CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Indentify whether convergence through post-processing pictures (

Kiffer September 4, 2012 02:16

Indentify whether convergence through post-processing pictures
I wonder are there some good ways to identify whether a simulation has reached good convergence through post-processing pictures. For example, if I look others' results, through the contour plot , the velocity and pressure trends looks
correct but how to evalute its accuracy and convergence based on this? are there some tricks? Thanks.

cfdnewbie September 4, 2012 03:10

I would strongly advise against this, as there are so many tricks to lie (to yourself and others) with pretty pictures... you could easily convince yourself that the lines are indeed identical, while there is a significant error you just can't detect with your axis settings, line width etc...

Especially with contour plots of 2 or 3-dim data it is impossible to really "see" accuracy or convergence. The only scenario under which I would accept something like this as a hint to accuracy or convergence is in the case where no other reference data was available, the physics of the system and or the numerics are not fully clear and all you have might be a plot or some lines from an experimentalist or some paper...

Why not stick with the numbers? They give you a much sharper and clearer indicator (and people won't frown at you at your next conference :) )

FMDenaro September 4, 2012 05:34

I totally agree ... convergence can be studied only by proper measure of the error ...
People often joke about contour plot, calling CFD -> Colour Fluid Dynamics

All times are GMT -4. The time now is 16:33.