MEC522 CONTROL ENGINEERING

2. SYSTEM MODEL REPRESENTATION

2.1 TRANSFER FUNCTION

Transfer function is the classical way of modelling linear systems to show input-output relations between
variables. Transfer function is defined as the ratio of the Laplace transform of the output and the input
by assuming that the initial conditions are all zero.

!(#) ! & # " '&# # "$# '⋯'&"

%(#)
= #% ') %$# '⋯')
(2.1)
!# %$# #')%

Consider a closed loop control system shown below;

R(s) E(s) Y(s)

+
G(s)
-
H(s)
3
The transfer function of the above block diagram is given as

!(#) *(#)
= (2.2)
%(#) +'*(#),(#)

This transfer function can also be written as

Y (s)
k åA s
k =0
k
k

= n
𝑛>𝑚 (2.3)
U (s)
s åA s
k =0
k
k

The classifications of the transfer functions are based on

a. Order: the highest power of s in denominator, n
b. Type: the power factor s in the denominator, l
c. Rank: the difference between the highest power of s denominator, n and the highest power of
s in the numerator, m, 𝑛 − 𝑚 ≥ 1

M.H.M.R
15
MEC522 CONTROL ENGINEERING

2.1.1 Characteristic Equation

The characteristic equation of a linear system (Linear Time Invariant LTI) is defined as the equation
obtained by setting the denominator polynomial of the transfer function to zero. Thus, the characteristic
equation of the system describe above in (2.1) is

𝑠 ! + 𝑎" 𝑠 !#$ + ⋯ + 𝑎!#$ 𝑠 + 𝑎! = 0 (2.4)

Or simply;

1 + 𝐺(𝑠)𝐻(𝑠) = 0 (2.5)

This characteristic equation characterizes and describes the behaviour of the system, while the roots of
the characteristic equation give information regarding the system’s stability.

2.1.2 Steps and Procedure in Determining Transfer Function?

a. Write a set of linearly independent integral-differential equation as shown in Eqn. (2.6)

&!' & !"# ' &' &$- & $"# -

&( !
+ 𝑎) &( !"# + ⋯ + 𝑎*+, &( + 𝑎* 𝑦 = 𝑏) &( $ + 𝑏, &( $"# + ⋯ + 𝑏. 𝑢 (2.6)

b. Convert the equations in frequency domain by taking Laplace transform assuming initial
conditions as zero.

&!' & !"# ' &' &$- & $"# -

ℒ 0 &( ! + 𝑎) &( !"# + ⋯ + 𝑎*+, &( + 𝑎* 𝑦1 = ℒ 0𝑏) &( $ + 𝑏, &( $"# + ⋯ + 𝑏. 𝑢1 (2.7)

c. Solve the resulting set of n linearly independent algebraic equations in the n dependent
variables for the Laplace transform of the output in terms of the Laplace transform of the
input.

(𝑠 * + 𝑎) 𝑠 *+, + ⋯ + 𝑎*+, 𝑠 + 𝑎* )𝑌(𝑠) = (𝑏) 𝑠 . + 𝑏, 𝑠 .+, + ⋯ + 𝑏. )𝑈(𝑠) (2.8)

d. Find the relation between Laplace transform of output and input eliminating intermediate
variables. The resultant equation is the transfer function.

!(#) !& # " '&# # "$# '⋯'&"

= #% ') %$# '⋯')
(2.9)
%(#) !# %$# #')%

M.H.M.R
16
MEC522 CONTROL ENGINEERING

2.1.2.1 Worked Example 1

Given the following differential equation with y(s) being the output variable and u(s) is the input variable
respectively, determine the corresponding transfer function.

&%' &' &-

+ 12 &( + 32𝑦 = 7 &( + 32𝑢(𝑡) (2.10)
&( %

Solution:

Take Laplace transform both sides of the equation and we obtain

&%' &' &-
ℒ 0 &( % + 12 &( + 32𝑦1 = ℒ 07 &( + 32𝑢(𝑡)1 (2.11)

(𝑠 / + 12𝑠 + 32)𝑌(𝑠) = (7𝑠 + 32)𝑈(𝑠) (2.12)

Now, take ratio between output 𝑌(𝑠) and input 𝑈(𝑠) of Eqn. (2.12)

!(#) &#'()
= (2.13)
%(#) # ! '*)#'()

2.1.3 Advantages of Transfer Function

Transfer function is a very powerful tool for analysing a control system and offers the following
advantages.
i. Provide the overall connection between a system’s components in terms of mathematical
relationship. Thus, it allows us to analyze the individual component of the system.
ii. Stability analysis of the system can be easily carried out using the concept of characteristic
equation. We get information about poles, zeros, and etc.
iii. The use of Laplace transform approach facilitate the conversion of integral-differential time
domain equation to simple algebraic equations.
iv. The Transfer function is expressed in terms of complex variable ‘s’, which is not a function
of the real variable time or any other variable, where it is used as the independent variable.
v. Transfer function is the property and unique equation of the system and its value depends
on the value of parameters (constants) of the system and is independent of the input.

