[Getdp] Second order element peculiarity

John_V jvillar.john at gmail.com
Tue May 25 00:56:05 CEST 2010


While searching for an answer to my problem getting a 2nd order 
electrostatic calculation to work 
(http://geuz.org/pipermail/getdp/2010/001304.html), I ran across this 
now several year old thread 
(http://www.geuz.org/pipermail/getdp/2006/000868.html) that bears on the 
problem.

All parties seemed satisfied with the solution, but I have run it 
myself, and I am not so sure. When I run it, it appears to me that the 
solution produced there is not second order. Here is my test: Upon 
completion I open the options, go to View[0], select Iso-values with 101 
intervals. This plots contours of constant potential from 0 to 100 V 
with 1 V increments. Now zoom in on any of the elements in the solution. 
What I observe is that the spacing between contours is constant within 
elements. This is characteristic of a constant slope, i.e., the 
potential is linear with position within each element. For this problem 
the slope should decrease (contour spacing increase) with increasing 
radius (i.e., to the right). It does, but the changes are all abrupt 
ones only at element boundaries. A second order solution should be able 
to describe slope changes within an element.

So why don't I see that? Is it something about my installation, or do 
others get the same?

John