Wind turbines are engineered to efficiently capture the kinetic energy of the wind, convert it to electrical energy and deliver it to the main power grid. For example, at the Jersey-Atlantic Wind Farm in Atlantic City, New Jersey, each wind turbine is equipped with three 120-foot rotor blades attached to a hub, a horizontal driveshaft and a nacelle similar in size to a city bus atop a 262-foot tower. Wind turns the blades and the driveshaft at speeds between 10 and 20 revolutions per minute.

Inside the nacelle, the driveshaft attaches to a gearbox that turns a second shaft at around 1,800 revolutions per minute, powering a generator that produces electricity for transmission to the grid. These horizontal shafts are supported by radial bearings, which are designed to bear loads perpendicular to the shaft. Each bearing resembles a pair of concentric rings separated by balls, or cylindrical or tapered rollers. The inner ring fits around a shaft and allows it to rotate while the outer ring remains stationary.

The gearbox accounts for approximately 13 percent of the cost of a wind turbine, according to the HTL Group, which provides engineering solutions to the wind turbine industry. The price tag for five wind turbines at the Jersey-Atlantic Wind Farm was $12.5 million in 2005.

Their cost notwithstanding, gearboxes are the Achilles heel of wind turbines. A quarter of gearboxes and generators need to be replaced within 10 years of commencing operation, according to a 2018 IHS Markit report cited by New Energy Update. Downtime to repair a gearbox disrupts electricity generation for days or even months, depending on availability of parts.

Steel bearings typically fail due to rolling contact fatigue, or cyclic variations in stress as the bearing turns while supporting a load. This phenomenon causes cracking, flaking and spalling, or small pieces breaking off of the bearing surface.

The HTL Group believes that the leading cause of premature gearbox failure is the unexpected formation of cracks in bearings that support shafts. The root cause is thought to be bearing damage due to variable loading and rollers skidding as a result of sudden relocation of the loaded zone during startup, shutdown, emergency stops and other transient events. Moisture and dirt that can contaminate the lubricant may also contribute to bearing failure.

Premature formation of cracks on the surface of shaft bearings in wind turbines can occur within the first one to three years of operation, or less than 10 percent of the expected service life, according to David Vaes of SKF, speaking at the companys 2016 Wind Turbine Management Conference. He reported that each case of premature failure was unique, independent of metallurgy and caused by one of several possible factors: accelerated fatigue due to higher than anticipated stresses such as tensile stress, shock loads or vibrations; lower than expected steel strength; and environmental factors such as lubrication, corrosion, electric currents and mixed regime friction.

Vaes reported that when bearings failed due to these reasons in benchtop tests, microscopic white etching areas formed. These now-infamous white etching cracks were observed more frequently in bearings undergoing premature failure and appeared to be a symptom rather than a cause of bearing failure.

Looking Inside Steel

Argonnes Gould explained that white etching cracks are localized, sub-surface microstructural alterations on an atomic scale that many researchers have observed within bearings that failed prematurely in wind turbines and other industrial applications.

WECs seem to be a universal issue, as they are observed in bearings made with various steel alloys, heat treatments and surface coatings that are lubricated with different formulations and additive packages. Gould emphasized that every example of WECs in laboratory tests had been produced by one or more of the factors noted above. It was unclear, he said, whether sub-surface WECs were in fact a cause, a symptom or a consequence of bearing failure.

To investigate, Gould applied one of the most powerful and largest instruments in the world: the Beamline 2-BM-A Micro Tomography Imaging unit, part of the Advanced Photon Source, a U.S. Department of Energy Office of Science user facility at Argonne National Lab. Tomography is a non-destructive form of spectroscopy that uses X-rays to penetrate solid objects and evaluate their structure. Computer programs are used to assemble the data into a set of cross-sectional images to visualize the interior of the object.

Gould used X-ray tomography to compare the micro­structure of samples of AISI 52100 steel from actual wind turbine gearbox bearings versus standard bench test specimens supplied for testing lubricants. Then, he used conventional metallographic methods, including sectioning, etching, polishing and optical microscopy, to examine the samples.

In this study, Gould took steel samples from a bearing pulled from the field due to failure. He cut sections of steel from the interior of part of the bearing that had not been loaded during use.