M.H.M.R
17
MEC522 CONTROL ENGINEERING

2.2 STATE SPACE REPRESENTATION

The trend of modern engineering systems is towards greater complexity and may have many inputs and
many output (MIMO) which targeted to achieve high precision and accuracy. To analyse such system,
it is essential to reduce the complexity of the mathematical expression, as well as to employ computers
for most of the tedious computations necessary in the analysis. The state-space approach is developed
to meet and satisfy the need of current trend and provides a best method for analysing complex control
systems.

2.2.1 GENERAL CONCEPTS

The state-space method generally derived from the input-output relationship where it is quite similar to
previous approach but represented in vector-matrix form which greatly simplifies the mathematical
representation of systems equations. The increase in the number of state variables, the number of inputs,
or the number of outputs does not increase the complexity of the equations.

Before any analysis can be carried out, we need to understand some basic terminologies, such state
variables, state vector, and state-space

State variables: the state of a dynamic system is the smallest set variables which describes the
behaviour of the system when inputs are known.

State vector: if n state variables are needed to completely describe the behaviour of a given
system, then these n state variables can be considered as the n components of a
vector x(t). Such a vector is called a state vector and it determines uniquely the
system state x(t) for any given time t and inputs.

State space: the n dimensional space whose coordinate axes consist of the 𝑥, axis, 𝑥/ axis,
..., 𝑥* axis, where 𝑥, , 𝑥/ , …., 𝑥* are state variables is called state space.

2.2.2 STATE SPACE GENERAL REPRESENTATION

For any particular control system, the state equation can be described as;
𝑥̇ = [𝐴]𝑥 + [𝐵]𝑢 (2.14)

And the output equation is;

𝑦 = [𝐶]𝑥 + [𝐷]𝑢 (2.15)

M.H.M.R
18
MEC522 CONTROL ENGINEERING

Where A is state matrix, B is input matrix, C is output matrix and D is direct transmission matrix.

State matrix describes the behaviour of the control systems, i.e.; system stability, response conditions,
controllability, observability, and so forth. In matrix form, stability of a particular system can be derived
from state matrix eigenvalues at which it represents the characteristic equation and thus, provide the
location of the roots for that system.

Meanwhile, the relationship between state matrix and input matrix gives rise to system controllability
factor as they relate the ability of system’s actuator. A system is controllable if it is possible to transfer
any state with any set of initial conditions to any final state in some finite time period. Alternatively, a
system is only controllable if every mode (or state) is connected to the control input. A system is referred
to as “stabilizable” so long as we can state control all unstable modes. This might mean that there are
some stable uncontrollable states.

2.2.2.1 Worked Example 1

Consider the following equation of motion;

𝑚 0 𝑥̈ , 𝑐,, −𝑐,/ 𝑥̇ , 𝑘,, −𝑘,/ 𝑥, 0

@ , B @ B + D−𝑐 𝑐// G @𝑥̇ / B + @−𝑘/, BD G = @ B (2.16)
0 𝑚/ 𝑥̈ / /, 𝑘// 𝑥/ 𝑓(𝑡)

Where 𝑐,, = 𝑐, + 𝑐/
𝑘,, = 𝑘, + 𝑘/
𝑐,/ = 𝑐/, = 𝑐/
𝑘,/ = 𝑘/, = 𝑘/

Determine the State Space Representation of the above equation.

M.H.M.R
19
MEC522 CONTROL ENGINEERING

Solution:

The State Space Representation of the above equation can be determined using the following process;

i. Define the state parameters, namely;

State variables Derivative of State variables

𝑧! = 𝑥! 𝑧̇! = 𝑥̇ ! = 𝑧"
1
𝑧" = 𝑥̇ ! 𝑧̇" = 𝑥̈ ! = [𝑘 (𝑧 − 𝑧! ) + 𝑐" (𝑧$ − 𝑧" ) − 𝑐! 𝑧" − 𝑘! 𝑧! ]
𝑚! " #
𝑧# = 𝑥" 𝑧̇# = 𝑥̇ " = 𝑧$
1 1
𝑧$ = 𝑥̇ " 𝑧̇$ = 𝑥̈ " − [𝑘" (𝑧# − 𝑧! ) + 𝑐" (𝑧$ − 𝑧" )] + 𝑓(𝑡)
𝑚" 𝑚"

ii. Rearrange the expression according to state and output equations;

State equation 𝑧̇ = [𝐴]𝑧 + [𝐵]𝑢

0 1 0 0
𝑧̇ 1 ⎡ (𝑘1+𝑘2) (𝑐1 +𝑐2 ) 𝑘2 𝑐2 ⎤ 𝑧1 0
𝑧̇ ⎢− 𝑚 −
𝑚1 𝑚1 𝑚1 ⎥ 𝑧2
0
K 2L = ⎢ 0 1 1 ⎥ 3
S𝑧 T + K 0 L 𝑓(𝑡) (2.17)
𝑧̇ 3 0 0 1
𝑧̇ 4
⎢ 𝑘2 𝑐2 𝑘
−
𝑐2 ⎥ 𝑧
⎣ 𝑚2
− 2 𝑚2 ⎦
4 𝑚2
𝑚2 𝑚2

