School of Chemical Engineering

The University of New South Wales

CEIC3006
Process Dynamics and Control

Week 5
Lecturer: Prof. Jie Bao
 This week’s lectures

 Introduction to feedback control: closed-loop transfer functions

 Proportional, Integral and Derivative Control (PID control)
 Closed-loop stability analysis
 Routh array method
 Control design (if time permits)

CEIC3006 Session 2, 2022

 Feedback Control

CEIC3006 Session 2, 2022

 Feedback Control
 Steam heated stirred tank  Steam heated stirred tank
– process schematic plot – control diagram

Tin, F
F,Tin Controller Valve
TT TC
e u ws T
+ C A P
IP
TR -
ws Tank
M
Steam

Condensate

F,T

CEIC3006 Session 2, 2022

 Block Diagram
 Components
 Function block (transfer function)
 Summing point
 Branch point
Tin
Summing
Summing point
point
Gd
Controller Valve Tank
P Ws + T
+ e +
Gc Gv Gp
TR -
Tm T Branch
Gm point

Thermocouple

CEIC3006 Session 2, 2022

 Proportional Control

 For proportional control, the controller output is proportional to the

error signal

CEIC3006 Session 2, 2022

 Proportional Control

In the steam heated stirred tank example, the controller decides the pressure of the
compressed air p(t) to manipulate the valve opening. Deriving the transfer function for an
ideal proportional controller

CEIC3006 Session 2, 2022

 Feedback Control: Components
Controller:
 hardware component provides calibrated signal for the actuator
 software component implements mathematical calculations

Actuator:
 physical (with dynamics) process triggered by the controller
 directly affects process

Sensor:
 monitors some property of the system and transmits signals back to
the controller

CEIC3006 Session 2, 2022

 Block Diagram Components
 Current-to-Pressure (I/P) Transducer
• Transducers are usually designed to have linear characteristics and
negligible (fast) dynamics
• The transducer transfer function merely consists of a steady-state gain
KIP:

CEIC3006 Session 2, 2022

 Block Diagram Components

 Control valve
Control valves are usually designed so that the flow rate through the
valve is a nearly linear function of the signal to the valve actuator
resulting a first-order transfer function:

CEIC3006 Session 2, 2022

 Block Diagram Components

 Sensors/Transmitters
A sensor’ transfer function can be determined from its dynamic model,
e.g., a thermal couple

CEIC3006 Session 2, 2022

 Closed-loop system
Tin

GD
Controller Valve Tank
+ P Ws T
e Gv GP
- GC
TR
Tm GM
Thermocouple Analysis of dynamics of standalone
The overall closed-loop systems systems can be applied to study:

Transient responses of overall closed-

TR T loop system!
Gcl(s)

Tin T
Gyd(s)
CEIC3006 Session 2, 2022
 Closed-Loop Transfer Function
D

For control, we need to identify closed-loop dynamics: GD

- Setpoint changes Servo R +
Controller
U
Valve
V
Tank
Y
e
GC Gv GP
- Disturbance rejection Regulatory -
Ym
GM

1. Closed-Loop Servo Response: Eliminate all variables except R(s) and Y(s) Thermocouple

 Transfer function relating Y(s) and R(s) when D(s)=0

 Isolate Y(s)
CEIC3006 Session 2, 2022
 Closed-Loop Transfer Function
D

2. Closed-loop Regulatory Response GD

Tank
 Transfer Function relating D(s) to Y(s) at R(s)=0 R +
Controller
e U
Valve
V Y
GC Gv GP
-
Ym
GM

Thermocouple

 Isolating Y(s)

CEIC3006 Session 2, 2022

 Closed-loop Transfer Function
Regulatory Response with Disturbance Dynamics

Servo Response:

Overall Closed-Loop Transfer Function

Servo

Regulatory
CEIC3006 Session 2, 2022
 Feedback Control
 Steam heated tank
F,Tin TT TC

IP

Ws
Steam

Condensate

F,T
Feedback control system: Valve is manipulated to increase flow of steam to control tank
temperature

Closed-loop process: Controller and process are interconnected.

