7F Users Group – preventing blowout trips

Preventing blowout trips of 7FAs

Technical presentations by vendor representatives are an important part of every 7F Users Group meeting. Fifteen presentations usually are made during five 45-min sessions. Having three presentations in each time slot virtually assures every user attendee will identify with at least one subject. These sessions are conducted in the afternoon, just ahead of the vendor fair, enabling follow-on discussion in the exhibit hall.

A presentation that got high marks from many users at the group’s 20th anniversary meeting was Mitch Cohen’s “Understanding and Reducing Lean Blowout Trips of 7FA DLN2.6 Combustors.” Cohen is a respected senior systems engineer for Orlando-based Turbine Technology Services Corp.

Intermittent blowout trips, often referred to by the acronym LBO (lean blowout), are said to be relatively common for some DLN2.6 combustors not equipped with an appropriate advance-warning system or alarm. Negatives of LBO include accelerated degradation of parts and the obvious loss in kilowatt-hour production.

In his introductory remarks, Cohen said that LBO trips, which typically are annunciated by the control system as either “High Exhaust Spread” or “Loss of Flame” trips, do not always have a well-defined cause or clear-cut corrective action. Consequently, operators often restart their units only to suffer another LBO relatively soon.

He identified these potential sources of DLN2.6 blowout:

  • Combustor tuned to a fuel/air ratio that is too lean.
  • Low-frequency chug dynamics.
  • Worn combustion-system hardware.
  • Instrumentation failure or shift in calibration.
  • Shift in control-valve calibration.
  • Continuous emissions monitoring system (CEMS) out of calibration.

Note that the first two causes of LBO are tuning-related; remainder are a function of hardware condition.
Continuing, Cohen identified the following as LBO influence factors:

  • Rich- or lean-PM1 operation.
  • Ambient temperature.
  • Unit configuration—that is, simple- or combined-cycle operation.
  • IGV part-load temperature control curve.

Cohen’s goal was to offer operators practical steps they could take during maintenance outages, combustion tuning, and normal operation to reduce the likelihood of an LBO trip. This included recommended guidelines for allowable emissions and dynamics levels to avoid blowouts and how the suggested values are influenced by split schedule and control curve adjustments. Cohen’s tuning experience spans more than 50 7FA turbines.

The information presented here does not disclose all of the tuning “tricks” the speaker shared with the three-dozen or so owner/operators who attended the presentation—this to protect his competitive advantage. However, it does provide operations personnel valuable insights on why blowouts occur and how to protect against them, as well as on how to distinguish between trips caused by LBO and dynamic instabilities.
A brief backgrounder on key combustion principles and the arrangement of DLN2.6 combustion systems precedes excerpts from Cohen’s presentation to allow those unfamiliar with this equipment to benefit from his experience, which offers valuable lessons for all operations personnel.

The ABCs

The DLN2.6 combustor, which is designed for 9 ppm NOx, demands very tight control of the fuel/air ratio over the entire load range. Just how close this combustion system must operate to “the edge” to assure that emissions and performance goals are achieved is illustrated in Fig 1.

Note how near the lean flammability limit (LFL)—the point at which the fuel/air ratio is too lean to support combustion and LBO occurs—is to the desired operating range. Such close proximity suggests that if any one of a 7FA’s 14 combustors operates to the left of the dashed line the turbine probably would trip.

Above 1.0, there is fuel in the mixture that can’t burn because the amount of air available is insufficient to support complete combustion. That’s why high CO levels are a characteristic of fuel-rich flames. Below an equivalence ratio of 1.0, oxygen remains after complete combustion of the fuel and it serves to suppress flame temperature. A ratio of 0.5 means there’s twice as much air as needed for complete combustion.Equivalence ratio is another important term to remember. It is the actual fuel/air ratio divided by the stoichiometric fuel/air ratio. Recall that the defining characteristic of a diffusion flame is that fuel and air are introduced into the combustor separately and the mixture burns at an equivalence ratio of 1.0, where flame temperature is highest and NOx production typically is highest as well.

High CO is not just a characteristic of fuel-rich flames. It also may be found in lean flames because they are relatively cool and the conversion of CO to CO2 slows as flame temperature decreases. CO ultimately limits engine turndown; the low combustion temperatures associated with low-power operation cause a rapid increase in CO level.

In the premixed combustion process used for DLE-equipped turbines, air and fuel are mixed upstream of the combustion chamber, allowing tight control of mixture stoichiometry and, therefore, flame temperature. Figs 2 and 3 show the arrangement of fuel nozzles for a typical DLN2.6 combustor. Note that the staging of multiple fuel streams through the six fuel nozzles is required to achieve turndown flexibility while maintaining flame stability, controlling NOx emissions within requirements, and minimizing destructive dynamics.

