Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering

Committee Chair

Jay P. Gore

Committee Co-Chair

Robert P. Lucht

Committee Member 1

Li Qiao

Committee Member 2

Pavlos Vlachos


Turbulent premixed combustion plays an important role in the development of gas turbine systems. Continuous work has been devoted to increasing the turbine inlet temperature regarding thermal efficiency and introducing combustion technology for suppression of pollutants such as nitrogen oxide (NOx). Exhaust gas recirculation (EGR) is recognized as a promising NOx inhibition technique mostly by reducing the combustion temperature. Recent industrial progress showed that EGR significantly reduced NOx emission even at the same turbine inlet temperature. An in-depth understanding of the combustion process is required for the development of low NOx combustors with EGR. In this work, a piloted axisymmetric reactor assisted turbulent (PARAT) burner was developed to study turbulent premixed combustion. The burner can be operated at both atmospheric and high-pressure conditions up to 20 bars. Carbon dioxide which is a major component of EGR in hydrocarbon combustion systems was added into the fuel stream to emulate dry EGR. The flames were operated at a nearly constant Reynolds number, Lewis number, and adiabatic flame temperature to minimize thermal and transport effects. The differences observed among these flames were expected to reveal the chemical effects of CO2 addition. The chemical effects on turbulent premixed flames are first investigated comprehensively with carefully controlled flame conditions. Time and space resolved measurements were performed under atmospheric conditions to investigate fundamentals of the combustion process with CO2 addition. Velocity boundary conditions at the burner exit were measured using particle image velocimetry. Simultaneous temperature and major species concentrations (O2 and CO2) were measured along the axial direction at the burner centerline and along the radial direction at representative axial locations above the burner. High-speed planar laser-induced fluorescence (PLIF) was applied to collect instantaneous images capturing emissions from OH radical which is an indicator of combustion products.

Progress variable was used to characterize turbulent premixed combustion behavior of the flames with different CO2 addition. In an instantaneous flame, temperature and species concentration are correlated using progress variable, and the flame structure is typically considered to be the same as a thin laminar flame without interference from turbulence. The relations between instantaneous flame temperature and major species concentrations, specifically O2 and CO2 in this study, were compared with Bray-Moss-Libby (BML) model based on progress variable and kinetic calculations. The thin flame assumption was examined using the correlation between the RMS temperature fluctuation values and mean temperature values. The measurements of the flames with varying CO2 addition collapse on the same curve with an identical deviation from the theory.

In the global flame structure analysis, the process occurring inside an instantaneous flame was neglected with the flamelet assumption. The turbulent flamelet structure was characterized using mean flame brush thickness (MFBT) and flame surface density (FSD). The mean progress variable within the flame brush at varying CO2 addition collapse on a universal curve with dimensionless coordinate associated with MFBT. The development of mean flame brush along the radial direction predominantly follows the turbulent diffusion law. A secondary effect introduced by CO2 addition was observed. The flame with higher CO2 addition shows a thinner MFBT. Flame surface density was computed within the flame brush. The FSD shows negligible difference in the upstream of the flame brush with varying levels of CO2 addition. At the downstream of the flames, a significant effect of CO2 addition on flame surface density was observed. The flames with higher CO2 addition show higher FSD.

Turbulent burning velocity represented by global consumption speed and local consumption speed was calculated. The global consumption speed decreases with increasing of CO2 addition due to a predominant effect of lower unstretched laminar flame speed. The global combustion intensity showing overall flame surface area increases with CO2 addition. The local consumption intensity was evaluated using two methods: one considers unburned and burned pockets, and one does not. The mean values of the local consumption intensity without considering pockets show little difference among the flames with varying levels of CO2 addition. The effect of CO2 addition was observed in local consumption intensity considering pockets. The flames with higher CO2 addition shows lower values as a function of axial locations. The fine-scale unburned pockets formation and burnout process showed significant effects on flamelet structure and turbulent burning velocity. The number of pockets formed and their size and consumption speed were evaluated at the downstream of the flames. The number of pockets increases with increasing of CO2 addition. The size and consumption speed show minimum influence from the CO2 addition.

Addition to the in-depth investigation of the effects of CO2 addition on turbulent premixed flames, efforts were also made to extend our capabilities in studying turbulent premixed combustion process with EGR. Planar laser-induced fluorescence measuring CH radical and simultaneously CH and OH radicals through CH C-X band was performed for flames with varying levels of CO2 addition. The measurement signal strength was evaluated with different wavelength and flame equivalence ratios. This technique exhibits excellent potential for more accurate flame measurements with rigorous indications of the primary reaction zone and post-flame zone. The challenges can be overcome by the development of higher energy kHz laser systems. Quantitative narrowband infrared radiation measurement was performed for turbulent premixed flames. The radiation intensity was computed using stochastic time and space series analysis in conjunction with different radiation models. The results from narrowband model and line by line model based on HITEMP database show good agreements with the measurements. Mean temperatures were computed using the onion peeling deconvolution method. Good agreements were observed between measurements and calculations. The quantitative narrowband infrared radiation measurement shows excellent potential for model validation in the absence of high-fidelity simulation codes of turbulent flames. Its capability can be advanced by the development of 3D time-dependent simulation tools and radiation models.