Determining quality of LES and DES by computing turbulent kinetic energy

    • FoxCFD

      Hello everyone,

      I'm doing simulations with WMLES S-Omega and SBES - WMLES S-Omega. According to the literature, to have a "good" LES simulation we should ensure that we're resolving 80% of the turbulent kinetic energy and 20% is being modeled by the subgrid-scale model.

      Computing Turbulent Kinetic Energy in WMLES

      This can be checked by computing the turbulent kinetic energy being resolved by the mesh and by the subgrid-scale model. Let's call k_res and k_sgs.

      Assuming k_res = 0.5 * (U_RMSE ^2 + V_RMSE ^2 + W_RMSE ^2 )

      Question 1: How do we compute k_sgs in the WMLES model ?

      Is there a formula to compute easily k_sgs using the subgrid quantities available by Fluent like

      • Subgrid Turbulent Viscosity
      • Subgrid Effective Viscosity
      • Subgrid Turbulent Viscosity Ratio
      • Subgrid Filter Length (is this the length computed by the equation 4 in the referenced paper?)

      After this, the total turbulent kinetic energy can be computed by k_total = k_res + k_sgs

      Now to assess the quality of the simulation we could compute the ratio k_res / k_total and check if it is more than 80% or not.

      Computing Turbulent Kinetic Energy in SBES

      Furthermore I wish to do the same for the SBES simulation. However, here we're also dealing with turbulence quantities from RANS.

      We have the turbulent kinetic energy directly available from it (let's call it k_RANS). I suspect this is from the areas where RANS is activated because k_RANS is very small compared to the k_res earlier computed.

      Question 2: Therefore, after computing k_sgs can we compute k_total as k_total = k_res + k_sgs+ k_RANS ?

      The referenced paper for the WMLES model is A hybrid RANS-LES approach with delayed-DES and wall-modelled LES capabilities

      Thank you for reading 😀

    • aitor.amatriain
      The calculation of the LES quality is a beta feature in the 2022R1 version:
      10.1. Estimation of the LES Resolution Quality (ansys.com)
      The characteristic length of the transition zones in SBES is negligible with respect to the characteristic length of the domain. In addition, these transition zones have no ¨physical¨ sense, so you can omit them. The LES part is treated in the same way as the pure LES.
Viewing 1 reply thread
  • You must be logged in to reply to this topic.