Learning Forum

[Thodorisdm]

Hello, 

 

I am running a dynamic mesh problem. The walls are having a sinusoidal motion and the layering method is employed. When I run the case for a frquency of 1Hz it runs normally. When I increase to 20Hz, the mesh is not moving (in the preview mesh at every timestep I have the same volumes). When I am then checking my dynamic zones, every rigid body zone has a nan y-coordinate in the cg definition. Is it a bug?

[Federico]

 

Hello, 

did you adjust your time step size accordingly when changing the frequency? 

Also, mesh preview will deform the mesh, hence you will need to re-read the case to a previous state with undeformed mesh. To a Mesh Check and make sure the CG coordinates are valid before you initialize the motion.

 

[Thodorisdm]

Yes everything has been adjusted. If my timestep was not adjusted, I think I would have a message for negative volumes, no?

[Federico]

Yes and no, depending on what your ymotion UDF does, time step size can bring in other issues as well.

As per my previous comment, do a mesh check prior to doing a Preview of Mesh Motion. Otherwise, you may share your UDF code here to give us more information.

[Thodorisdm]

I always check the mesh prior to the preview of the mesh motion. Hereby my UDF. Always works for frequency 1-5 but not for larger.

 

#include "udf.h"

 

#define amplitude 0.01

#define frequency 10

 

DEFINE_CG_MOTION(ymotion,dt,vel,omega,time,dtime)

{

real vmax;

real Tper; 

 

Tper = 1/frequency;

vmax = (amplitude * 2*M_PI) / Tper;

vel[1] = vmax*(cos(2 * M_PI * time / Tper));

}

[Federico]

what is your dt for this case with 10Hz?

[Thodorisdm]

dt = 4.7 e-4 sec. Ideal height on the stationary zones to calculate the split and compression is 8e-4 m. and the real height of my mesh there is 6e-4 m .

[Federico]

Ok, all your DM settings appear to be ok. 

However, Tper appears to be an integer division, which results in 0. Can you try with Tper = 1.0/frequency ?

[Thodorisdm]

yes, that solves everything! Thank you!