Project DNN-CFD - Deep Neural Networks for CFD Simulations

Project partners

• Complex Phenomena Unified Simulation Research Team, RIKEN-CSS, Japan
• Joint Laboratory for Extreme Scale Computing (JLESC)

Project description

Simulation-based optimization in engineering requires high computational power and may benefit from system heterogeneity. Highly scalable computational fluid dynamics (CFD) solvers have been developed at RIKEN-CCS and JSC to conduct large-scale simulations. RIKEN-CCS focuses on full-scale simulations of vehicles aerodynamics. JSC investigates respiratory flows to improve medical treatments. Due to massive amounts of data produced, the analysis of the results is becoming a time-consuming process. Data-driven approaches have the potential to speed up analyses or to even replace certain physics in simulations. Deep learning methods have proven to be a fast alternative to extract multi-scale features from high-dimensional data sets. Deep neural networks (DNN) are able to predict the steady-state flow and aerodynamic coefficients around bluff bodies and airfoils, and unsteady laminar vortex shedding over circular bodies. Autoencoders (AE) or generative models, like variational AEs or generative adversarial networks (GANs), have shown great potential in predciting 2D and 3D flow fields.

It is the aim of the proposed cooperation to tie in with these topics and to develop methods to efficiently compute and analyze complex optimization setups in engineering utilizing heterogeneous architectures. Physics-informed trained networks that allow for transfer learning are developed. That is, learning starts with pre-trained networks trained on data with similar features. It is investigated (i) how such such pre-trained DNNs adapt to the various flow configurations of interest for RIKEN-CCS and JSC, (ii) how they can speed up the simulation workflow, and (iii) how they overcome a shortage of training data.

Bilateral support activities will lead to knowledge exchange with respect to the different hardware available at the partners’ sites, in CFD methods, and in deep learning approaches. Hence, expertise of the centers in these fields are strongly promoted in the course of this project. To foster the cooperation, mutual short-time stays of the involved scientists are planned.

Links

Project website at JLESC: DNN-CFD