5. Smaller wind turbines.
The Evance R9000, Bergey Excel-S, Proven 6, Skystream and Honeywell RT6500.
Compared to their large turbine counterparts, small turbine manufacturers often do not have the same facilities and expertise to carry
out turbine tests to obtain the power curves. In consequence, the accuracy of their power curves and the subsequent claims for mean power and energy production
need to be treated with some caution and one of the aims of the WindPower program
is to enable a potential user to carry out checks on the consistency of small turbine performance claims. It should be noted that the
introduction of accreditation schemes like the US Small Wind Certification Scheme
and the UK's Microgeneration Certificate Scheme will do much to reduce inconsistent and inaccurate claims but
until such schemes are widely in use, manufacturer's data often needs to be treated with some scepticism.Even small wind turbines are too large to be tested in wind tunnels and so the steady power curve for a wind turbine is deduced from tests in a natural wind in which the power output is fluctuating due to turbulent variations in the wind speed. The problem here is that the power output always lags behind changes in wind speed. The determination of the steady state response of a system from tests with randomly fluctuating inputs is a classical problem in system identification and it is not in general a simple problem. To overcome this difficulty, the International Electrotechnical Commission recommends a testing procedure (IEC 61400-12) that, whilst not strictly correct, will give power curves generally of sufficient accuracy for practical purposes. The procedure requires that the wind speed and power output in a normal wind are effectively logged instantaneously at intervals of one second or less. The power output data is then 'binned' at speed intervals of 0.5 metre/second and then an average value of power output obtained from the 'binned' values. 'Binning' means, for example, that all the power readings that occurred between wind speeds of, say, 8.0 and 8.5 metres/second are collected and their average taken as the power output that would be obtained in a steady wind of 8.25 metres/second. There is an assumption in this that the time lag for a step downward change in wind speed is the same as the time lag for an upward step change in wind speed and that the occurence of increasing and decreasing speed events within the 'binned' data are the same. Generally, the IEC procedure will give power curves of acceptable accuracy for practical purpose although some errors will arise from large gust events. Errors will also occur for turbines that use a tail fin to furl the turbine away from the wind at high wind speeds. This furling action behaves quite differently between increases in wind speed and decreases in wind speed. However, this too is not likely to be significant unless the turbine is operating at a very windy site where the furling action is in play for a good proportion of the time.
0n this webpage, we will look at a number of popular small wind turbines to show the uncertainties that can arise in establishing the power curves and the subsequent errors that can arise in the mean powers and energy outputs that are calculated from them. The conclusion from these examples is that a potential user of a turbine should check the following three points:-
- Try to establish from the manufacturer how their power curves were obtained.
- Check that the turbine efficiencies calculated from their power curves do not give rise to absurd peak efficiencies. Even a well-designed small turbine is unlikely to achieve a peak efficiency of greater than 30-35%.
- Compare the mean power and energy outputs calculated from their power curves with their claimed values. In many instances, claims for mean power or energy output are over-optimistic.
Of course, there are many other factors that influence the choice of a turbine such as the cost, reliability, ease of installation, maintenance, noise levels and so on. However, if there are inconsistencies in the basic power data, it suggests that a potential user might need to examine other claims with care too. On the whole, the situation should get better with time because many countries now insist on proper accreditation trials which will include following the IEC's testing procedures.
The Evance R9000.
This turbine has received the UK's Microgeneration Certificate MCS WT0039. As a result, it will have been tested to IEC 61400-12 standards and so one can expect the data to be of a high standard and consistent.
The Evance R9000 is a three-bladed 5.5 metre diameter wind turbine with a tail fin. It uses a patented pitch control mechanism which it is claimed results in higher overall efficiencies than other small turbines and also limits the power output at high wind speeds to a constant level corresponding to the rated output of the generator of just above 5 kilowatts. It also has a electrical generator braking system.
