Use of an asynchronous motor as generator for a wind turbine

[Adriaan Kragten]

An asynchronous motor can be transformed into a PM-generator by replacing the original short-circuit armature by a PM-armature. This procedure is described in my public report KD 341. However, the price of neodymium magnets has increased substantial in the last years and manufacturing of a PM-armature is a lot of work. So it would be nice if an asynchronous motor could be used without modification or with only a small modification.

There are two ways how an asynchronous motor can be used as a generator. The most common way is to use a grid connected asynchronous motor at a rotational speed higher than the synchronous rotational speed. This way was used for the first Danish grid connected windmills. I have described this way in my public report KD 578 for the VIRYA-6.5 windmill. The main disadvantage of this system is that the rotational speed of a 4-pole asynchronous generator varies in between about 1500 and 1550 rpm depending on the load for a 50 Hz grid and so there is only one wind speed for which the rotor runs at its optimum tip speed ratio. This wind speed is called the design wind speed Vd. Mostly Vd is chosen about 7 m/s. At wind speeds lower than Vd, the rotor runs at a tip speed ratio which is higher than the optimum value and at wind speeds higher than Vd, the rotor runs at a tip speed ration which is lower than the optimum value. The average Cp is therefore substantial lower than the maximum value. Another disadvantage is that the connection in between the generator and the grid has to be broken if the frequency of an unloaded rotor is lower than 50 Hz because other wise the generator starts to work as a motor and energy is extracted from the grid. These are the reasons why this system is almost no longer used.

However, there is another way to use an asynchronous motor if the generator is not grid connected. For this use, the generator has to be made self starting. This is done by using an load build up from capacitors and resistors. The short-circuit armature has a certain residual magnetism and this magnetism triggers the R-C circuit. As the residual magnetism is very low, the rotational speed for which the triggering starts can be very high and may lie close to 1500 rpm for a 4-pole motor. This rotational speed can be lowered a lot by milling four small holes in the armature and by gluing four circular neodymium magnets in the holes such that two small north poles and two small south poles are created.

This system has some major problems. One is that the power is dissipated in the resistors. Another is that the required value of the resistance depends on the rotational speed. So one has to increase the resistance if the rotational speed becomes higher. So it is not possible to simply replace the resistors by a battery as the load resistance of a battery decreases at increasing rotational speed (see figure 4 report KD 595). But may be a clever electronic engineer can solve this problem. I am a mechanical engineer and don't know enough about electronics.

A description of this method and wire diagram of the R-C circuit is given at page 79 in the Dutch report Windbouwwerk 1 written by Fons de Beer already in 1982. I have translated this page into English and added this page as an attachment. One should realize that the asynchronous generator isn't used direct drive but that an accelerating gear box is used to get a sufficient high rotational speed. We are now 40 years later and it might be that at this time it would be done differently but this information might stimulate current electronic engineers to develop something with which it is possible to use an asynchronous motor for battery charging if only four small neodymium magnets are glued in the standard short-circuit armature.

 Translation page 79 Windbouwwerk 1.doc (165 kB - downloaded 53 times.)

[Adriaan Kragten]

There is a reason why the four magnets which can be added to the short-circuit armature of an asynchronous generator can only be small. Once these magnets have triggered the R-C circuit, large currents will flow in the aluminium short-circuit bars of the armature. These currents create a strong 4-pole magnetic field in the armature. This strong magnetic field is rotating slowly with respect to the armature. So at a certain moment, it coincides with the magnetic field of the four small magnets but later it works just opposite. So the final magnetic field is fluctuating and this means that also the torque is fluctuating. Only if small magnets are used, this fluctuation is limited. The rotational speed at which triggering starts, can also be reduced by taking a 6-pole or an 8-pole motor but this has as disadvantage that the voltage generated at a certain rotational speed is much higher than for a 4-pole motor.

[SparWeb]

Adriaan, in your write-up there is a reference to an optical coulping to the triacs.  I can't see that in the wiring diagram, however.  Are they internal to the SSRs?

[SparWeb]

Still curious about the SSRs.  I've tried looking them up on Siemens website.  Since the p/n in the book is 40 years old, the part is off the market, of course.
Do you know a way to specify an alternate.

