one page

894 IEEE Transactions on Power Delivery. Vol. 9, No. 2, April 1994

Sanjeev Gupta Tri State Engg. New Martinsville WV 26155

STRUCTURAL FAILURE ANALYSIS OF 345 KV TRANSMISSION LINE

Teny J. Wipf Fouad Fanous Mardith Baenziger Yang H. Hahm Midwest Power Des Moines, IA

Civil and Construction Engineering Department Iowa State University, Ames, IA 50011

Kewords - Structural, Transmission line, Failure, Finite element, Analysis, Static, Dynamic, Buckling, Galloping, Ice.

Abstract - An ice storm in Central Iowa on March 7, 1990, caused the domino failure of 69 H frame steel pole structures that were part of a 345 KV transmission line covering a distance of approximately 30.6 km (19 mi). This paper briefly documents the damage, and presents the nonlinear structural analyses performed on a portion of the above transmission line using the ETADS software, which is a structural finite element submodule of TLWorkstation software 111. T h e analysis included a characterization of the likely storm loads and their application to the transmission systems. The methodology used in trying to simulate the sequence of events leading to the failure is also presented.

INTRODUCTION

On March 7th. 1990, the southwest and central parts of Iowa experienced a severe ice storm. As a result of this event, a portion of the Lehigh-Sycamore 345 KV transmission line was damaged, causing the structural failure of 69 transmission structures. Fig. 1 shows the general layout of the damaged Lehigh-Sycamore transmission line. Only Structure-99 was left standing between Structures-51 to 119. The domino failure pattern spread toward the south on the south side of Structure-99 and toward the north on the north side of Structure-99.

The weather conditions that led to this event included heavy rainfall under freezing temperatures. This contributed to heavy ice formations on the conductors,

93 SM 441-6 PWRD A paper recommended and approved by the IEEE Transmission and Distribution Committee of the IEEE Power Engineering Society for presenta- tion at the IEEL/PES 1993 Summer Meeting, Vancouver, B.C., Canada, July 18-22, 1993. mitted September 1, 1992; made available for printing April 5, 1993.

PRINTED IN USA

Manuscript sub-

which along with moderate winds created unfavorable conditions for the line. The amount of radial ice recorded the following morning, 14 hours later and a t a temperature of 4.4O C (400 F), was between 32 and 38 mm (1.25 and 1.5 in). The average wind speed during the event was 5.4 m/s (12.1 mi/h) with a peak wind gust of 13.9 m/s (31 m a ) . Initial conditions of conductor galloping was observed by an experienced crew person during the storm.

This paper presents the results of an investigation which included documentation of the structural failures, and structural analyses to determine the most likely cause of the failure.

FAILURE DESCRIPTION

The structures damaged in the storm consisted of H frame steel pole structures ranging in height from 23 to 40 m (75 to 130 ft). The conductor and shield wire spans ranged from 267 to 472 m (875 to 1550 ft) and contained three bundled conductors corresponding to three phases of electric current and two shield wires. The bundled conductor for each phase had two conducting wires. The conductors were of the type 795 kcmil 26/7 ACSR (DRAKE) while the shield wires belonged to the category

Except Structure-99, Structures-100 through 119 had collapsed toward the North, while Structures-51 through 98 collapsed toward the South. Structure-99 was a suspension angle type structure, and Structure-100 was a tangent dead end type structure. After the failure, the conductors were lying on the ground and had separated from the failed structures a t several locations. The structural damage ranged from minor cracks observed at Structure-99 to complete collapse of the other failed structures. The failed structures showed minimal to significant torsionaldistortion. In addition, noticeable distress such as tension failures and birdcaging of the conductors and shield wires; separation of the conductors from the insulators and separation of the shield wires from the static masts; separation of the insulators from the structures a t the inboard and outboard arms; breaking of the insulators into pieces and shattering of the insulator glass bells, buckling and tearing of the bracing in one or two levels; and dragging of the crossarm and other components such as the insulators after they hit the ground, were observed throughdut t h e line a t several locations.

of 3/8”-7 WIRE EHS STEEL.

0885-8977/94/$04.00 @ 1993 IEEE

NORTH WOOD WARD A

STRUCTURE

THREE WOOD

895

O n the basis of the domino failure pattern of all the the Structure-% before the failure of the transmission h e

GRIMES A - POLKCOUNTY

THREE MADRID WOOD

A POLES

BOONE COUNTY-

Fig. 1. General layout of the damaged porbon of Lehigh-Sycamore transmission line.

structures, it was suspected that the failure initiated in the region of Structures-98, 99 and 100. Fig. 2 illustrates a detailed post-failure plan view of this region. It is worth noting the f i i a l position of the eastern conductor between Structures-98 and 100 which suggests that it separated from

was initiated. Fig. 3 shows Structure-100 after the failure. A more detailed post-failure description of this structure along with other failed structures in the line, is documented in [2].

