AbstractEffective distribution of hydrogen in refinery hydrogen networks is a major concern for refiners tackling the stringent specifications on maximum sulphur levels in middle distillates and the increasing global demand of diesel fuel. A major challenge is the implementation of a shift from conventional to ultra-deep methods of desulphurisation. Meanwhile, the capacity of secondary conversion processes such as fluid catalytic cracking (FCC) and hydrocracking in refineries has steadily increased in converting the bottom of the barrel into high-value lighter products resulting in increased levels of hydroprocessing, which exerts a higher demand on refinery hydrogen systems. Previous methodologies on hydrogen network optimization have been developed mainly based on the assumption of fixed hydroprocessing performance with constant hydrogen consumption and light hydrocarbon yields, in order to reduce the complexity of the optimisation problem. Consequently, critical interactions among feed and catalyst properties, hydroprocessor operating conditions, product quality and yields, and hydrogen consumption are usually neglected. This research work involves three major aspects: 1. Development of semi-empirical nonlinear lumped hydrodesulphurisation (HDS) and hydrocracker models that are robust and sufficiently detailed to capture the behaviour of the process with changes in feed characteristics and operating conditions. The formation of light hydrocarbons during HDS reactions have been accounted for. Hydrocracker conversion models and five/six-lumped product yield models for vacuum gas oil (VGO) and vacuum residue (VR) feedstocks have been developed from a combination of first principles and empirical methods based on several process parameters. The proposed models are validated with different feedstocks and shows good agreement with industrial data. 2. Integration of HDS and hydrocracker performance models into refinery hydrogen network models to explore existing interactions between processes and the hydrogen network, and their combined effect on the overall network objective. 3. Optimization of the overall superstructure under different operating scenarios to facilitate the efficient distribution and utilization of hydrogen and the maximization of clean high-value products. The integrated superstructure network model is developed and optimized within the General Algebraic Modelling System (GAMS). The model is representative of the dynamic interactions between hydrodesulphurisation and hydrocracking processes in the refinery hydrogen network as demonstrated by the reproducibility of industrial refinery data. Thus, this work presents a holistic and realistic implementation of refinery hydrogen management technique.