|
[Sponsors] |
Ahmed vehicle body tutorial RANS simulation gives dubious results |
![]() |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
![]() |
![]() |
#1 |
New Member
denzell
Join Date: Mar 2014
Posts: 12
Rep Power: 12 ![]() |
Dear all,
I found a tutorial on the Github page which contains the .SU2 meshfile and configuration file for an Ahmed vehicle body here: https://github.com/su2code/TestCases...rans/AhmedBody First of all I'd like to thank the SU2 development team for posting this (at least to me) valuable tutorial as it gives me a good indication on how to configure the config file for vehicle applications. I tried to run the tutorial simulation, where I was prompted that there was something wrong in the Config file, i.e. % Adaptive CFL number (NO, YES) CFL_ADAPT= NO % % Parameters of the adaptive CFL number (factor down, factor up, CFL min value, % CFL max value ) CFL_ADAPT_PARAM= ( 1.5, 0.5, 1.0, 100.0 ) I guess this is because of the fact that I am running a slightly older version of SU2 (3.2.7 "eagle" instead of the latest 3.2.9) I deleted these lines because the CFL_ADAPT was set to NO anyways, figuring it would have no effect. After running the simulation using the two files from Github I looked at the results, which are in the order of CL =0.053 and CD = 0.071. This is not quite what I expected for a RANS simulation on a car. I was wondering what could be wrong here..... Does anyone has a suggestion? Thanks in advance.. --Denzell |
|
![]() |
![]() |
![]() |
![]() |
#2 | |
Senior Member
Heather Kline
Join Date: Jun 2013
Posts: 309
Rep Power: 14 ![]() |
Quote:
You can check out older versions of the testcases to match with your code: look for a commit message which lists the version you want, and check out the test cases repository at that point. I can't tell from the information you give, but you may want to check whether the solution has converged. Without CFL adaptation (previously CFL ramp) it may be converging much slower and maybe reaching the maximum iterations before converging. |
||
![]() |
![]() |
![]() |
![]() |
#3 |
New Member
denzell
Join Date: Mar 2014
Posts: 12
Rep Power: 12 ![]() |
Hi Heather,
Thank you for your reply. You are right that the simulation took up to the maximum amount of iterations (=1000) and probably did not converge 'naturally' in that sense. I do have to note that both the CD and CL have been remaining stable for over 900 iterations though. After inspecting the .cfg file further I found that this specific config file has been written for the 3.2.0 release; slightly older than what I am running (3.2.7). I have changed the CFL value from 25 towards 5 to see if this has an effect on the drag and lift coefficients. It's confusing to me that a unchanged config file has this behavior.. I'll let you know if the different max CFL value has any effect. Thank you for your time! --Denzell Last edited by denzell; March 25, 2015 at 11:23. |
|
![]() |
![]() |
![]() |
![]() |
#4 |
New Member
denzell
Join Date: Mar 2014
Posts: 12
Rep Power: 12 ![]() |
Changing the CFL value did not resolve the problem...
|
|
![]() |
![]() |
![]() |
![]() |
#5 |
New Member
denzell
Join Date: Mar 2014
Posts: 12
Rep Power: 12 ![]() |
I think I resolved the problem, and believe the error to be within this line:
% Reference area for force coefficients (0 implies automatic calculation) REF_AREA= 0.677 The reference area for an Ahmed vehicle body is 0.389 x 0.288 = 0.112 square meter. Looks like the author did a sum instead of multiplication. Cheers Denzell |
|
![]() |
![]() |
![]() |
![]() |
#6 | |
Senior Member
Heather Kline
Join Date: Jun 2013
Posts: 309
Rep Power: 14 ![]() |
Quote:
If you are familiar with github, would you mind submitting a pull request with this change? That's the best way to make sure you're recorded as the person who contributed this improvement. Pull requests should be made by forking the develop branch, and submitting the pull request to the same branch. There is more information on this both on the github page and at http://su2.stanford.edu/develop.html |
||
![]() |
![]() |
![]() |
![]() |
#7 | |
New Member
denzell
Join Date: Mar 2014
Posts: 12
Rep Power: 12 ![]() |
Quote:
I'll try and look into implementing the config file modification, thanks for the advice. I'm still figuring out what the REF_AREA should be, as I have the feeling that because of the applied symmetry plane the projected area is not straightforward. The mentioned 0.22 square meter, which is the correct frontal projected area of an Ahmed body, also provides questionable results. At the moment I find that doubling the 0.22 towards 0.44 square meters yields force coefficients that are expected..... By the way, are you familiar with why the current V3.2.9 version of SU2 disabled the functionality of converting a CGNS grid towards native .SU2 format? Since I construct my meshes in CGNS I am now forced to only use a single processor.... --Denzell |
||
![]() |
![]() |
![]() |
![]() |
#8 | |
Senior Member
Heather Kline
Join Date: Jun 2013
Posts: 309
Rep Power: 14 ![]() |
Quote:
I do not know why CGNS conversion might not be working - as far as I know that capability should still be there. You can always revert to whatever version of the code works for you. I can't check it since I don't have CGNS on my system, but I think it might be possible to run a CGNS mesh in parallel as well. |
||
![]() |
![]() |
![]() |
|
|
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
Can't get good results on ahmed body research | Michael | Main CFD Forum | 8 | September 17, 2008 13:55 |
Backup results in steady simulation | Li | CFX | 0 | July 9, 2007 05:39 |
simulation results for k-w model and SST model | Li | CFX | 7 | June 29, 2007 04:19 |
Inquiry, post results on airflow over auto body | Roland | CFX | 1 | July 1, 2006 14:16 |
RANS results for LES | Li Yang | Main CFD Forum | 5 | June 17, 2002 05:34 |