Departments
of Engineering and Psychology
University of Aberdeen
Modeling & Managing Human Factors in Industrial Safety Using Virtual Reality Techniques
Work-package 7
Report D 7.2
Michael Baker, Kathryn Mearns & Emma Noble
Departments of Psychology & Engineering
University of Aberdeen
Executive Summary
The overall objective of this study is to carry out a series of investigative studies that will lead to a new methodological approach to identify and control for potential failures associated with human factors using state-of-the-art virtual reality as an integrated tool. The main aim of this work-package is to select a methodology for identifying and controlling potential failures associated with human factors. The methodology was selected in relation to the process industry. BP Exploration, Oil, and Gas at the Grangemouth site provided access to their extensive database relating to accident and incidents. By accessing the ‘Total Loss Control’ (TLC) database currently operating at the site, a number of accidents/incidents were selected and reviewed according to the selected methodology. Analysis of the selected accidents and incidents reviewed the physical hazards present and the human factor issues contributing to the accidents/incidents. Subsequent analysis considered the feasibility of simulating these accidents/incidents using immersive virtual reality.
Executive Summary 2
1. Introduction 5
1.1 Objectives of Work package 7 6
2. BP Grangemouth 1
3. What is an accident or incident? 1
4. Criterion for selection 1
4.1 Human Error 1
4.2 Physical Hazards 2
4.3 Person Injury 4
4.4 Location 4
4.5 Accident/Incident Type 4
4.6 Severity 4
4.7 Person Involvement 5
4.8 Injury sustained whilst working alone 5
4.9 Injury sustained whilst working in groups (two persons or more) 5
4.10 Incidents without loss 5
4.11 Accidents/incidents occurring during maintenance activities. 5
4.12 Accident/Incidents involving procedural tasks 6
5. Feasibility of accident/incident simulation 1
Table 1 – Physical Hazards, Human Factors Causes and Consequences of Fires 3
6 Conclusions 1
References 1
APPENDIX 1 2
‘SLIPS, TRIPS & FALLS’ INCIDENTS 2
‘Slips, Trips & Falls’ Incident 1 2
Plater fall incident 2
Summary 2
Physical Factors 2
Human Factors 2
Human Factors Investigation Tool (HFIT) 3
Capability to model 3
CRUSHING INCIDENTS 1
Crushing Incident 1 1
Inshore Marine Accident 1
Time 09:30 hrs 1
Summary 1
Physical Factors 1
Human Factors 1
HFIT 1
Capability to model 2
Crushing Incident 2 2
Serious Hand Injury 2
Summary 2
Physical Hazards 3
Human Factors 3
HFIT 3
Capability to Model 3
Crushing Incident 3 4
Injury to workshop employee 4
Time 14:40 4
Physical Hazards 4
Human Factors 4
HFIT: 4
Capability to Model 5
FIRE INCIDENTS 1
Fire Incident 1 1
Lab Technician Splashed with Concentrated Acid 1
Time: 11:45am 1
Summary 1
Physical Hazards 1
Human Factors 1
HFIT 2
Capability to model 2
Fire Incident 2 3
Fire during recatalysation 3
Summary 3
Physical Hazards: 3
Human Factors: 4
HFIT 4
Capability to Model 5
Fire Incident 3 5
Investigation into fire on Catalyst Filter 5
Summary 5
Physical Hazards 6
Human Factors 6
HFIT 6
Capability to Model 6
CRUSHING INCIDENTS 1
Crushing Incident 1 1
Inshore Marine Accident 1
Time 09:30 hrs 1
Summary 1
Physical Factors 1
Human Factors 1
HFIT 1
Capability to model 2
Crushing Incident 2 2
Serious Hand Injury 2
Summary 2
Physical Hazards 3
Human Factors 3
HFIT 3
Capability to Model 3
Crushing Incident 3 4
Injury to workshop employee 4
Time 14:40 4
Physical Hazards 4
Human Factors 4
HFIT: 4
Capability to Model 5
FIRE INCIDENTS 1
Fire Incident 1 1
Lab Technician Splashed with Concentrated Acid 1
Time: 11:45am 1
Summary 1
Physical Hazards 1
Human Factors 1
HFIT 2
Capability to model 2
Fire Incident 2 3
Fire during recatalysation 3
Summary 3
Physical Hazards: 3
Human Factors: 4
HFIT 4
Capability to Model 5
Fire Incident 3 5
Investigation into fire on Catalyst Filter 5
Summary 5
Physical Hazards 5
Human Factors 6
HFIT 6
Capability to Model 6
6
1. Introduction
The proposed project aims to define a new methodological approach to identify and control for potential failures associated with human factors. Using virtual reality as an integrated tool, the participating chemical process industries would be able to analyze the human factor aspects of accident causation, as well as model physical hazards and the consequences arising from them. The chemical processing industries were selected on the basis of the high potential hazards and risks that they impose. These industries are considered as ‘high-risk’ due to the social and environmental implications that they impose potentially transferring risks and hazards to society through raw products. Therefore, it is widely recognized that there is a need for improvement in safe working practices and accident prevention measures within the industry.
