A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case
Апстракт
Computational fluid dynamics (CFD) represents an attractive tool for estimating wind pressures and wind loads on high-rise buildings. The CFD analyses can be conducted either by low-fidelity simulations (RANS) or by high-fidelity ones (LES). The low-fidelity model can efficiently estimate wind pressures over a large range of wind directions, but it generally lacks accuracy. On the other hand, the high-fidelity model generally exhibits satisfactory accuracy, yet, the high computational cost can limit the number of approaching wind angles that can be considered. In order to take advantage of the main benefits of these two CFD approaches, a multi-fidelity machine learning framework is investigated that aims to ensure the simulation accuracy while maintaining the computational efficiency. The study shows that the accurate prediction of distributions of mean and rms pressure over a high-rise building for the entire wind rose can be obtained by utilizing only 3 LES-related wind directions. T...he artificial neural network is shown to perform best among considered machine learning models. Moreover, hyperparameter optimization significantly improves the model predictions, increasing the 𝑅��2 value in the case of rms pressure by 60%. Dominant and ineffective features are determined that provide a route to solve a similar application more effectively.
Кључне речи:
Wind loading / Machine learning / LES / RANSИзвор:
Building and Environment, 2023, 234Издавач:
- Elsevier
Финансирање / пројекти:
- "Fonds National de la Recherche, Luxembourg" (FNR), CORE Junior project DATA4WIND - "Data-Driven Approach for Urban Wind Energy Harvesting", C19/SR/13639741
Колекције
Институција/група
GraFarTY - JOUR AU - Šarkić Glumac, Anina AU - Jadhav, Onkar AU - Despotović, Vladimir AU - Blocken, Bert AU - Bordas, Stephane PA PY - 2023 UR - https://grafar.grf.bg.ac.rs/handle/123456789/3292 AB - Computational fluid dynamics (CFD) represents an attractive tool for estimating wind pressures and wind loads on high-rise buildings. The CFD analyses can be conducted either by low-fidelity simulations (RANS) or by high-fidelity ones (LES). The low-fidelity model can efficiently estimate wind pressures over a large range of wind directions, but it generally lacks accuracy. On the other hand, the high-fidelity model generally exhibits satisfactory accuracy, yet, the high computational cost can limit the number of approaching wind angles that can be considered. In order to take advantage of the main benefits of these two CFD approaches, a multi-fidelity machine learning framework is investigated that aims to ensure the simulation accuracy while maintaining the computational efficiency. The study shows that the accurate prediction of distributions of mean and rms pressure over a high-rise building for the entire wind rose can be obtained by utilizing only 3 LES-related wind directions. The artificial neural network is shown to perform best among considered machine learning models. Moreover, hyperparameter optimization significantly improves the model predictions, increasing the 𝑅��2 value in the case of rms pressure by 60%. Dominant and ineffective features are determined that provide a route to solve a similar application more effectively. PB - Elsevier T2 - Building and Environment T1 - A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case VL - 234 DO - 10.1016/j.buildenv.2023.110135 ER -
@article{ author = "Šarkić Glumac, Anina and Jadhav, Onkar and Despotović, Vladimir and Blocken, Bert and Bordas, Stephane PA", year = "2023", abstract = "Computational fluid dynamics (CFD) represents an attractive tool for estimating wind pressures and wind loads on high-rise buildings. The CFD analyses can be conducted either by low-fidelity simulations (RANS) or by high-fidelity ones (LES). The low-fidelity model can efficiently estimate wind pressures over a large range of wind directions, but it generally lacks accuracy. On the other hand, the high-fidelity model generally exhibits satisfactory accuracy, yet, the high computational cost can limit the number of approaching wind angles that can be considered. In order to take advantage of the main benefits of these two CFD approaches, a multi-fidelity machine learning framework is investigated that aims to ensure the simulation accuracy while maintaining the computational efficiency. The study shows that the accurate prediction of distributions of mean and rms pressure over a high-rise building for the entire wind rose can be obtained by utilizing only 3 LES-related wind directions. The artificial neural network is shown to perform best among considered machine learning models. Moreover, hyperparameter optimization significantly improves the model predictions, increasing the 𝑅��2 value in the case of rms pressure by 60%. Dominant and ineffective features are determined that provide a route to solve a similar application more effectively.", publisher = "Elsevier", journal = "Building and Environment", title = "A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case", volume = "234", doi = "10.1016/j.buildenv.2023.110135" }
Šarkić Glumac, A., Jadhav, O., Despotović, V., Blocken, B.,& Bordas, S. P.. (2023). A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case. in Building and Environment Elsevier., 234. https://doi.org/10.1016/j.buildenv.2023.110135
Šarkić Glumac A, Jadhav O, Despotović V, Blocken B, Bordas SP. A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case. in Building and Environment. 2023;234. doi:10.1016/j.buildenv.2023.110135 .
Šarkić Glumac, Anina, Jadhav, Onkar, Despotović, Vladimir, Blocken, Bert, Bordas, Stephane PA, "A multi-fidelity wind surface pressure assessment via machine learning: A high-rise building case" in Building and Environment, 234 (2023), https://doi.org/10.1016/j.buildenv.2023.110135 . .