Convergence and Design Verification with CFDesign

[sahinoz]

Hi,
I am new to CFDesign and was running some trial analyses on my computer the other day. Basically I was trying to compare two different valve geometries and measure the cfm. However, my results never converged. Called tech support and was told that results do not need to converge unless I am running the analysis to match lab datas, I am not trying to match lab data here, however, how do I trust a result that doesn't converge and compare it to another? With my solid FEA background, this didn't sound right at all. What do you all think about it?

 

[btrueblood]

Do the problems converge in inviscid mode?

Does the mass flux in/out balance on a time-averaged basis in viscid mode? What turbulence model are you using?

Remember you are solving a time-stepping equation, and the answer may be that large-scale vorticity causes the flow to fluctuate across the outlet (I assume here you have a fixed mass flux or fixed velocity at the inlet). Try modelling a long pipe on the outlet, to allow the flow to become a "fully-developed" pipe flow, this should reduce this problem, and you should see better convergence.