CEIC3006 Session 2, 2022

 Proportional Feedback
Example:
Given first order process:

for P-only feedback closed-loop dynamics:

Closed-Loop Larger Kc, smaller steady-state error
Steady-state gain

Closed-Loop
Time Constant
Larger Kc, quicker closed-loop response
CEIC3006 Session 2, 2022 18
 Proportional Feedback
Final response:

Larger Kc, smaller steady-state error

Note:
 for “zero offset response” we require
Tracking Disturbance rejection

 Possible to eliminate offset with P-only feedback (requires infinite controller gain)
 Need different control action to eliminate offset (integral)
 Proportional control is often insufficient!
Proportional, Integral and Derivative Control (PID control)
CEIC3006 Session 2, 2022
19
 Integral Control
Integrator is included to eliminate offset
 provides reset action: by summing the historical errors
 usually added to a proportional controller to produce a PI controller
• PI is the most widely used controller in industry
• optimal structure for first order processes

PI controller form

Infinite steady state gain!

Transfer function model

CEIC3006 Session 2, 2022 20

 PI Feedback
Example
PI control of a first order process

Closed-loop response

Note:
• Offset is removed, closed-loop is second order

CEIC3006 Session 2, 2022 21

 PI Feedback
Example (contd.)
Effect of integral time constant and controller gain on closed-loop dynamics

• natural period of oscillation

• damping coefficient

 integral time constant and controller gain can induce oscillation and change the
period of oscillation
CEIC3006 Session 2, 2022 22
 PI Feedback
 Effect of controller gain
Increasing controller gain:
 Increases speed of response
 Reduces control error quicker
 May cause oscillatory response or
causes instability

 Effect of integral time constant

Reducing integral time constant:
 Removes effect of disturbances (offset)
 Makes behavior more oscillatory, or
causes instability

CEIC3006 Session 2, 2022 23

 Derivative Action
 Derivative of error signal
 Used to compensate for trends in output
• uses speed of error signal change
• provides predictive or anticipatory action
 P and I modes only response to past and current errors
 Derivative mode has the form

• if error is increasing, decrease control action

• if error is decreasing, decrease control action

CEIC3006 Session 2, 2022 24

 PID Controllers
The acronym PID stands for:
• P - Proportional
• I - Integral: dealing sustained errors
• D - Derivative: predicting the trend of errors

PID Controllers:
• greater than 90% of all control implementations
• dates back to the 1930s
• very well studied and understood
• optimal structure for first and second order processes (given some assumptions)
• always first choice when designing a control system

CEIC3006 Session 2, 2022

 PID Control
PID Control Equation
Derivative
Proportional Action
Action

Integral action

PID Controller Parameters

Kc Proportional gain
τI Integral Time Constant
τD Derivative Time Constant
uR Controller bias
CEIC3006 Session 2, 2022
 PID Control

PID Controller Transfer Function

Or alternatively,

Note:
 denominator provides integration to remove possibility
of steady-state errors

CEIC3006 Session 2, 2022

 PID Feedback
Transfer Function

Closed-loop Transfer Function

 τ τ s2 + τ s + 1
G p ( s)Gv ( s) Kc  D I I 
 τIs 
Y ( s) = R( s) +
2
 τ τ s + τ s + 1
1 + G p ( s)Gv ( s) Kc  D I I  Gm ( s)
 τIs 
1
D( s)
2
 τ Dτ I s + τ I s + 1
1 + G p ( s)Gv ( s) Kc   Gm ( s)
 τIs 

Slightly more complicated than PI form

CEIC3006 Session 2, 2022 28
 PID Feedback
Example:
PID Control of a first order process
Kp
G p ( s) = , Gv ( s) = 1, Gm ( s) = 1
τs + 1
Closed-loop transfer function

τ Dτ I s2 + τ I s + 1
Y ( s) = R( s) +
 τ Iτ  2  1 + Kc K p 
 + τ Dτ I  s +   τ I s + 1
K K
 c p   K K
c p 
 τ Iτ  2  τ I 
  s +   s
K K
 c p K K
 c p
D( s)
 τ Iτ  2  1 + Kc K p 
 + τ Dτ I  s +   τ I s + 1
 Kc K p   Kc K p 

