Performance assessment of RBFMOpt, NSGA2, and MHACO on the thermal and energy optimization of an office building

Simulation-based Optimization processes (SBO) can be valuable methods in searching for efficient buildings. This study evaluates the performance of the multi-objective algorithms RBFMOpt, NSGA2, and MHACO facing the same SBO problem. The goal is to maximize thermal comfort while minimizing the energy consumption with HVAC systems for a typical Brazilian office building. We proposed a scoring method based on four algorithms’ performance metrics: hypervolume, variability, IGD+, and coverage. We also applied a Kruskal–Wallis test to determine whether the SBO process needs multiple runs to obtain the average performance of each algorithm. The results show that RBFMOpt presents the best performance, reaching a higher score, especially in situations with low budgets for the simulation and optimization process. The results also pointed out that the number of cycles for RBFMOpt impacts directly the quality of solutions, and a higher number of cycles provided better results.