GetDP

Samuel Kvasnica kvasnica at iaee.tuwien.ac.at
Wed Oct 6 16:43:56 CEST 1999


> >
> > Do I have to set any constraints on axis r=0 when using cylindrical coordinates ?
>
> yes : b_n = 0 -> a_t = 0
>
> -> in 2D, a = 0 will do
>

Thank you very much for help, I've set this boundary and corrected spherical transformation
and now it looks much more better ! (See picture). I just wonder why that 2 segments with -I
current around the middle segment look a bit different different. But in B plot there almost no
noticable differences.


> > Z-component of vector potential A seems to be useful when displaying 2D solenoidal
> > fields. But in case
> > of permanent magnets, I need to use also scalar potential (Bperm=grad phi). Correct ?
>
> why ? you can also impose the field with a vector potential... If nu =
> mu^(-1) and hc is the coercitive field, one way to proceed is

Yes, I was just playing with idea that scalar potential and Az are better to imagine than
3D vector potential.

> b = mu (h-hc)
> h = nu b + hc
> b = curl a
> for all a'

This is clear.

> => (curl h, a') = (j, a')
> => (h, curl a') - <n x h, a'> = (j, a')
> => (nu b, curl a') + (hc, curl a') - <n x (nu b), a'> - <n x hc, a'> =
> (j, a')
> => (nu curl a, curl a') + (hc, curl a') - <n x (nu curl a), a'> - <n x
> hc, a'> = (j, a')

Hmm I'm confused. This is not clear to me. What operator is (,) ? A scalar product ?
And what means <,> , it reminds me of symbolics in quantum mechanics... Did you mean
volume and surface integrals ? What stands n for, a normal vector ?


> Symmetries are just well set boundary conditions... If the symmetry
> implies e.g. that the normal component of b is zero (perpendicular
> current), then you should set the tangential component of a (or e) to
> zero. And so on: there is no trick behind symmetries...

ok, clear.

> Well this has changed in the new version : check the user's guide. To
> plot on a 2D grid (i.e. on a plane specified by 3 points {x1,y1,z1}
> {x2,y2,z2} and {x3,y3,z3}), you shoud use
>
> Plot[ b,   OnPlane {{x1,y1,z1}{x2,y2,z2}{x3,y3,z3}} {nbpts1, nbpts2},
> File "b_phi.cut"] ;
>
> > What is the difference between OnCut and OnGrid ?

> OnCut really cuts the mesh (computes intersections)

ok, I've got version 0.7. Seems to work corectly. Now the next question:
If OnCut computes intersections (or interpolates results between mesh points ?),
does OnGrid just report values on nearest mesh point ? And what's difference
then between OnCut and OnPlane ?

I need to generate a table of values containing x,y,Br,Bz like this:

x  y  Br  Bz
0  0  ..    ..
1  0  ..    ..
.....
9  0  ..    ..
0  1  ..    ..
.....

I tried both gmsh and gnuplot formats, last one is most suitable for me.
But I don't understand what does the first field 'type' mean. Then, why the
comments like # type x[1] y[1] z[1] value[1] must be generated for each line
and why do I get B as scalar (value[1]) at the boundary and correctly
as vector (value[3]) in volume. (in the same file, it should have always 3 components).


> > What is the meaning of Depth keyword ?
>
> Recursive subdivisions of elements in the postprocessing (e.g. for use
> with high order reconstructions)

Does it mean Depth 2 would divide each mesh element into 2 and use them to generate results ?


> > And finally: do you have any experience with use of GID or SolidMesh in combination with getDP ?
>
> No. Could you send me more information ?

I was looking for other mesh generators, geometry editors and visualization tools, since gmsh is
useful
for simple 2D geometries but behaves _very_ 'alpha' and slowly. Especially rendered graphics is not
saved into a pixmap so if you drag another window over gmsh it starts to render from beginnig which
is
very inconvenient. I think the fastest solution would be to port it to Qt (which supports also
Mesa/OpenGL
widget), and get rid of programming overhead in Motif that way. It's my (maybe subjective) opinion. 2

years ago I'd be also for Motif but things have changed. Anyway I've found following software and
wanted to ask you if you have some experience with it.

 http://gid.cimne.upc.es/intro/index.html
This one is commercial, but costs only ~170Euro. Demoversion (ready to download) is free and supports

700 2D elements and up to 3000 3D.

 http://www.erc.msstate.edu/research/thrusts/grid/solid_mesh/
This one seems to be free, but not available directly.I'll have to send an email to author to get a
copy.
It has been developed for flow simulations but that is not important is think.

Btw, would it be possible to use getDP also for fluidics and fluidics-thermal simulations ?

Best Regards,

Sam

--
--------------------------------------------------------------
Mag.Samuel Kvasnica
IAEE TU-WIEN
Gusshausstrase 27-29 / 359   email: kvasnica at iaee.tuwien.ac.at
1040 Wien                    tel:   +43-1-58801-35914
Austria                      fax:   +43-1-58801-35997
--------------------------------------------------------------