Complex Variables/Functions

Laplace Transformation
MCE 309 - System Dynamics and Control

Dr. Tolga Özaslan

Mechanical Engineering Department

Yıldırım Beyazıt University

Fall 2024

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 1 / 117

 Mathematical Modeling

Mathematical Model of a Dynamic System

Dynamic systems can be described using differential equations. While

there is no limit to the complexity of these equations, we prefer simpler
ones for easier analysis. For this we often make assumptions.
In its most general form, the dynamics equations of a system may look

like

f (c, ċ, c̈, ..., c (n) ) = g (r , r˙, r¨, ..., r (m) )

where c(t) is the output and r (t) is the input, c (m) is the mth time
derivative of c(t), f (...) and g (...) are some functions of their
corresponding input arguments.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 2 / 117

 Mathematical Modeling

Mathematical Model of a Dynamic System

A dynamic model may look like

r (t)
α(t) c 2 (t) =
r¨(t)

where α(t) is a time-varying coefficient.

This model is non-linear and time-varying hence very difficult analyze.

We usually prefer making assumptions to simply such equations so that

they become linear and time-invariant.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 3 / 117

 Mathematical Modeling

Linear Time-Invariant Differential Equation

Many systems can be approximately described by linear, time-invariant

differential equations of the form

d m c(t) d m−1 c(t)

+a m−1 + ... + a0 c(t) =
dt m dt m−1
d n r (t) d n−1 r (t)
bn + b n−1 + ... + b0 r (t)
dt n dt n−1
where c(t) is the output and r (t) is the input, a’s and b’s are the constant
coefficients of the system.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 4 / 117

 Mathematical Modeling

Linear Time-Invariant Differential Equation

After simplifications, the order of system may reduce. This reduces the
complexity of the model and gives more insight to the designer.

The system dynamics of a spring-mass-damper system reduces to

mẍ(t) + c ẋ(t) + kx(t) = F (t) (1)

assuming that the mass of the spring is negligible, dashpot behaves linearly
and the mass is rigid.
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 5 / 117
 Mathematical Modeling

Transfer Functions

Using transfer functions is another way of describing the dynamics of

linear, time-invariant systems. A transfer function is just a different
representation of the governing differential equation in the frequency
domain.

We use Laplace Transform to derive the transfer function of a given linear,

time-invariant different equation.

Transfer functions can be used only with linear systems which is a limiting
factor of its use cases. However they yield for insight about the system
compared to a differential equations.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 6 / 117

 Mathematical Modeling

Transfer Functions
The input, output functions and their derivatives appear throughout the
equation making it difficult for the designer to develop an insight.

d m c(t) d m−1 c(t)

+a m−1 + ... + a0 c(t) =
dt m dt m−1
d n r (t) d n−1 r (t)
bn + b n−1 + ... + b0 r (t)
dt n dt n−1
The block diagram of the system maybe as simple as

raising questions about whether there could be simpler mathematical

representations rather than the long differential equations.
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 7 / 117
 Mathematical Modeling

Transfer Functions

In addition to ease of representation, we would like to be able to

compound effects of cascaded subsystems easily as well.

Output of a block is the input to the next making the differential

equations more crowded and difficult to decipher to the designer.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 8 / 117

 Mathematical Modeling

Transfer Functions

A transfer function relates the input and output as

C (s) = G (s)R(s)

where all functions are in the s-domain, C (s) is the output, R(s) is the
input and G (s) is the system dynamics.
The input-output relation of a cascaded system can easily be obtained as

C (s) = G1 (s)G2 (s)...Gn (s)R(s)

where each Gi (s) is a subsystem.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 9 / 117

 Complex Numbers and Functions

Complex Variables

A complex number, s ∈ C, has a real and an imaginary part both of which

are constant.
√
s = σ + jω where σ, ω ∈ R, and j = −1

Real and imaginary components of s are σ and ω respectively.

Re(s) = σ and Im(s) = ω are shorthand notations to retrieve each

component of s.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 10 / 117

 Complex Numbers and Functions

Complex Functions
A complex function, G (s) : C → C, is a function of s ∈ C which has a real
and an imaginary part.

G (s) = Gx (s) + jGy (s) where Gx , Gy : C → R.

