Grid Refinement

Yamuna

Hi,

I am dealing with prediction of mixing regimes in a mixing tank ( having paddles). Currently, I am trying to do grid refinement for a simpler version of my case, so that I can increment the complexity step by step (such as paddles,multiple entry/exit ports, 3D geometry etc) and get a better feel of the situation. I am trying to predict the flow regime occurring inside a tank (2D- 780 mm X 575 mm), having a single entry and exit port (11mm dia each), velocity of 0.03 m/s and Reynolds no- 18975.

When I increase the grid density (4524, 6984, 12480, 28224- Uniform spacing), there is deviation in the velocity components (which I am monitoring at 150,300,450,600 & 750 mm depths). First three grids are showing the almost same pattern while the last grid is deviating. Since it is transitional flow, I am using RNG K-Epsilon model and also, SST K-Omega model (comparing)

Now, the problem is I don't have any set of data to validate which model or which grid is nearer to real data. So, can any one tell me how to overcome this situation?? Are there any way to know which one to go for??? I doubt if there is any numerical diffusion happening with higher density grids.

Since, I am completely new to this field, I am having trouble to understand the flow fields inside the tank. Can i do some manual calculations for say KE or velocity, to check whether the chosen model predicting right flow?

It would be of great help to me, if you could give me your valuable suggestion. I am really confused and frustrated doing grid independent study. I sincerely appreciate if any one can suggest me regarding this problem.

Thanks a lot!