Heavier crude oil, tighter environmental regulations and increased heavy-end upgrading in the petroleum industry are leading to the increased demand for hydrogen in oil refineries. Hence, hydrotreating and hydrocracking processes now play increasingly important roles in modern refineries. Refinery hydrogen networks are becoming more and more complicated as well. Therefore, optimisation of overall hydrogen networks is required to improve the hydrogen utilisation in oil refineries. In previous work for hydrogen management many methodologies have been developed for H2 network optimisation, all with fixed H2/Oil ratio and H2 partial pressure for H2 consumers, which may be too restrictive for H2 network optimisation. In this work, a variable H2/Oil and H2 partial pressure strategy is proposed to enhance the H2 network optimisation, which is verified and integrated into the optimisation methodology. An industrial case study is carried out to demonstrate the necessity and effectiveness of the approach. Another important issue is that existing binary component H2 network optimisation has a very simplistic assumption that all H2 rich streams consist of H2 and CH4 only, which leads to serious doubts about the solution's validity. To overcome the drawbacks in previous work, an improved modelling and optimisation approach has been developed. Light-hydrocarbon production and integrated flash calculation are incorporated into a hydrogen consumer model. An optimisation framework is developed to solve the resulting NLP problem. Both the CONOPT solver in GAMS and a simulated annealing (SA) algorithm are tested to identify a suitable optimisation engine. In a case study, the CONOPT solver out-performs the SA solver. The pros and cons of both methods are discussed, and in general the choice largely depends on the type of problems to solve.
|Date of Award||1 Aug 2011|
- The University of Manchester
|Supervisor||Nan Zhang (Supervisor) & Robin Smith (Supervisor)|