Experimental Investigation of the Stabilization and Structure of Turbulent Cool Diffusion Flames

Size: px

Start display at page:

Download "Experimental Investigation of the Stabilization and Structure of Turbulent Cool Diffusion Flames"

Rosanna Gibson
5 years ago
Views:

AIAA SciTech Forum 8 12 January 218, Kissimmee, Florida 218 AIAA Aerospace Sciences Meeting 1.2514/6.

1 AIAA SciTech Forum 8 12 January 218, Kissimmee, Florida 218 AIAA Aerospace Sciences Meeting / Experimental Investigation of the Stabilization and Structure of Turbulent Cool Diffusion Flames Christopher B. Reuter 1 and Omar R. Yehia 2 Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ 854 Sang Hee Won 3 Department of Mechanical Engineering, University of South Carolina, Columbia, SC 2928 and Matthew K. Fu, 4 Katherine Kokmanian, 5 Marcus Hultmark, 6 and Yiguang Ju 7 Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ 854 Although recent studies of laminar cool flames have provided important advances in understanding the low-temperature chemistry of both hydrocarbons and oxygenates, there has been limited experimental insight into how interactions between turbulence and chemistry occur in cool flames. To address this, a new Co-flow Axisymmetric Reactor- Assisted Turbulent (CARAT) burner has been developed and characterized in this investigation for the purpose of directly studying turbulent cool flames. A methodology for establishing stable turbulent cool diffusion flames under well-defined conditions is proposed. The structure of dimethyl ether flames is examined using both formaldehyde planar laserinduced fluorescence and Rayleigh scattering. It is found that weak turbulence produces wrinkled turbulent cool flames in which fluctuations occur mainly on the fuel side of the flame. However, at increased levels of turbulence, large pockets of unburned reactants appear in the vicinity of the cool flame, and structural fluctuations extend to both sides of the flame. This study offers a well-defined experimental platform for the study of turbulence-chemistry interactions at low temperatures. I. Introduction VER the past several years, experimental investigations of cool flame phenomena have proliferated. Cool Oflames have been studied under well-defined conditions in microgravity droplets 1, 2, counterflow burners 3-5, micro flow reactors 6, and Hencken burners 7. The results of these studies have shown that the low-temperature chain-branching sequence (peroxy chemistry) present in many hydrocarbons and oxygenates is capable of producing enough radicals and heat release to sustain stable cool flames across a variety of conditions. These experiments have also provided important validation targets for the development of chemical kinetic models at low temperatures 8. However, despite the ubiquity of turbulent combustion in real engines, well-defined turbulent cool flame experiments have been extremely limited in number. Gokalp 9 and Koliatis 1 studied premixed cool flames within reactors subject to various degrees of turbulence. Some of the authors previous experiments 11, 12 involved turbulent premixed flames that were strongly affected by low-temperature ignition, but no cool flame fronts were actually visible in those experiments. Recent numerical simulations have shown that cool flames can play an important role in the ignition 13-15, stabilization 16, and propagation 17 of turbulent flames. However, numerical prediction of cool flame behavior is extremely sensitive to the chemical kinetic model employed, and many widely used models that 1 Graduate Student, AIAA Student Member 2 Graduate Student, AIAA Student Member 3 Associate Professor, AIAA Member 4 Graduate Student 5 Graduate Student 6 Assistant Professor 7 Robert Porter Patterson Professor, AIAA Associate Fellow 1 Copyright 218 by Christopher B. Reuter. Published by the, Inc., with permission.

Turbulence Intensity, u'/u Integral Length Scale, l (mm) perform well for high-temperature flames are incapable of describing cool flames accurately 8.

