Performance evaluation of population-based metaheuristic algorithms and decision-making for multi-objective optimization of building design

Optimization algorithms and decision-making techniques are major components of multi-objective optimization. This study evaluated the performance of population-based metaheuristic algorithms and decision-making techniques in optimizing an unconventional building design - a lift-up design - to maximize the areas with wind and thermal comfort in a 'hot' and 'calm' climate. Four optimization algorithms (GA, PSO, GSA, FA) and three decision-making techniques (LINMAP, TOPSIS, Shannon Entropy) were employed to optimize the lift-up design. The effectiveness and efficiency of algorithms in optimization were measured using six metrics. The evaluation revealed a steady improvement of algorithms' performance as population and number of iterations increased up to the convergence at about 6000 evaluations without excessively increasing computational time. Although no algorithm scored best across all metrics, PSO was superior in many aspects. For the algorithms, the three decision-making techniques chose similar optimum designs with slight differences in a few design parameters. The optimum solution of multi-objective optimization was a better trade-off solution for the two objective functions than that of single-objective optimization. The study recommends conducting convergence tests using the performance metrics before optimization to decide a suitable population size and number of iterations for population-based metaheuristic optimization algorithms.