Artificial Intelligence-Based Prediction of Building Performance in Kit-of-Parts Prefabricated Residential Buildings

The growing demand for sustainable, affordable, and rapidly deployable housing has increased interest in prefabricated construction systems, particularly kit-of-parts (KoP) residential buildings. This need is especially evident in Australia, where housing shortages and increasing housing demand have intensified the need for faster and more efficient construction methods. KoP systems enable many design configurations through standardized components, offering a promising approach to accelerate housing delivery while maintaining design flexibility. However, the large number of possible design combinations creates challenges for evaluating building performance during the early design stages. Conventional simulation-based approaches for assessing energy efficiency, thermal comfort, and visual comfort are often time-consuming and require substantial modelling effort, limiting their application in fast-paced design processes.

This study presents a conceptual framework for the rapid prediction of building performance in residential KoP prefabricated buildings using artificial intelligence (AI). The framework integrates early-stage design parameters, including spatial configuration, building geometry, envelope characteristics, and climatic conditions, with key performance indicators such as energy demand, thermal comfort, and visual comfort. The framework was developed through a comprehensive review of the literature on prefabricated housing, building performance simulation, and AI applications in the built environment.

The proposed framework establishes the relationships between design attributes, simulation-generated performance data, and AI-based predictive models, enabling near-instantaneous performance assessment during conceptual design. By reducing dependence on computationally intensive simulations, the framework has the potential to support architects, engineers, and decision-makers in identifying high-performing design alternatives at an early stage. The study contributes to the development of AI-assisted design workflows and highlights opportunities for improving the sustainability, energy efficiency, and occupant comfort of future prefabricated residential buildings.