From hours to milliseconds: accelerating thermal performance with machine learning
DOI:
https://doi.org/10.46421/enarc.v9i1.6895Keywords:
Machine learning, Thermal performance, Energy simulation, Thermal load, Degree-hoursAbstract
This study developed Machine Learning models to predict the thermal performance of buildings, considering Thermal Load and Degree-Hours. Energy simulations were conducted in EnergyPlus to evaluate 35,280 combinations of walls, floors, and roofs in Curitiba, São Paulo, and Belém, representing different bioclimatic zones. The models were trained using XGBoost, validated through 10-fold cross-validation, and assessed using R², MAE, and MAPE. The results demonstrated high accuracy (R² > 0.99, MAPE < 6.1%), enabling rapid predictions in milliseconds. Even with only 2% of the dataset, the model maintained R² above 0.90, making it an efficient alternative to time-consuming energy simulations. This approach enables fast thermal-energy analyses, reduces time and computational costs, and contributes to the development of more sustainable and resilient buildings.
References
ABNT. NBR 15220 - Desempenho térmico em edificações. Associação Brasileira de Normas Técnicas. Rio de Janeiro. 2005.
ABNT. NBR 15220 - Desempenho térmico em edficiações. Associação Brasileira de Normas Técnicas. Rio de Janeiro. 2024.
ABNT. NBR 15575 - Edifícios Residenciais - Desempenho Térmico. Associação Brasileira de Normas Técnicas. Rio de Janeiro. 2024.
CLARKE, J. A. Energy Simulation in Building Design. 2. ed. [S.l.]: Butterworth-Heinemann, 2001.
CRUZ, A. S. et al. Multi-objective optimization based on surrogate models for sustainable building design: A systematic literature review. Building and Environment, v. 266, p. 112147, 2024. https://doi.org/10.1016/j.buildenv.2024.112147
FOROUZANDEH, N. et al. Room energy demand and thermal comfort predictions in early stages of design based on the Machine Learning methods. Intelligent Buildings International, 15, 2023. 3-20. https://doi.org/10.1080/17508975.2022.2049190
GARCÍA, I. K.; MORILÓN, D. G. ANNEXE: An open-source building energy design optimisation framework using artificial neural networks and genetic algorithms. Journal of Cleaner Production, 371, 2022. https://doi.org/10.1016/j.jclepro.2022.133500
IBARRA-VAZQUEZ, G. et al. Predicting open education competency level: A machine learning approach. Heliyon, 9, 2023. https://doi.org/10.1016/j.heliyon.2023.e20597
LAMBERTS, R.; DUTRA, L.; PEREIRA, F. O. R. Eficiência energética na arquitetura. 3ª. ed. Rio de Janeiro: ELETROBRAS/PROCEL, 2014.
MACHADO, R. M. E. S. et al. Metamodel to predict annual cooling thermal load for commercial, services and public buildings: A country-level approach to support energy efficiency regulation. Energy and Buildings, 301, 2023. https://doi.org/10.1016/j.enbuild.2023.113690
MENDES, V. F. et al. Sensitivity analysis of coating mortars according to their specific heat, specific gravity, thermal conductivity, and thickness in contribution to the global thermal performance of buildings. Sustainable Materials and Technologies, v. 31, p. e00381, 2022. https://doi.org/10.1016/j.susmat.2021.e00381
MENDES, V. F. et al. A systematic review of methods for evaluating the thermal performance of buildings through energy simulations. Renewable and Sustainable Energy Reviews, 189, 2024. https://doi.org/10.1016/j.rser.2023.113875
NGO, N.-T. Early predicting cooling loads for energy-efficient design in office buildings by machine learning. Energy e Buildings, 182, 2019. 264-273. https://doi.org/10.1016/j.enbuild.2018.10.004
RACKES, A.; MELO, A. P.; LAMBERTS, R. Naturally comfortable and sustainable: Informed design guidance and performance labeling for passive commercial buildings in hot climates. Applied Energy, 174, 2016. 256-274. https://doi.org/10.1016/j.apenergy.2016.04.081
SEYEDZADEH, S. et al. Tuning machine learning models for prediction of building energy loads. Sustainable Cities and Society, 47, 2019. https://doi.org/10.1016/j.scs.2019.101484
