Re: various induction machine questions

[David Meeker <dmeeker@xxxxxxxx>]

Jabulani Bembe wrote:
> Hi Guys,
>
> Iam trying to model  a three phase induction motor and following the
> example that David Meeker wrote. My motor is a 4pole machine and the
> stator has 48 slots and the rotor 36 slots. I followed the example but
> still the program is not running. If I use circuit property to define my
> stator current do I have to use the LUA SCRIPT files and do I have to
> difine bounderies for stator and rotor slots.
You don't have to use the lua script to define the currents via circuit properties.
 The idea with the lua script for defining currents in the example was that
the function defined in the script can be used to change all the currents
with one function call--otherwise, you'd have to change a lot of things manually
to change the current level in a consistent way.

Anyhow, without seeing an example, it's hard to say what's gone wrong.

---------------
Karakache Mohamed wrote:

> In the case of an asynchronous (induction) machine one iinterest with
> prelever the caracteristique one of the couple according to the slip.

> How can one make to vary the slip while working with femm

See the example at http://femm.berlios.de/2horse.zip  Since the configurations addressed in femm are all static (i.e. the rotor isn't really moving), you end up really doing the analysis at the slip frequency and inferring the behavior of the machine under load at the nominal operating conditions.

---------------
Douglas Stevenson wrote:

> I have a few conceptual questions regarding this model.
> 
> The example clearly demonstrates the "deep bar" effect on the rotor
at
> different frequencies. The current density in the rotor bars is clearly
> redistributed towards the outer edge of the rotor at reasonable starting
> frequencies of 50-60 Hz. 

If you are interested in having a circuit model that you can used over a
wide range of slip frequencies, you might have to add extra detail to the
motor model.  The form of the model presented in the example was intended
more as a starting point for induction motor modeling, rather than as adequate
for every situation.  

I'm usually only interested in field-oriented control of induction motors--in
that case, slip frequencies are generally low enough so that the simple model
is sufficient. However, there are plenty of situations in which a more complicated
model could be required.

> However, how would one assess the so-called "Proximity effect" in the
> stator windings? At higher frequencies of applied current (say a few
Khz
> caused by a current source with severe distortion), the presence of
a
> winding with parallel turns or conductors that are large w.r.t the skin
> depth, would create an effect similar to the deep bar situation. The
> ferromagnetic slot would cause the current density to be "redistributed"
> in those stator windings, once again producing an effective AC
> resistance far greater than the DC resistance, and a corresponding
> increase in undesirable losses.
>
> One would cleary need to define a large number of conductors/turns with
> their own circuit properties in FEMM, but how would one assess which
> regions of the individual conductors/turns have higher and lower
> resultant current densities? Each condcutor /turn has both its self
> inductance, causing the skin effect, and the mutual inductances with
all
> the surrounding conductors/turns as well as with the eddy currents in
> the core causing the proximity effect. How and which would one define
> boundary conditions for the individual turns/conductors?

To really nail this down, you end up having to explicitly model each wire
and define a circuit property for each wire that defines the total current
flowing in the wire. However, you don't have to define any special boundary
conditions for the induvidual conductors.  The "circuit" property makes sure
that a prescribed total current is carried in the wire, and the distribution
of the current within each wire is dictated automatically by skin and proximity
effects. 

There are a couple of examples that I had posted before that are intended
to model proximity effect losses.  Check out:
http://groups.yahoo.com/group/femm/message/1344
http://groups.yahoo.com/group/femm/message/1319

There also might be a way to infer proximity effect losses a posteriori from a field solution with nonconductive
bulk coil model.  It seems like I saw a paper about this in Transactions
on Magnetics not that long ago, but I wasn't able to fish up the reference
to it tonight--I suppose that I could be mistaken.

Dave.