Probabilistic Risk Assessment

Probabilistic Risk Assessment
Name	Probabilistic Risk Assessment
Discipline	Risk analysis
Developed	1960s–1970s

Contents

Probabilistic Risk Assessment.

Probabilistic Risk Assessment is a systematic approach to estimating the likelihood and consequences of adverse events by combining failure modes, event trees, and system models to produce quantified risk metrics, probability distributions, and decision-support outputs. Prominent in high-consequence domains, its development and use intersect with agencies, corporations, and academic centers associated with Oak Ridge National Laboratory, Sandia National Laboratories, Lawrence Livermore National Laboratory, Massachusetts Institute of Technology, and Stanford University. Practitioners draw on methods and standards promulgated by institutions such as International Organization for Standardization, American Society of Mechanical Engineers, National Aeronautics and Space Administration, Nuclear Regulatory Commission, and Department of Energy.

Overview

Probabilistic Risk Assessment integrates fault trees, event trees, and reliability data to estimate frequency distributions for accident scenarios, adopting probabilistic models common to work at Bell Labs, General Electric, DuPont, Westinghouse Electric Company, and Bechtel. Historically influenced by studies for United States Navy reactor safety, Three Mile Island accident reviews, and analyses following the Space Shuttle Challenger disaster, the approach synthesizes component failure rates from sources like MIL-STD-882 programs, IEEE reliability databases, and vendor records from firms such as Rolls-Royce and Boeing. Interdisciplinary teams often include experts affiliated with Harvard University, Princeton University, Yale University, Columbia University, and Cornell University.

PRA methodology commonly proceeds through system definition, initiating events identification, fault tree construction, event tree quantification, and consequence modeling using Monte Carlo simulation tools developed in contexts like Los Alamos National Laboratory projects, RAND Corporation studies, and software from commercial vendors. Quantification integrates data from component testing at centers such as Argonne National Laboratory, field failure databases maintained by Airbus, Siemens, ABB, and diagnostic programs used by Japan Atomic Energy Agency and Electric Reliability Council of Texas. Treatment of human actions often references protocols from Occupational Safety and Health Administration guidance and operator performance models derived in studies at Imperial College London and University of Cambridge.

PRA is applied in nuclear power licensing under frameworks used by the Nuclear Regulatory Commission and licensees like Exelon Corporation and Entergy Corporation, in aerospace certification for Federal Aviation Administration and manufacturers such as Lockheed Martin and Northrop Grumman, in chemical process safety programs for companies like Dow Chemical Company and BASF, and in offshore oil and gas integrity assessments for operators including Royal Dutch Shell, BP, and Chevron Corporation. It supports critical infrastructure resilience planning for agencies such as Federal Emergency Management Agency and utilities like Con Edison and Pacific Gas and Electric Company, and informs cybersecurity risk modeling in studies by National Institute of Standards and Technology, MITRE Corporation, and Kaspersky Lab collaborations.

Treatment of epistemic and aleatory uncertainty in PRA leverages Bayesian updating techniques used in research at Carnegie Mellon University and University of California, Berkeley, and sensitivity analysis methods championed in literature from Cornell University and University of Michigan. Techniques include Latin Hypercube Sampling, importance sampling, and global sensitivity measures similar to work by scholars at Imperial College London and ETH Zurich, with propagation of uncertainty through simulators developed in contexts such as European Organisation for Nuclear Research and National Renewable Energy Laboratory. Regulatory submissions often include uncertainty quantification guided by panels convened by Organisation for Economic Co-operation and Development, International Atomic Energy Agency, and World Health Organization.

Standards and guidance shaping PRA practice derive from documents issued by International Organization for Standardization, American National Standards Institute, Institute of Electrical and Electronics Engineers, Nuclear Regulatory Commission, and European Commission directives, and are implemented by operators like EDF Energy and contractors such as Fluor Corporation. Industry-specific frameworks include guidelines from International Maritime Organization for shipping, International Civil Aviation Organization for aviation, and consensus standards originating with American Petroleum Institute and Chemical Manufacturers Association affiliates. Oversight and review processes often involve consulting firms and audit bodies with ties to PricewaterhouseCoopers, Deloitte, Ernst & Young, and KPMG.

Critics cite overreliance on historical failure data from suppliers like GE Hitachi Nuclear Energy and service records maintained by Transocean as a weakness when novel configurations or rare events arise, echoing concerns raised after incidents involving Fukushima Daiichi Nuclear Power Plant and Deepwater Horizon. Other limitations highlighted in academic critiques from Oxford University and Yale University include model incompleteness, opaque assumptions in probabilistic models used by consultants affiliated with McKinsey & Company, and the challenge of capturing organizational and cultural factors emphasized in case work by Columbia University and King's College London.

Notable PRA-informed reviews and investigations include post-accident analyses for the Three Mile Island accident, accident investigations after the Space Shuttle Columbia disaster, safety cases for reactors such as Tokai Nuclear Power Plant and designs by Westinghouse Electric Company, offshore risk assessments following Deepwater Horizon, and aviation safety risk modeling for events studied by National Transportation Safety Board and Airbus. Research projects demonstrating PRA methodologies have been sponsored by organizations including Department of Energy, Defense Advanced Research Projects Agency, European Commission, and Bill & Melinda Gates Foundation.

Category:Risk assessment