Super-resolution of low-fidelity flow solutions via generative adversarial networks. (August 2022)

• Record Type: Journal Article
• Authors: Pourbagian, Mahdi; Ashrafizadeh, Ali
• Is Part Of: Simulation. Volume 98:Number 8(2022)
• Page Start: 645
• Page End: 663
• Publication Date: 2022-08
• Subjects: Fluid flow simulation -- low and high-fidelity CFD solution -- deep learning -- generative adversarial network; Computer simulation -- Periodicals; 003.3
• DOI: 10.1177/00375497211061260
• Languages: English
• ISSNs: 0037-5497

Abstract

While computational fluid dynamics (CFD) can solve a wide variety of fluid flow problems, accurate CFD simulations require significant computational resources and time. We propose a general method for super-resolution of low-fidelity flow simulations using deep learning. The approach is based on a conditional generative adversarial network (GAN) with inexpensive, low-fidelity solutions as inputs and high-fidelity simulations as outputs. The details, including the flexible structure, unique loss functions, and handling strategies, are thoroughly discussed, and the methodology is demonstrated using numerical simulations of incompressible flows. The distinction between low- and high-fidelity solutions is made in terms of discretization and physical modeling errors. Numerical experiments demonstrate that the approach is capable of accurately forecasting high-fidelity simulations.