See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320237264

Validity and consistency assessment of accident analysis methods in the

petroleum industry

Article  in  International journal of occupational safety and ergonomics: JOSE · October 2017

DOI: 10.1080/10803548.2017.1387400

CITATIONS READS

7 574

4 authors, including:

Omran Ahmadi Ali Khavanin

Tarbiat Modares University Tarbiat Modares University
43 PUBLICATIONS   245 CITATIONS    137 PUBLICATIONS   1,545 CITATIONS   

SEE PROFILE SEE PROFILE

Hamidreza Mokarami
Shiraz University of Medical Sciences
82 PUBLICATIONS   528 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Assessing the risk factors for developing work-related musculoskeletal disorders during dairy farming View project

Optimization of Electrospun Nanofiber for Application in Air Filtration View project

All content following this page was uploaded by Hamidreza Mokarami on 16 July 2018.

The user has requested enhancement of the downloaded file.

 International Journal of Occupational Safety and
Ergonomics

ISSN: 1080-3548 (Print) 2376-9130 (Online) Journal homepage: http://www.tandfonline.com/loi/tose20

Validity and consistency assessment of accident

analysis methods in the petroleum industry

Omran Ahmadi, Seyed Bagher Mortazavi, Ali Khavanin & Hamidreza

Mokarami

To cite this article: Omran Ahmadi, Seyed Bagher Mortazavi, Ali Khavanin & Hamidreza
Mokarami (2017): Validity and consistency assessment of accident analysis methods in
the petroleum industry, International Journal of Occupational Safety and Ergonomics, DOI:
10.1080/10803548.2017.1387400

To link to this article: http://dx.doi.org/10.1080/10803548.2017.1387400

Accepted author version posted online: 05

Oct 2017.

Submit your article to this journal

Article views: 8

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tose20

Download by: [UNIVERSITY OF ADELAIDE LIBRARIES] Date: 10 November 2017, At: 14:14
 Publisher: Taylor & Francis & Central Institute for Labour Protection – National Research Institute (CIOP-
PIB)
Journal: International Journal of Occupational Safety and Ergnomics
DOI: 10.1080/10803548.2017.1387400

Title:

Validity and consistency assessment of accident analysis methods

in the petroleum industry

Authors:
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Omran Ahmadi
Department of Occupational Health Engineering, Faculty of Medical Sciences, Tarbiat
Modares University, Tehran, Iran. email

Seyed Bagher Mortazavi

Department of Occupational Engineering, Faculty of Medical Sciences, Tarbiat Modares
University, Tehran, Iran.

Ali Khavanin

Department of Occupational Engineering, Faculty of Medical Sciences, Tarbiat Modares

University, Tehran, Iran.

Hamidreza Mokarami
Department of Ergonomics, School of health, Shiraz University of medical sciences, Shiraz,
Iran.

Corresponding Author: Seyed Bagher Mortazavi

Department of Occupational Engineering, Faculty of Medical Sciences, Tarbiat Modares
University, Tehran, Iran. mortazav@modares.ac.ir.

Acknowledgements
The authors would like to thank the assessor for participating in the present study.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This study is part of the first author’s PhD thesis and was conducted with financial support
from Tarbiat Modares University
 Validity and consistency assessment of accident analysis
methods in the petroleum industry
Omran Ahmadi a, Seyed Bagher Mortazavia*, Ali Khavanina, Hamidreza
Mokaramib
a
Department of Occupational Health Engineering, Faculty of Medical Sciences, Tarbiat
Modares University, Tehran, Iran.
b
Department of Ergonomics, School of health, Shiraz University of medical sciences,
Shiraz, Iran.
*
Corresponding Author: Seyed Bagher Mortazavi
Department of Occupational Engineering, Faculty of Medical Sciences, Tarbiat Modares
University, Tehran, Iran. mortazav@modares.ac.ir.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Abstract
Background: Accident analysis is the main aspect of accident investigation. It includes the
method of connecting different causes in a procedural way. Therefore, it is important to use
valid and reliable methods for the investigation of different causal factors of accidents,
especially the noteworthy ones.
Objective: The objective of this study was to prominent assess the accuracy (sensitivity
index-SI) and consistency of the six most commonly used accident analysis methods in the
petroleum industry.
Methods: In order to evaluate the methods of accident analysis, two real case studies
(process safety and personal accident) from the petroleum industry were analyzed by 10
assessors. Then, the accuracy and consistency of these methods were evaluated. The
assessors were trained in the workshop of accident analysis methods.
Results: The systematic cause analysis technique (SCAT) and bowtie methods gained the
greatest SI scores for both personal and process safety accidents, respectively. The best
average results of the consistency in a single method (based on 10 independent assessors)
were in the region of 70%.
Conclusion: This study confirmed that the application of methods with pre-defined causes
and logic tree could enhance the sensitivity and consistency of accident analysis.
Keywords: Accident analysis, accuracy, consistency, sensitivity index, petroleum industry.

