, gether with specifics of the numerous configurations that gender such singularities.32

With respect tonumerical methods, the presence of sin gularities can bedetectedby the divergence of peak stres values. Evidence of divergence requires a suitably refi sequence of discretizations. The sequence recomme here halves discretization intervals on a sequence of at l three analyses. With this approach in 2D FEA, element nu bers quadruple with grid refinement. In 3D FEA, eleme numbers increase by a factor of eight. Even with such lev of computational effort, there is no guarantee that a singu ity be detected. However, numerical experiments to date dicate that one is reasonably likely to unearth a singulari presence with the approach.

Once a discontinuity singularity is known to be prese the singular fields active require specialinterpretationif they are to be used. The foremost such interpretation in elasti takes the coefficient of the singularity, the stress intens factor K, as the parameter controlling brittle fracture a failure in general. This remains the basic tenet of linear e tic fracture mechanics~LEFM! even today. While LEFM is concerned primarily with the stress singularities at cracks is possible to consider extension of its basic tenet to ot singularities.

For the case of cracks within a single material, the pr tice of LEFM is quite accomplished at this time. In tw dimensions in particular, finite element analysis is most pable when it comes to calculating stress intensity facto The means favored here for resolving the crack-tip stresse via quarter-point elements~Section 4.4!, though certainly other possibilities are available. The most reliable means extractingK from such an analysis would appear to be v path-independent integrals~Section 4.5!. Companion testing is also well controlled and reproducible~Section 3.4!. How- ever, while predictions made by LEFM are typical trendwise correct, there are occasions when there is con erable room for improvement in their accuracy~Section 3.4!, and extension to other singularities may well face yet grea difficulties in making accurate predictions~Sections 3.2 and 3.3!. All told, there would appear to be a good case for

32The last stress of Eqs.~5.1! is not strictly singular, being bounded asr→0. However, it is undefined asr→0, and consequently shares some of the difficulties associated stress singularities.

s

f r

r

i

t

n

o ’

r

e t

d

u

e

e

-

ed ne ori-

lied

e

d eory

c-

d

d

ted

i- rta-

of

and

s

lar

e

in

e

I.

- nd-

e

le

a

294 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

tempting toimprove the modelingso that stress singularitie are replaced with physically sensible stresses.

For conforming contact problemsin elasticity, the re- moval of singularities is now well understood. This remov is accomplished by the policing of contact inequalities wh there are sufficient degrees of freedom in a problem to ef such policing~Section 2.4!. Commercial FEA codes are cu rently available to implement such analysis. Resulting fin stress fields continue to prove to be useful in enginee practice.

For other singular configurations, the removal of singu- larities is nowhere near as mature as it is for conform contact. However, the realsource of such singularitiesis emerging. These singularities do not really stem from discontinuities present, nor from the field equations of el ticity ~Section 2.1!. Rather, they stem from a probably u witting introduction of effectively infinite stiffnesses in co hesive laws. With this appreciation, it would appear to possible to remove most if not all of the discontinuity sing larities of elasticity by ensuring finite stiffnesses~Section 2.3!. Such removals can be pursued with or without rem ing the original discontinuity, indicating the discontinuity secondary role in the generation of stress singularities. T can also be undertaken without introducing plasticity or la strain effects, though such effects may merit inclusion loading progresses. Implementation of this type of approa however, faces some serious challenges. There are mod issues, analytical tractability concerns, and interpreta questions. Nonetheless, research in this area holds the p ise of significant improvements in the physical appropria ness of stress fields in classical elasticity in particular, an solid mechanics in general.

ACKNOWLEDGMENTS

I am grateful for the numerous comments received from c leagues during the course of preparing this review: JL Be Jr ~Carnegie Mellon University!, KL Johnson~Cambridge University, UK!, G Meda ~Corning, Inc!, TP Pawlak~AN- SYS, Inc!, BS Smallwood~Chrysler Corporation!, and PS Steif ~Carnegie Mellon University!. I am also grateful for the careful typing of the manuscript by M Gibb and R Kostya and the painstaking preparation of the drawings by K You ~all of Carnegie Mellon!. In addition, I am grateful for the thoughtful input of reviewers, and the incorporation of r sulting revisions by my wife, Della.

REFERENCES

@1# Thomson~Lord Kelvin! W ~1848!, Note on the integration of the equations of equilibrium of an elastic solid,Camb. Dublin Math. J.3, 87–89.

@2# Boussinesq J~1878!, Equilibrium of an elastic isotropic half-spac supporting different loads in the absence of gravity,Acad. Sci., Paris, C. R.86, 1260–1263~in French!.

@3# Cerruti V ~1882!, Studies of the equilibrium of isotropic elastic bod ies, Reale Academia dei Lincei, Serie3a, Memorie della Classe di Scienze Fisiche, Matematiche e Naturali13, 81–122~in Italian!.

@4# Mindlin RD ~1936!, Force at a point in the interior of a semi-infinit solid, Physics (N.Y.)7, 195–202.

@5# Poulos HG, and Davis EH~1974!, Elastic Solutions for Soil and Rock Mechanics, John Wiley and Sons, Inc, New York, NY.

@6# Michell JH ~1900!, Elementary distributions of plane stress,Proc. London Math. Soc.32, 35–61.

al en ect - ite ing

ng

he as- - - be u-

v- s hey ge as ch, ling

ion rom- te-

in

ol- th,

k, ng

e-

-

@7# Flamant M ~1892!, On the distribution of stresses in a two dimensional solid under transverse loading,Acad. Sci., Paris, C. R. 114, 1465–1468~in French!.

@8# Boussinesq J~1892!, On the local disturbances which are produc under concentrated loads, uniformly distributed in the out-of-pla direction, and acting on the upper surface of a half-space, either h zontally or as a pair vertically,Acad. Sci., Paris, C. R.114, 1510– 1516 ~in French!.

@9# Melan E~1932!, The state of stress due to a concentrated load app within a half-plane,Z. Angew. Math. Mech.12, 343–346~in German!.

@10# Kurshin LM ~1959!, Mixed plane boundary value problem of th theory of elasticity for a quadrant,J. Appl. Math. Mech.23, 1403– 1408.

@11# Sternberg E, and Eubanks RA~1955!, On the concept of concentrate loads and an extension of the uniqueness theorem in the linear th of elasticity,J. Ration. Mech. Anal.4, 135–168.

@12# Love AEH ~1944!, A Treatise on the Mathematical Theory of Elasti ity, 4th Edition, Dover Publ, New York NY.

@13# Turteltaub MJ, and Sternberg E~1968!, On concentrated loads an Green’s functions in elastostatics,Arch. Ration. Mech. Anal.29, 193– 240.

@14# Chowdhury KL ~1983!, Solution of the problem of a concentrate torque on a semi-space by similarity transformations,J. Elast. 13, 87–90.

@15# Chen T ~1992!, Some remarks on the solutions of a concentra torque and double forces on an elastic half-space,ASME J. Appl. Mech.59, 690–692.

@16# Timoshenko SP, and Goodier JN~1970!, Theory of Elasticity, 3rd Edition, McGraw-Hill Book Co, New York NY.

@17# Kolossoff G~1910!, On an application of the theory of complex var ables to the two-dimensional problem of elasticity theory, Disse tion, St Petersburg, Russia.

@18# Kolossoff G ~1914!, On some properties of the plane problem elasticity theory,Math. Phys.62, 384–409~in German!.

@19# Inglis CE ~1913!, Stresses in a plate due to the presence of cracks sharp corners,Inst. Nav. Archit. Mar. Eng., Trans.55, 219–241.

@20# Williams ML ~1952!, Stress singularities resulting from variou boundary conditions in angular corners of plates in extension,ASME J. Appl. Mech.19, 526–528.

@21# Williams ML ~1959!, The stresses around a fault or crack in dissimi media,Bull. Seismol. Soc. Am.49, 199–204.

@22# Sadowsky MA ~1928!, Two-dimensional problems of elasticity theory,Z. Angew. Math. Mech.8, 107–121~in German!.

@23# Abramov BM ~1937!, The problem of contact of an elastic infinit half-plane with an absolutely rigid rough foundation,C. R. (Dokl.) Acad. Sci. URSS17, 173–178.

@24# Zak AR ~1964!, Stresses in the vicinity of boundary discontinuities bodies of revolution,ASME J. Appl. Mech.31, 150–152.

@25# Dempsey JP, and Sinclair GB~1981!, On the singular behavior at the vertex of a bi-material wedge,J. Elast.11, 317–327.

@26# Thomson W, and Tait PG~1867!, A Treatise on Natural Philosophy, Cambridge Univ Press, Cambridge, UK~this book is also available in two parts asPrinciples of Mechanics and Dynamics, Dover Pub Inc, New York NY!.

@27# Brahtz JHA~1933!, Stress distribution in a reentrant corner,Trans. ASME Ser. E.55, 31–37.

@28# Knein M ~1926!, On the theory of pressure testing,Z. Angew. Math. Mech.6, 414–416~in German!.

@29# Dempsey JP, and Sinclair GB~1979!, On the stress singularities in th plane elasticity of the composite wedge,J. Elast.9, 373–391.

@30# Sneddon IN~1951!, Fourier Transforms, McGraw-Hill Book Co, New York NY.

@31# ABAQUS personnel~1997!, ABAQUS Standard User’s Manual, Re- vision 5.7, Vol I, Hibbitt, Karlsson and Sorensen Inc, Pawtucket R

@32# ANSYS personnel~1995!, ANSYS User’s Manual, Revision 5.2, Vol I, ANSYS Inc, Canonsburg PA.

@33# Sinclair GB, and Epps BE~2002!, On the logarithmic stress singu larities induced by the use of displacement shape functions in bou ary conditions in submodelling,Commun. Numer. Methods Eng.18, 121–130.

@34# Frisch-Fay R~1962!, Flexible Bars, Butterworth Inc, Washington DC. @35# Kondo M, and Sinclair GB~1985!, A simple substructuring procedur

for finite element analysis of stress concentrations,Commun. Numer. Methods Eng.1, 215–218.

@36# Cherepanov GP~1967!, Crack propagation in continuous media,J. Appl. Math. Mech.31, 503–512.

@37# Hutchinson JW~1968!, Singular behaviour at the end of a tensi crack in a hardening material,J. Mech. Phys. Solids16, 13–31.

@38# Rice JR, and Rosengren GF~1968!, Plane strain deformation near

t

p

-

f a

e

s

l

a

g

s

e

i

s

d

i

i

in

i

.

f

ri- ch

e

-

f e

e the

d

i-

b-

-

f a

-

te

ar

-

-

re,

er

ial

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 295

crack tip in a power-law hardening material,J. Mech. Phys. Solids16, 1–12.

@39# Ramberg W, and Osgood WR~1943!, Description of stress-strain curves by three parameters, Tech Note 902, Natl Advisory Commi on Aeronautics, Washington DC.

@40# Chao YT, and Yang S~1992!, Singularities at the apex of a shar V-notch in a linear strain hardening material,Int. J. Fract.57, 47–60.

@41# Rudge MRH, and Tiernan DM~1995!, Interfacial stress singularities in a bimaterial wedge,Int. J. Fract.74, 63–75.

@42# Zhang N, and Joseph PF~1998!, A nonlinear finite element eigenan alysis of singular plane stress fields in bimaterial wedges includ complex eigenvalues,Int. J. Fract.90, 175–207.

@43# Wong FS, and Shield RT~1969!, Large plane deformations of thin elastic sheets of neo-Hookean material,Z. Angew. Math. Phys.20, 176–199.

@44# Knowles JK, and Sternberg E~1973!, An asymptotic finite- deformation analysis of the elastostatic field near the tip of a crackJ. Elast.3, 67–107.

@45# Knowles JK, and Sternberg E~1974!, Finite-deformation analysis o the elastostatic field near the tip of a crack: Reconsideration higher-order results,J. Elast.4, 201–233.

@46# Geubelle PH, and Knauss WG~1994!, Finite strains at the tip of a crack in a sheet of hyperelastic material: I. Homogeneous casJ. Elast.35, 61–98.

@47# Geubelle PH, and Knauss WG~1994!, Finite strains at the tip of a crack in a sheet of hyperelastic material: II. Special bimaterial ca and III. General bimaterial case,J. Elast.35, 99–174.

@48# Knowles JK, and Sternberg E~1975!, On the singularity induced by certain mixed boundary conditions in linearized and nonlinear e ticity, Int. J. Solids Struct.11, 1173–1201.

@49# Ru CQ ~1997!, Finite deformations at the vertex of a bi-materi wedge,Int. J. Fract.84, 325–350.

@50# Duva JM ~1990!, The singularity strength at the apex of a wed undergoing finite deformation,ASME J. Appl. Mech.57, 577–580.

@51# Griffith AA ~1920!, The phenomena of rupture and flow in solid Philos. Trans. R. Soc. London, Ser. AA221, 163–198.

@52# Mansfield EH~1967!, On the stresses near a crack in an elastic sh Tech Report 67030, Royal Aircraft Est, Cranfield UK.

@53# Truesdell C~1952!, The mechanical foundations of elasticity and flu dynamics,J. Ration. Mech. Anal.1, 125–300;~1953!, 2, 593–616.

@54# Zheltov YuP, and Khristianovitch SA~1955!, On the mechanism of hydraulic fracture of an oil bearing stratum,Izvestiya Akademiia Nauk SSSR, Otd. Tekhn. Nauk5, 3–41~in Russian!.

@55# Tada H, Paris PC, and Irwin GR~1985!, The Stress Analysis of Crack Handbook, 2nd Edition, Paris Productions Inc, St Louis MO.

@56# Barenblatt GL~1959!, On the equilibrium of cracks due to brittle fracture,Dokl. Akad. Nauk SSSR127, 47–50~in Russian!.

@57# Barenblatt GI~1962!, The mathematical theory of equilibrium crack in brittle fracture,Adv. Appl. Mech.7, 55–129.

@58# Willis JR ~1967!, A comparison of the fracture criteria of Griffith an Barenblatt,J. Mech. Phys. Solids15, 151–162.

@59# Goodier JN~1968!, Mathematical theory of equilibrium cracks,Frac- ture, An Advanced TreatiseH Liebowitz ~ed!, Vol II, Academic Press, New York NY, 1–66.

@60# Peierls R~1940!, The size of a dislocation,Proc. Phys. Soc. London 52, 34–37.

@61# Hirth JP, and Lothe L~1968!, Theory of Dislocations, McGraw-Hill Book Co, New York NY.

@62# Cherepanov GP~1979!, Mechanics of Brittle Fracture, McGraw-Hill Book Co, New York NY.

@63# Lennard-Jones JE~1931!, Cohesion,Proc. Phys. Soc. London43, 461–482.

@64# Israelachvili JN~1992!, Intermolecular and Surface Forces, 2nd Ed tion, Academic Press, San Diego CA.

@65# LaméMG ~1852!, Lectures on the Mathematical Theory of the Ela ticity of Solid Bodies, Bachelier Pub, Paris, France~in French!.

@66# Sinclair GB, and Meda G~2001!, On some anomalies in Lame´’s solutions for elastic solids with holes,ASME J. Appl. Mech.68, 132– 134.

@67# Sinclair GB ~1996!, On the influence of cohesive stress-separat laws on elastic stress singularities,J. Elast.44, 203–221.

@68# Sinclair GB ~1999!, A bibliography on the use of cohesive laws solid mechanics, Report SM 99-8, Dept of Mech Eng, Carnegie M lon Univ, Pittsburgh PA.

@69# Cribb JL, and Tomkins B~1967!, On the nature of the stress at the t of a perfectly brittle crack,J. Mech. Phys. Solids15, 135–140.

@70# Needleman A~1994!, Computational modeling of material failure Proc of 12th US Nat Congress of Appl Mech, Seattle WA, S34–S42

@71# Sinclair GB, Meda G, and Smallwood BS~1995!, On the physical

tee

ing

,

nd

,

es,

as-

l

e

,

et,

d

s

-

s-

on

el-

p

,

stress field for the Griffith crack,Proc of 15th Canadian Congress o Appl Mech, Victoria, British Columbia,1, 210–211.

@72# Sinclair GB~2000!, Ridding elastic configurations of stress singula ties, Proc of 20th Southeastern Conf of Theoretical and Appl Me, Callaway Gardens, GA, pp SM93.1–8.

@73# Johnson KL ~1985!, Contact Mechanics, Cambridge Univ Press, Cambridge, UK.

@74# Hertz H ~1882!, On the contact of elastic solids,J. Reine Angew. Math. 92, 156–171 ~in German: for an account in English, se Johnson@73#, Ch 4!.

@75# Steuermann E~1939!, To Hertz’s theory of local deformations in compressed elastic bodies,C. R. (Dokl.) Acad. Sci. URSS25, 359– 361.

@76# Persson A~1964!, On the stress distribution of cylindrical elastic bod ies in contact, Dissertation, Chalmers Univ of Technology, Go¨teborg, Sweden.

@77# Gladwell GML ~1980!, Contact Problems in the Classical Theory o Elasticity, Sijthoff and Noordhoff Int Pub, Alphen aan den Rijn, Th Netherlands.

@78# Mossakovskii VI ~1954!, The fundamental mixed problem of th theory of elasticity for a half-space with a circular line separating boundary conditions,Prikl. Mat. Mekh.18, 187–196~in Russian!.

@79# Goodman LE~1962!, Contact stress analysis of normally loade rough spheres,ASME J. Appl. Mech.29, 515–522.

@80# Mossakovskii VI~1963!, Compression of elastic bodies under cond tions of adhesion~axisymmetric case!, J. Appl. Math. Mech.27, 630– 643.

@81# Spence DA~1968!, Self similar solutions to adhesive contact pro lems with incremental loading,Proc. R. Soc. London, Ser. AA305, 55–80.

@82# Spence DA~1975!, The Hertz contact problem with finite friction,J. Elast.5, 297–319.

@83# Dundurs J, and Comninou M~1979!, Some consequences of the in equality conditions in contact and crack problems,J. Elast.9, 71–82.

@84# Irwin GR ~1957!, Analysis of stresses and strains near the end o crack traversing a plate,ASME J. Appl. Mech.24, 361–364.

@85# Keating RF, and Sinclair GB~1995!, On the fundamental energy ar gument of elastic fracture mechanics,Int. J. Fract.74, 43–61.

@86# Eshelby JD~1951!, The force on an elastic singularity,Philos. Trans. R. Soc. London, Ser. AA244, 87–112.

@87# Sanders Jr JL~1960!, On the Griffith-Irwin fracture theory,ASME J. Appl. Mech.27, 352–353.

@88# Rice JR ~1968!, A path independent integral and the approxima analysis of strain concentration by notches and cracks,ASME J. Appl. Mech.35, 379–386.

@89# Irwin GR ~1948!, Fracture dynamics,Fracturing of Metals, Am Soc for Metals, Cleveland OH, 147–166.

@90# Orowan E ~1949!, Fracture and strength of solids,Reports on Progress in Physics12, 185–232.

@91# England AH ~1965!, A crack between dissimilar media,ASME J. Appl. Mech.32, 400–402.

@92# Comninou M~1977!, The interface crack,ASME J. Appl. Mech.44, 631–636.

@93# Sinclair GB~1980!, On the stress singularity at an interface crack,Int. J. Fract. 16, 111–119.

@94# Atkinson C ~1977!, On stress singularities and interfaces in line elastic fracture mechanics,Int. J. Fract.13, 807–820.

@95# He M-Y, and Hutchinson JW~1989!, Kinking of a crack out of an interface,ASME J. Appl. Mech.56, 270–278.

@96# Suo Z, and Hutchinson JW~1989!, Sandwich test specimens for mea suring interface crack toughness,Mater. Sci. Eng., AA107, 135–143.

@97# Malyshev BM, and Salganik RL~1965!, The strength of adhesive joints using the theory of cracks,Int. J. Fract. Mech.1, 114–128.

@98# Comninou M ~1990!, An overview of interface cracks,Eng. Fract. Mech.37, 197–208.

@99# Knowles JK, and Pucik TA~1973!, Uniqueness for plane crack prob lems in linear elastostatics,J. Elast.39, 223–236.

@100# Irwin GR ~1960!, Fracture mechanics,Structural Mechanics, Proc of 1st Symp on Naval Structural MechanicsJN Goodier and NJ Hoff ~eds!, Pergamon Press, Oxford, UK, 557–591.

@101# Rice JR~1968!, Mathematical analysis in the mechanics of fractu Fracture, An Advanced TreatiseH Liebowitz ~ed!, Vol II, Academic Press, New York, NY 191–311.

@102# Erdogan F, and Sih GC~1963!, On the crack extension in plates und plane loading and transverse shear,ASME J. Basic Eng.85, 519–527.

@103# Rice JR ~1988!, Elastic fracture mechanics concepts for interfac cracks,ASME J. Appl. Mech.55, 98–103.

@104# Hutchinson JW, and Suo Z~1991!, Mixed mode cracking in layered materials,Adv. Appl. Mech.29, 63–191.

,

,

g

6 g

-

r

f

f s

f s

A

c

c

r

ic ity,

ri-

ss

es

ta-

er-

r-

ox

e-

ics,

r-

ic

r-

- -

n-

- e-

n-

u- -

n-

in

h

han-

296 Sinclair: Stress singularities in classical elasticity–I Appl Mech Rev vol 57, no 4, July 2004

@105# Rooke DP, and Cartwright DJ~1976!, Compendium of Stress Intensit Factors, Hillingdon Press, Uxbridge, Middlesex, UK.

@106# Sih GC ~1973!, Handbook of Stress Intensity Factors, Vol 1, Inst of Fracture and Solid Mechanics, Lehigh Univ, Bethlehem PA.

@107# Murakami Y, Aoki S, Hasebe N, Itoh Y, Miyata H, Miyazaki N Terada H, Tohgo K, Toya M, and Yuuki R~1987!, Stress Intensity Factors Handbook, Vol 1, Pergamon Press, Oxford, UK.

@108# Murakami Y, Aoki S, Hasebe N, Itoh Y, Miyata H, Miyazaki N Terada H, Tohgo K, Toya M, and Yuuki R~1990!, Stress Intensity Factors Handbook, Vol 2, Revised Edition, Pergamon Press, Oxfor UK.

@109# Murakami Y, Hanson MT, Hasebe N, Itoh Y, Kishimoto K, Miyata H Miyazaki N, Terada H, Tohgo K, and Yuuki R~1992!, Stress Intensity Factors Handbook, Vol 3, Pergamon Press, Oxford, UK.

@110# Srawley JE, and Brown Jr, WF~1965!, Fracture toughness testin methods,Fracture Toughness Testing and Its Applications, STP No 381, Am Soc for Testing and Materials, Philadelphia, PA 133–19

@111# ASTM ~1998!, Standard test method for plane-strain fracture tou ness of metallic materials, E399-90~reapproved 1997!, 1998 Annual Book of ASTM Standards, Vol 3.01, Am Soc for Testing and Materi als, Philadelphia PA, 413–443.

@112# Heyer RH, and McCabe DE~1970! Evaluation of a method of test fo plane strain fracture toughness using a bend specimen,Review of Developments in Plane Strain Fracture Toughness Testing, STP No 463, Am Soc for Testing and Materials, Philadelphia PA, 22–41.

@113# McCabe DE~1972!, Evaluation of the compact tension specimen f determining plane strain fracture toughness of high strength mater J. Mater.7, 449–454.

@114# Underwood JH, and Kendall DP~1978!, Cooperative plane strain fracture toughness tests with C-shaped specimens,J. Test. Eval.6, 296–300.

@115# Hudson CM, and Seward SK~1978!, A compendium of sources o fracture toughness and fatigue-crack growth data for metallic allo Int. J. Fract.14, R151–R184.

@116# Hudson CM, and Seward SK~1982!, A compendium of sources o fracture toughness and fatigue-crack growth data for metallic alloy Part II, Int. J. Fract.20, R59–R117.

@117# Hudson CM, and Seward SK~1989!, A compendium of sources o fracture toughness and fatigue-crack growth data for metallic alloy Part III, Int. J. Fract.39, R43–R63.

@118# Hoysan SF, and Sinclair GB~1993!, On the variability of fracture toughness,Int. J. Fract.60, R43–R49.

@119# Sinclair GB, and Chambers AE~1987!, Strength size effects and frac ture mechanics: What does the physical evidence say?Eng. Fract. Mech.26, 279–310.

@120# Sinclair GB, and Pieri RV~1990!, On obtaining fatigue crack growth parameters from the literature,Int. J. Fatigue12, 57–62.

@121# Chamis CC~1969!, Failure criteria for filamentary composites, NAS Tech Note D-5367, NASA, Washington DC.

@122# Valliappan S, Kjellberg S, and Svensson NL~1980!, Finite element analysis of total hip prosthesis,Proc of Int Conf on Finite Elements in Biomechanics, Tucson AZ, Vol 2, 527–548.

@123# Belie RG, and Reddy JN~1980!, Direct prediction of fracture for two-dimensional plane stress structures,Comput. Struct.11, 49–53.

@124# Kim YJ, and Hsu TR~1982!, A numerical analysis on stable crac growth under increasing load,Int. J. Fract.20, 17–32.

@125# Chen C-N~1993!, Nonlinear fracture assessment by using the fin element method,Eng. Fract. Mech.46, 57–77.

@126# Wells AA ~1961!, Unstable crack propagation in metals—Cleava and fast fracture,Proc of the Crack Propagation Symp, Cranfield, UK, Vol 1, 210–230.

@127# Andersson H~1973!, A finite-element representation of stable crac growth,J. Mech. Phys. Solids21, 337–356.

@128# British Standards Inst~1979!, Methods for crack opening displace ment ~COD! testing, BS 5762, British Standards Inst, London, UK

@129# ASTM ~1998!, Standard test method for crack-tip opening displa ment ~CTOD! fracture toughness measurement, E1290-93,1998 An- nual Book of ASTM Standards, Vol. 3.01, Am Soc for Testing and Materials, Philadelphia, PA, 814–823.

@130# Cottrell AH ~1961!, Theoretical aspects of radiation damage a brittle fracture in steels,Steels for Reactor Pressure Circuits, Special Report No 69, Iron and Steel Inst, London, UK, 281–296.

@131# Burdekin FM ~1981!, Assessment of defects: The C.O.D. approa Philos. Trans. R. Soc. London, Ser. AA299, 155–167.

@132# Sinclair GB ~1985!, Some inherently unreliable practices in prese day fracture mechanics,Int. J. Fract.28, 3–16.

@133# Williams ML ~1951!, Surface stress singularities resulting from va ous boundary conditions in angular corners of plates under bend

y

d,

,

. h-

or ials,

ys,

-

k

ite

ge

k-

- . e-

nd

h,

nt

i- ing,

Proc of 1st US Natl Congress of Appl Mech, Illinois Inst of Tech, Chicago IL, 325–329.

@134# Kitover KA ~1952!, On the use of special systems of biharmon functions for the solution of some problems in the theory of elastic Prikl. Mat. Mekh.16, 739–748~in Russian!.

@135# Huth JH ~1953!, The complex-variable approach to stress singula ties,ASME J. Appl. Mech.20, 561–562.

@136# Williams ML ~1956!, The complex-variable approach to stre singularities—II,ASME J. Appl. Mech.23, 477–478.

@137# Bogy DB ~1968!, Edge-bonded dissimilar orthogonal elastic wedg under normal and shear loading,ASME J. Appl. Mech.35, 460–466.

@138# Tranter CJ~1948!, The use of the Mellin transform in finding the stress distribution in an infinite wedge,Q. J. Mech. Appl. Math.1, 125–130.

@139# Coker EG, and Filon LNG~1931!, A Treatise on Photo-Elasticity, Cambridge Univ Press, UK.

@140# Williams ML ~1957!, On the stress distribution at the base of a s tionary crack,ASME J. Appl. Mech.24, 109–114.

@141# Bogy DB ~1970!, On the problem of edge-bonded elastic quart planes loaded at the boundary,Int. J. Solids Struct.6, 1287–1313.

@142# Bogy DB, and Wang KC~1971!, Stress singularities at interface co ners in bonded dissimilar isotropic elastic materials,Int. J. Solids Struct.7, 993–1005.

@143# Dempsey JP~1981!, The wedge subjected to tractions: A parad resolved,J. Elast.11, 1–10.

@144# Ting TCT ~1984!, The wedge subjected to tractions: A paradox r examined,J. Elast.14, 235–247.

@145# Ting TCT ~1996!, Paradoxes puzzles, and dilemmas in mechan Chinese J. Mech.12, 25–32.

@146# Sinclair GB ~1980!, On the singular eigenfunctions for plane ha monic problems in composite regions,ASME J. Appl. Mech.47, 87– 92.

@147# Ting TCT ~1985!, Asymptotic solution near the apex of an elast wedge with curved boundaries,Q. Appl. Math.42, 467–476.

@148# Atkinson C~1979!, Stress singularities and fracture mechanics,Appl. Mech. Rev.32, 123–135.

@149# Hwang KC, Yu SW, and Yang W~1990!, Theoretical study of crack- tip singularity fields in China,Appl. Mech. Rev.43, 19–33.

@150# Murakami Y ~1992!, Stress singularity for notch at bimaterial inte face,Stress Intensity Factors Handbook, Vol 3, Murakamiet al Per- gamon Press, Oxford, UK, Ch 18, 963–1062.

@151# Sinclair GB ~1998!, FEA of singular elasticity problems,Proc of 8th Int ANSYS Conf, Pittsburgh, PA, Vol 1, 225–236.

@152# Westergaard HM~1939!, Bearing pressures and cracks,ASME J. Appl. Mech.6, A-49–A-53.

@153# Gallagher RH~1978!, A review of finite element techniques in frac ture mechanics,Proc of 1st Int Conf on Numerical Methods in Frac ture Mechanics, AR Luxmoore and DRJ Owen~eds!, Univ of Wales, Swansea UK, 1–25.

@154# Luxmoore AR, and Owen DRJ~eds! ~1978!, Proceedings of the First International Conference on Numerical Methods in Fracture Mecha ics, Univ of Wales, Swansea, UK.

@155# Luxmoore AR, and Owen DRJ~eds! ~1980!, Proceedings of the Sec ond International Conference on Numerical Methods in Fracture M chanics, Univ of Wales, Swansea, UK.

@156# Luxmoore AR, and Owen DRJ~eds! ~1984!, Proceedings of the Third International Conference on Numerical Methods in Fracture Mecha ics, Univ of Wales, Swansea, UK.

@157# Luxmoore AR, Owen DRJ, Rajapakse YPS, and Kanninen MF~eds! ~1987!, Proceedings of the Fourth International Conference on N merical Methods in Fracture Mechanics, Southwest Research Insti tute, San Antonio, TX.

@158# Luxmoore AR, and Owen DRJ~eds! ~1990!, Proceedings of the Fifth International Conference on Numerical Methods in Fracture Mecha ics, FhG Inst für Werkstoffmechnik, Freiburg, Germany.

@159# Henshell RD, and Shaw KG~1975!, Crack tip finite elements are unnecessary,Int. J. Numer. Methods Eng.9, 495–507.

@160# Barsoum RS~1976!, On the use of isoparametric finite elements linear fracture mechanics,Int. J. Numer. Methods Eng.10, 25–37.

@161# Wait R ~1978!, Finite element methods for elliptic problems wit singularities,Comput. Methods Appl. Mech. Eng.13, 141–150.

@162# Lim IL, Johnston IW, and Choi SK~1993!, Application of singular quadratic distorted isoparametric elements in linear fracture mec ics, Int. J. Numer. Methods Eng.36, 2473–2499.

@163# Meda G, and Sinclair GB~1994!, On the use of the H-integral to extract stress intensity factors,Proc of 6th Int ANSYS Conf, Pittsburgh PA, Vol 2, 6.39–6.60.

@164# Irwin GR ~1958!, Fracture,Handbuch der Physik, Springer-Verlag Ltd, Berlin, Germany, Vol VI, 551–590.

t

n

i

y

t s

- t,

r

s

2D

c le-

s,

Appl Mech Rev vol 57, no 4, July 2004 Sinclair: Stress singularities in classical elasticity–I 297

@165# Cooper DB, Meda G, and Sinclair GB~1995!, A comparison of crack- flank displacement fitting for estimatingK with a path independen integral,Int. J. Fract.70, 237–251.

@166# Chan SK, Tuba IS, and Wilson WK~1970!, On the finite element method in linear fracture mechanics,Eng. Fract. Mech.2, 1–17.

@167# Parks DM~1974!, A stiffness derivative finite element technique fo determination of crack tip stress intensity factors,Int. J. Fract. 10, 487–502.

@168# Rybicki EF, and Kanninen MF~1977!, A finite element calculation of stress intensity factors by a modified crack closure integral,Eng. Fract. Mech.9, 931–938.

@169# Stern M~1973!, A boundary integral representation for stress intens factors,Proc of 10th Anniversary Meeting of the Soc of Engineeri Science, Raleigh, NC, 125–132.

@170# Stern M, Becker EB, and Dunham RS~1976!, A contour integral computation of mixed-mode stress intensity factors,Int. J. Fract.12, 359–368.

@171# Stern M, and Soni ML~1976!, On the computation of stress intens ties at fixed-free corners,Int. J. Solids Struct.12, 331–337.

@172# Hong C-C, and Stern M~1978!, The computation of stress intensit factors in dissimilar materials,J. Elast.8, 21–34.

@173# Carpenter WC~1984!, Calculation of fracture mechanics paramete for a general corner,Int. J. Fract.24, 45–58.

@174# Sinclair GB, Okajima M, and Griffin JH~1984!, Path independen integrals for computing stress intensity factors at sharp notche

r

ity g

-

rs

in

elastic plates,Int. J. Numer. Methods Eng.20, 999–1008~see also ~1985!, Int. J. Fract.27, R81–R85!.

@175# Okajima M, and Sinclair GB~1986!, The C-integral: A path indepen dent integral for computing singularity participation at a butt join Proc of Int Conf on Computational Mech, Tokyo, Japan, Vol 1, V11– V16.

@176# Carpenter WC, and Byers C~1987!, A path independent integral fo computing stress intensities for V-notched cracks in a bi-material,Int. J. Fract. 35, 245–268.

@177# Banks-Sills L~1997!, A conservative integral for determining stres intensity factors of a bimaterial strip,Int. J. Fract.86, 385–398.

@178# Banks-Sills L, and Sherman D~1986!, Comparison of methods for calculating stress intensity factors with quarter-point elements,Int. J. Fract. 32, 127–140.

@179# Pang HLJ~1993!, Linear elastic fracture mechanics benchmarks: finite element test cases,Eng. Fract. Mech.44, 741–751.

@180# Pang HLJ, and Leggat RH~1990!, 2D test cases in linear elasti fracture mechanics, Report R0020, National Agency for Finite E ment Methods and Standards, Glasgow, UK.

@181# Banks-Sills L ~1991!, Application of the finite element method to linear elastic fracture mechanics,Appl. Mech. Rev.44, 447–461.

@182# Meda G, Messner TW, Sinclair GB, and Solecki JS~1996!, Path- independentH integrals for three-dimensional fracture mechanic Int. J. Fract.94, 217–234.

New ce in

e has ellon Pro- han- roving d me- ement

Glenn Sinclair received his undergraduate education from the University of Auckland in Zealand. He graduated with a BS in Mathematics in 1967 and a BE in Engineering Scien 1969. He then attended Caltech, graduating with a PhD in Applied Mechanics in 1972. H since served on the faculty of Yale University, the University of Auckland, Carnegie M University, and Louisiana State University, where he is currently the Francis S Blummer fessor of Mechanical Engineering. His research is primarily concerned with fracture mec ics, tribology, and numerical methods. Recent interests focus on finding means of imp modeling so that stress and pressure singularities are removed from both solid and flui chanics problems, and submodeling procedures and verification techniques for finite el analysis.

__MACOSX/stress/._Stress Singularities in Classical Elasticity I.pdf

stress/Stress Singularities in Classical Elasticity II.pdf

uch

Stress singularities in classical elasticity—II: Asymptotic identification

GB Sinclair Department of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803-6413; [email protected]

This review article~Part II! is a sequel to an earlier one~Part I! that dealt with means of re- moval and interpretation of stress singularities in elasticity, as well as their asymptotic and numerical analysis. It reviews contributions to the literature that have actually effected asymptotic identifications of possible stress singularities for specific configurations. For the most part, attention is focused on 2D elastostatic configurations with constituent materials be- ing homogeneous and isotropic. For such configurations, the following types of stress singu- larity are identified: power singularities with both real and complex exponents, logarithmic intensification of power singularities with real exponents, pure logarithmic singularities, and log-squared singularities. These identifications are reviewed for the in-plane loading of angular elastic plates comprised of a single material in Section 2, and for such plates comprised of multiple materials in Section 3. In Section 4, singularity identifications are examined for the out-of-plane shear of elastic wedges comprised of single and multiple materials, and for the out-of-plane bending of elastic plates within the context of classical and higher-order theory. A review of stress singularities identified for other geometries is given in Section 5, axisymmet- ric and 3D configurations being considered. A limited examination of the stress singularities identified for other field equations is given as well in Section 5. The paper closes with an overview of the status of singularity identification within elasticity. This Part II of the review has 227 references.@DOI: 10.1115/1.1767846#

1 INTRODUCTION

This article is a sequel to another one on stress singularities in classical elastostatics which considers their removal, inter- pretation, and analysis~Sinclair @1#—hereinafter referred to simply as Part I!. Both papers share the recognition that it is an exercise in futility to perform a stress analysis without appreciating the presence of a singularity when one occurs. In Part I, some methods for determining when a singularity is present, and possibilities for dealing with it when it is, are drawn from the literature and discussed. Here, in Part II, the literature is reviewed for contributions that have actually ef- fected determinations of when singularities may occur.

The means by which these determinations are made is asymptotic identification. It is therefore necessary, if Part II is to be fairly self-contained, that we recap key results at- tending the asymptotic identification of stress singularities. These are available in the literature and a description of their development is given in Part I, Sections 4.1 and 4.2. The particular approach considered in some detail there is via the Airy stress function and separation of variables~after Will- iams @2#!: There are other approaches which can lead to the same results~complex potentials, Mellin transforms!.

To fix ideas, we consider an angular elastic plate in exten-

sion ~Fig. 1!. The basic separable fields used to analyze s plates are given in Williams@2# and Part I, Section 4.1. In terms of cylindrical polar coordinatesr and u, the stresses s r , su , andt ru and displacementsur anduu in these fields are

s r52lr l21@c1 cos~l11!u1c2 sin~l11!u

1~l23!~c3 cos~l21!u1c4 sin~l21!u!#

su5lr l21@c1 cos~l11!u1c2 sin~l11!u

1~l11!~c3 cos~l21!u1c4 sin~l21!u!#

t ru5lr l21@c1 sin~l11!u2c2 cos~l11!u

1~l21!~c3 sin~l21!u2c4 cos~l21!u!# (1.1)

ur5 2r l

2m @c1 cos~l11!u1c2 sin~l11!u

1~l2k!~c3 cos~l21!u1c4 sin~l21!u!#

uu5 r l

2m @c1 sin~l11!u2c2 cos~l11!u

1~l1k!~c3 sin~l21!u2c4 cos~l21!u!#

Transmitted by Editorial Advisory Board member R. C. Benson

Appl Mech Rev vol 57, no 5, September 2004 38

© 2004 American Society of Mechanical Engineers5

he

a

e- lar s

t

386 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Fig. 1 Geometry and coordinates for the angular elastic pla

s

n

e t

e

c

- s

ard

er I,

m

e-

o- vial

in

y be

of xil- lds

e

em

ions

on-

for n51,..,nA2r A and l51 with

ĉ1 21 ĉ2

21 ĉ2 2Þ0 in the stress field attending~1.2!

s5O~cos~h ln r !!1O~sin~h ln r !! when D50

for complex l511 i h

Herein, l has taken on the role of an eigenvalue of t asymptotic problem,nA is the order of the matrixA, andr A

is its rank whenl is an eigenvalue. For a plate made of single material,nA54 at most; for a bimaterial plate,nA

58, and so on. The last stress in~1.3! is not singular asr →0, being bounded under this limit. However, it is und fined asr→0. Hence, to a degree, it shares with singu stresses some of the difficulties attending interpretation ar →0.

The conditions in~1.3! apply to angular plates in exten sion. Adaptation of~1.3! to states of antiplane shear follow directly ~see Sections 4.1 and 4.2!. Adaptation of~1.3! to bending is less direct but nonetheless fairly straightforw ~see Sections 4.3 and 4.4!. Adaptation of~1.3! to other con- figurations is discussed in Section 5.

