Fake it till you make it: using artificial turbulence to achieve swift converged turbulence statistics in a pressure-driven channel flow

TL;DR

This paper presents a novel, computationally efficient method using artificial turbulence to rapidly achieve statistically stationary conditions in pressure-driven channel flow simulations, significantly reducing simulation time.

Contribution

Introducing a synthetic turbulence generation method that accelerates the attainment of stationary flow conditions compared to traditional approaches.

Findings

Synthetic turbulence reduces spin-up time to 3 eddy turnovers.

Method costs less than 1 CPU hour for initial turbulence generation.

Achieves statistically stationary flow faster than existing methods.

Abstract

In this study, we introduced a simple yet innovative method to trigger turbulence in a channel flow to achieve statistically stationary flow conditions. We compare this new method based on synthetically generated three-dimensional turbulence with two other well-established methods, namely, linear profile superposed with random noise and descending counter-rotating vortices and log-law profile superposed with random noise and descending counter-rotating vortices. We found that synthetically generated three-dimensional turbulence provides a computationally cheap and effective way to reduce simulation spin-up to achieve statistically stationary flow conditions when a precursor turbulent initial condition is not available. At a one-time cost of less than 1 CPU hour to generate the synthetic turbulent initial condition, the flow becomes statistically stationary within 3 eddy turnovers for all the parameters of interest in wall-bounded pressure-driven channel flow simulations when compared to other alternatives that can take more than 10 eddy turnovers resulting in substantial savings in the computational cost.