The wind turbine industry has spent many years getting into gear, yet for a while now there has been almost as much talk about getting out. Dispensing with gearboxes in wind turbines has become all the rage. About a dozen companies are now following the path long taken by Germany's Enercon by choosing to couple the wind turbine rotor directly to a slow speed generator instead of using a gearbox to step up the low speed of the rotor shaft to a speed high enough to allow the use of a more conventional high speed generator. The direct drive pioneers, however, are mainly relative newcomers. Barring Enercon, a gearbox linked to a high speed generator remains an integral part of the technology supplied by all the world's leading wind turbine makers.
The main driving force behind the trend to do away with gearboxes is fear of gearbox failure. Market demand is for a simpler drive train with fewer components. The approach sidesteps both the risk of gearbox failure and the real chance that supply of gearboxes for wind turbines, at least in the short term, will not meet demand. At the same time, a notorious patent on variable speed wind turbine technology, taken out in America nearly 20 years ago and now owned by GE, has long had the wind industry seeking drive train configurations that avoid the need to accept a sub-optimal variable speed solution or do a deal with GE to stay out of court.
As wind turbines have grown in size over the past decade, gearboxes have failed -- and continue to do so -- across a range of turbine makes and gearbox suppliers (Windpower Monthly, November 2005). Although electrical system faults are the most frequent reason given for a wind turbine standing still, gearbox breakdowns head the list of failures, with the biggest economic consequences for plant owners.
The obvious appeal of direct drive technology is that no gearbox means no gearbox failures. Dispensing with it, however, brings its own set of problems. Slow speed generators are relatively rare on the market and, consequently, supplier choice is limited. They need to be big and strong to cope with all the torque and stress that a gearbox normally shoulders on behalf the generator. As a result, the mass of a slow speed generator is greater than that of a high speed generator and a gearbox combined. For many in the industry, the advantages of placing a heavy, and therefore more expensive, generator on top of a tall tower so that the gearbox can be left out have been questionable, particularly when those generators are nowhere near as readily available on the market as the more common high speed counterparts.
Recent technology developments connected with permanent magnet generators (PMGs), however, suggest there could be a third way. Some experts argue that use of a PMG in a direct drive turbine can reduce the overall mass of the drive train to that of a gearbox-plus-generator in a conventional wind turbine of equivalent size. At the same time, it has been demonstrated that a PMG provides an overall increase in efficiency since it produces more power at part load operation than achieved by a conventional generator, which reaches peak efficiency at full load. High efficiency at part load increases the output of wind turbines, which only operate in full load operation during periods when the wind is blowing hard enough for the machine to be operating at full rated capacity.
In essence, the fundamental difference between a PMG and a conventional generator is the source of the magnetism to create a magnetic field. In any generator, it is the relative motion of a magnetic field and a coil of wire that induces an electric current in the coil. The magnetic field can arise from a magnetised material, and this is the "permanent" magnet in a PMG. Alternatively, a current flowing in a coil induces a magnetic field, hence the term electro-magnetism associated with conventional generators in the wind industry.
Generators used in wind turbines are broadly of two types: synchronous and asynchronous (induction). The induction generator is rather simple in that the rotor (the rotating part of the generator) has windings in which induced currents circulate and magnetic fields are created, but there is no need for this to be connected to anything. The rotor will induce current in the stator coils that surround it and this current can be fed into the grid. The induction generator is extremely popular as both a generator and a motor and has been mass produced over a wide range of power ratings.
The synchronous generator, however, has technical advantages for operation on the grid -- specifically, control of power factor. But it requires either a special feed of current to the rotor (excitation) or use of permanent magnets to magnetise the rotor. Like the induction generator, the synchronous generator (without permanent magnets) is widely used in conventional power stations. It is also the generator type used by Enercon in its direct drive models. The PMG is a simple form of synchronous generator requiring no connections to the rotor. Instead, it carries permanent magnets around its periphery.
Whenever wind turbines operate in variable speed, which is the norm for modern megawatt scale systems, a power conditioning system, essentially complex power electronics, is required to convert the variable frequency electricity that results into electricity at standard network frequency. The doubly fed induction generator, more commonly referred to as DFIG, is much favoured in the wind industry for achieving this requirement. It is a clever form of asynchronous generator that can accommodate variable speed operation with a power conditioning system of only modest power rating -- perhaps only one-third of full rated power. The advantage is lower cost.
