Learning Forum

[skdubey]

Hi friends,

I am trying to run the model through a thin circular fluid domain with a single extended inlet in the radial direction and an outlet at the center. The flow pattern mostly resembles a source and sink flow. Due to the low cross section and unique boundary conditions, the flow appears to be turbulent.

a. I am a bit stuck on which model to use and what approach to take. Whenever I use the default settings for the K-E standard model, I always encounter a "floating exception" error. The same issue occurs when I use the K-E RNG model. However, when I use the K-E Realizable model, the simulation works fine. I thought I might be missing certain coefficient values required to run the K-E standard and RNG models. Do you have any suggestions on what might be causing this?

b. I have also tried using different websites to obtain initial values for the turbulent parameters. Do we really need these values for running the simulation in Fluent? I believe the default values in Ansys should be sufficient for running the simulation.

c. Additionally, I suspect there might be cavitation development in my flow. Currently, I am using the K_W_SST and K_W standard models. Should I continue with these models, or would it be better to use the K-E Realizable model along with cavitation modeling?

Thank you for your assistance.

[Rob]

Some images will help. It's very rare for rke to work and both standard and RNG ke to fail, so include images of the mesh and flow. 

All boundary conditions must be checked for a model. The turbulent values are less critical than velocity/pressure but still matter. If you mean the coefficients in the turbulence model panel I advise leaving those alone: they're exposed for a reason but only alter if you REALLY know why! 

Cavitation is another level of modelling. Do you want to see cavitation or just imply it's happening? 

[skdubey]

@Rob 

I hope the figure below helps:

a. Meshing: I have attached an image for your reference. Please take a look.

b. I have a quick question regarding cavitation. I discovered that the Zwart-Gerber-Belamri and Schnerr and Sauer models are compatible with all the turbulence models available in ANSYS FLUENT. Since my flow is turbulent, using any K-E or K-W model will affect the results for cavitation.

[Rob]

Difficult to see the mesh, but I suspect the jump in cell size & aspect ratio (centre, enlarged image, section) isn't helping the solver. O-grids (butterfly meshes) are only really used for block structured solvers now, and given Fluent has been unstructured for years they're not always helpful. 

So, check the mesh quality (cell quality and mesh quality are linked but aren't quite the same thing) and whether you're resolving all the flow features. 

[skdubey]

@Rob,

 

Hi Rob,

 

Thanks for the answer.  I think the meshing is something I am really struggling to work with the Ansys.

As we are focusing the flow mostly radially, the aspect ratio will be off course higher, for the mesh quality this is my results. In fact I check the meshing in the mesh component and mesh looks good.

 

 

Mesh Quality run:

A. Minimum Orthogonal Quality = 7.59539e-01 cell 341 on zone 151 (ID: 850082 on partition: 3) at location (-1.43727e-04 -4.92787e-02 5.58354e-02)

(To improve Orthogonal quality , use "Inverse Orthogonal Quality" in Fluent Meshing,

where Inverse Orthogonal Quality = 1 - Orthogonal Quality)

B. Maximum Aspect Ratio = 7.19138e+01 cell 8651 on zone 140 (ID: 189331 on partition: 0) at location (-1.08195e-02 -6.37903e-02 5.81116e-02)

[Rob]

OK, there are a couple of courses in the Learning section and tutorial in Help (click on Help in the solver). 

The aspect ratio is probably OK unless flow changes along the cell: read up on Aspect Ratio in the manual and literature. However, that just means you have OK cells. If there are too few cells to pick up the flow gradients the mesh quality may be poor. 

[skdubey]

@Rob, 

Are you suggesting that few cells, is creating issue for running the standard K-Epsilon Model and RNG model??

. I am bit surpise if the cell have issue with the gradient, How does the simulation works with other K-Epsilon (Realizable) and K-Omega (standard and sst) Model. 

[Rob]

I'd not expect the standard k-e to fail if rke is OK - but you may just have been lucky.

The mesh resolution check is fairly simple: plot contours with node values off. If the result looks "smooth" then it's probably OK. If the result looks very "blocky" then you need more cells.