Learning Forum

[seffa]

Background:

My company is using ANSYS Discovery AIM to study airflow in a ventilation pipe. The first attached picture (Fig 1) shows the duct. It has 1 inlet at 12ft/s and 28 outlets at 0 pascal. The material is air at steady state.

 

We have actual measured values for all 28 outlets and are comparing them to the values shown in the simulation. I have run three simulations with increasing mesh-element count and iterations to try and get better results. However, as I do this the results look worse and more unstable the finer the mesh and longer the iterations which is counter-intuitive to me. Please help me identify the cause of this. 

 

Procedure:

1.) My geometry is of the fluid volume.

2.) My mesh is shown in the second attachment (Fig 2). These settings were recommended to me by an ANSYS engineer last year.

3.)For my solution control an ANSYS engineer recommended removing the solution criteria and just adding the iteration count as the default convergence criteria was too small. 

4.) For results I took a calculated value of the average Z-velocity in each outlet. Z is the direction the outlet is facing (upwards). Then I graphed the velocity against the outlet number for the simulation and the measured values.

 

Results:

There are three attached images for the results. The simulated results are in yellow and the measured results in blue. X-axis is the outlet number (0 is inlet and 28 is the distal end). Y-axis is the velocity converted to ft/min.

        1.) Mesh count around 130,000 with 100 iterations (Fig 3).

        2.) Mesh count around 130,000 with 500 iterations (Fig 4).

        3.) Mesh count around 360,000 with 500 iterations (Fig 5).

 

I also calculated the total cubic feet/min (CFM) leaving all the exits and it did not match up with the CFM going into the duct which means the system did not display steady state which was a conditions that was checked in the geometry. The difference between these CFMs got worse (more different) as the mesh count and iterations increased.

CFM going into the duct is 928 which is shown in the actual measurements column at the bottom row. (12.6 ft/s * 60s/min * 1.23 ft^2 = 928ft^3/min)

1.23 ft^2 is the area of the duct.

 

Final Notes:

So to me as I tried to increase the accuracy of the simulation the results got worse and appeared to be less stable. This makes me question every simulation I have done in the last year using this software which is troubling (We have sold time using these modeling techniques).

 

Please let me know if you would like any other information about my setup.

I would love to have a screen share meeting with someone if that is appropriate.

 

Contact:

You can reach me at: seffa@techmark-inc.com or my cell 269-313-3048 during business hours here from 8am-5pm EST.

Thanks

Sam

[Subashni Ravichandran]

Hello seffa

I will be looking into your query. Would it be possible for you to share the Discovery AIM file with us?
I have shared an Ansys Secure Transfer link to you via email which can be used to transfer the files. 

[seffa]

Subashni I sent the file over.

 

Sam

[Subashni Ravichandran]

Hello  seffa

Unfortunately, I'm unable to open the file that you have shared. I will require the folder along with it. Instead, You can save your file "As Archive" and share the .wbpz file.

Please refer to this post regarding sharing AIM files.

[seffa]

Subashni I just sent the archive over.

 

Sam

[seffa]

seffa Please let me know if you had a chance to look at it.

 

THanks

Sam

[Naresh Patre]

Hello seffa

In Subashni's absence, I am following up on this query. Please let me know if you still need any assistance with this query.

[seffa]

Naresh Patre I would not worry about it anymore.

 

Thanks though

 

Sam