A novel plasma heater for auto-ignition studies of turbulent non-premixed flows

Abstract

In this paper, the development and characterization of a novel test rig for auto-ignition (AI) studies of a fuel jet propagating into a hot turbulent co-flow is reported. The test rig, based on microwave plasma heating, is capable of achieving co-flow temperatures up to 1300 K and velocities up to 40 $\hbox {ms}^{-1}$. Important boundary conditions at nozzle exit such as temperature, species, and velocity field were determined to prove the capabilities and limitations of the test rig. Liftoff height (LOH) measurements of $\hbox {CH}_4$, $\hbox {C}_2\hbox {H}_4$, and $\hbox {CH}_{4}/\hbox {H}_{2}$ jets, propagating into a turbulent heated air co-flow, were taken using chemiluminescence imaging. Effects of the temperature and Reynolds number (Re) of co-flow and jet were also studied. Results showed that the flame stabilization mechanism is supported substantially by AI rather than pure flame propagation. While the co-flow temperature dominates the AI process, the Re and temperature of the jet just have a small impact on the LOH.

Appendices

Appendix 1: Co-flow temperature profiles

See Fig. 11.

Fig. 11

Normalized mean temperatures of the co-flow for $r= 0,\,20,\,30,\,\hbox{and}\,40$ mm at different axial locations

Appendix 2: $\hbox {N}_2\hbox {O}$ spectroscopic measurement

The in situ spectrometric measurement of $\hbox {N}_2\hbox {O}$ concentration was based on a distributed-feedback continuous-wave quantum cascade laser (QCL) with single mode emission at $4.55\,\upmu \hbox {m}$. In order to match the R branch of the fundamental absorption of $\hbox {N}_2\hbox {O}$, the QCL wavenumber was varied from 2192 to $2198 \,\hbox {cm}^{-1}$ by applying laser heat sink temperatures from $-5$ to $-30\,^{\circ }\hbox {C}$. The QCL emission was fine-tuned via a triangular current modulation at 9.37 kHz. The modulation frequency was chosen that high due to intense beam steering disturbances in the hot turbulent co-flow stream. At the test rig, the QCL was placed at a distance of 40 cm beyond the outlet, perpendicularly to the flow direction, and it was shielded with ceramic insulation mats to prevent excessive heat transfer to the laser. For the measurement of $\hbox {N}_2\hbox {O}$ concentration, the collimated light beam was passed through the co-flow stream seven times by two additional gold mirrors (25.4 mm dia.), resulting in an absorption path length of 559 mm. The beam was finally focused using an off-axis parabolic mirror on a heat-shielded, liquid nitrogen-cooled $0.25\, \hbox {mm}^2$ HgCdTe detector (Kolmar Technologies, 50 MHz bandwidth). The signal was amplified with a low-noise transimpedance amplifier, digitized, and saved.

Resulting wavelength scans were analyzed with a LabVIEW-based algorithm, which fits a polynomial function to the data, including a term for time-dependent broadband transmission and thermal background radiation to yield an accurate baseline. The exact wavelength tuning of the scanned QCL was measured with a 25.4-mm germanium etalon, while the absolute position was verified with a Fourier transform spectrometer. The absolute gas concentration was calculated by integrating the absorbance spectrum using the extended Lambert–Beer law with a multiline Voigt fit and spectroscopic parameters from HITRAN (Rothman et al. 2013). The detection limit $\varDelta x_\mathrm{mol}$ was defined as:

$$\varDelta x_\mathrm{mol} = x_\mathrm{molec} \cdot \frac{\varDelta A_\mathrm{residual}}{A_\mathrm{peak}(\nu _0)}$$

(4)

where $A_\mathrm{peak}(\nu _0)$ describes the peak absorption at the line center, $\varDelta A_\mathrm{residual}$ the 1$\sigma$ residual of the fit over the complete tuning range, and $x_\mathrm{molec}$ the absolute mole fraction of the molecule. For the given boundary conditions (path length and acquisition rate), the detection limit of the spectrometer was estimated to 8.5 ppm using a $\hbox {N}_2\hbox {O}$ gas cell at controlled conditions of pressure, temperature, and $\hbox {N}_2\hbox {O}$ mole fraction.

