Owners and operators of GE aeroderivative gas turbines should make every effort to participate in meetings of the Western Turbine Users Inc. The information you gain access to and absorb at this conference is critical to the success of your company’s asset management program.
A particularly valuable service provided annually to user attendees by Strategic Power Systems Inc (SPS), Charlotte, is key performance indicators for the LM2500, LM5000, LM6000, and LMS100, as well as a review of the predominant causes of outages experienced by those machines the previous year.
SPS’s information “engine” is the company’s respected Operational Reliability Analysis Program (ORAP®). It gathers raw data from gas turbines as they operate and makes it more valuable by adding information from users on incident/outage details. SPS engineers crunch and analyze the stored material to produce insightful presentations for breakout sessions focusing on the individual gas turbines.
What follows is a summary of the four presentations made at last spring’s conference. The information is invaluable for those trying to identify where to look for potential problems before the gremlins rear their ugly heads. The last thing you want to do is to show up at next year’s meeting and have to say, “Oh, that happened to us as well,” when it could have been avoided.
The LM2500 is the most versatile engine in GE’s land and marine aero portfolio. The machine has been uprated and improved several times since its commercial introduction at the dawn of the 1970s and has racked up more than 70-million operating hours over the years. You can find the LM2500 in utility/IPP peaking, cogeneration, and combined-cycle plants, as well as in trailer-mounted emergency/standby packages, drilling-platform service, mechanical-drive applications (gas pipeline compressor drivers, for example), industrial combined heat and power, and marine main propulsion systems.
Gas fuel control and regulating valves were identified as the leading cause of forced outages among the more than 100 LM2500 engines in electric generation service that provided operating information to Strategic Power Systems Inc’s ORAP® database in 2013. Controls, controllers, and communication issues were second on the Top Ten list of forced-outage causes.
SPS’s Tom Christiansen and Brian Freeman presented on LM2500 performance at the Western Turbine meeting last spring. John Baker, plant manager, Clearwater Cogeneration Plant, Riverside Public Utilities, chairs the LM2500 sessions.
In round numbers, this fleet has a thousand engines dedicated to electricity production. Slightly more than half are base machines with single annular combustors (SAC), another 10% or so are base engines with DLE combustors. LM2500+ gas turbines account for about 25% of the fleet total; about a third of them have DLE combustors. LM2500+G4s represent less than 10% of the fleet and are split 50/50 SAC/DLE.
Christiansen and Freeman began their prepared remarks with a summary of LM2500 ORAP stats for the five-year period extending from Jan 1, 2009 through Dec 31, 2013. The information was presented by service duty: Peaking machines (service factor of less than 10%) compiled 75 unit-years of operation; cycling engines (service factor from 10% up to 50%), 154 unit-years; base-load units (service factor of 50% and above, 397 unit-years.
Highlights of the SPS analysis for peaking machines revealed the following:
- Fleet availability (simple-cycle plant) averaged 95.3% annually over the period, with the 2013 number an enviable 99.4%.
- Forced-outage factor, 0.4% in 2013, was much better than the annual 4.28% average from 2009 through 2013.
- Maintenance-outage factor, 0% in 2013; five-year average was 0.6% per annum.
- Planned-outage factor, 3.4% in 2013, was one percentage point above the annual average for the period.
- Service factor (the percentage of time units generated power), 2.3% in 2013, slightly below the annual average of 3%.
For cycling engines, the numbers were:
- Fleet availability, 94.9% in 2013, slightly higher than the 93.5% period average.
- Forced-outage factor, 3.2% in 2013, a tenth of a point above average.
- Maintenance-outage factor, 0.2% versus an average of 0.9%.
- Planned-outage factor, 1.8% compared to the five-year average of 2.4%.
- Service factor, 24.4% in 2013, significantly below the 30.9% average.
Results for base-load machines were well received with the 2013 numbers beating the annual averages in all categories:
- Fleet availability, 96.3% for 2013 versus 95.6% annual average from 2009 through 2013.
- Forced-outage factor, 1.1% versus 1.2%.
