Dr. Linda Bellamy, White Queen Safety Strategies The Netherlands

1. Development Of A Functional Model Of How Human Factors, Safety Management Systems And Wider Organisational Issues Fit Together Dr. Linda Bellamy, White Queen Safety StrategiesThe Netherlands Tim Geyer, ERM RiskLondon, UK John Wilkinson, Health and Safety ExecutiveBootle, UK

2. Human factors Human regulates and inspects the chemical industry Factors provides guidance A Safety Management System Safety Management System assesses safety reports Safety Case Risk Control Control Project overview Organisational factors HSE recognises are important in

3. Risk Control Human Factors SafetyManagement Organisation System But how do they fit together for Major Accident Prevention (MAP)? UNDER CONTROL LOSS OF CONTROL

4. Concept: Indicating the degree of hazard associated with a chemical • National Fire Protection Agency’s hazard diamond • Simple • Informative • Underlying taxonomy

5. MAP WARNING Triangle “Archetype” • Same idea as Hazard Diamond but using Major Accident Prevention (MAP) • All elements interface the same key risk control system elements • Want model that makes more explicit hidden HF aspects in long standing inspection approaches • ..together with organisation and SMS Safety Management System (SMS) Organisation Key Risk Control System (MAP) Human Factors

6. MAP elements • There could be many MAP elements to choose from and fit together (like Lego bricks) • Need to make a collection of these potential MAP bricks • And work out meaningful combinations Measuring performance Planning & implementing SMS Risk perception Competence / expertise Human Factors Leadership Complexity Organisation Design Modifications Key RiskControl System

7. Lego Analogy • Many bricks • Limited meaningful combinations

8. Meaningful combo sub elements = Theme What makes a vehicle What makes (Loss of) Control (LOC themes)

9. Steps • 1. Defined our “bricks”, a taxonomy of Major Accident Prevention (MAP) elements, from the literature, research etc. • 2. Found meaningful combinations (themes) by seeing which ones (re-)occurred together in major accidents

10. Taxonomy • Our taxonomy comprised 850 factors

11. e.g. Human Factors • Task demands (physical, mental) • Human capacities (physical, physiological, mental) • Behavioural outcomes (symptoms of mismatch, speed, accuracy..)

12. Accidents analysed Alloid Colloids (UK 1992) • Fire following misclassification of chemicals Associated Octel (UK 1992) • Fire due to poor awareness of risks in complex poorly maintained plant Cindu (NL 1992) • Explosion due to runaway reaction in a batch processing plant. Flixborough (UK 1974) • Explosion due to release from a temporary bypass assembly Grangemouth (UK 1987) • Fire due to passing valve (poor design) and inadequate isolation procedures Hickson & Welch (UK 1992) • Jet fire following runaway reaction during non routine vessel cleaning Longford (AUS 1998) • Explosion and fire due to incorrect valve operation. Texaco (UK 1994) • Explosion and fire due to failure to recover deviations

13. Model development – Accident contributors • Half the taxonomy factors (around 400) occurred at least once across the 8 accidents • A limited set appeared in all 8 accidents

14. RESULTS – Main Accident contributors (1) • Risk control failures in:… • Operations • ..Detection of departures from normal • ..Inspection, test and maintenance of plant, equipment, facilities • Design • ..Safety and environmental implications are identified • ..Necessary S & E control measures defined, documented, implemented • Modifications • As design (but less cases)

15. RESULTS – Main Accident contributors(2) • Safety Management System failures in: • ..Planning & Implementing: *Identifying hazards and assessing the risks • ..Measuring: *Ongoing assessment of compliance • ..Organising: Providing competences (specifically for MAP tasks) • ..Organising: Communicating information important to the control of MA hazards • ..Audit and review

16. RESULTS – Main Accident contributors(3) • Organisation failures • Organisational learning, memory and knowledge failures dominate • Personnel in job positions with key authorities and roles in MAP - Job allocation /provision of training failures • Norms and pressures • Chemical/process complexity • Actions on a single substance (drying, cooling, heating, distillation and separation) • Control complexity • Control Interactions

17. Model development – accident contributors (4) • Human Factors failures • Demands : • Workload** • man-machine interface • detecting deviations and process disturbances • Identification and recognition (what am I dealing with here, what does this situation require from me) • Capacities : • failures in understanding, interpretation & situational awareness** • learning & knowledge failures** • incorrect decisions or choices • selective attention, signal / pattern recognition failure • competence / expertise / skills

18. Symptoms of demand-capacity mismatch • Individual symptoms of strain • Reduced situational awareness • Self reported feelings of strain • Conflicts between safety and business processes • Unsafe group (collective) practices • Human error: deviations and omissions in doing task • * Mistakes [8 accidents] • Action physically impossible • Rule violations

19. Themes • “Warning" triangle –warning themes concerning hazardous combinations of the main components found in accidents • The warning triangle sits within a larger triangle (light blue) that shows the ‘theme’ under which the elements are linked • The theme is the name chosen to best describe what fails • The theme constrains the possible taxonomy elements that can appear in each of the parts of the warning triangle

20. Understanding Major Accident Prevention

21. Competence for tasks

22. Priorities, attention & conflict resolution

23. Assurance

24. PyraMAPS • The four primary chemical major accident prevention warning triangles can be joined to make a pyramid • The pyramid of chemical Major Accident Prevention (the PyraMAP). • The pyramid is a 3-D representation of the four main triangles identified and is a multidimensional archetype • It also represents the fact that the 4 triangles are linked by issues which cut across the 4 areas

25. PyraMAPs

26. WORKING MODEL • Run issue (s) through the PyraMAP • A kind of HAZOP/scenario generator tool

27. Example: non-routine maintenance for Triangle 1 3) Is selection and training used to deliver understanding of measures and effects of loss of control for MAP where [non-routine maintenance] could play a role? 2) Are there criteria & resources for including [non-routine maintenance] and possible consequences in a RA? 1) What risk control measures have been identified from the risk assessment to prevent a major accident resulting from [non-routine maintenance]? 4) Do people in jobs that could be related to MAP measures understand the risks of [non-routine maintenance] and the possible consequences?

28. The working model applied generally 3) Define constraints: limit the building components to be used in the archetype=themed combinations • Specify inputs: • issues of interest or concern/ • theme Warning Triangle archetype + a sector specific taxonomy for defining constraints 4) Outputs: Issues translated into model terms 2) Resources: users' knowledge (provides building components – can update & share)

29. The working model • Stakeholders have many professionals who are knowledgeable in specialist areas • The archetypes provide a basis for bringing this knowledge together • Flexible for the user • Specialist teams can generate themed instances (like the ones here for MAP) • Stakeholders can then transform issues into archetypes based on this expertise • The warning triangle archetype ensures that Human Factors can be integrated into major hazard activities in a meaningful way