NOTE: NOT TO SCALE NORTH c

0.5m (1.6f-t) CURLED UP (75 ft) INSUMTOR

"O GLASS 1 PIECES i-- .-.____ --Cl--.- -.-. 25 m (82 ft)

' BIRDCAGING OF b ,m CONDUCTOR

I I

I

A = 4 8 f}-.-.-.-.-,-'-.-'-'

EASTERN STATC MAST STRUCTURE-98

STRUCTURE-100 L,-.-. 7 A = 2

___._._.-.- \ -

EASTERN \- EASTERN CONDUCTOR w- 4 5 m (47.5 ft)

LEGEND

STRUCTURE CROSSARM BEFORE FAILURE I CONNECTION ASSEMBLY

CONDUCTORS - - - - - . STFNJCTURE CROSSARM AFTER FAILURE CONDUCTOR CONTINUATION AS SHOWN - .- -. - .- TO ACCOUNT FOR DISTORTED SCALE

A ANGLE OF TWIST, DEGREE

Fig. 2. Plan view of area near Structure-98,99 and 100 after transmission line failure.

896

Fig. 3. Structure-I 00 after the transmission line failure.

FINITE ELEMENT MODELING OF THE TRANSMISSION LINE

The structural analysis was limited to a portion in the vicinity of Structure-99, where the failure was speculated to have initiated based on the field observations. Fig. 4 illustrates the three dimensional finite element model which consisted of 245 nodes and 245 structural elements of various types. Five transmission structures along with the conductor and shield wire spans were included in the model. The structures were assumed to have fmed base supports at the ground level, though in reality, only Structure-99 was supported by concrete foundation while the other four structures were embedded approximately 4 to 7 m (13 to 23 ft) into the ground. The two end spans of conductors and shield wires on either side of the model,

were assumed to have their far ends as shown in Fig. 4. The bracings of the structures were assumed to carry only axial loads. A rigid joint was assumed at the connection of the inboard arm, outboard arm, static mast and the main pole of the structure. The insulator assembly was modeled as a catenary cable of uniform cross-section, with the breaking strength being defmed as the minimum ultimate capacity of the hardware components of which the assembly was composed.

A three dimensional beam element with polygonal cross-section and elastic-plastic material properties was selected from ETADS [l] software to model the poles, static masts, outboard and inboard arms. The element has 8 sides and a constant taper along its length. On an average, about 28 such elements were used to model the single structure.

A three dimensional catenary cable element of ETADS [I] software was used to model the conductors and the shield wires. This element can be used to simulate the characteristics of such flexible members since it responds to loads by deforming into a shape that permits the external loads to be resisted by an internal tensile force. A single element was used to model one span length of a conductor or shield wire. The design stringing tension in the conductors and shield wires were 27 kN (6050 Ib) and 8.9 IrN (2OOO Ib) respectively. Moreover, the standard material and section properties of 795 kcmil 26/7 ACSR (DRAKE) conductor and 3/8-7 WIRE EHS STEEL shield wire were used [3].

NONLINEARITY IN THE ANALYSIS

The material as well as geometric nonlinear behavior of the structures was considered in the structural analyses. A bilinear elastic-plastic material constitutive relationship, and the large displacement option of ETADS 11) software were used. The ETADS results were allowed to converge

5-STRUCTURE COMPUTER MODEL LEGEND 1 SHIELD WIRE CONDUCTOR IT INSULATOR 1 - - - - - - -

Y J cmi ipn IRE-oo

Fig. 4. Three dimensional finite element computer model used for ETADS analysis.

such that the residual nodal forces were less than 1.0% of the applied nodal forces. Hence, the loads were applied in several load increments, and enough number of iterations were allowed within each load increment until a converged solution was reached.

FAILURE CRITERIA

For the analysis, the insulators, conductors and shield wires were assumed to have failed when the axial tensile forces in them exceeded their rated strength [3]. For the structure members (e.g., static masts, outboard & inboard arms and the main poles), material yielding was considered indicative of failure. In addition, buckling of the main pole of the structure was the other failure condition considered in this study.

STRUCTURAL ANALYSIS

Based on the evidence and the information gathered from the failure site, several load cases were developed for the structural analysis. The type of loading considered included static ice loads on the conductors and shield wires, dynamic broken insulator loads and conductor galloping loads.

static Analysis This analysis was carried out to estimate the radial ice

thickness on the conductors and shield wires that could have initiated the transmission line collapse. In addition, the most stressed component was to be identified to determine if the failure could have been initiated due to the breaking of a hardware component or because of the local failure of the outboard arm, inboard arm, static mast or the main pole of the structures.

