Wind Turbines at the Boston Museum of Science: Production Data

[NOTE:  since I wrote the post below I have received email from David Rabkin, the project leader at the MOS.  David has commented on the post (and I will continue the conversation in the comments below) but he would like it made clear that the sensors have not been calibrated and the data is not to be considered entirely reliable.  And I want to make it clear that I have no official connection or financial interest in the MOS or any of the turbine manufacturers.  In the weeks prior to my visit I attempted to contact Mr. Rabkin by phone and email to get a tour, but as I understand he has a lot on his plate, and I didn’t hear back from him – and so what I write below are my personal observations and conjectures, with no inside information.  Now that we are in touch I hope to get the real scoop on what’s up with the turbines.]

This morning I had a holiday-related package to deliver to my aunt Lucy, who works at the Boston Museum of Science, and being a wind turbine enthusiast I took advantage of the opportunity to check out the exhibit and turbine installation that the museum has installed. They have a cool display that shows what each turbine is doing in real time, and an interface where you can access cumulative data for each machine as well.

First, here’s a view of the installation from the north bank of the Charles river:

From left to right, the turbines are 1. Mariah’s 1.2kW Windspire, 2. Southwest Windpower’s 2.4kW Skystream, 3. Swift’s 1kW 5 blade turbine from the UK, 4. five Aerovironment 1kW “Architectural Wind” turbines, and 5. Proven’s 6kW unit.

Here’s an image of the view to the west from the function room (row of big windows just under the museum tower) – there’s at least a two mile fetch upriver, past the Longfellow bridge and MIT:

I was most interested in the energy production data for the various turbines.  Here’s a snapshot of the top level display, taken at random while I was there. (As in the previous post, in the interest of full disclosure I should note that some years ago I worked for Southwest Windpower, where I did a substantial chunk of the mechanical design for the Skystream turbine.)

The display shows a current windspeed of 17 mph, out of the south.  I think the direction sensor must be miscalibrated, since the wind was out of the northwest.  The displayed windspeed ranged from around 10-20mph (updating roughly every second) while I was watching it.  Qualitatively it was a very windy day; in riding my bike back to Central Sq there were times when the headwind on Broadway just about brought me to a standstill.  If you look closely, you see that the Windspire, the Swift, and the Aerovironment turbines are listed as “spinning, but not enough wind”, and the Skystream and Proven turbines are listed as producing 575 and 685W respectively, at the instant I happened to snap the picture.  The story was the same for the entire time I was watching,  The Skystream and Proven machines were producing power, and the others were listed as spinning but not producing.

Then I clicked through for the cumulative data for each turbine, which I tabulated as follows – yellow numbers I took from the MOS display, green is my additions (you can click the image to get readable-size text):

The project apparently got started on October 9, or at least the display indicated that the cumulative production was taken starting from that date.  Since inception, the Proven has produced 538 kWh, the Skystream 344 kWh, the Aerovironment machines 67 kWh, the Windspire 57 kWh, and the Swift brings up the rear at 2.4kWh.  The MOS display also calculates capacity factor, which ranges from a high of 9.2% for the Skystream to 4.6% for the Proven, to 2.4% for the Windspire, 0.68% for the Aerovironment machines, and 0.08% for the Swift. Capacity factor is the amount of energy actually produced in a given site, as compared to what the turbine would produce if it operated continuously at its nameplate power rating – values of 20-40% are the norm for commercial installations, with many homeowner installations lower than that.  The MOS also calculates a “relative production” for each turbine, which I gather to be the actual production as a percentage of manufacturer’s claimed energy production for a site with the average windspeed at the Boston MOS.  This ranges from 61% for the Skystream down to 0.5% for the Swift.

The Swift shows zero production for the last month, and I have never seen it spin more than 1 rev per second, so I suspect it has either burned out (alternator shorted, which applies a continuous high torque to the rotor) or been shut down for some reason.  The Windspire (which appeared to be running when I last walked past the museum a few weeks ago) also appeared to be shut down or shorted out today (more on that later).  No such excuses can be made for the low productivity of the five Aerovironment machines, which had produced 2 orders of magnitude less energy on the day than the Skystream, despite having over 2X the aggregate nameplate rating.

