INTRODUCTION TO PROCESS

CONTROL:
Theory and Practice

FDA
Thursday 29 June 2006

Babatunde A. Ogunnaike
Chemical Engineering Department
University of Delaware

1
 OBJECTIVES

To provide an overview of, and an

introduction to, process control
systems and their role in chemical
(and biological) manufacturing
processes
2
 OUTLINE
1. INTRODUCTION TO THE MANUFACTURING
PROCESS
2. PROCESS CONTROL & INDUSTRIAL
PRACTICE: AN OVERVIEW
3. BLOCK DIAGRAMS & SYSTEMS ANALYSIS
4. ELEMENTS OF CONTROL THEORY
5. APPLICATION CASE STUDY (Granulation)
6. SUMMARY AND CONCLUSIONS
Process Control Lecture
Notes 3
0. ILLUSTRATIVE EXAMPLES

The Stirred Heating Tank

Process Control Lecture
Process Notes 4
 Example 2
Dual Temperature and Level Control in Stirred Mixing Tank

CW HW
Tc, Fc TH, FH

T h

Process Control Lecture

Notes 5
 PART 1

introduction

6
 1. INTRODUCTION
The Chemical Process
An Integration of Processing Units
for Converting Raw Materials and
Energy into Finished Products (and
Ener
Energy).
Raw materials
CHEMICA
gy

L
Finished Products
PROCESS

Energy

Process Control Lecture

Notes 7
Example: VINYL ACETATE
PROCESS

Process Control Lecture

Notes 8
 The Biologic Process
An Integration of Processing Units
for Converting Raw Materials and
Energy into Finished Products (and
Energy). Cytokines
Mammalian Cell
Cultures Growth Hormones
Ener Growth Factors
Raw materials
gy Monoclonal Antibodies
BIOLOGIC
PROCESS
Finished Products

Energy
Process Control Lecture
Notes 9
Example: Penicillin

Process Control Lecture

Notes 10
Example: Penicillin

PRODUCTION OF L-AMINO
ACIDS
BY AMINOACYLASE: SATO and
TOSA
Tanabe Seiyaku Company, Ltd.
Process Control Lecture Osaka, Japan
Notes 11
 Process Operation
 Fundamental Objectives of
Process Operation

must be operated safely;

specified production rates must be
maintained;

desired product quality specs must
be met.
 Main Obstacles to meeting
objectives:

Manufacturing processes are
dynamic: (Process variables are
always changing with time)
Process Control Lecture
Notes 12
 Process Operation
 Consequences
 Must be able to monitor, and
 induce change in key variables related to
safety,
production rate
and product quality.
Process Control System needed for these
purposes

Process Control Lecture

Notes 13
 PART 2

process control and industrial

practice: an overview

14
 2. PROCESS CONTROL: AN
OVERVIEW
 ANATOMY OF THE MANUFACTURING PROCESS (A
Control Viewpoint)

Consists of Input, Output, (and State) Variables

Input Variables (u):

Independently stimulate system;

can thus induce change in internal conditions

subdivided into MANIPULATED and DISTURBANCE Variables

Output Variables (y):

By which information about internal states of the process are
obtained

For commercial processes add End-use/Product Quality
Variables (z):

Provide information about product
Process Controlcharacteristics
Lecture
Notes 15
The stirred heating tank
process.

Process Control Lecture

Notes 16
Actual Process Examples
 Distillation Column
 Manipulated, u: Reflux flow rates; Reboiler Heat input;
 Disturbance, d: Feed composition;
 Output, y: Select Top and Bottom tray temperatures
 Product Quality, z: Pour Point; Overhead Mole fraction of
light material, …
 Polymer Reactor
 Manipulated u: Initiator flow rates; Jacket cooling water
flow;
 Disturbance, d: Feed composition;
 Output, y: Reactor temperature and Conversion;
 Product Quality, z: Co-polymer Composition; Molecular
weight distribution; Melt index; Mooney viscosity; …
Process Control Lecture
Notes 17
 The Process Control
System
 WHAT IT DOES
 Monitors outputs;
 Takes decisions about how best to
manipulate inputs to obtain desired behavior;
 Implements these decisions.

