D. Okrent

Comment on societal risk.

There is a need to measure societal risk more accurately; to examine and reevaluate our priorities in risk reduction; to determine the level of expenditure for risk reduction beyond which adverse economic and political effects may be overriding; and to develop a national approach to risk management.

Uncertainties in system analysis: Probabilistic versus nonprobabilistic theories

Abstract The theory of evidence and the theory of possibility have been suggested as possible alternatives to probability theory in safety analyses of engineering systems. This paper discusses three issues: (1) how formal probability theory has been relaxed to develop these nonprobabilistic models; (2) how degrees of belief are expressed in probabilistic and nonprobabilistic theories; and (3) the degree to which these nonprobabilistic models can be applied to system analysis in terms of their capability to combine knowledge.

A taxonomy of issues related to the use of expert judgments in probabilistic safety studies

Abstract Expert judgments are frequently used in probabilistic safety assessments (PSA). However, the methods employed in practice are very crude and a large gap exists between the theoretical methods available and actual practice. A taxonomy of issues related to the use of expert judgments in PSA was considered necessary to identify the needs of the practitioners and the applicability of existing models. In this paper, a taxonomy of issues related to the use of expert judgments in PSA is systematically reviewed with examples from case studies. Issues surrounding the expert judgement procces can be classified into two categories—(a) elicitation, and (b) the use of expert judgments. Various elements of these categories, such as model and parameter uncertainty, decomposition, the use of multiple experts, the selection of experts, expert training, elicitation, effect of information provided to experts, expert calibration, availability of evidence, opinion aggregation and dependence are then discussed. The issues of expert bias, calibration and dependence are of special concern. Sources of expert bias and dependence are discussed with some thoughts on overcoming them using examples from selected case studies.

A new methodology for the computer-aided construction of fault trees

Abstract A methodology for systematically constructing fault trees for general complex systems is developed. A means of modeling component behavior via decision tables is presented, and a procedure for constructing and editing fault trees, either manually or by computer, is developed. The techniques employed result in a complete fault tree in standard form. In order to demonstrate the methodology, the computer program CAT was developed and is used to construct trees for a nuclear system. By analyzing and comparing these fault trees, several conclusions are reached. First, such an approach can be used to produce fault trees that accurately describe system behavior. Second, multiple trees can be rapidly produced by defining various TOP events, including system success. Finally, the accuracy and utility of such trees is shown to depend upon the careful development of the decision table models by the analyst, and of the overall system definition itself. Thus the method is seen to be a tool for assisting in the work of fault tree construction rather than a replacement for the careful work of the fault tree analyst.

On the safety of Tokamak-type, central station fusion power reactors

Abstract A preliminary examination is made of several potential safety questions for Tokomak-type, central station fusion power reactors, using UWMAK-I as a reference concept. Larger tritium inventories than previously reported are estimated. The divertor loss-of-flow accident appears to require fast plasma shutdown. The large helium inventory in the cryogenic magnets affords a mechanism for generating large pressures within the containment. Lithium-air and lithium-concrete reactions have the potential for generating large amounts of heat, significant pressures, and active chemical species capable of volatilizing structural material such as a radioactive first wall.

Future risk from a hypothesized RCRA site disposing of carcinogenic metals should a loss of societal memory occur

Abstract The future risk of a hypothesized Resource Conservation and Recovery Act (RCRA) site disposing of carcinogenic metals, arsenic, chromium, nickel, cadmium, and beryllium in the U.S. is assessed. Societal memory is assumed to be lost regarding the site. A human intrusion scenario on the site and a residential scenario one kilometer down-gradient of the groundwater flow direction from the site are assumed, starting at 1000 years after the sites closure. For the human intrusion scenario, the exposure pathways considered are fruit and vegetable intake, soil ingestion, and dermal contact with soil. The quantitative results obtained for the three pathways are as follows: lifetime excess cancer risk due to fruit and vegetable intake is 0.18; isk due to dermal contact with the soil is 0.12; and risk due to soil ingestion is 2.6 x 10 ṫ . For the residential scenario, only qualitative discussion of exposure via groundwater is presented due to the large uncertainties. The U.S. EPA (Environmental Protection Agency) attention to and requirements concerning long-term risk from RCRA sites containing metal carcinogens, which never change due to radioactive decay, stand in sharp contrast to the stringent requirements over 10,000 years posed by EPA for geologic disposal of high level radioactive wastes, and the long-term requirements posed by the U.S. Nuclear Regulatory Commission for low level radioactive waste disposal sites.

An expert system approach for fault diagnosis to cope with spurious sensor signals and process state uncertainty

Abstract In recent years, there is an increasing interest in using expert systems to assist plant operators in fault diagnosis and decision making during and after accidents or trips. Since uncertainties in the knowledge base exist in most practical cases, how to deal with uncertainties, including spurious sensor signals, is an important issue that distinguishes different expert system approaches. This paper presents an approach that involves the derivation of the evidence-specific hypothesis space from the evidence (usually the sensor signals) collected on-line at a point in time. The knowledge base is constructed based on either fault trees or logic flowgraphs. Special attention is paid to the statistically independent initiating events. Spurious signals are treated by specifically including sensor failures and design defects. Based on a frequency-based calculation, the updated probability distribution over the hypothesis space is calculated. The approach is illustrated by a simple example. The practicality of application to very large systems remains to be explored.

A survey of expert opinion on low probability earthquakes

Abstract Seven experts in the field of earthquakes independently submitted their estimates of the probability per year of earthquakes of various sizes at each of eleven nuclear power reactor sites in the United States. In general, very large differences appeared among the predictions for every site, and the dominating seismic factors in each experts judgment varied widely. Comparison indicates that the Safe Shutdown Earthquake values accepted by the Atomic Energy Commission for peak horizontal acceleration correspond approximately to that acceleration having a median (or average) probability of occurrence per year at the site of about 10 −4 , if the opinions of these seven experts are used as the data base.

A knowledge-based prototype for optimization of preventive maintenance scheduling☆

Abstract A methodology has been developed, and a prototype tool, the Maintenance Advisor, has been designed and implemented based on this methodology which would assist the scheduling and decision-making for performance of preventive maintenance activities in a plant, based on probabilistic judgedment and probabilistic inference rules. Using data on failure rates, repair times, repair costs and indirect economic costs (e.g. power replacement and accident risk), and within the imposed deterministic constrainst, the program develops an optimum (minimum expected cost) maintenance schedule for the various pieces of equipment described by the model. The Maintenance Advisor is a frame-based object-oriented tool, programed in KEE and Lisp. Equipment and other objects are represented as complex units, containing a complete set of characteristics, data and functional capabilities. Functional relations between the units are described in terms of two relations: TYPE-OF and PART-OF. The hierarchies formed by these relations serve as the basis for probabilistic and other inferences.

Optimal safety goal allocation for nuclear power plants

Abstract This paper is on the development of a methodology for safety goal allocation given cost functions and Probabilistic Risk Assessment (PRA) models. The safety goal allocation problem is formulated as a constrained nonlinear optimization problem. Based on the decomposition principle, the safety goal allocation problem is first reduced to a lower order optimization problem. Then, a set of optimization algorithms is used to solve the decomposed optimization problem. In order to introduce expert opinion, weighting factors are incorporated in the objective function. Computer software was developed to perform the optimization and sensitivity analysis. The proposed method is demonstrated by using an industrially developed PRA model.