Daylighting performance prediction tool for early design stages using machine learning
Funding Sponsor
American University in Cairo
Author's Department
Architecture Department
Second Author's Department
Architecture Department
Third Author's Department
Architecture Department
Fourth Author's Department
Architecture Department
Find in your Library
https://doi.org/10.1016/j.jobe.2025.113496
Document Type
Research Article
Publication Title
Journal of Building Engineering
Publication Date
10-1-2025
doi
10.1016/j.jobe.2025.113496
Abstract
This study presents a novel daylighting performance prediction tool that aims at assisting designers in arriving at a range of design options for office spaces. The tool uses a machine learning module for application in different geographic locations. It facilitates rapid and reliable daylighting performance evaluation, particularly for early design phases, while offering user-friendly functionality for non-expert users. The study utilizes a three-stage methodology encompassing data collection from 100 cities, development of a machine learning model using an Artificial Neural Network model (ANN), and deployment of the web-based daylighting prediction tool interface. The ANN model achieved high predictive accuracy with R2 values exceeding 0.90 and MAE below 5 % across most city latitudes. The tool offers comprehensive insights into climate-based all year-round daylight performance metrics. It provides recommendations tailored to specific design scenarios through simulations conducted for various building orientations, window-to-wall ratios, room depths, and heights. The study emphasizes the significance of considering geographical parameters -such as the latitude and clearness index-in daylighting analysis and highlights the tool's potential in enhancing design decision-making processes while being simple-to-use. By empowering architects with actionable recommendations in the preliminary stages of design, the tool contributes to bridging the gap between daylight simulation research and practical design applications, ultimately enhancing the quality and efficiency of architectural design practice.
Recommended Citation
APA Citation
Mashaly, I.
El-Hussainy, M.
Sherif, A.
&
Tarabieh, K.
(2025). Daylighting performance prediction tool for early design stages using machine learning. Journal of Building Engineering, 111,
https://doi.org/10.1016/j.jobe.2025.113496
MLA Citation
Mashaly, Islam, et al.
"Daylighting performance prediction tool for early design stages using machine learning." Journal of Building Engineering, vol. 111, 2025
https://doi.org/10.1016/j.jobe.2025.113496