Mathematical Modelling and Simulation of a Trickle-Bed Reactor for Petroleum Feedstocks Hydrotreating

Petroleum refining is one of the most important industries worldwide with a continuous and increasing demand of higher quality products that requires novel frameworksto develop better technology. One of the most compelling processes is hydrotreatment, where light and heavy vacuum gas oils (VGO) that go out of theatmospheric distillation process of crude oil are converted into more valuable products under high hydrogen pressures taking place at trickle-bed reactors.
Trickle-bed reactor models to portray hydrotreating processes are frequently limited to study the performance at micro- and pilot-scale. Most of these studies take into account the conditions typical of industrial processes to report tendencies and behaviour of the model solutions. However, the reproducibility of industrial data is still a shortcoming of existing models, especially predicting the phase change of species. In addition, the vaporization of light ends due to the heat released by there actions is generally not addressed in literature due to the complexity of modelling and solving vapor-liquid equilibrium (VLE) in a trickle-bed reactor model.
The focus of this thesis is the simulation of a large-scale hydrotreating unit using a thorough reactor model based on first principles. The model is developed as a plug-flow reactor (PFTR) model, and alternatively, as a series of continuous stirred-tank reactors (CSTR). The mass and energy balance equations describe the transport of heat and species between the gas, liquid and the solid phase where the reaction takes place.
The performance of both models is tested against each other. The simulation of a hydrotreating reactor using the CSTR in series approach is quick but imprecise nonetheless. An additional optimization study is required to make this approachreliable. On the other hand, the plug-flow reactor model is an appropriate choice since it presents a good trade-off between the solution time and the consistency of the results.
The model parameters were taken from different sources in literature for similar systems, which presents an undeniable source of uncertainty. Therefore, the implementation of the plug-flow model is tested using a sensitivity analysis to determine the most influential parameters to the model solution. The results show that the selection of the kinetic model parameters is critical to obtain realistic results. This is especially relevant for the kinetics of aromatics saturation.
The plug-flow model is further developed to account for phase change, which is scarcely addressed in literature. A simulation framework is proposed and the methodology goes through the steps necessary to couple the reactor model, solved in Matlab, with a vapor-liquid equilibrium calculation from a process simulator such as ProII. Said simulation framework is able to handle the phase change in the reactor without increasing the complexity of the mathematical model. Moreover,the database available from ProII for petroleum streams allows us to take advantage of the use of pseudocomponents. The pseudocomponents provide supplementary attributes to simulated petroleum streams that resemble real feedstocks.
Using the simulation framework, the solution of the plug-flow reactor model obtained in MatLab is partitioned and coupled with a flash calculation carried outin ProII. The results are compared to the data available from a real large-scale hydrotreating trickle-bed reactor, demonstrating the capabilities of the simulation approach.