Experimental Characterization of Combustion Instabilities in Double-Stage Swirl Chamber Dener Silva de Almeida1, Cristiane Aparecida Martins1, Pedro Teixeira Lacava∗, 1 1
Aeronautical Technology Institute, ITA, São José dos Campos, SP, 12345-678, Brazil
Abstract The focus of the present work is a new configuration of Low-NOx combustor with especial application for gas turbine. For the concept proposed, the unfavorable combustion conditions for NOx formation are reached through the dynamic control of reactants mixing process into the double-stage combustor. In the first stage swirling air and gas fuel are injected and a sooting flame is observed in a region close to the flows interaction. From the first to the second stage, the combustor diameter increases and a pre-mixed lean combustion is stabilized into an intense recirculation zone. However, depending on the operational parameters, some acoustic oscillations happen for the lean combustion in the secondary chamber. Therefore, the present work is concerned about the influence of these operational parameters on the acoustic instabilities (equivalence ratio, fuel jet Reynolds number, swirler blades angle and primary chamber length/diameter ratio (L/D). The results show that increasing the swirler blades angle and reducing the fuel jet Reynolds number, the combustion oscillations can be attenuated. In addition, higher L/D ratios (maximum investigated here was 3) presented good results for instabilities attenuation. In spite of the difficulties to understand the complex phenomena of combustion instabilities, this work presents some recommendation to control the oscillations in this type of combustors. Introduction As result of the rapid increase of the air traffic and the large use of gas turbines to produce electric power, the reduction of pollutants has become the main objective in the design of gas turbine engines. To reduce CO2, one of the most important causes of the greenhouse effect, the tendency of the advance combustion chamber is to increase not only the inlet pressure and temperature but also the efficiency of the engine and to reduce the fossil fuel consumption. For aircraft gas turbines this is the only way to reduce CO2 emission. However, it increases the maximum temperature in the combustor and more NOx is produced. In this way, some efforts have been made to develop technologies to control NOx emissions for high performance gas turbines. One of the most promising ways to reduce emissions is to inject the fuel into premixing ducts at the exit of which, the fuel is pre-vaporized and the mixture is lean. In this case, the occurrence of stoichiometric zones is avoided and NOx production is reduced (Canepa et al., 2006). Unfortunately lean premixed combustors - LP suffer from serious problems related to combustion instabilities, flashback, and auto-ignition in the premixing ducts. Another technology to control NOx formation in gas turbine combustors is the Rich-Quench-Lean combustion – RQL. The RQL combustor designs incorporate an axially-staged approach in which the air injection is staged axially along the axis of the combustor. Although fuel-bound nitrogen species are converted to NOx very efficiently in an oxidizing environment (Sarv and Cernansky, 1989), if the nitrogen-bearing fuel is injected into a fuel-rich environment, a significant portion of the fuel-bound ∗
Corresponding author:
[email protected] Proceedings of the European Combustion Meeting 2009
nitrogen species will be converted to N2, instead of NOx. Once complete oxidation does not occur in the rich zone of a RQL combustor, additional oxidant is required to complete the combustion at an overall lean condition. The point at which this additional oxidant is injected is often called the quick-mix stage of an RQL combustor (Straub el al., 2005). This mixing zone is critical for minimizing thermal, or Zeldovich, NOx production; the air injected into the mixing zone should mix quickly and homogeneously and this is the main difficulty to design RQL combustors. In addition, in the rich zone considerable quantity of soot is formed, increasing dramatically the heat transfer by radiation to the combustor wall. Then, some complex system of air film cooling should be present. In fact, there are some improvements in LP and RQL original concepts to produce combustion devices with extremely low NOx emissions. For example: Lean Pre-mixed Burner with Spatially Periodic Recirculation of Combustion Products (Kalb and Sattelmayer, 2006), Double Swirler Lean Premixing Pre-vaporizing Burner (Canepa et al., 2006), Rich-Burn, Quick-Mix, LeanBurn Trapped Vortex Combustor (Straub et al., 2005), Rich-Catalytic Lean-Burn Combustion (Smith et al., 2005). Combinations of combustion zones unfavorable for NO formation or for conversion of fuel-bound nitrogen to N2 are the most important way to control NOx emissions by the modifications in the combustion process. Gas turbine combustion is also an object of research and technological development for the Combustion Group of the Aeronautical Institute of Technology – ITA in Brazil, as part of an institutional program with partners for gas turbines development. Low NOx combustor is also part of this program and
injected through the nozzle positioned in the center line of the primary chamber. Three different nozzle diameters were utilized, 2.35, 3.20 and 7.8 mm, to provide the fuel jet Reynolds numbers of 50,000, 40,000 and 15,000, respectively.