Torque, mass and cost
The power produced by a wind turbine is the product of torque and speed. As the turbine rotor speed is always comparatively low, the torque into the drive train is relatively high. It is torque above all that determines the physical size and cost of gearboxes and generators. The gearbox, if there is one, must deal with the high input torque. Its main function, however, is to increase speed: the same power emerging from the gearbox, except for some small losses, is then at high speed and fairly low torque. So a generator seeing high speed and low torque is commensurately less massive and less expensive than a generator designed to deal with the low speed, high torque output of the aerodynamic rotor in a direct drive configuration.
The bottom line is that all drive trains must deal with the high input torque of the low speed rotor. A lot of gearing renders the gearbox more complex and expensive, but makes life easy for the generator. No gearing avoids all the cost, efficiency and reliability penalties of a gearbox, but throws the whole job of dealing with the high input torque on to the generator.
Establishing the optimum drive train configuration is a major occupation of wind turbine companies and the wind research and development community. Should the torque handling duty be shared between the gearbox and generator? Should the gearbox be removed? Or is the answer to combine the best of both worlds and reduce the gearbox to two stages of gearing, where usually there are three? The motivation for the research is to raise energy production by improving reliability and limiting power losses, while reducing the mass of the equipment to cut capital costs.
The optimum drive train
Starting in 2000, a major drive train optimisation project, the WindPACT Advanced Drive Train Designs Study, got under way in the United States with the support of the National Renewable Energy Laboratory (NREL). To provide independent views, two separate main contractors, wind power consultants Global Energy Concepts and wind turbine maker Northern Power Systems, were selected. Each was required to evaluate drive train options and, in due course, develop a preferred option and detailed design that would be manufactured and tested.
Many systems were initially considered under WindPACT, including a multiple PMG solution, an unusual configuration now adopted by Clipper Wind in its Liberty turbine design. Significantly, both contractors finally preferred a system using a PMG, a direct drive PMG in the case of Northern Power Systems and one with a single stage of gearing in the case of Global Energy Concepts. A single stage gear is employed in the German Multibrid 5 MW turbine, which entered commercial production for offshore use in the summer. Multibrid is owned by French nuclear giant Areva.
Other drive train research has been conducted by British wind power consultancy company Garrad Hassan. In 2005, it undertook a major study of wind turbine design concepts with particular focus on the best drive train option. Again, the PMG surfaced as the most promising solution, although not specifically as a direct drive generator. At the same time as the Garrad Hassan study was concluding in favour of a PMG, GE Wind, quite independently, announced the introduction of its new wind turbine model, the GE 2.5 xl with a standard gearbox and high speed PMG.
GE's choice of a PMG was surprising, given the existence of the well established, and in most cases cheaper, DFIG and the more expensive power conversion equipment required. Two reasons emerge: better part load efficiency of the PMG and the advantage of a fully rated power converter, meaning the turbine can supply grids anywhere -- in Europe at 50 Hz or in America at 60 Hz -- with the same drive train hardware. It was also the PMG's greater efficiency in part load operation that led Garrad Hassan to conclude that it was the superior generator for economic wind energy conversion (figure 1).
Given the nature of the resource, a wind turbine operates in part load for most of its life. Increasing part load efficiency can mean energy gains of as much as 3% or 4% and these have a capitalised value far in excess of the extra cost of a PMG and its more expensive power electronics compared with a DFIG. In other words, a manufacturer should not shrink from making a generator substantially more expensive if the advantage achieved is part load efficiency gains. This is a perspective that is not that widely appreciated.
If a PMG is the best generator type, what should the remainder of the drive train look like for optimum wind energy conversion? Should it include a gearbox or not? If use of a gearbox brings economic advantages, how many stages of gearing is best?
Direct drive or not?
Beyond argument, no gearbox is the most reliable wind turbine configuration. Leaving it out means the machine is available for operation for more of the time, resulting in more captured energy to sell. Direct drive designs, however, have so far been heavier and by implication more expensive than conventional solutions (figure 2).
The latest PMG designs are aiming to narrow the gap and justify the concept with overall efficiency and reliability improvements. Even so, a gearbox and smaller PMG still comes in cheaper -- and results in cheaper energy production over the lifetime of the plant -- despite the possibility of more downtime. Where that may not be the case is offshore. The much higher cost of maintenance at sea, should things go wrong, could mean that more up-front investment in simplicity and reliability results in cheaper electricity over 20 years, or more, of operation.
On land, using a gearbox looks cheaper than no gearbox in the long run. That leaves the question of how much gearing is optimum, say for a 3 MW rated wind turbine. A single stage of gearing most typically produces a speed-up ratio of around 6:1, two stages 6x6=36, and so on. A wind turbine in the megawatt range with a rotor speed of 15 rpm will require an overall gear ratio of 1:100 to achieve 1500 rpm at the generator. This usually implies three stages of gearing -- and is the conventional solution -- but is not necessarily the most cost effective.
