SYSTEMIC APPROACH TO IMPROVE LEARNING FROM INCIDENTS

A Thesis

by

JULIANA ANDREA GUARGUATI ARIZA

Submitted to the Office of Graduate and Professional Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

Chair of Committee, M. Sam Mannan

Committee Members, Mahmoud El-Halwagi
S. Camille Peres
Head of Department, M. Nazmul Karim

May 2017

Major Subject: Safety Engineering

Copyright 2017 Juliana Andrea Guarguati Ariza

 ABSTRACT

Learning from incidents has always been very challenging for industry and

regulators due to the associated complexity in human and organizational behavior. In

spite of the efforts and resources spent on incident investigation processes and

organizational learning techniques, the outcome has not shown a significant

improvement. Chemical process incidents such as Texas City Refinery explosion, T2

Laboratories explosion, and West Fertilizer explosion continue to occur. It is evident that

as an industry, organizations are continually failing to learn from the past incidents. The

question is if organizations are ready to learn, are they following the right learning

approach? Organizations should adopt a systemic learning approach where the collected

knowledge is leveraged in order to enhance the performance of the safety management

systems and consequently prevent future incidents. The objective of this research is set

to enhance our understanding on how a company and the industry as a whole learn from

past incidents and define the key elements to improve the systemic learning. This study

was divided into four main phases: identification of the learning system, development of

a learning process and a proposed incident investigation process and finally the

validation of the proposed incident investigation process through a case study.

First, the learning system for the chemical and oil and gas industry, the different

types of learning, and the entities involved in it has been characterized. Based on the

results, some limitations on the system were identified and discussed. Secondly, a

systemic process for improving learning from incidents has been developed based on the

ii
identified limitations of the learning system. The proposed process provides a holistic

view of the learning process and explored the concept of knowledge management into

safety systems. Likewise, it provides guidelines of how learning systems can be

executed within the organization and how to support the implementation of safety

knowledge inside of it. Third, an incident investigation process has been proposed to

provide additional sources of information into the analysis, in order to support the

identification of root causes and the required changes in the management systems of the

organization. The process that has been developed enhances understanding of how to get

information and transform it into valuable recommendations that can be implemented in

their processes. Finally, the proposed process has been explained through a case study.

The obtained results provide a clear picture of how an incident investigation report can

be enhanced through this process.

iii
 DEDICATION

To

My parents: Samuel Guarguati and Martha Ariza, and

My boyfriend: Gabriel Ortiz,

for all your love and support.

iv
 ACKNOWLEDGEMENTS

First of all, I would like to express my sincere gratitude to my advisor and

mentor, Dr. Sam. Mannan. His guidance and continuous support gave me the confidence

to aspire and continually push myself to accomplish goals in my professional career. I

also want to thank him for giving me the opportunity of being part of the Mary Kay

O’Connor Process Safety Center at which I had the opportunity to meet inspiring and

wonderful people who taught me and supported me during this journey. I also would like

to thank my committee members, Dr. Mahmoud El-Halwagi and Dr. Peres, for taking

their time to serve on my committee. Especially, I would like to thank Dr. Peres for her

guidance, support, and the time we spent together trying to make this real.

I want to thank Dr. Noor Quddus for his support in all stages of my research. His

supervision, constructive comments and suggestions helped me shape this thesis. I am

also thankful to Valerie Green for all her administrative support, patience, and personal

guidance when I needed them. In addition, I want to extend gratitude to my sponsor

Colfuturo for providing me the financial support to accomplish my dream.

I also want to thank my friends at Texas A&M University for all the special

moments we spent together. I am grateful to my boyfriend for all his support and the

happiness he brings to my life. Finally, I would like to thank my parents who made my

dream real, for motivating me to be a better person, and their unconditional love,

guidance, and support.

v
 CONTRIBUTORS AND FUNDING SOURCES

This work was supervised by a dissertation committee consisting of Professor M.

Sam Mannan [advisor and chair] of the Department of Chemical Engineering and

Professor Mahmoud El-Halwagi of the Department of Chemical Engineering and

Professor S. Camille Peres of the Department of Public Health. All work for the

dissertation was completed independently by the student.

Graduate study was supported by the Harry West fellowship from Texas A&M

University and the sponsor COLFUTURO from Colombia.

vi
 TABLE OF CONTENTS

Page

ABSTRACT ...................................................................................................................... ii

DEDICATION .................................................................................................................. iv

ACKNOWLEDGEMENTS ............................................................................................... v

CONTRIBUTORS AND FUNDING SOURCES............................................................. vi

TABLE OF CONTENTS .................................................................................................vii

LIST OF FIGURES ............................................................................................................ x

LIST OF TABLES ............................................................................................................ xi

1. INTRODUCTION .......................................................................................................... 1

1.1. Motivation ............................................................................................................... 1

1.2. Background ............................................................................................................. 4
1.3. Objectives ................................................................................................................ 8
1.4. Thesis organization ................................................................................................. 9

2. ORGANIZATIONAL LEARNING ............................................................................. 12

2.1. Organizational learning ......................................................................................... 12

2.2. Single and double loop learning ............................................................................ 14
2.3. Organizational processes ....................................................................................... 16
2.3.1. Knowledge...................................................................................................... 16
2.3.2. Knowledge transfer ........................................................................................ 17
2.3.3. Organizational memory .................................................................................. 19
2.3.4. Organizational forgetting ............................................................................... 21
2.4. Barriers of organizational learning........................................................................ 21
2.5. Learning methods .................................................................................................. 25
2.6. Factors that influence learning .............................................................................. 31

3. INCIDENT INVESTIGATION ................................................................................... 34

3.1. Introduction ........................................................................................................... 34

3.2. Incident investigation terminology ....................................................................... 37
3.3. Incident investigation history ................................................................................ 40
3.4. Incident investigation team ................................................................................... 43

vii
 3.5 Incident investigation process ................................................................................ 45
3.5.1. Reporting ........................................................................................................ 46
3.5.2. Collecting evidence ........................................................................................ 47
3.5.3. Analyzing the evidence .................................................................................. 48
3.5.4. Developing recommendations ........................................................................ 49
3.5.5. Report and lessons learned ............................................................................. 50
3.6 Incident investigation methodologies .................................................................... 50
3.6.1. Fault tree analysis ........................................................................................... 50
3.6.2. STEP (Sequential Timed Events Plotting) ..................................................... 52
3.6.3. FRAM (Functional Resonance Analysis Method) ......................................... 54
3.6.4. Tripod BETA .................................................................................................. 55
3.6.5. IPICA (Integrated Procedure of Incident Cause Analysis) ............................ 56
3.6.6. TapRooT® System ......................................................................................... 57
3.6.7. MORT (Management Oversight and Risk Tree) ............................................ 58
3.6.8. STAMP (Systems-Theoretic Accident Model and Processes) ....................... 60
3.7. Developing quality incident investigation reports ................................................ 61

4. LEARNING PROCESS ............................................................................................... 64

4.1. Learning system .................................................................................................... 65

4.1.1. Limitations in the learning system ................................................................. 67
4.2. DIKM model ......................................................................................................... 72
4.3. Learning process ................................................................................................... 74
4.3.1. Corporate learning team ................................................................................. 76
4.3.2. Flow of information........................................................................................ 77
4.3.3. Learning process description .......................................................................... 79
4.4. Additional elements to achieve organizational learning ....................................... 86
4.4.1. Effective training programs ............................................................................ 88
4.4.2. Effective training courses ............................................................................... 88
4.4.3. Supporting unlearning processes .................................................................... 90

5. ENHANCE INTERNAL INFORMATION ................................................................. 93

5.1. Task analysis ......................................................................................................... 93

5.1.1. Hierarchical task analysis decomposition ...................................................... 94
5.1.2. Hierarchical task analysis results ................................................................... 94
5.1.3. Subtasks analysis .......................................................................................... 105
5.1.4. Task analysis remarks .................................................................................. 109
5.2. Proposed incident investigation process ............................................................. 110

6. CASE STUDY ........................................................................................................... 115

6.1. Incident description ............................................................................................. 115

6.1.1. Platform fire and explosion .......................................................................... 115
6.2. Incident analysis .................................................................................................. 119
viii
 6.2.1. Reviewing external incidents ....................................................................... 119
6.2.2. Reviewing internal incidents ........................................................................ 127
6.2.3. Identifying changes required in the management system and updating
training material ..................................................................................................... 133
6.2.4. Closing up the process .................................................................................. 135
6.2.5. The case study into the learning process ...................................................... 136

7. CONCLUSIONS AND FUTURE WORK ................................................................ 146

7.1. Conclusions ......................................................................................................... 146

7.2. Future work ......................................................................................................... 149

REFERENCES ............................................................................................................... 151

ix
 LIST OF FIGURES

Page

Figure 1 Thesis methodology ........................................................................................... 11

Figure 2 Units of learning ................................................................................................ 14

Figure 3 Single-loop and double-loop learning, adapted from [39] ................................. 15

Figure 4 Training methods, adapted from [27] ................................................................ 27

Figure 5 Swiss Cheese model, adapted from [71] ............................................................ 35

Figure 6 Safety pyramid, adapted from [75] .................................................................... 36

Figure 7 FTA symbols, adapted from [77] ....................................................................... 51

Figure 8 Basic representation of a FTA, adapted from [81] ............................................ 52

Figure 9 Basic STEP diagram, adapted from [93] ........................................................... 53

Figure 10 FRAM Function diagram, adapted from [94] .................................................. 54

Figure 11 Learning system, adapted from [90] ................................................................ 66

Figure 12 DIKW Model, adapted from [102] .................................................................. 73

Figure 13 Flows of information in the corporate learning system ................................... 78

Figure 14 Learning process .............................................................................................. 80

Figure 15 Improving organizational learning................................................................... 87

Figure 16 Task analysis decomposition part 1 ................................................................. 95

Figure 17 Task analysis decomposition part 2 ................................................................. 96

Figure 18 Task analysis decomposition part 3 ................................................................. 97

Figure 19 Proposed incident investigation process ........................................................ 112

Figure 20 US Gulf of Mexico offshore oil and gas pipeline types of failures over the
period from 1970 to 1999, adapted from [116] .............................................. 122

Figure 21 Reviewing internal incidents ......................................................................... 129

x
 LIST OF TABLES

Page

Table 1 Hierarchical task analysis part 1 ......................................................................... 98

Table 2 Hierarchical task analysis part 2 ......................................................................... 99

Table 3 Hierarchical task analysis part 3 ....................................................................... 100

Table 4 Hierarchical task analysis part 4 ....................................................................... 101

Table 5 Hierarchical task analysis part 5 ....................................................................... 102

Table 6 Hierarchical task analysis part 6 ....................................................................... 103

Table 7 Hierarchical task analysis part 7 ....................................................................... 104

Table 8 Action plan - Identification phase ..................................................................... 141

Table 9 Action plan - Implementation at a corporate level ............................................ 142

Table 10 Action plan - Implementation at a facility level.............................................. 143

xi
 1. INTRODUCTION

1.1. Motivation

In the chemical and oil and gas industry, organizations and regulators have spent

a significant effort into incident investigation processes and organizational learning

techniques in order to improve continually, avoid the same mistakes, and consequently

prevent incidents. In spite of these efforts, the outcome has not shown a significant

improvement and process safety incidents continue to occur. According to the Bureau of

Labor Statistics [1], over the five years period from 2010 to 2014, there were 758

fatalities in the chemical and oil and gas industry (the statistics covers the following

sectors: manufacturing, oil and gas extraction, drilling oil and gas wells, support

activities for oil and gas operations, and petroleum refineries) [1]. Likewise, in 2014, an

estimated 46,200 nonfatal injuries and illnesses were reported [1]. These statistics

suggest that the industry as a whole is failing to learn from the past incidents. Learning

from incidents provides organizations the required knowledge to improve safety

management systems and prevent future incidents, by understanding what went wrong in

past incidents, analyzing their lessons learned and implementing the necessary changes

into their processes.

During the last three decades, the industry has seen significant process safety

incidents that have been precursors of changes in regulations and standards in United

States and the rest of the world. However, there is a failure to successfully implement

and enforce those changes and make them transcend over the years. Thus, process safety

1
incidents become part of the history and the industry is doomed to repeat the same

mistakes. Without doubt, no process safety incident has been as harmful as Bhopal

disaster in 1984, due to the enormous impact in terms of fatalities, financial losses,

environmental damage, and industry reputation. This incident made organizations realize

the importance of learning from incidents not only inside the organization, but also

learning from external incidents that may be applicable to their operations. While no

other incident has been as devastating as Bhopal incident, there have been several

disasters over the last decades that have had significant impact on the industry and from

which the same mistakes can be identified over and over again.

During the same year as Bhopal disaster, in San Juan, Mexico City a catastrophic

fire and series of explosions at an LPG terminal killed over 500 people, injured 7,000

and more than 200,000 people were evacuated [2]. The day of the incident, LPG leaked

from a pipeline rupture, which formed a vapor cloud that dispersed into the surrounding

for about 10 minutes. The vapor cloud ignited and a flash fire resulted causing a violent

shock, resulting in a series of BLEVEs and minor explosions in nearby houses and

facilities in the area [3]. Some of the causes and contributing factors of this incident

were that no hazard identification process was carried out for the unit and there was a

lack of awareness regarding the associated hazard not only from the organization but

also the community and the emergency responders. Likewise, there was a lack of

effective land use planning due to the lack of regulations to control the construction of

residential houses around the facility. In addition, the incident showed an inadequate

mechanical integrity program as one of the root causes of the incident. Moreover, there

2
was a lack of emergency response planning and training regarding hazard identification.

Due to the substantial impact of this incident, the same causes and contributing factors

were expected to be learned collectively within the organization. However, this

organization reveals a failure to learn from its own history. The safety performance of

the organization over the last ten years exposes an estimated 197 incidents and 21

fatalities per year [4, 5]. In these process safety incidents mechanical integrity,

inadequate land use planning, and inadequate emergency response were identified as

contributing factors.

Similarly, process safety incidents such as West Fertilizer explosion (2013) in

which a fire and explosion of ammonium nitrate killed 15 people including 12

emergency responders, injured more than 260 people and severely damaged the nearby

community [6]. Another incident, Tianjin explosion (2015) in which a series of

explosions in a warehouse killed 173 people, 110 of them first responders, injured more

than 797 people and damaged around 17,000 houses [7]. In both incidents, the same

contributing factors as Mexico City disaster such as poor hazard awareness, inadequate

land use planning, and inadequate emergency planning were identified. Even though

these incidents do not have the same causes as in Mexico City incident, it is evident that

the same contributing factors increased their severity. Both examples illustrate a lack of

organizational memory and an inability to effectively share and implement lessons

learned.

As can be seen, organizations have failed to recognize and learn effectively from

past process safety incidents. In addition, the knowledge that is embedded into those

3
incidents has been lost over the years, making it difficult for new generations to

understand the reasons behind the implementation of those lessons learned. The

incidents showed in the previous examples reflect a lack of an effective management

system, which enables organizations to learn, adapt and grow [8]. This suggests having

an effective learning system that serves as a source of knowledge by the acquisition of

relevant information, the analysis of it and the use of this information to improve and

prevent process safety incidents. The question is if organizations have the required

information, are they following the right learning approach? Organizations should adopt

a systemic learning approach where the collected safety knowledge is leveraged in order

to enhance safety management systems and make the lessons learned part of their

culture.

1.2. Background

The study of how organizations learn from incidents and manage safety to

prevent process safety incidents is not a new topic within the process safety research

area. However, this field has gained more attention during the last sixteen years and

more disciplines such as psychology, sociology, and engineering have been studying it.

As a result, learning from incidents has become a fragmented field in which different

approaches have been proposed for specific scenarios or situations such as a certain part

of the learning process, the group of interest or the industry in which it is applied. Back

in the 70s, the theory on organizational learning proposed by Argyris and Schön

highlighted the importance of learning to detect and respond to undesirable events [9].

4
The theory suggested two modes of learning: single and double-loop learning. Even

though, both modes of learning are necessary, the authors emphasized the importance of

achieving double-loop learning to identify organizational and systemic failures [10].

This theory is still accepted and has been widely recognized in this field despite current

advances in organizational learning theories [11]. Lukic ‘research supports this theory by

the analysis of the type of learning that is adopted in organizations, in which incidents

usually are a combination of technical, human, and organizational factors, making single

and double-loop essential factors to accomplish successful learning [12]. Likewise,

Choularton demonstrated how double-loop learning could be to avoid superficial

learning [13].

Over the years, learning from incidents has been categorized into several areas of

study such as lessons learned, incident investigation and analysis, learning from incident

process, and conditions for learning [11]. Kletz has been one the pioneers of introducing

lessons learned from process safety incidents into organizational systems and

introducing real examples of how organizations failed and how to overcome these types

of failures. Furthermore, the author discussed how organizations have no memory and

the need to share and implement lesson learned [14, 15]. Aligned with this author,

several studies have also suggested that significant improvement have been made in the

development of lessons learned. However, the dissemination and implementation of

lessons learned inhibit the complete learning process [16-18]. Moreover, the literature

provides guideline of how to develop lessons learned, how to enhance the dissemination

process and how lessons learned can be embedded into the organization [17].

5
 Incident investigation and analysis represents one of the most critical steps in the

learning process due to the challenges that may arise from the identification of root

causes in process safety incidents. Lukic stated the importance of identifying appropriate

approaches to analyze incidents depending of the complexity of them and provided

solutions within the proper domain [12]. Similarly, research by Lindberg, Hansson and

Rollenhagen explored the literature with respect of incident investigation process and

proposed the CHAIN model based on six basic criteria for incident investigation. The

model argued the importance of studying the effectiveness and effects of different

incident investigation methods in order to apply appropriate techniques and improve the

experience feedback process [19]. In the same context, Fahlbruch and Schöbel presented

the Safety through organizational learning (SOL) method in which a standardized

process for analyzing incidents in a holistic socio-technical approach have been

proposed to support incident investigation analysis [20].

Regarding learning from incidents processes, several frameworks have been

proposed with the objective of increasing effectiveness in learning. Lukic proposed a

framework for development of learning initiatives in organizations. The framework

identified five elements that organizations should analyze to select the appropriate

approach for learning: participants, learning process, type of incident, type of knowledge

and learning context [21]. Although the framework proposes valuable initiatives, it does

not identify how its initiatives are linked to a system that can be implemented in

organizations. Similarly, Coze proposed a framework that integrated the different

approaches from which learning have been analyzed. The framework provided a bigger

6
picture of this field and highlighted the main elements that influence learning such as

actors, countries, steps, industry, disciplines and intensity of the event [22]. Conversely,

semi-quantitative studies have been conducted to determine the expected behavior of

learning from incidents within an organization over time. Cooke proposed a model to

show the dynamic of the system, how each element operates and serves as a continuous

improvement process. The model also suggested how incident-learning systems serves

as a bridge between accident causation theory and high reliability theory [8]. Likewise,

Avnet analyzed the expected behavior of shared knowledge over time by the validation

of share knowledge networks within the offshore oil and gas industry [23].

Literature regarding condition for learning described relevant factors that

influence the effectiveness of learning processes. Conditions such as trust, culture of no

blame, people involvement and information availability have been identified as

necessary conditions to increase reporting and enhance learning [11, 24]. In this context,

studies explored the potential impact of learning at multiple levels and the need of better

tools for incident investigation analysis [25]. Likewise, the literature emphasizes the

need to create resilience organizations in order to been able to adapt to and absorb

changes once an incident occurred [26]. In the same way, studies suggested the need of

involvement of new groups of experts to incorporate new insights into the field of

learning [27]. With respect to learning effectiveness measures within organizations,

Jacobson presented a method for evaluating the level of learning in terms of how broadly

the lessons learned are implemented, how much the organization is involved in this

process and for how long the lessons learned last within the organization [28]. Similarly,

7
studies suggested different learning criteria to analyze and identify potential areas of

opportunity within the learning process [29, 30]. Finally, Naot proposed a method to

analyze the effectiveness of organizational learning after an incident occurred. The

method incorporated twenty-two performance indicators of the learning process, which

determines the quality of the organizational learning system within the organization [31].

1.3. Objectives

This research aims to develop a comprehensive approach to integrated the main

stages of the learning process and enhance incident investigation processes. This is done

with the objective of providing a holistic and systemic view of the learning from

incidents process and exploring the concept of knowledge management into safety

learning systems. This research incorporates a psychology and engineering viewpoint in

order to provide an overall analysis of learning systems in the chemical and oil and gas

industry. Moreover, it provides more insights into the implementation of learning

systems within the organization. In this context, the specific objectives of this research

are:

Identify the limitations in the learning system for the chemical and oil and gas

industry and provide recommendations to guide organizations on how to overcome

the identified limitations on the learning system.

Enhanced understanding of how to learn form process safety incidents and how to

develop learning from incidents systems that enable organizations to improve

8
 internal learning. Additionally, to identify the key elements organizations should

take into account to improve this learning.

Develop a framework to determine how organizations can improve the internal

learning process and transferring knowledge. It will provide a systemic approach to

improve learning from incidents and will serves as a guideline for organizations on

how corporate learning systems can be executed within the organization and how to

support the implementation of safety knowledge inside of it.

1.4. Thesis organization

This thesis involves four main phases: identification of the learning system,

development of the learning process, development of a proposed incident investigation

process and finally the validation of the proposed incident investigation process through

a case study.

The first phase reviews the organizational learning theory and relevant concepts

for the development of the following sections. It also discusses the barriers that inhibit

organizational learning, factors that influence learning, and the different types of

learning methods that organizations can apply. Moreover, reviews the incident

investigation terminology and process. It also provides a general description of some

relevant incident investigation methodologies. Finally, this phase discusses the learning

system and the limitations that can be identified on it.

The second phase of this research introduces the proposed learning process and

examines each element of it. It also shows how a corporate learning system should

9
operate and the key elements that organizations needs to take into account during its

implementation.

The third phase of this thesis introduces a more detailed analysis of the first step

in the proposed learning process: enhance internal information. It examines the incident

investigation process and methodologies throughout the development of a task analysis

and a comprehensive analysis of the current incident investigation process. It also

presents the proposed incident investigation process and provides a description of each

of the incorporated steps.

Finally, the last phase analyzes an incident that occurred in the offshore industry

in North America. The incident has been examined through the proposed incident

investigation process showed in the previous section. The analysis comprises the

analysis of two external incidents with similar causes and three internal incidents that

occurred in the same organization. It should be noted that the incident investigation has

been performed by the organization at the time where the incident occurred. Therefore,

the objective of this analysis is to identify improving opportunities of the final report

through the analysis of similar incidents in the industry. The explained methodology is

presented in Figure 1.

