Quantification of margins and uncertainties: Alternative representations of epistemic uncertainty

In 2001, the National Nuclear Security Administration of the U.S. Department of Energy in conjunction with the national security laboratories (i.e., Los Alamos National Laboratory, Lawrence Livermore National Laboratory and Sandia National Laboratories) initiated development of a process designated Quantification of Margins and Uncertainties (QMU) for the use of risk assessment methodologies in the certification of the reliability and safety of the nation's nuclear weapons stockpile. A previous presentation, “Quantification of Margins and Uncertainties: Conceptual and Computational Basis,” describes the basic ideas that underlie QMU and illustrates these ideas with two notional examples that employ probability for the representation of aleatory and epistemic uncertainty. The current presentation introduces and illustrates the use of interval analysis, possibility theory and evidence theory as alternatives to the use of probability theory for the representation of epistemic uncertainty in QMU-type analyses. The following topics are considered: the mathematical structure of alternative representations of uncertainty, alternative representations of epistemic uncertainty in QMU analyses involving only epistemic uncertainty, and alternative representations of epistemic uncertainty in QMU analyses involving a separation of aleatory and epistemic uncertainty. Analyses involving interval analysis, possibility theory and evidence theory are illustrated with the same two notional examples used in the presentation indicated above to illustrate the use of probability to represent aleatory and epistemic uncertainty in QMU analyses.     

Ideas underlying the Quantification of Margins and Uncertainties

Key ideas underlying the application of Quantification of Margins and Uncertainties (QMU) to nuclear weapons stockpile lifecycle decisions are described. While QMU is a broad process and methodology for generating critical technical information to be used in U.S. nuclear weapon stockpile management, this paper emphasizes one component, which is information produced by computational modeling and simulation. In particular, the following topics are discussed: (i) the key principles of developing QMU information in the form of Best Estimate Plus Uncertainty, (ii) the need to separate aleatory and epistemic uncertainty in QMU, and (iii) the properties of risk-informed decision making (RIDM) that are best suited for effective application of QMU. The paper is written at a high level, but provides an extensive bibliography of useful papers for interested readers to deepen their understanding of the presented ideas.     

Quantification of margins and uncertainties: Example analyses from reactor safety and radioactive waste disposal involving the separation of aleatory and epistemic uncertainty

In 2001, the National Nuclear Security Administration (NNSA) of the U.S. Department of Energy (DOE) in conjunction with the national security laboratories (i.e., Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Sandia National Laboratories) initiated development of a process designated quantification of margins and uncertainties (QMU) for the use of risk assessment methodologies in the certification of the reliability and safety of the nation's nuclear weapons stockpile. A previous presentation, “Quantification of Margins and Uncertainties: Conceptual and Computational Basis,” describes the basic ideas that underlie QMU and illustrates these ideas with two notional examples. The basic ideas and challenges that underlie NNSA's mandate for QMU are present, and have been successfully addressed, in a number of past analyses for complex systems. To provide perspective on the implementation of a requirement for QMU in the analysis of a complex system, three past analyses are presented as examples: (i) the probabilistic risk assessment carried out for the Surry Nuclear Power Station as part of the U.S. Nuclear Regulatory Commission's (NRC's) reassessment of the risk from commercial nuclear power in the United States (i.e., the NUREG-1150 study), (ii) the performance assessment for the Waste Isolation Pilot Plant carried out by the DOE in support of a successful compliance certification application to the U.S. Environmental Agency, and (iii) the performance assessment for the proposed high-level radioactive waste repository at Yucca Mountain, Nevada, carried out by the DOE in support of a license application to the NRC. Each of the preceding analyses involved a detailed treatment of uncertainty and produced results used to establish compliance with specific numerical requirements on the performance of the system under study. As a result, these studies illustrate the determination of both margins and the uncertainty in margins in real analyses.     

