Dynamic modelling of an industrial polypropylene reactor and its application in melt index prediction during grade transitions

A dynamic model for propylene polymerisation in a continuous stirred tank reactor of a Hypol process is developed in order to infer the melt index of polypropylene on-line. The dynamic model contains a simplified mechanistic model covering gas-liquid equilibrium, kinetics of polymerisation, material balance and simplified energy equilibrium. Based on this dynamic model, melt index of polypropylene in the grade transition can be calculated using operating data from an industrial reactor. In order to eliminate the model error caused by uncertainties and disturbances of the industrial process, a recursive prediction error method has been applied to update model parameters in on-line applications. The predicted melt index is compared with the laboratory data. It is shown that this dynamic model has a strong adaptability during grade transitions and can be applied on-line to infer the melt index of polypropylene. Based on this dynamic model, optimisation and control of grade transitions of polypropylene can be implemented.