Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
ELLIPTIC GRID GENERATION USING PARALLEL COMPUTING
Submission Author:
Juan Carlos Assis da Silva , RJ
Co-Authors:
Juan Carlos Assis da Silva, Rômulo Bessi Freitas, Leonardo Santos de Brito Alves
Presenter: Juan Carlos Assis da Silva
doi://10.26678/ABCM.COBEM2023.COB2023-0664
Abstract
Through the last few years, numerical computations of fluid flow governing equations have been used to understand a wide class of engineering problems. In many such cases, the continuous governing equations are solved at discrete points in space and time, where partial differential equations are transformed into algebraic equations. Independently of which technique is employed to discretize the equations (e.g. finite volume, finite difference, finite element, and so on), it is necessary to know a priori the discrete points that represent the geometry of interest, called grid or mesh. Since the numerical solution quality is strongly related to the grid quality, this work focuses on generating two-dimensional grids based on nonlinear Poisson-type equations and algebraic transformations. Furthermore, boundary layer problems often require grid refinement, as well as the orthogonality of the grid elements near the wall to facilitate boundary condition implementation. Both requirements were also considered. As direct numerical simulation (DNS) usually requires a large number of grid points, serial computations of grids may, unfortunately, be too time-consuming. In order to reduce this cost, this work applies parallel computation techniques using a message-passing interface (MPI) paradigm through the PETSc library structure. In this way, the domain is split into different CPUs that compute each part of the solution, reducing the computational time required. The nonlinear Poisson equation was solved using a finite difference technique coupled with parallel Jacobi iterations. The results show the code's scalability, speed-up and efficiency.
Keywords
Grid Generation, Parallel Computing, Computational fluid dynamics (CFD)
