Accounting for system-level controls during early-stage design

Abstract

Designers of shipboard power systems must consider many tradeoffs when selecting generators in the early-stage setbased design. This is not just an "equipment problem" based on physical design requirements such as limits of size and weight. The operating strategy for generators must be considered when evaluating the goodness of the design. To this end, this paper develops a design approach to impose system-level controls onto generating plants to evaluate their optimality. Two critical design metrics-fuel consumption and quality of service (QOS) are defined to incorporate these control variables, and cooptimized to reflect mission-oriented system performance. The minimum fuel consumption is computed based on a new concept of economic dispatch algorithm, which takes active and reactive power balance, and system redundancy requirements into account. QOS metric is defined to reflect system-level control strategies based on the classical concept of mean-time-betweenfailure, which is defined to depend inversely on the power produced. We develop a multi-objective particle swarm optimization to effectively locate the Pareto fronts of system performance for all design alternatives over a whole given mission. Thus a set of optimal design alternatives can be rapidly selected to represent the non-dominated optimal compromises of the two metrics. This design approach is demonstrated and compared with a single-objective design approach in a design example of shipboard generating plant design. And more optimal alternatives are able to be found.I.