some experimental studies for new configurations of combustion chamber have been done. In particular the present paper is concerned about the studies of acoustic instabilities in a double-stage swirl combustor. This configuration of combustor uses part of RQL and LP concepts to control NOx emissions; however, without premixing in a preliminary duct as the LP combustors and without staged air addition as the RQL combustors. The combustion zones unfavorable to NO formation are established by the flow dynamic of the reactants and the burned gases into the chamber. Basically the combustor has two stages or two combustion chambers. In the first one, the total air flow emerges in a cylindrical chamber through a swirler; consequently, an intense circular flow with high tangential velocity takes place around the primary chamber wall. The fuel is injected in the primary chamber center line, and a yellow sooting flame is observed in the region of interaction between the fuel jet and the air circular flow. The air flow absorbs the intense heat transferred by radiation from the sooting flame and acts as a natural film cooling. In the transition region for the first to the second chamber, there is an abrupt increase of the combustor diameter and part of the air from the first chamber reverts to the chamber center line, creating an intense recirculation zone which is responsible for the quick-mix between the air, reminiscent fuel and combustion products. So, in the second stage a lean premixed flame is established. It is clear that the success to control NOx emissions depends on the main controlling parameters of the flow dynamic into the combustor, which are: the intensity of the air tangential velocity in the first stage, the Reynolds number of fuel jet injection, the ratio length/diameter of the first stage and the global equivalence ratio. In spite of the promising results about low NOx emissions using this configuration of combustion chamber, the present paper is concerned about the influence of the main controlling parameters of the flow dynamic on the appearance of combustion instabilities.
Figure 1 - Schematic diagram of the experimental setup. To detect the combustion instability, a Kistler 7261 piezoelectric pressure transducer was positioned at 3.0 cm above the swirler, at the primary chamber wall. This position was chosen due to the lower wall temperature, around 350 K at the external side, facilitating the transducer cooling. Additionally, some preliminary experiments have showed that in this position the transducer can easily detect the instabilities, mainly the first mode of oscillation around 75 Hz, which is the most pronounced frequency for all experimental conditions. The piezoelectric pressure transducer signals were amplified by a Kistler 5006 charge amplifier and monitored by a Tektronix 7633 oscilloscope. The transducer specifications are: measuring range -1 to 10 bar, maximum pressure 12 bar, resolution 1.5x10-5 bar, sensitivity 2200 pC/bar. The amplitude and frequency measurements uncertain were determined by calibration. The present work transducer and a standard transducer are coupled into a chamber fed by a pulsed flow, and results are compared to different frequencies and amplitude. The uncertain is calculated for the point of 95% probability in “Students t” curve, and for the results presented here the maximum error for frequency is 1% and for amplitude 5% of the measure. The acquisition rate was 3200 acquisitions per second and the analysis is done based on the Fourier transform. The parameters changed during the experiments were: L/D ratio for the primary chamber, fuel jet Reynolds number (Re), swirler blades angle (α) and the equivalence ratio (φ). As commented before, two different regions of combustion are observed into the combustor. Into de primary chamber, a cylindrical sooting flame is observed and after a quick transition, an intense lean blue flame is present in the secondary chamber. To illustrate it, Fig. 2 shows a photo where the stainless steel primary chamber is substituted by a cylindrical glass chamber and the secondary chamber is not present. The operation conditions for the photo presented in Fig. 2 are: natural gas mass flow rate mF =
Experimental Setup The experiments were performed in an atmospheric pressure laboratorial scale combustor, made in stainless steel and without refrigeration. As commented before, the air flow acts like a film cooling and the refrigeration is not necessary. The primary chamber diameter (D) is 10 cm and the length (L) may be 10, 20 or 30 cm (L/D ratio = 1, 2 or 3). The secondary chamber diameter and length are 20 and 50 cm, respectively. The air, which comes from air blowers, is conducted axially to the swirler positioned at the primary chamber entry. The swirler diameter is the same of the primary chamber and it has eight blades whose angle with the axial direction may be set from 0o to 80o. The Fig. 1 shows a schematic diagram of the experimental setup. The mass flow rate of the natural gas was kept constant at 1 g/s for all experiments, and maximum air mass flow rate was 100 g/s. Both mass flow rates were measured by calibrated orifice plate systems and the maximum error is 3% of the measure. The fuel was
2
1.0 g/s, air mass flow rate mair = 50 g/s, swirler blades angle α = 80o and L/D = 1.
