Improving Performance and Reliability with Condition Monitoring

Power plant operation is a bottom line driven business. Once an asset is built and providing power, every dollar spent on operation and maintenance affects that bottom line. Improvements to the plant reliability and performance can be made, but only with sufficent cost justitifcation to demonstrate return on investment. For condition monitoring (CMS), one factor is the known reliability of the particular wind farm, for example, take three onshore wind farms with MW class turbines with three point mounted drivetrain turbines with double spherical roller main bearings (DSRB). This type of turbine arrangement is known to have its issues, with failure of the downwind raceways and rollers. In this article we use a case study using site data to examine the bottom line. First from Weibull curves (a well known reliability analysis method) from several United States wind farms we have the data in Table 1. Note that main bearing failures are dependent on site conditions, bearing type, maintenance practices, etc. – this data is for problematic sites.

From these three farms we can construct a typical US wind farm with this issue for studying costs. The farm has 80 turbines, the size of turbine is approximately. 2MW and over ten years the average number of main bearing failures per year is 2.91 (based on the average data from Table 1), PPA is $55/MWhr, capacity factor is 0.35 and availability sits at 97% (excluding downtime from any main bearing failure). Each wind turbine brings $900 per day in revenue.

A question for the asset manager is: will condition monitoring improve the bottom line for main bearing issues on this wind farm? We paint a few scenarios in Table 2. First compare Scenario A and B, both with no condition monitoring, in each case the bearing fails and little warning was received. Only when the bearing availability is scarce (Scenario B), with three months lead time for repair, is the downtime a significant part of the cost. The cost is largely driven by crane mobilization and repair or replacement costs (costs quoted are typical but vary from site to site and are based on a refurbished bearing). Compare to Scenario C, the CMS system gives early warning of the failure, yet there are still separate crane mobilizations. Savings thanks to CMS are in the order of $40K where bearings are readily available and up to $300K when a lot of downtime is prevented (Scenario C versus A and B). More savings made when maintenance is scheduled more effectively due to the CMS (Scenario D). Given the early warning on the bearings (and it takes months for the DSRB failure mode to progress) repairs for the two failures are now scheduled for the same week, the crane is only mobilized once and savings are now in the order of $150-400K, depending on downtime saved (Scenario D versus A and B). Here the CMS system has added value with payback for retrofit installation and monitoring over three to four years, without considering potential benefits to the gearbox and generator. For different turbines and sites costs vary and where a known reliability issue exists, ROI calculations for CMS systems can be completed to evaluate benefits.

With the slow rotation it is often said that a damaged main bearing is difficult to detect with vibration; this is true where a standard envelope spectra or vibration RMS may not discern the damage from other vibration sources. However with the right techniques the condition monitoring is very effective for the main bearing and the progression of damage can be monitored over three to six months while the bearing deteriorates, finally when the temperature rises the bearing condition is very poor and replacement within a few weeks is recommended. Figure 1 provides an example where a low frequency accelerometer recorded time series data which was then processed in the Romax InSight iDS software for early detection of main bearing failure. This typical example illustrates how no special hardware is needed for main bearings, just good signal processing. By point D in the diagram the condition of the bearings is very poor, typical to a set (Figure 2) inspected in a factory teardown.

SCADA Monitoring

Aside from condition monitoring (via accelerometers, oil particle counters and so on) the turbine SCADA system is a valuable source of data for monitoring the turbine. From the 100’s of channels of data coming in (typically at a 10s sample rate) there are sets of parameters useful for checking the health of machine and also the performance. In many cases they are related, e.g. a high yaw misalignment will reduce performance, but also effect the health of the turbine longer term, as the rotor bending moments are increased and accelerate structural fatigue for all major components in the load path: pitch bearing, rotor hub, main bearing, main shaft, main frame, yaw bearing and depending on the drivetrain architecture also the gearbox. Analyzing SCADA data together with the maintenance records is also an effective way for the wind farm operator to monitor and control the maintenance quality.

Figure 3 shows a bearing temperature histogram from a European offshore wind farm analyzed in the Romax InSight SCADA software. After replacement of the high speed shaft bearing, the temperature histogram for the following months shifted to a higher temperature range with maximums up to 85°C. Unfortunately, this temperature shift was not identified at that time. As a result, the bearing had been running under this unhealthy condition and failed the gearbox within 12 months. Due to access and availability issues offshore, the gearbox wasn’t replaced until seven months later and the cost of downtime and repair was in excess of $1M.

Figure 4 provides the SCADA alarm history for another offshore wind farm. Here Turbine 78 has a large number of generator bearing temperature alarms and shutdowns. The total downtime for this machine from this one fault accumulated to 780 hours (an 8.9% availability reduction) over 12 months. This many production hours is around $75k at the site but as the alarms were more regular during high wind speeds, the real production loss was nearly $150k. In order to reduce the downtime, a reactive generator bearing replacement was carried out in September. However, the high temperature alarms continued. Another generator bearing replacement program was carried out two months later. The alarm problem still wasn’t solved after the second bearing exchange program, and with the issue prevalent on other machines at the site, an investigation was carried out by Romax to solve the problem. The replaced bearings were examined and a root cause analysis was conducted. This RCA determined the alarm wasn’t triggered by faulty bearings, but due to other factors. The misdiagnosis not only caused production loss, but also incurred unnecessary maintenance cost, with totals nearing $500k. Romax InSight SCADA brings together maintenance records, inspection reports and analysis reports with SCADA data to provide a powerful tool for assessing the site. This allows for a clear visualization (Insight) so the site manager or control center operators can easily see such correlations and make more informed decisions in operations and maintenance. Figure 5