Numerical simulation of fluid flows requires important computational efforts but it is essential in engineering applications. Reduced Order Model (ROM) can be employed whenever fast simulations are required, or in general, whenever a trade-off between computational cost and solution accuracy is a preeminent issue as in process optimization and control. In this work, the efforts have been put to develop a ROM for Computational Fluid Dynamics (CFD) application based on Finite Volume approximation, starting from the results available in turbulent Reynold-Averaged Navier–Stokes simulations in order to enlarge the application field of Proper Orthogonal Decomposition-Reduced Order Model (POD-ROM) technique to more industrial fields. The approach is tested in the classic benchmark of the numerical simulation of the 2D lid-driven cavity. In particular, two simulations at Re = 10^3 and Re = 10^5 have been considered in order to assess both a laminar and a turbulent case. Some quantities have been compared with the Full Order Model in order to assess the performance of the proposed ROM procedure i.e., the kinetic energy of the system and the reconstructed quantities of interest (velocity, pressure and turbulent viscosity). In addition, for the laminar case, the comparison between the ROM steady-state solution and the data available in literature has been presented. The results have turned out to be very satisfactory both for the accuracy and the computational times. As a major outcome, the approach turns out not to be affected by the energy blow up issue characterizing the results obtained by classic turbulent POD-Galerkin methods.
|Titolo:||POD-Galerkin method for finite volume approximation of Navier-Stokes and RANS equations|
|Autori interni:||LORENZI, STEFANO|
|Data di pubblicazione:||2016|
|Rivista:||COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING|
|Appare nelle tipologie:||01.1 Articolo in Rivista|
File in questo prodotto:
|Computer_Methods_in_Applied_Mechanics_and_Engineering_311_(2016)_151-179.pdf||Articolo principale||7.18 MB||Adobe PDF||Post-print||Accesso riservato|