In this dissertation, a series of experiments were carried out to investigate the auto-ignition process of transient fuel jets and sprays issuing into high-temperature, environments. Novel high-speed imaging and laser diagnostic techniques were developed and applied to characterize mixing and turbulent flow conditions prior to and at the onset of ignition. In addition, this research examines the topology and dynamics of ignition kernels as they grow and transition into a stable flame. Research was carried out primarily in canonical atmospheric pressure experiments, but a new high-pressure spray test facility is developed in this work with preliminary measurements presented, demonstrating new experimental capabilities. Specific contributions of this dissertation include: (1) characterization of the transient mixing processes of variable-density atmospheric pressure jets both before and after ignition, (2) determination of the most probable mixing and turbulent flow conditions leading to local auto-ignition, (3) statistical evaluation of the dynamic growth and transport of ignition kernels, (4) construction and characterization of a novel high-pressure, high temperature spray and combustion facility, and (5) demonstration of high-speed mixture fraction measurements in non-reacting and reacting sprays at realistic thermodynamic conditions.
First, a series of transient gas-phase fuel jets issuing into a high-temperature, vitiated environment at atmospheric pressure was investigated. A well-known jet-into-hot coflow configuration was utilized with the addition of a fast-acting solenoid valves to achieve pulsed fuel injection in an environment with well-defined boundary conditions. Four test conditions were studied to examine the effects of variations in jet Reynolds number, the fuel mixture composition, and coflow temperature. High-speed laser Rayleigh scattering (LRS) was performed at 10 kHz to measure the mixture fraction and temperature fields from fuel injection to auto-ignition. The measurements were facilitated by the high-energy pulse burst laser system (HEPBLS) at OSU which outputted pulse energies > 900 mJ for burst durations ranging from 8-11 ms. The high signal yield and advanced image processing and de-noising techniques yielded results with signal-to-noise ratios (SNR) greater than 200 and spatial resolution sufficient to measure the smallest turbulence spatial scales. Subsequently, highly accurate mixture fraction gradients and scalar dissipation rates are determined in the variable-density, transient jets. High-speed multi-view OH* chemiluminescence (CL) imaging was performed simultaneously to identify the first ignition kernels spatially and temporally and deduce global ignition statistics. Ignition kernels also were determined from the temperature measurements by locating appropriate temperature rises in the LRS images. Finally, simultaneous high-speed particle image velocimetry (PIV) and OH* CL were performed at 20 kHz to determine the most probable turbulent flow conditions leading to auto-ignition.
The most probable mixture fraction values at the initial ignition sites were determined for all cases. Probability density functions (PDFs) showed that ignition occurred over a broad range of mixture fraction values, but kernels preferentially form in lean regions in the jet periphery. The mean value of the mixture fraction at ignition, 〈 ξig 〉, for each case closely matches the most reactive mixture fraction (ξMR), which is calculated using a homogeneous reactor as prescribed within the literature. However, the PDF does appear to follow an exponential decay, with a non-negligible number of ignition kernels forming in rich regions. Joint statistics between the mixture fraction and scalar dissipation rate show the tendency for ignition kernels to form in both lean and low scalar dissipation rate regions. These quantitative results corroborate previous qualitative observations from direct numerical simulations (DNS) and experiments. The PIV/OH* results showed that ignition kernels tend to form in regions of low vorticity and strain rate. PDFs of vorticity and strain rate conditioned on ignition kernel location showed a narrowing of the PDF and decrease in the PDF tails (high magnitude vorticity and strain rate), indicating a general trend that the vorticity and strain rate are more likely to decrease at ignition locations compared to non-igniting fluid.
A comprehensive characterization of the scalar mixing process was performed within the transient, variable-density jets prior to ignition. The ensemble mean and RMS mixture fraction profiles showed that the jet reaches steady state prior to ignition for the low-temperature coflows, but not necessarily for the high temperature condition. Overall the process can be described as a starting vortex and subsequent shedding, followed by jet contraction and fluid entrainment, and finally reaching a steady state (for lower-temperature conditions) before igniting. Probability density functions of the mixture fraction and scalar dissipation, as well as the mean scalar dissipation rate conditioned on mixture fraction converge to common, single-mode distributions prior to ignition. The mixture fraction and scalar dissipation rate was examined in two separate regions within the jet - the so-called "inner" and "outer" shear layers of the jet. For regions closer to centerline (inner shear layer; 〈 ξ 〉 < 0.5), the scalar dissipation rate PDFs are more symmetric when compared to the PDFs of the scalar dissipation rate in the outer shear layer (0.05 < 〈 ξ 〉 < 0.5), which show clear asymmetry (negative skewness) for low values of χ.
Ignition kernel topology and dynamics were tracked from the first ignition kernel through flame stabilization using the high-speed OH* imaging. Kernel areas, growth rates, aspect ratios (defined as the kernel width over the kernel height), sphericity (defined as the ratio of kernel surface area over the squared perimeter), and the convective displacement velocity and angle (measured using the displacement of the kernel centroids) were determined as a function of time. The results showed that the ambient (coflow) temperature has the strongest influence over kernel dynamics, including overall growth rate and stretching. In general ignition kernels tend to stretch in the primary flow direction and move downstream with minimal lateral movement. Under auto-igniting cases, stable flame bases are formed by the merging of multiple expanding kernels.
This research also describes the fabrication and testing of a novel optically accessible high-pressure, high-temperature spray and combustion facility (HPTF) for studying the auto-ignition of vaporizing liquid fuels at relevant thermodynamic engine conditions. Facility boundary conditions including the ambient temperature uniformity and fuel injector tip temperature were well characterized and meet the necessary criteria for providing benchmark data to the spray community. The HPTF can operate over a range of oxidizing conditions, consisting of quasi-inert or non-reacting (5% O2
) to 21% O2. Initial studies considered n-dodecane (nC12H26) sprays as it is a single-component diesel surrogate and matches the well-known “Spray A” conditions described within the Engine Combustion Network (ECN). In the current dissertation, preliminary measurements are performed under non-reacting conditions with an ambient temperature of 900 K and reacting conditions, where the ambient oxygen content varies from 13 to 15% O2, and ambient temperatures range from 800 to 900 K. Measurement approaches consist of schlieren visualization, diffuse background extinction imaging (DBI), laser Rayleigh scattering, and OH* chemiluminescence. All imaging approaches are performed at a 100-kHz acquisition rate. Schlieren and DBI are used to characterize the liquid length and overall spray/ignition dynamics; OH* CL is used to characterize the ignition delay (ID) times and flame lift-off lengths (FLOL); and LRS is used to simultaneously measure mixture fraction and temperature in the gas-phase regions downstream of the spray. Liquid lengths, and mixture fraction profiles match existing results within the literature in the non-reacting sprays and ignition delay times and flame lift-off lengths match existing results in reacting sprays. The agreement with previous facilities experimental and computational results yield confidence in the boundary conditions, repeatability, and reliability of the HPTF. Finally, preliminary high-speed LRS (mixture fraction) measurements are presented in reacting sprays for the first time.