[sreehahaha]

Dear Foamers,

Hope you're having a great day.

I'm solving a scalar transport equation which happens to have a fvc::div(traction) in the governing equation. The field traction here is a 'surfaceScalarField'. For trial purposes, I initialised traction = scalar(1.0) uniformly. However, I'm using cyclic boundaries, and so I changed the values of one boundary (say leftPatch) to be the negative of its cyclic neighbourPatch (i.e. rightPatch). Now when I evaluate fvc::div(traction) to add to the governing equation, which will be a volScalarField, it is uniformly 0 everywhere. This is expected and I'm happy.

However, when I run the same code in parallel, the output of fvc::div(traction) is not uniformly 0 everywhere as it should be. I think this is because a given processor is not able to access boundaryField data that is located on another processor.

Is there a way to solve this issue? Please let me know.

Cheers,
S

[klausb]

"...I think this is because a given processor is not able to access boundaryField data that is located on another processor. ..."

You may have to write the code to process data across processes or to transfer data between processes. A good starting point to learn about it is the following programming tutorial:

https://github.com/UnnamedMoose/Basi.../OFtutorial5.C

[sreehahaha]

Thank you Klausb. The link you sent will be definitely helpful.

Furthermore,
1. Could you please tell me how I can change the values of the "surfaceScalarField traction" at processor boundaries? Preferably using a loop. I saw some posts here suggesting to use pos(), neg() functions that avoid using a loop. However, in my case, I will have to give certain values to 'traction' at run time. I'm not sure if this can be done without using a for loop.

2. What does correctBoundaryConditions() do exactly? I read that it corrects the pressure values at processor boundaries when you give p.correctBoundaryConditions(). However, in my case, I only have a surfaceScalarField. So, I cannot do traction.correctBoundaryConditions(). I'm a bit confused how it works.

Thanks again,
S

[klausb]

Dealing with processor boundaries is tricky.

Regarding 1: Read the following but you'll have to develop your own solution. Pout and Info are your friends to see what's going on at local and global level during develoment.

#4: Access patch points (on different processor) in parallel
http://www.wolfdynamics.com/training...g_openfoam.pdf
Flux at processor boundary
Neighbouring processor number for a patch
Access processor patches in parallel run

[sreehahaha]

Thank you so much Klausb, I will have a look at these.

Have a great day!
SP