Magnitude and angle of G (s) are defined respectively as

q
|G (s)| = Gx2 + Gy2
∠G (s) = atan2(Gy , Gx ).

Complex conjugate of G (s) is

Ḡ (s) = Gx − jGy

Complex functions may be single or multiple valued. In the linear control

context, these functions are single-valued.
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 11 / 117
 Complex Numbers and Functions

Visualizing Complex Functions

Since G (s) is a mapping from C to C, visualizing it is difficult. Both s and
G (s) have 2 real dimensions, hence the plot of (s, G (s)) would be 4
dimensional!

We instead plot (s, |G (s)|) and (s, ∠G (s)) separately. Assigning colors to
the values |G (s)| and ∠G (s) might save space and generate nice plots.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 12 / 117

 Complex Numbers and Functions

Using SymPy to Plot Complex Functions

from sympy import arg, I, symbols

from sympy.plotting import plot3d_parametric_surface, PlotGrid

u, v = symbols('u v') # Real and imaginary coordinates

uRange = (u, -.25, .25) # Ranges of u and v values
vRange = (v, -.25, .25)

# Plot magnitude and angle of 1 / (s + 1)

h1 = plot3d_parametric_surface(u, v, abs(1 / (u + I * v)), uRange, vRange,
title=r"$\left|\frac{1}{s+1}\right|$",
xlabel=r"$\sigma$ (real)", ylabel=r"$\omega$ (imag)",
show=False);

h2 = plot3d_parametric_surface(u, v, arg(1 / (u + I * v)), uRange, vRange,

title=r"$\angle\left(\frac{1}{s+1}\right)$",
xlabel=r"$\sigma$ (real)", ylabel=r"$\omega$ (imag)",
show=False);

PlotGrid(1, 2, h1, h2);

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 13 / 117

 Complex Numbers and Functions

Analytic Functions
A complex function G (s) is analytic in a particular region iff all of its
derivatives exist in that region.

dG (s) G (s + ∆s) − G (s)

= lim
ds ∆s→0 ∆s
∆G (s)
= lim
∆s→0 ∆s

There are infinitely-many ∆s’s,

i.e. directions of differentiation. This
can be written as

∆s = ∆σ + j∆ω

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 14 / 117

 Complex Numbers and Functions

Analytic Functions

We do not need to check the value of dGds(s) along all directions.

As long as derivatives along the two major directions are the same, we call
G (s) to be an analytic function (proof is skipped).

∆s = ∆σ
∆s = j∆ω

For ∆s = ∆σ, the path of differentiation is parallel to the real axis, and
parallel to the imaginary axis for ∆s = j∆ω.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 15 / 117

 Complex Numbers and Functions

Analytic Functions

Derivative along the real axis:

 
dG (s) ∆Gx ∆Gy
= lim +j
ds ∆σ→0 ∆σ ∆σ
∂Gx ∂Gy
= +j
∂σ ∂σ
Derivative along the imaginary axis:
 
dG (s) ∆Gx ∆Gy
= lim +j
ds j∆ω→0 j∆ω j∆ω
∂Gx ∂Gy
= −j +
dω dω

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 16 / 117

 Complex Numbers and Functions

Cauchy-Riemann Conditions

The sufficient condition to call a complex function as analytic is the

equality of its derivatives along the real and imaginary axes:
∂Gx ∂Gy ∂Gx ∂Gy
+j = −j +
∂σ ∂σ dω dω
or equivalently
∂Gx ∂Gy ∂Gy ∂Gx
= and =− .
∂σ dω ∂σ dω
These are known as the Cauchy-Riemann conditions.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 17 / 117

 Complex Numbers and Functions

Example

1
Consider G (s) = s+1 . After substitution s → (σ + jω) we get
1
G (s) = σ+1+jω .

σ+1 −ω
Gx = and Gy =
(σ + 1)2 + ω 2 (σ + 1)2 + ω 2

Derivatives exist everywhere except for s = −1, i.e. σ = −1, ω = 0.

