Chapter 3: Advanced control strategies

1. Cascade control
2. Time-delay compensation
3. Inferential control
4. Selective and override control
5. Nonlinear control (gain scheduling,
Fuzzy logic control)
6. Multivariable Control
7. Adaptive control
 Cascade control
2  Is used to solve drawback of feedback
control
 Uses a secondary measurement and
feedback loop
 When disturbances are associated with
manipulated variable or final control
element is nonlinear

Conventional furnace control

  Furnace control with cascade control
3

 Primary(master) – TT and TC loop

 Secondary(slave) – PT and PC loop
 Specific features of cascade control
4

 Output of master is reference for slave

 The Slave control loop is nested inside
the master control loop
 Two controlled variables
 Two sensors
 One manipulated variable
 cascade control of stirred chemical reactor
5
 Cascade control of stirred reactor
6

 Primary
 Controlled variable – reactor temperature
 Manipulated variable – cooling water inlet
stream
 Disturbance variable – feed temperature and
feed composition
 Secondary
 Jacket temperature – controlled variable
 Manipulated variable- cooling water
 Disturbance variable – cooling water
temperature
 Boiler level control
7
 boiler level control using cascade
8
 Cascade control of dryer
9
  parameters influencing the temperature of the outlet
air other than the moisture content of the drying
10
material are
 air flow,
 ambient air temperature,
 variations in steam temperature.

 If any of these parameters were to suddenly change,

the effect would be slow to register at the outlet
temperature even though there would be immediate
impact at the bottom of the dryer where the heated
air enters.
 Correspondingly, the control system would be slow to
correct for any of these changing loads.
 Cascade control of dryer
11
 Block diagram of general cascade
control system
12
Design considerations
 For the block diagram above, assuming
Ysp1=D1=0

 Which gives

13
 Design considerations
14
contd…
 Similarly

 For disturbances in D1
 Design considerations
15
contd…
 Characteristics equation of cascade loop

 If cascade is not there, characteristic

equation is same as normal feedback
 Allows to use larger values of Kc1
 Makes system less sensitive to errors
 Makes the system respond fastly
 Design considerations
16

 Primary control
 PI or PID
 Used to compensate small offsets of
secondary
 Secondary control
 P or PI is mostly used, not PD
 Small offsets are tolerated
 Design steps
17

 First design a P or PI controller for the

inner loop using frequency response
method or time domain analysis ( first
start with P and check if PI is needed
 Using the designed controller, simplify
the block diagram
 Design a PI or PID controller for outer
loop
 Example
18
 For the following cascade system,
determine the stability limits, time
constant and steady state error for the
cascade and normal feedback control
systems. Assume Gc2=Kc2=4
 Solution

19
 Inner loop of cascade system

 Time constant for cascade control is 0.2

 For conventional system, time constant is 1
 Characteristic equation comparison with
proportional controller
 Cascade control


Kc1u=43.3
  Conventional feedback

20

Kc1u=11.25
 Offset error for cascade control

 Offset error for conventional feedback

 Inferential Control
21

• Problem: Controlled variable cannot be

measured or has large sampling period.
• Possible solutions:
1. Control a related variable (e.g.,
temperature instead of composition).
2. Inferential control: Control is based on an
estimate of the controlled variable.
• The estimate is based on available
measurements.
– Examples: empirical relation, Kalman filter
• Modern term: soft sensor
 Inferential Control with Fast and
Slow Measured Variables
22
 Selective Control Systems & Override

For every controlled variable, it is very

desirable that there be at least one
manipulated variable.
Chapter 16

• But for some applications,

NC > N M
where:
NC = number of controlled variables

NM = number of manipulated
variables
• Solution: Use a selective control system or
an override to protect equipment from unsafe
23 operating conditions.
 24

• Low selector:
Chapter 16

• High selector:
 Example: High Selector Control System

25

Fig: control of reactor hot spot temperature by using a high selector

• multiple measurements
• one controller
• one final control element
  Constraint control

26

Fig: A selective control system to handle sand/

water slurry.
2 measurements, 2 controllers,
1 final control element
 Split range control
 Use multiple controllers and multiple control elements
27

2
measurment
2 controllers
2 manipulated variables
28

 Block diagram for sand/water slurry.

 Gain scheduling
29
 Controller parameters vary continuously with the
scheduling variable.
 Scheduling variables are divided into regions
where the process characteristics is quite
different. Different controller setting is assigned
to each region.

