This paper deals with the application of a recently developed parallel DG solver to compute complex three-dimensional flows modelled by the Reynolds-Averaged Navier-Stokes (RANS) and the k-omega turbulence model equations. Several implementation details of the model (including the use of the logarithm of omega rather than omega itself as unknown, a lower bound on omega deduced from realizability considerations and accurate wall boundary conditions for omega) have been recently presented in [2] and are included in the 3D code here developed. Starting from the implicit DG method introduced in [3] for the RANS equations coupled with the k-omega turbulence model, the code has been recently extended in several directions. First of all, it can now handle arbitrary hybrid three-dimensional grids containing hexahedra, pyramids, prisms and tetrahedra. It is worth mentioning that the DG method here employed can use the same polynomial approximation inside the elements, irrespective of their geometrical shape. The viscous flux discretization scheme, which is the straightforward extension to three dimensions of the scheme proposed in [3], can also be used on hybrid grids without any modification. Secondly, the code has been fully parallelized to run on any parallel system supporting the MPI standard for all message-passing communication, by using the Portable Extensible Toolkit for Scientific Computation (PETSc) software [1]. PETSc is a highly sophisticated library that allows code developers to manage communication at an higher abstract level than message passing. Moreover, PETSc provides access to a large variety of up-to date numerical algorithms for the solution of the large sparse linear systems arising from the implicit time discretization. To test the code on a complex flow configuration, we have considered a film cooled nozzle vane cascade. This application is characterized by a wide range of rapidly changing geometrical scales and localized flow phenomena which are well suited to be discretized by using hybrid grids coupled with the higher order DG approach.
A parallel high-order discontinuous Galerkin solver applied to complex three-dimensional turbulent flows
REBAY, Stefano;
2004-01-01
Abstract
This paper deals with the application of a recently developed parallel DG solver to compute complex three-dimensional flows modelled by the Reynolds-Averaged Navier-Stokes (RANS) and the k-omega turbulence model equations. Several implementation details of the model (including the use of the logarithm of omega rather than omega itself as unknown, a lower bound on omega deduced from realizability considerations and accurate wall boundary conditions for omega) have been recently presented in [2] and are included in the 3D code here developed. Starting from the implicit DG method introduced in [3] for the RANS equations coupled with the k-omega turbulence model, the code has been recently extended in several directions. First of all, it can now handle arbitrary hybrid three-dimensional grids containing hexahedra, pyramids, prisms and tetrahedra. It is worth mentioning that the DG method here employed can use the same polynomial approximation inside the elements, irrespective of their geometrical shape. The viscous flux discretization scheme, which is the straightforward extension to three dimensions of the scheme proposed in [3], can also be used on hybrid grids without any modification. Secondly, the code has been fully parallelized to run on any parallel system supporting the MPI standard for all message-passing communication, by using the Portable Extensible Toolkit for Scientific Computation (PETSc) software [1]. PETSc is a highly sophisticated library that allows code developers to manage communication at an higher abstract level than message passing. Moreover, PETSc provides access to a large variety of up-to date numerical algorithms for the solution of the large sparse linear systems arising from the implicit time discretization. To test the code on a complex flow configuration, we have considered a film cooled nozzle vane cascade. This application is characterized by a wide range of rapidly changing geometrical scales and localized flow phenomena which are well suited to be discretized by using hybrid grids coupled with the higher order DG approach.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
