Parameterization of Forced Isotropic Turbulent Flow using Autoencoders and Generative Adversarial Networks

TL;DR

This paper explores using autoencoders and GANs to efficiently generate and parameterize forced isotropic turbulent flows, reducing reliance on traditional CFD simulations by leveraging neural network models trained on statistical flow characteristics.

Contribution

It introduces a neural network-based approach to parameterize turbulent flows, enabling flow generation through learned statistical dependencies without classical CFD methods.

Findings

Neural models can generate turbulent flows with similar statistical properties.

Parameter variations effectively alter flow characteristics.

The approach reduces computational costs compared to traditional CFD simulations.

Abstract

Autoencoders and generative neural network models have recently gained popularity in fluid mechanics due to their spontaneity and low processing time instead of high fidelity CFD simulations. Auto encoders are used as model order reduction tools in applications of fluid mechanics by compressing input high-dimensional data using an encoder to map the input space into a lower-dimensional latent space. Whereas, generative models such as Variational Auto-encoders (VAEs) and Generative Adversarial Networks (GANs) are proving to be effective in generating solutions to chaotic models with high 'randomness' such as turbulent flows. In this study, forced isotropic turbulence flow is generated by parameterizing into some basic statistical characteristics. The models trained on pre-simulated data from dependencies on these characteristics and the flow generation is then affected by varying these parameters. The latent vectors pushed along the generator models like the decoders and generators contain independent entries which can be used to create different outputs with similar properties. The use of neural network-based architecture removes the need for dependency on the classical mesh-based Navier-Stoke equation estimation which is prominent in many CFD softwares.