Accident Investigation Techniques

SystemsModelofConstructionAccidentCausation.pdf

Home >Science homework help >Accident Investigation Techniques

arriers that d due to lack m factors these causal

ent causation idents—it analyzes the haviors—it uld” follow. edges the g he new requency of ts.

Systems Model of Construction Accident Causation Panagiotis Mitropoulos1; Tariq S. Abdelhamid2; and Gregory A. Howell3

Abstract: The current approach to safety focuses on prescribing and enforcing “defenses;” that is, physical and procedural b reduce the workers’ exposure to hazards. Under this perspective, accidents occur because the prescribed defenses are violate of safety knowledge and/or commitment. This perspective has a limited view of accident causality, as it ignores the work syste and their interactions that generate the hazardous situations and shape the work behaviors. Understanding and addressing factors that lead to accidents is necessary to develop effective accident prevention strategies. This paper presents a new accid model of the factors affecting the likelihood of accidents during a construction activity. The model takes a systems view of acc focuses on how the characteristics of the production system generate hazardous situations and shape the work behaviors, and conditions that trigger the release of the hazards. The model is based on descriptive rather than prescriptive models of work be takes into account the actual production behaviors, as opposed to the normative behaviors and procedures that workers “sho The model identifies the critical role of task unpredictability in generating unexpected hazardous situations, and acknowl inevitability of exposures and errors. The model identifies the need for two accident prevention strategies:~1! reliable production plannin to reduce task unpredictability, and~2! error management to increase the workers’ ability to avoid, trap, and mitigate errors. T causation model contributes to safety research by increasing understanding of the production system factors that affect the f accident. The practical benefit of the model is that it provides practitioners with strategies to reduce the likelihood of acciden

DOI: 10.1061/~ASCE!0733-9364~2005!131:7~816!

CE Database subject headings: Occupational safety; Construction site accidents; Accident prevention.

as a sures aised safety injury

the the

d on en-

that f the iors.” li-

, en- t li- ance. acci- ance

h. . It safety

zards of eased

dd table

afety ts are s to

ch as a less

d the ffec- ces in es are

ol of

ment 323.

um,

ssions te by ging pos- 2004. - 05/

Introduction

In recent years, construction accident rates have declined result of substantial effort by many parties. Increased pres from OSHA and owners, and increased cost of accidents r the contractors’ awareness. In turn, contractors increased training and enforcement. These efforts have reduced the and illness rate from 12.2 in 1993 to 7.9 in 2001. However, rate of fatalities has shown little improvement—since 1997, number of fatalities per year is consistently over 1,100~Bureau o Labor Statistics 2004!.

The current approach to accident prevention is base OSHA’s violations approach and focuses on prescribing and forcing “defenses;” that is, physical and procedural barriers reduce the workers’ exposure to hazards. The violations o defenses are called “unsafe conditions” and “unsafe behav This approach emphasizes~a! management commitment and po

1Assistant Professor, Arizona State University, Del E. Webb Scho Construction, Tempe, AZ 85287-0204. E-mail: mitro@asu.edu

2Assistant Professor, 207 Farrall Hall, Construction Manage Program, Michigan State University, East Lansing, MI 48824-1 E-mail: tabdelha@msu.edu

3Executive Director, Lean Construction Institute, Box 1003, Ketch ID 83340. E-mail: ghowell@leanconstruction.org

Note. Discussion open until December 1, 2005. Separate discu must be submitted for individual papers. To extend the closing da one month, a written request must be filed with the ASCE Mana Editor. The manuscript for this paper was submitted for review and sible publication on January 8, 2003; approved on September 2, This paper is part of theJournal of Construction Engineering and Man agement, Vol. 131, No. 7, July 1, 2005. ©ASCE, ISSN 0733-9364/20

7-816–825/$25.00.

816 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT

cies to prevent unsafe conditions and~b! workers’ training and motivation to prevent unsafe behaviors.

Safety programs, such as training, inspections, motivation forcement, penalties, etc., emphasize competency~“competen person” philosophy! and liability, and aim at increasing comp ance with safety rules and increasing the cost of noncompli The violations approach has contributed to the reduction of dent rates, but it also has limitations, as high levels of compli are costly and compliance does not ensure safety~Prichard 2002!. The following are some limitations of the traditional approac

(1) Reactive approach. The violations approach is reactive manages the hazards with defenses and relies on increased effort to reduce accidents. A proactive approach avoids ha ~e.g., by using a less hazardous method!, reduces the safety risk the production system and reduces accidents without incr safety effort.

(2) Conflict with production . The safety effort does not a value to production—it only replaces one type of unaccep loss ~human suffering and financial consequences! with a more acceptable cost. However, compliance requires significant s effort and resources and in the short term, safety requiremen in conflict with production and cost goals. This often lead noncompliance.

(3) Uncertainty limits the effectiveness of defenses. Com- pared to the well-structured, high-risk technical systems, su nuclear and process plants, airplanes, etc., construction is structured and loosely coupled system~Rasmussen 1997!. The ill-structured, dynamic nature of the construction process an large number of poorly defined situational hazards limit the e tiveness of such defenses, as they create many circumstan which the needed defenses are absent or existing defens

bypassed. Furthermore, safety defenses cannot address all types

ot erous

- ck of ives

e does What may

il- vio- f the ucture tion iola- oot

elop- ffec-

y and f the pe th

s and l pre

ased ts to in- era- not

eling and

cting by- s on e the

ls of duc-

that

ation sents scuss new

l .” A

left - ty of

ed in- t, not

s and es for erent ential

gy due

i- ele- s

ho-

causa- - man ory from amid

ing, The

ctions

e in-

tion -

sors, . The

fety s inci- of ided, d, or pre-

con-

of hazards~e.g., ergonomic hazards!, and in some cases cann overcome the legacy of design and the need to work in dang circumstances~e.g., roofwork!.

(4) Limited view of accident causality. The violations per spective attributes accidents to the managers’ or workers’ la safety knowledge and/or motivation. This approach perce safety as a problem of “right versus wrong”~safe versus unsaf! choice, and ignores the fact that the dynamic nature of work not involve conscious decision making or risk assessment. seems to be a rational act under a particular work situation, easily be judged as a unacceptable mistake on hindsight~Rasmus sen 1990!.

