RELIABILITY IMPROVEMENT ANALYSIS BASED ON FAULT TREES

Abstract

Fault tree analysis is a technique used in the study of the reliability of complex systems. Since its first introduction in the 1960's, fault tree analysis has evolved into one of the most powerful and widely used techniques for investigating the reliability of industrial systems. It is also used to quantify risks associated with potentially hazardous systems. Most of the studies carried out are related to how to construct a fault tree and how to determine the top event probability. A major obstacle encountered during fault tree analysis of complex systems is that the fault tree can be very huge. Due to the large number of gates and events required to be input, fault tree analysis can take a long time even when computers or workstations are available. If the fault trees can be simplified, large amount of time and effort can be spared. In order to overcome this problem, several methods have been developed. They are, however, rather tedious to apply and require one to have strong concepts and knowledge in order to correctly carry out the fault tree simplification. Another drawback of some of these methods is that they rely on truncation methodologies and approximations, which may not be appropriate at times. Another area of significance is determining the importance of basic events. Precise recognition of problem spots is a crucial step in order to effectively improve the reliability of a system. In most practical conditions, only limited resources are available for systems improvement. Therefore, there is a need to distinguish accurately the most critical parts of a system, so that effective actions can be taken to improve the system's overall reliability. To overcome this obstacle, the present method available is to use importance measures. There are quite a number of different importance measures available. However, they are not always easy to apply, especially when the system in question is very complex. Also, the user must know when different importance measures should be used. In this project, three principles that reduce the fault tree are introduced. These new reduction principles are simple to apply and they can also be used to reduce fault trees with repeated events. No calculations or formulations is required which means that reduction of the fault trees can be carried out very quickly in the absence of computers. Upon successful reduction of the fault trees, a new ranking approach is introduced to determine the significance of the basic events. This new ranking approach is a qualitative classification tool. Its main strength is that it is relatively accurate and also easy to apply. Again, as in the proposed reduction techniques, no calculations or formulations are involved in this new method. It can also be applied to fault trees with repeated events and undeveloped events.