 ITS DUAL TASKS

 Information gathering and transmission;
 Control action computation and
implementation.
Process Control Lecture
Notes 18
 Process Control
 TASKS AND HARDWARE
ELEMENTS

Measurement
[Sensors, Analyzers, etc]

Information Propagation
[Transmitters]
 Decision Making
[Controllers: Pneumatic,
Electronic Analog, Digital
computer]
 Corrective Action Implementation
[Final Control Elements, e.g.
Process Control Lecture
Valves, etc.] Notes 19
Structure of a process control
system
Control system goals

Process Control Lecture

Notes 20
 Control System
Configurations
 PRIMARY CONSIDERATIONS

Placement of the sensor in relation to the process and
the controller;

And how control decisions are made and implemented…
 POSSIBLE CONFIGURATIONS

Manual

Automatic

Feedback

Feedforward

Open Loop

Ratio

Cascade, ….

Process Control Lecture

Notes 21
The (automatic) feedback
configuration

Process Control Lecture

Notes 22
 Example of Feedback
Control

Concept of “Variability Transfer

From where it will “hurt”

To where it will NOT.

Steam Flow rate judiciously adjusted (absorbing variability)

to keep Tank temperature on target
Process Control Lecture
Notes 23
Furnace Temperature
Control

Fuel Flow rate judiciously adjusted

to keep product temperature on target
Process Control Lecture
Notes 24
The feedforward control
configuration

Process Control Lecture

Notes 25
The open loop control
configuration

Process Control Lecture

Notes 26
 The Effective Control System
 Main Objective
 To effect the expeditious transfer of
variability

away from product attributes (where we
cannot afford them)

to appropriate process variables (where
they are harmless)
 How
 By appropriate overall design and
implementation of all components
Process Control Lecture
Notes 27
 Control System & Process
Operation
 Fundamental Premise of Process Control
 The natural response of all dynamic
processes can be modified by the influence
of a controller;
 The objective is therefore to design and
implement the controller in such a way that
the dynamic response is modified
appropriately, in a desired fashion;
 But...

Process Control Lecture

Notes 28
 Process Control
 In reality:
 The extent to which the natural response
can be modified appropriately will usually
be determined by:

The nature of inherent process limitations;
(Time delays, inverse response, physical
constraints, etc)

The depth of our knowledge of intrinsic
process characteristics;

The versatility of the hardware elements
available for implementing the control system
Process Control Lecture
Notes 29
Comprehensive System
Components
 Data Acquisition
 Measurements of Process Variables and
 Product Characteristics and Attributes
 Information Processing
 Ascertaining the “true state’’ of the entire
process and the product from available
(often incomplete and noisy) data
 Decision Making
 Control action computation and
 Implementation
Process Control Lecture
Notes 30
 Historical Overview of
Industrial Process Control
 PRE 1940's [Manual Control]

Product and process demands not too strict;

Processes not too complex;

Task easy enough to accomplish by “manual
control”

 EARLY 1950's [Basic Feedback Control]


Increased production volume demands;

More complex process interconnections; more
“continuous” operation;

More compelling Process
needControl
for
Notes
Lecture
automation 31
 Historical Overview of
Industrial Process Control
 EARLY 1960's TO DATE [Computer Control]
 Increased process and product demands;

increased production volume demands now coupled with
tighter product specs;

increased emphasis on “low variability;”

increased stringency in environmental regulations;

tougher global economic competition; ...
 Even more complex process interconnections; with
energy integration; novel designs and operations
strategy; etc.
 Tasks dramatically more challenging; creating
need for more sophisticated, more powerful
techniques. Process Control Lecture
Notes 32
 Process Control Technology
Evolution

ELECTRONIC
MANUAL PNEUMATIC ANALOG DIGITAL
?
Pre 1940’s 1940’s-50’s 1950’s-60’s 1960’s to date

INFORMATION Pneumatic Electrical Electrical,

FLOW: Digital (Binary)