10
Figure 1 Thesis methodology

11
 2. ORGANIZATIONAL LEARNING

2.1. Organizational learning

Organizational learning needs to be understood from the social and systemic

perspective, since it emerges from the internal and external interactions among

individuals, technology and cultures. Thus, learning is not limited to the acquisition of

technical knowledge; it also requires the analysis and interpretation of the context where

the knowledge is going to be applied, and the understanding of human behavior and

social interactions that emerge in an organization [32]. As a result, this field has been

very challenging for the academia and industry, since the outcome is always going to be

different from organization to organization and even in the same organization, among

generations. Then, the real challenge is to transcend from the individual to the

organizational learning, in order to ensure that the knowledge has become part of the

culture of the organization regardless the people that are part of it. Individual learning

refers to the enhancement of individual mental models, through the acquisition of new or

modified knowledge, which subsequently guide human behavior and decision-making

[33]. Once the individual learning is achieved, this knowledge has to be stored,

communicated and implemented through the organization, so it can be available and

easily retrieved once it is required by another person, and consequently ensure

organizational learning.

Organizational learning is explained based on the individual learning theory, in

which learning is shown by the accumulation of individual learning that is reflected as

12
routine actions and values on behalf of the organization [33]. Therefore, individual

learning set the foundations of this theory, and then the organizational components are

added to gain a better understanding of the expected outcome. As stated by Fiol and

Lyes, organizational learning can be defined as a “change in the organization’s

knowledge that occurs as a function of experience” [34]. Likewise, in the technical view,

organizational learning is described as “the effective processing, interpretation of, and

response to, information both inside and outside the organization” 1999 [32]. Finally, as

assert by Huber, “an entity learns if, through its processing of information, the range of

its potential behaviors is changed… an organization learns if any of its unit acquires

knowledge that it recognizes as potentially useful to the organization” [32, 35] . Based

on the previous aforementioned, it can be concluded that organizational learning is the

process in which the knowledge is acquired for both outside and inside the organization,

is organized, analyzed and finally, applied in the organization. Moreover, it is expected

that this learning, changes human behaviors and gain experience for the organization.

Learning is a continuous process and varies depending of the size of the ‘unit’

that is getting the knowledge, i.e. employee, a department or an industry. Thus,

organizations gain knowledge in the following units of learning: individual, groups,

organizations, and inter organizations [36]. This learning is incremental as shown is

Figure 2. Individual and organizational learning have already explained in the previous

paragraphs. However, between them, the group learning is present, which refers to a

group that gains knowledge together through the interaction and experience with another

individual [37]. As an example of group learning we can identify a group people from

13
different departments, sharing the lessons learned from case study. On the other hand,

organizational learning is not the ultimate goal to understand the learning system as a

whole. The final level of learning is the inter-organizational learning, in which different

organizations share, communicate and learn together from experience. For the purpose

of this research this type of learning refers to the industry learning, more specifically, the

chemical and oil and gas industry learning.

Figure 2 Units of learning

2.2. Single and double loop learning

Learning from experience is also explained based on the concepts of single-loop

and double-loop. This model is based upon “theory of action” and “theory of use”,

developed by Argyris and Schön [38]. The model stated how people act based on their

mental maps and therefore how they plan, execute and evaluate their actions [38]. These

theories suggested three elements or stages of the learning experience, as shown in

Figure 3.

14
 Governing variables describes why people do what they do. Those are the main

values, beliefs and conceptual frameworks that are rooted to their cultural

background [39].

Actions describe what people do. Actions are the plans and strategies used by

people according to their governing variables. These strategies are executed in

order to keep their governing variables within acceptable limits [39].

Results describe the consequences or what people obtain from their actions. These

results can be both intended or unintended [39].

Figure 3 Single-loop and double-loop learning, adapted from [39]

Single-loop refers to the learning process in which the actions are modified based

on the expected results or consequences obtained. This type of learning involves the

recognition of an undesired outcome and the modification of the actions that leads to it.

It also can be defined as the correction of an error by changing the strategy of actions. In

the single-loop learning the governing variables remain unchanged. Conversely, double-

15
loop learning requires an additional step in which the governing variables are questioned

and evaluated in order to develop changes from the values, policies and objectives of the

organization [39].

The literature suggested that most of the organizations act according to single-

loop learning [40], in which operators spend their time focusing on how to correct

immediate actions to avoid the same mistake. But no actual analysis is performed to

detect the root causes of the incident and consequently, determine if a change in the

system is required or not. On the contrary, double-loop learning requires organizations to

develop critical thinking, in which subject matter experts should analyze the situation

and challenge the existing rules governing the organization. In conclusion, double-loop

learning allows organizations to identify root causes instead of intermediary causes and

therefore, prevent recurring in the same type of incidents. Moreover, single-loop

learning can limit organizations to only solve the symptoms of the actual organizational

causes.

2.3. Organizational processes

2.3.1. Knowledge

Knowledge is the outcome of learning [41]. Knowledge is defined as the

transformation of information, in which expert opinion, skills, and experience have been

accumulated. Knowledge is the synthesis of multiple resources of information over time,

in order to understand patterns and support decision-making [42, 43]. Thus, the

organizational learning objective is to enhance the knowledge that is acquired by the

16
employees and make sure that this knowledge is successfully stored and transferred

among the whole organization.

Knowledge can be categorized into two forms: explicit and tacit knowledge.

Explicit knowledge refers to what people knows and can be explained by individuals.

This type of knowledge is relatively easy to store, identify, and retrieve [44, 45]. Explicit

knowledge transfer can be verbal or written. Examples of explicit knowledge are reports,

procedures or case studies. The challenge with this type of knowledge is to ensure the

availability of the information as needed [46]. In contrast, tactic knowledge refers to

know-how [45, 47]. This type of knowledge is personal in nature; this implies that the

knowledge remains in the heads of individuals that are part of the organization.

Therefore, tacit knowledge is hard to extract and is mostly experience based. Tacit

knowledge can be transferred through training, practices, and experience [46].

Examples of tacit knowledge are the skills and expertise acquired by an employee or the

values and cultural beliefs of an employee.

2.3.2. Knowledge transfer

Knowledge transfer is no longer considered as an act [48]. Instead, it is

considered as a process in which, organizations gather valuable information and

experiences from one entity to another, within the organization and among them [49].

Those entities can vary depending of the number of people involved, such as groups,

departments, and facilities. The challenge with transferring knowledge is to transcend

from individual level to higher levels, in order to ensure that knowledge has been

implemented through the organization. Moreover, assessing how effectively

17
organizations are transferring knowledge is a difficult task, due to the complexity of

measuring changes in knowledge and performance of the recipient entity [50].

Transferring knowledge in organizations required the combination and

interconnection among three elements: members, tools, and tasks. By combining those

elements through the process, organizations can create knowledge management system,

where the knowledge can be stored in the organization through generations.

The process of transfer knowledge starts with the externalization, in which the

person or team must find the best method or approach to deliver the knowledge they

want to transfer. In this stage, the tacit knowledge has to be converting into explicit

knowledge so it can be delivered verbally or in written form. Second, the integration, in

which the knowledge is adapted and analyzed based on the context where it is going to

be applied. Finally, internalization, in which, the new knowledge is applied and

incorporated into the system [51, 52]. However, this process does not ensure that the

transfer knowledge is going to be successful.

Additionally, there are barriers involved in the process that makes it more

complex and therefore, makes the expecting outcome unpredictable in some cases. Some

of those barriers are explained below:

Receiver of the knowledge: the effectiveness of the transfer process depends on

who the intended receiver are, how familiar they are with the knowledge they will

get, and the context where it is applied. The similarity of task and context between

the transferring entity and the receiver entity determines how complex the process

and the type of technique they should implement. Likewise, it is important to

18
 understand the background of the receiver, in terms of the experience and technical

knowledge that they have, in order to define the best approach to deliver the

knowledge [52].

Nature of the task: another barrier that involved in the process is the nature of the

task, referring to how frequent a task is performed and if the task is a routine or not

activity. The nature of the task enables us to understand how familiar the

employees are with it and the type of system they are approaching [52].

Type of knowledge: the complexity of the process differs significantly depending

of the type of knowledge that will be transferred. The knowledge can be tacit or

explicit knowledge, and as mention before, explicit knowledge is easier to

communicate than tacit knowledge [52].

2.3.3. Organizational memory

Improving learning from incidents within an organization means improving the

organizational memory. Past incidents acquired by individuals become the source of

knowledge and is then analyzed, stored, and retrieved as a form of experience gained by

the organization as a whole. Improving the organizational memory of an organization

can increase productivity, decrease costs, and generally improve the know-how of the

organization due to accumulation of knowledge, implementation of new ideas, and

correction of errors, and consequently, lead to improve decision-making within the

organization.

Organizational memory is defined as the knowledge from the past that is exerted

upon present organizational tasks and routines [41, 53]. Likewise, organizational

19
memory is seen as the stored knowledge that an organization possesses [41]. This

concept has been evolving over the years, changing the approach and techniques applied

to gain and retain knowledge within the organization. The main change resides in the

evolution from individual memory to a corporate memory, with respect of the place

where the knowledge is stored. Nowadays, the objective is to enhance the mechanisms in

which knowledge is stored so it can be easily extracted through the years and

organizations do not have to rely only on individuals to maintain knowledge. These

mechanisms of storing are employees, organization’s culture, procedures and routines,

the management system, and the physical structure of the workplace [41]. The

combination of all of these mechanisms, ensure that organizations implement the

knowledge and make it part of the daily activities in the organization.

As Walsh and Ungson states, the organizational memory process involves three

stages: acquisition, retention, and retrieval [54]. The acquisition occurs when new

information is gained based on decisions and the evaluation of the consequences of those

decisions. The accumulation of information based on past decisions refers to the

organizational memory that is acquired by the organization [54]. Then, the information

is retained by the different storing mechanisms that the organization possesses, in order

to disseminate and implement the information across the organization. Finally,

organizations have to ensure that information can be retrieved as required. Thus,

employees can access the stored knowledge throughout the whole organization.

20
2.3.4. Organizational forgetting

Do organizations retain all the knowledge they get over the years? Does

knowledge change over time? Since employees leave, the technologies and best

practices change, and all the information cannot be retained. It is important that

organizations take into account the depreciation of knowledge over time and the amount

of knowledge that is forgotten by the organization once employees are no longer part of

it or whenever a new process is implemented.

Organizational forgetting is the intentional or unintentional loss of knowledge at

any level within an organization [55]. The intentional forgetting refers to process by

which individuals discard obsolete knowledge in order to receive updated or new

knowledge. By doing that, organizations are able to adapt to new changes in the

environment. Intentional forgetting is also called as unlearning process. Conversely, the

unintentional forgetting refers to the degradation of knowledge over time, due to lack of

storing mechanisms within the organization. Studies suggests that organizational

forgetting can be seen as positive and negative as the same time, considering unlearning

as the process that helps organizations to adapt and stay competitive in the market. In

contrast, it can be seen as negative when forgetting occurs with no reason and prevents

organizations to remember what need to be done to avoid the same mistakes [33, 55].

2.4. Barriers of organizational learning

The learning process becomes more complex for organization than for individual

due to a number of internal and external difficulties that inhibit organizations to learn

21
appropriately. Consequently, organizations have to focus their attention not only in the

elements that affect individual learning. They also have to understand the complexity of

learning interactions among individuals and the external elements that limit the learning

process. This research has been focusing in the understanding of eight barriers that

prevent or reduce the learning outcome:

Ambiguity about incident causation: refers to the tendency of the top management

level to select one interpretation instead of another. That means, that a leader is

going to choose to learn only from what it is familiar to him, and is aligned with

their preconceptions, actions, and goals [56] . This type of barrier, limited the

amount of lessons learned that are going to be communicated and implemented

within the organizations. Likewise, in some cases, this type of barrier can lead

organizations to interpret lessons learned incorrectly.

Political environments: refers to the limitations experienced by organizations once

an incident occurred. In which, the quality of the recommendations are going to

exposes the self-interests of some people. In this sense, the recommendations are

going to blame someone instead of pointing out failures in the management system

and consequently in the top-level management [56]. Moreover, political

environments restrict the amount of information that is communicated through the

organization and among them. In such politicized environments, learning is

reduced to a limited amount and low quality of lessons learned and partial learning

within the organization.

22
Competitive organizations: refers to the restrictions and limited knowledge that is

shared between and within the organizations. Learning is limited to the knowledge

and experience that is created inside the facility or organization. People from one

facility may not know what another facility is doing in terms of safety in order to

prevent the same mistakes. Competitive organizations are one the major barrier

that the chemical and oil and gas industry is facing nowadays. Sharing lessons

learned is restricted by lawyers inside and outside organizations, due the expose of

sensitive information that can reveal the know-how of organization. The chemical

and oil and gas industry needs to learn from industries such nuclear and aviation, in

which the information is shared among the whole industry to improve together and

prevent future incidents [56].

Lack of leadership: the role of leadership is key for organizations in order to learn

and have an open environment that helps employees to raise their concerns and

suggestions about how things are running. Leaders need to be involved in the

learning process, to motivate and support people throughout the process and give

them the required resources to achieve the expected outcome. Thus, leaders are

responsible for giving employees the spaces to learn and high quality resources to

get the appropriate knowledge. Additionally, leaders have to be shown as an

example, being committed to their own learning process [57].

Unlearning process: unlearning refers to process in which individuals discard

obsolete knowledge in order to receive updated or new knowledge [33]. This

process requires a high degree of reflection and understanding of what need to be

23
changed, making this process challenging for organizations because some people

are resistant to changes, act on the defensive, and have their own prejudices.

Therefore, organizations have to be creative, intuitive, and patients to face this

process [57].

Organizational culture: working in the culture of the organization is a continuous

journey in which all employees need to be engaged with the organization and they

have to believe the values and goals of the organization. A good organizational

culture allows companies to build trust among employees, motivate them to receive

knowledge from others and learn individually, in groups and in organizational

level. Furthermore, it helps organizations to be open to changes, innovation and

admission of mistakes when required. Finally, it increases near miss reporting,

since a culture of no blame is encouraged [50].

Inadequate communication: refers to the different communication channels that an

organization has and how much of their knowledge they are willing to share. This

also applies to the internal communication between departments and facilities. If an

organization is open to communicate and share their previous experiences and

knowledge, they can also receive knowledge from others and therefore, avoid the

same mistakes. Likewise, organizations have to ensure open spaces in which the

internal knowledge can be transferred from one facility to another. Finally, leaders

have to ensure an open environment in which employees feel free to communicate

with them to provide suggestions and receive feedback from their job.

24
 Short-term focus: refers to the tendency of some organizations to focus their effort

in solving short-term problems instead of looking at the big picture. Most of the

recommendations from incidents tend to focus just in the intermediary causes.

Thus, the management system failures remain hidden and the real problems are not

analyzed. This happens because root causes require more time to be identified and

addressed, and it requires a higher level of commitment for the top management

level. Similarly, this type of barrier refers to the tendency to focus in the single-

loop learning in which only actions are modified but no analysis of the root causes

is performed.

2.5. Learning methods

Knowledge is delivered to employees in form of training, which can vary

depending of the depth, frequency, and level of detailed of the task that need to be

learned. Edgar Dale developed the model “Cone of Experience”, which refers to the

different levels of abstraction to concrete experiences [58]. The model is explained with

a pyramid, showing in the top of the pyramid, the lowest level of abstraction: words.

Conversely, the bottom of the pyramid is showing the higher level of abstraction: real

life experiences [58]. Based on this model a whole new theory was developed in order to

explain the levels of retention with respect on the learning method that is applied.

Consequently, the pyramid was adapted and percentages were incorporated to show the

amount of knowledge that can be retained by an individuals depending on how this

information is presented [59]. Some researches claim that this new “learning pyramid”

25
lacks of evidence and proper validation with respect of the percentages that are shown

[58]. However, the learning pyramid can be seen as a guideline in order to understand

the different levels of retention, instead of a quantitated fact of individual retention. This

means, that teachers and organizations can guide their methods considering this

approach. But, they have also take into account the required background to perform each

method, the expected outcome, and the limitations of the target that is going to be taught.

Figure 4. presents the different average rates of retention developed by the National

Training Laboratory [60]. The pyramid can be divided into two levels: passive and

participatory teaching methods. Passive teaching methods refers to the different

techniques in which the individual is not participating or is not playing an active role

during the session, such as lectures, reading, audio-visual, and demonstrations.

Conversely, participatory teaching methods refers to the techniques in which the

individual plays an active role and is involved during the session, such as group

discussions, practices, and teaching others [60]. Similarly, passive and participatory

concepts can be described as cognitive and behavioral methods respectively.

26
 Figure 4 Training methods, adapted from [27]

Based on the pyramid, people can conclude that learning should be focusing just

in the bottom of the pyramid and forget the traditional ways of learning such as lectures

and reading. However, each method offers specific benefits that cannot be achieved only

by focusing on the bottom of the pyramid. Thus, organizations have to provide training

using a combination of two or more methods, in order to ensure that the knowledge is

transferred. Likewise, organizations have to balance the cost-benefit associated with

each technique and the appropriate scenarios to use them.

The cognitive approach is associated with transformation or addition of

knowledge, by generating relationships among the existing and new knowledge [61].

Therefore, these types of techniques are focusing in teaching the rules of how to do

something and providing the basic concepts and theories behind that [62]. In contrast,
27
the behavioral approach is associated with skill development (learn by doing) and a

change in behavior [61]. This type of technique focuses on providing practical training

and allows participants to behave and think how they would act in real life. Generally,

training objectives intend to achieve learning in knowledge, skills and attitudes. None of

those techniques can achieve all three objectives at the same time. For that reason,

training has to combine more than one method to perform a successful training program

[63].

The advantages and disadvantages of each method of the learning pyramid are

described below:

Lecture method: is designed to develop an understating about a specific topic. The

objective is to enhance theoretical knowledge through oral presentation. The

interaction between trainer and trainee is limited to the questions and answers that

can be raised during the presentation. Commonly, a lecture consists of an

introduction, body of the lecture, conclusions, and a summary. This type of method

is useful when you need to reach a large number of people. Moreover, lectures are

less expensive than others techniques and serves as a basis for other techniques

[63]. However, trainer should take into account the right balance between the

amount of material to be taught and the period of time they have, due to the

decreased effectiveness in retention [64].

Reading method: is designed as a self-study technique, which basically, has the

same objectives as lecture method, but there is no interaction between trainer and

trainee. This type of technique allows trainees go back and check the material as

28
many times as they need [65]. Commonly, reading method is used in organizations

to disseminate general knowledge regarding a specific topic, such as lessons

learned and best practices.

Audio-visual method: is designed to give trainees training through audio and visual

at the same time. Therefore, trainees can be more focused in what is being taught

[65]. This type of technique allows reaching large amount of people at the same

time. Moreover, audio-visual methods allow participants to learn by watching how

to do certain task and therefore, try to imitate what they see. Audio-visual

techniques such as videos are widely used in organizations to show past incidents

and their lessons learned. Videos have shown to improve quality of training

because of the advantages that they offered, such as flexibility in the speed they

want to learn, they can see events that are not easy to demonstrate, gives them

consistent instructions about the topic, and an objective feedback [63].

Demonstration method: is designed to show how to do something in order to teach

skills. The demonstration technique allows trainers to show the step-by-step of a

job task and the importance of each step involved in the process [63]. Moreover,

demonstrations are useful to make the training more meaningful and realistic.

Commonly, demonstration technique consists of a description of the main

objectives, practical demonstration, summary of the key learning points, and in

some cases the trainer allows the trainees to perform the same task, followed by

questions and discussion [66]. This type of method is usually applied to teach how

to follow a specific procedure [63].

29
Group discussion: is designed to allow trainees to share their opinions and

experiences about a certain topic. The objective of this technique is to improve

analytical skills, logical reasoning, and problem solving. Additionally, it helps

trainees to improve communications skills. Some of the most common types of

group discussion methods are case study and role-play [66]. Case study is designed

to apply theoretical concepts in applied context. The main objective is to give

trainees a detailed description of a situation, where an organization deals with a

difficult situation, and they have to analyze and identify the main causes of the

event and suggest recommendations. This type of technique aims to improve

analysis, synthesis and evaluation skills [63]. Similarly, in the role-play technique,

trainees have to perform a character for a specific situation that recreates real life.

The objective is to compare the performance with real life conditions and improve

problem solving skills and the identification of them [66].

Practice: refers to the type of training in which the trainee has the opportunity to

get fully involved in the activity that is being taught. Thus, the trainees are

expected to carry out the activity and demonstrate that they are able to do it

correctly. Usually, a complete training program ends with some practical training

in which the trainee is expected to develop specific skills regarding a specific job.

Moreover, the trainee is able to experience how the learned skills and behaviors are

transferred to the job, and how to deal with daily issues that arise during the job

[63]. Some of the most common techniques for practical training are Job

Instructional Technique (JIT) and simulators. Job instructional technique consists

30
 of an explanation of the task, instructional plan, a demonstration, try out, and

follow up. In this method, the trainee has to be able to explain how to perform the

task before the execution of it. Similarly, simulators are used to imitate real life

situations. This type of technique is widely used in the military and aerospace

industry.

Teaching others: the ultimate level of retention refers to teaching others, which

refers to level of understanding that you have to achieved in order to be able to

teach others. Once internalize a concept, they should be able to explain it to

someone else.

2.6. Factors that influence learning

The learning process can be affected for several factors that can lead to

unsuccessful learning outcomes. Those factors can be represented as physical and mental

process that occurred in an individual level at the same time they are getting trained.

Likewise, the different strategies implemented to transfer knowledge and the level of

understanding of the target can also influence learning.

First, a well-established training program has to take into account the

characteristics of the "adult learning" and the appropriate approach to reach them. Thus,

the training program should considers the following aspects:

Trainees need to know why they should learn: they have to understand the

objective and why that is important [61].

31
 Meaningful training: they are more willing to learn when the information is linked

to their job experiences [61].

Opportunities to practice: they need spaces to rehearse and show the learned

capability [61].

Commit training to memory: they need to move the information from the short-

term memory to the long-term memory. That means they need to receive detail

information and determine that it is important [61].

Feedback: they need to receive feedback regarding how well they are meeting the

objectives and how to fill the current gaps [61].

Spacing: teaching in various sessions instead of teaching all the material in short

period of time, will increase the long-term memory [67].

Test: testing increases learning and helps trainees to increase their long-term

memory [67].

Second, the age of the trainees has shown influence on the learning process as

certain mental capacities decrease over time. However, ages come with experience.

Thus, in some cases experience can compensate age. Third, the mental precondition that

the trainee brings to the training. This refers to the motivation and basic skills of the

trainee. Fourth, the perception that the trainee gets from the information that he is

receiving. This refers to the ability to organize and processed the information. Fifth, the

ability of the trainee to convey new information with the existing knowledge and uses

32
the information to influence behaviors. Finally, the ability of the trainee to adapts the

new knowledge to new situations [68-70].

