Modeling, Optimization, And Flowsheet Reconfiguration Of Polymerization Process With Embedded Molecular Weight Distributions

Increased worldwide competition requires economic efficiency from domestic polymerization processes to a significant extent. Therefore, a strong interest has motivated the achievement of higher productivity and quality of the polymer. Given that molecular weight distribution (MWD) is at the core in establishing key quality indices for polymers, there is a strong incentive to take MWD into account for the quality control and process operation optimization of polymer products. Nevertheless, due to the complexity of the physicochemical phenomena in polymerization reactions and the nonlinearities arising from the MWD calculation equations, modeling and optimization from MWDs in polymerization processes remains a complicated and time-consuming task. In this thesis, we focus on a multi-grade polymerization process. Based on a novel methodology of modeling, simulation, and optimization, we deal with simulation and optimization problems in polymerization processes on different levels, including reaction kinetics, unit operations, and flowsheet configurations.We first develop an equation-oriented (EO) rigorous model of a high-density polyethylene (HDPE) slurry process to describe the quantitative relation among process parameters, steady-state operating conditions, and end-use properties of the polymer. Based on this model, we conduct kinetic parameter estimation, steady-state process optimization and flowsheet reconfiguration with embedded MWDs. The optimal reactor network structure and operating policies overcome the limitations of conventional flowsheet configurations and lead to the improvement of the polymer end-use properties and productivity for the multi-grade polymerization process.The main framework and contributions in this thesis are listed as follows:(1) Develop an EO model framework for an industrial HDPE slurry process with an embedded MWD, including mass balance of the polymer components and MWD calculation based on the method of moments, thermodynamic surrogate kriging model based on the perturbed-chain statistical associating fluid theory equation of state (PC-SAFT EOS). The conservation relation among different process units is described under the rigorous kinetic mechanism.(2) Address kinetic parameter estimation for this HDPE slurry process model framework with MWD measurements. An estimability analysis using relative sensitivity coefficients and confidence intervals is conducted to determine a subset of estimable kinetic parameters. Then a novel multistep methodology is proposed to separate the ill-conditioned original estimation problem into well-conditioned subproblems, which is much less sensitive to initial values and can find optimal solutions very easily. Numerical results demonstrate the effectiveness of the optimization performance to obtain accurate parameter estimates in industrial cases.(3) Conduct process simulation and optimization under conventional flowsheet configurations with specified MWDs. Model is first validated by simulation with real plant data to show the good accuracy of calculated MWDs. With the help of the PAT (Parameter Automatic Tuning) strategy for NLP algorithms, we develop a systematic approach to solve the optimization problems based on the EO model. Productivity optimization with a specified MWD as a constraint is then proposed and solved by using this efficient simultaneous approach. Case studies of bimodal and unimodal MWDs show that the optimal steady-state operating policies maximize HDPE productivity with desired MWDs for different process configurations.(4) Propose a systematic theory and method of optimal reactor network synthesis of the HDPE slurry process with embedded MWDs. Based on conventional flowsheet configurations, a fully connected process superstructure of continuous stirred tank reactors (CSTRs) is first established through the introduction of splitters. Using this generalized superstructure as a basis, different optimal flowsheet configurations are generated to simultaneously maximize the monomer conversion and minimize the deviation between the calculated and target MWDs. Numerical results based on refined multiobjective optimization (MO) methods show that the optimal flowsheet configurations overcome the limitations of conventional reactor network structures and help to increase reactor productivity at the desired product quality.