Dear Mahaveer,

If you got the same trend of curves, then you should:

1- Check again the domain, the researchers sometimes solve for symmetry domain.

2- carefully check the boundary conditions.

3- If the problem is time dependent, try to vary the initial time and the time step.

4- Try another cases of the same paper.

5- If you do not get a consistency with these checks, you can try another paper, another cases, etc.

You potentially should consider the possibility, that the published result in the reference paper is simply wrong, in particular if the underlying research is published in a conference proceedings volume or in a scientific journal with not 100% well known excellence of there scientific paper review procedure.

In addition I would like to encourage you to:

a) Use a contemporary version of ANSYS Fluent, i.e. not a 6 years old version of ANSYS Fluent, but version 2019.R1. In the meanwhile dozens of known errors have been corrected in ANSYS Fluent over that period of time.

b) Vortex separators are known to be sensitive to the results of turbulence models. Which turbulence model are you using? If the flow has strong streamline curvature and swirl, use a k-omega-based BSL RSM turbulence model. If you can afford, try to do scale-resolving simulation using SAS-SST or SBES model for comparison with your first result. Are the results still the same (in time-averaged data with sufficient statistical reliability)?

c) Give it a trial and run the model in a different CFD code for verification. With only a little effort that can be done in ANSYS CFX. With a bit more effort in e.g. OpenFoam. Are the results the same? Trends the same? Is either one of the solvers giving an answer which corresponds with the reference publication?

By that you could try to narrow down the issue to its source.

Regards,

Dr. Th. Frank.

Mr. Muneer Ismael.. Right, I agree with your answer.

For the modeling of hydrodynamic vortex separator you need to try different turbulent models. With different turbulent model sometimes convergence issue also happens but if your results are getting the general trend as presented in the paper, then for accuracy you need to double verify the boundary conditions and suppositions. But the if the simulation cases are very well converged with all same mentioned conditions, then you need to consider trying on some other CFD tool as suggested by Thomas Frank

When you do C.F.D for any problem , do not trust in the paper you get you should depend on your self and make sure you do the best. Suppose you do not have any reference. I mean that you search for another method to check out.