33
 3. INCIDENT INVESTIGATION

3.1. Introduction

Incident investigation is one of the most powerful practices to learn from

experience within an organization. This process allows organizations to receive feedback

about their behavior and operational practices, in order to identify flaws in the system

and overcome the deficiencies to prevent similar incidents and avoid catastrophes. High

hazardous industries, such as, nuclear and chemical industries are more vulnerable to

experience “organizational accidents” as called by James Reason [71]. These types of

incidents are a combination of multiple failures that take place in complex systems. Even

if this type of incidents has very low probability of occurrence, once it happens it is often

catastrophic with devastating consequences that can involve people, infrastructure, and

environment [71]. Thus, organizations have to focus their attention on improving the

incident investigation program to detect latent failures that can lead to this type of

disasters and implement risk-reducing measures to prevent them.

Incident investigation is defined as the systematic approach for determining the

causes of an incident and the corresponding recommendations that may prevent or

mitigate future incidents [72]. Moreover, this process helps organizations to identify

failures in the management system that can reduce the risk of having another incident

with similar root causes at the same facility or other facilities within the organization

[72].

34
 The classical Accident Causation theory or “Swiss Cheese Model” developed by

Reason in 1997 makes a clear representation of the dynamics and path for an incident to

happen [71, 73]. The theory suggested how high-consequence incidents could occur

once latent and active failures are aligned, as depicted graphically in Figure 5. Latent

failures refer to deficiencies or flaws associated with organizational factors, supervision

or preconditions for unsafe acts. While, active failures refer to unsafe acts that act as the

triggering event [73, 74].

Figure 5 Swiss Cheese model, adapted from [71]

The interaction between latent and active failures that lead to a catastrophic

incident is hard to detect and understand, due to the complexity and limited number of

35
such incidents. Thus, learning from high consequences incidents limited the amount of

knowledge that need to be learned and only shows partial reality of the organization.

Research has shown that in order to unhide latent failures, it is necessary to investigate

all type of incidents that have significance and therefore, help to prevent disasters [75].

Figure 6 Safety pyramid, adapted from [75]

H.W. Heinrich was the first person to introduce the safety pyramid in his book

Industrial Accident Prevention in 1931. The safety pyramid refers to the ratio of the

number of minor incidents and near misses that occur before a major incident takes place

[76]. The pyramid has been validated and slightly modified by Frank E. Bird Jr. in 1996

and later on by ConocoPhillips Marine in 2003 [75]. The study made by ConocoPhillips

Marine suggests that for one fatality incident, an organization has 30 severe incidents,

300 minor incidents, 3,000 near misses, and 300,000 at-risk behaviors or unsafe acts

36
[75]. As a result, organizations has to focus their attention to the bottom of the pyramid,

where they can identify trends and find useful information to take the necessary

preventive actions and improve safety management systems. Figure 6. Presents a graphic

description of the safety pyramid.

3.2. Incident investigation terminology

The safety pyramid provides a general understanding of the different types of

incidents that an organization can experience depending of the severity of them. There is

not general agreement regarding incident classification. Even so, organizations recognize

the importance of classifying the nature of the incident in order to determine the

appropriate approach to investigate them [77]. Therefore, based on literature and

previous experience, a general description and classification is provided below:

Incident: a work-related unplanned event, which has or may have undesirable

consequence, such as harm to people, damage infrastructure or the environment [72,

78]. The definition of incident and accident are commonly mixed and used for the

same purpose. However, the definitions differ with respect of the capability to be

prevented or not. In this sense, it implies that an incident could be prevented if

something had been done differently.

Catastrophic incident: work-related incident, which has major consequences usually

involving fatalities, severe off-site environmental impact, and/or severe community

impact [77]. Some examples of catastrophic incidents are Bhopal in 1984 and

Macondo Blowout and explosion in 2010.

37
 Severe and minor incident: this type of classification varies depending on the risk

matrix applied in the company. Usually, a severe incident involves major injuries,

major interruptions to operation, and/or evacuation of personal. Conversely, minor

incidents usually refer to incidents that have no major consequences to people,

infrastructure or the environment.

Near miss: in terms of process safety, near miss is defined by the Mary Kay

O’Connor Process Safety Center as the event in which the loss of containment is

prevented by the last layer of protection in the process [79]. Likewise, near misses

refer to events with no undesirable consequences, but any changes in the

circumstances could have produced actual consequences [80].

The terminology used in incident investigations has been challenging over the

years because there is not a general understanding among the concepts and the

methodologies applied by organizations. As a result, organizations fail to understand the

difference among concepts such as root causes, contributing factors, and direct causes

and the importance to recognized and identifies each of them in order to develop

effective investigations and prevent incidents to happen again. For the purpose of this

research those concepts are defined as follow:

Causal factor: an event or condition necessary to produce or contribute to the

incident [81], i.e., that if eliminated would have prevented the incident or reduced

the consequences of it [72]. Causal factors can be classified into root cause,

contributing factor, and direct cause [81].

Direct cause: refers to the immediate events that lead to the incident [82].

38
 Root cause: refers to the underlying causes and the most basic causes of an

incident, that if removed, the incident would not have happened. Root causes are

associated with failures in the management system, which can be categorized into

failures in the system design or failures in system implementation [79].

Contributing factor: “factors that facilitate the occurrence of an incident such as

physical conditions and management practices” [72].

The level of detail of each incident investigation varies from company to

company but it is usually associated with the severity of the incident. Thus, high

potential incidents such as near misses with high significance could be excluded for

detailed analysis and the company may lose valuable findings that can lead to the

identification of failures in the management system. For that reason, companies should

take into account variables such as the nature, complexity, and the actual or potential

severity of the incident, in order to determine the appropriate team and the level of

analysis required for each incident [77].

It should be noted that although not all incidents will be investigated in detail or

investigated at all, they have to be reported and recorded in an internal database to

identify trends and keep track of all incidents within the organization. In this context,

organizations should investigate the following incidents [79]:

Catastrophic and major incidents.

Minor incidents with significance, i.e., incidents with high potential consequences.

Near misses with significance, i.e., incidents in which the outcome could have been

serious if the circumstances were slightly different [79].

39
 Minor recurring incidents or recurring near misses identified by trend analysis.

3.3. Incident investigation history

Although, incident investigation theory has been changing over the years and

new methodologies have been developed to address the complexity of new systems and

technologies in the industry. Incident investigation still conserves most of the basis and

assumptions developed in the earliest accident causation theories. Herbert W. Heinrich is

considered the pioneered of accident causation theory, introducing the domino theory of

accident causation back in 1931 [82]. The domino theory described the occurrence of an

incident based on the culmination of a series of events, which occur in a logical

sequence. Thus, an incident (called by the author as accident) can be prevented if the

series of events are disturbed [84]. The events are defined as dominos and are classified

into five categories: social environment and ancestry, fault of the person, unsafe act or

condition, incident, and injury. The theory is lined with the assumptions of incidents

prevention: focusing in people, which are responsible of the incidents and management,

where the incident can be prevented [82]. Moreover, Heinrich developed and set the

basis of the safety pyramid and created a general understanding about the importance of

accident theory.

The work done by Herbert W. Heinrich sets the starting point for the accident

causation theory, which over the years has evolved based on the level of complexity that

arises for an incident to occur. In this sense, the accident causation field has been divided

into three main ways of thinking. These three categories are: simple linear or sequential

40
model, complex linear or epidemiological model, and complex nonlinear model. Those

types of theories have been focused in the identification of single causes, multiple

causes, and complex outcomes, respectively [83]. It should be noted that the

combination of those theories and level of understanding of each way of thinking have

shaped our current understanding about incident investigation and risk analysis.

Simple linear model refers to Heinrich domino accident causation theory

explained previously, in which an accident can be prevented by eliminating or disturbing

one of the factors or dominos [83]. Similarly, Bird and Germain developed the Loss

Causation Model in 1985, in which a modified domino theory was established by the

incorporation of management factors in order to explain their relationship with the

causes and consequences of an incident [84].

Conversely, complex linear models focused in the identification and elimination

of root causes and the identification of the required barriers or controls in order to

prevent incidents. Moreover, these theories incorporated the interactions between the

individual and the system leading to unsafe conditions. Complex linear models can be

explained by a combination of theories such as energy damage theory, time sequence

model, epidemiological model, and systemic theories [85]. Complex linear theory is

highly recognized by the work done by James Reason [74] and Jens Rasmussen [57, 86],

which have had a significant impact in accident causation theory. Those theories are

based on a system-oriented approach and were able to change the perspective from the

traditional human error approach to a systematic organizational approach [84].

41
 Finally, complex non-linear models refer to the new tendency of thinking, in

which the non-linear incident causation is recognized. Thus, incidents are the result of a

combination of mutually interacting variables involved in complex environments. This

generation of thinking highlights the research done by Hollnagel and Leveson. Both of

them are recognized by the development of non-linear accident models in the 2000s

[78]. The models are: The Systems-Theoretic Accident Model and Process (STAMP)

developed by Leveson and The Functional Resonance Accident Model (FRAM)

developed by Hollnagel [83]. STAMP model considers systems in dynamic equilibrium

and analyzes the dysfunctional interactions among components that are part of the

system. Moreover, the model classified the different types of flaws that can lead to an

incident and analyzes the role of constraints in safety management systems [87].

Alternatively, FRAM model considers that system alterations lead to incidents when the

system cannot hold up such alterations. Thus, the model identified the different variables

within an organization and how to manage those alterations that can arise in order to

produce an incident [88].

These three generations of thinking, overlap over the years and combine theories

and models in order to suggest better approaches to understand real systems behaviors

[83]. Meanwhile, a considerable amount of methods have been developed for incident

investigation and those methods have been focused in the identification of root causes.

An older methodology, The Management Oversight and risk Tree – MORT, developed

by Johnson in 1973, provides a detail and comprehensive understanding for root causes

identification [89]. However, their application has been limited due to their complexity

42
and the required time to executed it. Even so, later methodologies have been developed

based on MORT methodology. Nowadays, organizations have spent significant amount

of resources in the development of computerized methodologies, which reduces time and

helps standardizing the process [90].

3.4. Incident investigation team

The incident investigation team is a critical element for the development of

successful incident investigations processes, because their expertise and teamwork will

shape the quality of the recommendations and the degree of analysis of the identified

causes. Usually, the top management along with the safety department are in charge of

selecting the incident investigation team based on the severity and the nature of the

incident [80]. The team is selected based on their experience, skills, and competencies

associated to the process, where the incident occurred, as well as, the competencies and

skills related with incident investigation methodologies. Since organizations usually are

in charge of multiple types of processes and products, it is not practical to have only one

team trained to perform incident investigations. Thus, organizations should have a pool

of trained employees across the company, who will be familiar with incident

investigation process and the methodologies that need to be applied [77]. The team

composition will vary depending of the specific required knowledge for the

investigation. Generally, an investigation team is composed but not limit to [77, 79, 80]:

Leader trained in incident investigations

Process operators, who will bring expertise of the process where the event

happened

43
 Process engineers and process safety specialists

Maintenance and technical specialists

Contractors representatives

Law representatives

Potential involvement of third parties depending of the nature of the incident.

In order to bring the objectivity to the investigation, it is a common practice to

involve employees from sister units or plants, who are familiar with the process and can

provide the necessary technical expertise in the investigation. Likewise, it is also a

common practice to include someone, who can help during the investigation with

specific technical inquiries [80]. Furthermore, some organizations try to avoid including

managers or people from the top level management to the team, because it can inhibit an

open communication within the members of the team [77]. Finally, it is also a common

practice to provide training to the team prior the investigation to provide an appropriate

background about the incident, required personal protective equipment, emergency

response, and incident investigation methodology that is going to be applied [80].

It should be noticed that the responsibilities assigned to the team are temporary

and required their full time in order to gather and analyze the information. Therefore, it

is important that managers understand their role and the temporary suspension of their

routine responsibilities [77, 80].

After the team has been selected the team members will have to go through the

whole incident investigation process, since gathering the information until the

development of the recommendations. One of the most challenging steps is the

44
identification and collection of evidence. Since, evidences can be easily disturbed and

manipulated by external sources. Moreover, based on the nature of the incident, there

may be more than one team investigating the incident, making even harder to perform

the job. Thus, the team has to act fast but at the same time be careful identifying

potential hazards to the team, trying to look the big picture of the scene, noting what is

missing and what is there that should not be there, and using all senses to get as much

evidence as possible. At the same time, they have to develop a list of preliminary

potential scenarios, which are going to be the main source for identifying the required

evidence and factors that contribute to the incident [80].

3.5 Incident investigation process

Once an incident occurs, the incident must be reported and classified

accordingly, in order to arrange the required resources and determine the best strategies

to approach the following investigation. This section addresses the different phases or

steps required to perform incident investigations, since the reporting of the incident until

developing and implementation of the recommendations [91].

To ensure a successful incident investigation, organizations have to ensure that a

proper incident investigation management system has been developed and implemented.

The incident investigation management system sets the basis and criteria of how an

incident is going to be handled and the procedures that need to be followed during this

process [72]. A common management system needs to consider at least but no limited to

the following elements [72, 85]:

45
 Organization’s responsibilities

Roles, relationships, and communication with others teams and departments

Regulatory issues and their role

Incident classification criteria

Team composition

Required documentation

Incident investigation methodologies

Required training

Reports

Resources

Incident investigation procedures

Interview forms

3.5.1. Reporting

The incident investigation management system should encourage and create a

reporting culture, where the employees feel confidents to report unsafe behaviors, near

misses and minor incidents. Likewise, a culture of no blame will allow employees to talk

and report without fear of potential consequences. Reporting must be made as quickly as

possible, in order to provide immediate response and preserve the evidence without

external alterations. There are two types of reporting to be taken into account once the

incident occurred. The first one is the internal reporting, which refers to reporting from

employees to the incident investigation management system and also the corresponding

report within the organization, to alert the departments and people that need to be

46
informed about the event. This type of reporting can be via paper, online or verbal. The

second type of reporting, depends of the severity of the incident, in which external

entities such as regulators and firefighters must be informed about the incident [91].

After the incident has been reported, the next step is to classify it, record it, and

determine how it is going to be approached. Then, the incident investigation team must

be selected to execute the investigation.

3.5.2. Collecting evidence

Gathering evidence is one of the most challenging and time consuming stages

during incident investigations, because in some cases, evidence is damaged,

documentation is not available or it is difficult to access, and witnesses may have

conflicting versions of what happened. Therefore, preserving and verifying evidence

requires experience and a rigorous plan that must be followed [81]. In this step, it is

important that prior to start collecting data, the team leader and team members available

at that time, develop a plan, which will guide the evidence collection. It is likely that the

plan is going to change as the investigation progresses [72].

The evidence or data can be classified into three types: human, physical or

documentary. Human evidence refers to witness statements or interviews, as well as

observations. Physical evidence refers to any physical source relevant to the incident

such as equipment, mechanical parts or chemicals. Documentary evidence refers to the

paper and electronic documentation, such as procedures, logs and reports [81].

Besides the information that can be collected inside the organization, there may be

additional sources that will support the investigation. Some of these of sources are:

47
equipment manufacturers, universities research centers, external databases, government

records, and companies with similar processes. As a result, not all evidence is collected

during the site visit, there may be data or information that will result from additional

analysis such as laboratory experiments, testing or external interviews [72].

The severity of the incident will determine how easily and quickly the incident

investigation team will have access to the evidence. In the case of major process safety

incidents, there may be external entities in place such as OSHA, EPA or insurance

companies that will limit the access to the site and the availability of information. At the

same time, the physical damaged of the site can affect the collection of data, and most of

the witness will be unavailable at the moment [72].

3.5.3. Analyzing the evidence

An objective investigation must consider all realistic scenarios that possibly lead

to an incident. Then, based on the collected evidence, each scenario will be tested and

consequently will be rejected or accepted. All scenarios must be documented and

reevaluated accordingly if new evidence is gathered [80]. The objective of this stage is to

identify the direct causes, root causes, and contributing factors in the incident, in order to

be able to answer the questions of what happened and why did it happen [81]. As

established in the incident investigation management system, a predefined methodology

is going to be applied to identify the root causes of the incident. Generally, root cause

methodologies start by identifying a timeline or sequence diagram, which set a basic

understanding of the incident, main events, and conditions [72]. The next steps vary

depending of the methodology that is going to be used. Some of the most relevant

48
methodologies for analysis of evidence and identification of root causes are further

discussed in next section.

3.5.4. Developing recommendations

The final outcome of incident investigations is the recommendation that are

generated based on the root causes and contributing factors identified through the

investigation. During this process, each cause has to be analyzed individually and a

proposed action or recommendation should arise from this analysis. The

recommendations have to be clearly defined, measurable and feasible. Likewise, the

objective and risk reduction action should be easily understood. Since the

recommendations will impact the management system of the organization, it is important

that recommendations go through a management of change evaluation before they are

approved [72]. Moreover, it is also a common practice that the recommendations are

risk-ranked to determine priority for implementation along with a cost-benefit analysis in

order to determine the feasibility of the recommendations. All recommendations need to

be evaluated and approved by the management and depending of the nature of the

incident, the legal department may need to review and approve the recommendations as

well. Once the recommendations are approved, the people in charge of the execution of

each recommendation are assigned and a timeline is defined. Finally, the management

system is responsible for tracking the implementation status of those recommendations

as well as document the resolution of each recommendation. Only until the

recommendations are implemented, organizations are reducing the risk and consequently

learning from incidents [80].

49
3.5.5. Report and lessons learned

The final stage of the incident investigation process involved the development of

a written report, which contains all documentation involved during the process. The

report covers all the detailed information used during the investigation, since evidence,

causes until recommendations. Generally, an incident report format includes at least the

following elements: summary, background, description of the incident, root causes,

recommendations, and appendix. Finally, the lessons learned are expected to be

communicated with all interested parties in order to ensured learning across the industry

[80].

3.6 Incident investigation methodologies

3.6.1. Fault tree analysis

Fault tree analysis is a graphic methodology, whose objective is to identify

“why” an incident happened [72]. The methodology uses logical reasoning to determine

the possible combinations or pathways that lead to an incident. Those combinations can

be associated with equipment or component failures, human factors, or organizational

factors [92]. Fault tree analysis uses starting point as the top event, which refers to the

outcome of the incident. Then, the process consists of going backwards in order to

identify the preceding causes or events, until the root causes are finally identified. The

level of detail of this technique will depend of the incident investigation team. A FTA

can be used as a qualitative method, quantitative method or both [81]. The methodology

50
uses symbols to guide readers to understand the different pathways and logic of the

diagram. Figure 7. Present the most relevant symbols for developing fault trees.

Figure 7 FTA symbols, adapted from [77]

The symbols AND and OR represent the gates that connect one event to one or

multiple events. The gate AND indicates that the outcome event occurs only if all the

input events occur. Conversely, the gate OR indicates that one or more input events have

to occur to produce the outcome event [72]. Figure 8. shows a basic representation of a

FTA.

51
 Figure 8 Basic representation of a FTA, adapted from [81]

3.6.2. STEP (Sequential Timed Events Plotting)

Sequential timed events plotting is a systematic methodology for incident

investigation (accident investigation as called by the author) developed by Hendrick and

Benner (1987) [93]. STEP is a multi-linear event sequence, which provides a detailed

description of the incident process. The methodology focuses in the identification of the

authors and the sequence of events or actions that lead to the incident. Thus, STEP

considers that multiple actions can take place at the same time by different authors. The

authors are considered as people or things directly involved in the incident [93].

The incident need to be analyzed looking for the big picture and then breaking

down into actors and actions. In order to accomplish that Hendrick ad Benner introduced

the term “making mental movies”, which refers to the visualization of each action
52
executed by each actor from the time the incident was expected to began, until the top

event of the incident. The methodology process is supported by a STEP worksheet,

which helps to visualize and link all the events together. The STEP worksheet is

described as a matrix, in which the authors are identified in the rows and the events in

the columns, as shown in Figure 9. To ensure logical sequence, the events are linked by

arrows that represent the flow of the incident [93].

Figure 9 Basic STEP diagram, adapted from [93]

After the diagram is completed, the incident investigation result is subject to

three different types of tests: row test, column test, and necessary and sufficient test. The

row test verifies that the actions and authors are broken down sufficiently. The column

test verifies that the sequence of events is consistent and coherent. Finally, the necessary

and sufficient test verifies that the previous event is sufficient to produce the outcome

53
event [82]. The STEP worksheet is then analyzed in order to identify the event or events

that generate safety problems. Each safety problem is then analyzed as candidate for

recommendation [82].

3.6.3. FRAM (Functional Resonance Analysis Method)

Functional resonance analysis method provides a socio-technical approach in

order to analyze and understand the functions of the system. The objective of the method

is to determine what should have happened that did not happen. To do that, FRAM

analyzes the potential variability and functional resonance in the system [94]. FRAM is

based on the development of four steps:

First, the identification of daily functions required in the system. Functions refer

to the activities required to produce a specific outcome. Thus, the first step required to

determine all possible functions involved in the system as well as the identification of

the attributes of each function. The attributes refer to the characterization of the

functions, which are described into six aspects as shown in Figure 10 [94].

Figure 10 FRAM Function diagram, adapted from [94]

54
 Second, the characterization of the variability of each function. This

characterization is based on eleven Common Performance Conditions (CPCs). Those

elements take into account human, technological, and organizational factors. The

variability is determined considering both the potential and actual variability of the

system [94]. Third, the identification of coupling functions and their interrelation in

order to identify unexpected outcomes. This step requires to link all the identified

functions and understand the potential variability that can be expected from that. Finally,

the last step refers to the identification of barriers for variability and performance

monitoring. Barriers can be divided into system and function barriers. System barriers

refer to organizational aspects and physical barriers. Conversely, functions barriers refer

to the way barriers can achieved their purpose. At the end, recommendations are going

to be develop to enhance the identified barriers [94].

3.6.4. Tripod BETA

Tripod Beta is a systematic method to perform incident investigations. The

method is based on the accident causation theory or Swiss Cheese Model. This

methodology was formalized by Shell International and provides an understanding of the

influences of human behavior and organizational factors in the system. Tripod theory

focuses in the identification of environment conditions (organizational factors) in which

humans work, rather than human conditions that lead to the incident. Thus, the ultimate

goal is the identification of failures in the management system and organizational

culture. Likewise, the methodology provides three main objectives: development of the

55
chain of events that lead to the incident, identification of the barriers that should have

prevented the incident, and the identification of the underling or root causes [95].

The first step, is the development of a “core diagram”, which refers to the

graphical representation of the events. The core diagram is developed based on the

identification of Tripod Beta trios: Agent, object, and the event. Then, the trios are

connected among them to provide a chain and sequence of events. The second step refers

to the identification of the immediate causes that allowed barrier failing. The third step

refers to the detailed understanding of why it happened. Thus, the preconditions and

underling causes of the incident are identified. Finally, the previous steps are repeated

until all the evidence is collected and analyzed [95].

