FEA, CFD Software Supports Artificial Intelligence for Simulation Workflows

The latest release of FEATool Multiphysics version 1.18 from Precise Simulation offers an improved GUI, export and conversion of simulation models, and support for artificial intelligence (AI) simulation workflows. FEATool Multiphysics is an FEA and CFD multi-solver simulation software.

According to the company, the latest release introduces a completely redesigned Graphical User Interface (GUI), making the toolbox more user friendly and easy-to-use. FEATool Multiphysics also features full support and integration with MATLAB and related toolboxes (such as for Optimization, Control Systems, and Machine Learning), and advanced features tailored for the evolving needs of engineers and researchers in industry and academia.

The FEATool toolbox now includes a Multiphysics Application Programming Interface (API), with one-click export functionality and automatic conversion of simulation models to MATLAB and Python script models. This enables users to quickly define and set up simulation models in a fully integrated and easy-to-use GUI, and later export, modify, and programmatically run them automatically for large scale parametric studies, and data generation and collection for Physics-Informed Neural Network (PINN), Machine Learning (ML) and AI simulation models.

Accoding to the company, an example of using FEATool Multiphysics to derive reference data as well as validation for machine learning CFD models can be found in several works on Deep Learning (DL) CFD methodology for flow prediction using with AI and machine learning by Prof. Thi-Thu-Huong Le and coworkers at the at the Blockchain Platform Research Center of the Pusan National University (PNU) in Korea.

Guodong Sa developed a Digital Twin (DT) framework for visualization and design of smart kitchens to facilitate improved kitchen and usability design. In the highlighted work, FEATool Multiphysics was used as a platform for programming and controlling simulation mesh and state-of-the-art CFD flow solvers such as OpenFOAM, and to script, automate, and programmatically generate thousands of sets of simulation data for training the digital twin framework.

And in the medical field, Prof. Zhang H. and colleagues have recently made use of FEATool Multiphysics to develop a method of denoising and improving vascular blood flow imaging for medical diagnosis using a PINN. The data to train the PINN on was automatically generated by simulating the Navier-Stokes equations for different flow conditions and geometries. See their preprint on Fluid Dynamics and Domain Reconstruction from Noisy Flow Images Using Physics-Informed Neural Networks and Quasi-Conformal Mapping for more information.

You can learn more here.

Sources: Press materials received from the company and additional information gleaned from the company’s website.