With inhomogeneous boundary conditions, further auxil- iary fields can participate. These fields follow from a furth differentiation with respect tol; stresses are given in Part Section 4.2. By way of example, thesu stress component in these fields is

su5r l21@~l ln2 r 12 ln r 2lu2!~ c̃1 cos~l11!u

1 c̃2 sin~l11!u!1~l11!~ c̃3 cos~l21!u

1 c̃4 sin~l21!u!1O~ ln r !1O~1!# (1.4)

asr→0. In ~1.4!, tildes atop constants distinguish them fro those of~1.1! or ~1.2!. All three sorts of field in concert lead to the following set of conditions for thesingular stresses that are possible with uniform tractions/ linear displac ments applied. For any stress components, asr→0:

s5ord~ ln2 r !1ord~ ln r ! when D50, ]nD

]ln 50

for n51,..,nA2r A with

c̃1 21 c̃2

21 c̃3 2Þ0 in the stresses attending~1.4!

s5ord~ ln r ! when D50, ]nD

]ln 50, for n51,.., nA2r A

(1.5) with c̃15 c̃25 c̃350 in the stresses attending~1.4!

s5ord~ ln r ! when D50, ]nD

]ln Þ0, for n5nA2r A

with ĉ1 21 ĉ2

21 ĉ3 2Þ0 in the stresses attending~1.2!

provided throughout~1.5!, l51 andr AÞr A8 , wherer A8 is the rank of the augmented matrix formed by combiningA with the nontrivial forcing vector associated with the inh mogeneous boundary conditions. Given such a nontri vector, the singularities in~1.5! occur irrespective of far-field boundary conditions. Hence the use of the ord notation

e

In ~1.1!, m is the shear modulus andk equals 324n for plane strain and (32n)/(11n) for plane stress,n being Poisson’s ratio. Further,ci ( i 51,...,4) are constants andl is the separation-of-variables parameter. This parameter ma complex. Then the real and imaginary parts of~1.1! each constitute acceptable fields which may have distinct set constants from one another. It is also possible to have au iary fields participate in the asymptotic analysis. These fie can be generated by differentiating with respect tol as in Dempsey and Sinclair@3#, and are given in Part I, Sectio 4.2. By way of example, thesu stress component in thes fields is

su5r l21@~11l ln r !~ ĉ1 cos~l11!u1 ĉ2 sin~l11!u!

1~2l111l~l11!ln r !~ ĉ3 cos~l21!u

1 ĉ4 sin~l21!u!2lu~ ĉ1 sin~l11!u2 ĉ2 cos~l11!u

1~l11!~ ĉ3 sin~l21!u2 ĉ4 cos~l21!u!!# (1.2)

In ~1.2!, the carets atop constants serve to distinguish th from those of~1.1!

Introducing the fields in~1.1! into a set of fourhomoge- neous boundary conditionsholding on the two edges of th angular plate results in a homogeneous system of equa in the four constantsci . Call the coefficient matrix of this systemA and its determinantD. Then, also entertaining th possibility of the participation of the fields attending~1.2! leads to the following set of conditions for thesingular stresses that are possible with homogeneous boundary ditions. For any stress components, asr→0:

s5O~r j21 cos~h ln r !!1O~r j21 sin~h ln r !! when D50

for complex l5j1 ih~0,j,1!

s5O~r l21 ln r !1O~r l21! when D50, ]nD

]ln 50

for n51, .., nA2r A and real l~0,l,1!

s5O~r l21! when D50 for real l~0,l,1! (1.3)

s5O~ ln r ! when D50, ]nD

]ln 50

s s

m u t

t n

m

h

p

of e

-

ped t to ndi-

lip ated

( d

first r by e ul ates ing ke ap-

ous In

the

s

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 387

~1.5! instead of the large orderO notation of ~1.3!.1 The singularity conditions in~1.5! apply directly to angular plate in extension: Adaptation to other configurations is discus in Sections 4 and 5.

In what follows, we review asymptotic analyses that e ploy ~1.3!, ~1.5!, or their equivalents to identify stress sing larities. We begin in Section 2 with angular elastic pla made of a single material under in-plane loading~ie, in ex- tension!. In Section 3, we review the singularities identifie when such plates are made of multiple materials. In Sec 4, we consider out-of-plane shear and bending. In Sectio we consider a variety of other circumstances: axisymme and 3D configurations within classical elasticity, and a li ited review of the effects of other field equations. Finally, Section 6, we close with some remarks on the general c acter of results, and the overall state of investigations i singularity identification.

2 STRESS SINGULARITIES FOR THE IN-PLANE LOADING OF AN ELASTIC PLATE MADE OF A SINGLE MATERIAL

2.1 Formulation and eigenvalue equations

Here we obtain the eigenvalue equations governing the sible stress singularities that can occur at the vertex of angular elastic plate subjected to different homogene boundary conditions on its edges.

To formally state the class of problems under consid ation, we continue to employ cylindrical polar coordinatesr andu with origin O at the plate vertex~Fig. 1!. In terms of these coordinates, the open angular region of interestR is given by

R5$~r ,u!u0,r ,`, 0,u,f% (2.1)

where f is the angle subtended at the vertex of the pla With these geometric preliminaries in place, we can form late our class of problems as follows.

In general, we seek the planar stress componentss r , su , and t ru and their companion displacementsur and uu , as functions of r and u throughoutR, satisfying: thestress equations of equilibriumin the absence of body forces,

]s r

]r 1

1

r

]t ru

]u 1

s r2su

r 50

(2.2) 1

r

]su

]u 1

]t ru

]r 1

2t ru

r 50

on R; the stress-displacement relationsfor a linear elastic plate which is both homogeneous and isotropic,

H s r

su J 5mF 2Q

k21 H 1

2J S ]ur

]r 2

1

r

]uu

]u 2

ur

r D G (2.3)

t ru5mF1

r

]ur

]u 1

]uu

]r 2

uu

r G

e

ed

- -

es

d ion

5, tric

- in ar-

nto

os- an

ous

er-

te. u-

with

Q5 ]ur

]r 1

1

r

]uu

]u 1

ur

r (2.4)

on R, whereinQ is the dilatation whilem continues as the shear modulus andk continues to equal 324n for plane strain and (32n)/(11n) for plane stress,n being Poisson’s ratio; any one of the admissible sets ofboundary conditions listed in Table 1 on the plate edge atu50, together with another such set on the edge atu5f or bisector atu5f/2 as appropriate, for 0,r ,`; and theregularity requirementsat the plate vertex,

ur5O~1!, uu5O~1!, as r→0 (2.5)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th plate vertexO.

The boundary conditionsof Table 1 merit some discus sion. Conditions I and II apply onu50 or f and are the classical conditions for a stress-free surface and one clam to a rigid attachment. The clamped conditions also admi interpretation as the homogeneous complement to the co tions attending indentation by a rigid punch with no s permitted. Such indentation is also sometimes associ with a ‘‘rough’’ or ‘‘adhesive’’ punch in the literature.

When the same conditions apply on both plate edgesu 50, f), it is useful to distinguish between symmetric an antisymmetric response about the plate bisector. In the instance, it is useful because the analysis can be easie virtue of leading to a 232 determinant for the eigenvalu equation instead of a 434. In the second instance, it is usef because it can restrict the number of singular stress st possible in a given global configuration before undertak its global analysis. Conditions III and IV enable one to ma this distinction. For the present plate configuration, they ply on u5f/2 when used in this role.2

Conditions III can also be interpreted as the homogene complement to indentation by a frictionless rigid punch. this role, they apply onu50 or f and are usually adjoined with the condition that the normal stress not be tensile in contact region. That is,

su<0 (2.6)

Table 1. Homogeneous boundary conditions for in-plane loading

Identifying Roman numeral

Boundary conditions

Physical description

I su50, t ru50 Stress free II uu50, ur50 Clamped III uu50, t ru50 Symmetry IV ur50, su50 Antisymmetry V uu50, t ru5 f su Contact with friction VI su5kuu , t ru5k8ur Cohesive stress-separation law

1A definition of ord is given in Part I, Section 1.2. The essential difference between andO is that, with the former, the coefficient of the related singularity cannot be z whereas with the latter it can.

ord ro,

2With symmetry,ur is an even function ofu aboutf/2, uu on odd: with antisymmetry, vice versa. Hence, on drawing on~2.3!, ~2.4!, the boundary conditions given in III and IV.

t

h

s

h

l

n

n

, v t

by that y of eld

any g in

le -

and lds un- the

to ro- is is ntly t is, can

ad- s in

lds re- y

ave ary ll- s

ue lds

ob- m

388 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

on u50 or f, for 0,r ,`. The indentor shape can lead further constraints outside the contact region to prevent terpenetration.

Conditions IV can also be interpreted as those for a t stiff reinforcement~Rao @4#!. The reinforcement is suffi- ciently relatively stiff to prevent extension (ur50), but not so stiff as to prevent bending because of its thinnessuu

Þ0). Conditions V extend the contact conditions of III to pe

mit finite friction via Amonton’s law.3 Herein f has the mag- nitude of the coefficient of friction. For these conditions, addition to seeking to apply the contact constraint~2.6! and any external displacement constraints, we must try to en that the shear stress opposes any slipping. This may be sible by selecting the sign off appropriately.

Conditions VI apply cohesive stress-separation laws. T k and k8 are the stiffnesses associated with relative tra verse and lateral displacements between material on the sides of the ray on which the conditions are applied. Wh applied onu50 in Fig. 1, bothk andk8 are positive: onu 5f, negative. In some instances it may be possible to one or the other of these stiffnesses to zero. For examplek8 can be taken as zero when loading is symmetric. In contr if k and k8 are let tend to infinity, Conditions II are recov ered. In general,k and k8 are of constant magnitude in th elastic regime and should both be consistent with the ela moduli of the surrounding continuum.

Conditions VI can also be interpreted as those for a p on an elastic foundation. Usually thenk8 is taken as zero giving Winkler conditions~Winkler @6#; Oravas@7# has that these conditions were given earlier in Euler@8#!.

In either role, Conditions VI differ from the others i Table 1 in that a single boundary condition involves both stress and a displacement. Such mixed boundary condit would seem to be fairly rare in elasticity. One further i stance occurs for the elastic angular plate reinforced b beam column—see Nuller@9#.

All of the foregoing boundary conditions are applie along radial rays emanating from the plate vertex. That is straight boundaries. If instead they are applied on cur boundaries that smoothly make tangents to the straight a vertex, the same singular eigenvalues can be expected. C panion eigenfunctions differ, however. See Ting@10#.

Some further comments on the preceding formulation also appropriate. First, regarding the absencerequirements at infinity on R. This renders fields complying with our formu lation nonunique. Since the principal attribute of these fie is the characterization of all possible responses at the p vertex, including especially all possible stress singularit there, such a lack of uniqueness is to be desired rather regulated against. In any configuration offinite extent locally containing one of the configurations admitted by our form lation, conditions on the other boundaries in the finite geo etry should make its solution unique.

Second, regardingdimensions. There is no length scale in

o in-

in

(

r-

in

ure pos-

us ns- two en

set , ast, - e stic

ate

a ions -

y a

d on ed the om-

are

- lds late ies than

u- m-

the problems formulated. This is also something provided associated finite problems. However, this does not mean there need be concern as to the dimensional consistenc the asymptotic analysis. To see this, observe that the fi equations~2.2!, ~2.3!, and ~2.4! are equidimensional inr , and thatur anduu occur ‘‘divided’’ by r . Thusr , ur , anduu

can be replaced byr /L, ur /L, and uu /L, whereL is any length scale, and leave the equations unchanged. Hence asymptotic solutions obtained can be regarded as bein terms ofr /L and thereby made dimensionally consistent.

Third, regarding material constants. These are con- strained to the physically applicable ranges, 0,m,` and 0<n<1/2. However, for plane strain with an incompressib material (n51/2), we have k51 and the stress displacement relations of~2.3! and ~2.4! are no longer di- rectly applicable. Under these circumstances we requireQ 50 in ~2.4! and modify~2.3! by removing theQ terms.

Fourth, regarding theregularity requirements, ~2.5!. These ensure bounded displacements at the plate vertex bounded forces on rays radiating from the vertex. Such fie definitely appeal as being more physical than those with bounded displacements or forces. This, though, is not reason for~2.5!. If physical appropriateness in itself were serve as sufficient justification, then we would want to p hibit unbounded stresses as well. We cannot do this. Th because then the formulation does not admit a sufficie broad class of fields to enable its solution in general: Tha the fields so admitted are incomplete. In contrast, we prohibit unbounded displacements because the fields so mitted are complete. This is explicitly shown for Condition I, and indicated for the remaining conditions in Table 1, Gregory@11#.4

For problems wherein the completeness of elastic fie with bounded displacements holds true, the regularity quirements of~2.5! are not just a nice option. Rather, the are essential if any companion finite problems are to h unique solutions. To explain further, consider the element problem of a circular elastic plate of unit radius under a around uniform pressurep. Absent regularity requirement as r→0, two solutions are possible:

s r5su52p, ur52 pr

4m ~k21! (2.7)

or

s r52su52 p

r 2 , ur5 p

2mr (2.8)

Requiring bounded displacements eliminates~2.8! and ren- ders the problem well posed by making it have a uniq solution. Analogously, uniqueness for singular stress fie with bounded displacements in completely formulated pr lems for finite regions occurs: The proof of this follows fro

4From the abstract and introduction in Gregory@11#, one might think that the original Williams’ eigenfunctions are complete for the boundary conditions in Williams@2#. As is demonstrated in Section 2.3, this is not so. Further reading of@11#, though, reveals that it recognizes the need to supplement the fields of~1.1! with those attending~1.2! for completeness.

3Also termed Coulomb’s law in the literature. See Ch 13, Johnson@5# for conditions under which there is some physical support for the use of this law.

x-

s th

in g

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 389

Table 2. Eigenvalue equations for symmetric response aboutuÄfÕ2

Boundary conditions on uÄ0,f

Eigenvalue equation

Equation number

I or VI-I or VI l sinf52sinlf ~2.9! II-II l sinf5k sinlf ~2.10! III-III cos f5coslf ~2.11! V-V f @(12k)sinlf1(11k12l)sinf#

5(11k)(coslf2cosf) ~2.12!

u

e

ac- e

tive

ess- tions

in ons

a e

the rise

ues

e to a

same sso- e

of

in- lly. w with s, be

ter-

t

sual

ce- as na- ms

an n- own f a

ts. in

u- in

rt I, t in ed are

re,

contact,~2.11! and ~2.15! in Tables 2 and 3, are given in Kalandiia@19#. Equations for these conditions which are e actly the same as~2.11! and~2.15! are provided in Seweryn and Molski @20#. Two further frictionless contact equation for such conditions in combination with free and wi clamped conditions can be obtained by settingf 50 in ~2.18! and ~2.19! of Table 4, respectively: These two are given Kalandiia@19#. The equation for contact with friction actin with itself symmetrically,~2.12!, would not appear to be readily available in the literature; the equation when this tion is antisymmetric,~2.16!, is essentially the same as th corresponding equation in Dempsey@21#. The contact with friction-free equation of~2.18! in Table 4 can be obtained from Gdoutos and Theocaris@22#. It follows on setting ‘‘G2’’ in @22# to infinity to reflect a rigid punch, and ‘‘q’ ’ 52 f because the friction conditions therein hold on a nega u-edge. The contact with friction-clamped equation of~2.19! in Table 4 is essentially given in Dempsey@21#. The equiva- lence of stress-free conditions with those for cohesive str separation laws as far as the foregoing eigenvalue equa are concerned is basically argued in Sinclair@23#.

When Conditions IV are interpreted as being for a th stiff reinforcement, eigenvalue equations for these conditi with others and a plate of vertex anglef/2 are given in Table 4. When Conditions IV act in this role on both edges of plate of vertex anglef, the eigenvalue equation can b formed as a product of~2.11! and ~2.15!.6

2.2 Power singularities with homogeneous boundary conditions

For the homogeneous boundary conditions of Table 1, associated eigenvalue equations of Tables 2–4 can give to stresses with power singularities when their eigenval are less than one—see~1.1!. To be in accordance with the regularity requirements~2.5!, these eigenvalues must not b less than zero. An eigenvalue equal to zero corresponds rigid body displacement in~1.1! and therefore is not of in- terest because associated stresses are not singular: The value leads to unbounded displacements for the fields a ciated with ~1.2! and therefore is not admissible. Thus th eigenvalue range forpower singularitiesis

0,l,1 (2.20)

We review eigenvalues within this range for a variety configurations in this section.

The solution of the eigenvalue equations within the s gular range typically cannot be done completely analytica Accordingly it usually proceeds numerically except for a fe select instances. The results so found are compared those in the literature. For all sources given in what follow they are consistent. Thus their calculation here may viewed as independently confirming values already de mined in the cited sources.

In presenting results we introduce thesingularity expo- nentg defined by

of

the boundedness of attendant strain energies and the Kirchhoff argument~see Knowles and Pucik@12#!.

On occasion, further support for the bounded displa ment conditions derives from solving a singular problem the limit of a sequence of nonsingular problems, the a logue of the approach adopted in concentrated load probl and for generalized functions in general in Lighthill@13#. An example is the plate under uniform remote tension with elliptical hole. As the height of the hole parallel to the te sion goes to zero, the nonsingular stress fields can be sh to recover the inverse-square-root stress singularity o stress-free mathematically-sharp crack~Kolossoff @14,15# and Inglis@16#!. This singularity has bounded displacemen The same is true of other singular configurations realized this way: see, for example, Neuber@17#.

The analysis of the class of asymptotic problems form lated proceeds routinely on using the approach outlined the Introduction here, and described in some detail in Pa Section 4.1. This yields the eigenvalue equations set ou Tables 2, 3, and 4 for symmetric, antisymmetric, and mix configurations, respectively. These eigenvalue equations typically available in the literature as described next: He they are independently derived largely as a check.

The free-free equations~2.9! and~2.13! in Tables 2 and 3, the clamped-clamped equations~2.10! and~2.14! in Tables 2 and 3, and the clamped-free equation~2.17! in Table 4 all effectively appear in Williams@2# and Kitover@18#.5 Equiva- lent equations to those for frictionless contact-frictionle

Table 3. Eigenvalue equations for antisymmetric response aboutu ÄfÕ2

Boundary conditions on uÄ0,f

Eigenvalue equation

Equation number

I or VI-I or VI l sinf5sinlf ~2.13! II-II l sinf52k sinlf ~2.14! III-III cos f52coslf ~2.15! V-V f @(12k)sinlf2(11k12l)sinf#

5(11k)(coslf1cosf) ~2.16!

Table 4. Eigenvalue equations for mixed problems

Boundary conditions on uÄ0,f

Eigenvalue equation

Equation number

I or VI-II 4@k sin2 lf1l2 sin2 f#5(11k)2 ~2.17! I or VI-V 2 f @(12k)sin2 lf2l(11k12l)sin2 f#

5(11k)(sin 2lf1l sin 2f) ~2.18!

II-V 2 f @k(12k)sin2 lf1l(11k12l)sin2 f# 5(11k)(k sin 2lf2l sin 2f)

~2.19!

n- ose in

ss

o

6That is, by rearranging~2.11!, ~2.15! so that they have expressions on one side the5, zero on the other, then setting the product of these expressions50. The so- obtained equation is given in Ro¨ssle@24#. This reference also gives nine further eige value equations. These equations are contained in, and are consistent with, th Tables 2–4.

5The eigenvalue equations in Kitover@18# are correct for plane strain, but appear have typographical errors for plane stress.

as

390 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

g512l (2.21)

Then ~1.1! has stresses which behave in accordance with

s5O~r 2g! as r→0 (2.22)

wheres is any stress component. That is, stresses are singu-

lar for g positive, and the largerg the more singular. The limits on the nature of this power singularity are, from~2.20! and ~2.21!,

0,g,1 (2.23)

In the event thatl is complex, we have stress singularities

Fig. 2 Singularity exponents for varying vertex angles:a! free-free and clamped-clamped~from ~2.9!, ~2.13! and ~2.10!, ~2.14!, respec- tively!, b! frictionless contact-frictionless contact~from ~2.11!, ~2.15!!, c! contact with friction-contact with friction~from ~2.12!, ~2.16!!, d! clamped-free~from ~2.17!!, e! contact-free~from ~2.18!!, f! contact-clamped~from ~2.19!!

. he- ray

ffec-

tric

e

- o

for

-

ed- if

ary ric

an- ot r re- r. are

on

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 391

in the first of ~1.3! with Re (12l)512j5g, Im l5h, and ~2.23! still applying tog. Results forg satisfying~2.23! are presented in Fig. 2a–f for varying vertex angles.

Included in Fig. 2a are the singularity exponents for the free–free plate, for both symmetric loading from~2.9!, and antisymmetric from~2.13!. The symmetric curve is given in Fig. 1, Williams @2#. It dominates singular character if load- ing is symmetric or mixed because the antisymmetric curve realizes weaker singularities with stress-free boundary con- ditions: Of course, it cannot dominate if loading is purely antisymmetric. The antisymmetric curve may be found in Fig. 9, Rösel @25# or Fig. 3a, Seweryn and Molski@20#.

For f5360° with free-free conditions, we have the tradi- tional, mathematically-sharp, stress-free crack with its inverse-square-root singularity for both symmetric and anti- symmetric loading (P1 , Fig. 2a!. For f5270°, we have a stress-free 90° reentrant corner with two possible singulari- ties, the stronger being for symmetric loading (P2 and P3 , Fig. 2a!. For f,257.5°, no further singularities are found for antisymmetric loading. Forf5180°, we have no singu- larity for symmetric loading. This is because, for this stress- free half-plane geometry, there is no discontinuity in bound- ary directions or conditions. Forf<180°, no further singularities are found for symmetric loading. Further, there are no complex eigenvalues with real parts in the singularity range for the free-free plate; this is shown in Karp and Karal @26#.

Given the equivalence ofcohesive stress-separation laws with stress-free conditions, the free-free curves of Fig. 2a also apply for these laws. Hence, the removal of some sin- gularities~noted in Part I, Section 2.3! can be confirmed. For a cracked configuration, putting cohesive laws ahead of a sharp crack as well as in back of it effectively gives a free-

free plate of vertex anglef5180°. Thus no singularities For a 90° reentrant corner under symmetric loading, co sive laws should be inserted ahead of the corner on the bisecting the plate to achieve bounded stiffnesses. This e tively gives a free-free plate withf5135°. Thus no singu- larities. The same sort of argument applies for antisymme loading ~see Sinclair, Khatod, and Rummel@27# for further explanation!.

Also in Fig. 2a are the singularity exponents for th clamped-clampedplate, for both symmetric loading from ~2.10!, and antisymmetric from~2.14!. These are for a rep resentative value ofk52, corresponding to Poisson’s rati n51/4 for plane strain, orn51/3 for plane stress. A very similar symmetric curve is given in Fig. 1, Williams@2#, for

k52 1 13: Some eigenvalues for antisymmetric response

the samek are given in Williams@28# from Ricci @29#. The actual symmetric curve fork52 may be found in Fig. 5a, Seweryn and Molski@20#, while the companion antisymmet ric curve is given in Fig. 8a, ibid. In Williams @2#, the sin- gularity associated with symmetric loading under clamp clamped conditions is claimed to be dominant. This is so loading is purely symmetric: Otherwise, for these bound conditions, the singularity associated with antisymmet loading is dominant.

For clamped-clamped conditions, both symmetric and tisymmetric curves atf5360° have an inverse-square-ro singularity. Forf5180°, both do not have a singularity fo similar reasons for this being so for free-free symmetric sponse. Forf,180°, no singularities are found for eithe No singularities associated with complex eigenvalues found for either.

Some indication of the influence of Poisson’s ratio

Fig. 2 Continued

t

e

r

i

s

l

r

t

u

- t

se any ent

on- h - s

up- ing

rily not to be

c- een

ad- rity

nta-

r

ur t

ch

lar on- f

the re

ari-

sen-

se. to ig.

392 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

singularities with clamped-clamped conditions is contain in Fig. 2a. This is because free-free with symmetry has same eigenvalue equation as clamped-clamped with antis metry andk51—see~2.9! and ~2.14!. Similarly free-free with antisymmetry is the same as clamped-clamped w symmetry andk51—see~2.13! and ~2.10!. Thus, ask de- creases corresponding to Poisson’s ratio increasing, the gularity for antisymmetric clamped-clamped conditions g stronger, while that for symmetric clamped-clamped g weaker. The trends thus evident in Fig. 2a are confirmed by singularity exponents for clamped-clamped conditions fok

51 2 3 and 3 in Seweryn and Molski@20#.

In Fig. 2b, singularity exponents are plotted for thefric- tionless contact-frictionless contactplate, for both symmetric loading from ~2.11!, and antisymmetric from~2.15!. These two eigenvalue equations are the simplest of all and adm analytical solution. Thus for symmetric configurations,

g522 2p

f ~p,f,2p! (2.24)

while for antisymmetric,

g5 p

f , 22

3p

f S 3p

2 ,f<2p D (2.25)

Expressions yielding these values ofg are given in equations ~36! and~41!, Seweryn and Molski@20#. These are the value plotted in Fig. 2b.

The ranges off in ~2.24! and ~2.25! bear comment. For symmetric loading, the absence of singular stresses whef 5180° is to be expected for the reasons put forward ear Given no singularities are found forf,180°, the range in- cludes all singularities for this loading. For antisymmet loading,f5270° terminates singular response on the low branch in a similar manner to the free-free antisymme case of Fig. 2a. For the upper branch, the same limit onf holds if antisymmetric singularities are not to be strong than those associated with symmetric loading. The reas for limiting singularity exponents in this way are as follow For contact on both plate edges and fields that are pu antisymmetric,su must be positive on one edge, negative the other. Where it is positive would be in violation of o contact stress constraint~2.6!. This means that antisymmetri loading needs to act in conjunction with sufficient symmet loading if compliance with~2.6! is to be achieved. Antisym metric singularity exponents cannot exceed symmetric if is to happen. Hence the limit in~2.25!. Observe, though, tha in the analysis of a given global problem, such complian with ~2.6! whence~2.25! does not have to be the case: needs to be checked for, and means sought to remedy situation if it does not occur.7

In Fig. 2c, singularity exponents for thecontact with friction-contact with friction plate are plotted. These ar from ~2.12! for symmetric configurations,~2.16! for antisym- metric. Values of friction coefficientf 51/2 and ofk52 are taken as representative. The general character of the e

n

ed he ym-

ith

sin- ts

ets

t to

n ier.

ic er ric

er ons s. rely on r

c ric

his t ce It the

e

xpo-

nents is similar to that of Fig. 2b for f 50, including the bounding of singularities for antisymmetric loading by tho for symmetric for the same reasons. That is, here too antisymmetric response must occur in concert with suffici symmetric participation if compliance with~2.6! is to be achieved.

There are some differences, however. For symmetric c figurations andf 51/2, there are two real branches whic merge together atf5252.5° into complex roots with a com mon real part~shown in Fig. 2c! and equal imaginary part of opposite sign~not shown!. Checking the companion eigenfunction for these real eigenvalues reveals that the per branch~shown! does have the contact shear oppos motion; the lower branch does not~hence not shown!. As previously, though, this removal here does not necessa mean that fields associated with the lower branch could be present in a problem. Again, singular stresses cease possible for symmetric configurations whenf5180°.

For symmetric loading, increasing the coefficient of fri tion f tends to reduce singularity exponents, as can be s by comparing Fig. 2c with Fig. 2b. For antisymmetric load- ing, results are mixed in this regard. For both types of lo ing, increasing Poisson’s ratio typically increases singula exponents.

In Fig. 2d, singularity exponents for theclamped-free plate are plotted. These exponents are from~2.17!. The real parts of all singular branches are shown for the represe tive valuek52; just the dominant singularity fork51. A similar curve to the upper branch fork52 is given in Fig. 1,

Williams @2#, for k52 1 13. The real parts of all branches fo

k52, as well as the most singular branch fork51, are given in Figs. 12a and 14a, Seweryn and Molski @20#, respectively.8

For f5360° and clamped-free conditions, there are fo possible singularities fork52: two for each complex roo indicated in Fig. 2d. For f5180° andk52, we have an oscillatory singularity as for an adhering, rigid, flat pun (P4 , Fig. 2d!. It is the presence of thesetwo roots as com- plex conjugates that precludes the removal of singu stresses in conforming contact problems when stick-free c ditions are assumed. Forf590° we have the singularities o P5 (n51/2, k51) andP6 (n53/8, k53/2) which, for ex- ample, apply to the edge of an adhering rubber tire and at outer surface of an epoxy-steel joint. No singularities a found for f,60° when k52, f,45° when k51. This trend of a larger range of vertex angles with stress singul ties with larger values of Poisson’s ratio~smallerk! is con- firmed by results for othern in Seweryn and Molski@20#.

In Fig. 2e, singularity exponents for thecontact-freeplate are plotted. These exponents are from~2.18!. The real parts of all singular branches are shown for the chosen repre tative case of contact with friction (f 51/2 andk52); just the dominant singularity is shown for the frictionless ca The exponents for contact with friction would not appear be available in the open literature: The values shown in F

s

7This is also the reason for excluding the further antisymmetric singularity expo g522p/f(p/2,f,p).

ent8The imaginary parts of singular eigenvalues fork52 and the other singular branche for k51 are also provided in Seweryn and Molski@20#.

h

a

d u

e g

i

e

p o

e

c- as

e 1

ct-

m- - un- hen

ses al qua- se

is a ex

r- e

at

- ju-

si-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 393

2e are confirmed in Klingbeil@30#. The exponents for fric- tionless contact are given in Fig. 2a, Seweryn and Molski @20#.9

There need be no restrictions on the branches include Fig. 2e as a result of contact constraints. This is because sign of the participation coefficient can always be such t the contact stress condition~2.6! is met, and a rigid body displacement can always be added so as to ensure the tional shear opposes slip. That is, the local fields associ with the singularity exponents of Fig. 2e can potentially par- ticipate in a global problem and all auxiliary contact con tions be met. Whether this actually happens for the partic global configuration of interest needs to be checked.

For f5360° and contact with friction-free conditions there are three singular stress fields possible. This is also case for frictionless contact, although this is not apparen Fig. 2e because only the most singular branch is includ For f5180°, the frictionless contact case gives the sin larity as for a tire at the edge of a pothole on an icy pavem (P7 , Fig. 2e!. For contact with friction andf5180°, the singularity that results is as for an adhering tire but w some slip permitted. Under these conditions there isonereal singularity ~at P78 , Fig. 2e! compared to the two for an ad hering tire with no slip. This enables the stress singularity be removed for conforming contact when there is cont with friction-free conditions. No singularities are found fo f,90° when f 50, f,116.6° when f 51/2. This is the trend in general, namely, asf becomes more positive, th range of vertex angles with stress singularities decreases the other hand, varying Poisson’s ratio while holdingf con- stant leaves the range of singular vertex angles unchang

In Fig. 2f, singularity exponents for thecontact-clamped plate are plotted. These exponents are from~2.19!. All sin- gular branches are shown for the representative case of tact with friction (f 51/2 andk52); just the dominant sin- gularity is shown for the representative frictionless casek 52). The exponents for contact with friction would not a pear to be available in the open literature: The values sh in Fig. 2f are confirmed in Smallwood@32#. The exponents for frictionless contact are given in Fig. 5a, Seweryn and Molski @20#. For the same reasons as for Fig. 2e, there need be no restrictions on the branches included in Fig. 2f as a result of contact constraints.

For f5360° and contact with friction-clamped cond tions, there are three singular stress fields possible. The s is true for frictionless contact-clamped conditions, thou this is not shown in Fig. 2f. For f5180° there is but one singularity for a given coefficient of friction. This enable singularities to be removed when transitioning from stick slip in contact problems. No singularities are found forf ,90° whenf 50, f,63.4° whenf 51/2. This is the trend in general here, namely, asf becomes more positive, th range of singular vertex angles increases. Conversely, constantf , increasing Poisson’s ratio reduces the range singular vertex angles.

d- th- n-

d in the at

fric- ted

i- lar

, the

t in d. u-

ent

th

- to

act r

. On

ed.

con-

( - wn

i- ame gh

s to

for of

While not strictly a power singularity, we close this se tion by noting instances of undefined oscillatory stresses in the last of~1.3!. These occur for mixed problems. For th clamped-free plate, their presence is indicated in Fig. 1a, Seweryn and Molski@20#, for k53. Solving ~2.17! for l 511 ih and k53 then gives oscillatory stresses whenf 5100.4°, 274.0° withh50.13, 0.02, respectively. Similarly for the other mixed problems—contact-free and conta clamped—solving~2.18! and ~2.19! for l511 ih leads to oscillatory stresses.

2.3 Log singularities with homogeneous boundary con- ditions

In addition to the singularities revealed for the real and co plex g of Fig. 2, there is the possibility of logarithmic con tributions to stress singularities. These may be produced der the homogeneous boundary conditions of Table 1. T they can take the form oflogarithmic intensificationof stress singularities. That is, stresses which behave as

s5O~r 2g ln r !1O~r 2g! as r→0 (2.26)

for g.0. For homogeneous boundary conditions and stres of the form of ~2.26!, at the outset these stem from re eigenvalues which are repeated roots of the eigenvalue e tion. This is a necessary but not sufficient condition for the stresses~see~1.3!!.

Repeated roots can be expected to occur where there transition from two real roots to roots which are compl conjugates. To see this, supposel is an eigenvalue ofD 50 for vertex anglef. Now perturbf by df while continu- ing to insistD50, and letdl denote the accompanying pe turbation inl. From Taylor’s theorem in two variables, w have

05 ]D

]l dl1

]D

]f df1

]2D

]l2

dl2

2 1

]2D

]l]f dldf

1 ]2D

]f2

df2

2 1...as df→0 (2.27)

wherein it is understood that all derivatives are evaluated l and f. If l is a repeated root, then]D/]l50 for l, f. Thus, provided]D/]fÞ0, ]2D/]l2Þ0 at l andf,

dl56A22df ]D

]fY ]2D

]l21O~df! as df→0

(2.28)

As df changes sign in~2.28!, we have the anticipated tran sition from two real roots to roots that are complex con gates.

This is what occurs atR1 in Fig. 2d. Further checking of the rank conditions in the second of~1.3! shows that they are satisfied for this repeated root. Forg.0, these rank condi- tions are necessary for the possibility of logarithmic inten fication of power singularities. In Dempsey@33#, such checks are carried out for the dominant singularity in the clampe free plate and consistently show the possibility of logari mic intensification of singularities wherever there is a tra

9England@31#, Fig. 5, gives values consistent with the exponents given here forf 50 and 0<f<180°, 270°<f<360°; the values ibid for 180°,f,270° do not apply to the frictionless contact-free plate.

394 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Table 5. Configurations with logarithmic singularities under homogeneous boundary conditions

Boundary conditions on uÄ0,f Configuration specifications

II-II f5f* , k51 V-V f5p, 2p, k51, f Þ0

k5cos 2f2f21 sin 2f, f 52cotf, fÞp, 2p I or VI-II f5p2fk , 2p2fk , k52f21 tanf I or VI-V k5112 cos 2f22f21 sin 2f, f 52cotf, fÞp, 2p II-V f5p/2, 3p/2, k53, f 523f/2

f5f̂k , f 5(k21)(32k)21cotf, fÞp, 2p, kÞ3

o

-

s

s

o

l

d

e

i

at- e - o f in

but

on ra-

fer- u-

r for

ri-

or m- ents s to ves.

ed ess lly

lex. or om

y

ge- ey

sition from complex to real roots. These transitions occur any f.101.4° exceptf5180° and 360°, and have 0,g ,0.75.

It can be expected that logarithmic intensification also curs for the less singular branch ofg under clamped-free conditions wherever there is a transition from complex real roots. Fork52, R2 in Fig. 2d is an example. For loga rithmic intensification being possible though, the rank con tions in the second of~1.3! need to be checked for thes configurations as well.

On occasion, repeated roots occur without a transit from complex to real values. This can be so if]D/]f50 for l andf in ~2.27!. Actual examples areR3 andP4 , both for k51, in Fig. 2d. This is not obvious from the figure becau the less-singular intersecting branch is not shown~see, though, Fig. 14a, Seweryn and Molski@20#!. For these points, however, Dempsey@33# has that the rank condition of ~1.3! are not satisfied and, consequently, logarithmic tensification is not possible.

Further configurations wherein logarithmic intensificati can be expected are where there are transitions from com to real eigenvalues for plates in contact with friction. The includeR4 of Fig. 2c, andR5 andR6 of Fig. 2e. Again, the rank conditions need to be checked to see if this is real possibility.

Typically logarithmic intensification of stress singularitie can be expected as stress singularities pass from being power singularities to oscillatory power singularities. some sense, the logarithmic intensification can be viewe a transition statebetween the two, resulting in stresses th are more singular than those with just power singulariti yet arguably less pathological than oscillatory singulariti We consider logarithmic singularities further in this sort role next when we review their occurrence without pow terms.

Pure logarithmic singularitieshave stresses which behav as

s5O~ ln r ! as r→0 (2.29)

For the pure logarithmic singularities of~2.29! under the homogeneous boundary conditions of Table 1, we need isfaction of the penultimate conditions in~1.3!. Only then can a log singularity occur. These are the weakest stress gularities possible in elasticity, and consequently the hard to detect absent an a priori appreciation of their poss participation. Accordingly, their asymptotic identification ca be of significant value.

for

c-

to

di- e

ion

e

in-

n plex se

y a

s pure In

as at es, s.

of er

e

sat-

sin- est ble n

Details of the application of the identification process tending the last of~1.3!, for the boundary conditions of Tabl 1, may be obtained from Sinclair@34#: Results are summa rized ibid. Every logarithmically singular configuration s identified complied with all of the conditions in the last o ~1.3!. Moreover, when situations arose during analysis which some of these requirements were complied with others not, no logarithmic singularities were found.

Given the importance of being aware of the participati of logarithmic stress singularities, we reiterate the configu tions so found here in Table 5. This table gives seven dif ent sets of specifications for configurations with log sing larities.

In Table 5,k is additionally constrained to the range fo physically applicable Poisson’s ratios. This is broadest plane strain. Hence

1<k<3 (2.30)

Further in Table 5, the vertex anglesf* , fk , and f̂k are such that

f* 5tanf* , fk5sin21 Ak11

2

kf̂k~~k21!214 cos 2f̂k!5~3k21!sin 2f̂k (2.31)

wherein the principal value of the arc sine is taken (0<fk

<p/2). The first of~2.31! realizesf* 5257.5°, a value pre- viously noted as that for the termination of power singula ties with free-free conditions and antisymmetry~Fig. 2a!. For k51, free-free eigenvalues coincide with those f clamped-clamped. Consequently this value for an inco pressible solid under clamped-clamped conditions repres a transition from stresses which are singular in themselve those which are bounded but have unbounded derivati We therefore distinguish it as the pointT1 in Fig. 2a. The same is true for all the other logarithmic configurations list in Table 5: They all represent transitions from power str singularities to no stress singularities. Further, they typica also represent transitions from real eigenvalues to comp

Local fields containing logarithmic stress singularities f all the configurations listed in Table 5 can be obtained fr Sinclair @34#. For the contact-clamped plate whenk51, f 50, andf̂k5f* /2, fields are also available from Dempse @21#. All of these fields demonstrate that the fields of~1.1! alone are, in general, incomplete for the plate with homo neous boundary conditions as in Table 1. In particular, th

e ary nts

ari- the

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 395

Table 6. Inhomogeneous boundary conditions for in-plane loading

Identifying Roman numeral

Boundary conditions

Physical description

I8 su52p, t ru5q Uniform tractions II 8 uu5rDf, ur5rDf8 Pinching with lateral constraint V8 uu5rDf, t ru5 f su Pinching with friction

o n

o

a

- i n a c s

i

g u

y

- der

are rm

na-

s is

n-

lip,

ary

dis- tions

e

ndi- s, in

ein u-

re

f di-

nal ous

- the

gu- ing og- ns

m re

If s by

ons s l es . e bed

@35#!. Again, they can be logarithmically intensified if th singularity coincides with that for homogeneous bound conditions. If, on the other hand, prescribed displaceme are continuously differentiable, generally no stress singul ties are produced. An apparent exception occurs when displacements are linear inr , the integral of uniform traction conditions in effect.10 For these conditions, we can similarl expect pure logarithmic singularities.

