Multi-objective design optimization of building energy retrofit using building energy simulation surrogate model

Abstract The urgent need to decarbonize existing buildings has led to an increased focus on energy retro6its as a crucial strategy for improving building performance and sustainability. While building energy simulation software offers valuable insights into the effects of various retro6it scenarios, determining the optimal energy retro6it solution remains a signi6icant challenge due to the prohibitive computational costs associated with simulation-based optimization. This study presents an alternative approach to multi-objective design optimization of building energy retro6its involving a building energy simulation surrogate model. A case study of a religious building in Metro Manila, Philippines was used to demonstrate the proposed methodology. The methodology begins with creating a comprehensive building energy simulation database using Latin Hypercube Sampling, followed by training regression models for energy consumption and thermal comfort using this database. Next, these regression models were coupled with a Genetic Algorithm to perform multi-objective optimization. Finally, ranking of solutions in the Pareto front was demonstrated using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). Optimization results were validated with the resulting accuracy and reliability of the surrogate model-based approach being within acceptable limits. The 6indings of this study suggest that regression surrogate models offer a computationally ef6icient and effective means of optimizing building energy retro6its. By providing a practical framework for multi-objective optimization, this research contributes to the advancement of sustainable building practices and supports the broader goal of decarbonizing the built environment.