[beattacz]

Hi,

I’m trying to do grid independence test of flow past cylinder using Ansys Fluent 17.0. I carried out calculation for several meshes (from 134 993 to 395 098), But i still don’t get coverage of results. The flow is 2D, so I think the numer of 395 098 elements is really high. And I should get coverage of results for mesh with less numer of elements. I know that I do sth wrong, but I don’t know what. Could you help me ? I’m beginner in CFD. I present the conditions of the test below.

Conditions of the test:
• Diameter of cylinder is 20mm; Domain size 900mmx300mm.
• Simulation type 2D
• Unsteady flow
• Turbulent model: k-omega SST
• Water as a fluid
• Reynolds numer: 19 904
• Boundary conditions:
o Inlet: Velocity inlet: 1m/s, turbulent intensivity 5%, Hydraulic Diameter 0,3m
o Outlet as outflow
o Top and bottom walls and cylinder : non slip walls
• Y+ =(0,2 -10)
• Time step: 0,0001s, Number of time steps: 100000

Below, I attached my results and image of my mesh

Thank you
Beata

[hwet]

how do you know your results are not converged?

[beattacz]

because cd and cl, and pressure aren't the same for each mesh
I attached my results in table below:

[hwet]

mesh independence testing is not the same as convergence of a simulation.

and mesh independence testing does not mean that the two meshes will have exact 0.0000001 and 0.0000001 cd or cl etc.

when you go from 1 million cells to 2 million and get a cd of 0.5 in the first and 0.7 in the second, then you can make a mesh with 3 million cells but if the cd goes to 0.72 now and for a 4 million cell mesh 0.74, you have to stick to the 2 million cell mesh, since the improvement you are getting is small compared to the double resources and computer power you are using.

Unless ofcourse you are in NASA and sending a space shuttle to pluto, then you can worry about the small difference.

[pradip.c1000]

hi beattacz
may i know y did u use K omega sst model???

[beattacz]

yes I dod use, k-w SST model

[hitzhwan]

Originally Posted by hwet

mesh independence testing is not the same as convergence of a simulation.

and mesh independence testing does not mean that the two meshes will have exact 0.0000001 and 0.0000001 cd or cl etc.

when you go from 1 million cells to 2 million and get a cd of 0.5 in the first and 0.7 in the second, then you can make a mesh with 3 million cells but if the cd goes to 0.72 now and for a 4 million cell mesh 0.74, you have to stick to the 2 million cell mesh, since the improvement you are getting is small compared to the double resources and computer power you are using.

Unless ofcourse you are in NASA and sending a space shuttle to pluto, then you can worry about the small difference.

You mean the mesh in this cases has been completed, but you see the average pressure of cylinder, it has changed a lot for the mesh models, why?