Different Post-Processor results by changing mesh refinement


#1

Hello together,
I’m doing for the workshop “Heat Distribution within the Extruder” the first homework session and I’ve got some different results which I don’t understand hundert percent:
For the first run, I gain a unsymmetrical heat distribution inside the extruder. But I compared all parametrics with the Workshop Video and with paper.
Then in my second run, I just changed in the mesh refinement the maximum and the minimum element edge length from 0.0034 to 0.00034 (maximum) and from 0.00034 to 0.000034 (minimum).
Then I simulated the whole thing with the same values, and now I’m getting a symmetric heat distribution.
For me, thats super weird, because I “just” changed the refinements, not a special value.
Plus I’m getting different maximum temperature values, as can be seen in the temperature legend.


Unsymmetrical Heat Distribution - Run 1


Symmetrical Heat Distribution - Run 2

I hope, someone can help me :slight_smile:

Regards,
Ulli


#2

Hi @ulrich_hartman,

Yes a change of your refinement will result in a finer mesh that allows better accuracy even if your starting parameters are the same.

Think of your mesh as a kind of function that gets more accurate each time an input is put in and after running through the function, the answer gets re-input back into function again to obtain another more accurate result.

Here is a small illustration:
Iteration 1: Input > function > Answer1
Iteration 2: Answer 1 > function > Answer2

As mentioned Answer2 will be more accurate than Answer1 because it has already run through your function once already. That is what a mesh refinement basically does, it allows more “iterations” with your results getting more and more accurate the finer it is. That is why you will see a “nicer” heat distribution for the refined run (run 2) versus the more coarse and inaccurate run (run1).

Hope I’ve explained it well enough!

Regards,
Barry


#3

Hi @Get_Barried, @ulrich_hartman, Yes absolutely we expect better results the finer the mesh. However, in this case it looks like no heat transfer has taken place in areas, which to me said that there was actually no contact between the surfaces. So I zoomed in and it looks like the coarseness of the mesh meant that the surfaces didn’t properly contact, thus not transferring any heat.

You can just see that on the snip below:

Was the original mesh set to the workshop settings?

Many thanks,
Darren


#4

Thank you for the quick response.
However for me it’s still surprising, that a change of tenth in mesh refinement has a changing max temperature of over 6K.
I already passed some numerical fluid dynamic courses in university, so I’m not this new to simulation. :slight_smile:

My main problem is the unsymmetrical temperature distribution.
The settings were in both cases the same. That’s the surprising thing…
The original mesh were set to the workshop settings

Quick explanation, how i got to the symmetric distrubition:
I got the wrong result (unsymmetric), so I compared all settings with the workshop settings. But I found no difference, ecept of the mesh refinemnts. So i changed this one a tenth down and adapted the settings to the new mesh and got the right distribution.

Greetings,
Ulli