Extract

Recently, there have been a number of researches on building design optimization by coupling multi-objective genetic algorithm with building simulation. The researches offered dozens of potential designs solutions as outcomes. However, little attention has been paid on the post-optimization process and how to use the optimization outcomes to facilitate the building designers and engineers to find near optimal solutions quickly and confidently. The GA-MLR-MCA approach presented in this study combined building simulation with a multi-objective Genetic Algorithm (NSGA-II) for optimization of thermal comfort and energy consumption for a typical residential house in five different cities across all the climatic regions in China. Results of the potential solutions based on the Pareto Front were then trained with multi-linear regression (MLR) models considering variables such as window-to-wall ratios, building orientation, heating air temperature setpoint, cooling air temperature setpoint, external wall insulation, roof insulation, and HVAC type. Typical R-square values for the MLR models both for thermal comfort and energy consumption were higher than 0.95. Monte Carlo approach was also applied to generate same amount of solutions and came out with results very close to the Pareto Front solutions.

PDF

Download