Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Reducing initial computational cost of complex turbulent flows simulations by using recorded data from a existent simulation
Submission Author:
Johnatas Freitas , MG
Co-Authors:
Johnatas Freitas, João Marcelo Vedovotto, ricardo serfaty, Millena Vale, Flavio de Oliveira Silva
Presenter: Johnatas Freitas
doi://10.26678/ABCM.COBEM2023.COB2023-0819
Abstract
In the computational simulation of complex turbulent flows, for instance, involving fluid-structure interactions and/or turbulent combustion with detailed chemistry, the computational cost can be cumbersome, even using massively distributed computing. Normally these detailed simulations are transient flows and as initial conditions, they retain some kind of unperturbed uniform flow, hence, one must wait until the simulation reaches a state of statistical stability in terms of statistical moments (normally up to the second). As a form of reducing this initial computational cost, we proposed a tool that used recorded data from an existing simulation to start a new simulation, thus enabling methods that required stabilized simulations to run from the beginning instead of waiting until a certain point. This process is usual in commercial CFD software, however, to the best knowledge of the authors, this procedure does not exist for software that relies on distributed dynamic adaptive meshes. Here, it is implemented with domain re-decomposition into a structured-block adaptive mesh. The procedure is tested and validated for numerous cases using the MFSim platform.
Keywords
Computational fluid dynamics (CFD), Computational Simulations, Adaptive mesh refinement