Premix (PM) burners of the same color in the diagram are internally manifolded together within the end cover. The quaternary, or fourth, fuel manifold (a ring around the outside of the combustor casing and upstream of the premix burners) serves the series of radial pegs shown in Figs 2 and 3. Quaternary is used on some units to reduce the amplitude of specific dynamic pressure frequencies responsible for wear and distress of the combustion hardware. The basic DLN2.6 layout in Fig 4 reveals four gas control valves arranged in parallel and served by a common stop/speed ratio valve. The three-way splitter valves found in older combustion systems are not used.

Cohen’s presentation to the 7FA users essentially began at Fig 5, Mode 6 split schedules for rich- and lean-PM1 combustors, which shows how fuel flow is distributed among the various nozzles/manifolds as load changes. Recall that the term “Mode 6” means that all six fuel nozzles are receiving fuel. The horizontal scale, combustion reference temperature, varies with gas-turbine load: The higher the temperature, the higher the load. At base load, TTRF1 typically is in the range of 2380F to 2420F.

Cohen told the group that lean PM1 was the method of varying load on the gas turbine with the original DLN2.6 combustion system. In the left-hand chart of Fig 5, PM1 was at its highest split percentage at base load (extreme right) and dropped as the unit was unloaded (moving to left). Load turndown to about 50% of rated capacity was possible using this approach. Rich PM1 was developed after lean PM1 for the purpose of achieving greater turndown. It typically enables the combustion system to achieve a turndown of 40% or less while maintaining CO emissions within the desired range.

The dashed line at 15% of the PM1 split schedule represents the optimal split for NOx in a DLN2.6 combustor—that is, the split at which minimum NOx emissions can be achieved. The terms “lean PM1” and “rich PM1” are used relative to this split value. At a fixed firing temperature, NOx increases as the split is modulated from this value in either direction—on the lean-PM1 side because the PM3 nozzles are getting richer and hotter; on the rich-PM1 side because the PM1 nozzle is getting richer and hotter.

This is illustrated in Figs 6 and 7 for base-load and part-load operation, respectively. In Fig 6, all nozzles operate with fuel/air ratios above the lean flammability limit. For the lean-PM1 case at the left, PM1 is slightly leaner than PM2/PM3 (orangey-yellow tint); for the rich-PM1 combustor at the right, PM2 is running slightly leaner than PM1 and PM3.

At part load during lean-PM1 operation, illustrated at the left in Fig 7, PM1 is the leanest nozzle and the PM1 and PM2 nozzles are operating below the LFL; the PM3 nozzles are the richest nozzles and anchor the flame. LBO is more likely to occur with this operating scheme than with the rich-PM1 scenario at the right where the single PM1 combustor anchors the flame.

Reason is that it’s much easier to keep one nozzle above the LFL than it is to keep three above the LFL.

LBO with lean PM1


Cohen showed the group a common LBO signature (Fig 8), pointing out the characteristic step change in exhaust spread with no other anomaly identified in the trip log or in historical PI data. In this case LBO occurred at 75 MW, just as the engine was transferring out of Mode 6. The trip was not anticipated and occurred so quickly there was no action operators could take to prevent it.

At base load on the lean-PM1 split schedule, Cohen continued, all nozzles operate at close to the same fuel/air ratio and all nozzles, individually, are above the LFL. As the unit unloads, PM1 and PM2 are leaned out—eventually dropping below the lean flammability limit. PM3 must operate at a sufficiently rich fuel/air ratio to prevent LBO. An indicator of insufficient LBO margin, Cohen said, is very low NOx values.

To provide adequate margin, he suggested not operating combustors at less than 6.5 ppm during part-load operation—even though in isolated instances, combustors with properly balanced, precision fuel nozzles can operate as low as 5 ppm without experiencing LBO.

Cohen illustrated the importance of having a generous “floor NOx level” by reviewing LBO events at a site with three simple-cycle 7FAs operating in the lean-PM1 mode. When he arrived at the plant, Cohen found one of the engines (Unit B) operating at less than 6 ppm and the other two above 6 ppm (Fig 9). Unit B was the only machine to suffer repeated LBOs.

Problem was corrected by changing the PM1 and PM3 split schedules to increase combustor stability and eliminate blowouts. The split changes increased NOx by 1-2 ppm (compare the Unit B as-found and as-left curves in the chart). The ability to make such precise tuning adjustments requires, among other things, an accurate NOx analyzer in the gas-turbine exhaust stream, capable tuning personnel (on staff or outside consultants), and combustion components in good condition.

The skill of the tuner and his/her knowledge of how the engine will be operated and how it behaves under changing ambient conditions are critical to success. Fig 10 illustrates the point. It shows that units can be tuned too lean. Here an LBO trip occurred on a gas turbine at only 3 MW below base load as the gas turbine was being unloaded. The exhaust-temperature spread changed from 57 to 274 deg F within 1 sec. What happens during load reduction is that the change in fuel flow is instantaneous but the inertia of the GT rotor keeps air flowing at a high rate—in this case reducing the fuel/air ratio below the lean flammability limit in at least one of the combustors.