The proposed project has been divided into a number of work packages in order to achieve the desired objective. The first work package reviews the existing methodologies for identifying and controlling failure associated with human factors. The second work package, which this report details, consists of defining criteria for selecting accidents/incidents and consequent identification of a selected number accident/incident scenarios for detailed study. Subsequent work packages involve the analysis of contributing factors associated with the selected accident/incident scenarios, evaluation of current methodologies, reviewing of the state-of-the-art virtual reality modeling, investigating the potential contribution of VR to improve the selected method, and preliminary evaluation of the revised methodology.
1.1 Objectives of Work package 7
The primary objective of this work package is to define criteria for selecting accidents and incidents, consequently identifying a number of scenarios to be simulated using state-of-the-art virtual reality. The aim is to produce a reasonable amount of data, selected from accidents/incidents to enable the evaluation process to be carried out.
The participating company specific to this work package is BP Grangemouth (UK). This site was selected on the basis of its unique location and infrastructure, integrating three business units on one site, exploration, oil and gas, and chemical processes. Further details of the site and its accident/incident history and databases are given below.
6
2. BP Grangemouth
BP Exploration, Oil, and Chemical processes make up 27 operations and 8 functions in Grangemouth, employing over 2000 people to ensure continuous production 24 hours a day. Central to the success of BP Grangemouth is the Forties Pipeline System (FPS). The FPS provides Grangemouth with direct access to oil and gas from the North Sea, carrying around one million barrels of oil a day, which is stabilized by removal of the gas content. The Refinery processes a proportion of the crude oil, producing a range of fuels and other products. The gas separated from the oil and refinery streams is subsequently used at the petrochemical plants where ethylene and other derivative products such as ethanol and polyethylene are produced. It is therefore essential that the products meet the qualitative and environmental requirements of the customer.
BP Grangemouth has shown a variable performance in health, safety and the environment, having experienced a number of accidents and incidents in recent years. The early summer 2000 saw three major incidents at the site, a power failure on 29th May, a medium pressure steam line rupture on 7th June and a fire on the Catalytic Cracker. No personal injuries occurred during these incidents. However, they were all potentially major incidents. Following these incidents a comprehensive safety review was carried out across the site whereby a total of 850 recommendations were made on improving the health, safety and environment of the workplace. Most of these recommendations are to be completed by the end of the year 2001. The site therefore offers an extensive and valuable amount of data with regard to accidents and incidents.
Currently operating at Grangemouth is an accident and incident database termed the TLC system. The TLC (Total Loss Control) system stores all reported accidents and near miss incidents on the site. The database holds information regarding the events and conditions leading up to the accident/incident, and the attributed causal factors. Entry of accidents and incidents into the TLC database is the responsibility if the line manager, supervisor or person responsible for safe working practices within the area at the time of the accident/incident. Entry of accidents/incidents requires a number of mandatory questions to be answered in order to determine if further investigation is required. The severity of the accident/incident is determined using a Boston Square, which has definitions for Minor, Serious and Major injuries, environmental or financial loss. During investigation of reported accidents and incidents all critical factors, immediate causes and root cause should be identified. Immediate causes are the conditions or actions occurring immediately prior to the event. Root causes are the underlying preconditions and can be categorised into personal factors and job factors. Full investigation reports should identify all immediate and root causes and further detail a list of recommendations for corrective action.
