Don’t accept poor starting reliability

By Brent A Gregory, Creative Power Solutions

Some power producers want to believe that gas-turbine-based generating facilities don’t require a rigorous engineering effort at the design/specification stage of a project. Given that the track record of gas turbines for land-based generation service industry-wide is excellent, and that GTs are “standard” facto­ry-assembled machines, it’s no won­der that executives who majored in business and finance think buying a gas turbine is not much different than buying a new corporate limo: Sign the papers, pay, and you’re ready to go.

But the “standard” engine often doesn’t quite match actual site and operating conditions and, conse­quently, the new plant may not meet the owner’s expectations regarding starting reliability, availability, effi­ciency, etc – critical factors in the competitive generation business. It’s rare that a modest investment in engineering and hardware improve­ments can’t correct the situation. The experience of the General Electricity Co of Libya (Gecol) that follows is a case in point.

The utility’s GT fleet includes 23 Alstom machines – including these models: 13D, 13E1, 13E1M, 13E2MXL, and 8C. One of its combined-cycle plants, a facility with three 2 × 1 power blocks and equipped with distillate-fueled 13E2 engines, faced recurring problems during wintertime starts. It turned to Creative Power Solutions (CPS) for a root-cause analysis and per­manent solution (Fig 1). Note that Libya, like most Middle Eastern nations, does not release the names and locations of important indus­trial facilities.

For readers unfamiliar with the 13E2, it is a robust frame engine with a 21-stage compressor, five-stage tur­bine, and a 72-burner annular com­bustion system. Last has two ring-type fuel manifolds, each with 36 burners. Arrangement of burners is such that there’s one located every 5 deg around the combustion chamber. The OEM claims this design assures even temperature distribution and avoids what it considers problem zones in competitive can-annular systems – such as cross-firing tubes and transition pieces.

The Libyan 13E2s have two igni­tion torches located 120 deg apart. They are ignited electrically and operate on bottled gas; flame is sup­ported with air from the instrument air system.

CPS engineers began the proj­ect by examining data provided by Gecol for both successful and failed starts. Based on this work, hypoth­eses were put forward and an action plan developed to test the hypoth­eses and identify possible solutions. Next, the preferred solution would be implemented on one engine and, after validation, would be installed on all six 13E2s.

How the engine is supposed to start (follow curves in Fig 2). First steps in a successful start: The ignition sequencer lights the igni­tion torch and later presets the oil control valves to the system filling stroke. When the flame is stable, as indicated by an increase in torch tip temperature, the trip shut-off valve (TSOV) opens and oil fills the system up to the sector (pilot) valves.

After a timed delay to fill the system, control valves go to ignition stroke, the burners ignite, and moni­tors detect the main flame. Then the torch is shut off and its tip tempera­ture drops.

That’s the plan. But analysis of startup data revealed the following scenarios which caused or contrib­uted to the failed starts:

• Ignition pilot flame either did not ignite or failed after ignition.
• Combustor ignited and the tur­bine exhaust temperature (TAT) rose to the alarm limit, accelera­tion dropped below its limit, and the engine tripped.
• Combustor ignited and operation was normal – that is, until the TSOV closed (data offered no clues as to why this happened) and the engine tripped on “no fuel.”
• Combustor ignited and opera­tion was normal – that is, until the flame monitors no longer saw the flame (weak signal) and the engine tripped on “flame-out.”

Several hypotheses proposed for the various startup failures observed were investigated in detail by analyzing the following:

• Fuel properties.
• Operating conditions.
• Ignition torch tests.

Here’s what engineers learned:

1. Low-quality atomization of liq­uid fuel during engine startup.
2. Wide variability – 36 deg F – in the flash point temperature of the light fuel oil burned.
3. Ignition torches produced a weak flame at the established operat­ing parameters.
4. Rapid depletion of bottled pilot gas caused a weak torch flame dur­ing startup. (Perhaps the gas bottles were too small or were not being refilled fully.

The details

Fuel nozzles for the Gecol GTs are designed to inject two fluids simulta­neously: distillate and water for NOx control. A drawback of this arrange­ment is that during ignition, NOx water is not used and the velocity of injected fuel is low, which is condu­cive to poor atomization (Fig 3).