There were significant differences in microstructure between the two types of the 400-micron steel cubes. Gould documented 529 atomic-scale defects-inclusions or voids-in the wind turbine cube versus only 129 in the standard specimen cube. The average of the areas of the ten largest inclusions in the gearbox piece was three times larger than in the standard specimen: 210.6 versus 72.1 square microns, respectively.

Goulds metallurgical analysis revealed that many of the defects in the wind turbine steel were dual-phase, non-metallic inclusions that contained particles rich in manganese, magnesium and sulfur, or aluminum and oxygen. Non-metallic inclusions form due to contamination and reactions during steelmaking. They disrupt the homogeneity of the steel microstructure and contribute to cracking and fatigue failure.

WECs tended to initiate in similar inclusions in Goulds other laboratory experiments with steel from wind turbine gearboxes. However, none of these inclusions were found in over one hundred samples of the standard sample steel. Both had comparable hardness and surface roughness values.

Gould concluded that these inclusions predispose AISI 52100 steel bearings to the formation of WECs.

I was very surprised by how different the steel in our [sample] specimens was from actual [wind turbine gearbox] steel, Gould told LubesnGreases. There are plenty of reasons. Firstly, the sources of the steel could be drastically different. Secondly, the manufacturing processes that are used to make the bearings can lead to differences in inclusion microstructure.

Yes, there are plenty of ways to get rid of dual-phase inclusions, but there are significant costs associated with these added manufacturing steps. For applications like wind turbines, where cost of energy is the bottom line, many of these techniques are rarely applied, he explained.

This finding has broad impacts for anyone trying to mimic large-scale component failure within small-scale tests, especially for fatigue based failure modes.

Lubricant Formulation

Gould used a laboratory-scale micro-pitting rig to study the effects of load, steel alloy and lubricant on bearing failure. The rig used three rings in a triangular arrangement to apply loads to a small cylindrical roller specimen at the center. A computer controlled the speed of rotation, load, lubricant temperature and other parameters. Rollers were tested for 300 million cycles or until failure.

First, ISO viscosity grade 150 commercial wind turbine gear oil was used in an experiment to compare rollers made from field and sample steels. Gould observed many long, branched WECs that had initiated at inclusions in the gearbox steel after it was tested at a higher load of 2.5 gigapascals of contact stress. Early-stage WECs, called butterflies, were present in rollers tested at a lower load of 1.9 GPa. This marked the first documented case of the formation of WECs in an accelerated benchtop test using fully-formulated wind turbine gearbox oils.

After sample rollers were tested with the same gear oil and conditions, Gould observed no WECs, although he saw typical surface failure due to rolling contact fatigue.

These results supported Goulds conclusion that inclusions present in the steel are part of the cause of bearing failure in wind turbine gearbox bearings and that formation of WECs is part of the failure mechanism.

Gould repeated this experiment with heavier commercial wind turbine gear oils.

With ISO VG 320 gear oil, he observed WECs in gearbox rollers that had been tested at the heavier load (2.5 GPa), similar to those observed with ISO VG 150 gear oil, and butterflies at the lighter load (1.9 GPa). Only surface failure occurred in sample rollers at both loads.

For the ISO VG 220 gear oil, butterflies were present in the gearbox steel at both loads. For the sample rollers, there was only typical surface failure and no sub-surface cracking. The average friction coefficient was higher and statistically significant in the tests where WECs formed.

Gould concluded that while a specific type of inclusion facilitates the formation of WECs, the choice of lubricant likely has an effect on the rate of WEC formation.

Even though this work shows how steel microstructure drives WECs, there are still plenty of things lubricant formulators can do to mitigate the damage to bearings, Gould said. We found a correlation between friction coefficient and WEC formation. Therefore, lubricants with lower friction may help lessen WECs.

Our current work focuses on the effect of electrical current on the formation of WECs. We have formed WECs with numerous fully formulated wind turbine lubricants, and we believe that there is a strong correlation between lubricant additive species and WEC formation, he concluded.

Mary Moon, Ph.D., is a professional chemist, consultant and technical writer and is technical editor of The NLGI Spokesman. Contact her at mmmoon@ix.netcom.com or 267-567-7234.