Wind MPPTs

[Yyrkoon]

Hello all,

As I have said in previous posts, my knowledge of circuitry/electronics is rather limited. I guess I may know some basic stuff, but basically what I know is enough to get myself into trouble concerning whether I really know what I think I know. I do however have a friend who is an Electronics guru, but his lack of interest/knowledge concerning wind generation; I need to explain a few things to him before he can help me design a controller. A bright side to this is that I do have some experience programming embedded devices(not a whole lot but some) So K.I.S.S. concerning using a programmable device, is not too much of an issue for me. Another problem I have is when people start talking technical I tend to get lost(lots to know concerning wind generation). So, with that said, let me lay out what I think I know, and then if someone would be kind enough to correct me where I am wrong, and fill in what I may have missed . . . any help would be greatly appreciated. SO here goes.

The idea of MPPTs is to maximize power output from a power source. The context I will be speaking of here concerns wind power. What I think I know is this:

A machine that is loaded 100% of the time depending on wind speed may not be working at its full potential. Since there is little power to be obtained in wind under 6-8mph, we may want to leave the machine unloaded until we obtain enough speed to keep our blade RPMs high enough to maintain a starting load. At this point, we figure what is an acceptable load to keep the RPMs up, letting through just enough power to maintain our present RPMs. Wind speed then increases slightly to allow us to increase the load, we then check what is an acceptable load while maintaining our RPM, and if so allow the load to be drawn. Our MPPT does this at set intervals until we approach an optimal/safe rotation speed.

Assuming the above is correct, how does this keep our TSR in range? How can a MPPT keep an alternator at high RPMs from losing efficiency thus keeping the stator from overheating ? On the latter here I am thinking that since a given machine could put out more power than it was designed to put out 'safely', at some point we decide on a 'current cut-off' and allow no more current to be drawn. Also I am thinking that this has potential for an over speed condition, where our concerns would be failing mechanical parts (blades, bearings, etc) rather than electrical. This is where I am assuming braking(if at all possible to be done safely/smartly), or more likely furling would come into play. I would also assume that rethinking a furling system because of a different load scheme would be necessary.

What have I missed ?

[Flux]

A true mppt will look for the maximum power point at each wind speed and adjust its load to keep the prop on the peak of the power curve.

In theory the tsr should remain constant for a given prop and the mppt should keep the prop rotational speed directly proportional to wind speed. If for some reason the maximum power point of the prop doesn't track with wind speed exactly proportional to rotational speed it will set the rotational speed to a value to match the maximum power point. It should work for any mill.

Wind changes speed frequently and violently and in less than a second the power of the prop can double, it is a tall order to track things that are so unpredictable especially as at each point you nave to measure power then change the controller pulse width then look again to see if it is better or worse. If worse you go the other way. if better you try again until you go too far then you move back. With solar this is reasonably easy as things change slowly.

I am not sure how far this has been perfected for wind. I suspect most schemes are very slow and track the average maximum power not the true peak.

The alternative that works for a given machine is to measure prop speed and adjust your controller to produce a given load for each rotational speed and you set it up so that the alternator power curve loads the prop to its optimum point all the way up the range. It only works for a given machine that it is set to track and if other things change such as battery volts it may not be a perfect track of the peak power point.

This is the method I use, it is not as convenient or versatile as a true seeking mppt but it works well enough as the prop doesn't have a sharp peak at the peak of the power curve and a little variation in tsr has no real effect.

These controllers use a buck or other variable ratio dc converter and altering the pulse width lets it work in a similar way to a transformer. If the alternator speed is high, the voltage will be high and it will produce the same power into a low voltage battery ( the amps go up in the battery circuit to maintain the same power at low volts).

Because the voltage is rising with speed the alternator efficiency stays well up so heating is much reduced. When eventually you reach the alternator limit you are likely near the comfortable mechanical limit of the prop and furling will keep under control. There is no need to limit the converter current and with furling it is not advisable to do so as it will run away with reduced load.

A speed based pitch control can be let to run at a speed that the alternator can handle but in this case current limiting although a waste of power would cause no trouble.

