Exploring Cross-Project Building Performance Prediction: Creation and Implementation of a Zone-Based Surrogate Model for Building Performance Optimization

Sustainable architectural design is increasingly recognized as an important aspect of building construction. However, not all architectural offices can readily achieve sustainability objectives in their projects, due to various constraints such as the lack of accessible tools, systematic approaches, time, and expertise. The described approach aims to contribute to the field of building performance optimization by developing a machine learning-based surrogate modeling framework that can be used for a wide range of projects without the need for project-specific simulations or model training. The proposed methodology utilizes a thermal zone-based dataset to train an artificial neural network-based surrogate model, which is integrated into a workflow that combines the model-to-data conversion algorithm and optimization algorithm. The approach allows architects and engineers to quickly assess the impact of different design parameters on building performance without the need for extensive simulations. A case study of a residential building in Tel Aviv, Israel demonstrated the effectiveness of the approach, with near-optimal solutions obtained in a shorter time than with simulation-based optimization.