2 Turbulence Intensity, u'/u Integral Length Scale, l (mm) perform well for high-temperature flames are incapable of describing cool flames accurately 8. Therefore, the experimental measurement of turbulent cool flames can provide valuable insights into the modeling of turbulencechemistry interactions at low temperatures. In the present study, a new Co-flow Axisymmetric Reactor-Assisted Turbulent (CARAT) burner is employed for the investigation of turbulent cool diffusion flames. Dimethyl ether (DME) is chosen as the fuel due to its relatively strong low-temperature chemistry and known Rayleigh scattering cross-section 18. The two-dimensional turbulent cool diffusion flame structure is examined through the formaldehyde planar laser-induced fluorescence technique, while one-dimensional radial temperature profiles are produced using Rayleigh scattering. The effects of increases in turbulence on the cool diffusion flame structure are also discussed. II. Experiment The Co-flow Axisymmetric Reactor-Assisted Turbulent (CARAT) burner, depicted in Figure 1, is a generational update to the Reactor-Assisted Turbulent Slot (RATS) burner 11, 12. Like the RATS burner, the CARAT burner is a Bunsen-type burner with an extended reactor section capable of controlling the pre-flame residence time and temperature. Two grids located in the reactor section generate turbulence within the main flow. The burner main exit is 15 mm in diameter (D). A small annulus (2-mm gap size) surrounding the main exit enables the use of a pilot flow. The CARAT burner also includes a large exterior channel that enables a heated or vitiated co-flow, similar to the well-known Cabra burner 19. The co-flow passes through a ceramic honeycomb mesh, which allows for flow laminarization and well-defined boundary conditions. The turbulence in the main flow has been characterized by a nanoscale thermal anemometry probe (NSTAP) 2 in combination with a Dantec Dynamics Streamware Pro Constant Temperature Anemometry system. Due to its small size (6 μm long, 2 μm wide, 1 nm thick), the NSTAP is capable of extremely high spatial and temporal resolution. The NSTAP was positioned at the centerline of the main exit and operated at a sampling frequency of 25 khz. Figure 2 shows the range of turbulence intensities (u /U) and integral length scales (l) measured by the probe as a function of the bulk Reynolds number (Re D = UD/ν). The integral time scales (and therefore the integral length scales) were determined by extrapolating the power spectrum to zero frequency. Since the turbulent flow field characterization was performed at room temperature (295 K), the effects of elevated temperatures on u /U and l are accounted for by modifying the mixture viscosity appropriately to determine Re D (T). To visualize the turbulent cool flame structure, formaldehyde (CH 2 O) planar laser-induced fluorescence (PLIF) is used. The third harmonic (355 nm) of an Nd:YAG laser (Quantel, Q-smart 85) excites the CH 2 O at an energy of approximately 2 mj/pulse. The laser operates at a frequency of 5 Hz. Two band filters, passing frequencies between 4 and 45 nm, are positioned in front of an ICCD camera (Princeton Instruments, PI-MAX 4) to isolate the CH 2 O fluorescence. The laser beam is expanded into an approximately 2-μm-thick vertical sheet and positioned at the centerline of the burner. Rayleigh scattering is performed in order to quantify the turbulent cool flame temperature. The second harmonic (532 nm) of a 1-Hz Nd:YAG laser (Spectra-Physics, lab-17-1) is directed through the flame at an energy of approximately 4 mj/pulse. Accounting for changes in the Rayleigh scattering cross-section within the flame is often the most challenging aspect of the technique 18. However, the cool flame s tendency to exhibit massive fuel leakage eases this difficulty. Using the example of a 1-D opposed-flow flame 21, 22, it can be seen in Fig. 3 that an Fig. 1. Photograph of the Co-flow Axisymmetric Reactor-Assisted Turbulent (CARAT) burner used in this study T main = 295 K u'/u l Bulk Reynolds Number, Re D Fig. 2. Results of the CARAT burner turbulent flow field characterization in terms of the turbulence intensity (u /U) and integral length scale (l)