CEIC3006 Session 2, 2022 29

 PID Feedback
Effect of derivative action
 Small value of τ D :
Increase τ D  reduction of
maximum deviation, response
time and degree of oscillation
 Large value of τ D :
Noise amplified
Response more oscillatory

τD should be small compared to

integral action

CEIC3006 Session 2, 2022 30

 Direct/Reverse Acting

CEIC3006 Session 2, 2022 31

 Direct/Reverse Acting
Air-to-open valve (K>0):
Controller output should decrease when ym increases
⇒ Reverse-acting Control (Kc>0)
A first order process with a proportional controller:

Air-to-close valve (K<0):

Controller output should increase when ym increases
⇒ Direct-acting Control (Kc<0)

CEIC3006 Session 2, 2022

 Block Diagram
Tin
 Open-loop transfer function
Ratio of feedback signal to the error signal GL
Controller Valve Tank
Fs T
+ e Pv
Gc Gv Gp
TR -
Tm Gm

 Feedforward transfer function Thermocouple

The ratio of the output to the error signal

 Close-loop transfer function

CEIC3006 Session 2, 2022

 Typical Block Diagrams
 Parallel Connection

In Matlab:
G = parallel (G1,G2)
 Series Connection

In Matlab:
G = series (G1,G2)
 Feedback Connection

In Matlab:
Gcl = feedback (G,H)
CEIC3006 Session 2, 2022
 Feedback systems
 General Expressions
 Negative Feedback Control System: the overall controlled system, the
closed-loop system that includes the controller, actuator, process and sensor:

Where GCL(s) is the closed-loop system, Gf(s) is the feedforward system and GOL(s) is the
open-loop system

 Positive Feedback Control System

CEIC3006 Session 2, 2022

 Closed-loop System

CEIC3006 Session 2, 2022

 Closed-loop System

CEIC3006 Session 2, 2022

 Closed-loop System

CEIC3006 Session 2, 2022

 Closed-loop Stability
 An important objective of process control is to ensure the stability of the
controlled process (closed-loop system), e.g., to stabilize an exothermic reactor.
 A linear system is said to be stable if the output response is bounded for all
bounded inputs. Otherwise, it is unstable.
Step Response
1.4
Amplitude

1.2

1
0.8
0.6
0.4

0.2
0 0 2 4 6 8 10 12

Time (sec.)
25
Step Response
20
15
Amplitude

10
5
0
-5
-10
-15
-20
0 10 20 30 40 50 60

Time (sec.)
12000
Step Response
10000
Amplitude

8000

6000

4000

2000

-2000
0 1 2 3 4 5 6 7 8 9 10

Time (sec.)

CEIC3006 Session 2, 2022

 Closed-loop Stability
 The dynamics of the overall controlled system can be very
different to that of the process. Due to the interactions among
the controller, actuator, process and sensors, the overall
closed-loop system can be unstable if the controller is not
properly designed.
 The closed-loop system can be unstable even if the controller,
actuator, process and the sensor are all stable!
 We need to analyse the closed-loop stability!

CEIC3006 Session 2, 2022

 Closed-loop Stability
Tin

GD Stability of a standalone system

Controller Valve Tank  If Re(pi)<0, then there are terms of the form y(t)
+ Pv Fs T
e Gv GP vanishes to a unique point – the system is stable.
- GC
TR  If any Re(pi)>0, then there is at least one term of the
Tm GM form i.e., y(t) will be unbounded – the system is
unstable.
Thermocouple
The closed-loop system

TR T
Gcl(s)
Stability analysis of dynamics of standalone
systems can be applied to determine the
Tin T stability of the closed-loop system!
Gyd(s)
CEIC3006 Session 2, 2022
 Closed-loop Stability
D(s)
 Closed-loop dynamics Gd
controller actuator process
R(s)+ Y(s)
Gc Gv Gp + +
-

sensor
Gm

GOL

 Characteristic Equation:
1 + GOL(s) =0
The roots of the characteristic equation is the poles of the closed-loop system.

