0% found this document useful (0 votes)
54 views220 pages

Mse 458 Process Dynamics and Control Students 2

The document discusses three textbooks related to process dynamics and control for chemical engineers. It then provides examples of dynamic responses for pure capacitive systems and first-order lag systems. For pure capacitive systems, the output grows linearly with time in an unbounded fashion, causing control problems. For first-order lag systems, the output approaches the final value exponentially and is self-regulating. The response depends on parameters like time constant and static gain. Systems can also interact in series, with non-interacting systems having independent transfer functions.

Uploaded by

mikemichaels267
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
54 views220 pages

Mse 458 Process Dynamics and Control Students 2

The document discusses three textbooks related to process dynamics and control for chemical engineers. It then provides examples of dynamic responses for pure capacitive systems and first-order lag systems. For pure capacitive systems, the output grows linearly with time in an unbounded fashion, causing control problems. For first-order lag systems, the output approaches the final value exponentially and is self-regulating. The response depends on parameters like time constant and static gain. Systems can also interact in series, with non-interacting systems having independent transfer functions.

Uploaded by

mikemichaels267
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 220

MSE 458 PROCESS DYNAMICS AND CONTROL

• Process Modelling, Simulation and Control for Chemical Engineers,


Luyben William L, McGraw Hill.

• Process System Analysis and Control, Coughanowr Donald, McGraw


Hill.

• Chemical Process Control, An Introduction to Theory and Practice,


Stephanopoulos George, Prentice Hall

1
Example: Pure Capacitive System
• Assume effluent flowrate F0 is determined by a constant –
displacement pump and not by hydrostatic pressure.

• Total mass balance around the tank yields


𝑑ℎ
• 𝐴 = 𝐹𝑖 − 𝐹𝑜 −−− −(13)
𝑑𝑡
• At steady-state
• 0 = 𝐹𝑖,𝑠 − 𝐹𝑜 −−− −(14)
• Subtracting equation (14) from (13) and writing as deviation variables
𝑑ℎ′
• 𝐴 = 𝐹𝑖′ −−− −(15)
𝑑𝑡

2
Example: Pure Capacitive System
• Yielding the following transfer function
ഥ′ (𝑠)
ℎ 1/𝐴
•𝐺 𝑠 = = −−− −(16)
𝐹ത𝑖′ (𝑠) 𝑠

3
Dynamic Response of a Pure Capacitive Process
• Transfer function for such a process is given by equation (5)

𝑦(𝑠) 𝐾𝑝′
•𝐺 𝑠 = ҧ = −−− −(5)
𝑓(𝑠) 𝑠

• Assuming f(t) undergoes a unit step change:


•𝑓 𝑡 =1 𝑓𝑜𝑟 𝑡 > 0
• For a unit step change
ҧ 1
•𝑓 𝑠 =
𝑠

4
Dynamic Response of a Pure Capacitive Process
• Therefore equation (5) becomes
𝐾𝑝′
• 𝑦ത 𝑠 = −−−− −(17)
𝑠2
• After inversion we have
• 𝑦 𝑡 = 𝐾𝑝′ 𝑡 −−−−−− −(18)
• We notice that the output grows linearly with time in an unbounded
fashion.
•𝑦 𝑡 ⟶∞ 𝑎𝑠 𝑡 ⟶ ∞

• Unbounded response of pure capacitive process also called Pure


Integrator

5
Dynamic Response of a Pure Capacitive Process
• Pure capacitive process can cause serious control problem
• System is said to be Non-self-regulating
• Processes with integrating action include: Tanks with liquids
• Vessels with gases
• Inventory systems for raw materials or products

6
Dynamic Response of First-Order Lag System
• Transfer function for such systems

𝑦(𝑠) 𝐾𝑝
•𝐺 𝑠 = ҧ = −−− −(3)
𝑓(𝑠) 𝜏𝑝 𝑠+1

• Assuming a unit step change in f(t), since 𝑓 ҧ 𝑠 = 1/𝑠, from (3) we


have
𝐾𝑝 𝐾𝑝 𝐾𝑝 𝜏𝑝
• 𝑦ത 𝑠 = = − −−−− −(19)
𝑠(𝜏𝑝 𝑠+1) 𝑠 𝜏𝑝 𝑠+1
• Inverting equation (19) we have
• 𝑦 𝑡 = 𝐾𝑝 1 − exp(−𝑡ൗ𝜏𝑝 𝑠) −−−− −(20)

7
Dynamic Response of First-Order Lag System
• For a step change of magnitude A the response would be
• 𝑦 𝑡 = 𝐴𝐾𝑝 1 − exp(−𝑡ൗ𝜏𝑝 𝑠) −−−− −(21)

8
Characteristics of the Response of First-Order
Systems
• (1) First-order lag process is self-regulating. It reaches a new steady-
state.

• (2) The slope of the response at t= 0 is equal to 1.


𝑑 𝑦(𝑡)Τ𝐴𝐾𝑝
• ฬ = 𝑒𝑥𝑝 −𝑡
ൗ𝜏𝑝 =1
𝑑 𝑡 Τ𝜏𝑝 𝑡=0
𝑡=0
• the response would reach its final value in one time constant.

9
Characteristics of the Response of First-Order
Systems
• Corollary
• The smaller the value of the time constant, the steeper the initial
response of the system.

• Also
• The time constant 𝜏𝑝 of a process is a measure of the time necessary
for the process to adjust to a change in its input.

10
Characteristics of the Response of First-Order
Systems
• (3) By four time constants, the response has essentially reached its
ultimate value
63.2% of final value by 1𝜏𝑝

86.5% 2𝜏𝑝

95.0% 3𝜏𝑝

98.0% 4𝜏𝑝

11
Characteristics of the Response of First-Order
Systems

12
Characteristics of the Response of First-Order
Systems
• The ultimate value of the response is equal to Kp.
• Hence the name steady-state or static gain given to the parameter Kp.

• ∆ 𝑜𝑢𝑡𝑝𝑢𝑡 = 𝐾𝑝 ∆ 𝑖𝑛𝑝𝑢𝑡 −−−−− −(22)

• For the same change in output we cause


• A small change in the input if Kp is very large (very sensitive system)
• A large change in the input if Kp is small

13
Example: Effect of Parameters on the Response
of a First-Order System
• Looking at the tank system
• It possesses 2 parameters
• (i) Cross sectional area of the tank, A
• (ii) Resistance to the flow of the liquid, R

• Looking at the problem another way


• (i) Time constant of process, τp Since 𝜏 = 𝐴𝑅 ; 𝜏𝛼𝐴
• (ii) Static gain, Kp 𝐾𝑝 = 𝑅

14
Example: Effect of Parameters on the Response
of a First-Order System
• Consider 2 tanks with different cross-sectional areas, A1 and A2,
• where A1 > A2 with same resistance, R.

• From equation (12)


• 𝜏 = 𝐴𝑅 ; 𝜏𝛼𝐴 −−−−−− −(12)
• ⟹ 𝜏𝑝1 > 𝜏𝑝2

• While the static gains remain the same.

15
Example: Effect of Parameters on the Response
of a First-Order System
• For same unit step change in the inlet flow rates
• Level of liquid in each tank responds according to Equation (21)
• 𝑦 𝑡 = 𝐴𝐾𝑝 1 − exp −𝑡
ൗ𝜏𝑝 𝑠 −−−− −(21)

16
Example: Effect of Parameters on the Response
of a First-Order System
• Assuming both tanks have different areas A1 and A2 and
• different flow resistances R1 and R2 such that

𝐴1 𝑅2
• = −−−− −(23)
𝐴2 𝑅1
• ⟹ 𝜏𝑝1 = 𝐴1 𝑅1 = 𝐴2 𝑅2 = 𝜏𝑝2

• Since A1 > A2 then from Equation (23)


• R2 > R1 which implies that Kp2 > Kp1.
17
Example: Effect of Parameters on the Response
of a First-Order System
• The responses are shown
• Both tanks have the same time constant,
• Therefore the same initial speed of response.
• However, as time goes on the tank with
• larger resistance R2, allows less liquid out of
the system.
• The larger the static gain of a process, the larger the steady-state
value of its output for the same input change.

18
First-Order Systems with Variable Time Constant
and Variable Gain
• Realistically, the Differential Equation describing processes has
variable time constants and gains.

• Example: Tank System with Variable Time Constant and Gain


• Assuming that the effluent flowrate F0 is not linear function, but
• 𝐹0 = 𝛽 ℎ 𝛽 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 −−−−−− −(24)
• Mass balance yields the following nonlinear equation:
𝑑ℎ
• 𝐴 + 𝛽 ℎ = 𝐹𝑖 −−−− −(25)
𝑑𝑡

19
First-Order Systems with Variable Time Constant
and Variable Gain
𝑑ℎ
• Linearizing Equation (25) 𝐴
+ 𝛽 ℎ = 𝐹𝑖 −−−− −(25)
𝑑𝑡
• around a steady state (S/S) yields
𝑑ℎ′ β
• 𝐴 + ℎ′ = 𝐹𝑖′ −−− −(26)
𝑑𝑡 2 ℎ𝑠
• Or
𝑑ℎ′
• 𝜏𝑝 + ℎ′ = 𝐾𝑝 𝐹𝑖′ −−−− −(27)
𝑑𝑡

• Where
2𝐴 ℎ𝑠 2 ℎ𝑠
• 𝜏𝑝 = 𝐾𝑝 =
𝛽 𝛽

20
First-Order Systems with Variable Time Constant
and Variable Gain
• System has a variable time constant and static gain.

21
Response of First-Order Systems in Series
• In industry, systems can interact in various ways

22
Response of First-Order Systems in Series
• For the Non-interacting system:
• (1) The outflow, q1, discharges into the atmosphere before spilling
into tank 2
• (2) The flow through R1 depends only on h1.
• (3) The variation in h2 in tank 2 does not affect the transient response
in tank 1.

23
Response of First-Order Systems in Series
• In the Interacting system, the flow through R1 depends on the
difference between h1 and h2.

