Direct numerical simulation of grid-induced turbulence
Year: 2021
Type: preprint
Abstract: Although grid-induced turbulence is one of the most important flow configurations in both fundamental and applied research, only few studies have so far tackled the problem by means of high-fidelity numerical simulations where all the grid geometrical features are properly modelled, resembling more directly the experimental framework and better describing both the turbulence generation and decay. In this work, we introduce and validate a numerical procedure, which employs a direct-forcing immersed boundary method and a GPU-based parallel implementation, to perform fully-resolved direct numerical simulations of grid-induced turbulent flows. First, a parametric study is carried out for different grid geometries (i.e., cross-sectional shape and grid solidity) and Reynolds numbers to explore numerically the influence of the upstream conditions on the resulting decay of the turbulent fluctuations. Thus, focusing on one representative case we perform a comparative analysis considering different measures of the relevant quantities (i.e. integral, Taylor and Kolmogorov lengthscales, turbulent Reynolds numbers and normalised energy dissipation rates), which are derived from the various velocity autocorrelation functions that can be defined in the problem. The main differences and similarities between these measures are highlighted, thus providing a complementary insight to understand in more detail the properties of the turbulent flow. Moreover, the numerical results appear in overall good agreement with the available experimental data in similar conditions. Finally, we discuss how the computational framework can be exploited for the investigation of other turbulence-related problems.
Source: arXiv (Cornell University)
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3137219111