CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Dissimilarity in compiling new BC in 32bit OpenSuSE 103 versus 64bit CentOS 50 versus 64bit SLED 100 (http://www.cfd-online.com/Forums/openfoam-solving/59096-dissimilarity-compiling-new-bc-32bit-opensuse-103-versus-64bit-centos-50-versus-64bit-sled-100-a.html)

sponiar February 25, 2008 16:36

Hello to OpenFOAM forum. We a
 
Hello to OpenFOAM forum.
We are working on implementation new BC in OF 1.4.1. New BC is depend on neigbour boundary centroid value of any simple scalar mathematics operation (dot product, gradient etc.)
We noted successful implementation in 32bit OpenSuSE 10.3 linux distribution with satisfactory convergence with simpleFoam and k-epsilon turblence models.

We copied same source c++ source (*C, *H, MakeFiles and MakeOptions) code of BC on another workstation with various linux distribution. In our case: 64Bit CentOs 5.0 and 64Bit SLED 10.0. We noted very individual behaviour.
I have to say, OpenFOAM 1.4.1 is same over all workstations. That means, we seting-up OF 1.4.1 same way something like: read README instalation setup, seting up enviroment, using precompiled and prepared GNU compiler from OF download.

Behaviours, which I was talking above are:
32Bit OpenSuSe 10.3.:
- compiling, implementation and solving is successful.

64Bit CentOS 5.0:
- compiling is successful,
- implementation new BC is successful,
- solving case:
- only components of velocity was solved, but scalar filed like p, k, epsion is not.
- FOAM FATAL ERROR say:gradientInternalCoeffs cannot be called for a defaultFvPatchField


64Bit SLED 10.0.:
- compiling is successful,
- implementation new BC is not successful,
- FOAM FATAL ERROR say that new boundary is not known.

Any idea, where could be any mistake. If I start 'foamInstallationTest' on every workstations, results are OK = no critical error, configurations are OK.

valery_reznic February 26, 2008 03:09

If you have one working system
 
If you have one working system, you can use
ermine (http://magicErmine.com) or statifier (http://statifier.sf.net) to pack executable with all shared libraries it's needed and move it to another box). Moving from 32-bit sytem to 64-bit is OK.

sponiar February 26, 2008 04:58

Hello Valery, thank you for y
 
Hello Valery,
thank you for your respondence. This is not, what I am looking for. I am not sharing OF directors tree. Every OF instalation is individual on invidual workastions. Differences is only in architecture and using linux distributions.

I am just testing new BC implemented on outlet BC. I don't understand this situation, when precompiled OF 1.4.1+gcc4.2.1+paraview+java were correct installed all over workastions with successful simulating some case (interFoam, simpleFoam, icoFoam etc.), I noted individual behaviour on each of linux distributions, even if new BC was compiled and implemented more or less successful.

My first guess was in some header were not included. Something like discussed in this thread:
http://www.cfd-online.com/OpenFOAM_D...ges/1/202.html
where Jarrod Sinclair contributed with FAOM FATAL ERROR, which is same like in my case. I thing, library libcfdTools is not using in OF version 1.4.1.

D.S.

valery_reznic February 26, 2008 07:31

Sure, my suggestion don't help
 
Sure, my suggestion don't help to understand the problem, but it can (I hope) avoid it: if you run exactly same executable (and it use exactly same libraries) it should work OK on all platforms.

Valery


All times are GMT -4. The time now is 15:15.