A efficient code to solve the incompressible Navier-Stokes equations on supercomputers
NEWS 12/03/2015 Google has just decided to slowly shut down google code (see http://google-opensource.blogspot.fr/2015/03/farewell-to-google-code.html). As a results, we are going to shut down this website very soon. We are more likely going to create a new platform for the large community of Incompact3d users. It is not clear yet under which form but we will keep you up to date with our decisions.
NEWS 03/03/2014 Incompact3d User Group Meeting, Imperial College London, Department of Aeronautics, Lecture Room 266, 24th April 2014, contact Sylvain Laizet s.laizet@imperial.ac.uk. Check the website of the meeting for more details http://www3.imperial.ac.uk/tmfc/conferences/incompact3d2014
NEWS 20/11/2013 Incompact3d and NAG Winners of Sixth HPC Innovation Excellence Awards http://uk.finance.yahoo.com/news/idc-announces-winners-sixth-hpc-160400051.html and http://www.hpcwire.com/off-the-wire/nag-awarded-hpc-innovation-sc13/
NEWS 15/11/2013: In the Downloads section, a new user guide is available with a new section on how to open a file generated by Incompact3d
NEWS 21/05/2013: In the Downloads section, you will find an archive of the last version of the code in a channel flow configuration. The configuration is the one from Laizet S. & Lamballais E., High-order compact schemes for incompressible flows: a simple and efficient method with the quasi-spectral accuracy, J. Comp. Phys., vol 228-15, pp 5989-6015, 2009.
NEWS 21/05/2013: The second version of the code is now available. Minor bugs have been fixed. Some new features have been added:
1- It is now possible to save sub 3D domains with the subroutine decomp_2d_write_subdomain(1,ux1,508,608,226,315,136,225,filename)
In this example, you will save in filename the 3D array ux1 from i=508 to 608, j=226 to 315 and k=136 to 225.
2- It is now possible to use halo-cell support when it is necessary for neighbouring blocks to exchange data explicitly. More information can be founded here: http://www.2decomp.org/halo.html
NEWS 29/10/2012: The first version of the code is now available, with a user guide.
NEWS 10/08/2012: We are hoping to release the first version of the code by the end of 2012. Thank you very much for your patience.
Incompact3d is a powerful numerical tool for academic research. It can combine the versatility of industrial codes with the accuracy of spectral codes. Thank to a very successful project with NAG and HECToR (UK Supercomputing facility), Incompact3d can be used on up to hundreds of thousands computational cores to solve the incompressible Navier-Stokes equations. This high level of parallelisation is achieved thank to a highly scalable 2D decomposition library and a distributed Fast Fourier Transform (FFT) interface. This library is available at http://www.2decomp.org and can be freely used for your own code.
Incompact3d is based on a Cartesian mesh. The use of such a simplified mesh offers the opportunity to implement high-order compact schemes in the code for the spatial discretisation whilst an Immersed Boundary Method (IBM) allows the implementation of any solid wall/bluff body geometry inside the computational domain. The main originality of the code is that the Poisson equation (to ensure the incompressibility) is fully solved in the spectral space via the modified wave number formalism, no matter what the boundary conditions are (periodic, free-slip, no-slip, inflow/outflow, etc.). Note finally that the pressure mesh is staggered from the velocity one by half a mesh to avoid spurious pressure oscillations that can be introduced by the IBM.
A priori, the combination of high-order schemes with the IBM might be problematic because of the discontinuity in velocity derivatives imposed locally by the artificial forcing term. However, even though the formal order of the solution can be reduced as a result of the IBM, the code has been demonstrated to be far more accurate with a sixth-order scheme than with a second order scheme both in statistics and instantaneous field. Note that there is an ongoing research project in Poitiers (France) on this topic. The idea is to use an innovative 1D approach to reduce the discontinuity in velocity derivatives at the wall of the solid body. This new strategy is not yet implemented in the released version of the code.
More information about the numerical methods can be found in:
Laizet S. & Lamballais E., High-order compact schemes for incompressible flows: a simple and efficient method with the quasi-spectral accuracy, J. Comp. Phys., vol 228-15, pp 5989-6015, 2009
Lamballais E., Fortune V. & Laizet S., Straightforward high-order numerical dissipation via the viscous term for Direct and Large Eddy Simulation, J. Comp. Phys., Vol 230-9, pp 3270-3275, 2011
More information about the parallel strategy of the code can be found in:
Laizet S. & Li N., Incompact3d, a powerful tool to tackle turbulence problems with up to 0(10^5) computational cores, Int. J. of Numerical Methods in Fluids, Vol 67-11, pp 1735-1757, 2011
Laizet S., Lamballais E. & Vassilicos J.C., A numerical strategy to combine high-order schemes, complex geometry and parallel computing for high resolution DNS of fractal generated turbulence, Computers & Fluids, vol 39-3, pp 471-484, 2010
Acknowledgement
Sylvain Laizet and Eric Lamballais would like to thank Imperial College London and The University of Poitiers for agreeing to make the code Incompact3d available for the scientific community.