24
Non-Interacting System
• Assume that
• (i) liquid has constant density, and
• (ii) tanks have uniform cross-sectional area and
• (iii) flow resistances are linear

• Transfer functions for tank 1 and tank 2 are


𝑄1 (𝑠) 𝐻2 (𝑠)
• 𝐺1 𝑠 = 𝐺2 𝑠 =
𝑄(𝑠) 𝑄1 (𝑠)

25
Non-Interacting System
• Taking a mass balance around each tank we have for
• Tank 1
𝑑ℎ1
• 𝑞 − 𝑞1 = 𝐴1 −−− −(1)
𝑑𝑡

• Tank 2
𝑑ℎ2
• 𝑞1 − 𝑞2 = 𝐴2 −−− −(2)
𝑑𝑡

26
Non-Interacting System
• The flow-head relationships for the two linear resistances is given as
ℎ1
• 𝑞1 = −−−− −(3)
𝑅1
ℎ2
• 𝑞2 = −−−−− −(4)
𝑅2

• Substituting (3) in (1) and finding the transfer function


𝑑ℎ1
• 𝑞 − 𝑞1 = 𝐴1 −−− −(1)
𝑑𝑡
𝑄1 (𝑠) 1
• = −−−− −(5)
𝑄(𝑠) 𝜏1 𝑠+1

27
Non-Interacting System
• Introducing deviation variables we have
• 𝑄1 𝑠 = 𝑞1 − 𝑞1,𝑠 𝑒𝑡𝑐 𝜏1 = 𝑅1 𝐴1

• Similarly combining (2) and (4) we have


𝑑ℎ2
• 𝑞1 − 𝑞2 = 𝐴2 −−− −(2)
𝑑𝑡
ℎ2
• 𝑞2 = −−−−− −(4)
𝑅2
𝐻2 (𝑠) 𝑅2
• = −−−− −(6)
𝑄1 (𝑠) 𝜏2 𝑠+1

28
Non-Interacting System
• And
• 𝐻2 = ℎ2 − ℎ2,𝑠 , 𝜏2 = 𝑅2 𝐴2

• Overall transfer function obtained as (5) X (6)


𝐻2 (𝑠) 1 𝑅2
• = −−−− −(7)
𝑄(𝑠) 𝜏1 𝑠+1 𝜏2 𝑠+1

29
Non-Interacting System
• Transient response of liquid-level system
1) Response is S-shaped
2) Response is sluggish for combined system
3) Sluggishness or delay sometimes called
• Transfer lag

30
Non-Interacting System
• For several non-interacting systems in series:
𝑥1 (𝑠) 𝑘1
• =
𝑥0 (𝑠) 𝜏1 𝑠+1
𝑥2 (𝑠) 𝑘2
• =
𝑥1 (𝑠) 𝜏2 𝑠+1
• ⋮
𝑥𝑛 (𝑠) 𝑘𝑛
• =
𝑥𝑛−1 (𝑠) 𝜏𝑛 𝑠+1

𝑥𝑛 (𝑠) 𝑛 𝑘𝑖
• = ς𝑖−1
𝑥0 (𝑠) 𝜏𝑖 𝑠+1

31
Non-Interacting System
• Step response of non-interacting first-order systems

32
Interacting Systems
• Mass Balance
𝑑ℎ1
• 𝑞 − 𝑞1 = 𝐴1 −−−− −(1)
𝑑𝑡
𝑑ℎ2
• 𝑞1 − 𝑞2 = 𝐴2 −−−− −(2)
𝑑𝑡

• Flow-Head relationship
• For tank 1
ℎ1 −ℎ2
• 𝑞1 = −−−− −(9)
𝑅1

33
Interacting Systems
• That for tank 2, R2 is as before
ℎ2
• 𝑞2 = −−−− − 4
𝑅2

• Expressing in terms of deviation variables


• Transforming the resulting equations
• Combining the transformed equations,

34
Interacting Systems
• In terms of deviation variables we have from (1) and (2)
𝑑𝐻1 𝑑ℎ1
• 𝑄 − 𝑄1 = 𝐴1 −−− −(10) 𝑞 − 𝑞1 = 𝐴1 −−−− −(1)
𝑑𝑡 𝑑𝑡
𝑑𝐻2 𝑑ℎ2
• 𝑄1 − 𝑄2 = 𝐴2 −−− −(11) 𝑞1 − 𝑞2 = 𝐴2 −−−− −(2)
𝑑𝑡 𝑑𝑡

• Writing (9) and (4) in terms of deviation variables


𝐻1 −𝐻2 ℎ1 −ℎ2
• 𝑄1 = −−−− −(12) 𝑞1 = −−−− −(9)
𝑅1 𝑅1
𝐻2 ℎ2
• 𝑄2 = −−−− − 13 𝑞2 = −−−− − 4
𝑅2 𝑅2
• Transforming (10) thru (13) gives

35
Interacting Systems
• 𝑄 𝑠 − 𝑄1 𝑠 = 𝐴1 𝑠𝐻1 𝑠 −−−− −(14)
• 𝑄1 𝑠 − 𝑄2 𝑠 = 𝐴2 𝑠𝐻2 𝑠 −−−− −(15)

• 𝑅1 𝑄1 𝑠 = 𝐻1 𝑠 − 𝐻2 𝑠 −−−− −(16)
• 𝑅2 𝑄2 𝑠 = 𝐻2 𝑠 −−−−−−−−− −(17)

• Four algebraic equations with five unknowns: (Q, Q1, Q2, H1 and H2)

36
Interacting Systems
• Combining to eliminate Q1, Q2, and H1 we have
𝐻2 (𝑠) 𝑅2
• = −−−− −(18)
𝑄(𝑠) 𝜏1 𝜏2 𝑠2 + 𝜏1 +𝜏2 +𝐴1 𝑅2 𝑠+1

• Comparing with the non-interacting system Equation (7)


𝐻2 (𝑠) 1 𝑅2
• = −−−− −(7)
𝑄(𝑠) 𝜏1 𝑠+1 𝜏2 𝑠+1

• Presence of A1R2 term in the coefficient of s for the interacting


system.

37
Interacting Systems
• Another expression for interacting is loading.
• The second tank is said to load the first tank.

• For effect of interaction on the transient response of a system


• Consider two-tank system with equal time constants

38
Interacting Systems
• For non-interacting system
𝑄2 (𝑠) 1 2
• = −−−− −(19)
𝑄(𝑠) 𝜏𝑠+1

• The unit step response for this transfer function comes out as
−𝑡Τ 1 −𝑡Τ𝜏
• 𝑄2 𝑡 = 1 − 𝑒 𝜏 − 𝑒 −−− −(20)
𝜏

• For interacting tanks, assuming 𝐴1 = 𝐴2


𝑄2 (𝑠) 1 𝑄2 (𝑠) 1
• = −− −(21) = −− −(22)
𝑄(𝑠) 𝜏2 𝑠 2 +3𝜏𝑠+1 𝑄(𝑠) 0.38𝜏𝑠+1 2.62𝜏𝑠+1

39
Interacting Systems
• Interaction has changed the effective time constant of the interacting
system.

• The response of Q2(t) to a unit-step change in Q(t) for the interacting


case Equation (22) is
• 𝑄2 𝑡 = 1 + 0.17 e −𝑡 Τ0.38𝜏 −1.17𝑒 −𝑡 Τ2.62𝜏 −−−− −(23)

• Non-interacting
−𝑡Τ 1 −𝑡Τ𝜏
• 𝑄2 𝑡 = 1 − 𝑒 𝜏 − 𝑒 −−− −(20)
𝜏

40
Interacting Systems
• Response for the interacting and non-interacting

41
Interacting Systems
• NB
• Interaction slows up the response

• Physically
1) For a unit step change, for non-interaction, flow q1 from tank 1 will
not be reduced by increase in h2 in tank 2
2) However, for the interacting case q1 will be reduced by the level in
tank 2
3) The effect of interaction on a system containing two first-order lags
is to change the ratio of effective time constants in the interacting
system
42
Interacting Systems
• Interacting system is more sluggish than the non-interacting system.

43
Dynamic Behavior of Second Order Systems
• Dynamic Behavior of Second Order Systems
• Second-order linear system
𝑑2 𝑦 𝑑𝑦
• 𝑎2 2 + 𝑎1 + 𝑎𝑜 𝑦 = 𝑏𝑓 𝑡 … … … … … ①
𝑑𝑡 𝑑𝑡
• If 𝑎𝑜 ≠ 0
we have
2
2𝑑 𝑦 𝑑𝑦
• 𝜏 + 2ζ𝜏 + 𝑦 = 𝐾𝑝 𝑓(𝑡) … … … … … ②
𝑑𝑡 2 𝑑𝑡

2 𝑎2 𝑎1 𝑏
•𝜏 = , 2ζ𝜏 = , 𝐾𝑝 =
𝑎𝑜 𝑎𝑜 𝑎𝑜

44
Dynamic Behavior of Second Order Systems
• τ = natural period of oscillation of the system
• ζ = damping factor
• Kp= Gain

• Transfer function
ȳ 𝑠 𝐾𝑝
•𝐺 𝑠 = = ……………③
𝑓(𝑠) 𝜏2 𝑠2 +2𝜁𝜏𝑠+1

45
Dynamic Behavior of Second - Order Systems
• Second order systems can arise from the following situations
1) Multi-Capacity Processes:
2) Inherently Second-Order Systems:
3) Processing systems with its controller:

• In the processing industry categories 1 and 3 dominate

46
Dynamic Response of a Second – Order System
• For a unit step change in the input f(t) equation ③
ȳ 𝑠 𝐾𝑝
•𝐺 𝑠 = = ……………③
𝑓(𝑠) 𝜏2 𝑠2 +2𝜁𝜏𝑠+1
• yields
𝐾𝑝
•ȳ 𝑠 = ……………… ④
𝑠(𝜏2 𝑠 2 +2𝜁𝜏𝑠+1)

• The poles of the second—order transfer function are given by the


roots of the characteristic polynomial

• 𝜏 2 𝑠 2 + 2𝜁𝜏𝑠 + 1 = 0
47
Dynamic Response of a Second – Order System
• The roots
−𝜁 𝜁 2 −1 −𝜁 𝜁 2 −1
• 𝑃1 = + 𝑎𝑛𝑑 𝑃2 = − ………...⑤
𝜏 𝜏 𝜏 𝜏

• Thus
𝐾𝑝 /𝜏2
•ȳ 𝑠 = ……………⑥
𝑠(𝑠−𝑃1 )(𝑠−𝑃2 )

48
Dynamic Response of a Second – Order System
• 𝜁2 − 1
• Case A
• When ζ>1 - 2 distinct roots and real poles

• Case B
• When ζ=1 - 2 equal roots (multiple pole)

• Case C
• When ζ<1 - 2 complex conjugate poles

49
Dynamic Response of a Second – Order System

50
Dynamic Response of a Second – Order System
• Case A -over-damped response when ζ>1
∆(𝑜𝑢𝑡𝑝𝑢𝑡 𝑠/𝑠)
• 𝑇ℎ𝑒 𝑔𝑎𝑖𝑛 𝐾𝑝 = ………………..⑦
∆(𝑖𝑛𝑝𝑢𝑡 𝑠/𝑠)

51
Dynamic Response of a Second – Order System
✓Curve resembles response of first order system (ζ=1 Critically
damped)
✓Response has initial delay
✓Response is sluggish
✓Response gets more sluggish as ζ increases

• Example: Responses of multi-capacity processes.

52
Dynamic Response of a Second – Order System
• Case C
• Underdamped responses when ζ<1

• Characteristics
✓The underdamped response is initially faster.
✓Underdamped is initially faster and reaches its ultimate value quickly.
Starts oscillating with progressively decreasing amplitude.
✓Underdamped responses in a chemical plant are caused by the
interaction of controllers with the process units they control.
✓Oscillating behavior becomes more pronounced with smaller values
of damping factor ζ.

53
Dynamic Response of a Second – Order System

54
Dynamic Response of a Second – Order System
✓Critically damped or over-damped responses seen as sluggish.

55
Characteristics of an underdamped Response

56
Characteristics of an underdamped Response
• (1) Overshoot: Ratio of A/B
• A: Maximum amount by which the response exceed its ultimate
value
• B: Ultimate value of response

−𝜋𝜁
• 𝐎𝐯𝐞𝐫𝐬𝐡𝐨𝐨𝐭 = 𝑒𝑥𝑝 ……………⑧
1−𝜁 2

57
Characteristics of an underdamped Response
• (2) Decay Ratio: is ratio C/A i.e. the ratio of the amounts above the
ultimate value of two successive peaks

−2𝜋𝜁
• Decay ratio = 𝑒𝑥𝑝 …………⑨
1−𝜁 2

• 𝐷𝑒𝑐𝑎𝑦 𝑟𝑎𝑡𝑖𝑜 = 𝑜𝑣𝑒𝑟𝑠ℎ𝑜𝑜𝑡 2

58
Characteristics of an underdamped Response
• Effect of damping factor on overshoot and decay ratio
• Overshoot
• Increases for decreasing ζ

• Decay ratio
• Larger ζ means greater damping
• Hence greater decay

59
Characteristics of an underdamped Response
• (3) Period Of Oscillation (T)[=] Time/Cycle: the time elapsed
between two successive peaks

• The radian frequency (rad/time) of oscillations of an underdamped


response is given by
1−𝜁 2
• 𝜔 = 𝑟𝑎𝑑𝑖𝑎𝑛 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 = ………………………….⑩
𝜏
• It is known that
1
• 𝜔 = 2𝜋𝑓 𝑎𝑛𝑑 𝑓=
𝑇
• Where f= cyclical frequency.

