[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Multiple group membership for nodes
--- In femm@xxxxxxxxxxxxxxx, "niallcaldwell" <niall.caldwell@xxxx> wrote:
> Firstly, many thanks to David for his advice over the slotted
> armature problem I had. I was not aware that the "lamination" feature
> essentially did this BH scaling that I was trying to do. The point
> about radial gap is well taken- this is the case in most of my
> designs. I presume that, if the material is working in the linear
> portion of its BH curve, and the gap is small in comparison with the
> major dimensions of the armature and the slot circumferential width,
> then a good approximation to obtain the same gap reluctance is to
> increase the radial gap by the same ratio as the "lamination factor"
> at the outside of the armature?
It might be reasonable to fudge the gap width in this circumstance.
However, I'd suggest using Carter's Coefficient as a guide for picking
the effective air gap. I'll have to look up the formula--I don't
remember it off the top of my head. Anyone have a good web reference
for Carter's Coefficient.
> One other point which has come up. I often want to parameterise
> models so that I can sweep through a design space to find an optimum
> of two or more variables. I want to use Lua with nested "for" loops
> to do this, for instance generating armature force for a range of
> diameters and lengths of the armature.
>
> To do this from the starting point of existing geometry I need to
> select nodes by group in Lua- all the nodes on the outside diameter
> are in group 1, while all the nodes at the top end are in group 2.
> The trouble is, there is always at least one node which wants to be a
> member of both groups (ie the outer top corner). Is there any way to
> assign multiple groups to entities?
Well, you can't assign multiple groups, but you could assign a
different group number for the elements that are supposed to be in
more than one group (e.g. group number 21 for elements that are
supposed to be in both groups 2 and 1 or something...). Then, you can do
selectgroup(1)
selectgroup(21)
to get group1 plus the shared section. When you do multiple
selectgroups, the previously selected group doesn't get unselected.
> The other option is to build up the entire problem in the lua script,
> bypassing the GUI, creating keypoints with coordinates calculated by
> functions of defined parameters. That way individual nodes
> coordinates can be a function of many parameters. This is the most
> powerful and general way of parameterising, and the way I used to
> work with Ansys becuase its GUI is such a dog compared to femm's, but
> really.... life is too short!
>
> BTW this same nested loop technique can be used to generate lookup
> tables of force and inductance at a range of currents and armature
> positions, for export to dynamic simulation packages (AmeSim, Easy5,
> Simulink) This allows the complete dynamic simulation of a solenoid
> taking into account damping, spring, inertia, colomb friction,
> electric circuit switching... I have heard that transient phenomena
> eg. eddy currents can be taken into account by running a frequency
> analysis in FEMM and inserting parasitic lumped parameters into the
> dynamic model.
>
> Niall