a transient simulation question
 

HI everyone,

I'm just working on a transient simulation for a water-filled enclosed domain(only wall and symmetry) which rotates around a fixed axis with a ramped up uniform speed (25rad/s, ie) from stationary status at t=0s, and it lasts for 0.5s, then it suddenly stops rotation at t=0.5s and goes back to the stationary status which also lasts for 0.5s, then it repeats the first cycle and starts rotating with the same speed (25rad/s, ie) from t=1.0s.

I could find double vortex or secondary flow in the first cycle from t=0 to t=0.5s, what puzzles me is that this phenomenon didn't repeat within the secondary cycle from t=1s to t=1.5s, and even the order of magnitude of the velocity within the 2nd cycle is much much smaller than that in the 1st cycle, it seems the 25rad/s rotation speed didn't create any considerable effect to the flow within the 2nd cycle.

I'm not sure if there are problems with my process. In the initial condition setup where I used cylindrical velocity type , for t=0s, I initialized each velocity component with 0m/s, then I started to run the transient simulation giving the domain 25rad/s speed and it generated a res file at t=0.5s, which I used as the initial value file for the next run(rotation speed =0) , then in the same way, I used the generated res file at t=1s as the inital value file for the next cycle from t=1s to t=1.5s when rotation speed=25rad/s.

I noticed in the res file at t=1s, the velocity is very close to 0, and it 's almost same as t=0, it means the initial value for these two runs are very similar and they also rotates with same speed, also no difference for other setups, I think the secondary flow should also repeat, at least the order of magnitude of velocity shouldn't be much smaller.

Any suggestions or input would be appreciated very much? Thanks for your help in advance.

Harry

Hi,

What Reynolds number is this running at? If the flow is turbulent and you are using k-e turbulence model it will probably kill secondary vorticies.

Glenn Horrocks

HI Glenn,

The Re is around 150, and I chose laminar model from the very beginning, and I didn't change it in the following run. Do you have any idea about it please?

Harry

Hi,

So I assume by that low Re number the flow is always laminar? In that case forget about the turbulence comment. Then you should just look at general accuracy issues - convergence, timestep and mesh sensitivity, discretisation schemes etc. When they are all working properly then you will get an accurate simulation.

Glenn Horrocks