Where

𝑧̇ 1 0 1 0 0
⎡ (/#0/$) (2# 02$ ) /$ 2$ ⎤ 𝑧! 0
𝑧̇ ⎢− 1# − 1 1# 1# ⎥ 𝑧" 0
𝑧̇ = K 2 L, 𝐴=⎢ #
, 𝑧 = D𝑧# E, 𝐵 = G 0 H and 𝑢 = 𝑓(𝑡)
𝑧̇ 3 0 0 0 1 ⎥ !
⎢ /$ 2$ /$
−1
2$ ⎥ 𝑧$
𝑧̇ 4 ⎣ 1$ 1$
− 1$ $⎦
1$

Output equation is obtained as 𝑦 = [𝐶]𝑧 + [𝐷]𝑢

𝑧!
1 0 0 0 𝑧"
𝑦=L M D E, (2.18)
0 0 1 0 𝑧#
𝑧$

𝑧!
1 0 0 0 𝑧"
where 𝐶=L M, 𝑧 = D𝑧# E𝐷 = 0
0 0 1 0
𝑧$

M.H.M.R
20
MEC522 CONTROL ENGINEERING

2.2.3 Conversion Between State Space Representation to Transfer Function

Consider state equation in Eqn. (2.14) and Output Eqn. in (2.15), i.e.,

𝑥̇ = [𝐴]𝑥 + [𝐵]𝑢
And the output is
𝑦 = [𝐶]𝑥 + [𝐷]𝑢

The Transfer function representation could be obtained by taking Laplace transform by

assuming all initial conditions are zero and with some matrix (algebraic operations) shown in
the following steps

i. Taking Laplace transform of both above equations yield

𝑠𝑋(𝑠) = [𝐴]𝑋(𝑠) + [𝐵]𝑈(𝑠)

(2.19)
𝑌(𝑠) = [𝐶]𝑋(𝑠) + [𝐷]𝑈(𝑠)

ii. Expressing state equation in terms of output equation, we obtain

𝑋(𝑠) = (𝑠 − [𝐴])+, [𝐵]𝑈(𝑠) (2.20)

where (𝑠 − [𝐴]) is known as transition matrix ∅

iii. Subtituting Eqn. (2.20) into Eqn. (2.19) we have

𝑌(𝑠) = [𝐶](𝑠 − [𝐴])−1 [𝐵]𝑈(𝑠) + [𝐷]𝑈(𝑠) (2.21)

iv. Now, the transfer function is obtained by taking ratio between 𝑌(𝑠) upon 𝑈(𝑠)

−1
𝑌(𝑠) = U[𝐶]∅ [𝐵] + [𝐷]V 𝑈(𝑠)

!(#) −1
𝑈(𝑠)
= ([𝐶]∅ [𝐵] + [𝐷]- (2.22)

M.H.M.R
21
MEC522 CONTROL ENGINEERING

2.2.3.1 Worked Example 1

Consider a mechanical system in which the state space representation is given as in Eqn. (2.14) and
(2.15) with matrices A, B, C, and D are as follow,
−2 0 1
𝐴=D G, 𝐵 = D G, 𝐶 = [1 1], and 𝐷 = 0
𝐾 1 0
Determine the transfer function for this system.

Solution:

i. The transfer function can be easily obtained using Eqn. (2.22), i.e.,

𝑌(𝑠)
= ([𝐶]∅+, [𝐵] + [𝐷])
𝑈(𝑠)
ii. First step is to solve for the transition matrix

𝑠 0 −2 0 𝑠+2 0
∅ = (𝑠 − [𝐴]) = D G−D G=D G (2.23)
0 𝑠 𝐾 1 −𝐾 𝑠 − 1
iii. The inversion of the transition matrix is obtained as

𝑠+2 0 +, , 𝑠−1 0
∅+, = D G = (34/)(3+,) D G (2.24)
−𝐾 𝑠 − 1 𝐾 𝑠+2

iv. Thus, the transfer function is

6(3) , 𝑠−1 0 1
= X[1 1] 0 D G1 D G + 0Y (2.25a)
7(3) (34/)(3+,) 𝐾 𝑠+2 0

6(3) , 𝑠−1 0 1
= X[1 1] 0 D G1 D GY (2.25b)
7(3) (34/)(3+,) 𝐾 𝑠+2 0

6(3) , 𝑠−1
= X[1 1] 0(34/)(3+,) D G1Y (2.25c)
7(3) 𝐾

!(#) W'(#X+)
= (#'Y)(#X+) (2.26)
%(#)

M.H.M.R
22
MEC522 CONTROL ENGINEERING

2.2.4 Conversion Between Transfer Function to State Space Representation

2.2.4.1 Worked Example 1

Consider transfer function of a process in Eqn. (2.27) and find its state-space representation

!(#) YZ
%(#)
= #8 '[#9 'Y\#'YZ (2.27)

Solution:

i. The first step is to convert the transfer function in Eqn. (2.27) into ODE form.
(𝑠 + + 9𝑠 , + 26𝑠 + 24)𝑌(𝑠) = 24𝑈(𝑠) (2.28)

Now, take inverse Laplace transform to obtain;

-3. -4. -.
-/ 3
+ 9 -/ 4 + 26 -/ + 24𝑦 = 𝑦⃛ + 9𝑦̈ + 26𝑦̇ + 24𝑦 = 24𝑢(𝑡) (2.28)

ii. Next, define the state parameters, namely;

