Learning Forum

[fuyu-m]

Hello, There is a problem that the customer's analysis causes a difference between the inlet and outlet volume flow rates. I verified this problem with a simple model, but the same problem occurred. The model has an inlet, an outlet and a rotating wall as attached. After 5 seconds, the volumetric flow rate at the inlet is -0.265 and the volumetric flow rate at the outlet is -0.030, resulting in a difference of about 10 times. ● In this analysis, the volumetric flow rate at the inlet and outlet is not balanced. Why is this? ● Is my boundary condition setting incorrect? Please tell me how to solve this problem. Thank you.

🛈 This post originally contained file attachments which have been removed in compliance with the updated Ansys Learning Forum Terms & Conditions

[Subashni Ravichandran]

Hello Fuyu-m

I'm trying a couple of iterations on the model that you have shared.

I will get back to you with a solution soon.

[fuyu-m]

Subashni 

Thank you for your quick reply.
I hope to find a solution.

Thank you.

[Subashni Ravichandran]

Hello Fuyu-m

I run the model shared by you without making any changes but at default fidelity, in my system equipped with NVIDIA Quadro P5000. the following are my observations:

1. You will see that I got the expected result. The inlet and outlet flow rate are almost the same. (please refer attached image).
    
2. Also, notice the velocity contour plot in the above image. You can clearly see the flow of water through the outlet. 
   In the image shared by you, we can see that the contour plot is different from this. There seems to be no flow from the outlet. 
    
3. I also run another iteration on a colleague's system, with a different graphics card to get an almost similar outcome. Thereby, we can tell it could not be an issue with Graphics.

It seems to me like there is some issue with the definition of the Outlet boundary condition in your model although I could not find anything wrong in the model you have shared. 

Could you please check your model or maybe try restarting the application and running the simulation again?

[fuyu-m]

Subashni 

Thank you for checking the model.

I saw that you and your colleague's machine did not have the problem of inflow and outflow volumetric flow rate mismatch.

 

I tried the following.
1. I updated my graphic driver to the latest version. (I use the P4000)

2. I checked if changing the precision would cause the same problem. (I moved the slider bar from precision Max to speed Max)

3. I confirmed that the same problem would occur on other machines.
(I verified with P5000 and M2000.)

 

And I got the following result.
1. Updating the graphics driver did not solve the problem.

2. Changing the precision may or may not cause problems. (Of course I understand that changing the precision will change the result.)
High accuracy does not mean that the problem does not occur, but with some accuracy it does not matter, but it does occur with higher or lower accuracy.
I changed the accuracy and noticed that the volumetric flow rate of the inflow hardly changed, and the volumetric flow rate of the outflow changed significantly.

3. Similar problems occurred on other machines.
The attached image is the result when it was verified with P5000. Please note that each has a different accuracy.

 

 

 

  

● Can I ask you the following?
1. Can you check if changing the accuracy on your or your colleague's machine does not cause any problems?

2. If the problem also occurs on your machine, can you tell me how to solve this problem?

3. If the problem does not occur on your machine, can you tell me what to do to solve this problem?

Thank you.

[Subashni Ravichandran]

Hello Fuyu-m

We observed similar issues of different graphics cards with different fidelities. 

The problem seems to be with the Geometry.

I run similar iterations on different models with similar BCs. I didn't find any issue. So the issue seems to be particular to this model alone.

I will have to consult my team on this issue.

[fuyu-m]

Subashni 

Thank you for checking the model again.
I'm glad to share the problem.

We are waiting for you and your team to find a solution to this problem and let us know.

Thank you.

[fuyu-m]

Subashni 

Hello,

Is the issue investigation in progress?
Could you tell me the situation?

This problem is also happening with our customers' models.
Our customers are considering deploying Discovery Live and hope to solve this problem.

We expect you and your team to solve this problem.

Thank you.

[Subashni Ravichandran]

Hello Fuyu-m

I have raised this issue with the development team. I will keep you posted on update from them. 

[Subashni Ravichandran]

Hello Fuyu-m

I had a few discussion with my team and I also run a few tests and the conclusions are as below:

1.  Discovery Live applies Rotating wall Boundary condition on only axially symmetric faces. 
   In the model shared by you, the Rotating wall BC has been applied to a half-cylinder face.
   Thereby resulting in inaccurate results.
    
2. Performing a similar analysis using an axially symmetric face gave me consistent results.
   The results did not fluctuate in any unexpected way with varying fidelity.
   Higher fidelity produced a more accurate result.
   The Flow rate at Inlet and Outlet were consistent at different fidelity.
   
       

[fuyu-m]

Subashni 

Thank you for verifying the problem.

I understand that the semi-cylindrical surface can not use the boundary conditions of the rotating wall.
I will explain this to customer.

Thank you.