B URNING V ELOCITY
5.2. RATIONALE FOR LEAN COMBUSTION IN GAS TURBINES
5.2.2. S TABILITY
Stability in gas turbines is a major design consideration. As suggested in Figure 5.10, the stability limit effectively defines the operability of a given design for a given inlet condition. As shown, any combustion system will have a characteristic“stability loop”
for a given fuel, temperature, and pressure. As the amount of air pushed through the system increases, the range of fuel/air ratios for which stable burning can be maintained will change. Figure 5.10 also shows a number of regions that are of particular interest.
The lower region defining the separation of stable and unstable combustion is the lean stability limit, sometimes referred to as the “static” stability limit. It is desirable to operate the system at as low a fuel/air ratio as possible because the lower fuel/air ratio areas will produce beneficial low temperatures. Similarly, the upper line demarks the rich stability limit. Along the stability loop generally lie seemingly random isolation areas of high frequency oscillations. These acoustic oscillations are referred to as
“dynamic stability” regions. The conditions giving rise to these relatively high fre- quency oscillations are difficult if not impossible to predict. Finally, as the air flow through the system reduces, the possibility of flashback occurs as the reaction has enough velocity to propagate upstream into injectors or premixers. Note that the most stable situation relative to airflow is the situation where the system is operated at stoichiometric fuel/air ratios (e.g., around fuel/air ratio of 0.01 in Figure 5.10). In the next sections, various aspects of stability and operability are discussed.
0.03
0.02
0.01
00 0.25 0.50 0.75 1.00
Air mass flow (kg/s)
Fuel/air ratio
High frequency oscillations Flashback
Stable burning
Static stability limit
Constant T3, P3 Constant fuel composition
Figure 5.10 Illustration of combustion operability issues for gas turbine combustor at fixed inlet temperature and pressure (Lieuwenet al., 2006).
5.2.2.1. Static Stability (Lean Blow-Off)
In many combustion systems, the static stability of the reaction is a key parameter relative to the overall operability and performance of the system. Over the years, significant effort (e.g., McDonell et al., 2001) has been directed at establishing rela- tionships between stability and conditions/geometry of the flame-holding device. As a result, basic design tools have been developed which can give insight into how changes in operating conditions and/or flame stabilizer geometry may impact the stability. These are relevant for both stationary and aviation type combustors.
For example, specific work by Lefebvre (Lefebvre and Ballal, 1980; Lefebvre and Baxter, 1992) has provided insight into the functional relationships among stability, equivalence ratio, free stream velocity, pressure, temperature, and bluff body geometry.
In that work, a cylindrical cone-shaped bluff body was introduced at the centerline of a larger diameter pipe within which either pure air or a mixture of gaseous fuel and air flowed. In order to study the stability of liquid fueled systems, pure air was used and liquid was introduced by a nozzle at the centerline in the wake formed downstream of the bluff body. The conclusion from the work conducted by Lefebvre and co-workers is summarized by the correlation expressed in Equation (5.1), which relates the weak extinction limit to the various parameters studied:
fLBO¼ 2:25½1þ0:4U(1þTu0) P0:25Toe(To=150)Dc(1Bg) 0:16
: (5:1)
The attractiveness of Equation (5.1) is that it provides a relationship of how the parameters studied influence the stability limits and also provides some guidance on how to design effective reaction holders for a given situation. It also provides a mechanism to scale results obtained at one condition to other conditions of interest.
The physical mechanism leading to Equation (5.1) is the ratio of reaction time to residence time, otherwise referred to as the combustor loading parameter. This time ratio is also captured through the non-dimensional group, the Damköhler number.
Characteristic time models are based on the concept of the reaction zone acting like a well-stirred reactor, which may be a reasonable representation in some cases. Other examples of this type of approach include the work of Mellor and co-workers (Plee and Mellor, 1979; Leonard and Mellor, 1981) which is summarized by the equation,
tslþ0:12t0fi¼2:21(t0hcþ0:011t0eb)þ0:095: (5:2) In these characteristic time models, characteristic values are needed for velocity as well as the reaction zone size. A characteristic dimension divided by a characteristic velocity provides a typical time scale associated with the reaction. As a result, some thought is needed to identify appropriate values.