In addition, Tripod Beta defined eleven mechanisms or Basic Risk Factors

(BRFs) to categorize and guide the identification of underlying causes. Those BRFs are:

design, tools and equipment, maintenance, housekeeping, error enforcing conditions,

procedures, training, communication, incompatible goals, organizations, and defenses

[82].

3.6.5. IPICA (Integrated Procedure of Incident Cause Analysis)

Integrated procedure of incident analysis is a method for incident investigation

based on the traditional Root Cause Analysis (RCA) method. The RCA is a complex

linear method, whose goal is the identification of casual factors and root causes of an

incident. The method mainly involved the following steps: development of incident

sequence, identification of causal factors, development of causal factor chart, and the

identification of root causes. The last step can be achieved by two different approaches:

56
using logic trees or using predefined trees. IPICA follows the second approach using

predefined trees [96].

IPICA suggests an innovative approach, which addressed the limitations of RCA

method identified by the author. The methodology is focused in the improvement on

three main areas: assumptions of the nature of incident causes, limited definition of root

causes and the methods applied for their identification and the complex non-linear effect

in incidents. IPICA identifies four incident cause levels: implementation, organizational,

management’s attitudes, and societal. The first level is considered as the direct causes of

the incident, which are associated with the meta-components, i.e., interaction of

hardware, software and personnel. The second level identifies failures in the

management system and safety culture of the organization. In order to identify this, a

root cause map is developed based on the CCPS guidelines to construct process safety

management systems [72]. The third and fourth levels are required when the causal

factors are associated with factors beyond the organizational boundaries. To identify

those factors, the implementation of non-linear techniques such as STAMP or safety

archetypes is needed [96].

3.6.6. TapRooT® System

TapRooT® System combines a process and set of tools to investigate incidents

and develop corrective actions. The system goes beyond the identification of the causes

of the problem, instead the system supports investigators through the whole incident

investigation process, since the collection of data, until the presentation of results and

recommendations to the top management level and interested parties. TapRooT® is a

57
software, which uses and combines multiple techniques in order to achieve detailed and

improved results [97].

The system consists in the execution of seven steps. The first step refers to the

collection of the required information. The second step is the development of a sequence

of events. In the third step the causal factors are identified. The second and third steps

are supported by techniques such as SnapCharT® and Equifactor®. The fourth step

refers to the identification of root causes, in which tools such as Root Cause Tree® and

dictionary are used to help investigators in the identification of all the root causes. The

root cause tree is divided into seven main categories and it is then break down into more

detailed. The seven categories are: procedures, communication, work direction, training,

management system, quality control, and human engineering. In addition, the root cause

tree includes a set of fifteen questions to help investigators analyze human factors in

more detail. Once the main root causes have been identified, the system guides the

investigator in the identification of more complex causes such as culture, systemic, and

organizational factors. The following steps refer to the analyzes of the root causes,

development of recommendations, reporting and implementation those

recommendations [98].

3.6.7. MORT (Management Oversight and Risk Tree)

Management Oversight and Risk Tree is an incident investigation methodology

developed by William Johnson. MORT consists of a predefined tree based on fault tree

analysis methodology. The main objective of the methodology is to go through each

branch in the tree and analyzes whether or not the associated causes are applicable to the

58
event. The branches end with the identification of failures in the management system

[92]. Thus, the methodology uses a schematic representation of a dynamic and ideal

management system model [89].

The logic diagram identifies 98 generic problems and over 1500 basic causes or

root causes associated with failures in the management system. The diagram starts with

the top event followed by an OR gate, which derived the first important analyses that

need to be done. The OR gate breaks down into two main branches: Management

oversight and omissions or assumed risk. Assumed risk refers to the risks that have been

analyzed and accepted by the management level. Conversely, unknown and unanalyzed

risks are considered in management oversight and omissions. The next step requires to

answer the questions ‘why’ and ‘what happened’. The ‘why’ refers to the management

system factors, while the ‘what happened’ refers to the specific controls that should be in

place [99].

In order to perform the analyses, the diagram is supported with a manual that

helps investigators to ask the right questions in each level of the analysis. Likewise, the

methodology uses a color-coding system to help investigators visualize the progress and

identify the areas that need additional information or analysis. The events that are not

applicable to the incident should be colored in black. For the remaining events, a

comprehensive evaluation needs to be done to understand if the event was adequate or

not. If the event is considered Less Than Adequate (LTA), it should be colored in red.

On the contrary, if the event is considered adequate, it is colored with green. In some

cases, events can not be classified into those two categories because of lack of

59
information or uncertainty about this specific event. Those events are colored in blue and

needs to be evaluated in more detail or more data need to be collected. The MORT

analysis finished when all the blue events has been evaluated and consequently decided

whether it is considered LTA or adequate [89, 99].

3.6.8. STAMP (Systems-Theoretic Accident Model and Processes)

Systems-Theoretic Accident Model and Processes is an accident model based on

system theory, developed by Nancy Leveson. STAMP describes accidents as inadequate

control or enforcement of safety constraints in the system, instead of components

failures. Thus, STAMP focuses in understanding why the control structure was

inadequate and which feedback loops failed in the system. Likewise, STAMP considers

safety as part of an adaptive socio-technical system, in which all the components are

interrelated and are kept in a dynamic equilibrium [87].

STAMP is based in the understanding and analysis of three basic concepts:

constrains, control loops and process models, and levels of control. These concepts

contextualize a classification of control flaws that help investigators to identify all

factors involved in the incident [100]. First, accidents are conceived as the identification

of constrains rather than events. Thus, STAMP identifies all the required constrains in

the system, including the social and organizational factors and no just system design

constraints. In addition, systems are conceived as hierarchical structures, in which each

level imposes constraints to the lower level in order to control the system. In this sense,

safety has two basic hierarchical levels of control: system development and system

operation. Those levels of control must interact and develop effective channels of

60
communication. Each level in the control structure must define the different channels of

enforcement and communication. Thus, each hierarchical level will impose constraints

downward and will receive feedback upward to verify the effectiveness of those

constraints. Finally, the concept of control loops and process models refers to the

consistency that must be achieved between the model of the process used by the

controllers (human or automated) and the actual process state. In this sense, the

controllers are able to supervise the actual state of the system [87].

STAMP developed a classification of flaws in the components of the system

development and operations to support investigators in the analyses process. The

classification is divided into three main branches: inadequate enforcement of constrains,

inadequate execution of constrains and inadequate feedback. Each branch is then divided

in more detail to identify al the factors involved in the incident [87].

3.7. Developing quality incident investigation reports

In incident investigations, all the effort and resources spent on it are going to

shape the quality of the final product of the process: the incident investigation report.

Therefore, identifying real root causes, developing quality recommendations, and

making sure the recommendations are implemented are key elements to success in this

process and reducing the risk in order to prevent similar incidents in the future. The

incident investigation process allows organizations to analyze their system behavior and

uncover hidden organizational, cultural, and technical flaws that are not visible in their

day-to-day work. The process has to be performed with rigorousness, time, and making

61
sure the organization understands the importance and valuable output they can get from

it. Otherwise, the process is going to be seen as a useless secondary work that needs to

be done to comply. Having a good incident report and an overall good incident

investigation program enables organizations to continually improve and over time

ensuring a reliable process safety management system and high standard operational

processes. In this sense, learning from incidents allows organizations to improve, be

more competitive, and safe money. As Klezt stated, if organizations think safety is

expensive, try an accident [14].

Quality incident investigation reports are the based on recommendations that are

clearly defined and are intended to solve the identified causes on a feasible and

measurable way. This means, that the documentation and analysis are not going to be

limited to generating information, but also to making sure that knowledge is created

upon that. Similarly, it is important to make sure that the created knowledge becomes

part of their management systems and operations. Once the organization creates and

accumulates knowledge within it, the next step is to collectively acquire and implement

this knowledge across the organization and industry, to make sure no one is doomed to

make the same mistakes. This can only be achieved by sharing the lessons learned and

ensuring those lessons are not forgotten. Trevor argued that achieving the first step

(spreading the message) is relatively easy. However, ensuring that this message would

become part of the organization is the real challenge [14]. Likewise, he asserted that

“organizations have no memory, only people have memories and they move on” [14].

62
Therefore, it is important to share lessons learned, create knowledge and make sure it

will not remain just on paper.

63
 4. LEARNING PROCESS

Learning can be defined as the process in which humans acquire or modify

knowledge through experience, education, or training [101]. Humans have the ability to

learn and this process is based on their previous knowledge. Thus, the results from the

same experience or training can be different for each person. Moreover, learning is not a

linear process. It follows a learning curve, which decreases over time if not adequate

reinforcement is performed. In order to ensure that learning has been achieved, a

modification of human mental models and/or changes in their routine behavior is

expected.

When it comes to learning inside an organization, human behavior cannot be

considered as a single component in the organizational system. It is required to consider

the interactions and relationships among them. Thus, the learning process is no longer

considered as an individual process, instead, it is considered as an organizational

process, in which, people try to acquire knowledge together and preserve it over time.

The challenge is to ensure that everyone within the organization manages, interprets, and

transfers knowledge effectively as needed. Likewise, companies have to ensure that

knowledge will remain available even if there are changes in personnel. As stated by

Kletz, “organizations have no memory, only people have memory and they move on”

[15]. In this context, organizations have been focusing on how to manage knowledge and

preserve it over time. These with the purpose of improve continually, avoid the same

mistakes, and consequently prevent incidents.

64
4.1. Learning system

In the learning system for the chemical and oil and gas industry, three entities can

be identified as components that have the ability to learn: individual, organization, and

industry. For the purpose of this research, the learning is based on previous incidents, the

implementation of lessons learned, and best practices. Learning occurs from the smallest

entity to the biggest one. Thus, once an incident occurs, the individuals involved in it are

going to challenge their mental models and identify the potential causes of the incident

and consequently learn from experience. However, due to the complexity of the human

behavior, it can be expected that a small portion of the people who experience an

incident, choose to ignore it.

Depending on the severity of the incident, reporting plays an important role in

the learning system. In the case of near misses, the safety culture of the organization is

challenged and the individuals involved in the near miss have to decide whether or not to

report the event. On the contrary, in the case of a major incident, the process safety team

is usually part of the immediate executive actions; therefore, reporting is not required

from the individuals that experienced the incidents to the safety team. Moreover,

organizations need to notify the corresponding regulatory agencies and determine their

role in the investigation process.

Once the incident is reported, the significance of the incident needs to be

determined. In this process, the safety team has to evaluate the impact of the incident and

based on those results, the safety team would determine whether or not the incident

would be investigated in detail. This means to perform an incident investigation process,

65
in which the root causes, lessons learned, and recommendations are determined.

Organizational learning occurs once each individual who is part of the organization has

acquired and processed the new knowledge that is being taught and collectively

implement the lessons learned. The ultimate level of learning is the industry learning, in

which the organizations share the lessons learned and transfer the existing knowledge

across the industry. This is done with the objective of having an updated knowledge and

improves collectively. Figure 11. shows the system explained before.

Figure 11 Learning system, adapted from [90]

66
4.1.1. Limitations in the learning system

The learning system presented in the previous section, gives an idea of how the

system works, the entities involved and their role in the learning process. Additionally, it

gives a bigger picture of the main steps and sequence of the learning process. Based on

this diagram, together with the available literature review in this field, it is evident that

organizations still see knowledge as temporal factor that is inherent to the people

involved in the organization, instead of a part of their management systems. Thus,

organizations fail to transcend from individual learning to organizational learning. In

addition, they fail to provide the required tools to enhance the individual learning

process in order to ensure higher levels of retention and sense of ownership. Finally, the

chemical and oil and gas industry fails to understand the need of putting all resources

together to improve as a whole. In this context, several limitations that inhibit learning

have been identified:

Identification of root causes: the incident investigation management systems set the

basis to ensure learning from incidents, because it ensures that the fundamental

causes of the incidents have been identified and managed properly. Thus, incident

investigation requires an appropriate level of expertise during all phases of

investigation, from the collection of evidence until the development of

recommendations. This would ensure that the collected data is reliable, adequate,

and complete. Moreover, that the analysis has been performed with sufficient detail

and the root causes have been identified. Organizations tend to identify direct

causes as the root causes of incidents. Thus, recommendations tend to solve

67
superficial problems that do not prevent similar incidents in the future. Direct

causes such as human error are usually identified as root cause, and investigation

process ends placing the blame on the operators instead of identifying failures in the

management system that contributed to the execution of the unsafe acts. This

misinterpretation of concepts inhibits organizations from developing high quality

incident reports and blocks the continuous improvement of the management

systems. Organizations need to ensure three key elements in order to be able to

identify root causes. First, they have to provide adequate tools to perform the

analysis of the investigation. The incident investigation team needs to be supported

with standardized methodologies and procedures to guide them through the whole

process. Second, organizations have to provide appropriate training regarding

incident investigation to ensure a pool of trained people who are familiar with the

incident investigation process and methodologies applied in the organization.

Finally, organizations have to ensure appropriate level of expertise based on the

nature of the incident to guarantee the required technical knowledge during the

analysis.

Promote reporting of near misses: learning should be promoted from every valuable

source of information. Therefore, organizations do not have to wait until a major

incident happened to improve their system. Acquiring data and information and

analyzing it, give the opportunity to learn and consequently prevent the occurrence

of major incidents and avoid losses. In this context, near misses give a significant

amount of information to identify trends and high potential consequence incidents.

68
It can also helps identifying failures in the management system and consequently

provide improvement opportunities. Promoting the report of near misses becomes a

crucial element for organizations to analyze more frequent incidents and extract

their knowledge. Achieving the expected results is challenging for organizations

because it requires a healthy reporting culture, in which the individual punishment

is avoided. This means, creating an open culture, where employees feel confident to

report incidents and everyone feel responsible for their own and their co-workers

safety. Moreover, promoting the report of near misses require the engagement of the

top management level to ensure the required resources and time.

Ensure implementation of lessons learned: Organizations usually spend a

significant amount of resources during incident investigations processes. However,

this effort decreases drastically once the recommendations have been developed and

the final report has been presented. Organizations fail to support the implementation

process and therefore, no actual learning is performed. In the majority of the cases,

learning is limited to the development of the report but actions are not executed.

Thus, organizations inhibit the organizational learning because no actual knowledge

has been transfer to the organization. Moreover, investigators play an important role

in the development of those recommendations, in order to ensure that they are clear,

measurable, and feasible; and consequently, people in charge of executing those

actions can easily understand what is intended and set the right timeline and

resources.

69
Ensure sharing knowledge within the organization: the ultimate goal of incident

investigations is to prevent similar incidents from happening again. In this sense,

the extracted learning from those investigations should to be spread out across the

organization in order to ensure that different facilities all around the world, can get

those lessons and extract their knowledge. Although organizations communicate

lessons learned, they usually fail to give the message in an appropriate way. This

means to understand the different types of targets across the organization and

unpack those lessons learned in a way that people can understand how those lessons

would impact them. Furthermore, this sharing is limited in most of the cases to

safety moments or emails, but there is no discussion or analysis of the applicability

of those lessons to their own process or facility.

Use and understanding of databases and academic resources: knowledge

management has become a key element in organizational learning, in order to ensure

that the data and information is stored and easily extracted when people need it.

Thus, databases play an important role to maintain information is available across the

organization. Additionally, academic resources for large organizations that have the

budget or external databases can help organizations to extract new information and

learn from external resources that handle similar processes. The main objective of

those external resources is that organizations learn and improve as a whole, as an

industry, instead of getting partial learning in few organizations. In this sense, the

industry fails in two different ways: First, the industry faces a competing and

politicized environment [56], in which organizations are not willing to share their

70
 lessons learned and best practices. Therefore, there is a lack of external information

from which others organizations could learn. Secondly, even in the cases where the

information is out there, organizations fail to understand the importance of those

external tools and therefore people are either not aware of there existence or are not

interested in using them. Finally, Organizations usually communicated external

lessons learned from big incidents, but people see that type of incidents as events that

could not happen in their own processes and therefore little attention is put into

analyzing how those lessons learned are applicable to their processes and how that

information could make their facility safer.

Validation of leading indicators: process safety indicators provide evidence about

the actual state of a particular process or activity. It helps organizations to get the

right signals before an incident materializes and therefore, take appropriate actions to

get the process back to normal. When an incident occurs, people involved in the

process spend little or no time analyzing why they failed to get the right signals. Was

the process giving the right signals? Were the indicators well defined? Why did

operators fail to understand those indicators? These types of questions have to be

answered after each incident investigation, in order to reevaluate the process, where

the incident occurred and make sure operators can get all the required information

and understand the process’s behavior.

This research has focused in the analysis of four of the limitations mentioned

above: ensure implementation of lessons learned, use and understanding of databases

and academic resources, ensure sharing knowledge within the organization, and

71
identification of root causes. The first three limitations have been addressed in this

section. The last limitation: identification of root causes is covered in section 5.

4.2. DIKM model

In order to improve learning from incidents, it is required to ensure a corporate

memory system within the organization, in which an organizational memory is present

and knowledge is effectively transferred and maintained through generations. The Data-

Information-Knowledge-Wisdom model (DIKM) fits the needs mentioned above and it

has been widely used for many organizations to manage knowledge. The model

describes the relationship among data, information, knowledge, and wisdom, and the

transformation of each element to a higher level [102, 103]. Thus, the model represents

how organizations can increase meaning and value as they go higher in the pyramid, as

shown in Figure 12. An extensive literature about this model and knowledge

management is available elsewhere [43, 104-106]. A summarized definition of each

element is provided to understand the structure of the model and its link with the

learning process.

Data can be defined as values without context or interpretation. They can be

symbols, signals or numbers with no structure [102, 103].

Information can be defined as data that has been processed for a purpose. It is

data that have been organized so it can have meaning for the user [102].

Knowledge can be defined as the transformation of information, in which expert

opinion, skills, and experience has been added. Knowledge is the synthesis of

72
 multiple resources of information over time, in order to understand patterns and

support decision-making [102].

Wisdom can be defined as accumulated knowledge. It requires an understanding

of principles, so it can be seen as an expert knowledge and judgment associated

with the meaning of life [102, 103].

Figure 12 DIKW Model, adapted from [102]

This research has been focused only in one type of knowledge: safety knowledge.

It converges technical, human and organizational aspects, that lead to influence people

behavior and values. Thus, the learning process goal is to enhance, capture, refine, and

implement safety knowledge so it can become part of the daily activities and culture of

the organization. The DIKW theory has been used as basis to construct a framework to

improve the learning process within an organization.

73
4.3. Learning process

Based on an extensive review of the available literature and benchmarking within

oil and gas organizations, consulting companies and subject matter experts, a framework

for the learning process inside an organization has been developed. The framework

suggests a standardized approach to leverage the safety knowledge within an

organization. It includes seven steps that should be followed in sequence in order to

improve learning from experience. The steps included in the framework are intended to

serve as a guideline for organizations that want to get a better understanding on how to

obtain information and make it valuable. That is, transforming information into

knowledge, by making it part of the organization and ensuring a corporate memory

system inside of it.

In order to leverage the safety knowledge inside the organization, it is required to

establish a corporate learning system, which should be owned by all employees in the

organization. The system should be easily understood and accessed when required. The

corporate learning system has two main objectives. First, make new information and

lessons learned part of the organization. This means, to make those lessons learned part

of the existing resources in the organization such as procedures and guidelines. In this

context, people can have access to lessons learned related with the information that they

are looking for and at the time they want it. This would help as a reminder about what

can go wrong with the specific activity they are intended to perform. Second, the

corporate learning system needs to support implementation and decision-making of new

alternatives with the objective of making their facilities safer. To ensure that the

74
information and knowledge that is being added will remain in the organization, it is

important to document the whole process, make sure a proper management of change

process is performed prior to any change, and link the reasons that support the

alternatives. The last point is intended to ensure that people will understand the reasons

of why the alternative was implemented at that time and the analysis behind it.

Most of the people involved in the organization are not even aware of where to

find information and the potential benefits they can get from it. On the contrary, some

people are aware about the available sources of information, but they do not know how

to take advantage of this information. People inside the organization have access to a

high volume of information but in many cases they do not have the time to identify

which information is important, which information is valuable and should to be

transformed into knowledge, and which information is not that relevant for them. As a

consequence, people lose relevant information that can be applied to their facilities.

People act based on what it is important for them and what they consider relevant for

their work and would have a potential benefit for them. Therefore, transferring

knowledge is a complex task, in which identifying the target and having a clear

understanding of why it is important to them, becomes a crucial step to achieve learning.

In this context, a lesson learned from an incident could be explained with technical detail

to a supervisor, whom has the engineering background to understand the concepts

involved on it. However, the same lesson learned has to be delivered in a different way

to an operator, so he can understand and process the same information as the supervisor,

without loosing the knowledge that is intended to be transfer.

75
 Some organizations have spent a lot of effort developing sophisticated systems to

store lessons learned and relevant information. Nevertheless, they fail to achieve the

complete cycle, in which people within the organization own the system and understand

it. The corporate learning system should be incorporated as part of the safety

management system of the organization, the top management should support it, and all

levels in the organization should be trained on it. The training would explain the benefits

of it, the role of the corporate learning team and the people involved on it. Moreover,

the training would give guidelines on how to use the Document Management System

(DMS), when to use it and the type of information that can be extracted from it.

4.3.1. Corporate learning team

A corporate safety team should be in charge of managing the corporate learning

system. The team has to be a group of subject matter experts with relevant knowledge in

the processes, products and types of facilities inside the organization. The people in

charge of the corporate learning system serves as a source of knowledge for the different

facilities and people in the organization. In this context, they would be the bridge

between the information and knowledge that need to be digested and implemented.

Moreover, the team has the responsibility of filtering the information in order to ensure

that the information that is being added is valid and reliable. It should be noted that the

team is not intended to impose or enforce the implementation of lessons learned in the

facilities, rather, the team is intended to guide facilities to get the right information and

direct them in the right direction. The objective of the team is to support decision-

making and provide the necessary resources to implement new alternatives. In this sense,

76
the corporate learning team would be responsible to acquire new information, validate

and analyze the applicability, disseminate the information to the intended targets, ensure

the availability of the information, and support implementation as required.

It is important to notice that a single team of experts it is not enough to determine

what information is relevant for the organization. In addition, it is not possible to have

the level of expertise in all areas of process safety. Thus, a pool of experts in different

areas within the organization should be linked with the corporate learning team, in order

to provide support in the analysis and selection of potential knowledge that should be

acquired by the organization. In this sense, the pool of experts would serve as a guideline

or consulting assistance for the corporate learning system.

