D. R. Snelling

Spectrally Resolved Measurement of Flame Radiation to Determine Soot Temperature and Concentration

A multiwavelength flame emission technique is developed for high spatial resolution determination of soot temperature and soot volume fraction in axisymmetric laminar diffusion flames. Horizontal scans of line-integrated spectra are collected over a spectral range of 500-945 nm. Inversion of these data through one-dimensional tomography using a three-point Abel inversion yields radial distributions of the soot radiation from which temperature profiles are extracted. From an absolute calibration of the flame emission and by use of these temperature data, absorption coefficients are calculated, which are directly proportional to the soot volume fractions. The important optical parameters are discussed. It is shown that a uniform sampling cross section through the flame must be maintained and that variations in sampling area produce inconsistencies between measurements and theory, which cannot be interpreted as spatial averaging of the property field. The variations in cross-sectional sampling area have the largest influence on the measurements at the edges of the flame, where the highest resolution is required. Emission attenuation by soot has been shown to have minor influence on the soot temperature and soot volume fraction for the soot loading of the axisymmetric flame tested. An emission correction scheme is outlined, which could be used for more heavily sooting flames. For a refractive index absorption function E(m) = Im[(m 2 - 1)/(m 2 + 2)] that is independent of wavelength, the soot temperatures and soot volume fractions measured with this technique are in excellent agreement with data obtained by coherent anti-Stokes Raman scattering nitrogen thermometry and two-dimensional soot extinction in the same ethylene coflow diffusion flame. The agreement of the results suggests a limit of the slope of the spectral response of E(m) to be between 0 and 20% over the spectral range examined.

Flame front surface characteristics in turbulent premixed propane/air combustion

The characteristics of the flame front surfaces in turbulent premixed propane/air flames were investigated. Flame front images were obtained using laser-induced fluorescence (LIF) of OH and Mie scattering on two Bunsen–type burners of 11.2-mm and 22.4-mm diameters. Nondimensional turbulence intensity, u′/SL, was varied from 0.9 to 15, and the Reynolds number, based on the integral length scale, varied from 40 to 467. Approximately 100 images were recorded for each experimental condition. Fractal parameters (fractal dimension, inner and outer cutoffs) and corresponding standard deviations were determined by analysis of the flame front images using the caliper technique. The fractal dimensions derived from OH and Mie scattering images are almost identical. However, inner and outer cutoffs from OH images are consistently higher than those obtained from Mie scattering. The self-similar region of the flame front wrinkling is about a decade for all flames studied. In the nondimensional turbulence intensity range from 1 to 15, it was found that the mean fractal dimension is about 2.2 and it does not show any dependence on turbulence intensity. This contradicts the findings of the previous studies that showed that the fractal dimension asymptotically reaches to 2.35–2.37 when the nondimensional turbulence intensity u′/SL exceeds 3. It is shown that the reason for this discrepancy is the image analysis method used in the previous studies. Examples are given to show the inadequacy of the circle method used in previous studies for extraction of fractal parameters from flame front images. The fractal parameters obtained so far, in this and previous studies, are not capable of correctly predicting the turbulent burning velocity using the available fractal area closure model.

Influence of hydrogen addition to fuel on temperature field and soot formation in diffusion flames

Overventilated, coflowing axisymmetric laminar diffusion flames of ethylene, propane, and butane were used to study the influence on soot of hydrogen addition to the fuel. The flame temperatures were measured by CARS along the flame axis as well as at off-axis radial locations. CARS spectra taken in heavily sooting regions exhibited poor fits due to C 2 absorbtion of part of the fundamental band of the nitrogen spectrum. It was found that C 2 absorbtion was confined to frequencies greater than 2313 cm −1 . We, therefore, implemented a strategy that fitted only the CARS spectra in the frequency range smaller than 2313 cm −1 . Measured soot concentrations and the flame temperature data with and without hydrogen and helium dilution were evaluated and the relative influences of dilution and direct chemical interaction on soot formation, as a result of hydrogen addition, are presented. It is shown that when hydrogen or helium is added to the fuel as a diluent in moderate quantities, the changes in the temperature field of the coflow diffusion flames are negligible. When allowance is made for the influence of dilution, addition of hydrogen to the fuel side of an ethylene diffusion flame reduces the soot formation. For propane and butane flames, hydrogen addition does not show any influence on soot formation apart from the dilution effect.

Characterization of flame front surfaces in turbulent premixed methane/air combustion

Abstract A detailed experimental investigation of the application of fractal geometry concepts in determining the turbulent burning velocity in the wrinkled flame regime of turbulent premixed combustion was conducted. The fractal dimension and cutoff scales were determined for six different turbulent flames in the wrinkled flame regime, where the turbulence intensity, turbulent length scale, and equivalence ratio were varied. Unlike previous reports, it has proved possible to obtain the fractal dimension and inner and outer cutoffs from individual flame images. From this individual data, the pdf distributions of all three fractal parameters, along with the distribution of th predicted increase in surface area, may be determined. The analysis of over 300 flame images for each flame condition provided a sufficient sample size to accurately define the pdf distributions and their means. However, the predicted S T S L , calculated using fractal parameters, was significantly below the measured values. For conical flames, a geometrical modification factor was employed to predict S T S L , however, this did little to improve the predictions. There appeared to be no dependence of the predicted S T S L on the approach flow turbulence. The cutoffs did not seem to vary significantly with any of the length scales in the approach flow turbulence, although the fractal dimension did appear to have a weak dependence on u′ S L and ReΛ. The probable reasons that fractal geometry does not correctly predict S T S L are that S T S L = A w A 0 does not hold in wrinkled turbulent premixed flames, that the flame front surface cannot be described by a single scaling exponent, or that these are not wrinkled flames even though they are within conventional definitions of the wrinkled flame regime.