Just for fun I looked up the swept area of each machine, so I could calculate energy production on a per area basis (also in the table above) and scrounged up very rough cost estimates for each machine (I cannot vouch for these numbers, they are just what I found in a few minutes poking around the web) and used these to estimate a best-case cost of energy assuming a 20 year life (which is way longer than any manufacturer is willing to warrant a turbine).  These COE numbers range from 35 cents per kWh for the Skystream to $50 per kWh for the Swift.

My aunt kindly showed me around and took me up to the top floor for a closer look at the machines.  Here’s a view of the Proven:


And Windspire:

I also noted from this vantage point that the anemometers for each of the large turbines were at about 10′ off the roof, well below hub height in each case.  This should be kept in mind if power curve data is ever reported from this installation.  I could not get close-up pictures of the Swift and Aerovironment turbines, because they were mounted just above my head, and we did not have access to the roof.   (On the plus side, I did not notice any noise from these machines being transmitted into the building – on the minus side, they were not generating any significant energy.)  I did get a clue as to what might be going on with the Mariah turbine, which continued to rotate slowly in stall.  Check out the following video at (apparently I have to pay wordpress to get it to embed.)

What I observed is that the tip of the tower was rotating eccentrically by close to the diameter of the tower tip – maybe 4″ as best I could estimate from that distance.  That much eccentricity would cause a hell of a vibration at full RPMs – enough to freak someone out and cause them to shut it off perhaps, and maybe enough to cause the rotor in the alternator to crash against the stator and short it out, leading to a permanent braked condition.

In conclusion, based on the data that the MOS has collected, the Skystream appears to come out on top in terms of capacity factor and energy capture per swept area, with the Proven machine a respectable second.  The Skystream has the best inferred cost of energy by a factor of 2.  The Swift has been out of commission for at least a month, and the Windspire seems to have suffered some kind of failure since I last inspected on December 6.  These apparent issues make it difficult to compare performance of these machines, but it is not a particularly good sign to have two turbines down for the count after less than 3 months.  It would seem advisable for the manufacturers of these machines to be closely monitoring an installation in such a prominent location and conducting prompt repairs, given the potential damage to their reputations (thousands of people tour the Boston MOS every day).    All 5 Aerovironment machines seem to be operating normally, which makes the extremely low production numbers and capacity factor troubling.  An inferred cost of energy of $9 per kWh is nothing to be proud of, and probably indicates that these machines will never pay back their embedded energy (that would be an interesting exercise for somebody to do).  Given the competent reputation that Aerovironment seems to have as an engineering company (they made working human powered aircraft etc.) and the fact that the turbines don’t seem grossly broken, it seems that the constrained-yaw, parapet-mounted turbine concept should be considered suspect until proven otherwise by published energy production data (which the manufacturer does not appear to offer.)  More generally, the “unconventional” designs do not seem to fare very well so far in the Boston MOS experiment, as compared to the 3 blade HAWT standard.  And it appears that the production data support my initial conjecture of December 9th – namely that putting wind turbines on buildings does not obviate the need for a tower of significant height.


10 Responses to “Wind Turbines at the Boston Museum of Science: Production Data”

  1. Mick Sagrillo Says:


    Great information. Would you consider gathering more information and doing a presentation on the MOS wind turbines at the Small Wind Conference in June in WI?


  2. David Rabkin Says:

    Dear readers interested in wind power:

    The best presentation of results from the Museum’s project will come from the Museum istelf, and will come when we feel we have sufficient and sufficiently well-analyzed data to report — probably in the Spring.

    Right now, our equipment and our data still have issues — from early calibration issues, to unplanned but needed downtime to replace parts from early prototypes with production-quality parts, to grounding problems, even an unexpected eddy around our oddly shaped building. The display in the Museum warns visitors that the process of calibration is not complete and thus the data have not yet been vetted.