1. Introduction
Annually, millions of occupational accidents occur in the world and cause considerable losses

to countries and organizations. Understanding the causes of these accidents will lead to the

implementation of better strategies to prevent their re-occurrence [1]. In other words,

understanding why accidents occur and how to prevent their reoccurrence is an essential part

of improving safety in any industry. Accidents that may seem as unfortunate incidents

(putting something in the wrong place at the wrong time), can be seen as the result of a chain

of failures and mistakes. To improve safety, conducting investigations plays a major role in
 efficient safety management and is fundamental to learning from accidents [2]. Accident

analysis is the core aspect of accident investigation and it includes the method of connecting

different causes in a procedural way. One of the most important steps in accident analysis is

the conscious selection of the accident analysis method. This is because of the fact that these

methods are based on the various models developed by the advances in the science of safety

over the past century. New insights into accidents have resulted in the development of novel

models and methods. The following factors or criteria are used to compare and select
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

accident analysis methods: it should be easy to understand and apply, guidance material

should be available and clear, should be present graphical format for the output, should be

valid (correspondence between the analysis findings and reality) and consistent (a reliable

method should facilitate the agreement between results and different assessors), etc. [3-10].

Based on recent studies, accuracy and consistency is one of the most important criteria for the

selection of accident analysis methods [11-16]. However, literature reviews have shown that

there is still a gap in studies on the accuracy and consistency of accident analysis methods

[5]. Therefore, the aim of this study was to determine the accuracy (sensitivity index-SI) and

consistency of six accident analysis methods: systematic cause analysis technique (SCAT),

sequentially timed events plotting (STEP), fault tree analysis (FTA), man, technology and

organisation (MTO), bowtie, and tripod-beta.

2. Materials and methods

2.1. Selection of accident analysis methods

Based on recent studies, the 16 most commonly used accident analysis methods are: accident

evolution and barrier function (AEB) [17], MTO [18], FTA [19], management oversight and

risk tree (MORT) [20, 21], SCAT [4, 5], tripod-beta [4, 22, 23], STEP [10], events and causal

factors analysis (ECFA) [4, 24], bowtie [25], Accimap [26-29], event tree analysis (ETA)
 [30], safety through organisational learning (SOL) [31], HSG 245 [5, 32], change analysis

(3CA) [32], systems theoretic accident modelling and processes model (STAMP) [33, 34],

and the human factors analysis and classification system (HFACS) [3, 35, 36]. Six of these

methods (SCAT, STEP, FTA, MTO, bowtie, and tripod-beta) were selected based on the

following criteria: the availability of software applications for software based methods, the

availability of methods manuals, and their application in the petroleum industry. The

characteristics and the major components of these six accident analysis methods are shown in
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Table 1.

2.2. Individuals training

The individuals chosen for this study lacked experience about the selected methods of

analysis. Hal et al. showed that factors like differences in knowledge and cultural experiences

can affect inter-observer reliability [37]. For this reason,10 post-graduate students of

occupational health, who were inexperienced but familiar with the background of the

accident analysis methods, were asked to participate in the study. Then, they were trained in a

workshop on accident analysis methods using 14 three-hour sessions. Two sessions were held

on accident investigation while the remaining sessions were on accident analysis methods.

2.3. Calculation of the accuracy and consistency of the selected methods

There are major components within each method which are the main sources of inaccuracy

and inconsistency. In this study, since the analytical component within each method was

different from each other, two components were used to calculate the SI and consistency of

the selected methods. The first component is the elements of analysis while the second

component is the decomposition tools employed by the methods (Table 2).

