Report KD 690 about 4-pole PM-generator available

[Adriaan Kragten]

Report KD 690 can be copied for free from my website: www.kdwindturbines.nl at the menu KD-reports. The tittle of this report is: "Ideas about a 4-pole permanent magnet generator for the VIRYA-2S windmill using the housing of a 4-pole, 3-phase, 0.75 kW asynchronous motor frame size 80 and 4 neodymium magnets size 80 * 20 * 10 mm. Design report of the rotor (lambda design = 4.5, B = 3, stainless steel blades)". Advantages of this generator are that the magnet costs are rather low (about € 28) but that the generator has a high maximum torque level and can therefore be used as a brake in combination with the VIRYA-2S rotor and the hinged side vane safety system. To give an impression of the generator, I have added figure 1 out of KD 690 as an attachment.

[SparWeb]

Hello Adriaan,

How much stronger do you think the air-gap flux could become if this were implemented for a 36-slot stator.  Since the skew angle would reduce from 6.5 to 4.3, can the air gap distance be significantly reduced?

On fabrication of the armature, you could consider including a slot or a keyway to physically lock the rotor to the shaft, rather than rely on an adhesive.  The tools to mill the magnet slots are the same tools that can be used to make these slots between rotor and shaft. 

Another question comes from the ligament of material between the bottom of the magnet and the hole for the shaft.  Will this be highly stressed?  A thicker ligament may also offer a better flux conduction path - an concern you have noted at the top of page 4.  By using a smaller magnet the slot would be less deep, and the improved flux path may partially make up for the reduction in magnet material.

[Adriaan Kragten]

Kienle and Spies also supplies a stator stamping with 36 slots for a 4-pole motor frame size 80. I have chosen 24 slots because then the winding is simpler. If you have 36 slots, the inclination angle of the magnet grooves becomes about 2/3 * 6.5 = 4.33°. But the calculation of the flux density in the air gap in between armature and stator doesn't change and so you will get the same calculated flux density of 0.99 T. If you look at the formula for the flux density in the air gap, you see that the flux density only depends on the number of armature poles p and not on the number of stator slots.

To my opinion, gluing of the armature to the shaft is the only practical option. If you calculate the shearing stress in the glue for the maximum torque, you will see that it is very low. You can't press the armature on the shaft because this results in a very high pulling stress in the bridge at the bottom of the armature grooves. A key in between the shaft and the armature makes things much more complicated and it isn't necessary if good quality glue is used. I have used epoxy glue and anaerobe glue. The advantage of anaerobe glue is that it can have a higher temperature. But as there are no short-circuit currents in the armature, like you have for a normal short-circuit armature with aluminium bars in it, the temperature isn't becoming high. Another advantage of anaerobe glue is that it hardens only in the gap and that is especially nice for gluing the magnets. So all superfluous glue can be wiped off easily.

The bridge must be that strong that the armature isn't damaged during milling of the grooves because of the side force of the cutter but a bridge with a thickness of about 1 mm is strong enough. Once the magnets are glued in the grooves, the armature becomes one piece and it becomes very strong. But I have calculated the stress in the bridge assuming that the magnets don't contribute to the strength of the armature. In this case every pole is pushed outwards by the centrifugal force. I have calculated the centrifugal force for the nominal rotational speed of 1500 rpm which is much higher than the real maximum rotational speed in the windmill. The centrifugal force acting on all four poles gives a pulling stress in the bridges. Even for this high rotational speed, the pulling stress in the bridge is very low.

I have built about six, 4-pole generators according to this principle and tested them until a rotational speed of 1500 rpm up to the maximum torque level and never got any problem with the bridges or with the magnets. The VIRYA-2.2S generator was tested almost ten years in the field.

[SparWeb]

Thank you
I completely neglected how the magnet poles would be oriented.  The flux gap would pass from steel quadrant to the stator, not from magnet to the stator.  Have I illustrated the paths as you intended?

[Adriaan Kragten]

Thank you
I completely neglected how the magnet poles would be oriented.  The flux gap would pass from steel quadrant to the stator, not from magnet to the stator.  Have I illustrated the paths as you intended?

(Attachment Link)

Yes, your picture is right. My figure 1 of KD 690 gives only the armature and not the stator stamping with the winding which is positioned around the armature. So the magnetic flux coming out of a north pole of the armature flows into the stator stamping, next it makes about an 270° turn and flows back into a south pole. A picture of a 4-pole armature plus stator (with 36 slots) is given in figure 1 of report KD 341.