 Uses different gain over different operating range

 Is useful for system where process dynamics
depends on known system variable
 Step responses can be used to categorize
 Multi variable control
30

 In practical control problems there

typically are a number of process variables
which must be controlled and a number
which can be manipulated (MIMO).
 Examples
 Inline blending
 Distillation column
 Gas liquid separator
 Boiler
 Note the "process interactions" between
controlled and
manipulated variables.
 In line blending
31

 Two controlled variables, input stream

flow rates affect the outputs w and x
 Distillation column
32

 Two manipulated variables-reflex flow

rate R and stream flow S
 Two controlled variables XD or XB
 Gas liquid separator
33

 Two manipulated variables – G and L

 Two controlled variables-P and h
 Multi loop control and process
interaction
34

 Multi loop control

 When multiple SISO loops are used to
control multi input multi output systems
 Need for study of process interactions
 One manipulated variable may affect many
controlled variables
 Block diagram analysis
35

 Consider a 2 input 2 output system

 Four possible transfer functions

 Block diagram analysis
36
contd…
 From superposition principle

 In matrix form

Where
 Multiloop Control Strategy
 Typical industrial approach

37 • Consists of using n standard FB controllers (e.g., PID), one for

each controlled variable.
 Control system design
1. Select controlled and manipulated variables.
2. Select pairing of controlled and manipulated variables.
3. Specify types of FB controllers.
Example: 2X2 System

Two possible controller pairings:

U1 with Y1, U2 with Y2 (1-1/2-2 pairing)
or
U1 with Y2, U2 with Y1 (1-2/2-1 pairing)
 1-1/2-2 method
38

 U1 has direct effect on Y1 and indirect

effect on Y2 through Gp21 and vice
versa
 1-2/2-1 method
39

 U1 directly affects Y2 and indirectly

affects Y1 and vice versa
 Stability of multi loop case
40

 The characteristic equation of the close

loop systems is

 Stability depends on
 4 TF’s
 Two controllers
 Exercise
41

 For the following system controlled by p

controller

 Simulate the system using Simulink in 1-

1/2-2 and 1-2/2-1 configuration. Which one
is better?
 Determine the value of gain k which gives
stable operation for the 1-1/2-2 connection
 Control-loop Interactions

42

 Process interactions may induce undesirable

interactions between two or more control loops.
Example: 2 x 2 system
 Control loop interactions are due to the
presence of a third feedback loop.
 Problems arising from control loop interactions
i. Closed-loop system may become destabilized.
ii. Controller tuning becomes more difficult.
43

 Figure: the hidden feedback control loop

in dark line for 1-1/2-2 control pairing
 Methods used to select multi loop
configurations
44

Relative Gain Array

Provides two types of useful information:
1. Measure of process interactions
2. Recommendation about best pairing of
controlled and manipulated variables.
Requires knowledge of steady-state gains but
not process dynamics.
 Methods used to select multi loop
configurations
45

 Bristol’s method of relative gain array

 =
 =1, closing or opening has no effect
 =0, no direct effect
 >1, loops interact proportionally
 0<<1, interaction is largest when =0.5
 <0, opposite effect
 Example of RGA Analysis: 2 x 2 system

Steady-state process model,

46

y1 K11u1  K12u2
y2 K 21u1  K 22u2

The RGA, L, is defined as:  11 12 

  

 21 22 

where the relative gain, ij, relates the ith

controlled variable and the jth manipulated
variable
 Scaling Properties:
i. ij is dimensionless

ii.    
i
ij
j
ij 1.0
Chapter 18

For a 2 x 2 system,
1
11  , 12 1  11 21
K K
1  12 21
K11K 22

Recommended Controller Pairing

It corresponds to the ij which have the largest

positive values that are closest to one.
47
 In general:
1. Pairings which correspond to negative pairings should
not be selected.
2. Otherwise, choose the pairing which has ij closest
to one.
Examples:
Process Gain Relative Gain
Chapter 18

Matrix, K : Array, L:

 K11 0   1 0
 0
 K 22 
  0 1
 
 0 K12 
 0 1
K
 21 0    1 0
 
 K11
 0
K12 
K 22 
  1 0
 0 1

 
 K11 0 
K
 21 K 22    1 0
 0 1
  48
 For 2 x 2 systems:
1
y1  K11u1  K12u2 11 
K12 K 21
, 12 1  11 21
1
y2  K 21u1  K 22u2 K11 K 22

Example 1:
Chapter 18

 K11 K12   2 1.5

K  
 K 21 K 22   1.5 2 

 2.29  1.29  Recommended pairing is Y1
 Λ   and U1, Y2 and U2.
  1 . 29 2 .29 

Example 2:

  2 1.5  0.64 0.36 

K    Λ  
 1.5 2   0 .36 0 .64 

Recommended pairing is Y1 with U1 and Y2 with U2.
49
Chapter 18 EXAMPLE: Thermal Mixing System

The RGA can be expressed in two equivalent forms:

Wh Wc Wh Wc
 T  Tc Th  T   Wc Wh 
W W
Th  Tc Th  Tc  Wc  Wh Wc  Wh 
K   and Λ  
 Th  T T  Tc   Wh Wc 
T  T 
 Th  Tc Th  Tc   Wc  Wh Wc  Wh 

Note that each relative gain is between 0 and 1. The recommended

controller pairing depends on nominal values of T, Th, and Tc.
50
 Example
51

A task is given to you to design an automated shower

in a given room. The design calls for the system that
can deliver 3 litre/m of water at 110F by mixing hot
water at 170F and cold water at 80F.flow and
temperature sensors are available with control valves
for adjusting hot and cold water flow rates.
 Calculate the required hot and cold water flow rates

assuming density and heat capacity of water are

constant.
 Calculate the relative gain array and recommend the

best pairing of controlled and manipulated variables.