An edited version of this article appeared in the Summer 1999 (Vol. 12, No. 3) edition of WindStats. For access to the tables and statistics mentioned in the text, please see the hard copy version of the article.
With more than US$1 billion of wind capacity newly installed in the United States, attention has begun to shift from the “wind rush” to its aftermath.
Will the turbines installed with such fanfare perform as well as promised.
For the wind industry in the United States this is more than a rhetorical question. The answer could determine the future of wind energy in America and could have repercussions on wind development around the globe. Much is at stake, not the least are the large sums of money invested in a region once considered to contain only a marginal–but vast–wind resource.
Are Midwestern performance projections aggressive? Yes. Meteorologists are projecting annual specific yields from 1,100 kilowatt-hours (kWh) per square meter (m²) of rotor swept area to a high of 1,280 kWh/m²–yields far in excess of average California production and existing wind turbines in the Midwest. Projected Midwestern yields are typically 50 percent greater than existing performance in post-1985 wind turbines in California, and from the few turbines previously operating in Iowa, and Ontario Canada.
Are projected Midwestern yields too aggressive? Possibly. Only time will tell, and only then if production figures from the new arrays become publicly available. Meteorologists make the case that the new projects take advantage of advances in Midwest wind resource assessments and wind turbine manufacturer’s ability to provide extremely tall towers. The results during the next few years will determine if widely used meteorological models are up to the standard that the public and financial communities now expect.
Specific yields, in units of electricity generation relative to the area of the wind stream swept by a wind turbine’s rotor (kWh/m²), are a more reliable means for comparing performance than capacity (plant) factor and generation relative to installed capacity (kWh/kW). Unlike capacity factor, specific yield is not dependent on generator size. Specific yield in kWh/m² is determined solely by the wind resource and the overall efficiency of the wind turbine in capturing it. Assuming that the wind turbines operate reliably, for example within 98 to 99 percent availability, the strength of the wind resource is the principal determinant of annual specific yields.
Because the conversion efficiencies of modern wind turbines are roughly comparable, lower than expected yields often indicate lower than expected availability, a wind resource less energetic than expected, or some combination of the two.
Projected Midwestern Specific Yields
WindStats has pieced together production estimates from various public and private sources for some of the Midwestern projects.
According to Enron Wind Corp.’s web site, their $235 million project for Mid-American and IES Utilities in Northwestern Iowa will produce a combined 600 million kWh per year. The 193 megawatt (MW) project will use 257 of Enron’s Z750 turbines. Enron’s web site says the Z750 series in Iowa uses a 50 meter diameter rotor. Thus, this project will yield about 1,190 kWh/m² annually according to Enron’s publicity.
Madison Gas & Electric’s joint project with Wisconsin Electric will produce 46 million kWh from 33 of Vestas’ V47 turbines for an annual specific yield of 800 kWh/m². The turbines will be installed along the Niagara Escarpment, in Wisconsin’s best wind resource.Disclosure: Paul Gipe has consulted to or written for NRG Systems, SeaWest, Aerovironment, USDOE, NREL, NASA, AWEA, KWEA, CanWea, DGW, EECA, Microsoft, the Izaak Walton League, the Minnesota Project, Independent Energy Magazine, WindStats, Windpower Monthly, Systemes Solaires, New Energy, and others. From 1984 to 1985 he worked for Zond Systems. Northern Alternative Energy installed several wind projects in Minnesota during the wind rush. NAE’s Micon project uses 56 meter towers, their Vestas project, 65 meter towers, according to data from NAE’s web site. The Micon turbines use a 48 meter diameter rotor, the largest outside Zond’s Z750, and are expected to produce 1,100 kWh/m². The Vestas turbines, with a slightly smaller rotor, are expected to produce a somewhat greater yield: 1,150 kWh/m².
Dan Juhl estimates his DanMar project near Woodstock, Minnesota will produce 33 million kWh per year. Juhl’s projections are the most aggressive of Midwestern estimates with an annual yield of 1,280 kWh/m² using 17 of Vestas’ V44s.
