It is important that SPS continue to provide industry benchmarks leading up to, and during the annual conference of the Western Turbine Users Inc, Salvatore A DellaVilla Jr, CEO of Strategic Power Systems® (SPS), told the editors at his company’s booth on the exhibition floor in Las Vegas, Mar 19, 2017. “Participation by owner/operators of LM engines in the ORAP® program allows us to aggregate data and provide meaningful analytics to the group,” he said.
“Our effort to provide the high-quality information required for decision-making demands automated collection of data from the control system,” DellaVilla continued.
“We bring this to your attention because our engineers and analytics team are seeing more and more issues related to the manual submittal of data. For example, you may have noticed that we do not provide any metrics for starting reliability. Our experts are concerned that the information submitted to SPS regarding starts is increasingly inaccurate and we are not confident in the starting reliability data to provide it as an industry benchmark.”
As stated in IEEE Standard 762, the document SPS uses to guide the processing of ORAP data, “a starting failure is an unsuccessful attempt to bring a unit from shutdown to the in-service state within a specified period (which may be different for individual units). Repeated initiations of the starting sequence without accomplishing corrective repairs are counted as a single attempt.”
Simply, if the unit does not reach breaker closure in a specified period of time, it is considered a failure to start.
Additionally, SPS is questioning much of the information reported to it from NERC GADS. The company says it continually sees issues in the fidelity of the data as that information relates to causes of downtime and duration.
“Our engineers have conducted a data-quality comparison referencing one of our long-time customers that recently moved from submitting data directly to us versus submitting NERC GADS data,” DellaVilla said.
“We found NERC GADS data are inherently high-level and do not have the same granularity of detail that ORAP requires. Plus, NERC GADS does not require users to identify component-level root causes to forced-outage events, a detail that creates issues when trying to compare this customer to the rest of the fleet.
“These data also raise concerns with the manufacturer. Periodically, we conduct quality reviews with the OEMs. During these reviews we often are questioned on the accuracy of events that have been submitted from NERC GADS reports. All the issues identified above make it difficult to use NERC GADS data to allow meaningful and accurate comparison with the rest of the fleet in ORAP.”
The easiest way to remedy these issues Team SPS says is by use of automated data collection from the control system, as DellaVilla had mentioned earlier in the conversation. By automating the data, the system records each mission, from startup to shutdown—including all major states from signal to start, through the permissives, to ignition, flame established, acceleration, breaker closure, through each change in load state, to shutdown, and then the cool-down period. This is the only way to eliminate human error and ambiguity and ensure data accuracy.
However, SPS still does need input from plant maintenance staffs regarding the symptoms, corrective actions, and eventually the root cause of outages to ensure that the full scope of the event is captured correctly. There always will be a human element to this reporting.
That said, SPS had prepared the latest RAM KPIs from ORAP for the user-group meeting (Table 1). The data were reviewed thoroughly and analyzed for accuracy. The information compiled in the table comes from 607 aero units for 2016 and 1,092 units for the 2011-2015 period. The aeroderivative gas turbines in the sample include engines from GE, P&W, and Siemens AGT (formerly Rolls-Royce) and represent units operating worldwide.
There was a minimal increase in annual operating (service) hours for peaking units from the 2011-2015 period to 2016; availability decreased by about 1.1% for 2016 and reliability was pretty consistent within the two time periods. Cycling units operated 163 hours less in 2016 than they averaged in 2011-2015; availability stayed exactly the same, while reliability decreased slightly (0.1%). Baseload units operated 176 less hours in 2016 versus 2011-2015 and annual starts decreased.
The regional analysis in Table 2 shows capacity factor was down by 6.9% in the West, but showed an increase of 10.5% and 10.3% in the Midwest and Northeast, respectively. Another interesting thing to note is that all regions with the exception of the West, had a reduction in reserve standby factor.