values for the tests performed. It was observed that changes in L/D, Re, α and φ have minor impact on the frequency (note the standard deviation also presented in the legends). In Fig. 4 the results show that most of instabilities happen for global equivalence ratios less than 0.3. The appearance of instabilities in quite low global equivalence ratios agrees with other studies such as Bradley et al. (1998), Lieuwen et al. (1998) and Cohen and Anderson (1996). In extremely lean combustion conditions, as expected for the secondary chamber in the present experiment, fluctuations in the equivalence ratio lead to spatial and temporal fluctuations of energy release which can induce pressure oscillations, especially when the flame is close to its limit of "blowout". In lean mixtures, the flame propagation velocity is quite low, which does not allow a recovery of energy release when fluctuations in equivalent ratio are present (Stone e Menon, 2002).
Figure 2 – Double-stage swirl flame. Results First of all, to analyze the origin of the combustion instabilities, the experimental conditions L/D = 2, Re = 40.000, α = 60o and several global equivalence ratios were investigated performing three different experiments: 1) combustion with the whole experimental setup (primary and secondary chambers); 2) combustion without the secondary chamber; 3) no combustion or non-reactive flow through the whole setup. The results are summarized in Fig. 3. The strong oscillations detected on the combustion with whole setup and for the global equivalence ratios less than 0.23, do not happen for the non-reactive flow and for the combustion without the secondary chamber. Then, it suggests that the origin of oscillations is the interactions between the energy released on the lean combustion into the secondary chamber, the structure of the reactive flow and the combustor acoustic characteristics.
15 10 5
Amplitude (mbar)
0
0.45
Amplitude (mbar)
0.35
0.40
0.45
0.50
L/D = 1 Re = 40,000 F = 79.82 Hz ± 9.07 Hz
15 10 5 0 0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
L/D = 1 Re = 15,000 F = 76.625 Hz ± 13.688 Hz
0 0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Equivalence Ratio (φ)
Figure 4 – Results for L/D = 1. Lieuwen et al. (1998) suggested that deficiencies on the reactants mixing process for lean global combustion produce different local equivalent ratios in the flame zone, and this distribution of equivalence ratio plays an important role on the mechanism that induces the appearance of combustion instabilities. Bradley et al (1998) performed experiments for methane/air flames stabilized by a swirler, keeping constant the swirl number and the axial velocity of injection, while the equivalence ratio was gradually reduced. They observed low frequency acoustics instabilities for global equivalence ratios less than 0.6. However, for equivalence ratios higher, the flame remained stable. Thus, in general, many studies have shown that lean combustion tends to present combustion instabilities and the main reason is the non-homogeneity of the equivalence ratio distribution through the reaction zone. For the present experiments, the parameters that control the dynamic of fuel and air flows (α, Re, L/D and φ), also control the mixing process in the secondary chamber between the reminiscent air and the partial combustion products from the primary chamber. The
0 0.40
0.30
5
5
0.35
0.25
10
10
0.30
0.20
15
Only primary chamber combustion No Combustion with the whole set Combustion with the whole set
0.25
0.15
20
20
15
0.20
25
25
L/D = 2 Re = 40,000 α = 60°
0.15
L/D = 1 Re = 50,000 F = 83.81 Hz ±7.01 Hz
20
25
20
40° 50° 60° 70° 80°
25
0.50
Equivalence Ratio (φ)
Figure 3 – Results for L//D = 2, Re = 40,000, α = 60o and several global equivalence ratio. Fig. 4 shows the results for L/D = 1, Re = 50,000, 40,000 and 15,000, α = 40o to 80o and several φ´s. The frequencies presented in the figures legend are the mean
3
and the fuel jet Reynolds number on the instabilities appearance. However, for the ratio L/D = 3 it can be clearly seen the strong attenuation of the amplitude, compared to L/D = 1 and 2. Except for the case Re = 50,000 in the Fig. 6, for the other two lower Reynolds number, the amplitudes were less than 5 mbar. For longer primary chamber, the jet axial momentum in the chambers transition region reduces, which facilitates the mixing process between the partial combustion gases and reminiscent air in the recirculation zone and, as commented before, previous works have pointed that the most homogeneous mixtures reduce the possibility of combustion instabilities. The combination L/D = 3 and Re = 15,000, Fig. 6, was the best condition analyzed in terms of preventing the instability induction. Independent on the swirl angle and the equivalence ratio, the higher primary chamber length and smaller Reynolds number studied here, provide combustion regimes in the secondary chamber unfavorable for instabilities occurrence. This fact makes clear the importance of the jet axial momentum in the transition region to the dynamic of the mixing process in the recirculation zone, resulting in the combustion instabilities occurrence.