(5) Limited learning . The focus on violations limits the ab ity to learn from accidents. Accident investigation focuses on lations and liability and does not increase understanding o accident phenomenon; rather, it perpetuates the current str by assigning blame. An evaluation of 17 accident investiga methodologies used by public agencies found that OSHA’s v tions approach is among the lowest in its ability to identify r causes~Benner 1985!.

Better models of accident causation are essential for dev ing effective accident prevention strategies. We argue that e tive causation models need to take a systems view of safet provide better understanding of how the characteristics o production system generate hazardous situations and sha work behaviors.

This paper develops a new causation model of the factor processes that generate construction accidents. The mode sented here, has the following attributes: • It focuses on the activity level, as opposed to event-b

models that focus on the incident level. This model attemp answer the question: “What causal factors and processes fluence the number of accidents during a construction op tion?” It is a conceptual model and at this stage it does operationalize or quantify the variables.

• It takes a systems view of accidents—it is a causal mod approach that moves away from looking at isolated events looks at the production as a system made up of intera variables~Sterman 2000!. Thus, accidents are viewed as products of the production system and the model focuse the characteristics of the production system that generat risks and shape the work behaviors.

• It is based on descriptive rather than prescriptive mode work behavior. That is, it takes into account the actual pro tion behaviors, as opposed to the normative behaviors workers “should” follow. The model is based on previous research in accident caus

human error, and construction safety. The next section pre the background literature. We then present the model and di its implications. Based on the model, the paper proposes directions for accident prevention.

Background Literature

Definitions

The National Safety Council~NSC! definessafetyas “the contro of recognized hazards to attain an acceptable level of risk hazard is defined as “an unsafe condition or activity that, if uncontrolled, can contribute to an accident.”Risk is a term ap plied to the individual or combined assessments of “probabili

loss” and “potential amount of loss.” NSC definesaccidentas “an

JOURNAL OF CONSTRUCT

occurrence in a sequence of events that produces unintend jury, death, or property damage. Accident refers to the even the result of the event.”

Accident Causation Models

Accident causation models attempt to understand the factor processes involved in accidents in order to develop strategi accident prevention. The different models are based on diff perception of the accident process. Some of the most influ accident causation models and methodologies are:~1! the Single Event concept;~2! the Determinant Variable concept;~3! the Domino Theory ~Heinrich 1936!; ~4! the Fault Tree analytica methodology; ~5! the Energy-Barriers-Targets model, which views the accident process as an unwanted release of ener to inadequate physical or procedural barriers;~6! the Manage- ment Oversight and Risk Tree, which focuses on “what” barr ers failed and “why” they failed—that is, what management ments permitted the barrier failure~DOE 1992!; ~7! Petersen’ Multiple Causation model ~1971!; and ~8! Reason’s ~1990! ‘ Swiss Cheese’ model of human error, and the “resident pat gens” or “latent failures.”

Construction Safety Literature

Construction researchers have proposed several accident tion models and root causes. McClay’s~1989! “universal frame work” identified three key elements of accidents: hazards, hu actions, and functional limitations. Hinze’s distraction the ~1996! argued that production pressures can distract workers the hazards and increase the probability of accidents. Abdelh and Everett~2000! identified management deficiencies, train and workers’ attitude as the three general root causes. “constraints-response” model~Suraji et al. 2001! argues tha project conditions or management decisions~distal factors! can cause responses that create inappropriate conditions or a ~proximal factors! that lead to accidents.

Organizational factors associated with safety performanc clude top management’s attitude toward safety~Levitt 1975!, or- ganizational culture~Molenaar et al. 2002!, safety climate~Mo- hamed 2002!, superintendent practices~Levitt and Samelso 1987; Hinze and Gordon 1979!, and turnover~Hinze 1978!. Hinze and Parker~1978! found that job pressures and crew competi are related to more injuries. Hinze~1981! found that good work ing relationships improve safety.

In terms of safety practices, Jaselskis et al.~1996! identified the frequency of formal safety meetings with project supervi and safety budget as factors related to lower incident rates Construction Industry Institute identified five best practices~Liska et al. 1993!: preproject and pretask planning for safety, sa orientation and training, safety incentives~the effect of this ha been debated!, alcohol and substance abuse programs, and dent investigations. Toole~2002! identified eight root causes accidents: lack of proper training, safety equipment not prov deficient enforcement of safety, unsafe equipment, metho condition, poor safety attitude, and isolated deviation from scribed behavior.

Few researchers have emphasized the role of design in struction safety~Hinze and Wiegand 1992; Gambatese 2000! and the importance of task characteristics and work method~Everett

1999! and proposed technological interventions to improve the

ION ENGINEERING AND MANAGEMENT © ASCE / JULY 2005 / 817

r than

n to ork-

n re- Rigby eeds dg- c- Rea- two

ith the

- her rrect s itua-

he hav- prac

. or

acci- iola-

by er- aints s ork- etc. loser

res-

ly the effort d- sible man

lim-

ed in from ward

rther- tation ehav- s of hat ac- rror- ntrol

k- s -

ds on causa- bles tion

signs rs; a s, the r ect

the the ctivi- ma-

per- s n-

ations l con-

e, as work- s to k- res zards. ” and zard ntrol

ut do

safety of specific construction operations~Bernhold et al. 2001!. These approaches focus on reducing the safety risks, rathe increasing the safety effort.

The construction safety literature has paid little attentio worker errors and effective ways to manage errors in the w place.

Human Error

Human error is a central element in accidents and has bee searched extensively by researchers of high-risk systems. ~1970! defines an error as a set of human actions that exc some level of acceptability. Traditionally, the standard of ju ment is the normative~prescribed! behavior. From this perspe tive, human error is a deviation from a normative procedure. son~1990! classified unsafe acts in three types of errors, and types of violations. Errors. Slips and lapsesare “skill-based” errors and occur w little or no conscious thought. A slip is an unintended error in execution of an otherwise correct plan. Mistakes~or decision er rors! involve the correct execution of a wrong plan. In ot words, mistakes are intentional behaviors that involve inco choice of action~inappropriate for the situation!. Perceptual error are actions that result from misinterpretation of the actual s tion.