 To calculate the accuracy and consistency of the SCAT method, three elements of analysis

were considered, namely: direct, immediate, and basic causes. The elements of analysis that

were considered for the FTA method are the basic cause, intermediate event, undeveloped

event and gates. This method employs a logic tree for the accident composition, and applies a

very important rule in accident analysis. This can also be considered as a major component of

the source of inconsistency. The elements of analysis in the STEP method are: actor, event,

and the cause. This method uses diagrams for the accident decomposition to show the
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

accident sequence. The elements of analysis for the MTO method are the causes (deviation),

events, and barriers. This method uses diagrams from events. Thereafter, causes or the

deviation were connected to any event in the diagram. All failed or missing barriers were

placed under the events in the diagram. The elements of analysis for the tripod-beta method

which were considered as major component factors of accuracy and consistency are barrier,

immediate cause, precondition, underlying cause, agent, object, and event. This method uses

diagrams for the accident decomposition. The accuracy and consistency of the bowtie method

have been analyzed by focusing on five major components of accident analysis namely:

barrier, escalation factor, cause, event, and consequence. This method also uses diagrams for

the accident decomposition.

2.4. Accident analysis

To evaluate the methods of analysis, two real case studies from the petroleum industry were

investigated. Two accidents were analyzed by individuals using the trained methods through

the workshop. The first case was a process safety accident (fire occurred in the gas

purification unit tower) while the second case was a personal accident (individual stuck

between a lift-track and a truck). These accidents were randomly selected among 65 accidents

on which the collected information was completed.

 The two accidents were already analyzed by the petroleum industry safety expert committee

and the results were selected as a gold standard. Besides, as the assessors carried out the

analysis, some new causes (in addition to those recognized by the expert group and those that

actually occurred in the accidents) were identified and judged by the investigators to be valid

and relevant. Thus, a total pool of causes or as another analysis component for each accident

was defined, against which each method could be tested for its sensitivity or accuracy. The

sensitivity of each participant’s analysis was calculated using the signal detection paradigm,
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

since it provides a useful framework for testing the power of accident analysis. Moreover, it

has been effectively used as a human error identification method in the past [14]. Equation 1

was used to calculate SI [38]. SI was calculated for both personal and process safety accidents

for all the methods.

( ) (1)

Where:

SI; sensitivity index

Hit; an analysis component identified by the analyst that was also reported by gold expert

analysis.

Miss; an analysis component that was reported by the gold analysis or pool but was not

identified by analyst.

False alarm (FA); analysis component identified by the analyst but not by the gold analysis

or pool (was added to the pool).

Correct Rejection; this represents the number of errors contained in the human error

identification methods error mode taxonomy that were correctly rejected by the analyst and

also not reported by the survey respondents [15]. Since the correct rejection part of the
 equation was not applicable to the accident analysis method which is the same as the human

error method, the sensitivity of the methods was calculated by considering the false alarm

and correct rejection as equal to zero for all the methods.

These two categories (Hit and Miss) were entered into the signal detection grid for each

expert. The signal detection paradigm was then used to calculate the SI. This produced a

figure between 0 and 1, the closer the SI is to 1, the more accurate the method analysis is (the
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

ability to identify relevant incident errors and potential errors which could have occurred).

The inter-rater consistency of a method, defined as a component percentage in an accident

with 10 independent assessors using particular methods, was calculated for the total

component pool. To calculate the consistency of the selected methods, the total number of

correct components between assessors for both personal and process safety accidents were

considered.

3. Results

Based on personal accident analysis, the mean SI score for each method is as shown in Figure

1. The SCAT method achieved the greatest SI score at 0.79. The bowtie method with an SI

score of 0.74 occupied second place, while the other methods, including FTA, MTO, tripod-

beta, and STEP were ranked lower at 0.70, 0.68, 0.63, and 0.56, respectively.

The mean SI score based on process safety accident analysis is presented for each method as

shown in Figure 2. The SCAT method reached the highest score of 0.76 as compared to the

other methods. The FTA method had the second highest SI score at 0.73, while the figures for

bowtie, MTO, tripod-beta, and STEP methods were lower at 0.67, 0.66, 0.64, and 0.59,

respectively.

Since the opportunity for inconsistencies is inherent in each of these steps, the mean level of

consistency across all pairs of assessors was used to calculate the inter-rater consistency.