The finite element model was subjected to gravity loads, stringing tensions, and gradually increasing ice loads on conductors and shield wires. These loads were applied in two loadsteps; the first loadstep included only the gravity and stringing tension loads. This yielded information regarding the forces and stresses induced in the system prior to the icing e v e n t In the second loadstep, 51 mm (2.0 in) of radial ice on the conductors and shield wires was also included. This was applied in several small load increments in order to ensure converged solution within each increment.

The results of the static analysis showed that the forces in the conductors and the shield wires were below their rated strength (140 kN (31.5 kips) for conductors and 68.5 kN (15.4 kips) for the shield wires) for loads up to 38 mm (1.50 in) of radial ice. At 44 mm (1.75 in) of radial ice, the tension in all the conductors was still below their rated strength. However, the tension in the shield wires was close to or exceeded the rated strength a t a few locations. At 51 mm (2.00 in) of radial ice, both the conductors and the shield wires were stressed close to or beyond their rated

897

strength. Insulators at Structures-97, 98, 99, 100 and 101 had

ultimate loadcarrying capacities of SO, SO, 160,222 and SO kN (18, 18,36,50 and 18 kips), respectively. The results of static analysis also showed that for up to 38 mm (1.5 in) of radial ice, the forces in all of the insulators of all the structures were below their ultimate loadcarrying capacity. However, as the radial ice thickness reached 44 mm (1.75 in), the forces in the insulators of Structures-97, 9 8 , s and 100 slightly exceeded their ultimate capacity. At 51 mm (2.0 in) of radial ice, the forces in the insulators of all the structures were significantly above their ultimate capacity. These results demonstrate that between 35 mm (1.5 in) and 44 mm (1.75 in) of radial ice possible failure could have been initiated in the insulators a t Structures-97, 98,99 and 100.

The ETADS software also calculates Von Mises effective stress, 0,. The value of the effective stress a t any point., obtained from the ETADS analysis, was compared with the material yield stress, 0,. , a t various levels of ice load. The analysis results showed that for Structure-9t3, no yielding took place at 25 mm (1.0 in) of radial ice; however, at 32 mm (1.25 in) of radial ice, yielding started in the upper portion of the main poles and in the two outboard arms of the structure. Furthermore, yielding became more concentrated, especially in the outboard arms of Structure- 98 a t 38 mm (1.50 in) of radial ice. The results illustrated similar behavior a t Structure-% as with Structure-98. In Structures-97, 101, and 100 the yielding started at 38 mm (1.50 in) of radial ice and was primarily concentrated in the outboard arms.

The results summarized above illustrate that failure in the transmission line could have initiated for radial ice thickness between 35 and 44 mm (1.5 and 1.75 in) because of the breaking of any of the insulators at Structures-97,98, 99 and 100. However, it is possible that the failure could have been initiated at smaller radial ice thickness if the effect due to galloping of the conductors was considered. Anjam [4] conducted a separate study of conductor galloping loads on the Lehigh-Sycamore transmission line. He concluded that at 35 mm (1.5 in) radial ice along with dynamic galloping effects, loads produced were similar in magnitude as that for 44 mm (1.75 in) radial ice without considering the galloping effects. Thus, one can reasonably assume that with 35 mm (1.5 in) of radial ice, there is suficient potential for initiating the failure. Hence, it was concluded that the transmission line collapse might have initiated due to the breaking of eastern insulator at Structure-99 for 35 mm (1.5 in) of radial ice.

Dvnamic Broken Insulator Analysis

To study the forces and stresses induced in the system due to the sudden breaking of the eastern insulator of Structure-99, a dynamic broken insulator analysis was performed. T o simulate a broken insulator effect of a transmission line, the previously described computer model was modified using the Fuse element available in the

898 ,* - - _ _ _ - \ I \ . , -v-l---- - - - - - - - - _ _ _ _ _ - -

DEAD END INSULATOR ASSEMBLY

“ 1 STRUCTURE-I 00

/ _ - - SUSPENSION INSULATOR ASSEMBLY

“ I OTHER STRUCTURES

“ 1 STRUCTU RE-99

Fig. 5. Dead end and suspension type insulator assemblies.

ETADS [l]. This element has only axial force carrying capability, and its stiffness can be set by the user to be zero at any time during the analysis.