COMPUTATION: Mechanical Electronic Digital

(Bellows, Springs)Analog Computer

DOMINANT Basic Feedback+ Feedforward/ + Model Predictive

TECHNOLOGY: (PID) Cascade Control

Process Control Lecture

Notes 33
 BASIC STRUCTURE of
Contemporary
Industrial Process Control
Prod
Planning,
Process and Operation Optimization
Sched. & Opt.
Performance Assessment
Monitoring
Quality Assurance
Product Quality Control
Process
Model-Based Control Control
Regulatory Control
Instrumentation
Sensors, Analyzers, Transmitters, Actuators
Design and
THE PROCESS Operating Procedur

Process Control Lecture

Notes 34
 Ideal Configuration for
Pharmaceuticals
Process Optimization PROCESS SPC
MONITORING/ 6-Sigma
IMPROVEMENT Quality By Design

PAT PCA/PLS
INFORMATION Filtering
PROCESS PROCESSING Process Model
Raw data
State estimation
Sensors/Analyzers

PROCESS & Advanced

PROD. QUALITYModel-based
rocess Variable Adjustments CONTROL Base Regulatory
Process Control Lecture
Notes 35
 History of Modern Control
 1960’s video clip documentary of the first
application of computer control in a refinery.

Process Control Lecture

Notes 36
 PART 3

block diagrams and process systems

analysis

37
 3. BLOCK DIAGRAMS &
PROCESS SYSTEMS ANALYSIS
 Process System Representation
 In terms of “flow of information”
 Convenient “block format”

Input, u Output, y
PROCESS
Perturbation Response

Process Control Lecture

Notes 38
Problems of Systems
Analysis

INPUT, u System OUTPUT, y

R(y,x,u,)

Given any 2 of 3 components, one can deduce

the third, consistent with given information:
Gives rise to 3 problems.
Process Control Lecture
Notes 39
Problem 1: Process Dynamics
and Simulation

INPUT, u System Model ?

M(y,x,u,)

The “Forward Deductive” Problem

Process Control Lecture

Notes 40
Problem 2: Process
Modeling/System
Identification

INPUT, u OUTPUT, y
?

The “Inductive” Problem

Process Control Lecture

Notes 41
Problem 3: Process Control

? System Model OUTPUT, y*

M(y,x,u,)

The “Inverse Deductive” Problem

Process Control Lecture

Notes 42
 Process Systems Analysis
 Process Dynamics
 Provides the platform for understanding
intrinsic process characteristics;
 Process Control
 Utilizes this knowledge in conjunction with
design principles for the design and
implementation of effective controllers;
 Process Modeling
 Provides the rational basis for carrying out
the tasks involved in “Process Dynamics”
and “Process Control”
Process Control Lecture
Notes 43
 Process Dynamics
 WHAT is “Process Dynamics”?

Analyzing and characterizing the dynamic behavior of
processes
 WHY study “Process Dynamics”?

Because the intrinsic characteristics of a process
determine how well we can control it

Each class possesses certain “signature” characteristics
that set it apart.
 HOW do we characterize process behavior?

By using various classes of process models

Study process responses to well-characterized inputs

Relate observed response characteristics to model
characteristics

The distinctive characteristics of a process that indicate

what behavior to expect from
Process it are
Control encoded in its
Lecture
MODEL Notes 44
Process Control Lecture
Notes 45
 Process Modeling
 Objective
 Develop a mathematical model

which represents the relevant aspects of a
process,

to be used for process analysis, understanding
and control;
 Premises
 Functional relationship (dictated by natural
laws) exists between the process variables
 Modeling exercise directed towards
uncovering this relationship
Process Control Lecture
Notes 46
 Process Modeling
 Approach Philosophies

Theoretical (Mechanistic) Modeling

Seeks to arrive at the relationship through
systematic application of the laws of nature
to the most important phenomena

End result: a Theoretical (Mechanistic)
model

Empirical Modeling (Model
Identification)

Seeks to approximate the unknown
functional relationship with simple
mathematical functions using experimental
data
Process Control Lecture

End result: Empirical model
Notes 47
 Process Control: Automatic
 Base (Regulatory) Control
 Control of basic process variables
(manipulating u—flow, levels, etc, to control
y—temperature concentrations, etc)
 Single variable; automatic and algorithmic

 Advanced Control
 Higher level control of product quality
 Model-based; multivariable

 Digital (Using the computer as controller)

 Sets setpoint for base regulatory controllers

Process Control Lecture

Notes 48
 Process Control
 Statistical Process Control
 Associated with ascertaining product quality
 On the basis of infrequent lab measurements

 Statistics invoked


To deal with noise (process variability as well as
measurement variability)

In the classical case, also to deal with the high
cost of taking unnecessary control action.
 Not “automatic,” or algorithmic.