[SparWeb]

In the future, if I want to make another motor conversion, I would like to try the slotted magnet method, as you've shown.

Here is the armature I made for the last of my motor conversions (in 2010 - how time flies).  It obviously mounts surface magnets.


When it was complete, I found that the magnets were slightly under-sized, and the air gap was larger than necessary. 

As you can see in the photo, I used a solid steel cylinder and I believe your armature must be solid, too.  Or do you think it is practical to make it from steel sheet (stampings) too?

[Adriaan Kragten]

I have changed figure 1 of KD 690 such that now the magnetic field lines are shown.

For my generator, the magnets are positioned radial. This has as advantages that the poles can have a radius which is the same as for the original short-circuit armature, that you get concentration of the magnetic flux because the flux coming out of one north pole is supplied by two magnets and that the magnet grooves can be inclined resulting in an almost non fluctuating sticking torque. The disadvantages are that you have to make rather deep grooves and that you need a stainless steel shaft.

The magnetic flux in the stator is fluctuating when the armature is rotating and so you need a laminated stator stamping to prevent strong eddy currents. The magnetic flux in the armature isn't fluctuating so the armature can be made from massive iron bar. However you can use a laminated stamping but not the original short-circuit armature as this armature contains aluminium short-circuit bars which don't guide the magnetic flux. Using laminated iron for the armature is only logic for mass production as then the iron coming out of the center of the stator stamping isn't lost and as it might also be possible to make the grooves for the magnets directly in the stamping.

If you want to make a new bigger 4-pole generator using this principle, it is important that you don't use a motor with a stator stamping with four outside grooves in which strips are laid to bind all sheets of the stamping together. I have done tests with an Indian stamping with these grooves and the armature had four strong preference positions per revolution. The preference positions occur when the magnets are just opposite the outside grooves as for this position, the magnetic loop flowing through the stator is almost not hindered by the grooves.

[Adriaan Kragten]

The VIRYA-2S generator uses magnets with a width of 20 mm and a thickness of 10 mm. The tolerance on the thickness is plus/minus 0.1 mm. The magnet grooves in the armature have width of 10.2 mm and a depth of 21.75 mm. The tolerance on the width is plus 0.1 mm. The grooves are made on a milling machine using a 10 mm finger cutter. First a 10 mm width groove is made in steps of some mm up to a depth of 21.5 mm. Next a new very sharp cutter is taken and the sides and the bottom of the groove are machined until the groove has the correct dimensions. During this last machining operation, the cutter has a tendency to bend and so the width of the groove can be a little less at the bottom of the groove than at the top of the groove. But the groove must have a width of at least 10.1 mm otherwise a 10.1 mm thick magnet can't be pushed up to the bottom. This effect of getting slightly tapered grooves isn't very significant for 21.75 mm deep grooves. However, if wider 10 mm thick magnets are used for a bigger 4-pole generator, it can give problems for deeper grooves. The problems can be solved by using thicker magnets when the magnet is wider. This allows the use of a larger cutter diameter which will bend less. But thicker magnets are more expensive and a thickness of 10 mm is very standard in Europe.

The problem can also be solved by using a 6-pole motor and an armature with 6 magnet grooves. I have looked to a 4-kW, 6-pole motor with a stator stamping of Kienle and Spiess. This stator stamping has an outside diameter of 170 mm, an inside diameter of 115 mm, a length of 140 mm and 36 slots. The original motor shaft has a diameter of 38 mm at the armature. Assume that the original motor shaft is used but the short-circuit armature is removed and replaced by a stainless steel bush with a diameter of 70 mm and a length of a little more that 150 mm. This bush is turned to a diameter of 69 mm over a length of 150 mm. A mild steel bush with a length of 150 mm and an inside hole with a diameter of 69.2 mm is glued to the stainless steel bush and turned to a diameter of 114.4 mm. So the air gap becomes 0.3 mm. Six 10.2 mm wide and 21.7 mm deep grooves are milled in the armature under an angle of 4.08°. Three magnets size 50 * 20 * 10 mm are glued in each groove. So a 150 mm long armature juts out 5 mm at each side of a 140 mm long stator but that isn't a problem.

