Risk and performance based techniques in system health and maintenance monitoring

Integrating the risk and reliability analysis, system topology, and database manipulation methodologies into a unified system-health and performance monitoring tool has a great potential for improving plant operation and safety. A methodology has been developed in this study to integrate three basic elements of a performance monitoring system: data, analytical models, and system relations. In this methodology, the plant topology plays the major role. It defines the relationships among plant components, systems and structures. It provides the system configurations and causal relationships needed for performing risk and performance analyses. The causal relationships modeled by the plant topology are very crucial for evaluating the plant condition, safety. Integrating performance analysis tool with plant topology and also integrating it with the various classes of information is another aspect of the methodology proposed by this research.; The use of plant topology is a replacement for arrays of logics which usually construct the knowledge-base of an inference engine. The expansion of the plant topology as the basis for explaining the relationships among plant components, systems, and structures, captures the dynamics of plant condition. It is one of the major contributions of this study.; In addition, a maintenance schedule optimization technique has been developed and integrated. The main objective of this optimizer, build based on the genetic algorithm methodology, has been to reduce the risk and the cost of maintenance operation. The genetic algorithm is a relatively new technology which mimics the concept of natural evolution processes. The general features of a traditional genetic algorithm have been modified in this study to better fit the requirements of maintenance activities and conditions.; The impact of each potential schedule on both plant safety and economy, has been evaluated using a disvalue function. The impact of schedule lineup on plant safety is modeled by "risk model". To estimate the direct maintenance cost and anticipated financial losses due to plant shutdown "cost" and "loss" models have been proposed. Comparing the GA's schedule with the results of Monte Carlo method, GA/Schedule Optimizer was found superior to the Monte Carlo technique.