On 15 July 2013, the gearbox of an 11-year-old wind turbine failed while it was running. The High Speed Shaft (HSS) bearings were determined to be the source of the problem – especially HSS‑B and ‑C which are its two downwind bearings, nearest the generator. This blog briefly presents the history of the failed high speed bearings, correlates these data, and briefly discusses likely causes of this failure. A more complete white paper is also available. It includes oil particle analyses, borescope inspections, vibration analyses, SCADA data, and photographs of them which were taken after they were removed.
The turbine discussed here is a GE 1.5s wind turbine with its original circa-2002 Eickhoff gearbox. There are three bearings on the High Speed Shaft (HSS) [figure 3]. HSS‑A and HSS‑B are cylindrical rolling element bearings. They are not “locating bearings,” meaning that they are designed to sustain only radial force, not axial force. (The figure shows a slightly different arrangement than is used in this gearbox, but the basics are the same.) Together, all three of these bearings support and locate the high speed shaft.
These high speed bearings are thought to be original to the gearbox.
Although this wind turbine gearbox operates in a desert and the desiccant breather looks to have been properly maintained, the water concentration was sometimes as high as 820 ppm. This may have been the saturation point for this fluid – which is not good. Studies have shown that concentrations higher than 100 ppm can shorten bearing life.
Oil sample analyses also often showed particle counts near the upper recommended limits for ISO cleanliness. Undesirable amounts of ferrous particles were found in the oil. Exactly where did these particles come from?
In December 2012, a borescope inspection of this gearbox was performed. Among the findings, medium to heavy levels of damage were detected on two of the three HSS bearings. Heavy micropitting was seen on the rollers and inner race of HSS‑B. A 2010 borescope inspection had identified only light micropitting on this bearing – quite a change in just two years.
As with most gearboxes of similar design, HSS‑C is inaccessible without partially disassembling the gearbox. Therefore, there is no borescope data from this bearing. Vibration is the only way to diagnose this bearing without partially disassembling the gearbox.
On 18 December 2012, this turbine’s drivetrain was permanently instrumented with UpWind Sentinel™, a permanently-installed vibration monitoring system. From its data, significant damage was observed on its rolling elements. On 28 June 2013, vibration signals indicated that HSS‑C was entering bearing failure stage 4 (of 4). Failure was imminent. Vibration data from that same date indicated that HSS‑B was already in its death throes: failure stage 4.
UpWind Engineering obtained and photographed all three failed high speed bearings (HSS‑A, ‑B, and ‑C) after the gearbox failure. The full white paper presents detailed photographs of these, bearing by bearing, and discusses those findings in some detail. This is a brief synopsis.
This bearing still had some life left in it, but it was on its way to failure.
The bearing showed obvious, heavy macropitting and significant micropitting. Many macropits had grown large enough to merge together.
The white paper mentioned above “unwraps” the outer race to show various forms of damage in the load zone (the area which supports most of the forces) and other zones of HSS‑B. The findings in the load zone included macropits and a deep, wide groove. Clearly, the clearances in this bearing were no longer what they were designed to be. This is a situation to be avoided.
Upon glancing at this bearing, a crack and liberated material (missing chunk) in the outer race was immediately obvious. The balls showed heavy macropitting, some flaking, and notable plastic deformation – they were no longer round. (The white paper illustrates and discusses this “cornucopia of failure modes.”)
It was noted that the vibration spectra presented show no classic vibration signals from HSS‑B: No inner race, outer race, nor ball spin frequencies can be found. This is because HSS‑B had entered into bearing failure stage 4 (of 4) before these vibration data were acquired. This is the final stage of bearing failure.
It is likely that the increased clearances in HSS‑B – it inability to carry the load it was designed for – lead to the failure of its neighbor, HSS‑C, when some of HSS‑B’s radial load was transferred to it. Four-point contact bearings like HSS‑C are not designed to carry much radial load. Between 10 and 28 June, HSS‑C also entered stage 4 bearing failure, a period which is usually relatively brief. There is no stage 5.
Prior to turbine shutdown, the gearbox “bearing A” temperature had been steadily rising, likely indicating a fault. At ~4 minutes prior before shutdown, HSS‑B and/or C probably suffered major failure. This caused significant, localized heating, as recorded in the CBM data. The turbine shut itself down based on just one signal: “bearing A” temperature. It’s scary to think what might have happened if that signal had malfunctioned!
Exactly what caused this gearbox’s high speed bearings to be damaged in the first place is unknown. Here are some potential causes, some of which have already been discussed:
- Wind conditions
- Design issue
- Torque reversals
- Other (other causes which are common to wind turbine generators, and especially to their HSS bearings)
The decision was made by the owners to run the high speed bearings to failure. This is sometimes an acceptable decision, since one must balance production with this risk – but the choice should be an informed decision. When operating a wind turbine farm, it is important to make maintenance and repair decisions based upon good data. In this particular case, it was determined that run-to-failure was the best option.
The high speed bearings’ failure was investigated analytically with oil analyses, visually with a borescope inspection, and electronically using SCADA data and vibration monitoring. Some of these potential causes are beyond the ability of the operator to address – but others can be mitigated.
In this gearbox, HSS‑C, one of the two bearings that failed, is not accessible via borescope inspection. The only way to monitor its health in situ is using temperature, oil, or vibration data:
- Oil analysis, when it does identify potential bearing problems, is not specific to any particular bearing in the gearbox.
- Borescope inspection would not have revealed the significant damage
- Bearing temperature data provided very, very little warning time. By the time the turbine faulted on that very bearing temperature and shut itself down, HSS‑C was in late failure stage 4. There is no stage 5. Had this single warning (the “bearing A” temperature) failed to provide warning, however late, the turbine may have failed catastrophically just minutes later – a significant risk.
In this case, inferences about the likely condition of HSS‑C could have been made from the known, bad condition of HSS‑B. However, short of gearbox disassembly, vibration data was the only useful way to positively monitor the health of HSS‑C.