The figure below shows the data from which the power curve (the green line) was obtained as an average of the binned power and wind speed readings. The peak power is obviously electronically regulated so that there is a sharp cut-off at 5.2 kilowatts. Because this turbine does not use a mechanical tail fin furling system to limit the power at high speeds, there is a lot less scatter in the data than, say, the Bergey turbine and so even in very wind situations, the mean power estimates calculated from the power curve should be reasonably accurate
The figure below shows the power curve and the efficiency graph obtained from the WindPower program. The peak efficiency is about 35% which is plausible for a well-designed small turbine and occurs at about 7 metres/second. As can be seen from the previous web page, the peak efficiencies of large turbines are over 40% but this is probably not attainable from small turbines because the blade Reynolds numbers are much lower.
.jpg)
The figure below shows the mean power obtained from the WindPower program for the Evance turbine with a Rayleigh distribution for the wind probablility density - see webpage 13. This is the distribution most commonly assumed for the probability function. Also shown is the mean power figures given in the manufacture's leaflet. The agreement is very close and they differ from the WindPower values by only fractions of a percent. It is almost certain that the manufacturer's mean power data is taken from their MCS certificate and so it is hardly surprising that the results agree with the WindPower program because a near identical integration scheme will have been used in the MCS certification process - see web page 14.
The table below shows the equivalent annual energy production in kilowatt-hours obtained by multiplying the mean power results by 8,760 - the number of hours in a year.
| Annual energy production in kilowatt-hours (Evance R-9000) | ||||||
| Mean wind speed (m/s) = | 5 | 6 | 7 | 8 | 9 | 10 |
| WindPower calculation | 8,669 | 13,101 | 17,378 | 21,222 | 24,544 | 27,341 |
It is not the purpose of this website to advocate one small turbine rather than another but simply to show the checks that should be carried out on manufacturer's performance claims. In any case, there are many other issues such as reliability, noise and costs that have to be taken into account in deciding on a particular turbine but, generally, one might have more confidence in a design if the data on it is consistent and does not lead to exagerrated claims for overall energy output.
Bergey Excel-S with Grid-Intertie.
The Bergey Excel is a popular US turbine design that has been around for some years and, in that period, it has undergone significant performance improvements. It has a blade diameter of 7 metres and comes with electrical regulators for connecting it either to a mains electricity system or for charging batteries. The Bergey unit uses a tail fin furling arrangement to turn the turbine away from the wind at high wind speeds. There is no cut-out speed.
The earlier Bergey Excel (circa 2000) has been the subject of a number of independent tests by the American National Renewable Energy Laboratory (NREL) in Colorado. The reports are all available from the NREL website. The tests have involved turbines with different aerofoil sections and different rotor diameters. Only one of the reports (NREL/EL-500-33546) seems to have been on a standard Excel S-60 turbine of the time - primarily to check the durability of the turbine over a period that altogether took about two years from 1999 through to 2001. Generally, the mechanical reliability of the turbine seemed of a high standard.
The figure below shows a scatter plot of NREL results obtained over 10 second averaging periods of output power against wind speed. Although this is not the same as the IEC testing procedure, it nicely illustrates the problems of extracting a 'steady' power curve from this type of data as discussed in the opening paragraphs of this web page. The tail fin furling mechanism comes into play at around 19 metres/second and it is clear that the inertia of the whole turbine system is creating significant time delays between the power output and the wind speed when this mechanism starts to operate. This results in a lot of scatter in the data. At lower speeds, the scatter is greatly reduced. The red line in the figure is the 'steady' power curve given in the manufacturer's literature which it is stated was obtained from 1 minute averages. Up to the speed at which the tail fin furling action comes into play, the agreement between the two sets of data is reasonably good.
Over the last few years, the design of the Excel turbine has been the subject of major revisions both in terms of the blade design and the electrical control system and a power curve for this newer design is available on an Excel spreadsheet from
The power curve was apparently obtained from tests to IEC 61400-12 standards and so should be to a good standard of accuracy.