The large droplets of fuel resulting from poor atomization vaporize slow­ly and mixing of the oil vapors with air likewise is poor. Such conditions produce an erratic flame which can self-extinguish. A strong pilot flame can help assure reliable ignition in such situations.

Engineers thought another reason for the failed starts might be high flash-point temperature, so they ana­lyzed distillate samples from many shipments for viscosity and flash point. Fig 4 shows that high-flash-point fuels also have high viscosities and, therefore, are more difficult to atomize than low-flash-point/low-vis­cosity oils. Variability in flash-point temperatures across the samples was 36 deg F, confirming in the minds of investigators that high viscosity was the underlying reason for at least some failed starts.

To sum up, the CPS investigators concluded that the primary reason for low starting reliability was poor ignition caused by weak gas pilot flames and poor atomization. The ideal oil for startup, they agreed, would be one with a low viscosity to achieve fine droplet sizes during atomization and a low flash point to facilitate ignition. Also that the pilot flame must be strongly “attached” to its lance and be sufficiently hot to vaporize and ignite all possible oils delivered to the plant.

Another benefit of prompt igni­tion and complete combustion dur­ing startup is the elimination of oil residue in the combustion chamber from failed starts. When that residue ignites during a successful start the additional heat may trip the unit on “high TAT.”

The solution

Obvious from the engineering evalu­ation was that igniter flame strength had to be increased to improve start­ing reliability. A more robust flame would help even large oil droplets vaporize and ignite quickly. A fully instrumented test rig that would allow CPS engineers to vary air and gas flow to one engine’s ignition torches was assembled in the plant workshop (Fig 5).

First step was to identify the opti­mum flow conditions, with no orifices installed, using the flowmeters on both the air and gas lines. The data gathered provided the basis for quan­titative analysis to determine the optimum operational envelope for the igniters. The parameters used to indicate flame condition included the following:

• Torch metal temperature. A ther­mocouple was used for this pur­pose.
• Combustion-zone temperature. Measured by a thermometer to assess the impact of gas-flame radiation on fuel-oil vaporization.
• Flame strength. A small-diame­ter, high-velocity jet of compressed air was blown across the flame to qualitatively determine its stabil­ity/strength. Photos were taken during the tests at various oper­ating conditions to record visual observations.

Mapping of test results identi­fied the air and fuel flows that pro­duced both high flame strength and temperature (dashed oval in Fig 6). This information was verified as suitable for conducting a full-scale proof-of-concept test on one of the station’s GTs with the expectation of success.

With an optimum operating win­dow identified, engineers conducted a quantitative analysis to demon­strate the importance of air and gas throughput on flame strength. This work was done by calculating flow rates for different orifice sizes based on air and fuel line pressures. Plots for theoretical air and fuel flow rates as a function of line pressure and orifice size is shown in Figs 7 and 8, respectively.

In Fig 7, the ellipse defined by the dashed line presents the optimal window for air flow. It was deter­mined using the minimum air pres­sure expected from the instrument air system – about 70 psig – and the blow-off point for the igniter flame, as determined during the workshop tests and plotted in Fig 6.

The takeaway from this effort was that an orifice in the air line of 4 mm or larger is undesirable. Further, that use of a small orifice (1 mm), as suggested by the OEM’s design crite­ria, also is not in your best interest.

Optimal window for fuel flow revealed that a moderate orifice size was the best option – too small and flame temperature would be too low, too large and the flame would blow off.

The ignition flame produced when using OEM-recommended orifice sizes in the gas and air lines is too lazy (Fig 9). Modest enlargement of operations team improved flame strength somewhat, but not enough for the required task.

Orifice sizes for the gas and air lines suggested by calculations per­formed by CPS engineers were then tested using the engine ignition torches. Stable flame conditions were obtained in the region marked in the upper right of Fig 10.

The stability of a flame depends on its thrust: The higher the thrust, the more heat available to vaporize the liquid fuel. For a given torch, thrust is increased by boosting the total flow of gas and air through the ignition system. The OEM igni­tion system performed poorly at this plant because the fuel did not meet the manufacturer’s spec and there was insufficient gas to create a strong flame.

Use of orifice sizes specified by CPS created the flame shown in Fig 11. Use of these orifices on all six engines restored starting reliability to fleet-wide numbers. All engines started on the first attempt. Ccj