WT Power Curve

[SparWeb]

For those of you who haven't found it yet, I have to recommend the book Wind Power by Paul Gipe.  He's been working on these things since they invented wind, and has a lot of experience to share.  www.wind-works.org

I've been thinking of writing a "review" of his book, but I have to finish it first!

Just one thing (of many) that I've been set straight on by his book is how to honestly report performance data.  Wow it's a hard standard to meet!

No wonder nobody does it.

I looked at my data from the PICLOG data logger that I built, and there's a wide gap between what I get and what is expected in a reputable performance analysis.  Just the amount of data that is required becomes enormous if you are expected to be calculating wind speed averages every minute or 10 minutes and correlating with simultaneous electrical power readings.

I've tried analyzing my PICLOG data in various ways, but trying his approach (as best I can with the data I have) is actually a lot simpler than I thought.  It's not the math, it's the reliability that's hard to achieve.

Here's what I can come up with for the old GE motor conversion I had last year, with an 8-foot prop on it.  The 4 curves represent power curves under different conditions, though the prop was the same each time.

The 4 lines differ so much because I was trying different wiring connections.  The first (August in blue) was Series-Star, and the power picked up in even the lightest of winds and the only penalty is that it peaks at about 250W.  The next is with the generator connected in Series-Delta (September in red).  Now more power can be collected, with nearly no penalty in range of useful wind speeds.  Later I re-connected it again, that time in Parallel-Star (December, 2008, in green).  The blue line from April the next year is also Parallel-Star.  Now I have to wonder about what was going on between December and April.  The curves are totally different!

This is where I get back to having reliable data!  On my site, there is a row of trees around the house, and in some seasons the wind has to blow through them to get to the wind turbine, at other times of the year it has a straight-on passage from the north.  The anemometer, only 12 feet from the ground, is exceptionally sensitive to this effect.  If on one day I collect data with a north wind, the resulting power curve will be very different from one I make based on wind from the west or south-west.

I'd like to post similar data from my new motor conversion - especially since it's flying with the same blades, but there's not enough wind to collect meaningful data!

[Flux]

Yes it is indeed a tricky subject.

For good results your anemometer needs to be a couple of prop diameters upwind of the prop and on its centreline axis. This means a different mounting for every wind direction. Just having an anemometer on a stick somewhere in the vicinity is no real use. If it is a very open site with no obstructions it may be of some relevance  but otherwise virtually meaningless.

You need plenty of close wind speed samples, those weather stations that dump a reading once every few minutes are useless.

Paul covers the process very well but if it is still there Hugh Piggott has an article somewhere on his site that gives a very clear indication of the process.

I was fortunate in having a friend with an interest in microprocessors and he built me a simple logger that dumps the 2 second speed readings into 1mph bins of wind speed and rejects those that are blatantly wrong. He has then done me a programme that sort the data out at the end and gives a plot of the power curve.

Even with all this and with the anemometer in the right place I still see variations from different wind directions ( those with lots of obstructions upwind are not to be trusted). I also see differences in the curve depending on the wind during the recording.

Ideally you need a period that covers all wind speeds reasonably well. If you do a run in high wind periods it skews the low wind performance and makes it look better. Similarly readings in predominantly lower winds will skew the high end low. You need representative samples in every bin. If there is not much high wind the samples you get are mainly gusts.

If I ignore the bad wind directions I tend to get fairly repeatable results but I have never had a site good enough to do more than comparative performance tests.

One thing I learned years ago is that if you take spot readings of wind speed and power you always seem to get over optimistic results,

Your performance curves are interesting and are much as I would expect, matching the loading is everything, alternator and prop efficiency matters little until things are correctly matched.

Flux

[Beaufort]

Very good info. on something that will become a very hot topic as people begin to realize they're not getting results from their store-bought systems.  I'm involved with a commercial project on the side and have churned over how to fairly rate a machine's performance; an obvious problem right now in the rush to adopt wind.  There are some standards coming from the SWCC that should hopefully describe some sort of a power curve test, but I'm skeptical it will contain what will really matter for "real world" performance.