Again in the ideal conditions where prop speed tracks wind speed the alternator volts will track wind speed and the power level will track wind speed cubed.

In a battery charging scheme where volts are effectively constant the battery current will track wind speed cubed for a perfect alternator. In real life the alternator inefficiency means that the track is nearer speed squared and I simply use a servo loop tracking the output current against a signal proportional to speed squared from a multiplier. By fiddling the loop gain it can be made to rise a bit above speed squared if the alternator is very efficient.

There you are nice and simple, so much easier to achieve than furling? Sadly no furling mills have been around for about a century, mppt is just around the corner?

You are new at this there is plenty left for you to perfect, it's not too late.

Flux

Flux

[Yyrkoon]

Ok, so what we're talking about is matching voltage to current at any given rotational speed for optimal power 'conversion'. Understanding this concept with solar seems to be easy for me(assuming I actually understand it fully), but not quite able to visualize this with a wind machine.

Maybe I should read some more before asking further questions.

[Flux]

I think you have got the idea more or less.

With solar you need a device that matches the peak power point of the cells to the battery voltage. The peak power point voltage changes with temperature and the battery voltage changes with state of charge but most of the time the changes in both are slow.

The converter requirement is that you extract power at the best available voltage and convert it to the best amps for the battery volts at that instant. The tracking bit just alters the ratio to cater for temperature battery volts and light level.

The converter requirement for wind is the same, transform power at one voltage to another with minimal loss.

The tracking problem is more complex as you are tracking a linear speed variation against effectively a cube law power requirement so the input voltage variation will be an order of magnitude greater than the solar equivalent.

The easy way is to preset the thing to suit your alternator and blade requirements and that is adopted by most of the grid tie units.

Full instantaneous peak power tracking is rather challenging. I see a mid way approach as likely to be the simplest solution where you programme a basic tracking curve from available information and you use the tracker bit to look for improvements over time and it can learn better characteristics than your initial guess as it goes along, much like the engine management things on cars. Your initial attempt may not be much better than "limp home mode" but it will get there and if something changes over time like blade damage affecting prop output it would again get back to the best it can in a short time.

The present method of dumping an alternator directly to a battery misses the optimum loading so badly that even a crude attempt at tracking without sampling or updating makes an amazing difference.

Flux

[Yyrkoon]

Flux,

I am not 100% sure, but I believe the device that we're using for our PV controller could do this sampling real time. The only problem I am seeing is 'regulating' the wild AC as it comes in from the alternator(for an accurate reading).

Read alt output voltage -> read battery voltage -> tell converter what voltage is required(if any), and send to batteries. This is what I am thinking the simplest form would be. Other variables such as battery temp, battery type, are we float charging, or equalizing, etc come into play if you need/want a more complex controller. I dont even think a RPM sensor would be needed.

[commanda]

Maybe I should read some more before asking further questions.

Flux wrote a very good treatise,

http://www.fieldlines.com/story/2006/3/17/185646/194

Oztules built a boost convrter

http://www.fieldlines.com/story/2007/11/25/22656/782

And I've done some work on this subject,

http://www.fieldlines.com/story/2006/3/12/14840/1315

http://www.fieldlines.com/story/2007/11/25/22656/782

I've just been slack and otherwise distracted to continue working on it at the moment.

Amanda

[oztules]

I think Commanda meant this story:

http://www.fieldlines.com/story/2006/5/4/31543/52124

Rather than this I think.

"Oztules built a boost convrter

http://www.fieldlines.com/story/2007/11/25/22656/782"

[Yyrkoon]

I still think I do not get the tracking part. What I am thinking is why even bother tracking input voltage, when all we need is to know the batteries voltage. That is, in context of a buck/step down converter.

I guess the part that I am stuck on is this: IF the maximum power point is tied to the max output in voltage, then whatever it is does not really matter. Our buck/step down converter should already be variable, and in this case it should take as broad of a range in voltage as possible. Unless this really is not possible, and we need to tell the input side of this converter "hey, x amount of voltage coming in, prepare . . .". On the output stage, we need only track/sample voltage of the batteries to make sure we are giving them their maximum power point.