recirculation zone in the secondary chamber is responsible for providing the lean mixture; increasing its intensity, also increase the trend of more homogeneous lean combustion. It can be seen in Fig. 4, that for higher swirl angles, for example α = 80o, there is an attenuation of the oscillations. The possible explanation for this behavior is that the intensity of the recirculation zone has an exponential dependence on the swirler angle (Couto et al., 1995); thus, the higher swirler angle intensify the mixing process and reduce the presence of non-uniformity in the lean combustion region, damping the oscillations. The recirculation zone intensification leading to more homogeneous mixtures and, consequently, reducing the trend of instabilities is reported in previous works. Lovett and Abuaf (1992) compared the flames established by swirlers with others established by “bluff bodies”, and in the first case the oscillations were more tenuous. The results of Tangirala et al. (1987) have shown that the mixture homogeneity and the combustion stability can be improved by increasing the swirl number. The experimental study of Broad el al. (1998) in a lean-premixed combustor, confirmed the oscillations attenuation when the swirl increases. In addition, stronger recirculation zone increases the flame velocity due to higher turbulence, and partially compensates the flame velocity reduction when the combustion becomes leaner. As a result, there is a faster recuperation of energy release fluctuations, inducing less pressure oscillation. Fig. 4 also shows the tendency of reduction or damping the oscillation when the combustion becomes close to the experimental limit of global equivalence ratio 0.15. In spite of the favoritism of leaner combustion to instabilities, in the case of swirler flames, the increase of air mass flow rate also increases the intensity of the recirculation zone. The behavior presented in Fig. 4 and confirmed in Fig. 5 and 6 for L/D = 2 and 3, respectively, pointed that leaning the global combustion, increasing the air mass flow rate and keeping α and Re constant, can induce the instabilities; however, in extreme lean global combustion, there is a tendency of damping the oscillations, probably as result of the intensification of the recirculation flow in the secondary chamber. Other point is the influence of the fuel jet Reynolds number. Comparing the results for 50,000, 40,000 and 15,000 in the Fig. 4, the reduction of Reynolds number to 15,000 attenuated the amplitude. A possible explanation for this fact is related to jet axial momentum in the transition region between the primary and secondary chambers. Thus, higher jet Reynolds number will also provide higher axial momentum in the transition region, which tends to weaken the recirculation zone in the secondary chamber, favoring the instabilities appearance. The Fig.5 and 6 show the results for the experiments with L/D = 2 and 3, respectively. Qualitatively the same trends are observed for the influence of the swirl angle, the global equivalence ratio
25
40° 50° 60° 70° 80°
L/D = 2 Re = 50,000 F = 80.45 Hz ± 4.74 Hz
20 15 10 5
Amplitude (mbar)
0 25
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
L/D = 2 Re = 40,000 F = 79.72 Hz ± 7.41 Hz
20 15 10 5 0 25
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
L/D = 2 Re = 15,000 F = 80.26 Hz ± 6.62 Hz
20 15 10 5 0 0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
Equivalence Ratio (φ)
Figure 5 – Results for L/D = 2.
40° 50° 60° 70° 80°
25 L/D = 3 Re = 50,000 F = 74.37 Hz ± 6.35 Hz
20 15 10 5
Amplitude (mbar)
0 25
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
L/D = 3 Re = 40,000 F = 72.94 Hz ± 8.02 Hz
20 15 10 5 0 25
0.15
0.20
0.25
0.30
0.35
20
0.40
0.45
0.50
L/D = 3 Re = 15,000 F = 76.15 Hz ± 7.84 Hz
15 10 5 0 0.15
0.20
0.25
0.30
0.35
0.40
Equivalence Ratio (φ)
Figure 6 – Results for L/D = 3.
4
0.45
0.50
tests were carried out at the Professor Feng Aeronautical Laboratory, Aeronautical Technology Institute.
