UC Berkeley

An ever increasing demand for energy coupled with a need to mitigate climate change necessitates technology (and lifestyle) changes globally. An aspect of the needed change is a decrease in the amount of anthropogenically generated CO2 emitted to the atmosphere. The decrease needed cannot be expected to be achieved through only one source of change or technology, but rather a portfolio of solutions are needed. One possible technology is Carbon Capture and Storage (CCS), which is likely to play some role due to its combination of mature and promising emerging technologies, such as the burning of hydrogen in gas turbines created by pre-combustion CCS separation processes. Thus research on effective methods of burning turbulent hydrogen jet flames (mimicking gas turbine environments) are needed, both in terms of experimental investigation and model development. The challenge in burning (and modeling the burning of) hydrogen lies in its wide range of flammable conditions, its high diffusivity (often requiring a diluent such as nitrogen to produce a lifted turbulent jet flame), and its behavior under a wide range of pressures. In this work, numerical models are used to simulate the environment of a gas turbine combustion chamber. Concurrent experimental investigations are separately conducted using a vitiated coflow burner (which mimics the gas turbine environment) to guide the numerical work in this dissertation. A variety of models are used to simulate, and occasionally guide, the experiment.

On the fundamental side, mixing and chemistry interactions motivated by a H2/N2 jet flame in a vitiated coflow are investigated using a 1-D numerical model for laminar flows and the Linear Eddy Model for turbulent flows. A radial profile of the jet in coflow can be modeled as fuel and oxidizer separated by an initial mixing width. The effects of species diffusion model, pressure, coflow composition, and turbulent mixing on the predicted autoignition delay times and mixture composition at ignition are considered. We find that in laminar simulations the differential diffusion model allows the mixture to autoignite sooner and at a fuel-richer mixture than the equal diffusion model. The effect of turbulence on autoignition is classified in two regimes, which are dependent on a reference laminar autoignition delay and turbulence time scale. For a turbulence timescale larger than the reference laminar autoignition time, turbulence has little influence on autoignition or the mixture at ignition. However, for a turbulence timescale smaller than the reference laminar timescale, the influence of turbulence on autoignition depends on the diffusion model. Differential diffusion simulations show an increase in autoignition delay time and a subsequent change in mixture composition at ignition with increasing turbulence. Equal diffusion simulations suggest the effect of increasing turbulence on autoignition delay time and the mixture fraction at ignition is minimal.

More practically, the stabilizing mechanism of a lifted jet flame is thought to be controlled by either autoignition, flame propagation, or a combination of the two. Experimental data for a turbulent hydrogen diluted with nitrogen jet flame in a vitiated coflow at atmospheric pressure, demonstrates distinct stability regimes where the jet flame is either attached, lifted, lifted-unsteady, or blown out. A 1-D parabolic RANS model is used, where turbulence-chemistry interactions are modeled with the joint scalar-PDF approach, and mixing is modeled with the Linear Eddy Model. The model only accounts for autoignition as a flame stabilization mechanism. However, by comparing the local turbulent flame speed to the local turbulent mean velocity, maps of regions where the flame speed is greater than the flow speed are created, which allow an estimate of lift-off heights based on flame propagation. Model results for the attached, lifted, and lifted-unsteady regimes show that the correct trend is captured. Additionally, at lower coflow equivalence ratios flame propagation appears dominant, while at higher coflow equivalence ratios autoignition appears dominant.