The flux density for this configuration is about 0.94 T, so the stator is saturated. So this option allows the use of the original motor shaft. This shaft has a diameter of 30 mm at the bearings but of 28 mm at the free shaft end. This shaft isn't as strong as a shaft with a tapered shaft end so the maximum allowable rotor diameter will be about 3.2 m. The generated voltage for a 6-pole generator will be about a factor 3/2 higher than for a 4-pole generator if the original 230/400 V motor winding is used. It depends on the diameter and the design tip speed ratio of the rotor and on the way of rectification if this winding can be used for 48 V battery charging.

[SparWeb]

These armature configurations are interesting, too:

https://www.kienle-spiess.de/kspm.html

If made with steel stampings (or laser-cut, possibly) this could be quite effective and cheap!
Compared to surface-magnets, the air gap would be much smaller, allowing smaller magnets to be used.
They would be difficult to make if done in a solid cylinder.  I don't think much would be lost by using laminations.

[Adriaan Kragten]

I know that these armature stampings with magnet holes in it exist. In fact they are also supplied by Kienle and Spiess for PM-motors. However, the used magnets are rather thin. As the poles are not inclined, one has used another trick to reduce the peak on the sticking torque. The trick is that the radius of the poles is made much smaller than the inside radius of the stator. This results in a small air gap at the heart of an armature pole but in a large air gap at the sides of a pole. So the average air gap is rather large and this reduces the maximum magnetic flux in the stator, especially for thin magnets. Another point is that the fluctuation of the sticking torque may be reduced enough for use as motor but I am afraid that it isn't reduced enough for use as wind turbine generator.

I have looked at a stator stamping of Kienle and Spiess for a 6-pole motor frame size 112 at their website. The code number of this stamping is: KSPM 112/6.115. The inside diameter of the stamping is 115 mm. The maximum outside diameter of the armature is 113.6 mm, so the minimum air gap is 0.7 mm at the heart of an armature pole but in the drawing it can be seen that at both sides of an armature pole, it is about 3 mm. The average air gap will be at least 1.5 mm. The magnets which have to be used are 20.3 * 3.5 mm. So the total magnet thickness is 7 mm in one magnetic loop. The total magnet thickness for my generator in one magnetic loop is 10 mm and the air gap is only 0.3 mm. So the magnetic flux which I get in the stator will certainly be much stronger and it is sure that the fluctuation of the sticking torque is negligible.

I have made a mistake in the calculation of the flux density in the air gap of the 6-pole generator. It is about 0.9 T in stead of 0.94 T but this is still high enough to get the stator stamping saturated.

[Adriaan Kragten]

A bigger 4-pole generator using frame size 100 is described in public report KD 503. I have just added a new chapter 4 to this report in which it is investigated if in stead of an Indian motor housing, it is possible to use a housing with a stator stamping according to IEC norm. It is possible but only if a new stainless steel shaft is used.

[SparWeb]

This is intriguing.
With world production of electric motors surpassing a billion, there are a staggering variety of configurations, and yet we still do not see skewed armatures in mass production.
A slight change to the armature that is shown in the attached picture above (which was also taken from K&S) will look like this:



(My quick drawing - many details omitted for clarity)

[SparWeb]

Between the magnet poles, grooves are machined to separate the poles.  There are such grooves shown on the photo above, but on that one, the grooves are aligned with the axis.  In my drawing, they are shown with overlapping lines because the cuts are NOT parallel to the axis of the rotor.  They are cut at a skew and hopefully it's shown clearly enough that the gap coincides with one stator tooth at one end, and an adjacent stator tooth at the opposite end. 

I can picture the methods needed to fabricate the entire thing, and it could be done with plates up to 25mm thick, thus allowing an assembly of a stack before the final cut of the skewed pole separation groove.

With the caveat that many details are omitted from the drawing, such as machine tool cutting relief, fillets, and consideration of flux path details to eliminate "short circuits", there do not appear to be any major obstacles.  That much detail simply isn't warranted on a conceptual drawing.

[Adriaan Kragten]

If the groove in between the poles is screwed, it may work well as wind turbine generator. Another option is to keep the grooves in parallel to the armature axis but to screw the slots in the stator stamping. But this is only possible if the stator stamping isn't welded. This methode of screwing the stator stamping has been used for years by the former Dutch manufacturer LMW.