Does this make sense ? What am I missing ?

Yes, I am looking at this from a programmers view point where it seems very simple. Maybe it is, and maybe it is not, but if I try and tackle this, I will be using a programmable device with an onboard PWM.

By the way, thanks for all the links guys(and gal?), I had not found Flux's discussion previous to his final diary entry. A good bit of the rest is over my head, but that is not your fault I will re-read all until it becomes clear(or my brain melts).

[commanda]

The point you are missing is this:

Instantaneous mill voltage varies as a function of rpm and load current. Added to that, increasing the load current will slow the mill's rpm. In and of itself, instantaneous mill voltage tells you precious little.

In my analog circuit, I have mill voltage and current, and rpm. Mill voltage times mill current gives me power. Differentiating this I can tell whether the power increased or decreased as a function of changing the pwm.  I also modulate the pwm as a function of rpm, to try and track rapidly changing wind conditions. It probably needs a non-linear (exponential) transfer function on the rpm input to more closely track the cubic nature of power vs wind velocity.

Alternatively, you could use an anemometer to give an input of present wind velocity, and alter the pwm to load the mill down to a given rpm for that wind velocity.

Clear as mud, I know. But hope it helps.

Amanda

[Flux]

Amanda's view on this makes perfect sense. I would have to do things the analogue way but I see no reason why it can't be done digitally, it's just that I don't understand digital things, I was brought up in the analogue era.

However you do it digitally the problem is the same. I think you have got hung up on the concept of a voltage regulator. This is not a buck switching voltage regulator. Your requirement for tracking maximum power doesn't want you to maintain the dc output voltage ( that would be useful with batteries fully charged in place of another charge controller).

However you do it, you need to measure the power going into the battery, alter the pwm and see if the power increases or decreases then change pwm to get a better result.

Amanda seems to be using input volts and current to get power and that should be good enough but the power you are interested in is to the battery and efficiency may vary.

For very precise results you may consider the true battery voltage but it will be near enough to assume battery volts are constant and that the output power is proportional to output current. Surely any tracking scheme must include output current or at least input power if you opt for that route.( you need input volts and current in that case)

Input volts are not much use as they change with loading. You need emf for speed and unless you try the solar trick of open circuiting you can't get it ( best not try that trick with wind). You can get rpm from frequency ( or the time interval if you want to work backwards, possibly easier with a processor). Speed is not needed for a true power point tracker but it is useful as when you are right the speed will rise linearly and the power follows its cube. When things are going crazy and you have trouble tracking fast enough you can set the current roughly from speed cubed. When conditions are stable then you can modulate the pwm and find the peak power point .

In the perfect world you can measure wind speed with an anemometer and if you track the prop speed to follow wind speed by altering the loading you will be very close and that would be a simple way. It probably works well enough on clean sites and with large machines with their high inertias, tracking average power against average wind speed may be as good as you can get.

With small machines with fast response and the severe difficulty of measuring the same wind as the prop sees it may be much more tricky and it is far from convenient having to move the anemometer with wind direction or have a bank of them. It would at best only give you a ball park figure and you would still need the pwm modulating method for a final track.

Flux

[oztules]

Flux,

I had thought that if we double the wind speed, we double the rpm. On an Ideal alternator, we would double the volts w=Esq/r... so power goes up with a square of the speed which is linear to the wind speed (I know about the cube for wind)..

However you say above that we are going to cube the output power........ where is this coming from....if the ideal alt is following a square....

"Speed is not needed for a true power point tracker but it is useful as when you are right the speed will rise linearly and the power follows its cube. When things are going crazy and you have trouble tracking fast enough you can set the current roughly from speed cubed. When conditions are stable then you can modulate the pwm and find the peak power point ."

sigh.......What did I miss this time.

lost again.....oztules

[commanda]

sigh.......What did I miss this time.

