Several makes of gearbox in several makes of turbine have been hit by the problems. The root cause, according to the Risø study, may be excessive wear within the planetary bearings at the front end of the gearbox, which are subjected to a highly complex system of loads. The problem came to light already in 1998 when extraordinary wear and severe pitting was discovered on the teeth of gearwheels within gearboxes that were only two years old or less (Windpower Monthly, December 1998).
Risø points out that the planetary bearings are subject to gravitational loads, fluctuating loads in the horizontal plane (especially when the rotor is yawed), a fluctuating thrust from the rotor, and a fluctuating torque. The weight of the gearbox assembly is sufficient to cause a slight deformation of the shaft, resulting in small misalignments in the bearing assembly, according to Risø's theory, which is outlined in a recently published technical report. The complex fluctuating forces, together with possible misalignments can, under certain circumstances, mean that some of the rollers become unloaded. Due to the effects of gravity and vibrations the bearings become wedged in such a way that they can no longer compensate for the misalignment.
The enormous inertia in the rotor means that it does not stop, despite the rollers having effectively seized, continues the report. At the same time, additional forces are imposed upon the bearing, which, in turn, lead to additional wear of the rollers and on the cage within which they rotate. Once wear has started to occur, it can progress rapidly, leading to failure.
The Risø report adds that the "locking" or "wedging" problem is much more likely to occur at low rotational speeds, or after the rotor has stopped -- and that the problem is accelerated by gravity forces due to the tilt of the shaft. This may be why turbines, rather than other machinery, are apparently particularly susceptible to a problem not encountered on such a huge scale by the bearing industry until now.
The report authors, Flemming Rasmussen, Kenneth Thomsen and Torben Juul Larsen, suggest two possible solutions to remedy the situation: either introducing some restraints so that the rollers cannot fall into the locked position; or the use of cylindrical roller bearings that may be less susceptible to the problem than spherical bearings, although they would not be able to compensate for the misalignment. Rasmussen says that Risø has devised a bearing that it believes would not seize up in the same way, for which it has already applied for a patent.
The report makes no judgement on where responsibility for the need to make hugely costly retrofits should lie, but implicitly throws three key questions into relief. Was a correct description of the full loads and forces passed on to the gear box manufacturer by wind turbine designers? Did the gear box manufacturers pass on the necessary information to the suppliers of the bearings? Do the spherical roller bearings in question have an inbuilt design fault that would always be revealed by a specific combination of operating conditions?