Surrogate-assisted optimization of refinery hydrogen networks with hydrogen sulfide removal

Abstract Hydrogen management is very important for the production of clean fuels such as low-sulfur gasoline and diesel. As a key contaminant in the hydrogen network, hydrogen sulfide not only deactivates catalysts but also corrodes equipments. Hence, it is necessary to integrate hydrogen sulfide removal process model into the optimization of hydrogen allocation network. However, a desulfurization process model with simplifications and assumptions may yield suboptimal results and a model based on rigorous process mechanism may require high computational cost. To solve this problem, this paper proposes a novel approach for simultaneous optimization of refinery hydrogen allocation network and hydrogen sulfide removal processes. The optimization is realized in two steps. First, surrogate models are developed as approximations to the rigorous process and thermodynamic model for the desulfurization processes. Second, the surrogate model is embedded into the mathematical programming model for hydrogen network optimization. The effectiveness of the proposed approach is illustrated by its application to a case study taken from a real refinery. Result comparison between the proposed method and a literature model based on simplified desulfurization models is carried out, proving that, with the proposed method, more practical optimization result can be obtained with much less computational effort.