Pyrolysis Kinetics and Chemical Structure Considerations of a Green River Oil Shale and its Derivatives

Pyrolysis Kinetics and Chemical Structure Considerations of a Green River Oil Shale and its Derivatives James Lyle Hillier Department of Chemical Engineering Doctor of Philosophy This work had the objective of determining both the kinetic parameters for the pyrolysis of oil shale and kerogen as well as using analytical techniques coupled with pyrolysis to shed light on the structure of a specific Green River oil shale. Because of the problems with linearized methods and disagreement among literature values and methods, a new method was developed to fit kinetic parameters to non linearized data. The method was demonstrated to determine the “correct” answer for mathematically generated data within a few percent error and was shown to have a lower sum squared error than the linearized methods. The curve-fitting methodology was then applied to pyrolysis kinetic data for kerogen and oil shale. Crushed samples were pyrolyzed at heating rates from 1 to 10 K/min and at pressures of 1 and 40 bar. The transient pyrolysis data were fit with a first-order model and a progressive Distributed Activation Energy Model (DAEM). An F-test was used to determine confidence regions and compare the kinetic parameters among the samples. The activation energies determined ranged from 173 to 226 kJ/mol, with most values around 200-220 kJ/mol. The kinetic coefficients determined for oil shale and the demineralized samples were statistically the same. Only small differences in kinetic coefficients were seen in the size-graded samples. The first-order and DAEM were shown to be statistically different, but a visual inspection of a graph of the model predictions and the data revealed that both models performed well. The largest effect on the kinetic parameters was between samples collected from different geographical locations. The pyrolysis products (and the parent kerogen sample) were analyzed by several chemical techniques to determine chemical structure information about the parent sample. The GC/MS data for the tars collected showed a distribution of alkenes/alkanes with 11 to 12 carbons in length being the most frequent. XPS analysis demonstrated that any chemical model must have pyridinic and pyrrolic nitrogens as well as carbonyls and carboxyl groups. Therefore a chemical structure model of kerogen should have heteroatoms of nitrogen in the aromatic region (i.e., the portions of the kerogen that are stable at moderate temperatures).