Help with parametric studies
Hi, I've been trying to use workbench to automate a parametric case study on Fluent.
I have a pipe and want to cover a range of Reynolds numbers, so I defined the velocity in my inlet as a parameter. After (On the Parameters cell on Workbench) I set some design points with the values I want the velocity parameter to take. After this I ran the simulation updating all design points.
I noticed my "base" case would take around 80 iterations to converge (a small time) and each design point would take a much larger time to update. So I opened the cases for each design point and when I plotted the residuals I noticed that each design point had run 1000 iterations (value I had set as maximum number of iterations), and the residuals had drop to my convergence target, but the simulations kept going and the residuals went up again to very large values. I checked that I used the convergence criteria to stop calculation.
I have no idea why the simulations didn't stop when the convergence criteria had been reached. I would be very grateful if anyone could help me on this matter :)
If the residuals rise again after dropping below your convergence target, the solution cannot be considered converged somewhere in between.
There must be an error in your setup.
As far as your original question is concerned: Are you sure that every residuum dropped below its unique limit?
The default values in fluent are not all the same.
Can you share an image of your plots?
I would not depend on residuals to judge the convergence. It is wise to create the monitor plots at appropriate places and ensure they flatten, before declaring convergence.
I've been working around this problem, and finally reached a solution.
Sorry for some lack of info that may have caused confusion. As for convergence I did set up the appropriate monitors it was just the residuals that were not satisfactory.
What I learned:
After watching the properties of the Fluent Solution cell, I noticed that although it was set to inherit setting from the Setup cell, it was not applying the same setting (I don't know why). For example, I was setting the solver for double precision and the Solution cell appeared as single precision. I'm not really sure as why this had an impact but the fact is that after changing setting on the solution cell to manual an enabling double precision the simulation ran as expected; using less than 100 iterations and converging with residuals below 10e-4 (also with the other monitors of velocity/mass flow flat).
Just wanted to share my solution and what I learned in case other run into this sort of problems :)
And thank you for your answers and help :)
how to initilize
well I have a much more general question.. how do I start autoamted parametric study??
I have 3 valves and there deflection are the design parameters and I´ve already run a static simulations.. now I want workbench to keep going with the next design point.. Is that just going to happen if fluent reached "convergence" and not the max. iteration steps. Or is there anything else I should remember? I am really clueless right now and would appreciate any help.
PS sorry for the language difference in the pic, but i think the general setup should be clear
|All times are GMT -4. The time now is 15:48.|