∂Gx ∂Gy ω 2 − (σ + 1)2

= =
∂σ ∂ω [(σ + 1)2 + ω 2 ]2
∂Gy ∂Gx 2ω(σ + 1)
=− =
∂σ ∂ω [(σ + 1)2 + ω 2 ]2

Hence G (s) is analytic along the entire complex plane except for s = −1.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 18 / 117

 Complex Numbers and Functions

Example (cont’d)

dG (s) ∂Gx ∂Gy

= +j
ds ∂σ ∂σ
ω − (σ + 1)2
2 2ω(σ + 1)
= 2
+j
2
[(σ + 1) + ω ]2 [(σ + 1)2 + ω 2 ]2
(σ + 1)2 − j2ω(σ + 1) − ω 2
=−
[(σ + 1)2 + ω 2 ]2
(σ + 1 − jw )2
=−
[(σ + 1 + jω)(σ + 1 − jω)]2
1 1
=− 2
=−
(σ + 1 + jω) (s + 1)2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 19 / 117

 Complex Numbers and Functions

Differentiation of G (s)

Derivatives of analytic complex functions, G (s), can be calculated simply

by differentiating w.r.t s.
 
dG (s) d 1 1
= =−
ds ds 1 + s (s + 1)2
s 2 +1
Example : Given G (s) = s 3 +2s 2

dG (s) 2s (s 2 + 1)(3s 2 + 4s)

= 3 −
ds s + 2s 2 (s 3 + 2s 2 )2
s 3 + 3s + 4
=− 3 2
s (s + 4s + 4)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 20 / 117

 Complex Numbers and Functions

Differentiation Using SymPy

from sympy import *

s = symbols('s')
G = (s**2 + 1) / (s**3 + 2 * s**2)

# Differentiate, then simplify to reduce clutter

res = simplify(diff(G, s))

# Use pretty printing function

# if your console supports unicode
pprint(res)

s2 + 1 s 3 + 3s + 4
 
d
=−
ds s 3 + 2s 2 s 3 (s 2 + 4s + 4)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 21 / 117

 Complex Numbers and Functions

Ordinary and Singular Points

Points in the s-plane where G (s) is analytic are ordinary points.

Points in the s-plane where G (s) is not analytic are singular points.

Singular points at which either G (s) or its derivatives goes to ±∞ are

called poles

Singular points where G (s) = 0 are called zeros

If G (s) → ±∞ as s → −p, and if G (s)(s + p)n → 0 for n ∈, Z+ , then the

order of the singular point s = −p is of order n. If n = 1, then s = −p is a
simple pole. If n = 2, then s is a 2nd order pole, 3rd order pole if n = 3,
etc.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 22 / 117

 Complex Numbers and Functions

Poles & Zeros of G (s)

Consider the complex function

K (s + 2)(s + 10)
G (s) =
s(s + 1)(s + 5)(s + 15)2

Zeros are at s = −2 and s = −10

Simple poles are at s = 0, s = −1 and s = −5

s = −15 is a 2nd order pole

K
As s → ∞, G (s) ∼ s3
→ 0. Hence s = ∞ is a triple zero

G (s) has 5 zeros and 5 poles.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 23 / 117

 Complex Numbers and Functions

Poles & Zeros of G (s) Using SymPy

from sympy import *

s = symbols('s')
G = (s+2) * (s+10) / (s) / (s+1) / (s+5) / (s+15)**2

# Get the numerator and denominator

num, denom = fraction(G)

# Solve for num = 0 and denom = 0

zeros = solve(num, s)
poles = solve(denom, s)

# Print the numerical values

print("Zeros : ", zeros)
print("Poles : ", poles)

# Factorization may give more insight

print("Factors of num. : ", factor(num))
print("Factors of denom. : ", factor(denom))

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 24 / 117

 Laplace Transform

Laplace Transformation

Define

f (t) := a function of time s.t. f (t) = 0 for t < 0

s := a complex variable
L := Laplace transformation operator
F (s) := Laplace transform of f (t)

Laplace transform of f (t) is calculated as

Z ∞
L [f (t)] = F (s) = f (t)e −st dt
0

This is very similar to Fourier transform with the additional e −σt term in
the integrand.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 25 / 117

 Laplace Transform

Inverse Laplace Transformation

The inverse operation is called the Inverse Laplace Transformation,

i.e. given a function is the s domain, we would like to find the
corresponding function in the time domain.
This is denoted as
Z c+i∞
−1 1
L [F (s)] = f (t) = F (s)e st ds for t > 0
2πj c−i∞

c is the distance of the line of integration from the imaginary axis. It is

chosen to be greater than the real parts of all singular points.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 26 / 117

 Laplace Transform

Inverse Laplace Transformation

Example : Line of integration

K
G (s) =
(s − 2)(s 2 + 2s + 2)

Poles are at s = −2, s = −1 ± j.

c > max(Re(poles)) = −1, c ∈ R.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 27 / 117

 Laplace Transform

Laplace Transform Pairs

Calculating inverse Laplace transformation might be difficult or time

taking. For this, we use partial-fraction expansion.

