This paper is concerned with a parallel system that sustains a time-independent load and consists of n components with exponential lifetimes. It is assumed that the total load is shared by the working components and the failures of components produce higher failure rates in the surviving components according to the relationship between the load and the failure rates. The power rule model among several load–failure rate relationships is considered. We consider the system efficiency measure as the expected profit earned by the system per unit time. The high load causes high gain but it also incurs frequent system failures. The expected profit per unit time is used as the criterion for evaluating the system efficiency. The goal of a system engineer is to determine the optimal load and identify redundant units, so maximizing the expected profit per unit time. First, the system reliability function is obtained and the optimization problem of the load-sharing parallel system is considered. Given the redundant units, the existence of the optimal load can be proved analytically and given the load, the optimal redundant units can be recognised (also analytically). The optimal load and redundant units are obtained simultaneously by numerical computation. Some numerical examples are studied.