We're approaching turbine performance in two parts:

1.  Controlled-condition power extraction:  This is the steady-state power draw from a clear windstream at a given speed.  Right now we're getting this from vehicle testing, and record the important variables (temperature, altitude, wind speed, power, batt. voltage, etc).  This results in what most publish as their "power curve", and is a good measure of efficiency versus swept area.

   
   

2.  Machine reaction to tower conditions:
   
   
   

  a)  Time to full power for change in wind speed:  This is simply measuring how fast the turbine reacts to a given ramp change in wind speed.  The time and power ramp can be used to calculate the "lost power" while the machine was reacting.

  b)  Time to full power for change in wind direction:  This is tougher to measure with vehicle testing, and tower testing this combines too many variables to isolate performance.  We've got a sort of vehicle test planned that examines how quickly the machine can turn and get up to speed given a mechanical offset under speed...at least this tells us something but not the entire dynamic picture.

As we can all guess, you can build a machine to conquer #1 but making a fast-reacting machine under #2 conditions can compromise performance under #1.  We've tested store-bought blades with "high performance" profiles that do well in ideal conditions that get close to #1, but cough and sputter under conditions that most of us face in the backyard.  We've also found the approach of building heavy machines that have to keep spinning in one direction, waiting for the wind to come back, works well for conditions closer to #1 but a different type of machine is needed to get good tower conditions under varying degrees of turbulence.  A car analogy works well here, where an 18-wheeler can get up to 70 mph just like a small passenger car, but the small car wins the race while the truck is slogging up to speed.

So we're going in a direction that seems to be very different from the rest of the small wind industry, who are happy to compete on wind tunnel performance.  We're looking to a day when site conditions can be measured to capture #1 and #2, and then a turbine can be matched to give the most power against those variables.  Then people will get performance closer to what they expect, and we're not wasting more resources on copper and magnets in the wrong application.

Comments welcome!

[Flux]

With wind tunnels or vehicle testing the wind is constant over the whole swept area.

I think much of what you are thinking about is far more influenced by the lack of constant wind across the swept area than the factors you are considering. I don't think response time to accelerate is a big issue, there seems little real difference between high and low inertia machines in performance.

The response to changing horizontal direction is mainly admitting that the wind is turbulent.How much this affects the wind velocity across the whole swept disc is again most likely the major factor.

It the wind is turbulent and the speed is not constant across the swept area then I don't think you will do much to improve things. Machines don't work well in turbulent areas. The only cure seems to be to mount them high enough to be above the turbulence.

From my testing I have found that when properly matched to the blades the thing can respond very rapidly to gusts and a doubling of power in a second is quite common. If the thing isn't matched then stall will prevent such quick changes. Blades running lightly loaded will not track the peak power and although the blade acceleration will be fast the load increase will be only a fraction of what it should be.

I find very large scatter in the power produced at different times in the same nominal speed and I am sure it is a function of how much of that measured speed is active across the blade area at the time of measurement.

It would be interesting to compare perhaps a dozen anemometers close together and look at the wind speed over an area the size of the prop on an instantaneous basis.

I look forward to your findings, any attempt at real measurement is a step in the right direction. Most commercial ratings are obviously suspect but the better manufacturers are probably fairly honest. Some of the others are in dreamland and use their dishonesty as a selling point.

Flux

[Beaufort]

Hi Flux, I agree with your point about variation across the swept area.  I'm guessing this is a bigger factor as the swept area increases?  We're looking at small turbines under 7 ft. dia. for now, and trying to tackle "macro" turbulence.  I'd love to put a grid of annemometers out there to quantify what this means, and to determine the size of "good wind packets" that are useful.  

I find all this very facinating that many of the challenges for small wind mirror those for big MW-class machines.  One of the bigger innovations in that area is deploying laser sensors to look out ahead of the turbine to react quicker to changes in wind stream.  And this is for wind farms in "ideal" locations.