I'm reading the rest of this and it's very interesting.  I think I understand what it's doing well enough that I would like to know more.  I especially like the controllability in both coarse and fine directions that it offers.

[SparWeb]

This might be an alternative, and currently available:

http://www.crydom.com/en/products/catalog/cw-series-24-ac-panel-mount.pdf

[Adriaan Kragten]

I have translated this page out of that book already in 2005 and recently I have added only the given figures. I have made this translation because an Indian company had also this idea and did some tests with small PM-magnets to reduce the rotational speed at which the R-C circuit is triggered. If I remember well, they found about 600 rpm for a 4-pole motor. I don't understand what is written and so I can have made mistakes during the translation. The author of the book is no longer alive and so I can't ask him questions about this wire diagram. I own the report but the remaining part tells only about the mechanical construction. The report is already 40 years old and the used electronics might be outdated but the idea is clever and it would be nice if it could be used. More information might be given at Google or Wikipedia. I have found the recent article: Assessment of Capacitance for Self-Excited Induction Generator ----- at: www.mdpi.com/1996-1073/11/10/2509/htm

When I made the translation in 2005, I had a discussion about the idea with an electronic engineer which has built his own windmill and I made some notes about this discussion. My question was if it would be possible to rectify the current and use a battery in stead of resistors. We found that the DC voltage was about 528 V for a phase voltage of 230 V and this is a very high battery voltage. This voltage can be reduced by a factor two by connecting the first and the second layer of the winding in parallel in stead of in series but this still requires a very high battery voltage. So my idea is to use the generator direct drive at a much lower rotational speed. However, then something must be done to reduce the rotational speed at which the R-C circuit is triggered. Using small permanent magnets to do this is a method which has been proven to work but I don't know if the rotational speed can be reduced enough for direct drive use.

[Mary B]

This might be an alternative, and currently available:

http://www.crydom.com/en/products/catalog/cw-series-24-ac-panel-mount.pdf

Looking at the schematic yes they will work, get the 3-32 vdc control voltage relays and match the current to your needs. The schematic version a 10 amp would be plenty so the CWD2410 would work. Get a heatsink for mounting, they do not like running hot!

[SparWeb]

Asynchronous motor/generator units need to be figured out by DIY builders.  The source motors are plentiful, and in my experience they are free for the asking in some places.  I personally would like the challenge.

I am not ready to say that rare-earth permanent magnets have had their day, but I am concerned that high prices will remain high, and only grow as demand for EV's drives the price upward.  Until 10 years ago, many if not most DIY alternator and motor-conversion projects featured here were built with less than a hundred dollars in magnets.  The same build would cost much more in Neo today.  To keep these projects accessible to tinkerers, we should look at new technical concepts. 

A bank of 3 starting caps (370V & 30uF) costs 60 USD.  If the motor self-excites without conversion, then this also eliminates re-machining the rotor. The project becomes vastly more accessible to the DIY builder.  I haven't explored this way to make a wind turbine because of my own ignorance and probably the lack of a trailblazer who published all the details of how they did it.  Maybe that trailblazer is already out there, I just haven't looked in the right places.

If it's of interest Adriaan (and if you can forgive me) here is a re-work of your translation.  I did have to reinterpret some phrases that originally used Dutch idiom, which cannot be translated literally, but rather by choosing a suitable English idiom instead.

 Translation page 79 Windbouwwerk 1.doc (164 kB - downloaded 52 times.)

[SparWeb]

A bit more research tonight...

This has (of course) been studied in detail over the years.
https://www.researchgate.net/publication/3462819_Wind_Energy_Conversion_Using_A_Self-Excited_Induction_Generator
https://www.academia.edu/20662997/Maximum_power_extraction_from_self_excited_induction_generators
https://my.ece.utah.edu/~bodson/pdf/Analysis%20of%20Triggered%20Self-Excitation%20in%20Induction%20Generators%20and%20Experimental%20Validation.pdf

I was somewhat intrigued to find a number of papers published in the 1980's by researchers at University of Calgary.  Local to me, but probably long gone, and I've never seen a trace of their handiwork up on poles around me here.  Maybe a little digging will turn up more interesting local connections.

