Parallel Implementation of Numerical Dissipation Rate and Viscosity Post-processing in Computational Fluid Dynamics

In simulations of Computational Fluid Dynamics (CFD), the numerical dissipation generated from discretization of governing equations can rarely be ignored. Estimating numerical dissipation rate and numerical viscosity is not only pressing for subgrid scale (SGS) modelling in Large Eddy Simulations (LES) and low order methods such as Finite Volume Method (FVM) in commerical codes, but also helpful for higher order methods in Direct Numerical Simulations (DNS) to assess the necessity of mesh refinement in specific regions. Following the recent work of Schranner et al. [1], we develop a parallel algorithm to evaluate numerical dissipation for existing CFD data, which allows to compute numerical dissipation rate and numerical viscosity in physical space for arbitrary sub-domains. The method will be tested for wallbounded turbulent flows, numerical dissipation of this postprocessing tool are compared with results computed directly inside the solver, the self-consistency can be verified using an accurate, pseudo-spectral solver for different mesh resolutions. The information computed by this post-processing step will enable us to determine the accuracy of simulations efficiently without time consuming trial and error validation process. Keywords-numerical dissipation; turbulent flow; parallel computing;