Two small projects were also installed in Nebraska as a peripheral part of the wind rush. The Nebraska Public Power District installed two Zond Z750s in Springview near the border with South Dakota. Lincoln Electric Service installed one Vestas V47 visible from I-80 near Lincoln in the Southeastern corner of the state. The Zond turbines are expected to generate 4.25 million kWh per year for a yield of about 1,110 kWh/m², while the Vestas turbine at Lincoln is expected to deliver 1.66 million kWh for a yield of about 950 kWh/m².
These projected yields are comparable to the best found in California’s windy passes. Few arrays in North America have been this productive, year in, year out.
Comparison with California
By far the most productive turbines in California have been those operated by Bill Adams’ San Gorgonio Farms atop Whitewater Hill in the throat of the San Gorgonio Pass. (See Table Specific Yield of Selected Turbines in California.) Of these, Adams’ DWT (Vestas) turbines are consistently the most productive. Average yields of the 400 kW turbines range from a low of 1,100 kWh/m² to a high of nearly 1600 kWh/m² for the cluster of 35 machines.
Turbines built by Britain’s Wind Energy Group have turned in surprisingly good results considering the generally less energetic wind resource in the Altamont Pass than elsewhere in California. The array of WEG’s 25 meter diameter MS2 turbines produced a striking yield of nearly 1,300 kWh/m² in 1995. The 250 kW turbines typically produced about 1,100 kWh/m² per year except when major gearbox repairs were needed and availability suffered.
Other projects have on occasion produced more than 1,100 kWh/m² in California. DanWin’s 23 meter, 160 kW turbines and Mitsubishi’s 28 meter, 250 kW turbines have exceeded the 1,000 kWh/m² mark in several years.
It’s clear that even the older technology can deliver high yields. But doing so consistently has been difficult, witness the uneven performance of WEG’s machines in the fairly benign environment of the Altamont Pass.
Two of California’s most modern large arrays are those developed in the Tehachapi Pass by the former Zond Systems, now a part of Enron: Sky River and Victory Garden IV. Both projects use Vestas’ V27. There are 98 of the 225 kW turbines installed in Victory Garden IV on the Busee Hills 5 kilometers southeast of the town of Tehachapi. There are another 342 turbines in Sky River on the crest of the Tehachapi Mountains 20 kilometers northeast of Tehachapi. (See Table Comparison of Projection to Actual for a Large California Project.)
Both the turbines and the meteorological techniques used represented state-of-the-art technology during the early 1990s. The wind resource assessment at the two sites was extensive. And the V27 design, a variable pitch machine which appeared in the late 1980s, reached the apex of its deployment during the early 1990s.
Between these two sites annual yields range from 850 to 1,200 kWh/m² according to data from the California Energy Commission’s Performance Reporting System. Overall annual production varied from 10 percent above to 14 percent below projections from 1993 through 1995. Unfortunately, only three full years of data are available from the CEC’s PRS. No annual data has been available since the 1995 report, though developers are required by law to report production to the CEC.
Comparison with Other Sites
Determining annual specific yields is a useful exercise for a quick “realty check” of performance projections. For example, when proponents of new wind turbine designs, such as New Zealand’s Vortec, begin bandying about unusually high production estimates a comparison with high performance projects nearby or elsewhere in the world will indicate the realm of what’s possible with existing technology.
One of the world’s most productive, if not “the” most productive wind turbine, has been the lone Vestas V27 overlooking Wellington, New Zealand. (See Table Sample of actual Production in the Midwest and Elsewhere) For the past four years the turbine has consistently produced nearly one million kWh annually. Though it once appeared that the seven turbines at La Ventosa in Mexico, would surpass performance in Wellington, they have not done so. Problems with maintenance quickly led to outages and poor availability. Performance suffered. But the V27 in Wellington has reliably delivered average annual yields of nearly 1,600 kWh/m². During one year the turbine generated more than one million kWh for a yield of 1,750 kWh/m², probably a world record.
One of the most productive cluster of wind turbines in Denmark are the three Hanstholme møelle in northwest Jutland. The port of Hanstholme is probably one of the windiest sites in Denmark. In 1997 the 36 meter diameter locally built turbines generated 4.29 million kWh for a yield of 1,400 kWh/m².
As seen at Whitewater Hill, the best sites in California are capable of delivering peak annual yields of nearly 1,600 kWh/m² during a good year. However, average yields at good sites range from 1,100 kWh/m² to 1,200 kWh/m² at best.