Given the importance of identifying logarithmically sin gular configurations, henceforth in this section we consi inhomogeneous boundary conditions as in Table 6. These the counterparts of those in Table 1 which include unifo tractions or linear displacements.

The boundary conditions of Table 6 merit some expla tion. In Conditions I8, p is an applied pressure whileq is as a constant shear. In Conditions II8, Df can be interpreted a the amount by which the vertex angle of an angular plate reduced by as a result of pinching contact with a rigid inde tor. With this interpretation,Df is positive on a negative u-edge, and vice versa. If such contact occurs with no s Df850: If it occurs with slip, we have Conditions V8. The inclusion of the possibility ofDf8Þ0 is so as to replicate displacement discontinuities which can occur in bound conditions in finite element analysis~FEA!. Such disconti- nuities occur in displacement derivatives at nodes when placement shape functions are used as boundary condi in submodeling with FEA, a practice implemented in som standard codes~eg, Chapter 14, ANSYS@36# and Section 7.3, ABAQUS @37#; see Sinclair and Epps@38# for further explanation!.

Using the conditions in~1.5!, instances of logarithmic stress singularities with the inhomogeneous boundary co tions of Table 6 can be identified. Typically by this mean logarithmic singularities in problems have been identified the literature as follows: Conditions I8– I in Kolossoff @15# and Dempsey@39#; Conditions I8– II in Sinclair @40#; all combinations of other conditions in Sinclair@34# and Sinclair and Epps@38#.

The configurations so found are given in Table 7: Ther there are twelve different sets of specifications for config rations with log singularities. In this table,k continues to be

demonstrate that the original Williams’ eigenfunctions a incomplete for the problems considered in Williams@2#.

2.4 Singularities with inhomogeneous boundary condi- tions

All of the preceding singularities for the in-plane loading an elastic plate occur with homogeneous boundary co tions on its radial edges. Here we consider what additio singular stress fields can be induced by inhomogene boundary conditions.

For applied tractionswhich are themselves singular, inte rior stresses are at least likewise singular. There is also potential of logarithmic intensification as in~2.26!. This can occur if the configuration of interest shares the same sin larity as in the applied tractions when under correspond homogeneous boundary conditions. This would mean l squared singularities in the event that the applied tracti were logarithmically singular. However, it would not see that either power or log singularities in applied tractions likely to be needed in practice.

What is more likely are nonsingular applied tractions. they are ord(r g) asr→0 andg.0, then the interior stresse are also nonsingular. This is so even if they get multiplied ln r becauser g ln r50 at r 50 wheng.0. Alternatively, if the applied tractions are ord(r 0) asr→0, we may see a tran sition between stresses which are nonsingular for tract that are ord(r g), to stresses which are singular for tractio that are ord(r 2g). Pure logarithmic singularities are natur candidates for such a transition: We look for further instan of their being induced by uniform tractions in what follow

For applied displacements, stress singularities can also b produced. In the first instance, these stem from prescr displacements which are not continuously differentiable~as in uu}Ar ). Then, singular stresses simply match the sin larity in displacement derivatives~see, eg, Browning and J

u-10‘‘Apparent’’ because displacements which are linear inr can have discontinuities in their derivatives whenr→0 on differentu.

Table 7. Configurations with logarithmic singularities under inhomogeneous boundary conditions

Boundary conditions on uÄ0,f Configuration specifications

I8 or VI-I f5p, 2p, qÞ0 f5f* , pÞ0 or qÞ0

II 8– II f5p, 2p, DfÞ0 or Df8Þ0 k51, DfÞ0 or Df8Þ0, fÞf*V8– V f5p, 2p, DfÞ0, kÞ1, f Þ0 f 50, DfÞ0

VI-VI f5f* , 2p I8 or VI– II 8 f5fk , p6fk , 2p2fk , pÞ0 or qÞ0 or Df8Þ0, kÞ2f21 tanf I8 or VI-V f5p, 2p, f pÞ0 or qÞ0

f 52cotf, fÞp, 2p, f pÞ0 or qÞ0, kÞ112 cos 2f22f21 sin 2f II 8– V f5p, 2p, DfÞ0

f5p/2, 3p/2, k53, DfÞ0 or f Df8Þ0, f Þ23f/2 f 5(k21)(32k)21 cotf, fÞp, 2p, f̂k , DfÞ0 or Df8Þ0, kÞ3

he e-

or - on of

e

ry

396 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

constrained as in~2.30! while f* , fk , andf̂k continue to be as in~2.31!. By suitably adjoining rigid body rotations, any combination of boundary conditions drawn from Tables 1 and 6 can be realized by the combinations given in Table 7.

A first instance of a logarithmic stress singularity in Table 7 occurs for a step shear on a half-plane (f5p andqÞ0). The full stress field is given in Kolossoff@15#. A related instance occurs for a constant shear on one side of a crack (f52p). Complete fields are given in Dempsey@39#. In both of these cases, the log singularity must participate ifq

Þ0 in the local boundary conditions. This is in contrast to t log singularities of Table 5 whose actual participation d pends on far-field conditions.

A further instance of a logarithmic stress singularity f Conditions I8– I in Table 7 occurs in Levy’s problem, al though such a log field is not included in the original soluti in Levy @41#. This problem entails an angular elastic plate vertex anglef subjected to a uniform pressurep on one edge while being free of stress on the other~Fig. 3a whereinf 5f* ). Levy’s traditional solution to the problem may b found in Article 45, Timoshenko and Goodier@42#. By way

Fig. 3 Examples of configurations with logarithmic stress singularities:a! Levy’s problem for a reentrant corner (f5f* ), b! pressure on a clamped acute corner (k52), c! symmetric indentation by a frictionless rigid sharp plate,d! displacement shape functions as bounda conditions for a submodel in FEA

i

t

t

p

x

t

i

e

i

i

d

- ey

in ied ns - n

om

n- out

ess

ities

es by can m ter- - -

in- end-

ally

the all m

e

ith

n

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 397

of example, the normal stresssu in Levy’s solution, in terms of the polar coordinatesr andu of Fig. 3a, may be expressed by

su52pF12 sinu cos~f2u!2u cosf

sinf2f cosf G (2.32)

In ~2.32!, it can be seen thatsu takes on the values of2p,0 at u50,f, respectively, and that there is no logarithmic s gularity in su . However, also clear in~2.32! is that the so- lution breaks down forf5f* of ~2.31!. This breakdown for the critical vertex angle off* is passed by without commen in Levy @41#. It is noted in Fillunger@43#, but perhaps is no as widely recognized today as it could be~eg, no mention of its existence is made in Timoshenko and Goodier@42#!. Nonetheless, it is serious and must be remedied if any ph cal sense whatsoever is to be made of elasticity treatmen a loaded plate which is as in Fig. 3a.

Supplementing the fields used to generate~2.32! by those attending~1.2! rectifies the situation. This is done in Dem sey @39#. The resultingsu , for example, may be expresse by

su52pF12 u

f* 2

cscf* 2f

* 2 ~2~sin~2u2f* !

2~2u2f* !cosf* !ln r 1~2u2f* !~cos~2u2f* !

2cosf* !!G (2.33)

for f5f* . Now there is a log singularity for this verte angle. Complete fields are given in Dempsey@39#. A reason- able transition between~2.32! and~2.33! is achieved in Ting @44#.

While it was once understandable to regard the bre down in the traditional solution to Levy’s problem as par doxical ~as in Sternberg and Koiter@45#!, armed with the analytical developments of Dempsey@39# and Ting @44#, it now would seem to be far less so. Thus here rather than f* in Levy’s problem a critical angle, we view it as atran- sition angleassociated with a logarithmic stress state wh is transitional much as in Section 2.3.

All of the foregoing examples occur for vertex angl where l51 is an eigenvalue. That is, for angles whereg 50 in Fig. 2. Suchf represent transition angles in the fo lowing sense. As the vertex anglef in angular elastic plates increases, there is a companion steady increase in the s lar character of stresses near the plate vertex~see Fig. 2!. These stresses go from power singularities in their der tives while being themselves bounded (g502), to having power singularities in themselves (g501). Transition angles with transitional log singularities demark the tw types of behavior.

We identify such transition angles with the letterT throughout Fig. 2. Hence for the free-free plate of Fig. 2a, we haveT1 , T2 , andT3 corresponding tof of f* , p, and

n-

t

ysi- s of

- d

ak- a-

erm

ch

s

l-

ngu-

va-

o

2p.11 In addition to a logarithmic stress singularity induce by the pressurep for f5f* here, we have one for the uniform shearq—a generalization of Levy’s problem in ef fect. Fields for this log singularity may be found in Demps @39#.

Another generalization of Levy’s problem is included Table 7. This occurs when the plate edge without appl tractions is clamped rather than free. That is, for Conditio I8– II. Typically there are four transition angles with loga rithmic stress singularities for this type of configuratio ~Table 7,~2.30! and ~2.31! for generalk; T4–T7 in Fig. 2d for k52). These angles can be less than 180°~eg, Fig. 3b!. Fields for associated log singularities may be obtained fr Sinclair @40#.

One other generalization of Levy’s problem is also i cluded in Table 7. This occurs when the plate edge with applied tractions is in contact. That is, for Conditions I8– V. There is a range of transition angles with logarithmic str singularities for this type of configuration~Table 7!; ex- amples are distinguished asT8–T11 in Fig. 2e. Again angles can be less than 180°. Fields for associated log singular may be obtained from Sinclair@34#.

Typically, the preceding logarithmic stress singulariti induced by uniform tractions can instead be produced cohesive laws. This is because cohesive law conditions admit rigid body translations which in turn produce unifor tractions. Thus Conditions VI are generally shown as al natives to Conditions I8 in Table 7. In this role, the condi tions given onp andq in Table 7 then apply to correspond ing uniform tractions within Conditions VI.

Turning to logarithmic stress singularities induced by homogeneous displacements, we first consider those att ing contact conditions. That is, Conditions V8– V in Table 7. For the case of an elastic angular plate being symmetric indented by a rigid frictionless plate with a sharp corner~Fig. 3c!, the finite rotations of one plate edge with respect to other produce log singularities. This is so even for sm rotations (0,Df!1). Local fields can be assembled fro those for~1.2!. Thus with ther andu coordinates of Fig. 3c,

H s r

su J 5

4m

11k

Df

f F2 ln r 1 H1 3J G (2.34)

with t ru50, and

ur5 2r

11k

Df

f @~k21!ln r 21#, uu52ru

Df

f (2.35)

Evident in ~2.34! is a log singularity which must participat for any DfÞ0.

Asymptotically the same log singularity as in~2.34! may be extracted from the global problem of indentation w

11The case off52p is not obviously a transition angle in Fig. 2a. This is because it is on a branch withg.0 only whenf.2p, a range of vertex angles not included i Fig. 2a. This branch can be seen in Fig. 2a, Seweryn and Molski@20#, near ‘‘a ’ ’ 5f/25p therein.

i

n l

i

h

u i e

t

r

i

l i

e t a

e

n

d

e

n

be

d in We ge- cal alue ese ish .2.

tes of

- site

398 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

rotation of a half-space. This problem is solved in Sect 48.4, Sneddon@46#.12 Such response can be expected to the case in other configurations wherein a plate vertex a gets extended or compressed. That is, that there is an iso logarithmic stress singularity with a coefficient proportion to the relative amount of rotation and the elastic moduli the material rotating—see Brock@47# and references therein

Other configurations with logarithmic stress singularit in response to contact conditions are identified in Table These can also be viewed as transition stress states as ated with transition angles~eg, T12 and T13 of Fig. 2c!. Fields can be obtained from Sinclair@34#.

Logarithmic stress singularities can be induced by in mogeneous displacements without contact conditions. T occurs for Conditions II8– II in Table 7. For the case of a straight boundary (f5p), these are the spurious log sing larities that can be introduced by the use of shape funct as boundary conditions in submodeling in finite elem analysis. An example involving four node elements is sho in Fig. 3d. Therein, log singularities at the node atO occur whenever there is a discontinuity in the derivatives of boundary displacementsu andv. That is, whenever the con stants are such thatc18Þc28 or c38Þc48 . Fields are given in Sinclair and Epps@38#. These spurious singularities whe shape functions are prescribed also occur for higher o elements and on any smooth submodel boundary~ibid!.

Other configurations with logarithmic stress singularit when Conditions VI and II8 occur in concert are identified in Table 7. These, too, are associated with transition angles~eg, wheng50 in Fig. 2f!. Fields can be obtained from Sincla @34#.

In closing this section we observe that most of the singularities identified in Table 7 stem from compliance w the last of~1.5! for nA54 whenr A53. Consequently, they do not require repeated roots of the eigenvalue equatio Indeed, for the most part, repeated roots are specifically cluded in Table 7. Just exactly when this is done in Tabl can be determined by comparing it with Table 5, every se specifications in the latter table corresponding to a repe root. Moreover, when such exclusions are relaxed and peated roots admitted, typically ln2 r stress singularities ar produced, in accordance with the first of~1.5!. The only exception is for the second set of specifications for Con tions V–V in Table 5 because the rank requirement is met.

As an example of a log-squared singularity, we consi symmetric indentation by a rigid sharp plate as in Fig. 3c, but now with lateral motion on the contacting edges complet constrained. That is, Conditions II8– II8 with DfÞ0 and Df850. Forf5f* of ~2.31! andk51, l51 is a repeated root ~see Table 5, cf Table 7!. The corresponding fields ca be assembled from those for~1.1!, ~1.2!, and~1.4!. Algebraic details can be obtained from Sinclair@34#. In terms of ther andu coordinates of Fig. 3c, the resulting fields have:

8

on be gle ated al of . es 7.

soci-

o- his

- ons nt

wn

he -

n der

es

ir

og th

n. ex- 7 of ted re-

di- ot

er

ly

H s r

su J 5

22mDf

f * 2 sinf*

@2 cosf* ~ ln2 r 12 ln r 2u2!

H 1

2J @2~cos 2u2cosf* !~ ln r 11!

22u sin 2u1f * 2 cos 2u#

t ru5 2mDf

f * 2 sinf*

@sin 2u~2 ln r 1f * 2 12!12u~cos 2u

2cosf* !# (2.36)

ur5 rDf

f * 2 sinf*

@2~cosf* 2cos 2u!ln r 12u sin 2u

2f * 2 cos 2u#

uu5 rDf

f * 2 sinf*

@2~sin 2u22u cosf* !ln r 12u~cos 2u

2cosf* !1f * 2 sin 2u#

for k→1. Other fields with log-squared singularities may obtained from@34,40#.

3 STRESS SINGULARITIES FOR THE IN-PLANE LOADING OF AN ELASTIC PLATE MADE OF MULTIPLE MATERIALS

3.1 Formulation and eigenvalue equations

Here we consider extension of the treatment presente Section 2 to plates made up of multiple elastic sectors. first formulate this extended class of problems for homo neous boundary conditions. Then we outline analyti means that can be used to derive companion eigenv equations. We stop short of actually presenting all th equations because of their relative complexity, but do furn references which contain them subsequently in Section 3

To begin, we continue to use cylindrical polar coordina r andu with origin O to describe the entire angular region interestR, with its complete vertex anglef. Now, though,R is comprised ofN subregions,Ri , i 51,2,...,N, andf of N subangles,f i ~Fig. 4!. Thus

R5 ø i 51

N

Ri , f5( i 51

N

f i (3.1)

where

Ri5$~r ,u!u0,r ,`, u i 21,u,u i% (3.2)

u i5 ( i 851

i

f i 8

with the understandingu050. With these geometric prelimi naries in place, we can formulate our class of compo problems as follows.

In general, we seek the planar stress componentss r , su , and t ru and their companion displacementsur and uu , as functions ofr andu throughoutR, satisfying: the appropri-

.4,

12There is a factor ofa21 missing from the stresses given at the end of Section 4 where ‘‘a’’ is as in Fig. 87 therein.

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 399

Fig. 4 Geometry and coordinates for the composite angular ela plate

e

s

i a h

r- the

en e-

la- f t it

ive - ra- is

iff- con-

als ive

ctly of

as tic

me

nt nts. ame. he dis- the ls

re

on ts

n- r a ith i- be

ons se e

Conditions C are from Rao@48#. They do not appeal as being particularly physically applicable. They might be inte preted as the conditions for a surface which is rough to point of locking and thereby prohibiting slip (ur matched!, yet on the point of separation (su→0). Such an interface is sketched in Fig. 5a.

Conditions D are also from Rao@48#. Essentially they are the same as conditions given in Erdogan and Gupta@49#. They are the composite counterpart of Conditions IV wh the latter are interpreted as being for a thin rigid reinforc ment. As such, they model a thin inclusion which is re tively stiff compared to its surrounding matrix: It is stif enough to restrain extension, however it is not so stiff tha restrains bending.

Conditions E are the composite counterparts of cohes stress-separation laws. Thusk andk8 are the stiffnesses as sociated with ‘‘springs’’ resisting normal and lateral sepa tion on an interface. This action for normal separation sketched in Fig. 5b where

uu 15 lim

u→u i

uu~u.u i ! (3.3)

with uu 2 defined analogously. In the elastic regime, the st

nesses in these laws should be chosen so that they are sistent with the elastic constitutive laws of the materi comprising the interface. When this is done, the adhes conditions are the physically appropriate ones for a perfe bonded interface: Conditions A are just a simplification them obtained by effectively lettingk andk8→` instead of their elastic values.

Conditions E also admit to other interpretations. One is a model of a flexibly bonded interface in studies of elas wave interactions in Jones and Whittier@50#. Another is as a model for an interface in a composite which permits so slip in Lene and Leguillon@51# ~for this latter interpretation, k is effectively taken to be infinite, thoughk8 is finite!.

As in Section 2.1, the preceding formulation is abse conditions at infinity and insists on bounded displaceme The basic reasons for these two aspects remain the s However, we are not aware of a formal extension of t completeness argument for elastic fields with bounded placements to composite configurations. Absent such, regularity conditions~2.5! must be viewed as provisiona when applied toN-material plates.

stic

Fig. 5 Sketches of interfaces;a! separating locking surfaces~Con- ditions C!, b! adhesive law action~Conditions E!

atefield equationsof elasticity;interface conditionson inter- nal plate edges;boundary conditionson external edges if the plate is open (f,2p), or further interface conditions if it is closed (f52p); and regularity requirementsat the plate vertex. The field equations hold onRi ( i 51,2,...,N) and are given by ~2.2!, ~2.3!, and ~2.4! with m andk in ~2.3! being replaced bym i andk i , wherem i is the shear modulus of th material comprisingRi andk i5324n i for plane strain, (3 2n i)/(11n i) for plane stress, withn i being Poisson’s ratio of this material. The admissible interface conditions a listed in Table 8 and hold onu5u i with i 51,2,...,N21, if the plate is open,i 50,1,...,N if the plate is closed (i 50 and N are for but one set of interface conditions!. The admissible boundary conditions continue to be as in Table 1 and hold u50,f if the plate is open. And the regularity requiremen are the same as~2.5! but now hold onRi , i 51,2,...,N.

The interface conditions of Table 8 merit comment. Co ditions A are the traditional conditions usually assumed fo perfectly bonded interface. Conditions B are for contact w friction governed by Amonton’s law. As such, to be phy cally applicable they further require that the normal stress nowhere tensile on the interface, as in~2.6!, and that relative lateral motion on the interface be opposed by shear tract there. Quite frequently in singularity analysis the special c of frictionless (f 50) contact is treated, so we distinguish t associated conditions by B0 in Table 8. Conditions A and B are the most common in singularity analysis.

Table 8. Interface conditions for in-plane loading

Identifying letter

Matched quantities

Additional conditions

Physical description

A su , t ru ur , uu

Perfectly bonded

B su , t ru uu

t ru5 f su Contact with friction

B0 su , t ru uu

t ru50 Frictionless contact

C su , t ru ur

su50 Separating locking surfaces

D su ur , uu

ur50 Thin rigid inclusion

E su , t ru su5k(uu 12uu

2) t ru5k8(ur

12ur 2)

Adhesive stress-separation law

d t

o e s

c

t t

e

d i

-

i o e

ld

o

ts. zero

nd-

ld

l is

lue

e

-

lf-

of he rial so to

for rched nd- is as

the e.

the

400 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Analysis follows that for single material plates. The co ditions for singularities with homogeneous bounda conditions/interface conditions remain as in~1.3!. Now, though, the order of the determinant involved is typica increased to

nA54N (3.4)

Hence the algebra involved in expanding determinants obtain eigenvalue equations in closed form can be consi ably more extensive. While the eigenvalues from the de minant could simply be numerically calculated without alg braic expansion, it is nonetheless useful to obtain simplified single expression for the eigenvalue equati Such expressions are more readily used than the raw d minant when the analysis of further specific configuration required. In addition, typically such expressions facilita checking by comparison with special cases/other indep dent algebraic analysis. To assist in obtaining them, so approaches for helping with the algebra entailed are offe in Dempsey and Sinclair@3# and Ying and Katz@52#.13

Once an eigenvalue equation is obtained for N-material problem (N>2), verification is important. This is a key concern because of the extent of the algebra volved. As previously mentioned, sometimes such verifi tion is afforded by other independent analysis. Otherwise addition to the obvious check of redoing the algebra, one also perform numerical checks. That is, evaluate the exp sion for the eigenvalue equation for diverse values of parameters involved, then compare with a direct calcula of the determinant from its originating matrix. Such compa sons need to take account of any factors removed in sim fying the expansion of the determinant to obtain the eig value equation. They should also be carried out for param values which do not, in themselves, realize simplifications the determinant.

Once checked, eigenvalue equations need to be solve singular eigenvalues. Generally this requires numer analysis. Such numerics are straightforward for the m part. The eigenvalues so computed can be verified by b substitution.

At this point, the entire analysis can be further checked consideringlimiting cases. For bimaterial plates with Condi tions A, one check is afforded by setting

m15m2 , k15k2 (3.5)

Then the eigenvalues for the corresponding single mate configuration should result.

A second check for bimaterials is to let one of the tw materials tend toward being rigid. Consider the fields in~1.1! under the limitm→`: The displacements go to zero. W therefore set displacements to zero in the interface condit of Table 8 to recover the corresponding boundary conditi of Table 1 for the one remaining deformable sector. Henc m i→`,

ber the

t

n- ry

lly

to er- er- e-

a n. ter- is

te en- me red

an

in- a-

, in can res- he ion ri- pli- n-

eter of

for cal ost ack

by

rial

o

e ons ns as

A→II, B→V, B0→III (3.6)

C→IV, D→II, E→VI

for R32 i and i 51 or 2. Again singular eigenvalues shou match single material values.

A third check for bimaterials is to let one of the tw materials become limp. Now consider the fields in~1.1! un- der the limit m→0, but first make the exchangesmc1 for c1 ,mc2 for c2 , and so on to avoid unbounded displacemen Now the stresses go to zero. We therefore set stresses to in the interface conditions to recover corresponding bou ary conditions. Hence asm i→0,

A→I, B→I, B0→I (3.7)

C→I, D→IV, E→I

for R32 i and i 51 or 2. Again, singular eigenvalues shou match single material values.

On occasion the eigenvalue equation for a bimateria insensitive as to whetherm1→` or m2→0, or vice versa. This simply means it should recover both of the eigenva equations for the corresponding boundary conditions in~3.6! and ~3.7! under either limit. For example, the eigenvalu equation for the interface crack can be written as

05sin2 lpF12 4m̂1m̂2

~m̂11m̂2!2 sin2 lpG (3.8)

where m̂15m11k1m2 and m̂25m21k2m1 . Equation~3.8! is insensitive as to whetherm1→` or m2→0. From~3.6! and ~3.7!, these limits correspond to A→II or A→I. Thus the interface crack~I-A-I ! becomes a half-plane with II-I or I-I. Under either limit,~3.8! recovers the product of the eigen value equation for a clamped-free half-plane~~2.17! for f 5p) with the eigenvalue equation for the free-free ha plane ((2.9)3(2.13) forf5p).

In addition to serving as checks, the limiting cases ~3.5!, ~3.6!, and ~3.7! enable a ready first assessment of t singular stresses involved when faced with a new bimate configuration which lacks any singularity analysis. It is al possible to extend the application of these types of limits configurations involving more than two materials.

For the general numerical analysis of eigenvalues other than special cases, the parameter space to be sea is now increased significantly in dimension over that atte ing configurations comprised of a single material. This because it now includes multiple vertex angles as well multiple pairs of elastic moduli.

For bimaterial plates, dimensional analysis reduces number of independent elastic moduli from four to thre This number can be further reduced by employing just two material constantsa andb defined by

Ha bJ 5

m2S k1H 1

2J 1D2m1S k2H 1

2J 1D m2~k111!1m1~k211!

(3.9)

The a andb of ~3.9! are given in Dundurs@53#. This article is a discussion which points out the reduction in the num of independent elastic constants that can be achieved by

s. At ently, the

13It is also possible to employ symbolic manipulation codes to expand determinan present this usually results in lengthy expressions for the determinant. Consequ such codes typically only provide an alternative to direct numerical treatment of original determinant.

c

i

l

e

o s

f

s

o

g’s s

e-

for if all this ach. and

for

e

ef- rities ith te i- in-

- ree bi- 6 n in w- a-

ns ey r if in- an is er- ed. na- vi-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 401

introduction ofa andb into the butt joint problem treated in Bogy @54#. Equivalenta andb were given earlier in Zak and Williams @55# to reduce the number of independent elas constants for the specific problem of a crack terminating p pendicular to a bimaterial interface. Dundurs@56# establishes general criteria for bimaterial configurations under whi such reductions can be made.

To demonstrate how the reduction is effected, we cons a perfectly bonded bimaterial with stress-free edges~I-A-I !. Absent a difference in materials, this plate’s stresses completely independent of elastic moduli, as is any singu ity exponent~see~2.9! and~2.13! for I-I !. Consequently, only the traditional matching conditions associated with perf bonding can introduce any dependence on elastic mod Without loss of generality, we take the perfectly bonded terface to occur atu50. Then, from~1.1!, the matching con- ditions result in the following sparse set of equations:

c1 11lc3

11c3 15c1

21lc3 21c3

2

c1 11lc3

12k1c3 15

m1

m2 ~c1

21lc3 22k2c3

2! (3.10)

c2 11lc4

12c4 15c2

21lc4 22c4

2

c2 11lc4

11k1c4 15

m1

m2 ~c2

21lc4 21k2c4

2!

In ~3.10!, the constants associated with the material ab u50 and modulim1 and k1 are distinguished with a plu sign, those with material below andm2 andk2 with a minus sign. Now subtracting the second of~3.10! from the first givesc3

1 in terms ofc1 2 andc3

2 and the two combinations o elastic moduli

m22m1

m2~k111! ,

m21k2m1

m2~k111! (3.11)

Back substituting into the first of~3.10! then givesc1 1 in

terms of c1 2 and c3

2 and the same two combinations. An performing the same operations on the third and fourth ~3.10! givesc4

1 andc2 1 each in terms ofc2

2 andc4 2 and the

same two combinations. Thus I-A-I stresses and singula exponents need only depend on the two combinations elastic moduli given in~3.11!. While these two combination are closer to those used in Zak and Williams@55# than those in Dundurs@53#, with some algebra they can be shown to equivalent toa andb of ~3.9!.

Similar analysis establishes that moduli dependence be reduced to just that ona andb for bimaterials and with any of the interface conditions A, B~and therefore B0), or C, under any combination of boundary conditions involving III, or IV ~Table 9!. Given the equivalence of cohesive la conditions with stress-free conditions as far as eigenva equations are concerned, singular eigenvalues with Co tions VI in bimaterials and with interface conditions A, B, C can also be expected to depend only ona andb.

The constantsa andb have seen widespread use for su configurations since Dundurs@56#, and have come to be known asDundurs parameters. They admit to physical inter-

tic er-

h

der

are ar-

ct uli.

in-

ve

d of

rity of

be

can

I, w lue ndi- r

ch

pretation to a degree. For plane stress, substituting form and k in terms of Young’s modulusE and Poisson’s ration gives:

a5 E22E1

E21E1 , bU5 n22n1

4 E25E1

(3.12)

Thus,a is a normalized measure of the mismatch in Youn moduli, while b reflects the difference in Poisson’s ratio when there is no difference in Young’s moduli. Similar r sults hold for plane strain ifE is exchanged forE/(12n2).

For bimaterials, the ranges 0,m i,`, 0<n i<1/2, andi 51,2 limit accompanyinga andb to within parallelograms ~Dundurs@53#!. These are given by

21,a,1

a21

4 <b<

a11

4 ...plane strain (3.13)

3a21

8 <b<

3a11

8 ...plane stress

The parallelogram for plane strain encompasses that plane stress and accordingly is the one to be searched possible singular eigenvalues are to be identified. Often search can be readily undertaken using an inverse appro That is, assuming a specific value of singular eigenvalue then solving fora andb.

It is also possible to use two sets ofa andb to reduce the number of independent elastic constants from five to four trimaterial plates: see Koguchi, Inoue, and Yada@57#. Then, too, the introduction ofas and bs can enable an invers approach to be adopted.

3.2 Power singularities identified in the literature

We now review contributions in the literature that have fected asymptotic assessments of possible stress singula for N-material plates under in-plane loading, starting w power singularities. We carry out our review in approxima order of increasing analytical complexity. We begin with b material plates and arguably the simplest of these, those volving ‘‘cracks’’ ~Fig. 6!: Here ‘‘cracks’’ means mathemati cal slits which may or may not have the traditional stress-f conditions of fracture mechanics. Next we consider open material plates~Fig. 7!: Altogether, the geometries in Figs. and 7 are the ones which have received the most attentio the literature. Thereafter we conclude the section by revie ing contributions for other bimaterial plates and some trim terial ones.

There is considerable duplication within the investigatio reviewed. We include later references for problems if th represent a means of verification of earlier research, o they provide further information on the singular stresses volved. We do this irrespective of whether or not we c envisage a situation in which the singular configuration physically appropriate. We exclude later references oth wise: A significant number of references are thus exclud In particular, we do not include later references which a lyze a global problem whose singular character was pre

ei- ogy in

rm. ry t is

s

er n

402 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Fig. 6 Bimaterial ‘‘crack’’ geometries analyzed for stress sing larities: a! interface crack,a8) interface crack ending at a kink o the interface,b! crack ending orthogonal to an interface,b8) crack ending obliquely to an interface

i a o

en-

n

nd d -

on

s in

i-

e

n-

ular

till ink

ace

n n is

edn

n- l n.

ons

from Tables 1 and 8 are I-A-I. Williams@58# provides both the eigenvalue equation and resulting complex singular genvalue in closed form. These results are confirmed in B @59#. The eigenvalue equation is equivalent to that given ~3.8!, while the associated singularity exponent is

g5 1

2 , h5

1

2p ln

m̂1

m̂2 (3.14)

wherem̂1 and m̂2 are as in~3.8!. For the interface crack~Fig. 6a! with clamped conditions

~II-A-II !, Theocaris and Gdoutos@60# give an eigenvalue equation and complex singular eigenvalue in closed fo Ting @61# furnishes a different expression for the imagina part of the complex singular eigenvalue: This latter resul confirmed in Ballarini@62# and elsewhere.14 The singularity exponent from Ting@61# for the clamped interface crack i similar to that for stress-free flanks. It has

g5 1

2 , h5

1

2p ln

k2m̂1

k1m̂2 (3.15)

Thus the imaginary part differs by at most60.175 from that in ~3.14!.

For the interface with one flank free and the oth clamped~I-A-II !, Theocaris and Gdoutos@60# give an eigen- value equation. Closed-form expressions for singular eig values are given in Ting@61#.

For the interface crack with contact with friction betwee the crack flanks~B-A!, Comninou@63# provides an eigen- value equation. This equation is confirmed in Dempsey a Sinclair @64#. Singular eigenvalues follow by inspection an are furnished in Comninou@63#, as is the companion eigen function. The simpler frictionless case (B0-A) is treated in the Appendix of Comninou@65#.

For the interface crack when there is contact with fricti on the interface ahead of a stress-free crack~I-B-I !, Gdoutos and Theocaris@22# provides an eigenvalue equation in term of Dundurs parameters. This equation is confirmed Comninou@66#. An expression for the resulting singular e genvalues is given in Gdoutos and Theocaris@22#. The sim- pler frictionless case (I-B0-I) is treated in Dundurs and Le @67#.

Finally, for the interface crack when an inextensible i clusion is inserted into the crack~D-A!, Dempsey@21# gives an eigenvalue equation. Closed-form expressions for sing eigenfunctions are given in Wu@68#.

We next considerkinked interface cracks. Here the geom- etry for these cracks is taken to be such that the ‘‘crack’’ s lies between the two materials but now terminates at a k on their interface~Fig. 6a8!. This geometry may be viewed as a generalization of that for the previous straight interf crack ~Fig. 6a!.

For the kinked interface crack~Fig. 6a8! with stress-free flanks ~I-A-I !, Bogy @59# furnishes the eigenvalue equatio in terms of Dundurs parameters. This eigenvalue equatio confirmed in Dempsey and Sinclair@64#. In addition, Bogy @59# provides singular eigenvalues for a variety of kink

14It is also implicit in Erdogan and Gupta@49#.

u-

gu-

ously well appreciated—the contribution of this genre of vestigation lies in the implications of the glob configuration analyzed, rather than singularity identificati

For the interface crackof Fig. 6a with stress-free crack flanks, the corresponding boundary and interface conditi

Fig. 7 Open bimaterial plate geometries analyzed for stress si larities: a! butt joint, a8) oblique butt joint,b! two plates of equal vertex angles,c! angular plate on a half-plane

t

o

i

s

e n

k

m

p

e

r

n ei-

alue

ace

are

ick-

nge of -

an a- -

air er

en

well ns. s n, nd ner h

ei-

fi-

s

nce

er-

im- ers ct

c re-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 403

interface cracks. These eigenvalues are numerically de mined for the most part. Further numerical eigenvalues given in Chen and Hasebe@69#. Theocaris and Gdoutos@60# and van Vroonhoven@70# also treat kinked interface crack with stress-free flanks, but do not use Dundurs paramete

For the kinked interface crack~Fig. 6a8! with clamped conditions either on one flank~I-A-II ! or both ~II-A-II !, Theocaris and Gdoutos@60# gives eigenvalue equations an some singular eigenvalues.

For the kinked interface crack with crack flanks perfec bonded~A-A !, equation~19! of Bogy and Wang@71# is the eigenvalue equation in terms of Dundurs parameters. T equation is confirmed in Dempsey and Sinclair@64# and else- where. In addition, Bogy and Wang@71# provides singular eigenvalues for quite a variety of such kinked configuratio Chen and Nisitani@72# provides the associated eigenfuncti as well as further eigenvalues. Van Vroonhoven@70#, Pageau, Joseph, and Biggers@73#, and Chaudhuri, Xie, and Garal @74# also treat the same kinked configuration without us Dundurs parameters.

For the kinked interface crack when there is contact w friction between the crack flanks~B-A!, an eigenvalue equa tion may be found in Dempsey and Sinclair@64# in terms of Dundurs parameters. Corresponding singular eigenvalue a variety of such configurations are numerically determin in Dempsey@21#. If contact with friction also occurs on th interface ahead of the crack~B-B!, an eigenvalue equatio may be found in Dempsey and Sinclair@64# in terms of Dun- durs parameters.

For the kinked interface crack when Conditions C Table 8 hold, eigenvalue equations in terms of Dundurs rameters for A-C, B-C, and C-C may be found in Demps and Sinclair @64#. Finally, for the kinked interface crac when Conditions D of Table 8 hold, eigenvalue equations A-D, B-D, C-D, and D-D are given in Dempsey@21#.

We now consider ‘‘cracks’’terminating at an interface rather than lying along it. The simplest such configuration when the crack impinges at a right angle~Fig. 6b!, because then the geometry is symmetric enabling symmetric and tisymmetric loading to be analyzed separately. As a con quence, this special case has received attention by a nu of investigators in the literature.

For a crackterminating normalto an interface~Fig. 6b! and having stress-free flanks~I-A-A-I !, Zak and Williams @55# furnishes an eigenvalue equation for loading which symmetric about the crack. This equation is in terms of rameters which are equivalent to those of Dundurs. It is c firmed in Dempsey and Sinclair@64#.15 Zak and Williams @55# provides singular eigenvalues. Further singular eig values are given in Khrapkov@75# and Bogy@76#. The ei- genvalue equation for antisymmetric loading is given

a

e

ould

ter- are

s rs.

d

ly

his

ns. n

a ng

ith -

for ed

of pa- ey

for

is

an- se- ber

is a-

on-

n-

in

Dempsey and Sinclair@64#. This equation is the same as fo symmetric loading.16

For a crack terminating normal to an interface~Fig. 6b! with contact with friction between the flanks~B-A-A !, Comninou and Dundurs@77# furnish an eigenvalue equatio in terms of Dundurs parameters. Corresponding singular genvalues are provided: These are independent of the v of the coefficient of friction.

For a stress-free crack terminating normal to an interf which is itself in contact with friction~I-B-B-I !, eigenvalue equations when loading is symmetric or antisymmetric furnished in Dempsey and Sinclair@64# in terms of Dundurs parameters. These equations are confirmed in Wijeyew rema, Dundurs, and Keer@78# which in addition provides singular eigenvalues for both modes of loading and a ra of values of the coefficient of friction. The simpler case frictionless contact (I-B0-B0-I) is treated in Gharpuray, Dun durs, and Keer@79#.

Finally, for a stress-free crack terminating normal to interface on which Conditions C or D hold, eigenvalue equ tions are available as follows: for I-C-C-I with either sym metric or antisymmetric loading, from Dempsey and Sincl @64# in terms of Dundurs parameters; for I-D-D-I and eith symmetric or antisymmetric loading, from Dempsey@21#.

For the more general instance of a crackterminating ob- liquely ~Fig. 6b8!, several investigations are available. Wh the crack is free of stress~I-A-A-I !, Bogy @76# furnishes the eigenvalue equation in terms of Dundurs parameters, as as singular eigenvalues for a variety of such configuratio Fenner@80# and Yong-Li @81# compute singular eigenvalue directly from the determinant without algebraic expansio though Fenner@80# does establish that eigenvalues depe on only two material constants. The eigenvalues in Fen @80# and Yong-Li@81# include ones which agree closely wit corresponding values in Bogy@76# ~provided a state of plane stress is assumed in Yong-Li@81#!. Wang and Chen@82# treats the same configuration: On occasion, the singular genvalues in Wang and Chen@82# agree with corresponding values in Bogy@76#, but in some instances there are signi cant discrepancies between the two.

For a crack terminating obliquely at an interface~Fig. 6b8! with flanks in contact with friction~B-A-A !, Comninou and Dundurs@77# furnishes an eigenvalue equation in term of Dundurs parameters. Comninou and Dundurs@77# also provides singular eigenvalues for varying angles of incide of the crack and different coefficients of friction.

For a stress-free crack terminating obliquely to an int face which is itself in contact with friction~I-B-B-I !, Wijeyewickrema, Dundurs and Keer@78# furnishes an eigen- value equation in terms of Dundurs parameters. When s plified for special instances, this equation agrees with oth in the literature. The simpler case of frictionless conta (I-B0-B0-I) is treated in Gharpuray, Dundurs, and Keer@79#.

ms

16Taken together, the eigenvalue equation for both symmetric and antisymmetri sponse is given as a simple squared term in Bogy@76#. This equation appears to hav an extraneous factor of sin2 lp—see ~28! and ~18! et seq ibid. The same factor is present in the eigenvalue equation for when the crack terminates obliquely. It w not appear to lead to errors in eigenvalues reported.

15There would appear to be a typographical error in the equation in Zak and Willi @55#. All that is needed to correct this error is to replace cosp with coslp.

f

a

n e n

s

e

n

a

h

tes

am- in

gles

re-

lf- are

am- and le

Er-

am-

ts. iven

tic rial

lair ec-

Sin- tz ti-

- n ,

in- n-

fer-

Sin- ed

re

ntian tated

404 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

There are some further generalizations for stress- cracks terminating at an interface~I-A-A-I ! which are ana- lyzed in the literature. If the crack flanks in Fig. 6b8 are allowed to subtend a finite angle at their tip and thus beco a reentrant corner, an analysis may be found in Tan Meguid @83#. If the interface in Fig. 6b8 is allowed to have a kink at the point where the crack terminates, an analysis m be found in Pinsan and Zhuping@84#.

