Dynamic Simulation And Optimization Of Molecular Weight Distribution For Ethylene Slurry Polymerization Process

Polymer molecular weight distribution(MWD) determines the properties and the performance of the polymer. As M WD is an extremely important quality index for the polymer, simulation and optimization of the MWD of the ethylene slurry polymerization process is of great significance.Based on the dynamic model of the ethylene slurry polymerization process, MWD is calculated through model decomposition and sequential variable decoupling method, Optimizating the operation of grade transition is also under studied.The main research contents and contributions of this thesis are sumarized as follows:1. Due to the very large dimension of the MWD, which can be up to105, direct simultaneous solving of the dynamic MWD model is very difficult. In this thesis, a decoupling method is proposed for rigorous model calculation of the MWD. The MWD model in large scale differential-algebraic equations (DAEs) is decoupled to a moment model in small scale DAEs and population balance model for chain length based MWD in large scale ordinary differential equations (ODEs). To solve the large scale ODEs, the sequential variable decoupling method is further proposed via numerical computation.2. To speed up the solving process of the large scale ODEs population balance model, the parallel sequential variable decoupling method has been implemented on the multi core processor based parallel computing platform, which enbles the simulation speed of the MWD to be significantly improved. Moreover, the first order derivatives of the MWD with respect to the operating condition variables are also calculated in parallel method, which can greatly reduce the time of each iteration during the optimization process.3. Reducing the grade transition time can save the transition material, which is also important for the industrial production. Optimization of the HDPE grade transition process based on MWD is also under studied. The optimization scheme employs a two-layer structure. The inner layer uses hydrogen quantity for optimization variables and the target MWD curve for the optimization goal, and the outer layer gradually reduces the feed time of hydrogen. The relative error between the optimizated MWD curve and the target curve along the full chain length is less than3%, while the switching time can be reduced by about30%.