Using a single method, the results obtained for the 10 assessors are as shown in Figure 3.
 The best mean results for a single method (based on 10 independent assessors) were observed

in the region of 70%. Assessors using SCAT method were the most consistent at accident

analysis with 70%. The intra-rater consistency for the 10 assessors using FTA, STEP, tripod-

beta, bowtie, and MTO was 0.68, 0.64, 0.53, 0.58, and 0.52, respectively.

4. Discussion

The accuracy of the analysis methods in both personal and process safety accidents did not
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

differ considerably and the SI for the entire cases was above the average (0.50). This study

confirmed that the SCAT method is capable of providing accurate evidence in the accident

analysis. This can be attributed to the pre-defined causes in method for the accident analysis

that consequently guides in accident analysis. Furthermore, the FTA method offered more

accurate results as compared to the other methods; this is mainly due to the logic tree used by

this method. This logic tree can navigate the assessor by selecting logic components in the

accident analysis. The SI of these two methods were higher for process accident than personal

accident which can be assigned to the inherently easy uses of logic tree to process the fault in

contrast to the personal failure or fault.

Another objective of this study was to determine the inter-rater consistency of the selected

methods. It was observed that the assessors who used SCAT recorded the greatest

consistency of the accident analysis at 70%, while the figure for the FTA method was the

second highest as compared to the other methods. These findings revealed that the utilization

of pre-defined causes and logic tree by these methods can lead to similar results for accident

analysis by different assessors.

Although the SCAT and FTA provided more accurate results when compared with other

studied methods; there are still many other factors that should be taken into consideration [3-

10]. Although, this study did not provide enough information concerning the optimal method

of selection, it only offered acceptable evidence proving the positive effect of the pre-defined
 causes and logic tree on the result accuracy in the accident analysis. Therefore, these can be

incorporated in future studies in order to select the most appropriate accident analysis

methods. Depending on the type of organization, the weight and importance of criteria for

selecting the accident analysis method may differ. Also, the results of this study showed that

although the SI and consistency of tripod-beta was lower as compared to the other methods, it

had a greater ability to analyze petroleum industry accidents compared to the SCAT, FTA,

MTO, bowtie, and STEP methods [39].

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

In a study, Katsakiori et al. [5] showed that there were no validation studies in literature to

investigate the validity and reliability of the accident analysis methods, except for the tripod-

beta method. Munson compared accident investigation methods for wild land firefighting

incidents using the case study method with five assessors. After evaluating the case study, the

methods were rated by the assessor. The calculated realistic scores for the STEP and FTA

analysis methods were 9 (90%) and 8 (80%), respectively. The figures for the consistency of

the STEP and FTA analysis methods were 10 (100%) and 9 (90%), respectively. The realistic

and consistency figures for the STEP and FTA methods were higher than those calculated in

the present study. This result shows that the obtained values for the validity and consistency

in the present study, based on the assessors’ analysis results, were close to reality [10].

The limitation of this study lies in the fact that it was conducted using past data (occurred

accidents). With the purpose of the accurate prediction of future accident scenarios in mind,

the used past data was based on the assumption that nothing would change. However, after

analyzing the accidents, it was found that things changed. Therefore, it seemed unattainable

to provide suitable conditions to test the validity of the accident analysis method, for the

prediction of future accident scenarios [40].

5. Conclusion
 This study confirmed that the application of methods with pre-defined causes and logic tree

could enhance the sensitivity and consistency of accident analysis. Examples of methods with

pre-defined causes and logic tree include SCAT and FTA respectively. These features of

accurate methods can be used for further investigation with the purpose of improving the

accuracy and consistency of the methods. It is recommended that further research needs to be

conducted to enhance accident analysis accuracy. Moreover, further applications of the

accident analysis method are encouraged within the petroleum industry. Although, the
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

present research findings confirmed that methods with pre-defined causes or logic tree can

increase the sensitivity of accident analysis; further research is still required to enhance the

accuracy of methods that cover the other organizations criteria. Using the present research

findings which corroborate the higher accuracy and consistency of some methods, it is

possible to select the accident analysis method in a more efficient way.

