Deep learning-assisted prediction of viscoplastic flow in superhydrophobic channels
Physics of Fluids · 01.05.2025 · 37 (5)
DOI: 10.1063/5.0268110 ↗Kurz gefasst
Fachartikel in der Zeitschrift Physics of Fluids, online erschienen am 12.05.2025. Verfasst von A. Joulaei und 3 weiteren Autor:innen. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 121 Literaturangaben. Gefördert von: Canada Research Chairs, Natural Sciences and Engineering Research Council of Canada, Canada Foundation for Innovation.
Abstract (Originalsprache)
Viscoplastic fluid flows in channels with asymmetric superhydrophobic (SH) walls are analyzed using computational fluid dynamics simulations, while a machine learning framework is developed to enhance computational efficiency and accuracy. A deep learning framework, based on a convolutional neural network (CNN), is trained on numerical simulation data to reconstruct velocity fields and identify yield surfaces. Additional models, including multiple linear regression (MLR), support vector machine (SVM), extreme learning machine (ELM), and adaptive neuro-fuzzy inference system (ANFIS), estimate key flow variables such as effective slip length and normalized plug area, with their predictions compared to CNN results, using standard performance metrics. Our results reveal that the CNN demonstrates strong predictive accuracy, closely matching the performance of ELM and ANFIS while significantly outperforming MLR and SVM. The trained CNN also enables the optimization of SH wall characteristics, capturing the non-monotonic effects of SH wall groove misalignment (ε) and slip area fraction (φ) on velocity asymmetry. Optimal misalignment shifts from ε=0.25 at φ1=φ2=0.5 to ε≈0.35 for φ1=φ2=0.1 or φ1=φ2=0.9, where subscripts 1 and 2 refer to the lower and upper SH walls, respectively. In addition, CNN results reveal that the optimal slip area fraction φ2=0.5 remains unchanged for smaller upper wall groove periodicity lengths (ℓ2), but decreases as ℓ2 increases, leading to a ∼10% reduction in flow velocity asymmetry.
Förderung
- Canada Research Chairs
- Natural Sciences and Engineering Research Council of Canada
- Canada Foundation for Innovation
Zitieren
APA
Joulaei, A., Pal, A., Rahmani, H., & Taghavi, S. M. (2025). Deep learning-assisted prediction of viscoplastic flow in superhydrophobic channels. Physics of Fluids, 37(5). https://doi.org/10.1063/5.0268110
Vancouver
Joulaei A, Pal A, Rahmani H, Taghavi SM. Deep learning-assisted prediction of viscoplastic flow in superhydrophobic channels. Physics of Fluids. 2025;37(5). doi:10.1063/5.0268110
BibTeX
@article{Joulaei2025,
author = {Joulaei, A. and Pal, A. and Rahmani, H. and Taghavi, S. M.},
title = {Deep learning-assisted prediction of viscoplastic flow in superhydrophobic channels},
journal = {Physics of Fluids},
year = {2025},
volume = {37},
number = {5},
doi = {10.1063/5.0268110}
}
EUVIMED