|
| Case Studies 9 Months, 1 Week ago | Karma: 1 | I'm new to this forum so still trying to figure things out. Is there an area where I can upload V&V cases? For example, I ran a simple freestream flow through case for sonicFoam where I initialize the interior field to a value other than freestream and would like to upload the results somewhere to document it. www.cfd-online.com/Forums/openfoam/91082...low.html#post320676. I thought it would be interesting for others, especially those with limited experience, to see what happens for such a case. Of course, this is assuming I ran it correctly. |
| | | |
|
| Re:Case Studies 9 Months ago | Karma: 4 | Hi Martin, you could provide your test case to the test harness (see the link on the left menu panel). If you're logged in you should see the upload form. This will be provided to Martin Beaudoin, who maintains the harness. Moroever, there are indeed plans to provide V&V cases as a database and public 'show room' for others. This would be more visible. Users would be asked to provide their case and some result data to assess the quality of results. The uploads would be manageable from the web. If you are ok with it, we can discuss some prerequisites here and I set that up over the next few days!? best regards, Holger |
| | | |
|
| Re:Case Studies 9 Months ago | Karma: 1 | Hi Holger If I understand the Test Harness idea correctly, this case probably is not suitable for it since I ran the sonicFoam case for many iterations. It took a long (wall clock) time. So, I don't consider it a quality assurance case. But, I'm willing to do it if the case fits into the concept. However, IMO, a V&V database would be perfect. Wow, the topic of prerequisites is complex since OpenFOAM® can do so much. But, for my case it would be good to specify the category (2D external aerodynamics), solver (sonicFOAM), viscous category (laminar), turbulence model (N/A), time stepping approach (time accurate, constant dt), time step, approximate max and min CFL number, Mach number, Reynolds number, angle of attack, boundary conditions, a description, and a means of uploading input decks, results, and images. A description field would probably be enough, rather then separate entry fields, to make web coding easier. This forum setup is enough since I can upload images and files. However, if there is a database at the back end, then separate fields would allow for more refined searches. But, more work on administrator's part. It would be good to allow people to supply comments. Again, just like this forum. Since I have not come across a user driven V&V site, I guess this will be a learning experience. It would also be nice to see a todo list of generic, simple, V&V cases. One could even provide the input decks and let users run the case on their machines and upload the results. Once the case is done, it is taken off the todo list. Sincerely, Martin |
| | | |
|
| Re:Case Studies 9 Months ago | Karma: 4 | Hi Martin, thanks for your feedback. Indeed, a user driven V&V site would be great and I do know other OSS projects, which maintain similar databases for community contributions. So we could start such a base in a breeze (the components are there). Note that there is also the documentation project on its way again, as there has been quite some user demand for it. This naturally will have some interfaces to V&V (test case data base) and to teaching (tutorial collection). However, I think I could perfectly set this up on this site to get it launched now. Would you (and maybe some more people) volunteer and test to get a comprehensive lists of needed features? Moreover, it would be great to have a number of test cases as a start base? Any volunteers? best regards, Holger Edit: Forgot to stress: Such a V&V community data base should in a second step feed the automatic test harness. Therefore, I suggest a rating and comment system. Moreover, user should be able to tag a case as 'results affirmed'. This way, the test harness (which currently is about 'CFD code integrity') could be extended towards 'CFD quality assessement'. What do you think? |
| | Last Edit: 2011/08/20 19:32 By admin. | |
|
| Re:Case Studies 9 Months ago | Karma: 1 | Hi Holgar, I'm willing to assist where I can in regards to creating V&V test cases and running them. I do have free CPU time here and there. But nothing big. I think your ideas are good. It would be good to also see a rating system that would categorize the result as either a best practice or worst practice. Something along the line of a scale between 10 and -10. Hopefully there is not a stigma against posting worst practice results. It is one of the best ways to learn. Or, maybe, an effort should be undertaken to hunt those worst case practices down. I would like to suggest that flat plate results could be a good way to start this. Along the lines of turbmodels.larc.nasa.gov/flatplate.html. I've got some, so I'll be glad to upload those. For those in academia, would it be possible to have students run some cases as homework or extra credit? It would also be good for a field to exist for the user to input the OpenFOAM® versions used. |
| | | |
|
| Re:Case Studies 9 Months ago | Karma: 4 | Hi Martin, ok... I suggest we 'collect' volunteers and ideas to get a reasonable basis for a start. So every interested party: Drop a note here and become involved! I have done some web search. IMO there are two interesting components doing the job: Both are open source and GPL. The first one looks just great and is free in its core function (I have just run the demo)! However, we would need to wait 4 weeks for the stable release. Moreover, I would need a club membership (it is partly commercial for addons). The second one looks quite professonal (fully commercial). Let's use the time and prepare for the first community-driven CFD/V&V test case database! best regards, Holger |
| | Last Edit: 2011/08/21 17:49 By admin. | |
|