A wind turbine derived from an induction motor (or asynchronous motor as the academics has put it) has advantages and disadvantages. 
Good: It is simple, cheap and robust.  It's easily connected to a variety of loads and regulation systems.
Good: it can be operated in variable speed or constant speed modes, depending on the system that loads and controls it.
Good: the electrical circuits can be excited by capacitors.
Meh:  Its self-excitation also depends on variable magnetism in the rotor.
Bad:  Common 4-pole motors self-excite above 1800 RPM.  That's too fast for a wind turbine.
Bad:  If the load changes, the upstream effect on the wind turbine is unstable.
Bad:  It responds to a resistive load very differently from a reactive load.

The typical solution for the speed mis-match is a gearbox.  Those of us who've been watching wind turbine technology for years have seen the WT graveyards in California, many of which suffered from inadequate gearboxes.  The ideal DIY solution does not have a gearbox.  Matching direct drive rotors that need to run between 200 - 600 RPM to an induction motor suggests that somehow the induction machine needs a 36-pole rotor.  I don't know how to do that.

There is a silly idea in my head that maybe the magnetizing current can be stimulated by a step-up transformer on the capacitor bank.  Even if there is some sense to the thought, my intuition says it will blow up the capacitors if something isn't done to protect them.

Variable speed isn't a problem.  Many DC and rectified-AC wind turbines (like mine) are variable speed, and well-matched to the cubic wind power curve due to this freedom.

Adriaan has already pointed out the solution to pre-excitation: insert a few small magnets on the rotor to guarantee that a magnetizing current in the rest of the rotor is generated by the capacitor bank.  They don't have to be large or expendive, though it is a permanent modification to the induction machine.

It feels tantalizingly close doesn't it?

[SparWeb]

For readers' convenience, these figures are provided in the chapter Adriaan has translated:

This diagram shows the induction generator, a cut-out switch, and the capacitor bank.  The output terminals are labeled R,S,T.
Note that the generator is connected in Wye, and the capacitors are in Delta.


This diagram shows the system controls starting from the R,S,T terminals.  It accommodates the variable speed of the wind turbine by progressively switches on resistance loads.  Note that load B1 seems to activate first while B3 is last, so during the transition the 3 phases are unbalanced.  Given the difference between the three 500-ohm resistors and the 50K resistor, all in series, there isn't a very wide transition period.  For the human observer they would seem to come on almost simultaneously.  At the turbine, however, the increase in load by 3 steps would allow the rotor speed to continue increasing rather than switch in a high load on all 3 phases at once.

[Adriaan Kragten]

I still believe that a PM-generator made from an asynchronous motor is a good option but the rising price of neodymium magnets is a problem. The costs of the magnets can be reduced if thin magnets are used. In my report KD 718, I give an armature construction for which only 5 mm thick magnets are used. The armature has 16 poles mechanically but four poles physically. A big advantage of a permanent magnet armature is that the magnetism is always there and so the generator can be used as a brake. If the original short-circuit armature is used it might be possible to trigger the R-C circuit at a sufficient low rotational speed but below this speed, no torque can be supplied and therefore the generator can't be used as a brake. The advantage of using no magnets is that the generator will have a very low sticking torque which is only caused by the friction of the bearings and the seal on the generator shaft. So starting of the rotor won't be a problem if a rotor with a low starting torque coefficient is used.

Another way to prevent neodymium magnets is to go back to the much cheaper ferroxdure magnets. However, these magnets have a much lower remanence (about a factor four lower) than neodymium magnets and a sufficient high magnetic flux in the air gap can only be gained if the magnetic flux is concentrated. One of my first PM-generators was using ferroxdure magnets. It was a 12-pole generator and the magnets were placed in radial groves. So the magnetic flux out of two rows of magnets is flowing through one armature pole and this gives enough concentration to get saturation of the stator stamping for a 12-pole armature. A disadvantage of a 12-pole armature is that now it is no longer possible to use the standard motor winding. This was one of the reasons why I switched to a 4-pole motor but then only neodymium magnets give a sufficient strong armature.

[mbouwer]

In the 80's I made, together with 2 windmill friends, this 5 meters blade diameter windmill.
It was a model of the Newecs 45 with 220/380 asynchronous motor as generator.
Each phase fed a 500 Watt heater.

[SparWeb]

That is a beast, mbouwer!
The tower for that would be a tremendous project itself.

Do you have any records of the controls it used?  Did it have capacitors to supply magnetizing current?