We also use tables for both forward and inverse Laplace transforms.

L [δ(t)] = 1 (unit impulse) t n−1

 
1
L = n
1 (n − 1)! s
L [1(t)] = (unit step)
s ω
L [sin(ωt)] =
1 s 2 + ω2
L [t] = 2 s
s L [cos(ωt)] = 2
s + ω2
See the text and reference books for a longer list of pairs.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 28 / 117

 Laplace Transform

Step Function Transform

Find the Laplace transform of the step function, f (t) = α1(t).

f (t) = α1(t)
(
0, if t < 0
=
α, if t ≥ 0

Z ∞
F (s) = αe −st 1(t)dt
0
∞
αe −st −a a
=− =0− =
s t=0 s s

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 29 / 117

 Laplace Transform

Ramp Function Transform

Find the Laplace transform of the ramp function f (t) = βt1(t)

f (t) = βt1(t) Z ∞
βe −st t1(t)dt
(
0, if t < 0 F (s) =
= 0
βt, if t ≥ 0
R R
Use integration by parts : udv = uv − vdu

Choose u = βt and dv = e −st dt. Then du = βdt and v = − 1s e −st .

1 −st ∞
  Z ∞ 
1 −st
F (s) = (βt) − e − − e (βdt)
s 0 0 s
β −st ∞
 
βt −st β
= − e − 2e = 2 valid for Re(s) > 0
s s 0 s
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 30 / 117
 Laplace Transform

Impulse Function Transform

Impulse function, a.k.a delta dirac function, is a narrow and spiky
distribution such that

(
+∞, if t = 0
δ(t) =
0, if t 6= 0

and
Z ∞
δ(t)dt = 1
−∞ xp (t) → δ(t) as τ → 0.

Z ∞ Z 0+
−st
L [δ(t)] = δ(t)e dt = δ(t)e −st dt = 1
0− 0−

Note that e −st = 1 within the range of integration.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 31 / 117
 Laplace Transform

Sinusoid Transform
The Laplace transform of the sinusoid function, i.e. f (t) = sin(ωt) for
ω∈R

Z ∞
L [sin(ωt)] = sin(ωt)e −st dt
Z0 ∞
1 jωt
e − e −jωt e −st dt


=
0 2j
1 ∞ (jω−s)t
Z
= e − e −(jω+s)t dt
2j 0
  ∞
1 1 (jω−s)t 1 −(jω+s)t
= e + e
2j (jω − s) (jω + s) 0
 
1 1 1 ω
=− + = 2
2j jω − s jω + s s + ω2

F (s) is valid in the entire complex plane except for the poles at s = ±jω.
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 32 / 117
 Laplace Transform

Exponentially Decaying Function Transform

Given f (t) = Ae −at 1(t) find F (s) = L [f (t)]

Z ∞
F (s) = f (t)e −st dt
Z0 ∞
= Ae −at 1(u)e −st dt
Z0 ∞
= Ae −(s+a)t dt
0
∞
A −(s+a)t
=− e
s +a 0
A
=
s +a

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 33 / 117

 Laplace Transform

Laplace Transform Using SymPy

from sympy import *

s = symbols('s')

# Transform of a Sinusoidal
t, w = symbols('t, w', real=True)
f = sin(t * w)
F = simplify(integrate(f * exp(-s*t), (t, 0, oo), conds='none'))
print("\nf(t) = "), pprint(f)
print("\nF(s) = "), pprint(F)

# Transform of a Decaying Function

a, A = symbols('a, A', real=True)
f = A * exp(-a*t)
F = simplify(integrate(f * exp(-s*t), (t, 0, oo), conds='none'))
print("\nf(t) = "), pprint(f)
print("\nF(s) = "), pprint(F)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 34 / 117

