CURRENT RESEARCH AND INDUSTRIAL APPLICATIONS OF INTEGRATED SRA

CURRENT RESEARCH AND INDUSTRIAL APPLICATIONS OF INTEGRATED SRA AND QRA MODELS Philip Smedley

ASA SRA QRA HFA

http: //mar. ist. utl. pt/saferelnet Thematic Network on Safety and Reliability of Industrial Products, Systems and Structures OBJECTIVE To provide: consistent, safe & cost-effective solutions for a range of industrial systems across different industrial sectors throughout the system’s life-cycle.

PROGRAMME Steering Committee Liaison Committee

PROGRAMME WP SCOPE LEADER 1 Management, Dissemination & Exploitation IST 2 Risk Assessment Methodology EQE 3 Human & Org. Factors in Risk Assessments DAP 4 Integration of Risk & Reliability Formulations ETHZ 5 Reliability Based Design RCP 6 Assessment of Existing Structures & Life Extension BUW 7 Risk & Cost Based Inspection & Maintenance Planning 8 Standardisation and Codes PAFA 9 Training and Education NTNU 10 Strategy in the Various Industrial Sectors Atkins IST

UK PARTNERS EQE International PAFA Consulting Engineers Atkins BOMEL Limited Petrellus Limited Corr. Ocean Ltd Liverpool John Moores University of Liverpool The University of Surrey Network Rail Highways Agency Health and Safety Executive

INTEGRATION ASA SRA QRA HFA

ADVANCED STRUCTURAL ANALYSIS STRENGTHS • Solutions to complex / time-dependent problems • Speed – cost-effective solutions • System’s redundancy and reserve strength • Uncertainty analysis – parametric variations WEAKNESSES • Difficult to estimate accuracy in results • Potential errors or inadequacies in programs • Potentially inadequate user skill levels

STRUCTURAL RELIABILITY ANALYSIS STRENGTHS • ‘Complete’ representation of loading and resistance uncertainties in design problems • Fully quantified reliability estimates • Updated estimates as new data added or improved by expert opinion (Bayesian updating) WEAKNESSES • Better for empiric rather than parametric formulae • If human factors are included they are generally fairly crude or simplistic estimates.

QUANTIFIED RISK ASSESSMENT STRENGTHS • Causes and consequences of hazard modelled • Strong for operational and accident problems • Quantification of underlying issues - based on incident data and expert opinion (frequentist) WEAKNESSES • Lack of data or understanding of problem or inaccurate data due to biased opinions • Uncertainty only considered in the underlying statistics rather than the model • Not good for time-dependent problems

HUMAN FACTOR ASSESSMENT STRENGTHS • Most (80%) incidents caused by human error therefore essential element in our understanding • Human behaviour often very predictable • Includes individual and corporate behaviour WEAKNESSES • Cynicism - knowledge of HFs generally from specialists outside the engineering industry • High uncertainties in models and data (for now) • Difficult issues of cultural/society differences

SRA-QRA-HFA INTEGRATION IS IT FEASIBLE? A Qualified - Yes. A number of common issues: • Mathematical models are of a similar format • All seek to achieve a target level of safety (Annual target reliability or risk acceptance criteria) • Need quality, unbiased data (historic or opinion)

SRA-QRA-HFA INTEGRATION INITIAL INTEGRATED MODELS 1. Reliability distribution replaces deterministic quantification in risk analysis - fault tree 2. Human factor Bayesian Probabilistic Networks can readily be reformulated into fault trees

INTEGRATION – Example 1 INST. FOR ELECTRIC POWER RES. (HUNGARY) 1. Process Analysis – Deterministic Assessment 1. Initiating event identification 2. Event tree development 2. System Analysis – Reliability Assessment 1. Fault tree development 2. Hardware failure data estimation 3. Human failure data estimation 3. Structural Analysis – Fragility Assessment

INTEGRATION – Example 1

INTEGRATION – Example 2 SWALE CROSSING : Kent – Isle of Sheppey

INTEGRATION – Example 2 PAFA CONSULTING ENGINEERS 1. Risk Analysis – AASHOTO Guidelines 1. Number of Ships subdivided into 6 classes 2. Probability of aberrance (human error, mechanical failure, severe environmental loading) 3. Probability of collision with bridge pier 4. Probability of exceeding bridge pier strength 2. To Probability of Aberrance add: 1. Mechanical reliability of bridge lift mechanism 2. Avoidance of other vessels in area (esp. yachts)

PROBLEM: ACCEPTANCE CRITERIA from Faber/Schneider Objective Hazard Potential Correct model Accepted Risk Accurate Risk Assessment Not adequate Adequately quantified (good data) Wrong Accepted Risks modelled Neglected Taken into account Not Realised Subjectively realised Inaccuracies due to Human Errors Not known Objectively known

SRA-QRA-HFA INTEGRATION IS IT DESIRABLE? Sometimes • Expanding a reliability model, for example, to account for poorly defined human factors will add time and cost but not improve the overall understanding of the system. • The three approaches have been developed to solve specific problems. Each approach has many models each with specific strengths and weaknesses. One integrated approach is likely to be less rigorous in some instances.

SRA-QRA-HFA INTEGRATION SAFERELNET APPROACH • Seeking to develop a consistent mathematical model that may be used to integrate some of the strengths of SRA – QRA – HRA. • If such an integrated approach can be developed, to consider the strengths and weaknesses within such a model. • Discuss and develop thinking for a consistent risk and reliability acceptance criteria. http: //mar. ist. utl. pt/saferelnet