Modeling Mild Thermal Cracking of Heavy Crude Oil and Bitumen with VLE Calculations

Abstract

The current shortage of crude oil from conventional sources has increased interest in developing unconventional resources such as oil sands. Heavy crudes and bitumen are found in Northern Alberta and their exploration, processing, and transport to market pose challenges in the use of these resources. Part of the solution to these challenges involves the reactive thermal processing of heavy crudes and bitumen. This thesis focused on mild thermal cracking reactions, and two studies regarding these reactions were presented. The first was an experimental study performed in a pilot-scale semi-batch reactor. The three crude oils were heated to 350, 400, 425, and 450°C at 1240 kPa. A five-lump reaction model combined with a process simulator with VLE calculations was fitted with the experimental data obtained. The goodness of fit between the model predicted values and experimental values for the Hardisty (MBL), Albian Heavy Synthetic (AHS), and Christina Lake Dilute Bitumen (CDB) were determined to be 0.99, 0.99, and 0.98, respectively. Moreover, 80, 85, and 89% of the optimized model’s predicted values had less than 10% error for MBL, AHS, and CDB, respectively. The second study described the implementation of a mild thermal cracking reaction model to the development of a train car fire-model for the assessment of safety aspects in the design of train cars used to transport crude oil. Case studies were conducted using the UniSim® depressuring utility and a previously developed mild thermal cracking reaction model to demonstrate the effect of compositional change. Three crude oils with varying properties and representative of the types of crudes transported by rail in Canada were used here: MBL, AHS, and CDB. The case studies conducted showed the performance of a train car fire-model to be dependent on the crude oil characteristics: up to -57% and -99% difference in model predicted variables for AHS and CDB, respectively, when compared to MBL. Furthermore, the model’s performance was also shown to be affected by the compositional change of a given crude oil due to mild thermal cracking reactions: up to 42% difference in model predicted variables when compared to the base case.