You are on page 1of 21

CH3050: Process Dynamics and Control

Quick Overview of Control Problems and Principles

Arun K. Tangirala
Department of Chemical Engineering
IIT Madras

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 1


Expected Learning Outcomes

Expected Outcomes
The learner will be able to
1 Define the two fundamental problems in control.
2 Explain the two natural control principles (strategies).
3 List the paraphernalia for carrying out control.
4 Explain the hierarchy of process operations.

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 2


Two control problems
Maintain variable(s) at desired value (or within Drive variable(s) from its present (steady)
a range) despite disturbances. value (range) to a new target value (range).

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 3


Two control problems
Maintain variable(s) at desired value (or within Drive variable(s) from its present (steady)
a range) despite disturbances. value (range) to a new target value (range).

▶ Goal is to both drive the variable to its final value and shape its trajectory!
▶ All control problems are a mix of the above two
Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 4
Four virtual experiments

Fi 1 Study the process dynamics


(gain, rate of evolution, delay)
2 Implement “open-loop” control and
h study its merits / demerits
Fo
3 Introduce feedback, implement a new
controller and study its performance
Objective: Control the liquid level by
manipulating the inlet flow rate (valve
4 Insert uncertainty in the sensor and
position) study its impact on feedback control

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 5


Two natural control strategies
1 Feedback control: Sense the response now and take action instantly or later. Has
a compensatory outlook since inputs are manipulated after the effects are felt.
Disturbance
Manipulated
variable

+
Set-point +
Controller Actuator +
- Process

Controlled
Measured variable
variable
Sensor

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 6


Two natural control strategies
1 Feedback control: Sense the response now and take action instantly or later. Has
a compensatory outlook since inputs are manipulated after the effects are felt.
Disturbance
Manipulated
variable

+
Set-point +
Controller Actuator +
- Process

Controlled
Measured variable
variable
Sensor

▶ Intuitive and very natural.


▶ Inevitable strategy in real situations (why?)
▶ Deviations from set-point have to be tolerated for some finite duration in the least.
Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 7
Two natural control principles . . . contd.
2 Feedforward control: Predict the response at a later time (to future
disturbances) and take action now. Has an anticipatory outlook since inputs are
manipulated before the effects occur.
Sensor Disturbance

+ Process
Feedforward +
Set-point Actuator Process variable
Controller

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 8


Two natural control principles . . . contd.
2 Feedforward control: Predict the response at a later time (to future
disturbances) and take action now. Has an anticipatory outlook since inputs are
manipulated before the effects occur.
Sensor Disturbance

+ Process
Feedforward +
Set-point Actuator Process variable
Controller

▶ Predictive and requires good foresight.


▶ Instrumentation and accurate knowledge of future events and effects are essential.
▶ Deviations from set-point can be kept to zero.
Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 9
Control: Inversion Problem

Regardless of the strategy adopted, every control problem is an inversion problem


since it is about determining or back-calculating the input (profile) for a given
output / set-point (profile).

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 10


Control: Inversion Problem

Regardless of the strategy adopted, every control problem is an inversion problem


since it is about determining or back-calculating the input (profile) for a given
output / set-point (profile).

▶ The steady-state value of the input for a desired output change can be computed
relatively easily (recall the notion of gain!)

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 11


Control: Inversion Problem

Regardless of the strategy adopted, every control problem is an inversion problem


since it is about determining or back-calculating the input (profile) for a given
output / set-point (profile).

▶ The steady-state value of the input for a desired output change can be computed
relatively easily (recall the notion of gain!)
▶ The challenge is in the computation of input trajectory, i.e., the transient phase.

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 12


Control: Inversion Problem

Regardless of the strategy adopted, every control problem is an inversion problem


since it is about determining or back-calculating the input (profile) for a given
output / set-point (profile).

