Reliability: Improving by design
Historically, the weakest link of a wind turbine has been its gearbox. As turbine sizes have increased, designing gearboxes able to handle the forces generated by longer and heavier blades has become problematic. Making matters worse, turbine loading is variable and hard to predict. It is not uncommon to have a gearbox fail in less than two years of operation.
Most gearbox failures have been due to movement of the machine chassis, which causes misalignment of the gearbox with generator shafts and leads to failure. Such failures typically occur in the high-speed rear gearing portion of the gearbox when the bearings become faulty. The frequency of failures can be reduced by regular, once-a-year turbine realignments.
Most manufacturers have made their turbines more reliable by improving the lube-oil filtration system in the gearbox so it can remove all particles larger than seven microns across. If a particle of that size breaks free of the meshing gears, it can damage other gears and bearings.
Some wind turbine vendors, like Enercon GmbH (www.enercon.de), are experimenting with increasing the number of poles in their machines, making it possible to eliminate the gearbox. Though most turbines have four or six magnetic fields from windings (pole pairs) and use a gearbox, if the generator has 50 to 100 pole pairs, the use of electronic control also can eliminate the need for a gearbox. Coupling the blades directly to the generator in machines without a gearbox also eliminates the mechanical or tonal noise produced by conventional turbines.
Other manufacturers, like Clipper Windpower Plc (www.clipperwind.com), have further improved reliability by using distributed gearing using multiple paths and multiple generators (see Cover Story, Steel Winds). The company claims this approach will ensure continued turbine operation even if one of the generators fails. But the claim will have to be verified by operating experience of the first plant to use eight of Clipper’s production-model, 2.5-MW Liberty I machines—the Steel Winds project near Buffalo, N.Y., which only entered service in April 2007.
Reactive power needs
Wind turbines typically drive an asynchronous generator that consumes, rather than generates, reactive power. Consequently, their power factor must be corrected before the wind farm can be connected to the transmission system. Compensation for the reactive power corresponding to no-load conditions is typically done using fixed capacitors within the facility. Any remaining reactive power consumption in excess of that must be compensated for by other methods.
Most turbines have integral inverters that can convert the turbine’s output to direct current and then back to alternating current at any desired power factor. However, using inverters for power factor correction may create problematic stray currents in the generator rotor. If a stray current is drawn to ground by arcing over the generator bearings, the generator will fail. Newer designs use permanent-magnet rotors to eliminate these stray currents and prevent failure.
Although placing an inverter inside a wind turbine’s nacelle (the hub housing, where the generator resides) is the rule in the rest of the world, in the U.S., General Electric has a patent on this technology (when used as a doubly fed induction generator to meet low-voltage ride-through requirements) that won’t expire until 2010. As a result, other manufacturers and developers have had to resort to various volt-amperes reactive (VAR) compensation schemes, some of which are company secrets. However, the patent restriction is not too limiting, because at wind farms with capacities larger than 20 MW it is more economic to collect all the output from the wind turbines and provide VAR support at one location.
A standard way to provide VAR support outside the turbine is to dynamically compensate for reactive power using static VAR compensators (SVCs). SVCs comprise parallel banks of capacitors and a reactor, some or all of which are controlled by thyristors. Their control circuitry performs a number of functions, such as determining the best time to switch in the capacitors to avoid unnecessary voltage stress on the system.
Using SVCs for VAR support costs about half as much as using inverters. However, SVCs operate more slowly than inverters (in one cycle, vs. ¼ cycle), and not over the full voltage range needed for uniform VAR support. What’s more, SVCs have substantially less low-voltage ride-through capability—not enough to meet FERC Order 661. In addition, any SVC installation needs to be thoroughly analyzed for possible harmonic resonance in the system. This analysis is made more complicated because SVCs modify the resonance frequencies of a system, depending how the SVC is operated.
Consider a typical 100-MW installation that requires 30 MVAR for dynamic stability. Because the MVARs are only required for dynamic support, the overload capability of the power inverters can be used to meet this criterion. For example, DSTATCOM inverters from American Superconductor (www.amsuper.com) are designed to handle 2.6 times the rated VAR output for 2 seconds. Twelve MVARs of DSTATCOMs meet the dynamic requirement, and another 38 MVAR of capacitor banks will be required to meet the power factor requirements.
Email
Comments
Print
Reuse