Cohen stressed the point that LBO can occur at any load within Mode 6 if the unit is not well-tuned. When he arrived at the site, he found NOx in the engine exhaust at less than 6 ppm at all loads and minimum NOx at less than 5 ppm just below base load, where the LBO occurred (Fig 11). Solution, as before, was to increase combustor stability by making changes to the PM1 and PM3 split schedules to increase NOx by 1 to 2 ppm.

Cohen noted that sites having SCRs often do not pay close attention to engine NOx emissions; operators focus on stack NOx levels, which are reported to environmental authorities. In such cases, it’s easy to overlook conditions conducive to LBO. The diligent tuning specialist, he continued, reviews unit historical data to see how NOx has varied with ambient temperature changes over the life of the unit, and what influence such variables as the shape of split schedules and the shape of the IGV part-load temperature control curve have on the NOx/ambient relationship. Some units, Cohen pointed out, are able to operate throughout the year without requiring seasonal tuning, but this is exception rather than the rule.

Chug-induced blowout with lean-PM1 combustors

So-called “chug” describes a low-frequency (15-18 Hz) dynamic mode capable of causing a blowout trip. Cohen said many 7FA users refer to this phenomenon as LBO dynamics; however, the mechanism of blowout really is a different phenomenon than LBOs caused by burners falling below the lean flammability limit, as described above. Chug-induced blowout is a trip driven by a dynamic instability which can occur at elevated NOx levels—7 to 8 ppm–that do not suggest operation below the LFL.

These two modes of blowout can occur independently of each other. A distinguishing characteristic of chug tones is that when they are present at amplitudes as low as 0.5 psi peak to peak, they can be heard as a low rumble and felt as floor vibrations when standing near the turbine. This is in contrast to 140-Hz hot-tone dynamics (the primary tone responsible for component wear) which cannot be sensibly heard or felt—even at amplitudes of 5-10 psi peak to peak.   

Another characteristic is that both lean- and rich-PM1 combustors are susceptible to chug-induced blowout. Cohen said he could not recall seeing or hearing of any LBOs on rich-PM1 combustors in which the combustor falls below the lean flammability limit as indicated by very low NOx emissions. The PM1 nozzle in the rich-PM1 configuration operates richer than the PM3 nozzles do in the lean-PM1 configuration, he added, more reliably anchoring the combustor and preventing LBO trips.

Fig 12 illustrates a chug-induced blowout of a lean-PM1 combustor. Comparison with Figs 8 and 10 shows it does not resemble classic LBO caused by burners falling below the LFL. Note the intermittent and unsteady increases in NOx and CO preceding the trip, coupled with highly unusual oscillations in compressor discharge pressure.

CDP oscillations of up to 5 psi max to min correlate closely with increases in NOx and CO (Fig 13). Although not shown in the graph, detailed data review showed that the CDP oscillations were not being driven by oscillations in the inlet guide vanes, the inlet bleed heat system, or the fuel system. When the unit was tuned, relatively low-amplitude chug dynamics (Fig 14) were present and were found to be the driver of both the CDP oscillations and the blowout of the combustor.

Cohen said amplitudes on the order of 1 psi peak-to-peak are capable of causing the combustor to blow out. Clearly, having a combustion dynamics monitoring system (CDMS)—which was not installed on this unit—can help identify chug before blowout occurs.  But  plant personnel must be trained to identify the phenomenon.  Fig 15 shows that for the lean-PM1 case illustrated, increasing chug amplitude prevented operation at loads lower than 135 MW. Note the abrupt end of the chug as-found curve. Retuning of the unit significantly reduced the chug amplitude (to less than 0.2 psi peak to peak) and restored the turndown capability to 50% load.

Chug-induced blowout with rich-PM1 combustors

As noted previously, rich-PM1 combustors, while virtually immune to LBO trips caused by combustors falling below the LFL, are susceptible to chug-induced trips. Fig 16 describes the chug-induced trip of a rich-PM1 combustor. Two 7FAs at this simple-cycle site experienced three such trips over a three-day period, each characterized by a sudden increase in exhaust spread. There were no obvious causal indications in the trip log and NOx was in the 7-8 ppm range—above the level one would expect classic LBO.

Historic dynamics data (15-sec average) in Fig 17 offer a closer look at the chug-induced trip described in Fig 16. The data plot for combustor can 9 on one of the affected engines shows repeated bursts of chug tone shortly before the trip. Can 9 had the highest amplitude among the 14 installed on the 7FA (0.8 psi peak to peak) when the trip occurred, although all cans exhibited chug tones. Because the data are presented as a 15-sec average, it is possible—perhaps likely—that the instantaneous amplitude at the moment of trip was greater than 1 psi.

As with the lean-PM1 case history that preceded this one, retuning of the unit eliminated the susceptibility to chug-induced blowout by lowering dynamics from 0.6 psi to less than 0.2 psi peak to peak (Fig 18). Note that NOx emissions were about the same before and after tuning—between about 7 and 8 ppm. CCJ