We next consideropen bimaterial plateswhich do not, for the most part, involve cracks~Fig. 7!. We begin with prob- ably the simplest such configuration, thebutt joint of Fig. 7a. When the outside surfaces are free of stress and the joi perfectly bonded~I-A-I !, Bogy @85# furnishes the eigenvalu equation in terms of Dundurs parameters. This equatio confirmed in Dempsey and Sinclair@64#. Bogy @85# also pro- vides singular eigenvalues: These are consistent with co sponding values in Hein and Erdogan@86#. The ratio of the shear moduli for which a power singularity first starts appear is given in Kubo, Ohji, and Nakai@87#.

For the butt joint~Fig. 7a! with stress-free outside surfac and contact with friction on the interface~I-B-I !, Theocaris and Gdoutos@88# furnishes the eigenvalue equation in term of Dundurs parameters. This equation is confirmed in Dem sey and Sinclair@64# ~the sign of the friction coefficient ha to be changed because the friction condition is applied o negativeu-face in Theocaris and Gdoutos@88#, a positive u-face in Dempsey and Sinclair@64#!. In addition, Theocaris and Gdoutos@88# provides singular eigenvalues for varyin coefficients of friction.

The more generaloblique butt jointhere has the interfac meet the outside free surface at an angle other than 90°~Fig. 7a8!. When the joint is perfectly bonded~I-A-I !, Bogy @59# furnishes the eigenvalue equation in terms of Dundurs rameters. This equation is confirmed in Dempsey and S clair @64#. Singular eigenvalues are also provided in Bo @59# for several angles of incidence of the interface with t outside surface. Further singular eigenvalues are give Hein and Erdogan@86# and Rao@48#. Geometries for which a power singularity first starts to appear are given in R @48#, and Kubo, Ohji, and Nakai@87#.

For the oblique butt joint~Fig. 7a8! when the interface is in contact with friction~I-B-I !, Theocaris and Gdoutos@88# furnishes the eigenvalue equation in terms of Dundurs rameters. This equation is confirmed in Dempsey and S clair @64# ~again, the sign of the friction coefficient has to b changed!. In addition, Theocaris and Gdoutos@88# provides singular eigenvalues for several angles of incidence varying coefficients of friction.

A further open bimaterial geometry investigated in t literature is that of twoplates with equal vertex angles~Fig. 7b!. When the outside edges of the plates are stress free they are perfectly bonded along their interface~I-A-I !, Bogy @59# furnishes the eigenvalue equation. This equation is c firmed in Dempsey and Sinclair@64#. Singular eigenvalues are also provided in Bogy@59# for several plate vertex angles. Further singular eigenvalues are given in Rao@48#. Geometries for which a power singularity first starts to a pear are given in Rao@48# and Kubo, Ohji, and Nakai@87#.

ree

me nd

ay

t is

is

rre-

to

e

s p-

n a

g

pa- in-

gy he

in

ao

pa- in- e

nd

e

and

on-

p-

The stress-free bimaterial plate, with constituent pla with equal vertex angles~Fig. 7b!, can have the interface in contact with friction~I-B-I !. Theocaris and Gdoutos@88# fur- nishes the eigenvalue equation in terms of Dundurs par eters under these conditions. This equation is confirmed Dempsey and Sinclair@64# ~again, the sign of the friction coefficient has to be changed!. Theocaris and Gdoutos@88# also provides singular eigenvalues for several vertex an and any value of the coefficient of friction.

The last open bimaterial geometry investigated quite f quently in the literature is that of aplate sector on a half- plane~Fig. 7c!. When the outside edge of the plate and ha plane surface exterior to it are free of stress, and the two perfectly bonded along their interface~I-A-I !, Bogy @59# fur- nishes the eigenvalue equation in terms of Dundurs par eters. This eigenvalue equation is confirmed in Gdoutos Theocaris@22#. Singular eigenvalues when the vertex ang of the plate is 90° are provided in Bogy@59#. Singular ei- genvalues for other vertex angles are given in Hein and dogan@86# and Gdoutos and Theocaris@22#.

For a plate on a half-plane~Fig. 7c! when the plate is in contact with friction~I-B-I !, Gdoutos and Theocaris@22# fur- nishes the eigenvalue equation in terms of Dundurs par eters. This equation is confirmed in Comninou@66#. Singular eigenvalues are provided in Gdoutos and Theocaris@22# for plate angles of 60° and 90° and varying friction coefficien Singular eigenvalues for some other vertex angles are g in Theocaris and Gdoutos@89#. The simpler case of friction- less contact (I-B0-I) is treated in Rao@48# and Dundurs and Lee @67#.

There are some other bimaterial plates with asympto analysis in the literature. For the perfectly bonded bimate plate with stress-free edges~I-A-I ! and arbitrary vertex angles, an eigenvalue equation is given in Aksentian@90#. In terms of Dundurs parameters, it is furnished in Bogy@59#. This latter equation is confirmed in Dempsey and Sinc @64#. If one or both of the edges are clamped instead, resp tive eigenvalue equations are furnished in Dempsey and clair @64#. These equations are confirmed in Ying and Ka @52#.17 The second configuration is also investigated in Ave sian and Chobanian@91#. If both edges have rigid thin rein forcements~IV-A-IV !, the eigenvalue equation is given i Dempsey and Sinclair@64# in terms of Dundurs parameters and some eigenvalues are given in Rao@48#.

Eigenvalue equations for bimaterial plates with other terface conditions are available as follows. In terms of Du durs parameters for stress-free bimaterial plates with dif ent interface conditions~I-B-I and I-C-I! and arbitrary vertex angles, eigenvalue equations are given in Dempsey and clair @64#. In terms of Dundurs parameters for further clos bimaterial plates~A-C, B-C, and C-C!, eigenvalue equations are also given in Dempsey and Sinclair@64#. Eigenvalue equations involving Conditions II and Conditions B or C a given in Dempsey@21#. Eigenvalue equations involving Conditions D are also given in Dempsey@21#.

17In addition, these equations appear to be consistent with those given in Akse @90# providedmi is taken to be Poisson’s number rather than Poisson’s ratio as s on p 193. That is, providedmi51/n i .

-

s a e

e

s

a

h c

a

e

r

a

a -

a

we

face r. he a-

en is

er- er

his

ro

ps -

ite ss he

in-

or

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 405

Turning to thetrimaterial plate, the simplest of such con figurations occurs when the plates are all comprised of same material. Picu and Gupta@92# treats a closed plate o this type with frictionless contact on its interface (B0-B0-B0). Singular eigenvalues are independent of ela moduli and are given for a range of vertex angles in Picu Gupta@92#. Other degenerate trimaterial plates wherein th are not three distinct materials include the crack geomet of Figs. 6b andb8 reviewed earlier.

When a trimaterial plate is actually comprised of thr distinct materials, analysis can be extensive. Nonethel there are some true trimaterial plates investigated in the erature. For an open trimaterial plate with bonded interfa and stress-free/clamped exterior edges~I-A-A-I, I-A-A-II, or II-A-A-II !, Ying and Katz@52# derives eigenvalue equation An eigenvalue equation for the first of these configuratio ~I-A-A-I ! is given in terms of pairs of Dundurs parameters Koguchi, Inoue, and Yada@57#, as are some resulting singu lar eigenvalues. Further singular eigenvalues from the s equation are presented in Inoue and Koguchi@93#. Additional singular eigenvalues are given in Pageau, Joseph, and gers @73#, together with some singular eigenvalues for t closed and bonded trimaterial plate~A-A-A !. The nature of associated singular eigenfunctions for I-A-A-I is consider in Pageau et al@94#.

In closing, we comment on the one remaining set of terface conditions in Table 8, Conditions E. With these ad sive stress-separation laws instead of the classical perfe bonded conditions, some reduction in the occurrence stress singularities is to be expected. This is indicated limiting cases with single-material plates. However, this yet to be formally established in general.

3.3 Log singularities identified in the literature

Here we review contributions to the literature that have ymptotically established the possibility of logarithmic term in stress singularities forN-material plates under in-plan loading. We start with when such singularities can occur w homogeneous boundary conditions, then consider their currence with inhomogeneous boundary conditions. We cus on bimaterial plates and follow the same order of geo etries as previously in Section 3.2.

Before beginning this review, we recap the requireme for logarithmic participation in bimaterial plates becau these continue to be incorrectly stated/applied in the lite ture. Forhomogeneous boundary conditionsas in Table 1, conditions for logarithmic intensification of power singula ties are as in the second of~1.3! with nA58 for bimaterials. For the case of pure logarithmic singularities, conditions as in the penultimate of~1.3! with nA58. For inhomoge- neous boundary conditionsas in Table 6, conditions for a log-squared singularity are as in the first of~1.5! for nA58. For the case of pure logarithmic singularities, conditions as in the last two of~1.5! for nA58. Throughout these con ditions for bimaterials, corresponding inequalities forĉs and c̃s are to hold on at least oneRi ( i 51,2), while equations for c̃s are to hold on both.

For a pure logarithmic singularity, conditions other th

the f s tic nd re

ries

e ess, lit-

ces

. ns in - me

Big- he

ed

in- e- tly- of

via is

s- s

ith oc- fo- m-

nts se ra-

i-

re

re

n

the preceding continue to be advanced in the literature~eg, Murakami @95# and Wijeyewickrema et al@78#!. Typically these have

D5 ]D

]l 50 for l51 (3.16)

While appealing in its simplicity,~3.16! is not sufficientfor a log singularity withhomogeneousboundary conditions, and it is not necessaryfor a log singularity withinhomogeneous boundary conditions. To remove any doubt that this is so, furnish some demonstrations.

As a first demonstration of~3.16! not being sufficient with homogeneous boundary conditions, we consider the inter crack ~Fig. 6a! with crack flanks perfectly bonded togethe That is, Conditions A hold both ahead of and in back of t ‘‘crack’’ tip. The determinant for this case is given in equ tion ~25!, Bogy and Wang@71#. In terms of the eigenvaluel, this has

D52~12b2!2 sin4 lp (3.17)

ClearlyD of ~3.17! satisfies~3.16!. The coefficient matrixA which leads toD can be assembled from~1.1! on applying Conditions A onu50,p. Checking the rank of this matrix reveals that it drops to four whenl51. Thus~1.3! requires that the first four derivatives ofD be zero whenl51. TheD of ~3.17! has just its first three derivatives being zero wh l51. Consequently, no log singularity is possible for th configuration despite the fact that~3.16! is met. This is what one would expect because this configuration has two p fectly bonded half-planes with no discontinuities in eith boundary geometry or boundary conditions.

As a second demonstration of~3.16! not being sufficient, we consider the interface crack~Fig. 6a!, but now with the crack flanks in frictionless contact. The determinant for t configuration is given as equation~54!, Comninou@65#. This equation has a multiplicative factor which cannot be ze removed, and otherwise is

D5sin3 lp coslp (3.18)

Clearly D of ~3.18! satisfies~3.16!. However, checking the rank of the corresponding coefficient matrix reveals it dro to five whenl51. Thus~1.3! requires the first three deriva tives of D to be zero whenl51. TheD of ~3.18! has only the first two of its derivatives zero whenl51. Consequently, no log singularity is possible for this configuration desp the fact that~3.16! is met. This absence of logarithmic stre singularities is consistent with the results in Table 5 for t two limiting cases of~3.6! and ~3.7!.

As a third and final demonstration of~3.16! not being sufficient, we consider a crack terminating normal to an terface~Fig. 6b with I-A-A-I !. The determinant in this in- stance may be obtained from Dempsey and Sinclair@64# as

D5sin2 lp@a1b222l2~a2b!~12b!

1~12b2!coslp#2 (3.19)

ClearlyD of ~3.19! complies with~3.16!. This leads Koguchi et al @57# to conclude that log singularities are possible f

e t

y

-

a

t

t

r

s

ure p- ck

e g

g.

t,

l lt

gu- li. oc- of a so.

406 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

any values ofa and b. However, checking the rank of th coefficient matrix reveals that it drops to ten. Because bimaterial is a degenerate trimaterial (nA512), this means that the first two derivatives ofD must be zero atl51 for a log singularity in~1.3!. The D of ~3.19! has only its first derivative zero atl51. Consequently, no log singularit is possible here. This absence of logarithmic stress singu ties is consistent with results in Table 5 for limitin cases.18

Demonstrations of~3.16! not being necessary for log sin gularities with inhomogeneous boundary conditions abou They can be found in Table 7 as limiting cases.

In the literature,~3.16! is typically used with homoge neous boundary conditions. The configurations identified this way may admit the possibility of a log singularity, o they may not. The other requirements in~1.3! need to be further checked to decide. Absent such checks, the situa remains ambiguous in this regard and, accordingly, we o configurations so identified in the review that follows. As result, to date in the literature the number of bimaterial pla identified as having pure logarithmic singularities is few than that for plates comprised of a single material. In fa one would expect the opposite to be the case given the e parameters available with bimaterials. This probably me that there are a significant number of bimaterials which have log singularities that are, as yet, not identified explici

We begin with instances oflogarithmic intensificationof power singularities underhomogeneous boundary cond tions. Quite a variety of such instances are identified Dempsey@33#. Typically they occur at transitions from com plex to real eigenvalues~cf, Section 2.3!. For the traditional conditions for perfect bonding on the interface while ou edges are stress free~I-A-I !, the following bimaterial geom- etries are determined as having the possibility of pow logarithmic stress singularities in Dempsey@33#: Figs. 6a8 andb8, and Figs. 7a8, b, andc. Other instances are identifie for closed bimaterial plates. These are for perfect bond ~A-A ! and the geometry of Fig. 6a8, and for frictional contact with perfect bonding~B-A-A ! and the geometry of Fig. 6b8. The last is really a degenerate trimaterial.

Turning to pure logarithmic singularitieswith homoge- neous boundary conditions, there are few instances identifie in the literature wherein~1.3! is known to be satisfied fo bimaterials. Two such are for two sets of specifications the oblique butt joint~Fig. 7a8! with stress-free conditions ~I-A-I ! which are given in Chen@96#. Some further instance are given in Dempsey@33# for the following configurations: Fig. 6a8 with A-A, and Fig. 7a8 with I-A-I. Additionalin- stances may be inferred from Dempsey@33# for the configu- ration of Fig. 6b8 with B-A-A.19

ear. ity tedi

his

lari- g

- nd.

in r

tion mit a tes er ct, xtra ns

do ly.

i- in -

er

er-

d ing

d

for

Of course there have to be many more instances of p logarithmic singularities than this for bimaterials. This is a parent from limiting cases. For example, for the kinked cra ~Fig. 6a8! with stress-free flanks~I-A-I !, the limits in ~3.6! and ~3.7! lead to I-II and I-I, respectively, for a plate of on material. Then Table 5, for I or VI-II, shows there is a lo singularity for the first of these limits~the actual vertex angle involved isf5101.2° whenk52.85). Other configurations with pure logarithmic singularities for limiting cases are: Fi 6a8 with I-A-II, II-A-II, I-B-I, A-A, A-B, A-D, B-B, B-C, B-D, and D-D; Fig. 6b8 with I-A-A-I, B-A-A, and I-B-B-I; and Fig. 7a8, b, andc with I-A-I and I-B-I.

For pure logarithmic singularitieswith inhomogeneous boundary conditions, Bogy @85# provides an example for the butt joint subjected to tractions. Asymptotically at the join the configuration is as in Fig. 8 whereinpi and qi are the constant pressure and uniform shear traction on materiai ( i 51,2). In Bogy@85#, a pure log singularity is found to resu if

a50 or a52bÞ61

q1Þq2 (3.20)

In addition, a pure log singularity is found to result if

a52bÞ0

p1 21p2

2Þ0, p1~11a!Þp2~12a! (3.21)

Complete stress fields corresponding to~3.20! and~3.21! are given in equations~4.6! and ~4.7!, Bogy @85#.

The configurations admitted by~3.21! include ones with continuous tractions across the joint. Here, then, the sin larity is associated with the discontinuity in material modu Generally this added material discontinuity increases the currence of stress singularities over that for a plate made single material. However, this does not always have to be For the limited cases ofa51 andb5 1

2 or a521 andb 52 1

2, there is no log singularity associated with a step sh Here, then, the singularity associated with the discontinu in the applied shear traction is being offset by that associa with the discontinuity in material moduli.

ulari-

c can

Fig. 8 Butt joint subjected to uniform tractions

18For D of ~3.19!, l51 can be a root of multiplicity 3 for special values ofa andb: Further checking of these special cases reveals that they too do not have log sing ties. 19The thrust of Dempsey@33# is to gain an appreciation of when power-logarithm singularities occur. Hence pure log singularities are not explicitly identified and only be inferred as limits in this paper.

e e

y

o

f

l

d

l f

a

a

l r

e:

in- 6

-

rre-

or gle dge ider

ri-

er- o n s ear

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 407

The pure logarithmic singularities associated with~3.20! and ~3.21! both occur whenl51 is a repeated root of th eigenvalue equation and~3.16! is indeed satisfied for thes instances. So how is this consistent with the conditions ~1.5! for a log singularity with inhomogeneous bounda conditions? For that matter, because any configuration w inhomogeneous boundary conditions can also include the sponse with corresponding homogeneous boundary co tions, how is it consistent with the conditions of~1.3! for a log singularity with homogeneous boundary conditions?

Answering the second question first, we considera50 anda52b in turn. Fora50, assemblingA reveals that its rank is seven. Hence the first part of the conditions in~1.3! for a log singularity is actually satisfied. However, asse bling associated fields reveals thatĉ1 , ĉ2 , andĉ3 of ~1.2! are all zero. Therefore~1.3! has that there is no log singularit for homogeneous boundary conditions whena50. For a 52b, the rank ofA drops to six. Then~1.3! requires that, in addition to ~3.16! being met,]2D/]l250. This is not the case for theD here. Therefore~1.3! has that there is no log singularity for homogeneous boundary conditions whena 52b. Thus there is no pure logarithmic singularity whats ever for this configuration with homogeneous boundary c ditions. The fields given in Bogy@85# in equations~4.6! and ~4.7! are consistent with this conclusion.

Answering the first question second, we consider~3.20! and~3.21! in turn. For~3.20!, we find that it is an instance o compliance with the second of~1.5!. Then sincer A can equal nA21 ~for a50), ~3.16! can be satisfied too. For~3.21!, we find that it is an instance of compliance with the last of~1.5!. Then, sincer A5nA22, ~3.16! can be satisfied too. The field given in Bogy @85# are again consistent with these conc sions.

With respect to inhomogeneous boundary conditions, response of the butt joint is analogous to that of a plate w uniform shear tractions. For this last, as the plate ver angle varies, regular solutions withr 0 stresses break down This results in their requiring auxiliary fields for the verte angle with the breakdown. This in turn leads to a logarithm stress singularity for this angle. A transition between the t types of solutions can be achieved by suitably supplemen the regularr 0 stresses for inhomogeneous boundary con tions with stresses for corresponding homogeneous boun conditions. In effect, this is the approach developed Dempsey@39# and Ting@44#. For the butt joint, the only rea difference is that material moduli are varying instead o vertex angle: Otherwise the same evolution occurs.

Apparently only two further instances of pure logarithm singularities for bimaterials with inhomogeneous bound conditions are identified in the open literature. These are the oblique butt joint~Fig. 7a8! and may be found in Chen @96#. Of course, there have to be many more instances of singularities for bimaterials with inhomogeneous bound conditions than the total reported here. Again, this is app ent from limiting cases. For the kinked crack~Fig. 6a8! with tractions applied to its flanks (I8-A-I 8), the limits in ~3.6! and ~3.7! lead to I8– II and I8– I, respectively, for a single material plate. Then Table 7 has logarithmic stress singu ties for both limiting cases. Other kinked crack configu

in ry ith re-

ndi-

m-

o- n-

s u-

the ith tex . x ic

wo ting di- ary in

a

ic ry for

log ry ar-

ari- a-

tions with pure log singularities for limiting cases ar I8-IC-I, I8-IC-II, I 8-IC-V, II 8-IC-II, II 8-IC-V, V8-IĈ-V, where IC denotes interface conditions A, B, C, or D, ICˆ the same set minus C. Similarly, other limiting cases of log s gularities can be identified for further geometries in Fig. and Fig. 7.

Finally, for log-squared singularitieswith inhomogeneous boundary conditions, two instances are identified for the ob lique butt joint ~Fig. 7a8! in Chen @96#. Quite a number of other instances can be identified as limiting cases co sponding to~3.6! and ~3.7! via Tables 5 and 7~see the dis- cussion at the end of Section 2.4!.

4 STRESS SINGULARITIES FOR OUT-OF-PLANE LOADING

4.1 Out-of-plane shear of an elastic wedge made of a single material

Here we follow the order of presentation in Section 2 f in-plane loading when we treat out-of-plane shear of sin material wedges. Thus we begin by considering a we under homogeneous boundary conditions, then we cons inhomogeneous boundary conditions.

The elastic wedge of interest can be framed with cylind cal polar coordinatesr , u, andz with origin O ~Fig. 9!. It has indefinite extent in ther andz directions while subtending an anglef at its vertex. The only existing displacement ent tained is in thez directionuz . This displacement is taken t be independent ofz. Consequently field equations hold o the 2D regionR of ~2.1!. With these geometric preliminarie in place, we can formulate the class of out-of-plane sh problems of initial interest as next.

In general, we seek the out-of-plane shear stressest rz and tuz , and their companion out-of-plane displacementuz , as functions of r and u throughoutR, satisfying: thestress equation of equilibriumin the absence of body forces,

]t rz

]r 1

t rz

r 1

1

r

]tuz

]u 50 (4.1)

on R; the stress-displacement relationsfor a linear elastic wedge which is both homogeneous and isotropic,

Fig. 9 Geometry and coordinates for the elastic wedge

te- w

408 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Table 9. Boundary conditions for out-of-plane shear

Identifying Roman numeral

Boundary condition

Physical description

Is tuz50 Stress free II s uz50 Clamped III s tuz5kuz Cohesive stress-separation la Is8 tuz5q Uniform shear II s8 uz5rDfs Linear displacement

o s

I

o

o

,

h

ns h t the

ries

t for sen- ce-

e i-

m

nti-

em

n

al

d in

r

e- ntal

of s

n

of e

ut- is

ther ent

. In r

d- tic

re- n- rre- the

et

in use To

is

H t rz

tuz J 5lr l21S H c1

2c2 J sinlu1 H c2

c1 J coslu D

(4.4)

uz5 r l

m ~c1 sinlu1c2 coslu!

We are now in a position to discuss further the comple ness of fields complying with~4.3!. Given that uz does in- deed admit to representation by combinations of functio which are separable inr and u, the completeness of suc functions complying with~4.3! can be argued as follows. A the outset we draw on Sturm-Liouville theory to establish completeness of the fields in~4.4! for homogeneousbound- ary conditions whenl is real.20 Then we observe thatt rz or uz can equally well be represented on a circular arc by se from ~4.4! with eitherl never negative orl never positive. As a result, we must have a complete representation jus l never negative. Hence we must have a complete repre tation with bounded displacements, provided these displa ments are separable.

Eigenvalue equationsare obtained on introducing th fields of ~4.4! into pairs of homogeneous boundary cond tions drawn from Table 9. This leads to, fornonmixed prob- lems(Is or III s– Is or III s , IIs– IIs),

sinlf50 (4.5)

and formixed problems(Is or III s– IIs),

coslf50 (4.6)

Equation~4.5! for Is– Is has an associated torsion proble which is analyzed in Saint-Venant@98#. This problem is an- tisymmetric about the wedge bisector so that only the a symmetric contribution to~4.5! is involved ~viz, coslf/2 50). However, this is the part of~4.5! which leads to sin- gular eigenvalues. Moreover, given that the torsion probl Is– Is can be solved via the warping displacement~Neumann problem! or via a stress function~Poisson’s equation with Dirichlet conditions!, this equation also holds for IIs– IIs . Equation~4.6! for Is– IIs can also be viewed as for a torsio problem if a vertex angle of 2f is taken. Equation~4.6! for Is– IIs is explicitly obtained as a limiting case for a bimateri wedge in Aksentian@90#. The equivalence of IIIs with Is as far as both eigenvalue equations are concerned is argue Sinclair @99#.

In accordance with~4.3!, the range of eigenvalues fo admissible power singularities of the form of~4.4! is as pre- viously ~ie, 0,l,1). Such singular eigenvalues can be d termined in closed form for the elementary transcende equations represented by~4.5! and ~4.6!. Reintroducing the singularity exponentg512l, we thus have the following admissible power singularities: for nonmixed problems,

g512 p

f ~p,f<2p! (4.7)

and for mixed problems,

g512 p

2f S p

2 ,f<2p D

20See, eg, Ch V, Courant and Hilbert@97#.

t rz5m ]uz

]r , tuz5

m

r

]uz

]u (4.2)

on R, whereinm continues as the shear modulus; any one the first three admissiblehomogeneous boundary condition in Table 9 ~identified as Is , IIs , and IIIs therein! on the wedge face atu50, together with another such condition o the wedge face atu5f, for 0,r ,`; and theregularity requirementat the wedge vertex

uz5O~1! as r→0 (4.3)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th wedge vertexO.

Several comments on the preceding are in order. The of-plane displacement admitted with its shear stresse sometimes termed a state ofantiplane shear. This state is physically representative of the response at cracks and o geometric features under Mode III loading. The displacem uz is also physically representative of thewarping produced when noncircular prismatic bars are subjected to torque this role, it complements theuu displacement component fo pure torsion~see, eg, Ch 10, Timoshenko and Goodier@42#!.

The homogeneousboundary conditionsof Table 9 have in-plane counterparts in Table 1 in accordance with: I fors , II for II s , and VI for IIIs . It is possible as well to interpret Is

as the analogue of III, and IIs as the analogue of IV. If the stiffnessk in III s is let to tend to zero, Is is recovered, while if it is let to tend to infinity, IIs is recovered. Otherwisek is positive onu50, negative onu5f.

As in Section 2.1, there are noconditions at infinityor length scalepresent in the formulation. For the reasons a vanced in Section 2.1, this is appropriate in an asympt treatment. Further, regarding theregularity requirement ~4.3!, we remark that this can be included provided the sulting formulation can be shown to be complete. We c sider this completeness issue further once we have co sponding basic fields established. Given completeness singular fields admitted by~4.3! have unbounded stresses y bounded displacements.

Analysis is straightforward and parallels that outlined Section 1. Indeed, it is simpler than that in Section 1 beca the problem at hand is harmonic rather than biharmonic. see this, substitute~4.2! into ~4.1!. This showsuz to be har- monic. Therefore it admits to separation of variables. T leads to, as ourbasic fields for out-of-plane shear,

ith

y a-

are

,

i

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 409

Fig. 10 Singularity exponents in out-of-plane shear for vary wedge angles

t r

i

p

i

do

ry mic

do

lf-

d y-

,

n sla-

as- tor- ait

he ee ion

a rve to -

les

ith

are no repeated roots. Hence~1.3! ~with nA52 therein! has that there are no singularities other than those of Fig. 10 w homogeneous boundary conditions.

This need not be the case forinhomogeneous boundar conditions. For uniform tractions/linear displacements, log rithmic stress singularities are possible.21 The specific inho- mogeneous boundary conditions considered to this end included in Table 9, distinguished by primes. Hereinq con- tinues as a constant shear traction for Is8 , while Dfs is the out-of-plane angle rotated through by IIs8 . For these condi- tions, use of the basic fields of~4.4! leads to systems which in general, cannot be solved whenl51 is an eigenvalue. Thus we need auxiliary fields. These follow from~4.4! on differentiating with respect tol. For the stress componentt rz

this leads to, as an example of ourauxiliary fields for out-of- plane shear,

t rz5r l21@~11l ln r !~ ĉ1 sinlu1 ĉ2 coslu!

1lu~ ĉ1 coslu2 ĉ2 sinlu!# (4.9)

In ~4.9!, carets atop constants continue to indicate they not have to be the same as in~4.4!. Using the full fields associated with~4.9!, in conjunction with those of~4.4!, then enables solution. Hence a log singularity forl51 ~see ~4.9!!. This occurs when the last of~1.5! holds for nA52. Now, though, the conditions on the constants within auxilia fields can be dispensed with. This is because a logarith stress singularity attends any nontrivialc in the auxiliary fields for out-of-plane shear~see~4.9!!. Configurations that do comply with the foregoing requirements and thereby have log singularities are given in Table 10.22

In Table 10, the logarithmic stress singularity on a ha plane with Is8– Is or IIs8– II can be anticipated from the asymptotic analysis in Wasow@103#. For Is8– Is and bothf 5p andf52p, these log singularities are fully develope in Ting @104#, together with a reasonable transition for var ing vertex angles throughp and 2p effected by means of the approach of Ting@44#. For other configurations in Table 10 a similar analysis may be found in Sinclair@105#. For the most part, IIIs is equivalent to Is8 in Table 10 because it ca produce a uniform shear in response to a rigid body tran

ng

Table 10. Single material configurations in out-of-plane shear with logarithmic singularities

Boundary conditions on uÄ0,f Configuration specifications

Is8 or III s– Is f5p, 2p, qÞ0 II s8– IIs f5p, 2p, DfsÞ0 III s– III s f52p Is8 or III s– IIs8 f5p/2, 3p/2, qÞ0, DfsÞ0

ts to

ould

ed

le

omo-

g512 3p

2f S 3p

2 ,f<2p D (4.8)

The singularity exponents of~4.7! and ~4.8! are plotted in Fig. 10 whereint denotes either shear.

For nonmixed problems, stress singularities are only sociated with reentrant corners. For prismatic bars under sion, this is recognized in Section 710, Thomson and T @100#, and in Saint-Venant@101#. Forf5360° with free-free conditions, the nonmixed curve of Fig. 10 recovers inverse-square-root singularity of a traditional stress-f crack under Mode III loading. The associated eigenfunct is given in Irwin and Kies@102#. For f5270°, the non- mixed curve produces the singularity as for a keyway in shaft transmitting torque. For other vertex angles, the cu for nonmixed problems in Fig. 10 is similar in character the upper curves in Fig. 2a which are for corresponding non mixed problems with in-plane loading.

For mixed problems, a broader range of vertex ang leads to stress singularities~Fig. 10!. This is similar to the corresponding situation with in-plane loading~Fig. 2a cf 2d!. Furthermore, the general character of the mixed curves w out-of-plane shear~Fig. 10! is quite similar to those forg for k52 with in-plane loading~Fig. 2d!.

In light of the preceding discussion regarding comple ness, it would seem to be unlikely for there to be anyth other than real power singularities for out-of-plane shear o wedge under homogeneous boundary conditions. This ex tation is in fact met by the eigenvalue equations~4.5! and ~4.6!. Separating real and imaginary parts in these equat reveals that there are no complex eigenvalues. In addit differentiating these equations with respect tol reveals there

te- ng f a ec-

ons ion,

21As in Section 2.4, it is possible for nonsingular inhomogeneous displacemen produce other than logarithmic singularities, eg,uz}Ar . Further, if these singularities coincide with ones for corresponding homogeneous boundary conditions, they c possibly be logarithmically intensified. 22The full fields for Eq.~4.9! do give rise to a logarithmic displacement field associat with homogeneous boundary conditions. This occurs forl50, c150, andc25F/f and is for a line-load of strengthF in out-of-plane shear. The correspondingtuz is zero, so that stress-free conditions are obeyed by this stress field for any vertex angf. However, the associated logarithmic displacement field is not in compliance with~4.3!, so that our original statement concerning the absence of logarithmic terms with h geneous boundary conditions still holds.

nt nts. . The can on- -

m- w

410 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Table 11. Interface conditions for out-of-plane shear

Identifying letter

Matched quantities

Additional conditions

Physical description

As tuz , uz Perfectly bonded Bs tuz tuz50 Frictionless contact Ds uz uz50 Thin rigid inclusion Es tuz tuz5k(uz

12uz 2) Adhesive stress-separation la

d

-

i h

e

o -

r

e

u

o

n

in di- e

ce

of

m- ces

eter.

the

in n-

in re

to ifi- I

e ing a

on

the We we nt ons. o-

i-

- f

u- est

- e

to

n-

f

s

As in Section 4.1, the preceding formulation is abse conditions at infinity and insists on bounded displaceme The basic reasons for these two aspects remain the same limited completeness argument advanced in Section 4.1 be extended to composite configurations with interface c ditions As , Bs , or Ds : for these conditions, composite con figurations are merely equivalent to multiple regular Stur Liouville problems.

Conditions As have received by far the most attention the literature. Accordingly, we focus attention on these tra tional conditions for a perfectly bonded interface next. W comment briefly on the nature of results for other interfa conditions at the end of the section.

We begin our review of perfectly bonded wedges made multiple materials by consideringbimaterials. Analysis for this class of composite wedge is straightforward and co pact. This is because the order of the coefficient matri involved is only four~cf 8 for most in-plane bimaterials!, and eigenvalue equations depend on a single material param Here

m̄5m1 /m2 (4.10)

serves as this single material constant. Eigenvalue equationsfor bimaterial wedges in out-of-

plane shear are available in the literature as follows. For open bimaterial wedge and Conditions Is2As2Is , IIs2As

2II s , and Is2As2II s eigenvalue equations are obtained Aksentian @90#. The first two of these equations are co firmed both in Rao@48# and in Sinclair@99#. The last eigen- value equation for Is2As2II s is confirmed both in Sinclair @99# and in Ma and Hour@106#. For a crack terminating at an interface~Fig. 6b8! and Conditions Is2As2As2Is , an ei- genvalue equation is given in Fenner@80#. For the closed bimaterial wedge and Conditions As2As , an eigenvalue equation is given in Sinclair@99#. This eigenvalue equation in confirmed in Pageauet al @73#.

Eigenvaluesfor the dominant power singularity for Is

2As2Is are given in a compact graphical form in Rao@48# for any values off1 and f2 , but for a somewhat limited range ofm̄. Eigenvalues for a more extensive range ofm̄ but limited values off1 andf2 are given in Sinclair@99#: The f1 and f2 treated therein correspond to the geometries Fig. 6a, a8, and Fig. 7. Secondary power singularities a also given in Sinclair@99#.

There is a certain duality between Is2As2Is and IIs 2As2II s which enables singular eigenvalues for the latter be directly obtained from eigenvalues for the former. Spec cally this is done by entering graphs of eigenvalues fors

2As2Is with the truem̄ replaced by 1/m̄ ~see Sinclair@99# for further explanation!. For both types of configuration, th discontinuity of an abrupt change in shear modulus attend Conditions As means it is no longer necessary to have reentrant corner for singular stresses to be possible.

Singular eigenvalues for Is2As2II s and geometries as in Fig. 6a, a8, and Fig. 7 are given in Sinclair@99#. Some fur- ther eigenvalues for Is2As2II s are provided in Ma and

tion. Under these circumstances, the requirements place q for Is8 apply to the uniform shear within IIIs .

4.2 Out-of-plane shear of an elastic wedge made of mul tiple materials

In this section we consider extension of the treatment in preceding section to wedges made of multiple materials. first formulate this extended class of problems. Thereafter review contributions to the literature which identify attenda stress singularities under homogeneous boundary condit We then discuss the further singularities possible with in mogeneous boundary conditions.

To begin, we continue to use cylindrical polar coord nates,r , u, andz with origin O to describe the entire wedg of interest with its complete vertex anglef. Now, though, the wedge is comprised ofN prismatic subwedges with ver tex anglesf i , i 51,2,...,N. Each of these subwedges is indefinite extent in both ther andz directions. We also con tinue to entertain displacement in thez direction alone, with this displacement being independent ofz. Consequently, field equations hold on the 2D regionR of ~3.1! ~Fig. 4!. With these geometric preliminaries in place, we can form late the class of out-of-plane shear problems of initial inter as next.

In general, we seek the out-of-plane shear stressest rz and tuz and their companion out-of-plane displacementuz as functions ofr andu throughoutR, satisfying: the appropri- atefield equationsof elasticity;interface conditionson inter- nal wedge faces;boundary conditionson external faces if the wedge is open (f,2p), or further interface conditions if it is closed (f52p); and a regularity requirementat the wedge vertex. The field equations hold onRi of ~3.2!, i 51,2,...,N, and are given by~4.1! and ~4.2!, with m in the latter being replaced bym i , the shear modulus of the mate rial comprisingRi . The admissible interface conditions a listed in Table 11 and hold onu5u i of ~3.2!, with i 51,2,...,N21 if the wedge is open,i 50,1,...,N if the wedge is closed (i 50 andN are for but one set of interfac conditions!. The admissible boundary conditions continue be as in Table 9 and hold onu50, f if the wedge is open. And the regularity requirement is the same as~4.3!, but now holds onRi , i 51,2,...,N.

The interface conditions of Table 11 have in-plane co terparts in Table 8 as follows: A for As , B0 for Bs , D for Ds , and E for Es . There is no counterpart to Conditions C Table 8 in out-of-plane shear. In Conditions Es , k is the stiffness in the adhesive stress-separation law anduz

1 anduz 2

are defined analogously touu 1 of ~3.3!. In reality, Conditions

As for perfect bonding are just a simplification of Conditio Es obtained on lettingk tend to infinity.

in ns

so

ies

nd

- ion

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 411

Table 12. Bimaterial configurations in out-of-plane shear with loga- rithmic singularities

Boundary conditions on uÄ0,f Configuration specifications

Is8 or III s– Is8 or III s f15f25p, q1Þq2 f15f25p/2, q1Þ2m̄q2 f15p/2, f253p/2, q1Þm̄q2 m̄52cotf1 tanf2, q1 cosf2Þq2 cosf1

II s8– IIs8 f15f15p, Df1ÞDf2 f15f25p/2, m1Df1Þ2m2Df2 f15p/2, f253p/2, m1Df1Þm2Df2 m̄52tanf1 cotf2, m̄Df1 sinf2Þ2Df2 sinf1

Is8 or III s– IIs8 f15p, f25p/2, q1Þm2Df2 f15p/2, f25p, q1Þm1Df2 m̄5cotf1 cotf2, q1 sinf2Þ2m2Df2 cosf1

t

e

u

c

a

the ed ant is-

ner- rial

rial

-

ec- I

e gen- n- 0°

e

ns

, the For

hat es or-

ut- as- e- of a ns, ul- .

i

ns

er-

n. m er n- ase

the

a- lly

ble

can act like Is8 by virtue of uniform shears being induced response to a rigid body translation. In this role, restrictio on q1 andq2 in Table 12 then apply to the uniform shears induced.

By way of examples of the logarithmic stress singularit of Table 12, Fig. 11 illustrates Is82As2Is8 for a half-space and for a bimaterial wedge with vertex angles of 30° a 120°. For the first~Fig. 11a!, Table 12 hasm̄51 and q1

Þ2q2 when f11f25p. Hence there is no material dis continuity. Here, then, the discontinuity in the shear tract by itself has an associated log singularity~this is the configu- ration analyzed in Ting@104#!. For the second~Fig. 11b!, Table 12 hasm̄53 and q1Þ2A38q2 when f1530° and f25120°. Hence we can takeq152q and q25q so that there is no discontinuity in the shear traction. Here, then, discontinuity in material moduli by itself has an associat log singularity. Notice, too, that there need not be a reentr corner present for a log singularity with either of these d continuities.

There are a few analyses oftrimaterial wedgesin out-of- plane shear available in the literature. The simplest dege ate trimaterial treated is a stress-free crack in one mate terminating normal to an interface with a second mate (Is2As2As2Is and a cross section as in Fig. 6b!. Singular eigenvalues are determined in closed form in Barnett@108#. These eigenvalues are confirmed in Fenner@80#. When the crack is other than perpendicular to the interface (Is2As

2As2Is and Fig. 6b8!, the eigenvalue equation for this de generate trimaterial is given in Sendeckyj@109#. This equa- tion is confirmed in Fenner@80# which also furnishes some singular eigenvalues. True trimaterial wedges with cross s tions as in Fig. 6b and either stress-free crack flanks (s

2As2As2Is) or bonded ones (As2As2As) are analyzed in Pageauet al @110#. This reference provides eigenvalu equations, singular eigenvalues, and accompanying ei functions. A further true trimaterial wedge with each co stituent single-material wedge having a vertex angle of 9 and with outside stress-free faces (Is2As2As2Is) is ana- lyzed in Keer and Freeman@111#. This reference provides th eigenvalue equation.