▶ The steady-state value of the input for a desired output change can be computed
relatively easily (recall the notion of gain!)
▶ The challenge is in the computation of input trajectory, i.e., the transient phase.
▶ Requires the knowledge of process dynamics and inversion of the same

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 13


Control: Inversion Problem

Regardless of the strategy adopted, every control problem is an inversion problem


since it is about determining or back-calculating the input (profile) for a given
output / set-point (profile).

▶ The steady-state value of the input for a desired output change can be computed
relatively easily (recall the notion of gain!)
▶ The challenge is in the computation of input trajectory, i.e., the transient phase.
▶ Requires the knowledge of process dynamics and inversion of the same
▶ How do we invert an ordinary differential equation (ODE)?

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 14


The paraphernalia

Essentials:

1 Mathematical model of process dynamics


2 Controller and a method of design
3 Instrumentation (sensors, comparators, actuators, etc.)

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 15


The paraphernalia

Essentials:

1 Mathematical model of process dynamics


2 Controller and a method of design
3 Instrumentation (sensors, comparators, actuators, etc.)

In addition:

▶ Control objectives (including performance specifications)


▶ Control strategy (including configuration and control law)

Optionally, constraints and robustness (to uncertainties) specifications may be provided.


Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 16
Mathematical concepts for control

Classical control requires foundations in:

1 Differential calculus (or of difference equations)


2 Complex numbers
3 Transforms (Laplace, Fourier)
4 Linear (vector / matrix) algebra
5 Optimization

In addition, optimal control and modern problems of control-loop performance as-


sessment (how good are the control loops?) also involve concepts from statistics.

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 17


Mathematical concepts for control

Classical control requires foundations in:

1 Differential calculus (or of difference equations)


2 Complex numbers
3 Transforms (Laplace, Fourier)
4 Linear (vector / matrix) algebra
5 Optimization

In addition, optimal control and modern problems of control-loop performance as-


sessment (how good are the control loops?) also involve concepts from statistics.

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 18


Road Map
Process Knowledge
Model Development Control Objectives
(First-principles / Empirical) Operating Constraints

Dynamical Analysis
(Stability, Responses, Process
Characteristics)

Controller Design
(Control Law, Tuning, Performance,
Robustness)

Performance Monitoring
(Benchmarking, Assessment,
Diagnosis)

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 19


Road Map
ODEs with algebraic laws Process Knowledge
State-space models
Linear time-invariant models Model Development Control Objectives
Transform-domain models (First-principles / Empirical) Operating Constraints
(Transfer functions)
Model conversions
Data-driven models

Stability concepts
Response analysis Dynamical Analysis
(Impulse, Step & Frequency (Stability, Responses, Process
response functions)
Characteristics)
Time-domain characteristics
Filtering characteristics
Model approximations

Control principles
Feedback control philosophy
Closed-loop stability
PID controllers
Controller Design
Design methods (Control Law, Tuning, Performance,
Model-based control Robustness)
(IMC and MPC)
Feedforward, cascade
control
Robustness analysis
Instrumentation

Performance Monitoring
Performance assessment
Minimum variance index (Benchmarking, Assessment,
Causes of poor performance Diagnosis)

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 20


Bibliography

Bequette, B. W. (1998). Process Dynamics: Modeling, Analysis and Simulation. Upper Saddle River,
NJ, USA: Prentice Hall.
Dorf, R. C. and R. H. Bishop (2010). Modern Control Systems. 12th edition. Upper Saddle River, NJ,
USA: Prentice Hall.
Ogunnaike, B. A. and W. H. Ray (1994). Process Dynamics, Modeling, and Control. Oxford University
Press.
Seborg, D. E., T. F. Edgar, D. A. Mellichamp, and F. Doyle (2011). Process Dynamics and Control. 3rd
edition. New York, USA: John Wiley & Sons, Inc.
Stephanopoulos, G. (1984). Chemical Process Control: An Introduction to Theory and Practice. Upper
Saddle River, NJ, USA: Prentice Hall.

Arun K. Tangirala (IIT Madras) CH3050: Process Dynamics and Control 21

You might also like