cfx fluent comparison on data simulation

[enrico1979]

Bye-bye.

I am simulating a somewhat peculiar venturi with internally an additional obstacle.
I have performed simulations with both ansys fluid dynamic software (cfx and fluent ).
the values that I get are quite discordant (actually the chromatic scale relative to the speed that in the post is created is quite similar, less than absolute values different than about 30%, upgraded with cfx).
the simulations I made both of them in 2d (although cfx must richer at boundary sym, which does not concern flunet) and in particular for flunet I did the following:
-k-epsilon (in model)
-the simulation is velocity inlet based (or so I set it)
-inlet condition pongo gauge pressure 101325 pa, and operating pressure the set equal to 0 (I recommend that I found in a tutorial on a wing profile, and that's what I kept for outlet)
-venturs and obstacle as no slip wall
-the high ends (the sides of the rectangle representing the domain) I fixed it once with stress sheer and values of 0 (so as to avoid the influence of the upper walls and simulate the free slip wall of cfx, software that as you have understood better)

then after setting the solution according to the fact for the tutorial of the wing profile of which I mentioned I launched the calculation and the comparison results say the above (their maximum absolute value, for the speed profile is about 30% different).
Can you tell me if I did something wrong in the set up of fluent?
is it inultile to ask which software you have to believe, given that the answer is surely fluent?? or both are reliable and depends on the experience of who uses the software? ?

thanks in advance for availability and good work

 

[Vmax]

Unless you're using particularly shady mesh, it doesn't come back.
in my experience, all those times that I have made comparisons between different solutors (commercial but also open source) I have always obtained differences in the order of 3-5% maximum, and this value I recommend to keep in mind in the fluent vs cfx comparison.
as you will know cfx is quite abnormal as it is a vertex-based, while fluent is cell-centered, so the same mesh is not the same discretization for both, especially when you have unstructured mesh.
the first step is to evaluate the results for an adequate resolution, possibly to climb up.
if they don't come back obviously the setup is different, unless they always hang.