Application of chemometric and experimental tools for monitoring processes of industrial importance

• Author / Creator

  Sivaramakrishnan, Kaushik

• The common theme of this work was to investigate the relationships between multi-dimensional experimental data and the physical and chemical properties of certain reacting systems that have industrial relevance. Three particular systems of increasing complexity in terms of composition of the reaction mixture were considered, namely: catalytic oligomerization of an aliphatic olefin (propylene), oxidation of a naphthene-aromatic (tetralin) and thermal conversion of Canadian oilsands bitumen, which is a mixture of a compound classes. The overall objective of my thesis was to evaluate the statistical and analytical information obtained from a variety of chemometric and experimental tools to understand the chemical behavior of each system from a reaction chemistry viewpoint, with simultaneously validating the potential for online monitoring. The entire research conducted in four phases and though the first two studies were independent from those on bitumen, all works were aimed at addressing gaps in related literature.Propylene, obtained as a by-product of pyrolysis of organic material is usually converted to heavier hydrocarbons in the gasoline or distillate boiling range. In the first phase of my research, the ability of nonlinear least squares-support vector regression (LS-SVR) adopting kernel transformations to predict concentrations of hydrocarbon products at the outlet of a flow reactor from H-ZSM-5 catalyzed propylene oligomerization was assessed in comparison with linear partial least squares (PLSR) and its variant, interval-partial least squares regression (i-PLSR). The classification of product concentrations through dimension reduction and pattern recognition techniques like principal component analysis (PCA) and hierarchical clustering analysis (HCA) reflected different elements of the oligomerization chemistry quite well. The possibility of eliminating offline concentration measurement and online tracking of product concentration distribution was the key contribution of this work. The role of mass transfer was considered to be important in the liquid-phase oxidation of hydrocarbons to produce essential petrochemicals. In the second phase, the use of simple and multiple linear regression in identifying the most significant parameter related to oxygen availability in the tetralin oxidation process was investigated. It was conducted in a microfluidic reactor under Taylor flow conditions, where the relative importance of mass transfer over hydrodynamic parameters in controlling product selectivity was established. The challenge was to tackle the multicollinearity that was detected among the explanatory variables through F-tests, t-tests, partial and Pearson correlations, changes in regression coefficient estimates and their standard errors. A pathway to predict product selectivity and tetralin conversion from the only user-controlled variable (inlet flowrate of tetralin) was developed to potentially eliminate the use of gas chromatography for offline measurements of the outputs.Thermal conversion of Athabasca bitumen was studied in detail with the intention of providing scientific reasoning for the peculiar increases in viscosity observed by certain previous researchers at less severe reaction conditions than employed in industry. A number of product characterizations by straightforward and hyphenated experimental techniques including vibrational and nuclear magnetic resonance spectroscopy, elemental analyses, distillation profiles, compositional analysis of gaseous products density and refractive indices were conducted. A comprehensive review and analysis of the plausible factors that could have affected the viscosity was provided. The third phase of my research was critical in identifying that post-thermal reaction procedures like solvent extraction and rheological conditions of viscosity measurement had a major impact on viscosity. This study highlighted some detrimental effects of the presence of halogens in bitumen and was important for simulation of partial upgrading at the laboratory-scale. The final study of my thesis comprised of deriving a credible reaction pathway from the quantitative statistical results of multivariate curve resolution and Bayesian clustering and learning methods on infrared spectra of the products from thermal conversion of Athabasca bitumen over a range of temperatures from 300 – 420 ºC. Minimum external chemical knowledge was used for the chemometric part, while the consistency of the conversion chemistry along with the proposed reaction mechanisms as indicated from the chemometric results was inspected and compared with that of Cold Lake bitumen as well. In summary, explanations for chemical changes in each of the three systems were derived from process data in different forms like spectra and process conditions by means of chemometric and experimental tools with a potential for online control.

• Subjects / Keywords

  • Chemometrics
  • Reaction network
  • Oligomerization
  • Bayesian learning
  • Athabasca bitumen conversion

• Graduation date

  Fall 2019

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-7zbj-af38

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.