References
[1] Johnson C, Holloway CM. A survey of logic formalisms to support mishap analysis.
Reliab Eng Syst Safety. 2003;80(3):271-91.
[2] Stoop JA. Independent accident investigation: a modern safety tool. J Hazard Mater.
2004;111(1):39-44.
[3] Salmon PM, Cornelissen M, Trotter MJ. Systems-based accident analysis methods: A
comparison of Accimap, HFACS, and STAMP. Safety Sci. 2012;50(4):1158-70.
[4] Sklet S. Comparison of some selected methods for accident investigation. J Hazard Mater.
2004;111(1):29-37.
[5] Katsakiori P, Sakellaropoulos G, Manatakis E. Towards an evaluation of accident
investigation methods in terms of their alignment with accident causation models. Safety Sci.
2009;47(7):1007-15.
[6] Kontogiannis T, Leopoulos V, Marmaras N. A comparison of accident analysis techniques
for safety-critical man–machine systems. Int J Ind Ergon. 2000;25(4):327-47.
[7] Center for Chemical Process Safety. Guidelines for investigating chemical process
incidents. 2nd ed. New York (NY): Wiley; 2003.
[8] Nivolianitou Z, Leopoulos V, Konstantinidou M. Comparison of techniques for accident
scenario analysis in hazardous systems. J Loss Prev Process Ind. 2004;17(6):467-75.
[9] Morrison LM. Best practices in incident investigation in the chemical process industries
with examples from the industry sector and specifically from Nova Chemicals. J Hazard
Mater. 2004;111(1):161-6.
 [10] Munson S. Assessment of accident investigation methods for wildland firefighting
incidents by case study method [master’s thesis]. Montana (MT): University of Montana;
2000.
[11] Kirwan B. The validation of three human reliability quantification techniques—
THERP, HEART and JHEDI: part iii—Practical aspects of the usage of the techniques. Appl
Ergon. 1997;28(1):27-39.
[12] Kirwan B. Validation of human reliability assessment techniques: part 1—validation
issues. Safety Sci. 1997;27(1):25-41.
[13] Kirwan B. Human error identification in human reliability assessment. Part 2:
Detailed comparison of techniques. Appl Ergon. 1992;23(6):371-81.
[14] Baber C, Stanton NA. Human error identification techniques applied to public
technology: predictions compared with observed use. Appl Ergon. 1996;27(2):119-31.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

[15] Stanton NA, Salmon P, Harris D, Marshall A, Demagalski J, Young MS, et al.
Predicting pilot error: testing a new methodology and a multi-methods and analysts approach.
Appl Ergon. 2009;40(3):464-71.
[16] Power C, Fox J. Comparing the comprehensiveness of three expert inspection
methodologies for detecting errors in interactive systems. Safety Sci. 2014;62:286-94.
[17] Svenson O. On models of incidents and accidents. Paper presented at: 7th European
Conference on Cognitive Science Approaches to Process Control; 1999 September 21-24;
Villeneuve d’Ascq, France.
[18] Andersson O, Rollenhagen C. The MTO concept and organisational learning at
Forsmark NPP, Sweden. Paper presented at: IAEA International Conference on Safety Culture
in Nuclear Installations; 2002 Dec 2-6; Rio de Janeiro, Brazil.
[19] Ferry TS. Modern accident investigation and analysis. New York (NY): Wiley; 1988.
[20] Johnson WG. MORT safety assurance systems. New York (NY): Marcel
Dekker.1980.
[21] Knox NW, Eicher RW. Mort User's Manual: For use with the Management Oversight
and Risk Tree analytical logic diagram. Idaho Falls (ID): System Safety Development Center;
1992. (DOE/ SSDC-76-45/4-Rev.3; no. DE92013785).
[22] Wagenaar WA, Groeneweg J, Hudson P, Reason J. Promoting safety in the oil
industry. The ergonomics society lecture presented at the ergonomics society annual
conference, Edinburgh, 13-16 April 1993. Ergonomics. 1994;37(12):1999-2013.
[23] Gower-Jones A, van der Graf G. Experience with Tripod BETA Incident Analysis.
Paper presented at: Society of Petroleum Engineers: International Conference on Health,
Safety, and Environment in Oil and Gas Exploration and Production; 1998 June 7-10; Caracas,
Venezuela.
[24] Buys J, Clark J. Events and causal factors charting. Idaho Falls (ID): SCIENTECH,
Inc.; 1995.