An approximate epistemic uncertainty analysis approach in the presence of epistemic and aleatory uncertainties

Epistemic uncertainty analysis is an essential feature of any model application subject to ‘state of knowledge’ uncertainties. Such analysis is usually carried out on the basis of a Monte Carlo simulation sampling the epistemic variables and performing the corresponding model runs.In situations, however, where aleatory uncertainties are also present in the model, an adequate treatment of both types of uncertainties would require a two-stage nested Monte Carlo simulation, i.e. sampling the epistemic variables (‘outer loop’) and nested sampling of the aleatory variables (‘inner loop’). It is clear that for complex and long running codes the computational effort to perform all the resulting model runs may be prohibitive.Therefore, an approach of an approximate epistemic uncertainty analysis is suggested which is based solely on two simple Monte Carlo samples: (a) joint sampling of both, epistemic and aleatory variables simultaneously, (b) sampling of aleatory variables alone with the epistemic variables held fixed at their reference values.The applications of this approach to dynamic reliability analyses presented in this paper look quite promising and suggest that performing such an approximate epistemic uncertainty analysis is preferable to the alternative of not performing any.     

Summary from the epistemic uncertainty workshop: consensus amid diversity

The ‘Epistemic Uncertainty Workshop’ sponsored by Sandia National Laboratories was held in Albuquerque, New Mexico, on 6–7 August 2002. The workshop was organized around a set of Challenge Problems involving both epistemic and aleatory uncertainty that the workshop participants were invited to solve and discuss. This concluding article in a special issue of Reliability Engineering and System Safety based on the workshop discusses the intent of the Challenge Problems, summarizes some discussions from the workshop, and provides a technical comparison among the papers in this special issue. The Challenge Problems were computationally simple models that were intended as vehicles for the illustration and comparison of conceptual and numerical techniques for use in analyses that involve: (i) epistemic uncertainty, (ii) aggregation of multiple characterizations of epistemic uncertainty, (iii) combination of epistemic and aleatory uncertainty, and (iv) models with repeated parameters. There was considerable diversity of opinion at the workshop about both methods and fundamental issues, and yet substantial consensus about what the answers to the problems were, and even about how each of the four issues should be addressed. Among the technical approaches advanced were probability theory, Dempster–Shafer evidence theory, random sets, sets of probability measures, imprecise coherent probabilities, coherent lower previsions, probability boxes, possibility theory, fuzzy sets, joint distribution tableaux, polynomial chaos expansions, and info-gap models. Although some participants maintained that a purely probabilistic approach is fully capable of accounting for all forms of uncertainty, most agreed that the treatment of epistemic uncertainty introduces important considerations and that the issues underlying the Challenge Problems are legitimate and significant. Topics identified as meriting additional research include elicitation of uncertainty representations, aggregation of multiple uncertainty representations, dependence and independence, model uncertainty, solution of black-box problems, efficient sampling strategies for computation, and communication of analysis results.     

An approximate sensitivity analysis of results from complex computer models in the presence of epistemic and aleatory uncertainties

This paper focuses on sensitivity analysis of results from computer models in which both epistemic and aleatory uncertainties are present. Sensitivity is defined in the sense of “uncertainty importance” in order to identify and to rank the principal sources of epistemic uncertainty. A natural and consistent way to arrive at sensitivity results in such cases would be a two-dimensional or double-loop nested Monte Carlo sampling strategy in which the epistemic parameters are sampled in the outer loop and the aleatory variables are sampled in the nested inner loop. However, the computational effort of this procedure may be prohibitive for complex and time-demanding codes. This paper therefore suggests an approximate method for sensitivity analysis based on particular one-dimensional or single-loop sampling procedures, which require substantially less computational effort. From the results of such sampling one can obtain approximate estimates of several standard uncertainty importance measures for the aleatory probability distributions and related probabilistic quantities of the model outcomes of interest. The reliability of the approximate sensitivity results depends on the effect of all epistemic uncertainties on the total joint epistemic and aleatory uncertainty of the outcome. The magnitude of this effect can be expressed quantitatively and estimated from the same single-loop samples. The higher it is the more accurate the approximate sensitivity results will be. A case study, which shows that the results from the proposed approximate method are comparable to those obtained with the full two-dimensional approach, is provided.     

