FEMM model of the 10 hp conversion

[dinges]

After reading this story by Sparweb

http://www.fieldlines.com/story/2007/10/14/43956/594

I gave the software a little try myself and was surprized by how easy it is to use. It's just 1.4 MB to download too. After fooling around with it a bit I decided to try to model the 10 hp conversion (still mainly in the drawing stage) as accurately as possible. First step was to measure the stator and draw a projection in AutoCAD. The .dxf drawing can then be imported into FEMM for analysis.

(details: the motor is a 7.5 kW (10 hp) one, that will be converted with 36 pcs. N40 magnets, 2"x1"x.5" in 'JAM' style.)

After running the simulation I realized (just as Sparweb did) that, despite what I -thought- was a large airgap (1 mm at the edge of the magnets, 2 mm in the middle), the stator teeth still were driven (well) into saturation, as the images below show:

I will have to remodel: create a larger airgap and see how large it has to be before (massive) saturation stops.

I think that slight local saturation isn't much of an issue, but when entire parts of the stator tooth are above 1.6 T that can't be good I expect.

The preliminary conclusions:

• air gap should NOT necessarily be as small as possible (as I more or less thought it should be in motor conversions)
  
• grade of Neos isn't very important in conversions (I use N40); any premium magnet grade would just mean you'd have to introduce extra airgap to reduce flux density in the stator.
  
• FEMM is pretty easy to work with
  
• FEMM is 2D package so may not be very suitable to modeling Zubbly style conversions (round magnets with helical skew)
  
  

Any comment/insight on the issue of tooth saturation (especially from Flux if he reads this) would be greatly appreciated. My main question now is just how much saturation is a problem... A little local saturation is likely not an issue, but just how much is 'too much'...

More detailed images can be found here:

http://www.anotherpower.com/gallery/dinges/blue_monster_accurate_FEMM_airgap_1_2mm_N40_detail?full=1

http://www.anotherpower.com/gallery/dinges/blue_monster_accurate_FEMM_airgap_1_2mm_N40_overview?full



=1

Regards,

Peter.

[wdyasq]

" air gap should NOT necessarily be as small as possible (as I more or less thought it should be in motor conversions)

  grade of Neos isn't very important in conversions (I use N40); any premium magnet grade would just mean you'd have to introduce extra airgap to reduce flux density in the stator."

Another option may be to use less magnetic material. This would further reduce cost and, if one is using the original stator turned down, less chance of going the rough the laminates.

You have started a mess.

Ron

[Bobbyb]

Hi,

Interesten software!!

why is saturation a problem? Of coarse it's a waste of magnets i.e. money, but are there other disadvantages?

[boB]

I wonder what would happen if you had a real small air gap AND the entire magnet and  laminations were all made out of Neodymium Fe Bo, besides being really expensive and hard to work with.

boB

[RP]

Peter, if you can it might be worth modeling the effect of stator current in the windings.  This will "push back" against the magnetizing force and may drop the teeth out of saturation.

[oztules]

Thanks for that Peter,

At last some of what Flux has talked about with regards to teeth, iron behind the teeth etc is making some kind of sense.

A great many thanks in fact.

.......oztules

ps if you wish to spend some of your nordic nights to produce an animation of 1 degree movement over say 36 degrees... it would become all the more clear how tooth shape and magnet interaction and iron behind the teeth really works.

[dinges]

Yes, that makes sense.

And that's really the problem with what I've done; it's just a partial model of the motor. Doesn't show the entire picture.

Peter.

[ghurd]

Hi Peter,

"My main question now is just how much saturation is a problem..."

I never figured it out that far.

My answer "Bad magnets don't cog, good magnets don't cog, but great magnets cog- a lot."

From #7, http://www.fieldlines.com/story/2006/8/23/144336/344

Might be something helpful in there.

Thats why I got the one rotor stuck, just seeing what was what.

G-

[Flux]

Does that programme actually manage to deal with saturation? when the teeth are saturated then the flux path from magnets to stator core is via a very long air gap equal to the true gap and tooth length. I suspect that it doesn't take this into account.

