<![CDATA[Energy efficiency has become one of the most important aspects in smart building design, especially considering the gap that has increasingly been noted between simulated energy performance and actual consumption. Even though digital design tools like BIM have enhanced design capabilities, there is still a big gap in energy performance, usually rooted in the static nature of traditional simulations. This research tries to respond to this challenge by proposing a conceptual framework linking predictive AI models with BIM for enhanced accuracy in early design stage forecasting. Other than a few studies that revolved around optimization in the post-occupancy phase, this study applies a conceptual-simulative methodology by using a synthetic BIM model of a medium-sized office building. Machine learning algorithms, such as random forest and gradient boosting, were trained on parameterized design data for predicting EUI. Strong predictive consistency was identified with an R² of 0.89 between the predicted and simulated EUI and a conceptual reduction of the performance gap of about 18%. The model also shows robust logical correspondence to the concepts of energy efficiency within a wide range of design scenarios. This research concludes that predictive AI can significantly improve energy performance forecasting in smart building design and provides a proactive data-driven approach toward overcoming the energy efficiency gap in support of more sustainable architectural practices without immediate physical field testing.]]>
Copyrights © 2025
