What does "tolerance" stand for?

Nene Lee · May 9, 2016, 04:50

Hi foamers!
Two issues, which have been puzzled me a lot, i.e.,
1. What dose the entry 'tolerance' in fvSolution::solvers of every equations stand for? I mean that, for U equation, if the tolerance 1e-5 is specified and the final iteration can achieve a residual below 1e-5, what is the maximal relative error for U?

2. I find nearly all the toelerances in tutorials of OpenFOAM are taken as 1e-5~1e-8. Is this a general rule for every case? Are there any general rules to determine the tolerances?

Can anybody give me any tips? Thanks in Advance!

All the best!

Bazinga · May 9, 2016, 10:56

1. "tolerance" describes the global, normalized residual of the matrices A phi = b for each transport equation. From the residuals you can not conclude to the error of your flow value at the various positions in your domain because other errors such as the modeling error are not considered.

2. It depends on your simulation - I would recommend doing your own analysis by changing the residuals validate the solution to see the impact on your results.

May 9, 2016, 04:50	What does "tolerance" stand for?	#1
Nene Lee New Member Nene Lee Join Date: Mar 2015 Location: Jinan, China Posts: 2 Rep Power: 0	Hi foamers! Two issues, which have been puzzled me a lot, i.e., 1. What dose the entry 'tolerance' in fvSolution::solvers of every equations stand for? I mean that, for U equation, if the tolerance 1e-5 is specified and the final iteration can achieve a residual below 1e-5, what is the maximal relative error for U? 2. I find nearly all the toelerances in tutorials of OpenFOAM are taken as 1e-5~1e-8. Is this a general rule for every case? Are there any general rules to determine the tolerances? Can anybody give me any tips? Thanks in Advance! All the best!

May 9, 2016, 10:56		#2
Bazinga Senior Member Join Date: Jun 2012 Location: Germany, Bochum Posts: 230 Rep Power: 15	1. "tolerance" describes the global, normalized residual of the matrices A phi = b for each transport equation. From the residuals you can not conclude to the error of your flow value at the various positions in your domain because other errors such as the modeling error are not considered. 2. It depends on your simulation - I would recommend doing your own analysis by changing the residuals validate the solution to see the impact on your results.