This thesis utilizes time-resolved particle image velocimetry (PIV) measurements to provide a detailed statistical characterization of the kinematic properties from a set of standard non-premixed flames, which serve as benchmark test cases within the turbulent combustion community. The flames under consideration are the well-characterized DLR CH4/H2/N2 jet flames operating at two Reynolds numbers, DLR A (Re = 15,200) and DLR B (Re = 22,800), where DLR A represents a robust flame condition and DLR B represents a flame near blowout with high degrees of local flame extinction. A qualitative and quantitative comparison between DLR A and DLR B is presented to understand the effect of Reynolds number variations and local flame extinction on turbulence statistics. Detailed data processing techniques to analyze the time-resolved, 2D PIV data have been developed and presented in this work. Radial profiles of the mean and RMS fluctuations of the velocity are presented and results from DLR A agree with previous point-based measurements giving confidence in the measurements and analysis. While the DLR flames are highly utilized test cases, no published velocity data from the DLR B flames is available and thus the current measurements extend the existing data set available for modelers.

A series of measurements were taken of the shock-boundary layer interaction (SBLI) in a Mach 2.1 continuously operated wind tunnel. The SBLI was generated by a small (~1.1mm tall) 20° wedge located on the top wall, and data were taken both in the region near the compression wedge and in the area where this shock impinged on the bottom wall. PIV was the primary measurement tool in both locations, though pressure data were also acquired near the compression wedge. Data were acquired at 4 spanwise locations to study the three-dimensionality of the flow. Both interactions were found to be highly 3-D, with a stronger interaction observed near the channel centerline. Evidence of a corner vortical structure in the compression corner was observed, and substantiated by CFD. Intermittent flow reversal was seen in the reflected shock interaction near the channel centerline, though not in the corners. The data suggest the presence of vortical structures generated near the channel centerline and pushed towards the sidewalls. Following the characterization of the base case, a Monte Carlo experiment was performed in which geometric perturbations were installed along the bottom wall of the wind tunnel and their effect on the flow was studied. The Monte Carlo device was designed and installed at the location predicted to be most sensitive by CFD. The majority of the locations initially tested displayed minimal sensitivity, with only the largest and most upstream quasi-2D cases showing significant effects on the flow at the corner. The perturbation device was redesigned and moved upstream, and additional quasi-2D cases were tested. It was found that some configurations accelerated the flow and strengthened the primary shock, while others slowed the flow and weakened the shock. Overall, the flow was observed to be very sensitive to some perturbations, but only to those located within a limited range of streamwise positions, and with a wide variety of system responses possible.

This research aims to investigate the nonlinear dynamics of the non-reacting jets and non-premixed lifted jet flames. The goal is to understand better how the flow system dynamics change over time and identify the path toward unwanted conditions such as flashback, extinction, or blowout to limit combustors' dynamical failure. The existence of these undesirable conditions is bound to the fluid's history, meaning that initiated perturbation may persist in the system for time scales comparable to large-scale flow timescales. Hence, the notion is to utilize jet and jet flames as a study test case to work out how the flow evolves dynamically with the hope of understanding how to limit occurrences of the chaotic unwanted condition. Initially, planar particle image velocimetry has been used for the development of the methodologies. I have used planar data to investigate the nonlinear dynamics of non-reacting turbulent jets, with a low-to-moderate Reynolds number using the single-trajectory framework and ensemble framework. I have used Lyapunov exponents to calculate the spectra of scaling indices of the attractor. Then, I used Lagrangian Coherent Structures (LCSs), which are defined as manifolds that are locally Euclidean and invariant, to study the relationship between Lyapunov exponent changes with flow topological features. These LCSs behave as hypersurfaces with maximally repelling or attracting properties. These various methodologies were used to investigate flame-turbulence interaction in lifted jet flames. The Lagrangian framework is shown to be effective at revealing the kinematics associated with flame-turbulence interaction. The LCSs' time history represents how eddy structures interact with the flame and highlight their role in the dynamics of the lifted jet flames. Finally, I have investigated the flame and turbulence interaction using high-speed luminosity imaging and simultaneous three-dimensional particle image velocimetry. The three-dimensional Lagrangian structures provide us a more detailed flow-flame interaction. It is shown that the flow features associated with attracting LCSs can create a barrier attracting the flame that makes the flame move upstream. In contrast, the presence of repelling LCSs near stationary flames breaks the balance between the gas velocity and flame propagation speed, causing the flame to become non-stationary and move downstream. It was also found that the repelling LCSs induce negative curvature on the flame surface whereas pushing the flame toward the products. However, the attracting LCSs induce positive curvature on the flame surface and draws the flame toward the reactants

As part of an ongoing research program aimed at developing techniques capable of quantitative imaging of mixture fraction and scalar dissipation in turbulent flames, three-scalar measurements were made in a turbulent nonpremixed flame. The use of nitrogen Raman scattering to detect a passive conserved scalar made it possible to increase confidence in the two-scalar technique based on simultaneous imaging of Rayleigh scattering and fuel Raman scattering. These experiments showed that proper parameterization of mixture fraction-dependent terms appearing in the expression for mixture fraction can improve accuracy for lean values of mixture fraction as well as those near stoichiometric. Additionally, a new experimental technique was investigated which allows the extraction of velocity information from laser-based scalar imaging in turbulent flows. Preliminary results from simultaneous particle-imaging velocimetry (PIV) and optical flow velocimetry showed that this technique has potential for unseeded velocity measurements compatible with the mixture fraction imaging. The optical flow approach was based on extracting velocity vectors from the intensity variations naturally present in inhomogeneous turbulent flow images of a mixing-dependent scalar quantity.

Simultaneous images of the CH and OH reaction zones are reported for Intensely Wrinkled nonpremixed flames, to determine whether reaction zones retain their thin laminar flamelet structure or become distributed reaction zones. Intensely Wrinkled Flames (IWFs) were achieved by using a special burner with large coflow air velocities to obtain a normalized turbulence intensity of 3.6, which is 10 times greater than the turbulence intensity within jet flames. The images were used to measure profiles of the flame surface density and the average CH layer thickness; it is argued that these parameters are the ones that should be used to assess new large eddy simulations (LESs), rather than insensitive parameters such as mean concentrations. In the regime of IWFs, the CH reaction zones remained as thin as those measured in laminar jet flames (i.e., less than 1 mm thick) and had the appearance of flamelets. These thin reaction zones were extinguished before they became thickened by intense turbulence, which provides experimental evidence to support laminar flamelet modeling concepts. Shredded flames occurred, within which the reaction zones were short, discontinuous segments, and the degree of flame wrinkling was significantly larger than in jet flames. Shredded flames have not been observed previously. There is no evidence of small-scale wrinkling of the reaction zones at scales less than half the integral scale. The images showed where the instantaneous stoichiometric contour is located, since it exists at the boundary between the CH and OH layers. Flame surface densities were typically 0.3 mm to the negative 1 power.