 Laplace Transform

Laplace Transform Using SymPy

# Laplace transforms using library functions

t, w = symbols('t, w', real=True)
f = sin(t * w)
F = laplace_transform(f, t, s)
print("\nF(s)"), pprint(F)

a, A = symbols('a, A', real=True)

f = A * exp(-a*t)
F = laplace_transform(f, t, s)
print("\nF(s)"), pprint(F)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 35 / 117

 Laplace Transform

Inverse Laplace Transform

Example : Find the inverse Laplace transform of F (s) = 1s .

Z c+i∞
−1 1 1 st
L [F (s)] = f (t) = e ds
2πj c−i∞ s
c+j∞
= Ei(st)
s=c−j∞

Ei(t) is the exponential integral function which is a non-elementary

function. i.e. it cannot be written in terms of sums, products or
compositions of finitely many polynomials, rationals, trigonometric,
hyperbolic or exponential functions.

Evaluation of Ei(t) is possible using numerical methods or can be

approximated using elementary function.
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 36 / 117
 Laplace Transform

Inverse Laplace Transform

Cauchy’s integral formula can be used to calculate such integrals and this
is a topic in complex analysis.

Some graduate level advanced engineering math courses cover this.

We use tables to find the corresponding time-domain function for a given

function in s-domain.

Ogata’s book has a list of ∼ 30 transformation pairs in Appendix-A.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 37 / 117

 Laplace Transform

Inverse Laplace Transform

Find the inverse Laplace transforms by comparing the given functions with
the transformation pairs tables (Table A-1 in Ogata).

K
Example : F (s) = s .

The 2nd pair is 1(t) ↔ 1s . Hence f (t) = K 1(t).

1
Example : F (s) = (s−a)2
.

The 7th pair is te −at ↔ 1

(s+a)2
. Hence f (t) = te at .

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 38 / 117

 Laplace Transform

Inverse Laplace Transform

s+3
Example : F (s) = (s+1)(s+2) .

There is no pair in the table that exactly matches this pattern. But we can
obtain F (s) by combining multiple pairs. Pairs #15 and #16 are (resp.)

1  −at  1
e − e −bt ↔
b−a (s + a)(s + b)
1  −bt  s
be − ae −at ↔
b−a (s + a)(s + b)

F (s) = F16 (s) + 3F15 (s). Hence f (t) = f16 (t) + 3f15 (t).

f (t) = 2e −t − e −2t for t ≥ 0

The systematic way of decomposing a relatively complex fraction into

simpler fractions is called partial fraction expansion (or decomposition).
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 39 / 117
 Properties of Laplace Transform

Superposition
Definition : The response of a linear system to the sum of two different
inputs equals the sum of the outputs of each input applied individually.
F (a + b) = F (a) + F (b) for some linear system F
Laplace transform is a linear operation, hence superposition property
applies.
Z ∞
L [αf1 (t) + βf2 (t)] = [αf1 (t) + βf2 (t)] e −st dt
0
Z ∞ Z ∞
−st
=α f1 (t)e dt + β f2 (t)e −st dt
0 0
= αF1 (s) + βF2 (s)
Scaling property follows superposition naturally
L [αf (t)] = αF (s)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 40 / 117

 Properties of Laplace Transform

Superposition
Example : Find the Laplace transform of f (t) = a + b sin(ωt) for t ≥ 0

We know Laplace transforms of 1(t) and sin(ωt) from Laplace transform

pairs table
1
L [1(t)] =
s
ω
L [sin(ωt)] =
s2 + ω2
Laplace transform of the whole expression writes as
a bω
L [a + b sin(ωt)] = + 2
s s + ω2
as 2 + bωs + aω 2
=
s 3 + ω2s

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 41 / 117

 Properties of Laplace Transform

Time Delay
The usual practice is to assume that the system is turned on at t = 0. In
certain cases this might happen at a later time t = λ > 0. Such cases can
be handled using the time delay property of Laplace transform.

