Enhancing strategy promoted by large step length for the structure optimization of heat exchanger networks

Abstract

Random Walk algorithm with Compulsive Evolution can simultaneously optimize the continuous (heat load) and integer (structure) variables in heat exchanger networks, by giving a random step length to heat load and setting the minimum heat load. Meanwhile, the structural mutation is allowed, as imperfect solutions may be accepted, in the low probability event that it may avoid entrapment in the local optima. Consequently, the step length influences the efficiency of both continuous and integer variables optimization significantly. Specifically, this study observes the distribution of step length during the evolution process, while analyzing its effects on the structural mutation and optimization. The observation found that a sufficiently large step length can stimulate structure optimization and enhance the efficiency of escaping local optima. On this basis, the study proposes an enhancing strategy promoted by large step length to establish a rational distribution curve of step length during the evolution while also advance the structure optimization of heat exchanger networks. Finally, the analysis of four different cases verifies the validity of the presented enhancing strategy, showing that the proposed approach promoted by large step length facilitates the structure optimization and achieves more cost-effective results, compared to other existing strategies.