Appendix 3: OH-PLIF measurement

For the detection of OH radicals, a high-speed dye laser (Sirah Lasertechnik GmbH, Credo) was operated with Rhodamine 6G and pumped by a 10-kHz frequency-doubled Nd:YAG laser (EdgeWave GmbH, INNOSLAB CX16 II-E, 8 mJ/pulse). The dye laser radiation was tuned to 565.854 nm and frequency-doubled to 282.927 nm using an arrangement of two BBO crystals. A set of four Pellin-Broca prisms separated the UV from the fundamental beam. The UV beam had an energy of  0.3 mJ/pulse. It was expanded using a plano-concave lens ($f=-50$ mm) and two plano-convex lenses ($f=400$ mm and 600 mm) into a sheet of 30 mm height and ${\sim }300\,\upmu \hbox {m}$ thickness at the exit of the burner head.

The fluorescence OH emission was collected at 310 nm and imaged onto a high-speed intensifier (LaVision GmbH, HS-IRO) and a corresponding high-speed camera (LaVision, HSS6) using an UV lens (Halle B. Nachfl. GmbH, $f\# = 2.5$) and a high-transmission, band-pass filter (Laser Components, UV-B, $308 \pm 10\,\hbox {nm} >80\,\%$ transmission). The intensifier was gated at $100\,\upmu \hbox {s}$ to reduce the interference of chemiluminescence emission and 9900 single-shot images were recorded for each tested condition.

Equilibrium calculations using detailed chemistry in CHEMKIN-II (Kee et al. 1989) were performed to estimate the OH content of an air co-flow at 1400 K. The chemical species included N, $\hbox {N}_2$, O, $\hbox {O}_2$, $\hbox {O}_3$, $\hbox {H}_2\hbox {O}$, OH, NO, $\hbox {NO}_2$, H, $\hbox {H}_2$, $\hbox {HO}_2$, NH, $\hbox {NH}_3$, $\hbox {N}_2\hbox {H}$, and HNO. Calculations with 10,000 ppm NO, 1000 ppm $\hbox {NO}_2$, and 7000 ppm $\hbox {H}_2\hbox {O}$ resulted in an OH content of 7 ppm. To quantify the detection limit of the PLIF system, reference measurements were taken in a laminar $\hbox {CH}_4/\hbox {air}$ Bunsen flame at stoichiometric condition. The maximum OH content at this condition was calculated to 0.728 vol% using complex chemistry with 53 species (Somers 1994). The ratio between the fluorescence signal from the reference measurements to the standard deviation of the signal from the measurements in the co-flow of the test rig was 258/1. Therefore, the fraction of OH radicals in the co-flow for all the tested conditions is $\le$28 ppm.

Appendix 4: Stereo PIV measurement

The measurement of the instant velocity field of the co-flow in a plane at the exit of the burner head was taken by means of stereoscopic particle image velocimetry (S-PIV). This technique allows determination of the three components of the velocity vectors in a two-dimensional plane. Alumina particles (Albemarle Corporation, Martoxid$^{\textregistered}$ MR70) with a diameter ${\sim }1.5\,\upmu \hbox {m}$ were seeded to the fuel and co-flow streams using fluidized bed seeders. The particles were illuminated using a frequency-doubled pulsed Nd:YAG dual-cavity laser (EdgeWave GmbH, 2 INNOSLAB IS4 II-DE, 100 W at 532 nm). The repetition rate of the laser was 10 kHz, and the temporal separation between the two laser beams (PIV 1 and PIV 2) was set to $3\,\upmu \hbox {s}$. The laser beams were overlapped using a polarization beam splitter. The perpendicular-polarized beam (PIV 1) and the parallel-polarized beam (PIV 2) were converted into circular polarization using a quarter-wave plate. The beams were expanded and focused at the measurement volume into laser sheets of 35 mm height and ${\sim }1.5\, \hbox {mm}$ thickness using a plano-concave lens ($f= -50$ mm) and two plano-convex lenses ($f = 400$ and 2000 mm).