State variables Derivative of State variables

𝑧, = 𝑦 𝑧̇, = 𝑦̇ = 𝑧/
𝑧/ = 𝑦̇ 𝑧̇/ = 𝑦̈ = 𝑧:
𝑧: = 𝑦̈ 𝑧̇: = 𝑦⃛ = −9𝑧: − 26𝑧/ − 24𝑧, + 24𝑢(𝑡)

iii. Rearrange the expression according to state and output equations;

State equation 𝑥̇ = [𝐴]𝑥 + [𝐵]𝑢

𝑧̇, 0 1 0 𝑧, 0
𝑧̇
^ /_ = ^ 0 0 1 _ ^𝑧/ _ + ^ 0 _ 𝑢(𝑡) (2.29)
𝑧̇: −24 −26 −9 𝑧: 24

Output equation is obtained as 𝑦 = [𝐶]𝑥 + [𝐷]𝑢

Obviously, the system output is 𝑦 or simply 𝑦 = 𝑧, , so that
𝑧,
𝑦 = [1 0 0] ^ / _ 𝑧 (2.30)
𝑧:

M.H.M.R
23
MEC522 CONTROL ENGINEERING

2.2.4.2 Worked Example 2

Consider transfer function of a process in Eqn. (2.31) and find its state-space representation

!(#) # 9 ']#'Y
= #8 '[#9 'Y\#'YZ (2.31)
%(#)

Solution:

i. The first step is to separate the upper and lower polynomial into cascaded form

U(s) 1
X1(s) Y(s)
𝑠 " + 7𝑠 + 2
𝑠 & + 9𝑠 ' + 26𝑠 + 24

ii. Convert the transfer function into ODE form (Taking inverse LT with I.C zero)
- 3 05 - 4 05 -05
-/ 3
+9 -/ 4
+ 26 -/
+ 24𝑥$ = 𝑢(𝑡) (2.32)

- 4 05 -05
𝑦(𝑡) = -/ 4
+7 -/
+ 2𝑥$ (2.33)

iii. Next, find the state equations for block containing denominator, i.e., from Eqn. (2.32)
a. define the state parameters, namely;

State variables Derivative of State variables

𝑧, = 𝑥1 𝑧̇, = 𝑥̇ 1 = 𝑧/
𝑧/ = 𝑥̇ 1 𝑧̇/ = 𝑥̈ 1 = 𝑧:
𝑧: = 𝑥̈ 1 𝑧̇: = 𝑥⃛, = −9𝑧: − 26𝑧/ − 24𝑧, + 𝑢(𝑡)

b. Rearrange the expression according to state and output equations;

State equation 𝑥̇ = [𝐴]𝑥 + [𝐵]𝑢

𝑧̇, 0 1 0 𝑧, 0
^𝑧̇/ _ = ^ 0 0 1 _ ^𝑧/ _ + ^0_ 𝑢(𝑡) (2.34)
𝑧̇: −24 −26 −9 𝑧: 1

M.H.M.R
24
MEC522 CONTROL ENGINEERING

Output equation is obtained as 𝑦 = [𝐶]𝑥 + [𝐷]𝑢

But, from the cascaded form, we obtain the output as

𝑌(𝑠) = (𝑠 / + 7𝑠 + 2)𝑋, (𝑠) (2.35)

And the time domain form is obtained as in Eqn. (2.33). But, the state variables were defined as

𝑧, = 𝑥1
𝑧/ = 𝑥̇ 1
𝑧: = 𝑥̈ 1

Thus, the system output is 𝑦 or simply 𝑦 = 𝑏/ 𝑧: + 𝑏, 𝑧/ + 𝑏) 𝑧, , so that

𝑧, 𝑧,
𝑦 = [𝑏) 𝑏, 𝑏/ ] ^𝑧/ _ = [2 7 1] ^𝑧/ _ (2.36)
𝑧: 𝑧:

M.H.M.R
25
MEC522 CONTROL ENGINEERING

2.3 BLOCK DIAGRAMS MANIPULATION

A control system can be represented in block diagrams to show the relationships that exist among its
various components and their functions. It can be used together with the transfer function to describe
the cause-effect relationships throughout the system. In general, block diagrams can be used to model
linear as well as nonlinear systems.

Transfer Function
Input, r(t) G(s)
Output, c(t)

Figure 2.1: Element of a block diagram

2.3.1 Simple Rules of Block Diagram Algebra

i. Transfer operator in series.

Consider a system with two blocks in series

α1 α2 α3
G1 G2
3

The simplified block diagram is

α1 α3
G1 G2

Mathematically,
a 2 = G1a1 (2.37)

a 3 = G2a 2 (2.38)
Substituting (2.37) into (2.38) gives rise to;
a 3 = G2 (G1a1 ) (2.39a)

a3
= G1G2 (2.39b)
a1

ii. Transfer operator in parallel

Consider a system with two blocks in parallel

M.H.M.R
26
MEC522 CONTROL ENGINEERING

α1 α2 α4
G1(s) ++

α3
G2(s)
3

The simplified block diagram is

α1 α4
G1(s)+G2(s)

Mathematically,
a 2 = G1a1 (2.40)

a 3 = G2a1 (2.41)

a 4 = a 2 + a3 (2.42)

