ERROR - invalid argument / not a number
hi all
i'm running external aero simulation with a basic set-up (uncompressible air, stand. k-e ...). the computation is ok and seems to CV normally. but at one point, without any DV in the residuals, i get this error message: Error: > (greater-than) invalid argument (wrong type): not a number. Error Object: nan i don't understand where does it come from and if it is linked to my model, since it's working properly at the beginning... does anyone have an idea? is my file corrupted? i really need some info about that error. thanks spepsicho |
Re: ERROR - invalid argument / not a number
this usually indicates a large or infinite number such as a division by zero. might occur in a UDF if you are using one
|
Re: ERROR - invalid argument / not a number
I understand that some value is to big to be supported by the solver, but it's quite unclear why..
I don't use UDFs and especially, why would the case be OK for a quite long time and then crash like this without any apparent DV (which would clearly indicates increasing numbers) ?? I'm really stucked because I can't know what's the origin. If you have more details about the possible reasons, don't hesitate ;o)) spepsicho |
Re: ERROR - invalid argument / not a number
Maybe you should try to start your simulation a few iteration steps earlier before this error showed up (hope you use the autosave function). Then try to change one/some of your URF and start your iteration again. Sometimes it helps but I am also facing this problem and I don`t find the reason for that. Claud
|
hello,
I know your messages are old, but since I've just met this type of error, I think it can be usefull for anyone to know in what way it can occur : in my case, it's a mistake in the fluent journal file, I've defined an inlet with these lines : /define/boundary-conditions/velocity-inlet INLETthe error must be : L=0.0% !!!! the solution is to write k=0.0 and eps=0.0 : /define/boundary-conditions/velocity-inlet INLET |
I had the same problem, however, I found 2 solutions for this issue (might only work in my case, implicit density based simulation, single precision).
The described problem only occured when I ran a job with multiple cores (parallel). The message did not pop up when I ran the job in serial mode (on just one processor). Hence I ran a couple of iterations in serial mode, then switched back to parallel processing. The second solution I found was to simply switch to double precision (hence the division by a value close to zero could be resolved). Played around for quite a while with this, hope this will help others! |
Try reducing the CFL no., hopefully it will help...
|
I also had the same problem. very very high numbers (nan)
I just changed from single precision to double, and the problem was solved :)) thanks to enigma for the great advice. by the way parallel settings does not contribute to this error. |
All times are GMT -4. The time now is 04:31. |