Multi-objective optimization of building energy performance and indoor thermal comfort: A new method using artificial bee colony (ABC)

Abstract

The aim of this paper is to present a powerful simulation-based multi-objective optimization of building energy efficiency and indoor thermal comfort to obtain the optimal solutions of the comfort-energy efficient configurations of building envelope. The optimization method is developed by integrating a multi-objective artificial bee colony (MOABC) optimization algorithm implemented in MATLAB with EnergyPlus building energy simulation tool. The proposed optimization approach is applied to a single office room; and the building parameters, including the room rotation, window size, cooling and heating setpoint temperatures, glazing and wall material properties are considered as decision variables. In the present study, single-objective and multi-objective optimization analyses of the total annual building electricity consumption and the Predicted Percentage of Dissatisfied (PPD) are investigated to bring down the total energy cost as well as the thermal discomfort in four major climate regions of Iran, i.e. temperate, warm-dry, warm-humid and cold ones. In the results part, the achieved optimal solutions are presented in the form of Pareto fronts to reveal the mutual impacts of variables on the building energy consumption and the thermal discomfort. Finally, the ultimate optimum solution on the Pareto fronts are selected by TOPSIS decision-making method and the results of double-objective minimization problem are compared with the single-objective ones as well as the base design. The results of double-objective optimization problem indicate that in different climates, even though the total building electricity consumption increases a bit about 2.9–11.3%, the PPD significantly decreases about 49.1–56.8% compared to the baseline model. In addition, the comparisons of single-objective and double-objective optimization approaches clearly show that multi-objective optimization methods yield more appropriate results respect to the single ones, mainly because of the lower deviation index value from the ideal solution.