In closing, we comment on the other interface conditio in Table 11. We observe that Conditions Bs and Ds act like the boundary conditions Is and IIs of Table 9. Consequently bimaterials with these interface conditions simply have same singular character as two single-material wedges. Conditions Es , some reduction in singular stresses over t for As is to be expected. This is indicated via limiting cas with single-material wedges. However, this is yet to be f mally established in general.

4.3 Out-of-plane bending: Classical theory

Here we consider the singularities that can occur in the o of-plane bending of an elastic plate when treated within cl sical fourth-order theory. We follow the same order of pr sentation as previously. Thus we first treat plates made single material under homogeneous boundary conditio then inhomogeneous conditions, then plates made of m tiple materials under these two types of conditions in turn

es:

Hour @106#, which also furnishes associated eigenfunctio for all three types of configuration Is2As2Is , IIs2As 2II s , and Is2As2II s .

Singular eigenvalues for the crack terminating at an in face with Is2As2As2Is are given in Fenner@80# for a wide range ofm̄ and all angles of incidence.

Singular eigenvalues for As2As are given in Sinclair @99#. Some further eigenvalues for As2As are provided in Pageau et al@73#, together with the associated eigenfunctio

In view of the discussion of completeness, it would se to be unlikely for there to be anything other than real pow singularities for out-of-plane shear of a bimaterial wedge der homogeneous boundary conditions. That this is the c is confirmed in Sinclair@99#. Further, there is no logarithmi participation under homogeneous boundary conditions Is and II s ~ibid!.

Again, this absence of other singularities need not be case forinhomogeneous boundary conditions. For the uni- form traction/linear displacement conditions of Table 9, log rithmic stress singularities are possible. Following basica the same steps as in Section 4.1, Sinclair@107# identifies instances of such log singularities. These we present in T 12.

In Table 12 it is understood thatf1 andf2 are required to be such that positive shear moduli are involved with

0,m̄,` (4.11)

Further in Table 12,q1 is q andDf1 is Dfs onR1 , while q2

is q andDf2 is Dfs on R2 . As previously, Conditions IIIs

Fig. 11 Examples of wedges with logarithmic stress singularit a! half-space with discontinuous shear traction,b! bimaterial wedge with continuous shear traction

e

412 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Fig. 12 Geometry and coordinates for the angular elastic plat bending

i

ith

-

in

Table 13. Boundary conditions for out-of-plane bending with classical theory

Identifying Roman numeral

Boundary conditions

Physical description

Ib Mu50, Qu2 ]Mru

]r 50 Stress free

IIb uz50, ]uz

]u 50 Clamped

III b ]uz

]u 50, Qu50 Symmetry

IVb uz50, M u50 Simply supported

Vb Mu5kt

]2

]r2 S1r ]uz

]u D, Qu2

]Mru

]r 5kb

]4uz

]r4

Elastically restrained

Ib8 M u5M 01M 08r ,

Qu2 ]Mru

]r 5V0

Applied moment/shear

IIb8 uz5u0r 3,

]uz

]u 5u08r

3 Applied displacement/ rotation

IVb8 M u5M̂ 01M̂ 08r , uz5û0r 3 Applied moment/ displacement

t

The angular elastic plate to be bent can be framed w cylindrical polar coordinatesr , u, andz with origin O at the vertex of the mid-plane of the plate~Fig. 12!. It has indefinite extent in ther direction, thickness 2h in thez direction, and subtends an anglef at its vertex. The displacement of pr mary concern is that in thez directionuz . This displacement has associated moment resultantsMr , M u , andMru shown acting in a positive sense in Fig. 13a, and shear resultantsQr

andQu shown likewise in Fig. 13b. All of these field quan- tities are taken to be independent ofz. Consequently, field equations hold on the 2D regionR of ~2.1! and ~3.1! for single-material and multiple-material plates, respective With these preliminaries in place, we can formulate the o of-plane bending problems of initial interest as next.

In general, we seek the out-of-plane displacementuz , and its associated resultantsMr , M u , Mru , Qr , and Qu , as functions of r and u throughoutR of ~2.1!, satisfying: the equations of equilibriumin the absence of body forces an loading on the plate faces atz56h,

]

]r ~rQr !1

]Qu

]u 50

]Mr

]r 2

1

r

]Mru

]u 1

Mr2M u

r 2Qr50 (4.12)

1

r

]M u

]u 2

]Mru

]r 2

2Mru

r 2Qu50

of e

in

te- ies

ly. ut-

d

on R; the resultant-displacement relationsfor a linear elastic plate which is both homogeneous and isotropic,

H Mr

M u J 52mbF H n

1J ¹2uzH 1

2J ~12v ! ]2uz

]r 2 G Mru5mb~12v !

]

]r S 1

r

]uz

]u D (4.13)

H Qr

Qu J 52mbH ]

]r

1

r

]

]u

J ~¹2uz!

on R, wherein mb54mh3/3(12n) is the flexural rigidity and ¹2 the Laplacian operator inr and u coordinates; any one of the first five sets of admissibleboundary conditionsin Table 13~identified as Ib– Vb therein! on the plate edge a u50, together with another such set on the edge atu5f or the bisector atu5f/2 as appropriate, for 0,r ,`; and the regularity requirementat the plate vertex,

uz5O~r ! as r→0 (4.14)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th plate vertex,O.

Several comments on the preceding formulation are order. When the plate haslateral loadingon aface, the right- hand side of the first of~4.12! is no longer zero. Provided this lateral loading is continuous or, if singular, has in grable singularities, it in itself does not produce singularit in any of the resultants.

The resultants are related to thestressesin the plate by

Fig. 13 Stress resultants on plate elements:a! positive moment resultants~element viewed fromz5h face!, b! positive shear result- ants

H Mr

M u J 5Eh H s r

su J zdz, H Qr

Q J 5Eh H t rz

t J dz

o

n

n

o

h

o

o

r

iry

d-

nts s

ons s in y to n in .

ped

m

th- n as

ed. to a lds,

s of e 1:

r ri- s

nt a-

ds n- pe- al

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 413

ns

ses

is-

ts can

ry. s

-

th ym- nse

- or ey

in ,

i- r

d- tic

he en t out em -

biharmonic. Hence we can simply use the biharmonic A stress function used to generate~1.1! ~from Williams @2#! as the displacement in bending~as in Williams@115#!. Thus we obtain, as ourbasic displacement solution for classical ben ing theory,

uz5r l11L~l, u! (4.16)

L~l, u!5c1 cos~l11!u1c2 sin~l11!u

1c3 cos~l21!u1c4 sin~l21!u

where ci( i 5124) continue as constants. Stress resulta follow from ~4.13!. Now we can apply the homogeneou boundary conditions of Table 13. These boundary conditi then turn out to have mathematically analogous condition Table 1 for in-plane extension. This enables us here simpl use the eigenvalue equations for in-plane extension give Section 2.1 for out-of-plane bending with classical theory

To explain themathematical analogyfurther, the simplest bending boundary conditions to consider are for a clam or built-in edge. Applying Conditions IIb of Table 13 onu 5f to uz of ~4.16! implies

L50, ]L

]u 50, at u5f (4.17)

These requirements onL are the same as would result fro a stress-free edge for a plate in extension~see~1.1! on re- placing (l11)c3 and (l11)c4 therein withc3 andc4). It follows that clamped conditions under bending are ma ematically analogous to stress-free conditions in extensio far as eigenvalue equations are concerned.

Similarly, other mathematical analogies can be develop For example, a free edge under bending is analogous clamped edge with extension. This last analogy only ho however, whenk in extension conditions is replaced bykb

5(31n)/(12n). All told, the following mathematical analogies hold between the bending boundary condition Table 13 and the extensional boundary conditions of Tabl

Ib→II ~l21! with k→kb , IIb or Vb→I (4.18)

III b→III, IV b→IV

Without the (l21) factor, the first of~4.18! is developed in a general context in Southwell@116#, which also observes that the second for IIb was well known circa 1950. The facto (l21) is not significant when considering power singula ties, but could play a role in identifying log singularitie whenl51. The analogy for IIIb follows directly from com- paring conditions as done for~4.17!, while the analogy for IVb is noted in Rao@48#. The equivalence of Vb and IIb follows from an adaptation of the argument in Sinclair@99# for boundary conditions which have terms with a differe r -dependence within a single condition. While the equiv lence of elastically restrained conditions with built in hol for anyl, similar arguments show elastically restrained co ditions to be equivalent to stress-free conditions for the s cial case ofl51, and to symmetry conditions for the speci

ular

Mru 2h 2t ru

u 2h uz

(4.15)

These relations are consistent with the sign conventi shown in Fig. 13. Assumings r , su , andt ru to be linear in z, it is possible to invert the first of~4.15!. Likewise, assum- ing t rz andtuz to be parabolic inz and zero atz56h, it is possible to invert the second of~4.15!. It follows that any singularity in the moment resultantsMr , M u , andMru gives rise to the same singularity in the stressess r , su , andt ru , respectively, while any singularity in the shear resultantsQr

andQu gives rise to the same singularity in the shear stres t rz andtuz , respectively.

Turning to the boundary conditions of Table 13, the h torical introduction in Love@112# credits Kirchhoff@113# as being first to advance Conditions Ib for a free edge. While it would be physically natural to insist that all three resulta be zero on a free edge, the fourth-order classical theory only admit two conditions per edge. Conditions Ib are the two that arise out of a variational development of the theo Conditions IIb are the counterparts of built-in end conditio in beam theory. Conditions Ib and IIb , respectively, are physically closest to Conditions I and II of Table 1 for in plane loading.

As previously, when the same conditions apply on b plate edges it is advantageous to distinguish between s metric and antisymmetric response. Symmetric respo about the plate bisector implies thatuz is an even function of u aboutu5f/2. Conditions IIIb ensure that this is so. Con ditions IVb physically correspond to a simply supported hinged edge: If applied on the plate bisector, however, t take on the role of antisymmetry conditions by ensuringuz is an odd function ofu aboutu5f/2.

Conditions Vb are for a plate attached to an elastic bar torsion and bending. Hencekt is the bar’s torsional stiffness kb its bending stiffness. With the present resultants,kt is positive on a positiveu edge and vice versa, whilekb is negative on a positiveu edge and vice versa. These cond tions are physically closest to Conditions VI of Table 1 f in-plane loading.23

As earlier, there are noconditions at infinityor in-plane length scalepresent in the formulation. For the reasons a vanced in Section 2.1, this is appropriate in an asympt treatment. Further, regarding theregularity requirement ~4.14!, we remark that this can be included provided t resulting formulation can be shown to be complete. Giv the analogy between the extensional case and bending se subsequently in this section, completeness would se likely, although it is not formally established. Given com pleteness, the singular fields admitted by~4.14! have bounded displacements.

Analysis proceeds on using the second and third of~4.12! to eliminateQr and Qu from the first, then substituting fo the moment resultants from~4.13!. This establishes thatuz is

23A development of the conditions for an elastically restrained plate in rectang coordinates may be found in Art 22, Timoshenko and Woinowsky-Krieger@114#.

o

t m

d

r

are

ed ,

s are

no

lar nse

nd i- of

-

For

and

re agi- gu-

ven. re

r in

ct ribu-

sses rack mal

in rmal gni- yed ess

414 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

case ofl52. This last can lead to log singularities wit elastically restrained conditions, something we investig subsequently.

From ~4.18!, then, we have the following eigenvalu equations for generall, and free (Ib), built-in (IIb), simply supported (IVb), and elastically restrained (Vb) conditions, directly from Tables 2–4 of Section 2.1: Ib2Ib ~2.10! (l21) with k→kb

for symmetric response ~2.14! (l21) with k→kb for antisymmetric response

IIb or Vb– IIb or Vb

~2.9! for symmetric response

~2.13! for antisymmetric response Ib– IIb or Vb ~2.17! (l21) with k→kb Ib– IVb ~2.14! (l21) with k→kb , f→2f IIb or Vb– IVb ~2.13! with f→2f

(4.19)

For simply supported conditions on both plate edges, corresponding eigenvalue is merely indicated at the end Section 2.1. This is because the corresponding extensi configuration is not that physically significant and cons quently has received little attention. Here, with bending, i physically important, so we give its symmetric and antisy metric equations explicitly:

IVb– IVb coslf1cosf50 for symmetric response

(4.20) coslf2cosf50 for antisymmetric.

The equations in~4.20! are consistent with~4.18! and the combined eigenvalue equation indicated at the end of S tion 2.1.

For the most part, the eigenvalue equations of~4.19! and ~4.20! are basically available in the literature. Dixon@117# gives an equation for IIb– IIb and a 90° corner. Carrier an Shaw @118# gives an equation for Ib– Ib and antisymmetric response. These equations are confirmed in Williams@115#, which also gives equations for all combinations of Ib , IIb , and IVb . When conditions are nonmixed, symmetric and a tisymmetric equations are not distinguished in William @115# but are both included in a single equation. Combined this way, corresponding equations in~4.19! and ~4.20! are either exactly the same as, or equivalent to, the equation Williams @115#.24

From ~4.13! and ~4.16!,

M5O~r l21!, Q5O~~c3 21c4

2!r l22!, as r→0 (4.21)

on R, where M is any moment resultant,Q either shear resultant. Thus provideduz is not purely harmonic~ie, pro- vided c3Þ0 or c4Þ0), the general range of eigenvalues f power singularitiesis

0<l,2 (4.22)

Singular eigenvalues in this range comply with the regula requirement~4.14!.

Singular eigenvalueswithin the range~4.22! are deter- mined in the literature as follows. For all possible combin

h ate

e

the of nal

e- is -

ec-

n- s in

s in

or

ity

a-

tions of the boundary conditions Ib , IIb , and IVb and vertex angles not exceeding 180°, Williams@115# furnishes the real parts of dominant singular eigenvalues. Typically these determined numerically.

For IVb– IVb , singular eigenvalues can be determin analytically from~4.20!. Thus for symmetric configurations

l5~2n21! p

f 21 ~~2n21!,f,np,n51,2!

l516 p

f ~p,f<2p! (4.23)

while for antisymmetric,

l5 2p

f 21 ~p,f<2p! (4.24)

For the minus sign in the second of~4.23!, the range off can be extended to includep. Some of the limits on the range for f in ~4.23! and~4.24! are because the shear resultants identically zero for these eigenvalues and thus there are singularities when 1,l,2.25 The dominant singularity for f<p comes from the first of~4.23! with n51. This eigen- value is plotted in Williams@115#.

For IIb– IIb and any vertex angle, the real parts of singu eigenvalues for both symmetric and antisymmetric respo are given in Morley@119#. Forf5270° with Ib– Ib , IIb– IIb , and IVb– IVb , singular eigenvalues are given in Hrudey a Hrabok @120#, including real and imaginary parts when e genvalues are complex. For all possible combinations boundary conditions Ib , IIb , and IVb and vertex angles be tween 180° and 360°, Leissa, McGee, and Huang@121# fur- nishes the real parts of dominant singular eigenvalues. Ib– Ib whenn50, 1/3, 1/2, IIb– IIb , and Ib– IIb whenn50, singular eigenvalues may be obtained from Seweryn Molski @20#, on using the analogies in~4.19!. In Seweryn and Molski@20#, symmetric and antisymmetric responses a distinguished and provided separately, both real and im nary parts of complex eigenvalues are furnished, and sin lar eigenvalues other than just the dominant ones are gi For Ib– IIb and f590°, complex singular eigenvalues a tabulated in Gregory, Chonghua, and Wan@122# for n 50,1/4,1/3,1/2.

The singular eigenfunctionsfor a cracked plate unde bending within fourth-order classical theory are derived Williams @123#. Under symmetric loading one might expe the tension side of the plate to have the same stress dist tion as for a Mode 1 crack in extension~developed in Will- iams @124# and Irwin @125#!. While both have inverse- square-root stress singularities and tensile normal stre acting transverse to the crack and directly ahead of the c tip, classical bending theory predicts a compressive nor stress acting parallel to the crack and ahead of it. This is marked contrast to the extensional case which has the no stress parallel to the crack being tensile and equal in ma tude to the transverse component. Given the key role pla by boundary conditions in influencing the character of str

f

24None of the (l21) factors in~4.19! are present in equations in the literature. Po sibly this is because the basic fields attending~4.16! are nonsingular forl51. Auxil- iary fields are singular however.

s- 25The Kirchhoff shear,Qu2]Mru /]r , would however be singular for wider ranges o f with 1,l,2.

i t

c

i

e

-

n

i

o

o

a

g t

a

ap-

s

is is

in - m-

s

s to late, le is lly

t its

d- ifi- nner ith tly ua-

u- in u-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 415

singularities, the arguably less physical result of class bending theory may be because the crack flanks are not stress free with fourth-order theory.

Turning to stress singularities involving logarithmi terms, at the outset these stem from auxiliary bending fie generated by differentiating~4.16! with respect tol. The ensuring development is outlined in Sinclair@126#. As for the extensional case,logarithmic intensificationof power singu- larities can be expected to occur when eigenvalues trans from complex to real. Such instances have yet to be fu checked out in the literature. Rather than logarithmic int sification, here we concentrate onpure logarithmicsingulari- ties.

There are two eigenvalues which result in what might termed pure logarithmic singularities. First forl51, pure log singularities are possible in moment resultants. Typica these are accompanied by 1/r singularities in shear result ants. That is, we have

M5O~ ln r !, Q5O~~ ĉ3 21 ĉ4

2!/r !, as r→0 (4.25)

on R whenl51, whereĉ3 andĉ4 are constants in auxiliary fields corresponding toc3 andc4 in the basic fields attending ~4.16!. Even whenĉ35 ĉ450 and the shear resultants va ish, the Kirchhoff shear has a 1/r singularity. Thus these singularities are associated with some form of concentra shear loading.

Second forl52, pure log singularities are possible shear resultants. That is, we have

M5o~1!, Q5O~~ ĉ3 21 ĉ4

2!ln r !, as r→0 (4.26)

on R when l52. These are the weakest singularities p sible and consequently the least readily detected with merical methods. Accordingly their asymptotic identificati can be of significant value.

Conditions for singularities as in~4.25! with homoge- neous boundary conditions are as in the next to last of~1.3! with nA54. Conditions with inhomogeneous boundary co ditions are as in the last two of~1.5! with nA54. Examples of corresponding boundary conditions are those associ with the constantsM0 and M̂0 in Ib8 and IVb8 of Table 13, respectively, with the other constants in these conditions ing set to zero.

Conditions for singularities as in~4.26! with homoge- neous boundary conditions are as in the next to last of~1.3! with nA54 except that nowl52. Conditions with inhomo- geneous boundary conditions are as in the last of~1.5! with nA54, but now with l52 instead of 1. Correspondin boundary conditions are those associated with the cons M08 , V0 , u0 , u08 , M̂08 , andû0 in Ib8 , IIb8 , and IVb8 , with M0

andM̂0 being set to zero~see Sinclair@126#!. As a first example of a singularity as in~4.25!, we have

the out-of-plane line load on the edge of a half-plane pla This has

M5ord~ ln r !, Qr5ord~1/r !, Qu50, as r→0 (4.27)

on R(f5p). Full fields may be obtained from Article 49 Nadai @127#. A second example is the half-plane plate ag but now under a step moment on its edge~Fig. 14a!. This has

cal ruly

lds

tion lly n-

be

lly,

-

ted

n

s- nu- n

n-

ted

be-

ants

te.

, in

fields as in~4.27! but with the roles ofQr andQu reversed. A third example is the hinged quarter-plane plate under plied moments~Fig. 14a8!. This has

M5ord~ ln r !, Qr5Qu50, as r→0 (4.28)

on R(f5p/2). The Kirchhoff shear, though, still behave like 1/r as r→0.

The presence of 1/r terms make it unlikely that any of the foregoing could pass undetected in a stress analysis. Th not the case for singularities as in~4.26!. Accordingly, we list all configurations known to have singularities of this form Table 14~from Sinclair@126#!. For elastically restrained con ditions, some of these stem from their equivalence with sy metry conditions for the special case ofl52.

In Table 14,fb is such that

4kb sin2 fb5~kb11!~26A42kb! (4.29)

If in addition to ~4.29!, kb52tanfb /fb cos 2fb , by defini- tion, thenk5k̂b fb5f̂b . Two examples of singularities a in ~4.26! are shown in Figs. 14b and b8. The first is for a half-plane plate with a step shear and thus quite analogou the extensional case. The second is for a quarter-plane p hinged on one edge with shear on the other. This examp perhaps less obvious than the first, although it is rea equivalent to it. Even less obvious in Table 14 is Ib8– Ib for the half-plane whenM u5M08r : Here the moment resultan actually varies continuously along the plate edge, though derivative does not.

Turning to plates made of multiple materials under ben ing, there are relatively few instances of singularity ident cation compared to the extensional case. However, Fe @80# shows that perfectly bonded conditions in bending w fourth-order theory are effectively equivalent to perfec bonded conditions in extension as far as eigenvalue eq

Fig. 14 Examples of configurations with logarithmic stress sing larities:a!, a8! plates with applied moments and log singularities s r , su , andt ru ; b!, b8) plates with applied shears and log sing larities in t rz andtuz .

416 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Table 14. Configurations with logarithmic singularities in shear resultants

Boundary conditions on uÄ0,f

Configuration specifications „mÄ1,2…

Ib8– Ib f5p or 2p, M 08Þ0 or V0Þ0 kb56secf, and

M08~kb12!fStan f

2D61

Þ6V0~22kb!

IIb8– IIb or Vb f5p or 2p, u0Þ0 or u08Þ0 IVb8– IVb8 f5p or 2p, M 08Þ6û0(12n)mb Ib– IIb or Vb f5f̂b , kb5k̂b Ib8– IIb8 f5fb , kbÞk̂b and

(M0826u0(12n)mb cos 3f)(3 sin 3f2(kb12)sinf) Þ(V012u08(12n)mb cos 3f)(3 cos 3f1(kb22)cosf)

Ib8– IVb8 f5~2m21!

p

2 , and

24û0(12v)mbÞM̂ 08(kb15)2( – )mV0(kb11) f5mp, M08Þ( – )mM̂ 08

Ib8– Vb8 f5p or 2p, V0Þ0 f5p/2 or 3p/2, M08Þ0 kb5sec 2f, M 08(kb12)tanfÞV0(22kb)

IIb8– IVb8 f5~2m21!

p

2 , and

2M̂08Þ(12v)mb(3(32kb)û02( – )m(kb11)u08) f5mp, û0Þ( – )mu0

IIb8– Vb f5p/2 or 3p/2, u0Þ0 f5p or 2p, u08Þ0

IVb8– Vb f5p/2 or 3p/2, M̂ 08Þ6û0mb(12n)

e

n

a

n

n

te as

tions are concerned. Herein perfect bonding in bend matchesuz , ]uz /]u, M u , and Qu2]Mru /]r . Then the equivalence holds provided

k→kb , m1 /m2→m2 /m1 (4.30)

wherem1 andm2 are the shear moduli for material on eith side of the interface. Thus using the analogies in~4.19!, ei- genvalue equations for perfectly bonded interfaces can obtained from extensional counterparts~Section 3!.

There are two instances of bimaterial plates under be ing explicity treated in the literature. The first concerns crack meeting an interface~Figs. 6b and b8). Fenner@80# determines singular eigenvalues for any angle of incide and a range of ratios. The singular eigenfunctions for crack parallel to the interface may be obtained from Sih a Rice @128#. The second concerns a bonded bimaterial pl Huang@129# computes singular eigenvalues for a variety geometries and a range of moduli ratios.

4.4 Out-of-plane bending: Higher-order theory

In this section we consider the singularities that can occu the out-of-plane bending of an elastic plate when trea within sixth-order theory. The particular theory considered due to Reissner@130#. This is the sixth-order theory that ha received most attention when it comes to singularity ide fication. We do comment briefly, though, on a similar theo in Hencky @131#.

The angular elastic plate to be bent continues to be a Fig. 12. As for classical theory, the out-of-plane displac ment of the plate isuz , its moment resultants areMr , M u , and Mru , and its shear resultants areQr and Qu ~positive resultants shown Fig. 13!. In addition, the plate has rotation v r and vu . All of these field quantities are taken to be i

ing

r

be

nd- a

ce the nd te. of

r in ted is

s ti- ry

s in e-

s -

dependent ofz so that field equations continue to apply onR of ~2.1!. With these preliminaries in place, we can formula the class of out-of-plane bending problems of interest next.

In general, we seek the out-of-plane displacementuz , its rotationsv r andvu , and its associated resultantsMr , M u , Mru Qr , and Qu as functions ofr and u throughoutR, satisfying: theequations of equilibriumin the absence of body forces and loading on the plate faces,~4.12! on R; the resultant-displacement/rotation relationsfor a linear elastic plate which is both homogeneous and isotropic,

H Mr

M u J 52mbF H n

1J ¹2uzH 1

2J ~12n! ]2uz

]r 2 G H 1

2J 4h2

5

]Qr

]r

Mru5mb~12n! ]

]r S 1

r

]uz

]u D2 2h2

5 S 1

r

]Qr

]u 1r

]

]r S Qu

r D D (4.31)

H Qr

Qu J 5

5mb~12n!

4h2 H v r1 ]uz

]r

vu1 1

r

]uz

]u

J Table 15. Boundary conditions for out-of-plane bending with Reiss- ner’s theory

Identifying Roman numeral

Boundary conditions

Physical description

IB M u50, Qr50, Qu50 Stress free IIB uz50, v r50, vu50 Clamped III B vu50, Mru50, Qu50 Symmetry IVB uz50, v r50, M u50 Simply supported

e

e r

d

e

o

e

e

a

ts -

n

ble

nd a

qua- ided

on

lair

- lue

the

rton

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 417

on R, whereinmb remains as the flexural rigidity; any one o the sets ofboundary conditionsIB , IIB , or IVB in Table 15 on the plate edge atu50, together with another such set o the edge atu5f, or one of Conditions IIIB and IVB on the plate bisector atu5f/2 as appropriate, for 0,r ,`; and the regularity requirementat the plate vertex,

uz5o~1!, v r5O~1!, vu5O~1!, as r→0 (4.32)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th plate vertexO.

Several comments on the preceding formulation are order. First, as with classical theory, the presence of lat loading has no effect on the nature of any singularities p vided the loading is integrable. Second, the resultants c tinue to be related to stresses as in~4.15! so that singularities in resultants lead to singularities in corresponding str components. Third, the boundary conditions now presc three quantities per edge~cf, Table 13!: Accordingly, they enable the physically natural conditions attending a stre free edge to be enforced. Fourth, when the same boun conditions apply on both plate edges, Conditions IIIB and IVB enable one to distinguish between symmetric and a symmetric response—in this role, Conditions IVB are equivalent to antisymmetry conditions. Fifth, there are conditions at infinity nor should there be in this asympto formulation. Sixth and last, the regularity requirement is co sistent with classical theory and therefore analogous to lier such requirements: However, absent a formal proof completeness for Reissner’s theory, it is provisional at t time.

The theory in Hencky@131# has the same equations equilibrium and boundary conditions as Reissner’s theo Differences occur in the resultant-displacement/rotation r tions. In the absence of loading on the plate faces, th differences are confined to the numerical coefficients ofh2

terms in ~4.31!. Moreover, these differences are consist throughout~4.31!. As a result, Reissner’s theory can be tran formed into that of Hencky simply by making the transfo mationh→A5/6h in ~4.31! whereverh occurs explicitly~ie, there is no change tomb). It follows that we can expect the singularities in Hencky’s theory to stem from the same genvalue equations as for Reissner’s theory once this tr formation is implemented.

Analysis proceeds on introducing a stress functionx so that the first of~4.12! is satisfied. Thus

Qr5 1

r

]x

]u , Qu52

]x

]r (4.33)

on R. Substituting~4.31! and ~4.33! into the last of~4.12! then gives

]

]r S x2 2h2

5 ¹2x D5

mb

r

]

]u ~¹2uz!

(4.34)

1

r

]

]u S x2 2h2

5 ¹2x D52mb

]

]r ~¹2uz!

f

n

of e

in ral

ro- on-

ss ibe

ss- ary

nti-

no tic n- ar- of

his

f ry. la- ese

nt s- r-

ei- ns-

on R. In solving~4.34!, we need solutions with six constan sharing a common power ofr in order to meet the six bound ary conditions, three to an edge, which hold for allr . To this end, Burton and Sinclair@132# use a series approach with

uz5r l11L~l,u!1O~r l13! (4.35)

x5r l11L̂~l,u!1O~r l13!

on R. HenceL is as in ~4.16!, L̂ likewise with ci ( i 51 24) exchanged forĉi . Then relating the dominant term o the left-hand side of~4.34! asr→0, the¹2x term, to that of the right-hand side relatesĉ3 and ĉ4 to c3 and c4 . This leaves six free constants as desired (c12c4 , ĉ1 , and ĉ2). The resulting fields for the dominant terms asr→0 are given in @132#.

What is overlooked in Burton and Sinclair@132# is the possibility ofx terms on the left-hand side of~4.34! interact- ing with the right whenx is purely harmonic. This omission is corrected in Yen and Zhou@133#, the resulting additional field being given in@133#.

Substituting the fields in Burton and Sinclair@132# into the various combinations of boundary conditions availa from Table 15 leads to associatedeigenvalue equations. With the exception of the equation for IVB– IVB , these equations are confirmed in Yen and Zhou@133#. This confirmation oc- curs because the additional solution available in Yen a Zhou @133# does not actively participate other than for plate simply supported on both edges. These confirmed e tions then are the same as extensional counterparts prov k takes on its value for plane stress (k5(32n)/(11n)). Accordingly, drawing on results from Tables 2–4 of Secti 2.1, we have:26

IB– IB ~2.9! for symmetric response ~2.13! for antisymmetric response

IIB– IIB ~2.10! for symmetric response ~2.14! for antisymmetric response

IB– IIB ~2.17! IB– IVB ~2.13! with f→2f IIB– IVB ~2.14! with f→2f

(4.36)

For IVB– IVB , the additional field of Yen and Zhou@133# is active and the eigenvalue equation in Burton and Sinc @132# has an additional factor. Accordingly we have:

IVB– IVB ~coslf6cosf!cos~l21! f

2 50 (4.37)

The plus sign in~4.37! is associated with symmetric re sponse, the minus sign with antisymmetric. The eigenva equations in~4.36! and ~4.37! are independent ofh. Conse- quently, the singular eigenvalues in Hencky’s theory are sameas those in Reissner’s theory.

For ~4.35! from ~4.31! and ~4.33!,

M5O~r l21!,Q5O~r l! (4.38)

v5O~r l!,uz5O~r l11!,as r→0

26There is a typographical error in the second eigenvalue equation in Table 1 of Bu and Sinclair@132#. The correct result is given in~4.36!.

n

e

n

e

e

t

l y

l ay

ent ults ks aly-

As

-

nd or en at in

s

ions der ion l-

p se log

t is hear

e- ld

ang up-

in al sed r. d to

lf- e

re

418 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

on R, wherev is either rotation component. From~4.32!, admissiblepower singularitiesin moment resultants then oc cur for

0<Rel,1 (4.39)

For the additional solution of Yen and Zhou@133#,

M5O~r l23!, Q5O~r l22! (4.40)

v5O~r l22!, uz5O~r l11!, as r→0

on R. Admissible power singularities in moment resultan occur for

2<Rel,3 (4.41)

Observe that, with either~4.38! and ~4.39! or ~4.40! and ~4.41!, M is singular whileQ is not, in contrast to classica theory whereQ is typically more singular thanM . Singular eigenvaluesas in ~4.39! for ~4.38! can be obtained directly from their extensional counterparts~see Section 2.2!.

For simple supported conditions, singular eigenvalues in ~4.39! for ~4.38! from the first factor in~4.37! are included in ~4.23! and ~4.24!. Singular eigenvalues as in~4.41! for ~4.40! from the second factor in~4.37! are:

l5 p

f 11 ~p/2,f<p!

(4.42)

l5 3p

f 11 ~3p/2,f<2p!

These eigenvalues are given in Yen and Zhou@133#. There is a further singular field for simply supported co

ditions not included in@132,133#. This has

M5O~r p/f!, Q5O~r p/f21! (4.43)

v5O~r p/f11!, uz5O~r p/f!, as r→0

on R. Thus forf.p, this field hasQ singular,M not, in contrast to the results in@132,133#. This singularity is iden- tified in Huang, McGee, and Leissa@134#.

Turning to companion eigenfunctions for Reissne theory, for the fields associated with~4.39!, both the r -dependence and the individual functions ofu in eigenfunc- tions can be shown to be the same as extensional cou parts~see Williams@2# and Burton and Sinclair@132#!. Thus all that is needed for eigenfunctions to coincide complet to within a multiplicative factor is that the weighting of th individual functions ofu be the same in bending as in exte sion.

That this in fact can occur is demonstrated for the cas a crack in a plate of vanishing thickness in Knowles a Wang@135#. For a plate of finite thickness, it is demonstrat for a crack in Hartranft and Sih@136# and Wang@137#. Hence the tensile side of the plate in bending behaves as is a crack in a plate in extension, a physically reasona result in contrast to that of classical theory.

On the compression side of the plate, the eigenfunc for the crack in Reissner’s theory leads to interpenetration overlapping of the crack’s flanks. While this is not physica possible, it is nonetheless possible in an elastic anal

-

ts

l

as

n-

r’s

ter-

ly e -

of nd d

if it ble

ion or

ly sis

within Reissner’s theory~it is also possible with classica theory!. In some instances, overlapping displacements m be negated by the addition of an in-plane tension of suffici magnitude. Otherwise, more physically appropriate res can only be obtained by entertaining contact of the flan and tracking this contact as loading proceeds. Such an an sis is really 3D, as well as being geometrically nonlinear: such, it is outside the scope of this section.

For logarithmic terms underhomogeneousboundary con- ditions, conditions like~1.3! are indicated in Burton and Sin clair @132# for fields stemming from~4.35!. No actual in- stances of these singularities are identified in Burton a Sinclair @132#. Logarithmic terms can be generated f simply-supported conditions and the additional fields of Y and Zhou@133#, although it remains to be determined wh additional conditions these logarithmic fields must meet order to participate.

There is also the possibility of logarithmic singularitie being induced byinhomogeneousboundary conditions. This type of response can be expected to include configurat that correspond to instances of log stress singularities un inhomogeneous boundary conditions for plates in extens ~Section 2.4!. Some support of this expectation being fu filled can be found in Hartranft@138#. There a log singularity is identified inMr in response to a step inMru on a plate edge (f5p), whereas no log singularity is found for a ste in M u . Thus this situation is analogous to the in-plane ca wherein a step in shear on a half-plane plate produces a singularity ~Table 7, Section 2.4!, while a step in normal stress does not. A further log singularity inQr in response to a step inQu on a plate edge is identified in Hartranft@138#. While this configuration has no counterpart in extension, i analogous to the antiplane shear case wherein a step in s on a half-space wedge produces a log singularity~Table 10, Section 4.1!.

For plates made of multiple materials treated within R issner’s theory, the only singularity identifications that wou appear to be available in the literature are those in Hu @129#. These are for closed bimaterial plates and simply s ported bimaterial plates.

5 STRESS SINGULARITIES FOR OTHER ELAS- TIC CONFIGURATIONS

5.1 Axisymmetric configurations

A representative axisymmetric configuration is sketched Fig. 15. This depicts a right circular cylinder with a conic cap bonded into a half-space which in general is compri of a material with distinct elastic moduli from the cylinde The end of the cylinder above the half-space is subjecte an applied torqueT and an axial forceF. Stress singularities can be expected at the vertex of the conePs , and at points where the perimeter of the cylinder is bonded to the ha space~eg, Ps8). For each location, we wish to consider th singularities that may be induced by either the torqueT or the forceF. Thus we have four configurations: the inner co

ur-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 419

Fig. 15 Singular axisymmetric configurations

i

s

of e

on.

l

ents ts

l- at

ly ri-

r-

e

In general, we seek the axisymmetric shear stressestru

andtuc and their companion displacementuu , as functions of r and c throughoutR, satisfying: thestress equation of equilibrium absent body force,

r ]tru

]r 1

]tuc

]c 13tru12tuc cotc50 (5.3)

on R; the stress-displacement relationsfor a linear elastic cone which is both homogeneous and isotropic,

tru5mS ]uu

]r 2

uu

r D , tuc5 m

r S ]uu

]c 2uu cotc D (5.4)

on R, whereinm continues as the shear modulus; thebound- ary conditionfor either a clamped or a stress-free cone s face,

uu50 or tuc50 at c5f/2 (5.5)

for 0,r,`; and theregularity requirementat the cone ver- tex,

uu5O~1! as r→0 (5.6)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th cone vertexO.

To solve the preceding problems, first substitute~5.4! into ~5.3!. Then seeking a separable solution foruu of the form rl f (c) leads to Legendre’s associated differential equati Hence, for boundeduu whenc50,

uu5clrlPl 1~cosc! (5.7)

Here cl is a constant coefficient, andPl 1 is an associated

Legendre function of the first kind of degreel and order one.27 The eigenvalue equations attending~5.4!, ~5.5!, and ~5.7! are developed and solved numerically in Bazˇant and Keer @140#. No roots are found in the range 0,l,1 irre- spective of cone vertex anglef. Consequently,no power singularitiesare found for the cone vertex under torsion.

For acone vertex under axial loading, the same spherica polar coordinates are appropriate~Fig. 16!. Now there are two displacements,ur anduc . However, both are still only functions ofr andc, so that the region of interest remainsR of ~5.2!. On this region we can formulate ourinner axial problemsas next.

In general, we seek the axisymmetric stress compon sr , sc , su , andtrc , and their companion displacemen ur anduc , as functions ofr andc throughoutR, satisfying: the stress equations of equilibriumabsent body forces,

r ]sr

]r 1

]trc

]c 12sr2sc2su1trc cotc50

(5.8) ]sc

]c 1r

]trc

]r 13trc1~sc2su!cotc50

on R; the stress-displacement relationsfor a linear elastic cone which is both homogeneous and isotropic,

27The functionPl 1 is as defined in Ch 8, Abramowitz and Stegun@139#.

vertex with torsion or with axial loading, and the outer cy inder boundary with torsion or with axial loading. We tre each of these in turn in what follows.

For a cone vertex under torsion, spherical polar coordi- natesr, c, andu enable the asymptotic problem to be read formulated~Fig. 16!. These coordinates share a common o gin O with the rectangular Cartesian coordinatesx, y, andz, and are related to them by:

x5r sinc cosu, y5r sinc sinu, z5r cosc (5.1)

for 0<r,`, 0<c<p, and 0<u,2p. Under pure torsion, the only displacement is that in theu direction,uu , which is a function ofr andc alone. Hence the open region of inte estR is

R5$~r, c !u0,r,`, 0,c,f/2% (5.2)

where f is now the vertex angle of the cone. With the geometric preliminaries in place, we can formulate ourinner torsion problemsas next.

Fig. 16 Spherical polar coordinates for a cone vertex

n

t

-

i

h -

i

u

m- or

or

w stic ar

n

n

and

y

420 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

sr52mF nQ

122n 1

]ur

]r G sc52mF nQ

122n 1

1

r S ]uc

]c 1urD G

(5.9)

su52mF nQ

122n 1

1

r ~ur1uc cotc!G

trc5mF1

r

]ur

]c 1

]uc

]r 2

uc

r G with

Q5 ]ur

]r 1

2ur

r 1

1

r

]uc

]c 1

uc

r cotc (5.10)

on R, whereinQ continues as the dilatation,n as Poisson’s ratio; theboundary conditionsfor either a clamped or stress free cone surface,

ur5uc50 at c5f/2 (5.11)

or

sc5trc50 at c5f/2 (5.12)

for 0,r,`; and theregularity requirementsat the cone vertex,

ur5O~1!, uc5O~1!, as r→0 (5.13)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th cone vertexO.

Thompson and Little@141# uses Papkovich-Neuber pote tials to develop solutions for the preceding field equatio and also derives the eigenvalue equation for~5.12!. Using the same solutions, Bazˇant and Keer@140# derives the eigen- value equation for~5.11!, and solves both eigenvalue equ tions numerically. Power singularities are found for bo clamped and stress-free conditions for reentrant cone ver ~ie, p,f<2p). The singularity exponents involved are a real and depend on the value of Poisson’s ratio. Singula exponents are given forn ranging from 0–0.499 in incre ments of 0.1 in Bazˇant and Keer@140#. Exponents forn 50.3 are confirmed in Beagles and Sa¨ndig @142#.