Quantification of epistemic and aleatory uncertainties in level-1 probabilistic safety assessment studies

There will be simplifying assumptions and idealizations in the availability models of complex processes and phenomena. These simplifications and idealizations generate uncertainties which can be classified as aleatory (arising due to randomness) and/or epistemic (due to lack of knowledge). The problem of acknowledging and treating uncertainty is vital for practical usability of reliability analysis results. The distinction of uncertainties is useful for taking the reliability/risk informed decisions with confidence and also for effective management of uncertainty. In level-1 probabilistic safety assessment (PSA) of nuclear power plants (NPP), the current practice is carrying out epistemic uncertainty analysis on the basis of a simple Monte-Carlo simulation by sampling the epistemic variables in the model. However, the aleatory uncertainty is neglected and point estimates of aleatory variables, viz., time to failure and time to repair are considered. Treatment of both types of uncertainties would require a two-phase Monte-Carlo simulation, outer loop samples epistemic variables and inner loop samples aleatory variables. A methodology based on two-phase Monte-Carlo simulation is presented for distinguishing both the kinds of uncertainty in the context of availability/reliability evaluation in level-1 PSA studies of NPP.     

An approach to combining unreliable pieces of evidence and their propagation in a system response analysis

The paper describes an approach to representing, aggregating and propagating aleatory and epistemic uncertainty through computational models. The framework for the approach employs the theory of imprecise coherent probabilities. The approach is exemplified by a simple algebraic system, the inputs of which are uncertain. Six different uncertainty situations are considered, including mixtures of epistemic and aleatory uncertainty.     

Latin hypercube sampling and the propagation of uncertainty in analyses of complex systems

The following techniques for uncertainty and sensitivity analysis are briefly summarized: Monte Carlo analysis, differential analysis, response surface methodology, Fourier amplitude sensitivity test, Sobol' variance decomposition, and fast probability integration. Desirable features of Monte Carlo analysis in conjunction with Latin hypercube sampling are described in discussions of the following topics: (i) properties of random, stratified and Latin hypercube sampling, (ii) comparisons of random and Latin hypercube sampling, (iii) operations involving Latin hypercube sampling (i.e. correlation control, reweighting of samples to incorporate changed distributions, replicated sampling to test reproducibility of results), (iv) uncertainty analysis (i.e. cumulative distribution functions, complementary cumulative distribution functions, box plots), (v) sensitivity analysis (i.e. scatterplots, regression analysis, correlation analysis, rank transformations, searches for nonrandom patterns), and (vi) analyses involving stochastic (i.e. aleatory) and subjective (i.e. epistemic) uncertainty.     

Quantification of margins and uncertainties of complex systems in the presence of aleatoric and epistemic uncertainty

Performance assessment of complex systems is ideally done through full system-level testing which is seldom available for high consequence systems. Further, a reality of engineering practice is that some features of system behavior are not known from experimental data, but from expert assessment, only. On the other hand, individual component data, which are part of the full system are more readily available. The lack of system level data and the complexity of the system lead to a need to build computational models of a system in a hierarchical or building block approach (from simple components to the full system). The models are then used for performance prediction in lieu of experiments, to estimate the confidence in the performance of these systems. Central to this are the need to quantify the uncertainties present in the system and to compare the system response to an expected performance measure. This is the basic idea behind Quantification of Margins and Uncertainties (QMU). QMU is applied in decision making—there are many uncertainties caused by inherent variability (aleatoric) in materials, configurations, environments, etc., and lack of information (epistemic) in models for deterministic and random variables that influence system behavior and performance. This paper proposes a methodology to quantify margins and uncertainty in the presence of both aleatoric and epistemic uncertainty. It presents a framework based on Bayes networks to use available data at multiple levels of complexity (i.e. components, subsystem, etc.) and demonstrates a method to incorporate epistemic uncertainty given in terms of intervals on a model parameter.     

数据稀缺与更新条件下基于概率密度演化-测度变换的认知不确定性量化分析

工程设计中往往需要同时处理固有不确定性与认知不确定性。对于固有不确定性分析与量化,国内外已有诸多研究,例如Monte Carlo方法、正交多项式展开理论和概率密度演化理论等。而对认知不确定性、特别是固有不确定性与认知不确定性耦合情况下的研究,则还相对缺乏。该文中,针对数据稀缺与数据更新导致的认知不确定性,首先分别引入Bootstrap方法和Bayes更新方法进行不确定性表征。在此基础上,结合基于概率密度演化-测度变换的两类不确定性量化统一理论新框架,提出了存在认知不确定性情况下的不确定性传播与可靠性分析高效方法及其具体数值算法。由此,给出了基于数据进行工程系统不确定性量化、传播与可靠性分析的基本途径。通过具有工程实际数据的3个工程实例分析,包括无限边坡稳定性分析、挡土墙稳定性分析和屋面桁架结构可靠性分析,验证了该文方法的精度和效率。     