Another class of model involves a comparison of reaction propagation to the flow speed. This type of model may be more physically representative of a reaction stabilized by a vortex breakdown. In this class of flow, the ability of the reaction to anchor itself does seem related to the reaction speed propagating toward the flame holder versus the axial velocity of the flow approaching the stagnation point. This type of model has been typically implemented through the use of Peclet numbers (Putnam and Jensen, 1949) and has been utilized recently to correlate stability of swirl stabilized flames
(Hoffmanet al., 1994; Kroneret al., 2002). A challenge with existing Peclet number approaches is the reliance upon turbulent flame speeds. As has been illustrated in Chapter 2 and in the literature, the typical dependency of turbulent flame speed on laminar flame speed is not intuitive and can be sometimes misleading (Kido et al., 2002). Furthermore, recent work has suggested that neither the characteristic time nor propagation based models are sufficient for describing more subtle effects such as fuel composition due in part to their dependency upon the turbulent flame speed (Zhang et al., 2005).
In summary, although extensive work has been done on developing tools for static stability limits, some inconsistencies have been observed which may or may not be dependent upon the type of stabilizing geometry used or upon the fuel composition.
This is an area which needs further refinement.
5.2.2.2. Dynamic Oscillations
Dynamic oscillations are potentially significant problems for both stationary and propulsion engines. A review of these issues is available in both recent (Lieuwen and McManus, 2003) and earlier reviews (McManuset al., 1993), and is discussed further in Chapter 7. The manifestation of the dynamic oscillations is tied to circumstances where perturbations in the heat release and pressure field couple together in a resonant manner. If the perturbation in heat release is in phase with an acoustic wave, the results can be destructive. Figure 5.11 illustrates an example of the type of destructive force combustion oscillations can cause to gas turbine engines.
Oscillations have been studied extensively for the past decade relative to lean gas turbines due to the problems that have been encountered (Candel, 2002). In most cases, solutions have been found through trial and error approaches. Considerable effort has been directed at predicting whether or not oscillations will occur for a given system, but a comprehensive model has yet to be fully developed (Ducruix et al., 2003; Dowling and Stow, 2003; Lieuwen, 2003). Despite this, progress has been tremendous, and solutions ranging from altering fuel schedules or physical injection locations to the inclusion of damper tubes or hole patterns have been attempted successfully (Richards et al., 2003). These so-called passive solutions are attractive because they are less complicated compared to more exotic closed loop active control approaches, but there
Figure 5.11 Transition failure in GE-F gas turbine due to dynamic stability issues (Tratham, 2001).
has also been effort directed at actively monitoring and controlling combustion oscilla- tions (Docquier and Candel, 2002; Cohen and Banaszuk, 2003).
Stationary gas turbines seem to suffer more from dynamic stability issues than do aero engines. This is because stationary turbines are designed to operate very near the static stability limit. At some points near the static limit (see Figure 5.10), a situation may arise in terms of fuel loading, aerodynamics, and heat release that lead to the onset of an oscillatory behavior. Operating near the static stability limit nearly guarantees the presence of finite perturbations in the combustor.
Stationary combustion systems have a few features that make them especially susceptible to these problems (Lieuwen and McManus, 2003). For example, to minim- ize CO emissions, the use of cooling air, especially along walls (“hot wall strategies”), is minimized. Elimination of extensive cooling passages and jets results in“stiff”bound- aries containing the acoustic field which provide minimal damping. Also, to achieve a high degree of premixing, the combustion system is set up to have the reaction sit downstream of some“dump plane”which potentially situates the reaction at an acoustic pressure maximum point. Finally, stationary combustors have some freedom relative to their length, and to ensure CO burnout, long burnout zones are common which leads to a relatively long dimension relative to the heat release zone and thereby the possibility of exciting organ pipe modes.
Nearly every gas turbine OEM involved in dry low emissions combustion has had to deal with acoustic oscillations. Both passive (e.g., moving fuel injection location, adding damper tubes, modifying “hard” boundaries in the combustion chamber, adjusting fuel split between a non-premixed pilot and a fully premixed main) and active (e.g., pulsed fuel) approaches have been demonstrated on fielded engines. Some very interesting experiences are described in the literature that illustrate the difficulty that the industry has faced with this issue (Hermannet al., 2001; Lieuwen and Yang, 2006).
Stationary engines have been a viable platform for the evaluation of closed loop control systems. Examples of full scale implementation of closed loop control on a 240 MW central station power plant can be found (Hermannet al., 2001).
In aviation engines, due to the focus on safety, engines will not operate as close to the lean stability limit, thereby resulting in some margin relative to the onset of oscillations.
However, acoustic oscillations can occur in situations where the system is operated
“conventionally” and even at stoichiometric combustion zones (Janus et al., 1997;
Mongia et al., 2003; Bernieret al., 2004). For aviation applications, where reliability and safety are key, passive approaches are highly preferred. Despite this, examples of implementing closed loop control on liquid fueled systems have appeared (Cokeret al., 2006). Applying active control on aviation applications could be possible if reliability and safety can be demonstrated conclusively. In advanced engines today, extensive sensor arrays are already in place, but comprehensive testing of any active control system would be needed before applying it in practice. Unmanned aircraft represent a viable test bed for such an application.