CEIC3006 Session 2, 2022

 Closed-loop stability
 General Stability criterion:
“A closed-loop feedback control system is stable if and only if all roots
of the characteristic polynomial are negative or have negative real parts.
Otherwise, the system is unstable.”

 Unstable region is the right half plane (RHP) of the complex plane.

 Valid for all linear systems.

 For nonlinear system, if its linearized system is stable, the stability holds only
locally. Moving away from the point of linearization may cause instability.
CEIC3006 Session 2, 2022
 Closed-loop Stability

Root locations

Closer to imaginary axis :

slower dynamics

Closer to real axis:

less oscillatory

CEIC3006 Session 2, 2022

 Closed-loop Stability

 Example

Roots:
Kc=2: -1.25, -0.22 + 0.44i, -0.22 - 0.44i
Kc=6: -1.48, -0.11 + 0.68i, -0.11- 0.68i
Kc=16: -1.79, 0.04 + 0.97i, 0.04 -0.97i

Step Response
3

2.5

2
Kc=16
1.5
Amplitude

0.5

Kc=6 Kc=2
0

-0.5

-1

-1.5
0 2 4 6 8 10 12 14 16 18 20
Time (sec.)
CEIC3006 Session 2, 2022
 Example

𝐾𝐾𝑐𝑐 𝐾𝐾𝑣𝑣 𝐾𝐾𝑝𝑝

1+ =0
𝜏𝜏𝑝𝑝 𝑠𝑠 + 1

1+𝐾𝐾𝑐𝑐 𝐾𝐾𝑣𝑣 𝐾𝐾𝑝𝑝

Pole = −
𝜏𝜏𝑝𝑝
CEIC3006 Session 2, 2022
 Example
 The closed-loop system will be stable if this root is negative.
 Because time constants are always positive (τp > 0), the feedback control system
will be stable if KcKvKp > −1.
 This means that as long as the controller has the correct control action (i.e.,
reverse- or direct-acting), then the system will be stable.
 For example, if Kp > 0 and Kv > 0 then the controller must be made reverse-acting
so that Kc > 0.
 By contrast, if Kp < 0, then a direct-acting controller (Kc < 0) is required.

CEIC3006 Session 2, 2022

 Closed-loop Stability
 Step-by-step Procedure
 Derive closed-loop transfer function
 Obtain the characteristic equation
 Determine whether the roots are in LHP

Problem reduces to finding roots of a polynomial equation

CEIC3006 Session 2, 2022

 Example
 Consider the feedback control system:

 Determine the range of Kc values that result in a stable closed-

loop system
 The characteristic equation of close loop transfer function is:

CEIC3006 Session 2, 2022

 Example

 The close loop characteristic polynomial:

 Poles of the polynomial:

 To have a stable system, both roots of this characteristic equation must have negative real parts.
 The roots will be negative if 40(Kc + 1) > 0.
 The square root will have a value less than 7.
 If 40(Kc + 1) > 0, then Kc + 1 > 0, and Kc > −1.
 The closed-loop system will be stable if Kc > −1.

CEIC3006 Session 2, 2022

 Example
 Consider a process, Gp = 0.2/(−s + 1), that is
open-loop unstable. If Gv = Gm = 1, determine
whether a proportional controller can stabilize
the closed-loop system.
 The characteristic equation for this system is:

 It has the single pole p = 1 + 0.2Kc.

 The stability requirement is that Kc < −5.
 Feedback control can be used to stabilize a process
that is not stable without control.

CEIC3006 Session 2, 2022

 Closed-loop Stability

 Analysis problems
 For known closed-loop systems
1. Known process model
2. Known controller
MATLAB function ROOTS

 Design problems
 Unknown controller parameters
 Routh array

CEIC3006 Session 2, 2022

 Routh Array

 Determination of negative real roots

Polynomial characteristic equations

• A necessary condition for stability is that all of the coefficients in the

characteristic equation are positive
• If any of the coefficients is less than or equal to 0
Unstable poles
• If all coefficients are positive  possible stable and unstable poles!