Substituting (2.40) and (2.41) into (2.42) gives rise to;

a 4 = G1a1 + G2a1 (2.43a)

a4
= G1 ( s ) + G2 ( s ) (2.43b)
a1

iii. Transfer operator in feedback

Consider a system with two blocks in feedback

α1 α4
α2
+- G(s)
α3
H(s)
3

The simplified block diagram is

α1 𝐺(𝑠) α4
1 + 𝐺 (𝑠)𝐻(𝑠)

M.H.M.R
27
MEC522 CONTROL ENGINEERING

Mathematically,
a 4 = Ga 2 (2.44)

a 3 = Ha 4 (2.45)

The error signal is

a 2 = a1 - a 3 (2.46)

Substituting (2.46) and (2.45) into (2.44) gives rise to;

a 4 = G(a1 - a 3 ) (2.47a)

a 4 = Ga1 - Ga 3 (2.47b)

a 4 = Ga1 - GHa 4 (2.47c)

(1 + GH )a 4 = Ga1 (2.47d)

a4 G(s)
= (2.47e)
a1 1 + G ( s ) H ( s )

M.H.M.R
28
MEC522 CONTROL ENGINEERING

iv. Important rules for block diagram reduction

Table 2.1 depicts block diagram representations and their equivalents such that efficient block
diagram manipulations could be achieved.

Table 2.1: Some Block diagram representations and their equivalents.

AG x1
A AG-B A AG-B
G(s) +- +- G(s)
B
1/G(s) B
3

A AG A AG
G(s) G(s)

AG AG
G(s)
3

A AG A AG
G(s) G(s)

1 A 1/G(s) A

A B A
X1 X1 B
+- G1(s) 1/G2(s) +- G1(s) G2(s)
3

G2(s)
3

M.H.M.R
29
MEC522 CONTROL ENGINEERING

2.3.1.1 Worked Example 1

Consider a mechanical system shown in Fig. 2.2. Using the rules of block diagram manipulation,
obtain the transfer function of the system.

R(s) 1 1 X(s)
+
𝑚 - 𝑠"

𝑘
𝑚

Figure 2.2: A Block diagram representation of a simple mechanical system.

Solution:

i. The first step is to identify the condition of operators, i.e., in series, parallel, cascaded, and
etc.

R(s) 1 1 X(s)
+
𝑚 - 𝑠"

𝑘 operators in feedback

ii. Solve the identified condition of operators

1
R(s) 1 𝑠" X(s)
𝑚 1 𝑘
1+ "𝑚
𝑠

1 1
R(s) ( ) X(s)
𝑠" 𝑚
1 𝑘
1+ "𝑚
𝑠

M.H.M.R
30
MEC522 CONTROL ENGINEERING

R(s) X(s)
𝐺(𝑠)

iii. The transfer function is obtained by solving for 𝐺(𝑠)

1 1
( ) 1
𝑠2 𝑚
𝐺(𝑠) = =
1 𝑘 𝑚𝑠2 + 𝑘
1+ 2
𝑠 𝑚
Thus, the transfer function is
𝑋(𝑠) 1
= 𝐺(𝑠) = 2
𝑅(𝑠) 𝑚𝑠 + 𝑘

2.3.1.2 Worked Example 2

Consider a control system shown in Fig. 2.3. Using the rules of block diagram manipulation, obtain
the transfer function of the system.

R(s) X(s)
+ 𝐺1 𝐺2 ++ 𝐺# 𝐺$
-
l
𝐻!

𝐻2

Figure 2.3: A Block diagram representation of a control system.

Solution:

i. The first step is to identify the condition of operators, i.e., in series, parallel, cascaded, and
etc.

M.H.M.R
31
MEC522 CONTROL ENGINEERING

R(s) X(s)
+ 𝐺1 𝐺2 ++ 𝐺# 𝐺$
-
l
𝐻! operators in feedback

operators in series
𝐻2

ii. Solve the identified condition of operators

R(s) 𝐺# 𝐺$ X(s)
+- 𝐺1 𝐺/ 1 − 𝐻! 𝐺# 𝐺$

operators in feedback
operators in series

𝐻2

R(s) 𝐺! 𝐺" 𝐺# 𝐺$ X(s)

+- 1 − 𝐻!𝐺# 𝐺$

𝐻2

𝐺! 𝐺" 𝐺# 𝐺$
R(s) 1 − 𝐻!𝐺# 𝐺$ X(s)
𝐺! 𝐺" 𝐺# 𝐺$
1 + (𝐻") 1 − 𝐻! 𝐺# 𝐺$

M.H.M.R
32
MEC522 CONTROL ENGINEERING

R(s) 𝐺(𝑠) X(s)

iii. The transfer function is obtained by solving for 𝐺(𝑠)

𝐺! 𝐺" 𝐺# 𝐺$
1 − 𝐻! 𝐺# 𝐺$ 𝐺! 𝐺" 𝐺# 𝐺$
𝐺(𝑠) = =
𝐺! 𝐺" 𝐺# 𝐺$ 1 − 𝐻! # $ + (𝐻" )𝐺! 𝐺" 𝐺# 𝐺$
𝐺 𝐺
1 + (𝐻" ) 1 − 𝐻 𝐺 𝐺
! # $

Thus, the transfer function is

𝑋(𝑠) 𝐺! 𝐺" 𝐺# 𝐺$
= 𝐺(𝑠) =
𝑅(𝑠) 1 − 𝐻! 𝐺# 𝐺$ + (𝐻" )𝐺! 𝐺" 𝐺# 𝐺$

2.3.1.3 Worked Example 3

Consider a mechanical system shown in Fig. 2.4. Using the rules of block diagram manipulation,
obtain the transfer function of the system.