60
Characteristics of an underdamped Response

61
Characteristics of an underdamped Response
• Thus
2𝜋𝜏
•𝑇= ……………⑪
1−𝜁 2

• (4) Natural Period of Oscillation: A second order system with ζ=0 is a


system free of any damping.
• Its transfer function has two purely imaginary poles and that it will
oscillate continuously with a constant amplitude and a natural
frequency

62
Characteristics of an underdamped Response
• Transfer Function (ζ=0)
𝐾𝑝
•𝐺 𝑠 =
𝜏2 𝑠2 +1
𝐾𝑝 Τ𝜏2
•𝐺 𝑠 = 1 1 …………………….⑫
𝑠−𝑗 𝑠+𝑗
𝜏 𝜏

• Natural radian frequency


1
• 𝜔𝑛 = …………………⑬
𝜏
1
• −𝑵𝒂𝒕𝒖𝒓𝒂𝒍 𝒄𝒚𝒄𝒍𝒊𝒄𝒂𝒍 𝒇𝒓𝒆𝒒𝒖𝒆𝒏𝒄𝒚 𝑓𝑛 =
𝑇𝑛

63
Characteristics of an underdamped Response
• Natural period of oscillation Tn

• 𝑇𝑛 = 2𝜋𝜏 … … … … … … … … . ⑭

• (5) Response Time: The time needed for the response to reach
within +5% of its final value and stay there.
• Underdamped system will reach its ultimate value in an oscillatory
manner as t→∞.

64
Characteristics of an underdamped Response
• (6) Rise time: It is defined as the time required for the response to
reach its final value for the first time.

• It characterizes the speed with which an underdamped system


responds.
• The smaller the value of the ζ the shorter the rise time (i.e. the faster
the response of the system) but at the same time the larger the value
of the overshoot.

65
Characteristics of an underdamped Response
• Objective during the design of a controller will be proper selection of
ζ and τ values, so that

I. The overshoot is small


II. Rise time is short
III. Decay ratio is small
IV. Response time is short
V. Not possible to achieve all these objectives for same values of ζ and
τ. An acceptable compromise will be defined.

66
Inherently Second-Order Processes
• Associated with the motion of liquid masses or the mechanical
translation of solid parts possessing
I. Inertia to motion
II. Resistance to motion
III. Capacitance to store mechanical energy. e.g. flywheel

• Resistance and capacitance are characteristic of first order systems


• Inherently second-order systems are characterized by their inertia to
motion.

67
Inherently Second-Order Processes
• Newton’s law applied on a given system yields:
• (Balance of Forces on the system) = (Mass of System) × (Acceleration)-
----------(15)

• Since
d(Velocity)
• 𝑎𝑐𝑐𝑒𝑙𝑒𝑟𝑎𝑡𝑖𝑜𝑛 =
dt
d(spatial displacement)
• 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 =
dt

68
Inherently Second-Order Processes
• (Balance of Forces on the system) = (Mass of System) ×
d2 (spatial displacement)
2 ----------(16)
dt
• 2nd term gives rise to second-order behavior of the system.

• Example
• First-Order Processes with Second-Order Dynamics Due to the
Presence of a Control System.

69
Inherently Second-Order Processes
• Consider the tank as shown below
• First-order system transfer function given as
h(s) K𝑝
•𝐺 𝑠 = = −−− −(17)
Fi (s) 𝜏𝑝 𝑆+1

• Solution
• The dynamic mass – balance around the tank gives
Adh
• = F𝑖 − F𝑜 −−− −(18)
dt

70
Inherently Second-Order Processes
• At steady state we have
Adh
• 0 = F𝑖,𝑠 − F𝑜,𝑠 −−− −(19) = F𝑖 − F𝑜 −−− −(18)
dt

• Subtract (19) from (18) to have


Adh′
• = F𝑖′ − F𝑜′ −−− −(20)
dt

• When ℎ′ ≠ 0 , deviation or error used by the controller to increase or


decrease the effluent flow rate according to the relationship

71
Inherently Second-Order Processes
K𝑐 𝑡 ′
• F𝑜 = F𝑜,𝑠 + K 𝑐 h′ + ‫׬‬0
ℎ 𝑡 𝑑𝑡 −−− −(21)
τ𝐼
• Where K 𝑐 , τ𝐼 are constant parameters with positive values.

• According to (21)
1) When h′ = 0, then F𝑜 = F𝑜,𝑠 , the valve V stays where it is.
2) When h′ < 0 (i.e. level is down) from Eqn. (21) F𝑜 < F𝑜,𝑠
• (i.e. controller reduces the effluent rate and the level starts
increasing)

72
Inherently Second-Order Processes
3) When h′ > 0 (i.e. level is up) from eqn (21) F𝑜 > F𝑜,𝑠
• (i.e. controller increases the effluent rate and the level decreases)

• Replacing F𝑜′ in Eqn (20) with Eqn (21) we have


Adh′
• = F𝑖′ − F𝑜′ −−− −(20)
dt

𝑑h′ K𝑐 𝑡 ′
• 𝐴 + K𝑐 h′ + ‫׬‬0
ℎ 𝑡 𝑑𝑡 = F𝑖′ −−− −(22)
𝑑𝑡 τ𝐼

73
Inherently Second-Order Processes
• Laplace transform of Eqn (22) gives

ത ത 𝐾𝑐 1 ത ′
′ ′
• 𝐴𝑠ℎ 𝑠 + K 𝑐 ℎ s + ℎ 𝑠 = Fത 𝑖′ (𝑠)
𝜏𝐼 𝑠
• Or
Aτ𝐼 2 ത τ𝑖 𝑠 ′
• 𝑠 ′
+ 𝜏𝐼 𝑠 + 1 ℎ 𝑠 = 𝐹ത 𝑠 −−− −(23)
K𝑐 K𝑐

• From Equation (23) the transfer function is


output ഥ ′ (s)
ℎ 𝐾𝑝 𝑠
•𝐺 𝑠 = = ഥ′ (s) =
input F 𝑖 𝜏2 𝑠 2 +2𝜁τs+1

74
Inherently Second-Order Processes
• Where
Aτ𝐼 τ𝐼
• 𝜏2 = , 2𝜁τ = τ𝐼 , K𝑝 =
K𝑐 K𝑐

• From the above equations we find


Aτ𝐼 1 K𝑐 τ𝐼
•𝜏= , 𝜁=
K𝑐 2 A

• Depending on the values of the control parameters K 𝑐 and τ𝐼 we may


have the following cases

75
Inherently Second-Order Processes
K𝑐 τ𝑖
1) < 2. Then 𝜁 < 1 and the response ℎത ′ (s) to a step input in
A
Fത 𝑖′ (s) is that of underdamped system

K𝑐 τ𝑖
2) A
= 2. Then 𝜁 = 1 and system is critically damped.

K𝑐 τ𝑖
3) A
> 2. Then 𝜁 > 1 and the system is overdamped.

76
Inherently Second-Order Processes
• Resulting second-order behavior of the liquid level.

• First order dynamic behavior of a tank


changes to a second-order
when a proportional-integral controller
is added to the process.

77
Analyses and Design of Feedback Control System
• For a process as shown known as Open Loop.
• A disturbance, d (load or process load) changes
• Our control objective is to keep the value of
y at the desired level ‘‘𝑦𝑠𝑝 ‘’ (set point)

• Feedback control action takes the following steps


1) Measure the value of the output, 𝑦𝑚 indicated by the measuring
sensor

78
Analyses and Design of Feedback Control System
2) Compare indicated value 𝑦𝑚 to the desired value ‘‘𝑦𝑠𝑝 ‘’. Let the
deviation or error be 𝜖 = 𝑦𝑠𝑝 − 𝑦𝑚
3) Deviation, ‘’ 𝜖 ‘’ is supplied to the main controller.
The controller changes the value of the manipulated variable “m” so
as to reduce the magnitude of the deviation, 𝜖.

• Controller affects manipulated variable through final control element


(usually a control valve).

79
Analyses and Design of Feedback Control System
• These steps are summarized below ( Closed Loop.)

80
Analyses and Design of Feedback Control System
• Typical feedback control systems
1) Flow control

81
Analyses and Design of Feedback Control System
2) Pressure Control

82
Analyses and Design of Feedback Control System
3) Liquid Level Control
• Used for the control of liquid levels at the bottom of a distillation
column and its condenser accumulation tank.

83
Analyses and Design of Feedback Control System
4) Temperature control
• Control the temperature of the exiting hot stream at the desired
value, TSP.

84
Analyses and Design of Feedback Control System
5) Composition Control
• Composition is the controlled variable in the blending system.

85
Analyses and Design of Feedback Control System
• To simplify the presentation of the feedback control system replace
diagram with a simple circle with one of the following
characterizations
1) FC for flow control
2) PC for pressure control
3) LC for liquid level
4) TC for temperature
5) CC for concentration

86
Analyses and Design of Feedback Control System
• For transmission

1) LT- Level Transmission,


2) TT- Temperature Transmission,
3) PT- Pressure Transmission,
4) FT- Flow Transmission,
5) CT- Concentration Transmission.

87
Analyses and Design of Feedback Control System
• Simplified representation of flow feedback loop

88
Analyses and Design of Feedback Control System
• Simplified representation of liquid-level Feedback Loops

89
Analyses and Design of Feedback Control System
• Basic hardware components of a feedback control loop are the
following:
1) Process: Material equipment where physical or chemical
operations take place (tank, heat exchangers, reactors, separators,
etc.)

2) Measuring Instrument/Sensor: e.g.


• -Thermocouple(temperature)
• - Bellows, diaphragms (pressure or liquid level)
• - orifice plate (flow)
• - gas chromatograph, spectroscopic analyzers (composition)

90
Analyses and Design of Feedback Control System
3) Transmission lines: used to carry the measurement signal from
sensors to the controller and to final control element. These lines
can be either pneumatic (compressed air), liquid or electrical.
4) Controller: it also includes the function of the comparator. This unit
with logic decides how much to change the value of the
manipulated variable. It requires the specification of the set-point.
5) Final Control Element: This device receives the control signal from
the controller and implements it physically adjusting the value of
the manipulated variable. Usually a control valve
Each system with an input and an output is described with a differential
equation or a transfer function.