Saturation is fairly progressive and beyond about 1.4T the steel will be close to saturation and more flux will fringe through the slots.

I think it unlikely that anything beyond 1.6T is possible without an enormous length of magnet.

What real effect it has I don't know, certainly the iron loss will increase with peak flux density in the teeth and will make the no load loss worse. beyond saturation it will not increase cogging and may reduce it ( in the extreme case if the flux in the teeth is the same as in the slot then it would vanish)

The higher the flux, the lower the copper loss so you may have to trade losses at cut in for different full load efficiency.

There could be additional surface losses in the magnets with very small gaps, it doesn't seem to be as bad as solid pole shoes but the effect must be there.

With magnets not matched to the stator arc then I wouldn't go less than 1/2mm at the closest point. With arc magnets then I wouldn't use less than 1mm gap.

If the program does properly predict things into saturation then shorter magnets or longer air gap would seem a good idea, but I suspect that with 1/2" thick rectangular magnets you can keep to the 1/2mm at closest point and not worry too much.

Jerry always seem to have found that arc magnets worked better than rectangular, suggesting that a short gap was not a problem. ( I think his criteria was output rather than start up loss so it may not be the whole answer).

Flux

[dinges]

Thanks !

Ghurd, that link is very informative; I've been googling the board all evening yesterday but missed that one.

Flux,

I think it models saturation too; I simulated using S-2 (0.024" thick) stator iron; the magnetization curve for this particular iron is a material property that's defined in FEMM. Also the magnetic properties of air and the rotor iron (1020 = C20 steel) are defined and taken into account. At least, if it didn't, it would be utterly worthless as a FEM program.

Also, one can see in the simulations the effect you remarked about: when the tooth is saturated one can observe flux lines jumping out and taking another part, through the air.

But I've only limited knowledge of electro-magnetics and electro-magnetic FEM simulations. And I have nothing to test/verify my simulation results against. So the question, is how true-to-real are these pretty pictures it produces.

That's also why I wonder: with windings in the core, the reaction-EMF from them would 'push' the fieldlines back and reduce saturation of the core. Only problem is that below cut-in speed there is no reaction-EMF to do this. So the core would be saturated below cut-in speed and reduce start-up performance.

As always there's compromises. My main worry I have w.r.t. to stator losses is start-up performance. This is going to be a generator rated for about ~3 kW peak, I expect. But what I want is something that generates a trickle (50-100 W ?) in low winds but does it nearly continuously. I'm not looking for a genny that produces 10 kW in the twice-a-year storms.

So maximum rated output of the generator wouldn't be of much concern to me, but low-wind performance would be.

What is making me doubt about all this large airgap: most people on this board build their conversions with as small an airgap as is possible and they seem to get decent results... So the safe route might just be to go that route myself as well ?

Decisions decisions decisions...

Peter.

[Countryboy]

Peter,

 To determine how critical air gap is, you may want to do some real life experimenting.   I figure you will remove the rotor laminations and make a new solid steel rotor, so it would be easy to do a little experimenting then.  Here is how I would do it if I were you:

  I would turn down the old rotor diameter so that I could install one or two magnets on the rotor, leaving a very small air gap.  Then, install that rotor and spin it to test low rpm output.  

  Then, turn the rotor diameter smaller a tenth of a mm, and retest.  Then another tenth mm more air gap and test again.

  This would give you the most accurate real life answers of how critical air gap is in a motor conversion, although the test may be a little time consuming.

  Yes, I know this test using only one or two magnets would cog, but I don't feel it would impact test results.  For test results, we are only interested in the change in output proportional to change in air gap, where as cogging would impact total output at low rpms.

[FishbonzWV]

Peter

I did a conversion on a 3/4 hp 3ph motor with 4 #29 mags. In my haste to get it done I forgot to offset one set of mags and this thing cogged badly. I tried to remove one set but the JBweld held fast and I didn't want to break the mags. So after a lot of trial and error with more JBweld putty and some 1/4" x 1/4" mini neos I got the cogging down to a 'two finger spin it with all your might and get almost two revolutions'.

