Evolutionary many-objective optimization for retrofit planning in public buildings: A comparative study

There has been an increasing movement toward retrofitting existing (in-use) buildings to achieve a significant reduction in energy consumption and greenhouse gas emissions in the building sector. When planning retrofits for public buildings, decision-makers are required to make rational decisions that will achieve four critical objectives: minimize energy consumption, reduce CO2 emissions, mitigate retrofit costs, and maximize thermal comfort. This study aims to solve this four-objective optimization problem (so-called the problem of many-objective optimization) for retrofit planning in public buildings via an evolutionary many-objective optimization (EO) algorithm that handles these objectives at the same time. This study involves the application of EO algorithms (NSGA-II, MOPSO, MOEA/D, and NSGA-III) and the evaluation of their performance. A description of these algorithms is presented, and each algorithm is implemented in a public-building retrofit project. The algorithms' performances were analyzed, and the results were compared based on two aspects: diversity and convergence. The results indicated that NSGA-III can be used to derive a comprehensive set of trade-off alternatives from possible retrofit scenarios, thereby serving as a useful reference for retrofit planners. These decision-makers can then utilize the provided references to select optimal retrofit strategies and satisfy stakeholders.