1. Modeling & control of Reactive Distillation
Jianjun Peng
Supervisors: Dr. Edgar Dr. Eldridge

2. Outline Background information Research objectives Modeling
Experimental plans
Conclusions
02/19/2001
TWMCC

3. Reactive Distillation: Example
A+B
Conventional process C
A, B B A
reactor E
A,B,E A C B
Reactive distillation
Reactive section
02/19/2001
TWMCC

4. The Good vs. the Bad
The good + higher conversion + reduced capital cost, energy
The bad more difficult to design & control poor understanding of the process
02/19/2001
TWMCC

5. AspenPlus Simulations
AspenPlus Radfrac
Equilibrium model
Tert-Amyl Methyl Ether (TAME) system
Steady state
Objective: How different is reactive distillation comparing to ordinary distillation?
02/19/2001
TWMCC

6. Reflux Ratio Influence
Modeling & Control of Reactive Distillation
Reflux Ratio Influence
Reactive distillation is quite different from ordinary distillation because of the complex interactions between mass transfer and reaction. The process model is highly nonlinear with multiplicity. Therefore it is quite difficult to model, design and control the process.
There are growing industrial interests in reactive distillation. Several commercial applications of reactive distillation are very successful and there are many potential applications.
Rate-based modeling is a new way to model chemical processes. It has several advantages over the traditional equilibrium modeling. However, the model is more complex and more different to solve efficiently.
NMPC can handle nonlinearity, multiplicity and constrains and therefore is an attractive strategy for controlling reactive distillation. Model reduction and efficient optimization techniques are needed.
There are very few experimental data in the open literature. Therefore the experimental data in this work will be very useful for model and controller validation and improvement, and for understanding the reactive distillation process in depth.
02/19/2001
TWMCC

7. Modeling & Control of Reactive Distillation
Pressure Influence
Reactive distillation is quite different from ordinary distillation because of the complex interactions between mass transfer and reaction. The process model is highly nonlinear with multiplicity. Therefore it is quite difficult to model, design and control the process.
There are growing industrial interests in reactive distillation. Several commercial applications of reactive distillation are very successful and there are many potential applications.
Rate-based modeling is a new way to model chemical processes. It has several advantages over the traditional equilibrium modeling. However, the model is more complex and more different to solve efficiently.
NMPC can handle nonlinearity, multiplicity and constrains and therefore is an attractive strategy for controlling reactive distillation. Model reduction and efficient optimization techniques are needed.
There are very few experimental data in the open literature. Therefore the experimental data in this work will be very useful for model and controller validation and improvement, and for understanding the reactive distillation process in depth.
02/19/2001
TWMCC

8. Product Rate Influence
Modeling & Control of Reactive Distillation
Product Rate Influence
Reactive distillation is quite different from ordinary distillation because of the complex interactions between mass transfer and reaction. The process model is highly nonlinear with multiplicity. Therefore it is quite difficult to model, design and control the process.
There are growing industrial interests in reactive distillation. Several commercial applications of reactive distillation are very successful and there are many potential applications.
Rate-based modeling is a new way to model chemical processes. It has several advantages over the traditional equilibrium modeling. However, the model is more complex and more different to solve efficiently.
NMPC can handle nonlinearity, multiplicity and constrains and therefore is an attractive strategy for controlling reactive distillation. Model reduction and efficient optimization techniques are needed.
There are very few experimental data in the open literature. Therefore the experimental data in this work will be very useful for model and controller validation and improvement, and for understanding the reactive distillation process in depth.
02/19/2001
TWMCC

9. Research Objectives Dynamic model - for the purpose of control
Model predictive control - PID may not be adequate
Controller implementation - pilot plant with Delta V control system
Experimental validation
02/19/2001
TWMCC

10. Equilibrium Models
Vapor-liquid equilibrium at each stage (section for packed column)
Tray efficiency or HETP
One mass balance for each stage
02/19/2001
TWMCC

11. Rate-based Models Mass transfer equations
Vapor-liquid equilibrium only at interface
Transport properties mass transfer coefficients heat transfer coefficients
NO tray efficiency or HETP
02/19/2001
TWMCC

12. Rate-based Models(2) Vk yi,k Lk-1 xi,k-1 fLi,k fVi,k Vapor Liquid
Catalyst
N, E
N, E
QVk
Vk+1 yi,k+1
Lk xi,k
QLk
02/19/2001
TWMCC

13. Equilibrium or Rate-based?
Equilibrium models Rate-based models
- Not rigorous Rigorous
+ Simple Complicated
? Tray efficiency or HETP ? Mass transfer
02/19/2001
TWMCC

14. Model Comparison
Jin-Ho Lee etc. (1998) individual efficiency hard to predict rate-based model is preferred
R. Baur (2000) smaller window for multiplicity in rate-based model rate-based model is preferred
No experimental validation
No details about mass transfer
No details about the behavior of reactive distillation
02/19/2001
TWMCC

15. Modeling
Mass transfer - Maxwell-Stefan equations - Overall mass transfer? - Empirical mass transfer coefficients
Reaction - heterogeneous or pseudo-homogeneous?
Dynamics - vapor holdup? - energy holdup?
02/19/2001
TWMCC

16. Model Assumptions Overall mass transfer Pseudo-homogeneous reaction
Pseudo-steady state energy balances
negligible vapor holdup
02/19/2001
TWMCC

17. Model Solution
Aspen Custom Modeler (ACM) * custom models * built-in DAE solvers * built-in property models * integrated PID controllers * modeling language
02/19/2001
TWMCC

18. Simulation Plans Comparison with equilibrium model
Comparison with more rigorous rate- based model (Sebastien Lextrait)
Parameter influence - reflux ratio, boil-up ratio, pressure, feed composition
Dynamic response - feed, reflux ratio, boil-up ratio
02/19/2001
TWMCC

19. Experimental Plans 6 inch reactive distillation pilot plant
Catalytic
Packing
Structured
Reactive Distillation
Column
6 in. diameter
34 ft. T-T
Backcracking
Reactor
6 inch reactive distillation pilot plant
TAME system
Experiments steady state dynamic controller implementation
02/19/2001
TWMCC

20. Future Work Solving the model with ACM Simulations and comparisons
Experiments: steady state and dynamic
MPC and NMPC implementation
02/19/2001
TWMCC

21. Concluding Remarks
Reactive distillation is advantageous, but poorly understood.
A dynamic rate-based model has been developed.
Future contributions
Solving the rate-based dynamic model
Controller development using simulations
MPC implementation on Delta V
Experimental validation
02/19/2001
TWMCC