3 unburning mixture with boundary temperatures of T = K and a cool flame with boundary temperatures of T = 6 K have nearly the same mixture properties. The presence of a hot flame, on the other hand, causes significant changes which must be accounted for through extensive corrections of the Rayleigh signal. Therefore, in this study, the cool flame temperature is determined by directly comparing its Rayleigh scattering signal to an unburning flow field with identical mixture properties at a known temperature. III. Stabilization of Turbulent Cool Diffusion Flames In this study, turbulent cool diffusion flames are established using dimethyl ether (DME) as the fuel. Under the conditions given in Fig. 4, cool flames are able to be stabilized in the recirculation zone between the main flow (DME/N 2 ) and the pilot flow (O 2 ). The range of conditions for which a cool flame can be stabilized is highly temperature dependent. Specifically, DME turbulent cool diffusion flames were seen to form readily for T = 6 K boundary temperatures but not for T = 55 K or T = K. The use of pure oxygen in the pilot flow extends the range of conditions for which turbulent cool diffusion flames can be stabilized but is not strictly necessary, as cool flames were also able to be formed with air in the pilot flow. Notably, increasing the main flow velocity (U) or decreasing the DME mole fraction (X DME ) does not result in a sharp transition between stable turbulent cool diffusion flames and blowoff. Rather, it was seen that a range of conditions exists for which lifted turbulent cool diffusion flames result, some of which transiently reach blowout over the course of a several minutes. For all measurements in this study, the pilot flow velocity is 1. m/s, and the co-flow velocity is approximately.55 m/s Distance (cm) Fig. 3. Computed mixture properties for hot flames, cool flames, and unreacting mixtures in a laminar counterflow configuration. IV. Flame Structure Measurements through CH 2 O PLIF Due to its unique chemistry, DME is an extremely attractive fuel for the CH 2 O PLIF technique. Multiple reaction pathways are available for DME to decompose into CH 2 O, and as a result the signal-to-noise ratio for CH 2 O PLIF of DME flames is typically quite high. DME cool flames offer even stronger CH 2 O PLIF signals than DME hot flames 23, as the lack of consumption pathways makes CH 2 O one of the main low-temperature oxidation products. Figure 5 shows the DME turbulent cool diffusion flame structure for a main flow velocity of U = 1.5 m/s and a DME mole fraction of X DME =.2. This condition corresponds to a turbulence intensity of u /U =.16. All three streams (main, pilot, co-flow) exit the burner at a temperature of T = 6 K (as elucidated in Fig. 4). The turbulent cool flame is clearly wrinkled but unbroken (Fig. 5a), with the majority of the wrinkling occurring on the fuel side of the flame. The higher likelihood of fuel-side fluctuations is confirmed quantitatively by the values of the standard deviation of the CH 2 O signal in Fig. 5c. When the main flow velocity is increased to U = 2.75 m/s, pockets of unburned reactants and flame islands become much more common (Fig. 6a). The strength of the average CH 2 O signal (Fig. 6b) decreases noticeably compared to the U = 1.5 m/s case, especially in the regions closest to the burner exit. Fig. 6c also reveals that the standard deviation of the CH 2 O signal generally decreases on the fuel side of the flame. However, fluctuations are seen to substantially increase on the oxidizer side, particularly for the regions furthest from the burner exit. Thus, at Mixture Molecular Weight (g/mol) N 2 /DME vs O 2 a = 1 s -1 X DME =.2 O 2 pilot flow at 6 K Ceramic honeycomb mesh Fuel/N 2 main flow at 6 K Hot MW (T=6 K) Cool MW (T=6 K) Mix MW (T= K) Hot DME (T=6 K) Cool DME (T=6 K) Mix DME (T= K) Laser sheet Fig. 4. Experimental schematic for producing turbulent cool diffusion flames in the CARAT burner DME Mole Fraction Air co-flow at 6 K Turbulencegenerating grid 3

18 16 14 12 8 6 4 (a) (b) (c) D D D Fig. 5. CH 2 O PLIF measurements for DME turbulent cool diffusion flames at U = 1.5 m/s and X DME =.2. Instantaneous values are shown in (a), mean values in (b), and standard deviations in (c).

4 (a) (b) (c) D D D Fig. 5. CH 2 O PLIF measurements for DME turbulent cool diffusion flames at U = 1.5 m/s and X DME =.2. Instantaneous values are shown in (a), mean values in (b), and standard deviations in (c). increased levels of turbulence, two distinct maxima in the fluctuating values occur, which resembles measurements of DME turbulent hot diffusion flames 24. V. Flame Structure Measurements through Rayleigh Scattering Figure 7 shows the DME turbulent cool diffusion flame temperature distribution at x/r = 2.2 for a main flow velocity of U = 1.5 m/s and a DME mole fraction of X DME =.2 (identical conditions to Fig. 5). As alluded to previously, the cool flame temperatures are determined by comparing the raw Rayleigh scattering signal at T = 6 K (cool flame) to the raw signals from a T = K unburning case and a T = 55 K unburning case. Besides the exit temperatures, all three cases have identical boundary conditions (X DME, U, U pilot, etc.). The temperature of the central N 2 /DME stream at an axial location of x/r = 2.2 is determined to be T 595 K, which is quite close to the value of T = 6 K measured by thermocouple at x/r =. At x/r = 2.2, the maximum mean temperatures occur near r/r ±.9. The maximum temperature is slightly asymmetric, peaking at T 725 K on left side and T 7 K on right side (a) (b) (c) 2 D D D Fig. 6. CH 2 O PLIF measurements for DME turbulent cool diffusion flames at U = 2.75 m/s and X DME =.2. Instantaneous values are shown in (a), mean values in (b), and standard deviations in (c). 4