91
Types of Feedback Controllers
• Feedback controllers can be classified as
1) Proportional Controller
2) Proportional-Integral Controller
3) Proportional-Integral-Derivative Controller

• The controller produces an actuating signal 𝑐(𝑡) in such a way as to


return the output to the desired value,𝑦𝑠𝑝 .
• 𝐼𝑛𝑝𝑢𝑡: 𝑒𝑟𝑟𝑜𝑟 𝜖 𝑡 = 𝑦𝑠𝑝 − 𝑦𝑚 𝑡
• 𝑂𝑢𝑡𝑝𝑢𝑡: 𝑐 𝑡 (pneumatic or electrical)

92
Proportional Controller (P-Controller)
• Its actuating output is proportional to the error.
• 𝑐 𝑡 = 𝐾𝑐 𝜖 𝑡 + 𝑐𝑠 1
• 𝐾𝑐 –proportional gain of the controller
• 𝑐𝑠 – controller’s bias signal (actuating signal when 𝜖 = 0)

• Proportional controller described by its proportional gain 𝐾𝑐 or by its


proportional band (PB) where
100
• 𝑃𝐵 = (2) Usually 1 ≤ 𝑃𝐵 ≤ 500
𝐾𝑐

93
Proportional Controller (P-Controller)
• The higher the gain 𝐾𝑐 or the smaller the PB, the higher the sensitivity
of controller actuating signal to deviation 𝜖 will be.

• The transfer function of a proportional controller is given as


• 𝐺𝑐 𝑠 = 𝐾𝑐 3

94
Proportional-Integral Controller (PI Controller)
• Commonly called proportional-plus-reset controller.
• Its actuating signal is related to the error thus;
𝐾𝑐 𝑡
• 𝑐 𝑡 = 𝐾𝑐 𝜖 𝑡 + ‫׬‬0
𝜖 𝑡 + 𝑐𝑠 4
𝜏𝐼

• 𝜏𝐼 –integral time constant or Reset time in minutes


• 𝜏𝐼 is an adjustable parameter
• 0.1 ≤ 𝜏𝐼 ≤ 50𝑚𝑖𝑛

95
Proportional-Integral Controller (PI Controller)
• Reset time is the time needed by the controller to repeat the initial
proportional action change in its output.
• The integral action causes the output 𝑐(𝑡) to change as long as an
error exists in the process output.
• Therefore such a controller can eliminate even small errors.

• Transfer function of a PI controller is given as


1
• 𝐺𝑐 𝑠 = 𝐾𝑐 1 + 5
𝜏𝐼 𝑠

96
Proportional-Integral Controller (PI Controller)
• What is the Integral Windup?

• The integral term of a PI controller causes its output to continue


changing as long as there is a non-zero error.
• Often the errors cannot be eliminated quickly.
• Given time they produce larger and larger values for the integral term
which in turn keeps increasing the control action until it is saturated.
• The valve is then either completely open or closed.

97
Proportional-Integral-Derivative Controller (PID
Controller)
• The output is given by
𝐾𝑐 𝑡 𝑑𝜖
• 𝑐 𝑡 = 𝐾𝑐 𝜖 𝑡 + ‫𝜖 ׬‬ 𝑡 𝑑𝑡 + 𝐾𝑐 𝜏𝐷 + 𝐶𝑠 6
𝜏𝐼 0 𝑑𝑡

• 𝜏𝐷 –derivative time constant

• The derivative term anticipates what the error will be


• It applies a control action proportional to the current rate of change
in the error.

98
Proportional-Integral-Derivative Controller (PID
Controller)
• Called anticipatory control.

• Major drawbacks to derivative control are


1) For a response with constant non-zero error it gives no control
action since 𝑑𝜖Τ𝑑𝑡 = 0
2) For noisy response with almost zero error it can compute large
derivatives and thus yield large control action, although it is not
needed.
1
• 𝐺𝑐 𝑠 = 𝐾𝑐 1 + + 𝜏𝐷 𝑠 7
𝜏𝐼 𝑠

99
Measuring Devices
• Accurate measuring device is key to proper control

• Flow sensors
• Pressure sensors
• Temperature sensors
• Composition analyzers
• Final control element

100
Dynamic Behavior of Feedback – Controlled
Processes
• Behavior of feedback control systems is dictated by
1) The value of the disturbance (load) d
• Or
2) The change in the desired value of the set-point 𝑦𝑠𝑝 .

• Block Diagram and the Closed-Loop Response


• The generalized closed-loop system is basically made up of
• (i) Process (ii) Measuring device (iii) Controller
(iv) Final Control Element

101
Dynamic Behavior of Feedback – Controlled
Processes
• From the generalized closed-loop as shown below

102
Dynamic Behavior of Feedback – Controlled
Processes
• Determine the transfer function of each component.

• For the process


• 𝑦ത 𝑠 = 𝐺𝑝 𝑠 𝑀 ഥ 𝑠 + 𝐺𝑑 𝑠 𝑑ҧ 𝑠 1

• Measuring device
• 𝑦ത𝑚 = 𝐺𝑚 𝑠 𝑦ത 𝑠 2

103
Dynamic Behavior of Feedback – Controlled
Processes
• Controller Mechanism
• Comparator:
• 𝜖 ҧ 𝑠 = 𝑦ത𝑠𝑝 𝑠 − 𝑦ത𝑚 𝑠 (3𝑎)
• Control action:
• 𝑐ҧ 𝑠 = 𝐺𝑐 𝑠 𝜖 ҧ 𝑠 (3𝑏)

• Final Control Element


•𝑀ഥ 𝑠 = 𝐺𝑓 𝑠 𝑐ҧ 𝑠 4

104
Dynamic Behavior of Feedback – Controlled
Processes
• The blocks between the comparator and the controlled output
constitute the forward path (𝐺𝑓 , 𝐺𝑐 , 𝐺𝑝 ) and
• 𝐺𝑚 is on the feedback path.

• Taking the process


• 𝑦ത 𝑠 = 𝐺𝑝 𝑠 𝑀 ഥ 𝑠 + 𝐺𝑑 𝑠 𝑑ҧ 𝑠 −−− −(1)
• With some analysis we obtain
𝐺𝑓 𝐺𝑝 𝐺𝑐 𝐺𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 (5)
1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚 1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚

105
Dynamic Behavior of Feedback – Controlled
Processes
• From the generalized closed-loop as shown below

106
Dynamic Behavior of Feedback – Controlled
Processes
• This is the closed-loop response of the process.
𝐺 𝐺𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 6
1+𝐺𝐺𝑚 1+𝐺𝐺𝑚

• 𝑦ത 𝑠 = 𝐺𝑠𝑝 𝑦ത𝑠𝑝 𝑠 + 𝐺𝑙𝑜𝑎𝑑 𝑑ҧ 𝑠 7

• For feedback control system we have two types of control problems


Servo and Regulator problem.

107
Dynamic Behavior of Feedback – Controlled
Processes
• Servo problem: Disturbance is constant and set-point is changing
• 𝑦ത 𝑠 = 𝐺𝑠𝑝 𝑠 𝑦ത𝑠𝑝 𝑠 8

• Regulator Problem: set-point is constant and changing load


• 𝑦ത 𝑠 = 𝐺𝑙𝑜𝑎𝑑 𝑑ҧ 𝑠 (9)

• NOTE: closed-loop overall transfer functions 𝐺𝑠𝑝 and 𝐺𝑙𝑜𝑎𝑑 depend


not only on the process dynamics but also the dynamics of the
measuring sensor, controller and final control element.

108
Effect of Proportional Control on the Response
of a Controlled Process
• The closed-loop response of a process is given by Equation (5).
𝐺𝑓 𝐺𝑝 𝐺𝑐 𝐺𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 (5)
1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚 1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚
• Let
• 𝐺𝑚 𝑠 = 1, 𝐺𝑓 (𝑠) = 1

• For proportional controller


• 𝐺𝑐 𝑠 = 𝐾𝑐

109
Effect of Proportional Control on the Response
of a Controlled Process
• Equation (5)
𝐺𝑓 𝐺𝑝 𝐺𝑐 𝐺𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 (5)
1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚 1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚

• Becomes

𝐺𝑝 𝑠 𝐾𝑐 𝐺𝑑 𝑠
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 −−− − 10
1+𝐺𝑝 𝑠 𝐾𝑐 1+𝐺𝑝 𝑠 𝐾𝑐

110
• First-Order Systems
• First-order systems are defined by the following equation

𝑑𝑦
• 𝜏𝑝 + 𝑦 = 𝐾𝑝 𝑚 + 𝐾𝑑 𝑑
𝑑𝑡

• 𝑤𝑖𝑡ℎ 𝑦 0 = 𝑚 0 = 𝑑 0 = 0

111
Effect of Proportional Control on the Response
of a Controlled Process
• Laplace transform gives
𝐾𝑝 𝐾𝑑
• 𝑦ത 𝑠 = 𝑚
ഥ 𝑠 + 𝑑ҧ 𝑠
𝜏𝑝 𝑠+1 𝜏𝑝 𝑠+1

• For the uncontrolled system


• Time constant: 𝜏𝑝
• Static gains : 𝐾𝑝 for the manipulation
• 𝐾𝑑 for the load

112
Effect of Proportional Control on the Response
of a Controlled Process
• If
𝐾𝑝 𝐾𝑑
• 𝐺𝑝 𝑠 = 𝑎𝑛𝑑 𝐺𝑑 𝑠 =
𝜏𝑝 𝑠+1 𝜏𝑝 𝑠+1

• Substituting into Equation (10) gives


𝐺𝑝 𝑠 𝐾𝑐 𝐺𝑑 𝑠
• 𝑦ത 𝑠 = 𝑦ത 𝑠 + 𝑑ҧ 𝑠 −−− − 10
1+𝐺𝑝 𝑠 𝐾𝑐 𝑠𝑝 1+𝐺𝑝 𝑠 𝐾𝑐

𝐾𝑝 𝐾𝑐 𝐾𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠
𝜏𝑝 𝑠+1+𝐾𝑝 𝐾𝑐 𝜏𝑝 𝑠+1+𝐾𝑝 𝐾𝑐

113
Effect of Proportional Control on the Response
of a Controlled Process
• Rearranging we have
𝐾𝑝 ′ 𝐾𝑑′
• 𝑦ത 𝑠 = ′ 𝑠+1 𝑦ത𝑠𝑝 𝑠 + ′ 𝑠+1 𝑑ҧ 𝑠 −−− −(11)
𝜏𝑝 𝜏𝑝
• Where
𝜏𝑝
• 𝜏𝑝′ = −−− −(12 𝑎)
1+𝐾𝑝 𝐾𝑐
𝐾𝑝 𝐾𝑐
• 𝐾𝑝′ = −−− −(12 𝑏)
1+𝐾𝑝 𝐾𝑐
𝐾𝑑
• 𝐾𝑑′ = −−− − 12 𝑐
1+𝐾𝑝 𝐾𝑐

114
Effect of Proportional Control on the Response
of a Controlled Process
• 𝐾𝑝′ 𝑎𝑛𝑑 𝐾𝑑′ − 𝐶𝑙𝑜𝑠𝑒𝑑 𝑙𝑜𝑜𝑝 𝑠𝑡𝑎𝑡𝑖𝑐 𝑔𝑎𝑖𝑛𝑠

• Characteristics of the closed-loop response of a first-order system


1) It remains first-order with respect to load and set-point

2) Time constant is reduced i.e 𝜏𝑝′ < 𝜏𝑝 meaning that closed-loop


response has become faster than open-loop response

3) Static gains decrease

115
Effect of Proportional Control on the Response
of a Controlled Process
• Assuming a unit step change in the set point (Servo problem) –
Constant disturbance
1
• 𝑦ത𝑠𝑝 𝑠 = 𝑑ҧ 𝑠 = 0
𝑠
• Then Equation (11)
𝐾𝑝 ′ 𝐾𝑑′
• 𝑦ത 𝑠 = ′ 𝑠+1 𝑦ത𝑠𝑝 𝑠 + ′ 𝑠+1 𝑑ҧ 𝑠 −−− −(11)
𝜏𝑝 𝜏𝑝
• becomes
𝐾𝑝 ′ 1
• 𝑦ത 𝑠 = ′ 𝑠+1 𝑠
𝜏𝑝

