No abstract

No abstract

Recent nuclear power plant events (e.g., Chernobyl, Diablo Canyon, and Vogtle) and U.S. Nuclear Regulatory Commission (NRC) reports (e.g., NUREG-1449) have led to concerns regarding human reliability during low power and shutdown (LP&S) conditions and limitations of human reliability analysis (HRA) methodologies in adequately representing the LP&S environment. As a result of these concerns, the NRC initiated two parallel research projects to assess the influence of LP&S conditions on human reliability through an analysis of operational experience at pressurized water reactors (PWRs) and boiling water reactors (BWRs). These research projects, performed by Brookhaven National Laboratory for PWRs, and Sandia National Laboratories for BWRs, identified unique aspects of human performance during LP&S conditions and provided a program plan for research and development necessary to improve existing HRA methodologies. This report documents the results of the analysis of LP&S operating experience and describes the improved HRA program plan.

Where the papers in these proceedings have been authond by contractors ofthe United States Government, neitherthe UnitedStatesGovemmentnoranyagencythelrtof, noranyoftheiremployees, makesanywmanty, expressed or implied, or assumes any legal liability or responsibility for any third party's use, or the results of such use, of any information, apparatus, product, or process disclosed in these proceedings, or represents that its use by such third party would not infringe privately owned rights. The views expressed in these proceedings are not necessarily those of the U.S. Nuclear Regulatory Commission.

Exlstlng software technology f o r buudlng expert systems provides m naechanlslns to support simUar design and Impkmentation f o r closely-related applications. Such mechunisms could provide significant advantage f o r lowering the cost of expert systems by redudng both the effort required f o r development and the trcllnlng and expe&nce required by system developers. I n this paper, we consider how an appropdate organization of knowledge in expert systems can support more costeffective development f o r famules of appUcat&ns. W e describe the design and implementation of a kmwledge base organization f o r troubleshooting applhtions involving a f d y of wmplex electrical-mecM-hydr& equipment. Expert systems'* ' * ' solve problems in narrow, well-defined application domains by applymg simple, general inference procedures to substantial. knavledge bases of applicationdependent symbolic knowledge. During the past 15 years research and development projects following this basic approach have created dozens of prototype expert systems. These efforts have produced a distinct, new software technology for supporting the implementation of expert systems. They have also explored a range of applications in the general areas of diagnceidinterpretation, configuratioddesign, and planning.Despite advances in software technology and a growth in understanding about how to approach general classes of applications, the development process for expert systems remains too w t l y to justify their application to many otherwise suitable problems. M g n and implementation typically require many revision cycles that contribute to high manpower costs. One approach to improving cost-effectiveness involves obtaining more leverage from prior system development efforts.' , Currently, valuable know-how acquired through experience has not been disseminated effectively.mechanism, except personal experience, for significantly reducing the effort required to build systems for closely related applications.In this paper, we discuss how knowledge common to a family of applications can be organized to extend currentThere is no effective D. Whitehead FMC Corp. Northern Ordnance DlolslonMtnneapoUs, M N , U S A expert systems software tedmology. As an example, we describe an "application-generic" expert systems software package we have designed and implemented for fault diagn& a p plications involving complex electrical-hydraulic-mechanical (E3IM) equipment. This work has had two motivations. On the theoretical side, we would like to advance our understanding of how knowledge representation affects problemsolving performance. On the practical side, we would like to increase the accessibility and cost-effectiveness of solutions incorporating the expert systems approach. Knowledge Representation in Expert SystemsExpert systems commonly represent application applicationdependent knowledge in knowledge bases containing hundreds or thousands of implic* that relate situations and conclusions. In most current designs, the situations and conclusions are typically ...