Comparison of metaheuristic algorithms on robust design optimization of a plain-fin-tube heat exchanger

Robust optimization methodologies become necessary due to the growing demand for better and more realistic optimization design outcomes. Optimizers are crucial parts of these methodologies, comprised of a search and selection strategy in order to locate the global optimum. Particularly, metaheuristic optimizers are well-known for their capabilities on identifying global optimum solutions in rather complex landscapes, while their field is actively researched and a handful of new optimizers are developed. To bridge the gap between robust optimization and new metaheuristics algorithms, several optimizers, such as cuckoo search and firefly algorithm, are investigated and compared in terms of their performance and optimal design solutions. The study focuses on the robust optimization design of a plain-fin-tube heat exchanger while deterministic results are also obtained for the sake of comparison. Furthermore, these two optimizers are compared to the more well-known genetic algorithm and particle swarm optimization in order to establish a baseline for their performance. The optimizers used, are coupled with polynomial chaos to propagate effectively the defined uncertainties. The results show that each optimization scheme is capable of delivering well-formed Pareto fronts. The robust design solutions are similar for the various optimizers, however differences are identified in the diversity of the optimal solutions and the convergence speed of the optimizers. Multi-objective cuckoo search optimization is considered as the most performing optimizer due to the improved optimal Pareto front obtained and its fast convergence.