116
Effect of Proportional Control on the Response
of a Controlled Process
• After inversion we have
−𝑡
′ ൗ𝜏′
• 𝑦 𝑡 = 𝐾𝑑 1 − 𝑒 𝑝 −−− −(14)

• Following is the response to a unit step change in load

117
Effect of Proportional Control on the Response
of a Controlled Process
• Similarly, there is an offset from the set-point
• 𝑜𝑓𝑓𝑠𝑒𝑡 = 𝑠𝑒𝑡 𝑝𝑜𝑖𝑛𝑡 − (𝑢𝑙𝑡𝑖𝑚𝑎𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒)
−𝐾𝑑
• 𝑜𝑓𝑓𝑠𝑒𝑡 = 0 − 𝐾𝑑′ =
1+𝐾𝑝 𝐾𝑐

• The response is much closer to the desired set point than would have
been with no control
• Although the offset tends to zero as 𝐾𝑐 → ∞, we never use extremely
large 𝐾𝑐 for proportional control; because of stability issues

118
Second-Order Systems (Servo Problem)
• Transfer function for a second-order process is
𝑦(𝑠) 𝐾𝑝
• 𝐺𝑝 𝑠 = =
𝑚(𝑠) 𝜏2 𝑠2 +2𝜁𝜏𝑠+1

• Putting this expression into Equation (10) and knowing that for servo
problem 𝑑 𝑠 = 0,

𝐺𝑝 𝑠 𝐾𝑐 𝐺𝑑 𝑠
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 −−− − 10
1+𝐺𝑝 𝑠 𝐾𝑐 1+𝐺𝑝 𝑠 𝐾𝑐
𝐾𝑝′
•𝑦 𝑠 = 𝑦 𝑠 −−− −(14)
𝜏′ 2 𝑠2 +2𝜁 ′ 𝜏′ 𝑠+1 𝑠𝑝

119
Second-Order Systems (Servo Problem)
• Where
𝜏
• 𝜏′ = −−− −(15, 𝑎)
1+𝐾𝑝 𝐾𝑐
𝜁
• 𝜁′ = −−− −(15, 𝑏)
1+𝐾𝑝 𝐾𝑐
𝐾𝑝 𝐾𝑐
• 𝐾𝑝′ = −−− −(15, 𝑏)
1+𝐾𝑝 𝐾𝑐

120
Second-Order Systems (Servo Problem)
• Observations
• The closed-loop response of a 2nd Order system with proportional
control has the following characteristics
1) It remains 2nd Order
2) Static gain decreases
3) Natural period decreases
4) Damping factor decreases
• Implication: Overdamped process, with proportional control and
appropriate value of Kc may become underdamped (oscillatory).

121
Second-Order Systems (Servo Problem)
• Consider a unit step change in the set point i.e 𝑦𝑠𝑝 𝑠 = 1Τ𝑠
𝐾𝑝′
• Eqn (14) 𝑦 𝑠 = 𝑦𝑠𝑝 𝑠 −−− −(14)
𝜏′ 2 𝑠2 +2𝜁 ′ 𝜏′ 𝑠+1

• becomes

𝐾𝑝′ 1
•𝑦 𝑠 = −−− −(14)
𝜏′ 2 𝑠2 +2𝜁 ′ 𝜏′ 𝑠+1 𝑠

122
Second-Order Systems (Servo Problem)
• Depending on the value 𝜁 ′ the inverse of the expression above may
be given by
• Overdamped case 𝜁 ′ > 1
−𝜁𝑡ൗ 𝑡 𝜁 𝑡
• 𝑦 𝑡 = 𝐾𝑝 1 − 𝑒 𝜏 𝑐𝑜𝑠ℎ 𝜁2 − 1 + 𝑠𝑖𝑛ℎ 𝜁2 − 1 −
𝜏 𝜁 2 −1 𝜏
− (14, 𝑎)

• Critically damped 𝜁 ′ = 1
𝑡 −𝑡Τ
• 𝑦 𝑡 = 𝐾𝑝 1 − 1 + 𝑒 𝜏 −−− −(14, 𝑏)
𝜁

123
Second-Order Systems (Servo Problem)
• Underdamped 𝜁 ′ < 1
1 −𝜁𝑡ൗ
• 𝑦 𝑡 = 𝐾𝑝 1 − 𝑒 𝜏 𝑠𝑖𝑛 𝜔𝑡 + 𝜑 −−− −(14, 𝑐)
1−𝜁 2

• where
1−𝜁 2
•𝜔=
𝜏
−1 1−𝜁 2
• 𝜑 = 𝑡𝑎𝑛
𝜁

124
Second-Order Systems (Servo Problem)
• Independently of 𝜁 ′ the ultimate value of y(t) is given by the final
value theorem, thus

𝐾𝑝 𝐾𝑐
• 𝑦 𝑡 → ∞ = lim 𝑠𝑦ത 𝑠 = 𝐾𝑝′ =
𝑠→0 1+𝐾𝑝 𝐾𝑐

• There is an offset in this case as well.


• 𝑂𝑓𝑓𝑠𝑒𝑡 = 𝑁𝑒𝑤 𝑆𝑒𝑡 𝑃𝑜𝑖𝑛𝑡 − (𝑈𝑙𝑡𝑖𝑚𝑎𝑡𝑒 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑅𝑒𝑠𝑝𝑜𝑛𝑠𝑒)

125
Second-Order Systems (Servo Problem)
𝐾𝑝 𝐾𝑐 1
• 𝑂𝑓𝑓𝑠𝑒𝑡 = 1 − =
1+𝐾𝑝 𝐾𝑐 1+𝐾𝑝 𝐾𝑐

• ⟹ 𝑂𝑓𝑓𝑠𝑒𝑡 → 0 𝑓𝑜𝑟 𝐾𝑐 → ∞

• Remarks
• (1) Depending on the value of 𝜁 for the uncontrolled 2nd Order
system Equation (15, b)
′ 𝜁
•𝜁 = −−− −(15, 𝑏)
1+𝐾𝑝 𝐾𝑐

126
Second-Order Systems (Servo Problem)
• Can have
• 𝜁 ′ < 1, 𝜁 ′ = 1, 𝜁′ > 1

• For 𝜁 ′ > 1 we have overdamped response of the closed loop system


which is sluggish.
• We can increase 𝐾𝑐 and make 𝜁 ′ < 1.
• We then have closed-loop response which is faster but becomes
oscillatory.
• Also with increased 𝐾𝑐 the offset decreases.

127
Second-Order Systems (Servo Problem)
• Remarks (continued)
• (2) Increase in the speed of the system’s response and decrease in the
offset, come at the expense of
• (i) higher overshoot (maximum errors) and
• (ii) longer oscillatory response.

• Thus as K c increases, causes 𝜁 ′ to decrease.

128
Second-Order Systems (Servo Problem)
• Remarks
−𝜋𝜁
• 𝑎 𝑂𝑣𝑒𝑟𝑠ℎ𝑜𝑜𝑡 = 𝑒𝑥𝑝 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠
1−𝜁 2
• 𝑏 𝐷𝑒𝑐𝑎𝑦 𝑅𝑎𝑡𝑖𝑜 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠
−2𝜋𝜁 2
• 𝐷𝑒𝑐𝑎𝑦 𝑟𝑎𝑡𝑖𝑜 = 𝑒𝑥𝑝 = 𝑜𝑣𝑒𝑟𝑠ℎ𝑜𝑜𝑡
1−𝜁 2
2𝜋𝜏
• 𝑐 𝑃𝑒𝑟𝑖𝑜𝑑 𝑜𝑓 𝑜𝑠𝑐𝑖𝑙𝑙𝑎𝑡𝑖𝑜𝑛, 𝑇 = 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠
1−𝜁 2

129
Second-Order Systems (Servo Problem)

130
Effect of Integral Control Action
• Applying Integral control action to a 1st order system and a servo
problem [d(s) = 0] i.e. changing set point, Eqn (5)

𝐺𝑓 𝐺𝑝 𝐺𝑐 𝐺𝑑
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 𝑠 + 𝑑ҧ 𝑠 (5)
1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚 1+ 𝐺𝑓 𝐺𝑐 𝐺𝑝 𝐺𝑚

• becomes
Gp GF GC
• 𝑦ത 𝑠 = 𝑦ത𝑠𝑝 −−− −(16)
1+ Gp GF GC G𝑚

131
Effect of Integral Control Action
• For simplicity let G𝑚 = GF = 1

• For 1st Order Process


Kp
• Gp =
𝜏p 𝑠+1

• For simple integral control action


1
• Gc = Kc
𝜏𝐼 𝑠

132
Effect of Integral Control Action
• Substitute Gp GF GC G𝑚 and after manipulations
1
• 𝑦(𝑠) = ysp 𝑠 −−− −(17)
τ2 S2 +2𝜁𝜏𝑠+1

• where
𝜏𝐼 𝜏𝑝
•𝜏= −−− −(18𝑎)
K𝑝 K𝑐

1 𝜏𝐼
•𝜁= −−− −(18𝑏)
2 τ𝑝 K𝑝 K𝑐

133
Effect of Integral Control Action
• Remarks
1) Integral control action increases the order of the dynamics for
closed loop response:
2) A 1st order system behaves like a 2nd order system
3) The response becomes more sluggish by increased order of process

• For a unit step change in the set point we have (Servo problem)
1 1
• 𝑦(𝑠) =
τ2 S2 +2𝜁𝜏𝑠+1 𝑠

134
Effect of Integral Control Action
• The shape of the curve depends on whether
• 𝜁>1 – overdamped
• 𝜁=1 – critically damped
• 𝜁<1 – underdamped
• The ultimate value of the response can be obtained from the final
value theorem
• 𝑦 𝑡 → ∞ = lim 𝑠𝑦(𝑠) ത
𝑠→0
1
•𝑦 𝑡→∞ = lim 2 2 =1
𝑠→0 τ S +2𝜁𝜏𝑠+1

135
Effect of Integral Control Action
• 𝑜𝑓𝑓𝑠𝑒𝑡 = 𝑠𝑒𝑡 𝑝𝑜𝑖𝑛𝑡 − (𝑢𝑙𝑡𝑖𝑚𝑎𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒)
• Therefore, the offset = 1 -1 = 0
• Integral control action eliminates offset.
• It can also eliminate offset in Regulator problem.

• Notes
1 𝜏𝐼
• (1) Eqn (18-b) 𝜁= −−− −(18𝑏)
2 τ𝑝 K𝑝 K𝑐

• Closed loop response ζ<1, ζ = 1, ζ >1 depends on 𝐾𝑐 and 𝜏𝐼

136
Effect of Integral Control Action
• NOTE (2)
• From Eqn (18-b) we notice that as 𝐾𝑐 increases, ζ decreases.
• Consequences of decreasing ζ are
• (a) Response moves from sluggish overdamped to faster but
oscillatory underdamped behavior.
• (b) Overshoot and decay ratio of the closed-loop response both
increase.
• We can improve the speed of the closed-loop response at the
expense of higher deviation and longer oscillations.

137
Effect of Integral Control Action
• Effects of gain on the closed loop response of 1st order systems with
integral control

138
Effect of Integral Control Action
• NOTE 3
• From Eqn (18.b) we observe that as 𝜏𝐼 decreases, ζ decreases
therefore the behavior is as the forgoing in (2) with the response as in
the figure above.