CEIC3006 Session 2, 2022

 Routh Array
Routh array for a polynomial equation
is

1 an an − 2 an − 4 
2 an − 1 an − 3 an − 5 
3 b1 b2 b3 
4 c1 c2 
 
n +1 z1

a a −a a a a −a a
b1 = n −1 n − 2 n − 3 n , b2 = n −1 n − 4 n −5 n ,
an −1 an −1
where ba −b a ba −b a
c1 = 1 n − 3 2 n −1 , c2 = 1 n −5 3 n −1 ,
b1 b1
All elements of left column must be positive to have roots with negative real parts

CEIC3006 Session 2, 2022

 Examples
 Determine the stability of a system that has the
characteristic equation

 Because the s term is missing, its coefficient is zero.

 Thus, the system is unstable.
 A necessary condition for stability is that all of the
coefficients in the characteristic equation must be
positive

CEIC3006 Session 2, 2022

 Example
 Find the values of controller gain Kc that make the feedback control system with the
following characteristic equation stable.

 All coefficients are positive

1 an an − 2 an − 4 
2 an − 1 an − 3 an − 5 
3 b1 b2 b3 
4 c1 c2 
 
n +1 z1

17×8−10 1+𝐾𝐾𝑐𝑐 126−10Kc

𝑏𝑏1 = = > 0 → 𝐾𝐾c < 12.6
17 17
c1=1+Kc>0

CEIC3006 Session 2, 2022

 Dynamics of FOPTD Model and a PI Controller

 Unit-step disturbance responses for the candidate controllers (First order plus time delay
(FOPTD) Model: K = 1, θ = 4, τ = 20) and G = Gd
More aggressive

Unstable
More aggressive

Best
performance

 Controller settings can be adjusted to achieve the desired closed-loop system performance, a
procedure referred to as controller tuning.
CEIC3006 Session 2, 2022
 Controller Design
 Control design criteria
 Closed-loop system is stable.

 The effects of disturbances are minimized

 Rapid, smooth responses to setpoint changes

 Offset is eliminated

 Excessive control action should be avoided

 Robust, insensitive to changes in process conditions and to inaccuracies in the process
model.
D(s)
Gd
controller actuator process
R(s)+ + + Y(s)
- Gc Gv Gp

sensor
Gm
CEIC3006 Session 2, 2022
 Controller Design Methods

 PID controller settings can be determined by a number of

alternative techniques:

1) Direct Synthesis (DS) method

2) Internal Model Control (IMC) method
3) Controller tuning relations
4) Computer simulation
5) On-line tuning after the control system is installed

CEIC3006 Session 2, 2022

 Control Design

 Model based control design

A dynamic process model is used to

predict the behaviour of the process

determine appropriate control action

CEIC3006 Session 2, 2022

 Direct Synthesis Method
D

G
process d
R + controller actuator + C
Gc Gv Gp +
-
sensor
Gm

Isolate :

~ , controller is given by
For a desired trajectory (C/R)d and plant model G

CEIC3006 Session 2, 2022

 Direct Synthesis
 Remarks
 not necessarily in the PID form
 inverse of process model to yield pole-zero
cancellation (often inexact because of process
approximation)
 used with care with unstable process or
processes with RHP zeroes

 Perfect Control
 cannot be achieved, requires infinite gain
e.g.

CEIC3006 Session 2, 2022

 Direct Synthesis
 Closed-loop process with finite settling time
 1st order process

 Leads to a PI controller

CEIC3006 Session 2, 2022

 Direct Synthesis
 Closed-loop process with finite settling time
 For 2nd order G(s),

 We have a PID controller

CEIC3006 Session 2, 2022

 Systems with Dead time
Delayed transfer functions

Many chemical processes can be approximately modeled as

First order plus time-delay

Problem: the dead time makes analysis (poles and zeros) more difficult

CEIC3006 Session 2, 2022

 Dead-time Approximations
 Taylor expansion

 Linear term in Taylor expansion

 Approximate dead-time by a rational (polynomial) function

 Most common is Padé approximation:

CEIC3006 Session 2, 2022

 Direct Synthesis
 First order plus time delay (FOPTD)

 Desired closed-loop:

 Leads to a PI controller

CEIC3006 Session 2, 2022

 Direct Synthesis
 Second order plus time delay (SOPTD)

CEIC3006 Session 2, 2022

 Direct Synthesis

 The above are just examples.

