Hi All,
I am also trying to use the ConductionFiniteDifference method in one test case I am running. Is there a way to know the error tolerance which is used to control convergence when using this method? Is the error tolerance applied to all variables or to a specific variable? If the latter, then I would like to know the variable which is used as well as the error tolerance value.
Thanks a lot!
Thierry
The conduction finite difference model has an inner loop, for each surface, and an outer loop for all the inside face surfaces (which can interact with each other by radiation).
If you are asking about the inner Gauss-Sidel loop, then the convergence criteria looks like this
sum( SurfaceFD( Surf ).TDT - SurfaceFD( Surf ).TDTLast ) / sum( SurfaceFD( Surf ).TDT ) ) < 0.00001
Which is looking at normalized temperature results for a surface from one GS iteration to the next for all the nodes in that surface.
The outer loop is controlled by the user input field called Inside Face Surface Temperature Convergence Criteria in the HeatBalanceSettings:ConductionFiniteDifference object, with a default of 0.002. This is looking at how the results change for the temperature right at the interface between zone air and the surface. It will recalculate all the surfaces until none change more than the criteria.
Thanks a lot for the detailed answer! That helps a lot. Thierry
That's a comment, not an answer. :) And as thanks, you should upvote! :) :)
Asked: 2014-11-10 16:46:03 -0600
Seen: 359 times
Last updated: Nov 11 '14
