Reliability analysis and optimization of complex systems

Most of the earlier literature on system reliability optimization consider only the simple series-parallel systems subject to oneror two constraints. Practical systems have complex rather than the simple series parallel configurations. With a view towards solving these complex system reliability optimization problems, an efficient computer algorithm based on the path enumeration method has been developed. An important feature of the method is the module representation of the reliability graph which considerably simplifies the calculation of reliability and sensitivity functions of complex systems. A modified integer gradient method is used for system optimization. Although the method does not insure a global optimum, it does find various near-optimum solutions. From a practical consideration, this could provide for a wider choice during the design phase. In this research an effort has also been made to apply basic reliability concepts to process plants. A new formulation of the optimal reliability design of process plants which takes into account the quantitative aspects of systems throughput is proposed. It is based on the k-out-of-n configuration instead of the conventional parallel redundancy configuration. The problem is so formulated that determining the optimum configuration also determines the optimum capacity of units to be used at each stage of the system. A computer program based on a pseudo-Boolean algorithm is used to solve this non-linear integer programming problem.