A Moment Approach to Evaluation and Optimization of Complex System Reliability

A moment approach to evaluating reliability of complex systems is described. The advantages of this type of model have been discussed in relation to models based simply on probability of failure. The main difficulty of a moment approach is the evaluation of first and second moments of the system failure-time distribution. A proposed theorem considerably alleviates this difficulty. In order to facilitate applying this theorem, approximate expressions for the mean and variance of parallel systems are suggested. The problem of system optimization is then discussed and the technique is illustrated with a bridge network. If the system s-expected life remains unchanged when redundancy is added, the variance of system life always decreases.