The confined TECFLAM swirling natural gas burner

Application Challenge 2-06 © copyright ERCOFTAC 2004

Overview of Tests

Experimental results were obtained from non-intrusive Laser based techniques and probe measurements respectively. The velocity data include the complete stress tensor providing data for comparison of various turbulence models. Beside the velocity measurements carried out by EKT Darmstadt test data include major species (CH4, H2, O2, N2, H2O, CO2 and CO) and temperature measured with RAMAN spectroscopy by DLR Stuttgart as well as 2D distributions of temperature measured with Rayleigh scattering and NO and OH concentrations obtained with PLIF carried out by PCI Heidelberg. Simultaneous temperature and concentration measurements allow to generate cross correlations (Joint PDF's) that are of particular interest for the validation of combustion models. Data were obtained for different swirl numbers and equivalence ratios. The setup of the optical devices at the combustion chamber is shown in figure 4.

Figure 4: Setup of optical measurement techniques

The consistency of the velocity and the major species concentrations was cross checked by sampling probe measurements carried out by EBI Karlsruhe. The test data include radial profiles of the velocity for several distances from the burner exit as well as 2-D distributions of main and interim species and temperature. Beside the mean and rms values of species, mixture fraction and temperature, the data of simultaneous measurements can be correlated in form of scatter plots that provide essential information for the comparison to the numerical simulation. The complete data sets including all cross correlations and the stress tensor for various operating points also including the isothermal flow fields can be obtained from the TECFLAM web page via http://www.csi.tu-darmstadt.de/institute/rsm/validierungsdaten/index.de.jsp.

Test case for Velocity measurements

Description of Experiment

A two component fibre optic LDV (Dantec) was used for the velocity measurements. A 4 W Argon Ion Laser served as light source. The beam was submitted to a transmitter box. A Bragg cell divided the incoming beam into different colours. The two beam pairs of the wavelengths lg = 514.5 nm (green) and lb = 488 nm (blue) were selected to survey the axial and the radial or circumferential velocity component, depending on the traversing direction. The first order beam of each colour was frequency shifted at 40 MHz. The incoming Laser beams (diameter 2.2 mm) were then coupled into a fibre optic probe and collimated in the measurement volume with a front lens of f1 = 600 mm focal length. A beam expander (m = 1.9) was applied to reduce the size of the measurement volume to allow to analyse regions of high velocity gradients in the flow. The distance between the beam pairs was DS = 72 mm. With these values, the size of the probe volume could be calculated to a length of lm = 0.780 mm and to a diameter of dm = 0.094 mm. Photomultipliers observed the measurement volume through interference filters in backward scatter mode. The signal was electronically down mixed depending on the measured Doppler frequency. A digital auto-correlator evaluated the Doppler signals. Statistical averaging was transit time weighted to minimise velocity bias.

Boundary Data

At the burner outlet 1 mm above the gas exit, a complete set of profiles of first and second order moments of velocity has been measured.

Measurement Errors

An estimate of the statistical error concerning mean velocity was 5 %, whereas fluctuations were accurate within 6 %. Data obtained from the shear zone were even more accurate.

Measured Data

Further down stream axial, radial as well as circumferential velocity components and the respective fluctuations were measured for a large variety of axial and radial positions.

Velocity measurements with a five-hole probe have also been carried out at two different locations to demonstrate the intended reproducibility of the combustor configuration. An example for the velocity data measured with LDV and the 5-hole probe at a distance of 90 mm from the burner exit is displayed in the figures 5 to 7.

Figure 5: Measurements of mean velocity (h = 90 mm)

Figure 6: Measurements of velocity fluctuations (h = 90 mm)

Figure 7: Measurements of velocity cross correlations (h = 90 mm)

References

[1] T. Landenfeld, A. Kremer, E.P. Hassel, J. Janicka, T. Schäfer, J. Kazenwadel, C. Schulz, J. Wolfrum: 27th Symposium (Int'l) on Combustion (The Combustion Institute), Pittsburg (1998), p. 1023

[2] S. Böckle, J. Kazenwadel, T. Kunzelmann, D.-I. Shin, C. Schulz:, J. Wolfrum: 28th Symposium (Int'l) on Combustion (The Combustion Institute Pittsburgh) (2000), in press

[3] P. Schmittel, B. Günther, B. Lenze, W. Leuckel, H. Bockhorn: 28th Symposium (Int'l) on Combustion (The Combustion Institute Pittsburgh) (2000), in press

Test Case EXP-1

Description of Experiment

The Raman system is based on a flashlamp pumped dye Laser (l = 489 nm, 2 µs pulse duration, 3 J pulse energy) whose beam is focused into the combustion chamber by a lens and retroreflected on the other side by a spherical mirror. The scattered light from the focal region is collected at 90° by an achromatic lens (Æ = 100 mm, f = 300 mm) and relayed to the entrance slit of a spectrograph. After spectral separation, the Raman signals from the major species CH4, H2, O2, N2, H2O, CO2, and CO are detected simultaneously by an array of 10 photomultiplier tubes, transferred to gated integrators, and finally stored and processed in a PC. The spatial resolution of the measurement is determined by the focal diameter of the Laser beam and the slit width of the spectrograph and is 0.6 mm in each dimension.