    When you read Ben’s comments, hear anyone else’s thoughts, or view the turbines for yourself, be scientific about drawing conclusions. In thinking about the Mariah, for example, Ben notes “That much eccentricity would cause a hell of a vibration at full RPMs.” Actually, the wobble appears only at certain resonant frequencies (70 rpm and the one he most likely observed at 280 rpm). These resonances are inherent in the unit’s design and do not cause vibration problems, at least not if the unit is properly installed.

    Ben’s observations of the Swift may be correct. But not necessarily. It’s located in a weird eddy, and experiences clean wind only from a few very specific wind directions. It is tremendously handicapped by it’s location. It either needs to be moved or any power curve analysis performed during the brief periods of time when it experiences the cleaner wind that one might expect at a better location.

    Rooftoop wind is extremely complicated. We have encountered many surprises during the project, learning all the way. Sharing this learning is our ultimate goal. And if you want to learn best from our experiences, then please be patient and wait until we can present a complete picture of what is going on. In the Spring, our web site will include detailed information on the project.

    In the meantime, I encourage you all to observe and learn what you can from the data we present. But like any scientist, you need to recognize that your conclusions are always tentative, always limited by the quality of your data and also by your inability to identify and rule out alternative hypotheses.

    The Museum will share everything we’ve got, but not until we feel our data are sound and that we’ve collected enough to tell our story with some reasonable degree of confidence.

    – David Rabkin, Museum of Science

  3. fiveislandsorchard Says:

    Regarding Mick’s comment, I would love to go to the conference, but I wouldn’t presume to report the MOS data for them, at least not without their permission. David’s note above makes it clear that he intends to communicate the results of their testing once they have their data logging and other issues sorted out, and I look forward to that report – the Small Wind conference would be a fine place to present it.

    Regarding the performance of the turbines, I understand David’s caution about jumping to conclusions, and naturally all of the sensors should be calibrated properly. In my own experience, though, calibration of commercially-available sensors is about getting from +/-5% to +/-1% – and the performance of the turbines at the MOS is varying by two orders of magnitude. I would have more concern about sensors and calibrations except for the fact that the data the system was collecting lines up qualitatively with the observed turbine performance. Specifically, the Skystream and Proven machines were continuously running at speed, and the system reported reasonable output values for them; the Windspire and the Swift never came out of stall in the time I was there despite extremely windy conditions, and the system reported no energy output from those turbines; and the AV turbines were constantly going in and out of stall, and the system reported minimal (but nonzero) output. Of course the calibrations should be checked, but it seems to me that David may be selling himself and his test hardware a bit short.

    Regarding the question of whether the Windspire was functioning properly, the turbine is running in relatively clean air (cleaner air than the Proven, for the wind direction) but was rotating much too slowly for the given windspeed; blades spinning that slowly typically aren’t even producing enough voltage in the alternator to wake up the electronics. I used the video that I posted on YouTube to determine the speed, and it was approximately 23 RPM. (Sometimes videos of wind turbines spinning fast are aliased such that the blades appear to be turning much slower than they really are; in this case what you see is the actual speed.) This is well below the resonant frequencies that David reports are apparently a normal part of the operation of the Windspire. I don’t have the design details of the turbine that I would need to calculate what the resonances would be, but I think it was way too slow for dynamics that severe to show up, and what I saw didn’t look like resonance to me; it wasn’t smooth and the amplitude didn’t vary with rotational speed – it just looked bent. And resonance or not, the turbine should not have been deep in stall in winds gusting above 20 mph. Per David’s comments about the scientific method, obviously I am aware of the possibility that I am wrong and something else is going on, but based on my engineering intuition and years of experience with wind turbines, I will lay dollars-to-doughnuts odds behind my original conjecture – that turbine is either broken or shut down.

