Learning Forum

[Snow1Runner]

Hello everyone! For the research group I've been working on, I'm trying to replicate the results of a simulation one of the Master's student left after graduating. I'm doing this both to get familiarized with Icepak and learn as much as I can from the setup of the simulation. Sadly, due to the nature of the work, I won't be able to share my files but I'll do my best to explain it as best as I can. 

I believe I have the whole setup of my simulation identical to his. I started with the same CAD model, made the same simplifications and set every single setting available in Icepak identical to his model. I do have access to his files so I'm using his Icepak model as a basis for my simulation. Our mesh settings are almost identical with around the same number of nodes. To keep the story short, our solutions are virtually identical expect for the pressure contours. I have checked multiple areas of interest for both temperature contours and velocity profiles and they are spot on so I do believe I'm on the right track. Our convergence plots and residuals both look very close and given that the everything is very similar to his published work, I am inclined to say that it has converged properly. 

For the pressure contour plots though, things are not identical. From a qualitative perspective, the contours look very similar with the "high" and "low" pressure spots located in the correct areas. However, when looking at the magnitudes, it seems like my values are shifted by about 90 Pa negative. I am very interested in the pressure drop of a cavity that has airflow thanks to a couple of fans. I have looked at the fan settings a million times and they also look identical (I apologize for using the word identical 10000 times in this post).  

At this point, the real simple answer would be to run his files and move on. The problem is that the files do not yield the expected results from the correct solution he left, the one used for his publication. My solution files are in agreement with the solution from his files but not with the correct solution. 

Skip here for a TLDR

Summary of the situation: 

Solution 1: My solution files 

Solution 2: Solution I ran from the Masters student files 

Solution 3: Solution that the Masters student left used for his publication

Solution 1 and 2 agree with each other but these pressure contour plots do not agree with Solution 3. 

 

I have looked extensively at the .cas files and .problem flies and compared Solution 2 and 3 against each other. Before I discuss my findings, I would love to hear what other methods you guys use to diagnose a problem like this where the solutions don't agree with seemingly identical settings. 

So far, for the .cas files, the only difference between Solution 2 and Solution 3 is under ";;; node section

---

: 

(10 (0 1 51f0a0 1))
(10 (1 1 51f0a0 1 3) (

vs

(10 (0 1 4fbed3 1))
(10 (1 1 4fbed3 1 3) (

For the .problem file, the situation is different. Solution 2 has the following lines that are non-existent on Solution 3. 

- set grid_hdm_aniso_uniform 0

- set grid_hdm_per_object_size_func_0

- set solve_hybrid_init 0

- set gridcut_opengl_saveimage 0

Please let me know if there is more info I can give to help find the answer to my question. 

Thanks a lot in advance! 

 

[Amine Ben Hadj Ali]

What is the difference in the pressure contours if all other quantities are same? Can you insert the screenshots?

[Snow1Runner]

Thanks for the response Amine. I believe I have included 2 attachments in the post with a comparison of the pressure contours and fan settings. Please let me know if you can't see them. 

Cheers!

[Amine Ben Hadj Ali]

Ansys staff do not look into attachments. Insert the pictures.

[Snow1Runner]

Pictures added. Apologies for the confusion! 

[Amine Ben Hadj Ali]

Solution 1 is equal to Solution 2 but both differs in pressure magnitude from the pictures the student left. I would first contact the masterand. Better export the settings of each case file and use diff on Linux or similar on Windows and compare the settings. Also use other initialization. Possible that there is a grid change but I doubt why it does only affect pressure.

[Amine Ben Hadj Ali]

My above statements assuming all cases are converging and running the same version.

[Snow1Runner]

Thank you Amine for your help. I have been looking around and Meld seems like a good option to compare the documents in Windows. As I understand it, Icepak creates a bunch of documents but the ones I should compare are the .cas files and .problem files. Is this correct? I am not entirely sure how to 'export the settings for the case files' within Icepak. 

Thanks in advance for your help!