Existing sites in the Midwest are far less productive. The Tacke 600 on a 50 meter tower at the Bruce Nuclear plant near Kincardine, Ontario has delivered an average of 1.15 million kWh during its four years of operation for a yield of 800 kWh/m². The Vestas V44 above Traverse City, Michigan has produced an average of only 800,000 kWh per year at its perch on the east side of Lake Michigan.
Possibly more representative of performance in the Upper Midwest are the two NEG-Micon 600 kW turbines near Sibley, Iowa. (See Table Sibley Iowa Production.) 1998 was the first full calendar of operation. According to data provided by NAE to WindStats, the turbines generated from 1.1 to 1.3 million kWh that year. The average yield of both turbines at a 46 meter hub height was about 800 kWh/m², well short of the 1,100 to 1,200 kWh/m² per year projected for the new projects in Iowa and Minnesota.
There are three aspects of performance projections that entail a degree of technical uncertainty: assessments of the wind resource, accuracy of wind turbine power curves, and estimates of losses from various sources (availability or down time, array effects, intra project transmission losses, and so on). Calculating typical losses that a project can expect, especially those of array effects, introduces considerable uncertainty.
It’s also no secret in the wind industry that despite international measurement standards not all power curves are created equal. Some manufacturers produce power curves “more aggressive” than others. An exasperated engineer at WindWorld once asked how a competitor could produce more power than WindWorld’s then current model when they both used an identical rotor and similar pitch settings. The answer was simple, the marketing department of WindWorld’s competitor “selected” the power curve they preferred.
Still, it’s estimating the amount of fuel available to a wind power plant, the wind, that introduces the greatest unknown. For this reason wind developers undertake extensive resource assessments before taking their projects to Wall Street or Zurich. And it’s the unusual wind resource in the Midwest that has led to such aggressive performance estimates.
One reason for the wind boom in Minnesota is the state’s decades old program monitoring the wind resource and making this data available to the public. It was the state of Minnesota’s Wind Resource Assessment Program or WRAP that first called attention to the enormous potential on Buffalo Ridge in the southwestern corner of the state. Previous nationwide surveys had lumped the area into Battelle Pacific Northwest Laboratories’ Class 4 category. A class 4 resource in the Battelle system has average annual wind speeds of 6.5 to 7.0 m/s at 30 meters above the ground. While good, it’s far from excellent.
The ridge is actually a low-lying hill 100-kilometers (60-miles) long running southeastward from Sica Hollow, South Dakota to Spirit Lake, Iowa. The “ridge” is more correctly termed a low rise that forms the drainage divide between the Mississippi and Missouri watersheds. The strong winter winds that sweep across this subtle topographic feature, give the ridge a resource comparable to that found in some of California’s mountain passes.
Minnesota went further than most other state monitoring programs and has used tall instrumented towers with measurements at 30, 60, and 90 meters above ground. And what they found was startling: wind shear exceeding 0.40 (_, [editor note that this is the Greek letter alpha]) at some tower heights. Meteorologists had suspected that such a strong wind sheer existed from the so-called nocturnal jet but it wasn’t until data was collected that it became apparent how beneficial it might be.
One reason for the aggressive performance projections may be due to developers natural tendency to use the best sites first. For example, Minnesota wind development through June 1999 has used the “best of the best” says Rory Artig, an engineer with the state’s Department of Public Service. Artig notes that Great River Energy’s three Vestas’ V47s near Chandler, Minnesota are near the “peak of Buffalo Ridge,” as are Northern States Power’s Phase II and Phase III projects.
Though the WRAP tower at Chandler found wind shears from the 30 to 50 meter heights typical of the Great Plains, at heights from 50 to 70 meters shear jumped to 0.42. (See Table Minnesota Department of Public Service Chandler Tall Tower.) “The resource is very strong,” say Artig. “It’s quite different from that in California. . . You see quite high shear at the upper levels. From 1996 through 1998 Artig found that the 2-1/2 year average annual wind speed at Chandler was 6.9 m/s at 30 meters, 7.6 m/s at 60 meters, and 8.2 m/s at 90 meters above ground level.
Meteorologists who have examined the Minnesota data as well as that form private sources agree with Artig. “We’ve seen high shear, particularly in the wintertime,” says Jack Kline, a principal in RAM Associates. The Upper Midwest is “not a particularly robust wind regime otherwise.”