Garrad Hassan evaluated a series of PMG designs for a fixed output power of 3 MW, ranging from direct drive turbines with no gearbox to a traditional high speed generator configuration. The hundreds of parameters of each PMG design were optimised using genetic algorithms to produce a high efficiency, minimum cost generator corresponding to gear ratios from 1:1 to 1:120.
Direct drive generators tend to function best as a comparatively thin disk shape, or pancake, as clearly seen in the early Enercon turbines. Much later the characteristic Enercon pancake went through a cosmetic face lift and re-appeared in the recognisable egg-shape of Enercon's megawatt designs today. Conversely, high speed generators tend to become long and thin, resulting in the sausage-shaped nacelles that make them easily distinguishable from direct drive turbines. So which of these two extremes is the optimal for wind energy conversion?
What emerged from Garrad Hassan's number crunching is that the most efficient generator, both from the point of view of minimising relative structure cost and maximising overall electrical efficiency, is an intermediate ratio, where the generator is a compact cylinder -- neither pancake nor sausage. With that settled, attention turned to the choice of gearbox, or not, best suited to this type of generator, a complex process as different kinds of gearbox suited different overall ratios.
The aim was to determine a drive train configuration for minimum cost of energy, which required taking into account not only the capital cost of both the gearbox and the generator, but also the efficiency and reliability of both in combination. An optimum gear ratio was found to be in the mid-range, around 1:40, where a two speed gearbox would be appropriate. This suggested that a two speed gearbox combined with a PMG are the key ingredients of the optimum drive train.
Hybrid wins through
This hybrid solution, with a gearbox but less than three stages of gearing, represents roughly where companies such as Finland's Winwind, Multibrid, and most recently Innovative Windpower and Gamesa have arrived -- Innovative Windpower with its Falcon 1.25 MW machine designed by German engineering group Windforce and Gamesa with its prototype 4.5 MW turbine. The original Multibrid was developed by Aerodyn, a long established German wind turbine design company. The integrated Multibrid concept, comprising a single stage gear unit and PMG, has since evolved into systems that sometimes have two stages of gearing.
But whether a single stage gearbox is likely to be more reliable and result in cheaper energy than using two or three stages of gearing is not a foregone conclusion. Specialists at the gear design unit at Britain's Durham University are not convinced. Having more stages is the most economic solution, they argue, which allows for some of the cost difference to be used to pay for increased design margins, meaning increased reliability. The approach means a two stage gearbox may be both more economic and more reliable than a single stage unit.
Overall, the Garrad Hassan study plumped for the hybrid solution: a two stage gearbox and PMG. But the question of how much gearing, if any, was much harder to resolve than the preference for a PMG. Whether that is the right track to be on, time will tell. Winwind, Multibrid and now Gamesa have endorsed the hybrid drive train solution. Gamesa has chosen a two stage planetary gearbox (ratio 1:37) and multi-pole PMG in its new G10X 4.5 MW wind turbine design.
Small wind turbines run at comparatively high speed and for that reason direct drive PMG designs are long established among wind turbines with capacity ratings of around 100 kW or less. For utility scale wind turbines, the potential benefits in efficiency and reliability of direct drive turbines, with or without a PMG, are only now being tried out commercially by more than a very select few (table next page). Among these, Enercon has long demonstrated that direct drive technology can be commercial, although it has not employed a PMG. Its all-electric turbine identity, with a minimisation of other technologies like gear mechanisms and hydraulics, has been unique.
Others are now entering the market with variations on the direct drive theme. Enercon's preferred generator design, with a wound field rotor, has been deserted by American company Northern Power Systems and Japanese Harakosan, a company that emerged out of Zephyros, a progeny of the pioneering Dutch Lagerwey company from the 1990s. Both favour a PMG. Northern Power Systems, which makes the Northwind 100 kW turbines, particularly for use in cold climates, has developed a 2.2 MW turbine with a direct drive PMG on the heels of its WindPACT work for NREL. Two prototypes will be made next year, the company says. Harakosan markets a 2 MW wind turbine, a group of which operate in a small wind farm in Finland.
A new direct drive design, the small 75 kW U50, has been launched by Korean wind turbine supplier Unison. Instead of the usual front heavy design, with generator and rotor located upwind of the tower centreline, the Unison generator is set at the rear of the nacelle, counterbalancing the rotor and producing an almost spherical wind turbine top.
Others believe in the Enercon concept without permanent magnets. A new 1 MW design by Croatian company Koncar is direct drive with a wound rotor field, a concept also being followed by Emergya Wind Technologies, a Dutch company that acquired the earlier Lagerwey direct drive technology based on a wound rotor.