Process Control Lecture

Notes 49
 PART 4

elements of control theory

50
 4. ELEMENTS OF CONTROL
THEORY
 Control Systems Design
 Base level

Feedback (with PID control)

More complex strategies (Cascade/
Feedforward…)
 Advanced Strategies

Multivariable Control

Optimization-Based: MPC

Process Control Lecture

Notes 51
 A. BASIC PROCESS
CONTROL
 Feedback Control
 Conventional Feedback Controller Design
 Controller Tuning
 More Complex Structures
 Cascade,
 Feedforward and other enhancements
 Difficult Dynamics
 Time Delay
 Inverse Response
 Open Loop Instability
Process Control Lecture
Notes 52
 Feedback Control: An
Example
 Stirred Heating Tank Process
 Characteristics
 1 input (manipulated) variable
(Q); u = Q – Qs
 1 output (controlled) variable (T);
y = T – Ts
 1 disturbance variable (Ti);
d = Ti - Tis
 Operating Objective
 Maintain tank temperature at
desired value Td
 Consider feedback control
strategy Process Control Lecture
Notes 53
 The Feedback Control
Strategy
(i) Measure T (y, as a deviation from steady state
Ts) to produce Tm (ym) using a measuring device
(thermocouple)
ym = Tm – T s
(ii) Compare with set-point to generate feedback
error
ε = yd – ym (yd = Td - Tds)
(iii) Send ε to controller where corrective action u (in
the form of a command signal c(t)) is computed
and sent to the valve
(iv)Implement as u by the control valve
Process Control Lecture
Notes 54
(v) Measure y again and repeat
 Overall Block Diagram
Closed-loop block
Physical diagram
Process

Process Control Lecture

Notes 55
 Classical Feedback
Controller
The Proportional + Integral + Derivative (PID)
Controller
p(t) = Kc [ ε(t) + 1/I ε(t)dt + Ddε/dt ]
 + ps
Transfer function
gc(s) = Kc [1 + 1/(is) + Ds ]

Proportional Gain Integral time Derivative time

Controller Design: Concerned primarily with determining

appropriate values to use for controller parameters
Process Control Lecture
Notes 56
 Conventional feedback
controller design
 Preliminary Considerations
 Controller: A modifier of process dynamics
 Controller design: Choosing controller type
and controller parameters so that the closed
loop behavior is modified appropriately
 Overall Control system elements also critical
to successful controller design

Sensors and transmitters

Valves and other final control elements

Appropriate commercial control hardware
components
Process Control Lecture
Notes 57
 Controller Design
Principles
 Feedback Controller Problem Statement

Choose controller type and controller parameters

Such that (objective) criteria of closed-loop
performance are satisfied
 Performance Criteria

Stability: Closed-loop system should

Be stable

Drive output to desired set-point and keep it there; not
too fast, not too slow

Steady State: Should show no steady state offset
(preferably)

Dynamic: Typically subjective, but some objective
criteria such as Time-Integral criteria are used
Process Control Lecture
Notes 58
 Controller Type decision
 Characteristics of classical feedback
controllers

P Controller

Accelerates control system response

Leaves a non-zero steady-state offset (except for pure
capacity processes)

PI Controller

Eliminates offsets

But response becomes more oscillatory

Added integral action tends to increase propensity
towards instability as Kc increases

PID Controller

Integral action eliminates offset

Oscillatory action curbed somewhat by derivative action

Presence of derivative action tends to amplify noise
components of noisy signals
Process Control Lecture
Notes 59
 Guidelines for selecting
classical feedback controllers
 P Controller
 When steady state offsets are
unimportant
 When process possesses a natural

integrator
 Common Industrial application: Many

liquid level control loops

Process Control Lecture

Notes 60
 Guidelines for selecting
classical feedback controllers
 PI Controller
 When steady state offsets cannot be
tolerated
 When oscillatory responses are OK.

