dynamic grid error when running viscous flow test

[YaojieYu]

Hi, Jacob. Recently, I am running a unsteady deforming cylinder case. I found the Euler equations works pretty well. However, the viscous test is unstable. I guess there may be some wrong in viscous module when the grid move. Have your ever tested a viscous dynamic grid case?

[JacobCrabill]

Hi Yaojie,

I haven't been running any viscous cases lately, so it's entirely possible there's a bug somewhere in the viscous portions of the code. One issue I can think of right now is that the time step estimate in the code is for inviscid cases only currently, so you will have to use a fixed time step (dtType = 0).

What's the deforming cylinder case that you're running? I'll be out of town for a while, but I'd be curious to try out your test case when I get back.

-Jacob

[YaojieYu]

Hi Jacob,
The test I want to run is a rigid moving cylinder. So, I add the following code into motion switch in geo.cpp.

case 4: {
      /// Rigid oscillation in a circle
      if (gridID==0) {

        #pragma omp parallel for
        for (int iv=0; iv<nVerts; iv++) {
          xv(iv,0) = xv0[iv].x;
          xv(iv,1) = xv0[iv].y+ 0.2*sin(2.*pi*0.2145*rkTime);

          gridVel(iv,0) = 0;
          gridVel(iv,1) = 0.4*pi*0.2145*cos(2.*pi*0.2145*rkTime);
        }
      }
      break;
    } 

All the grid points move in the same sinusoidal manner, with fr=0.2145 and Ay = 0.2. I even set dt = 1.e-7, which still burn down.