Approaching Dynamic Reliability By Monte Carlo Simulation

Abstract

Abstract. The event tree/fault tree methodology is probably the most widely recognized and adopted method for assessing the reliabilty characteristics of a plant. However the methodology might suffer some limitations when treating scenarios in which the dynamics of the system play an important role. In these cases a dynamic approach seems appropriate: such analysis should take into account the mutual effects of the process physical evolution, the operator's performance and the system hardware configuration,. In general this analysis requires large efforts both in the implementation of the relating models and in their evaluation as a function of time, which in tum leads to large CPU times. Therefore, a research effort aiming at developing fast models and more efficient techniques for their evaluation is needed. Moreover the effects of the input uncertainties on the resulting output should be considered, this being particularly true for the Human Reliability Analysis (HRA) both for the operator model and its input data. In this regard the Monte Carlo approach seems to enjoy the desirable features.In this paper we present a Monte Carlo dynamic approach to reliability and compare it to a classical static analysis in which the time-dependent response of the plant and the control are not followed. A simple example shows how in those cases in which the reliability characteristics of the plant strongly depend on the process, the results obtained by the two methodologies are quite different. The simulation allows to take into account some important issues such as components' aging, failures on demand, operator's actions. Here we present a simple model of failures on demand and the effects that these failures have on the reliability of the system. Biasing of such failures is also introduced and in this regard we point out that these techniques in general require a high degree of care and judgment.

Supervised dynamic probabilistic risk assessment: Review and comparison of methods

“…Therefore, it should be possible to search it exhaustively in polynomial or linear time. Examples of this approach with MCS for scenario generation is shown in [54,55,56]. Memorizing interesting parts of the event sequence and then using them in the generation of interesting scenarios via MCS is the approach shown in [23,22].…”