- Maintenance-outage factor, 0.5% versus 0.6%.
- Planned-outage factor, 2.1% versus 2.7%.
- Service factor, 89.8% versus 87.3%.
Engine removals, both scheduled and forced, also were reported by Christiansen and Freeman. In the five-year period, LM2500s reporting data to SPS accumulated about 3.5-million service hours and experienced 27 forced and 79 scheduled removal events. Note that no peaking unit forced out of service during the period required a shop visit, and no peaker was sent to the depot for a scheduled overhaul from 2009 through 2013. Cycling units performed almost as well, experiencing only one forced removal event; there were five scheduled removals. The mean time between forced outages of base-load engines was more than 116,000 hours.
Analysis of forced-outage incidents
Next, the speakers presented details on the Top Ten causes of forced-outage incidents recorded from Jan 1 through Dec 31, 2013, giving both numbers of occurrences and their duration in hours. In sum, 477 total incidents were reported to ORAP by LM2500 owner/operators last year, slightly more than half of those are represented on the Top Ten list (actually Top 11 because of a tie for tenth place).
SPS said outages caused by gas fuel control and regulating valves lasted less than five hours on average. Breaking down the category, 15 of the 53 incidents were traced to issues with metering valves, half related to the actuator with most of them reported by one unit. Eight incidents on seven units concerned the failure of the gas valve to shut off; another eight were attributed to sticking gas valves, but six of these incidents were at one site. Other failures were traced to gas-valve wiring, calibration issues, pressure fluctuations, etc.
Controls/controllers/communication were involved in three-dozen forced-outage incidents, a third of them related to issues with the controller or input modules on 10 units. Most outages in this category were of short duration, averaging less than four hours. It doesn’t take much time to replace I/O cards (eight incidents), provided you have replacements onsite.
Combustion system was in third place with 30 incidents reported, but again, most quickly resolved. The total outage time for this category was 117 hours. One-third of the outages were attributed to “loss-of-flame, causes unknown.” Most of the upsets occurred on startup; three were automatic trips, two of those on one unit. Flame-detector failures should be expected, particularly where legacy equipment continues in service. Last year, users reported eight incidents on seven units. Bad igniter fuse, acoustic sensor failure, combustion mapping issues, broken fuel nozzle, and high acoustic pulsations were among the other causes.
Substation failures/grid instability came in at No. 4 with two-dozen incidents, half of those attributed to issues with connected solar plants at two sites. Offsite substation failures and six incidents of grid instability at one plant accounted for the other trips. Outages in this category averaged more than a day.
Trips attributed to lube-oil systems are to be expected and usually can be cleared during the shift that they occur. Sump RTDs are a source of some problems—broken cable/wiring, for example. Oil leaks are another sure bet, as are incidents caused by clogged filters. Weather events are virtually guaranteed to make the Top Ten. Last year, a dozen outages were caused by lightning strikes, another six attributed to the effects of cold weather (freezing equipment, sensor lines, etc).
Steam injection finds application in LM2500s for NOx control and/or power boost. Steam leaks and valve issues always are a possibility. Station transformers were ranked No. 8 in terms of incidents, but this category accounted for the second-most number of outage hours.
Interestingly, the step-up transformers serving seven units at two sites were reported as “failed, unknown causes.” Plus, there was damage to LV winding insulation on three units at one site and temperature excursions were experienced by three units at another site. The bottom line: Of 16 total incidents, the 13 noted above occurred at only four sites. Transformer issues abound in the industry, particularly where there is limited on-staff electrical know-how.
Fuel-gas compressor trips (ninth on the list by number of incidents) typically were categorized as “general.” Tied for tenth place were power turbine instrumentation and HP compressor instrumentation and VSV actuation, with 14 incidents each.
The longest outage ran six months, for repair, in the US, of damaged blading in the sixth stage of the power turbine on one international unit. Most of the outage time was caused by having a lease engine tied up in customs. Station transformers came in second with 1437 outage hours. Together, these two incidents totaled nearly two-thirds of the total hours for the Top Ten contributors to outage time. No. 3 was 818 hours assigned to seven boiler-feed-pump motor issues—six at one site.