Further singularities for a cone vertex under axial load occur when the cone isrigid andindentsan elastic half-space ~cf, Fig. 15 nearPs). Herein the boundary conditions in th contact region are

uc5u02r cotf/2, trc5 f sc , at c5p/2 (5.14)

whereu0 is the penetration of the cone vertex andf contin- ues as the coefficient of friction. The frictionless case of t configuration is analyzed in Love@143# and leads to a loga rithmic stress singularity at the cone vertex. The friction ca also has a log singularity~Hanson@144#!.

Bimaterialcone vertices~as atPs in Fig. 15! are analyzed in Keer and Parihar@145#. Perfect bonding on the interface assumed~ie, sc , trc , ur , anduc are matched atc5f/2). Power singularities are identified for varying elastic mod and cone vertex angles. In contrast to the single mate

-

of e

- ns,

a- th ices ll rity

ng

e

is

se

s

li rial

cone, some of the singularity exponents involved are co plex. However, the magnitude of the singularity exponent, of its real part if complex, is bounded from above by that f clamped conditions in Bazˇant and Keer@140#.

By way of an example of acylindrical boundary under torsion, we reconsider the configuration in Fig. 15, but no with the cylinder and half-space comprised of a single ela material. For this and like configurations, cylindrical pol coordinates,r , u, andz enable ready formulation~Fig. 17!. These coordinates share a common originO with the rectan- gular Cartesian coordinatesx, y, and z and are related to them by

x5r cosu, y5r sinu, z5z (5.15)

for 0<r ,`, 0<u,2p, and 2`,z,`. Under pure tor- sion, the only displacement is that in theu direction, uu , which is a function ofr andz alone. Hence the open regio of interestR becomes

R5$~r , z!u0<r ,R,0<z,` or 0<r ,`,2`,z,0% (5.16)

whereR is the radius of the cylinder. On this region we ca formulate our sampleouter torsion problemas next.

In general, we seek the axisymmetric shear stressest ru

andtuz , and their companion displacementuu , as functions of r and z throughoutR, satisfying: thestress equation of equilibrium absent body force,

]t ru

]r 1

2t ru

r 1

]tuz

]z 50 (5.17)

on R; the stress-displacement relationsfor a linear elastic cylinder and half-space which are also homogeneous isotropic,

t ru5mS ]uu

]r 2

uu

r D , tuz5m ]uu

]z (5.18)

on R; the stress-freeboundary conditions,

Fig. 17 Cylindrical polar coordinates for a cylindrical boundar

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 421

Fig. 18 Section through the cylinder and the half-space with lo coordinates at the reentrant corner

t

o

t e

e ss-

rt

r

tion

ny nd- se

are of s s,

in rack

cted

et

of e

, a . ti-

re- any ut

gu-

. ew

s

em. s,

]

]r 52sinû

]

] r̂ 2

cosû

]

]û (5.23)

]

]z 5cosû

]

] r̂ 2

sinû

]

]û

Then ~5.18!, ~5.22!, and~5.23! give

t r̂ ẑ5m ]uẑ

] r̂ 1O~R21! as R→`

(5.24)

tu ẑ5 m

]uẑ

]u 1O~R21! as R→`

In ~5.24!, the limit R→` corresponds to approaching th corner atÔ. Thus under this limit we have the same stre displacement relations as for antiplane shear~cf, ~4.2! of Section 4.1!.

Turning to the stress equation of equilibrium, we inve ~5.22! to obtaint ru and tuz in terms oft r̂ ẑ and tû ẑ , then substitute into~5.17!. This gives, using~5.23!,

]t r̂ ẑ

] r̂ 1

t r̂ ẑ

r̂ 1

1

]tû ẑ

]û 1O~R21!50 as R→` (5.25)

Again, we have the same equation as for antiplane shear~cf, ~4.1! of Section 4.1!. Further, the boundary conditions fo our sample problem simply are

tû ẑ50 at û50, 3p/2 (5.26)

for 0, r̂ ,`. Thus we have in~5.24!–~5.26! an antiplane shear problem belonging to the class formulated in Sec 4.1. Consequently, from~4.7!,

t5O~ r̂ 21/3! as r̂→0 (5.27)

wheret is either shear stress. There is nothing special about our sample problem. A

other feature on the cylindrical boundary can have its bou ary conditions transformed so that locally they match tho of Table 9, Section 4.1. Further, when multiple materials involved, interface conditions can be matched with those Table 11, Section 4.2. It follows that all of the singularitie identified in Sections 4.1 and 4.2, including log singularitie apply to corresponding outer torsion problems.

Demonstrations of this correspondence are available the literature. Early examples are the penny-shaped c under torsion in Section 5.4, Neuber@17#, and torsion of a rigid disk on a half-space in Reissner and Sagoci@146#. Both have inverse-square-root singularities as would be predi from ~4.7! with f52p and from~4.8! with f5p, respec- tively. For a general V-notch in a pipe under torsion, Tsuji al @147# obtains singularities as in~4.7!. This paper also

cal

t ru50 at r 5R ~0,z,`! (5.19)

tuz50 at z50 ~R,r ,`!

and theregularity requirementat the reentrant corner,

uu5O~1! as A~R2r !21z2→0 (5.20)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th reentrant corner atr 5R andz50.

One expects that in the local vicinity of greatest interes state ofout-of-planeor antiplane sheardominates response If this is so, we can simply draw on the singularities iden fied for antiplane shear~Sections 4.1 and 4.2! to identify the singularities possible in outer torsion problems.

To show that antiplane shear indeed characterizes sponse for the case of our sample problem, as well as for other outer torsion problem, we proceed as follows. With loss of generality we consider a section through the confi ration of Fig. 17 on they axis for y>0. For this section we introduce local cylindrical polar coordinatesr̂ , û, andẑ with origin Ô at the reentrant corner~Fig. 18!. These are related to rectangular Cartesian coordinatesx̂, ŷ, and ẑ sharing the same originÔ as in ~5.15! with carets: Consistent with a right-handed system,ẑ is positive out of the plane of Fig. 18 Then the original coordinate system is related to the n local system in the plane of Fig. 18 by:

r 5R2 r̂ sinû, z5 r̂ cosû (5.21)

For the shear stresses of the local coordinate system,t r̂ ẑ and tû ẑ , equilibrium of a pair of appropriately oriented triangle leads to

t r̂ ẑ5t ru sinû2tuz cosû (5.22)

tû ẑ5t ru cosû1tuz sinû

Now we wish to substitute~5.18! into ~5.22! to determine the stress-displacement relations in the local coordinate sys In order for the results to be in terms of the local quantiti we make the exchangeuu52uẑ and invoke the chain rule to obtain

s t

n

a

e

s

nd b-

ar- t be-

ns, nd-

the t is ms ar-

hat d- d-

in rack

to

rre- f a nd of

ial y

in n. re-

he

re- We sses

rve

422 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

treats a bimaterial corner and finds the same stress singu ties as for antiplane shear. Other examples of outer tor problems with bimaterials leading to the same singulari as for antiplane shear may be found in Freeman and K @148#, Westmann@149#, and Keer and Freeman@150#. A tri- material example is given in Keer and Freeman@111#. Still further examples exist in the literature: The preceding is an extensive list but merely intended to reflect the variety configurations displaying the correspondence.

By way of example of acylindrical boundary under axial loading, we continue to use our sample geometry of a sin material version of Fig. 15, but now under an axial lo instead of a torque. The cylindrical polar coordinates,r , u, and z of ~5.15!, Fig. 17, then continue to be appropriat Now, though, we have the two displacements,ur and uz . However, spatial dependence continues to be just onr andz so thatR of ~5.16! continues to be the region of interest. O this region we can formulate our sampleouter axial problem as next.

In general, we seek the axisymmetric stress compon s r , su , sz , andt rz , and their companion displacementsur

anduz , as functions ofr andz throughoutR, satisfying: the stress equations of equilibriumabsent body forces,

]s r

]r 1

]t rz

]z 1

s r2su

r 50

(5.28) ]sz

]z 1

]t rz

]r 1

t rz

r 50

on R; the stress-displacement relationsfor a linear elastic cylinder and half-space which are also homogeneous isotropic,

s r52mS nQ

122n 1

]ur

]r D , su52mS nQ

122n 1

ur

r D (5.29)

sz52mS nQ

122n 1

]uz

]z D , t rz5mS ]ur

]z 1

]uz

]r D with dilatation

Q5 ]ur

]r 1

ur

r 1

]uz

]z (5.30)

on R; the stress-freeboundary conditions,

s r5t rz50 at r 5R ~0,z,`! (5.31)

sz5t rz50 at z50 ~R,r ,`!

and theregularity requirementsat the reentrant corner

ur5O~1!, uz5O~1!, as A~R2r !21z2→0 (5.32)

on R. In particular, we are interested in the local behavior the fields complying with the foregoing in the vicinity of th reentrant corner atr 5R andz50.

One expects that, in the local vicinity of greatest intere a state ofplane straindominates response. This is becau for a section such as that on they axis, plane strain has]/]u being a null operator, in common with axisymmetry~u as in Fig. 17!. If, in fact, such a correspondence holds, we c

lari- ion ies eer

ot of

gle d

e.

n

nts

and

of e

st, e,

an

simply draw on the singularities identified in Sections 2 a 3 to identify the singularities possible in outer axial pro lems.

That indeed this correspondence occurs is argued in B ton @151# and Zak @152#. The approach is similar to tha presented here for the simpler torsion problem—simpler cause fewer field quantities are involved. While Zak@152# explicitly treats stress-free and clamped boundary conditio the correspondence is equally applicable to the other bou ary conditions in Tables 1 and 6 of Section 2, and to interface conditions in Table 8 of Section 3. The end resul that not just our sample problem but all outer axial proble have corresponding plane strain configurations which ch acterize their stress singularities. Conversely, it follows t all of the singularities identified in Sections 2 and 3, inclu ing those involving logarithmic terms, apply to correspon ing outer axial problems.

Demonstrations of this correspondence are available the literature. Early examples are the penny-shaped c under transverse tension in Sneddon@153#, and the rigid, right circular cylinder, with a flat lubricated end, pressed in a half-space in Harding and Sneddon@154#. Both have inverse-square-root singularities as is predicted by co sponding plane strain configurations. A further example o clamped-free right-angled corner is given in Benthem a Minderhoud@155#, and shares the same singularity as that ~2.17! in Section 2.1. An example for a bonded bimater cylinder is given in Agarwal@156#, and leads to essentiall the same eigenvalue equation for singularities as given Bogy @85# for the corresponding plane strain configuratio Again, the references listed here are merely intended to flect some of the variety of configurations displaying t correspondence.

5.2 Three-dimensional geometries with continuous ver- tex paths

In this section, we continue to be interested in angular gions, but now these regions are 3D rather than 2D. distinguish three classes of such geometries. These cla

Fig. 19 Bimaterial elastic wedge with vertex locus a smooth cu

t a ct- an

s in w ex-

tri- is. ely the eri-

be cal this

to cal

h

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 423

Fig. 20 Constant pressure on a surface sector of an elastic space

i

of

t , e- re-

r

sly in

nti- 4.2. ed per-

r- tic

in- ck rse-

e- ar-

- ch in-

i- r is

ght cta-

t a le as

t a its ed n

iii ! Geometries wherein the vertex involved traces ou discontinuous path. An example is the crack interse ing a free surface in Fig. 21: Here the vertex has angle of 2p and its path ceases abruptly atO.

We review singularities identified for each of these classe turn. We begin with the first two in this section, then revie the third in a separate following section because of the tensive number of contributions for this class.

In undertaking these reviews, we do admit some con butions entailing a significant amount of numerical analys Previously, in two dimensions, analysis was almost entir analytical: Numerical analysis was essentially confined to calculation of eigenvalues from equations derived, a num cal process that need have no errors effectively, as may verified by back substitution. Here we entertain numeri methods that do entail numerical approximations. We do because of the greater intractability of 3D geometries purely analytical approaches. We still focus, though, on lo singularity identification rather than on global analysis singular problems.

For thefirst class of3D geometries, the key general resul is established in Aksentian@90#. For the geometry of Fig. 19 Aksentian@90# proves the asymptotic equivalence of 3D r sponse at the vertex to the combination of plane strain sponse~in the yz plane in Fig. 19! and out-of-plane shea response~in the x direction in Fig. 19!.28 Thus, when the vertex in a 3D configuration has a path with a continuou turning tangent, all the plane strain singularities identified Sections 2 and 3 can participate, together with all the a plane shear singularities identified in Sections 4.1 and Although eigenvalue equations are only explicitly deriv for stress-free and clamped boundary conditions and fectly bonded interface conditions in Aksentian@90#, the ap- plicability of the equivalence for other boundary and inte face conditions follows immediately from the asympto governing equations,~1.5!–~1.7!, @90#, provided footnote 17 of Section 3.2 is observed.

An example of this equivalence between 2D and 3D s gularities occurs for the 3D problem of a flat elliptical cra under transverse tension. This shares the same inve square-root singularity of a crack in plane strain~Sadowsky and Sternberg@158# and Green and Sneddon@159#!. When this elliptical crack is loaded in shear as well, the invers square-root singularity of a crack in antiplane shear also p ticipates~Kassir and Sih@160#!. Another example is the con tact of an elastic half-space by a flat, frictionless, rigid pun of elliptical cross section. Again, the inverse-square-root s gularity of plane strain is present~Galin @161# and Green and Sneddon@159#!. When a torsional shear is applied in add tion, the inverse-square-root singularity of antiplane shea added~Mindlin @162#!.

alf-

an

are arranged in order of decreasing continuity. As one m expect, consequently they are in order of decreasing tra bility.

The three classes of 3D geometries are as follows:

i! Geometries wherein the vertex involved traces ou path with a continuously turning tangent. An examp is the bimaterial wedge in Fig. 19: Here the vertex h an angle off11f2 and its path follows a smooth curve.

ii ! Geometries wherein the vertex involved traces ou path that is continuous but has a discontinuity in direction. An example is the half-space with a load surface sector in Fig. 20: Here the ‘‘vertex’’ has a angle ofp and its path turns abruptly through 2p2f̂ at O.

Fig. 21 Elastic half-space with a crack terminating at its surfa

28For the case of a single material (f250) and a crack (f152p), the same result is given in Hartranft and Sih@157#. There, however, the result is assumed rather th proven.ce

a

t

o

h

n

n t

x

,

e

the n

- Also

that

h it

ibu-

a

u-

f

di-

c- the

e

nd-

424 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

As an initial instance of thesecond class of3D geom- etries, we consider an elastic half-space loaded with a pr surep on a surface sector which subtends an anglef̂ ~Fig. 20!. Hence in this problem, in general, stresses and displ ments are sought throughout the half-space satisfying: 3D field equations of elasticity,29 and theboundary condi- tions, on z50,

sz5H 2p for 2f̂/2,u,f̂/2 0 for other u,

(5.33)

tyz5tzx50 for 2p,u<p

all for 0,r ,` ~see Fig. 20 for coordinates!. In particular, in the surface atz50, the stress componen

in rectangular coordinates are given by

H sx

sy J 5 H 2

1J p~122n!

2p sinf̂ ln r as r→0 (5.34)

with txy50, wheren is Poisson’s ratio of the half-space. F f̂5p/2, this state of pure shear with a log singularity consistent with results derived in Love@164#. For generalf̂, the log singularity of~5.34! is identified in Turteltaub and Wheeler@165#.

Notice that whenf̂52p, no singularity occurs in~5.34!. This is as it should be when the half-space surface is loa throughout with a constant pressure. Further, whenf̂5p, no singularity occurs. This is consistent with the 2D problem a step normal pressure on a half-space in plane strain w has no singularity~Michell @166#!. It is also consistent with the teaching of Aksentian@90# since then the configuration i Fig. 20 is a member of the first class of 3D geometries.

The means of singularity identification in Turteltaub a Wheeler@165# is via asymptotics on line integral represen tions and is quite analytically sophisticated. With further a plication, no doubt it could produce results for other norm loadings. Too, it can treat shear tractions on the half-sp surface~see later!. Here, instead, we next develop a mo limited approach. While this approach cannot treat sh tractions, it is simple for normal loadings.

A potential representation of the stresses within an ela half-space free of surface shear tractions is given in Sec 5.7, Green and Zerna@167#. This representation can be e pressed in terms of a single harmonic functionC 5C(r ,u,z), with r , u, and z as in Fig. 20. AssumingC admits to a Taylor’s series expansion inz, the nonzero stresses in the surfacez50 then are given by:

H s r

su J 5F H2n

1 J ¹2C H 1

2J ~122n! ]2C

]r 2 G (5.35)

sz5¹2C, t ru5 122n

r

]

]r S 1

r

]C

]r D In ~5.35!, ¹2 is the Laplacian inr andu coordinates. Further becauseC is harmonic inr , u, andz, ¹2C52]2C/]z2, a

ear r

es-

ce- the

s

r is

ded

of ich

d a- p- al ace re ear

stic tion -

result used in~5.35!. Now with a view to representing a constant normal pressure over a sector, we take

C5r 2F ( n50

`

~an cosnu!1â2 ln r cos 2uG (5.36)

The r -dependence in~5.36! realizes asz from ~5.35! which is independent ofr , as desired. By suitably selecting th Fourier coefficientsan in the summation in~5.36!, it would appear to be possible to represent a step pressure on sector with2f̂/2,u,f̂/2. However, when just the terms i the summation in~5.36! are substituted into~5.35!, thesz so produced lacks a contribution froma2 . Needed, therefore, is the â2 term in ~5.36!. With this addition, a complete repre sentation for a constant pressure on the sector results. with this addition, the log singularity terms of~5.34! result ~on transformings r , su , andt ru into their counterparts in rectangular coordinates!.

One consequence of the foregoing development is any pressure distribution which is constant inr and even inu on a sector in Fig. 20 has a log singularity associated wit (f̂5p,2p). Thus, for example, if in~5.33!

sz5H 2p cos pu

f̂ for 2f̂/2<u<f̂/2

0 for other u

(5.37)

there is an associated log singularity. This pressure distr tion has no jumps atu56f̂/2, but still has a jump asr →0 for 2f̂/2,u,f̂/2. A similar development leads to log singularity if the pressure distribution is odd inu on the surface sector, provided there is still a jump in the distrib tion as asr→0 (f̂Þp,2p). On the other hand, if in~5.33!

sz5H 2pr for 2f̂/2<u,f̂/2 0 for other u

(5.38)

there is no log singularity (r 3 is now the factor outside the brackets in~5.36!, and anâ3 term is required instead ofâ2). This pressure distribution does have jumps atu56f̂/2, but none asr→0. Clearly, then, log singularities like that o ~5.34! are associated with jumps inr rather thanu. Away from O, this outcome is consistent with Aksentian@90# and the absence of singularities with pressure jumps in two mensions.

If the sector in Fig. 20 is loaded via uniform shear tra tions rather than pressures, log singularities result along edges of the sector. Forf̂5p/2, these log singularities ar contained in results in Smith and Alavi@168#, Shah and Kobayashi@169#, and Liao and Atluri@170#. For generalf̂, they are identified in Turteltaub and Wheeler@165#. For the component of the shear traction normal to the sector bou ary, the log singularities are as for plane strain~Kolossoff @15#!. For the component parallel, as for antiplane sh ~Ting @104#!. Again, therefore, away fromO a realization of the equivalence in Aksentian@90#.

di- ylord

29A convenient compendium of the 3D field equations of elasticity in all three coo nate systems eventually used in this section is provided in Ch 2, Hughes and Ga @163#.

r ace

d

en i- gle

, .

l in- as ro-

i

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 425

Fig. 22 a! Rigid punch with a wedge-shaped flat base pressed an elastic half-space,b! dual crack problem

r d ary n- thed

u-

o ults. b-

h

ies r a

an nd isfy-

all for 0,r ,` ~see Fig. 22a for coordinates!. The homoge- neous part (u050) of this punch problem has a dual o equivalent crack problem. The latter has a full elastic sp with a crack on thexy plane whenf̂/2,uuu<p, a crack ligament in the same plane whenuuu,f̂/2 ~Fig. 22b!. Under symmetric loadingF, this crack configuration can be treate as a half-space with boundary conditions as in~5.39! with u050.

For either configuration, the potential representation giv in Green and Zerna@167# reduces analysis of the singular ties present to the determination of those attending a sin harmonic functionC, with C50 within the surface sector ]C/]z50 without. This is a relatively simple 3D problem Consequently, and also because of the multiple physica terpretations harmonic functions admit to, this problem h been the subject of quite a number of investigations. In ch nological order, these include: Galin@171#, Rvachev@172#, Noble @173#, Aleksandrov and Babeshko@174#, Bažant @175#, Walden @176#, Morrison and Lewis@177#, Brothers @178#, Keer and Parihar@179#, Ioakimidis @180#, Takakuda @181#, Xu and Kundu@182#, and Glushkov, Glushkova, an Lapina@183#. The means of analysis in these references v from primarily numerical to largely analytical. There is ge erally good agreement as to the stress singularities unear with these means between@173#, @175–181#, and@183#.

The analysis that stands out in its efforts to verify sing larity exponents is Morrison and Lewis@177#. Therein, in addition to comparing with a full set of earlier analyses, tw independent approaches are employed to check res These two agree to typically within 0.1%. Furthermore, su sequent analyses in Brothers@178#, Keer and Parihar@179#, and Takakuda@181# all display excellent agreement wit

nto

Fig. 23 Singularity exponents for varying wedge angles

Another instance of the second class of 3D geomet involves the elastic half-space again, but this time unde flat rigid punch on the surface sector~Fig. 22a!. The punch is frictionless or lubricated, and indents the half-space by amountu0 . Hence in this problem, in general, stresses a displacements are sought throughout the half-space sat ing: the 3D field equations of elasticity, and theboundary conditions, on z50,

uz52u0 for 2f̂/2,u,f̂/2

sz50 for f̂/2,uuu<p (5.39)

tyz5tzx50 for 2p,u<p

t

2 t

e

ric- nts. s a ar

r

t is in

t- rities nch se- in

ries

ther, a

ra a

dary

mi-

426 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

singularity exponents calculated in Morrison and Lew @177# ~to within 1%!. Accordingly these are the results sum marized here.

Singularity exponentsg for varying wedge anglesf̂ are presented in Fig. 23. Thereg is as in

szu z50

5O~r 2g! as r→0 (5.40)

for 2f̂/2,u,f̂/2. For f̂52p there is no singularity (g 50). This is as it should be when the punch indents entire half-space surface. Forf̂5p, g51/2. This, in con- junction with the u-dependence insz on z50, realizes inverse-square-root singularities as the edges of the pu are approached. That is, for example,

szu z50

5OS 1YAr S f̂

2 2u D D as u→ f̂2

2 (5.41)

These inverse-square-root singularities on the edges of punch are also present for all wedge anglesf̂Þ2p. Away from O, they represent a further demonstration of t equivalence with plane strain response of Aksentian@90#.

Of course, the dual crack problem has the same singu ity exponents as shown for the frictionless punch in Fig. In contrast to the 2D situation, however, its antisymme counterpart does not necessarily share the same expon This antisymmetric crack problem can also be treated a half-space problem. Then the boundary conditions are, oz 50,

ux5uy50 for2f̂/2,u,f̂/2

sz50 for 2p,u<p (5.42)

tyz5txz50 for f/2,uuu<p

all for 0,r ,`. Associated singularity exponents are calc lated in Parihar and Keer@184#. These exponents depend o Poisson’s ratio. Forn50, they are the same as for the sym metric crack. FornÞ0, they differ. Exponents forn50.5 differ most and are plotted in Fig. 23 with a dotted curv Exponents forn50.3 are also given in Parihar and Ke @184#.

A further configuration involving the wedge-shape punch takes the punch to adhere to the half-space rather to allow frictionless slip. Then the boundary conditions b come, onz50,

uz52u0 , ux5uy50, for 2f̂/2,u,f̂/2

sz5tyz5tzx50 for f̂/2,uuu<p (5.43)

all for 0,r ,`. This punch problem also has a dual cra problem. The latter is for an interface crack with one ma rial being rigid.

is -

he

nch

the

he

lar- 3.

ric ents. s a n

u- n -

e. r

d than e-

ck te-

The adhering punch problem is less tractable than the f tionless case, and leads to complex singularity expone The real part of such exponents is included in Fig. 23 a broken line. For 0,f̂<45°, these results are from Parih and Keer @185#: for f̂590°, from Brothers@178#. These exponents are forn50.3. Forn50.5, results for the adhering punch are the same as for the frictionless case. For othen, see Parihar and Keer@185#, which also gives the imaginary parts of exponents. While the magnitude of the real par less than the singularity exponent for the frictionless case the range 0,f̂<90° in Fig. 23, the oscillatory nature a tending these complex exponents makes these singula arguably more pathological. Again, at the edges of the pu (u→6f̂/2), plane strain response occurs. That is, inver square-root singularities with oscillatory multipliers as Abramov @186#.

Our final instance of the second class of 3D geomet involves an elastic solid with a 3D reentrant corner. When this corner has faces which are perpendicular to one ano the configuration is tantamount to removing an octant from full elastic space~Fig. 24; sometimes termed the Fiche vertex!. This geometry can be viewed as a wedge with vertex with an angle of 3p/2 and which follows a path which turns abruptly throughp/2 at O. Hence it qualifies for our second class. When the corner is stress free, the boun conditions are:

sx5txy5tzx50 at x50 for y.0 and z.0

sy5tyz5txy50 at y50 for x.0 and z.0 (5.44)

sz5tzx5tyz50 at z50 for x.0 and y.0

where the coordinate system used is as in Fig. 24. The do nant singularities that can be present atO for this stress-free corner are estimated in Abdel-Messieh and Thatcher@187#,

Fig. 24 Three-dimensional reentrant corner

e r r

e r

,

r

c

o

, a r m

nd ight

e

m- ty is,

n

ess -

bles nd e e

ns

ten-

with

in ther

n-

em

en-

i es

lues

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 427

Schmitz, Volk, and Wendland@188#, and Glushkov, Glush- kova, and Lapina@183#. There is good agreement betwe the first and third sources of numerical values of singula exponents. Forn ranging from 0.2 to 0.5, dominant singula ity exponents from both sources exceed 0.58. This is stron than the corresponding 2D reentrant corner~dominant g 50.46). Two-dimensional plane strain and antiplane sh singularities can be expected to be possible away fromO ~from Aksentian@90#!.

5.3 Three-dimensional geometries with crack-surface intersections

As an initial instance of thethird class of3D geometries, we consider an elastic half-space loaded transversely to a c within it, with the crack intersecting the half-space’s fr surface~Fig. 21!. We begin with when the crack front inte sects this surface at right angles (f̃5p/2 in Fig. 21!. Under Mode I loading, symmetry enables attention to be confin just a quarter-space (y.0, z,0 in Fig. 21!. Hence in this problem, in general, stresses and displacements are so throughout the quarter-space satisfying: the 3D field equa- tions of elasticity; stress-free crack conditionson the crack flank,

sy5tyz5txy50 on y50 (5.45)

for x.0 andz,0; symmetry conditionsahead of the crack

v50, tyz5txy50, on y50 (5.46)

for x,0 and z,0; and free-surface conditionson the quarter-space surface,

sz5tzx5tyz50 on z50 (5.47)

for all x and fory.0 ~see Fig. 21 for the rectangular coo dinate system used!. As this particular 3D configuration is going to merit extensive discussion, hereafter we sim term it the 3D crack problem.

Over the years, there have been numerous contribution the literature which address various aspects of the 3D c problem—see Panasyuk, Andrejkiv, and Stadnik@189,190# for reviews which together cite some 500 related referen Focusing on singularity identification at the crack-surface tersection point (O in Fig. 21!, Sih @191# provides a review through the 1970s. In chronological order, contributions this aspect since include: Folias@192#, Kawai, Fujitani, and Kumagai @193#, Benthem @194#, Bažant and Estenssor @195–197#, Sinclair@198#, Benthem@199#, Yamada and Oku- mura @200#, Burton et al @201#, Takakuda@181#, Shaofu, Xing, and Qingzhi@202#, Shivakumar and Raju@203#, Zhu @204#, Barsoum and Chen@205#, Ghahremani@206#, Leung and Su@207,208#, Su and Sun@209#, and Glushkov, Glush- kova, and Lapina@183#. Together these papers are testimo to the challenge of the preceding asymptotic problem. Wh none of these papers solves the 3D crack problem comple analytically and correctly for all values of Poisson’s ratio stress singularity at the intersection point has been cle identified at this time. This identification relies on seve papers, and on both largely analytical treatments and pri

n ity - ger

ear

rack e

-

ed

ught

r-

ply

s to ack

es. in-

on

ny ile tely a rly

al a-

rily numerical ones. We summarize the singularity so fou next, then offer some comments on other analyses that m appear to disagree with it to varying degrees.

For f̃5p/2 in Fig. 21, the crack front coincides with th negativez axis. At the outset, then, we note thatfor z,0 the 3D crack of Fig. 21 belongs to our first class of 3D geo etries. Thus Aksentian@90# applies and the stress singulari for 2D plane strain should result at such locations. That

s5O~r 21/2! as r→0 for z,0 (5.48)

whereins continues as any stress component andr is the radial coordinate in Fig. 21. It follows that any investigatio of the singular response atO should include the singularity in ~5.48! if the crack front is approached away fromO.

Returning attention to the singular response right atO, the investigation that has led to a clear identification of str singularity there is Benthem@194#. In @194#, stresses are as sumed to be separable in spherical polar coordinates~Fig. 21!, with

s5r2g f ~u!g~c! as r→0 (5.49)

Using Boussinesq-Papkovich-Neuber potentials then ena the 3D field equations of elasticity to be complied with a yields trigonometric functions forf (u), associated Legendr functions forg(c). Suitably selecting and combining thes solutions satisfies exactly the symmetry conditions~5.46! and the stress-free crack conditions~5.45!. The only remain- ing conditions, the stress-free surface conditions~5.47!, are then satisfied approximately with sums of series of solutio complying with all other requirements~see@194# for details of the numerical method adopted to this end!. Hence, largely an analytical approach which could be viewed as an ex sion to three dimensions of that in Williams@2# for two di- mensions.

Results recover~5.48! on the crack front away fromO as they should. They also recover the plane strain stresses their inverse-square-root singularity (g51/2) when n50, the one value of Poisson’s ratio for which plane stra stresses satisfy the stress-free surface conditions. For o values ofn,g,1/2 and the singularity is weaker. This ge eral tend ofg51/2 for n50 with g,1/2 for n.0 is con- firmed in a number of investigations subsequent to Benth @194#: Bažant and Estenssoro@195–197#, Benthem@199#, Yamada and Okumura@200#, Burton et al@201#, Takakuda @181#, Shaofu, Xing, and Qingzhi@202#, Shivakumar and Raju @203#, Barsoum and Chen@205#, Ghahremani@206#, and Glushkov et al@183#.

The precise values of the singularity exponentg for n .0 in Benthem@194# are confirmed in Benthem@199# with what is in essence a direct numerical analysis. The indep dent approach in Benthem@199# leads to values that typically

differ by 1/3% and have a maximum difference of 11 2%.

Further confirmation of the precise values ofg for n.0 in Benthem@194# is provided by the analyses in Bazˇant and Estenssoro@195,197#, Takakuda @181#, and Ghahreman @206#. The average difference between numerical valu given in the first three references and corresponding va

rms

e - ont

r

in s hem

pro- un- rtu-

r

428 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Fig. 25 Singularity exponents for varying Poisson’s ratios fo quarter-plane crack in an elastic half-space

r

s a

r

un-

eir

en

lib-

ress the , r me - h st

a-

e

, a

- f

as

ere No for

- om- 3D ed

in ri-

a-

ite ee vi-

ce, the

rac-

terization of the stress and displacement fields there. In te of the spherical polar coordinates of Fig. 21, the first has

sz5K cosc cos~~112n!~p2c!!

r2nAr sinc cos

u

2 as r→0 (5.50)

whereK is a constant~independent ofr, c, andu!, while the second has

s5O~r21/222n!, u5O~r1/222n!, as r→0 (5.51)

where u is any displacement component. Away from th crack-surface intersection point,~5.50! recovers the inverse square-root singularity that must occur at the crack fr ~that is, for c→p, rÞ0). For n50, ~5.50! recovers an inverse-square-root singularity asr→0, the same singula character as in Benthem@194#. For nÞ0, the singularity in ~5.50! is stronger. This is in contrast to the singularity Benthem@194#. Indeed forn.1/4, even the displacement are unbounded. This has led to some discussion: Bent and Koiter@211#, Folias@212,213#.

Unbounded displacements are even less physically ap priate than singular stresses. While singular fields with bounded displacements are possible for the 2D crack, fo nately one can prove that they need never participate~via the completeness argument in Gregory@11#!. Unfortunately, no such proof currently exists for 3D cracks. Therefore, the bounded displacements of Folias@192# cannot be ruled out as possible participants in the 3D crack problem, despite th lack of physical appeal.

There is, though, a valid objection to the singularity giv in Folias @192# in its present form. By virtue ofz50 being free of shear tractions, the third stress equation of equi rium has]sz /]z50 at z50. Equivalently, in terms of the spherical polar coordinates of Fig. 23,

1

r

]sz

]c 50 at c5p/2 (5.52)

From ~5.50!,

]sz

]c 5K

sinpn

r2nAr cos

u

2 at c5p/2 (5.53)

Aside for the casen50, then, equilibrium is not complied with by the explicit singular stress given in Folias@192#. Of course, the method of solution construction adopted in@192# assures satisfaction of the equilibrium equations by the st fields in toto. Thus there must be further contributions to stress field in Folias@192#, not to date explicitly extracted that combine with~5.50! to restore this compliance. In orde to do this, these further contributions must share the sa dependence onr as in~5.50!. Consequently there is the pos sibility they may completely remove singular fields whic behave as in~5.51!. Not to say that this has to happen, ju that it could. As a result, Folias@192# cannot be relied on for singularity identification in the 3D crack problem.

Kawai, Fujitani, and Kumagai@193# attempts to identify local stress singularities for the 3D crack problem. This p per assumes the stresses can be represented as in~5.49! and Benthem @194#. Thereafter it determines forms for th

a

in Benthem@194# is less than 1/20%, while the maximum difference is less than 1/10%. In the fourth reference graphical comparison is made and shows all theg values in Benthem@194# lying on ag-value curve computed in Ghah emani@206#. All told, there is now excellent confirmation o the singularity exponents in Benthem@194#. Accordingly, we present these singularity exponents here in Fig. 25.

While, at this time, there would appear to be no doubt to the existence of a stress singularity of the form of~5.49! with exponents as in Fig. 25, this does not mean that th cannot be other singular fields for the 3D crack problem. completeness argument is advanced in the literature stresses of the form of~5.49!: Absent such, other singulari ties are not precluded. Conversely, absent a companion c pleteness argument, another form of singularity for the crack problem does not invalidate the singularity identifi in Benthem@194#.

There are, in fact, quite a number of other approache the literature aimed at identifying alternative stress singul ties to that of Benthem@194#. We review these efforts in chronological order next.

Folias@192# attempts the ambitious task of finding an an lytical solution for a truly 3D,global, crack configuration. This configuration entails a through crack, in a plate of fin thickness, with crack fronts orthogonal to the stress-f plate faces and under transverse far-field tension. In the cinity of where one of the crack fronts intersects a plate fa the 3D crack problem is contained. The analysis employs symbolic method of Lur’e~Section 3.2,@210#!. Results in- clude an explicit expression for the singular part ofsz at the crack-surface intersection, as well as an asymptotic cha

-

r

e

o

w

t

s

r

o

l

u

D

e se

d- gu- e-

ely ile r of

per n by

es

the

in

or sepa-

- e

ck ar- nk sur- con- de- ce hat or

v- e-

he

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 429

stresses and displacements so that the field equations of ticity are complied with: These forms are similar to those Benthem@194#. Then it proceeds to satisfy the symmet conditions~5.46! and the free-surface conditions~5.47! ex- actly, the stress-free crack conditions~5.45! approximately: This is in contrast to Benthem@194# which satisfies the free surface conditions approximately. Results include one sin larity which is similar to that in Benthem@194# in that it shares diminishing strength with increasingn having started from an inverse-square-root singularity whenn50. Expo- nent values for this singularity, however, do differ apprec bly from those in@194# ~by 40% whenn51/2). Results also include a stress singularity which is stronger than an inve square-root singularity for all values ofn. If a valid result, this last would represent an additional and distinctly differ singularity from that identified in Benthem@194#.

The issue of validity in Kawai et al@193# stems from its use of series of associated Legendre functions to satisfy stress-free crack conditions. In terms of the spherical p coordinates of Fig. 21, the series involve, for example,

Pl 2l12n12~2cosc!5O~~p2c!l22n22! as c→p

(5.54)

for n51,2,..., andl,2.30 Stresses with such terms cann converge to zero on a plane includingc5p. Moreover, such terms lead to singular stress behavior at the crack frontc →p) away from the crack-surface intersection that is kno not to occur. Hence, the stress singularities in Kawai et @193# need to have it established that they are comple free of such terms in order for them to be admissible. To d this would not appear to have been done.

Sinclair@198# attempts to identify local stress singularitie for the 3D crack problem. This paper assumes that stre are separable in cylindrical polar coordinates. It satisfies field equations of elasticity exactly with forms comprised elementary functions. It also satisfies the stress-free c and symmetry conditions,~5.45! and ~5.46!, exactly. How- ever, it only attempts to satisfy the free-surface conditio ~5.47! for sz with a term which is itself asymptotically zer as the crack tip on the surface is approached~the actual re- sidual beingO(r 3/2) asr→0 thereon!. Results for the domi- nant singularity have

ŝ5O~z2/Ar ! as r→0 (5.55)

whereinŝ is any stress component other thantuz and t rz , these last being nonsingular~see Fig. 21 for the cylindrica polar coordinates used!. Away from the crack-surface inter section point, the appropriate inverse-square-root singula is recovered at the crack front. Forn50, the known inverse- square-root singularity is not recovered in the surface az 50. The general trend of a weakening of singular respo as the free surface is approached is reflected in~5.55!, but now by a reducing singularity coefficient rather than a red ing singularity exponent.

elas- in ry

gu-

ia-

se-

nt

the lar

ot

( n al

ely ate

s ses the of ack

ns

- rity

t nse

c-

The absence of the plane strain singularity forn50 does not necessarily invalidate a singularity identified for the 3 crack. While completeness~Gregory @11#! and uniqueness ~Knowles and Pucik@12#! mean that any loading of the 3D crack which is independent ofz must produce these plan strain fields whenn50, this does not have to be the ca when loading is not independent ofz.

The real objection to the simple analysis in Sinclair@198# lies in its satisfaction of the free-surface conditions. Boun ary conditions are known to effect the nature of stress sin larities considerably in elasticity. Hence, satisfying the fre surface conditions in only an asymptotic sense is quite lik to change the nature of any singularity found. Thus, wh results in Sinclair@198# may indicate a possible trend fo singularities in the 3D crack problem, they fall far short actually identifying a possible stress singularity.

Shivakumar and Raju@203# attempts to identify two local stress singularities for the 3D crack problem. In this pa the singular stresses are assumed to admit representatio

s5F~u,z!r 21/21G~c,u!r2g as r→0 or r→0 (5.56)

whereF and G are continuous functions.31 Analysis is via finite elements with fitting used to estimate singulariti present. This fitting is undertaken for each term in~5.56! separately. Away from the crack-surface, results recover inverse-square-root singularity that must occur. Forn50 they recover the known plane strain singularity~automati- cally, by the superposition employed!. For n→0, they indi- cate that the functionF is zero or negligibly small. Forn .0, they also estimate singularity exponents which are good agreement with Benthem@194# ~average difference 2/3%, maximum difference3/2%).

Zhu @204# attempts to identify local stress singularities f the 3D crack problem. This paper assumes stresses are rable in cylindrical polar coordinates. Analysis is via a com bination of two solutions. The first is for a crack in plan strain. The second removes surface tractions atz50 from the first. It is derived from a single harmonic potential~after Section 5.7, Green and Zerna@167#!. The 3D aspects of the analysis are limited to the planez50. Within this plane, all field equations and boundary conditions in the 3D cra problem are satisfied. Unfortunately, in meeting the she free conditions on the crack flanks, continuity of crack fla displacements is required in the second solution. When face tractions are applied to a cracked half-space, such tinuity is generally not the case. Hence the analysis is monstrably incomplete. Away from the crack-surfa intersection point, the inverse-square-root singularity t must occur is recovered automatically by construction. F n50, the known plane strain singularity is likewise reco ered. Forn→0, results show a persistence of the invers square-root singularity in the free surface atz50. The coef- ficient of the singularity is reduced from that away from t

31Observe, therefore, that the second term in~5.56! is not the same as in Benthem @194#. In @194#, G(c,u) contains terms which areO((p2c)21/2) asc→p.