In order to determine the number density of each species, the Raman signals are calibrated in cold and heated flows and in the exhaust gas of premixed laminar flames [2, 3]. The temperature is determined from the total number density via the ideal gas law, and the mixture fraction is calculated using Bilger’s definition [4], which is based on the measured atomic mass fractions of O, H, and C. The data evaluation includes corrections for crosstalk between different Raman channels and background from Laser induced fluorescence from water and polycyclic hydrocarbons (PAHs). The background from Laser excited PAH emissions is corrected for by using the signals from additional photomultiplier tubes installed in Raman-free regions of the spectrum [2]. In the flame investigated the PAH concentrations are significantly smaller than in fuel-rich regions of jet diffusion flames [2], probably due to fast and efficient mixing of fuel and air by the swirling flow field which diminishes the formation of (large) PAHs.

Boundary Data

The same boundary data are applicable as described in section 2.1. These have been a complete set of profiles of first and second order moments of velocity

Measurement Errors

The accuracy achieved for the mean values is typically 2 - 3 % for the temperature, 2 % for N2, 4 % for CO2 and decreases for smaller mole fractions. The accuracy of a single-pulse measurement is reduced due to photon statistics and is on the order of 5 % for the temperature, 5 - 7% for H2O (with a mole fraction of 0.2 at 2000 K), 12 - 15% for O2 (mole fraction 0.03, T»2000 K), and 2 % for CH4 (mole fraction = 1, T»1000 K). The accuracy of the CO detection is lower than for the other species because of corrections for cross talk and interferences stemming from PAHs. For a CO mole fraction of 0.06 the accuracy of the measurements is 20 %, for a mole fraction of 0.02 it is 50 %.

Measured Data

Spontaneous Raman scattering has been applied to simultaneously determine the temperature and the species concentrations of CH4, H2, O2, N2, H2O, CO2, and CO in pointwise single shot measurements with a spatial resolution of 0.6 mm. Because all major species have been detected, the mixture fraction could also be deduced from the experimental data. The experimental setup and the calibration and correction procedure have been optimised to achieve highly accurate and reliable data. The flames were investigated at typically 120 locations. At each location 300 single-pulse measurements were performed from which the joint PDFs were determined. Eight different heights above the nozzle (h = 10 to 300 mm) have been regarded with radial locations ranging from -10 to 150 mm.

The consistency of the RAMAN measurements was cross checked by probe measurements of the stable species and temperature that ware carried out by EBI Karlsruhe. The Figures 8 and 9 show the radial profiles for the two different heights 10 and 90 mm above the burner exit.

Figure 8: Measurements of species concentrations at h = 10 mm

Figure 9: Measurements of species concentrations at h = 90 mm

As an example the two-dimensional field of the mean temperature and its fluctuations are display in figure 10.

Figure 10: Measurements of rms (left) and mean (right) temperature

Beside mean values of the mole fraction and the temperature (i.e. radial profiles) also cross-correlations of the mole fraction of the temperature with the mixture fraction were measured. These so-called scatter plots for the CH4 mole fraction and the CO mole fraction 10 mm above the burner exit are shown in the figures 11 an 12. The scatter plots for the temperature at different distances from the burner exit are displayed in the figures 13 to 15.

Figure 11: Scatter plot of CH4 mole fraction

Figure 12: Scatter plot of CO mole fraction

Figure 13: Scatter plot for temperature at h = 10 mm

Figure 14: Scatter plot for temperature at h = 20 mm

Figure 15: Scatter plot for temperature at h = 40 mm

References

[1] W. Meier, S. Prucker, M.-H. Cao, W. Stricker: Combust. Sci. Technol. 118, 293 (1996)

[2] V. Bergmann, W. Meier, D. Wolf, W. Stricker: Appl. Phys. B 66, 489 (1998)

[3] S. Prucker, W. Meier, W. Stricker: Rev. Sci. Instrum. 65, 2908 (1994)

[4] R.W. Bilger: 22nd Symposium on Combustion, The Combustion Institute, Pittsburgh (1988) p. 475

[5] F. Holzäpfel, B. Lenze, W. Leuckel: 26th Symposium on Combustion, The Combustion Institute, Pittsburgh (1996) p.187

Test Case EXP-2

Description of Experiment

Temperature:

Rayleigh imaging is used for thermometry. For excitation a KrF excimer Laser with a wavelength of 248 nm is employed. The calibration of the Rayleigh cross-sections is based on measurements in cold ambient air and on measurements in a calibration flame.

NO concentration:

NO concentration fields are measured by Laser-induced fluorescence (LIF) imaging. For the excitation of NO the transition at 225.2 nm is used. Laser light of this wavelength is generated by H2-Raman-shifting of a tuneable KrF excimer Laser. Fluorescence is detected in a range from 230 - 255 nm. Calibration is performed by using a lean calibration flame doped with different concentrations of NO. The correction for quenching is based on the assumption of a gas composition according to completely burned gases or using published quenching cross sections [5]. Quantification of signal intensities is feasible as long as information about local gas composition is available. Simultaneously measured local temperatures are used to correct for temperature-dependent effects (absorption and fluorescence quantum yield).