The Top Ten contributors to LM2500 forced-outage hours in 2013 accounted for 81% of the outage hours charged against the fleet last year. The power turbine issue alone contributed nearly half that amount.
The countdown to extinction of the LM5000 began when GE stopped manufacturing the engine several years ago; only about 50 units continue in service worldwide. In 2010, Air New Zealand Gas Turbines became the sole parts supplier for LM5000s worldwide, as well as the depot repair facility for GE’s LM5000 lease fleet. Today, Western Turbine meetings may be the only place owner/operators can still meet as a group to share experiences and maintain a “life-support” network for this engine.
Nearly half of the machines in the fleet are arranged for simple-cycle operation; 40% for cogeneration service. Balance of the units serve in combined-cycle systems. Most engines (70%) burn gas only; 20% are equipped for dual-fuel service. The high-time machine has approximately 200,000 operating hours.
The gas fuel system was the largest contributor to forced-outage incidents in 2013, as it was for the LM2500 fleet, Strategic Power Systems Inc’s Karl Maier and Daniel Ralph told LM5000 owner/operators at the 2014 meeting. They began with a summary of LM5000 ORAP stats for the five-year period extending from Jan 1, 2009 through Dec 31, 2013. The information was presented by service duty: Peaking machines aggregated 24 unit-years of operation; cycling engines, 30 unit-years; base-load units, 19 unit-years.
Highlights of the SPS analysis for cycling machines revealed the following:
- Fleet availability (simple-cycle plant) averaged 93.7% annually over the five-year period, with the 2013 number slightly below that at 92.9%.
- Forced-outage factor, 4.6% in 2013, was significantly above the annual 1.9% average from 2009 through 2013.
- Maintenance-outage factor, 0% in 2013 as it was for the LM2500, beat the 0.3% annual average.
- Planned-outage factor, 2.4% in 2013 was less than the 4% annual average for the period.
- Service factor (the percentage of time units generated power) was 35%, significantly above the 28% annual average.
Results for base-load machines included the following:
- Fleet availability, 84.7% in 2013, averaged 88.2% from 2009 through 2013.
- Forced-outage factor of 0.5% in 2013 was well below the five-year average of 4.5%.
- Maintenance-outage factor, 0.2% last year, was half the annual average of 0.4%.
- Planned-outage factor, which averaged 6.9% over the five-year period, was 14.6% in 2013.
- Service factor was 63.4% in 2013, down significantly from a 72.1% period average.
Engine removals, both scheduled and forced, also were reported by Maier and Ralph. In the five-year period, LM5000s reporting data to SPS accumulated only 230,000 service hours (round numbers) while experiencing 18 forced and nine scheduled removal events. As for the LM2500, no peaking unit forced out of service during the evaluation period required a shop visit; plus, no peaker was sent to the depot for a scheduled overhaul from 2009 through 2013. Cycling units experienced seven forced removal events (one about every 10,000 service hours) and six scheduled removal events. The mean time between forced outages of base-load engines was about 14,000 hours.
Analysis of forced-outage incidents
Next, the speakers presented details on the Top Ten causes of forced-outage incidents recorded from Jan 1 through December 31, 2013, giving both numbers of occurrences and their duration in hours. In sum, 100 total incidents were reported to ORAP by LM5000 owner/operators last year, 75% of them are included in the Top Ten categories.
There were 19 incidents reported against the gas system, more than a third attributed to faults and failures with the fuel control/regulating valves—almost a mirror image of the LM2500 data. And as it was for the LM2500 fleet, the outages averaged just under five hours each. Lube-oil system issues were second on the Top Ten list with 11 incidents, including the usual suspects: engine oil leaks, sump and scavenge pipe leaks, problems with oil filters, etc. Outages averaged just over 14 hours for this category.
The power turbine was third in number of incidents, with eight, but first by a wide margin in outage duration. In three cases the power turbine required replacement. Reasons were vibration, oil leak, and “unknown.” The 1520 hours of downtime required to correct PT problems accounted for nearly 60% of the total forced-outage hours for LM5000 engines reporting data to ORAP last year.
