CFD Advice, Tips and Tricks, Best Practices
Computational Fluid Dynamics is often said to be mishmash between science and art. Navier-Stokes equations can be elegantly derived, precisely discretized and efficiently coded. However, once a scientist or an engineer is faced with unexpected, persistent and inexplicable convergence problems associated with numerical, iterative methods of CFD, one need not only help of exact science but also extensive experience, subtle intuition and a dash of fine art of fluid flow modeling. The right advice, tip or trick from fellow modeller can save hours and days of frustration and futile drudgery. So if you have any CFD tips or tricks to contribute - please share with others.
Currently available CFD tips and tricks:
1. Assumptions and approximations at model building stage (pre-processing):
2. Errors and uncertainties related to model mesh building stage (pre_processing):
3. Numerical errors/uncertainties at model running stage (solution)
4. Post-processing and Interpretation or Results
e.g., AIRFLOW DISTRIBUTION TO A LOW EMISSIONS COMBUSTOR
http://www.faqs.org/patents/inv/658945Copy the number 20090222187, go to
http://www.freepatentsonline.com/20090222187.pdfEnjoy full pdf document that can be viewd and/or printed.
http://uspto.gov/) use plugins that do not work on most computers. Anyway, there is a relatively simple workaround, just use Apple web-browser Safari.
The steps are as follows:
1. Copy decomposer file into your working directory
let's say your working directory is
then copy decomposeParDict, decomposition parameter file, to your working folder
and edit decomposeParDict (use nedit or gedit or vi depending on Linux Distros)
$ nedit system/decomposeParDict
and set parameters (number of processors 2, method simple)
n ( 2 1 1 );
2. Prepare mesh
$ blockMeshand run decomposition script
and if need to run again
$ decomposePar -force(-force to overwrite previous folders and files)
3. Now you are ready to run mpirun script and save output data in case.log. You might want to add & at the end command line to run it in background.
$ mpirun -np 2 simpleFoam -parallel >case.log &
4. Reconstruct final results (stored in subfolder processor0, processor1 etc)
5. Postprocess results using your favorite postprocessing software eg ParaView on Windows or Linux
$ foamToVTKcreates VTK folder result ready to be used by ParaView