4.3.2. Flow of information

Information and knowledge are important components in the management of any

safety system. The system has to ensure the availability, accuracy and updated

knowledge, in order to support the learning process and decision-making. Inside the

system two different types of information can be identified: internal and external

information. Internal information refers to the information that is created inside the

organization, such as incident investigation reports, best practices and performance

indicators reports. Conversely, external information refers to the information that is

obtained from agencies, databases, academic resources and other organizations. This

information can be lessons learned, new standards, best practices, recommended

practices, etc.

77
 The corporate learning system has the challenge to effectively manage all

channels and types of information at the right time and to the right target. Therefore, the

system should be supported by a Document Management System (DMS), in which all

the information would be uploaded easily and extracted as needed. In addition, the DMS

needs to keep the information up to date and link the new information with the existing

documents of the organization.

Figure 13 Flows of information in the corporate learning system

In this context, the corporate learning system would acquire internal and external

information, which is going to be analyzed and validated by the corporate learning team.

The relevant information would then be incorporated into the DMS and linked to the

78
existing documentations as required. Then, the team together with representatives of

each facility, in which the knowledge is intend to be transferred, would work together to

determine the best approach to implement that knowledge into the facility. After

implementation, the facility would provide some feedback to the corporate learning team

to ensure continuous improvement and verified the effectiveness of the process. At the

same time, each facility is responsible for providing internal information to the system to

assure that the information is analyzed for different facilities across the organization and

it is not just communicated. This flow of information is shown in Figure 13.

4.3.3. Learning process description

The learning process consists of seven steps: enhance internal information,

acquire information, assess alternatives, define metrics, refine alternative, implement

alternative, and monitor performance. The process describes how the corporate learning

system would work and the key elements that need to be considered in each step. The

first two steps refer to the acquisition of relevant data and information into the system

and the last five steps refer to the transformation of that information into knowledge. A

diagram of the process is shown in Figure 14.

79
 Figure 14 Learning process

4.3.3.1. Enhance internal information

Enhance internal information refers to the process in which the incident

investigation process and final product of is improved by the incorporation of additional

steps in the process. Its goal is to provide a more detailed analysis of the incident and the

safety management system of the area where the incident occurred. The process is

enhanced through the investigation of all incidents with significance, the review of

previous internal incidents and similar external incidents, the revalidation of

performance indicators of the process, and the identification of failures in the

80
management system. Additionally, this step support organizations to effectively identify

organizational, design and cultural failures instead of human errors as root cause of

incidents. This step is explained in detail in section 5.

4.3.3.2. Acquire information

Once the organization can rely on the internal information that is created inside

of it, the next step is to acquire additional information from different sources. In this

sense, it is important to identify the potential sources from which relevant information

can be extracted. As explained previously, there are two types of sources: internal and

external. Internal information would be acquired from the different facilities within the

organization. Likewise, internal information would be gained from all branch offices

around the world that are part of the same organization. This information would be

primarily lessons learned from incident investigations. However, information such as

best practices and trend analysis reports can be considered as well. External information

opens a wide window of opportunities from where new information can be acquired. The

primary source of information would be available databases. Nowadays, there is still no

single database available, which can provide full access to a large number of incidents in

the chemical and oil and gas industry and, at the same time, have all the desirable

features such as easily searchable, public, with multiple filters options, and with detailed

analysis reports. Therefore, it is necessary to use a combination of multiple databases.

Regulatory agencies such as the Environmental Protection Agency (EPA) have their own

database, from which information regarding toxic releases can be found. The Chemical

Safety Board [6] provides detailed analyses of process safety incidents. However, the

81
amount of incidents investigated is limited. Some public databases such as eMARS

developed by the Major Accident Hazards Bureau (MAHB) provides incidents reports

involving highly hazardous substances. Likewise, private databases such as Process

Safety Incident Database (PSID) created by CCPS, provides incidents reports for

organizations participating on it. Additional information can be acquired from external

associations like the American Petroleum Institute (API) and the National Fire

Protection Association (NFPA), which provides guidelines for standards and best

practices. Finally, large organizations that have the budget also have the opportunity to

established partnership with academic entities in order to be involved and have access to

trending investigations from first hand. This type of partnership can be established with

universities like Texas A&M, which houses Mary Kay O’Connor Process Safety Center.

It is evident the wide area from which safety information can be extracted. Thus, it is

important that the corporate learning team set the basis and defines the scope and

objectives of it. Moreover, those objectives have to be aligned with the organization’s

goals and business priorities.

4.3.3.3. Assess alternatives

Once the team has prioritized the alternatives that could be valuable for the

organization and fit the needs of it, they have to identify the potential processes,

facilities or business areas, where the alternative could be applied and would gain

benefits from it. Those relevant installations are the potential target, where the

information could be translated into knowledge and consequently be implemented into

their processes. During this process the corporate learning team has the responsibility of

82
determining and classifying the potential targets and the channels of communication that

are going to be used during the process. The information can be communicated into two

different approaches: sharing the information or transferring the information. Sharing the

information refers to the communication to an arbitrary receiver. In this case, it would be

facilities that might be interested in the information but no action is required from them.

Conversely, transferring the information refers to the communication to a specific

receiver, who in this case would be facilities in which the new or alternative information

could be implemented and the potential knowledge is intended to be transferred.

Transferring the information requires a high level of commitment from the

different facilities that have been selected as potential targets. From this point, a member

or members of the corporate learning team would meet with some representatives of the

potential target and would work together in the analysis of the alternative. Through

meetings the objective is to communicate the alternative and the assumptions that have

been made in order to determine why it is applicable for them. Then, they would

evaluate together the applicability of this alternative to their process, by analyzing the

potential benefits, costs and risk associated with the implementation or not of the

alternative. This process has to be documented and the final decision has to be

supported. It is important to mention, that the objective of those meetings is to take the

time to really analyze the alternative, rather than enforce the implementation of it. At the

end, the final decision would be made by each installation. Not all alternatives have to be

implemented because in some cases the alternative would be not feasible to implement

83
to that specific process or they are just handling the risk with different alternatives that

are valid as well.

Some alternatives can be applied across the organization and do not require the

identification of potential targets. Therefore, for those alternatives the corporate learning

team should develop an action plan and execute it as part of their job. One example for

this type of alternatives can be a modification to a procedure that is applicable to all

facilities. In this context, the team together with a subject matter expert would be

responsible for making the required changes, documenting the changes, incorporating

them into the DMS, communicating it and ensuring training as needed.

4.3.3.4. Define metrics

This step requires asking the questions of how they can validate what they want

to achieve? How they can know the alternative is operating as expected? Defining

performance indicators allow people to verify the system behavior and measure how

much it has been changed with respect to the initial set point. Indicators should be

determined prior to implementation in order to have a clear picture of what they want to

achieve with it. Likewise, it would help to verify the results since initial phases.

Indicators would change depending of the actual phase of the alternative and would give

real data to determine if the alternative is giving the expected results.

Conversely, indicators for the corporate learning system should be determined as

well, to evaluate the performance of the system and identify improving opportunities for

it. In this sense, indicators to analyze how the system operates, personnel performance,

feedback and results of the system should be defined and evaluate periodically.

84
Indicators such as the number of identified alternatives vs. the number of alternatives

implemented, percentage of people trained on how to corporate learning system works

and how to take advantage of it, percentage of alternatives that are rejected by the

facilities, the number of deficiencies identified in the system, or number of management

review of the system can be used as metrics for the system.

4.3.3.5. Refine alternative

Defining metrics and refining the alternative should be performed at the same

time, because once the facility is fitting the alternative to their own processes, some

modifications or alterations to the initial idea might be necessary. However, initial

metrics have to be determined to set the general goal of the alternative. Refining the

alternative refers to the process in which the facility has to define the limitations,

assumptions, boundaries and required resources based on their own needs and capacity.

Similarly, refining the alternative requires to perform a management of change process

in order to determine the potential impact to the process and people and the additional

considerations that need to be made prior implementation. Finally, an action plan has to

be developed, in which a detailed list of tasks, timeline and resources should be

evidenced.

4.3.3.6. Implement alternative

Implement the alternative refers to the actual execution of the action plan that has

been proposed by the facility. Each facility would be responsible for the execution and

verification of it. However, the corporate learning team should verify periodically its

85
current status. Likewise, the corporate learning team would support the whole process

based on what has been agreed during the evaluation phase of the alternative.

4.3.3.7. Monitor performance

Monitor performance refers to continuous improvement to validate and refine the

implemented alternatives and corporate learning system. At this point, a detailed review

should be performed to verify if the knowledge was successfully transferred, if the

knowledge was appropriate, was analyzed in detail, accurate criteria was used, and the

expected benefits were achieved. Moreover, the team has to analyze if the

implementation was performed as expected and what can be improved from it. Based on

the improvement opportunities that have been found during this analysis, the alternative

can be improved and refined as needed or can be modified if the expected results were

not successfully achieved. Additionally, the corporate learning system should be

validated in order to determine if the process has been successfully implemented across

the organization. In this analysis the corporate learning team should evaluate if the

objectives have been achieved, if people are fully committed with the system and feel

part of it, if the organization is getting the right knowledge, if the level of competence in

the team is adequate, and how the cycle and each phase of it can be improved.

4.4. Additional elements to achieve organizational learning

In the learning system presented in the first section of this section, three types of

learning were identified: individual, organizational and industrial learning. Learning is

achieved through the small entity to a bigger one. In this case, industry learning only can

86
be achieved once and organizational learning has been achieved. Similarly,

organizational learning is achieved once individual learning has been achieved for

people involved in the organization. Even though this research has been focused on the

understanding and improvement of organizational learning, it is important to mention

that achieving individual learning becomes an essential part of achieving organizational

learning. Therefore, organizations need to provide the necessary tools and resources in

order to support and enhance individual learning for all employees. As a result,

individual learning collectively is transformed into an organizational learning and at the

same time, an effective organizational learning is able to support each phase of

employee's individual learning as presented in Figure 15.

Figure 15 Improving organizational learning

Supporting the individual learning can be achieved through the development of

corporate learning systems in which the knowledge is being managed within the

organization and a learning environment is encouraged, as been discussed in the

previous sections. Conversely, transcending the individual learning to an organizational

one requires the implementation of adequate training programs and supporting

87
techniques, which enables people to individually acquire new information, analyze it and

apply it into their daily jobs. Some of these elements are briefly discussed below.

4.4.1. Effective training programs

Developing and delivering effective training programs ensure the basis for a

successfully corporate learning system, which enables employees to gain an

understanding of the main goals, structure, and expectations of the organization. Thus,

an effective training program should address the following elements:

Detailed training needs matrix for all employees’ position in the organization.

Specific qualification criteria for trainer’s selection. These criteria should include

the necessary technical knowledge that is intent to be transferred as well as the

necessary soft skills such as good communicator, presentation skills, able to open

healthily discussions, etc.

Expectations, goals and final product of each training course.

Up to date training material, which covers all relevant technical knowledge and

can be easily understood for the trainees.

Required tools, resources and infrastructure to perform the predefined training

courses.

4.4.2. Effective training courses

As discussed in section two, organizations should apply different training

techniques in their training programs based on the predefined expectations of the course

and the current needs of the organization. Although lectures methods are the most

common applied technique in most organizations, this method is not able to support

88
skills and attitudes development. Thus, it is important to recognize the use of multiple

training techniques in order to be able to provide a successful training program, which

enables employees to acquire and develop knowledge, skills, and attitudes. Similarly,

higher levels of retention’ techniques such as group discussions and practices cannot be

fully achieved without the understanding and analysis of the theory behind it. For

instance, organizations can implement the use of simulators to teach operators how to

perform a specific task and achieved the necessary skills to perform it. However, if the

use of simulators is not supplemented with additional training techniques such as

lectures, readings, and discussions, it is likely that the operator is not going to be able to

make decisions on his own once he experiences an abnormal situation during his real

job. The operator would not have the required technical knowledge and analysis to

understand the process behavior and potential outcomes of it.

Once the training techniques have been defined, it is also important to structure

the development of the course and the elements that trainers must take into consideration

during the preparation, execution, and evaluation of the course. First, trainers have to

determine the objectives of the course; this requires the understanding of the target and

their motivation. In this sense, the challenge is to identify why is this new knowledge

relevant to them, instead of thinking why it is important for the trainer or the

organization. This with the objective of answering the questions: do they want to know

that? Or do they need to know that? And based on it, trainers can take the best approach

for creating ownership about the knowledge that is being taught. Secondly, the

development of the course should be problem-solving oriented in which they are able to

89
analyze situations and make decisions about it. In addition, it should encourage trainees

to seek for more information and create spaces where they have to explain to others what

they are learning. This would allow trainees to digest the information and take the time

to consolidate this knowledge and fitting it into their existing knowledge. Thirdly, each

lesson has to clearly identify the benefits trainees can get for it in order to maintain their

motivation. Likewise, it is always a good approach teaching with examples because this

enables them to be familiar with the situation and recreate this new knowledge into their

own situations, and at the same time, it increases trainer’s credibility. Finally, trainees

should receive constant feedback during the development of the course and an

assessment should be incorporated as part of the process to help trainees refresh concepts

and identifying the key elements of the course. This assessment should be problem-

solving oriented [107-109].

4.4.3. Supporting unlearning processes

As Pighin and Marzona stated, “unlearning is defined as throwing away concepts

learnt in the past to give space for possible new learning” [110]. In this sense,

organizations should support employees in the transition from the old knowledge to the

new one and provide the required tools and resources to make the transition easier at

both individual and organizational level. In the process safety field more attention have

been given to the development of management of change programs in order to ensure

that risks are controlled once the organization makes changes in their facilities,

operations or personnel. However, less attention has been given to the cultural and

organizational factors, as well as the individual impact that may be influenced by those

90
changes. Therefore, management of change programs should incorporated individual,

organizational and cultural aspects that may be also influenced by the changes that are

made in the organization. The unlearning process can help organizations to identify the

potential impact of those changes and provide the corresponding plan to overcome the

resistant to changes attitudes that may arise from learning processes.

Some of the elements that people who are intended to transmit the new

knowledge should take into account are briefly discussed below. First, the acquisition of

new knowledge is not always an easy process for some employees, and it is even harder

when they have been doing their job in a particular way for many years. Therefore, it is

important to identify the potential impact of this change (new knowledge) and develop

mitigation actions to overcome these factors. Secondly, it is essential to understand that

there is going to be resistant of change because it is part of human behavior and it cannot

be prevented. But at the same time, it is important to note that the resistant of change can

be temporary if proper actions are taken to manage it. Thus, instead of having a negative

attitude of what employees are going to say or how they are going to complain about it,

organizations should be focusing on predicting those behaviors and developing

corrective actions for those attitudes. Thirdly, once the person in charge of transmitting

the new knowledge is in contact with the target, he has to explain to them the

requirements of the process. This means, to explain why this change is necessary for the

organization and why this new knowledge is important for them, instead of start

explaining how to perform the new task or going in detail to the technical part. This step

gives them the opportunity to analyze how these new requirements fit into their existing

91
knowledge and potentially identify some gaps during this process. Finally, the new

knowledge can be transmitted, follow by the identification of resistant to change

attitudes and the implementation of corrective actions. This process should be performed

over and over again until the new knowledge has been consolidated in all employees

involved in the process [110].

92
 5. ENHANCE INTERNAL INFORMATION

In this section, the first step of the learning process enhance internal

information is introduced in detail. This section presents the incident investigation

process throughout the development of a task analysis in order to analyze the

relationship among the people, the task, and the environment for a specific task. The

corresponding results were then used to enhance the existing incident investigation

process by the incorporation of additional steps into the process. Finally, a description

of each of the incorporated steps is provided.

5.1. Task analysis

A task analysis is performed with the objective of enhancing the understanding

of how the people, environment, and task work together and how the system is

supporting this process [111-113]. A hierarchical task analysis has been developed for

the incident investigation process to analyze the task’s objectives in this process, the

people involved in it, the complexity of the task and the expected goals for each stage in

the process. The general structure of an incident investigation process involves the

following steps: reporting, collecting evidence, analyzing evidence, identifying the

causes of the incident, developing recommendations, and sharing lessons learned. A

more comprehensive description of the incident investigation process has been presented

in section 3. The incident investigation process has been divided into subtasks and the

most challenging tasks have been selected for further analysis. Each category has been

93
broken down into more detail tasks to understand specific goals of each step of the

process. Nine subtasks have been selected for detailed analysis in which the common

errors were determined. Finally, the general characteristics of the people and

environment for the selected subtask were discussed.

5.1.1. Hierarchical task analysis decomposition

Figures 16-18 present the hierarchical task decomposition for the incident

investigation process. The steps highlighted in gray represent the subtasks that have been

considered the most complex and challenging in the process. Thus, a further analysis has

been performed in the next section. The following subtask were selected:

Report incident

Select investigation team

Visit scene:

- Interview witnesses

- Gather preliminary evidence

Analyze evidence

Apply incident investigation methodology

- Identify root causes and contributing factors

Develop recommendations

Define responsible and resources

Monitor status of recommendations

5.1.2. Hierarchical task analysis results

Tables 1-7 present the results obtained from the hierarchical task analysis.

94
Figure 16 Task analysis decomposition part 1
95
Figure 17 Task analysis decomposition part 2

96
Figure 18 Task analysis decomposition part 3

97
Table 1 Hierarchical task analysis part 1

Task People Environment Why is Challenging? Common Errors

Requires a strong safety culture within

Varies depending of the organization Tendency to cover up
the safety culture mistakes
Employees have to be engaged with
Mechanism for their work Employees fail to
Report reporting understand the benefits
Employees
incident The organization has to ensure an open they can get from the
Level of engagement and healthy reporting environment in investigation of incidents
of the top which individual blame is avoided
management Top management tends
Employees feel that they are going to to blame employees
lose their jobs

Select investigation team

Pressure against Requires a pool of trained people in based on availability of
time incident investigation personnel
Safety
Select Level of engagement Availability of personnel (100% of their Selected people have
Department
investiga of the top time for a short period of time) little or no training in
tion team management incident investigation
Management
Based on the complexity of the system process
Working with others where the incident occurred, specific
to make decisions expertise is required Required expertise in the
team is ignored

98
Table 2 Hierarchical task analysis part 2

Task People Environment Why is Challenging? Common Errors

Pressure against time

External entities involved in the Inexperienced people

process Witnesses’ availability performed interviews

Working with new people to Memory recalls change over Tendency to blame
make decisions time employees

High volume of pressure, Witnesses may be affected by a The interviewer

Incident emotions, and uncertainty combination of emotions attempts to get what he
Interview
investigation thinks happened
witnesses
team The incident investigation team The information may not be instead of listening
set aside their routine work accurate and witnesses may what the witness have
have different versions of the to say
The investigation is a temporary same incident
job Witnesses bias: they
Witnesses can make wrong say what they think
Different types of witnesses: assumptions happened or hide facts
Employees, usually operators they do not want to tell
that are part of the process

99
Table 3 Hierarchical task analysis part 3
Common
Task People Environment Why is Challenging?
Errors

Pressure against time

External entities involved in the The site may be severely

process damaged

Working with new people to make External regulatory

decisions agencies may have control
of the site making access
Gather Incident High volume of pressure, emotions, difficult Loss of
preliminary investigation and uncertainty relevant
evidence team Some physical evidence evidence
The incident investigation team set degrades with time
aside their routine work
Degree of difficulty of
The investigation is a temporary job understanding of the
incident during the analysis
Complexity of the system where the phase
incident occurred

100
Table 4 Hierarchical task analysis part 4

Task People Environment Why is Challenging? Common Errors

Pressure against time Requires expertise in incident

investigation methodologies
External entities involved in
the process Requires expertise in the
process where the incident Incorrect assumptions
Working with new people to occurred, the chemical
make decisions involved, and the organization Insufficient evidence to
select the most likely
High volume of pressure, It is hard to make assumptions scenario
Incident emotions, and uncertainty about what happened
Analyze
investigation Inadequate incident
evidence
team The incident investigation Some evidence may be investigation
team set aside their routine missing. methodologies and
work procedures
Pressure from different
The investigation is a departments Insufficient analysis
temporary job and time spent on it
Degree of difficulty of
Complexity of the system understanding of the incident
where the incident occurred during the analysis phase

101
Table 5 Hierarchical task analysis part 5

Task People Environment Why is Challenging? Common Errors

Pressure against time

Requires expertise in
External entities involved in the
incident investigation
process
methodologies and analysis
Working with new people to
Requires expertise in the
make decisions Identification of
process where the incident
intermediary
occurred, the chemical
Identify High volume of pressure, causes as root
Incident involved and the
contributing emotions, and uncertainty causes of the
investigation organization
factors and incident
team
root causes The incident investigation team
Pressure from different
set aside their routine work Improper analysis
departments
of results
The investigation is a temporary
Degree of difficulty of
job
understanding of the incident
during the analysis phase
Complexity of the system where
the incident occurred

102
Table 6 Hierarchical task analysis part 6

Task People Environment Why is Challenging? Common Errors

Pressure against time

External entities involved in

the process Reach an agreement with Superficial
all the different groups of recommendations
Working with new people to interest
make decisions Unrelated
Incident
Recommendations have to recommendations
investigation
Develop High volume of pressure, respond to each identified with the identified
team
recommendations emotions, and uncertainty cause in an effective causes of the
manner incident
Management
The incident investigation
team set aside their routine Investigators can write Ambiguous or
work ambiguous or unclear unclear
recommendations recommendations
The investigation is a
temporary job

103
Table 7 Hierarchical task analysis part 7

Task People Environment Why is Challenging? Common Errors

Employees’ availability
Inadequate risk-ranked
Pressure against time Budget is limited
of recommendations
and resources
There is no time for extra
Safety Inadequate top
Define Level of engagement work
Department management
responsible and of the top management
engagement
resources Top management has to be
Management
Working with others engaged.
Inadequate distribution
to make decisions
of resources
Requires a strong safety
culture within the
organization

Recommendations’ status is
Pressure against time Recommendations are
easily forgotten
and resources forgot and therefore are
never implemented
Safety Top management and
Level of engagement
Monitor status of Department employees have to be
of the top management Recommendation’
recommendations engaged
status is only
Management
Working with others responsibility of the
Requires a strong safety
to make decisions safety department
culture within the
organization

104
5.1.3. Subtasks analysis

Since each incident investigation is a new process, the people involved and the

environment is always changing depending on the severity of the incident and the unit

where the incident occurred. In addition, the number of people in charge of the

investigation, their backgrounds and previous experience regarding incident

investigations would vary as well. Furthermore, the frequency of this task cannot be

determined, since it would depend on the number of reported incidents and the type of

incident. In the incident investigation process three groups of people were identified:

employees who perform the task of reporting incidents, safety department and

management, and the incident investigation team. Each of this group of people executes

different stages during the incident investigation process and has different attributes.

Therefore, they need to be considered separated.

5.1.3.1. People: employees

Task: report incident

- People that work within the organization and perform a specific task on it.

