Modeling of vacuum gas oil hydrocracking process using lumping approach

Abstract

Conversion of heavy feedstock into lighter products in a hydrocracking process has a strong influence on the profitability of petroleum refining. To facilitate initiatives toward improving the process efficiency, a good process model is required. This thesis discusses the development and application of lump modeling strategies in a pilot plant and industrial scale hydrocracking processes. For the pilot plant, two different four-lump models are considered, i.e., combined bed and dual bed models. Unlike the simpler combined bed model, the dual bed model includes hydrogen consumption and hydrotreating reactions, and the reactor is subdivided into two different layers so that the effect of hydrotreating reactions can be integrated. To extend the application to a commercial refinery, a full-lump model is configured to predict the product yields of a commercial hydrocracking unit known as Isomax. Using bed temperatures, flow rate of fresh vacuum gas oil (VGO), recycle rate and catalyst life as process variables, the model is proven capable of predicting the yield of all products. The model also provides improvement to previous works by considering liquefied petroleum gas (LPG), light gases, fresh VGO and recycle feed as separate lumps. This model is then used in optimizing the reactor operation where the bed temperature, flow rate of fresh VGO and combined feed ratio are adjusted to increase the plant profitability whilst maintaining all process limitations and operating constraints. To extend the model for dynamic applications, a modified space-time conservation element and solution element method (CE-SE) is introduced. The results obtained from simulation of the pilot plant based on fourlump model using CE-SE method are comparable to those obtained using finite difference method, thus providing opportunities for further works involving dynamic behavior of the process. Although the study is focussed on hydrocracking, this thesis has proved the practicality of using lumped modeling technique to address model development requirement for complex industrial processes.