We know that the relaxation factors are very significant in the closure equation. Although during interaction with some researchers, I know that they usually modify the relaxation factor to get the proper simulation results as well while going through Ansys coding documentation. I know that they set up the optimum relaxation factors which mostly handle all kindsof simulations, as well as ones that guide suggested changes in the relaxation factors according to the turbulence phenomena, to get a stable solution.
When I went through some published research papers (enclosed one paper here please refer to it ), I came to know that the relaxation factors are the effect the turbulence flow phenomena during numerical simulation.
This confuses me because both are different and violate each other.
This confusion births the curiosities in my mind which are:
What is the main importance of these factors in basic CFD equations?
Is this the right way to modify that factor (although research shows this produces an error)?
If there is an error produced during the numerical simulation of turbulence flow then how can we overcome this error, and how can we get the suitable value of these factors for a particular flow condition?
Thanks for your reply ! i know about the relaxation factors but ones i came across some published results as i mentioned, that leads to create the confusion in my mind so i just want to discuss it with researches over here and i am happy to get away my confusion.
This is to Sargon A Gabriel and all others in the discussion.
As we all know that in fluent we provide the number of iterations per time step suppose by default 20 max and absolute convergence criteria of 0.001. Suppose the solution is not converging after 10 time steps and residual starts oscillating heavily, so i reduce the under relaxation factors. now what i see is that residuals solution becomes stable and started decreasing and after the 20 time steps, resuduals for all variables goes below the specified convergence criteria.
My observation is 1) residuals reaches convergence quickly with reduced underrelaxation factors.
2) As one of you said, reduced under relaxation factors takes more iteration to converge. I did not see that. it takes same number of 20 iterations per timestep and reaches convergence in 20 time steps. With out low URF residuals dont even reach convergence within 20 iterations. why?
3) if you take low URF, you basically take low residuals, and consequently you reach your convergence criteria quickly.
So how it takes large no of iterations?
Thanks
Vivek
- Badges
- Science topic