Fluctuating error when doing convergence study - what is the cause?
 

Hi,

Firstly, new to CFD so please keep that in mind!

I am doing a mesh convergence study. Apart from ofc looking at residuals I have also been monitoring a surface force and a clearance flow which are both points of interest in my analysis.
I've refined my mesh in about seven steps and computed the error for each step by finding the percentage difference in magnitude by comparing all to the magnitude of the finest step.
The general trend is that the error is decreasing before reaching 'zero', i.e. when the finest mesh (last step) is compared to itself.
When the error drops bellow 1% I see small fluctuations in the error though (for surface force) and the error is not steadily decreasing but jumping a bit. Not back above 1% though.
I tried to create two meshes with almost the same amount of cells and they give quite different errors despite having almost identical mesh.

Is this to worry about, and is there something smart to be said about this fluctuation and its cause?

I've tried to attach the error plot, hope you can see it.
Thanks!

First of all, what kind of problem are you solving?
Then, if you assume a solution to be "exact" on a fine grid, that does not mean it is actually the exact solution. Thus, oscillations can appear. I suggest to read this issue in the textbook of Peric & Ferziger.