This thread gave me an idea about removing the tops of the T's of the lams...making them 'l' to reduce the cogging even more so.

Did you have to cad the lams and mags or were they part of the software package?

Could you run a simulation with 'l's ?

It would be interesting to see the difference in saturation levels.

I just might have to break out the hand grinder and attack those nasty lams!

Bonz

[vawtman]

Heres a post i did awhile back when thinkiin the same thing.

http://www.fieldlines.com/story/2006/6/27/1321/01546

Flux kinda put the idea to rest has is common with my ideas.

[oztules]

Yep,

most of my scatterbrained ideas suffer a similar fate... your not alone in this

.........oztules

[dinges]

Good idea.

It would be a good option if I'd have NdFeB magnets with holes in them, for screwing them down. Also, it would only work (if using one or two magnets) when the stator was skewed.

When installing only one magnet using the JAM method, the thing would cog so badly that it would completely drown out the effect of the iron loss. There's an easy way to determine iron loss (I've described it in the decogging tutorial) by a simple measurement and some calculations, but it requires a non-cogging conversion for it to work.

In my case, the rotor will be made using the JAM offset method, which means I'd have to install one full belt of magnets (12 for a 12-pole) for it not to cog.

But the experiment you describe would be a very interesting one to make. One could learn a lot from it, if done properly.

And there's the fact that I've already removed the original rotor from the shaft.

Thanks for the tip. I'll be keeping it in mind.

[FishbonzWV]

So by the link it looks like bridging in the slots might be beneficial to reducing the cog. If I can get some spare time I could try some JBweld putty on another identical motor and test it with the converted rotor.

The Zubbly is on a test stand about head high and when it is spinning I can hear a vibration that I assume is the cogging effects.

As to taking off the points of the T...I thought this might be a way to reduce the saturation level since the FEMM model showed over saturation.

[SparWeb]

Peter,

I'm both ahead of you, and still catching up.  I've been sourcing arc-segment magnets that will be more appropriate for my next 3HP conversion project, planned for this winter.

I've been working with FEMM, like you, and obviously we're getting the same results when we put in the same variables.  I have played around with gaps and magnet ratings, it doesn't really matter.  It always saturates the stator.

I don't think FEMM is properly modelling the saturation because the flux simply shouldn't be as high as 2.1.  Other things I've noticed is that you can confine the flux in a thin steel ring that holds the magnets on the rotor.  Then assume that the core of the rotor is aluminum.  I built my first conversion that way - perhaps I was unwittingly reducing the flux in the stator by mounting the magnets on a press-fit ring.  No way to tell.  FEMM isn't modelling the saturation, so you can't tell if the saturated ring prevents saturation of the sator.

Something else that bothers me about FEMM is that you are supposed to set the system frequency to 60 Hz by default.  Well these motor conversions are really churning away when the output current is 10 Hz.  Lower electrical frequencies give more time for laminations to saturate, from what I've learned here and there.

And who's to say saturation of the stator is a bad thing?  Maybe, as Flux mentioned, it's reducing some of the cog...  And I do appreciate the ability to model the generator with FEMM.  It's free and the visual result is VERY informative.

I gave up on the fat block magnets and I'm going to try arc-segments.  I modeled them in FEMM and since they're thin they don't drive the flux up so much.

Since I haven't been able to find arc-segment magnets of the correct size for my next motor-conversion project, I decided to order some from a manufacturer.

The magnet specs:

NdFeB Arc Segment Magnet

2.219" o.r. X 1.969" i.r. X 1" X 75 degree arc

One #8 screw hole with countersunk , Grade N42, Nickel-plated

Qty 16 @ $38.75/pc (8 N-pole on o.r. & 8 S-Pole on o.r.)

The magnets are designed to fit inside a 4.5" 3-phase motor stator with 1/32" clearance.  Each arc is 1/4" thick.  Notice that I want the screw hole, too.

The price they quoted is rather high, but I'm looking for other interested parties on the odd-chance that I can increase the order quantity.  That might push down the unit cost.