Hello
Roman,
What I
do in this situation is to subdivide the airgap between static and moving parts
with lines that trace a contour right round the moving part. I can
then control the element size within the airgap to ensure that I get an accurate
force result. (read the part in the manual that recommends using at least
two elements between a line integral and a change in material boundary:- it
is important!) See the attached femm model that I
used to plot the force on a steel block as it moves into the gap between an
electromagnet. (I have also used other lines to subdivide the
air into different areas, so that I can control the element size where
necessary.)
These lines can then be picked in the post
processor. I use the Lua lines below in my post-processor
script;
--
obtain forces
clearcontour() seteditmode("point") ncpts = getn(cpts.x) for k = 1,ncpts do selectpoint(cpts.x[k],cpts.y[k]) end selectpoint(cpts.x[1],cpts.y[1]) fx,
fxi, fy, fyi = lineintegral(3)
The x
and y coordinates of the line end points are stored in 'cpts.x' and 'cpts.y'
which I define in my preprocessor model script, and read in from a
file at the beginning of my post-processor script. You dont actually
need to draw lines in your femm model, but you need the points there to use the
method described.
Furthermore, I use this line midway in the airgap to help define my
element size within the airgap (using the line properties). This way I ensure
that I only have really small elements where I need them (this is useful if you
have a model with a lot of travel, or a very small airgap). The
important obvious thing is to move these contour points with the moving geometry
within your Lua pre-processor script (the model attached was designed with
static contour lines but this is an exception to what I usually
do)
This works well for me, and results agree well with
physical systems (to date Femm has predicted forces to within 10-15% of the
actual measured forces for 3-4 systems that I have built, which I consider
to be excellent for 2D - FEA)
Regards,
Finlay Evans
Finlay Evans -----Original Message-----
From: Roman Beresik [mailto:beresikr@xxxxxxx] Sent: 24 July 2002 14:26 To: femm@xxxxxxxxxxxxxxx Subject: [femm] Define contour
|
Attachment:
bin00054.bin
Description: Binary data