Dear everyone:

I am using DPM method to simulate drops distribution in a compartment. The residuals are always oscillatory and don't converge. Could you please tell me how to make my solution convergence ? Thank you very much for your help.

I am looking forward to your reply.

With best regards,

You

I have exactly the same problem if it is any consolation!

hi

Hi you,

can you give more detail on your case : coupled or segregated solver ? two way coupled dpm calculation or not ? with stochastic particle track ? does the calculation for the continuous phase converge without particle, ...

best regards

Alain

Hi Alain,

Thank you very much for your reply. I am using segregated solver and coupled dpm calculation with stochastic particle track. The calculation for the continuous phase converges without particle. The drop diameter is 1.2 mm.

Do you provide enough information ?

best regards, You

Can it have something to do with how often you inject particles ?

Christian

Hi,

Thank you very much for your reply. I have tried inject particles every 5 to 20 iterations.

With best regards, You

Dear you

1) Drop diameter is fairly big compared to the general DPM application. So, I think, it's momentum source is fairly large. Because you use stochastic partilce tracking, small change in momentum source in local area(local cell) can cause fairly big effect. So, how about to try mean particle tracking(without stochastic option) ?

2) When I try DPM and convergence is difficult, I increase gas phase iteration, that is, 50~100 iteration per each DPM iteration. At that time, I set residual convergence criteria very low value, e.g., 1.e-05. Then, iteration is continued. After many iterations(at least 50~100 DPM injection iteration), I check every residual whether they are enough low or not. With this approach, in general(not every case), I was successful.

I hope that it might be helpful to you.

Sincerely, Jinwook

Dear you

I agree with jimwook.

Your lack of convergence is certainly do to stochastic tracking.

Note that this can also be the result of a physical phenomenon ! Trochastic tracking introduce implicitly a temporal term in the momentum equation due to fluctuation of the term of exchange.

This can be observed experimentaly in dense two phase flow such as pneumatic transport or fluidized bed.

Take a look at your momemtum source term and your flow pressure and velocity field. Your source term can fluctuate but your field can be steady. In this peculiar case your flow is "converged" even if your residuals remains high.

At last, if your flow diverge, you can also underrelax the source term coming from DPM.

Best regards

Alain

Add to previous comment :

Increasing injection number might be helpful for convergence because momentum source of each injection is relatively small with many injection number.

Sincerely, Jinwook