The loads on the modified model were applied in two loadsteps. In the first step, 38 mm (1.5 in) of radial ice was applied statically on the conductors and shield wires and the solution was carried out until convergence was attained. In the second loadstep, the stiffnesses of all the Fuse elements were set to zero and a dynamic analysis was carried out. Both geometric and material nonlinearities were included in the analysis. The fundamental period of vibration of the structures was calculated as 2.5 sec. Hence, in order to minimize the effect of numerical damping, an integration time step size was limited to 1/50 of this period.

The static analysis of the first loadstep was successfully completed, however, immediately after the Fuse elements were broken, the analysis was terminated because of numerical problems. The ETADS program indicated that local instability had occurred a t several locations in the system. The f m t loadstep results showed that longitudinal force imbalances of 200 kN (45.0 kips) existed at Structure- 98 (in the direction of Structure-97) and at Structure-100 (in the direction of Structure-101). It was concluded that sudden release of these forces on to the system following the insulator break at Structure-99 might be beyond the load carrying capacity of Structures-98 and 100. Next, a buckling analysis of these structures was performed to further supplement the above conclusion.

Buckling Analvsis of Structures-98 and 100 A buckling analysis for individual structures (98 and

100) was performed by applying a horizontal force a t the eastern insulator attachment point, The magnitude of this force was increased until the buckling load was reached. No force was applied a t the middle and western conductor attachment points on the structure since these conductors were intact on both sides of the structure and therefore no force imbalance was created a t these points.

The analysis results showed that a force imbalance of 29 kN (6.50 kips) in Structure-98 was sufficient to cause instability and hence buckling failure of the structure. Similarly the horizontal force imbalance for Structure-100

was evaluated as 67 kN (15.00 kips). From the above results it was concluded that the force imbalance of 200 kN (45.0 kips), to which the Structures-98 and 100 were subjected after the eastern insulator failure at Structure-99, was considerably beyond the buckling strength of these two structures. Thus this force imbalance could easily result in the buckling failure of Structures-98 and 100.

Galtouina Analvsis of Structure-100 A dynamic and buckling analysis for Structure-100

(using a single structure model) was performed in order to investigate the possibility of transmission line failure under conductor galloping loads. As mentioned earlier in the paper, galloping behavior had been observed during the storm. The analysis was performed for Structure-100 only since it was a dead end for the conductors. Hence, it had almost horizontal dead end insulator assemblies on both sides as compared to other structures that contained either vertical or inclined suspension insulator assemblies (see Fig. 5). Note that Structure-99 has inclined insulators since it was located at an angle change (see Figs. 1 and 2). The suspension insulator assemblies can swing under the horizontal force imbalance, thereby, reducing the magnitude of galloping loads transferred directly to the structure. Thus, Structure-100 was most susceptible to horizontal force imbalance created by conductor galloping loads.

A software using an algorithm developed by Baenziger [5,6], and modified by Li Li [7] and Anjam [4], was used to idealize the galloping loads for Structure-100 a t 38 mm (1.5 in) of radial ice. These galloping loads were expressed as net horizontal and vertical time varying forces at the insulator attachment points, assuming that the conductors on one side of Structure-100 gallop with no conductor on the other side (Loadcase 1); conductors in adjacent spans of Structure-100 galloping out of phase (Loadcase 2) and conductors in adjacent spans of Structure-100 galloping in phase (Loadcase 3).

The condition in which the conductors gallop only on one side of the structure might arise under rare circumstances such as uneven ice deposition on adjacent spans, drastic differences in adjacent span characteristics, or wind approaching the conductors in an oblique direction.

899

stressed a t 0.1 sec., with material yielding occurring mainly in the two outboard arms of the structure.

A buckling analysis of Structure-100 was also performed to investigate if these galloping loads could cause the buckling failure of Structure-100. Peak values of horizontal and vertical time varying galloping forces were used as input for the buckling analysis. The analysis results showed that for out-of-phase galloping (Loadcase 2), 7 0 8 of the peak values of galloping loads was sufficient to cause instability and hence buckling failure of the structure. On the other hand, for in-phase galloping (Loadcase 3), the analysis showed that the buckling failure of the structure would not occur.

The dynamic analysis for galloping loads indicated that substantial damage in the form of material yielding could be induced in Structure-100. In addition, the buckling analysis under galloping loads indicated that the structure could buckle because of out-of-phase galloping loads. These results show that if a t 38 mm (1.5 in) of radial ice the conductors in adjacent spans of Structure-100 had galloped out of phase, it is possible that the buckling failure of Structure-100 could have initiated the transmission line failure.

I \ 1

STRUCTURE-I 00

Fig. 6. Galloping Loadcases 2 and 3: Resultant horizontal & vertical forces at Structure-I 00.

