Detecting Variable Speed Performance Improvements

By Paul Gipe

An edited version of this article appeared in Vol 8 No 4, (Autumn 1995)

“If there are any performance benefits to variable speed, they are difficult to find” says WindStats’ contributing editor Paul Gipe.

Proponents of variable speed have argued that wind turbines using this technology should capture 8-15% more energy than constant speed machines. In a side-by-side test of turbines using both technologies, the variable speed turbine should yield about 100 kWh/m2/yr more than the constant speed machine in a wind regime capable of producing an annual specific yield of 1,000 kWh/m2. Such a noticeable difference in performance should begin appearing in industry statistics now that variable speed technology has several years of operating history in Europe and North America. However, publicly available production records on both continents reveal that the performance benefits of variable speed, if any, are ambiguous at best.

 Performance or annual specific yield depends not only on the wind turbine, but also on the wind regime. With the data currently available it is nearly impossible to screen performance records for wind regime and no attempt has been made to do so. The results are presented as though all turbine models were equally distributed across all Northern European wind regimes. Of course they are not. Accordingly the results can only be used to gauge general trends in specific yields and should not be used to compare one wind turbine model against another to any degree of precision.

Interestingly, as more performance data becomes available, variable speed advocates are shifting their emphasis. They now down play the performance advantage of variable speed, instead stressing variable speed’s presumed ability to reduce loads on the drive train. Ironically, it was variable speed’s performance advantage over constant speed machines that put the sizzle in Kenetech’s aggressive marketing campaign in the U.S. They trumpeted it as a “breakthrough” in the popular press, promising higher yields in moderate speed wind regimes than constant speed machines.

Engineers widely agree that variable speed can reduce shock loads to the drive train in gusty winds. This benefit translates into increased longevity or the ability to use lighter drive train components than otherwise possible. Both benefits produce obscure results. There is no way to measure the veracity of these claims with publicly available data, certainly not today. Twenty years from now we can count the number and cost of gear box replacements between variable speed and constant speed machines. In the meantime, though, we only have differences in specific yields to verify product claims.

 Related to wind regime as a complicating factor in performance comparisons is tower height. The new class of 500-600 kW machines are all installed on 40 and 50 meter towers. Thus their performance should be superior to older turbines installed on 30-35 meter towers. To minimize this effect in the following analysis, only turbines of similar sizes were compared with each other.

Another limitation is the small number of turbines in the larger size classes in northern Europe. Though their numbers have mushroomed, many were only recently installed and have few operating hours. A similar problem occurs when analyzing data from North America. The performance of Kenetech’s early projects of its KVS 33 is marred by widely reported teething problems. Thus, it will be another year or two before analysts can examine complete records for the performance of the KVS 33 in Minnesota, Alberta, and California.

Kenetech Performance Outside California

The performance results of Kenetech’s KVS 33 outside of California are inconclusive. There are no other turbines nearby with which to compare them. Clearly Kenetech suffered teething problems with its technology in Alberta during 1994. 1995 production atop Cowley Ridge in southwestern Alberta is on track with Kenetech’s projections and Kenetech fully expects to reach their target of 55 million kWh. If they do, Kenetech’s yields in Alberta will rival those of Vestas’ V27 at Zond’s Sky River project and SeaWest’s Mitsubishi turbines in the Tehachapi Pass, all constant speed machines. Yet WEG’s MS2 turbines in the relatively low winds of the Altamont Pass produced 1,214 kWh/m2 in 1991. Kenetech’s turbines in Minnesota turned in 900 kWh/m2 in 1994, a respectable but not earth shattering yield for a Midwestern site.

Kenetech Performance in California Over the entire year Kenetech’s KVS 33 in the Altamont Pass under performed not only Kenetech’s now obsolete model 56-100 technology but also that of Altamont’s perennial poor performer, Howden. Though Howden had one of its best years ever. Even after discounting the 1st quarter for installation of several new turbines and the repair of Kenetech prototypes, the KVS under performed the 56-100, WEG’s MS2, and Howden’s 33 meter turbine. The KVS 33 may have finally showed its stuff in the 4th quarter in comparison to the 56-100 and Howden’s 33/300. But here again WEG’s constant speed 25 meter turbine is equally productive without any fancy whiz-bang technology.

 Table: 1994 Quarterly Specific Yield of KVS33 and Selected Other Turbines
See WindStats Volume 8, No. 4, Autumn 1995

It should be noted that the specific yield shown here for Howden’s 330/33 is more representative of Howden’s performance than that in the CEC’s summarized statistics. The CEC report lists the Howden 330/33 with the swept area of the old 31 meter model, which has long since been replaced by the 33 meter version. The CEC reports substantially overstate Howden’s average yields.

 The KVS 33’s record for one full quarter in the San Gorgonio Pass doesn’t offer the company much solace. During the last quarter the KVS 33 under performed the aging 56-100s on the Whitewater Wash and expectedly fell far short of California’s best producers, the Vestas DWT turbines on Whitewater Hill.