Error and uncertainty in modeling and simulation

This article develops a general framework for identifying error and uncertainty in computational simulations that deal with the numerical solution of a set of partial differential equations (PDEs). A comprehensive, new view of the general phases of modeling and simulation is proposed, consisting of the following phases: conceptual modeling of the physical system, mathematical modeling of the conceptual model, discretization and algorithm selection for the mathematical model, computer programming of the discrete model, numerical solution of the computer program model, and representation of the numerical solution. Our view incorporates the modeling and simulation phases that are recognized in the systems engineering and operations research communities, but it adds phases that are specific to the numerical solution of PDEs. In each of these phases, general sources of uncertainty, both aleatory and epistemic, and error are identified. Our general framework is applicable to any numerical discretization procedure for solving ODEs or PDEs. To demonstrate this framework, we describe a system-level example: the flight of an unguided, rocket-boosted, aircraft-launched missile. This example is discussed in detail at each of the six phases of modeling and simulation. Two alternative models of the flight dynamics are considered, along with aleatory uncertainty of the initial mass of the missile and epistemic uncertainty in the thrust of the rocket motor. We also investigate the interaction of modeling uncertainties and numerical integration error in the solution of the ordinary differential equations for the flight dynamics.     

Characterization of subjective uncertainty in the 1996 performance assessment for the Waste Isolation Pilot Plant

The 1996 performance assessment (PA) for the Waste Isolation Pilot Plant (WIPP) maintains a separation between stochastic (i.e. aleatory) and subjective (i.e. epistemic) uncertainty, with stochastic uncertainty arising from the possible disruptions that could occur at the WIPP over the 10,000 yr regulatory period specified by the US Environmental Protection Agency (40 CFR 191, 40 CFR 194) and subjective uncertainty arising from an inability to uniquely characterize many of the inputs required in the 1996 WIPP PA. The characterization of subjective uncertainty is discussed, including assignment of distributions, uncertain variables selected for inclusion in analysis, correlation control, sample size, statistical confidence on mean complementary cumulative distribution functions, generation of Latin hypercube samples, sensitivity analysis techniques, and scenarios involving stochastic and subjective uncertainty.     

Challenge problems: uncertainty in system response given uncertain parameters

The risk assessment community has begun to make a clear distinction between aleatory and epistemic uncertainty in theory and in practice. Aleatory uncertainty is also referred to in the literature as variability, irreducible uncertainty, inherent uncertainty, and stochastic uncertainty. Epistemic uncertainty is also termed reducible uncertainty, subjective uncertainty, and state-of-knowledge uncertainty. Methods to efficiently represent, aggregate, and propagate different types of uncertainty through computational models are clearly of vital importance. The most widely known and developed methods are available within the mathematics of probability theory, whether frequentist or subjectivist. Newer mathematical approaches, which extend or otherwise depart from probability theory, are also available, and are sometimes referred to as generalized information theory (GIT). For example, possibility theory, fuzzy set theory, and evidence theory are three components of GIT. To try to develop a better understanding of the relative advantages and disadvantages of traditional and newer methods and encourage a dialog between the risk assessment, reliability engineering, and GIT communities, a workshop was held. To focus discussion and debate at the workshop, a set of prototype problems, generally referred to as challenge problems, was constructed. The challenge problems concentrate on the representation, aggregation, and propagation of epistemic uncertainty and mixtures of epistemic and aleatory uncertainty through two simple model systems. This paper describes the challenge problems and gives numerical values for the different input parameters so that results from different investigators can be directly compared.     

Relative contributions of aleatory and epistemic uncertainty sources in time series prediction

This paper develops a novel computational framework to compute the Sobol indices that quantify the relative contributions of various uncertainty sources towards the system response prediction uncertainty. In the presence of both aleatory and epistemic uncertainty, two challenges are addressed in this paper for the model-based computation of the Sobol indices: due to data uncertainty, input distributions are not precisely known; and due to model uncertainty, the model output is uncertain even for a fixed realization of the input. An auxiliary variable method based on the probability integral transform is introduced to distinguish and represent each uncertainty source explicitly, whether aleatory or epistemic. The auxiliary variables facilitate building a deterministic relationship between the uncertainty sources and the output, which is needed in the Sobol indices computation. The proposed framework is developed for two types of model inputs: random variable input and time series input. A Bayesian autoregressive moving average (ARMA) approach is chosen to model the time series input due to its capability to represent both natural variability and epistemic uncertainty due to limited data. A novel controlled-seed computational technique based on pseudo-random number generation is proposed to efficiently represent the natural variability in the time series input. This controlled-seed method significantly accelerates the Sobol indices computation under time series input, and makes it computationally affordable.     