Yes, total agreement that a well-matched system will give the biggest results in most conditions.  But a well-matched system could have blades that weigh 200 lbs and it will never reach cut-in unless it gets 60 seconds of steady wind, and then once at speed it can hardly turn due to the gyroscopic forces.  This extreme example may work well on a very tall tower in a good wind location, but few of us can do this where we live.  And comparing moderate differences in inertia in a fair or high wind location will probably not show a significant difference (as you state).

So there is a spectrum of turbulence that occurs for home wind power, and right now there is a cutoff (that nobody can define) below which current commercial turbines have an excessive payback.  All locations have some amount of wind variation...we're just looking at ways to better suit that variation because the current machines aren't designed for it (for good reason...500 ft. above wind-swept plains yield the most amount of power but I don't live there).  

[Flux]

I think we pretty much agree on this. I suspect that at 6ft diameter the wind distribution across the disc is a factor if there is any significant turbulence. On my tests with a 6ft machine I got very consistent results with winds that had no obstructions upwind for 1/4 mile. Other directions were not consistent and power curves were not repeatable on different days.

I am sure that many small improvements are possible but the only thing I ever found to make a major difference was mppt. The 6 ft machine I mentioned produces 3 times the power of other similar sized machines I have tried at near 30 mph.

The power is also up in the band between cut in and 15mph where we always assume conventionally loaded machines are running fairly efficiently. The electrical efficiency is not a factor in this region as it is high with conventional loading on light loads and my converter is not having a great effect in this region, so it must be stall that we always assume is not a big factor in this region.

In high winds then it is partly electrical and partly prop matching, I can hold an overall electrical efficiency of 70% in this region where it has to be below 40% to avoid drastic stall without the converter. I can also keep furling off to higher wind speeds as stator heat is not the limiting factor.

The problem as always is that you can get a cheaper solution by just increasing blade size on a conventional machine. If it is first cost that sells a machine then a simpler bigger one will win the day. I don't know how you square performance against cost in the commercial world.

Flux

[Beaufort]

Ahh, yes.  And then there's MPPT.  Gains in all conditions...still waiting for Midnight Solar's box to come out. 70% electrical efficiency over 23 mph is doing very well indeed...I'm impressed.  So when will the "Flux Classic" be available?

Performance versus cost is usually worked out to $/kW at some chosen MPH.  And this drives me nuts, because everyone is using different windspeeds and drawing comparisons between machines when it suits their needs.  MPPT can be factored into this, but at some point there has to be a swept area dimension.  I'm guessing the cost/output gains of MPPT will win above a certain windspeed, for a given swept area.  Below that windspeed, the gains won't outweigh the cost.  

So without MPPT, we're looking at picking up gains in the bell curve of conditions where others falter.  One good test was a Pepsi Challenge with side-by-side machines in less-than-ideal conditions and it was clear there is something very wrong with selling certain types of HAWT's into any kind of turbulent conditions.  This is where VAWT's win the perception game, and they are.  Mariah has sold 4000 of their Windspires to date and are gearing up for 1000/month(!!??!??).  And tax rebates are going for all that??  These things aren't going out in the middle of a wind-swept plain.  That nonsense is a great example of how power curves (back to the original topic) and lack of real-world data can lead to misinformed consumers.  Our mission is to get our hands around real world conditions for residential windpower and to bring about some sense.  And hopefully a Pepsi Challenge is in the future with these VAWT's (and that crazy thing from Honeywell...talk about high inertia!).

[SparWeb]

I will look up Hugh's article if I can find it.

When I finished the data logger last year and started using it, I was at a bit of a loss as to what to do with the data collected.  The research I did led me down a very different path, and only since reading Gipe's book have a realized that I've been going about it all wrong.  I was focused on "filtering" the scattered data, which took my eye off collecting good data in the first place.  This is the first time that I have made simple sense of all that data.

I was also trying to design a "bigger better" data logger recently, so I'm lucky to have found this before trying to finish it.  Now I can build in some of the things you mentioned, like sorting the data on-board into bins.

