Dynamic Modeling and Model Predictive Control for Packed

Dynamic Modeling and Model Predictive Control for Packed Reactive Distillation Processes Presented by: Sebastien Lextrait Advisor: Dr. Thomas F. Edgar February 18, 2000

Overview • • • Introduction Control Issues My Research Objectives Implementation Aspects Conclusions February 19, 2001

Introduction • Drawbacks – Complex dynamic behavior – Reduced operability and controllability • Active areas of research – Design methods – Analyses of behavior – Control schemes February 19, 2001

Control Issues • Objectives – Reactant conversion – Final product purity • Problems – Transition between multiple steady states – Non-minimum phase behavior February 19, 2001

Control Issues • Control strategy literature – Difficulties with traditional controllers • Purity and reactant conversion • Multiple steady states • Limitations of traditional feedback controllers – Advanced control schemes • Dynamic matrix control method • Output feedback control scheme for some high index DAEs • Nonlinear model predictive control strategy February 19, 2001

Control Issues • Investigation of MPC control strategy – Performance dependence on model complexity – Problem formulation • On-line implementation not addressed – Computational time • Experimental validation of proposed control schemes February 19, 2001

My Research Objectives • Four main objectives – Develop dynamic models of different complexity – Model validation and behavior analysis – Investigation of MPC strategy – On-line controller implementation and experimental validation February 19, 2001

My Research Objectives • Develop dynamic models of different complexity – Equilibrium versus rate-based – Physical properties – Hydraulic model (liquid holdup, vapor pressure drop) – Catalyst description (diffusion limitations) February 19, 2001

My Research Objectives • Model validation and behavior analysis – Experimental data • Steady-state • Dynamic – Study the effect of modeling assumptions on the overall model behavior • Computational time • Response accuracy February 19, 2001

My Research Objectives • Investigation of MPC strategy – Nonlinear programming problem (simultaneous approach) – Investigate state estimation and observer issues for this process – NMPC versus PID and linear MPC – Commercial Software Packages February 19, 2001

My Research Objectives • On-line controller implementation and experimental validation – Numerical simulations – Real-time optimization issues – Pilot Plant Testing – Interface NOVA to Fisher Rosemount Delta. V distributed control system February 19, 2001

Implementation Aspects • NOVA optimization and DAE system – Flexible environment – Optimization and simulation of DAEs – Nonlinear constrained optimization problems – Large-scale problems – Method of weighted residuals on finite elements February 19, 2001

Implementation Aspects • On-line controller implementation February 19, 2001

Conclusions • Dynamic modeling of packed reactive distillation required • Investigation of MPC strategy for this nontraditional chemical process • Adapt control scheme for on-line use • Validation with experimental data February 19, 2001