I think that inclined magnet grooves are not used in PM-motors because the magnets are not fixed mechanically. So one is afraid that the magnets are pushed out of the grooves by the centrifugal force if the motor runs at a high rotational speed. But this doesn't count for a direct drive PM-generator as the maximum rotational speed in the wind turbine is much lower than the nominal rotational speed of a 4-pole motor (1500 rpm at f = 50 Hz). But if you calculate the shearing stress in the glue for a rotational speed of 1500 rpm, it is very low. All my PM-generators with this kind of armatures have been tested up to a rotational speed of 1500 rpm and never the magnets came loose. But it is important that the groove is cleaned well before gluing and that a good quality anaerobe glue is used which hardens in an air gap of 0.2 mm. I have used Threebond TB 1132.

[Adriaan Kragten]

A disadvantage of screwing the grooves in between the poles and having straight grooves for the magnets is that now all armature sheets become different and that they must be mounted in the right sequence. This problem can be solved by using square magnets and by dividing the armature in different sections. Assume one uses magnets size 20 * 20 * 5 mm which are supplied by Enesmagnesi. Assume that the armature has a length of 100 mm and that it is divided in five 20 mm long sections. Assume that the stator has 36 slots, so the stator pole angle is 10°. Every armature section is now rotated 2° with respect to its neighbour. Assume that the armature sheets are connected to the shaft by two keys like it is also done for the stampings on Kienle & Spiess. This means that the key grooves of all stampings must be in line and this means that you will get only five different armature stampings. The magnets are mounted first into each 20 mm long armature section and each section is pressed to the shaft. Only the magnets in the two outer sections have to be glued. The really required magnet dimensions depend on the real armature geometry.

I have built an armature with three rotated armature sections and with tangentially mounted magnets size 20 * 20 * 5 mm for a housing frame size 71. However, I used an Indian housing and that housing had a stator stamping with four outside grooves. These grooves caused four strong preference positions per revolution but in between these four preference positions, the sticking torque was almost constant. So it has been proven that the idea of using several armature sections, works if the stator has no outside grooves. The more sections you take, the better the sticking torque is flattened.

[SparWeb]

Indeed, that addresses an assumption too easily made.  One either assumes that the armature would be made with a solid cylinder (difficult to manufacture) or with thin sheets like the stator (about 1-2mm thick each).

However, a third option is available where the armature is still segmented, but the segments can be thick (eg. 12 or 25mm) and rotated a few degrees with respect to one another.  The thickness corresponds to the depth of each magnet that will be inserted, and the rotation corresponds to the number of layers and angle between respective stator teeth.  A total of 4 layers each 25mm thick would yield an armature 10cm long, have a desirable de-cogging skew, be very robust, yet not too expensive to fabricate since the machine cutting tools can easily reach through.

I know I am just re-stating what you just wrote, in my own words, but thank you all the same.  And I also agree that the more sections made, the flatter the torque ripple.  I do like how this appears to be both practical and accessible to the builder with modest machining abilities.  Even more practical if one wants to provide CAD files to a shop with laser-cutter tables.

[Adriaan Kragten]

There is another aspect which is not yet discussed and that is the exact width of an armature pole. If the stator has 36 slots, it means that the stator pole angle is 360 / 36 = 10°. I think that the width of an armature pole has the be chosen such that it covers exactly a whole number of stator poles. The armature pole angle of the hearts of a 4-pole armature is 90° but the armature must be made smaller other wise there will be a large magnetic short-circuit in between the north and the south poles. Assume that an armature pole is made that wide that it covers exactly eight stator poles. This means that an armature pole has a width such that it covers an angle of 80° and that the groove in between the armature poles covers an angle of 10°.

The fluctuation of the sticking torque is caused by the fact that there is a small slot in between the stator poles to be able to mount the coils in the stator. If an armature pole has an exact angle of 80°, there will be always the same overlapping area in between the eight stator poles and an armature pole and so the sticking torque will fluctuate only a little depending on the position of the armature. But if the width of the armature pole is for instance 80° minus the width of a stator slot, there is a position for which all eight stator poles are overlapped but there is also a position for which the overlapping area is less. In this case, a preference position will occur when the overlapping area is maximal because the magnetic flux flows easiest for this position. 

[SparWeb]

An important adjustment to be made to my diagram above, before it can be used.  Thank you AK!