One could experiment with an array of anemometers, but you'd have to be careful to not cause mutual interference!  (You can't win).

On the subject of inertia, it sounds to me like a rotor that "misses" some power because it responds too slowly doesn't suffer from too much inertia, it's installed at a site that's too turbulent.  Any gust with enough potential to double the kinetic energy (cube root of 2 equals 1.26) is more likely due to large objects in the airstream shedding turbulence.

Sure there can be major differences in the inertia of different rotors.  Consider carbon fiber blades, wooden blades or solid aluminum blades with the same chord.  Pilots of fixed-pitch propeller aircraft are aware of this inertia when they fly similar airplanes with different material props, too.  

The other factor is the authority the blades have to produce the aerodynamic forces that accelerate them.  Since the effort is to make WT blades as light as possible, the selection of aluminum WT blades is rare.  The weight distribution in the blade then becomes a moot point if the heavy type isn't available.  The difference between wood and CF seem to be a wash, since CF is denser but less is used because it is stronger.

Efforts have been made to model the turbulence under certain conditions.  I've seen such correction factors used in technical reports that attempt to validate the wind potential at certain sites.  Here is a report from the US that wrangled with the issue of under-performance:

http://www.masstech.org/renewableenergy/sm_renew/Small%20Wind%20Progress%20Report%20061208.pdf

Do I need to bring up wind shear, or is that the 1000-pound gorilla in the discussion?

[Beaufort]

That's a great report, and very sad at the same time.  That sort of thing is going on all over the place.  The results obtained in the report remind me of what it would be like to buy a car with 26 MPG on the sticker and getting only 6 MPG.  And furthermore, nobody can explain why.  So the manufacturers point fingers at the installers and the installers point fingers at the manufacturers...and so on.  There is quite a range of turbines examined in the report, which doesn't make analysis easy either.  It's like we're re-living the 70's again in a mad rush to get off oil, and we didn't learn anything.  Ugh!

I like how they had to remove the one turbine that was installed by the homeowner from some of the charts because it performed too well.

[Perry1]

Interesting post. I enjoy the analytical aspect of this even  though I have seen enough power curves to make me throw up. In large scale wind our contracts are usually based on 'availability' and 'power output'. Availability is a measure of how reliable the machine is. In other words, if the wind is blowing, is the machine ready. We shoot for 97-98%. The rest of the time is lost to breakdowns and maintenance. Power output comes from, you guessed it, power curves. We have 10,000 turbines in the field, each generating a PC once a month against which our contract people have to prepare billing reports against. So, as you can imagine, lots of power curves.

I had to use them mostly for troubleshooting purposes. One customer in Maine had weird signals at the crown of the PC that was costing them production, so we had to pay them for the lost energy. After a lot of research and testing we were able to determine a signal that indicated it was icing. It was a big deal since we were losing 1/4 million a month on it.

Anemometer placement is a big deal as you can imagine. Ours are on the top of our nacelle, aft of and in the wake of the blades. We have a transfer function which we believe tells us the windspeed presented to the blades but there are obvious flaws in  that method.

Velocity gradients across the blades is a big deal too, especially with a 75 meter rotor. Mostly occurs from top to bottom.

On my modest home built turbines I use a Datek system for daq. Record turbine voltage, current, battery voltage and current and wind speed. I log 1 second averages for all data. started with a 10 minute avg like we use at work but with turbulent sites this is utterly useless. Currently I am working on a automatic power curve generator for my GUI but it is a little slow going since I am teaching myself VB6 as I go. Also, as I often complain about my site gets very little if any wind. It kills me because it is so hard to get good data. I pretty much just build these because I think they are cool.

Perry

[SparWeb]

Even the experts are wrestling with this.  The Cadmus group that wrote that report seem to be trying to validate their energy modelling software (which they call "SWEET"), yet their estimates were even more rosy than the installers are.  When your fudge factor grows to 60 or 77%, you know your model isn't working properly.