R(s) 1 1 1 X(s)
𝑚
+- 𝑠
+-
𝑠

𝑐
𝑚
𝑘
𝑚

Figure 2.4: A Block diagram representation of a mechanical system.

Solution:

This problem is quite similar to the first Worked Example, first is to solve for operators in feedback in
the inner-loop, then solve the operators in feedback in outer-loop. The final reduction is

𝑋(𝑠) 1
= 𝐺(𝑠) = 2
𝑅(𝑠) 𝑚𝑠 + 𝑐𝑠 + 𝑘

M.H.M.R
33
MEC522 CONTROL ENGINEERING

2.3.2 Conversion between Transfer Function and Block Diagram

Consider an equation of motion of a mechanical system in (2.48).

&%; &;
𝑚 &( % + 𝑐 &( + 𝑘𝑥 = 𝑟(𝑡) (2.48)

Rearrange Eqn. (2.48) into configuration form, we obtain

&%; < &; = ,

&( %
= − . &( − . 𝑥 + . 𝑟(𝑡) (2.49)

Taking Laplace transform (with zero initial conditions) of (2.49), give rise to

< = ,
𝑠 / 𝑋(𝑠) = − . 𝑠𝑋(𝑠) − . 𝑋(𝑠) + . 𝑅(𝑠) (2.50)

a. Case 1: 𝑐 = 𝑘 = 0

,
𝑋(𝑠) = 𝑅(𝑠) (2.51)
.3 %

, ,
It is obvious that 𝑋(𝑠) depends on ., 3% , and input 𝑅(𝑠). Block diagram representation of

Eqn. (2.51) could be established as shown in the following figure.

R(s) 1 1 X(s)
𝑚 𝑠"

b. Case 2: 𝑐 = 0, 𝑘 > 0

= ,
𝑠 / 𝑋(𝑠) = − 𝑋(𝑠) + 𝑅(𝑠) (2.52)
. .

= ,
X1 + .3% Y 𝑋(𝑠) = .3% 𝑅(𝑠) (2.53)

# #
> ?
(% $
𝑋(𝑠) = # )
𝑅(𝑠) (2.54)
@,4 % > ?A
( $

M.H.M.R
34
MEC522 CONTROL ENGINEERING

,
It can be seen that Eqn. (2.54) has a configuration of operators in feedback with 𝐺(𝑠) = 3%
=
and 𝐻(𝑠) = ..

#
B(3)> ?
$
𝑋(𝑠) = ,4B(3)C(3) 𝑅(𝑠) (2.55)

B(3) ,
𝑋(𝑠) = ,4B(3)C(3) X.Y 𝑅(𝑠) (2.56)

Thus, it is clear that Case 2 is a block diagram representation of operators in feedback and
in series as shown in the following figure.

R(s) 1 𝐺(𝑠) X(s)

𝑚 1 + 𝐺(𝑠)𝐻(𝑠)

1
R(s) 1 𝑠/ X(s)
𝑚 1 𝑘
c1 + / X𝑚Yd
𝑠

Referring to rules of block diagram manipulation, the above block could be expanded to the
following figure

R(s) 1 1 X(s)
𝑚
+- 𝑠"

𝑘
𝑚

c. Case 3: 𝑐 > 0, 𝑘 > 0

< = ,
𝑠 / 𝑋(𝑠) = − . 𝑠𝑋(𝑠) − . 𝑋(𝑠) + . 𝑅(𝑠) (2.57)

< = ,
𝑋(𝑠) = − .3% 𝑠𝑋(𝑠) − .3% 𝑋(𝑠) + .3% 𝑅(𝑠) (2.58)

M.H.M.R
35
MEC522 CONTROL ENGINEERING

< = ,
X1 + .3% 𝑠 + .3%Y 𝑋(𝑠) = .3% 𝑅(𝑠) (2.59)

#
(% ,
𝑋(𝑠) = # ) # * X.Y 𝑅(𝑠) (2.60)
,4 % > ?4 % 3> ?
( $ ( $

Observing the pattern of Eqn. (2.60), it suggests that, the block diagram is a combination of operators
= <
in feedback (two layers, one with . and the other with .), and operators in series. The resultant block

diagram is as shown in the following diagram

R(s) 1 1 1 X(s)
𝑚
+- 𝑠
+-
𝑠

𝑐
𝑚
𝑘
𝑚

M.H.M.R
36
MEC522 CONTROL ENGINEERING

2.4 SIGNAL FLOW GRAPHS

A signal flow graph/diagram (SFG) may be regarded as a simplified version of a block diagram. The
common application of signal flow graph is in system diagramming. This method was developed by S.
J. Mason. This method does not require any reduction process to obtain the transfer function of a
particular system.

This representation graphically shows the relationship between variables of a set of linear algebraic
written in the form of cause-and-effect relations. It consists of network in which nodes representing each
of the system variables are connected by directed branches which have associated branch gains. A signal
can be transmitted through a branch only in the direction of the arrow as illustrated in Fig. 2.5.

