A low-storage method consistent with second-order statistics for time-resolved databases of turbulent channel flow up to $Re_\tau=5300$

TL;DR

This paper introduces a low-storage, GPU-accelerated method for creating time-resolved turbulent flow databases that efficiently capture large-scale dynamics up to high Reynolds numbers.

Contribution

A novel low-storage approach that preserves second-order statistics and large-scale flow features while significantly reducing storage and computational costs.

Findings

Enables long-time, large-scale turbulent flow simulations.

Retains full second-order statistics with reduced storage.

Achieves significant speed-up using GPU co-processors.

Abstract

In this paper a novel low-storage method for time-resolved databases is presented. This approach reduces the storage cost of time-resolved databases by storing filtered flow fields that target the large and intermediate scales, while retaining all the information needed to fully reconstruct the flow at the level of filtered flow fields and complete second-order statistics. This is done by storing also the filtered turbulent stresses, allowing to recover the exact effect of the small scales on the large and intermediate scales. A significant speed-up of the computations is achieved, first, by relaxing the numerical resolution, which is shown to affect only the dynamics close to the wall, but not the large scales stored in the database, and, second, by exploiting the \change{computing power and efficiency} of GPU co-processors using a new high-resolution hybrid CUDA-MPI code. This speed-up allows running for physically meaningful times to capture the dynamics of the large scales. The resulting temporally resolved large-scale database of a turbulent channel flow up to $Re_\tau=5300$, in large boxes for long times, is briefly introduced, showing significant indicators of large-scale dynamics with characteristic times of the order of up to eight eddy turnover times.