[Adriaan Kragten]

The book Windbouwwerk 1 of Fons de Beer describes two windmills, both with 2-bladed rotors with blade pitch control. The smallest rotor had a diameter of 3 m and the biggest one had a diameter of 5 m. The construction of generator, the gear box and the rotor shaft bearing housing of the 5 m one was very similar to the 5 m one of Bouwer as presented in his photo. I have made a copy of the windmill given at the front page of the book and added this as an attachment. The head is turned in the wind by a wind servo. The central tower pipe could be lifted with a winch in the lower 6 m high lattice tower.

[mbouwer]

@ Sparweb,
Height of the mast was 9 m.
The capacitor value was calculated with the booklet of J.P.Covent: De Asynchrone Generator.

[kitestrings]

That tower to mast design, in Adriaan's photo, looks familiar.  I think SparW's & ours are very similar.

Was the winch for some sort of braking or pivoting out of the wind?  ~ks

[joestue]

What if..

You take 3 variac transformers. 0 to 140 volts.

3 Capacitors sized for self excition gets connected across the 0 to 140v coils. The 0v connection point of all three transformers goes to the Y point of a 220v motor.

Movable tap goes to the motor coils.

Self starting should be far easier.

Need some creative way to self regulate.

Maybe ditch the variac and use relays and multi tapped transformer

[joestue]

feed the 3 phase output without capacitors directly into a vfd, then use an Arduino or some other mechanism to measure the rpm of the turbine, and drive the vfd's commanded frequency to be some value less than the wind turbines rpm.

in fact you could have an aux tachometer on the turbine drive a small transformer, rectify it, and send it to the vfd's 0-10v input.

the small transformer would prevent emi from destroying the vfd in the event of a transient.

a high voltage tachometer on the turbine could be used to power up the vfd.

then take the 400 volts from the vfd and run it through a fixed ratio buck converter such that the batteries are kept at 52 volts or whatever.

the vfd doesn't care what the dc bus voltage is, the power transferred into the battery will be a function of the difference in rpm between the vfd and the turbine.

a diode between the buck converter and the vfd will ensure that the vfd can power the buck converter but not the other way around. this way start up is dependent on wind to drive the turbine, not the battery to drive the vfd.

the vfd brake resistor could be configured in a way to turn on the buck converter, that way its not running all the time.


a mechanical brake is a requirement for this system in the event the vfd burns up.

[Adriaan Kragten]

That tower to mast design, in Adriaan's photo, looks familiar.  I think SparW's & ours are very similar.

Was the winch for some sort of braking or pivoting out of the wind?  ~ks

The winch is used to lower the tower pipe such that the head is at the top of the 6 m high lattice tower. In this case one can adjust the safety system of the rotor using a 6 m high ladder. So the winch isn't used to rotate the tower pipe.

[kitestrings]

Oh yes, I see it now.  So, is that is a pulley mounted on the bottom of the pipe?  Is there an inner and outer pipe perhaps?

[SparWeb]

Today I spent some time doing the math.  The research I did the other day turned up some methods to calculate capacitor sizes needed to do this.

One thing that kept getting in the way is my desire to self-excite the generator at low speed, low frequency.  All of the papers, articles, and books I find keep the frequency constant.  Whatever domain they are in (50Hz, 60Hz, or 400Hz) they don't work out or even consider the effect of altering the frequency at which the motor operates as a generator.

That took some work to get past, but once I did, things made sense and I could work out the rest.  However, I also found a problem.

When working out the capacitor values for a typical induction motor at, say 60 Hz, the inductance of the windings is about 10-100 milliHenrys and that can be compensated with capacitors in the range of 10-1000 microFarads.  This stuff is pretty common, even in varieties that can resist 200 or 300 volts.  This quickly goes out of whack when the frequency drops.  The inductance in the motor winding is constant, and this crowbars the capacitor value to a very high range.  Whatever capacitor value you had at 60Hz will increase to the power of 2 as frequency is reduced.