Turbine package controls was tied with PT issues for third place on the Top Ten list with eight incidents. However, outages to resolve these problems averaged the equivalent of only one shift. Loose wiring and vibration monitoring challenges often were the culprits. The steam injection system was fifth in number of incidents. Gremlins were in the steam flow control valve and injection flowmeter.
Ranking sixth through tenth were grid instability, HP compressor adjustments, compressor discharge air components, LP steam valves, and turbine instrumentation and cables.
At the end of 2013, the operating LM6000 fleet numbered 815 units, 40% of them participating in ORAP. Steve Giaquinto, and his SPS co-presenter, Dennis Russell, broke down their statistical analysis of fleet performance based on duty cycle, as their colleagues had done for the other GE aeros. Here are the highlights of that effort:
Peaking machines. Over the five-year evaluation period, this class of LM6000 assets totaled 598 unit-years of service. The performance summary for 2013 is presented below; the numbers in parentheses are averages for the 2009-2013 period:
- Forced-outage factor, 2.7% (3.04%).
- Maintenance-outage factor, 0.4% (0.76%).
- Planned-outage factor, 3.9% (3.26%).
- Service factor, 5.4% (4.64%).
- Availability, 92.9% (92.9%).
No real surprises in these data. The higher-than-average service factor in 2013 makes sense given grid requirements to fill in for must-take intermittent renewables these days. However, the 5.4% figure is seven-tenths of a percentage point lower than the service factor recorded in 2012. This is understandable given the 10.5% reduction in natural-gas consumption by the electric power sector (2012 versus 2013) reported by the US Energy Information Administration.
Cycling units totaled 462 unit-years of service during the 2009-2013 evaluation period. Performance data are the following:
- Forced-outage factor, 3.6% (2.38%).
- Maintenance-outage factor, 1.1% (1.32%).
- Planned-outage factor, 3.1% (3.62%).
- Service factor, 24.2% (25.7).
- Availability, 92.3% (92.7%).
The data show the 2013 forced-outage rate up significantly over the 2009-2013 average; it also is higher by a percentage point or more than the forced-outage rate for every other year in the period of interest. The 2013 service factor was less than the five-year average, as it also was for the peaker and base-load categories.
Base-load engines Engines in base-load service compiled 493 unit-years of service from Jan 1, 2009 to Dec 31, 2013. Here are the performance metrics for this group of GTs:
- Forced-outage factor, 1.3% (2.0%).
- Maintenance-outage factor, 1.0% (0.96%).
- Planned-outage factor, 2.7% (3.44%).
- Service factor, 76.1% (79.2%).
- Availability, 95.0% (93.6%).
Interesting to note is that the service factor in 2013 was lower, by from 2.4 to 5.5 percentage points, than it was for every other year in the period studied. Perhaps less operation contributed to the favorable 95% availability figure, which was higher in 2013 than any in other year from 2009 through 2012.
Engine removals, both scheduled and forced, also were reported by Giaquinto and Russell. In the five-year period, the LM6000s reporting data to SPS accumulated more than 4.7 million service hours, experienced 72 forced and 168 scheduled engine removal events. Mean times between depot visits were 20,603 hours for the base-load engines, 18,373 hours for cycling units, and 14,091for peakers. However, the mean time between forced outages requiring depot visits was 72,769 hours, or 8.3 years, for base-load units, 55,118 hours for cycling assets, and 39,924 for peakers.
Analysis of forced-outage incidents
Engine performance review complete, Giaquinto dug into the causes of forced-outage incidents, reporting both numbers of occurrences and their duration in hours. The LM6000 Breakout Chair Andrew Gundershaug, who manages several peakers for Calpine Corp in California, urged his user colleagues to listen carefully to what Giaquinto was about to share because that information would point to engine spares they should have on hand at their plants.