Under normal conditions of ice and wind, the probability of occurrence of Loadcase 1 is very low and hence Loadcases 2 and 3 were selected for the galloping analysis of Structure-100. The resultant horizontal and vertical galloping force-time curves (at the insulator attachment points) for Loadcases 2 and 3 were sinusoidal in nature. For Loadcase 2 , the resultant horizontal force had a maximum and minimum values of 15 kN (3.4 kips) and -14 kN (-3.2 kips); and the resultant vertical force had a maximum and minimum values of 34 kN (7.6 kips) and 32 kN (7.2 kips). Similarly, for Loadcase 3, the corresponding figures for the resultant horizontal force were 1.3 kN (0.3 kips) and -0.66 kN (-0.15 kips); and for the resultant vertical force were 44.5 kN (10.0 kips) and 22.2 kN (5.0 kips).

A dynamic analysis for one full cycle of the sinusoidal galloping loads (duration of 6.0 sec.) was performed by applying the load a t three conductor locations as illustrated in Fig. 6. Both geometric and material nonlinearities were included in the analysis.

For Loadcase 3, i.e., in-phase galloping, the analysis was successfully completed up to a time of 6.0 sec. However, for Loadcase 2 , i.e., out-of-phase galloping, the analysis was completed only up to 0.7 sec., showing that excessive material yielding had taken place at various locations in the structure. Since, material yielding was considered as an indication of failure, it was not necessary to continue the analysis beyond this time.

The stresses induced in various members of Structure- 100 because of the out-of-phase galloping loads (for 38 mm (1.5 in) of radial ice) showed that the structure was critically stressed a t times of 0.1 sec. and at 0.7 sec. Results showed that a t 0.7 sec., the material yielding occurred in the two main poles of the structure, while at 0.1 sec. yielding had taken place mainly in the two outboard arms of the structure.

The stresses induced in various members of Structure- 100 because of the in-phase galloping loads (for 35 mm (1.5 in) of radial ice) showed that the structure was critically

CONCLUSIONS

The work summarized in this paper was devoted to analyzing a portion of the Lehigh-Sycamore 345 KV transmission line located in t h e central part of the state of Iowa. Failure of this line resulted from the ice loads that were experienced during an ice storm that swept through the state of Iowa on March 7, 1990. The objective of the summarized work was to quantlfy potential loading leading to different failure possibilities. Two potential initiators of the observed failure have been investigated based on physical evidence a t the site, eyewitness accounts and observed weather conditions. The forces required to cause failure based on these conditions have been quantified. This study presents information to enhance the analysis of future failures and to provide important information for designers.

ACKNOWLEDGMENTS

The study presented in this paper was based upon research conducted for the utility owners of the 345 KV LehighSycamore Transmission Line in Central Iowa. Special thanks are accorded to personnel a t Iowa Power Co. who served as the contact for obtaining the necessary information and to the utility owners who provided useful commenrs and directions during the conduct of the study.

The authors extend sincere appreciation to Mr. Tip Goodwin and Mr. Vichien Nopratvarakom of Sverdrup Technology, Inc. for the technical assistance provided related to ETADS (TLWorkstation) software.

REFERENCES

Electric Power Research Institute's TL Workstation Sofrware, module E T m S , Palo Alto, California, July 1990. T. Wipf, M. Baenziger, F. Fanous, S. Gupta, and R. Anjam, Ice Storm Damag Assessment o f the &high- Sycamore 345 KV Transmhsion Line, tinal report submitted to Iowa Power Company, Ames, Iowa, July 1991. E. S. Doocy, A. R. Hard, C. B. Rawlins, and R. Ikegami, Transmission Line Reference Book, EPRI, Palo Alto, California, 1979. R. Anjam, Galloping and Broken Conductor Anal'h of Transmhsion Lines, Thesis, Iowa State University, Ames, Iowa, 1991. M. B. Thomas, Broken Conductor Loads on the Transmhsion Line Structures, Dissertation, University of Wisconsin-Madison, Madison, Wisconsin, 1981. M. B. Thomas and A. H. Peyrot, "Dynamic Response of Ruptured Conductors in Transmission Lines," IEEE, vol. PAS 101, September 1952. Li Li, Dynamic Loads on Pole Transmhsion Line Structures from Galloping Conductors, Thesis, Iowa State University, Ames, Iowa, 1990.

BIOGRAPHY

Sanieev GUDU was born in New Delhi, India, on February 25, 1965. H e received a B. Tech. degree in Civil Engineering from Indian Institute of Technology, New Delhi, India, and a M.S. degree in Structural Engineering from Iowa State University (ISU), Ames, Iowa, in 19S9 and 1991, respectively.