But it’s a different story in Solano County. There the KVS 33 outproduced Kenetech’s 56-100 by 15% in the 3rd quarter and by 80% during the low winds of the 4th quarter. Possibly the watchful eye of the Sacramento Municipal Utility District, whose order for 45 additional MW depends on the success of the first 17 turbines, encouraged Kenetech to stay ahead of maintenance problems that have hindered operations of the KVS 33 elsewhere.

Solano County could represent the wind regime where the KVS 33 will perform best. By far it’s not the windiest region in California. The 56-100 have produced yields of about 670 kWh/m2/yr in two of the last four years and in 1991 produced 770 kWh/m2. The 56-100 can deliver exceptional performance at good sites. The 74 turbines on the Whitewater Wash near Palm Springs delivered a yield of nearly 1,300 kWh/m2 in 1994, nearly as good as any other machine in California. So it is possible that the KVS 33 will deliver on its promise of better performance in moderate wind regimes. It is just too soon to tell with certainty.

European Performance of Variable Speed Turbines

Now turning to Europe, no country has more installed variable speed wind turbines in the 300-500 size class than Germany. By the end of the 1st quarter of 1995 there were nearly 300 variable speed wind turbines, Enercon’s 32, 33, and 40 meter models, in operation.

 Frank Bourbeau of Enerpro, a California manufacturer of soft-start electronic controllers for constant speed machines, has suggested summarizing the quarterly data to determine trends between variable speed and constant speed wind turbines. If variable speed machines are successful at improving overall yields, they will lessen demand for Enerpro’s soft-start controllers.

For various reasons, however, the quarterly report’s statistics for average monthly yields don’t lend themselves to annual summaries. Most importantly are the substantial differences in wind resources from one month to the next within each quarter. Thus, the only way to examine the average monthly yields per quarter is quarter by quarter. By examining the specific yield of the summarized data reported by WindStats for eight quarters from 1993 to the 1st quarter of 1995, some trends do emerge.

Most noticeable is that Enercon’s E40 outperforms all 600 kW turbines in Germany by about 3%. This isn’t the whole story, though, because Vestas 500 kW outperforms Enercon’s E40 by about a similar amount. Nordtank’s 500 kW model outperforms both. But there are only half as many Nordank turbines as there are Vestas and Enercon and the data may reflect the location of the Nordtank turbines at superior sites.

 Enercon’s 33 meter model shows clear superiority to the earlier 32 meter model. But there’s no clear difference, at least in the statistics, between Enercon’s 33 meter model and the direct-drive E40.

 The best performance of all was turned in by the Micon 400 kW model, a constant speed machine. The 17 turbines in this class must be clustered at a particularly windy site.

 If there are any performance advantages to variable speed operation in Germany, they are lost in the statistical noise.

Table: Monthly Specific Yield in Germany by Quarter See WindStats Volume 8, No. 4, Autumn 1995

What about broadening the scope to include selected projects in Denmark (DK), Germany (D), Belgium (B), and the Netherlands (NL)? The results are similar. In 1994 Enercon E40s at four selected sites in Germany produced about 1,000-1,100 kWh/m2. At certain sites in Denmark and Germany the constant speed machines of Vestas, Bonus, and Micon did just as well.

 Table: Specific Yields of Selected Projects in Europe
See WindStats Volume 8, No. 4, Autumn 1995

The performance of Enercon and WindMaster turbines in the low countries may be telling. The 50 WindMaster turbines in the accompanying table are located on the enclosing dike of the Noordoostpolder on the east side of the IJsselmeer. They are downwind of the dike, but they do enjoy the unobstructed fetch across the IJsselmeer. The Windmaster turbines in Belgium are well known. They stand on the breakwater at Zeebrugge. Again they enjoy an obstructed fetch across the English Channel. The Enercon turbines are at one of the best sites in the Netherlands. They are on a dike of the outer delta works (Oosterscheldekering) and, like the WindMaster turbines at Zeebrugge, get the full brunt of winds sweeping across the English Channel. The WindMaster turbines represent fairly old technology. The turbines at Zeebrugge and the Noordoostpolder were designed in the mid 1980s and were installed in the late 1980s. Despite this, they are just as productive as the variable speed Enercon turbines at the delta works. The turbines, both variable speed and constant speed, all produce about 1,200 kWh/m2/yr.

 In spite of nearly two years of data on the performance of variable speed turbines, the results remain ambiguous. Certainly, variable speed turbines are as productive as constant speed machines. But whether they offer performance benefits, as once touted, remains to be seen.

Kenetech may resolve this problem. They have completed the largest project ever in the Netherlands, a 35 MW linear wind plant on a old dike near Eemshaven in the northeastern province of Groningen. The 94 Kenetech turbines abut a string of 40 Micon turbines, affording one of the best opportunities yet to compare the performance of the variable speed KVS 33 side-by-side with a constant speed Danish turbine. The wind industry will simply have to wait for the results before it can answer the question of whether variable speed turbines out perform constant speed machines.