Say you start with an induction motor whose rated speed is just below 1800 RPM.  Actually lower, with slip.  It's a 4-pole motor, and this is the speed it takes when fed 60Hz.  Now, you want to use this induction motor on a wind turbine and have it cut-in at 180 RPM.  Direct-drive, no gearbox, unlike the example above.  Then the frequency is reduced by 1800/180=10 therefore only 6 Hz.  With a frequency reduced by a factor of ten, the induction in the windings will have a lower reactance, by a factor of 10.  When calculating the capacitor value, the matching lagging reactance (10 times lower) multiplied by the frequency (also 10 times lower) requires you to choose a capacitor 100x higher rating than before.

Your 200 uF capacitor just became a 20mF capacitor.  Good luck finding that with a 200 Volt rating, or higher.

[SparWeb]

But maybe...

Adriaan suggested that adding a few permanent magnets can help get things going.  That would put a small field into the windings, and it may also benefit from cogging tooth to tooth.  The magnetizing current wouldn't be entirely reactive.  There's a thin thread of hope.  I believe this can be tested, but I barely understand what I'm writing about.   

[joestue]

so the volts per hz at the induction motor is fundamentally limited by the saturation curve and the air gap inductance, and copper losses.

but the energy available to drive the air gap flux is limited the inverse cube of the rpm assuming constant tip speed ratio.... so.. simply plot that curve and find the volts per hz curve that should actually generate power.

a typical 5hp induction motor is going to be 200-300 watts of losses just to idle at 230v 3 phase, 2-4 amps. full load current is 12-14 amps.

but when running backwards you need about 10% more rotor watts than that, because of the optimizations and geometry made for operation as a motor, is now running backwards. so you are going to burn up, call it 300 watts, of copper losses in the stator, just to reach net zero  at nominal rpm and flux density. so if you size the generator, rotor, gear box, etc, such that your 5hp motor is consuming 6hp from the wind, at 1750 rpm, it should work fine and will deliver 5hp into the grid at 230v 60hz at 80% efficiency. which is actually pretty damn good.

but at lower rpm you need to drop both the frequency, and the voltage very quickly, due to the power dropping out with the cube of the windspeed.

at half the nominal rpm the power available is but 1/8th. instead of 6hp of wind you have .75 hp, or just 500 watts. in theory you could make this work, burning up 200-300 watts and you would get just 200-300 watts output, but that's really not good enough. assuming the losses are all copper losses you would want to drop the voltage about 25% such that 50% less energy is stored in the air gap inductance, thus dropping the copper losses from 200-300 watts down to 100-150. this would increase your output back to 250-300 watts.

at 1/3rd nominal windspeed you have but 1/27th the power available and that may not overcome the friction in the gear box!

[mbouwer]

Would it be possible to make your own slow running asynchronous generator?
Starting with a lamination package with lots of teeth.

[joestue]

the frequency doesn't matter, its the air gap that has to be magnetized, and to do that takes current flowing in copper coils.

with magnets the "current" flowing in them, continuously, is free.

[Adriaan Kragten]

The frequency can be halved by taking an 8-pole asynchronous motor. So then the synchronous frequency is 50 Hz for n = 750 rpm. But if you want that the R-C circuit is triggered at for instance n = 150 rpm, the frequency is only 10 Hz and it might be that this requires capacitors and resistors of extreme value and may be also eight rather big PM magnets.

I remember that I have visited a guy very long ago which had a 5 m windmill on a houseboat, about like the one as described in the report. So it had an accelerating gearbox and a 4-pole asynchronous generator with capacitors and resistors as load but not with an automatic system which regulates the load. He told me that resistors (and may be also capacitors) have to be added at increasing wind speed. One day he wasn't at home and a thunder storm came over. His wife was at home but she didn't add resistors. So at a certain wind speed the rotor torque became too high and the rotor started running almost unloaded. This resulted in blade flutter. He used a 2-bladed rotor made out of one wooden beam. The rotor survived but the blade tips were cracked that much that they looked like a paint brush.

[SparWeb]

Electrically, this would work better with a high pole count motor, I think.  As Adriaan points out, using an 8-pole motor instead of 4-pole makes the reduction factor better by a factor of 4.  If possible to find a 12- or 16-pole motor, then the problem would become more reasonable.  But those are quite rare. 

Also, for more poles in an induction motor, the size of the frame gets larger.  If you have two 3 HP motors, the 8-pole version will weigh 2x more than the 4-pole version. If you were making a choice between either a high pole-count motor or a gearbox+low pole-count only on the basis of mass, you would choose the gearbox and a 2-pole motor.