As a graduate student at ISU, he was involved in projects dealing in structural analysis using finite element method. Since April, 1992, h e has been with Tri State Engineering & MC, Inc., working in the area of structural analysis and design.

Mr. Gupta is a member of the American Society for Civil Engineers (ASCE). He holds an "Engineer in Training" certificate in the state of Iowa.

Term J. WiDf received B.S. and M.S. degrees in Civil Engineering from the University of Nebraska-Lincoln, and a Ph.D. degree in Engineering Mechanics from the University of Nebraska-Lincoln.

Dr. Wipf joined the faculty at Iowa State University in Structural Engineering in 1953 where he is now an Associate Professor of Civil Engineering. H e has directed several projects in the general area of transmission line structures and structural failure analysis.

He is a registered professional engineer and a member of ASCE, ACI and ASEE.

Fouad Fanous received a B.S. from Cairo University, Egypt, in 1%9, M.S. and Ph.D. from Iowa State University in 1950 and 1982, respectively. Since then, he has been teaching and conducting research in the areas of reliability analysis of nuclear containment buildings under severe accident loading, strengthening, life study, and management of bridge structures. H e has published numerous papers related to his work

Dr. Fanous is a member of ASCE, ACI, ASEE and is a registered professional engineer in the state of Iowa and Emt.

Mardith Baenziger was born in 1945 in Iowa. She received a Bachelor of Architectural Engineering in 1968, a M.S. degree in Nuclear Engineering in 1%9 from Iowa State University; a M.S. in 1 9 9 and a Ph.D. in 1981 from University of Wisconsin.

Dr. Baenziger worked as a consulting engineer for a firm in Akron, Ohio from 1969 to 1975. She joined the Iowa State University, Department of Civil and Construction Engineering in 1951; she is currently a n associate professor there.

She is a member of the American Society for Civil Engineers, American Society for Engineering Education, Society of Women Engineers, Sigma Xi, Tau Beta Pi and Chi Epsilon. She received the Iowa Governor's Science Medal for science teaching in 1991. Dr. Baenziger is a registered professional engineer.

Yane H. Hahm received a BSEE from Iowa State University in 1959, and has been actively engaged in the electric power industry as an engineer with consulting engineering companies and later with a utility company. H e has had various capacities in the areas of transmission, substation and construction engineering. H e became a Principal Engineer in 19%. H e is now a Principal Engineer in Substation, Transmission and Distribution Services of Midwest Power, a division of Midwest Power Systems Inc.

Mr. Hahm is a member of I E E E P E S , and is a registered professional engineer in the field of electrical engineering in the state of Iowa.

90 1

DISCUSSION H. BRIAN WHITE, Transmission Line Consultant. Hudson, Quebec,Canada. It is truly a delight to get a report, such as this, of a significant line failure in which all parties, the owner utility as well as consultants and advisers all co-operate to present their best combined effort at explaining what happened. Improvements in line design practices are needed in several areas and proaress will only come when many of us are given the chance to understand what leads to some of these too frequent and sometimes embarrassing events. The authors have made a commendable effort to reach the conclusion, with which we agree, that the failure of the east conductor support at structure 99 triggered the collapse of 98 and a lengthy cascade resulted in each direction from 99, which remained standing. The cascades themselves are not a proper subject for discussion at this time for this paper confines itself to the question of finding the trigger event which was somewhere in the section 98 to 100 as shown in Fiq.2. However F i g 2 does not contain many of the important data needed to understand what happened. We found need for and obtained separately span lengths, wire tensions. tower heights and specially the line angle at 99 so that we could calculate for ourselves what we believe happened in this area. A key issue throughout is the thickness of ice deposited on the wires in the March I storm and in this matter we note the fact that after countless icing events, (usually associated with failures 1 , there is rarely a good measure of the actual ice deposit. That is specially true in this event when so much effort had to be spent in working backwards from the component strengths to try to determine what ice thickness o r ice load existed at the time of the failure. Precipitation icing is seldom circular and measured thicknesses are almost always suspect as they are taken at the thickest and most dramatic point. Such thickness measurements are not a good indication of the unit ice load. A s wind on ice loadings are of secondarv interest to a line designer, we should focus on the vertical loading of ice and this can best be found by taking a 12" or any specific length of ice on the wire, breakine o r scraping it off and putting it in a bottle o r plastic bag for subsequent weighing. It is the load o r weight of ice that is of importance. The evidence left by an ice storm is so much easier to measure than that of a severe wind event but the icing evidence is seldom measured and recorded in a precise and useful manner, We have several reasons to believe that the equivalent radial ice thickness did not exceed 1.25 " at the time that the failure was triggered and the most compelling piece of evidence is the author's own statement on the fourth page that yielding of the upper poles and outer crossarms of 98 and similar structures will take place at 1.25" of radiai ice. The authors themselves state that material yielding should be considered indicative of structural failure.

