solver tolerance in fvSolution
 

Hello,

i found a lot of posts in the forum where some solver tolerances within openFOAM are suggested as accurate. For example in the fvSolution file for solving the p-equation is "tolerance 1e-6", "1e-4" or "1e-8".
I wanted to figure out how this tolerance number is calculated to estimate a number for my simulations, but i can't get it out of the code:confused:. Is there any explanation what this residual number / norm means or which norm is taken from the residual vector!?

A little example would be very nice (for example in relative error if i would have only one cell ... or something else)

Thanks a lot.
rupert

Well,

the relTol is the relative tolerance between the initial and the final residual. Both are reported by the solver:
GAMG: Solving for p, Initial residual = 0.0334352, Final residual = 0.000941545, No Iterations 17

if you specify a relTol of 0.1 the solver will stop iterating if the final residual falls below 0.1*0.0334352=0.00334352.
The absolute tolerance you refer to with "tolerance" in fvSolution is the value the Final residual need to fall below.

Regards Bastian

Thanks for the reply Bastian,

but the question is still how the (total) residuum is calculated or what it means..
Do you know that?

Please check here: http://www.cfd-online.com/Forums/ope...d-solvers.html

Google knows ;-)

thank you alberto

tolerance relTol question
 

Thanks about this post, But I have 2 question.

1- O.F uses relTol or tolerance for stop iteration?
2- What is difference between tolerance in for example "U" and "UFinal"?

Thanks.

I think both of them are used. Both conditions have to be respected.
(as read in this article).

tolerance relTol question
 

Hi amin_jalalian, I'm a bit late, but a new user might appreciate the answer. About your second question, I believe UFinal is used for transient algorithms for the final iteration of each inner loop (inside the same timestep). Not 100% sure, though.