Physics-Informed Neural Networks for Low Reynolds Number Flows over Cylinder
Open Access
Abstract
Physics-informed neural network (PINN) architectures are recent developments that can act as surrogate models for fluid dynamics in order to reduce computational costs. PINNs make use of deep neural networks, where the Navier-Stokes equation and freestream boundary conditions are used as losses of the neural network; hence, no simulation or experimental data in the training of the PINN is required. Here, the formulation of PINN for fluid dynamics is demonstrated and critical factors influencing the PINN design are discussed through a low Reynolds number flow over a cylinder. The PINN architecture showed the greatest improvement to the accuracy of results from the increase in the number of layers, followed by the increase in the number of points in the point cloud. Increasing the number of nodes per hidden layer brings about the smallest improvement in performance. In general, PINN is much more efficient than computational fluid dynamics (CFD) in terms of memory resource usage, with PINN requiring 5–10 times less memory. The tradeoff for this advantage is that it requires longer computational time, with PINN requiring approximately 3 times more than that of CFD. In essence, this paper demonstrates the direct formulation of PINN without the need for data, alongside hyperparameter design and comparison of computational requirements.Keywords
Funding Information
- Nanyang Technological University, Singapore
- Singapore Centre
- National Research Foundation, Prime Minister’s Office, Singapore
This publication has 27 references indexed in Scilit:
- Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equationsJournal of Computational Physics, 2019
- Deep learning of vortex-induced vibrationsJournal of Fluid Mechanics, 2018
- Application of Convolutional Neural Network to Predict Airfoil Lift Coefficient2018
- Reduced-order modelling of non-linear, transient aerodynamics of the HIRENASD wingThe Aeronautical Journal, 2016
- The Finite Volume Method in Computational Fluid Dynamics2016
- Computational aerodynamics of low Reynolds number plunging, pitching and flexible wings for MAV applicationsActa Mechanica Sinica, 2008
- A systematic approach to the numerical calculation of fundamental quantities of the two-dimensional flow over a circular cylinderJournal of Fluids and Structures, 2007
- Approximation capabilities of multilayer feedforward networksNeural Networks, 1991
- Approximation by superpositions of a sigmoidal functionMathematics of Control, Signals, and Systems, 1989
- Experiments on the flow past a circular cylinder at low Reynolds numbersJournal of Fluid Mechanics, 1959