Bayesian parameter estimation in probabilistic risk assessment1
Introduction
A probabilistic risk assessment (PRA) is an exercise aimed at estimating the probability and consequences of accidents for the facility or process being studied. Although not required by this definition, PRA often involves the analysis of low-probability events for which few data are available. Bayesian parameter estimation techniques are useful because, unlike classical statistical methods, they are able to incorporate a wide variety of information types, e.g. expert judgement as well as statistical data, in the estimation process.
In addition to their ability to deal with sparse data, Bayesian techniques are appropriate for use in PRA because they are derived from the framework of subjective probability. This framework, which holds that probability is a subjective (internal) measure of event likelihood, is an integral part of current theories on decision-making under uncertainty (e.g. [1]). A further, practical advantage of the subjective probability framework in PRA applications is that propagation of uncertainties through complex models is relatively simple. On the other hand, it is very difficult, and intractable in `real' problems, to propagate classical statistical confidence intervals through PRA models to estimate a confidence interval for a composite result of interest (e.g.reactor core damage frequency).
For these reasons, Bayesian techniques, first championed for use in PRA in the late 1970s 2, 3, have become widespread in that field. This paper presents a tutorial on Bayesian techniques especially relevant to PRA. It outlines the motivation behind the Bayesian approach, summarizes the general parameter estimation problem, presents methods for constructing likelihood functions and prior distributions, discusses some useful results, and ends with a variety of cautions and comments regarding practical applications. The intent is to provide an expository review with simple but realistic examples. A number of references are provided for readers interested in exploring various topics in greater detail.
Section snippets
Motivation and framework
The following discussion provides an introduction to the issues motivating the use of Bayesian methods in PRA and outlines the parameter estimation problem that is the focus of this paper. More detailed discussions can be found in [4] and [5].
Bayes's theorem
According to the theory of subjective probability, Bayes's theorem is the only way in which a coherent analyst, i.e. an analyst whose probabilities behave according to the laws of probability [8], can update his/her state of knowledge. This section describes the general features of Bayes's theorem and introduces a simple application of it. Additional material, both introductory and advanced, can be found in numerous texts (e.g. [9], [10]).
The likelihood function
The construction of an appropriate likelihood function requires engineering/scientific knowledge specific to the process being modeled, as well as probabilistic modeling expertise. This section raises a number of general points on modeling considerations and provides some functions widely used in PRAs. A special topic relevant to the construction of the likelihood function, i.e. the treatment of uncertain data, is discussed in Section 7.
Prior distributions
Development of an appropriate prior distribution is, sometimes with just cause, the most controversial part of a Bayesian analysis. It can often be the most resource-intensive as well. This section discusses some of the issues to be considered when creating an informative prior distribution (i.e. a prior distribution that at least partly reflects the analyst's state of knowledge) as well as a number of so-called `objective', noninformative prior distributions.
A numerical example
We now illustrate the concepts presented in the preceding sections with a simplified example. Assume that our population of interest consists of 10 emergency diesel generators (EDGs), similar to those used by commercial nuclear power plants in the USA as sources of emergency power. Assume that our task is to estimate φ1, the probability that the diesel at our plant (designated EDG no.1) will fail to start on demand. At our plant, there have been 140 successful demands (i.e. no failures). The
Implementation notes and other special topics
With the development of a prior distribution and a likelihood function, the estimation process is fairly straightforward for problems of common interest in PRA; it often requires only the solution (analytical or numerical) of Bayes's theorem for a single unknown parameter. This section briefly discusses a number of topics relevant to Eq. (5), including some common pitfalls. It also introduces an advanced topic relevant to current PRA applications: the treatment of uncertain data.
Concluding remarks
To many PRA analysts, `Bayesian analysis' can mean a mechanical updating of either a roughly defined informative prior distribution (i.e. a prior distribution which only roughly approximates the analyst's belief concerning the parameter of interest) or some form of `objective' prior distribution (e.g. a noninformative or maximum entropy prior). While Bayes's theorem is straightforward in principle, and while practical Bayesian analysis for PRA applications is often a fairly simple process,
Acknowledgements
The authors gratefully acknowledge the helpful comments of G. Apostolakis, A. Mosleh, T. Leahy and the anonymous referees, and the assistance of D. Siu in preparing the manuscript. This paper was supported by the Idaho National Engineering Laboratory Center for Reliability and Risk Assessment under the US Department of Energy (DOE) Idaho Field Office Contract DE-AC07-76ID01570. The opinions, findings, conclusions and recommendations expressed herein are those of the authors and do not
References (54)
Hidden sources of uncertainty: judgment in the collection and analysis of data
Nuclear Engineering and Design
(1986)- et al.
Data specialization for plant specific risk studies
Nuclear Engineering and Design
(1980) More comprehensive approaches for determining risk sensitivities to component aging
Nuclear Engineering and Design
(1988)The two-stage Bayesian method used for the T-Book application
Reliability Engineering and System Safety
(1996)Constrained non-informative priors in risk assessment
Reliability Engineering and System Safety
(1996)- et al.
Response: On the `true' value of a failure rate and related statistical issues
Reliability Engineering and System Safety
(1989) A Monte Carlo method for multiple parameter estimation in the presence of uncertain data
Reliability Engineering and System Safety
(1990)- Keeney, R. L. and Raiffa, H., Decisions with Multiple Objectives. Cambridge University Press,...
Probability and risk assessment: the subjectivistic viewpoint and some suggestions
Nuclear Safety
(1978)- et al.
Characterization and evaluation of uncertainty in probabilistic risk analysis
Nuclear Safety
(1981)
The concept of probability in safety assessments of technological systems
Science
On the quantitative definition of risk
Risk Analysis
Expert opinion and statistical evidence: an application to reactor core melt frequency
Nuclear Science and Engineering
The assessment of probability distributions from expert opinions with an application to seismic fragility curves
Risk Analysis
A statistical model for combining biased expert opinions
IEEE Transactions on Reliability R-
On a `two-stage' Bayesian procedure for determining failure rates
IEEE Transactions on Power Apparatus and Systems PAS-
Bayes empirical Bayes
Journal of the American Statistical Association
Cited by (211)
Reliability modeling in a predictive maintenance context: A margin-based approach
2024, Reliability Engineering and System SafetyA Bayesian approach for reliability estimation for non-homogeneous and interval-censored failure data
2024, Process Safety and Environmental ProtectionL-moments and Bayesian inference for probabilistic risk assessment with scarce samples that include extremes
2023, Reliability Engineering and System SafetyDynamic-risk-informed safety barrier management: An application to cost-effective barrier optimization based on data from multiple sources
2023, Journal of Loss Prevention in the Process Industries
- 1
The views and conclusions in this paper are those of the authors and should not be interpreted as necessarily representing the views or official policies, either expressed or implied, of the US Nuclear Regulatory Commission or the US Department of Energy.