 The Direct Synthesis method can be applied to any process transfer
functions (but may not lead to PI/PID controllers)

 If there is a RHP zero, keep that zero in (C/R)d.

 Make sure the steady state gain of (C/R)d to be 1 to ensure offset free
control!

CEIC3006 Session 2, 2022

 Example
 Design a controller using Direct Synthesis method

 Process parameters:

 Servo transfer function:

𝜏𝜏𝑐𝑐 = 12

CEIC3006 Session 2, 2022

 Example
Servo response to a step Regulatory response to a step
change with magnitude of 20 change with magnitude of 20
25 2.5

20 2

15 1.5
y(t)

y(t)
10 1

5 0.5

0 0
0 100 200 300 0 100 200 300
Time (seconds) Time (seconds)

CEIC3006 Session 2, 2022

 Example
 Use the DS design method to calculate PID controller settings for the process

 Consider three values of the desired closed-loop time constant: τc = 1, 3, and 10. Evaluate
the controllers for unit step changes in both the set point and the disturbance, assuming
that Gd = G. Repeat the evaluation for two cases:
a. The process model is perfect (𝐺𝐺� = 𝐺𝐺).
� = 0.9 instead of the actual value, K = 2.
b. The model gain is incorrect, 𝐾𝐾

CEIC3006 Session 2, 2022

 Example

τc = 1 τc = 3 τc = 10
� = 2)
𝐾𝐾𝑐𝑐 (𝐾𝐾 3.75 1.88 0.68
� = 0.9)
𝐾𝐾𝑐𝑐 (𝐾𝐾 8.33 4.17 1.51
τI 15 15 15
τD 3.33 3.33 3.33

CEIC3006 Session 2, 2022

 Example

τc = 1 τc = 3 τc = 10
� = 2)
𝐾𝐾𝑐𝑐 (𝐾𝐾 3.75 1.88 0.68
� = 0.9)
𝐾𝐾𝑐𝑐 (𝐾𝐾 8.33 4.17 1.51
τI 15 15 15
τD 3.33 3.33 3.33

Correct model gain Incorrect model gain

CEIC3006 Session 2, 2022

 Example

CEIC3006 Session 2, 2022

 Field Controller Tuning
 What is controller tuning?
To tune controller settings until the control system performance is considered to be
satisfactory
 Why?
 The process information may be incomplete or not accurate – not possible to design
controller based on the models

 The process information may not be accurate - field tuning may still be required to
fine tune the controller
 Advantages:
 Process models are not required
 Disadvantage:
 Tedious and time consuming
 Hard to get the best controller parameters

CEIC3006 Session 2, 2022

 Trial and Error Tuning
 Basic Idea
Trading-off between stability and performance

 Ultimate Gain Kcu

 Definition: The ultimate gain Kcu is the largest value of the controller gain Kc that results
in closed-loop stability when a proportional-only controller is used.
 How to estimate Kcu
• Theoretically
• Experimentally:
Increase Kc until the
control loop results in
continuous cycling –
a sustained oscillation
with constant amplitude.

CEIC3006 Session 2, 2022

 Trial and Error Tuning
 Procedure:
1. Eliminate integral and derivative action by setting τD at its minimum value and τI at
its maximum value.
2. Set Kc at a low value and put the controller on automatic.
3. Increase the controller gain Kc by small increments until continuous cycling occurs
after a small set-point or load change.
4. Reduce Kc by a factor of two.
5. Decrease τI in small increments until continuous cycling occurs again. Set τI equal to
three times this value.
6. Increase τD until continuous cycling occurs. Set τD equal to one-third of this value
 Issues:
 Not applicable to unstable process
 Some simple processes do not have ultimate gain (for example?)

