Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering

First Advisor

Vic A. Cundy


A comprehensive study of rotary kiln incineration is ongoing at Louisiana State University. Through experimentation at all levels and numerical modeling, the underlying physical processes are searched out and studied with the intent to improve the understanding of how rotary kiln incinerators process waste with the eventual goal of creating a fully predictive numerical model. The experimental work presented here focuses on mapping combustion gas temperature and, for the first time, velocity fields of a field-scale, industrial incinerator. Measurements are made at multiple points across an upper quadrant of the kiln near its exit using a bidirectional pressure probe, suction pyrometer, and a newly designed, lighter yet stiffer, positioning boom. The kiln is directly fired using natural gas in a steady state mode without waste processing. Results indicate insignificant horizontal variation, but strong vertical stratification, with the highest values of temperature and velocity corresponding to the top of the kiln. Access restraints prevented the lower region from being mapped. Operating conditions were varied by adjusting the amount of ambient air added to the front of the kiln. Increasing this air flow reduced temperatures as expected, but did not have as significant an effect on velocities. The quality of the results is examined by performing mass balances and by comparing with an existing numerical model. Both methods indicate that the experimental results are reasonable. A new steady state numerical model for the rotary kiln segment of this incinerator is then presented. This model builds on previous LSU work by including radiation and soot in the heat transfer analysis, switching to an adiabatic kiln wall boundary condition, and including a more accurate geometry and better fitting grid. These changes improve agreement with data taken from this rotary kiln by up to two orders of magnitude compared with previously developed models at LSU. In most instances, prediction is within repeatability limits of the experiments. Grid dependency is demonstrated near the kiln front where gradients are very steep. Near the exit, however, where experimental data are available, both grids produce very similar results. Parametric and sensitivity studies using the developed model are reported.