5 Mean Temperature (K) STDEV Temperature (K) Like the CH 2 O PLIF measurements at the same condition (Fig. 5), the maximum standard deviation in the temperature (T 9 K) occurs closer to fuel side (r/r ±.8) than the maximum mean temperature. Interestingly, the maximum temperature fluctuations in Fig. 7 are very close to the values obtained from thermocouple measurements in an n-heptane turbulent cool premixed flame 9. In this study, the minimum temperature fluctuations (T 2 K) occur in the unburning regions away from the flame. Together, these measurements give T /T.3 in the unburning regions and up to T /T.13 in the cool flame itself. VI. Conclusion A new Co-flow Axisymmetric Reactor-Assisted Turbulent (CARAT) burner has been utilized to study r/r Fig. 7. Measured mean and standard deviation of temperature for a U = 1.5 m/s DME turbulent cool diffusion flame at x/r = 2.2. turbulent cool diffusion flames. Measurements with a nanoscale thermal anemometry probe (NSTAP) reveal that the burner is capable of producing turbulence intensities from.16 < u /U <.2 over the range of conditions studied. By heating the co-flow (air), pilot (oxygen), and main (nitrogen/dimethyl ether) streams to T = 6 K, turbulent cool diffusion flames are able to be established over a range of conditions. Formaldehyde planar laserinduced fluorescence measurements indicate that weak turbulence causes wrinkling to occur primarily on the fuel side of the cool flame. However, increased turbulence is seen to produce significant fluctuations on both sides of the flame. Measurements of the cool flame temperature, obtained through Rayleigh scattering, confirm these structural observations for weak turbulence. The experimental methodology described in this study provides a well-defined platform for the study of turbulence-chemistry interactions at low temperatures. Acknowledgments The authors recognize the support of the NETL UTSR Program under DOE Award No. DE-FE CBR and MKF are supported by the NDSEG fellowship program of the U.S. Department of Defense. ORY is grateful for support from the Daniel and Florence Guggenheim Foundation Fellowship at Princeton University. References 1 Nayagam, V., Dietrich, D. L., Ferkul, P. V., Hicks, M. C., and Williams, F. A. "Can cool flames support quasi-steady alkane droplet burning?," Combustion and Flame Vol. 159, No. 12, 212, pp Farouk, T. I., Hicks, M. C., and Dryer, F. L. "Multistage oscillatory Cool Flame behavior for isolated alkane droplet combustion in elevated pressure microgravity condition," Proceedings of the Combustion Institute Vol. 35, No. 2, 215, pp Won, S. H., Jiang, B., Diévart, P., Sohn, C. H., and Ju, Y. "Self-sustaining n-heptane cool diffusion flames activated by ozone," Proceedings of the Combustion Institute Vol. 35, No. 1, 215, pp Reuter, C. B., Won, S. H., and Ju, Y. "Experimental study of the dynamics and structure of self-sustaining premixed cool flames using a counterflow burner," Combustion and Flame Vol. 166, 216, pp Deng, S., Han, D., and Law, C. K. "Ignition and extinction of strained nonpremixed cool flames at elevated pressures," Combustion and Flame Vol. 176, 217, pp Kikui, S., Kamada, T., Nakamura, H., Tezuka, T., Hasegawa, S., and Maruta, K. "Characteristics of n-butane weak flames at elevated pressures in a micro flow reactor with a controlled temperature profile," Proceedings of the Combustion Institute Vol. 35, No. 3, 215, pp Hajilou, M., Ombrello, T., Won, S. H., and Belmont, E. "Experimental and numerical characterization of freely propagating ozone-activated dimethyl ether cool flames," Combustion and Flame Vol. 176, 217, pp Reuter, C. B., Lee, M., Won, S. H., and Ju, Y. "Study of the low-temperature reactivity of large n-alkanes through cool diffusion flame extinction," Combustion and Flame Vol. 179, 217, pp Gökalp, I., Dumas, G. M. L., and Ben-Aïm, R. I. "Temperature field measurements in a premixed turbulent cool flame," Symposium (International) on Combustion Vol. 18, No. 1, 1981, pp Kolaitis, D. I., and Founti, M. A. "A tabulated chemistry approach for numerical modeling of diesel spray evaporation in a stabilized cool flame environment," Combustion and Flame Vol. 145, No. 1, 26, pp Won, S. H., Windom, B., Jiang, B., and Ju, Y. "The role of low temperature fuel chemistry on turbulent flame propagation," Combustion and Flame Vol. 161, No. 2, 214, pp X DME =.2 U = 1.5 m/s u'/u =.16 x/r = 2.2 Mean STDEV

