CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Automate solving with slight changes (http://www.cfd-online.com/Forums/fluent/49152-automate-solving-slight-changes.html)

Kai Wohlfahrt September 1, 2008 07:37

Automate solving with slight changes
 
I'm using fluent for a physics assignment, and testing different wing profiles at different angles of attack, and I have many similar set ups (the same wing, only the direction of the flow changes a little in each).

I'd like to know if i can set fluent to save the result, adjust the angle (by say +5 degrees for a range of 30 degrees-so six variations) of the flow, and then solve again as soon as one has converged.

Many thanks,

Kai

Jason B September 3, 2008 11:48

Re: Automate solving with slight changes
 
Use a journal. Open one, and then start a journal (file->write->start journal). Then set it up, and when you're done setting it up, stop the journal (file->write->stop journal). Then you can use this journal as a starting point and copy/modify whatever you want. Do them all in the same journal. Look at the TUI commands, as well as the part of the manual on journals, both of these are available online.

Be careful of how you're judging convergence. Especially as your AoA gets steep. I highly recommend monitoring the Cl and Cd (Reports->Monitors->Forces).

Hope this helps, and good luck, Jason

Kai Wohlfahrt September 4, 2008 09:30

Re: Automate solving with slight changes
 
Ok, I'll try it out when I get my hands on the school computers over the weekend.

For convergence I suppose I'll have to use residuals and set a max number of iterations, and do any with problems over again the next session, because I don't think Fluent can monitor forces automatically to find convergence (please correct me if I'm wrong of course!).

Jason B September 4, 2008 14:45

Re: Automate solving with slight changes
 
You're correct, which is why I pointed it out. You'll just have to look over your cases and see which ones converged and which weren't. And I wouldn't rely on the residuals by themselves, so even if the residuals get below 1e-3 (or whatever you set) I would still check the forces to make sure they've resolved.

Also if you've got the computational time, you may want to consider smaller deviations than 5 degrees. Especially when you get near the region where the flow wants to separate.

Good luck, and hope this helps, Jason

Will September 4, 2008 17:59

Re: Automate solving with slight changes
 
Fluent can use custom monitors to check for convergence via a scheme file. You can either write your own if you know how or use the one fluent provides in their support site if you have an account. Just search for "custom residual monitor" in the search cases and solutions field.

Kai Wohlfahrt September 5, 2008 12:29

Q2: Setting up parallel Fluent on a network domain
 
Ok, I'll add a second question to my first, are there any special steps i need to take to run fluent on a domain type network?

I have MPICH2 (not from the ANSYS site though, do I need to get the version from there?), and I belive I need to install it on all the computers I intend to run Fluent on.

Then get the name of the computers from our ICT manager, log on to each computer (after installation of MPICH2 do I need an administrator account?), enter the names in my host file so it will look like

Homelaptop School-1 School-1 School-2 School-2 ...

and then run fluent (partition, solve, adapt etc... this part shouldn't be so much of a problem I hope).

Sorry for pestering you with this simple kind of question, but I can't find any specific guide in the docs and I won't have much time to play around with settings.

Kai


All times are GMT -4. The time now is 07:46.