30See, eg, Ch 8, Abramowitz and Stegun@139#, for the asymptotic behavior given in ~5.54!.

l o

a

o n

a

t

l

o

r a

,

t

na- t

of ce rity e or the

the em d

21

s. e- the

ve nce in- -

is m. ob-

a r to

an tron-

d

ic se

by

es, ns.

430 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

surface by a factor of (12n22n2). This factor is 1 when n50, appropriately, 5/8 whenn51/4, and 0 whenn51/2.

While all of analysis in Zhu@204# is correct, the 3D so- lution found is really only valid in the surfacez50. If in- stead it held for allz in the half-space, it would be possib to simply take the stresses it produced on the surface subregion within the half-space as prescribing tractio thereon and so pose a problem for which the fields in Z @204# are applicable. Absent a solution for other thanz50, however, there is no guarantee that the fields in Zhu@204# ever participate in an actual 3D crack problem. They cou though, in which case they would represent an additional complementary singularity to that found in Benthem@194#.

Leung and Su@207# attempts to identify local stress sin gularities for the 3D crack problem. This paper superimpo the singular crack-tip stresses in plane strain with stres that are assumed to be separable in spherical polar co nates. The latter have to have an inverse-square-root si larity to effect the removal of the stresses from the former the surface atz50. They are analyzed with finite element By construction, appropriate singular behavior results aw from the crack-surface intersection point and forn50. For n→0, a drop in the coefficient of the inverse-square-ro singularity is indicated at the free surface.

Leung and Su@208# attempts the same identification a @207#, but primarily by analytical means rather than nume cal. After superposing the plane strain fields, the appro for the residual problem follows that in Zhu@204# and uses a single harmonic potential. In fact, the approach in Leung a Su @208# could be interpreted as an attempt to extend results of Zhu@204# for the surface into the interior. In mak ing this attempt, however, the approach follows that Kawai et al @193# rather than that in Benthem@194#. As a result, in its present form it suffers from the same lack convergence and from the introduction of inadmissible s gular stresses on the crack front. Consequently, the resu Leung and Su@208# cannot be accepted at this time.

Su and Sun@209# attempts to identify local stress singu larities in a global configuration entailing a through crack a plate~the same geometry as in Folias@192#!. This paper employs an interesting decomposition of the fields involv into a plane stress state, a shear stress state, and a Papk Fadle state. Each of these states is assumed to be separa cylindrical polar coordinates. Series of solutions are e ployed. The analysis is analytical with the minor excepti of the routine numerical determination of the eigenvalu used in the Papkovich-Fadle expansion. Results recove appropriate singular behavior away from the crack-surf intersection point and forn50. Forn.0, the dominant sin- gular character identified is the same as in~5.55!. As noted in ~5.55! et seq,tuz andt rz are nonsingular. Hence they are n explicitly given in Su and Sun@209#. They are needed though, to ensure satisfaction of the equilibrium equations the singular stress components. Unfortunately, when th shear stress components are derived from the displacem given in Appendix I of Su and Sun@209#, it transpires that they are not zero on the plate faces. This is in violation of free-surface conditions. If this shortcoming in the promisi

e f a ns hu

ld, nd

- ses ses rdi- gu-

on s. ay

ot

s ri- ch

nd he - in

of in- ts in

- in

ed vich-

ble in m- on es the ce

ot

by ese ents

he ng

approach in Su and Sun@209# were to be rectified, it would appear that this paper would lead to the first essentially a lytical solution for a singularity in the 3D crack problem. A this time, however, it cannot be accepted as such.

In sum for the 3D crack problem, the current state research findings is as follows. Away from the crack-surfa intersection point, an inverse-square-root stress singula occurs, and only it occurs. ‘‘Away’’ includes arbitrarily clos to the point, but not at it. Consequently, all singularities f the 3D crack problem may be viewed as characterizing participation of this inverse-square-root singularity as free surface is approached. Viewed in this light, Benth @194# provides the only truly confirmed singularity identifie to date, with

s5OS r1/22g

Ar sinc D as r→0 (5.57)

In ~5.57!, the spherical polar coordinates are as in Fig. and the singularity exponentg as in Fig. 25. Forn50, g 51/2. Then the crack-tip singularity for plane strain applie For nÞ0, g,1/2. Hence the participation of the invers square-root singularity goes to zero in the free surface for singularity identified in Benthem@194#. There may be other singularities for the 3D crack, some of which may not ha this participation go to zero in the free surface. The existe of these alternative singularities would not invalidate the s gularity in Benthem@194#. As of now, any such other singu larities have yet to be properly identified.

A further instance of the third class of 3D geometries the antisymmetric counterpart of the 3D crack proble Herein the formulation is the same as for the 3D crack pr lem except that the symmetry conditions~5.46! are ex- changed for antisymmetry conditions:

u5w50, sy50, on y50 (5.58)

for x,0 andz,0, whereu andw are displacements in thex andz directions, respectively.

Benthem@199# analyzes the antisymmetric 3D crack via finite difference approach. Singularity exponents appea have converged to within about 2% in Benthem@199#. Two branches of singularity exponents for varyingn are identified by this means. One branch is for a stronger singularity th the symmetric case, one weaker. The exponents for the s ger singularity are confirmed in Bazˇant and Estenssoro @196,197# ~typically to within 2%!. They are also confirmed in Ghahremi @206#. Exponents for both the stronger an weaker singularities from Benthem@199# are included in Fig. 25 for varying Poisson’s ratios.

Alternative singularities may exist for the antisymmetr 3D crack problem. Again these would not invalidate tho identified in Benthem@199#. An indication of a possibility in this regard is given in Meda et al@214# which uses the very limited approach of Sinclair@198# to arrive at singular char- acter as in~5.55!. The same singular character is obtained different means in Appendix II, Su and Sun@209#, but insuf- ficient details are furnished therein to enable checking.

As two last instances of the third class of 3D geometri we consider two further crack-intersection configuratio

on ern of

lin- s re- at the

ety ns

las-

the less ing ing ent lar t ith

e rge

i

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 431

Fig. 26 Singularity exponents for varying angles of intersect for a symmetrically-loaded crack in an elastic half-space

e

t i

-

a i

e

- for one

to

ua- a-

ilib- and in

qui-

rst

d. t

ess by

wo a

ck ( -

s

ur-

ion th for

s- ng er the ten re

Some remarks on the effects of other field equations singularities are given in Part I, Section 2.1. These conc the possible removal of stress singularities by relaxing any the three linearizations of classical elasticity. These three earizations are: the small stress assumption that stresse main below yield levels, the small strain assumption th strains are linearly related to displacement gradients, and small deflection assumption that the loads act in their entir on the undeformed state. Relaxing the first two linearizatio entails switching to the field equations ofelastoplasticityand large strain elasticity, respectively. A relaxation of the third can be made by applying the field equations of classical e ticity, together with loading conditions,on deformed states instead of undeformed. For each of these modifications to field equations, analysis is nonlinear and consequently tractable than that for classical elasticity. The general find of such analysis is that relaxing any of the three lineariz assumptions of classical elasticity does lead to a differ singular character. Further, typically the resulting singu character is less nonphysical~for example, the replacemen of oscillatory stress singularities for the interface crack w a nonoscillatory singularity!. Occasionally a singularity is even removed~for example, the singularity at an adhesiv butt joint present in classical elasticity is absent from a la strain treatment!.32 However, the great majority of singulari ties in classical elasticity persist, albeit in altered forms, any of these modifications to the field equations. Thus n of these modifications is really successful when it comes removing stress singularities from classical elasticity.

Here we consider some other changes to the field eq tions of classical elasticity. We distinguish these modific tions as follows: changes in the stress equations of equ rium, changes in the stress-displacement relations, changes to both. We consider each type of modification turn next.

As a first simple change to the stress equations of e librium, we consider the effects of introducingbody-force fields, heretofore taken as null. Then, for example, the fi equilibrium equation for in-plane loading in~2.2! becomes

]s r

]r 1

1

r

]t ru

]u 1

s r2su

r 1Fr50 (5.59)

where Fr is the radial component of the body-force fiel What is apparent from~5.59! is that for body forces to effec stress singularities which behave asO(r 2g) as r→0, they themselves have to behave asO(r 2g21). Such body forces would not seem likely to be needed in practice. Hence str singularities in elasticity can be expected to be unaffected the presence of body-force fields.33

There is one possible exception to the foregoing in t dimensions. This is the line-load body force. For such body-force field,

Fr5FA /r , Fu50 (5.60)

on

not

face

The first further crack-intersection configuration entails cra fronts which are not orthogonal to the free surfacef̃ Þp/2 in Fig. 21!. Motivated by a search for an energy r lease rateGI satisfying 0,GI,` in the free surface, Bazˇant and Estenssoro@195–197# seek an anglef̃ such thatg 51/2. This leads tof̃.p/2 for the symmetric case. Value of such f̃ are given in Bazˇant and Estenssoro@197# for n ranging from 0.0 to 0.4. These values are confirmed in B ton et al@201# and Takakuda@181# ~typically agreement is to within 1%!. The antisymmetric case is also treated in Bazˇant and Estenssoro@197#. This leads tof̃,p/2. Singularity ex- ponents other thang51/2 are tabulated in Takakuda@181# for 0<n<1/2 and otherf̃ under symmetric loading. Forn 50.0, 0.4, these results are illustrated in Fig. 26.

The second and final, further, crack-surface intersec configuration is as for the 3D crack problem but now w two materials comprising the half-space. Singularities this 3D interface crack are identified in Bazˇant and Estens soro @197#, Barsoum and Chen@205#, and Ghahremani and Shih @215#.

5.4 Other field equations

While it falls outside the stress singularities in classical el ticity reviewed so far, it is nonetheless appropriate in clos this review to offer a few comments on the effects of oth field equations on singularities. It is appropriate because singularities attending other field equations are quite of directly related to those in classical elasticity. The intent h is to indicate this sort of connection, rather than extensiv explore it. Accordingly, references cited here are by way example, rather than anything approaching a comprehen listing.

ely of sive

32As explained in Part I, Section 2.1, the introduction of perfect plasticity does really qualify as a modification that removes a singularity. 33Plate bending singularities are similarly unaffected by the presence of plate loading: see Sections 4.3 and 4.4.

r

c .

d

-

o i t

d -

m

o

t

m

h e

e

ns, re

tic-

ast

er- ded ed. ion

n-

dy- by

ro- ibil- ce ails a-

on- ature Sec-

ies

in

nt ch ions c-

e

ts ular ity. stic . If stic

ngth on in ce

in- ace

hen

432 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

whereFA is a force per unit area. Then directly from~5.59! and the second of~2.2!,

s r5su52FA ln~r /r c!, t ru50 (5.61)

wherein r c is some characterizing radius introduced to e sure dimensional consistency. The field in~5.61! also satis- fies the stress equations of compatibility: It therefore co plies with all the field equations of elasticity. Analogously three dimensions, a point-load body force leads to a st singularity which behaves asO(r21) as r→0. Aside from like instances, however, stress singularities in elasticity be expected to be the same with body forces as without

As a second change to the stress equations of equilibr we consider the introduction of inertial terms. Then the equations becomeequations of motion. This change can be viewed as mathematically equivalent to introducing a bo force field. For example, in~5.59!, set

Fr52rm

]2ur

]t2 (5.62)

whereinrm is the mass density andt is time. Then the equa tion of motion in the radial direction is recovered. Similar the equation of motion in the angular direction can be rec ered. Accordingly we can expect singularities attend equations of motion to be similar to those for elastosta with body forces.

As an initial instance of dynamic response, we consi the case ofvibrationsof elastic media. If the vibratory mo tion has frequencyv, then one may take

ur5ûr~r ,u!sinvt (5.63)

Assuming the same vibratory dependence for the other placement component and the stresses enables sinvt and the time dependence to be factored out of the equations of tion. Then we have exactly the same equations as for introduction of body forces. Consequently no changes stress singularities from those in classical elasticity are to expected when vibrations are introduced. That this is s demonstrated for antiplane shear in Sagochi@216#. It is fur- ther demonstrated for out-of-plane bending within four order theory in Leissa, McGee, and Huang@121#, and within sixth-order theory in Huang, McGee, and Leissa@134#.34

For the more general case ofelastodynamicresponse wherein motion is transitory rather than vibratory, the sa sort of correspondence should occur whenever the stress displacement fields are separable in their spatial and tem ral dependences. This separable nature need not be wit spect to a stationary coordinate system for a correspond to hold. A demonstration is given in Achenbach and Bazˇant @218# for propagating cracks. For both antiplane shear a plane strain, the inverse-square-root stress singularity of e tostatics is recovered. Now, though, theu-dependence is only the same as the elastostatic case in the limit as the spe crack propagation goes to zero.

s at- . See,

n-

m- in ess

an

ium se

y-

ly v-

ng ics

er

dis-

o- the to be is

h-

e and po- re-

nce

nd las-

d of

Turning to changes to the stress-displacement relatio we first consider the effects of introducing temperatu fields. The resulting field equations ofthermoelasticitycan be couched so that they differ from those of classical elas ity only in the stress-displacement relations~see, eg, Section 1.3, Nowacki@219#!. The same field equations can be rec using the Duhamel-Neumann analogy~Section 1.9,@219#!. Then they reveal that the singularities in stationary th moelasticity are the same as in classical elasticity provi two possible additional sources of singularity are admitt The first additional source is the action of a normal tract of magnitudecTT, wherecT is a material constant~propor- tional to the material’s linear coefficient of thermal expa sion and its bulk modulus!, and T is the temperature field present. The second additional source is an effective bo force field. In two dimensions, the latter can be expressed

Fr52cT

]T

]r , Fu52

cT

r

]T

]u (5.64)

whereFu is the u-component of the body force. It follows that, in two dimensions, the additional singularities so p duced come from constant normal tractions and the poss ity of a line-source temperature field. The first can produ logarithmic singularities as in Section 2.4. The second ent T of O(ln r) asr→0, hence a line-load body force and log rithmic singular stresses as in~5.61!. In three dimensions analogous results hold. For an elastic half-space with a c stant temperature on a surface rectangle and zero temper elsewhere on the surface, stresses can be obtained from tion 2.3, @219#. These stresses have logarithmic singularit at the corners of the rectangle~cf, @164#!. At a point source in three dimensions, the temperature and stresses areO(r21) as r→0 ~see, eg, Section 2.12,@219#!. In addition, all the other singularities in classical elasticity can be present thermoelasticity.35

A different type of modification to the stress-displaceme relations results from varying elastic moduli. Three su variations are entertained to a limited degree here: variat with time, variations with position, and variations with dire tion.

Elastic moduli can vary with time so as to reflect th physical phenomena of creep and relaxation. When suchvis- coelastic variations are consistent with the constrain needed for the correspondence principle to hold, the sing character in classical elasticity carries over to viscoelastic If the singular stresses in elasticity are independent of ela moduli, identical singular stresses occur in viscoelasticity the singular stresses in elasticity are dependent on ela moduli, singular stresses have the same singularity stre or exponent in viscoelasticity. Now, though, the participati of different parts of the singularity coefficient can vary different ways with time. A statement of the corresponden principle and a demonstration of its implications for the s gular stresses attending a point load normal to a half-sp

ence

35The situation is more complex than this limited discussion would indicate w multiple materials are present in thermoelasticity. Then the added discontinuitie tending jumps in thermal conductivities can have associated stress singularities for example, Yan and Ting@220# and Yang and Munz@221#.

34While the inclusion of vibratory response leaves singularities unaltered, the pres of singularities can alter vibratory response: see Leissa@217#.

r l

s c

h f

a

n r

l

s

e

c i

hat lar

ne the n of - erg in- for

are

ne- n-

int rom this

ted is-

ur en-

ary ies es

ag

ces- , for

no pite

d on of an

e no

f a ons,

ite e real

on- ess- of

me I,

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 433

are provided in Lee@222#. Further demonstrations of the co respondence between elasticity and viscoelasticity singu ties are given in Williams@223# for a crack.

When elastic moduli vary with position as piecewise co stants, Sections 3, 4.2, and 4.3 summarize the numerou lated studies in the literature for suchinhomogeneous elasti media. The general finding of these studies is that the int duction of additional discontinuities attending abru changes in elastic moduli increases both the occurrence the strength of stress singularities. This does not have to so, though. Occasionally the singularity associated wit discontinuity already present in a configuration can be of by the singularity associated with an added discontinuity elastic moduli~see, eg, the butt joint in Section 3.3!.

When elastic moduli vary with position other than piecewise constants, there are relatively few studies in literature. However, for the simple case of antiplane shea is straightforward to bound the effects of a radial depende of the shear modulus. Taking the shear modulus to vary

m5m0~r /r c! « (5.65)

for u«u!1, provides two extremes. For«.0, m→0 as r →0: For «,0, m→` asr→0. Then, following the analyti- cal path laid out in Section 4.1, leads to

g512 p

f 2

«

2 , g512

p

2f 2

«

2 (5.66)

as «→0, for the dominant singularity exponents for no mixed, mixed problems, respectively. Hence the singula exponent is reduced when the modulus softens to zero« .0), and it is increased when the modulus stiffens to infin («,0). This type of response is consistent with findings general in this review, namely that increasing stiffness ty cally increases singular character. It is, though, for an treme variation in moduli. And it is not that dramatic give this extreme variation. This suggests that the dependenc stress singularities on more realistic radial variations of e tic moduli may be slight if any.

To investigate this suggestion further, we take

m5m01m1

r

r c (5.67)

This choice requiresuz be taken as a series of separab functions with increasing powers ofr in implementing the analytical approach in Section 4.1 instead of just a sin separable term. Even so, the same results for exponent obtained as in Section 4.1~viz, as in~5.66! with «50). That is, the linear radial dependence of the shear modulus ~5.67! leaves singularity exponents unchanged from those a constant shear modulus. By a like means, the same r can be expected to hold for elastic plates in extension.

The third and final variation in elastic moduli that w consider is to allow them to change with direction. Hence admit anisotropiceffects. Ting@224#, Chapter 9, furnishes a clear exposition of singularity identification in anisotrop elasticity, together with an extensive set of related referen The general finding is that the additional discontinuit which can attend anisotropy can have associated stress

- ari-

n- re-

ro- pt and be a

set in

s the r, it nce as

- ity (

ity in pi- ex- n e of as-

le

gle are

in for sult

e we

ic es.

es sin-

gularities which increase the singular character over t found in classical elasticity. Occasionally additional singu stresses can offset those in isotropic elasticity~cf, a bimate- rial versus a single material!. An example of such offsetting is given in Ting@225#.

As our last modification to the field equations, and o which effects both the stress equations of equilibrium and stress-displacement relations, we consider the introductio couple stresses. Field equations for a linearized theory in cluding couple stresses may be found in Muki and Sternb @226#. The general finding for such a theory is that the s gularity strength remains the same as in classical theory corresponding stresses, but dependence onu is modified. Demonstrations of this persistence of singular stresses given in Muki and Sternberg@226# for the half-space under a discontinuous shear traction, normal and tangential li loads, and a flat, lubricated, strip punch. A further demo stration is given in Sternberg and Muki@227# for a crack in plane strain.

6 CONCLUDING REMARKS

In classical elasticity, stress singularities occur under po loads, lineloads, and so on. They can also occur away f any such concentrated loading. It is the occurrence of latter type of singularity that is reviewed here.

When stress singularities occur away from concentra loading, they do so in concert with a discontinuity: no d continuity, no singularity. Hence we term themdiscontinuity singularities. The discontinuities for such singularities occ on boundaries. In classical elasticity, these discontinuities tail abrupt changes in boundary directions/bound conditions/elastic moduli. In general, such discontinuit flag the possibility of singularities. In particular, step chang in uniform tractions or first derivatives of displacements fl the possibility of logarithmic singularities.

The presence of a discontinuity, however, does not ne sarily mean that there is a stress singularity. For example the in-plane loading of an angular elastic plate, there are singularities when the vertex angle is less than 180°, des the presence of a sharp corner~see Sections 2.1 and 2.2!. For the same plate as a half-plane with a step pressure applie its edge, there are no singularities, despite the presence abrupt change in boundary conditions~Section 2.4!. For the same plate with one face clamped the other free, there ar singularities when the vertex angle is less than 45°~Sections 2.1 and 2.2!. This last example is despite the presence o sharp corner, and an abrupt change in boundary conditi and an abrupt change in elastic moduli~clamped conditions being attributable to attachment to a material with an infin Young’s modulus!. Thus, while discontinuities flag possibl stress singularities, they are not in themselves the sources.

The real sources of discontinuity singularities are disc tinuities in the stiffnesses in the cohesive or adhesive str separation laws which underlie the constitutive relations elasticity. This may not be immediately apparent for so singularities. Some further explanation is given in Part Sections 2 and 5.

434 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

Table 16. State of the art of stress singularity identification in classical elasticity

Configuration

Single material Bimaterial

Power Log Power Log

In-plane loading of a plate C C C o Antiplane shear of a wedge C C C C Plate bending, 4th order theory C C c o Plate bending, 6th order theory c o o O Axisymmetric torsion of a cylinder

C C C C

Axisymmetric axial loading at vertex

C c c O

Axisymmetric axial loading at a cylindrical boundary

C C C o

Three-dimensional away from 3D vertex

C C C o

e

a

a

a

u t

t

h

w

ent

in

s m-

Fig. 27 Classes of configurations that are effectively equival with respect to singularity identification:a! configurations equiva- lent to plates in extension,b! configurations equivalent to wedges antiplane shear

d out nce

is ons

ct

w ch ac- uite er, lar pen

re

- als rity

In two dimensions, the various discontinuity singulariti actually identified to date in classical elasticity may be su marized as follows. For any stress components, as the sin- gular point is approached, elasticity can have:

s5O~r 2g cos~h ln r !!1O~r 2g sin~h ln r !!

s5O~r 2g ln r !1O~r 2g!

s5O~r 2g!

s5ord~ ln2 r !1ord~ ln r ! (6.1)

s5ord~ ln r !

s5O~ ln r !

s5O~cos~h ln r !!1O~sin~h ln r !!

as r→0, whereing is the singularity exponent (0,g,1), and h is the imaginary part of the eigenvalue involved. ~6.1!, O is associated with locally homogeneous bound conditions, ord with locally inhomogeneous~ord is defined in Part I, Section 1.2!. For the former, the singularity may o may not participate depending on other far-field bound conditions: hence theO notation. Typically, though, once such a singularity is identified as possible, it does particip For the latter, the singularity’s participation is guaranteed the inhomogeneous part of the local boundary conditio hence the ord notation.36

Numerous such singularities are identified in the literat for classical elasticity. Table 16 summarizes the state-of- art of these identifications for the various, essentially 2 configurations that are reviewed here and involve one or materials.

In Table 16,power singularitiesinclude the first three of ~6.1!. For the most part, the singularities involved are as the third of~6.1!. There are, though, quite a few instances singularities as in the first of~6.1!. There are relatively few as in the second of~6.1!. In Table 16, then,logarithmic sin- gularities include the fourth through sixth of~6.1!. These are all weaker than power singularities. Accordingly they can harder to detect absent an asymptotic appreciation of t

are

In ry

r ry

te. by ns:

re he- D, wo

in of

be eir

possible presence. This is the reason they are separate from power singularities in Table 16. Of course, the prese of either type of singularity needs to be recognized if one to avoid the futile exercise of stress-strength comparis once either occurs.37

In Table 16, the following notation is adopted with respe to the state-of-the-art of identifications:

C5 largely closed,c5partly closed (6.2)

O5 largely open, o5partly open

In ~6.2!, largely closed means there are few, if any, ne singular configurations to be identified. Moreover, any su new configurations are not expected to occur often in pr tice. In contrast, partly closed means there are, in fact, q a few more singular configurations to be identified. Furth in ~6.2!, largely open means the great majority of singu configurations have yet to be identified, whereas partly o means just a majority.

The bulk of singularity identifications in the literature a for in-plane loading of an elastic plate~Sections 2 and 3!. As a result, identification for this configuration is largely com plete. The partly open area of log singularities in bimateri is explained in Section 3.3. While there are fewer singula identifications in the literature for antiplane shear, they

n

36The last stress of~6.1! is not strictly singular, being bounded asr→0. However, it is undefined asr→0, and consequently shares some of the difficulties associated stress singularities.

ith37The few known instances of the last of~6.1! occurring are given at the end of Sectio 2.2.

3

h t i

l

a

e

e

tle

a

ic

e- ,

-

-

i- ral

of

and

ic ity

s

r- un-

ity d,

e

ion

s r

g ex-

in

i-

ion,

ic e-

RI. - nd-

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 435

nonetheless largely complete~Sections 4.1 and 4.2!. This is a consequence of analysis for this configuration being re tively simple.

These two configurations lead to singularity identific tions for a number of other configurations. The various wa they do this are illustrated in Fig. 27. Therein the followin notation is used:

AE5asymptotically equivalent configuration

EM5eigenvalue equations match (6.

MA 5mathematically analogous configuration

For further explanation, see Sections 4.3, 4.5, 5.1, and Also in Fig. 27, an arrow with a solid line denotes that t correspondence holds for both single materials and bima als, whereas one with a broken line just for single mater ~to date, anyway!.

There are a few singularity identifications for trimateria These are mentioned in Sections 3.2 and 4.2.

There are 3D configurations other than those of Table and Fig. 27 for which singularities are identified. An indic tion of the state-of-the-art with respect to singularity iden fication for these configurations is given in Sections 5.2 a 5.3.

A discussion of stress singularities for field equatio other than those of classical elasticity may be found in S tion 5.4. Typically, if a stress singularity occurs in classic elasticity, singular stresses persist with other field equatio Sometimes singularities persist with modified strengt sometimes with the same. Examples of the former inclu elastoplasticity and large strain~nonlinear! elasticity. Ex- amples of the latter include elastodynamics, viscoelastic thermoelasticity, and couple stress theory.

ACKNOWLEDGMENTS

I am grateful for helpful input received from colleagues du ing the course of preparing this review: DB Bogy~UC Ber- keley!, JP Dempsey~Clarkson University!, NW Klingbeil ~Wright State University!, BS Smallwood~Chrysler Corpo- ration!, and TCT Ting~University of Illinois, Chicago!. I am also grateful for the careful typing of the manuscript by Gibb and R Kostyak, and the painstaking preparation of drawings by DC Little~all of Carnegie Mellon University!. In addition, I am grateful to R.C. Benson~Penn State! for overseeing the extensive review process, as well as for thoughtful input of reviewers, and the incorporation of r sulting revisions by my wife, Della.

REFERENCES

@1# Sinclair GB ~2004!, Stress singularities in classical elasticity—I: R moval, interpretation, and analysis,Appl. Mech. Rev.57, 251-298.

@2# Williams ML ~1952!, Stress singularities resulting from variou boundary conditions in angular corners of plates in extension,ASME J. Appl. Mech.19, 526–528.

@3# Dempsey JP, and Sinclair GB~1979!, On the stress singularities in th plane elasticity of the composite wedge,J. Elast.9, 373–391.

@4# Rao AK ~1958!, A study into the problem of diffusion of loads, Doc toral Dissertation, Univ of London, London, UK.

la-

a- ys g

)

5.2. e eri- als

s.

16 -

ti- nd

ns ec- al ns.

hs, de

ity,

r-

M the

the e-

-

s

-

@5# Johnson KL ~1985!, Contact Mechanics, Cambridge Univ Press, Cambridge, UK.

@6# Winkler E ~1867!, Theory of Elasticity and Strength, ~German!, H Dominicus Press, Prague, Austria.

@7# Oravas GA~1964!, Review Number 4988,Appl. Mech. Rev.17, 684. @8# Euler L ~1774!, The pressure in planes from weights that set

~Latin!, Novi Commun Acad Petropolit18, 283–329. @9# Nuller BM ~1974!, Deformation of an elastic wedge reinforced by

beam,J. Appl. Math. Mech.38, 823–829. @10# Ting TCT ~1985!, Asymptotic solution near the apex of an elast

wedge with curved boundaries,Q. Appl. Math.42, 467–476. @11# Gregory RD~1979!, Green’s functions, bi-linear forms, and complet

ness of the eigenfunctions for the elastostatic strip and wedgeJ. Elast.9, 283–309.

@12# Knowles JK, and Pucik TA~1973!, Uniqueness for plane crack prob lems in linear elastostatics,J. Elast.3, 155–160.

@13# Lighthill MJ ~1975!, Introduction to Fourier Analysis and General ized Functions, Cambridge Univ Press, Cambridge, UK.

@14# Kolossoff G~1910!, On an application of the theory of complex var ables to the two-dimensional problem of elasticity theory, Docto Dissertation, St Petersburg, Russia.

@15# Kolossoff G ~1914!, On some properties of the plane problem elasticity theory~German!, Z. Math. Phys.62, 384–409.

@16# Inglis CE ~1913!, Stresses in a plate due to the presence of cracks sharp corners,Soc. Nav. Archit. Mar. Eng., Trans.55, 219–241.

@17# Neuber H 1946,Theory of Notch Stresses, JW Edwards Co, Ann Arbor MI.

@18# Kitover KA ~1952!, On the use of special systems of biharmon functions for the solution of some problems in the theory of elastic ~Russian!, Prikl. Mat. Mekh.16, 739–748.

@19# Kalandiia AI ~1969!, Remarks on the singularity of elastic solution near corners,J. Appl. Math. Mech.33, 127–131.

@20# Seweryn A, and Molski K~1996!, Elastic stress singularities and co responding generalized stress intensity factors for angular corners der various boundary conditions,Eng. Fract. Mech.55, 529–556.

@21# Dempsey JP~1978!, On the stress singularities in the plane elastic of the composite wedge, Doctoral Dissertation, Univ of Aucklan Auckland, New Zealand.

@22# Gdoutos EE, and Theocaris PS~1975!, Stress concentrations at th apex of a plane indentor acting on an elastic half plane,ASME J. Appl. Mech.42, 688–692.

@23# Sinclair GB ~1996!, On the influence of cohesive stress-separat laws on elastic stress singularities,J. Elast.44, 203–221.

@24# Rössle A~2000!, Corner singularities and regularity of weak solution for the two-dimensional Lame´ equations on domains with angula corners,J. Elast.60, 57–75.

@25# Rösel R ~1987!, On the wedge/notch eigenvalues,Int. J. Fract. 33, 61–71.

@26# Karp SN, and Karal FC~1962!, The elastic-field behavior in the neighborhood of a crack of arbitrary angle,Commun. Pure Appl. Math. 15, 413–421.

@27# Sinclair GB, Khatod HN, and Rummel CW~2003!, On the removal of stress singularities from sharp corners,Proc of the 19 Canadian Con- gress of Applied Mechanics, Vol. 1, Calgary, Alberta, 1, 94–95.

@28# Williams ML ~1953!, Discussion of ‘Stress singularities resultin from various boundary conditions in angular corners of plates in tension,’ASME J. Appl. Mech.20, 590.

@29# Ricci L ~1951!, Table of Minimum Roots of an Equation of Interest Solid Mechanics~Italian!, Pubblicazioni No 296, Istituto per le Ap- plicazioni del Calcolo, Rome, Italy.

@30# Klingbeil NW ~1996!, private communication. @31# England AH~1971!, On stress singularities in linear elasticity,Int. J.

Eng. Sci.9, 571–585. @32# Smallwood BS~1996!, private communication. @33# Dempsey JP~1995!, Power-logarithmic stress singularities at b

material corners and interface cracks,J. Adhes. Sci. Technol.9, 253– 265.

@34# Sinclair GB ~1999!, Logarithmic stress singularities resulting from various boundary conditions in angular corners of plates in extens ASME J. Appl. Mech.66, 556–560.

@35# Browning RV, and Ju FD~1977!, Stress singularities at an elast crack with arbitrary surface loads, including possible fracture m chanics applications,J. Math. Anal. Appl.58, 482–496.

@36# ANSYS personnel~1995!, ANSYS User’s Manual, Revision 5.2, Vol I, ANSYS Inc, Canonsburg, PA.

@37# ABAQUS personnel~1997!, ABAQUS Standard User’s Manual, Re- vision 5.7, Vol I, Hibbitt, Karlsson and Sorensen, Inc, Pawtucket,

@38# Sinclair GB, and Epps BE~2002!, On the logarithmic stress singu larities induced by the use of displacement shape functions in bou

o

o

g

r

r

n

-

a

-

l

t

d

,

e

g

e

a

r

r

es-

- tion,

ck

t

two

an

h

h

c

es

l

- ear- 61st of

l ugh

s

te

n tic ory

s

l ,

r-

e trac-

e be

em

436 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

ary conditions in submodelling,Commun. Numer. Methods Eng.18, 121–130.

@39# Dempsey JP~1981!, The wedge subjected to tractions: A parad resolved,J. Elast.11, 1–10.

@40# Sinclair GB~1998!, Further paradoxes in generalized Levy problem J. Elast.50, 253–259.

@41# Levy M ~1898!, On the elastic equilibrium of a masonry dam triangular cross section~French!, Acad. Sci., Paris, C. R.127, 10–15.

@42# Timoshenko SP, and Goodier JN~1970!, Theory of Elasticity, 3rd Edition, McGraw-Hill, New York, NY.

@43# Fillunger P~1912!, Three important plane states of stress for wed shaped bodies~German!, Z. Angew. Math. Mech.60, 275–285.

@44# Ting TCT ~1984!, The wedge subjected to tractions: A paradox examined,J. Elast.14, 235–247.

@45# Sternberg E, and Koiter WT~1959!, Discussion of ‘‘The wedge unde a concentrated couple: A paradox in the two-dimensional theory elasticity,’’ ASME J. Appl. Mech.26, 472–474.

@46# Sneddon IN~1951!, Fourier Transforms, McGraw-Hill, New York, NY.

@47# Brock LM ~1979!, Frictionless indentation by a rigid wedge: Th effect of tangential displacements in the contact zone,Int. J. Eng. Sci. 17, 365–372.

@48# Rao AK ~1971!, Stress concentrations and singularities at interfa corners,Z. Angew. Math. Mech.51, 395–406.

@49# Erdogan F, and Gupta GD~1972!, Stresses near a flat inclusion i bonded dissimilar materials,Int. J. Solids Struct.8, 533–547.

@50# Jones JP, and Whittier JS~1967!, Waves at a flexibly bonded inter face,ASME J. Appl. Mech.34, 905–909.

@51# Lene F, and Leguillon D~1982!, Homogenized constitutive law for a partially cohesive composite material,Int. J. Solids Struct.18, 443– 458.

@52# Ying X, and Katz IN~1987!, A uniform formulation for the calcula- tion of stress singularities in the plane elasticity of a wedge compo of multiple isotropic materials,Comput. Math. Appl.14, 437–458.

@53# Dundurs J~1969!, Discussion of ‘‘Edge-bonded dissimilar orthogon elastic wedges under normal and shear loading,’’ASME J. Appl. Mech.36, 650–652.

@54# Bogy DB ~1968!, Edge-bonded dissimilar orthogonal elastic wedg under normal and shear loading,ASME J. Appl. Mech.35, 460–466.

@55# Zak AR, and Williams ML~1963!, Crack point stress singularities a a bi-material interface,ASME J. Appl. Mech.30, 142–143.

@56# Dundurs J~1967!, Effect of elastic constants on stress in a compos under plane deformation,J. Compos. Mater.1, 310–322.

@57# Koguchi H, Inoue T, and Yada T~1996!, Stress singularity in three phase bonded structure,ASME J. Appl. Mech.63, 252–258.

@58# Williams MI ~1959!, The stresses around a fault or crack in dissimi media,Bull. Seismol. Soc. Am.49, 199–204.

@59# Bogy DB ~1971!, Two edge-bonded elastic wedges of different ma rials and wedge angles under surface tractions,ASME J. Appl. Mech. 38, 377–386.

@60# Theocaris PS, and Gdoutos EE~1977!, Stress singularities in cracke composite full-planes,Int. J. Fract.13, 763–773.

@61# Ting TCT ~1986!, Explicit solution and invariance of the singularitie at an interface crack in anisotropic composites,Int. J. Solids Struct. 22, 965–983.

@62# Ballarini R ~1990!, A rigid line inclusion at a bimaterial interface Eng. Fract. Mech.37, 1–5.

@63# Comninou M~1977!, Interface crack with friction in the contact zone ASME J. Appl. Mech.44, 780–781.

@64# Dempsey JP, and Sinclair GB~1981!, On the singular behavior at th vertex of a bi-material wedge,J. Elast.11, 317–327.

@65# Comninou M~1977!, The interface crack,ASME J. Appl. Mech.44, 631–636.

@66# Comninou M ~1976!, Stress singularity at a sharp edge in conta problems with friction,Z. Angew. Math. Phys.27, 493–499.

@67# Dundurs J, and Lee M-S~1972!, Stress concentration at a sharp ed in contact problems,J. Elast.2, 109–112.

@68# Wu K-C ~1990!, Line inclusions at anisotropic bimaterial interfac Mech. Mater.10, 173–182.

@69# Chen YZ, and Hasebe N~1993!, Singularity eigenvalue analysis of crack along a wedge-shaped interface,ASME J. Appl. Mech.60, 781– 783.

@70# Van Vroonhoven JCW~1992!, Stress singularities in bi-materia wedges with adhesion and delamination,Fatigue Fract. Eng. Mater. Struct.15, 159–171.

@71# Bogy DB, and Wang KC~1971!, Stress singularities at interface co ners in bonded dissimilar isotropic elastic materials,Int. J. Solids Struct.7, 993–1005.

x

s,

f

e-

e-

of

e

ce

sed

l

es

t

ite

ar

e-

s

,

ct

e

,

l

-

@72# Chen D-H, and Nisitani H~1993!, Singular stress field near the corne of jointed dissimilar materials,ASME J. Appl. Mech.60, 607–613.

@73# Pageau SS, Joseph PF, and Biggers Jr SB~1994!, The order of stress singularities for bonded and disbonded three-material junctions,Int. J. Solids Struct.31, 2979–2997.

@74# Chaudhuri RA, Xie M, and Garala HJ~1996!, Stress singularity due to kink band weakening a unidirectional composite under compr sion,J. Compos. Mater.30, 672–691.

@75# Khrapkov AA ~1968!, First fundamental problem for a piecewise homogeneous plane with a slit perpendicular to the line of separa J. Appl. Math. Mech.32, 666–678.

@76# Bogy DB ~1971!, On the plane elastostatic problem of a loaded cra terminating at a material interface,ASME J. Appl. Mech.38, 911– 918.

@77# Comninou M, and Dundurs J~1979!, A closed crack tip terminating a an interface,ASME J. Appl. Mech.46, 97–100.

@78# Wijeyewickrema AC, Dundurs J, and Keer LM~1995!, The singular stress field of a crack terminating at a frictional interface between materials,ASME J. Appl. Mech.62, 289–293.

@79# Gharpuray VM, Dundurs J, and Keer LM~1991!, A crack terminating at a slipping interface between two materials,ASME J. Appl. Mech. 58, 960–963.

@80# Fenner DN~1976!, Stress singularities in composite materials with arbitrarily oriented crack meeting an interface,Int. J. Fract.12, 705– 721.

@81# Yong-Li W ~1992!, Crack tip stress singularities in a bimaterial wit an inclined interface,Int. J. Fract.54, R65–R72.

@82# Wang W-C, and Chen J-T~1994!, Singularities of an arbitrarily in- clined semi-infinite crack meeting a bimaterial interface,Eng. Fract. Mech.49, 671–680.

@83# Tan MA, and Meguid SA~1997!, Stress singularities of a sharp notc terminating at a bimaterial interface,Proc of 16th Canadian Congress of Applied MechanicsVol 1, Laval Univ, Quebec, Canada, 17–18.

@84# Pinsan C, and Zhuping D~1991!, Singular behavior and asymptoti stress field of a crack terminating at a bimaterial interface,Acta Mech. Sin.7, 344–350.

@85# Bogy DB ~1970!, On the problem of edge-bonded quarter-plan loaded at the boundary,Int. J. Solids Struct.6, 1287–1313.

@86# Hein VL, and Erdogan F~1971!, Stress singularities in a two-materia wedge,Int. J. Fract. Mech.7, 317–330.