The figure below on the left is taken from the WindPower program and shows the newer Bergey power curve and the resulting efficiencies (aka power coefficients) of the turbine. The power curve in the Bergey literature extends only to 20 metres/sec so that the power curve has just been extrapolated downwards from 20 to 30 metres/second as a rough reflection of the decrease in power due to the furling action of the tail fin. The efficiencies (i.e. power coefficients) of the earlier Bergey Excel were quite low with peak values of around 13% but, with the newer blade design, the efficiencies are now close to 30% over quite a wide range of speeds. This is a measure of the improvements in performance that can be achieved through careful blade design.
The curve on the right shows the mean power that can be expected from a Bergey Excel turbine for a range of mean wind speeds.
For those interested in annual energy production, this is easily obtained by multiplying the mean power by the number of hours in a year (i.e. 8,760). The table below shows the annual energy production based on the mean power curve.
| Annual energy production in kilowatt-hours (Bergey Excel S-60) | ||||||
| Mean wind speed (m/s) = | 5 | 6 | 7 | 8 | 9 | 10 |
| WindPower calculation | 13,171 | 21,098 | 29,756 | 38,257 | 45,996 | 52,644 |
From the Bergey website, a figure of 13,277 kWh is quoted as an annual energy output for a mean speed of 5 metres/second and this is consistent with the present calculation. This is roughly twice the annual energy production of the earlier model.
Finally, it should be noted that Bergey have applied for both US and UK accreditation certificates and no doubt the power curves from these tests will become available in due course.
Proven 6.
The Proven 6 is a small turbine manufactured in Scotland. It has a 5.5 metre diameter rotor and a ingenious hinged blade mechanism that allows the blades to deflect in high winds and, in this way, both damage avoidance and power regulation at high speeds are achieved. There is no cut-out speed.
Once again, the Proven 6 has been the subject of independent tests. The first figure shows another scatter plot of power against wind speed obtained from a Proven 6 on top of a residential block in London - the Ashenden report (see references). The results were obtained apparently in accordance with the IEC 61400-12 testing procedure.
The red line shows the mean power obtained from averaging the binned results at wind speed intervals of 0.5 metres/second. The Ashenden site had a mean wind speed of only 3.8 metres/second so that there are not many data points at the higher speeds.
In addition to the Ashenden results, the National Engineering Laboratory (NEL) in East Kilbride also carried out some tests on the Proven 6 and their results are shown on the left below along the Ashenden results and the manufacturer's curve obtained from a Proven brochure. The figure on the right shows the turbine efficiencies calculated from the power curves. It will be noted straightaway that the efficiency figures from the manufacturer's data at the lower speeds are too high to be realistic. For good aerodynamic reasons, small turbines cannot achieve the same efficiency values as their larger counterparts and a good design would be doing well to achieve peak efficiencies in the 30-35% range. The NEL and Ashenden results are in reasonable agreement and give peak efficiencies close to 30%. This is far more plausible. The Proven blades seem to have taper and twist which suggests that they may have been designed with some care.
Using the steady power curves above, the mean power output has been calculated using the WindPower program with the default wind speed standard deviation of 62% of the mean wind speed. The graph below shows the results of these calculations for the Proven power curve data and the NEL test data. The more plausible results using the NEL power curve lead to mean power outputs that are around a third less than those obtained from the Proven power curve.
Most manufacturers seem to prefer presenting potential performance figures for their turbines in terms of the energy produced by a turbine over some period of time - usually a year. From the WindPower program results shown above, the energy output in a year for the Proven 6 is easily calculated by multiplying the turbine mean power by the number of hours in a year i.e. 8760. The table below shows the kilowatt-hours for a range of mean wind speeds. The figures based on the NEL test data are almost certainly a better guide to the energy output but, even here, it should be stressed that these figures assume there is no down time for the turbine and also that there are no other losses in the system.
| Annual energy production in kilowatt-hours (Proven 6) | ||||||
| Mean wind speed (m/s) = | 5 | 6 | 7 | 8 | 9 | 10 |
| WindPower calculation using Proven power curve | 12,352 | 17,160 | 21,637 | 25,492 | 28,733 | 31,273 |
| WindPower calculation using NEL power curve | 7,998 | 11,721 | 15,216 | 18,308 | 20,936 | 23,126 |
Proven have only a temporary UK Microgeneration certificate (as at October 2010) and it will be interesting to see the power curve obtained from the full approval tests.
