Double feeding slashes inverter size

Dramatic improvements can now be made in the cost effectiveness of both wind turbines and variable speed drives, reports Tom Shelley

Variable speed brushless generators and motors can be made to work with inverters - about one third the size normally required - if they have two stators with different numbers of poles, fed and working at two different frequencies. One set of stator windings is directly attached to the mains supply and the other to the inverter. The basic idea has been around for more than a century, but it is only with growing use of wind turbines and variable speed drives, coupled with the decreasing cost of power electronics and advances in modelling, that the technology has had an opportunity to come into its own. The machines are called Brushless Doubly Fed induction Machines (BDFMs) and are being developed primarily, but not exclusively, as generators for wind turbines by Wind Technologies, a company spun out of the University of Cambridge Department of Engineering. Wind turbine generators have to work with inverters, if they are going to feed in to mains power supplies, because wind and wind generator speed is variable, yet mains power is associated with a fixed frequency and voltage. With doubly fed machines, one winding is connected to the mains, so most of the generated power is fed directly in to the mains. The inverter associated with the second stator winding only has to handle part of the total power, down to as little as one third. The idea is already being successfully applied to slip ring AC generators, but applying it to brushless generators can be expected to double the mean time between failures and greatly reduce maintenance costs. Any technology that allows inverters to be made at one-third their usual size and avoids the need for brushes has to be immensely advantageous. The origins of the base idea for doubly fed induction machines goes back 1902 when the Siemens Brothers and a Mr F Lydall were granted British Patent 16839 to protect the idea of connecting together two induction motors with different numbers of poles. Their machine had two stator windings and two rotor windings, each brought out on slip rings. The next big advances were made by another Briton, LJ Hunt, who came up with the idea of doing away with slip rings and having a single rotor winding, but two stators. He developed his machines between 1907-1914 and it appears they had a limited commercial life. No further significant advances seem to have been made until the 1980s, when some serious US government funded research was undertaken at the University of Oregon. The results were patented, but have now lapsed. Research in the UK was restarted by Professor Stephen Williamson at Cambridge in the 1990s and, since 1999, is being undertaken by a team led by Dr Richard McMahon. "We started our research with a 2kW machine and Laurence, Scott & Electromotors in Norwich then donated a 10kW machine," recalls co-researcher Dr Ehsan Abdi. "Several masters and doctoral students have worked on the project, but significant progress was not made until the last three or four years. Only now can we design with confidence. The rotor is similar to a standard cage rotor, in that the bars are shorted at the ends, but they are not quite the same." The stators couple to each other via the rotor, and rotor speed = (?1 ± ?2)/(p1 ± p2) where ?1 is mains frequency and ?2 is that delivered to the inverter in the case of a generator, or received from the inverter in the case of a motor. "Drive applications have huge potential, but at the moment the company is focusing on large wind turbines", he adds. Development has been assisted since 2003 by use of Bluetooth wireless instrumentation on the rotors to measure rotor bar currents. CSR donated the original modules. The team is presently concentrating on improving stability and control of the machine over a range of speeds. Matlab and Simulink from MathWorks are used for modelling and control. The test rig has the 10kW machine coupled to a DC machine that can be used either as a motor - in which case the BDFM acts as a generator - or as a generator itself. The inverter associated with the second stator winding of the machine uses IGBTs (Insulated Gate Bipolar Transistors). The machine can be run at any speed from 0 to 1,500 rpm. "It has a special torque speed characteristic, not similar to an induction machine," adds Abdi. A PC is used to generate the PWM signals in the inverter. Researcher Shiyi Shao comments: "I want to make the dynamic performance better. It is presently possible to speed up 10rpm in one second. We are working on improving performance, both as a generator and a motor. We need to control the load angle very accurately." Failure to do so makes the machine liable to loss of synchronism - crucial to the efficient running of any brushless machine that requires switching electromagnet coils in response to rotor position - and it will then become unstable. The project received the Scientific Instrument Makers Award and the Cambridge University Entrepreneurs Business Idea Award in 2004. In 2005, the Institution of Electrical Engineers, now the Institution of Engineering and Technology, added its Innovation in Engineering Award. The next stage, says Abdi, would be to establish a 20kW to 30kW wind turbine generator, mounted and monitored on the West Cambridge site, which is where the laboratory has recently been relocated. The stage after that would be to build a 600kW generator, retrofit it to a wind turbine in a wind farm in Germany and test it there. Pointers * Brushless induction generators with two stators and different numbers of poles can produce mains frequency output by combining the output of one set of windings with the inverted output of the other. The net result is that the inverter need only be one-third the size of one required to invert full current from a single stator * The method also works in reverse, allowing a variable speed drive to power a machine that is power rated up to three times that of the inverter * Although the base idea is more than a century old, it is still being perfected. At present, it is not possible to cope satisfactorily with very rapid speed changes