The Laplace transform of f (t) with a delay of λ > 0 is (i.e. L [f (t − λ)])

Z ∞
F̄ (s) = f (t − λ)e −st dt
0
Substituting t̄ = t − λ gives
Z ∞
F̄ (s) = f (t̄)e −s(t̄+λ) d t̄
Z−λ∞
= f (t̄)e −s(t̄+λ) d t̄
0
Z ∞
−sλ
=e f (t̄)e −s t̄ d t̄ = e −sλ F (s)
0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 42 / 117

 Properties of Laplace Transform

Time Delay

Example : Find the Laplace transform of a

sinusoidal with a non-zero phase angle,
i.e. f¯(t) = A sin(ωt − θ).

Note that the shift is not in time. So we

have to rewrite f¯(t) as
  
¯ θ
f (t) = A sin ω t −
ω


= A sin ω(t − θ̄)

f¯(t) is both scaled in time by ω and delayed by θ̄.

For simplicity we assume ω = 1 to eliminate the time-scaling effect.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 43 / 117

 Properties of Laplace Transform

Time Delay
Example (cont’d) (note θ̄ = θ since ω = 1)

Z ∞
F̄ (s) = A sin(t − θ)e −st dt
0
Z ∞
=A [sin(t) cos(θ) − cos(t) sin(θ)] e −st dt
0 Z ∞ Z ∞ 
−st −st
= A cos(θ) sin(t)e dt − sin(θ) cos(t)e dt
0 0
 
1 s
= A cos(θ) 2 − sin(θ) 2
s +1 s +1
= ... how do I get from here to the next line?
A
= e −sθ 2 ← this is what the time delay property gives
s +1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 44 / 117

 Properties of Laplace Transform

Time Delay
Example (cont’d)

Laplace transform assumes f (t) = 0 for t < 0. But in the first line, we
started the integration from 0 although it should have been θ.

Z θ
F̄ (s) = A sin(t − θ)e −st dt
Z 0∞
+ A sin(t − θ)e −st dt
θ

The first term evaluates to 0.

We can either substitute t̄ = t − θ or pursue the strategy in the previous

slide.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 45 / 117

 Properties of Laplace Transform

Time Delay
Example (cont’d)

from sympy import *

s = symbols('s')

# Transform of a Sinusoidal
th, t = symbols('th, t', real=True)
f = sin(t - th)

F1 = integrate(f * exp(-s*t), (t, 0, +oo), conds='none')

print("This is the result of integration if the lower boundary is '0'")
pprint(simplify(F1))

F2 = integrate(f * exp(-s*t), (t, th, +oo), conds='none')

print("... and if the lower boundary is 'th'")
pprint(simplify(F2))

F 1 gives the same expression as in slide 44.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 46 / 117
 Properties of Laplace Transform

Time Delay
Example (cont’d) :

The output of the Python code

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 47 / 117

 Properties of Laplace Transform

Time Delay

Example (cont’d) : Let’s perform the integration with proper limits and
not using substitution t̄ = t − θ (this is what we used in the proof of
time-delay property)

Z ∞
F̄ (s) = A sin(t − θ)e −st dt
θ
Z ∞
=A [sin(t) cos(θ) − cos(t) sin(θ)] e −st dt
θ
 Z ∞  Z ∞ 
= A cos(θ) sin(t)e −st dt − sin(θ) cos(t)e −st dt
θ θ

Use integration by parts twice to calculate the integrals

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 48 / 117

 Properties of Laplace Transform

Time Delay
sin(ωt)e −st dt
R
Example (cont’d) : Consider a more general case I =

Define u = sin(ωt) and dv = e −st

   
−1 −st −1 −st
Z
I = sin(ωt) e − (ω cos(ωt)) e dt
s s
Z
1 ω
= − sin(ωt)e −st + cos(ωt)e −st dt
s s
Again define u = cos(ωt) and dv = e −st
   Z   
1 ω −1 −st −1 −st
I = − sin(ωt)e −st + cos(ωt) e − (−ω sin(ωt)) e dt
s s s s
ω2
Z
1 ω
= − sin(ωt)e −st − 2 cos(ωt)e −st − 2 sin(ωt)e −st dt
s s s

