snGrad(rho) blows up my calculation - problems with the BC
 

Hi everybody!

I have the problem, that in some configurations (e.g. same case on 1 or 4 CPUs) my calculations blow up. I identified the calculation of rho as a possible source of my problems.

The solver I use is interFoam in OF version 2.2.x. I use a modified boundary condition for alpha, which is derived from the mixed BC. So besides e.g. the type and the value for a zeroGradient BC I have to provide some additional fields (refGradient, refValue, valueFraction). If i write out the rho field for the first timestep, everything seems to be fine until these additional fields are listed. They have values of xxxe+235 or in the next step nan for the mentioned refGradient or refValue or valueFraction.
I would just be glad if these fields are not touched for rho. The actual "value" of the BC is correct again.

Any ideas or which additional information can I provide?

Thanks in advance!
Nicklas

I provided a Boundary Condition for rho - so now it does not blow up the fields..

But this BC is inconsistent with the one I provide for alpha1 - so the results get strange. Somehow the calculation of rho with a non zeroGradient BC blows up my calculation..

And the BCs for pressure and velocity where "wrong" in the provided testCase. I'll update them as soon as possible.

Greetings

By the way. I had none of these problems in the OF1.6-ext Version. There, it works perfectly..