Dual rotor axial flux motor design

[Miles]

This is interesting...
http://www.set.ugent.be/docs/TO_HEAF_v2.pdf

That makes 3 ways to equalise the radial flux density:

- Equalise, per lamination, the ratio of the cross sections of the core "head" to core "neck" .

- Reduce magnet area as radius decreases (only applicable in the case where the no. of poles is less than no.of slots).

- Reduce the airgap distance as radius increases (Ghent university proposal).

[Miles]

Axial-flux permanent-magnet machine modeling, design, simulation and analysis.
A. Mahmoudi*, N. A. Rahim and W. P. Hew

[Miles]

Adopting the US vernacular....: Woot!

Comparison of Nonoriented and Grain-Oriented Material in an Axial Flux
Permanent-Magnet Machine
Damian Kowal, Peter Sergeant, Luc Dupré, and Alex Van den Bossche

V. CONCLUSION
For an AFPMSM whose stator flux is flowing in axial direction
in the major part of the stator, GO material was compared
with NO material. With GO material, the machine has about 7
times less iron loss at the same speed, and a 10% higher torque
for the same current. Nevertheless, the EMF at no-load is almost
the same for both materials. For a given torque, the GO material
causes a 10% higher torque-to-current ratio which makes
it possible to reduce the copper losses—quadratic with the current—
by about 20%. Alternatively, because of the lower iron
losses, it is possible with GO material to allow larger copper
losses without increasing the temperature of the machine. This
means a higher stator current and more torque.We conclude that
for the considered type of PMSM, it is worth the extra cost to
use GO material.

[Miles]

Optimized Design Considering the Mass Influence of an Axial
Flux Permanent-Magnet Synchronous Generator
With Concentrated Pole Windings
Hendrik Vansompel, Peter Sergeant and Luc Dupré

[bearing]

Very nice finds!

Considering the amount of academic research done on axial flux motors, and the obvious advantages they have, I'm a bit confused to why there are so few axial flux motors available/produced/used commercially.

[kenkad]

Miles,
The end of the third paragraph on the Ghent papers says:
"contentrated pole windings----A disadvantage is that for a three-phase machine several windings should be connected in series. When connecting n concentrated windings in series, the total electromotive force is not n times the voltage of a single winding, because of phase shift in the neighboring windings. An atlternative is to make a machine with many phases [3]."

This is what I had tried to bring up earlier. I had see this paper before but had not copied it at that time. Thank you for posting it here so now I can make a copy and give it to the PhD student that hopefully will do my modeling/simulation soon. I believe that a nine phase motor cited in the reference above is bsically the same as three separate three-phase windings. I do know in my design that it reduces the required drive of each three-phase group.
kenkad

[bearing]

When connecting n concentrated windings in series, the total electromotive force is not n times the voltage of a single winding, because of phase shift in the neighboring windings

This is the so called "winding factor". In a good design the phase difference won't be very big. I think Miles had about 0.95, so it won't do much to increase it to 1.

[Miles]

Have you seen this one, Ken?

[Miles]

When connecting n concentrated windings in series, the total electromotive force is not n times the voltage of a single winding, because of phase shift in the neighboring windings

This is the so called "winding factor". In a good design the phase difference won't be very big. I think Miles had about 0.95, so it won't do much to increase it to 1.

As they say:

Although a multiphase system for concentrated pole windings
is proposed in [5], a regular three-phase system is used in
this paper. In order to obtain a high winding factor, the number
of teeth is set to 15 resulting in a winding factor of 0.951 for the
16-pole machine

[kenkad]

Miles,
Yes, I had this copy. This is a paper that the PhD students Professor cited to me as a probable issue of the harmonics. Of course, that is something to find out. I am not a motor designer as you and I have discussed before. There are a lot of tradeoffs in any design problem. I was trying to find a way create a motor design that would allow shutting down of some of the three-phase groups when a vehicle is crusing. At that point, cogging effects as less of an issue, etc. Sort of like running an 8-cyl engine on 4 cylinders to save fuel. It really makes a difference in finding a full bridge driver for each three-phase group! As usual, it will all come out in the end one way or another. If the simulation is of any consequence, then we go on to a prototype, if I live long enough or if we live through Dec 21, 2012 (ha ha).
kenkad

[Miles]

I was trying to find a way create a motor design that would allow shutting down of some of the three-phase groups when a vehicle is crusing. At that point, cogging effects as less of an issue, etc. Sort of like running an 8-cyl engine on 4 cylinders to save fuel.

I still can't see the logic of that one. I don't think it's an appropriate analogy. Switching off phase groups will lower the efficiency.