The Skystream.
The Skytream is a small US three-bladed downwind turbine (no tail fin) manufactured by Southwest Windpower in Arizona. It has been developed over a period of several years in conjunction with the National Renewable Energy Laboratory in Colorado. It was called the Storm turbine originally but, with modifications, it was renamed the Skystream. It is a 3.7 metre diameter turbine with a peak power output of 2.4 kilowatts and has received the UK's Microgeneration Certificate. This certificate is available from the Skystream website and in the figure below is shown a comparison between the power curve from this certificate compared to the power curve from earlier manufacturer's leaflets. It was not stated in these leaflets how the earlier power curve was obtained but, by reference to the NREL report on the Skystream, it may have been obtained by taking either 20 second averages or possibly 1 minute averages. It demonstrates how the processing of analysing the fluctuating data influences the apparent power curves. In this case, the effect is not large but it does show how important it is that a consistent testing procedure should be followed if valid comparisons are to be made between different wind turbines.
Control of the rotational speed and the power output of the Skystream is by electronic control of the generator which can also be used to stop the turbine altogether either because the wind speed has risen above 25 metres/second or because there has been a failure of the grid to which the Skystream is connected. It also has a back-up safety braking system if the normal braking system should fail for any reason. Generally, the turbine seems to have been carefully developed as one might expect from an association with a large government research and testing laboratory.
The figure below shows the WindPower plots of the power curve and efficiency on the left and the resulting mean power on the right using a Rayleigh probability distribution for the wind speed. The peak efficiency of the Skystream is just under 30% which is a very reasonable value and it maintains this level of efficiency over quite a broad speed range from, say, about 5 to 11 metres/second.
The table below shows the mean power results from the WindPower program converted into annual energy production. Also shown is the energy production figures taken from the UK's Microgeneration Certificate. Up to 7 metres/second, the agreement with the WindPower estimates is within a percent but the differences increase until the WindPower results are just over 10% higher than the Microgeneration Certificate values at 10 metres/second. Without knowing in detail how the Microgeneration Certificate calculations for this particular certificate were carried out, the reason for this difference cannot be established but it should be noted that the accuracy of the present numerical integration scheme has been carefully checked and one should not expect differences of greater than one percent - see web page 14 on the calculation of mean power.
Also shown are the annual energy figures from an earlier Skystream leaflet. Once again, the figures are in good agreement at the lower speeds but diverge more at the higher mean speeds of 9 and 10 metres/second. So, even before the issue of the Microgeneration Certificate, the Skystream data was reasonably consistent.
| Annual energy production in kilowatt-hours (Skystream) | ||||||
| Mean wind speed (m/s) = | 5 | 6 | 7 | 8 | 9 | 10 |
| WindPower calculation | 3,393 | 5,338 | 7,291 | 9,064 | 10,556 | 11,729 |
| Microgeneration Certificate data | 3,416 | 5,349 | 7,207 | 8,737 | 9,814 | 10,439 |
| Skystream leaflet | 3,840 | 5,448 | 6,828 | 8,136 | 9,156 | 9,744 |
The Honeywell Windgate RT6500 wind turbine
This is a new wind turbine in the small wind turbine market. Honeywell is a large and reputable company and so one would expect this turbine to be well manufactured and backed up by reliable performance data. Unfortunately, the initial advertising material (including a launch video that can be seen on YouTube) does not seem to provide this. This is a shame because there are aspects of this machine which seem attractive but uncertainties about its power data casts doubt on its likely performance.The turbine is unusual in that it is a horizontal axis turbine of 1.7 metres diameter with apparently twenty blades. Counter to intuition, it should be noted that having a large number of blades does not confer any significant improvement in aerodynamic efficiency compared with a three-bladed or even a two-bladed turbine. The generator is housed in an annulus around the blades. It is claimed that this results in a low frictional resistance to rotation so that the cut-in speed is close to 1 metre/second. It is argued that this is a good thing in that the turbine will start to generate power at much lower wind speeds than a more conventional wind turbine. However, this is actually not very significant in itself because the power generated at these low wind speeds is miniscular - literally just a few watts. A more significant point is that the turbine seems able to operate over wide speed range up to to about 20 metres/second without any mechanical feathering of the blades. What happens above this speed is not altogether clear. The figure below shows an image of the turbine with what is apparently a steady speed power curve obtained from an advertising pamphlet Energy-Output-Power-Curve.pdf.