Thus we take this 1.25" as an upper limit of icing at least at the time of the collapse. The evidence also shows that structure 98 did not fail by yielding of upper masts or of the outer crossarms. The authors have assured themselves. and we concur in this, that no other line components were close to being critically loaded at that thickness. It becomes necessary to conceive a scenario that will instigate a failure at o r below about 1.25" of ice. The direction of the wind that accompanied the ice storm could incite galloping in some spans from 83 to 99 but maybe not from 9 9 towards 100 because of the fairly large line angle (almost 3 0 degrees) at Y 9 . kind direction relative to a line section will account for some of the vagaries of the galloping phenomena. If the east phase of 98 to 99 started to gallop, we could have the same mechanism that was deduced at structure 121 of the M G and E 345 kV line in 1974 in which the top 1 5 ' o r so of a guyed mast running angle structure was broken off and flung several hundred feet into the line angle. The static load in the mast at the time of failure was about 20% of capacity and the mechanism of the system did the rest. This mechanism was described in IEEE paper No. A 79 106-6 by this discussor. A running angle suspension can turn the vertical galloping forces at the clamp into very complex motions of the conductors on both adjacent spans and result in a resonant beat of large magnitude. Thus galloping adjacent to the runnine angle suspension at 99 could lead to rather violent shock loads to the suspension strlnes. violent enough to break the assembly and cause the conductor to be flune 2 8 5 ' into t h e angle. (It will also be found in CIGRE paper 22-04 of 1974 that the Japanese authors and investigators found load fluctuations of UP to 2:l with galloping wires at both suspension and dead end positions: dynamic loads that could start failures at far below the ice thicknesses discussed in t.his current paper. ) Accepting the separation of the East phase from 99, the progress of the collapse of 98 and 100 and the cascadings readily follow. The dropping of this east phase at 99 will introduce about 90' of slack into that phase from 98 to 100 and the cross arm of 98 will then rotate to the south, as not.ed in F i q . 2 , causing complete collapse of the structure. The collapse of 98 will cause a laree increase in tension of the two still intact phases between 91 and 99, an increase which will pass with no trouble through 99 to the strain assemblies fitted to the suspension structure 100. There the center phase insulator assembly broke under the increasea tension and then the conductor pulled hack through 99 where the aluminum was stripped (birdcaging 1 . We work partly through this scenarlo to demonstrate that only a few elementary wire rope principles are needed to understand what happened and why the cascades started. All can be explained with an understanding of slack of which there was only about 4 to 6' in each span before the failure started. With the release of the east phase at 99 and the introduction of about YO' of slack, there

902

Structures

Span Length (ft) (m)

would be almost complete relaxation and verv large outward p u l l s . However f o r the intact phases 97 to 99, the dropping of the mid structure 98 would lead to an almost douhllne' of tension (depends on the amount that reaches the groundJ and the initial loadlne: on adjacent structures would be inward until something gave, as it did at l U 0 . -It this point there was now too much s l a c k in all phases and the cascading went outwards in both directions. Cascades start when too mlich s l a c k is introduced to the wire svstem and structures are not adequate t o r the resultlne longitudinal loads. Because we seldom have precise knowl-dqe o r the ice load in such events, there is little need for more than first order analvsis n t the failure scenario. I However careful analvsis must be made n f th- qtrenc+h n t critical components.) Thus our main disappointment in this v q n e r has t o do with the great reliance on computer programs to obtain understanding O T relatively simple situations. Primitive appraisals could be ~ustifi~d onpp + h - limiting capacitv of the suspension towprs was established at about o r even le=? than 1.25" radial ice and all wire load calculations and the gallopinq analvses iind-r thicker ice would seem to be but theoretical exercises. Furthermore with so much time and ettort spent in trying to apply the programs.the insight1 s I that might prevent recurrence ot a similar event was not uncovered and little attention was directed t o the prevention or cascades, bv far the more important issup to be settled. In our opinion the kev issue was t h e separation of the insulator assemblv at the east phase o t 99 at far below its itatir strength rating and the contribution that thr running angle torm of suspension I incliner! string) may have made to the problem. O u r experiences and studies leads us r o he cautious in inserting these running angle suspensions at line angles o f more than ahoiit 7 degrees in conditions conducive to galloping- 7 degrees being the line angle a t which slack introduced bv failure of the suspension would not introduce large ~ n w a r d o r outward forces on adJacent towers. This anqle is dependent on span lengths. We find more points in this paper with which to agree than to differ. The cerv p o s ~ t i v p consideration is that it is only with papers such as this one will we gain from the experiences of others and learn to develop forensic and analvsis techniques that will help us understand 'what really happened'.

