Integrated Process and Control System Design of Polymerization Reactors under Uncertainty. Optimal Grade Transition Operation

ASTEASUAIN, MARIANO; BRANDOLIN, ADRIANA; SARMORIA, CLAUDIA; BANDONI, ALBERTO

Rio de Janeiro

Congreso; 2nd Mercosur Congress on Chemical Engineering - 4th MERCOSUR Congress on Process Systems Engineering (ENPROMER 2005); 2005

Comité Organizador

Unlike traditional, large-scale petrochemical industries that focus on a single product, polymer manufacturers produce a number of polymer grades in order to achieve the property requirements of different applications. These grades are produced over a wide range of conditions, and changing from one grade to the next in the most time- and cost-efficient manner is a main objective of a well-designed process. The particularly difficult control problem of a grade transition in a polymerization reactor should be able to cope with process perturbations and uncertainty in the operating conditions (i.e. feed compositions and temperatures) and model parameters (i.e. global heat transfer coefficient). However, this has seldom been considered in previous optimal grade transition studies. Moreover, in the last decades, the significant benefits of integrating control aspects in the early stages of process design have been shown. However, polymer engineering is just now incorporating this important methodology. In this work, a simultaneous process and control system design is performed for optimal grade transition operation in a styrene polymerization reactor. A previous model is extended to consider uncertainties in operating conditions and model parameters, as well as perturbations in process variables. The process design, which includes reactor unit and initiator type selections, and steady-state operating points, is performed simultaneously with the reactor control system. This task involves finding the best combination of controlled ? manipulated variables, and the optimal controllers? tuning parameters. Discrete design decisions are incorporated by means of discrete optimization variables. The simultaneous process and control system design problem results in a Mixed Integer Dynamic Optimization (MIDO), which is solved with the software gPROMS/gOPT (Process Systems Enterprise, Ltd.). An optimal process ? control system design is obtained, which minimizes off-specification product during grade transition, and at the same time guarantees feasible operation in the full range of uncertain parameters for the considered process perturbations.