[mbouwer]

So make an asynchronous rotor in this lamination package??
(60 teeth)

[SparWeb]

That's one way to do it.
For now I'm thinking about ways that the induction motor can be used without even disassembling it.

I've discussed this with a friend who knows a lot more about motors than I do.  He reminded me of the Volt/Hz ratio that should be kept constant when changing the operating frequency.  This helps me a lot because dropping from 60 Hz to 6 Hz is equally proportional to dropping from 240V to 24V.  Since 24V RMS is exactly the operating frequency of my current wind turbine as it is rectified to 28VDC, there might be a very lucky coincidence, here.

There may in fact be a convergence of these factors, allowing for the more reasonable values of capacitance per phase to work.

We are planning to meet in the coming week or two and attempt a test using my lathe as the prime mover to drive an induction motor.  The motor hasn't been modified at all, and as a pleasant coincidence it happens to be a 6-pole motor, which should help.  The motor also has wiring that allows for connections at either 240VAC or 480VAC, allowing us to change voltage ranges and (implicitly) change the inductance per phase, looking for combinations with the greatest advantage.

Stay tuned!

[joestue]

I found a paper that covers self excited induction generators very well, at a level I think a few people here can understand

https://unescochair.bntu.by/sites/unescochair.bntu.by/files/energy/edu_docs/chapter_16.1_asynchronous_genertators.docx

I don't think a diy high pole count induction generator is a good idea as the air gap is going to be too large, unless you have a lathe with a tool post grinder that can grind your diy lamination stack to match an existing induction motor rotor from a large motor, and get the air gap down to .005".

methods to center the rotor in the ground, diy stator can be as "dumb" as wrapping a sheet of thick paper around the rotor, fitting it into the stator and using epoxy to set the bearings in place. can this hold up? probably. you could drill and ream for roll pins to hold the thing together.


anyhow i'm reasonably confident that a 5hp motor or larger could be adequately "controlled" using a 60$, 2 hp vfd to provide the reactive current needed to excite the machine. the issue is parasitic losses at low windspeeds and programming the vfd.

you can incorporate a star delta switch to transition between high and low windspeeds.

in fact i would theorize that the series star, series delta, double star, double delta connection which gets you a 3.4, 2, 1.73, 1 volt ratio can satisfy self excitation into a rectifier load across a 3:1 ratio of wind speed.

big failure point for a lots of relays solution to change the motor windings is.. it might work with a pmg. but with a self excited induction machine, a stuck relay is going to kill the excition of the machine before the relay contact burns up and open circuits. in a pmg there is some redundancy and the system would work with a few stuck relays or open coils except on the highest voltage (in which case the machine is now single phase)

[mbouwer]

@ joestue,

Turning out the laminations ( en afterwards shaping the teeth ) did not cause any problems.
And it seems to me that then it is also possible to make a rotor yourself.
I think someone with expertise can design and calculate that.

[Adriaan Kragten]

Electrically, this would work better with a high pole count motor, I think.  As Adriaan points out, using an 8-pole motor instead of 4-pole makes the reduction factor better by a factor of 4.  If possible to find a 12- or 16-pole motor, then the problem would become more reasonable.  But those are quite rare. 

Also, for more poles in an induction motor, the size of the frame gets larger.  If you have two 3 HP motors, the 8-pole version will weigh 2x more than the 4-pole version. If you were making a choice between either a high pole-count motor or a gearbox+low pole-count only on the basis of mass, you would choose the gearbox and a 2-pole motor.

The maximum torque level depends on the armature volume. If you compare a 4-pole asynchronous motor and an 8-pole asynchronous motor of the same frame size you will see that the 8-pole motor has a larger inside diameter of the stator stamping. So an 8-pole motor has a larger armature volume at the same length of the stator and therefore a larger maximum torque level. The power of a 4-pole motor is much higher than for an 8-pole motor of the same frame size because it runs at about the double rpm. So if both are used without a gear box, you can better take an 8-pole motor. The frequency at a certain rpm of an 8-pole generator is a factor two higher than for a 4-pole generator

[Mary B]

With the number of ECM motors out there now that are PM is it worth the effort for small wind? I could see doing this if you want 2000 watts or more...

(edited to make it ECM... caught that this morning...)