ECCOMAS 2024

Uncertainty Quantification for Lattice Boltzmann Method on Large Eddy Simulations based Turbulent Channel Flow

  • Xue, Xiao (University College London)
  • McCullough, Jon (University College London)
  • Coveney, Peter (University College London)

Please login to view abstract download link

Large Eddy Simulations (LES) play an essential role in accurately simulating aero/hydrodynamic applications in industrial settings. The Lattice Boltzmann Method (LBM), known for its parallelization friendly nature, has become increasingly popular for complex flow simulations. In high Reynolds number flows, LES approximates effective viscosity on top of turbulent viscosity to simulate turbulent flows, offering computational efficiency over direct numerical simulation. This study focuses on uncertainty quantification for wall resolved LES using LBM in turbulent channel flow at a friction Reynolds number Reτ =180. We employ EasyVVUQ, a tool for verification, validation, and uncertainty quantification, to deploy uncertainty quantification campaign. Our investigation encompasses spatial uncertainties in the channel flow by varying streamwise and spanwise configurations, along with temporal uncertainties through alterations in the statistical start point and sample duration. Critically, we examine uncertainties introduced by the choice of the Smagorinsky constant in LBM simulations. This investigation enhances the understanding of how parameter choices influence turbulent flows. Furthermore, our findings provide valuable insights into parameter selection for LES-based LBM which is also beneficial to a wide range of high-Reynolds number practical applications.