I haven't looked at the report for a year (since I first found it) and now I discover the last page.  Now that's a story!  The report is talking about 19 systems, but the last page lists 33 (thirty-three!) wind turbines installed under the Massachusetts funding program, but only 19 had data to report!  There is a reliability issue, and working with WT installers that haven't put up more than one or two probably doesn't help.  I also want to find out why all of the WindTechCo machines (about 5 of them) were removed from service during the time of the study.

[Beaufort]

I was hoping someone in the big wind market would join into this...is anybody setting themselves apart from other turbines by being able to react quicker to changes in conditions?  The laser look-ahead systems seem like the next level of control over the anemometers like on yours..are you seeing that the additional cost produces results?  I spoke to several of the companies involved with these systems at Windpower '09 in Chicago and it sounded very cutting edge, with many still tweaking their control schemes as they go along.  It was hard to get anyone to say what the actual improvement is with these in place.  If there is a significant improvement in look-ahead control, that indicates something positive about being able to react much faster to wind changes.

Regarding the velocity gradients across the MW-class blades, it leaves me wondering if a bunch of smaller blade systems would be able to produce more power.  I'm sure the cost/MW wouldn't even compare, but when one big one goes down..it's a huge amount of power lost at one time.  Taken over the life of a site and the "averaging" effect of velocity gradients on huge blades (that are getting even bigger), one has to wonder how it would compare.  

[Perry1]

Hey Beaufort,

Yes, we've looked at other anemometry systems. There's the laser guys which is relatively new and more often you see ultrasonic systems, mounted on either the nacelle or spinner. The main selling part for these type of systems is that it solves some of the ice buildup issues and they remove some of the mechanical failure parts bearings/etc). To date they have proven too expensive to retrofit on to turbines and have long ago been 'cost reduced' out of the existing designs. I work in aftermarket turbine upgrades and we passed on these as there was no quantifiable value proposition. I think you found this out. Were any of the people selling these able to tell you what their value prop was?

As for your other question, I am not aware of any companies that differentiate themselves by being able to yaw into the wind quicker than others. Big turbines are different than small turbines in that respect. They really don't spend that much time yawing. Site selection puts them in places with predominant wind directions and the wind dynamics at 80 meters is quite different than we see on our 10 meter poles. Useless trivia warning, it takes about 20 kW to yaw the turbine. I am aware of some yaw optimization projects underway at various manufactures. Funny idea but there is some data indicating that pointing right into the wind may not be the optimum position for harnessing power. I have not seen that data though.

As far as many smaller turbines being more effective. Unfortunately for that argument, and for us small wind enthusiasts as a whole, turbines are only cost effective when they are larger. Up to about 2 MW wind power can only compete with the help of government subsidies. 2-4 MW has a better chance. Maintenance and repair costs are usually not factored in and those costs are very high. Cost $1000 every time you have to climb a turbine, then there are replacement part costs on top of that. Imagine if you had 1000 turbines to maintain vs 100.

Perry

[Beaufort]

Hi Perry,  thanks for the info.  Yeah, the laser systems at the show didn't appear to have what I would call a value proposition.  Yet there is an ad from one of them in a trade mag with a guy looking in a rear-view mirror and the caption, "you don't drive your car by looking behind you, why would you control your turbine the same way?".  I didn't think that quick yaw adjustment is where the value is, but rather blade pitch adjustment to suit changes ahead in the windstream.  And this makes sense given the mechanical stuff involved to twist something so large at speed.  

I'm always looking for new ideas that would translate to the small stuff.  Given the amount of research going into big wind, and the scale of payback it would make sense that there will be some good innovation.  My overall gut feel is that there are only small gains by tweaking blade profiles and steady-state efficiency, but the real attention will shift toward overall uptime (as you elude to), dealing with variable conditions, and being able to predict output accurately given the grid challenges.  I see similarities between big and small wind in this respect.

[anteror]

Read my experiences about this;

http://www.thebackshed.com/windmill/forum1/forum_posts.asp?TID=2034&PN=1

Antero

Finland