[Getdp] Second order element peculiarity

David Colignon David.Colignon at ulg.ac.be
Thu Aug 24 16:19:54 CEST 2006


Hi,

the condition on the coefficients associated to the second order basis functions must be homegeneous ( 0 value ) otherwise the imposed condition considers a boundary condition other than constant ( with second degree variation ). This is due to the fact that the second order elements are built with hierarchical basis functions.

So with 

{ Name ElectricScalarPotential_d2 ; Type Assign ;
    Case {
      { Region Ground; Value 0. ; }
      { Region electrode ; Value 0. ; }
    }
  }

in con_cyl.pro it works very well

Cheers,

Dave

-- 
David Colignon, Ph.D.
ELAP - Service d'Electricité Appliquée
Institut Montefiore B28
Université de Liège
4000 Liège - BELGIQUE
Tél: +32 (0)4 366 37 32
Fax: +32 (0)4 366 29 10
http://elap.montefiore.ulg.ac.be



Jim Davis wrote:
> Hello getdp aficionados,
> 
> I am having trouble using second order elements in the solution of an 
> electrostatics problem and I was hoping someone could tell me what I am 
> doing wrong.  It seems that when I have a region surrounded by 
> constrained surfaces (Dirichlet and homogeneous Neumann) the solution 
> getdp gives me for second order elements contains nodal potentials 
> higher than any of the constrained potentials in the grid.
> 
> As an example, I have attached the input files for a concentric cylinder 
> problem.  The inner radius is 0.1 and is set to 0 volts while the outer 
> radius is 0.5 and is set to 100 v.  The endcaps have no boundary 
> conditions set and therefore default to the natural b.c. (homogeneous 
> Neumann), right? Looking at the solution generated by getdp, there is a 
> distinct rise in potential (117 v) prior to the outer radius.  If I use 
> linear elements, I do not see this behavior.  That is, the potential 
> increases monotonically from the inner radius to the outer radius as 
> expected for linear elements.
> 
> In going to second order elements, I took the following steps:
> 
>              1.  Increased the number of Gaussian integration points for 
> triangles from 4 to 6.
>              2.  Increased the fill-in parameter in solver.par to cover 
> the max bandwidth
>              3.  Added the basis functions associated with element edges 
> and added the corresponding constraints.
> 
> What am I missing?
> 
> Thanks very much for any insight you can provide on this problem.
> 
> 
> Jim
> 
> 
> ---------------------------------------------------------------------------- 
> 
> 
> Jim Davis                              Telephone: (404) 894-7231
> Signature Technology Laboratory              Fax: (404) 894-8515
> Georgia Tech Research Institute            email: jd4 at prism.gatech.edu
> Georgia Institute of Technology
> Atlanta, GA  30332-0800
> 
> 
> ------------------------------------------------------------------------
>