Normally, we would size the alternator to produce 100% of available wind power at the furling point, then run the alternator at less than what it is capable of at a given rpm to match the power available from the blades.

I think Flux presented some good graphs back in "Matching the Load".

Amanda

[Flux]

I thought we were discussing mppt for battery charging. For heating loads directly connected then power does follow speed squared, that is why no single resistive load works properly for heating. You need the load to come on faster as the speed rises.

Going back to battery charging, with direct connection and a highly efficient alternator the speed will not be able to rise significantly, if you were 100% efficient the speed couldn't rise and power would just try to shoot up if the prop could manage it.

With a tracker and converter you need to start phasing the converter back right above cut in to avoid stall. The load needs to come on in such a way as to keep the prop speed rising with wind speed. If you have an ideal prop and you stay on top of its power curve then the power absorbed from the prop should rise as speed cubed. Because of losses the power into the battery will not quite follow speed cubed but should be aiming for near it. With the sort of losses I have about and the characteristics of my props which seem to want to run at lower tsr on higher loads the thing seems to work out to tracking power with speed squared but the actual figure will depend on the prop and alternator and converter loss. If you aim for a curve rather than a true seeking mppt you probably won't get perfection but so far I haven't seen a true wind mppt tracker actually working to the point where I could say it is a commercial proposition.

It seems almost there and when it comes it will be simple to use and will work with any machine without set up but for now my approximate scheme has been working for about a year. Sadly I haven't been able to collect reliable data from it for reasons I didn't expect. Without accurate data I am reluctant to say much except that it is a 6ft machine and manages over 30A into a fully charged 24v battery in winds that are claimed to be about 25mph from the local airport and that is higher than the mill site.

It seems to start producing at about 6 mph.

Flux

[Yyrkoon]

Well . . . I know this is slightly off subject but; after talking to my buddy for a while about mppt's in the context of a PV array, he is telling me "do not believe everything that you read", and then told me that this is not how PV panels work. He also says you(as in me) need to understand batteries. Batteries per him, are self regulating, will clamp the voltage down, and convert excess voltage into amperage. Then of course if you put too much voltage into a battery, you risk 'cooking' it. All this I pretty much already knew, but was under the impression that if you have a panel that puts out 20v to charge a battery at 13v, 7 volts of this would be wasted if not converted to current(under ideal conditions close to 2 amps if the panel is rated 20v @ 3.35 amps). This after reading Blue Sky Energy INC's mppt spiel at the following link.

http://www.blueskyenergyinc.com/pdf/Blue%20Sky_What%20is%20MPPT.pdf

As an example, he tells me a voltage rating of PV panel are measured open circuit, amperage is measured basically shorted(shunted I think he said). If you measure voltage open circuit, amperage will fall, and if you measure current shorted, voltage will fall. Now, since he is the Electronic guru, I have to defer to his judgment, and assume I have read too much into this whole mppt 'thing'. Perhaps, what I did not think about is that even though a PV panel can be rated at 20v @ 3.35 amps, there is still a happy medium there somewhere, its just not 20v @ 3.35 amps. This would probably be where the 'tracking' part kicks in. If this is true, then his statement "this is probably not worth the hassle" is probably not far off the mark. At least concerning PV panels.

Back on topic now, I can not help but wonder if the 'hassle' is justified concerning wind turbines. I still can not help but think that just putting in a smart(possibly battery tracking) buck converter would be where I should call it good enough. Sure, I can still see some good coming from using a true mmpt with a wind turbine, but is that added cost, time, and effort really worth it ?

[GWatPE]

The MPPT problem is well understood.

I tackled the problem with a boost maximiser.  

I built an analogue version for my axial flux mill.  This used a current sensing voltage controlled shunt to regulate the pulse width modulator.  The shunt was controlled by the windmill rectified output voltage.  This cct gives an 8fold increase in battery current for a doubling of windmill output voltage, across a 6fold change in windmill rpm into a 24V nom battery bank.  I have published output graphs on "www.thebackshed.com".