R(s) 1 1 X(s)
𝑚
+- 𝑠"

𝑘
𝑚

gain

1 1
1 𝑚 𝑠2 1
R(s) X(s)

nodes
𝑘 loop
−
𝑚

Figure 2.5: A Block diagram representation of a simple mechanical system and its corresponding Signal Flow
Diagram.

M.H.M.R
37
MEC522 CONTROL ENGINEERING

2.4.1 SFG Terminology

gain

1 1
1 𝑚 𝑠2 1
R(s) X(s)
n1 n2 n3 n4 n5

nodes
𝑘 loop
−
𝑚

Figure 2.6: A signal flow diagram of a mechanical (mass-spring) system.

Some Definitions
i. Node: node is a point to represent a variable (n1, n2, n3, n4, n5)
, , =
ii. Transfer: A gain between two nodes (1, .
, 3%
, 1, − .)

iii. Entry node: this node is an independent variable, i.e., n1

iv. Exit node: this node represents dependent variable, i.e., n5
v. Forward path: a network with connected nodes in straight path, e.g., n1→n2→n3→n4→n5
vi. Loop: Loop is closed path, e.g., n3→n4→n3
vii. Non-touching loop: loops are not touching when nodes on one loop are not shared with
another loop (e.g. as illustrated in Fig. 2.7)

𝑎 𝑏 𝑐 𝑑 𝑒
R(s) X(s)
n1 n2 n3 n4 n5 n6

𝑔 𝑓

Figure 2.7: A signal flow diagram of a control system with two non-touching loops

The terms used to represent a particular signal flow graph are as below;
i. Path Gain, Pk: Value from node n-1 to node n of a particular forward path
ii. Loop Gain, Lk: Value in all path gain
iii. Determinant, ∆: 1 − ∑=D, 𝐿= + ∑=D, 𝐿= 𝐿E + ∑=D, 𝐿= 𝐿E 𝐿F + ⋯, or simply 1 − sum of
individual loop gain, 𝐿= + sum of product of two non-touching loop gains, 𝐿= 𝐿E + sum of
product of three non-touching loop gains, 𝐿= 𝐿E 𝐿F + ⋯ +

M.H.M.R
38
MEC522 CONTROL ENGINEERING

iv. Individual determinant, ∆= : The difference between 1 and sum of individual loop gain that
is not touching with current kth forward path
,
v. The transfer function, is known as M (Mason’s Rule): 𝑀 = ∆ ∑= 𝑃= ∆= = 𝐺(𝑠)

2.4.2 Signal Flow Graph and Block Diagram

Table 2.2 illustrates a couple of differences between signal flow graph and block diagram.

Table 2.2: Some Block diagram representations and their equivalent in signal flow graph.

R 𝐺 C
R C
G

R x C R 𝐺, x 𝐺/ C
G1 G2

R1
R1 𝐺,
G1
x C
+ G3 x 𝐺: C
R2 +
G2
𝐺/
R2

R E C
+ G1 R 1 E 𝐺, C
-
G2 −𝐺/

L L
R E C
+ G1 ++ G2 R 1 E 𝐺, 𝐺/ C
-
H −𝐻

M.H.M.R
39
MEC522 CONTROL ENGINEERING

2.4.2.1 Worked Example 1

Consider a mechanical system shown in Fig. 2.5. Using the rules of signal flow graph, obtain the
transfer function of the system.

Solution:

i. The first step is to identify how many forward paths and loops. From the Fig., only one
forward path is present and there are five (5) nodes on the forward path and only one loop
starting from n3→n4→n3.

R(s) 1 1 X(s)
+
𝑚 - 𝑠"

𝑘
𝑚

Forward path

1 1
1 𝑚 𝑠2 1
R(s) X(s)
n1 n2 n3 n4 n5

𝑘 loop
−
𝑚

ii. Now, let’s find out the gain for each connecting node in the forward path and the loop.
a. Pk: start from node n1 and ends at node n5, i.e., n1→n2→n3→n4→n5. Between these
connecting nodes, we need to multiply the gains, i.e., from n1 to n2; the gain is 1, next
, ,
from n2 to n3; the gain is ., from n3 to n4; the gain is 3% , and lastly from n4 to n5; the
, ,
gain is 1. So, the product of all gains on this forward path is 𝑃, = 1 × . × 3% × 1.

b. Lk: start from node n3 and ends at node n3, i.e., n3→n4→n3. Between these connecting
,
nodes, we need to multiply the gains, i.e., from n3 to n4; the gain is 3% , next from n4 to
= , =
n3; the gain is − .. So, the product of all gains on this loop is 𝐿, = 3% × − ..

M.H.M.R
40
MEC522 CONTROL ENGINEERING

c. ∆= : Since there is only one loop the SFG and is touching with kth forward path, thus
∆= = ∆, = 1 − 0

=
d. ∆: ∆= 1 − 𝐿, = 1 + .3%

iii. The transfer function G(s) is obtained by solving the following equation

1 1 1 1 1
𝑀= q 𝑃= ∆= = (𝑃, ∆, ) = r s=
∆ 𝑘 𝑘 𝑚𝑠 / 𝑚𝑠 /+𝑘
= 1+ 1+
𝑚𝑠 / 𝑚𝑠 /

1
𝐺(𝑠) =
𝑚𝑠 / +𝑘

2.4.2.2 Worked Example 2

Consider a mechanical system shown in Fig. 2.8. Using the rules of signal flow graph, obtain the
transfer function of the system.
Hint: you are required to transform the block diagram into the signal flow graph to solve for the
transfer function, use Table 2.2 to assist the transformation process.

