Steady state detection for computational fluid dynamics

TL;DR

This paper introduces a steady state detection feature in OpenFOAM for fluid dynamics simulations, enabling automatic termination once saturation is reached, thus simplifying large parameter studies.

Contribution

It implements and compares two statistical methods for steady state detection within a CFD library, enhancing simulation automation.

Findings

Both methods effectively identify steady state conditions.

The t-test and f-test have distinct areas of application.

Demonstrated usefulness with a simple test case.

Abstract

Large parameter studies of fluid dynamic instabilities can crucially be simplified, if the user does not need to specify the simulation time. For this purpose, a steady state detection is implemented in the CFD library OpenFOAM. It terminates simulations automatically if characteristic measurements do not change over time, i.e. when saturation is reached. For that purpose, recent data is compared to previous one using two selectable methods. Both calculate a value used as steady state indicator. The first method performing a two sample Student's t-test examines the difference between the means. Similarly, the second method utilises a two sample f-test checking for changes between the variances. Both methods are briefly described, compared and their specific area of application is discussed. Their usefulness is demonstrated with a simple exemplary test case.