Keywords: OpenFOAM, scalability, drag coefficient, RAS, LES, DNS.

In this paper the application of Open Source Field Operation and Manipulation (OpenFOAM) C++ libraries for solving engineering problems on parallel platforms is presented. The main objective of this paper is to present scalability of OpenFOAM on parallel platforms solving real engineering problems of fluid dynamics.

Three turbulence models i.e. Reynolds-average simulation (RAS), large eddy simulation (LES) and detached eddy simulation (DES) are investigated in terms of scalability of the code. LES and RAS turbulence models are also investigated in terms of suitability for drag coefficient calculations. For each of used turbulence models a different model to resolve turbulence properties is used. The standard K-epsilon model for RAS and the Smagorinski model are used to resolve subgrid-scale stresses which results in a filtering operation for LES; and Spalart-Allmaras is used to resolve small scale turbulence for DES.

The finite volume method (FVM) is used for spatial discretization and the creation of a numerical model which is then solved. The domain decomposition method is used to run OpenFOAM in parallel on distributed processors. Several numerical experiments were run to investigate scalability of LES, RAS and DES turbulent models implemented in OpenFOAM. The same numerical model with a grid consisting of approximately 4.5 million cells (96% tetrahedral and 4% prisms) was used for all of them. All runs were transient and all were initialized from same converged steady state solution using RAS K-epsilon turbulence model. All numerical experiments were run on a "Beowulf Class Cluster Computer" using 8, 16, 32 and 48 cores respectively. A mesh convergence study was performed and its results are presented and compared with published data [1,2]. Good correlation between the results calculated for the LES turbulence model and published data was found. Numerical experiments with a fixed number of time steps to study the scalability of all three turbulence models were performed and data are presented in this paper. It was found that all three turbulence models show high scalability up to 32 cores. After that a sharp decrease in scalability was observed. The reasons for this behavior are discussed in the conclusion section of the paper.

Another set of simulations were carried out to compare the drag coefficient of the LES and the RAS turbulence models. The results of those simulations are also presented in this paper. It was found that the LES turbulence model provides reasonably good agreement with the published data in contrast with RAS which was found to be much faster but also less accurate.

1

http://www.grc.nasa.gov/WWW/k-12/airplane/dragsphere.html, accessed (Dec 2012).

2

F.A. Morrison, "Data Correlation for Drag Coefficient for a Sphere", Department of Chemical Engineering, Michigan Technological University, Houghton, MI, URL, accessed (Dec 2012).

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £40 +P&P)