Integrating Machine Learning with Parametric Modeling Environments to Predict Building Daylighting Performance

Abstract

Buildings account for 30% of global energy consumption. To support the development of energy-efficient built environments and cities, architects, urban planners, and engineers have begun to utilize building performance simulation (BPS). Supporting decision-making and steering design toward high performance are crucial in the early design phase, when decisions have the biggest impact on the final product’s energy consumption and costs. Architects often assess the energy performance of hundreds of different design solutions and configurations in parametric environments to find the most energy-efficient solution. To adhere to strict project due dates, architects usually shorten the time required for BPS by reducing the number of variables or parameters for the building and facade design. This practice usually results in the elimination of design parameters that could potentially contribute to an energy-optimized design configuration. To achieve time-efficient BPS processes, recent research has focused on developing machine learning (ML) algorithms that can instantly predict the energy performance of buildings. Although ML has demonstrated high levels of accuracy, ML techniques are difficult to implement because they require technical knowledge of programming languages. Furthermore, these ML algorithms were developed as standalone applications, making them inappropriate for geometric modeling environments. This paper introduces a new ML tool that can be easily accessed within Grasshopper, a parametric modeling environment. The proposed tool can instantly predict the daylighting performance of building configurations.