(146e) Title TBD

In 2025, polyolefins are expected to be a 348 billion-dollar industry. This value corresponds to an annual production far surpassing 150 million-tons of products such as polyethylene, polypropylene, among others. The ever increasing polyolefin production incurs, however, in increased production costs, energy usage, and environmental impact. The objective of this study is to show that control strategies such as dynamic optimization, data reconciliation, and nonlinear predictive control can reduce the amount of off-spec products and provide faster reactor transition between product grades. GEKKO and a simplified polypropylene fluidized bed reactor were used for this study. GEKKO is an optimization software for mixed-integer and differential algebraic equations. The simplified reactor model comprises of a nonlinear empirical correlation of the polypropylene melt index (MI) subject to monomer, comonomer, and hydrogen flow rates. GEKKO, therefore, continuously estimates the model's parameters and optimize (through model predictive control) the best flow rates combination that yields the highest production rate while minimizing grade transition time and off-spec products (optimizer's constraints) following given MI set points. GEKKO works as a dynamic estimator and controller of the reactor. In conclusion, GEKKO's ability to constantly update its model's parameters and use nonlinear model predictive control to optimally meet set points represents a huge benefit for the industry by minimizing waste, energy consumption, and production costs.