Learning Forum

[IRobin]

Hello,

I have been experimenting with dynamic meshing (in fluent) for several days now and I encountered (what looks like) a problem. It concerns the movement of a solid through air which is suddenly stopped. In other words, the velocity profile is given as (written and compiled in a UDF):

v_x = 0.0 m/s and v_y = 1.0 m/s when t <= 0.3s

v_x = 1.0 m/s and v_y = 0.0 m/s when t > 0.3s

It is at this sudden change in body velocity, something strange happens. Below are the velocity vectors at t = 0.3 s and t = 0.33 s respectively.

After the change in direction, the velocity vectors in the first figure above the moving solid have dissappeared completely in the second figure. This is quite strange as one would expect them to continue for some time longer before slowly dissipating. I have already varied the time step size and the max number of iterations. Furthermore, all residuals seem to convergence.

I am using he standard k-eps model and refined the geometry locally fine (I also tried k-omega SST but with no succes either). Does anyone have an idea why the velocity is suddenly dissipated so quickly? Might it maybe related to the wall functions? Or is it not allowed to have such abrupt velocity changes (I did tried a linear decreasing velocity profile but it also showed the same behavior) or is it related to the dynamic mesh?

Regards,

Robin 

[Rob]

What other boundaries are there?  How big a time step did you use? At 1m/s I'd not expect compressibility to be an issue, but check the local pressures to be sure. 

[IRobin]

Dear rwoolhou,

I used the following boundary conditions:

Furthermore, after reading your reply, I lowered the time step to 0.001s with a max iteration of 40 (although this seems too low, it converges rather quickly), hoping this might change something, but unfortunately it did not.. Below are the local pressures given at respectively time step t=0.3 and t=0.31s:

The abrupt change in pressure is still clearly visible.

 

Regards,

Robin

[IRobin]

Could some one maybe help me further on this? I have been trying many different options already. I have switched between the realizable, standard and RNG k-epsilon model as well as the k-omega SST (+changed the wall functions). Furthermore, I reduced the time step a lot, but still with no change. The solution converges as I observe the residual as well as the velocity at a certain point vs the iterations. But still, the velocity experiences a major jump when the solid has stopped moving... Below is the velocity vs the time at a location roughly 30 cm above the initial position of the box.

It is seen that the velocity increases as expected (since the box/wall moves closer) but drastically drops when the box has stopped moving. Shouldn't one ecpect it to be more gradual rather than so abruptly?

Any suggestions are welcome,

 

Robin