[25] Khakzad N, Khan F, Amyotte P. Dynamic risk analysis using bow-tie approach.
Reliab Eng Syst Safety. 2012;104:36-44.
[26] Hopkins A. Lessons from Longford: the Esso gas plant explosion. Sydney (Australia):
CCH Australia; 2000.
[27] Johnson C, de Almeida IM. Extending the borders of accident investigation: applying
novel analysis techniques to the loss of the Brazilian space launch vehicle VLS-1 V03. Safety
Sci. 2008;46(1):38-53.
 [28] Svedung I, Rasmussen J. Graphic representation of accident scenarios: mapping
system structure and the causation of accidents. Safety Sci. 2002;40(5):397-417.
[29] Rasmussen J, Suedung I. Proactive risk management in a dynamic society. Karlstad
(Swed( : Swedish Rescue Services Agency; 2000.
[30] Energy Institute. Guidance on Investigating and Analysing Human and Organisational
Factors Aspects of Incidents and Accidents. London (UK): Energy Institute; 2008.
[31] Fahlbruch B, Schöbel M. SOL–Safety through organizational learning: A method for
event analysis. Safety Sci. 2011;49(1):27-31.
[32] Doidge JP. Successful health and safety management. Perfusion. 1997;12:217-20.
[33] Qureshi ZH. A review of accident modelling approaches for complex socio-technical
systems. in: Cant T, editor. Proceedings of the twelfth Australian workshop on Safety critical
systems and software and safety-related programmable systems; 2007 Aug 30-31; Adelaide,
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Australia. Darlinghurst (Australian): Australian Computer Society, Inc. 2007;86:47-59.

[34] Syvertsen R-AH. Modeling the Deepwater Horizon blowout using STAMP [master’s
thesis]. Trondheim: Norwegian University of Science and Technology; 2012.
[35] Shappell SA, Wiegmann DA. A human error approach to aviation accident analysis:
The human factors analysis and classification system. Ann Arbor (MI): Ashgate Publishing
Limited; 2003.
[36] Olsen NS. Coding ATC incident data using HFACS: Inter-coder consensus. Safety
Sci. 2011;49(10):1365-70.
[37] Bird FE, Germain GL, Veritas N. Practical loss control leadership. Loganville (GA):
International Loss Control Institute; 1985.
[38] Stanton NA, Harris D, Salmon PM, et al. Predicting design induced pilot error using
HET (Human Error Template)–A new formal human error identification method for flight
decks. The Aeronautical J. 2006;110(1104):107-15.
[39] Ahmadi O, Mortazavi SB, Khavanin A. [Selection of the optimal method for analysis
of accidents in petroleum industry using fuzzy ANP and TOPSIS multi criteria decision
methods]. Iran Occupational Health. 2017;14(2):148-57. Persian
[40] Wagenaar WA, Van Der Schrier J. Accident analysis: The goal, and how to get there.
Safety Sci. 1997;26(1-2):25-33.
[41] Hendrick K, Benner L. Investigating accidents with STEP. New York (NY): CRC
Press; 1986.
[42] Groeneweg J. Controlling the controllable. The management of safety. 3rd ed. Leiden
(Netherlands): DSWO Press;1996.
[43] Chevreau FR, Wybo JL, Cauchois D. Organizing learning processes on risks by using
the bow-tie representation. J of hazard mater. 2006; 31;130(3):276-83.
[44] Pilot S. What is fault tree analysis. Quality Progress. 2002;35(3):120.
 Fig. 1. Sensitivity Index score (based on personal accident analysis)
Note: FTA = fault tree analysis; MTO = man, technology and organisation; SCAT =
systematic cause analysis technique; STEP = sequentially timed events plotting.

Fig. 2. Sensitivity Index score (based on process accident)

Note: FTA = fault tree analysis; MTO = man, technology and organisation; SCAT =
systematic cause analysis technique; STEP = sequentially timed events plotting.

Fig. 3. Accident analysis methods intra-rater consistency

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Note: FTA = fault tree analysis; MTO = man, technology and organisation; SCAT =
systematic cause analysis technique; STEP = sequentially timed events plotting.

