![]() ![]() The study has used a log-normal distribution for all the uncertainties assigned to component failure rates. Based on these uncertainties, a Monte Carlo technique was used to calculate the uncertainty of the overall system failure probability. In a few cases the uncertainty is a factor of 1000 (± 30). In general, the failure rates used in this study have uncertainties of a factor of 10 or 100 (± 3 or ± 10). Based on these assessments, component failure rates were modified to account for the deficiencies found.Ī common criticism of the fault tree method is that the system failure probabilities are not meaningful because of uncertainties in the knowledge of applicable failure rates. To ascertain how likely such design errors may be, this study carefully reviewed the components in a selected number of important safety systems to determine how well the design specifications had, in fact, been satisfied. It is, of course, possible that certain components may fail to meet these special design conditions. The design specifications of the components of the ESF's require that they be qualified to operate under a variety of accident conditions. Based on their assessments, component failure rates and their uncertainties were increased for the extreme environments. In the process of determining applicable failure rates, the study employed specialists in component reliability to assess the effect of such conditions on system components. Foremost among these environmental factors are radiation and high temperature steam. The compilations of such industrial experience are the basic source of most of the failure rate data used in the study.Ĭertain components of nuclear systems may be subjected to rather unique environments, particularly during serious accidents. The conditions of service of most of the components in reactors are similar to conditions in many other applications, such as those in fossil-fueled plants and chemical processing plants. Each source has been investigated to determine its appropriateness for application to nuclear plants. In this study extensive searches have been made for sources of failure rate data. It has therefore been necessary to also use data from a much broader base of industrial experience. In the case of reactors the experience of a few hundred reactor-years is not sufficient by itself to provide statistically meaningful probabilities for most of the required component failure rates. The accuracy of the fault tree method is highest when component failure rates are based on data obtained from failures in systems identical to the one under analysis. ![]() These failure data included estimates of component failures, human errors, and testing and maintenance contributions. However, the majority of failure data was utilized as input to the fault trees so that the probabilities of the system failures could be determined. Such uses included the determinations of the probabilities of initiating events such as pipe breaks and reactor vessel ruptures. They were used directly to establish the probabilities of major events (failures) for which fault trees were not constructed. The study utilized failure rate data in two principal ways. Weaver, in Nuclear Power Safety, 1976 Failure Rate Data. If, however, our reliability based on the best case scenario is unacceptable, we have to take corrective action by looking for improved designs.Ģ.0 Constraining, confining, containing elementsĥ.3 Temperature effects (see under aging)Ī REVIEW OF ACCIDENT RISKS IN LIGHT-WATER-COOLED NUCLEAR POWER PLANTS If that is the case, we need not worry because the actual quality will be closer to the expected value. We will then investigate if the worst case is viable. In order to make things less complicated we will calculate reliability based on two qualities – best and worst. This reflects the fact that machinery parts and components can have different qualities. In using failure rate data for machinery reliability assessment it is a good idea to work with “worst,” “best,” and “expected” concepts. Basic failure modes and the failure mechanisms associated with them play a central role in machinery failure analysis. Earlier we looked at some basic failure modes in connection with failure distributions. Failure modes have distinct failure rates and the component or part failure rate is the sum of its mode failure rate.įailure modes are typically first a description of loss of function or malfunction and then a more detailed expansion in terms of the basic failure mode, namely the appearance of the failure (see Table 7-1). Another important aspect of reliability prediction using failure rates is the consideration of failure modes. ![]()
0 Comments
Leave a Reply. |