A study of turbulent lean-premixed prevaporized combustion with emphasis on fuel dispersion
Abstract
An experimental/computational investigation of turbulent spray processes affecting premixing-prevaporization in liquid-fueled lean-premixed combustors is reported. Low combustor emissions levels and high combustion efficiency require optimum premixer performance. Complete drop evaporation and spatially and temporally homogeneous fuel-air mixtures at the inlet to the combustor are desirable. An experimental study with a tubular premixer, tubular combustor operating at atmospheric pressure is conducted to evaluate the importance of drop dispersion, equivalence ratio, premixer residence times, and atomization characteristics on premixing-prevaporization. Premixing-prevaporization effectiveness is evaluated primarily based on measurements of gas concentrations in the exhaust. Ultra-low NO$\rm\sb{x}$ operation is possible for equivalence ratios less than 0.5. The equivalence ratio is the dominant variable. Mie-scattering measurements are conducted at the exit of the premixer to determine qualitative levels of premixing-prevaporization. These measurements show considerable spatial inhomogeneities across the premixer without significant adverse effects on NO$\rm\sb{x}$ formation. Drop dispersion is found to be the most critical parameter for high combustion efficiency and low NO$\rm\sb{x}.$ An experiment is designed to conduct a fundamental study of drop dispersion in a well characterized confined turbulent flow to obtain measurements of the velocity statistics of both phases, drop size distributions, and liquid dispersion. These measurements provide the gas-phase flow field for the dispersion calculations, and provide initial conditions for statistical simulations of drop dispersion. Simulations of drop dispersion are performed using a correlated Monte-Carlo method. This approach avoids selection of a single turbulent length scale for drop-turbulence interactions, and an assumption of isotropic gas-phase turbulence. Predictions of rms velocity fluctuations, and PDFs of drop velocity and drop size are compared to measurements to provide a stringent evaluation of the simulation. The simulations provide reasonably good quantitative agreement with measurements. A new statistical approach called the Discrete Probability Function (DPF) method is evaluated for drop dispersion calculations using available dispersion measurements for single size particles. The DPF method leads to a reduction in statistical noise when compared to Monte-Carlo methods. However, the increase in computational expense with the DPF method requires that more efficient algorithms need to be developed before the DPF method can be applied to practical dispersion calculations.
Degree
Ph.D.
Advisors
Gore, Purdue University.
Subject Area
Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.