Ensemble Kalman method for learning turbulence models from indirect observation data

TL;DR

This paper introduces an ensemble Kalman method to train neural network-based turbulence models from indirect velocity data, avoiding the need for full Reynolds stress data and adjoint RANS solvers, and demonstrates its effectiveness on duct and hill flow cases.

Contribution

First application of ensemble Kalman method to turbulence model learning from indirect data, enabling robust training without adjoint RANS development.

Findings

Successfully learned turbulence models from velocity data in duct flow.

Model generalizes to predict separated flows over periodic hills.

Ensemble Kalman method effectively trains neural network turbulence models.

Abstract

In this work, we propose using an ensemble Kalman method to learn a nonlinear eddy viscosity model, represented as a tensor basis neural network, from velocity data. Data-driven turbulence models have emerged as a promising alternative to traditional models for providing closure mapping from the mean velocities to Reynolds stresses. Most data-driven models in this category need full-field Reynolds stress data for training, which not only places stringent demand on the data generation but also makes the trained model ill-conditioned and lacks robustness. This difficulty can be alleviated by incorporating the Reynolds-averaged Navier-Stokes (RANS) solver in the training process. However, this would necessitate developing adjoint solvers of the RANS model, which requires extra effort in code development and maintenance. Given this difficulty, we present an ensemble Kalman method with an adaptive step size to train a neural network-based turbulence model by using indirect observation data. To our knowledge, this is the first such attempt in turbulence modelling. The ensemble method is first verified on the flow in a square duct, where it correctly learns the underlying turbulence models from velocity data. Then, the generalizability of the learned model is evaluated on a family of separated flows over periodic hills. It is demonstrated that the turbulence model learned in one flow can predict flows in similar configurations with varying slopes.