Modeling and Control of Polymerization Reactors: An Industrial Perspective

1
Modeling and Control of Polymerization Reactors
An Industrial Perspective

• Dynamics and Control Seminar Series
• College of Engineering
• University of Delaware
• May 12, 1999

John R. Richards John P. Congalidis DuPont
Central RD
2
Outline

• Motivation
• Current trends
• Practical considerations
• Illustrative example
• Conclusions

3
Motivation

• Safety
• Global competition
• Cost pressure
• Quality improvements
• Environmental concerns
• Energy minimization
• Minimum investment designs
• Reduce inventories
• Benchmark study

4
Evolution of Process Control Technology

ELECTRONIC ANALOG
?
MANUAL
PNEUMATIC
DIGITAL
Pre 1940s
1940s-50s
1950s-60s
1960s to date
INFORMATION FLOW
Pneumatic
Electrical
Electrical, Digital (Binary)
Electronic Analog
Digital Computer
Mechanical (Bellows, Springs)
COMPUTATION
DOMINANT TECHNOLOGY
Basic Feedback (PID)
Feedforward/ Cascade
Model Predictive Control
5
Current Trends

• Industry is replacing pneumatics
• Distributed control systems (DCS) becoming
  standard
• Smart instruments
• Increased computer power available online
• Process data stored on computer are being
  routinely analyzed
• Optimization of business supply chain

6
Practical Considerations

• Measurement
• Nonexistent?
• Infrequent?
• Poorly maintained?
• Relationship between reactor operating conditions
  and customer related final polymer properties
• Uncontrolled disturbances
• Equipment variation
• Multiproduct plants
• Maintenance of control systems

7
DuPont Benchmark Study

• Estimated stake of about 500 million/year in
  DuPont from the application of process control
• About 50 of the stake dependent on improved
  measurements
• Only 20 of the applications require advanced
  process control
• Represents 50 of the savings
• Not using enough new technology
• Reorganization of process control into short-term
  tactical and long-term strategic groups
• Dont have the 500 million yet!
• As quality improves, plant pushes harder to
  capture stake, it becomes harder to achieve
  remaining fraction of stake!

8
Polymerization Control The Literature

• Relatively small number of very active academic
  researchers
• Excellent review articles and a book
• Congalidis and Richards (1998)
• Dimitratos et al. (1991)
• Eliçabe and Meira (1988)
• Kiparrisides (1996)
• Macgregor et al. (1984)
• Penlidis (1992, 1994)
• Ray (1989, 1992)
• Schork et al. (1993, 1994)
• Very limited number of publications by industrial
  practitioners

9
Polymerization Control The Literature

• Methyl methacrylate is the most common monomer
  used either in homopolymerization or in
  copolymerization with other acrylates or vinyl
  acetate
• Wide range of topics studied both experimentally
  and theoretically
• Temperature control
• Programmed addition of initiator and reactants
• State estimators mainly using extended Kalman
  filters
• Advanced feedback controllers (adaptive,
  nonlinear, MPC)
• Reported industrial applications limited to
  ethylene and a-olefin polymerization reactors

10
Polymerization Control The Literature

• Heavily focused on free radical polymerization
  systems in bulk, solution, and emulsion
• Less emphasis on condensation polymerization
  systems
• Investigation and evaluation of novel sensors for
  online measurement and estimation of conversion
  and polymer properties
• A healthy mix of experimental and simulation
  studies in both CSTR, batch, and semibatch
  reactors including some reactor trains

11
Process Control Hierarchy
Process Control Hierarchy
Prod Planning, Sched. Opt.
Monitoring
Product Quality Control
Model-Based Control
Regulatory Control
Sensors, Analyzers, Transmitters, Actuators
THE PROCESS
12
Illustrative Example

• The process
• Congalidis, J. P and J. R. Richards, Process
  control of polymerization reactors An industrial
  perspective, Polymer Reaction Eng., 6(2), 71
  (1998).
• Large adiabatic reactor (under pressure) used to
  manufacture polymer latex from monomers A, B,
  and C
• Along with separator (at lower pressure) and
  monomer recovery units
• Main production objectives
• Production rate
• Copolymer composition
• Inherent viscosity (related to MWD)

13
Schematic Diagram
14
The Emulsion Polymerization Reactor
Monomer
Droplet
Monomer
Droplet
15
Particle Formation Mechanisms
16
Particle Growth Mechanisms
17
Model Development and Utilization

• First principles Emulsion Polymerization Model
  (EPM)
• Based on particle formation and growth mechanisms
  of emulsion polymerization
• Detailed material balances and thermodynamic
  relations
  
• Nonlinear system of 110 differential algebraic
  equations
• Richards, J. R., J. P. Congalidis, and R. G.
  Gilbert, Mathematical modeling of emulsion
  copolymerization reactors, J. App. Poly. Sci.,
  37, 2727 (1989).
• Predicts concentration particle size colloidal
  characteristics of latex particles monomer
  conversion copolymer composition molecular
  weight and branching averages

18
Model Validation Against Laboratory Data
19
Model Validation Against Laboratory Data
20
Model Utilization

• Process understanding
• Extensive simulation to study effects of
  operating conditions (reactor temperature,
  monomer feed composition, monomer feed rate) on
  reactor performance
• Operating condition/control structure
  specification
• Numerical sensitivity study provided guidance
  for determining appropriate operating conditions
  and control structure

21
Model Predictions of Reactor Operation
22
Model Predictions of Reactor Operation
23
Model Predictions of Reactor Operation
24
Process Understanding

• Sensitivity studies indicated
• Temperature control crucial
• Inherent viscosity extremely sensitive to
  temperature variations
• Monomer conversion sensitivity increases at high
  temperatures
• Locate reactor operating temperature away from
  low conversion region (T gt 120 C)
• One way to decouple product quality control
• Copolymer composition naturally very sensitive to
  changes in reactor feed composition but
• Inherent viscosity relatively insensitive to
  reactor feed composition changes!

25
Control System Overview

• Reactor temperature control
• Aqueous feed temperature reactor
  temperature
• Copolymer composition control
• Monomer feed rates reactor feed comp
  copolymer composition
• Inherent viscosity control
• Monomer to inherentchain transfer
  viscosityagent ratio

26
Copolymer Composition Control Strategy
27
Inherent Viscosity Control Strategy
28
Hardware for Control System Implementation
29
Impact of Control System on Plant Productivity

• Results
• Improved product quality yield (in terms of
  significant increase in first-pass,
  first-quality yield)
• Significant reduction in product quality
  variability as measured by the process
  performance index, Ppk (values gt 1.3 considered
  very good)
• Ignificant avings financial returns

30
Assessing Control System Performance
31
Assessing Control System Performance
32
Design of Control Scheme for Emulsion
Copolymerization Reactor

• A challenging problem because of highly
  interactive and nonlinear behavior
• First principles process model was used for
  control design
• Hierarchical approach
• Feedforward control eliminates measured
  disturbances
• Feedback control handles unmeasured disturbances
• Complex analysis, simple implementation

33
Future Trends

• Sophisticated computing capabilities will be
  standard in DCS
• Model based control will be more prevalent
• Focus on nonlinear, multivariable control
• Large part of DuPont businesses are solids
• Control particle size distribution
• Importance of optimization, scheduling, and
  transitions
• Incorporate process control into process design
• In the past christmas tree approach
• Make sure the plant is controllable before
  construction
• Plant-Wide Control
• New installations will be developed quicker
  because of new platform developments like IDCOM
• Close industrial partnerships with academic
  researchers