Study On Lumping Kinetic Modeling Of Pyrolysis For Heavy Vacuum Gas Oil And EPSOS Perfecting

Ethylene industry is an important pillar of the petrochemical industry, the level of ethylene production is a measure of a country’s main indicator of the development of petrochemical industry. Due to the low ethylene self-sufficiency rate in China, the capacity of ethylene production remains to be enhanced in the next few years. Since domestic raw materials of cracking is increasingly heavier, in order to meet the demand for pyrolysis process development and optimization with heavy raw materials, an in-depth research for cracking kinetics and development of the corresponding model are urgently needed.Firstly, this paper analyzes the status of pyrolysis reaction model, scans the theory of lumping and its application in cracking, which help to summarize the overall criteria of lumping division. In comparing with naphtha, light diesel oil and other raw materials which are commonly used in cracking, we evaluate basic cracking performance of HVGO, which stands for Heavy Vacuum Gas Oil.Secondly, this paper proposes a nine-lumping model of HVGO pyrolysis. According to the reaction network, a mathematical model based on easily measured industrial data is established, which makes the lumping model more practical. In addition, several sub-models are set off to simplify the difficulty of parameters fitting. The premise of sub-model method is also proposed. The idea of hierarchical lumping model, which combines the molecular model, can enhance the analytical capacity with limited lumps.Thirdly, the parameters regression method is improved. After analyzing the effect of residual inconsistency on the simulation accuracy, the objective function is optimized and a Marquardt method with weight coefficient is established. Through an example of five-lumping model, we verify the necessity of weight coefficient in the new algorithm. Finally, during the application of EPSOS in an straight-run naphtha, this paper analysis the deficiency of Kumar’s molecular model and the software itself. The material characterization, enthalpy calculation and pressure prediction are hence improved.