I have made a digital version that has sensing of the total battery current and voltage.  There is battery SOC compensation.  I adopted the parallel boost cct configuration with the main mill rectifier.  This was set up on an iron cored mill to lower the cutin rpm.  The digital version has been checked at 100 readings per second, and only exhibits 1Vpp, oscillatory behavior at the 1-3W mill output power level[approx 12V rectified mill voltage output, nom 24V battery, approx 550W maximum power].  All of the maximum calculating type prototypes I have designed, hunted for a maximum, and overloaded, or underloaded the windmill.  I eventually settled on a windmill loading profile programmed into the micro.  

I would design a windmill for the load, at the upper power levels, and only fix the bottom end with a boost maximiser, where the windmill is turning, but would not normally be producing power.  

A MPPT can easily over, or under load a windmill.  The rapid changing of power levels with wind gusts, and the effects of furling mechanisms compound the control process.  My estimate would be that digital MPPT would require over 1000 calculations per second, with an output of at least 10bit step resolution.  A boost maximiser reduces the complexity and number crunching required approx 8fold.

Gordon.

[Yyrkoon]

Yes, I understand that, and above what commanda was saying is what I believe what would have to be done; Have a tach sensor and voltage readings so you know for a fact that voltage increase was from a wind increase, etc. Same for lowering voltages. The more thought I put into this 'real time mppt' concept, the more complex it seems to be, hence my post just above.

I am still wondering if tracked battery voltage + buck converter would not be just as good; assuming we(or I) felt that a real time mppt was just too much hassle. Then like what Flux sugested above with a 'pseudo mppt', you program it to change settings every so often. 10, 20 steps, whatever you feel like. This may not be elegant, but if you only have one mill would it matter ? Even if you had planned on building a small wind farm it would not be too much hassle to replicate if you kept good notes. Anyways, I do not see why this would not help the mill maintain speed, and keep it more efficient than just dumping the load into a controller with optional dump load . . .

[ghurd]

Maybe someone recalls who I am talking about, and can post links?

College EE kid in TX, made and sold a PCB for wind MPPT, tested it in real life before posting, posted here and at gotwind, not very long ago.

I teased him about a cloud over his part of TX on weather channel radar, and spending the rest of his life working at McDonalds, and how I was too old to know what he was talking about.

I think of him as "SW-44"-something.  May have been ham radio related.  

How I think of him is "S&W .44 Mag".  Good luck with that!

I am not so good with names.  

His circuit may be exactly what you seek.

Sorry to S&W-44.  It'll happen when you get past 45?

G-

[GWatPE]

The issue with buck converters is all of the power has to be processed at the highest power level.  The parallel boost cct only requires the boost maximiser to function at the bottom 30% or so, of the wind energy curve, where normally the windmill emf would be insufficient to charge a battery load anyway.  It is more efficient to design the windmill for the load, than to design a cct to fix a loading problem.  A windmill can produce relentless power during storm events, and I have found that rectifier diodes working with a battery and a windmill designed for direct connection to the load at the high power levels, have a better chance of survival than a buck cct having to process the same power.

Gordon.

[TomW]

G;

Personally, I like the Colt .44

My purpose for posting is to say that user is S4W2099 and his Diaries with the ongoing saga are over here:

http://www.fieldlines.com/user/s4w2099/diary

Tom

[ghurd]

Sorry S4W2099!  Still looks like a ham thing to me.

Ouch.  The SW99 is a poor excuse for an abomination from S&W and Walther.

I got a 422 for the wife, if that is any consolation.

But got her the duo-tone .380 Bersa Thunder for CCW.

I'll stick with my 7.5" Super Blackhawk in .44, and it is in blue because I am old school.

Sometimes my blueing wears a bit thin.

G-

[Yyrkoon]

Tom,

Personally, I like the .45 Long colt

Anyhew . . . thanks for the link, sw4's posts are interesting, even if they are over my head mostly

To Tom, and everyone else. Previously I had been under the impression that winds here top out at around 60 mph, but recently learned that winds here during storm events can exceed 80mph(even if the weather channel says otherwise). This is probably how a lot of the wind turbines in this area end up dead door nail. Not just the cheesy Mallards, etc. Winds here are not always very high, but when they do blow I would have to say that 20-25mph would be fairly common. Monsoon season is altogether a different story . . . and winter here can be rough too.