 Common Industrial application: A large

number of typical industrial loops

Process Control Lecture

Notes 61
 Guidelines for selecting
classical feedback controllers
 PID Controller
 When it is important to compensate for
natural process sluggishness, and signals
are reasonably noise free;
 When there are time delays or other

dynamics requiring “anticipation”

 Common Industrial application: Temperature

control loops with significant measurement

lag and/or delay
 AVOID USING FOR FLOW LOOPS (or other

excessively noisyProcess
signals)
Control Lecture
Notes 62
 Controller Tuning
 Basic Principles
 Concerned with rationally choosing controller parameters
 Always based on (explicit or implicit) process
characterization
 Tuning rules or techniques depend on the type of process
characterization employed (frequency domain or time
domain; explicit model or implicit model)
 Typical Practical Procedure
 Obtain process characterization
 Use favorite tuning rule (usually available in Tabular
form)
 Test controller in simulation and refine tuning before
implementing on actual process
Process Control Lecture
Notes 63
 Conventional Controller Tuning

With a Fundamental Without a Fundamental

Process Model Process Model
No
Mo
d el
at
all
Optimization
Approximate Model Relay Autotuning
Model Following
Kcu; Pu;
Stability Margins (Table 15.1)
(Table 15.1) Step Response Frequency Response

(Process Reaction Curve) Bode Stability

Kcu; Pu
Ziegler-Nichols (15.2) (Table 15.1)
Cohen-Coon (15.3)
ITAE (15.4)
Process Control Lecture
IMC (15.6) Notes 64
 Autotuning
 What
 A procedure for determining process stability
characteristics in the closed loop, experimentally, during
regular process operation;
 And (automatically) determining from these, appropriate
controller tuning parameters
 How
 Using a simple on-off relay controller (such as is found in
the home furnace)
 Introduce low-amplitude input, (with magnitude h),
 Generate oscillatory process output with amplitude A
 Requirements
 A relay controller (in hardware or software)
 Provided as autotuning cards in electronic controllers
Process Control Lecture
 Provided as software in distributed
Notes control systems 65
Design of more complex
structures
 Motivation
 Cascade Control
 Feedforward Control
 Ratio Control

Process Control Lecture

Notes 66
 More Complex Structures
 Motivation
 Feedback (alone) inadequate under
certain conditions

(everyday example: Driving!)
 Significant disturbances affecting y
and/or u cause performance to
deteriorate
 Improve performance by additional

augmentation of feedback
Process Control Lecture
Notes 67
 Popular techniques
 Cascade Control

For significant disturbances affecting u
 Feedforward Control

For significant (measurable)
disturbances affecting y directly
 Ratio Control

For coping with a wildly varying
disturbance by maintaining a constant
ratio between it and a manipulated
variable Process Control Lecture
Notes 68
 Cascade Control
 Example Illustration

Figure 16.2. Block diagram for Fig 16.1 process

igure 16.1. Conventional feedback control system (with two disturbances)
f a distillation column's bottom temperature.

Process Control Lecture

Notes 69
 Cascade Control
 Example Illustration
 Original configuration: Temperature controller
manipulates control valve directly
 Primary problem:

Steam flow rate subject to disturbances in header pressure

Identical valve openings yield different steam flow rates

Result: Inconsistent and unacceptable performance
 Solution: Install a flow controller whose responsibility
is to ensure the right flow rate demanded by the
temperature controller is achieved (regardless of the
fluctuations in header pressure)

Process Control Lecture

Notes 70
 Cascade Control

Figure 16.5. Block diagram for Fig 16.2 process

gure 16.4. Cascade control of a distillation
(showing inner and outer cascaded loops)
olumn's bottom temperature

Process Control Lecture

Notes 71
 Cascade Control
 Main Features
 Two nested controllers

Outer (Primary) loop has the MASTER controller

Inner (Secondary) loop has the SLAVE controller
 For effectiveness, inner loop dynamics must
be faster
 Controller Tuning
 Start with inner loop; tune tightly (can be
pure proportional control)
 With inner loop in operation, tune outer loop
Process Control Lecture
Notes 72
Base Regulatory Control
System for a Fermentor

