[Gmsh] integration, spikes

[t.eschner at dokom.net]

  Thank you very much (I am very late, but I hope not too late).

  After a forced pause I made a new attempt and found that a naive cut  
produced a quite distorted mesh that caused numeric problems, I  
assume. I am happier with plugin(CutGrid) now.

  After fixing the parameters, plugin(Intergrate) started to provide  
values with actually sufficient accuracy, but I will switch to global  
quantities later, as you suggested.

  Kind regards, Thomas

Zitat von Christophe Geuzaine <cgeuzaine at uliege.be>:

>> On 5 Aug 2020, at 10:12, t.eschner at dokom.net wrote:
>>
>> Dear all,
>>
>> trying to get acquainted with gmsh/getdp, I stumbled upon two  
>> problems that I am not able to resolve.
>>
>> I am trying to model a simple conductive cell with two electrodes,  
>> a simple one and a structured one. To me, the results look quite  
>> realistic so far (I am still in the stage of checking), with two  
>> exceptions:
>>
>> The first: When I try to calculate the overall current by calling  
>> plugin(Integrate) on views slightly above the simple electrode, I  
>> get two different results (for the vector view -8,88 and the  
>> x-component view -39,9), and neither is close enough to the  
>> expected value of -2x(4,94..4,99).
>
> Be careful with brute-force integration. If you use GetDP would  
> recommend using global quantities, which define such integrals in  
> the "variationally correct" way: see e.g.  
> https://gitlab.onelab.info/doc/tutorials/-/wikis/Electrostatics-with-floating-potentials
>
>> The second: The current density field has few spiky values that I  
>> do not know how to cure. Their position seems to be constant, the  
>> values seem to change with the grid parameters, and it might be  
>> getdp-related.
>
> With a (scalar) potential formulation your current density is  
> defined through the gradient, whose accuracy will depend quite  
> strongly on the mesh quality. You can give the new "HXT" 3D  
> algorithm a try, whose optimizer produces slightly better elements,  
> and see if this improves things.
>
> Christophe
>
>> Please, have a look and give me a hint on how to correct/proceed. Thank you.
>>
>> Kind regards, Thomas
>>
>> <porous_electrode.zip>_______________________________________________
>> gmsh mailing list
>> gmsh at onelab.info
>> http://onelab.info/mailman/listinfo/gmsh
>
> —
> Prof. Christophe Geuzaine
> University of Liege, Electrical Engineering and Computer  
> Sciencehttp://www.montefiore.ulg.ac.be/~geuzaine
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://onelab.info/pipermail/gmsh/attachments/20201026/6ea263ba/attachment.html>