Employees have different levels of education and backgrounds. However, employees

are expected to have the same training regarding policies, values and goals of the

organization. Additionally, employees are expected to have some preliminary

training and/or awareness campaign regarding incident reporting.

- Employees spend eight or more hours per day in their jobs. Reporting incidents is

considered as an extra task besides their routine work.

105
 - Employees need to be motivated and understand the benefits of it in order to report

an incident.

- Employees need to have general knowledge regarding safety and how to report an

incident to perform the task.

- No skills are required to perform this task.

5.1.3.2. People: safety department, management

Tasks: Select investigation team, define responsible and resources, monitor status of

recommendations

- The tasks mentioned above are specific tasks that are performed in conjunction with

two different groups of people: the safety department and the top management. The

combination of both gives the necessary background in process safety and

management to support decision-making. Thus, the safety department is in charge

of providing expertise in the selection of personnel and determining the required

resources to perform the investigation. Conversely, the top management is in charge

of providing the required resources and ensuring personnel availability.

- Their role is primarily decision-making to ensure that the investigation is executed

with the right people, resources and time.

- The safety department should have expertise in process safety and incident

investigation in order to determine the severity of the incident and resources needed.

- The top management should have knowledge in business and safety priorities, as

well as the available budget in the organization to support decision-making.

106
 - Both groups of people should have strong communication and leadership skills to

coordinate different groups of people and make decisions within a short period of

time.

5.1.3.3 People: incident investigation team

Tasks: Interview witnesses, gather preliminary evidence, analyze evidence, identify

contributing factors and root causes, develop recommendations

- The people that are part of the incident investigation team are usually employees

who have to suspend their routine job and perform the investigation as a temporary

job. Thus, people have to perform the task within a short period of time and under a

lot of pressure.

- The team is composed of different background and experience, which enables the

understanding of the incident by the combination of expertise. Depending on the

nature of the incident the incident investigation team would vary and different types

of background would be involved, such as engineers, safety specialist, maintenance

operators, contractors, lawyers, etc. Therefore, the people involved usually have

never worked together before.

- Each member of the team needs to be trained in incident investigation and if

possible, having previous experience in incident investigations.

- People within the team should have communication and leadership skills in order to

collect evidence, interview people and support the analysis process.

- The team has to have a strong technical knowledge about the process where the

incident occurred and the chemicals involved.

107
 - At least one of the members of the team should have a strong background in incident

investigation methodologies and root causes identification.

5.1.3.4. Environment

To perform this task successfully, it is required to have a strong safety culture

within the organization in which employees and the top management are engaged with it

and an open and healthy reporting environment is encouraged. Similarly, depending on

the level of engagement of management and employees, the right amount of resources

and time would be spent during this task.

The environment where the task is performed is the facility where the incident

occurred. These are chemical and oil and gas facilities such as refineries, platforms,

onshore facilities, etc. The investigation is executed at the same time that employees are

performing their job or some of them may be recovering from the consequences of the

incident. Furthermore, the unit where the incident occurred may be destroyed or heavily

damaged by the incident, making it difficult to have access to the information.

Reporting the incident is executed depending on the available tools that the

organization has for reporting incidents. This can be verbal reporting, written forms or

electronic forms. Likewise, the investigation analysis is executed depending on the

available tools and training within the organization. These tools refer to the incident

investigation methodologies that the organization has established depending on the

severity of the incident. Additional tools are the available procedures and guidelines for

the protocol prior the investigation, the development of the investigation, and required

post-activities.

108
5.1.4. Task analysis remarks

The task analysis developed for the incident investigation process gives more

relevant insights of the process. First, the quality of the investigation is proportional to

the level of training, previous experience performing the same task, and level of

expertise of the incident investigation team. Thus, organizations should have a pool of

experts in incident investigation methodologies as well as incident investigation

processes. Likewise, it is important to gather the right level of technical knowledge

based on the type of incident and its significance. Secondly, the safety culture of the

organization plays a fundamental role in the success of investigations, because it

determines the number of incidents that are going to be reported, the resources

availability, time spent, and engagement of the top management, as well as employees

who are in charge of the execution of recommendations. Similarly, the safety culture will

determine if recommendations are implemented and the whole incident investigation

process is completed. Third, communication, teamwork, and leadership skills are

significant contributing factors in the process. These skills enable the team to organize

relevant information, interview people, analyze the incident and work together in the

development of recommendations. In addition, these skills enable the team to complete

the investigation on time and with sufficient analysis. Finally, developing

recommendations was identified as one critical task in the process because it converges

all the effort and knowledge spent into the analysis. Moreover, recommendations are the

final output of the investigation. Therefore, the quality of it is going to be measured

based on the feasibility, clearness, and adequacy of the recommendations.

109
 Additionally, the task decomposition and analysis of it identified some of the

limitations of the process and improving opportunities on it. Since the environment and

people involved in the investigation are always changing, the organization miss valuable

information that can be used in future incident investigations. As a result, each

investigation is a new process in which the learning curve has to start from zero each

time. This gives the opportunity to enhance the process by the incorporation of

knowledge from previous incidents and extracts this knowledge for the incident that is

being analyzed. Similarly, the current incident investigation process does not support the

identification of required changes in the management system, which enables

organizations to identify cultural, management conditions, and system deficiencies that

increase the risk in the organization. Furthermore, the training material is not updated in

order to ensure a complete learning cycle of the incident. Moreover, the re-evaluation of

leading indicators of the process is not considered in the incident investigation process.

5.2. Proposed incident investigation process

Enhance internal information refers to the process in which the incident

investigation process is conducted and the final product of it have been analyzed and

additional steps have been incorporated into the process. The objective is to provide a

more detailed analysis of the incident and the safety management system, where the

incident occurred. Since incident investigations are one of the most powerful practices to

learn from experience, organizations need to focus their attention on improving their

incident investigation management programs to provide high-quality recommendations

110
and ensure that organizational and cultural failures are identified [77, 91]. Incident

investigation process refers to the sequence of stages addressed after an incident

happens. Starting from the report of the incident, followed by the collection of evidence,

incident investigation analysis, development of recommendations, communication of

lessons learned, and follow-up of recommendations [80, 81]. Regardless of the incident

investigation method used, the main structure remains constant. Thus, organizations miss

valuable information that should be analyzed during this process such as previous

incidents, leading indicators of the process, how the identified gaps fit into the current

training material, and how to incorporate this learning into the management systems of

the organization. Figure 19 presents a modified incident investigation process, in which

additional steps have been incorporated into the traditional process with the objective of

providing additional sources of information into the investigation. The grey boxes in the

figure represent the steps that have been included, while the white boxes refer to the

original structure of the incident investigation processes.

After an incident occurs and is reported, the significance of the incident has to be

determined. This refers to the categorization of the incident in which the potential impact

of the incident is evaluated. Based on this result, the level of detail in the analysis and

the complexity of it are determined. This means if simple methodologies and/or trend

analysis are sufficient or a detailed analysis is required. Then, the collection of

information and evidence has to be performed. This stage would determine the quality of

the identified causes and recommendations [80]. Once the information has been

collected, the incident investigation analysis is developed. In this stage incident

111
investigation methodologies are implemented for the reconstruction of events, the

identification of critical and causal events and the identification of root causes of the

incident.

Figure 19 Proposed incident investigation process

The following steps suggest a detailed analysis of previous internal and external

incidents from which valuable information can be extracted. The analysis of external

incidents gives the opportunity to retrieve missing lessons learned that the organization

failed to learn. Thus, the applicability of those lessons learned can be evaluated into the

112
process. The objective of reviewing external incidents is not only to analyze the

applicability of their recommendations to the incident that is being analyzed. It is also to

analyze the current practices that others organizations are implementing, compare their

own technology with respect to others organizations, and extend their existing

knowledge regarding national and international standards and best practices that may be

applicable to their facilities. Moreover, review of external incidents serves as support to

explain the causes of the incident to the interested parties. This means helping build trust

and credibility with regard to the results that the incident investigation team is

presenting.

Similarly, the analysis of internal incidents gives an opportunity to analyze

incidents from the same organization, the process where the incident occurred and the

safety performance of it. The analysis is based on the following criteria: are there any

similar incidents? Or is there a significant amount of incidents/near misses in the same

unit? The answer to the first question should be followed by a trend analysis and the

evaluation of the applicability of previous recommendations into the incident that is

being investigated. Conversely, the answer to the second question should be followed by

a revalidation of the process hazard analysis for the process where the incident occurred.

Then, the root causes that has been previously determined in this process have to

be re-evaluated in order to ensure a complete picture of the analysis in which

organizational and cultural flaws can be identified. Subsequently, the required changes

in the management system should be identified. This stage of the investigation aims to

provide more insights into the identification of root causes and successfully uncover

113
hidden faults of the system and avoid the identification of human error as root causes of

the incident. The following step refers to the identification of the technical gaps that

have been identified during the investigation. These gaps should be compared with

respect to the current training material and provide recommendations to ensure that the

training material is up to date and cover all relevant knowledge for each specific role in

the organization. The last additional step incorporated into the process refers to the re-

evaluation of the leading indicators of the process to determine if the process is showing

the right signals, the indicators are well defined and to ensure that operators understand

the data and information the system is giving. The main objective of the additional steps

is to ensure that root causes are identified and enhance the quality of the

recommendations. Finally, the recommendations can be developed, implemented and

lessons learned communicated.

114
 6. CASE STUDY

This section analyzes an incident that happened in the offshore industry in North

America. The incident has been examined from the perspective of the proposed incident

investigation flow chart presented in the previous section. The analysis comprises the

analysis of two external incidents with similar causes and two internal incidents that

occurred in the same organization. The data and information presented is real and

verified. However, data have been anonymized in order to protect the identity of the

organization. It should be note that the incident investigation has been performed by the

organization at the time where the incident occurred. Therefore, the objective of this

analysis is to identify improving opportunities of the final report through the analysis of

similar incidents in the industry.

6.1. Incident description

6.1.1. Platform fire and explosion

A high-pressure fuel gas line ruptured on a platform releasing high quantities of

high-pressure gas, which detonated within next 5 seconds. The explosion and resulting

fire killed 7 people and destroyed a significant part of the platform. The line was

designed as a fuel gas separator bypass in order to provide operational flexibility. Thus,

the line was used occasionally.

Prior to the incident, the line had been inspected twice, once two years ago and

another eight years ago. In both occasions, visual inspection, thickness measurement,

115
and materials characterization were performed. For those inspections, the results were

satisfactory and no reduction of the internal thickness was detected, with the exception

of a “weldolet” that was changed due to a severe external corrosion. The investigation

found that the line rupture was due to a severe localized reduction of the internal

thickness of the line. The laboratory analysis determined that the reduction of the

internal thickness was the result of a combination of a microbiologically influenced

corrosion (MIC) and corrosion associated with high levels of hydrogen sulfide (H2 S).

The identified causes of the incident, associated with prevention and detection of

the incident, determined by the organization in the final report are briefly summarized

below:

The fuel gas line was designed assuming that the potential presence of corrosion

bacteria was zero

Bacteria corrosion mechanisms were not considered during the process hazard

analysis performed in the unit

Safety, maintenance, and operational personnel were unaware regarding the

presence of bacteria corrosion mechanisms in the platform

The fuel gas line was designed assuming that there would be insufficient quantities

of H2 S in the line to cause significant corrosion

There was no system in place for real-time monitoring to identify high levels of

H2 S in the fuel gas lines of the unit

The fuel gas line was not identified as a bypass line in the mechanical integrity

program to design and perform the corresponding inspection program

116
 Personnel in the platform were unaware about the responsibility of communicating

to the inspectors about bypass lines and dead legs to put special attention during

the inspection

The potential combination of microbiologically influenced corrosion (MIC) and

corrosion associated with high levels of hydrogen sulfide (H2 S) was not considered

into the corrosion inspection program

The supervisor in charge did not have the required knowledge to analyze the

inspection results given by the inspection company

Based on the identified causes of the incident some of the recommendations are

briefly summarized below.

1. Identify and monitor the type, origin, and effect of presented bacteria in the platform

and design a tolerance level in order to establish control methods in the fuel gas

system

2. Update material selection criteria for fuel gas lines based on the types of damages

mechanisms that in combination can influence corrosion velocities

3. Analyze the operational feasibility of eliminating dead legs. Where applicable,

implement the use of corrosion inhibitors in the fuel gas system for the corrosion

mechanisms mentioned above

4. Ensure the participation of a certified and experienced person in offshore corrosion

mechanisms during the development of PHA’s

5. Establish communicating procedures of bacteria detection between the platform

personnel and inspectors. Additionally, identify all sporadic fuel gas lines in order to

117
 update the inspection program and communicate the required planning and

inspection

6. Implement a real-time monitoring system to identify high levels of H2 S in the fuel

gas lines of the unit

7. Implement corrosion coupons in the fuel gas system within the offshore platforms

8. Validate H2 S operational limits and evaluate the operational and maintenance

strategies for gas processing plants to ensure that permissible levels of H2 S are not

exceeded

9. Redesign the corrosion inspection program to ensure that special conditions in the

process such as bypass lines and dead legs and combination of corrosion

mechanisms such as MIC and H2 S are considered

10. Ensure that all personnel associated with planning, supervision, and execution of

activities related to inspection, non-destructive tests, and mechanical integrity are

certified and trained based on the applicable national and international regulations

(being enunciated but no limit API 570, 572, 510, 574, 580, 581, 653)

11. Develop and/or update static equipment inspection procedures to ensure that they are

developed based on the applicable national and international regulations.

Additionally, detection and control mechanisms for sporadic lines are included.

Finally, detection of corrosion mechanisms associated with the combination of a

MIC and corrosion associated with high levels of H2 S

12. Implement additional inspection techniques for the fuel gas lines that allows the

identification of corrosion mechanisms such as MIC and H2 S

118
6.2. Incident analysis

The first step incorporated into the incident investigation process refers to the

preliminary analysis in which the significance of the incident is determined. This means

to categorize the incident based on the potential impact that the incident may have with

the objective of determining the level of detail and complexity of the investigation. In

this study, the significance of the incident was determined as very high due to the

number of fatalities, injuries and property damaged. Thus, the incident was subject to

further analysis during the investigation. The following seven steps were developed

during the investigation performed by the organization: report incident, determine the

significance of the incident, gather information, develop sequence diagram, identify

critical events, identify causal factors, and identify root causes. The next step is the

review of previous incidents, which has been developed in this section.

6.2.1. Reviewing external incidents

Review process of previous incidents takes into account two different types of

incidents: internal and external incidents. Two incidents were considered for the external

incident review process: the first incident is BP North Slope oil spill in 2006 and the

second incident is natural gas pipeline rupture and fire in Carlsbad, New Mexico in

2000.

BP North Slope oil spill, 2006: On March 2nd of 2006, a BP operator discovered a

leak in a 34-inch transit pipeline. The leak turned out to be the largest spill ever

experienced by Alaskan North Slope. The cause of the leak was due to an internal

119
 corrosion that caused a hole at the bottom of the pipeline. The leak was associated

with a microbiological corrosion caused by sulfate reducing bacteria [114].

Natural gas pipeline rupture and fire in Carlsbad, New Mexico, 2000: On August 19th

of 2000, a 30-inch natural gas transmission pipeline ruptured. The released gas

consequently ignited and burned for almost an hour. The incident killed twelve

people and extensively damaged the infrastructure around. The investigation

determined that the major safety issues were associated with pipeline design, the

internal corrosion control program, the lack of federal safety regulations for natural

gas lines, and the inadequate federal oversight [115].

The objective of reviewing external incidents is not only to analyze the

applicability of their recommendations to the incident that is being analyzed. It is also to

analyze the current practices that others organizations are implementing, compare their

technology with respect to others organizations, and extend their existing knowledge

regarding national and international standards and best practices that may be applicable

to their facilities. Moreover, reviewing external incidents serves as support to explain the

causes of the incident to the stakeholders. This means helping build trust and credibility

with regard to the results that the incident investigation team is presenting.

6.2.1.1. Corrosion statistics in the offshore industry

The final report of the incident developed by the organization presented a

detailed analysis regarding the microbiologically influenced corrosion (MIC) and

hydrogen sulfide corrosion (H2 S), supported by laboratory analysis and subject matter

experts. However, due to the nature of the combined corrosion mechanisms, the

120
assimilation process from the operators, the supervisor and managers presented a

challenge. Even though, the evidence was there, it was hard for them to believe “such

unexpected event”. Therefore, some representative statistics in the offshore industry has

been presented in order to highlight the frequency of this type of incidents in the

offshore industry and consequently failure to learn from them.

According to the US Department of Interior Minerals Management Service,

almost 3000 offshore pipeline incidents were reported from 1970 to 1999, of which 51%

of those were associated with corrosion as shown in figure 20 [116].

Corrosion represents the leading cause of failure in the Gulf of Mexico offshore

pipelines, counting for more than 1483 offshore corrosion incidents over the past years.

Of these incidents, 35% were associated with internal corrosion and the remaining 65%

with external corrosion [116]. Over the period from 1989 to 1999, 75% of the total

number of reported corrosion incidents was associated with internal corrosion failures,

showing a direct connection between internal corrosion and pipeline aging. Statistics

indicates that additional factors such as pipeline infrastructure growing and changes in

operating conditions also increase the likelihood of internal corrosion failures [116].

Likewise, statistics show that over the period from 1970 to 1999, the majority of internal

corrosion failures occurred in natural gas pipelines, representing 67% (518 incidents) of

the reported incidents [116]. Moreover, according with the U.S. Department of

Transportation Pipeline and Hazardous Materials Safety Administration, even though the

number of reported corrosion incidents in gas gathering lines is not that high, more than

90% of those are caused by internal corrosion failures [117].

121
 US Gulf of Mexico offshore oil and gas
pipeline types of failures
3% 3% Corrosion
4% Natural Hazard
Impact
6% Structural
Unknown
8% Other
51% Material
21% Weld defect
Anothoring
Erosion
Construction

Figure 20 US Gulf of Mexico offshore oil and gas pipeline types of failures over the

period from 1970 to 1999, adapted from [116]

6.2.1.2. Corrosion monitoring program

BP North Spill incident report briefly outlined two key aspects that all pipeline

corrosion management programs in the oil and gas industry may have such as a

corrosion monitoring program and a pipeline leak detection system, which prevent a

pipeline from developing leaks and ensuring prompt response once a leak occurs.

At the time of the incident that is being analyzed, the organization did not have

an effective internal corrosion-monitoring program for gas lines in place. Likewise, the

platform did not have a pipeline leak detection system. The corrosion monitoring

program that the organization had at that time involved visual inspection, thickness

measurement, and hydrostatic testing. The investigation report recommend the

implementation of a real-time monitoring system in order to identify high levels of H2 S

and the implementation of corrosion coupons in the fuel gas system. Additional

122
corrosion monitoring system can be considered based on the analysis of the current

technology applied for different organizations. For example, BP North Spill incident

report highlights the use of “smart pigs” in order to provide more coverage and the

detection of smaller defects. Similarly, the investigation report developed by El Paso

Natural Gas Company highlights the implementation of ultrasonic testing on the non-

piggable portions of the pipeline and the gas quality monitoring to ensure that predefined

limits are not exceeded. Furthermore, the report briefly outlined some of the-state-of-the-

art technology described by the National Association of Corrosion Engineer (NACE)

such as the implementation of Molecular Microbiological Methods (MMM) to

monitoring internal corrosion mechanisms in the pipeline. Finally, BP North Spill

incident report also emphasizes the technology that was implemented after the incident

with respect to their pipeline leak detection system. The organization increased the

number of field inspections and implemented infrared heat detectors to improve leak

detection.

Based on the results of the previous review of the two external incidents with

regards to internal corrosion monitoring program and a pipeline leak detection system,

the following recommendations can be incorporated into the final report:

Recomendation1: Evaluate the implementation of the most suitable technology for

internal corrosion monitoring in the gas pipeline system. The implementation of smart
pigging, ultrasonic testing, and gas monitoring should be analyzed with respect to the
existing technology.

Recomendation2: Evaluate the feasibility of the implementation of the-state-of-the-art

technology such as molecular microbiological methods in order to monitor bacteria
activity.

123
 Aligned with the previous recommendations and some of the recommendations

stated in the report, the organization failed to effectively perform a complete

management of change program for the incorporation of new technologies or systems

into their existing safety management system. Thus, the required resources, expertise,

and management commitment were not sufficient to fully prevent and mitigate this type

of incident. In this particular case, the organization failed to provide the required training

to the personnel in charge of analyzing and providing corrective actions during the

inspections that were part of the corrosion-monitoring program. In this sense, even if the

inspections were performed based on the schedule, there was a lack of expertise to

conclude what was a good or bad result and provide the technical knowledge to

determine the following corrective and preventing actions. Moreover, during the

inspections performed in the unit, the replaced pieces were never subject to any type of

further analysis in order to analyze the identified corrosion mechanism and the potential

presence of additional corrosion mechanisms. Thus, bacteria activity and high levels of

H2 S were not identified during monitoring activities. El Paso Natural Gas Company

incident report discussed some of the post-inspections activities that the organization

performed as part of the corrosion-monitoring program. The organization selected some

fractions of pipeline that was changed during inspection activities and performed

metallurgical examinations to identify additional mechanisms of corrosion.

Based on the previous discussion, the following recommendations can be

incorporated into the final report:

124
Recommendation 3: Enforce the management of change program for the
implementation of any new technology or system in which the required training,
responsibilities, procedures and channels of communication must be specified.

Recommendation 4: Performed laboratory analysis of some of the portions of pipeline

that have been replaced during inspections, in order to identify additional potential
mechanisms of corrosion.

6.2.1.3. Applicable standards

The incident investigation report highlights some relevant applicable standards

that must be incorporated into employees and contractors training. Likewise, the report

recommends the inclusion of those standards into their existing inspection procedures.

Those standards mentioned in the report are: being enunciated but no limit, API 570,

572, 510, 574, 580, and 581, 653. However, additional standards can be taken into

consideration.

