LONDON — As ever in turbine design, generator selection is inextricably linked to many other design decisions: the drivetrain speed, the main bearing arrangement, the nacelle structure and top head mass which, in turn, all feed into the ultimate formula: the lowest possible lifetime cost of energy for each turbine. For now, with efficiency and grid compliance the top demands, what are the market trends in generator selection?
Manufacturers have used a variety of generator designs in their variable speed turbines. The Dual-Fed Induction Generator (DFIG), an industry standard since the late 1990s, currently rules the roost in volume terms but its need for a high-speed gearbox, extra maintenance and difficulty in complying with grid codes means turbine manufacturers have been looking for new directions. Today, it’s permanent magnet generator (PMG) technology that looks most promising.
In a high-speed DFIG drivetrain, a slow-turning shaft from the rotor (10-20 rpm) drives a gearbox whose output shaft, rotating at up to 2000 rpm, drives the generator. In a DFIG, both rotor and stator use electrically excited copper windings to create magnetic fields. As the rotor spins, interaction between these fields generates electricity. DFIGs must spin at 750-1500 rpm to operate, hence they are restricted to high-speed applications.
The rotor circuit is controlled by a power electronics converter, while the stator is connected directly to the grid. This converter controls voltages and currents, keeping the DFIG synchronised with the grid while turbine rotor speed varies (typically the range is +/- 30% of the synchronous speed or 60% to 110% of the DFIG’s rated speed).
The great advantage of the DFIG is that it only requires a ‘partial’ — roughly 35% of the generator’s rated capacity — converter because only 25%-30% of the input mechanical energy is fed to the grid through the converter from the rotor, the rest going directly to the grid from the stator. The efficiency of the DFIG is very good for the same reason; little power is lost via the converter.
Controlling the rotor circuit in this way also allows the generator to import and export reactive power to support the grid during outages — Low Voltage Ride-Through (LVRT). However, today’s more demanding grid codes stretch this to the limit and many existing DFIGs have had to be retrofitted with extra electronics to cope.
In PMGs and in other synchronous designs like the EESG where the electrical energy is generated at a variable frequency related to the rotational speed of the rotor, the output must be converted to match the frequency of the grid. Here the electronics must deal with the full power output, demanding full power converters which are considerably more expensive than partial converters – around three times as much according to Indar — and which also have greater electrical losses.
But as turbines become larger and more advanced, vendors are looking to these PMG designs to enhance reliability and serviceability, reduce weight and comply with grid codes. For those manufacturers looking to eliminate the gearbox, compact PMGs are particularly attractive. Slow rotation speeds typically demand much larger diameter generators to accommodate the increase in the number of magnetic poles on the rotor for direct drive applications.
PMGs operate in much the same way as EESGs except, as their name suggests, they employ magnets in the rotor instead of windings to create the magnetic field required. This means no slip rings or brushes, and so reduced maintenance and greater reliability. The high energy density of permanent magnets (a 15 mm-thick segment of permanent magnets can generate the same magnetic field as a 10-15 cm section of energised copper coils) also helps to deliver a lighter, more compact unit.
PMGs are almost as efficient at full-load generation as standard DFIGs, but are more efficient at part-loads – the most common conditions that wind turbines operate in. DFIGs are more efficient in high, steady winds, but must have electrical current injected into the rotor at low speeds, resulting in lower efficiency. Companies such as GE and Vestas have used PMGs for some years in various models and have more recently been joined by the likes of Alstom and Siemens.
A key attraction for manufacturers is that a full power converter (FPC) confers greater ability to comply with the latest grid codes, of which LVRT is the main element. To support grid voltage during a voltage dip, the turbine drive train and its power converter must inject reactive current.
Because it is completely decoupled from the grid, full power converters can support longer, lower dips than a standard DFIG whose otherwise efficient partial converter works against it here. This full decoupling between a PMG and the grid can also potentially lengthen gearbox life due to reduced loads on the drivetrain and does away with the parasitic currents found in DFIGs which can damage generator bearings.
So why doesn’t everyone use PMGs?
With these advantages, why isn’t everyone rushing to PMG designs? One reason is cost when compared to established technologies. Manufacturing large PMGs for direct drive turbines is challenging. A tiny air gap of a few millimetres between rotor and stator demands tighter tolerances and maintaining these standards when machining components 6 metres in diameter is a serious production challenge.
It also means turbine designers have to make generator supports sufficiently rigid in order to prevent potentially fatal distortions. Originally developed for geared machines, Alstom’s Pure Torque design uses an elastic coupling to isolate the massive 8 metres diameter PMG from non-rotational loads in its Haliade machine.
Another disincentive for OEMs is that large PMGs require expensive rare earth magnets whose price volatility has been well documented. Other issues include NdFeB magnets’ susceptibility to corrosion and their sensitivity to heat: go much above 80°C and electrical losses climb rapidly. Worse, there’s a risk of reversed polarity or permanently losing magnetic field strength. Manufacturers such as The Switch use special coatings to prevent corrosion and incorporate proprietary hybrid air-and-liquid cooling systems in their generator designs.