• NOTE 4
• Summarizing, increasing the integral control action makes the
response of the closed loop system more sensitive

139
Effects of Derivative Control Action
• For derivative control action alone we have 𝐺𝑐 = K c 𝜏D 𝑠

• For simplicity, let G𝑚 = GF = 1 the closed loop response of a first-order


system is given by
KP Kc 𝜏𝐷 𝑠
• 𝑦 𝑠 = ysp 𝑠 −− −(19)
(𝜏𝑃 + KP Kc 𝜏𝐷 )𝑠+1

• Observations from Eq (19)


• (1) Derivatives control does not change the order of the response.

140
Effects of Derivative Control Action
• (2) Effective time constant is 𝜏𝑃 + 𝐾𝑃 𝐾𝐶 𝜏𝐷 > 𝜏𝑃
• The response of the controlled process is slower than original
process.
• As K c increases, the effective time constant increases and response
becomes progressively slower.

141
Derivative Control on a Second (2nd) Order
System
• Let G𝑚 = GF = 1

• (1) The closed-loop response for a 2nd Order system is


KP Kc 𝜏𝐷 𝑠
• 𝑦(𝑠) 2 2 ysp (𝑠)
τ S +(2𝜉𝜏 + KP Kc 𝜏𝐷 )𝑠+1
• Observation
• (a) The natural period of the closed-loop remains the same
• (b) New damping factor 𝜉’ can be found from the following eqn.
• 2𝜁 ′ 𝜏 = 2𝜁𝜏 + 𝐾𝑝 𝐾𝑐 𝜏𝐷
142
Derivative Control on a Second (2nd) Order
System
• That is 𝜁 ′ > 𝜁

• Implies that the closed-loop response is more damped and damping


increases as K c or 𝜏𝐷 increases
• Produces a more robust behavior by the controlled process.

• (2) The decrease in the speed of the response and increase in


damping demonstrates that the derivative control action produced
more robust behavior by the controlled process.

143
Effect of Composite Control Actions
Effect of PI Control
• Usual controllers are a combination of P, I and D.

• Combination of PI modes leads to the following effects on the


response of a closed-loop system:
• (1) Order of the response increases (effect of Integral mode)
• (2) Offset is eliminated (effect of Integral mode)
• (3) (i) As K 𝑐 increases, response becomes faster (Proportional and
Integral mode) and (ii) more oscillatory to set point changes
[overshoot and decay ratio increase (effect of Integral mode )].

144
Effect of Composite Control Actions
Effect of PI Control
• Large values of K 𝑐 create very sensitive response and may lead to
instability.

• (4) As τ𝐼 decreases, for constant K 𝑐 , response becomes faster but


more oscillating with higher overshoot and decay ratio (effect of
Integral mode)

145
Effect of PID Control
• For a PID controller the derivative action brings a moderative effect
on the wild offshoot and oscillation of large K 𝑐
• Increasing K 𝑐 leads to faster response, however overshoot remains
almost the same and the settling time is shorter.

146
Stability Analysis of Feedback Systems
• Dynamics of an uncontrolled process changes when hooked up to
measuring devices, controllers and final control elements etc

• Non-oscillatory 1st Order processes may acquire oscillatory behavior


with PI control.

• Also 2nd order processes may become unstable with a PI controller


with wrong selection of K 𝑐 and τ𝐼

147
The Concept of Stability
• A system is considered unstable if after it has been disturbed by an
input change and does not return to the initial state.

• For our purpose, we define stability as:


• A dynamic system is considered to be stable if for every bounded
input it produced output regardless of its initial state.
• “Bounded” is an input that always remains between an upper and a
lower limit e.g. sinusoidal, step
• Unbounded output means very large

148
The Concept of Stability
• (a) Stable and (b) unstable responses

149
The Concept of Stability
• The location of the poles of a transfer function gives us the first
criterion for checking the stability of a system

• Even one pole with positive real part renders the system unstable.

• Example
• Stabilization of an Unstable Process with P Control
• Consider a process with the following response.

150
The Concept of Stability
• Where the system has an input m.
10 5
•𝑦 𝑠 = 𝑚 𝑠 + 𝑑 𝑠
𝑠−1 𝑠−1

• Solution
• Solution of the function possesses a pole at s = 1 > 0 implying
Instability
• For a unit step change in the load, the response is curve α

151
The Concept of Stability
• Curve α, open-loop unstable response;
• Curve β, closed-loop stable response with P control

152
The Concept of Stability
• Introducing a feedback control system with proportional control only
and assume G𝑚 = GF = 1
• Block diagram of closed-loop system

153
The Concept of Stability
• The closed-loop response of the system is given by Equation (5)
𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑑
•𝑦 𝑠 = 𝑦𝑠𝑝 𝑠 + 𝑑(𝑠)
1+𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚 1+𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚

• For this system comes out as


10𝐾𝑐 5
•𝑦 𝑠 = 𝑦𝑠𝑝 𝑠 + 𝑑(𝑠)
𝑠− 1−10𝐾𝑐 𝑠−(1−10𝐾𝑐 )
• The closed-loop transfer functions therefore are
10𝐾𝑐 5
• 𝐺𝑠𝑝 = ; 𝐺𝑙𝑜𝑎𝑑 =
𝑠− 1−10𝐾𝑐 𝑠− 1−10𝐾𝑐

154
The Concept of Stability
• These transfer functions have a negative pole if Kc > 1/10, leading to a
stabilized system.

• Curve β is for unit step change in load for Kc = 1

155
The Characteristic Equation
• Consider the generalized feedback control system shown below

156
The Characteristic Equation
• Closed-loop response is given by Equation (5) which can be written in
an equivalent form
• 𝑦 𝑠 = 𝐺𝑠𝑝 𝑦𝑠𝑝 𝑠 + 𝐺𝑙𝑜𝑎𝑑 𝑑(𝑠)
𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑑
•𝑦 𝑠 = 𝑦𝑠𝑝 𝑠 + 𝑑(𝑠)
1+𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚 1+𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚
• Stability characteristics of the closed-loop response determined by
the poles of the transfer functions Gsp and Gload
• Poles are given by the solution of the equation
• 1 + 𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚 = 0 20
• Equation (20) is known as the Characteristic Equation
• (Denominator in Equation (5)

157
Stability Criterion
• A feedback control system is stable if all the roots of its characteristic
equation have negative real parts
• Stability criterion is independent of changes being in the set-point or
load
• Roots of the Characteristic Equation are poles of closed-loop transfer
functions Gsp and Gload hence called closed-loop poles.

158
Stability Criterion
• Example
• Stability Analysis of Feedback loops
• For the example looked at earlier we have
10
• 𝐺𝑝 = , 𝐺𝑓 = 1, 𝐺𝑚 = 1 , 𝐺𝑐 = 𝐾𝑐
𝑠−1

• Characteristic Equation is
10
• 1 + 𝐺𝑝 𝐺𝑓 𝐺𝑐 𝐺𝑚 = 1 + . 1. 𝐾𝑐 . 1 = 0
𝑠−1

159
Stability Criterion
• Roots p 𝑝 = 𝑠 = 1 − 10𝐾𝑐
• System is stable if 𝑝 < 0 𝑖𝑚𝑝𝑙𝑦𝑖𝑛𝑔 𝐾𝑐 > 1Τ10

• Example
1
• 𝐺𝑝 = , 𝐺𝑓 = 1 , 𝐺𝑚 = 1
𝑠2 +2𝑠+2
1
• 𝐺𝑐 = 𝐾𝑐 1 +
𝜏𝐼 𝑠
• Characteristic Equation
1 1
• 1 + 𝐺𝑝 𝐺𝑓 𝐺𝑚 𝐺𝑐 = 1 + . 1 . 𝐾𝑐 1 + =0
𝑠2 +2𝑠+𝑠 𝜏𝐼 𝑠

160
Stability Criterion
• For 𝐾𝑐 = 100 , 𝜏𝐼 = 0.1
• 𝑠 3 + 2𝑠 2 + 102𝑠 + 1000 = 0

• Roots
• 𝑃1 = −7.185, 𝑃2 = 2.59 + 𝑗 11.5 , 𝑃3 = 2.59 − 𝑗 11.5

• For the real part of the roots being positive the system is unstable.

161
Stability Criterion
• Without the controller the 2nd Order system with transfer function
1
• 𝐺𝑝 𝑠 =
𝑠2 +2𝑠+2

• Has solution or poles: 𝑝1 = −1 + 𝑗 𝑝2 = −1 − 𝑗

• The negative real part means the open system is stable.

162
Routh- Hurwitz Criterion for Stability
• Routh-Hurwitz procedure allows us to test if any root is to the right of
the imaginary axis
• First expand the characteristic equation into a polynomial form thus
• 1 + 𝐺𝑝 𝐺𝑐 𝐺𝑓 𝐺𝑚 = 𝑎0 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛 = 0
• Let 𝑎0 be positive. If it is negative, multiply both sides by -1.

• First Test
• If any of the coefficients 𝑎1 , 𝑎2 , 𝑎3 , … , 𝑎𝑛−1 , 𝑎𝑛 is negative, one root
has positive real part, hence system is unstable. No further analysis

163
Routh- Hurwitz Criterion for Stability
• Second Test
• If all of the coefficients 𝑎1 , 𝑎2 , 𝑎3 , … , 𝑎𝑛−1 , 𝑎𝑛 are positive, we cannot
conclude
• Therefore form the Routh Array
𝑅𝑜𝑤 1 𝑎0 𝑎2 𝑎4 𝑎6 …
𝑅𝑜𝑤 2 𝑎1 𝑎3 𝑎5 𝑎7 …

• 𝑅𝑜𝑤 3
𝑅𝑜𝑤 4
𝐴1 𝐴2 𝐴3 𝐴4 …
𝐵1 𝐵2 𝐵3 𝐵4 …
𝑅𝑜𝑤 5 𝐶1 𝐶2 𝐶3 𝐶4 …
: : : : : …
𝑅𝑜𝑤 𝑛 + 1 𝑊1 𝑊2 𝑊3 𝑊4 …

164
Routh- Hurwitz Criterion for Stability
• Where
𝑎1 𝑎2 −𝑎0 𝑎3 𝑎1 𝑎4 −𝑎0 𝑎5 𝑎1 𝑎6 −𝑎0 𝑎7
• 𝐴1 = ; 𝐴2 = ; 𝐴3 = ……
𝑎1 𝑎1 𝑎1

𝐴1 𝑎3 −𝑎1 𝐴2 𝐴1 𝑎5 −𝑎1 𝐴3 𝐴1 𝑎7 −𝑎1 𝐴4


• 𝐵1 = ; 𝐵2 = ; 𝐵3 = ……
𝐴1 𝐴1 𝐴1

𝐵1 𝐴2 −𝐴1 𝐵2 𝐵1 𝐴3 −𝐴1 𝐵3
• 𝐶1 = ; 𝐶2 = … … 𝑒𝑡𝑐
𝐵1 𝐵1

• Examine the elements of the 1st column of the array


• 𝑎0 , 𝑎1 , 𝐴1 , 𝐵1 , 𝐶1 , … , 𝑊1
165
Routh- Hurwitz Criterion for Stability
• Routh-Hurwitz Criterion
1) If any of these elements is negative, we have at least one root to
the right of the imaginary axis
2) The number of the sign changes in the elements of the first column
is equal to the number of roots to the right of the imaginary axis.

• System is stable if all the elements in the first column of the Routh
array are positive.

