Can anyone shed any light onto why running exactly the same model in 5.4 and 5.5 is giving different solutions?

Check Fluent users website for bug updates. The differences are probably due to fixes in 5.5 that are not in 5.4, ie there are bugs in 5.4.

One would think that later versions are better, but this isn't always so. To be honest, if you can't find a reason from the bug fixes, its going to be a hard slog to work out the cause. To some extent choose the one you're happy with (your validation) and stick to it. Its a never ending saga with updates, otherwise.

Greg

Cheers Greg. I couldn't find anything obvious, it could I suppose be differences in the platforms, I am running 5.4 on unix and 5.5 on NT.

Yeah - who knows!?

I know there were a few bugs - though nothing major in 5.4.18 (I think).

It depends on how different - if you're using single precision on one and double on another computer then that is likely to have an effect.

If the difference is large then it probably is something in the code - but its almost impossible to work out what.

Check if any default values have changed too...although I don't remember anything major.

I suppose, you can see how different 6 is now!!!!!!! That will give you three answers to consider!

Greg

I guess if I run it in 3 versions one of them is likely to kick out the answer I want! Not sure thats the best modelling strategy though!