10.

Dynamic Behavior of
Closed-Loop Control Systems
• Feedback control loop( or closed-loop system) ; the combination
of the process, the feedback controller, and the instruments.

10.1 Block Diagram Representation

Figure 10.1. Temperature control system or stirred-tank heater with

steam heating.( ----- electrical signals; pneumatic signal)
10.1.1 Process

Ti
T
Process
Ps

Figure 10.2. Inputs and output of the process.

• Dynamic model of a steam-heated, stirred tank:

dT
mC  wC (Ti  T )  h p Ap (Tw  T ) (10.1)
dt
dTw
mw C w  hs As (Ts  Tw )  h p Ap (Tw  T ) (10.2)
dt
Where the subscripts w, s, and p refer respectively to the wall of
the heating coil, and to its steam and process sides.
Assume that the dynamics of the heating coil are negligible since
the dynamics are fast compared to the dynamics of the tank
contents.  Left side of (10.2) equal to zero!
0  hs As (Ts  Tw )  h p Ap (Tw  T ) (10.3)
hs AsTs  h p ApT
 Tw  (10.4)
hs As  h p Ap
Substituting (10.4) into (10.1) gives
dT
mC  wC (Ti  T )  U A (Ts  T ) (10.5)
dt
Where U A is defined by
h p Ap hs As
UA  (10.6)
hs As  h p Ap
Assume that Ts  a . bPs  (10.5)
dT
mC  wC (Ti  T )  U A (a  bPs  T ) (10.7)
dt
Apply Laplace transform after introducing deviation variables.
mCsT ' ( s )  wC[Ti ' ( s )  T ' ( s )]  U A [bPs ' ( s )  T ' ( s )] (10.8)
K1 K2
 T ' ( s)  Ti ' ( s)  Ps ' ( s ) (10.9)
s  1 s  1
mC wC U Ab
where   , K1  and K 2  .
wC  U A wC  U A wC  U A
T ' (s) K1
 for Ps ' ( s )  0 (10.10)
Ti ' ( s ) s  1
T ' (s) K2
 for Ti ' ( s )  0 (10.11)
Ps ' ( s ) s  1

Figure 10.3. Block diagram of the process.

10.1.2 Thermocouple and Transmitter
Assume that the dynamic behavior of the thermocouple and
transmitter can be approximated by a first-order transfer function:
Tm ' ( s ) Km
 (10.12)
T ' (s)  m s  1

Figure 10.4. Block diagram for the thermocouple

and temperature transmitter.

• If  m    negligible dynamics.
• Steady-state gain K m depends on the input and output ranges of
the thermocouple-transmitter combination.
10.1.3 Controller
Assume that a proportional plus integral controller is used

P' (s)  1 
 K c 1   (10.13)
E (s)  Is 
where P' ( s ) and E (s ) are the Laplace transforms of the controller
output p ' (t ) and error signal e(t ). Note that p' (t ) and e(t ) are electrical
signals which have the units of [mA] while K c is dimensionless.

Figure 10.5. Block diagram for the controller.

 10.1.4 Current-to-Pressure(I/P) Transducer
Since transducers are usually designed to have linear characteristics
and negligible(fast) dynamics  Assume that the transducer transfer
function merely consists of a steady-state gain : K IP
Pt ' ( s )
 K IP (10.14)
P' ( s)
10.1.5 Control Valve
The flow through the valve is a non-linear function of the signal to
the valve actuator. However, a first-order transfer function usually
provides an adequate model for operation of an installed valve in
the vicinity of a nominal steady-state.
Ps ' ( s ) Kv
 (10.15)
Pt ' ( s )  v s  1

Figure 10.6. Block diagram for Figure 10.7. Block diagram for
the I/P transducer. the control valve.
 Complete block diagram

Figure 10.8. Block diagram for the entire control system.

10.2 Closed-Loop Transfer Functions
 The standard notations.
C = Controlled variable
M = Manipulated variable
L = Load variable
P = Controller output
E = Error signal
B = Measured value of C
R = Set point
~
R = Internal set point(used by controller)
X 1 = Change in Cdue to L
X 2 = Change in C due to M
Gc = Set point
Gv = Transfer function for final control element
G p = Process transfer function
GL = Load transfer function
Gm = Transfer function for measuring element and transmitter
K m = Steady-state gain for Gm
 Figure 10.9. Standard block diagram of a feedback control system

• In Figure 10.9. each variable is the Laplace Transform of a deviation variable.

– Feedforward path: the signal path from to through blocks ,
and . E C Gc
– Feedback path: the signal path from to the comparator through .
Gv Gp
C Gm
 Figure 10.10. Alternative form of the standard block diagram
of a feedback control system

• For these two block diagrams to be equivalent, the relation between

Cand Lmust be preserved. Thus, and
G L GLmust
* be related by the
expression.

GL  GL*G p (10.16)
 10.2.1 Block Diagram Reduction
 It is often convenient to combine several blocks into a single
block.
• Example

Figure 10.11. Three blocks in series.

X1  G 1 U, X 2  G 2 X1 , X 3  G 3 X 2 (10.17)

By successive substitution,
X 3  G 3G 2 G1 U  GU (10.18)

Figure 10.12. Equivalent block diagram.