CEIC3006 Session 2, 2022

 Continuous Cycling Method
Ziegler Nichols Continuous Cycling Method (1942)
 Procedure
Step 1: Experimentally determine the ultimate gain Kcu (trial-and-error)
Step 2: Determine the period of the resulting sustained oscillation – ultimate period Pu.
Step 3: Calculate PID controller settings from Kcu and Pu:

CEIC3006 Session 2, 2022

 Continuous Cycling Method
Controller Settings based on the Continuous Cycling Method of Ziegler and Nichols (1942)

Controller Kc τI τD
P 0.5Kcu — —
PI 0.45Kcu Pu/1.2 —
PID 0.6Kcu Pu/2 Pu/8

 Remarks
 Widely considered to be an "industry standard"
 Advantages:
 Only need one trial-and-error search
 Large safety margin for Kc
 Disadvantages:
 Slow dynamics results the trial-and-error determination of Kcu and Pu be time consuming.
 For many applications, continuous cycling is objectionable because the process is pushed to a stability limit.
CEIC3006 Session 2, 2022
 Continuous Cycling Method
 The Z-N controller settings are based on a 1/4 decay ratio.
 The settings tend to produce oscillatory responses and large overshoots for
set-point changes.

 Modification
Modified Ziegler-Nichols Settings for PID

Controller Kc τI τD
Original (1/4 decay ratio) 0.6Kcu Pu/2 Pu/8
Some overshoot 0.33Kcu Pu/2 Pu/3
No overshoot 0.2Kcu Pu/2 Pu/3

CEIC3006 Session 2, 2022

 Continuous Cycling Method

 Example

 By experiments: Kcu=0.95 and Pu=12

 Calculate PID settings:

CEIC3006 Session 2, 2022

 Continuous Cycling Method
 Example

CEIC3006 Session 2, 2022

 Cohen-Coon and Ziegler Nichols Tuning Relations

Controller Ziegler-Nichols Cohen-Coon

Proportional KK C = τ( θ) ( θ) 3
KK = τ + 1
C

Proportional
+
( θ)
KK C = 0.9 τ KK = 0.9(τ )+ 0.083
C θ
τ θ [3.33 + 0.33( τ )]
θ
= 3.33(θ )
Integral τ I
1.0 + 2.2(θ )
I
=
τ τ τ
τ
Proportional KK = 1.2(τ ) ( θ )+ 0.270
KK C = 1.35 τ
+ C θ
θ [32 + 6(θ τ )]
= 2.0(θ )
Integral τ τ =
13 + 8(θ )
I
+ τ I

Derivative
τ τ
τ
= 0.5(θ )
D τ 0.37(θ )
τ
τ τ D
τ 1.0 + 0.2(θ )
=
τ
Example:

PI: Kc=10.27 τI=18.54

PID: Kc=15.64 τI=19.75 τd=3.10

CEIC3006 Session 2, 2022

 PID Control Implementation Issues

 Reset Windup
 Sustained error causes the integral term to become very large and the
PID controller output eventually saturated.

 Anti-reset windup
• Temporary halting the integral action whenever the controller output
saturates.

CEIC3006 Session 2, 2022

 PID Control Implementation Issues

Eliminating Reset Windup - Implementation

Actuator Error Integral summation

upper limit + stopped
upper limit - active
normal + active
normal - active
lower limit + active
lower limit - stopped

CEIC3006 Session 2, 2022

 PID Control Implementation Issues
Realization of Derivative Action
 Ideal PID control:

 Physically unrealizable:
Needs future information

 Infinite gain for step change input

CEIC3006 Session 2, 2022

 PID Control Implementation Issues

 Commercial PID controllers:

where α is a very small number

 Proper systems
A transfer function whose order of numerator is no greater (or less) than its
order of denominator is a proper (or strictly proper) system.
Non-proper systems cannot be realized

CEIC3006 Session 2, 2022

 Derivative Kick
 Ideal PID controller D(s)
controller actuator process
R(s)+ + +Y(s)
- Gc Gv Gp
E(s)
sensor
Ym(s)
Gm

Sudden changes in set point, thus e(t), will cause very large derivative term.
 Eliminating the derivative kick
Take the derivative action on the measurement ym(t) rather than the error signal e(t).

CEIC3006 Session 2, 2022