The aims of this work package were to review accident and incident reports recorded on the TLC database at the BP Grangemouth site. By identifying common human factors and physical hazards that give rise to accidents/incidents criteria can be made for selecting accidents to simulate using immersive virtual reality. The TLC database was accessed over the duration of a three-day visit to the site and a number of more detailed accidents and incident reports were selected according to a variety of criteria. The feasibility of modeling the selected accidents and incidents was subsequently reviewed.
In an effort to select a number of accidents and incidents for virtual reality simulation, five years of retrospective accident/incident data from the BP Grangemouth database was studied. In attempting to establish an understanding of the underlying human errors that result in accidents and incidents, it was decided to study the most frequently occurring accident and incident types at the site.
During the study of the database in order to gain an overall feel for types of accidents and incidents occurring on site, it became apparent that there were three distinct accident types. These were slips/trips and falls, crushing incidents, and fires. Furthermore, it was evident that these accident types were occurring most frequently during periods of ‘turnaround’ whereby maintenance activities and numbers in the workforce increase. Thus, these three accident/incident types were considered as a starting point for the selection of accidents and incidents. Using virtual reality and its limitations as the driving force for selection, the feasibility of modeling accidents/incidents was then determined.
After considering possible accidents and incidents according to the above criteria selection was determined by the limitations of the virtual reality system. Whilst the most frequent accidents and incidents in the petro-chemical process industry are caused by ‘slips, trips and falls’, and the most serious and potentially major accidents result in crushing incidents, the most feasible scenario within the limits of the virtual reality system is fire.
6
3. What is an accident or incident?
‘Accidents’ can be defined as being unplanned and undesirable deviations in systems operations. The consequences of accidents can take the form of personal injury, loss of life, property damage, environmental damage, or the loss of the system itself. The term ‘Incident’ tends to adopt the definitive meaning of a near-miss accident or an event without injury (Kontogiannis etal, 2000). Near-miss incidents are significant to the process of organisational learning because they result from the same pre-conditions that underlie many accidents, typically caused by the combination of latent failures and erroneous human interventions. Analysing near-miss incidents therefore enables recovery processes to be assessed. Subsequently any weaknesses in the system can be identified and eliminated, preventing future accidents of its form occurring (Reason, 1997). It was therefore important that as an initial criterion for selecting appropriate scenarios for simulation that both accident and incident reports were reviewed.
There are a variety of factors that significantly impact on safety performance. These factors can directly cause accidents and incidents or be contributory factors in the causation of accidents. Direct and indirect influences on safety performance include personal characteristics of the employee performing the task, aspects of the task itself, and the equipment used directly or indirectly in the task (Wickens, Gordon & Liu, 1998). When considering factors for the simulation of accident and incident scenarios both human components and physical properties of the environment and task have to be considered. Human factor representations and physical system representations of accident causation should be complementary to one another. This project aims to achieve this and in doing so will gain a holistic view of accident causation. Physical system representations alone neglect cognitively derived human behaviour, thus failing to accurately represent the real world. Human perceptions and cognitive capacities enable the extraction and interpretation of information from the environment, determining actions and behavioural responses in given situations (Groner, 2001). Fundamentally, human behaviour is governed by an information processing system, by which information is extracted, given meaning, used to consider responses and applied behaviourally. Thus, by assessing human behaviour within given virtual reality situations, an understanding of the cognitive capabilities and perceptions will be achieved. Subsequently, an insight will be gained as to how individuals use information to make decisions and of the physical properties of the working environment that lead to human error.
6
4. Criterion for selection
4.1 Human Error
Human factor assessment in accident analysis is becoming more widely recognised within process industries. ‘Human error’ is a term often found in root cause statements, implying that underlying the causes of all accidents and incidents human actions have played a contributory role. Human error is characterised by a combination of overload, decision to ‘err’, and poor work designs/conditions (DeJoy, 1990) and incorporates individuals, group and organisational failings. It is needless to say that the safety behaviour, attitudes and beliefs of individual(s) significantly influence the safety performance of organisations and collectively define the safety culture of an organisation at a given time. Often defined as ‘the way we do things around here’, the prevailing safety culture has serious implications on how health, safety and environment issues are managed within the organisation.