Reliability analysis of structures based on a probability‐uncertainty hybrid model

The traditional reliability analysis method based on probabilistic method requires probability distributions of all the uncertain parameters. However, in practical applications, the distributions of some parameters may not be precisely known due to the lack of sufficient sample data. The probabilistic theory cannot directly measure the reliability of structures with epistemic uncertainty, ie, subjective randomness and fuzziness. Hence, a hybrid reliability analysis (HRA) problem will be caused when the aleatory and epistemic uncertainties coexist in a structure. In this paper, by combining the probability theory and the uncertainty theory into a chance theory, a probability‐uncertainty hybrid model is established, and a new quantification method based on the uncertain random variables for the structural reliability is presented in order to simultaneously satisfy the duality of random variables and the subadditivity of uncertain variables; then, a reliability index is explored based on the chance expected value and variance. Besides, the formulas of the chance theory‐based reliability and reliability index are derived to uniformly assess the reliability of structures under the hybrid aleatory and epistemic uncertainties. The numerical experiments illustrate the validity of the proposed method, and the results of the proposed method can provide a more accurate assessment of the structural system under the mixed uncertainties than the ones obtained separately from the probability theory and the uncertainty theory.     

Verification of the calculation of probability of loss of assured safety in temperature-dependent systems with multiple weak and strong links

Two approaches to the calculation of probability of loss of assured safety (PLOAS) in temperature dependent weak link/strong link systems are described and compared on the basis of three test problems. The approaches differ in that the first approach permits a separation of epistemic and aleatory uncertainty in the calculation of PLOAS and the second approach combines epistemic and aleatory uncertainty before the calculation of PLOAS. A discrepancy in the results obtained with the test problems led to the identification of an implementation error for one of the approaches. The importance and efficacy of well-designed verification test problems are demonstrated.     

Decision making with epistemic uncertainty under safety constraints: An application to seismic design

The problem of accounting for epistemic uncertainty in risk management decisions is conceptually straightforward, but is riddled with practical difficulties. Simple approximations are often used whereby future variations in epistemic uncertainty are ignored or worst-case scenarios are postulated. These strategies tend to produce sub-optimal decisions. We develop a general framework based on Bayesian decision theory and exemplify it for the case of seismic design of buildings. When temporal fluctuations of the epistemic uncertainties and regulatory safety constraints are included, the optimal level of seismic protection exceeds the normative level at the time of construction. Optimal Bayesian decisions do not depend on the aleatory or epistemic nature of the uncertainties, but only on the total (epistemic plus aleatory) uncertainty and how that total uncertainty varies randomly during the lifetime of the project.     

Characterization of stochastic uncertainty in the 1996 performance assessment for the Waste Isolation Pilot Plant

The 1996 performance assessment (PA) for the Waste Isolation Pilot Plant (WIPP) maintains a separation between stochastic (i.e. aleatory) and subjective (i.e. epistemic) uncertainty, with stochastic uncertainty arising from the possible disruptions that could occur at the WIPP over the 10,000-yr regulatory period specified by the US Environmental Protection Agency (40 CFR 191, 40 CFR 194) and subjective uncertainty arising from an inability to uniquely characterize many of the inputs required in the 1996 WIPP PA. The characterization of stochastic uncertainty is discussed, including drilling intrusion time, drilling location, penetration of excavated/nonexcavated areas of the repository, penetration of pressurized brine beneath the repository, borehole plugging patterns, activity level of waste, and occurrence of potash mining. Additional topics discussed include sampling procedures, generation of individual 10,000-yr futures for the WIPP, construction of complementary cumulative distribution functions (CCDFs), mechanistic calculations carried out to support CCDF construction, the Kaplan/Garrick ordered triple representation for risk, and determination of scenarios and scenario probabilities.