OH concentration:

OH concentration fields are recorded by Laser-induced fluorescence imaging. OH is excited at the transition at 248 nm by use of a tuneable KrF excimer Laser. Detection of fluorescence is performed at 295 ± 5 nm. For calibration a calibration flame and literature data of OH concentrations is used.

Formaldehyde distribution:

Qualitative CH2O distributions are measured by Laser-induced fluorescence imaging using an excitation wavelength of 353.2 nm. For this purpose a tuneable XeF excimer Laser is used. Fluorescence is detected from 295 to 450 nm. Measurements are showing qualitative CH2O-LIF-intensity distributions only. Nevertheless formaldehyde distribution fields have the potential, in combination with OH concentration fields, to visualise the heat release distribution and therefore give an optimal visualisation of flame-front positions [6]. The extended areas where formaldehyde was detected in the swirl flame indicate the presence of low temperature chemistry in preheated gas pockets before ignition.

Boundary Data

Again the same boundary data are applicable as described in section 2.1. These have been a complete set of profiles of first and second order moments of velocity

Measurement Errors

No estimate of the accuracy of the measurement is available.

Measured Data

The present study is focused on the Laser-based investigation of the NO distribution within the reacting flow field of a strongly swirling, confined 150 kW natural gas flame (swirl number S = 0.9, equivalence ratio j = 0.83). Simultaneous quantitative measurements of NO-and OH- concentration fields (using Laser-induced fluorescence imaging, LIF) and temperature distribution (by Rayleigh scattering) allow the determination of concentration distributions as well as the analysis of correlations between all three scalars, respectively.

Up to now only a limited amount of temperature and species distribution data are available for swirling flames. Especially the NO distribution is difficult to observe due to the intrinsically low levels of NO produced in this flame type. Whereas for comparison with computational fluid dynamics simulations (CFD) temporally averaged information about temperature and species concentrations are sufficient, for probability density function approaches (PDF) correlations between different scalars are of interest. Simultaneous two-dimensional measurements of these scalars by Laser-based imaging techniques allow to assess the necessary information.

Both, time- and Favre averaged concentration fields are obtained from the simultaneous measurements. Correlations between all three simultaneously measured scalars show significant differences in the different parts of the reactive turbulent flow. Whereas the overall correlation of NO concentration and temperature shows a linear trend no correlation between NO and OH was found in the observed area. Independent experiments assessing the mean fields of temperature, NO- and OH-concentration throughout the whole reaction zone were carried out. Maximum NO concentrations are localized in the lower part of the inner recirculation zone. In contrast, maximum average OH concentrations are found close to the shear layer near the path of the injected fresh gases.

Mixing properties of the unburned gases have been investigated for the isothermal and combusting flow using tetrahydrothiophene (THT) as a new fluorescing tracer. This compound is present at concentrations of 10 mg/m3 in the natural gas delivered by the gas supply network where it is added as an odor marker for safety reasons. The experiments show the distribution of natural gas in the isothermal non-reactive and in the reactive flow directly above the exit of the swirl burner. For the non-reacting case the occurrence of high concentrations of natural gas is limited to a restricted area. The relative standard deviation reveals the presence of two shear layers at both sides of the main gas flow where fluctuations are increased. In the reactive flow, THT-LIF still indicates the presence of unburned natural gas. However, the direction of the fresh gas flow is slightly changed as compared to the non-reactive case due to volumetric variation during the reaction and to variations in gas viscosity in the presence of temperature gradients.

References

[1] T. Landenfeld, A. Kremer, E.P. Hassel, J. Janicka, T. Schäfer, J. Kazenwadel, C. Schulz, J. Wolfrum: Laser diagnostic and numerical studies of strongly swirling natural gas flames, Proc. Combust. Inst. 27 (1998) 1023-1030

[2] S. Böckle, J. Kazenwadel, T. Kunzelmann, D.-I. Shin, C. Schulz: Single-shot Laser-induced fluorescence imaging of formaldehyde with XeF excimer excitation, Applied Physics B 70 (2000) 733-735

[3] S. Böckle, J. Kazenwadel, T. Kunzelmann, D.-I. Shin, C. Schulz, J. Wolfrum: Simultaneous single-shot Laser-based imaging of formaldehyde, OH and temperature in turbulent flames, Proc. Combust. Inst.28 (2000)

[4] S. Böckle, J. Kazenwadel, C. Schulz: Laser-diagnostic multi-species imaging in strongly swirling natural gas flames, Appl. Phys.B (2000).

[5] P.H. Paul, J.A. Gray, J.L. Durant Jr., J.W. Thoman Jr.: Appl.Phys. B 57 (1993) 249-259

[6] P.H. Paul and H.B. Najm: 27th Symposium on Combustion ,The Combustion Institute, Pittsburgh (1998) 43-50

Contributors: Stefan Hohmann - MTU Aero Engines

Site Design and Implementation: Atkins and UniS