Precision of multiplex CARS temperatures using both single-mode and multimode pump lasers

The noise level in single-pulse resonant nitrogen CARS spectra is shown to decrease with increasing pump laser bandwidth. This is the reverse of the trend observed for nonresonant CARS spectra. The precision of single-pulse CARS temperature measurements is shown to be dramatically increased by performing a weighted fit of theoretical and experimental CARS spectra using the measured detector noise coefficients as weighting parameters. The inclusion of collisional narrowing and cross-coherence in the CARS theory calculations and their effect on best fit temperatures are discussed. These temperatures, measured in a flat-flame burner, are compared with those obtained by Na line-reversal.

Surface density measurements of turbulent premixed flames in a spark-ignition engine and a bunsen-type burner using planar laser-induced fluorescence

Surface densities of turbulent three-dimensional premixed flames determined from planar flame images and flame surface densities in a spark-ignition engine are both reported for the first time. The flame images were obtained using laser-induced fluorescence of biacetyl in the engine and of OH radicals in a Bunsen flame. Images from orthogonal planes have been used to determine the mean orientation angle along the line of intersection of the planes in the engine experiments. Because of the axisymmetric flow field of the Bunsen flame, it has been assumed that the flame front has a symmetric mean orientation behavior about the axis, and the normal to the flame front can be determined from images in one plane. The flame surface density (Ω) profiles and orientation angle that produced them are presented as a function of the progress variable. The directional cosines for the engine flame are almost identical in both orthogonal planes, indicating the isotropic nature of the flame front. A typical value of 0.7 has been found for the directional cosine for the Bunsen flame and the engine flame at 1200 rpm. At a lower engine speed, the directional cosine has increased slightly. The bray-Moss-Libby (BML) model coefficients obtained at maximum flame surface density in this work agree with the previously reported results and display an inverse power-law dependence on the integral length scale when normalized by the laminar flame thickness. The BML coefficients obtained from direct numerical simulation (DNS) studies are in qualitative agreement with the present results. In the range of u′/SL investigated in this work, the flame-surface density has not shown much variation.

Noise in single-shot broadband coherent anti-Stokes Raman spectroscopy that employs a modeless dye laser.

The noise in single-shot coherent anti-Stokes Raman (CARS) spectroscopy that employs a broadband modeless dye laser (MDL) is examined and the results are compared with those of a conventional dye laser. The noise of the dye-laser, the nonresonant CARS, and the resonant N(2) CARS signals are determined. The use of a MDL is shown to result in substantially reduced CARS noise when the CARS signal is generated with a single-mode pump laser, but only a marginal reduction of noise is observed with a multimode pump source The noise measurements are compared with theoretical predictions that are based on models that assume modes of random amplitudes and phases in the multimode laser sources. The combination of a MDL and a single-mode pump laser is shown to increase the precision of single-shot N(2) CARS temperature measurements.

Effect of detector nonlinearity and image persistence on CARS derived temperatures

The image persistence of self-scanning photodiode arrays (IPDA) incorporating P-20 phosphor-based intensifiers is shown to make them unsuitable for single-pulse CARS temperature measurements in turbulent combustion. Correcting CARS flame spectra for the nonlinear response of the IPDA detectors increases CARS derived temperatures approximately 3-6%. This error is partially offset by correcting for the perturbations in the N(2) vibrational population resulting from stimulated Raman pumping. The effect of these population perturbations on CARS-derived temperatures is determined. CARS flame spectra obtained with uncorrelated pump beams that are corrected for IPDA nonlinearity and stimulated Raman pumping are shown to give temperatures in good agreement with combined thermocouple/sodium line-reversal measurements.

Multichannel light detectors and their use for CARS spectroscopy.

A broadband CARS system designed to measure flame temperature from hot nitrogen CARS spectra is described, and a source of temperature error is identified and attributed to nonlinear behavior of the optical multichannel diode array detector. Detector sensitivity, linearity, uniformity, and noise are discussed. Data are presented for the Tracor-Northern TN-1223-4GI and TN-6132 detectors. It is shown that the nonlinearity observed with CARS signals is a function of radiation density at the detector input plane. This nonlinearity is associated with saturation in the microchannel plate intensifier and should be observable with short duration high intensity light pulses other than CARS, and with intensified detectors (both diode array and vidicon) other than the Tracor-Northern. The nonlinearity observed with the TN-1223-4GI detector can be eliminated by reducing the radiation density. At maximum sensitivity the detector shot noise contributes a significant component to the shot-to-shot variation of the spectral profile of the CARS signal.

Nonlinearity and image persistence of P-20 phosphor-based intensified photodiode array detectors used in CARS spectroscopy

Several self-scanning photodiode arrays (IPDA) used for CARS spectroscopy are shown to exhibit a greater image persistence than has generally been realized, and to exhibit a falloff in sensitivity that is logarithmic with decreasing output signal. These effects are attributed to the P-20 phosphor based intensifiers used in the IPDAs and are probably generic to all such detectors. A strategy for minimizing the image persistence in CARS spectroscopy is presented. A prototype detector incorporating a much faster rare earth phosphor is evaluated and shown to be more suited to single pulse CARS measurements in turbulent combustion than the IPDAs incorporating P-20 phosphors.