Violations. Routine violationsare habitual departures from t rules and often tolerated by supervision. This may involve be iors that are established practice as opposed to the specified tice, such as driving 5 – 10 mph faster than the speed limitEx- ceptional violationsare neither typical of the individual n condoned by management.

Rasmussen’s Descriptive Model of Work Behavior

Descriptive models of work behavior attempt to understand dents without reference to normative concepts of errors or v tions. An important descriptive model is the one proposed Rasmussen et al.~1994!. According to Rasmussen, workers op ate within a work system shaped by objectives and constr ~economic, functional, safety related, etc.!. A worker searche freely within those boundaries guided by criteria such as w load, cost effectiveness, risk of failure, joy of exploration, Figure 1 illustrates how the work behaviors tend to migrate c to the boundary of functionally acceptable performance~limit of loss of control! due to two primary pressures: the production p

Fig. 1. Rasumussen’s work behavior model~adapted from Rasmussen et al. 1994!

sures for increased efficiency, and the tendency for least effort,

818 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT

which is a response to increased workload. Managers supp “cost gradient” and the worker searches and finds a “least gradient.” The result is a “systematic migration toward the boun ary of acceptable performance, and when crossing an irrever boundary, work will no longer be successful due to a ‘hu error’” ~Rasmussen et al. 1994, p. 149!. A breakdown in work performance indicates an operation too close to its capability its and/or the limits of the ability to recover control.

Safety programs attempt to counter the pressures outlin the Rasmussen model and prescribe “safe behaviors” away the boundary. However, the pressures that push workers to the boundary require that safety efforts are continuous. Fu more, efforts to improve system safety lead to human adap that compensates for safety improvements. Thus, the work b ior is likely to be maintained close to the boundary of los control. To address these problems, Rasmussen proposes t cident prevention efforts should focus on development of e tolerant work systems that make the boundary of loss of co visible and reversible.

Based on Rasmussen’s framework, Howell et al.~2003! iden- tify three zones of operation:~a! the “safe zone,” where the wor ers’ behaviors are within the boundary defined by safety rule~b! the “hazard zone”~or “near the edge”!, and~c! the “loss of con trol” zone.

Accident Causation Model

Model Overview

Figure 2 presents the accident causation model, which buil the Rasmussen model and previous construction accident tion models. This conceptual framework identifies the varia that influence the likelihood of accidents during a construc activity. The arrows indicate cause-effect relationships. The indicate the direction of the relationship between the facto positive sign indicates that when the causal factor X change effect Y changes in the same direction~X increase→Y increase o X decrease→Y decrease!. A negative sign indicates that the eff changes in the opposite direction~X increase→Y decrease, X decrease→Y increase!.

The characteristics of the activity and work context, and task unpredictability shape the work situations within which workers operate, and create hazardous situations. Different a ties involve different hazardous situations, depending on the terial, tools, location, etc. Furthermore, the same activity formed under different method or context~physical condition and surrounding activities! involves different hazards. Task u predictability leads to unplanned tasks and unexpected situ that also create hazardous situations. Safety efforts to contro ditions reduce the hazardous situations.

Workers’ behaviors determine both the production outcom well as the exposure to hazards. Production pressures and load and the tendency for competent action drive worker adopt more efficient work behaviors~such as working faster, ta ing shortcuts, or working without the required safety procedu!, which increase production, but also increase exposure to ha The shaded section where the “hazardous work situations “work behaviors” overlap indicates work behaviors in the ha zone that expose workers to hazards. Safety efforts to co workers’ behaviors reduce exposures to hazards.

Exposures to hazards create the potential for incidents, b

not necessarily lead to incidents. For an incident to occur, the

dition igger ny er con- ectio

ver- lease pends tors. “near

and s an e not

with t and

haz- onal r one etect situ-

ion is and

dent. work

rent or

haz-

b- tools

ent ity - -

c- fined

ate bris,

may

nges efore

tivity n-

ed as ated, s may

hazard must be released. Human errors and changes in con create a “mismatch” between conditions and actions and tr the release of hazards. Not all errors release hazards—ma rors are inconsequential, while other errors are “trapped” and trol is recovered before the hazard is released. The shaded s where “Exposures” and “Errors & Changes in conditions” o lap, indicates the errors under condition of exposure that re hazards and generate incidents. The likelihood of errors de on the task, the environment, and the workers’ capacity fac Depending on the consequences, an incident may be a miss,” an injury accident, or a fatality.

The causation model in Fig. 2 depicts the key factors processes that lead to accidents. Several other relationship feedback loops exist, which are discussed briefly as they ar as critical for understanding the accident process.

Hazardous Work Situations

In this paper, hazardous situations are defined as “situations the potential to cause injury, unless the worker can detec avoid the hazard, without exposing themselves to a greater ard.” This definition acknowledges the subjective and situati nature of many hazards. In other words, what is a threat fo person may not be for another, depending on the ability to d and avoid the hazard. Furthermore, what is a hazard in one ation may not be in another. The hazardous situations definit consistent with NSC’s definition, because the ability to detect avoid a hazard reduces its potential to contribute to an acci

As shown in Fig. 2, the nature and number of hazardous situations during an activity depend on the following factors:~1! characteristics of the activity and context,~2! safety efforts to control conditions, and~3! task unpredictability. A discussion these factors follows.

Activity and Context Characteristics Different activities involve different hazards. Surveying in

Fig. 2. Accid

empty site involves few hazards, while steel erection includes

JOURNAL OF CONSTRUCT

many more. Furthermore, the same activity involves diffe hazardous situations if performed with a different method under different conditions. There are three main sources of ards:~a! the work technology,~b! the physical conditions, and~c! the surrounding activities.

(a) Work technology. The work technology includes the o jects and actions necessary to perform the task, such as the and equipment~scaffolds, power tools, cranes, heavy equipm!, the material~e.g., heavy or sharp objects, chemicals, electric!, the physical tasks required~material handling!, and the by products of the production task~scrap metal, etc.!. The same ac tivity may be performed with different tools and equipment~lad- ders, scaffolds, or mechanical lifts!, or at different locations~on the ground or at elevation! and involve different hazards.

(b) Physical conditions. The physical environment of the a tivity ~such as high elevations, floor openings, trenches, con space, overhead power lines, and housekeeping conditions! cre- ates another set of hazards. Environmental conditions~such as cold, heat, illumination, vibrations, noise, vapors, etc.! may create additional sources of hazards.

