Abstract:Multiphysics problems that are characterized by complex interactions among fluid dynamics, heat transfer, structural mechanics, and electromagnetics, are inherently challenging due to their coupled nature. While experimental data on certain state variables may be available, integrating these data with numerical solvers remains a significant challenge. Physics-informed neural networks (PINNs) have shown promising results in various engineering disciplines, particularly in handling noisy data and solving inverse problems in partial differential equations (PDEs). However, their effectiveness in forecasting nonlinear phenomena in multiphysics regimes, particularly involving turbulence, is yet to be fully established. This study introduces NeuroSEM, a hybrid framework integrating PINNs with the high-fidelity Spectral Element Method (SEM) solver, Nektar++. NeuroSEM leverages the strengths of both PINNs and SEM, providing robust solutions for multiphysics problems. PINNs are trained to assimilate data and model physical phenomena in specific subdomains, which are then integrated into the Nektar++ solver. We demonstrate the efficiency and accuracy of NeuroSEM for thermal convection in cavity flow and flow past a cylinder. We applied NeuroSEM to the Rayleigh-Bénard convection system, including cases with missing thermal boundary conditions and noisy datasets, and to real particle image velocimetry (PIV) data to capture flow patterns characterized by horseshoe vortical structures. The framework's plug-and-play nature facilitates its extension to other multiphysics or multiscale problems. Furthermore, NeuroSEM is optimized for efficient execution on emerging integrated GPU-CPU architectures. This hybrid approach enhances the accuracy and efficiency of simulations, making it a powerful tool for tackling complex engineering challenges in various scientific domains.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively integrate experimental data and numerical simulation methods in multi - physics problems. Specifically, the paper focuses on multi - physics problems with complex interactions among fluid dynamics, heat transfer, structural mechanics and electromagnetics, which are extremely challenging due to their coupling nature. Although experimental data may be available for some state variables, integrating these data with numerical solvers remains a significant challenge. In particular, although existing physics - informed neural networks (PINNs) perform well in handling noisy data and solving inverse problems of partial differential equations (PDEs), their effectiveness in predicting nonlinear phenomena, especially those involving turbulence, has not been fully established. To this end, the paper introduces NeuroSEM, a hybrid framework that combines PINNs and the high - precision spectral element method (SEM) solver Nektar++. NeuroSEM provides a robust method to solve multi - physics problems by leveraging the data assimilation capabilities of PINNs and the high - precision solving capabilities of SEM. Specifically, PINNs are trained to assimilate data in specific sub - domains and model physical phenomena, and then these results are integrated into the Nektar++ solver. The paper demonstrates the efficiency and accuracy of NeuroSEM in thermal convection in cavity flow and flow around a cylinder problems, and it can effectively handle data assimilation problems, especially in sub - domains and state variables where data are available. In addition, the paper also explores the application of NeuroSEM in the Rayleigh - Bénard convection system, including cases with missing thermal boundary conditions and noisy data sets. Finally, the researchers apply NeuroSEM to actual particle image velocimetry (PIV) data to capture flow patterns with horseshoe - vortex structures. The results show that NeuroSEM can accurately model physical phenomena and assimilate data within specified sub - domains. The plug - and - play nature of this framework makes it easy to extend to other multi - physics or multi - scale problems, and it is optimized for emerging integrated GPU - CPU architectures, improving the accuracy and efficiency of simulations and becoming a powerful tool for solving complex engineering challenges.

NeuroSEM: A hybrid framework for simulating multiphysics problems by coupling PINNs and spectral elements

NeuroSEM: A hybrid framework for simulating multiphysics problems by coupling PINNs and spectral elements

A Spatial-Temporally Adaptive PINN Framework for 3D Bi-Ventricular Electrophysiological Simulations and Parameter Inference

PINNs-MPF: A Physics-Informed Neural Network Framework for Multi-Phase-Field Simulation of Interface Dynamics

A mixed formulation for physics-informed neural networks as a potential solver for engineering problems in heterogeneous domains: Comparison with finite element method

NVIDIA SimNet^{TM}: an AI-accelerated multi-physics simulation framework

An Efficient Framework for Solving Forward and Inverse Problems of Nonlinear Partial Differential Equations Via Enhanced Physics-Informed Neural Network Based on Adaptive Learning

Mixed formulation of physics-informed neural networks for thermo-mechanically coupled systems and heterogeneous domains

Physics-Guided, Physics-Informed, and Physics-Encoded Neural Networks and Operators in Scientific Computing: Fluid and Solid Mechanics

A Novel Paradigm in Solving Multiscale Problems

Simulation of multi-species flow and heat transfer using physics-informed neural networks

Interfacing finite elements with deep neural operators for fast multiscale modeling of mechanics problems

Physics-Informed Neural Networks with Complementary Soft and Hard Constraints for Solving Complex Boundary Navier-Stokes Equations

Neural network-augmented differentiable finite element method for boundary value problems

Enhancing Arterial Blood Flow Simulations through Physics-Informed Neural Networks

DMIS: Dynamic Mesh-based Importance Sampling for Training Physics-Informed Neural Networks

Physics-informed neural network simulation of thermal cavity flow

Solving a class of multi-scale elliptic PDEs by means of Fourier-based mixed physics informed neural networks

Physics-informed neural networks (PINNs) for fluid mechanics: A review

From Mesh to Neural Nets: A Multi-Method Evaluation of Physics-Informed Neural Networks and Galerkin Finite Element Method for Solving Nonlinear Convection-Reaction-Diffusion Equations

CoolPINNs: A Physics-informed Neural Network Modeling of Active Cooling in Vascular Systems