10.2.2 Set Point Changes( = Servo Problem)

Figure 10.9. Standard block diagram of a feedback control system

C  X1  X 2 (10.19)
X 1  GL L  0 (since L  0) (10.20)
X 2  Gp M (10.21)
 C  G p M  G~p Gv P  G p Gv Gc E
 G p Gv Gc ( R  B )  G p Gv Gc ( K m R  Gm C ) (10.22)
~ ~
[M  Gv P, P  Gc E , E  R  B, R  K m R and B  GmC.]
 Desired closed-loop transfer function,
C K m Gc Gv G p
 (10.23)
R 1  Gc Gv G p Gm
10.2.2 Load Changes( = Regulatory Problem)

C  X 1  X 2  GL L  G p M  GL L  G p Gv Gc Gm C (10.24)
 Desired closed-loop transfer function,
C GL
 (10.25)
L 1  Gc Gv G p Gm

 A comparison of (10.23) and (10.25) indicate that both closed-lo

op transfer functions have the same denominator.
The denominator is written as 1  GOL
where GisOLthe open-loop
transfer function, 1  GOL  1  Gc Gv.G p Gm

GL K m Gc Gv G p
 C L R (10.26)
1  GOL 1  GOL
 10.3 Closed-loop Responses of Simple Control Systems
In this section, we consider the dynamic behavior of several
elementary control problems for load variable and set-point changes.

• For the liquid-level control system

Figure 10.10. Liquid-level Control System

 q1: the load variable.
q2: the manipulated variable.
Assumption:
1. The liquid density  and the cross-sectional area of the tank A are
constant.
2. The flow-head relation is linear, q3 = h/R.
3. The transmitter and control valves have negligible dynamics.
4. Pneumatic instruments are used.
• Mass balance for the tank contents.
dh
A  q1  q2  q3 (10.27)
dt
• Transfer Function
H ( s ) Kp H ( s ) Kp
 G p (s)  (10.28)  GL ( s )  (10.29)
Q2 ( s ) s  1 Q1( s ) s  1

Where, KP = R, = RA
 Assuming that the dynamics of the level transmitter and control
valve, the corresponding transfer functions can be written as Gm
(s) = Km and Gv(s) = Kv .

Block diagram for level control system

Figure 10.11. Block diagram for level control system

 • Proportional Control and Set-Point Change
If a proportional controller is used, then Gc(s) = Kc .
The closed-loop transfer function for set-point changes is given by
H ( s) K c K v K p K m /(s  1) K1
  (10.30)
H R ( s ) 1  K c K v K p K m /(s  1)  1s  1

K OL 
where, K1  (10.31) 1  (10.32)
1  K OL 1  K OL

KOL is the open-loop gain, KOL = Kc Kv Kp Km (KOL > 0 for stability → chapte
r 11)

Thus since 1 < , one consequence of feedback control is that it enables t
he controlled process to respond more quickly than the uncontrolled proce
ss.  The reason for the introduction of feedback control
The closed-loop response to a unit step change of magnitude M in set point is
given by

h(t )  K1M (1  exp(t /  1 )) (10.33)

Figure 10.12. Step response for proportional control (set-point change)

The offset ( steady-state error) is defined as

offset  hR ()  h()

M
 M  K1M  M (1  K1 )  (10.34)
1  K OL

Since KOL = Kc Kv Kp Km

Kc : , KOL : , offset : 

if Kc  , offset  0
• Proportional Control and Load Changes
The closed-loop transfer function for load changes is given by
H ( s ) K p /(s  1) K2
  (10.35)
Q1( s ) 1  K c K v K p K m /(s  1)  1s  1

Kp
where, K 2  (10.36)
1  K OL

The closed-loop response to step change of magnitude M in load

h(t )  K 2 M (1  exp(t /  1 )) (10.37)
hR (t )  0
K pM
offset  0  h()   K 2 M  
1  K OL
The same situation can be observed as set point change case
• PI Control and Load Changes
 1 

Gc ( s )  K c 1  
 Is 

The closed-loop transfer function for load change is given by

H ( s ) K p /(s  1) K p I s
 
Q1( s )  1   I s (s  1)  K OL ( I s  1)
1  K OL 1   /(s  1)
 Is 
This transfer function can be rearranged as a second-order one.
H ( s ) K ps
 2 2 (10.38)
Q1( s )  3 s  2 3 3 s  1

where, K3 = I / KcKvKm  3   I / K OL

1  1  K OL 
3   I /

2  K OL  
For a unit step change in load,

h(t ) 
K3
3 1   2

exp( 3t /  3 ) sin 1   32 t /  3  (10.39)
3

h()  0 because of the exponential term in (10.39).

For set-point change, offset will be zero too!

Addition of Integral action

 Elimination of offset for step changes of load and set-point
• PI Control of an Integrating Process

Figure 10.13. Liquid-level control system with pump in exit line

This system differs from the previous example in two ways

1. The exit line contains a pump
2. The manipulated variable is the exit flow rate rather than
an inlet flow rate.
The process and load transfer functions are given by
H ( s ) 1
G p ( s)   (10.40)
Q3 ( s ) As

H ( s ) 1
GL ( s )   (10.41)
Q1( s ) As

The closed-loop transfer function for load changes

H ( s ) GL K4s
  2 2 (10.42)
Q1( s ) 1  GcGvG pGm  4 s  2 4 4 s  1

where, K4 = -I / KcKvKm  4   I / K OL  4  0.5 K OL I

KOL = KcKvKpKm , Kp = - 1/A

• The effect of I

I:    4  0.5 K OL I : 

 closed-loop responses: less oscillatory

• The effect of Kc
Kc:   4:   closed-loop responses: less oscillatory

• The effect of Kc for the stable process except the integrating

process
Kc:   closed-loop responses: more oscillatory