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 49 / 117

 Properties of Laplace Transform

Time Delay
Example (cont’d) : Note the recurrence of the initial integral

ω2
Z
1 −st ω −st
I = − sin(ωt)e − 2 cos(ωt)e − 2 sin(ωt)e −st dt
s s s
1 ω ω2
= − sin(ωt)e −st − 2 cos(ωt)e −st − 2 I
s s s
Solving for I gives
s sin(ωt) + ω cos(ωt) −st
I=− e
s 2 + ω2
lim I = 0
t→∞

Using a similar procedure we can find that

−s cos(ωt) + ω sin(ωt) −st
Z
cos(ωt)e −st dt = e
s 2 + ω2
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 50 / 117
 Properties of Laplace Transform

Time Delay

from sympy import *

s = symbols('s')

# Transform of a Sine Function

t, w = symbols('t, w', real=True)
f = sin(t * w)
F = simplify(integrate(f * exp(-s*t), t, conds='none'))
print("f(t) = "), pprint(f)
print("F(s) = "), pprint(F)

# Transform of a Cosine Function

f = cos(t * w)
F = simplify(integrate(f * exp(-s*t), t, conds='none'))
print("f(t) = "), pprint(f)
print("F(s) = "), pprint(F)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 51 / 117

 Properties of Laplace Transform

Time Delay

Example (cont’d) : Back to our question

Z ∞
F̄ (s) = A sin(t − θ)e −st dt
θ
 Z ∞  Z ∞ 
−st −st
= A cos(θ) sin(t)e dt − sin(θ) cos(t)e dt
θ θ
  −sθ 
e
= A cos(θ) 2 (s sin(θ) + cos(θ)) −
s +1
 −sθ 
e
sin(θ) 2 (s cos(θ) − sin(θ))
s +1
A
= e −sθ 2 this completes the proof at slide 44X
s +1
Note that both integrals evaluate to 0 as t → ∞
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 52 / 117
 Properties of Laplace Transform

Time Scaling
Change of scale might be necessary in certain cases such as converting the
units of an equation from km/h to m/s.

For a given function f (t), find the Laplace transform of f (at) in terms of
L [f (t)].

Z ∞
F̄ (s) = f (at)e −st dt
0

Substitute t̄ = at to get
Z ∞ s t̄ 1
F̄ (s) = f (t̄)e − a d t̄
0 |a|
1 s 
= F
|a| a

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 53 / 117

 Properties of Laplace Transform

Time Scaling

Example : Find the Laplace transform of a sinusoidal with a non-unitary

time scale, i.e. f¯(t) = A sin(ωt).

We know from earlier slides (or from Laplace transformation tables) that

1
L [sin(t)] =
s2 +1
Using the time-scaling property, we get
A 1
L [A sin(ωt)] =
|ω| s 2


ω +1
Aω
=
s2 + ω2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 54 / 117

 Properties of Laplace Transform

Shift in Frequency

When a function is multiplied in the time-domain with an exponentially

decaying function, the corresponding Laplace transform is shifted, i.e. a
frequency shift happens.

Suppose f¯(t) = e −at f (t). We are asked to find L f¯(t) in terms of

 

L [f (t)].

Z ∞
e −at f (t) e −st dt


F̄ (s) =
Z0 ∞
= f (t)e −(s+a)t dt
0
= F (s + a)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 55 / 117

 Properties of Laplace Transform

Shift in Frequency

Example : Find the Laplace transform of f (t) = Asin(ωt)e −at

We know from the time-scaling property that

ω
L [sin(ωt)] =
s2 + ω2
Incorporating the effect of exponentially decaying coefficient, we get

L [f (t)] = F (s + a)
Aω
=
(s + a)2 + ω 2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 56 / 117

 Properties of Laplace Transform

Differentiation
The Laplace transform of the derivative of a function is related to the
Laplace transform of the original function as
  Z ∞ 
df df
L = e −st dt
dt 0− dt
We define du = df −st and use integration by parts as follows
dt and v = e

∞
Z ∞  Z ∞
df
e −st dt = e −st f (t) +s f (t)e −st dt
0− dt 0− 0−

= −f (0− ) + sF (s)
Note that we used
lim e −st f (t) = 0
t→∞
lim e −st f (t) = (1)f (0− )
t→0−

