OpenFOAM is numerical library for Continuum Mechanics simulations. Written in C++ and using an object-oriented approach, OpenFOAM provides a flexible simulation platform by mimicking the form of partial differential equations in software. The library provides extensive capabilities in pre-configured solvers, utilities and libraries, with massive parallelism in a domain decomposition model. OpenFOAM is freely available and open source, licensed under the GNU General Public Licence. For more information about OpenFOAM, see the OpenFOAM home page by OpenCFD Ltd.

OPENFOAM is registered trademark of OpenCFD Ltd.

Current version and installations

Current version of OpenFOAM is 2.0 and it is available at CSC on Louhi, Hippu and Vuori servers. Previous version 1.7 is also available on the servers.

Initialization of OpenFOAM

An environment for OpenFOAM usage can be created by following the instruction given here.

How to get started

User should read the manual available on OpenFOAM home page (see over).  These manuals include all required information for the start, such as tutorials.

For to test the OpenFOAM environment on Hippu, one may run the Lid-driven cavity flow tutorial (see the OpenFOAM User Guide) following these instructions.

Compiling your own solver

Instructions for how to create an own solver by modifying an existing solver, are given here.

Running OpenFOAM in batch mode

Full instructions for usage of CSC's computing servers, including detailed information about the batch job executions, are available in here. An important notice! Hippu server is dedicated to interactive jobs, and no batch job environment exist.

Examples of running OpenFOAM cavity tutorial in sequential and parallel mode are available here.

Comparison of OpenFOAM parallel efficiency on Murska and Louhi (Cray XT) servers

Parallelization efficiency test was done using the laminar cavity flow tutorial expanding it to three dimension and using three meshes - 8, 13,8 and 22  milj. cells.  The test were run with 4, 8, 16, 32, 64, 124, 256, 512 and 1024 cores.  On Murska server with OpenMPI message passing library and Infiniband communication link.  The results of parallel efficiency on Murska and Louhi are seen here.

OpenFOAM scalability on Louhi (Cray XT) server

The test was done using the same case as the comparison between Murska and Louhi, see over.  Scalability results are seen here.

Parallel efficiency of OpenFOAM on Murska server was shown to be good (> 75 %) with enough large number of degree of freedoms per a parallel process. Notice that here the test runs were done with the laminar flow solver.  Turbulence models etc. will naturally increase the degree of freedoms per a cell.

On Cray XT OpenFOAM exhibits super-linear scalability.