Modeling of Reactive Distillation John Schell Dr. R. Bruce Eldridge Dr. Thomas F. Edgar

Outline Outline Overview of Reactive Distillation Project Overview Tower Design Steady-State Models Dynamic Models and Control Individual Work Column Design and Operation Validation of Models Preliminary Dynamics and Control Studies Future Work

Reactive Distillation Homogeneous or Heterogeneous/ Catalytic Distillation First Patents in 1920s Applied in 1980s to Methyl Acetate Common applications: Ethylene Glycol MTBE, TAME, TAA

Favorable Applications Westerterp (1992) Match between reaction and distillation temperatures Difference in relative volatility between product and one reactant Fast reaction not requiring a large amount of catalyst Others: liquid phase reaction, azeotrope considerations,exothermic reactions

Subawalla Approach (Dissertation) 1. Decide on a Pre-reactor - Rate of reaction - >1/2 of initial reaction rate at 80% of equilibrium conversion 2. Pressure 3. Location of Zone 4. Estimate Catalyst - Isothermal Plug-flow reactor with ideal separators 5. Design Tower - Size reaction zone • Catalyst requirements • Column diameter - Determine reactant feed ratio - Feed location - Reflux ratio • High reflux rate - 2-3 times non-rxtive column - Diameter • Through-put • Catalyst density

Project Overview Design and Construct TAME Column Validate Steady State Models Develop Dynamic Models Test Control Algorithms

TAME Chemistry TAME Chemistry Exothermic Equilibrium Limited 45-62% at 50-80 C Azeotropes Catalyst: Amberlyst-15 Methanol can inhibit rates. Rihko and Krause (1995)

Pilot Plant (SRP) Pilot Plant (SRP) 0.152-meter diameter column Finite reflux 7 meters of packing in 3 sections Fisher DeltaV Control Koch’s Katamax packing Makeup MeOH C5 from Cat Cracker Pre-Reactor Reactive Distillation Column Mixing Tank Back - Cracking Reactor Recycle TAME Unreacted C5, MeOH 3.7 atm

SRP Pilot Plant SRP Pilot Plant Koch – Spool section, Katamax, Catalyst SRP - $145K

Steady-State Multiplicity Bravo et al. (1993) Observed multiple steady-states in TAME CD Hauan et al. (1997) dynamic simulation provided evidence in MTBE system Nijuis et al. (1993) found multiplicity in MTBE system Jacobs and Krishna (1993)

Steady-State Distillation Models Trayed Tower: Equilibrium Model Rate Model Packed Tower: Continuous Model

TAME Reaction Rates TAME Reaction Rates

TAME Concentration Profile

Effective Reaction Rate Traditionally simulations use intrinsic reaction rate. Effective rate is a function of intrinsic rate and diffusion limitations. Molefraction Effective Rate

Control for TAME Tower Control for TAME Tower Fisher DeltaV Visual Basic Matlab, Visual Studio State Estimation Temperature Profiles Online Analyzers Control Algorithms PID Linear MPC Non-Linear MPC

Individual Work Design and Construct RD Column for Novel System Steady State Model Validation Dynamic Models and Control Study

Novel System A + B C1 C3 C2 Kinetic Reaction Exothermic Not Equilibrium limited Equilibrium Isomers Exothermic Kinetics from CSTR Experiments Feed is dominated by inerts Replace hazardous heterogeneous catalyst

Novel System Data Novel System Data

Novel System Data Novel System Data

Simulation Validation - 50 psig

Simulation Validation – 35 psi

Effect of Pressure

Effect of Varying Feed Rate

Dynamic Modeling and Control Study Aspen Custom Modeler/ Aspen Dynamics Validate Steady State Solution Validate Dynamic Studies Develop Control Algorithms PID Linear MPC NLMPC

Equations vs. Variables Aspen Custom Modeler Aspen Custom Modeler Formerly Speed-Up and DynaPlus Equation Solver Aspen Properties Plus Tear Variables automatically selected Solves Steady-State and Dynamic Dynamic Events and Task Automation Equations vs. Variables

Validation of Dynamic Simulator

Feed Disturbance With Manual Control

Control of Reactive Distillation Configurations DB LV BV, LB… Goals Conversion Product Purity F R D B V L Duty

Control of Reactive Distillation Bartlett and Wahnschafft (1997) Simple Feed-Forward/ Feed-Back PI Scheme Sneesby et al. (1999) Two point control with linear conversion estimator Kumar and Daoutidis (1999) Showed linear controllers unstable for ethylene glycol systems Demonstrated possible Nonlinear MPC scheme

Dependency of Conversion on Reboiler Duty and Reflux Ratio

Conversion vs Reboiler Duty

Single Tray Conversion Estimation

Single Tray Purity Estimation

Feed Disturbance With Manual Control

Feed Disturbance with Simple PID Control

Conclusion and Future Work TAME Tower Collect Data Validate Models Developing Advanced Models Improvements New chemical system Adjust for better dynamic studies Novel System Validate Dynamic Models Develop Control Algorithms