6 12 Windom, B., Won, S. H., Reuter, C. B., Jiang, B., Ju, Y., Hammack, S., Ombrello, T., and Carter, C. "Study of ignition chemistry on turbulent premixed flames of n-heptane/air by using a reactor assisted turbulent slot burner," Combustion and Flame Vol. 169, 216, pp Krisman, A., Hawkes, E. R., Talei, M., Bhagatwala, A., and Chen, J. H. "A direct numerical simulation of cool-flame affected autoignition in diesel engine-relevant conditions," Proceedings of the Combustion Institute Vol. 36, 217, pp Echekki, T., and Ahmed, S. F. "Turbulence effects on the autoignition of DME in a turbulent co-flowing jet," Combustion and Flame Vol. 178, 217, pp Dahms, R. N., Paczko, G. A., Skeen, S. A., and Pickett, L. M. "Understanding the ignition mechanism of high-pressure spray flames," Proceedings of the Combustion Institute Vol. 36, No. 2, 217, pp Krisman, A., Hawkes, E. R., Talei, M., Bhagatwala, A., and Chen, J. H. "Characterisation of two-stage ignition in diesel engine-relevant thermochemical conditions using direct numerical simulation," Combustion and Flame Vol. 172, 216, pp Minamoto, Y., and Chen, J. H. "DNS of a turbulent lifted DME jet flame," Combustion and Flame Vol. 169, 216, pp Fuest, F., Barlow, R. S., Chen, J.-Y., and Dreizler, A. "Raman/Rayleigh scattering and CO-LIF measurements in laminar and turbulent jet flames of dimethyl ether," Combustion and Flame Vol. 159, No. 8, 212, pp Cabra, R., Myhrvold, T., Chen, J. Y., Dibble, R. W., Karpetis, A. N., and Barlow, R. S. "Simultaneous laser raman-rayleighlif measurements and numerical modeling results of a lifted turbulent H2/N2 jet flame in a vitiated coflow," Proceedings of the Combustion Institute Vol. 29, No. 2, 22, pp Bailey, S. C. C., Kunkel, G. J., Hultmark, M., Vallikivi, M., Hill, J. P., Meyer, K. A., Tsay, C., Arnold, C. B., and Smits, A. J. "Turbulence measurements using a nanoscale thermal anemometry probe," Journal of Fluid Mechanics Vol. 663, 21, pp Lutz, A. E., Kee, R. J., Grcar, J. F., and Rupley, F. M. "OPPDIF: A Fortran program for computing opposed-flow diffusion flames," Sandia National Laboratories Report SAND , Yang, X., Shen, X., Santner, J., Zhao, H., and Ju, Y., Princeton HP-Mech. Mech.html, Reuter, C. B., Won, S. H., and Ju, Y. "Flame structure and ignition limit of partially premixed cool flames in a counterflow burner," Proceedings of the Combustion Institute Vol. 36, 217, pp Gabet, K. N., Shen, H., Patton, R. A., Fuest, F., and Sutton, J. A. "A comparison of turbulent dimethyl ether and methane non-premixed flame structure," Proceedings of the Combustion Institute Vol. 34, No. 1, 213, pp

Erratum to: High speed mixture fraction and temperature imaging of pulsed, turbulent fuel jets auto igniting in high temperature, vitiated co flows

$Erratum to: High speed mixture fraction and temperature imaging of pulsed, turbulent fuel jets auto igniting in high temperature, vitiated co flows$ DOI 10.1007/s00348-015-2101-9 ERRATUM Erratum to: High speed mixture fraction and temperature imaging of pulsed, turbulent fuel jets auto igniting in high temperature, vitiated co flows Michael J. Papageorge