(c) Surrounding activities. Surrounding activities also gener “threats,” such as falling objects, heavy equipment traffic, de etc. The project schedule~sequence and timing of activities! af- fects the work context. For example, out-of-sequence work increase the difficulty and the hazards of the activity.

The number and nature of hazardous work situations cha during the course of an activity. Some hazard sources exist b the activity starts, while other hazards develop during the ac ~e.g., a trench is excavated!, tools and equipment wear out, co ditions change, and surrounding activities start and end.

Task Unpredictability Task unpredictability means that the work cannot be complet planned—the scope of the task may be different than anticip the actual conditions are different than expected, or resource

usation model

ent ca

be missing~information, tool, material, etc.!. Workers’ errors also

ION ENGINEERING AND MANAGEMENT © ASCE / JULY 2005 / 819

reate bil-

d the

quired e.g.,

re an , the , un- more han-

tasks,

ute” ction s” of , eve ure. uces

r ex- indi-

on r. on- e the

e the as p area

s are rts to tify o pro s and pro-

age- e the ted.

ence azard e

nter cient

s y d hav-

s and mea- tion. ovid- quip-

vior- ry of ures. time

“risk- m-

s and uch iolate ally ce the der lead

afety zards

afety with

way

d by y

educe s to g in cam- -

ad- . First, ision amic tion dy-

ures t- olved ely

to- unter x and

any

duc-

increase task unpredictability as errors by one crew may c unpredictable conditions for a following crew. Task unpredicta ity increases both the likelihood of hazardous situations, an production pressures and workload.

Unpredictability Generates Hazardous Situations. First, the resources, equipment, manpower, or safety measures re may not be available for the unexpected tasks or conditions, the crew planned for a 6 ft. ladder, but some locations requi 8 ft. ladder. Even if a crew performs safety pretask planning plan will be inadequate if the task is unpredictable. Second predictable tasks and conditions require increased effort, movement of workers and equipment, increased material dling, increased need to improvise, more out-of-sequence and involve much chaos and confusion.

Unpredictability Increases Workload and Production Pres- sures. This generates interruptions and “urgent/last min problems that have to be resolved promptly, otherwise produ can be significantly disrupted. This creates temporary “peak production pressure and sudden changes to production pace if the overall activity is not under particular schedule press Furthermore, resolving the interruptions takes time and red the time available for the planned task.

The increased production pressures may lead the worke~or supervisors! to do the work in any way they can~“make do”! without the appropriate safety measures or resources. Fo ample, lack of adequate manpower may lead a worker to vidually perform tasks that normally require two people~e.g., move heavy material, enter confined space, etc.!. Or, if a ladder is not tall enough for all the work locations, the worker may step the last two steps, rather than look for an appropriate ladde

Much of the complexity and dynamism of the work in c struction is caused by a failure to reliably plan and coordinat work activities.

Safety Effort to Control Conditions Safety measures to control conditions are barriers that confin hazard sources, and prevent exposure to the hazards, such rimeter cable, support of deep trenches, and closing-off the under steel erection. OSHA regulations define what condition hazardous and what safety barriers are needed. Safety effo control conditions include training and inspections to iden hazardous conditions, and the time and resources needed t vide and maintain the safety measures. Economic pressure time or personnel shortage may prevent management from viding and maintaining the required safety measures. Man ment commitment and policies that support safety increas likelihood that the safety resources and effort will be commit

Efficient Work Behaviors

Efficient work behaviors increase production, but in the pres of hazards such behaviors also bring the workers in the h zone~expose workers to the hazards!, which in turn increases th likelihood of incidents that may disrupt production and cou any prior gains. Thus, under hazardous conditions, less effi work behavior is required to prevent exposure.

Efficient work behavior is shaped by:~1! production pressure and workload, which increase efficient behavior,~2! the tendenc for competent action, which increases efficient behaviors, an~3! safety efforts to control behaviors, which reduce efficient be

iors and consequently exposures.

820 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT

Production Pressures and Workload Rasmussen’s framework illustrates that production pressure efforts to reduce workload may lead workers to avoid safety sures, or not follow the safety rules if they slow down produc Similarly, cost pressures may prevent management from pr ing the required safety measures or appropriate tools and e ment.

Tendency for Competent Action The tendency for efficient, competent action is another beha shaping factor, which pushes workers close to the bounda loss of control, even in the absence of high production press Workers take shortcuts or exert excessive effort to reduce the to perform a task. These behaviors are typically considered taking.” However, from the perspective of the worker, it is co petent action—experienced professionals develop shortcut tricks of the trade as efficient ways to perform the work. S behaviors often are established trade practices that may v prescribed procedures—they are “routine violations” typic tolerated by supervisors. Such behaviors protect and enhan workers’ feeling of competency. They are very efficient un normal conditions, but under special circumstances they may to accidents.

Safety Effort to Control Behaviors When hazards cannot be contained with physical barriers, s procedures are prescribed to prevent exposure to the ha ~lockout-tag out, or testing the air in confined spaces!. “Unsafe behaviors” are those acts that violate prescribed procedure. S programs and campaigns attempt to increase compliance safety rules, and maintain work behaviors in the “safe zone” a from the boundary.

The prevailing view is that unsafe behaviors are cause lack of knowledge of the hazards~competent person philosoph! or poor safety attitude. As a result, management actions to r unsafe behaviors focus on training and motivating worker comply with the safety rules. Such practices include trainin safety rules and procedures, incentives and motivational paigns~such as safety culture and value-based safety!, enforce ment and punishments, and behavior-based safety.

The main limitation of these practices is that they do not dress the systemic forces that push workers near the edge the dynamic nature of work does not involve conscious dec making or risk assessment; workers immersed in the dyn flow of work do not make decisions based on careful situa analysis but on know-how, heuristics, and a perception of namic control, and they cannot follow prescriptive proced prepared by outside experts~Rasmussen 1997!. Second, shor term conflicts between safety and production are usually res in favor of production, as efforts for production have relativ certain outcomes and receive rapid and rewarding feedback~Rea- son 1990!. Finally, because the workers’ behaviors migrate ward the boundary, safety needs to maintain continuous co pressures. Last, but not least, in an unstructured, comple dynamic environment like a construction jobsite, there are m hazardous situations not covered by work rules.