The Top Ten contributors to forced-outage incidents in 2013 focused on control systems and related components—including controllers, cards, power supplies, communications links, thermocouples, gas fuel modulating valves, gas staging/control valves, flame detectors, etc. Outages involving these components represented nearly one-third of the 886 incidents reported in the Top 10. Although control systems was second overall in number of incidents with 172, it was first among the issues over which plant personnel had some measure of control.
In first place were the more than 250 incidents aggregated under external causes. Nearly three-quarters of those were attributed to gas supply interruptions, grid problems, and substation and line work; they contributed more than 5000 of the 5290 total outage hours for the category. Environmental issues—lightning, wind, heavy rain, snow, etc—accounted for most of the remaining outage time from external causes.
Third on the list was 136 incidents reported against SCR and oxidation catalyst systems, but they totaled only 118 hours of outage time. More than 100 of the incidents were related to high CO indications at one site.
In fourth place was burners and nozzles annular and general combustion issues. More than half of the 73 incidents reported in this category were caused by flameout or failure to ignite. Gas compressors have their share of O&M challenges and this year finished fifth with 47 mentions, 17 of those related to gas pressure.
Station equipment transformers were charged with 43 incidents (sixth place), but this was the high-hours group, recording a total outage time of 7951 hours—including more than 5000 hours to replace one transformer. Second in forced-outage hours was external causes. Third was one crane incident that caused an outage of 4858 hours. Next on the list: Repairs to two generator rotors which took nearly 4000 hours.
Failure of a third-stage HP compressor blade on one unit and replacement of a cooling tower for another plant each caused outages in excess of 2000 hours, good for fifth and six places, respectively. An HV breaker failure at one site and replacement of a cracked second-stage HP turbine nozzle at another each were charged more than 2000 hours of outage time, ranking these incidents seven and eight in the hierarchy. Four generator neutral ground faults totaling more than 1300 outage hours and damage caused by the curling of a second-stage HP compressor shroud (nearly 1200 hours) were ninth and tenth on the list.
The LMS100 fleet rang up all-star performance stats in 2012, but key performance indicators (KPIs) were not nearly as good in 2013, SPS’s Sal and Tripp DellaVilla told attendees at the LMS100 breakout session. For example, 2013 forced-outage factors of 9.1% for peaking units and 5.6% for cycling units both were well above comparable 2012 data and also above the average annual FOF for the last five years.
Two unrelated events—the unavailability of replacement compressor blades for one unit and a bearing problem at another—accounted for nearly half of the fleet’s forced-outage hours last year. Eliminating these two outliers from the data would bring the fleet’s KPIs within the range of expectations.
Here are the LMS100 fleet metrics for 2013 based on ORAP data:
- Availability, 87.5% for peaking units, 89.6% for cycling units. In 2012 the numbers were 93.4% and 95.1%, respectively. Annual averages from January 2009 and December 2013 were 90.8% and 90.4%, respectively.
- Forced-outage factor, 9.1% peaking, 5.6% cycling. For 2012: 1.9%, 2.4%. Annual averages: 5.3%, 4.6%.
- Maintenance-outage factor, 0.9% peaking, 1.3% cycling. For 2012: 2.5%, 0.5%. Annual averages: 0.8%, 2.5%.
- Planned-outage factor, 2.4% peaking, 3.4% cycling. For 2012: 2.2%, 2.0%. Annual averages: 3%, 2.4%.
- Service factor, 7.1% peaking, 17.1% cycling. For 2012: 6.9%, 25.6%. Annual averages: 6.7%, 20.7%.
At the end of 2013, the LMS100 simple-cycle fleet numbered 51 units, 34 of them (67%) participating in the ORAP program. For the five-year period from January 2009 through December 31, DellaVilla told the group, peaking machines providing operating data to ORAP had accumulated 29 unit-years of service; cycling, 49; and base load, nine. The mean time between engine removals for the fleet was 8844. There were nine forced removal events and nine scheduled removal events during the period investigated. Interestingly, the mean time between engine removals for peaking machines was less than half the fleet average at 3893 hours.