166
Routh- Hurwitz Criterion for Stability
• Example

• Stability Analysis with the Routh-Hurwitz Criterion


• Considering the feedback control of a 2nd order system, the
characteristic equation is
𝐾𝑐
• 𝑠3 + 2𝑠 2 + 2 + 𝐾𝑐 𝑠 + =0
𝜏𝐼

• The Routh array is

167
Routh- Hurwitz Criterion for Stability
• Routh array is
𝑅𝑜𝑤 1 1 2 + 𝐾𝑐
𝐾𝑐
𝑅𝑜𝑤 2 2
• 𝜏𝐼
2 2+𝐾𝑐 −𝐾𝑐ൗ𝜏𝐼
𝑅𝑜𝑤 3 0
2
𝐾𝑐
• 𝑅𝑜𝑤 4 ൗ𝜏𝐼

2 2+𝐾𝑐 −𝐾𝑐 Τ𝜏𝐼 𝐾𝑐


• Elements of the first column are 1, 2, ,
2 𝜏𝐼

168
Routh- Hurwitz Criterion for Stability
• All are always positive except the third which can be positive or
negative, depending on Kc and τI.
1) If 𝐾𝑐 = 100, 𝜏𝐼 = 0.1, the 3rd element of the 1st column becomes (-
398 < 0), implying an unstable system. Also we have 2 sign changes
in the elements of the 1st column. Therefore we have 2 roots with
positive real points.
2) If 𝐾𝑐 = 100, 𝜏𝐼 = 0.5, the 3rd element is equal to +2 > 0, implying a
stable system since all the elements of the first column are positive.
3) In general the system is stable if Kc and τI satisfy the condition
𝐾𝑐ൗ
2(2 + 𝐾𝑐 ) > 𝜏𝐼

169
Routh- Hurwitz Criterion for Stability
• Limitations of Routh array
1) It cannot handle systems with dead time. It tells if the system is
stable or unstable, but gives no information about how close to
instability a system may be.

2) Another limitation of the Routh method is the need to express the


characteristic equation explicitly as a polynomial. It gets complex in
high order systems.

170
Design of Feedback Controllers
• Critical questions
1) How do we select the type of feedback controller (i.e. P, PI or PID)?
2) How do we adjust the parameters of the selected controller (Kc, τI,
τD) to achieve ‘optimum’ response for the controlled process?

• Outline of the Design Problems


• Consider block diagram of a closed loop system as shown below.

171
Design of Feedback Controllers
• Block diagram of generalized closed-loop system

172
Design of Feedback Controllers
• With a change in load or set-point, controller tries to bring the output
again close to the desired set-point

• Response of a system to unit step change in load with no control, and


various types of feedback controllers

173
Design of Feedback Controllers
• First Design Question
• Q1. What type of feedback controller should be used to control a
given process?
• Assuming a PI controller was selected, we need to select values for Kc
(Gain) and (reset time) τI, as depicted on the Figures below.

174
Design of Feedback Controllers
• For set-point changes, we can improve the speed of the closed-loop
response at the expense of higher deviation and longer oscillations,
large Kc

• Q2. How do we select the best values for the adjustable parameters
of a feedback controller? (Controller tuning)

• Question 1 and 2 can be best answered in a quantitative manner to


be able to compare. Use of a performance criterion.

175
Design of Feedback Controllers
• Q3. What performance criterion should be used for the selection and
the tuning of the controller?
• Variety of performance criteria that can be used include;
1) Keep the maximum deviation (error) as small as possible
2) Achieve short settling times
3) Minimize the integral of the errors until the process has settled to
its desired set-point etc.

• Different performance criteria lead to different control designs.

176
Simple Performance Criteria
• Consider two different feedback control systems
• If criterion for design is
• Return to the desired level of operation as soon as possible
• Then controller A is selected

177
Simple Performance Criteria
• However, if the criterion is to

• Keep the maximum deviation as small as possible


• Or
• Return to the desired level of operation and stay close to it in the
shortest time.

• Then controller B will be selected

178
Simple Performance Criteria
• For every process control application we can distinguish between
1) Steady state performance criteria
2) Dynamic response performance criteria

• Steady State Performance Criteria


• Principal performance criterion is zero-error at steady-state.
• Proportional controller cannot achieve zero steady-state error
• PI controller can
• For Proportional control the steady-state error (offset) tends to zero
as 𝐾𝑐 → ∞

179
Simple Performance Criteria
• Dynamic Performance Of Closed- Loop Systems
• This is based on two types of criteria.
1) Criteria that use only a few points of the response. Usually simpler
but approximate.
2) Criteria that use the entire closed-loop response from time t = 0
until t = very large. These are more precise but more cumbersome.

180
Simple Performance Criteria
• Approximate Simple Performance Criteria
• Simple performance criteria are based on some characteristic
features of the closed-loop response such as;
1) Overshoot
2) Rise time
3) Settling time
4) Decay ratio
5) Frequency of oscillation of the transient.

181
Simple Performance Criteria
• We could design for the controller to have
• Minimum overshoot, or
• Minimum settling time
• By adjusting and selecting appropriate 𝐾𝑐 or 𝜏𝐼 , 𝜏𝐷 , 𝑒𝑡𝑐.

• For desired dynamic response, require that more objectives be


satisfied (i.e. minimize overshoot/ minimize settling time etc.)
• Unfortunately, such demands result in conflicting response
characteristics

182
Simple Performance Criteria
• For example, for PID controller: by decreasing the value of overshoot
we increase the settling time.
• Effect of gain on the closed-loop response of first-order systems with
PID control

183
Simple Performance Criteria
• Most popular practical performance criterion: Decay ratio
• The one-quarter decay ratio: reasonable trade-off between a fast rise
time and a reasonable settling time

• Example
• Consider the servo-control problem of a 1st order process with a PI
controller. The closed-loop response is given by the following
equation, when 𝐺𝑚 = 𝐺𝑓 = 1
𝜏1 𝑠+1
•𝑦 𝑠 = 𝑦𝑠𝑝 𝑠
𝜏2 𝑠2 +2𝜁𝜏𝑠 +1

184
Simple Performance Criteria
• Where
𝜏𝐼 𝜏𝑝 1 𝜏𝐼
•𝜏= 𝜁= (1 + 𝐾𝑝 𝐾𝑐 )
𝐾𝑝 𝐾𝑐 2 𝜏𝑝 𝐾𝑝 𝐾𝑐

• For best values for 𝐾𝑐 and 𝜏𝐼 we can use the simple criterion of an
underdamped response of a 2nd order system.
• Selecting the 1Τ4 decay ratio criterion and knowing that
−2𝜋𝜁
• 𝑑𝑒𝑐𝑎𝑦 𝑟𝑎𝑡𝑖𝑜 = exp
1−𝜁 2

185
Simple Performance Criteria
• For the problem, we have
𝜏
−2𝜋1Τ2 𝜏 𝐾 𝐼 𝐾 1+𝐾𝑝 𝐾𝑐
𝑝 𝑝 𝑐 1
• 𝑒𝑥𝑝 1 𝜏
=
1−4𝜏 𝐾 𝐼 𝐾 (1+𝐾𝑝 𝐾𝑐 )2 4
𝑝 𝑝 𝑐

• Which simplifies to
𝜏𝐼 1
• −2𝜋 2 1 + 𝐾𝑝 𝐾𝑐 = ln
4𝜏𝑝 𝐾𝑝 𝐾𝑐 −𝜏𝐼 1+𝐾𝑝 𝐾𝑐 4

186
Simple Performance Criteria
1
• Several controller settings can satisfy the decay ratio criterion.
4
• For 𝐾𝑝 = 0.1 and 𝜏𝑝 = 10
• The following solutions apply
𝑲𝒄 1 10 30 50 100

𝝉𝑰 0.153 0.464 0.348 0.258 0.153

• Usually select 𝐾𝑐 so that the controller has the necessary ‘strength’ to


push the response back to the desired set-point

187
Time Integral Performance Criteria
• This involves the shape of the complete closed-loop response from
time t = 0 until steady state has been reached.
• Criteria in this category are based on the entire response of the
process.

• Most often used are:


1) Integral of square error (ISE)
2) Integral of the absolute value of the error (IAE)
3) Integral of time-weighted absolute error (ITAE)

188
Time Integral Performance Criteria
• (1) Integral of square error (ISE)
∞ 2
• Where 𝐼𝑆𝐸 = ‫׬‬0 𝜖 𝑡 𝑑𝑡 (1)

• (2) Integral of the absolute value of the error (IAE)



• Where 𝐼𝐴𝐸 = ‫׬‬0 𝜖(𝑡) 𝑑𝑡 (2)

• (3) Integral of time-weighted absolute error (ITAE)



• Where 𝐼𝑇𝐴𝐸 = ‫׬‬0 𝑡 𝜖(𝑡) 𝑑𝑡 (3)

189
Time Integral Performance Criteria
• Note that ∈ 𝑡 = 𝑦𝑠𝑝 𝑡 − 𝑦 𝑡 is the deviation or error of the
response from the desired set-point.

• Knowing which of the 3 criteria to use depends on characteristics of


the system we want to control

• (1) To strongly suppress large errors ISE is better than IAE because
the errors are squared and thus contribute more to the value of the
integral

190
Time Integral Performance Criteria
• (2) For suppression of small errors IAE is better than ISE because
when we square small numbers (smaller than one) they become even
smaller.

• (3) To suppress errors that persist for long times, ITAE tunes the
controller better, presence of large time, t amplifies effects of even
small errors in the value of integral.

191
Time Integral Performance Criteria
• Closed-loop response using various time integral criteria

192
Time Integral Performance Criteria
• Example
• Consider the following feedback system

193
Time Integral Performance Criteria
• Closed loop response given as
𝜏𝐼 𝑠+1 𝜏𝐼 Τ20𝐾𝑐 𝑠
• 𝑦(𝑠)
ത = 𝜏𝐼 1 𝑦ത𝑆𝑃 𝑠 + 𝜏𝐼 2 1 𝑑ҧ 𝑠 (4)
𝑠2 +𝜏𝐼 1+ 𝑠+1 𝑠 +𝜏𝐼 1+ 𝑠+1
20𝐾𝐶 20𝐾𝑐 20𝐾𝑐 20𝐾𝑐

• Which can be rewritten as


𝜏𝐼 𝑠+1 𝜏𝐼 Τ20𝐾𝑐 𝑠 ҧ
• 𝑦ത 𝑠 = 𝑦ത 𝑠 + 𝑑 𝑠 (5)
𝜏2 𝑠2 +2𝜁𝜏𝑠+1 𝑠𝑝 𝜏2 𝑠 2 +2𝜁𝜏𝑠+1
• Where
𝜏𝐼 1 𝜏𝐼
•𝜏= (6) 𝜁= 1 + 20𝐾𝑐 (7)
20𝐾𝑐 2 20𝐾𝑐

194
Time Integral Performance Criteria
• Assuming ISE (Integral of square error) criterion
∞ 2
• 𝐼𝑆𝐸 = ‫׬‬0 𝜖 𝑡 𝑑𝑡 (1)
• and
• Unit step change in step-point,
• from Equation 5
𝜏𝐼 𝑠+1 𝜏𝐼 Τ20𝐾𝑐 𝑠 ҧ
• 𝑦ത 𝑠 = 2 2 𝑦ത𝑠𝑝 𝑠 + 𝑑 𝑠 (5)
𝜏 𝑠 +2𝜁𝜏𝑠+1 𝜏2 𝑠 2 +2𝜁𝜏𝑠+1

• We have
𝜏𝐼 𝑠+1 1
•𝑦 𝑠 =
𝜏2 𝑠2 +2𝜁𝜏𝑠+1 𝑠

195
Time Integral Performance Criteria
• Inverting the above equation and solving for (𝜁 < 1)
−𝜁𝑡ൗ
𝑒 𝜏 𝜏𝐼 𝑡 𝑡
• 𝑦 𝑡 =1+ 2
ቈ sin 1− 𝜁 2 − sin ቆ 1 − 𝜁 2 +
1−𝜁 𝜏 𝜏 𝜏
1−𝜁 2
tan−1 ቇ቉ (8)
𝜁

∞ 2
• Solving the optimization problem; Minimize 𝐼𝑆𝐸 = ‫׬‬0 𝑦𝑠𝑝 − 𝑦 𝑡

196
Time Integral Performance Criteria
• By selecting values of τ and 𝜁, y (t) is given by Equation (8) above.
• For optimization the following equations (conditions) hold
𝜕 𝐼𝑆𝐸 𝜕 𝐼𝑆𝐸
• = =0
𝜕𝑡 𝜕𝜁

• Let τ* and 𝜁 * be the optimal values.