Results The present paper investigated experimentally the influence of the swirler angle, the fuel jet Reynolds number, the primary chamber length/diameter ratio and the global equivalence ratio on the appearance of combustion instabilities in a laboratorial scale doublestage swirl combustor. It was observed that the global equivalence ratio has influence on the mechanisms that induces the instabilities. Combustor operations with quite lean global condition can be easily induced to combustion instabilities. As well as reported in the literature, lean combustion associated with non-homogeneous distribution of equivalence ratio through the reaction zone causes fluctuations of energy release which can induce pressure fluctuations and, consequently, combustion instabilities. However, the results presented here showed that the instabilities attenuation or suppression is possible by choosing suitable values for the swirler angle, fuel jet Reynolds number and primary chamber length/diameter ratio. Higher swirler angle, 70o and 80o, for example, tends to attenuate the oscillations, due to the intensification of the recirculation zone, enhancing the reactants mixing process and, in addition, increasing the flame velocity due to the higher turbulence. On the other hand, the jet Reynolds number and the primary chamber length play an important role on the axial momentum in the transition region between the primary and secondary chambers. Higher axial momentum weakens the recirculation flow in the secondary chamber and increases the possibility of nonhomogenous equivalence ratio in the reaction zone. By reducing the jet Reynolds number and increasing the primary chamber length the oscillations were attenuated. The results presented here have pointed some observations that may be taken into account for the design of double-stage swirl combustor for gas turbine applications; however, these observations must be confronted with other important design requirements, for example, low total pressure drop, homogeneous temperature profile in the combustor exit, flammability for the whole engine operation map and, of course, low pollutants emissions. Additional theoretical and experimental works must be done for the better understanding about the interaction between the flows in the primary and secondary chambers with thermalacoustic excitation, for example, using OH – PLIF images for the secondary chamber or PIV analysis for the transition region, but the present work provided which are the important parameters that must be investigated in future works.
References Bradley, D., Gaskell, P.H., Gu, X.J., Lames, M., and Scott, M.J., 1998, “Premixed Turbulent Flame Instability and NO Formation in a Lean-Burn Swirl Burner”, Combustion and Flame, 115, pp. 515-538. Canepa, E., Di Martino, P., Formosa, P., Ubaldi, M., and Zunino, P., 2006, “Unsteady Aerodynamics of an Aeroengine Double Swirler Leas Premixing Prevaporizing Burner”, Journal of Engineering for Gas Turbine and Power, 128, pp. 29-39. Cohen, J.C., and Anderson, T, 1996, “Experimental Investigation of Near-Blowout Instabilities in a Lean, Premixed Step Combustion”, In: 34 th Aerospace Sciences Meeting and Exhibit, jan. 15-18, 1996, Reno, Nevada. AIAA paper # 96-0819. Couto, H.S., Muniz, W.F., and Bastos-Netto, D., 1995, “Geometrical Parameters for Flows Across Axial Swilers”, Proceedings of the 3rd Asian-Pacific International Symposium on Combustion and Energy Utilization, 1, p. 255-260. Kalb, J.R., and Sattelmayer, T., 2006, “Lean Blowout Limit and NOx Production of a Premixed Subppm NOx Burner with Periodic Recirculation of Combustion Products”, Journal of Engineering for Gas Turbine and Power, 128, pp. 247-254. Lieuwen, T., Neumeier, Y., and Zinn, B.T., 1998, “The Role of Unmixedness and Chemical Kinetics in Driving Combustion Instabilities in Lean Premixed Combustors”, Combustion Science and Technology, 135, pp. 193-211. Lovett, J.A., and Abuaf, N., 1992, “Emissions and Stability Characteristics of Flameholders for Lean Premixed Combustion”, ASME paper # 92 – GT- 120. Sarv, H., and Cernansky, N. P., 1989, ‘‘NOx Formation From the Combustion of Monodisperse nHeptane Sprays Doped With Fuel-Nitrogen Additives”, Combustion and Flame, 76, pp. 265–275. Smith, L.L., Karim, H., Castaldi, M.J., Etemad S., Pfefferle, W.C., Khanna, V., and Smith K.O., 2005, “ Rich-Catalytic Lean-Burn Combustion for Low-SingleDigit NOx Gas Turbines”, Journal of Engineering for Gas Turbine and Power, 127, pp. 27-35. Stone, C., and Menon, S., 2002, “Swirl Control of Combustion Instabilities in a Gas Turbine Combustion”, Proceedings of Combustion Institute, 29, pp. 155-160. Straub, D.L., Castelon, K.H., Lewis, R.E., Sidwell, T. G., Maloney, D.J., and Richards, G.A., 2005, “Assessment of Rich-Burn, Quick-Mix, Lean Burner Trapped Vortex Combustor for Stationary Gas Turbines”, Journal of Engineering for Gas Turbine and Power, 127, pp. 36-41. Tangirala, V., Chen, R.H., and Driscoll, J.F., 1987, “Effect of Heat Release and Swirl on the Recirculation within Swirl-Stabilized Flames”, Combustion Science and Technology, 51, pp. 75-95.
Acknowledgement This work was sponsored by the Brazilian National Research Council - CNPq (research resource and grant of Mr. Lacava) and by The State of Maranhão Research Foundation – FAPEMA (grant of Mr. Almeida). The
5