Process Control Lecture

Notes 73
 Feedforward Control
 Basic Premises
 When significant disturbance variations affect y
directly, and
 A reasonable model of the effect of d on y exists, and
 It is possible to compensate for this effect before the
process is affected (i.e. the physics allows it)
 Illustration

Figure 16.7. The stirred mixing

tank under feedforward control

Process Control Lecture

Notes 74
 Feedforward Control
 Practical Considerations
 Strategy is not possible if:

d is not measurable;
 The transfer functions gd and g are not known
sufficiently accurately
 gd/g is unrealizable (for example, if the delay in g is
longer than the delay in gd –think about the physics in
such a case!)
 When strategy is possible, implement with lead/lag
element or else with a computer
 Feedforward control does not affect closed loop
stability
 Typical to augmentProcess
feedforward with feedback
Control Lecture
Notes 75
B. ADVANCED PROCESS
CONTROL
 Introduction to Digital Control
 Introduction to Model-based and
Multivariable Control

Process Control Lecture

Notes 76
 Intro to digital control
 The Digital Computer: Key attributes of
interest
 High speed execution of complex
computations
 Tremendous capacity for mass storage and
rapid retrieval of data
 On-line data acquisition capability
 Consistent, sustained drop in cost
Perfectly suited for use as a controller
Process Control Lecture
Notes 77
 Introduction to digital
control
 The Digital Computer as a controller
 A controller: simply a device for computing
control action according to a pre-specified
control law;

See digital computer attributes above
 Offers greater flexibility and versatility than
the classical controllers

The control “law” need no longer be restricted to a
hardwired analog circuit (e.g., for PID control):

any control law, no matter how unconventional or
complicated, can be programmed on a computer.
Process Control Lecture
Notes 78
 Intro to Digital Control
 Key characteristics of Digital Control
 Computer receives and sends information only
at discrete points in time (not continuously)
 Process acquires a new characteristic as a
DISCRETE-TIME or SAMPLED-DATA system
 Implications
 Continuous y must be “sampled” (Analog-to-
Digital conversion)
 Discrete u(tk) must be made continuous.
(Digital-to-Analog conversions)
Process Control Lecture
Notes 79
 Intro to Digital Control

Figure 23.1. A continuous signal discretized.

Figure 23.2. A discrete signal from the computer and the intended continuous version.
Process Control Lecture
Notes 80
Block Diagram
Components

Figure 23.3 Typical block diagram for a computer control system.

Process Control Lecture

Notes 81
 Intro to Digital Control
 Critical Points to consider
 Choosing sampling time t
 Choosing “hold” mechanism

 Appropriate model form and analysis

techniques
 Controller design strategies

Process Control Lecture

Notes 82
 Sample and Hold
 Key Results of Interest
 Choose t ≈ 0.2 dom; (dominant time
constant)
 Zero-order hold most common

hold u constant until new value arrives

“staircase” appearance

Gives rise to apparent delay. An important, but
often ignored characteristic of digital control!
Figure 23.7. Apparent signal
delay
induced by the ZOH element.

Process Control Lecture

Notes 83
 Modeling and Controller
Design
 Appropriate models

Discrete-time difference equation models (equivalent
of continuous ordinary differential equation models)

Discrete z-transfer functions g(z) (equivalent of
continuous s-transfer functions, g(s))
 Controller Design Strategies

Full-scale digital control designs

Approximation of continuous controller design in
discrete form

Very popular; gc(s) translated to gc(z)

Especially for PID controllers must take into account the
apparent delay introduced by sampling.

