Hi everyone,

I am solving the flow in a cuboidal shape tank with rectangular c/s inlet & outlet. There is one problem

The area avg velocity@inlet & outlet are different from each other and also from actual velocity given as B.C at inlet.

(1)When i use the unstructured mesh i noticed that the avg velocity @inlet is less while @outlet is more than the specified B.C e.g. From POST, area avg V @inlet=0.002371 and @outlet=0.00261 while B.C is v=0.0025 by parabolic profile.

When i increase the surface resolution by refining mesh at both inlet & outelt, then velocity differences w.r.t B.C. is become more than earlier.

(2)For the structured mesh, when mesh at inlet & outlet surface is refined then velocity values come closer to but not equal to B.C. This happens upto certain level of refinement of mesh, beyond which solver ends with ERROR FINEMESS, fatal overflow.

Can anybody suggest what will be the ultimate solution? Is it necessary to refine mesh in rhe domain, rather making very fine at inlet & outlet.

IMP:: The profile@inlet, plotted by post is slightly different than given by CEL expression.

Thanks in advance

tuks

Hi,

It sounds like you have not yet got a fully converged solution, and possibly not yet reached grid independance. What are your residuals and imbalances? What physics are you using?

Regards, Glenn

hi Glenn,

The solution is converged well, the solver finished as RMS values are below 1e-08. For grid independent solution, i had used two mesh sizes, the sloution was quantitatively different but qualitatively same .

What is mean by grid independence? Is it using different mesh sizes (i.e edge lengths) or using same mesh size but refining the mesh in the region where variables variation is more? Can you specify clearly? I am trying to get avg values of velocity @inlet & @outlet to be same and be closer to the B.C. given at inlet..

Thank you again Glenn.

Tuks

Hi,

What do you mean by "quantitatively different but qualitatively the same"? It sounds like you are not yet grid convergent. Try a finer grid.

Glenn Horrocks

Thaks Glenn for your good insight. i have been trying with fine grids and i will inform you later, once let me finish with the mesh analysis.

Thak you once again

tuks