how i can be sure i reached convergence?

[Anonymous]

Hi all,

i have some problem with a simulation, i run at the beginning only 100 iterations just to have an idea if it works properly or if there were some big problmes in the geometry. and i noticed that the convergence plot reached horizontal line on pressure, TKE, TED and was near to horizontal for Vy, V, instead Vx was reducing.

my structure is an air duct with an inlet and 4 section of outlet going down 

you can find the results with vector on the picture.

But when i run more iteration the behaviour change a lot and i didn't reached convergence but it seems i get worst results. i decided to run 5000 iteration but the system stop at 2745 without any message in the second run 

what can be the problem? the results in the #100 iteration can be considered correct or not?

[Anonymous]

i couldn't load the last 2 pictures. you can see that in the first case i have a recirculation in front of the inlet, instead in the second one i have something like an inlet instead of an outlet. using the traces it seems that the air that flow through inlet go only on the right side, instead the other section are reached by the air that flow thanks to this "new" inlet

[Jon.Wilde]

If your outlets are short enough to cause recirculation over them, this will happen. You likely need longer outlets.

As a first guess without seeing the model 😉

[matt.bemis]

Hi @Anonymous,

Thanks for reaching out on the forums! 100 iterations is not enough to get good results. That being said, if you don't have "good results" at 2000+ iterations, then more iterations will probably not help. 

I would start by using the default convergence and setting the simulation for somewhere between 500-1000 iterations. I personally like to see if the Autodesk CFD will auto-detect convergence during that period. If the residuals flatten and meet a certain predefined criteria then the simulation will stop iterating. You will see all this information in the status file. 

As Jon said, there is probably recirculation across the inlets/outlets. If this is the case then the simulation will NEVER converge. I like to check the mass/volume flow rate on the inlet/outlets and see if they match. The best way to do this is probably with a plane at the inlet/outlet and then use the bulk calculator. You should be able to get them almost identical. 

Thanks,

[Anonymous]

Ok, thanks for your suggestions.

i increased the outlet region as suggested by Jon and then i run for 1000 iterations. the software stop at 632 reaching convergence and seems better now.

i still have a problem, inlet and outlet mass/volume flow rate doesn't match. i have recirculation inside the duct.

Hope you have some suggestions to solve also this problem

[Jon.Wilde]

Do you have recirculation over the inlet or outlet though? If you do, they won't match 🙂

[matt.bemis]

hi @Anonymous,

How much is the mass flow rate off by? Is it within 1%? 5%? Chances are there's recirculation across an inlet/outlet. It's a good sign that the simulation auto converged but it is important that there is not recirculation.

Thanks,