Production

Efficient work behaviors increase production. Increased pro

tion reduces production pressures, which in turn reduce efficient

hen s re-

s the s of r can

l rk-

rmal t the

hout ning

ive

ith wer etc.

t does cted

ons ciden n con e is a

l to re re-

l z- cci-

fety mad

is in f loss error error

ystem the e ap-

be the

age- rrors

a- n -

rs i-

d to wrong he tor ider- med ent

orkers Error

devel- -

are- s to ision-

s the , etc., tive- amo- s is pera-

se of ho is

ip- b- sce- quent

near fast then n the very situ- and

work behaviors. Production is also affected by accidents—w the number and severity of incidents increase, production i duced.

Exposure to Hazards

Exposure to hazards exists when the work behavior bring worker in the “hazard zone” and near the boundary of los control, where events can take place faster than the worke detect and avoid the danger. Exposure are the result of:~a! effi- cient behavior that leads to routine violations,~b! exceptiona violations,~c! proper actions that are near the limits of the wo er’s ability ~such as physical effort and ergonomic exposures!, or ~d! unrecognized hazards: if a hazard is not identified, a no work behavior may expose a worker to the hazard withou worker’s knowledge.

Examples of exposures are working in a deep trench wit sloping or trench protection, working near an unprotected ope at high elevation without fall protection, working with defect tools and equipment~knowingly or unknowingly!, performing electrical work without lockout-tag out, working in an area w heavy equipment traffic, operating equipment close to po lines, working under another crew, handling heavy material,

Exposure to hazards creates the potential for accident, bu not automatically lead to accident: a worker near an unprote edge will not necessarily fall. In other words, “unsafe” conditi and actions are not sufficient to cause an accident. For an ac to occur, the hazard must be released. Errors and changes i ditions trigger the release of the hazard: in both cases ther “mismatch” between the situation and the action.

Errors and Changes in Conditions

Unlike violations, errors are unintended actions that fai achieve their intended outcome. As discussed in the literatu view, human error involves slips~unintentional loss of control!, mistakes~selection of incorrect course of action! and perceptua errors. Huang and Hinze~2003! found that misjudgment of ha ardous situation was a significant factor in over 30% of fall a dents.

Not all errors release hazards. An error will have no sa consequences if there is no exposure to a hazard; e.g., errors in flight simulators are inconsequential. However, if a worker the hazard zone, an error may push him over the boundary o of control, and release the hazard. If the worker detects the soon enough to “trap” the error and recover control, then the will not release the hazard.

Hazards are also released by changes in the state of the s such as mechanical failures, loss of soil stability, etc. If worker can react fast enough and adjust the behavior to on propriate for the new conditions, then the accident can avoided. If the change in the conditions is too sudden for worker to recover control, it results in an accident. Error man ment and situation awareness increase the ability to correct e detect changes, and prevent or avoid the release of hazard

Error Inducing Factors The likelihood of errors depends on the task demands~complex- ity, dynamism, pressures!, the environment, and the workers’ c pacity. According to Rasmussen et al.~1981!, causes of huma “malfunction” include external events~such as distractions, com ponent failures, or the physical environment!, excessive task d

mands ~due to task characteristics and situation!, performance

JOURNAL OF CONSTRUCT

t -

shaping factors~such as work load, skills, and stress facto!, reduced capacity~due to fatigue, etc.!, or intrinsic human var ability.

Natural drive toward economy of cognitive effects may lea wrong assessment of situations and task demands, and rules to be applied~Reason 1990!. Many perceptual errors are t result of “cognitive confusion;” that is, the selection of a mo program to execute a previously learned task while not cons ing the new conditions of the environment the task is perfor in or the new dimensions or design of the tools or equipm being used.

Error Management Error management is a set of strategies that enable the w detect and correct errors before onset of consequences. management strategies for individuals and team have been oped in other sectors~primarily in military and commercial avia tion! and have focused primarily on improving situation aw ness~Endsley 1988!, and developing effective team processe increase a team’s collective situation awareness and dec making ~Helmreich et al. 1999!.

In construction, the primary error-management strategy i use of warnings, such as signs, spotters, backup alarms which alert workers when approaching a hazard. The effec ness of such warning measures is limited. Blackmon and Gr padhye~1995! found that the effectiveness of backup alarm low because of the general noise level of the jobsite, the o tors’ reliance on the alarms, and reduced attention.

Incidents and Consequences

An incident is the undesired event that results from the relea the hazard. A hazard may be released by the same worker w exposed to the hazard~e.g., a worker loses control of the equ ment!, or by another worker~e.g., another worker drops an o ject!. Table 1 lists examples of typical exposure to hazard narios found on construction sites, and possible subse scenarios of the release of hazards.

Depending on the consequences, an incident can be a miss, an injury, or a fatality accident. If the worker can react enough to avoid the hazard or interrupt the flow of events, the incident is a near miss. The ability to react depends o speed of the hazard release: when the loss of control is sudden, there is no time to react. Worker’s experience and ational awareness is critical in anticipating hazard release

Table 1. Hazard Scenarios on Construction Operations

Hazard Exposure to hazard Incident

Unprotected edge ~physical condition!

Worker near unprotected edge

Worker slips and falls

Saw with dull blade ~undermaintained tool!

Worker using saw with dull blade

Saw jams and kicks back

Material handling ~activity element!

Worker lifts material A muscle exceeds functional limit

Surrounding activities Worker in same area w/excavator

Excavator turns and hits worker

Surrounding activities Working under another crew

Object falls and hits worker

Unsecured power source~physical condition!

Working without lockout-tag out

Another worker turns on the power

taking appropriate action.

ION ENGINEERING AND MANAGEMENT © ASCE / JULY 2005 / 821

miti- equip - e te or d

d on azard from

the lease ening king hese r per- e

son posed eavy nally rker lt to

. Al- afe,” d by te the the ns. es . The

n have

ow to since

ani- lso they

nset

h are ety h in-

m, as r fo-

under keli- onal

model. ues-

uture

the es the

nd the to could

rther may t

5 n 1 ame tices.

ning duces

and s con- kers keep ing:

ffi-

work ss of ome be-

the , we s of es,

Protective measures are “the last line of defense” and can gate the consequences of an incident. Personal protective ment increases the error tolerance of the system~e.g., fall protec tion equipment reduces the consequences of falls!. The magnitud of injury also depends on situational factors that may aggrava mitigate the injury ~including the individual’s tolerance, an luck!.