    Regarding the Swift, it is indeed located in a non-ideal spot, although several of the product photos on the manufacturer’s website show the turbines installed at or below the height of nearby obstacles. I would expect the energy production to be low in such a site – but zero energy in a whole month, after producing 2.4kWh at some point in the Oct/Nov time period? And never coming out of stall in the solid fraction of a windy hour that I’ve spent watching it so far? Again, I’m quite willing to be convinced by data but my engineering intuition tells me that something is the matter with it.

    In any case, maybe I’m wrong, but I hope not – those turbines should have been producing energy yesterday, and they weren’t. So the optimistic view is that there are some glitches in these particular units that are preventing normal operation, and that once repaired and appropriately sited, the MOS will be able to record some reasonable output data.

    In the case of the bank of AV turbines, I’m very interested to know if there’s a known calibration issue, since the output from those machines has been consistently nonzero but very low.

    All in all, I think the MOS is doing a real service by installing these turbines and displaying their output for the public. This sort of independent testing is crucial to help the industry mature and gain the confidence of the public. I do think a couple of small things could add significantly to the value of the information that’s displayed to the public. Specifically, a category called “malfunction – down for repair” could be added to the realtime display, such that if the machines are not running in a strong wind, the public won’t conclude that that particular turbine is incapable of producing energy. The corollary to that is that specific information about how many hours of run time are represented in the cumulative production of each machine should be displayed. And the single most useful thing that could be added to the public display at the MOS is average wind data for one or more of the several anemometers on the rooftop test site; this would allow an order-of-magnitude cross-check against manufacturer data.



  4. Alex DePillis Says:

    Ben and David (and Mick),

    It would be good to hear an update, if there hasn’t already been one.

    And yes, this might make a good topic for the June wind conference. I will be in Massachusetts in two weeks, and could work with the Mr. Rabkin to create a presentation, and save him the trip, since I am based here in Wisconsin, and plan on attending the conference.

    There is starting to be a pretty good body of data and reporting of small, building-mounted turbines. See, for example, the Warwick Wind Trials report.

    Best regards,


  5. Boston Museum of Science holds conference on small wind turbine project « Five Islands Orchard Says:

    […] of data from their small wind turbine experiment, and I posted some of the publicly available data here in this blog. Some of the turbines appeared to be working better than others, and I reported my […]

  6. Brother Donald Paul Says:

    We paid $14,000.00 for two Windspires in May of 2009. Sixteen months later, the largest amount of electricity we generated was .80 cents a month. You read that correctly, EIGHTY CENTS a month. As early as three months after the install, the company acknowledged that the turbines and inverters were “defective” (their words, not mine) and offered replacements “when available”. The turbines were doing exactly what this article mentioned. In ideal wind conditions, they appeared stalled and continue to do so. A year and dozens of empty promises later, our Windspires are still not replaced. The corporation talks a good game but is unable and/or unwilling to deliver. This has turned into a sad and disappointing, not to mention very expensive experience. We would advise anyone considering small scale windpower to look elsewhere. Caveat emptor!

    • Tom Gallery Says:

      Windspire is quickly becoming a failed company. They got a $400,000 grant from the local municipality for opening the plant in Manistee, MI. The promise was 120 full time jobs by 2011. They now have 15 employees and were closed for 10 weeks last summer. They’re trying to blame the problems on China and magnet availability!
      It’s a flawed product from every standpoint. Yet the internet is full of glowing reports based on no facts.
      NREL issued the final report on Windspire recently. Take a look.

  7. Boston trials data | Hugh Piggott's blog Says:

    […] Wind Turbines at the Boston Museum of Science: Production Data […]

  8. Architectural Wind Turbines | Wind And Solar Says:

    […] cost of energy by a factor of 2. The Swift has been out of commission for at least a month, and the Windspire seems to have suffered some kind of failure since I last inspected on December 6. Regarding […]

  9. San Diego Says:

    It is for holding your collection of different players on the island as low
    cost flights started in November 2005 from London’s Luton Airport, and in nerve-racking situations. The initial meal centered on a holiday plus experience oneself!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: