questions on transient simulation
 

Hi everyone,

I have a case that doesn't converge in a steady simulation and so I have to make a transient simulation. But after I read the chapter about time-dependent calculations (users giude) I still have some questions and I hope you can help me.

1. the default value for the "max iterations/time step" is 20. My question is that can't it be sometimes more, e.g 50 , or doesn't it depend on the case? Because sometimes my soultion converges after 25-30 iterations and I think that a value of 20 is too strict. What is your opinion, should I take the default value (20 iterations) and trust fluent's users guide, or can I increase this value? My time step is very small (delta_t=1.7e-06).

2. I want to simulate a discrete phase with my transient continious phase. Do I have to simulate the discrete phase as transient too, and do I have to set the same time step for the continous phase (flow) and the discrete phase?

3. everytime I want to change the time step, the calculation begins to iteratate from the first iterataion (all number of time steps again). How can I make small changes in my case setup and just go on with my iterations and not having to iterate all 1000 iterations again?

Thanks in advance for help.

regards,
spring

Hi, Spring,
1 Every time step the residuals jump and then tenuate in the following iterations. The number of max itertions per time steps should be so as to let the residuals drop three orders of magnitude. If it is not, it is not recommended to increase this number, but to decrease the time step.
The time step is not evaluated based on its absolute size, but is according to the characteristic time of the flow phenomenon.

2 Though not obligatory, it is recommended to simulate the discrete phase as transient if contiuous phase in transient calculated. Again the time step for particle tracking is determined by the characteristic time of the particle phenomenon, and it is not necessary same as the for continuous phase.

3 This is not a problem. Every time you restart the iteration is just a continuation to the last step before the the interruption.

i hope these will help but not misleading.

Hi xrs333,

thank you very much for your helpful reply. My problem is that I don't have any experiance in transient simulation and that my simulation won't converge no matter what I do :confused:. So I have other questions on transient simulations.

1. in the users guide the PISO solver is recommended for transient simulations (especially for large time steps) and the SIMPLE/C solver is recommended for transient LES simulations with (because of their small time steps). Well I think that a time step in the range of 1.xxxe-06 is small but my model is laminar. Should I PISO anyway?

2. I choose a convergence criteria to three ordes of magnitude, but I "relative" not absolut, because I think absolute doesn't make that sense, right?

regards,
spring