Loading...
Mesh Adaptation for Large Eddy Simulation with the FR/CPR Method
Ims, Jeremy Hopper
Ims, Jeremy Hopper
Citations
Altmetric:
Abstract
The computational mesh for a Computational Fluid Dynamics (CFD) simulation must provide sufficient cell density at proper locations in the domain in order to resolve the flow physics impactful to the targeted engineering parameters. The key locations for resolution are often imprecisely known and so must be found by trial and error. This dissertation discusses CFD mesh adaptation—the computer guided adjustment of the mesh in response to the simulation. Specifically, this document addresses mesh adaptation for Large Eddy Simulation (LES) with the Flux Reconstruction / Correction Procedure via Reconstruction (FR/CPR) method. It presents a computer program to adaptively refine 3D unstructured hexahedral meshes, guided by the distribution of error within the flow field, estimated by an error indicator algorithm integrated into the flow solver. Furthermore, it introduces four error indicator algorithms for tracking the location and the amount of under-resolution in turbulent flow fields. The error estimators are derived, mathematically analyzed, and numerically tested upon two well-known benchmark LES simulations. The analysis leads to a performance evaluation of the error indicators, judged on their ability to drive the CFD simulation toward truth. The four error indicators are: 1) The Unsteady Residual Indicator (unStdE), based on the unsteady residual from the FR/CPR calculation, 2) The Smoothness Indicator (smthE), based on a local smoothness indicator, 3) The adapted Toosi-Larson Indicator (T.L.errE), based on the estimated small scale turbulent kinetic energy, and 4) The Average Toosi-Larson Indicator (avgT.L.errE), conceptually the same as T.L.errE but formulated to be less costly to compute. Upon LES simulations that model transitional flow past the T106 low pressure turbine blade, all of the error indicators demonstrate ability to boost resolution of the flow field, improving simulation accuracy of force coefficients, vortex structure, Reynolds stresses, and the energy spectra. unStdE and T.L.errE are found to be the fastest to bring improvement to coarse-meshed simulations. Of the two, unStdE is mathematically simpler and easier to compute.
Description
Date
2022-12-31
Journal Title
Journal ISSN
Volume Title
Publisher
University of Kansas
Collections
Research Projects
Organizational Units
Journal Issue
Keywords
Computational physics, Aerospace engineering, Fluid mechanics, Computational Fluid Dynamics, FR/CPR, High-Order Method, Large Eddy Simulation, Mesh Adaptation, Turbulent Error Estimation