The noteworthy feature of the power curve is that it shows an ever increasing power output with speed and apparently the turbine will produce a power of 2 kilowatts or more at higher speeds approaching 20 metres/second. However, the figure below shows the turbine efficiency based on this power curve. Not only is the Betz limit exceeded but the efficiency rises above 100% at speeds below about 3.5 metres/second! Clearly, this casts doubt on the whole performance of the turbine.
.jpg)
Initially, it was a puzzle to know how such data could have been produced. However, in some advertising material, the manufacturer mentions that a wind tunnel was available in which they tested the turbine. As mentioned at the beginning of this webpage, it would need to be a very large tunnel in order to avoid problems over blockage effects. From a later video, it now seems that the turbine was tested in what amounts to a duct. The figure below shows a still image taken from this video.
.jpg)
Testing a turbine in a duct bears little relationship to testing it in free air and the power curve and efficiencies obtained in this way will differ greatly from those obtained in a free air test. As the figure below shows, the boundary conditions in the two case are completely different.
However, if we make the assumption that the flow through the turbine itself is similar in both cases then we could use the duct test results if only a relationship could be determined between the wind speed through the turbine Ut and the higher wind speed in the stream tube far upstream in a free air test Uu.
The analysis given on the Betz limit on web page 13 shows that the wind speed through the turbine is the mean of the upstream and downstream speeds (see the above figure) so that
Using the above power curve, the WindPower program has been used to calculate the annual energy outputs shown in the table below for the standared Rayleigh wind speed probability distribution. The price of the turbine is quoted at around $4,500 which seems quite attractive. However, as this turbine is aimed at installation on buildings in urban environments where wind speeds would probably be only around 4 metres/second, the payback period would be very long - in excess of ten years even with generous feed-in tariffs.
| Annual energy production in kilowatt-hours (Honeywell WT6000) | ||||||
| Mean wind speed (m/s) = | 5 | 6 | 7 | 8 | 9 | 10 |
| WindPower calculation | 777 | 1,224 | 1,761 | 2,377 | 3,046 | 3,724 |
This turbine has been included on this webpage as an example of how careful a potential buyer of a small turbine must be when faced with a manufacturer's claims about their wind turbine performance. It is a good illustration of how important it is that turbines should be subjected to a proper accreditation process.
J. T.G. Pierik, R.W. Dunlop, W.K. Lee, J. Gabriel (2001). Performance evaluation methods for autonomous, applications orientated wind turbine systems. Energy Research Centre of the Netherlands Report ECN-RX--01-062. Presented at European Wind Energy Conference and Exhibition, Copenhagen, Denmark, 2-6 July, 2001.
J van Dam,M Meadors (2003). Duration test report for the Bergey Excel-S60.US National Renewables Energy Laboratory report NREL/EL-500-33546.
Southwark Council (2008). Ashenden Wind Turbine Trial; Phase 1 results. Southwark Council, London South Bank University, Brian Dunlop Associates report.
P.Migliore, J. Green, D. Calley, J. Lonjaret (2005). Balancing performance, noise, cost and aesthetics in the "Storm" wind turbine. NREL/CP-500-38157 (Conference paper for WindPower 2005)
Next is a discussion on obtaining estimates of mean wind speed.
