Learning Forum

[LoïcS]

Hi Ansys user,

What percentage difference can be observed in results from one Ansys version to another? Let's take the case of fluent 2016 and fluent 2021.

[Rob]

2016 would have been Fluent R17 if I remember the release dates correctly. 

Depending on models, mesh etc I'd expect differences to be negligable, and we run tests to confirm this. However, as solver defaults, numerical schemes and the like are improved some differences are possible, but I would expect these to be small. Note, changes such that flow features that were previously not able to be captured would cause a larger variation. 

[LoïcS]

I am trying to rebuild a simulation from a paper. The simulation consisted of 2 rigid plates, one oscillates normally to the fixed one. A sinusoidal displacement is imposed to the plate and I collect the generated pressure. I got a gap around 5% between our 2 results. So I don't know if such variation from 2 fluent versions is possible.

[Rob]

It's possible, but you'd need to check the exact mesh and then all of the settings including solver numerics. Was there any mention of convergence criterion?