97-98 98-99 99-100 100-101

1348 1305 1230 1175 411 398 375 358

Manuscript received August 2, 1993

RON CARRINGTON, Chief Transmission Engineer, POWER Engineers, Inc., Hailey, Idaho 83333:

This interesting report of the analysis of a line failure raises several questions. Radial ice of 1.25 to 1.5" was recorded 14 hours after the event.

How was this ice measured, what was the density and is there any assurance that the ice loading was equivalent to that of a continuous symmetrical cylinder of such thickness? We have found that the shield wires often play an important role in the trigger event and sometimes a very significant role in a subsequent cascading. It would be instructive if the position of the shield wires and associated hardware were noted in the area of structures 98 to 100.

Manuscript received August 10, 1993.

SANJEEV GUPTA, TERRY J. WIPF, FOUAD FANOUS, MARDITH BAENZIGER & YANG H. HAHM:

The authors extend their sincere appreciation to the discusser, H. Brian White, for his interest in the subject of the paper and thank him for the time and efforts he took to share his own valuable experiences and knowledge related to the subject. The paper was written with a view to share the failure event with the readers, the investigative efforts and the results obtained through a detailed structural analysis. The paper also presented the application of existing state-of-art computer software in the structural analysis of transmission lines.

A range of geometrical data/ properties of the structures are mentioned in the text of the paper rather than specific values for each structures in Fig. 2 (refer to page 1 of the paper). The stringing tensions in the conductors and shield wires are mentioned on page 3 of the paper. In response to discusser's comment, the authors present additional information in Table I below:

Table I. Geometrical data for the failed transmission line.

Structure I 9 7 I 98 I99 I 100' I 101 Structure Height**

Structure-99 '* From ground to cross arm. * Also the conductor elevation.

As mentioned in the paper, the radial ice thickness of 32 to 38 mm (1.25 to 1.50 in) was recorded in the field. It should be noted that this reading was taken 14 hours after the failure and at a temperature of 4.4' C (40" F). Hence, the possibility of radial ice being more than 1.5 inches at the time of failure can not be completely ruled out. The structural analysis was carried out for various ice thicknesses assuming a continuous symmetrical cylinder of ice around the conductors and shield wires. The authors agree with the approach of weighing a specific length of ice immediately after the failure to get a reasonable measure of the ice density.

903

The static analysis revealed that the material yielding for Structure-98 started in the upper portions of the main poles and the two outboard arms of the structure at 32 mm (1.25 in) of radial ice. However, this yielding was very localized and progressed across the cross sectional area of the members as the ice load was increased. It should be noted that at Structures-97, 100 and 101, material yielding in fact started at 38 mm (1.5 in) of radial ice load and was primarily concentrated in the outboard arms. The analysis did not show any gross yielding of any section. Hence, a failure could not be assumed. Although material yielding may be considered as indicative of structural failure, the debatable question still remains as to what percentage of cross sectional area of a member should yield before it can be considered to have failed.

The failure scenario presented by the discusser, which is based on the first order analysis using few elementary wire rope principles, seems acceptable. In fact, similar scenarios were developed by the authors and the utility owners the very next day after the failure. However, an objective of this study was to gain additional insight into the possible failure scenarios through a detailed structural analysis of the transmission line system. The authors used the ETADS software to analyze the transmission line considering different probable load cases to calculate the resulting forces and stresses induced in the system. In addition, the purpose of the paper was to summarize the

methodology and the procedure followed during the structural failure analysis of the Lehigh-Sycamore transmission line using ETADS software. The authors hope that the analysis results reported in the paper present a useful source of information for carrying out further research involving issues such as "preventing the recurrence of similar events" and "prevention of cascades".

The authors are equally appreciative of the time and interest taken by Ron Carrington in the paper and thank him for sharing his views. The radial ice thickness was measured by Midwest Power Engineering personnel in the field using a measuring tape. The ice deposited on the conductors and shield wires was almost symmetrically radial.

The authors agree with the discusser's statement about the importance of shield wires in triggering the cascading event described in the paper. The shield wires were included in the finite element model and were also subjected to ice loading. The possible role of the shield wire in the failure event was also studied during field documentation subsequent to transmission line failure. Neither the analysis results nor the field evidence suggested that the transmission line cascading could have been triggered by the failure of shield wires or associated hardware.

Manuscript received October 1, 1993.