Complementary to the recommendation described in the report, El Paso Natural

Gas Company incident report provides a more compressive description of the applicable

standards regarding corrosion issues. This can help the organization to expand their

knowledge and be aware of more sources of applicable information. Some of the

applicable standards are mentioned below:

- NACE SP0106-2006: Internal Corrosion Control in Pipelines

- NACE Standard TM0106- 2006: Detection, Testing, and Evaluation of

Microbiologically Influenced Corrosion (MIC) on External Surfaces of Buried

Pipelines

- NACE SP0206-2006: Internal Corrosion Direct Assessment Methodology for

Pipelines Carrying Normally Dry Natural Gas

125
 - ASME B31.8 Gas Transmission & Distribution Piping

- ASME B31.8S Managing System Integrity of Gas Pipelines

- ANSI GPTC Z380.1 Guide for Gas Transmission, Distribution, and

Gathering Piping Systems, 2015 Edition

Additionally, it is important to mention that those standards should not be used

only for training and procedures purposes. It is important to verify that the current

corrosion management program and internal policies and guidelines are aligned with

those applicable standards. Thus, a complementary recommendation can be proposed:

Recommendation 5: Revise and update as required the corrosion management program,

policies and internal guidelines in order to ensure that are aligned and comply with all
the applicable national and international standards. Some of the applicable standards are
mentioned below:
- API 570, 572, 510, 574, 580, 581, 653
- NACE SP0106-2006: Internal Corrosion Control in Pipelines
- NACE Standard TM0106- 2006: Detection, Testing, and Evaluation of
Microbiologically Influenced Corrosion (MIC) on External Surfaces of Buried
Pipelines
- NACE SP0206-2006: Internal Corrosion Direct Assessment Methodology for
Pipelines Carrying Normally Dry Natural Gas
- ASME B31.8 Gas Transmission & Distribution Piping
- ASME B31.8S Managing System Integrity of Gas Pipelines
- ANSI GPTC Z380.1 Guide for Gas Transmission, Distribution, and
Gathering Piping Systems, 2015 Edition

6.2.1.4. Corrosion management program

Both external incidents emphasize the main elements involved in any corrosion

management program and the required procedures and guidelines that may be in place in

order to ensure that effective actions are performed with respect to prevention,

mitigation and control. A corrosion management program should clearly address the

126
following elements: definition of policies and objectives, organizational structure and

responsibilities, planning, procedures, and implementation, measuring system

performance and finally reviewing system performance [91].

Based on the incident report, the investigation revealed that the organization

presented weaknesses in their existing corrosion management program. Analyzing some

previous incidents in the same platform it can be seen that the platform present failures

in their mechanical integrity program in general, not only with respect to corrosion.

Although, inspections and corresponding activities are usually performed in accordance

with the schedule (implementation phase of the corrosion monitoring program). In their

program, some flaws can be identified in the planning and measuring system

performance phases. The incident investigation reveals that during the planning phase,

not all corrosion threats were identified and the required actions in case of the

identification of any failure were not addressed effectively in the program. Likewise, the

organization failed to effectively review and analyze the system performance in order to

validate the effectiveness of the current corrosion prevention and monitoring methods.

Based on this analysis, the following recommendation can be incorporated into the final

report.

Recommendation 6: Validate the corrosion management program to ensure that all

types of corrosion threats are assessed and a risk assessment is performed. The program
should incorporate the following threats: Internal corrosion threat, external corrosion
threat, safety/hazard threat, environmental threat and operability threat.

6.2.2. Reviewing internal incidents

For the internal incident review, two incidents were considered. Both incidents

occurred on the same platform 3 months and one-year back to the incident that is being
127
analyzed. The incidents were selected based on the criteria of cause similarity and

incident unit. Additionally, documents such as previous process safety performance

indicators, reliability report, and the mechanical integrity manual, were considered into

the analysis.

Incident 1: A bitter water leak was detected in a 6” line from the low-pressure gas

rectifier FA-3104 to the pressurized system, due to a material loss in the line. The line

was out of service at the time of the incident. The investigation determined that the

leak was caused by severe corrosion in the line. This due to the following causes:

presence of trapped water on the line, there was not an action plan in place for

removing off-line interconnections, and there was a failure to properly analyze and

take actions of the inspections results developed by third parties.

Incident 2: During maintenance work in a mechanical valve, a minor gas leak was

detected in the stem of the valve. The leak was controlled half an hour later. There

weren’t any injured, environmental damaged or production losses. The investigation

determined that the cause of the incident was due to a deterioration of the stem valve.

Additionally, a severe corrosion was detected in the valve.

While reviewing internal incidents, two approaches can be taken as presented in

Figure 21. First, analyze if there are a significant number of incidents in the same unit

where the incident occurred. The objective of the analysis is to find the need for re-

evaluating the process hazard analysis of the process. Secondly, to analyze similar

incidents that had occurred in the same organization, to find trends and the applicability

of those recommendations to the incident that is being investigated. Finally, the incidents

128
investigation team should go back to the identified root causes and re-evaluate them

based on the identified findings.

Even though, there was no record of a significant number of incidents in the same unit.

The reported incidents have been classified with high significance. Moreover, the

investigation report determined the Process Hazard Analysis (PHA) performed in that

process did not consider all different types of corrosion threats that can be present in fuel

gas systems. Thus, it is highly recommended to re-evaluate the performed process

hazard analysis, analyze those results and provide mitigation actions as required.

Moreover, it is important to ensure that is executed by qualified personnel with relevant

experience in offshore operations, corrosion mechanisms, and hazard identification

methods.

Figure 21 Reviewing internal incidents

129
 In the second part of the analysis, similar incidents with similar causes were

considered, along with additional process safety documentation that helped to provide

more evidence into the analysis. The following improving opportunities to the existing

investigation report were proposed:

Recommendation 7: Re-evaluate the process hazard analysis performed in the fuel gas
system of the platform, ensuring that is executed by qualified personnel with relevant
experience in offshore operations, corrosion mechanisms, and hazard identification
methods.

6.2.2.1. Incident investigation program

A third party hired by the organization was in charge of the development of the

investigation report of the incident that is being analyzed. While both the internal

incidents reports that were selected for this analysis, were developed by people inside of

the organization. Although, the severity of both internal incidents is relatively low

compared with the incident for the case study. Some inconsistencies were identified with

respect to the level of detail of the analysis, the real identification of root causes, and

quality of the recommendations. The analysis exposed that the organization had an

inadequate incident investigation program, which inhibits the organization to learn from

previous incidents. In this sense, it was unlikely that the organization would prevent a

catastrophe like the one in this case study, because the identified causes in previous

incidents were limited to the resolution of direct causes such as physical or human

causes (changing a valve, replacing a section of the equipment or an employee

temporary suspension). Aligned with this, a culture of blame was identified on the

previous reports in which some recommendations tended to punish the operator instead

130
of analyzing the causes behind that behavior. Moreover, even if human and

organizational causes were identified as root causes of those incidents, the following

recommendations were focused just on immediate and physical causes. Hence, flaws

were reflected in the incident investigation program and the provided training in this

topic. In conclusion, a significant gap was identified between incident investigation

reports developed in-house and incident investigation reports developed by a third party.

Recommendation 8: Validate the current incident investigation procedure and

methodologies applied by the organization, to guarantee that the identification of root
causes is achieved during these processes. Likewise, ensure that the organization has a
pool of experts in incident investigation methodologies as well as the overall incident
investigation process.

Recommendation 9: Validate and update their current incident investigation training

material in order to ensure that the incident investigation process, requirements, and
applied methodologies are explained with sufficient detail.

Recommendation10: Ensure the investigation team receives a refreshing training prior

any incident investigation, in order to guarantee that people in the team understand the
main steps in the process, their role and responsibilities, and the expected outcomes of
the investigation.

6.2.2.2. Implementation of recommendations

During the analysis of previous incidents, both incidents emphasize a failure to

properly analyze and take actions of the inspections results developed by third parties.

The incident used for this case study exposes the same cause in which the supervisor in

charge was not able to analyze inspections results due to a lack of knowledge and

training regarding this type of data. This is an illustration of how a lack of commitment

with respect to the implementation of recommendations can lead to big catastrophes

131
such as the one that is being analyzed. Furthermore, this reflected a lack of enforcement

and management commitment regarding the complete achievement of process safety

management programs. Thus, even if activities were performed based on the program,

no actual actions were performed with respect to the analysis and recommendations of

those results, making it hard for the organization to improve and make their processes

safer.

Additionally, in incident 1, one of the recommendations suggested the

development of an action plan for removing off-line interconnections in the platform.

The recommendation was ranked with high priority. Despite the associated risk, the

recommendation was never implemented and the same recommendation was identified

in the final report of the incident used in this case study.

Recommendation 11: Ensure that recommendations from incident investigations,

inspections or audits are easily understood for the operators and supervisor in charge of
the process. Likewise, ensure that the personnel in charge of the implementation of those
recommendations have the required knowledge and training to interpret this type of
information. Moreover, ensure an adequate follow-up program in which the periodic
status of recommendations is reported to the management level.

6.2.2.3. Safety management system

In analysis of the internal documentation mentioned previously and the incident

in this case study, it can be concluded that there is a lack of proper channels of

communications among the different groups of people that interact in the same process.

This means, that people in charge of the development of process hazard analysis,

inspections, and the daily operators are not aligned with respect of the identified hazards

in the process and how they have to work together in order to define the best approach to
132
take for the process. With all things considered, it was concluded that the different

elements of the safety management system of the organization do not aligned properly

and do not interact with each other. Thus, the mechanical integrity, training or

management of change programs are not fully aligned with the process hazard analysis

and compliance audits elements of the safety management system in the platform.

Recommendation 12: Ensure that safety management programs of the organization are
aligned with the process hazard analysis and audit programs performed in the facility.
Ensure appropriate channels of communication, training and qualified personnel to
defined effective prevention and mitigation measures.

6.2.3. Identifying changes required in the management system and updating

training material

Following the review of previous incidents, the next step in the proposed process

is the identification of the changes needed in their management systems, which are

exposed during the development of the incident investigation. This stage of the

investigation aims to provide more insights into the identification of root causes and

successfully uncover hidden faults of the system. Through the analysis of the identified

causes of the incident, review of previous internal and external incidents, and an analysis

of the current state of their management system, relevant changes in the management

system were identified for this case study.

The analysis suggested that changes in the current training need-matrix are necessary

to ensure that personnel have the required knowledge and experience to perform

their responsibilities on the platform. This means to validate that the required

training for a specific role is up to date and the identified gaps (Corrosion

133
management program, applicable corrosion standards, incident investigation

program) are filled by the organization.

The management of change program should be re-evaluated to ensure that changes in

technologies or processes take into account the potential impact on training needs,

roles responsibilities, and channels of communications among the organization.

The mechanical integrity program should be validated to assure that the planning

phase of the program is developed for a group of qualified personnel with relevant

experience of the process and products in the platform. Likewise, the analysis

suggested validating the current personnel selection criteria for the development of

process hazard analysis.

The investigation suggested an overall change in the safety culture of the

organization in which the level of engagement and awareness of the top management

need to be re-evaluated, as wells as the operational discipline and incident

investigation program.

Within the suggested changed in the safety culture of the organization, the

operational discipline program should be re-evaluated in order to ensure that the

organization is taking an appropriate approach for communicating and enforcing

procedures, standards and policies. Additionally, the audit program must be validated

to ensure that the auditors are qualified to perform the job and that audit procedures

are adequate, available and understood by employees.

Together with the two previous arguments, the incident investigation program should

be re-evaluated to ensure that appropriate methodologies are applied and that the

134
 incident investigation process, procedures, and roles and requirements are well

established and understood.

Aligned with the analysis performed in this step, it is also important to evaluate

which training material should be updated based on the technical gaps that were

identified during the investigation. For instance, during the development of this case

study, technical and organizational gaps such as internal corrosion monitoring program,

identification of corrosion threats, and applicable corrosion standards and best practices

we identified as current gaps in their training material. Similarly, the incident

investigation training material should be validated to ensure that proper methodologies

for root cause identification are implemented within the organization and the basis of an

effective incident investigation program are covered in this training.

6.2.4. Closing up the process

The next step in the analysis is to re-evaluate the leading indicators of the process

where the incident occurred. This with the objective of identifying the fact that if the

process is showing the right signals and making sure those operators understand the data

and information the system is giving. The case study identified that appropriate

indicators were in place. However, performance indicators reports were usually

presented without any analysis behind it, making difficult to understand the actual state

of the process. The presentation of these indicators was limited to the comparison of the

result with respect to the predefined goal of the year. In this sense, the results are

presented in a very high level and no analysis of the data was performed. Thus, the

learning process cannot be achieved because no actual transformation of the data into

135
information and knowledge is performed. Given these arguments, the following

recommendation can be incorporated into the final report.

Recommendation 13: Re-evaluate the objective of the existing leading indicators in the
process, in order to verify how the results can help to understand the process behavior
and what actions can be taken with respect to these results. Similarly, ensure that
performance indicators reports are followed by the corresponding analysis of the
obtained results.

Thirteen additional recommendations were developed to the final incident

investigation report. Additionally, six changes in the existing management system and

the required changes in their current training material were suggested. The

recommendations need first to be validated for people inside the organization and the

management's approval is required. The next steps in the process are to assign the

recommendations and finally share the lessons learned. Those steps are out of the scope

of this case study. However, the results are going to be presented to the organization and

the following steps will depend on the decision made by them.

6.2.5. The case study into the learning process

The first step in the learning process has been explained to illustrate how incident

investigations can be enhanced and quality information can be incorporated into the

existing corporate learning system. The following steps in the process require the

implementation of a pilot test inside the organization in order to determine the actual

performance of the potential alternative and the system. Therefore, a hypothetical case

has been developed for one of the recommendations proposed on the previous analysis.

The steps two to seven were developed for recommendation 5.

Second step: acquire information

136
 Enhancing internal information also gives the opportunity to identify additional

sources of information that can be used in the future for different facilities or groups of

people. In this sense, the selected external incidents (BP North Slope oil spill, 2006 and

Natural gas pipeline rupture and fire in Carlsbad, New Mexico, 2000) can be grouped

together based on the type of incident or the associated causes (Internal corrosion). Then,

this information can be linked to the document management system, so people would

have access in case similar information is needed for future investigations or operational

inquiries.

Third step: Assess alternative

During this stage, the corporate learning team identifies the potential scope of the

recommendation, the benefits, potential cost, and the associated risk. Based on this

analysis, the recommendation would be considered as a potential alternative that should

be incorporated across the organization.

- Scope: the recommendation that is being analyzed involves the review of the

mechanical integrity program and corrosion guidelines of the platform where the

incident occurred. Since some of these documents are transversal for the

organization, the initial identification should be performed at a corporate level,

follow by the adjustment of internal procedures and implementation of specific

activities at a facility level.

- Benefits: the implementation of the alternative would ensure that the

organization is operating in a legal and safe manner. In this sense, the

organization would implement international corrosion standards, which enables a

137
 consistent interpretation and implementation of the corrosion management

program. Likewise, it would support external and internal auditing process to be

more consistent with respect to results of inspections and process indicators of

the process. Moreover, this would enable the organization to potentially

minimize legal liability and be more competitive around the world.

- Cost: the cost would be determined in terms of the number of people required in

the initial identification of gaps with respect to the applicable standards and the

current corrosion program and guideline, and the cost of a subject matter expert

who would serve as an advisory during this phase. Similarly, the time and

resources needed to update the documentation, the training required, and finally,

the time and resources needed for implementation of specific activities at each

facility level.

- Risk: the likelihood of having another similar incident associated with internal

corrosion would potentially decrease if appropriate corrosion management

guidelines are in place, people are trained, and the documentation is consistently

interpreted and implemented within the organization.

For the initial identification, a work team is required. In this team the corporate

learning team would work together with three representatives of the process safety

department (who are part of mechanical integrity team) and an external subject matter

expert with relevant prior experience in corrosion management as well as a strong

knowledge in international corrosion standards suggested in this recommendation.

138
 A preliminary plan has been developed to determine how the recommendation

would be implemented and the people in charge of the execution of each task. The plan

was divided into three phases: identification phase, implementation at a corporate level,

and implementation at a facility level. Table 8 to 10 present the preliminary plan for the

revision and updating of the corrosion management program and guidelines based on the

applicable international standards.

Fourth step: Define metrics

A set of indicators has been proposed to track the performance and progress of

the alternative. The indicators were defined based on the three phases of the action plan.

Identification phase:

- Percentage of standards reviews completed according to the schedule

- Number of non-conformances to non-regulatory corrosion standards per year

Implementation at a corporate level:

- Percentage of activities completed according to the schedule

- Percentage of people trained on the incorporated standards

Implementation at a facility level:

- Percentage of activities completed according to the schedule

- Percentage of people trained on the incorporated standards

- Number of facilities that identified the required changes and developed an action

plan

- Number of facilities that performed the required changes

- Number of audits in which the work team participated

139
- Number of incidents associated with corrosion per year

- Number of repeat non-conformances in all facilities per year

- Number of inconsistencies or issues identified in all facilities during the

implementation phase

140
Table 8 Action plan - Identification phase
Timeline
Item Activity Responsible
M1 M2 M3 M4 M5 M6

Identification phase

1 Review current corrosion management program Work Team

Review inspection and maintenance guidelines for

2 Work Team
pipeline systems

Review NACE SP0106-2006 and NACE Standard

3 Work Team
TM0106- 2006

Review NACE SP0206-2006 and ANSI GPTC

4 Work Team
Z380.1

5 Review ASME B31.8 and ASME B31.8S Work Team

Identify gaps in the current corrosion program and

6 inspection and maintenance guidelines for pipeline Work Team

systems

141
Table 9 Action plan - Implementation at a corporate level
Timeline
Item Activity Responsible
M1 M2 M3 M4 M5 M6 M7 M8

Implementation at a corporate level

Update corrosion management program and

7 inspection and maintenance guidelines for Work Team
pipeline systems
Define the appropriate risk tolerance criteria
8 Work Team
or guidance

9 Perform a management of change process Work Team

10 Document sources for potential changes Work Team

Define technical and regulatory knowledge

11 Work Team
needed for compliance

12 Define expectations for each facility Work Team

13 Define audit schedule to verify compliance Work Team

14 Communicate to all appropriate personnel Work Team

Provide training needed for compliance and
15 Work Team
auditing

142
Table 10 Action plan - Implementation at a facility level
Timeline
M M M M M M M M M M M M M
Item Activity Responsible M M
M
1 2
9 1 1 1 1 1 1 1 1 1 2 2 2 2
5 0
0 1 2 3 4 6 7 8 9 1 2 3 4

Implementation at a facility level

Update inspections and maintenance
16 Each Facility
procedures
Perform a management of change
17 Each Facility
process
Provide training needed for
18 Each Facility
compliance of procedures

19 Perform audits based on schedule Each Facility

Participate in audits as necessary to

20 ensure that auditors are qualified and Work Team

verify appropriate information

21 Perform review meeting Work Team

143
 Fifth step: refine alternative

Since the alternative has been designed transversal for all the organization. The

execution of the first two phases is required first in order to determine how the

alternative is expected to be implemented for each type of facility and consequently

determine the specific subtasks required for each of them. For the execution of the third

phase, the work team would define the expectations for each facility. This would

facilitate the level of detail in auditing and documentation for each facility.

Sixth step: implement alternative

This step involves the execution of each of the task defined in the action plan.

During the first phase of the plan, gaps were identified with respect to the integrity

management program elements. More specifically, gaps associated with the integrity

threat classification and the identification of the potential pipeline impact by threat.

Thus, the standard “ASME B31.8S Managing System Integrity of Gas Pipelines” was

the primary source of information for the following phases. During the second phase of

the plan, the corrosion management program and inspection and maintenance guidelines

were updated, communicated and the mechanical integrity personnel were trained. For

the last phase, each facility had to developed a sub plan in which activities such as the

reevaluation of the process hazard analysis need to be performed based on the results of

the management of change process. Moreover, each facility was in charge of defining

the people who need to be trained such as supervisor and operators.

Seventh step: monitor performance

For the last step of the process, review meetings were scheduled to verify the

144
progress for each facility, identified common difficulties, and analyzed preliminary

results of the alternative. Similarly, results of the proposed indicators for the alternative

were analyzed, refreshing training was provided and actions were taken based on these

results. Finally, the work team participated in some of the audits performed in the

facilities in order to verify that auditors are qualified and used the appropriate standards.

This case study was developed with the objective of explaining the proposed

incident investigation process and providing a clearer picture of how it can be

implemented within the existing incident investigation programs. Moreover, it illustrates

how investigations can be enhanced through a more detailed analysis to identify the root

causes and consequently achieve high-quality recommendations. Finally, it explains how

the case study fits together in the complete learning process through a hypothetical

implementation of one of the proposed recommendations.

145
 7. CONCLUSIONS AND FUTURE WORK

7.1. Conclusions

In this thesis, a systemic process for improving learning from incidents has been

developed for the chemical and oil and gas industry based on the identified limitations of

the learning system. It consists of seven steps that enable organizations to acquire

relevant safety information, absorb it and transform it into valuable safety knowledge

that can be incorporated into the organizational management systems. The framework is

intended to provide a holistic view of the learning from incident process and explore the

concept of integration of knowledge management into safety learning systems.

Additionally, the limitations of the learning system and the main elements that

organizations should take into account to improve learning have been discussed in order

to enhance understanding of this field. Finally, the proposed incident investigation

process was explained through the development of a case study in which an incident

investigation report was enhanced by the identification of improving opportunities for

the existing safety management system of the organization.

This work provides a holistic view of the learning from incident process, which

comprises the development of a corporate learning system, providing guidelines of how

learning systems can be executed within the organization and how to support the

implementation of safety knowledge inside of it. The proposed learning process is

intended to serve as a tool to disseminate and analyze safety knowledge across the

organization. Likewise, it is intended to support single and double-loop learning since it

146
converges a detailed analysis of the underlying causes of incidents, provides the

opportunity for analysis from different perspectives and examines of potentially valuable

information for the organization. Moreover, the proposed learning system gives some

empirical and theoretical insights into the implementation of learning incidents into real

organizations for the chemical and oil and gas industry. In this context, the framework

aims to provide a more integrated perspective to this field by the incorporation of

psychological and engineering inputs into the analysis. Finally, this work give a bigger

picture of the learning from incidents process and overcome some of the limitations that

the industry faced with regard to this field.

Knowledge management theory has been applied for the purpose of defining the

role of safety knowledge in an organization, developing a knowledge culture in which

knowledge transfer is encouraged, and finally developing a knowledge structure that

incorporated technical and human elements into the system [104]. The study of this

research is restricted to enhance internal learning in an organizational level. Therefore,

the scope of this thesis was limited to an organizational learning inside the organization.

The identified limitations in the learning system have been addressed as follow:

The implementation of lessons learned and sharing knowledge within the

organization have been the focus of this study. Even though, the available literature

highlights the importance of sharing and implementing lessons learned [15, 18], the

purposes of this study is focused ensuring the implementation of lessons learned in

all facilities where it may be applicable, and not just in the facility where the incident

occurred. Thus, the implementation of a corporate learning system inside the

147
organization aims to ensure that the lessons learned are not just communicated and

disseminated, but also analyzed and documented for all the facilities. Similarly,

sharing knowledge within the organization is encouraged in the corporate learning

system by the flows of information in and out that are established from and to all

facilities that are part of the organization.

The use and understanding of academic resources and databases are encouraged in

the second step of the proposed learning process: acquire information. The

identification of external resources has been identified as the key feature of this

process in order to get valuable safety information that may be applicable to the

organization. Thus, the importance of using databases, external associations, and

academic resources has been discussed.

