convergence in fluent!!!
 

hi all

i just wanted to know that is it necessary for any simulation to converge the iterations.....

thanks Rashmi

Re: convergence in fluent!!!
 

Yes. otherwise you will not acheive a numerical solution. On top of it I would advise to look for grid independence too....monitor the variable you are looking for.. CFD without a converged and grid indeoendent solution does not make much sense. Good luck in your runs.

Re: convergence in fluent!!!
 

hi

thanks for response.i have already run my simulation for 1000 iterations so do i increase them till i get the converged solution

thanks Rashmi

Re: convergence in fluent!!!
 

Typically, yes, keep itterating till you reach "convergence". There is no magic number, but residuals in the 10e-6 range should give reasonable values.

In one case I am running, even though my residuals are all down in the 1e-6 range, a parameter that I am tracking (mass-flow throught a duct) is still changing, and hence the solution is not converged.

Re: convergence in fluent!!!
 

Obviously, what happens is some times your paramters look like they have converged, however after some hundred iterations they start to diverge or converge! Well, patience is the key here and keep iterating untill your parameter is flat.

Thanks,

Matt

Re: convergence in fluent!!!
 

i do agree with kp that a case without convergent and grid independent is not make sense.

we always need a STABLE solution.

my way to check the convergency is to monitor volocity or pressure changes at few points in the critical regions.

i always put the residual range where the case will never acheive.

once the velocity or pressure at those points are not changing.

that means your case is converged.

grid independence is to make sure the grid size won't influence the final predictions.

Re: convergence in fluent!!!
 

thanx iam new to the fluent...yes i agree but how do u select the points..or monitor the velocity or pressure...plz guide

Re: convergence in fluent!!!
 

how to select the points? common sense. select the points in the high gradient region where will change rapidly. if even that kind of region being stable, no reason other regions are not stable.

monitor velocity or pressure? Depend on what flow properties you want to investigate. If your project focus on velocity prediction, then monitoring velocity. If pressure or temperature, then monitoring pressure or temperature. Just monitor the flow properties which are important to you.