Wound field rotor designs were initially favoured in the 1990s, when high strength neodymium-iron-boron magnets were much more expensive. Designs employing a PMG with ferrite magnets would have been very heavy. Compared to a PMG, the wound field rotor with its separate set of electrical windings gives extra control of the rotor magnetic field, including the ability to switch it off. It is more complex than a PMG, however, requiring a power supply to the windings involving slip rings. Another potential disadvantage is that a direct drive generator with wound rotor seems to be generally heavier than a PMG.
Goldwind and Siemens
China's Goldwind, because of the rapidly escalating scale of its wind turbine business, is probably contributing most to the market penetration and operational experience of PMGs in wind turbines. It makes both a 1.2 MW and 1.5 MW PMG turbine, having acquired the technology under licence from Vensys Energiesysteme, a small German wind turbine engineering company. Thirty-three of the 1.5 MW turbine supplied electricity to this year's Olympic Games in Beijing. Last year, Goldwind lay eighth in the world rankings of wind turbine suppliers with just over 1% of the global market, according to BTM Consult's annual World Market Update. It may well have improved on that position during 2008.
Perhaps even more significantly is the interest being shown in direct drive PMG technology by the world's sixth ranking wind turbine manufacturer, Siemens Energy, traditionally known for having one of the most conservative approaches to wind turbine design. Using its 3.6 MW turbine as a starting point, Siemens, which acquired Denmark's Bonus line of turbines, is conducting a major investigation of PMG direct drive technology.
Technical director Henrik Stiesdal, who admits the Siemens research is a possible "game changer" (Windpower Monthly, September 2008), clearly has an open mind as yet. He argues that there is no fundamental problem with using gearboxes in wind turbines. Of 1100 Bonus machines sold into California between 1983 and 1987, all with gearboxes, he has since traced 1060 of these machines, with a mean age of 23 years, of which 96% are still running.
Siemens may have its eye on using PMGs offshore to reduce maintenance by having a drive train with fewer parts. Like its current offshore technology, the experimental 3.6 MW will have a wholly enclosed nacelle. On the topic of maintenance, concerns have been expressed about the practicalities of working with the high strength permanent magnetic fields of the PMG, but Stiesdal sees no specific problems. His view is shared by engineers at The Switch, a supplier of PMGs, although both acknowledge that the factory assembly of PMGs demands special tools and procedures for safe handling of strong magnets.
British company Converteam is providing the first direct drive technology to Siemens. According to the company's Derek Grieve, permanent magnets enable the mass of the direct drive generator to be reduced to approximately that of an equivalent gearbox plus conventional generator, plus a substantial 5% gain in overall efficiency. Even so, there is little evidence of direct drive designs leading to lighter drive trains than conventional designs (figure 2).
Siemens, in rejecting a DFIG in favour of a PMG in its direct drive design, was particularly focused on large offshore turbines. Although the DFIG is cheaper than a PMG and full power converter, converter technology is young and prices are dropping rapidly. Moreover, Stiesdal mistrusts any system with slip rings, especially offshore, even though he acknowledges the technology may be perfectly satisfactory. Grid compatibility is also more difficult for the DFIG compared with a synchronous generator, with or without magnets, and thus adds complexity and cost to meeting prevailing grid codes. Lastly, the DFIG has a significant efficiency penalty compared with the PMG's part load efficiency advantages.
As experience with PMG technology in wind turbines increases, thanks to their use in increasing quantities of turbines made by Goldwind and others, such as America's Clipper Windpower, the knowledge base will grow and more refinements are likely.
There is also growing interest in superconducting generators, using high temperature superconductor (HTS) wire. Superconductivity, which means conduct of power with negligible losses, can now be achieved at typical temperatures of liquid nitrogen, around -200¡C. When superconductivity was first discovered, it was only observed at yet lower temperatures approaching absolute zero, putting it out of reach for commercial use. The HTS wire is claimed to carry approximately ten times the current of a comparable copper wire, resulting in dramatic reduction in machine size and a substantial increase in drive train efficiency, due to the near elimination of electrical losses in HTS materials. In 2005, Siemens recorded the successful initial operation of the world's first generator for marine technology, rated at 4 MW, based on HTS designed and manufactured by European Advanced Superconductors.
A company active in the wind industry, American Superconductor Corporation (AMSC), claims it has designs "coming soon" for wind turbine systems that will halve losses and reduce generator weight to one-third of today's typical sizing. AMSC, however, is renown among investors for making claims that have yet to see the light of day. While the superconductor technology is certainly worth watching, it may be a way off yet. While it promises large mass reductions in a direct drive generator, the need for cryogenic systems to maintain the low temperatures in the conducting elements is a counterbalancing complexity and cost.