Coordinated robust optimization of building and surrounding microclimate in early-stage design under uncertainty

In early-stage building design, energy efficiency and outdoor thermal environment constitute two fundamental concerns for sustainability-driven optimization. However, many uncertainties affect building performance and the surrounding microclimate, making the quantification and robust optimization more challenging. In this study, a coordinated robust optimization method is proposed for building and the surrounding microclimate in early-stage design, efficiently handling their uncertainties and mutual impacts. Major uncertainties affecting building performance in subtropical regions are identified through a comprehensive sensitivity analysis. A LightGBM-based surrogate model is developed for building energy performance quantification under uncertainty, which greatly improves the computational efficiency. Alternative objective functions are analyzed and compared to achieve the optimal energy and environmental performance. The proposed method is validated through a case study in Hong Kong. The results show the necessity of considering uncertainties in coordinated building–microclimate optimization to achieve robust and energy-efficient envelope design solutions. Neglecting the uncertainties could result in an energy waste of over 20% in subtropical climates. Results also show that the surrogate-assisted method enables coordinated robust design optimization within a practically feasible timeframe, reducing the computation time from 4.9 × 106 h to just 2.68 h, compared to a conventional simulation-based approach.