Analysis of forced-outage incidents
There were 348 incidents reported across the fleet in 2013 with the Top Ten contributors to forced outages accounting for 43% of them—about the same as in 2012. In terms of number of incidents, the category controllers and software was No. 1 on the list (42 incidents) as it was last year (33). But with 11 more engines in the ORAP program this year than last, there were fewer incidents per machine. However, the outage time per incident was much higher in 2013 (11.3 hours) than in 2012 (3.8).
ORAP data do not indicate why outage hours increased so dramatically year over year (YOY), but it may be associated with a slower response time in addressing failures. Consider that staffing usually does not increase linearly with the number of units at a given site and, given the decrease in service factor, there may have been little incentive to pay overtime and premium shipping costs to correct deficiencies as quickly as in the past.
A breakdown of the controllers and software category showed that 27 of the 42 incidents, involving nine sites, attributed the outages experienced to controller failures. Controller calibration issues, and loss of communication among turbine package controllers, the data acquisition system, and HMI, each accounted for three additional incidents at three sites.
Second among the Top Ten causes of forced outages in 2013 was gas fuel with 32 events (88 total hours). This category was not among last year’s Top Ten; however, it also made the Top Ten list this year for the LM6000. Given that the majority of trips involved fuel metering, test failure, logic issues with new software, sticking valves, and wiring issues, it may be that recent implementation of industry-wide safety initiatives concerning the fuel gas system are linked to a significant number of the short-duration unit trips.
No. 3 on this year’s list, plant air compressor, also did not appear among the 2012 top 10. All 16 incidents (52 outage hours) occurred at just three sites. One of the affected plants reported that its engines tripped on loss of air pressure and were forced into a 4-hr lockout. It may be that with several new units reporting this year, the lockout was an unknown consequence of an air compressor upset.
SCR issues came in at No. 4 with 13 incidents (23 hours). Nine incidents were reported at one site because of high NOx and/or CO; two incidents at another site were attributed to failure of tempering-air fans; operator error and a catalyst issue caused the remaining incidents. No. 5 was 11 incidents (eight hours) at one site involving issues with variable bleed valves that caused unit trips. Calibration corrected the problem.
Faulty PLC logic/data translation and calibration issues accounted for half of the eight incidents (17 hours) reported in the No. 6 category, emissions monitoring. No. 7, combustion system, tallied eight forced outages (30 hours); reasons included failure to light off (low T48 temperature), flameout, and excessive temperature spread. No. 8, lube-oil chip detectors, caused seven trips, five at one site.
Variable-bleed-valve actuators, No. 9, failed seven times at one plant because of low oil pressure, VBV1 position mismatch, or general malfunction. No. 4 ball bearing came in No. 10 with five incidents, all at one site. But the outage hours totaled a whopping 2699, putting it No. 2 on the Top Ten list by number of forced-outage hours. Three of the five incidents involved downtime caused by high vibration during startup, later attributed to a broken bearing. The chip detector alarm forced operators to take the unit out of service twice for bearing inspection.
First on the Top Ten list by hours, HP compressor blades concerned the lack of available parts to replace damaged airfoils. The single unit affected was out of service for more than five months (3721 hours). No. 3 was LP compressor issues at three different sites. In all three cases, damage to blades and vanes was in evidence. One outage, totaling 1008 hours, was caused by a stall event; the root cause is under investigation. Another machine was out of service for 988 hours; the third unit was down for 551 hours. In the last case, the damaged hardware was identified during a borescope inspection.
Coming in at No. 4 on the Top Ten was 900 hours of outage time required to repair broken generator leads on one unit. Thus, the first four spots on the Top Ten, involving six units at different sites, totaled 9868 hours—or more than two-thirds of the forced-outage time suffered by the entire fleet in 2013.
No. 5 in terms of hours was variable-bleed-valve actuators (521), No. 6 was controllers and software (476 hours)—both covered previously. No. 7 was a 412-hr unplanned outage to repair intercooler leaks on one engine, No. 8 was 333 outage hours incurred when the transmission operator shut down a line to one site for repairs, No. 9 was the 290 hours it took to deal with a HP compressor casing crack, and No. 10 was 257 hours to address package piping oil leaks—one of which was implicated in a fire. CCJ