Table 1. Summary of the characteristics of the selected six accident analysis methods
Method Characteristic
STEP STEP [41] views accidents as multiple ways of causal factors that react to the previous factors and may interact with
others throughout the system to finally result in an accident. STEP is a comprehensive approach for reconstruction of an
accident. STEP methods used worksheet which is the documentation that organizes data gathering and analysis. It
explains the beginning and end of the accident sequence along columns that represent time. The actors are listed in rows
of the worksheet. Actors defined as people or things acted to produce the harmful outcome. Each actor does an action,
 called an event, that shows the interactions among actors and events when displayed on worksheet. The method
accommodated events that occur at the same time subsequently. Each event is displayed by a block diagram. Finally,
STEP indicates safety problems and development of safety recommendations.
SCAT SCAT method is based on the ILCI Loss Causation Model [7,37]. It facilitates investigation of events through the use of
chart. Chart acts as pre-defined causes to ensure that investigator looks at all aspects of the accident. There are five
blocks in the chart. Therefore, the primary block contains space to explain the accident. The second block indicates the
common contact with energy, substance and other parameters that result in event. The third block indicates the
immediate causes of accident and the fourth block of the chart indicates the basic causes. Finally, the last block
generally indicates the successful control program that is important in losses.
Tripod-beta The origin of Tripod-beta analysis method is Tripod that is based on Reason model [42]. Based on this model,
organizational failures are the main factors in accidents as latent factors remain for a long time in the system and always
followed by a number of technical and human errors when contributing to an accident. In Tripod method, occurrence of
an accident is because of missing or failing controls and barriers. Unsafe acts (active failures) are visible and easily
identified in case barriers failed. Unsafe acts are generated by underlying mechanisms acting in organizations. These
mechanisms are called basic risk factors (BRF) and they cover human, organizational and technical problems. The
Tripod-beta is a software tool that merges two models, HEMP and the Tripod. In the Beta model, the barrier and
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

controls are directly linked to unsafe acts, preconditions and latent failures. In Tripod model, barrier and controls are
directly linked to the BRF.
Bowtie Bowtie is one of the best graphical methods models that complete accident scenario from initiate causes to end
consequences. It consists of fault tree and event tree models. These two methods are attached to each other through
FTA top event as an initiating event tree [43]. Bowtie method develops more realistic understanding of the relationship
between causes in the accident, its consequences and barriers that could prevent accruing event in any stage of an
accident. In addition to determining the root causes of accidents, bowtie method is fully efficient in determining the
critical tasks to ensure the integrity and effectiveness of current controls. It also has important role in identifying the key
performance indicators of HSE tasks and their improvements.
MTO MTO analysis method is based on three methods including change analysis, event and cause diagram and barrier
analysis. The bases for MTO method is human, organizational and technical factors considered in an accident [4]. The
first step in MTO is creating diagram from events. Then, possible human and technical factors are identified and
vertically connected to any event in the diagram. Finally, technical, organization or human barriers are analyzed which
failed or missing following the accident. All failed or missing barriers are placed under the events in the diagram. The
last important step is to provide recommendation.

FTA FTA is a logical method for evaluation of an accident resulting from sequences and combinations of events. This
method was first introduced by the Bell Laboratories [44]. FTA first defines a top event which is a failure or accident.
Then, it builds the sequence of events leading to that failure or accident through applying a top-down logic (deductive
approach) by two logic gates (the AND OR gate). Finally, FTA method represents graphical combination of causes for
the accident. FTA analysis may be applied in quantitative, qualitative, or both manners [4].
Note: FTA = fault tree analysis; MTO = man, technology and organisation; SCAT = systematic cause analysis technique; STEP =
sequentially timed events plotting.

Table 2. Major components for the selected six accident analysis methods
Component Method

Bowtie Tripod-beta MTO STEP FTA SCAT

Elements of - Barrier - Barrier - Cause - Actor - Basic cause - Direct cause

analysis - Escalation - Immediate cause (deviation) - Event - Intermediate event - Immediate
factor - Precondition - Event - Cause - Undeveloped cause
- Cause - Underlying cause - Barrier event - Basic causes
- Event - Agent - Gate" or" & "and"
- Consequence - Object
- Event
 Decomposition Diagram Diagram Diagram Diagram Logic fault tree -
(worksheet)
Note: FTA = fault tree analysis; MTO = man, technology and organisation; SCAT = systematic cause analysis technique; STEP =
sequentially timed events plotting.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Sensitivity index score

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9

FTA
SCAT
STEP
Tripod-beta
Accident analysis method
Bowtie
MTO
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

Sensitivity index score

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8

FTA
SCAT
STEP
Tripod-beta
Accident analysis method
Bowtie
MTO
 0.8

0.7
Intra-rater consistency

0.6

0.5

0.4

0.3

0.2

0.1
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 14:14 10 November 2017

0
FTA SCAT STEP Tripod-beta Bowtie MTO
Accident analysis method

View publication stats