Taxonomy of Incidents

The model identifies three different types of accidents, base the source of exposure and the action that releases the h This taxonomy is based on etiology of accidents, and differs OSHA’s classifications.

~a! Loss of control.The first type involves situations where person who is exposed to the hazard is also the one who re the hazard. For example, a worker near an unprotected op may slip or not see the opening and fall. Or a crane is wor near a power line and comes in contact with the line. In t cases, the release of the hazard is due to loss of control, o ceptual failure to detect the boundary~the edge of the slab, th distance from the power line!.

~b! Coordination.In the second type of incidents, the per who releases the hazard is different than the one who is ex to the hazard. For example, a worker is working near h equipment, and the equipment operator turns and unintentio strikes him. Or a worker is under another crew, and the wo above drops an object. What makes such accidents difficu prevent is that they happen under normal work behaviors though the actions of both workers may be independently “s they fail to adjust their behavior in the new conditions create the presence of the other worker. Safety rules try to separa crews and determine rules of “priority” and responsibility, but dynamic conditions on the jobsite create many such situatio

~c! Unrecognized hazard.Another type of accidents involv situations where the worker is exposed to a “hidden” hazard hidden hazard can be a component near its functional limit~such as an unsecured deck! and a normal behavior~such as walking o the deck! may release the hazard. The hidden hazard may been created by an error of a previous crew. The issue of h address unrecognized hazards is a particularly difficult one many things are not recognized as hazards until after they m fest themselves~Prichard 2002!. Ergonomic hazards may be a included in this category, as the workers cannot recognize if are approaching the limit of physical tolerance~the limit of how much weight the worker can safely lift, the time before the o of cumulative trauma disorders such as tendonitis, etc.!.

Other Causal Relationships

Several other relationships and feedback loops exist, whic not depicted in Fig. 2, such as~a! safety incidents increase saf efforts, as management typically responds to incidents wit creased efforts to control conditions and behaviors; and~b! safety incidents reduce efficient behaviors, at least in the short ter workers behave more carefully after an accident. The pape cused on the causal factors and relationships necessary to stand the main forces at work. A model that considers the li hood of accidents over time should take the additi

relationships into account.

822 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT

Model Discussion and Implications

The next section discusses some of the issues raised by the It provides some supporting evidence, but mostly it raises q tions, proposes hypotheses, and identifies directions for f work.

Importance of Task Unpredictability

The model argues that the unpredictability of the task and environment leads to increased accidents because it increas number of hazardous situations, the production pressures, a likelihood of errors. As a result, “unlikely events are likely happen, because there are so many unlikely events that happen” ~Per Bak 1996!.

The effect of task unpredictability on accidents requires fu investigation. However, there is some initial evidence that it be significant. A recent study~Thomassen et al. 2003! found tha projects that used the Last Planner System®~LPS! for production control ~Ballard and Howell 1998! had an incident rate of 7.8 ~12 incidents in 305,604 labor-hours!, while the incident rate o projects that did not use LPS was 14.13~41 incidents in 580,37 labor-hours!. Both groups of projects were performed by the s contractor, were of similar nature, and used same safety prac These initial findings suggest that effective production plan can have significant effect on safety, possibly because it re the task unpredictability.

Task Unpredictability, Work Behaviors, and Safety

The model highlights the conflict between safety measures the pressures for efficient behavior. Current safety measure straint productive behaviors with a set of rules to keep wor away from the hazard zone. Instead of asking how can we workers from acting in more efficient ways, we should be ask “ How can we make it safe for workers to work more e ciently?”

Figure 3 proposes a relationship between task conditions, behaviors, and accidents. The curve indicates the limit of lo control. The figure illustrates that when task conditions bec more complex and unpredictable, the efficiency of the work havior must be reduced to avoid an accident.

In a simple example, the safe speed on a road~for a specific driver and vehicle! depends on the condition of the road, visibility, the traffic load, etc. To safely increase the speed need to create conditions that reduce the likelihood of los control. This may include increasing visibility around curv

Fig. 3. Work conditions, work behavior, and safe operation

widening the road, reducing the potholes and obstacles, etc. It is

den

st ef- ting ated, e of age- cope

sures

d. mili- in- pro

s and

enti- with

and

i- rdless s

a- tive

op- n to t

here h as

ntrol

ASA es in es in that ity of di-

ty of was

heir d on age-

. pro-

s e ilities

- tions r can fore and nti-

way n or s- ve und- nd it

influ- given ction

t. Of t -

ions

not re- ment

haz- res- ap- any s the pre- ards

npre- d, the

ions

ing and

rk so ated ined

r re-

rew the

important to stabilize the work conditions and prevent sud deterioration when workers are at a point of high efficiency.

Error Management

As discussed earlier, production pressures, tendency for lea fort and task unpredictability often result in the workers opera in the hazard zone. In addition, human error cannot be elimin especially in complex and dynamic work situations. Becaus the inevitability of exposures and errors, effective error man ment is necessary in order to increase the workers’ ability to with hazardous situations.

Error management provides a set of error countermea with three lines of defense:~1! error avoidance,~2! error trapping ~to prevent errors from propagating!, and ~3! error mitigation to reduce the consequences of those errors that are not trappe

Error management strategies have been developed first in tary and commercial aviation and have focused primarily on creasing situation awareness, and establishing effective team cesses to increase a team’s collective situation awarenes decision-making.

Situation Awareness In aviation, problems with situation awareness have been id fied as the leading causal factor in accidents associated human error. Situation Awareness~SA! is the perceptionof the elements in the current environment within a volume of time space, thecomprehensionof their meaning~the “forces at work”!, and the projection of their status in the near future~Endsley 1988!. SA is affected by individual factors~capabilities, exper ence, fatigue/stress, set expectations, and “press-on-rega mentality!, task factors~task overload and underload, objective!, and environmental factors~complexity, interruptions, and d graded operating conditions!.

