Maintainability analysis using Q-GERT network simulation

Abstract

In reliability analyses of nuclear reactor safety systems, time- dependent phenomena (such as component failure during op eration and the ensuing repair activity) are often disregarded or treated quasi-statically because the time dependencies involved are not amenable to closed-form analytic formulation. In addi tion, conditional phenomena (such as a failure mode that depends on the failure mode in a previous event) and phenom ena that must be characterized by arbitrary probability distribu tions are often ignored. This paper reports the initial applica tion of the computer program Q-GERT, an existing code that uses dynamic Monte Carlo modeling, to reactor safety-system simulation and analysis. Using Q-GERT with its Monte Carlo techniques, the above shortcomings can, in principle, be dealt with, and the analysis can be elevated from the level of a reliabil ity analysis to that of a maintainability analysis. This new applica tion of Q-GERT, called the Maintainability Analysis Procedure (MAP), is illustrated by simulating and evaluating the behavior of a simple safety system as this system is subjected to each of four possible failure processes. These processes are distinguished by the number and complexity of the included phenomena, and the effects on estimates of the probability of system failure are shown to be significant as the processes become more complex.