[magjo]

Good day foamers!

I have a problem with groovyBC when referencing field values in parallel. I'm running openfoam 6 with swak4foam developer version, and the case has been tested on both linux and mac with docker giving the same results.

The domain is reconstructed blood vessels, and has one inlet and several outlets. For the outlets I'm using a resistance analogy for outflow bounadry condition , and relating the outlet flow to the pressure and resistance in the blood vessels.

The setting of groovyBC in U is as follows:

 
    outlet1 //outlet 1 surface4
    {
      type            groovyBC;
      valueExpression "(p/resistance*2/(pi*pow(radius,2))*(1-sphere/pow(radius,2)))*normal()";
      variables
8
(
"frac=0.255235606748;"
"Q=1e-6;"
"xc=0.5*(min(pts().x)+max(pts().x));"
"yc=0.5*(min(pts().y)+max(pts().y));"
"zc=0.5*(min(pts().z)+max(pts().z));"
"radius=sqrt(sum(mag(Sf()))/pi);"
"sphere=(pow((pos().y-yc),2)+pow((pos().x-xc),2)+pow((pos().z-zc),2));"
"resistance=46751015.7676;"
)
;
}

This continues for 6 more outlets. The only difference from each outlet is that the resistance is changing. When running the case in serial everything works fine. When running it in paralllel the simulation starts exploding after about 100 iterations. The sphere part is to induce a parabolic flow profile, and gave no difference when excluding this part and just using pressure and resistance.

I found this note on the contrib/swak4Foam site:

9.6.3 Bugfixes
9.6.3.1 Variables not working for parallel computations
    If a variable is defined and the patch which it is defined on
    doesn't have any faces the variable is reported on that processor
    as not existing and the whole run fails

But it is not the variables that are undefined, if that had been the case, the resistance part would have been divided by zero and it would have exploded in iteration 1 and not 100. I also tried it without variables but the same thing happens then as well.

    outlet1 //outlet 1 surface4
    {
      type            groovyBC;
      valueExpression "(average(p)/17028966.9573*2/sum(mag(Sf()))*(1-(pow((pos().y-0.5*(min(pts().y)+max(pts().y))),2)+pow((pos().x-0.5*(min(pts().x)+max(pts().x))),2)+pow((pos().z-0.5*(min(pts().z)+max(pts().z))),2))/(sum(mag(Sf()))/pi)))*normal()";
}

I haven't found any recent issues of this problem, so I sincerely hope that someone knows how it is possible to fix it.

[Adriana Correia]

Hi Magnus,


I'm facing the same problem by using groovyBC in parallel. In my case, I need to access x position. Everything works well in serie, but then in parallel it blows up at the first iteration.


Did you find a solution for your case ?


Thank you and Best regards,
Adriana

[magjo]

Hi! I did not find a solution. There seems to have been a bug on the bug tracker previously:
https://sourceforge.net/p/openfoam-e...swak4foam/123/

This bug was related to a specific version, and not followed after 2014. You can also look at the thread which lead up to it, maybe it has some ideas that might help.

Bug in groovyBC in parallel computation

Unfortunately it seems that little has happened since 2014.