The Port Alma and Chatham Kent wind farms in Ontario have been monitored by ONYX InSight for many years. There are 88 Siemens SWT2.3 wind turbines between the two sites and there has been no history of bearing failures in any of the 88 generators, until now.
This very interesting case was the “Tale of Two Bearings” (until we caught a third one).
It all started when Turbine T30 was flagged by our monitoring engineers with a strong indication of an outer race fault on the drive end (DE) generator bearing.
Figure 1. Turbine T30 DE Generator Bearing Outer Race Fault Frequency
Because this was the first of this type of failure at an older site, there wasn’t a rush to replace it. When the repair finally took place, the failure was very progressed. You can see the evidence of this progression in the images below after the fan and end cover were removed.
Figure 2. Photos of Turbine T30 Generator Drive End
The ONYX standard for any new failure mode is to perform a deep dive into the data with fleetMONITOR, and then to set up an array of automated analysis methods to identify the fault as early as possible. At ONYX, we strive for 6-18 months warning on all bearing faults. When the monitoring engineers have implemented a number of effective methods to model the failure in fleetMONITOR, all historical data is reprocessed, and alarms are generated with new health trends, based on all of the methods now available in the software. In this case, we immediately picked up an early stage fault trending on a generator bearing on Turbine T01, this time from the non-drive end (NDE).
Figure 3. Turbine T01 Non-Drive End Generator Bearing Health Index
In the case of T30, the failed generator bearing required complete generator replacement. This major component exchange was unplanned and caused extended downtime and additional costs. With these new lessons learned, the site manager was quick to arrange an up-tower bearing repair on T01 and shipped the parts to ONYX in Boulder, CO for dis-assembly and inspection. The disassembly process was quick at the ONYX shop where we drilled out the cage, compressed the race, and disassembled the bearing into its components.
Figure 4. T01 Generator Bearing Disassembly at the ONYX shop in Boulder, CO
After disassembly, the inspection showed clear evidence of electrical fluting (see our failureATLAS: https://goo.gl/grtMYz). The fluting can be seen clearly in the correct lighting, as seen shown in the photos below.
Figure 5. Visual Evidence of Electrical Fluting Damage on the Generator Bearing From T01
With lessons learned from T30, this damage was now very simple to detect with vibration analysis in fleetMONITOR, but the physical defect can only be felt by running your fingernail over the fluting marks. Could the bearing have been run longer? Possibly. By applying our empirical model, Vibration Based Life, to more and more of these failures, we are building the answers to that question.
This was the conclusion to “A Tale of Two Bearings”, but then there was a third….
Figure 6. T61 NDE Generator Health Index in fleetMONITOR
Once the faulty bearing in T61 showed a consistent excursion above the yellow alarm threshold, the site was notified. This allowed management to schedule an up-tower repair as a planned maintenance action, conveniently scheduled and at the best possible cost.
The figure below shows how the earliest stages of damage have once again been detected; had this been a less convenient time of year for repair, the bearing could have run for many months while under close observation of vibration and temperature.
Figure 7. Visual Evidence of Damage on the Generator Bearing From T61