THE DESIGN OF THE I~2S-LWR STEAM GENERATION SYSTEM USING MULTI-OBJECTIVE OPTIMIZATION SCHEMES

摘要

Multi-objective optimization is an advantageous tool used in many fields of research and industry. It has been successfully applied to engineering design, scientific experimentation and business decision-making. However, so far it has seen minimal utilization in the development of nuclear power plant design. Multi-objective optimization applied to product design involves manipulating key design parameters in order to develop a globally optimal design(s) in the face of differing and possibly opposing objectives. This paper introduces the use of multi-objective optimization in the development of a steam generation system (SGS)for the Integral, Inherently Safe Light Water Reactor (I~2S-LWR) nuclear power plant, including complex versions of genetic algorithms (GA) and particle swarm optimization (PSO). The best solutions from each of these algorithms form a Pareto front, from which an ideal system or component can be developed. Each potential SGS concept is subject to its own constraints and therefore merits its own population, called a cluster in this paper, during the multi-objective optimization. Three different multi-objective optimization schemes were tested and compared against the first three Zitzler-Deb-Thiele (ZDT) functions. The optimization method that performed the best was a version of Non-dominated Sorting Particle Swarm Optimization (NSPSO). The NSPSO method was then used as the basis for developing a cost-efficiency Pareto front of optimized design of the SGS utilized by the PS-LWR currently being developed. This design optimization resulted in a Pareto front of different viable power cycle designs with varying efficiencies and costs that all meet the specific constraints of the I~2S-LWR concept. This design optimization method can be applied to any variation of special nuclear power plant constraints, resulting in a number of possible options of varying complexity, efficiency and cost. A brief discussion on the considerations and limitations of such applications of the NSPSO to other systems and components is included at the conclusion of this paper.