Physics-constrained Random Forests for Turbulence Model Uncertainty Estimation

TL;DR

This paper introduces a physics-constrained machine learning method using Random Forests to quantify turbulence model uncertainties, providing a data-driven approach that estimates confidence levels even with limited data.

Contribution

It presents a novel physics-constrained Random Forest framework for turbulence uncertainty quantification that requires no user input and offers a priori confidence estimation.

Findings

Effective uncertainty quantification in turbulence models

Data-driven approach reduces reliance on user input

Provides confidence estimates with scarce data

Abstract

To achieve virtual certification for industrial design, quantifying the uncertainties in simulation-driven processes is crucial. We discuss a physics-constrained approach to account for epistemic uncertainty of turbulence models. In order to eliminate user input, we incorporate a data-driven machine learning strategy. In addition to it, our study focuses on developing an a priori estimation of prediction confidence when accurate data is scarce.