Robust multi-objective building design optimization approach using NSGA-II algorithm considering weather and occupant uncertainties

Abstract

A novel methodology is presented to evaluate how real-world operational uncertainties impact multi-objective building design optimization process. NSGA-II algorithm is applied across nine distinct operational scenarios, considering diverse weather datasets and energy usage patterns to optimize energy efficiency and thermal comfort in a residential building within India's hot and dry climate. The analysis considers 382,400 possible design options for each scenario. Using (10.5–12.5) % of model evaluations, NSGA-II significantly reduces Annual Cooling Energy Demand (ACED; 46.7%–58.7%) and Cooling Set-point Unmet Hours (CSUH; 70.92%–86.93%) to identify nine Pareto–Optimal solution sets. The final robust optimal design is confirmed by selecting the most repeated parameter settings, representing the highest robustness. Low thermal transmittance walls, double-glazed windows with a 10% window-to-wall ratio (WWR), deep overhangs, and a low solar heat gain coefficient are recognized as robust optimal building components. However, Air Conditioner (AC) sizing lacks robustness in minimizing energy demand across scenarios, requiring case-specific evaluation due to its sensitivity to weather and operational variations. Further, PAWN sensitivity analysis highlighted the substantial impact of WWR (KS_WWR ∼ (0.46–0.66)) and wall type (KS_Wall ∼ (0.31–0.46)) on ACED and CSUH in chosen climate. AC operation is more important for predicting CSUH (KS_AC ∼ 0.35) than ACED (KS_AC ∼ 0.10).

KEYWORDS:

• Robust building design optimization
• NSGA-II algorithm
• operational uncertainties
• PAWN sensitivity analysis

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Department of Science and Technology, Government of India [grant number DST/TMD/UKBEE/2017/17]; Engineering and Physics Research Council EPSRC [grant number EP/R008612/1].