@87# Kubo S, Ohji K, and Nakai Y~1991!, Free-edge singularities in edge bonded elastic dissimilar wedges—geometrical conditions for app ance and disappearance of stress singularities, Report of the Meeting of the Committee on Fracture Mechanics, Vol 39, Soc Materials Science, Japan, 13–17~this work is summarized in Mu- rakami @95#!.

@88# Theocaris PS, and Gdoutos EE~1976!, Stress singularities at equa angle biwedges and two-material composite half planes with ro interfaces,ASME J. Appl. Mech.43, 64–68.

@89# Theocaris PS, and Gdoutos EE~1976!, Stress singularities at vertice of composite plates with smooth or rough interfaces,Arch. Mech. Stosow.28, 693–704.

@90# Aksentian OK~1967!, Singularities of the stress-strain state of a pla in the neighborhood of an edge,J. Appl. Math. Mech.31, 193–202.

@91# Avetisian AG, and Chobanian KS~1972!, On the nature of stresses i the fixed vicinity of the junction surface brink of a composite elas body loaded in conditions of a plane problem of the elasticity the ~Russian!, Dokl. Akad. Nauk Arm. SSR25, 13–25.

@92# Picu CR, and Gupta V~1996!, Stress singularities at triple junction with freely sliding grains,Int. J. Solids Struct.33, 1535–1541.

@93# Inoue T, and Koguchi H~1996!, Influence of the intermediate materia on the order of stress singularity in three-phase bonded structureInt. J. Solids Struct.33, 399–417.

@94# Pageau SS, Gadi KS, Biggers Jr SB, and Joseph PF~1996!, Standard- ized complex and logarithmic eigensolutions forn-material wedges and junctions,Int. J. Fract.77, 51–76~See also: ibid.,80, 97!.

@95# Murakami Y ~1992!, Stress singularity for notch at bimaterial inte face, Stress Intensity Factors HandbookVol 3, Y Murakami et al ~eds!, Pergamon Press, Oxford, UK, 963–1062.

@96# Chen D-H~1996!, Logarithmic singular stress field in a semi-infinit plate consisting of two edge-bonded wedges subjected to surface tions, Int. J. Fract.75, 352–378.

@97# Courant R and Hilbert D~1953!, Methods of Mathematical Physics, 1st English Edition, Vol I, Interscience Publ, New York, NY.

@98# de Saint-Venant B~1878!, On the torsion of prisms with bases mad of circular arcs and radial lines, and on a singularity which can found in certain uses of the logarithmic cylindrical coordinate syst of Lamé ~French!, Acad. Sci., Paris, C. R.87, 849–854.

-

i

-

t

,

c

i

r

i

f

t

e

r

,

he

es

s

-

p-

n

n igh

d

i-

n

l

der

in

of

y by

r is-

a

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 437

@99# Sinclair GB ~1980!, On the singular eigenfunctions for plane ha monic problems in composite regions,ASME J. Appl. Mech.47, 87– 92.

@100# Thomson W, and Tait PG~1867!, A Treatise on Natural Philosophy, Cambridge Univ Press, Cambridge, UK~This book is also available in two parts asPrinciples of Mechanics and Dynamics, Dover Publ, New York, NY!.

@101# de Saint-Venant B~1878!, Sample calculations for the torsion o prisms with bases made of circular arcs and radial lines~French!, Acad. Sci., Paris, C. R.87, 893–899.

@102# Irwin GR, and Kies JA~1957!, Fracture,Proc of ONR Symp, Wash- ington DC, 563–571.

@103# Wasow W ~1957!, Asymptotic development of the solution of Di richlet’s Problem at analytic corners,Duke Math. J.24, 47–56.

@104# Ting TCT ~1985!, Elastic wedge subjected to antiplane she tractions—A paradox explained,Q. J. Mech. Appl. Math.38, 245– 255.

@105# Sinclair GB ~1999!, A note on the removal of further breakdowns classical solutions of Laplace’s equation on sectorial regions,J. Elast. 56, 247–252.

@106# Ma C-C, and Hour B-L~1989!, Analysis of dissimilar anisotropic wedges subjected to antiplane shear deformation,Int. J. Solids Struct. 25, 1295–1309.

@107# Sinclair GB ~2000!, Logarithmic stress singularities in two dimensional and three-dimensional elasticity,Proc of 20th Southeast- ern Conf of Theoretical and Applied Mechanics, Callaway Gardens, GA, SM94.1–8.

@108# Barnett DM ~1967!, The effect of shear modulus on the stress dis bution produced by a planar array of screw dislocations near a metallic interface,Acta Metall.15, 589–594.

@109# Sendeckyj GP~1973!, Analysis of cracks in heterogeneous solid Fracture Mechanics of Ceramics, Vol 1, RC Bradt, DPH Hasselman and FF Lang~eds!, Plenum Press, New York, NY, 269–282.

@110# Pageau SS, Joseph PF, and Biggers Jr SB~1995!, Singular antiplane stress fields for bonded and disbonded three-material junctions,Eng. Fract. Mech.52, 821–832.

@111# Keer LM, and Freeman NJ~1970!, Load transfer problem for an embedded shaft in torsion,ASME J. Appl. Mech.37, 959–964.

@112# Love AEH ~1944!, A Treatise on the Mathematical Theory of Elasti ity, 4th Edition, Dover Publ, New York, NY.

@113# Kirchhoff G ~1850!, On the equilibrium and deflection of an elast plate ~German!, J. Reine Angew. Math.40, 51–88.

@114# Timoshenko SP, and Woinowsky-Krieger S~1959!, Theory of Plates and Shells, 2nd Edition, McGraw-Hill, New York, NY.

@115# Williams ML ~1951!, Surface stress singularities resulting from va ous boundary conditions in angular corners of plates under bend Proc of 1st US Nat Congress of Applied Mechanics, Illinois Inst of Tech, Chicago, IL, 325–329.

@116# Southwell RV~1950!, On the analogues relating flexure and extens of flat plates,Q. J. Mech. Appl. Math.3, 257–271.

@117# Dixon AC ~1934!, The problem of the rectangular plate,J. Lond. Math. Soc.9, 61–74.

@118# Carrier GF and Shaw FS~1949!, Some problems in the bending o thin plates,Proc of 3rd Symp in Applied Mathematics of the A Mathematical Soc, Vol 3, Univ of Michigan, Ann Arbor, MI, 125– 128.

@119# Morley LSD ~1964!, Bending of clamped rectilinear plates,Q. J. Mech. Appl. Math.17, 293–317.

@120# Hrudey TM, and Hrabok MM~1986!, Singularity finite elements for plate bending,J. Eng. Mech.112, 666–681.

@121# Leissa AW, McGee OG, and Huang CS~1993!, Vibrations of sectorial plates having corner stress singularities,ASME J. Appl. Mech.60, 134–140.

@122# Gregory RD, Chonghua G, and Wan FYM~1996!, The cantilever strip under torsion, bending or flexure at infinity,J. Elast.43, 109–136.

@123# Williams ML ~1961!, The bending stress distribution at the base o stationary crack,ASME J. Appl. Mech.28, 78–82.

@124# Williams ML ~1957!, On the stress distribution at the base of a s tionary crack,ASME J. Appl. Mech.24, 109–114.

@125# Irwin GR ~1957!, Analysis of stresses and strains near the end o crack traversing a plate,ASME J. Appl. Mech.24, 361–364.

@126# Sinclair GB ~2000!, Logarithmic stress singularities resulting from various boundary conditions in angular corners of plates under b ing, ASME J. Appl. Mech.67, 219–223.

@127# Nádai A ~1925!, The Elastic Plate~German!, Julius Springer Publ, Berlin, Germany.

@128# Sih GC, and Rice JR~1964!, The bending of plates of dissimila materials with cracks,ASME J. Appl. Mech.31, 477–482.

r-

f

ar

n

ri- bi-

s,

-

c

i- ing,

on

f m

a

a-

f a

nd-

@129# Huang CS~2002!, Corner singularities in bi-material Mindlin plates Compos. Struct.56, 315–327.

@130# Reissner E~1945!, The effect of transverse shear deformation on t bending of elastic plates,ASME J. Appl. Mech.12, A-69–A-77.

@131# Hencky H ~1947!, On the treatment of shear rotations in flat plat ~German!, Oesterr. Ing.-Arch.16, 72–76.

@132# Burton WS, and Sinclair GB~1986!, On the singularities in Reissner’ theory for the bending of elastic plates,ASME J. Appl. Mech.53, 220–222.

@133# Yen DHY, and Zhou M~1989!, On the corner singularities in Reiss ner’s theory for the bending of elastic plates,ASME J. Appl. Mech. 56, 726–729.

@134# Huang CS, McGee OG, and Leissa AW~1994!, Exact analytical so- lutions for free vibrations of thick sectional plates with simply su ported radial edges,Int. J. Solids Struct.31, 1609–1631.

@135# Knowles JK, and Wang N-M~1960!, On the bending of an elastic plate containing a crack,Z. Agnew Math. Phys.39, 223–236.

@136# Hartranft RJ, and Sih GC~1968!, Effect of plate thickness on the bending stress distribution around through cracks,Z Agnew Math. Phys.47, 276–291.

@137# Wang N-M ~1968!, Effects of plate thickness on the bending of a elastic plate containing a crack,Z Agnew Math. Phys.47, 371–390.

@138# Hartranft RJ ~1966!, On some problems involving singularities i Reissner’s theory of plate bending, Doctoral dissertation, Leh Univ, Bethlehem, PA.

@139# Abramowitz M, and Stegun IA~1965!, Handbook of Mathematical Functions, Dover Publ, New York, NY.

@140# Bažant ZP, and Keer LM~1974!, Singularities of elastic stresses an of harmonic functions at conical notches or inclusions,Int. J. Solids Struct.10, 957–964.

@141# Thomson TR, and Little RW~1970!, End effects in a truncated sem infinite cone,Q. J. Mech. Appl. Math.23, 185–195.

@142# Beagles AE, and Sa¨ndig A-M ~1991!, Singularities of rotationally symmetric solutions of boundary value problems for the Lame´ equa- tions,Z. Angew. Math. Mech.71, 423–431.

@143# Love AEH ~1939!, Boussinesq’s problem for a rigid cone,Q. J. Math. 10, 161–175.

@144# Hanson MT~1993!, The elastic field for a sliding conical punch on a isotropic half-space,ASME J. Appl. Mech.60, 557–559.

@145# Keer LM, and Parihar KS~1978!, Elastic stress singularity at conica inclusions,Int. J. Solids Struct.14, 261–263.

@146# Reissner E, and Sagoci HF~1944!, Forced torsional oscillations of an elastic half-space: I,J. Appl. Phys.15, 652–654.

@147# Tsuji T, Noda N, Shibuya T, and Koizumi T~1990!, Stress singularity at the bottom of a V-notch in the inner surface of a circular pipe un torsion,Int. J. Pressure Vessels Piping42, 135–155.

@148# Freeman NJ, and Keer LM~1967!, Torsion of a cylindrical rod welded to an elastic half space,ASME J. Appl. Mech.34, 687–692.

@149# Westmann RA~1968!, Discussion of ‘‘Torsion of a cylindrical rod welded to an elastic half space,’’ASME J. Appl. Mech.35, 197–198.

@150# Keer LM, and Freeman NJ~1969!, Torsion of a finite elastic cylin- drical rod partially bonded to an elastic half space,Q. Appl. Math.26, 567–573.

@151# Barton MV ~1941!, The circular cylinder with a band of uniform pressure on a finite length of the surface,ASME J. Appl. Mech.8, A-97–A-104.

@152# Zate AR~1964!, Stresses in the vicinity of boundary discontinuities bodies of revolution,ASME J. Appl. Mech.31, 150–152.

@153# Sneddon IN~1946!, The distribution of stress in the neighborhood a crack in an elastic solid,Proc. R. Soc. London, Ser. AA187, 229– 260.

@154# Harding JW, and Sneddon IN~1945!, The elastic stresses produced b the indentation of the plane surface of a semi-infinite elastic solid a rigid punch,Proc. Cambridge Philos. Soc.41, 16–26.

@155# Benthem JP, and Minderhoud P~1972!, The problem of the solid cylinder compressed between rough rigid stamps,Int. J. Solids Struct. 8, 1027–1042.

@156# Agarwal VK ~1978!, Axisymmetric solution of the end-problem fo semi-infinite elastic circular cylinder and its application to joined d similar cylinders under uniform tension,Int. J. Eng. Sci.16, 985–998.

@157# Hartranft RJ, and Sih GC~1977!, Stress singularity for a crack with an arbitrarily curved front,Eng. Fract. Mech.9, 705–718.

@158# Sadowsky MA, and Sternberg E~1949!, Stress concentration around triaxial ellipsoidal cavity,ASME J. Appl. Mech.16, 149–157.

@159# Green AE, and Sneddon IN~1950!, The distribution of stress in the neighborhood of a flat elliptical crack in an elastic solid,Proc. Cam- bridge Philos. Soc.46, 159–164.

n

s t

y

g

w

e

c p

l

e

a

e

-

or

s,

ite

r lids, . - gi- iv,

- -

r-

e

in a

k,

-

g

f

a

s-

-

al

f-

nd

s pic

,

s-

438 Sinclair: Stress singularities in classical elasticity–II Appl Mech Rev vol 57, no 5, September 2004

@160# Kassir MK, and Sih GC~1966!, Three-dimensional stress distributio around an elliptical crack under arbitrary loadings,ASME J. Appl. Mech.33, 601–611.

@161# Galin LA ~1948!, Determination of displacements in three dimensional contact problems of the theory of elasticity~Russian!, Prikl. Mat. Mekh.12, 241–250.

@162# Mindlin RD ~1949!, Compliance of elastic bodies in contact,ASME J. Appl. Mech.16, 259–268.

@163# Hughes WF, and Gaylord EW~1964!, Basic Equations of Engineering Science, Schaum Publ Co, New York, NY.

@164# Love AEH ~1929!, The stress produced in a semi-infinite solid b pressure on part of the boundary,Philos. Trans. R. Soc. London, Se A A228, 397–420.

@165# Turteltaub MJ, and Wheeler LT~1976!, Line integral representation for the traction problem of the elastic halfspace and their applica to a study of the singular effects of load jumps,J. Elast.6, 161–178.

@166# Michell JH ~1902!, The inversion of plane stress,Proc. London Math. Soc.34, 134–142.

@167# Green AE, and Zerna W~1968!, Theoretical Elasticity, 2nd Edition, Oxford at Clarendon Press, UK.

@168# Smith FW, and Alavi MJ~1971!, Stress intensity factors for a penn shaped crack in a half space,Eng. Fract. Mech.3, 241–254.

@169# Shah RC, and Kobayashi AS~1973!, Stress intensity factors for an elliptical crack approaching the surface of a semi-infinite solid,Int. J. Fract. 9, 133–146.

@170# Liao CY, and Atluri SN~1989!, Stress intensity factor variation alon a semicircular surface flaw in a finite-thickness plate,Eng. Fract. Mech.34, 957–976.

@171# Galin LA ~1947!, Pressure of a stamp with a flat base in the form an infinite wedge on an elastic half-space~Russian!, Dokl. Akad. Nauk SSSR58„2….

@172# Rvachev VL~1959!, The pressure on an elastic half-space of a sta with a wedge-shaped planform,J. Appl. Math. Mech.23, 229–233.

@173# Noble B ~1959!, The potential and charge distribution near the tip a flat angular sector, Report NYU EM-135, New York Univ, Ne York, NY.

@174# Aleksandrov VM, and Babeshko VA~1972!, On the pressure on an elastic half-space of a wedge-shaped stamp,J. Appl. Math. Mech.36, 78–83.

@175# Bažant ZP~1974!, Three-dimensional harmonic functions near term nation or intersection of gradient intersection lines: A general num cal method,Int. J. Eng. Sci.12, 221–243.

@176# Walden H~1974!, Solution of an eigenvalue problem for the Lapla operator on a spherical surface, Report X-582-74-41, Goddard S Flight Center, Greenbelt, MD.

@177# Morrison JA, and Lewis JA~1976!, Charge singularity at the corne of a flat plate,SIAM (Soc. Ind. Appl. Math.) J. Appl. Math.31, 233– 250.

@178# Brothers PW~1976!, The rigid rectangular punch on the elastic ha space, Masters thesis, Univ of Auckland, Auckland, New Zealand

@179# Keer LM, and Parihar KS~1978!, A note on the singularity at the corner of a wedge-shaped punch or crack,SIAM (Soc. Ind. Appl. Math.) J. Appl. Math.34, 297–302.

@180# Ioakimidis NI ~1985!, Determination of the order of singularity at th apex of a wedge-shaped crack,Eng. Fract. Mech.22, 369–373.

@181# Takakuda K~1985!, Stress singularities near crack front edges,Bull. Japan Soc. Mech. Engineers28, 225–231.

@182# Xu L, and Kundu T~1995!, Stress singularities at crack corners,J. Elast.39, 1–16.

@183# Glushkov E, Glushkova N, and Lapina O~1999!, 3D elastic stress singularity at polyhedral corner points,Int. J. Solids Struct.36, 1105– 1128.

@184# Parihar KS, and Keer LM~1978!, Stress singularity at the corner of wedge-shaped crack or inclusion,ASME J. Appl. Mech.45, 791–796.

@185# Parihar KS, and Keer LM~1979!, The singularity at the apex of a bonded wedge-shaped stamp,ASME J. Appl. Mech.46, 577–580.

@186# Abramov BM ~1937!, The problem of contact of an elastic infinit half-plane with an absolutely rigid rough foundation,C. R. (Dokl.) Acad. Sci. URSS17, 173–178.

@187# Abdel-Messieh YS, and Thatcher RW~1990!, Estimating the form of some three-dimensional singularities,Commun. Appl. Numer. Meth ods6, 333–341.

@188# Schmitz H, Volk K, and Wendland W~1993!, Three-dimensional sin- gularities of elastic fields near vertices,Numer. Methods Partial Dif- fer. Equ.9, 323–337.

@189# Panasyuk VV, Andrejkiv AE, and Stadnik MM~1980!, Basic me- chanical concepts and mathematical techniques in application three-dimensional crack problems: A review,Eng. Fract. Mech.13, 925–937.

-

y r.

ion

of

mp

of

i- ri-

e ace

r

f- .

to

@190# Panasyuk VV, Andrejkiv AE, and Stadnik MM~1981!, Three- dimensional static crack problems solution~A review!, Eng. Fract. Mech.14, 245–260.

@191# Sih GC~1971!, A review of the three-dimensional stress problem f a cracked plate,Int. J. Fract. Mech.7, 39–61.

@192# Folias ES~1975!, On the three-dimensional theory of cracked plate ASME J. Appl. Mech.42, 663–674.

@193# Kawai T, Fujitani Y, and Kumagai K~1977!, Analysis of singularity at the root of the surface crack problem,Proc of Int Conf on Fracture Mechanics and Technology, Hong Kong, 1157–1163.

@194# Benthem JP~1977!, State of stress at the vertex of a quarter-infin crack in a half-space,Int. J. Solids Struct.13, 479–492.

@195# Bažant ZP and Estenssoro LF~1977!, General numerical method fo three-dimensional singularities in cracked or notched elastic so Proc of 4th Int Conf on Fracture, Vol 3, Waterloo, Ontario, 371–383

@196# Bažant ZP, and Estenssoro LF~1977!, Stress singularity and propa gation of cracks at their intersection with surfaces, Structural En neering Report 77-12/480, Dept of Civil Eng, Northwestern Un Evanston, IL.

@197# Bažant ZP, and Estenssoro LF~1979!, Surface singularity and crack propagation,Int. J. Solids Struct.15, 405–426~See also: Addendum ~1980!, ibid. 16, 479–481, and Erratum~1983!, ibid. 19, 661!.

@198# Sinclair GB~1979!, Asymptotic singular eigenfunctions for the three dimensional crack,Proc of 7th Canadian Congress of Applied Me chanics, Sherbrooke, Quebec, 295–296.

@199# Benthem JP~1980!, The quarter-infinite crack in a half-space: Alte native and additional solutions,Int. J. Solids Struct.16, 119–130.

@200# Yamada Y, and Okumura H~1981!, Analysis of local stress in com- posite materials by the 3D finite element,Proc of Japan-US Conf on Composite Materials, Tokyo, Japan, 55–84.

@201# Burton WS, Sinclair GB, Solecki JS, and Swedlow JL~1984!, On the implications for LEFM of the three-dimensional aspects in som crack-surface intersection problems,Int. J. Fract.25, 3–32.

@202# Shaofu W, Xing Z, and Qingzhi H~1988!, Functional variable dis- placement method in analysis of singularity near the corner point three-dimensional cracked solid,Eng. Fract. Mech.31, 191–200.

@203# Shivakumar KN, and Raju IS~1990!, Treatment of singularities in cracked bodies,Int. J. Fract.45, 159–178.

@204# Zhu WX ~1990!, Singular stress field of three-dimensional crac Eng. Fract. Mech.36, 239–244.

@205# Barsoum RS, and Chen T-K~1991!, Three-dimensional surface sin gularity of an interface crack,Int. J. Fract.50, 221–237.

@206# Ghahremani F~1991!, A numerical variational method for extractin 3D singularities,Int. J. Solids Struct.27, 1371–1386.

@207# Leung AYT, and Su RKL~1996!, Order of the singular stress field o through-thickness cracks,Int. J. Fract.75, 85–93.

@208# Leung AYT, and Su RKL~1996!, Analytical solution for mode I crack orthogonal to free surface,Int. J. Fract.76, 79–85.

@209# Su XM, and Sun CT~1996!, On singular stress of the crack tip of thick plate under in-plane loading,Int. J. Fract.82, 237–252.

@210# Lur’e AI ~1964!, Three-Dimensional Problems of the Theory of Ela ticity, Interscience Publ, John Wiley and Sons, New York, NY.

@211# Benthem JP, and Koiter WT~1976!, Discussion of ‘‘On the three- dimensional theory of cracked plates,’’ASME J. Appl. Mech.43, 374.

@212# Folias ES~1976!, Authors closure to discussion of ‘‘On the three dimensional theory of cracked plates,’’ASME J. Appl. Mech.43, 374–375.

@213# Folias ES~1980!, Method of solution of a class of three-dimension elastostatic problems under mode I loading,Int. J. Fract. 16, 335– 348.

@214# Meda G, Messner TW, Sinclair GB, and Solecki JS~1998!, Path- independent H integrals for three-dimensional fracture mechanics,Int. J. Fract. 94, 217–234.

@215# Ghahremani F, and Shih CF~1992!, Corner singularities of three- dimensional planar interface cracks,ASME J. Appl. Mech.59, 61–68.

@216# Sagochi HF~1944!, Forced torsional oscillations of an elastic hal space: II,J. Appl. Phys.15, 655–662.

@217# Leissa AW~2001!, Singularity considerations in membrane, plate a shell behaviors,Int. J. Solids Struct.38, 3341–3353.

@218# Achenbach JD, and Bazˇant ZP~1975!, Elastodynamic near-tip stres and displacement fields for rapidly propagating cracks in orthotro materials,ASME J. Appl. Mech.42, 183–189.

@219# Nowacki W ~1968!, Thermoelasticity, 2nd Edition, Pergamon Press Oxford, UK.

@220# Yan G, and Ting TCT~1993!, Ther 21/2(ln r) singularities at interface cracks in monoclinic and isotropic bimaterials due to heat flow,ASME J. Appl. Mech.60, 432–437.

@221# Yang YY, and Munz D~1995!, Stress intensity factor and stress di

-

n

e

Appl Mech Rev vol 57, no 5, September 2004 Sinclair: Stress singularities in classical elasticity–II 439

tribution in a joint with an interface corner under thermal and m chanical loading,Comput. Struct.57, 467–476.

@222# Lee EH~1955!, Stress analysis in visco-elastic bodies,Q. Appl. Math. 13, 183–190.

@223# Williams ML ~1962!, The fracture of viscoelastic materials,Fracture of Solids, Proc of an Int Conf, Maple Valley, WA, 157–188.

@224# Ting TCT ~1986!, Anisotropic Elasticity, Oxford Univ Press, New York, NY.

e-@225# Ting TCT ~1998!, The remarkable nature of cylindrically uniform anisotropic elastic materials,Proc of 13th US Natl Congress of Ap plied Mechanics, TA7.

@226# Muki R, and Sternberg E~1965!, The influence of couple-stresses o singular stress concentrations in elastic solids,Z. Angew. Math. Phys. 16, 611–648.

@227# Sternberg E, and Muki R~1967!, The effect of couple-stresses on th stress concentration around a crack,Int. J. Solids Struct.3, 69–95.

New ce in

e has ellon Pro- han- roving d me- ement

Glenn Sinclair received his undergraduate education from the University of Auckland in Zealand. He graduated with a BS in Mathematics in 1967 and a BE in Engineering Scien 1969. He then attended Caltech, graduating with a PhD in Applied Mechanics in 1972. H since served on the faculty of Yale University, the University of Auckland, Carnegie M University, and Louisiana State University, where he is currently the Francis S Blummer fessor of Mechanical Engineering. His research is primarily concerned with fracture mec ics, tribology, and numerical methods. Recent interests focus on finding means of imp modeling so that stress and pressure singularities are removed from both solid and flui chanics problems, and submodeling procedures and verification techniques for finite el analysis.

__MACOSX/stress/._Stress Singularities in Classical Elasticity II.pdf

stress/Term Project (Cantilever Beams)-Part I (2018-03-11).pdf

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 1/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

�� TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

CANTILEVER BEAMS

PART I: AXIAL TENSION

Date: 03-11-2018

The main goal of this term project is to compare solutions obtained by the 1-D Mechanics-of-Materials (MoM) approach, which are approximate yet more practical, with the 3-D finite element (FE) solutions. Whenever the exact and rigorous Elasticity solutions are available, they should also be included for comparison.

All FE simulations must be performed using SolidWorks 2016 version. Students are strongly encouraged to use the computer stations in ME CAD Room (Steinman Room ST-213). Once executed successfully, be sure to save all key results (e.g., stresses and displacements in ASCII/text, MS Excel and/or graphic formats), which are necessary for comparison with the theoretical MoM/Elasticity predictions. When presenting results graphically, students are suggested to use the deformed geometry. If possible, also superimpose them the undeformed meshes. Otherwise, juxtaposing the deformed and undeformed geometries for easy comparison. It should be noted that SolidWorks, just like almost all commercial general-purpose FE software, allows users to adjust scale factors to control Deformation Shape and Stress Fringes for better viewing.

In the term project, each student is assigned with a different set of load values, although all students will work on the same geometries and boundary conditions of the 3-D bars. All student are expected to add his/her own proper loading and boundary conditions, material properties and other information/data, if so required by SolidWorks and abstract all needed results (graphical, texted and/or tabulated) for the report.

In order to compare the SolidWorks FE simulations with theoretical MoM/Elasticity results numerically and/or graphically, the analytical solutions should first be formulated, derived and/or simply cited with proper references. When an analytical solution is adopted from a know source, the equation(s) needed for comparison with the FE results should only be included with concise explanation in the main text of the report and the formula can be simply, yet clearly, referred to the textbook or class notes (e.g., Eq. so and so in Page so and so, Section so and so, etc.). If a equation for comparison is not available for lifting off directly from the textbook or class notes, the student should derive it by him/herself. Details of the formulation/derivation should be given in the Appendix of the report.

PART I: AXIAL TENSION

I.1 Cantilever Beams of Circular Solid Sections (Required for all students)

As shown in Fig I.1, a prismatic cantilever beam of length L and circular solid cross-section of radius 0R is fixed at one end (A) while loaded axially at the other (B). The x-axis is through the centroidal axis (C.A.) of the axial bar, which is made of a linearly elastic material of Young’s modulus E and Poisson’s ratio Q. The bar is loaded axially at the free end by (a) a uniform tension 0V and (b) a concentrated axial force

2 0 0 0 0P A RV S V through the centroid of the end cross-section. In this term project, each student is assigned

with a different 0V value (thus, of course a different 0P ) although all students will work on the same geometries, boundary conditions and material properties. Refer to the file: Term-Project Case Assignment (2018-02-27) for the load values you need to work on and Table I.1 for the material and dimensional data.

yousef ha
yousef ha
yousef ha
yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 2/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

z

y

R0

y

x

L

V0

A BC

2x L

z

y

R0

P0

L

y

x P0BC

2x L

A

(a) uniform axial tension (b) concentrated axial force through C.A.

FIGURE I.1 An axially loaded circular cylindrical bar.

TABLE I.1 Material and dimensional data of bars of circular cross-section.

E Q 0R L 1R 2R

> @GPa > @- > @m > @m > @m > @m

2 0.25 0.2 06R 00.5R 01.5R

The afore-mentioned bar geometry has been prepared in a SolidWorks Assembly file: Assembly 2_1.SLDASM, which represents a quarter of the circular bar in the 1st octant � �0, 0, 0x y zt t t of the Cartesian x-y-z coordinate system. Each student is expected to add his/her proper loading and boundary conditions, material properties and other information/data, if so required by SolidWorks.

For Case (a): Uniform Axial Tension, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Roller/Slider option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x , the

horizontal � �0z and vertical � �0y symmetric/central planes.

x A uniform tension of 0V value should be applied on the free end face � �x L through the Pressure option of SolidWorks “External Loads”.

x An element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes (approximate 202,000 elements) through the Create Mesh option of SolidWorks “Mesh”.

Similarly, for Case (b): Concentrated Tensile Force, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Fixed Geometry option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x whereas, again, the Roller/Slider option should be applied in SolidWorks “Fixtures” to the horizontal � �0z and vertical � �0y symmetric/central planes.

x A concentrated tensile axial force of 0P value should be applied at the center of on the free-end

� � � �, , ,0,0 mx y z L . x Once again, an element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes

(approximate 202,000 elements).

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 3/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

Report: (minimum items to be included) 1. Summarize the MoM solutions of the problem, including the two field variables: stress tensor

� �, 1,2,3ij i jW and displacement vector � �1,2,3iu i . 2. For each FE case, show the mesh with loading and restraint B.C.’s. Be sure to include the coordinate system

used in the FE simulation. 3. Compare the MoM solutions with the two FEM results for the following cases:

a) Plot on the same graph the axial extensions u in [mm] from both FE simulations and MoM prediction along the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

b) Plot on the same graph the transverse contraction v in [mm] from both FE simulations and MoM

prediction along the vertical diameter of the mid-section 0;0 ; 0 2 Lx y R z§ · d d ¨ ¸

© ¹ . In a separate plot,

demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

c) Display 3 normal stress contour plots for � �, ,x y zV V V , respectively, on the mid-span cross-section

2 Lx§ · ¨ ¸

© ¹ on the same page from the Axial-Tension Case. Repeat the same for the Concentrated-Force

Case. Compare the results with the MoM predictions. Hint: The MATLAB command meshgrid can be used to defined the contour plot area whereas, as the

name implies, the command contour is for contour plots.

d) Display 3 shear stress contour plots for � �, ,xy yx zxW W W , respectively, on the mid-span cross-section 2 Lx§ · ¨ ¸

© ¹

on the same page from the Axial-Tension Case. Repeat the same for the Concentrated-Force Case. Compare the results with the MoM predictions.

e) Consider the area: 0 0;0 ; 0 2 2 R RL x L y z§ ·� d d d d ¨ ¸

© ¹ , which is a vicinity within the x-y plane around the

loading point, i.e., � � � �, , ,0,0x y z L . Display 3 xV stress contour plots in the area from the Concentrated-Force Case FE result and two Elasticity solutions based on the 2-D Flamant and 3-D Boussinesq normal-force solutions:

A. 2-D HALF-PLANE NORMAL FORCE (FLAMANT SOLUTION)

� � � �

� � � �

� � � �

3 22 0 0 0

2 2 22 2 22 2 2

2 2 2

0

x y xy

z xz yz

P L x P L x y P L x y

L x y L x y L x y V V W

S S S

V W W

� � � �

ª º ª º ª º� � � � � �¬ ¼ ¬ ¼ ¬ ¼

/ (I.1)

where the out of x-y plane line-load 0P has a unit of force

out-of-plane thickness ª º « » ¬ ¼

. For this term project,

simply choose 0 0P P .

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 4/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

B. 3-D HALF-SPACE NORMAL FORCE (BOUSSINESQ SOLUTION)

� �

� � � � � � � �

� � � � � � � �

� �

� �

3 0

5

2 2 220

3 2

2 2 220

3 2

2 0

5

2 0

5

0 3 5

3 2

3 21 2 2

3 21 2 2

3 2

3 2

3 2

x

y

z

xy

xz

yz

P L x R

y L x y R L xP R R L x L x R R R L x R L x

z L x z R L xP R R L x L x R R R L x R L x

P y L x R

P z L x R

P yz x R R

V S

QV S

QV S

W S

W S

W S

­ ½ª º� � ��° ° � � � � � �® ¾« »� � � �° °¬ ¼¯ ¿ ­ ½ª º� � ��° ° � � � � � �® ¾« »� � � �° °¬ ¼¯ ¿

� �2

21 2 R L x R L x

Q

­ ° ° ° ° ° ° ° °° ® ° ° ° ° ° ° ° � �ª º° � �« »° � �¬ ¼¯

(I.2)

where � �2 2 2R L x y z � � � .

f) Repeat e) for yV , xyW , maxV , minV , vMV and TrW (called Stress Intensity in SolidWorks), respectively.

g) Consider the area: 0 0 00 ; ; 0

2 2 R Rx y R z§ ·d d d d ¨ ¸

© ¹ , which is a vicinity within the x-y plane around the

bottom edge point of the fixed end, i.e., � � � �0, , 0, ,0x y z R . Display the stress contour plots of xV , yV ,

xyW , maxV , minV , vMV and TrW in the area from the Concentrated-Force Case FE result. h) (Option) Compare the FE results in 7) with available Elasticity solutions, e.g., Sinclair, GB “Stress

singularities in classical elasticity - I. Removal, Interpretation and Analysis and II. Asymptotic identification,” Applied Mechanics Review, Vol. 57, No. 4, pp. 251-297 and No. 5, pp. 385-439, 2004.

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 5/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

I.2 Truncated Conical Cantilever Beam (Required for all students)

Figure I.2 shows a truncated conical bar of length L and end radii � �1 2,R R is fixed as a roller/slider boundary at

the restrained end (A) while loaded axially by a uniform tension: 2 0

1 02 1

R R

V V at the free end (B). In the figure

the x-axis is through the centroidal axis (C.A.) of the bar, which is made of a linearly elastic material of Young’s modulus E and Poisson’s ratio Q.

R1

R2

z

y L

x

y

V1

2x L

CA B

FIGURE I.2 A truncated conical bar loaded by uniform axial tension.

Refer to the file: Term-Project Case Assignment (2018-02-27) for the load values you need to work on and Table I.1 for the material and dimensional data. Finally, the following boundary and loading conditions should be applied and the mesh size should be chosen:

x Roller/Slider option should be imposed in SolidWorks “Fixtures” to the restrained end face � �0x , the

horizontal � �0z and vertical � �0y symmetric/central planes.

x A uniform tension of 1V value should be applied on the free end face � �x L through the Pressure option of SolidWorks “External Loads”.

x An element size of 0.01m with a tolerance of 0.0005m should be chosen to create the meshes (approximate 215,000 elements) through the Create Mesh option of SolidWorks “Mesh”.

x

Report: (minimum items to be included) 1. Show the mesh with loading and restraint B.C.’s. Be sure to include the coordinate system used in the FE

simulation. 2. Find the MoM solution of the axial normal stress � �xV distribution.

Hint: Force balance. 3. Find the MoM solution of the displacement field � �1,2,3iu i .

Note: The governing eq of the axial displacement � �1u u of this axially loaded bar with variable cross-section is

� � � �, 0

du x tdE A x dx dx

ª º « »

¬ ¼ (I.3)

4. Due to its axi-symmetrical nature, the conventional MoM approach concludes this axially loaded circular object will not have transverse shear stresses � �,xy xzW W , hence � �,xr xTW W , where � �, ,r xT constitutes a cylindrical coordinate system. Inspecting the equilibrium equations in cylindrical coordinate system, Eq (1.8-C3), to prove this conclusion is wrong. Indeed, an Advanced MoM approach taking into account the tapered-beam effect, does prove the existence of transverse shears. The results are summarized in the file:

yousef ha

SPRING 2018 STATEMENT CANTILEVER BEAMS Page 6/6

ME 54100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS TERM PROJECT: THEORETICAL/FE BEAM ANALYSES

Transverse Shear Stresses in Tapered Beams (2017-12-06)

5. Compare the above MoM/Advanced MoM solutions with the FEM results for the following cases: a) Plot on the same graph the axial extensions u in [mm] from the FE simulation and MoM prediction along

the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

b) Plot on the same graph the transverse contraction v in [mm] from the FE simulation and MoM prediction

along the vertical diameter of the mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate plot,

demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

c) Plot on the same graph the axial normal stress xV in [MPa] from the FE simulation and MoM prediction along the C.A. � �0 ; 0; 0x L y zd d . In a separate plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the C.A. approximately every 0.1m.

d) Plot on the same graph the transverse shear stress xyW in [MPa] from the FE simulation and MoM

prediction along the vertical diameter of the mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate

plot, demonstrate how the data are abstracted using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

e) Plot the transverse normal stress yV in [MPa] from the FE simulation along the vertical diameter of the

mid-section 1 2;0 ; 0 2 2

R RLx y z�§ · d d ¨ ¸ © ¹

. In a separate plot, demonstrate how the data are abstracted

using the SolidWorks Probe option in Plot Tools. Probe the FE results along the diameter approximately every 0.05m.

yousef ha

__MACOSX/stress/._Term Project (Cantilever Beams)-Part I (2018-03-11).pdf

stress/Transverse Shear Stresses in Tapered Beams (2017-12-06)(1).pdf

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 1/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

SUPPLEMENT: TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS (Self-Study)

transverse shear stress:    

   

 

   

 

* , ,1 ,

z z

xy

z

P x A x y M x Q x yd x y

b x dx A x I x 

    

 

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 2/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

(A) Linear Tapered, Circular Cross-Section

  1 2 2 2

R R R x x R fx R

L

        

 where 1 2R RdR f

dx L

   (A1)

  2A x R ,   4

4 z

R I x

  ,   2 22b x R y  (A2)

  2

* 2 2 2 1, sin 2

R y A x y y R y R

R

     ,     3

2 2 22 ,

3 zQ x y R y  (A3)

Case A1: axial tension ~   0P x P &   0zM x 

  0 0

3 ,xy

P f P f y x y y

R A R 

   ~ linear (A4)

Case A2: end vertical force ~   0P x  &    0zM x V L x 

      

 

2 2 2 20

5

2 2

0

4 , 4

3

4 1 1 4

3

xy

V x y R R y f L x R y

R

f L xV y y

A R R R

 

       

                   

           

~ parabolic (A5)

Case A3: end moment ~   0P x  &   0zM x M

    2

2 20 0

5

4 4 , 4 1 4

3 3 xy

M f M f y x y R y

R AR R 

          

   

~ parabolic (A6)

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 3/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

(B) Linear Tapered, Rectangular Cross-Section

  1 2 2 2

h h h x x h fx h

L

        

 where 1 2h hdh f

dx L

   (B1)

 A x bh ,   3

12 z

bh I x  (B2)

 * , 2

h A x y b y

    

  ,  

2 2,

2 4 z

b h Q x y y

    

  (B3)

Case B1: axial tension ~   0P x P &   0zM x 

  0 0

2 ,xy

P f P f y x y y

bh A h    ~ linear (B4)

Case B2: end vertical force ~   0P x  &    0zM x V L x 

      

 

2 2 2 20

4

2 2

0

3 , 4 12

2

3 2 2 1 1 3

2

xy

V x y h h y f L x h y

bh

f L xV y y

A h h h

        

                   

           

~ parabolic (B5)

Case B3: end moment ~   0P x  &   0zM x M

    2

2 20 0

4

3 3 2 , 12 1 3

2 2 xy

M f M f y x y h y

bh Ah h 

          

   

~ parabolic (B6)

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 4/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 5/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

SUPPLEMENT TRANSVERSE SHEAR STRESSES IN TAPERED BEAMS PAGE 6/6

ME 44100: ADVANCED STRESS ANALYSIS ME I4200: APPLIED STRESS ANALYSIS

CHAPTER 5 –BENDING OF BEAMS

Reversed Tapered Beam

__MACOSX/stress/._Transverse Shear Stresses in Tapered Beams (2017-12-06)(1).pdf