Team Processes (Crew Resource Management) Crew Resource Management~CRM! was first developed in avi tion to reduce accidents due to “pilot error.” CRM is an ac process by crewmembers to identify significant threats to an eration, communicate them to the pilot and carry out a pla avoid or mitigate each threat~Helmreich et al. 1999!. In recen years, CRM has been adopted by other high-risk operations w effective work performance requires coordinated action, suc hospital operating teams, oil platforms, and power plant co centers.

Based on analysis of over 28,000 aviation accidents, N discovered that over 70% of the accidents were due to failur team communication and coordination rather than deficienci technical proficiency. Further simulator studies confirmed crew performance was more closely associated with the qual crew communication than with the technical proficiency of in vidual pilots. No differences were found between the severi the errors made by effective and ineffective crews, rather, it the ability of the effective crews to communicate that kept t errors from snowballing into undesirable outcomes. Base these findings, NASA developed the Cockpit Resource Man ment system~later called Crew Resource Management! to im- prove the crews’ situational awareness and decision making

CRM emphasizes the key nontechnical skills and team cesses that affect operational safety. CRM training addressesitu- ation awareness, contingency planningto identify ahead of tim the proper response to abnormal situations, assign responsib

for handling problems, and predetermine crew roles for high-

JOURNAL OF CONSTRUCT

”

workload phases of flight,stress management, and crew commu nication such as cross checking and communicating inten before the execution of actions, so that another crew membe identify an inappropriate intention or action and correct it be the error happens, soliciting input from all crew members, assertivenessin alerting team members and supervisor of ide fied threats and errors.

Other Error Management Strategies Error proofing involves design of tools and equipment in a that they can detect an abnormal condition and shut dow independently act to prevent failure~such as stability control sy tems in cars!. Rasmussen~1997! proposed that an alternati strategy to safety rules is to increase the visibility of the bo ary. The concept of boundary however is not well defined a may include a physical boundary~the edge of the slab!, point of loss of control~the crane’s point of loss of stability!, or a func- tional limit ~the load a muscle can take!.

Directions for Accident Prevention

The model identifies several potential interventions that can ence the safety outcome of a construction operation. For a activity, the hazards can be reduced by changing the constru method, the work sequence or the physical environmen course, the activities required depend on design choices~e.g., cas in place concrete versus steel structure!. However, given a con struction method, the model identifies two important direct for accident prevention:~1! reduce task unpredictability, and~2! increase error management capability. These strategies do place the safety defenses and technical training but comple them.

Reduce Task Unpredictability

Reducing unpredictability will reduce unexpected tasks and ardous situations, interruptions and ‘short-term’ production p sures, and will reduce the likelihood of errors. The current proach does not deal with the fact that workers face m unpredictable situations. Safety pretask planning addresse predictable hazards involved in an activity. However, in an un dictable task or environment, there will be situations and haz that safety pretask planning will not address. When task u dictability is reduced, the task can be executed as planne hazards will be predictable, and the defenses can be set.

This strategy shifts safety’s focus from controlling the act of management and workers~follow the rules! to stabilizing the work conditions.

Unpredictability can be reduced if the production plann system produces high quality work assignments. Ballard Howell ~1998! identify five requirements for high quality wo assignments:~a! definition~work assignment is specific enough that the right material can be provided, work can be coordin with other trades, and at the end of the week it can be determ if the assignment was completed!, ~b! soundness~all design in- formation, material, prerequisite work, work area and othe sources needed to complete the assignment are available!, ~c! se- quence~work is released in the correct sequence!, ~d! size ~the assignments are sized to the productive capability of the c!, and ~e! learning ~assignments not completed are tracked and reasons identified!.

Only assignments that meet the quality criteria should be in-

ess viors

task

indi- com- but

team . The

vi- om-

at it rac- , and

een e- con- to

con- , the ment ch an ward s and

igated push

tions ener- sures

egies; ent

not that

ropo- nder- ping

th-

- -

en- .,

ion.”

ites, The

r .

ruc- rd

.” s. ent

ts.”

ro-

zed

ty.” om/

cluded in the weekly work plans. In addition, the work proc and conditions must be controlled to support efficient beha ~clear access, layout, location and sequence of subtasks!. Finally, during the task, the crew needs to periodically re-evaluate the and conditions to prevent degradation.

Increase Error Management Capability

Construction has not developed systematic ways to train vidual and teams in error management. The most developed ponent is training in hazard identification which is essential, not sufficient. The skills related to situation awareness and processes are developed incidentally through experience need for awareness and effective team processes~within a crew and between different crews! becomes more important on acti ties and projects with high uncertainty and complexity, c pressed schedules, and limited work areas.

CRM’s main difference from existing safety practices is th focuses on critical nontechnical aspects of the workers’ inte tion that enable the crew to successfully recognize, cope with recover from hazardous situations and errors.

A simple version of CRM adapted for construction has b proposed by Mitropoulos et al.~2003!. In this approach manag ment requires crewmembers to speak up when they identify ditions that exceed their “comfort zone” in terms of ability perform the work effectively and safely. Members can raise cerns about the work method, the conditions in the work area sequence of the work, the lack of safety measures and equip lack of appropriate equipment for the task, etc. To support su approach, management must take a non punitive policy to errors, and must create a team environment that support develops workers’ assertiveness.

Conclusions

The accident causation model presented in this paper invest the production factors that generate hazardous situations and the workers into the hazard zone. It also examined the limita of current safety strategies in counteracting the problems g ated by the production factors. The paper argued that expo and errors are inevitable and proposed two alternative strat reducing task unpredictability and improving error managem capabilities. A limitation of the model is that it may have considered other important factors and causal relationships contribute to accidents. The model should be considered as p sitions for testing. Future research should focus on better u standing the effect of task unpredictability, and on develo error management strategies.

References

Abdelhamid, T., and Everett, J.~2000!. “Identifying root causes of con struction accidents.”J. Constr. Eng. Manage., 126~1!, 52–60.

Ballard, G., and Howell, G.~1998!. “Shielding production: Essential st in production control.”J. Constr. Eng. Manage., 124~1!, 11–17.

Benner, L., Jr.~1985!. “Rating accident models and investigation me odologies.”J. Safety Res., 16, 105–126.

Bernold, L., Lorenc, S., and Davis, M.~2001!. “Technological interven tion to eliminate back injury risks for nailing.”J. Constr. Eng. Man age., 127~3!, 245–250.