Mie scattering from the particles was detected using two high-speed CMOS cameras (Vision Research, Phantom v711, double-frame exposure) coupled with macrolenses (Sigma, 105 mm, $f\# = 2.8$) in Scheimpflug arrangement at an angle of $55^{\circ }$ with respect to the axial axis. Images (512 $\times$ 512 $\hbox {px}^2$) were recorded at 10 kHz with an aperture of $f\# = 5.6$. Short band-pass filters (Edmund Optics GmbH, 532nm) were coupled to the camera lenses to avoid the interference from diffused light and chemiluminescence. Images were postprocessed, and the corresponding velocity fields were calculated using the commercial software Davis (LaVision GmbH) and a calibration target (Type 7, LaVision GmbH).

Appendix 5: Thermal decomposition calculations and measurement of the inner surface temperature of the fuel lance

Zero-dimensional simulations of the thermal decomposition process were performed using a continuously stirred reactor (CSTR) model in Cantera 2.1 (Goodwin et al. 2014) and the GRI 3.0 reaction mechanism (Smith et al. 1997). The fuel ($\hbox {CH}_4$, $\hbox {CH}_{4}/\hbox {H}_{2}$, or $\hbox {C}_2\hbox {H}_4$) was exposed to a fixed temperature in the CSTR for a residence time of 30 ms. The selection of the residence time was based on the transit time of the fuel inside the lance during the injection for the slowest velocity case $(30\, \hbox {ms}^{-1})$ and the 700 mm section of the lance that is exposed to the hot co-flow. The onset of the thermal decomposition was found by increasing the reactor temperature by steps of 5 K starting from 600 K. Using this scheme, the beginning of the thermal decomposition was defined as the temperature where 0.1 % of the initial mole fraction of the fuel was decomposed.

The inner wall temperature of the fuel lance was measured for the case of highest fuel temperature (operational point H, $T_\mathrm{jet} = 892\,\hbox {K}$). Measurements were taken by means of phosphor thermography using the zero-dimensional luminescence lifetime method (Brübach et al. 2013) at the inner radius of the front end of the fuel lance, which is expected to be the hottest spot. As thermographic phosphor, $\hbox {LiAL}_5\hbox {O}_8{:}\,\hbox {Fe}^{3+}$ was used. The phosphor was excited with the fourth harmonic (266 nm) of a non-focused, pulsed, Q-switched Nd:YAG laser (Quanta Ray, INDI, repetition rate 10 Hz, pulse width 5–8 ns). The pulse energy density was adjusted to 1 mJ (measured after an aperture of $d = 1$ mm) by combining a half-wave plate and a Glan polarizer. For spatial filtering and for minimizing the influence of blackbody radiation and other interfering light, the emitted luminescence light was imaged onto a pinhole $(d = 100\,\upmu \hbox{m})$ by a 85-mm Nikkor camera lens. In order to enable homogeneous illumination of the photomultiplier tube (PMT Hamamatsu H6780), the signal was collimated by a plano-convex lens $(f = 30\,\hbox {mm})$ behind the pinhole. The PMT current was read out by an oscilloscope (Tektronix 5032B, 350 MHz) at an input resistance of 512 $\Omega$ and a cable capacity of $C = 300$ pF, resulting in a threshold lifetime of 153 ns. The measured decay signals of the phosphor were approximated by a mono-exponential decay and converted to scalar lifetimes, using an iterative fitting algorithm (Brübach et al. 2009), combined with a linear regression of the sum (Fuhrmann et al. 2014). The measured lifetimes were compared with a calibration curve of the thermographic phosphor to determine the surface temperature.