The identification of root causes is enhanced by the incorporation of additional steps

into the traditional incident investigation process. These steps provide some

guidelines to the incident investigation team in terms of identifying more insights

during the analysis, by the incorporation of additional sources of information that can

give more clarity and detail with respect to the organizational, technical and cultural

flaws of the organization. Similarly, the proposed process gives a holistic view by

the identification of the required changes in the management system and current

training material of the organization. Additionally, the proposed incident

investigation process encourages the validation of leading indicators of the process in

order to ensure that operators and supervisors are getting the right signals to identify

technical failures in the system.

148
 The applicability of the proposed process has been explained through the analysis

of an investigation report of an incident in the offshore industry. The obtained results

provide a clear picture of how the process can be implemented within existing incident

investigation programs. Moreover, it explained how investigations can be enhanced

through a more detailed analysis in the identification of root causes and how high-quality

recommendations can be achieved.

7.2. Future work

Based on the limitations and the scope of this research, the performed study can

be complemented by the analysis in the following areas:

The DIKW (Data, Information, Knowledge, Wisdom) theory has been used in this

study in order to leverage the safety knowledge within an organization. The

proposed learning process provides a clear understanding of how to transcend data

and information into knowledge. However, the concept of wisdom has not yet been

clearly defined in this research. Thus, the analysis of why organizations are not

getting wisdom, which elements are necessary to achieve it? and analyzing the

relationship between learning and wisdom in organizations can be a new area of

study in this field.

The concept of individual learning has been discussed in this research in a high

level. Therefore, the necessary elements to achieve individual learning, best

approaches to learn, and the relationship between learning and individual

149
characteristics such as age, motivation, and perception can be further analyzed to

understand how these factors can influence the level of retention in individuals.

The concept of sharing knowledge within the organization has been discussed and

how learning systems can be implemented in organizations. However, tools and

methods for delivering this knowledge have not been covered in this research.

Evaluating the benefits and effectiveness of different types of tools would help

organizations to enhance learning at both individual and organizational level.

A case study was developed to explain how the proposed incident investigation

process can be implemented in organizations. Likewise, a theoretical insight was

presented of how the case study fits into the complete learning process. The

proposed framework of improving learning from incidents should be tested through

the development of a pilot in a real organization for a particular group of facilities.

This would give the opportunity to validate the complete process and refine it.

150
 REFERENCES

1. Bureau of Labor Statistics, Nonfatal and fatal occupational injuries and illnesses
by industry. United States Department of Labor, 2016, September 7 [Data file].
Retrieved from: http://www.bls.gov/iif/oshsum1.htm [Accessed 20 Noviembre
2016].

2. Pietersen, C., Analysis of the LPG-disaster in Mexico City. Journal of Hazardous

Materials, 1988. 20: pp. 85-107.

3. Johnson, K., State and community during the aftermath of Mexico City's
November 19, 1984 Gas Explosion. Florida Mental Health Institute (FMHI),
1985. 58: pp. 1-44.

4. El Horizonte, Pemex y su negro historial en 2015. El Horizonte, 2016. Retrieved

from http://elhorizonte.mx/mexico/estados/598171/pemex-y-su-negro-historial-
en-2015 [Accessed 2 March 2016].

5. Gonzalez, S., Pemex reporta promedio de 153 accidentes con 21 muertes por
año. La Jornada, 2015. Retrieved from
http://www.jornada.unam.mx/ultimas/2015/04/01/pemex-reporta-promedio-de-
153-accidentes-con-21-muertes-por-ano-7763.html [Accessed 2 March 2016].

6. CSB, West fertilizer final investigation report in west fertilizer explosion and
Fire. U.S. Chemical Safety and Hazard Investigation Board, 2016. Retrieved
from http://www.csb.gov/west-fertilizer-explosion-and-fire-/ [Accessed 5 March
2016].

7. BBC, China explosions: what we know about what happened in Tianjin. BBC
News, 2015, August 17. Retrieved from http://www.bbc.com/news/world-asia-
china-33844084 [Accessed 5 March 2016].

8. Cooke, D.L. and T.R. Rohleder, Learning from incidents: from normal accidents
to high reliability. System Dynamics Review, 2006. 22(3): pp. 213-239.

9. Argyris, C. and D.A. Schön, Organizational learning: a theory of action

perspective. 1978. Addison-Wesley, Reading, MA.

151
10. Argyris, C. and D.A. Schön, Organizational learning II: theory, method, and
practice. 1996. Addison-Wesley, Reading, MA.

11. Drupsteen, L. and F.W. Guldenmund, What is learning? A review of the safety
literature to define learning from incidents, accidents and disasters. Journal of
Contingencies and Crisis Management, 2014. 22(2): pp. 81-96.

12. Lukic, D., A. Margaryan, and A. Littlejohn, How organisations learn from safety
incidents: a multifaceted problem. Journal of Workplace Learning, 2010. 22(7):
pp. 428-450.

13. Choularton, R., Complex learning: organizational learning from disasters.

Safety Science, 2001. 39(1): pp. 61-70.

14. Kletz, T.A., Lessons from disaster: How organizations have no memory and
accidents recur. 1993. IChemE, UK.

15. Kletz, T.A., Learning from experience. Journal of Hazardous Materials, 2004.
115(1): pp. 1-8.

16. Keegan, A. and J.R. Turner, Quantity versus quality in project-based learning
practices. Management Learning, 2001. 32(1): pp. 77-98.

17. Duffield, S. and S.J. Whitty, Developing a systemic lessons learned knowledge
model for organisational learning through projects. International Journal of
Project Management, 2015. 33(2): pp. 311-324.

18. Gordon, H.J. Integrating learning into safety. In ASSE Professional

Development Conference and Exposition, American Society of Safety Engineers,
2006, June. Seattle, Washington.

19. Lindberg, A.-K., S.O. Hansson, and C. Rollenhagen, Learning from accidents–
what more do we need to know?. Safety Science, 2010. 48(6): pp. 714-721.

20. Fahlbruch, B. and M. Schöbel, SOL–Safety through organizational learning: a

method for event analysis. Safety Science, 2011. 49(1): pp. 27-31.

21. Lukic, D., A. Littlejohn, and A. Margaryan, A framework for learning from
incidents in the workplace. Safety Science, 2012. 50(4): pp. 950-957.

152
22. Le Coze, J.C., What have we learned about learning from accidents? Post-
disasters reflections. Safety Science, 2013. 51(1): pp. 441-453.

23. Avnet, M.S., A network-based approach to organizational culture and learning

in system safety. Procedia Computer Science, 2015. 44: pp. 588-598.

24. Pidgeon, N. and M. O'Leary, Man-made disasters: why technology and

organizations (sometimes) fail. Safety Science, 2000. 34(1): pp. 15-30.

25. Carroll, J.S. and B. Fahlbruch, “The gift of failure: new approaches to analyzing
and learning from events and near-misses.” honoring the contributions of
Bernhard Wilpert. Safety Science, 2011. 49(1): pp. 1-4.

26. Huber, S., et al., Learning from organizational incidents: resilience engineering
for high risk process environments. Process Safety Progress, 2009. 28(1): pp. 90-
95.

27. Sanne, J.M., Learning from adverse events in the nuclear power industry:
organizational learning, policy making and normalization. Technology in
Society, 2012. 34(3): pp. 239-250.

28. Jacobsson, A., Å. Ek, and R. Akselsson, Method for evaluating learning from
incidents using the idea of “level of learning”. Journal of Loss Prevention in the
Process Industries, 2011. 24(4): pp. 333-343.

29. Drupsteen, L., J. Groeneweg, and G.I. Zwetsloot, Critical steps in learning from
incidents: using learning potential in the process from reporting an incident to
accident prevention. International Journal of Occupational Safety and
Ergonomics, 2013. 19(1): pp. 63-77.

30. Hovden, J., F. Størseth, and R.K. Tinmannsvik, Multilevel learning from
accidents–case studies in transport. Safety Science, 2011. 49(1): pp. 98-105.

31. Naot, Y.B.-H., R. Lipshitz, and M. Popper, Discerning the quality of

organizational learning. Management Learning, 2004. 35(4): pp. 451-472.

32. Easterby-Smith, M., L. Araujo, and J. Burgoyne, Organizational learning and

the learning organization: Developments in theory and practice. 1999. Sage
Publications, London; Thousand Oaks, California; New Delhi.

153
33. Easterby-Smith, M. and M.A. Lyles, Handbook of organizational learning and
knowledge management. 2011. John Wiley & Sons Ltd Publications, United
Kingdom.

34. Fiol, C.M. and M.A. Lyles, Organizational learning. The Academy of
Management Review, 1985. 10(4): ppp. 803-813.

35. Huber, G.P., Organizational learning: The contributing processes and the
literatures. Organization Science, 1991. 2(1): pp. 88-115.

36. Baum, J., The Blackwell companion to organizations. 2002. Blackwell, Malden,
MA.

37. Turner, M.E., Groups at work: Theory and research. 2001. Lawrence Erlbaum
Associate, Mahwah, NJ.

38. Argyris, C. and D.A. Schon, Theory in practice: increasing professional

effectiveness. 1974. Jossey-Bass, Oxford, England.

39. Argyris, C., Single-Loop and Double-Loop Models in Research on Decision

Making. Administrative Science Quarterly, 1976. 21(3): pp. 363-375

40. McNamara, C., Field guide to consulting and organizational development with
nonprofits: A collaborative and systems approach to performance, change and
learning. 2005. Authenticity Consulting, Minneapolis, MN.

41. Argote, L., Organizational learning: Creating, retaining and transferring

knowledge. 2012. Springer Science & Business Media, New York City, NY.

42. Grant, R.M., The development of knowledge management in the oil and gas
industry/El desarrollo de la dirección del onocimiento en la industria del
petroleo y gas. Universia Business Review, 2013. (40): p. 92.

43. Rowley, J., The wisdom hierarchy: representations of the DIKW hierarchy.
Journal of Information Science, 2007. 33(2): pp. 163-180.

44. Wellman, J., Organizational learning: how companies and institutions manage
and apply knowledge. 2009. Palgrave Macmillan, New York, NY.

154
45. Frost, A. The different types of knowledge. Knowledge Management Site, 2017.
Retrieved from http://www.knowledge-management-tools.net/different-types-of-
knowledge.html [Accessed 20 February 2017].

46. Sanchez, R., Tacit knowledge versus explicit knowledge. Department of

Industrial Economics and Strategy, Copenhagen Business School, 2004,
Retrieved from http://openarchive.cbs.dk/cbsweb/handle/10398/7224 [Accessed
14 November 2016].

47. Brown, J.S. and P. Duguid, Organizing knowledge. California Management

Review, 1998. 40(3): pp. 90-111.

48. Szulanski, G., The process of knowledge transfer: A diachronic analysis of

stickiness. Organizational Behavior and Human Decision Processes, 2000. 82(1):
pp. 9-27.

49. Montazemi, A.R. and A. Esfahanipour, Application of cognitive map in

knowledge management. Encyclopedia of Information Science and Technology,
IGI Global, 2009. (2): pp. 169-176.

50. Argote, L. and P. Ingram, Knowledge transfer: A basis for competitive advantage
in firms. Organizational Behavior and Human Decision Processes, 2000. 82(1):
pp. 150-169.

51. Malhotra, Y., Knowledge Management and Business Model Innovation. 2001.
IGI Global, Hershey, PA.

52. Dixon, N.M., Common knowledge: How companies thrive by sharing what they
know. 2000. Harvard Business School Press, Boston, MA.

53. Stein, E.W. and V. Zwass, Actualizing organizational memory with information
systems. Information Systems Research, 1995. 6(2): pp. 85-117.

54. Walsh, J.P. and G.R. Ungson, Organizational memory. Academy of Management
Review, 1991. 16(1): pp. 57-91.

55. Fernandez, V. and A. Sune, Organizational forgetting and its causes: an

empirical research. Journal of Organizational Change Management, 2009. 22(6):
pp. 620-634.

155
56. Mohun, A. and S.D. Sagan, The Limits of Safety: organizations, accidents, and
nuclear weapons. 1995. American Association for the Advancement of Science,
Washington, DC.

57. Rasmussen, J., Risk management in a dynamic society: a modelling problem.

Safety Science, 1997. 27(2): pp. 183-213.

58. Lalley, J.P. and R.H. Miller, The learning pyramid: Does it point teachers in the
right direction. Education, 2007. 128(1): p. 64.

59. Davis, B. and M. Summers, Applying Dale's cone of experience to increase

learning and retention: a study of student learning in a foundational leadership
course. Engineering Leaders Conference 2014, Qscience Proceedings. 2014. pp.
1-7.

60. Magennis, S. and A. Farrell, Teaching and learning activities: expanding the
repertoire to support student learning. Emerging Issues in the Practice of
University Learning and Teaching, 2005. 1: pp.45-54.

61. Deubel, P., An investigation of behaviorist and cognitive approaches to

instructional multimedia design. Journal of Educational Multimedia and
Hypermedia, 2003. 12(1): pp. 63-90.

62. Basu, P., Training methods for making a training course more lively. LinkedIn,
2015. Retrieved from https://www.linkedin.com/pulse/training-methods-making-
course-more-lively-pabitra-basu [Accessed 20 January 2016].

63. Noe, R.A., Employee training and development. Seventh Edition, 2002. McGraw
Hill Education, New York, NY.

64. Blanchard, P.N., Training delivery methods. Reference for Business, 2016.
Retrieved from http://www.referenceforbusiness.com/management/Tr-
Z/Training-Delivery-Methods.html#ixzz4E7CQAxeU [Accessed 20 January
2016].

65. Amit, S., Choosing the right eLearning methods:factors and elements. eLearning
Industry, 2015. Retrieved from https://elearningindustry.com/choosing-right-
elearning-methods-factors-elements [Accessed 02 November 2016].

66. UNESCO, Training guide and training techniques. 2004. UNESCO Asia and
Pacific Regional Bureau for Education, Bangkok, Thailand.
156
67. Pichee, D., 6 ways to use science to improve your employee training program.
Training Industry, 2016. Retrieved from
https://www.trainingindustry.com/webinars/6-ways-to-use-science-to-improve-
your-employee-training-program.aspx [Accessed 01 December 2016].

68. Taylor, G.S., G.F. Templeton, and L.T. Baker, Factors influencing the success of
organizational learning implementation: A policy facet perspective. International
Journal of Management Reviews, 2010. 12(4): pp. 353-364.

69. Park, J.-H. and H.J. Choi, Factors influencing adult learners' decision to drop
out or persist in online learning. Educational Technology & Society, 2009.
12(4): pp. 207-217.

70. Bos, C.S. and S. Vaughn, Strategies for teaching students with learning and
behavior problems. 2002. Allyn & Bacon, A Pearson Education Company, 75
Arlington Street, Boston, MA.

71. Reason, J., Achieving a safe culture: theory and practice. Work & Stress, 1998.
12(3): pp. 293-306.

72. CCPS, Guidelines for risk based process safety. 2007. Center for Chemical
Process Safety of the American Institute of Chemical Engineers and John Wiley
& Sons, Inc, Hoboken, New Jersey.

73. Reason, J., Managing the risks of organizational accidents. 1997. Ashgate,
Hampshire, England.

74. Reason, J., Human error. 1990. Cambridge University Press, Cambridge, United
Kingdom.

75. Freibott, B., Incident management as a cornerstone for a successful safety

culture. 2012. Safety and Security Engineering V. WIT Press, United Kingdom.

76. Heinrich, H.W. and E. Granniss, Industrial accident prevention: a scientific

approach. Fourht Edition. 1959. McGraw-Hill, New York, NY.

77. CCPS, Guidelines for investigating chemical process incidents. Second edition.
2003. Center for Chemical Process Safety of the American Institute of Chemical
Engineers, New York, NY.

157
78. OSHA, Incident [Accident] Investigations: A Guide for Employees. United States
Department of Labor Occupational Safety and Health Administration, 2015.
Retrieved from https://www.osha.gov/dte/IncInvGuide4Empl_Dec2015.pdf
[Accessed 10 October 2016].

79. Texas A&M University, Incident Investigation, in Process Safety course,

Chapter 13. 2015. Mary Kay O’Connor Process Safety Center, College Station,
TX.

80. Mannan, S., Lees' Loss prevention in the process industries, hazard
identification, assessment and control. 2005. Butterworth–Heinemann,
Burlington, MA.

81. Sklet, S., Methods for accident investigation. 2002. Norwegian University of
Science and Technology, Trondheim, Norway.

82. Hosseinian, S.S. and Z.J. Torghabeh, Major theories of construction accident
causation models: a literature review. International Journal of Advances in
Engineering & Technology, 2012. 4(2): pp. 53-66.

83. HaSPA (Health and Safety Professionals Alliance), The core body of knowledge
for generalist OHS professionals. 2012. Safety Institute of Australia Ltd,
Tullamarine, Victoria, Australia.

84. Bird, F.E., G.L. Germain., Practical loss control leadership. 1986. Institute
Publishing, Loganville, GA.

85. OSHA, How to conduct incident investigation. Alliance, an OSHA cooperative

program. Incident Investigation, United States Department of Labor
Occupational Safety and Health Administration, 2014. Retrieved from
https://www.osha.gov/dcsp/products/topics/incidentinvestigation/ [Accessed 10
october 2016].

86. Svedung, J.R.I. and J. Rasmussen, Proactive risk management in a dynamic

society. 2000. Karlstad: Swedish Rescue Services Agency, Karlstad, Sweden.

87. Leveson, N., A new accident model for engineering safer systems. Safety
Science, 2004. 42(4): pp. 237-270.

158
88. Hollnagel, E. and O. Goteman, The functional resonance accident model.
Proceedings of Cognitive System Engineering in Process Plant. 2004. pp. 155-
161.

89. Johnson, W.G., Management Oversight and Risk Tree-MORT. 1973. Aerojet
Nuclear Co., Scoville, ID.

90. Pasman, H.J., Risk analysis and control for industrial processes-gas, oil and
chemicals: a system perspective for assessing and avoiding low-probability,
high-consequence events. 2015. Butterworth-Heinemann, Waltham, MA.

91. Energy Institute, Guidance on investigating and analysing human and

organisational factors aspects of incidents and accidents. 2008. Energy Institute,
London, UK.

92. Livingstone, A., G. Jackson, and K. Priestley, Root causes analysis: literature
review. Health & Safety Executive, Contract Research Report 325. 2001. HSE
Book, Norwich, UK.

93. Hendrick, K. and L. Benner, Investigating accidents with STEP. Occupational

Safety and Health Vol 13. 1986. Marcel Dekker, Inc, New York, NY.

94. Hollnagel, E., FRAM, the functional resonance analysis method: modelling
complex socio-technical systems. 2012. Ashgate Publishing, Ltd., Farnham, UK.

95. Energy Institute, Guidance on using Tripod Beta in the investigation and analysis
of incidents, accidents and business losses. 2015. Energy Institute, London, UK.

96. Ferjencik, M., An integrated approach to the analysis of incident causes. Safety
Science, 2011. 49(6): pp. 886-905.

97. Paradies, M. and L. Unger, TapRoot: The system for root cause analysis,
problem investigation, and proactive improvement. 2000: System Improvements,
Knoxville, TN.

98. TapRooT, Using the TapRooT® System for Chemical Industry Incident
Investigation. How Does the TapRooT® System Work?. TapRooT® Website,
2015. Retrieved from http://www.taproot.com [Accessed 10 October 2016].

159
99. Knox, N.W. and R.W. Eicher, MORT user's manual for use with the management
oversignt and risk tree analytical logic diagram. 1983. EG and G Idaho, Inc.,
Idaho Falls, ID.

100. Leveson, N. G., Daouk, M., Dulac, N., & Marais, K., Applying STAMP in
accident analysis. 2003. Massachusetts Institute of Technology , Cambridge,
MA.

101. Oxford, Learning definition. Oxford Dictionary Website, 2017. Retrieved from
https://en.oxforddictionaries.com/definition/learning [Accesed 12 October 2016].

102. Baskarada, S. and A. Koronios, Data, information, knowledge, wisdom (DIKW):

a semiotic theoretical and empirical exploration of the hierarchy and its quality
dimension. Australasian Journal of Information Systems, 2013. 18(1): pp. 5-24

103. Rowley, J.E., The wisdom hierarchy: representations of the DIKW hierarchy.
Journal of Information Science, 2007. 33(2): pp. 163-180.

104. Davenport, T.H. and L. Prusak, Working knowledge: how organizations manage
what they know. 1998. Harvard Business Press, Boston, MA.

105. Grant, K.A. and C.T. Grant, Developing a model of next generation knowledge
management. Issues in Informing Science and Information Technology, 2008.
5(2): pp. 571-590.

106. Lambe, P., The unacknowledged parentage of knowledge management. Journal

of Knowledge Management, 2011. 15(2): pp. 175-197.

107. TPUB, Instructional Methods and Techniques. Military requirements for Petty
Officers Third and Second Class. 2013. Retrieved from
http://navyadvancement.tpub.com/14504/css/Instructional-Methods-And-
Techniques-25.htm [Accesed 18 November 2016].

108. Wood, E., Problem-based learning: exploiting knowledge of how people learn to
promote effective learning. Bioscience Education, 2004. 3(1): pp. 1-12.

109. Noe, R.A., Employee Training and Development. Fifth Edition, 2010. McGraw-
Hill, New York, NY.

160
110. Pighin, M. and A. Marzona, Unlearning/relearning in processes of business
information systems innovation. Journal of Information and Organizational
Sciences, 2011. 35(1): pp. 59-72.

111. Annett, J., Hierarchical task analysis. Handbook of Cognitive Task Design,
2003. 2: pp. 17-35.

112. Stanton, N.A., Hierarchical task analysis: developments, applications, and

extensions. Applied Ergonomics, 2006. 37(1): pp. 55-79.

113. Salmon, P., Jenkins, D., Stanton, N., & Walker, G., Hierarchical task analysis vs.
cognitive work analysis: comparison of theory, methodology and contribution to
system design. Theoretical Issues in Ergonomics Science, 2010. 11(6): pp. 504-
531.

114. Bailey, A., BP: Learning from oil spill lessons. Petroleum News, 2006. 11(20).
Retrieved from http://www.petroleumnews.com/pntruncate/573947058.shtml
[Accessed 15 November 2016].

115. National Transportation Safety Board. Natural gas pipeline rupture and fire near
Carlsbad, New Mexico, august 19, 2000. Pipeline Accident Report NTSB/PAR-
03/01, 2003. National Transportation Safety Board, Washington, D.C.

116. Garber, J.D., A. Alvarado, and R.H. Winters, Study tracks internal-corrosion
trends in aging gulf pipelines. Oil and Gas Journal, 2000. 98(13): pp. 68-73.

117. Baker, M. and Fessler, R.R., Pipeline corrosion, Final report. 2008. US
Department of Transportation Pipeline and Hazardous Materials Safety
Administration, Washington, D.C.

161