Again, let me reiterate; I am very electronics newb. Also, it probably looks as though I am bouncing around here from one idea to the next(which I probably am), but I am looking for a decent output 'bulletproof' design. What I consider decent output is ~500w at ~25mph, and something that will give me ~10 amps in winds of ~15mph(less than 15mph would be good too). How the buck controller figures in here, is that I have read that you can use them to help stall the machine, which I am thinking coupled with a small-ish machine, and furling could help avoid a catastrophic failure. Also, I am under the impression that a buck converter could help keep an alternators efficiency higher than just dropping the output directly into a regular charge controller(we use Xantrex C60's here) Maybe a boost controller could be used to stall an alternator ? I honestly do not know, but I do see the appeal of running a delta configured coil set, with a boost controller to extract low speed winds.

My very basic concerns are that I do not want to build something(especially from scratch), spending hard earned cash, only to see a storm event rip it to shreds. As Flux has said in one of my previous posts, perhaps I am over thinking things, and worrying too much ? I always like to err on the side of caution though . . .it is just in my nature. Extracting every last bit of power that I can is of course a plus.

Either way, and whatever I do; I am learning much from the comments this post is receiving (and many, many more from searches), and I do appreciate every-single-comment from you all.

[s4w2099]

Yeah, thats me, I didnt make that many of those and it is not TRUE MPPT but its cheap and works. Covered cost of development and bought some books for school. I need to make them more user friendly, they were too crude for my taste.

I am working now on true MPPT in buck mode and it is looking great so far. I am waiting for some solar panels to make a high voltage array and actually test it hard. I ordered 4 of those cheap Chinese harbor freight like panels to connect in series. That would allow long thin wire runs with very little losses.

Some PICs here:

http://www.fieldlines.com/comments/2008/9/1/15516/83721/19#19

Right now I have made tests with bench top power supply and single 12V panel into discharged batteries. There are increases but not that much with that input voltage because when the batteries are approaching a decent charge state the MPP at that temperature can be below battery voltage or very near, making the gain minimal or even loose some. It was a very hot day when I tested with the solar panel.

New panels get here on monday, so sometime next week I will make a post about it. One thing I did noticed is that the buck converter approach was much more expensive than boost converter because of the driving that needs to be done. But regardless of the topology being used the concept of finding the max power point is the same with the "Perturb and Observe" algorithm. That is the formal name for it and it does have the speed problem that flux was talking about. There are other methods of doing this that can be used for fast changing conditions, but I still know very little about them. I will be researching that soon.

[Yyrkoon]

I would be interested in seeing your data, especially if comparing buck vs boost. One thing I am thinking though is that this could not be a true apples to apples comparison unless you used two different machines(cut in differences), and then who sets the standard for a true apples to apples comparison . . .lol. I think I just confused myself(again).

[Yyrkoon]

Point taken on the buck vs boost circuits, and I do probably want to use boost here. It is all slowly sinking in. Today the wind here was an easy 40-50mph steady. My neighbors AirX was constantly braking every couple of minutes(meaning the winds were exceeding its braking point a lot).

Also, according to my neighbor, average winds here are actually 16mph, not the 5-7mph the weather monitoring websites claim(because they count no wind days, where to me this does not matter). So . . . I am thinking I need a slow machine, and if I need a slow machine; I am thinking boost makes much more sense. Before I was thinking about making a fast/small machine, but I am currently rethinking this.

[jimjjnn]

My favorite is .44 mag Desert Eagle. 4 lbs of steel with rotating bolt lockup. Heavy but handles smooth as silk.

[ghurd]

I carry a 9x18 CZ82 and a PA63 (at the same time, w/~50rds total),

maybe a Mak or SW39 too.

DE stuff is just too bulky and heavy.

Smaller output 24/7 is better than larger output 1/25?

G-