Learning Forum

[burgreen]

Fluent 2019.R1 will read in legacy mesh interfaces fine, but it does not properly create mesh interfaces any more. Instead, it creates non-overlapping wall zones. A non-overlapping wall zones does not communicate between two fluid domains. Even if I delete a mesh interface in a legacy case and recreate it, non-overlapping wall zones are created. I've tried every combo of mesh interface options.

Has anyone else experienced this? Does anyone have a solution to this issue?

[Raef.Kobeissi]

HI
We need to understand more the nature of your geometry. It is hard to give an answer without that.

[burgreen]

Tet mesh. Three domains. OpenFOAM creates interfaces and solves it like a champ.

[DrAmine]

After initialization just list the interface: what is written in the console?

[burgreen]

What do you mean by "initialization" and by "just list the interface"?

[DrAmine]

Initialization of the flow field and open the interfaces panel and hit list.

[burgreen]

Description Name ID Area Area Percentage

---

---

---

---

---

Interface-Zone-Side-1 ai-fb-ih 2 4.43215377e-03

Interface-Zone-Side-2 ai-ih-fb 12 4.43416615e-03

Non-Overlapping-Zone-Side-1 ai-fb-ih-non-overlapping 20 9.26657029e-09 0.0002%

Non-Overlapping-Zone-Side-2 ai-ih-fb-non-overlapping 21 1.31230284e-06 0.0296%

Interface-Interior-Zone z-ai-fb-ih-interior-1-1 19 4.43283618e-03 100.0154%

 

So, should look great right?

 

Residuals:

 

 

 

 

Velocity ought to be passing through the interface. It is not. I clipped away the velocities in the 3 - 2500 m/s range.

 

 

 

[burgreen]

If I create the interface with any combo of different options, I get basically the same results as above, namely, no flow through the interface and an unstable non-converging solution.

[DrAmine]

Not perfect overlapping but almost is interpreted as interior. Can you recreate with option matching.
On both sides of interface does the mesh look dissimilar?

[burgreen]

Both interfaces have been recreated using the matching option. And as previously stated, I get basically the same results as above, namely, no flow through the interface and an unstable non-converging solution.

The mesh of each interface is shown below.

Description Name ID Area Area Percentage

---

---

---

---

---

Interface-Zone-Side-1 ai-fb-ih 2 4.43215377e-03

Interface-Zone-Side-2 ai-ih-fb 12 4.43416615e-03

Interface-Interior-Zone z-ai-fb-ih-interior-1-1 28 4.43283638e-03 100.0154%

 

[DrAmine]

But the information you are posting tells me that the interface is almost 100% overlapping.

[burgreen]

No offense, Amine, but really? Really? This is the extent of help one receives on this forum? I have been running interface overlapped cases in Fluent for almost 20 years. I think I can be trusted when I say they are not working properly in this version. Without them, I am dead in the water. Until they are fixed, I am in the process of migrating my entire CFD workflow away from Ansys to OpenFOAM.

[DrAmine]

Do you mean that this worked in older version? I just interpret the output from Fluent you are sharing and I just to gave an idea as I do not know your case. If it is a bug or an issue we need to look into it.

[burgreen]

It has worked fine in all Fluent versions from 2002 up until 2019 with the 2019.R1 release installed on our system. From my experience, this is a bug or an issue that needs attention from Ansys devs. I thought that was made clear in the OP when I used the word "broken", or maybe not.

And even in 2019.R1, the interfaces work fine (i.e., the solution is stable and converges) if I read in an older case file. The issue arises when interfaces are created in 2019.R1. As we have established, everything looks fine, but they are not communicating between domains.

[DrAmine]

We will take care.

[burgreen]

Turns out my case was read in as an inviscid case. Therefore, the viscous resistance inputs were not enabled when defining the porous media parameters for the fluid. I mistakenly input incorrect values into the inertial resistance inputs, thinking they were the viscous resistance block. Later, when laminar viscosity option was set. The viscous resistances were automatically initialized to usually large values. This combo of errors lead to the previous results.