𝐻"

R C
+- 𝐺1 𝐺2 ++ 𝐺# +
-
𝐺$
l

𝐻1

Figure 2.8: A Block diagram representation of a process in a control system.

Solution:

i. The first step is to establish the equivalent signal flow graph for the given problem, then
identify how many forward paths and loops. From the following figure, there are two
forward paths;

M.H.M.R
41
MEC522 CONTROL ENGINEERING

𝑃, = 𝑛1 → 𝑛2 → 𝑛3 → 𝑛4 → 𝑛5 → 𝑛6 → 𝑛7

𝑃/ = 𝑛1 → 𝑛2 → 𝑛3 → 𝑛5 → 𝑛6 → 𝑛7

and three loops; 𝐿, = 𝑛4 → 𝑛5 → 𝑛6 → 𝑛4, 𝐿/ = 𝑛2 → 𝑛3 → 𝑛4 → 𝑛5 → 𝑛6 → 𝑛2, and

𝐿: = 𝑛2 → 𝑛3 → 𝑛5 → 𝑛6 → 𝑛2

𝐻"

R 1 𝐺1 𝐺2 𝐺# 𝐺$ 1 C
n1 n2 n3 n4 n5 n6 n7
f f hc

−1

−𝐻1

ii. Now, let’s find out the gain for each connecting node in the forward path and the loop.
a. Pk: both forward paths start from node n1 and ends at node n7, i.e.,

𝑃, = 𝑛1 → 𝑛2 → 𝑛3 → 𝑛4 → 𝑛5 → 𝑛6 → 𝑛7 = 𝐺, 𝐺/ 𝐺: 𝐺H

𝑃/ = 𝑛1 → 𝑛2 → 𝑛3 → 𝑛5 → 𝑛6 → 𝑛7 = −𝐺, 𝐺H

b. Lk: the first one starts from node n4 and ends at node n4, i.e.,
𝐿, = 𝑛4 → 𝑛5 → 𝑛6 → 𝑛4 = 𝐺: 𝐺H 𝐻/
The second one starts from n2 and ends at n2, i.e.,
𝐿/ = 𝑛2 → 𝑛3 → 𝑛4 → 𝑛5 → 𝑛6 → 𝑛2 = −𝐺, 𝐺/ 𝐺: 𝐺H 𝐻,
The third starts also from n2 and ends at n2, i.e.,
𝐿: = 𝑛2 → 𝑛3 → 𝑛5 → 𝑛6 → 𝑛2 = 𝐺, 𝐺H 𝐻,
c. ∆= : Since all the loops are touching each other at namely node 𝑛6, thus ∆= = ∆, =
∆/ = 1 − 0

d. ∆: ∆= 1 − ∑ 𝐿= = 1 − (𝐿, + 𝐿/ + 𝐿: )

M.H.M.R
42
MEC522 CONTROL ENGINEERING

∆= 1 − 𝐺: 𝐺H 𝐻/ + 𝐺, 𝐺/ 𝐺: 𝐺H 𝐻, − 𝐺, 𝐺H 𝐻,

iii. The transfer function G(s) is obtained by solving the following equation

1 (𝑃, ∆, + 𝑃/ ∆/ )
𝑀= q 𝑃= ∆= =
∆ 1 − 𝐺: 𝐺H 𝐻/ + 𝐺, 𝐺/ 𝐺: 𝐺H 𝐻, − 𝐺, 𝐺H 𝐻,
=

𝐺, 𝐺/ 𝐺: 𝐺H − 𝐺, 𝐺H
𝑀=
1 − 𝐺: 𝐺H 𝐻/ + 𝐺, 𝐺/ 𝐺: 𝐺H 𝐻, − 𝐺, 𝐺H 𝐻,

𝐺, 𝐺/ 𝐺: 𝐺H − 𝐺, 𝐺H
𝐺(𝑠) =
1 − 𝐺: 𝐺H 𝐻/ + 𝐺, 𝐺/ 𝐺: 𝐺H 𝐻, − 𝐺, 𝐺H 𝐻,

2.4.3.1 Unsolved Problems

Consider block diagrams shown in Fig. Q(a1) to Fig. Q(a3). Using the rules of signal flow graph, obtain
the transfer function for each system.
Hint: you are required to transform the block diagram into the signal flow graph to solve for the transfer
function, use Table 2.2 to assist the transformation process.

G3

R + C
+ G1 + + G2 + G4
- - -
3 3

H2

H3

Figure Q(a1)

M.H.M.R
43
MEC522 CONTROL ENGINEERING

H3
R
3
C
+ -
G1 + G2 + G3
- -

H1

H2

Figure Q(a2)

H2
3
R - C
+ + G1 + G2 G3
- -

H1

H3

Figure Q(a3)

M.H.M.R
44