The tallest tilt-up tower used for Midwest resource assessments is 50 meters. For heights above 50 meters meteorologists rely on the WRAP data or on data from regional telecom towers. But Kline cautions that determining shear “beyond you’re measuring point” relies on extrapolation, in effect an educated guess.
Minnesota’s Artig is optimistic. “If they (NSP’s phase II & III) can solve their problems, they can very likely meet their projections,” he says. “They’ve had some start up problems. . . but it can be done.”
High shear may be a regional phenomena. If so, this augers well for wind development throughout the upper Midwest. It’s characteristic of Buffalo Ridge according to meteorologist Ron Nierenberg. This contrasts with California’s Tehachapi Pass where shear on Cameron Ridge is near zero. At exposed sites in Minnesota, says Nierenberg, wind shear is often double that of the 1/7 power law, from 0.2 to 0.3. It’s similar in Iowa and Wisconsin. Trees, he says, have this effect (on wind shear) throughout the Plains’ states. For comparison, Cameron Ridge in California is virtually treeless.
During summer months when wind speeds are typically low in a continental wind regime, a “nocturnal jet” may occur at a certain height above ground where wind shear can reach 0.4. This “jet” has nothing to do with the jet stream, Nierenberg explains, it’s simply a layer of fast moving air. “There are lots of places in the world where there’s a localized zone of high winds, a so-called jet,” he says. “Buffalo Ridge, at 600 meters above sea level, is possibly just high enough to pierce it.”
There’s a similar “jet” about 300 meters above California’s enormous San Joaquin Valley. In part it is this jet which produces the Tehachapi Wind Resource Area. However, within the Tehachapi Pass the ambient shear, or the shear before wind turbines were installed, was near zero. With the addition of the wind turbines, for example on Cameron Ridge, there’s now a “wake induced shear,” says Nierenberg.
One of wind energy’s natural oddities is the negative shear in California’s Altamont Pass. The wind resource of the Altamont Pass is characterized by a relatively thin zone of high winds. Above this lens, wind speeds decrease. This negative shear was made famous by Pacific Gas & Electric’s installation of a Boeing Mod-2 in Solano County. The turbine was so tall that the upper part of the rotor poked through the high-wind lens and partly operated in the low winds aloft–when it was in service.
Meteorologists comment that the rotor on the Mod-2 would sometime see winds aloft not only of a different speed from that near ground level, but in the opposite direction as well. These unexpected conditions may have contributed to the mechanical problems the turbine suffered.
Another plus for the Midwest wind resource, says Nierenberg, is that it’s “a much more benevolent environment” than that in the Tehachapi Pass. Wind speeds “rarely exceed 20 m/s in the Midwest whereas in Tehachapi there are periods of extremely high and low winds.” It’s the extremely high winds that cause the increased fatigue cycles on turbines that are characteristic of Tehachapi sites, he explains.
There’s also 15 percent less air density at 1500 meters in the Tehachapi Pass than in the Midwest. “They’ve simply got more air to work with” in the Midwest, Nierenberg says. During the Midwest’s cold winters air density is near that of sea level. “If the machines hold up, they should be able to do it (meet projections),” says Nierenberg. But he quickly adds that only “if they don’t put them too close together.”
For the most part developers are installing about half the wind capacity per hectare than typical of California. The Midwestern arrays are more open than those in California with 4 rotor diameter by 8 rotor diameter spacing or 5 rotor diameter by 7 rotor diameter spacing. This contrasts with the 2 by 6 spacing sometimes found in California.
Clearly, performance projections for Midwestern projects are aggressive. But no one knows with certainty whether these projections are too aggressive. The burden is on developers, manufacturers, and meteorologists to prove that indeed the turbines can consistently deliver such performance under Midwestern conditions. Only time will tell if the turbines and those who manage them can meet such lofty projections.
Disclosure: Paul Gipe has consulted to or written for NRG Systems, SeaWest, Aerovironment, USDOE, NREL, NASA, AWEA, KWEA, CanWea, DGW, EECA, Microsoft, the Izaak Walton League, the Minnesota Project, Independent Energy Magazine, WindStats, Windpower Monthly, Systemes Solaires, New Energy, and others. From 1984 to 1985 he worked for Zond Systems.