There are also some concerns over the reliability of full-power converters. Studies such as the EU’s Reliawind point to power electronics rather than gearboxes as the most prone to faults. Frequency converters had a failure rate of 12.96% (failures/turbine/year) and contributed 18.39% to average time lost (hours/year) while gearboxes failed at a rate of 5.66% and contributed 4.66% of lost time. REpower has said that gearbox failures only contribute between 4% and 8% of total turbine failures set against around 30% for electrical systems and power electronics.
In the largest direct-drive machines, manufacturers are now using multipart stators, each with a dedicated converter. This means that the generator will continue to function even if one element fails. For example, the Haliade’s PMG (manufactured by GE Power Delivery) can lose one of its three converters but continue to generate at up to 4 MW. It’s also possible to isolate individual stator coils if required. In Siemens’ SWT-6.0-120 the stator is electrically split into two halves, each with its own converter, allowing it to operate at 50% capacity if a generator section or converter fails.
Looking to the future
So when are PMGs going to take over? The short answer is, not just yet. DFIGs are a cheap, well-proven technology that will be around in volume for years to come. For example, REpower’s latest 6.15 MW 6M model employs a similar high-speed, non-integrated geared drive system to its predecessor the 5M and sticks with the DFIG. The vast majority of today’s turbines still employ this generator design and whether it or the PMG is the most efficient over the full operational range of a turbine is still debatable.
‘There are now DFIG solutions in the market with maintenance conditions similar to the PMG machines,’ says Xabier Irure, Indar’s wind power & series division export manager, noting improvements to historical DFIG challenges like brushes, bearings and insulation. ‘Today, both partial and full converters can be designed to satisfy 100% of grid codes. We believe that the DFIG machine has still a market but it is very limited to high-speed solutions.’
Other generator types are still valid too. Leroy Somner recently released a new ‘high efficiency’ brushless EESG intended to compete with PMGs in medium and high-speed applications. This employs a full-power converter integrated with the generator and the company claims that the machine can match or beat PMG efficiency even at part loads by varying the air-gap magnetic flux — something that PMGs are unable to do.
There are also evolutions of the DFIG such as Indar’s xDFM. This combines a small PMG with a large DFIG within the same machine. In this design, the partial converter is connected between the rotor of the DFIG, and the stator of the PMG and so is isolated from the grid; only the stator is connected to the grid. This greatly enhances its LVRT capabilities while retaining the cost advantages of a partial converter.
The drivetrain speed is also fundamental to generator selection; a DFIG is not going to work with a direct drive while a PMG does not appear to offer any great advantage over a modern DFIG when used with a high-speed drivetrain. The qualities of lightness, compactness, cost, maintenance, reliability and so on also feed into generator selection, as does whether the turbine will be used on- or offshore.
‘A conventional high-speed drivetrain with 3-4 stage gearbox and DFIG generators is well proven and relatively compact,’ says Panu Kurronen, product manager at The Switch. ‘But gearboxes can give trouble, and traditional designs can be bulky and heavy. A direct drive means no gearbox but has a very large generator, lots of (potentially unreliable) electronics and expensive rare earth materials are needed for PMG generators.
‘As far as I can see, the medium speed with one or two stages is the best compromise for larger (4-6 MW) turbines,’ continues Kurronen. ‘It gives reasonable size and weight with high efficiency and no high-speed gears which are most vulnerable to damage. This solution also enables high level of integration.’ Medium-speed PMG drivetrains are appearing, for example, in Vestas’ V-164. Vestas cites magnet supply as a reason to select a geared solution.
With lower wear than a high-speed gearbox, it avoids the very bulky, magnet-hungry generators that ultra-low, direct-drive speeds entail. According to Indar, a direct-drive PMG will use 12 times as many magnets as a high-speed PMG while a medium-speed design will use between 1.4 and three times as many.
A hybrid or integrated design whose gearbox and generator share the same frame, bearings and shaft also helps to reduce weight and bulk. The Areva Multibrid M5000, which combines a medium-speed gearbox with a PMG, is the best-known turbine design of this type. However, this approach requires greater collaboration between suppliers, particularly where there are shared components, and also reduces the manufacturer’s supplier alternatives.
The latest integrated development is the medium-speed (400 rpm), 3-6 MW FusionDrive from The Switch and Moventas. Built as one unit, it is claimed to be the smallest and lightest combination of gearbox and PMG on the market.
Light and compact best describes ABB’s medium-speed PMG generator, released late last year. For a 7 MW PMG generator at a nominal speed of around 400 rpm, ABB quotes a diameter of 3 metres and a weight of under 30,000 kg. It claims over 98% efficiency with partial load efficiencies of 98% even at 20% load. Rated at up to 8 MW, the generator uses a full-power converter and its ‘semi-integrated’ modular design employs a flange to connect the gearbox and generator.
Beyond PMGs, there has been little recent innovation in generator technology. For now, and for all the reasons outlined above — with efficiency and grid compliance at the top – the market trend is definitely towards PMGs. As ever in turbine design, generator selection is inextricably linked to many other design decisions: the drivetrain speed, the main bearing arrangement, the nacelle structure and top head mass which, in turn, all feed into the ultimate fomula: the lowest possible lifetime cost of energy for each turbine.
James Lawson is a freelance journalist focusing on the energy sector.