FIRST ORDER SYSTEM UNDER P-ONLY CONTROL IS

ALWAYS STABLE IN CONTINUOUS-TIME; CAN BE
UNSTABLE IN DISCRETE-TIME
Process Control Lecture
Notes 84
 “Classical” Digital Control
 Digital PID Controller
 “Position” form (approximate integral with sum;
derivativewith finite
t kdifference)
 
u (k ) K c  (k ) 
I
  (i)  t [ (k )   (k  1)]
D

 i 1 
P
I
D
 Transfer function
 t 1 D 1 
u ( z ) K c 1  1
 [1  z ] ( z )
  I (1  z ) t 
u ( z )  g ( z  1 ) ( z )
Process Control Lecture
Notes 85
 “Classical” Digital Control
 Digital PID Controller
 “Velocity” form
 t D 
u (k ) K c  (k )   (k )  [ (k )   (k  1)]
 I t 
u (k ) u (k  1)  u (k )

 Sometimes more convenient form for

implementation
Process Control Lecture
Notes 86
 “Classical” Digital Control
 Digital PID Controller Design
 Design for continuous system, adding
extra t/2 delay
 Convert to digital form

 Implement either in direct “position”

form or velocity form

See Chapters 24-26 of O & R for
details
Process Control Lecture
Notes 87
Intro to Model-based and
Multivariable Control

Intro to MV Control

Process Control Lecture

Notes 88
 PART 5

application case study

89
CASE STUDY

CaseStudy-Granula
tion

Process Control Lecture

Notes 90
 SUMMARY AND
CONCLUSIONS
 Necessity of Process Control
 Processes are dynamic: (Variables are always
changing with time)
 To meet fundamental objectives of Process Operation

must be operated safely;

specified production rates must be maintained;

desired product quality specs must be met.
Requires process control system

to monitor, and induce change in key variables
related to
safety,
production rate
and product quality.
Process Control Lecture
Notes 91
 Summary
 Designing effective control systems
 Requires knowledge of intrinsic process
characteristics (and the associated control problems
they pose)
 Along with design principles
 Affiliated Topics
 Process Dynamics

Provides the platform for understanding intrinsic process
characteristics;
 Process Modeling

Provides the rational basis for carrying out the tasks
involved in “Process Dynamics” and “Process Control”
Process Control Lecture
Notes 92
 Summary
 Control System Implementation platforms
 Manual
 Automatic

Analog

Digital
 Control Configurations
 Feedback
 Feedforward
 Cascade
 Ratio …
 Controller Design
 Choosing controller configuration, controller type and
controller parameters
 Choosing controller parameters:
Process ControlTuning
Lecture
Notes 93
 Summary
 Feedback controller tuning

For Continuous PID Controllers

Obtain process characterization (model)

Consult tuning rules

To improve performance when disturbance
effects are significant

Consider Cascade Control when disturbances affect
u,

Consider Feedforward Control when disturbances
affect y, and reasonably reliable models are
available, and the feedforward controller is
realizable

For systems with difficult dynamics

Classical controllers may be effective in some
cases;
Process Control Lecture

In general, use compensators
Notes 94
 Summary
 Feedback controller tuning
 For Digital PID Controllers

Design continuous PID controller

Obtain digital approximation, accounting for additional
delay due to sample-and-hold operations
 For multivariable systems

Try multiple single loop designs first, if interaction analysis
indicates effectiveness; otherwise

Employ multivariable approach
 Additional Topics
 Industrial Granulator Control system discussed as a
Case Study

Process Control Lecture

Notes 95
Conclusions
 Process Control for Chemicals and
Pharmaceuticals
 Many similarities so that what works in one
can be translated (appropriately) to the
other (especially for “small molecules”)
 Many issues to be resolved with “biologics”

Measurements (Process and Product Quality)

Inference from available process measurements

Process model development

Control theory suitable to such processes

Implementation details
Process Control Lecture
Notes 96
 Ideal Configuration for
Pharmaceuticals
Process Optimization PROCESS SPC
MONITORING/ 6-Sigma
IMPROVEMENT Quality By Design

PAT PCA/PLS
INFORMATION Filtering
PROCESS PROCESSING Process Model
Raw data
State estimation
Sensors/Analyzers

PROCESS & Advanced

PROD. QUALITYModel-based
rocess Variable Adjustments CONTROL Base Regulatory
Process Control Lecture
Notes 97
 Going on from here
 Depth on Selected topics
 Modeling and Identification
 Digital Control
 Multivariable Control
 Additional Topics not covered
 Statistical Process Control
 Process monitoring and diagnostics
 Other application case studies discussed
in detail
Process Control Lecture
Notes 98