5.2.3. I
GNITION/A
UTOIGNITIONThere are two ignition issues associated with gas turbines: the first is ignition of the engine for startup, and the second is autoignition of the fuel–air mixture. Both affect the design of systems and are described briefly below.
5.2.3.1. Stationary
For stationary engines, ignition at startup is not really a critical design issue.
However, the occurrence of ignition within the premixer is a concern. Figure 5.12 illustrates the situation that exists in typical lean premixed combustion approaches. If reaction occurs within the premixer, it will result in high NOxemissions and can also potentially damage the injector/combustor. As a result, if the ignition delay time is shorter than the premixer residence time, the system will have operability issues.
For gaseous fuels, one approach to quantifying the potential for this problem is carrying out a chemical kinetic calculation using an appropriate mechanism.
Such mechanisms are commonly available for natural gas type fuels (http://www.me.
berkeley.edu/gri mech/; Ribaucouret al., 2000), although there is some debate as to the appropriateness of these mechanisms for ignition calculations.
Alternatively, global expressions for ignition delay have been developed which are convenient to apply. One example is shown in Equation (5.3) (Li and Williams, 2002):
t¼2:610 15½O2o4=3½CH41=3o
To0:92exp (13180=To) : (5:3) Recent work has shown that the few global expressions for low temperature ignition delay times have a wide range (2–3 orders of magnitude) of predicted values (Chen et al., 2004), so they should be applied and verified with care.
Experience has shown that autoignition for natural gas is not a major issue for lean premixed combustion. However, with increasing interest in alternative fuels, this issue has again been raised. Compounding this matter, while expressions for ignition delay for non-methane gaseous fuels are available, far less evaluation of their accuracy has been completed. Of increasing recent interest is operation on hydrogen containing gases. In this case, although extensive work has been done on hydrogen/oxygen reaction systems, only limited results are available for lower temperature regimes that are relevant to gas turbine premixers. To illustrate, Figure 5.13 presents a comparison of measured ignition delay times with calculated delay times using a well-accepted hydrogen/air reaction mechanism. As shown, a significant lack of agreement in the low temperature regions (e.g., <1000°F) is evident. As a result, this is another area that is in need of further work. Chapter 8 includes some additional discussion of hydrogen use in gas turbines.
Fuel
Air Mixing length
Reaction
Figure 5.12 Lean premixed combustion strategy.
5.2.3.2. Aviation
Ignition is a particularly critical issue for aviation engines, and altitude relight is always an important performance criterion. The ignitability of a mixture depends strongly upon the equivalence ratio in the vicinity of the igniter. Like static stability, a“loop”can be created for a given combustor which describes its ignition range. The ignition loop always falls within the static stability loop. This is because once the system is operating, energy from hot walls and radiation from the reaction help to widen the range of fuel/air ratios which will sustain the reaction. For ignition, cold fuel is on the walls and the entire combustion system is relatively cold. The differences between the static stability loop and the ignition loop give some characteristics of the combustion system. Ideally, these loops would be close to each other, but both must achieve requirements associated with overall stability requirements for a given combustor air flow.
Autoignition is a concern for lean premixed prevaporized (LPP) systems. For liquid fuels, the ignition delay time is generally shorter than it is for gaseous fuels. Because aviation fuels are so complex, it is difficult to identify a suitable kinetic mechanism which can be used for estimating ignition delay. However, global expressions for ignition delay are available in the literature for Jet-A fuel, and an example (Guin, 1998) is shown in Equation (5.4):
t¼0:508e(3377=T)P 0:9: (5:4)
This expression suggests that at 30 bar and 700 K inlet temperature, ignition delay times are around 3 ms. This is on the order of the typical premixer residence times. This is consistent with other work done on ignition delay for liquid fuels (Lefebvreet al., 1986).
Mueller mechanism versus experimental findings
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
1000/T (1/K)
Ignition delay time
Mueller et al., 1 atm Mueller et al., 3 atm Mueller et al., 5 atm Mueller et al., 10 atm Mueller et al., 15 atm Mueller et al., 20 atm Current Study, H2/CO (5 atm) Current Study, adjusted to 20 atm Peschke et al. (12–23 atm) Current Study, H2 (6 atm) Boleda, H2/CO/CO2 (5 atm) 10 µs
1 ms 1 s 10 s 100 s
100 µs 10 ms 100 ms
725 790 865 945 1040 1145 1270 1415 1585 1790 2040 2350
Temperature (F)
Figure 5.13 Comparison of ignition delay times for detailed mechanisms and measurements (Beereret al., 2006). (See color insert.)