[Arlo1]

I was trying to find a way create a motor design that would allow shutting down of some of the three-phase groups when a vehicle is crusing. At that point, cogging effects as less of an issue, etc. Sort of like running an 8-cyl engine on 4 cylinders to save fuel.

I still can't see the logic of that one. I don't think it's an appropriate analogy. Switching off phase groups will lower the efficiency.

Yeah electric motors don't work that way. Instead the controller only puts the required power to move the motor for the load and speed required. The controller uses PWM to control this so in essence at cruise it is turning all the phases on and off really fast to save energy.
And ICE with cylinder deactivation on the other hand will not just stop putting fuel into the deactivated cylinders it will also close the valves so the pressure inside those cylinders will neutralize and become an air spring for the piston to bounce off of on the end of each stroke. So if you wanted to simulate this with an electric motor (which will not likely gain you anything at all) you would have to find a way to remove the eddy current losses and Hysteresis losses from the non powered coils.

[bearing]

Yes, a better approach to decrease idle losses could be to make the rotor discs able to move further apart at light loads, which weakens the field and decreases iron losses. But it will only help at very light loads, since that would decrease the torque constant, hence increase resistive losses.

Maybe it's more beneficial in a performance/cost relationship to buy better steel, than to make the rotor discs moveable.

[Miles]

Regarding access to the center of the motor for the three phase wires.....
The copper winding strip is 14mm x 1mm so that's 14 sq. mm in cross sectional area. Equivalent to a 4.2mm diameter round wire...... Compromising a bit on the total cross sectional area, I could use 2.6mm dia. wire in a loop, sharing the current. There's just room for this to pass between the core heads and the coils, at each end.

But.........How seriously will this proximity to the coil/core unbalance the electromagnetics?

Should I feed in from the other side of the core modules to maintain the coil winding direction?
This will reinforce the armature field on 2 of the 6 modules, for each phase?

Should I reconsider the original idea of linking the coils on the outside of the motor?

[fechter]

If the wire passes through the slot, it will be 1/2 turn, negating 1/2 turn of the strip. Running it the other way will add 1/2 turn, which would probably be better but you have to be sure it won't fight the BEMF from the other windings. Making the connection on the outside to maintain an even turns count would be the best bet.

[Miles]

If the wire passes through the slot, it will be 1/2 turn, negating 1/2 turn of the strip. Running it the other way will add 1/2 turn, which would probably be better but you have to be sure it won't fight the BEMF from the other windings.

Thanks Richard. That's what I was thinking. Feeding it through on the other side, the current will be running in the corresponding direction for the adjacent coil, too (AaA, BbB etc..) . I'm worried that, with only 4 turns per slot, an extra 1/2 turn is a relatively large difference...

Making the connection on the outside to maintain an even turns count would be the best bet.

But, on the other hand, connecting the modules on the outside increases the overall diameter [and one of the reasons for exposing the endturns through the case was to maximise the mean airgap radius...], it also increases the length of the links between the phase groups.

Hmmmm........

[Miles]

Ok. I'll model the "Pompidou" version, to see what possibilities it has....

[iti_uk]

Surely if you spun the windings 180 degrees around the laminations, you would not be adding any turns while having the connections on the outside of the motor...? Or am I missing something?

Chris

[Miles]

Hi Chris,

Surely if you spun the windings 180 degrees around the laminations, you would not be adding any turns while having the connections on the outside of the motor...?

Indeed.

Making the connection on the outside to maintain an even turns count would be the best bet.

But, on the other hand, connecting the modules on the outside increases the overall diameter [and one of the reasons for exposing the endturns through the case was to maximise the mean airgap radius...], it also increases the length of the links between the phase groups.

Hmmmm........

Ok. I'll model the "Pompidou" version, to see what possibilities it has....

Or am I missing something?

[iti_uk]

Lol I guess I should refresh the page before I post.

[Miles]

[fechter]

I'm still trying to grasp what happens if the turn count is a little uneven. Since each phase is a bunch of windings in series, as long as the phases are equal, it may not make much difference if one winding has an extra half turn.

The current in a series circuit is equal everywhere, but the half turn difference will result in a tiny bit of extra torque on the core with the extra turn. Since the rotors are solid, the torque from each core adds to the total, so a slight imbalance should not do anything bad.

If there was an imbalance between phases, that would be bad, as one could slightly fight the other and cause a major efficiency loss.

[Miles]

I was worried that it might be a cause of vibration in the rotors.

How about making the connection along the length of the first half turn (from the edge)? You'd end up with a compromised start but it would be better than nothing?

[fechter]

I don't think the imbalance will cause a significant vibration in the rotors. The rotors are pretty stiff and the forces from normal running would be much greater than the imbalance force.

I'm not picturing your last idea.

[Miles]

Something like this. I doubt if it has any advantage over simply missing out the first half turn?