Combustion Measurements Using Femtosecond Coherent Anti-Stokes Raman Scattering

Levi M Thomas, Purdue University

Abstract

This work builds upon the significant contributions of previous researchers to further advance the chirped-probe pulse femtosecond coherent anti-Stokes Raman scattering (CPP fs CARS) technique. The CPP fs CARS laser system was reconfigured for spectral isolation of the signal from the CARS input beams. Previously, scatter from the pump beam was frequency degenerate with the signal beam, resulting in signal noise. In preparation for measurements in sooty flames and spray flames, the optical parametric amplifier was reconfigured to output the frequency-doubled idler at 982 nm instead of the doubled signal at 675 nm. The CARS signal, generated at 675 nm, then had complete spectral isolation from the other sources of intense laser radiation. The spectral fitting routine was optimized. Computational processing time for an entire data set previously required several weeks. Using basic estimates as well as more complicated ultrafast laser measurement techniques, sensible ranges were established for xviii the floating variables in the fitting algorithm used to determine laser parameters which yielded faster and more consistent convergence. A new methodology was developed to directly measure the instrument response function (IRF) of the data acquisition system. This frequency-dependent IRF was applied to each synthetic spectrum prior to fitting with an experimental spectrum. Lastly, the code was modified to allow for batch processing of multiple sets of single-shot signal spectra. Laser parameters are first determined through measurements at known conditions in a nearly-adiabatic flame, and then used in spectral fitting to determine single-shot temperature measurements of experimental data. A new data processing technique, based on the statistical method of maximum likelihood, was developed and implemented to combine results from multiple sets of laser parameters in an error-weighted mean. A new testing protocol, that characterizes measurement performance with randomized validation spectra of known temperature, was performed 25 times over a period of four months. Aggregated results showed mean accuracy and precision of 2.7% and ±3.5% across flame temperatures, and 9.9% and ±6.1% at room temperature. CPP fs CARS was initially unsuccessful in the highly-luminous ethylene-nitrogen diffusion flame due to the strong background levels. Exploiting the incoherent nature of soot incandescence, a ~26-foot extension was implemented in the signal beam collection path which reduced the recorded background level by ~45%. Single-laser-shot spectra were able to fit for temperature determination, and a full set of radial scans was performed at seven axial locations. This flame was extremely bright and exhibited visibly-unsteady behavior. Analysis of the temperature fields showed increased RMS fluctuation levels, but the average temperature profiles were largely symmetrical. Fourier analysis revealed the presence of coherent structures in the temperature spectrum of all flames, and was attributed to a Kelvin-Helmholtz-type instability driven by density gradients in the shear layer between the pilot flame and coflow. This instability peaked at 3 diameters axially and decayed thereafter in all flames except in the ethylene-nitrogen diffusion flame, where it increased in strength moving radially outward with height. Conjecture offered to explain why the temperature instability seemed to amplify only in xix the diffusion flame, is that the potential for hydrodynamic-thermodynamic coupling is stronger in a diffusion flame compared to a premixed flame. Vorticity and eddies created by a shear layer instability can increase the reaction zone surface area, thereby increasing the rate of heat release. After the OPL extension, the CARS technique was successful in performing single-shot temperature measurements in this highly-luminous flame. CPP fs CARS was demonstrated for the first time in spray flames. The Sydney Needle Spray Burner (SYNSBURN™) produced acetone and ethanol spray flames of variable density and fuel loading using. Radial and axial probe volume locations were chosen targeting the region where the coaxial pilot ignites and anchors the spray flame. The CPP fs CARS technique was most successful in the dilute spray flames. Signal dropout was most severe on centerline, but typically below 10% on average. Examples of interferences from hydrocarbon fuel in the probe volume are presented. The interferences were attributed to two separate phenomena and categorized based on the probable phase of the fuel—liquid or gas. Interference caused by liquid fuel was considered unavoidable, but easily identified because the spectra could not be analyzed. Interference from vapor fuel was more problematic as the nitrogen signal was only moderately corrupted in the high-frequency region. The spectra could be fit but temperature was over predicted. Rejecting individual signal spectra, based on a fitting error threshold, was shown effective in excluding shots with significant interference from fuel droplets, but shots with only minor interference would require a more-advanced rejection criterion. Analysis of the temperature fields shows that the spray flames with a higher degree of fuel atomization are effectively ignited and anchored by the pilot. (Abstract shortened by ProQuest.)

Degree

Ph.D.

Advisors

Lucht, Purdue University.

Subject Area

Mechanical engineering

COinS