Re: NEC-LIST: What is the best way to model a complex surface?

Abraham Rubinstein,
Maybe I can help. It sound like a very big piece of simulation, the real
question is maybe about how close are the sides (wire frame) that made up
the patches, this is about singularities and such likes, if you have lots
of long and narrow triangular patches, the simulation may be error prone.
Provided you have not violated other rules, a rule of thumb ratio of 30
times between wire radius and length will be sufficient. So for 1 cm
wires, a radius of 0.33 mm will do. However, the length of any segment
of a wire can be reduced to its radius if the wire segment is part of a
piecewise model of a larger structure eg. as in a spiral or solenoid. :-)

 Here is what I suppose is some of the usual rules which I compile from
somewhere some years ago. . . .
The length, l to width, w ratio of any one patch should not exceed 30.
That is, the patch cannot be too long and narrow.
l/w < 30

The segments in the vicinity of fine features have to be segmented to
account for the expected normal of the Electric field.
A quasi static model of the Electric field distribution would help e.g.
sharp corners will lead to finer discretization.
This has to be done with proper judgement from experience to create nearly
constant blocks of charges in each patch.
This point is very important when it comes to discretisation at feed
points, a possible clue about insufficient segmentation is the real part of
the input impedance going negative. Structural parts far away from the
source can be discretized larger, perhaps going down to 8 segments per
wavelength.

BTW, how many segments are there in your model for the car?

Tim Foo,
(or just call me 'Tim')
                                                E-mail: fwj_at_np.edu.sg
ECE, School of Engineering,
http://www.np.edu.sg/ece/ Tel: + 65 460 6143
Ngee Ann Polytechnic, Fax: + 65 467 1730
535 Clementi Road,
Singapore 599489

|---------+---------------------------->
| | "Abraham |
| | Rubinstein" |
| | <abraham.rubinste|
| | in_at_epfl.ch> |
| | Sent by: |
| | nec-list-admin_at_gw|
| | eep.ca |
| | |
| | |
| | 01/29/02 07:21 PM|
| | |
|---------+---------------------------->
>-----------------------------------------------------------------------------------------------------------------|
  | |
  | To: <nec-list_at_gweep.ca> |
  | cc: (bcc: Wan Juang Foo/ece/staff/npnet) |
  | Subject: NEC-LIST: What is the best way to model a complex surface? |
>-----------------------------------------------------------------------------------------------------------------|

Ok, here is my problem:

I have CAD data which represents a vehicle. The mesh is made of triangular
pieces that conform the surfaces of the whole geometry. I've been able to
convert the sides of the triangles into GW structures of 1 segment each so
that this information is now a NEC file. The shortest segment is about 1 cm
long and the longest goes up to 14.5 cm. I can't really cut the longer
wires
into shorter segments since this will increase the total number of segments
to a figure impossible to run.

What should be the radius to assign to these segments taking into account
the lengths?

Has anyone done this before with success (reading CAD data not originally
conceived as a NEC input file and obtaining good results)?

Thanks in advance,
-----------------------------------------------------------------
Abraham Rubinstein
Power Systems Laboratory
Swiss Federal Institute of Technology
EPFL-DE-LRE
CH-1015-Lausanne
Tel: (+41 21) 693 2667
E-mail: abraham.rubinstein_at_epfl.ch
http://emcwww.epfl.ch
-----------------------------------------------------------------

--
The NEC-List mailing list <nec-list_at_gweep.ca>
http://www.gweep.ca/mailman/listinfo.cgi/nec-list
-- 
The NEC-List mailing list <nec-list_at_gweep.ca>
http://www.gweep.ca/mailman/listinfo.cgi/nec-list