Blackmon, R. B., and Gramopadhye, A. K.~1995!. “Improving construc

tion safety by providing positive feedback on backup alarms.”J. Con-

824 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT

str. Eng. Manage., 121~2!, 166–171. Bureau of Labor Statistics.~2004!. “National census of fatal occupation

injuries in 2003.” U.S. Department of Labor, Washington, D.C. Department of Energy.~1992!. “Root Cause Analysis Guidance Doc

ment.” DOE-NE-STD-1004-92, Washington, D.C. Endsley, M.~1988!. “Design and evaluation for situation awareness

hancement.”Proc., HF S, 32nd annual meeting, Santa Monica, Calif 97–101.

Everett J.,~1999! “Overexertion injuries in construction.”J. Constr. Eng Manage., 125~2!, 109–114.

Gambatese, J. A.~2000!. “Safety constructability: Designer Involveme in Construction Site Safety.”Proc., Construction Congress VI, Febru- ary 20–22, Orlando, Fla., 650–660.

Heinrich, H. W. ~1936!. Industrial accident Prevention, McGraw-Hill, New York.

Helmreich, R. L., Merritt, A. C., and Wilhelm, J. A.~1999!. “The evolu- tion of crew resource management training in commercial aviat Int. J. Aviat. Psychol., 9~1!, 19–32.

Hinze, J.~1978!. “Turnover, new workers, and safety.”J. Constr. Div. Am. Soc. Civ. Eng., 104~4!, 409–417.

Hinze, J.~1981!. “Human aspects of construction safety.”J. Constr. Div. Am. Soc. Civ. Eng., 107~1!, 61–72.

Hinze, J.~1996! “The distraction theory of accident causation.”Proc., Int. Conf. On Implementation of Safety and Health on Constr. Sites, CIB Working Commission W99: Safety and Health on Construction S L. M. Alvez Diaz and R. J. Coble, eds., Balkema, Rotterdam, Netherlands, 357–384.

Hinze, J., and Gordon, F.~1979!. “Supervisor-worker relationship affec injury rate.” J. Constr. Div., Am. Soc. Civ. Eng., 105~3!, 253–262.

Hinze, J., and Parker, H. W.~1978!. “Safety, productivity and job pre sures.”J. Constr. Div., Am. Soc. Civ. Eng., 104~1!, 27–35.

Hinze, J., and Wiegand, F.~1992!. “The role of designers in constructi worker safety.”J. Constr. Eng. Manage., 118~4!, 677–684.

Howell, G. A., Ballard, G., Abdelhamid, T., and Mitropoulos, P.~2003!. “Rethinking Safety: Learning to Work near the Edge.”Proc., Con- struction Research Congress, March 19–21, Honolulu, Hawaii.

Huang, X., and Hinze, J.~2003!. “Analysis of Construction Worker Fa Accidents.”J. Constr. Eng. Manage., 129~3!, 262–271.

Jaselskis, E. J., Anderson, S. D., and Jeffrey, S. R.~1996!. “Strategies fo Achieving Excellence in Construction Safety Performance.”J. Constr Eng. Manage., 122~1!, 61–70.

Levitt, R. ~1975!. “The effect of top management on safety in const tion.” Technical Report No. 196, The Construction Institute, Stanfo University, Calif.

Levitt, R. E., and Samelson, N. M.~1987!. Construction safety manag ment, McGraw-Hill, New York.

Liska, R. W., Goodle, D., and Sen, R.~1993!. “Zero accident techniques Source Document 86, Construction Industry Institute, Austin, Texa

McClay, R. E.~1989!. “Toward a more universal model of loss incid causation.” Professional Safety, January, 15–20.

Mitropoulos, P., Howell, G. A., and Reiser, P.~2003!. “Workers at the Edge: Hazard Recognition and Action.”Proc., 11th Annual Confe ence on Lean Construction, IGLC, June 22–24, Blacksburg, Va.

Mohamed, S.~2002!. “Safety climate in construction site environmen J. Constr. Eng. Manage., 128~5!, 375–384.

Molenaar, K., Brown, H., Caile, S., and Smith, R.~2002!. “Corporate culture: A study of firms with outstanding construction safety.” P fessional Safety, July, 18–27.

Per Bak. ~1996!. “How nature works—The science of self-organi criticality.” Springer, New York.

Petersen, D.~1971!. Techniques of safety management. McGraw-Hill, New York.

Prichard, R.~2002!. “Debunking the 13 myths of construction safe International Risk Management Institute, May, http://www.irmi.c expert/articles/prichard010.asp

Rasmussen, J.~1990!. “Human error and the problem of causality analysis of accidents.”Philos. Trans. R. Soc. London, Ser. B, 327,

449–462.

od-

, M., nts Kir- ent

trol

al .

in a on

Rasmussen, J.~1997!. “Risk management in a dynamic society: A m eling problem.”Safety Sci., 27~2/3!, 183–213.

Rasmussen, J., Pesdersen, O. M., Mancini, G., Carnino, A., Griffon and Gagnolet, P.~1981!. “Classification system for reporting eve involving human malfunction.” Riso-M–2240, as referenced in wan, B. ~1994!. A Guide to Practical Human Reliability Assessm, Burgess Science, Basingstoke, U.K.

Rasmussen, J., Pejtersen, A. M., and Goodstein, L. P.~1994!. Cognitive system engineering, Wiley, New York.

Reason, J. T.~1990!. Human error, Cambridge University Press, Ne York.

Rigby, L. ~1970!. “The nature of human error.”Annual Technical Con

JOURNAL OF CONSTRUCT

ference Transactions of the American Society for Quality Con, 475–566.

Sterman, J.~2000!. Business Dynamics, McGraw-Hill, New York. Suraji, A., Duff, A. R., and Peckitt, S. J.~2001!. “Development of caus

model of construction accident causation.”J. Constr. Eng. Manage, 127~4!, 337–344.

Thomassen, M. A., Sander, D., Barnes, K. A., and Nielsen, A.~2003!. “Experience and Results from Implementing Lean Construction Large Danish Contracting Firm.”Proc., 11th Annual Conference Lean Construction, July 22–24, Blacksburg, Va, 644–655.

Toole, M. ~2002!. “Construction site safety roles.”J. Constr. Eng. Man age., 128~3!, 203–210.