AI-driven design optimization for sustainable buildings: A systematic review

Abstract

Buildings are major contributors to global carbon emissions, accounting for a substantial portion of energy consumption and environmental impact. This situation presents a critical opportunity for energy conservation through strategic interventions in both building design and operational phases. Artificial Intelligence (AI) has emerged as a transformative approach in this context, enhancing the efficiency and precision of energy management efforts. In the operational phase, AI is extensively utilized as smart controllers for Heating, Ventilation, and Air Conditioning (HVAC) systems and passive energy gains, as well as for fault detection. In the design phase, AI is pivotal as a surrogate model, enabling rapid and accurate evaluation of design options and allowing designers to optimize building performance with minimal computational resources. As the early-stage optimization is more cost-effective than post-construction modifications, design phase optimization has a great potential. Consequently, this paper examines recent advancements in surrogate-assisted design optimization for sustainable buildings, providing a comprehensive overview of the entire optimization process, from data preparation and surrogate model training to final optimization. The review categorizes studies based on experimental approaches and methodologies, identifying trends, gaps, and opportunities in the field. Notably, it highlights how modern AI techniques can incorporate previously unexplored dimensions into surrogate-assisted optimization, broadening the scope and potential of surrogate models. Therefore, this study provides guidance for future research and practical applications of AI-driven strategies in sustainable building practices.