• Determine the corresponding optimal values of 𝐾𝑐 and 𝜏𝐼 using
𝜏𝐼 1 𝜏𝐼
Equations 6 and 7 𝜏 = (6) 𝜁 = 1 + 20𝐾𝑐 (7)
20𝐾𝑐 2 20𝐾𝑐

197
Time Integral Performance Criteria
• Otherwise using ITAE criterion instead we have to minimize ITAE, i.e.

• Minimize 𝐼𝑇𝐴𝐸 = ‫׬‬0 𝑡 𝑦𝑠𝑝 − 𝑦(𝑡) 𝑑𝑡

• Where y (t) is given by Equation 8.



𝜕(𝐼𝑇𝐴𝐸) 𝜕 𝐼𝑇𝐴𝐸
• The solution τ* and 𝜁 * is given by = =0
𝜕𝜏 𝜕𝜁
• Leading to calculating optimal 𝐾𝑐 and 𝜏𝐼 from Equations (6) and (7)
above as before

198
Time Integral Performance Criteria
• Considering unit step change in Load. Equation (5) yields;
𝜏𝐼 Τ20𝐾𝑐 𝑠 1
•𝑦 𝑠 =
𝜏2 𝑠2 +2𝜁𝜏𝑠+1 𝑠

• After inversion
−𝜁𝑡ൗ
𝜏𝐼 Τ20𝐾𝑐 𝑒 𝜏 𝑡
•𝑦 𝑡 = sin 1− 𝜁 2 (9)
𝜏 1−𝜁 2 𝜁
• We can further determine τ* and 𝜁 * as before by minimizing the
equation.

199
Time Integral Performance Criteria
• Comparing Equations (8) and (9)
−𝜁𝑡ൗ
𝑒 𝜏 𝜏𝐼 𝑡 𝑡
• 𝑦 𝑡 =1+ 2
ቈ sin 1− 𝜁 2 − sin ቆ 1 − 𝜁 2 +
1−𝜁 𝜏 𝜏 𝜏
1−𝜁 2
tan−1 ቇ቉ (8) (ISE)
𝜁

−𝜁𝑡ൗ
𝜏𝐼 Τ20𝐾𝑐 𝑒 𝜏 𝑡
•𝑦 𝑡 = sin 1− 𝜁 2 (9) (ITAE)
𝜏 1−𝜁 2 𝜁
• Therefore expect the optimal 𝐾𝑐 and 𝜏𝐼 to be different

200
Selecting the Type of Feedback Controller
• On deciding which of the 3 popular feedback controllers to use, we
can go through the following steps:

1) Define an appropriate performance criterion (e.g. ISE, IAE or ITAE)

2) Compute the value of the performance criterion using a P or PI or


PID controller with the best settings for the adjusted
parameters 𝐾𝑐 , 𝜏𝐼 and 𝜏𝑝 .

201
Selecting the Type of Feedback Controller
3) Select that controller which gives the “best” value for the
performance criterion.

• This procedure is mathematically rigorous, but tedious and has


serious practical drawbacks.

202
Summary of Use of the Three Types of Feedback
Controllers
• Following are the characteristics of the three types of controllers
1) Proportional Control
a) Accelerates response of a controlled system
b) Produces an offset (i.e. non-zero steady-state error) for all processes except
those with terms 1Τ𝑠 (integrators) in their transfer functions such as;
i. Liquid level in tank
ii. Gas pressure in vessel

203
Summary of Use of the Three Types of Feedback
Controllers
2) Integral Control
a) Eliminates any offset
b) Elimination of offset is at expense of the maximum deviations
c) Produces sluggish, long oscillating responses
d) If we increase gain 𝐾𝑐 to produce faster response, the system becomes
more oscillatory and may lead to instability.
3) Derivative Control
a) Anticipates future errors and introduces appropriate action
b) Introduces stabilizing effect on the closed-loop response of a process
• PID controller gives the best response; it comes with attendant complexity in
tuning when one has to adjust three parameters

204
Summary of Use of the Three Types of Feedback
Controllers
• To balance quality of desired response against tuning difficulty
• Adopt the following rules in selecting most appropriate controller.

1) If possible, use simple proportional controller


• Proportional controller can be used if
a) We can achieve acceptable offset with moderate values of 𝐾𝑐
b) The process has an integrating action (i.e. a term 1Τ𝑠 in its transfer function)
for which the P control does not exhibit offset. Therefore for gas pressure or
liquid-level control, we can use only P controller.

205
Summary of Use of the Three Types of
Feedback Controllers
2) If a simple P controller is not acceptable, use a PI.
• PI is to be used when P control alone cannot provide sufficiently
steady-state errors (offset).
• PI is therefore seldom used for liquid-level or gas pressure control
system but rather (almost always) on flow control.
• Response of a flow system is rather fast. Speed of closed-loop systems
remains satisfactory despite slowdown caused by integral control
mode.

206
Summary of Use of the Three Types of
Feedback Controllers
3) Use PID to increase speed of closed-loop response and retain
robustness
• PI eliminates offset but reduces the speed of the closed-loop
response.
• For multi-capacity process with sluggish response, PI will make it even
more sluggish.
• Addition of derivative control action allows use of higher gains which
produces faster response without excessive oscillations.

207
Summary of Use of the Three Types of
Feedback Controllers
• Derivative action is recommended for
• (1) temperature and
• (2) composition control

• Where we have sluggish, multi-capacity processes.

208
CONTROLLER TUNING
• Process of selecting the appropriate adjustable parameters
1) Use simple criteria like the;
a) One-quarter decay ratio, minimum
b) Minimum setting time
c) Minimum largest error etc.

2) Use time integral performance criteria such as ISE, IAE, or ITAE


which is cumbersome and time-consuming
3) Use semi-empirical rules which have been proven in practice.

209
Empirical Tuning Method by Cohen and Coon
• This is known as the Process Reaction Curve Method.
• Consider the “opened” control system below;

210
Empirical Tuning Method by Cohen and Coon
• Introduce a step change of magnitude A
• Record the value of the output with respect to time
• The curve generated 𝑦𝑚 (𝑡) is called the Process Reaction Curve.

• Between 𝑦𝑚 and c, we have the following transfer function.


𝑦𝑚 𝑠
• 𝐺𝑃𝑅𝐶 𝑠 = = 𝐺𝑓 𝑠 𝐺𝑝 𝑠 𝐺𝑚 𝑠 10
𝑐 𝑠
• process reaction is affected by the dynamics of the main process,
measuring sensors and final control element.

211
Empirical Tuning Method by Cohen and Coon
• Cohen and Coon observed that the response of most processing units
to an input change has a sigmoidal shape
• (a) Process reaction curve; (b) its approximation with a first-order
plus dead-time system

212
Empirical Tuning Method by Cohen and Coon
• Sigmoidal shape can be adequately approximated by a response of a
1st order system with dead time (dashed curve in Figure (b).)

𝑦𝑚 (𝑠) 𝐾 𝑒 −𝑡𝑑 𝑠
• 𝐺𝑃𝑅𝐶 = ≅ (11)
𝑐 𝑠 𝜏𝑠+1
• With 3 parameters:
• Static gain, K
• Dead time, 𝑡𝑑
• Time constant, 𝜏

213
Empirical Tuning Method by Cohen and Coon
• From the figure (b), we can estimate the values of the 3 parameters,
thus
𝑂𝑢𝑡𝑝𝑢𝑡 𝑎𝑡 𝑠𝑡𝑒𝑎𝑑𝑦 𝑠𝑡𝑎𝑡𝑒 𝐵
• 𝐾= =
𝐼𝑛𝑝𝑢𝑡 𝑎𝑡 𝑠𝑡𝑒𝑎𝑑𝑦 𝑠𝑡𝑎𝑡𝑒 𝐴
𝐵
• 𝜏=
𝑆
• where S is slope of the sigmoidal response at point of inflection
• 𝑡𝑑 = time elapsed until system responded

214
Empirical Tuning Method by Cohen and Coon
• Cohen and Coon used the approximate model of Equation (11) and
estimated the values of the parameters K, 𝑡𝑑 , 𝜏 as indicated.
𝑦𝑚 (𝑠) 𝐾 𝑒 −𝑡𝑑 𝑠
• 𝐺𝑃𝑅𝐶 = ≅ (11)
𝑐 𝑠 𝜏𝑠+1

• They then derived expressions for the “best” controller settings using
load changes and various performance criteria such as
a) One-quarter decay ratio
b) Minimum offset
c) Minimum Integral Square Error (ISE)

215
Empirical Tuning Method by Cohen and Coon
• Results of their analysis are summarized thus
1) For proportional controllers use
1 𝜏 𝑡𝑑
• 𝐾𝑐 = 1+ (12)
𝐾 𝑡𝑑 3𝜏

2) For Proportional Integral (PI) controller


1 𝜏 𝑡𝑑
• 𝐾𝑐 = 0.9 + (13 𝑎)
𝐾 𝑡𝑑 12𝜏
30+3𝑡𝑑 Τ𝜏
• 𝜏𝐼 = 𝑡𝑑 𝑡 (13 𝑏)
9+20 𝑑ൗ𝜏

216
Empirical Tuning Method by Cohen and Coon
3) For PID controller
1 𝜏 4 𝑡𝑑
• 𝐾𝑐 = + 14 𝑎
𝐾 𝑡𝑑 3 4𝜏

32+6𝑡𝑑 Τ𝜏
• 𝜏𝐼 = 𝑡𝑑 (14 𝑏)
13+8𝑡𝑑 Τ𝜏

4
• 𝜏𝐷 = 𝑡𝑑 (14 𝑐)
11+2𝑡𝑑 Τ𝜏

217
Empirical Tuning Method by Cohen and Coon
• Remarks
1) Controller setting given by Equations (12), (13) and (14) are based
on the assumption that the 1st order plus dead time system is a
good approximation.

2) It is noticed that almost all physical processes encountered in a


chemical plant are simple 1st Order or multi-capacity processes
whose response has the general overdamped shape.

218
Empirical Tuning Method by Cohen and Coon
1) From Equations (12), (13 a) and (14 a) giving the value of the
proportional gain 𝐾𝑐 , we notice that
a) Gain of the PI controller is lower than that of P controller. This is because
the integral control mode makes the system more sensitive and thus the
gain value needs to be more conservative.

b) Stabilizing effect of derivative control allows use of higher gains in PID


controller (higher than the gain for P or PI controllers).

219
•C’est Fini

220

You might also like