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 57 / 117

 Properties of Laplace Transform

Differentiation

For many systems we assume that f (0) = 0 for t < 0 hence f (0− ) = 0
resulting in
 
df
L = sF (s)
dt

This can be generalized for higher-order time derivatives of f (t) as follows

 2 
d f df
L 2
= s 2 F (s) − sf (0− ) − (0− )
dt dt
..
.
 n  2
d f n (n−1) − (n−2) d f − d (n−1) f −
L = s F (s) − s f (0 ) − s (0 ) − · · · − (0 )
dt n dt 2 dt (n−1)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 58 / 117

 Properties of Laplace Transform

Differentiation
The expression for higher-order time-derivatives might look messy, however
we just recur the same rule onto the (n − 1)st derivative to get the
expression for the nth derivative
 
df
L (t) = sL [f (t)] − f (0− )
dt
 2   
d f df df
L (t) = sL (t) − (0− )
dt 2 dt dt
 3  2
d 2f −
 
d f d f
L (t) = sL (t) − (0 )
dt 3 dt 2 dt 2
..
.
 n " #
d (n−1) f d (n−1) f −

d f
L n
(t) = sL (t) − (0 )
dt dt (n−1) dt (n−1)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 59 / 117

 Properties of Laplace Transform

Differentiation
Example : The mass-spring-dashpot system’s equation of motion is

mẍ + c ẋ + kx = u(t)

Using the differentiation property of Laplace transformation we can easily

write

m s 2 X (s) − sx(0− ) − ẋ(0− ) + c sX (s) − x(0− ) + kX (s) = U(s)

   

Assuming all initial conditions are 0, this simplifies to

(ms 2 + cs + k)X (s) = U(s)

Note that rearranging the terms gives the transfer function of this system
1
X (s) = G (s)U(s) where G (s) =
ms 2 + cs + k
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 60 / 117
 Properties of Laplace Transform

Differentiation
d
Example : Find the Laplace transform of g (t) = dt cos(ωt)

From earlier slides we know that

s
L [cos(ωt)] =
s 2 + ω2
Using the differentiation property we can write
 
d s
− cos ω0−


L cos(ωt) = s 2 2
dt s +ω
s2
= 2 −1
s + ω2
ω2
=− 2
s + ω2
Note that
 
d
L cos(ωt) = −ωL [sin(ωt)] X
dt
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 61 / 117
 Properties of Laplace Transform

Differentiation
d2
Example : Find the Laplace transform of g (t) = dt 2
cos(ωt)

Using the differentiation property twice we can write

 2 
d g dg −
L (t) = s 2 L [g (t)] − sg (0− ) − (0 )
dt 2 dt
which gives
 2 
d 2 −

d −


L cos(ωt) = s L [cos(ωt)] − s cos ω0 − cos ω0
dt 2 dt
s
= s2 2 − s cos ω0− + ω sin ω0−



s +ω 2

s3
= 2 −s
s + ω2
sω 2
=− 2 note this equals − ω 2 L [cos(ωt)] X
s + ω2
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 62 / 117
 Properties of Laplace Transform

Integration
The Laplace transform of the time integral of a function is related to the
Laplace transform of the original function in a similar way to how its time
derivative relates
Rt
Given f¯(t) = f (τ )dτ , find F̄ (s) = L f¯(t)
 
0
Z ∞ Z t 
F̄ (s) = f (τ )dτ e −st dt
0 0
Rt
Using integration by parts and setting u = 0 f (τ )dτ , dv = e −st dt
 Z t  ∞ Z ∞
1 −st 1
F̄ (s) = − e f (τ )dτ − − e −st f (t)dt
s 0 0 0 s
1
= F (s)
s
Rt
Note that limt→∞ e −st = 0 and limt→0 0 f (τ )dτ = 0
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 63 / 117
 Properties of Laplace Transform

Integration
Example : The mass-spring-dashpot system’s equation of motion is
ma(t) + cv (t) + kx(t) = u(t)
where
a(t) = ẍ(t)
Z t
v (t) = a(τ )dτ
0
Z t
x(t) = v (τ )dτ
0

assuming 0 initial conditions i.e. a(0) = 0, v (0) = 0 and x(0) = 0

Taking the Laplace transform and using the superposition property we get
mA(s) + cV (s) + kX (s) = U(s)
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 64 / 117
 Properties of Laplace Transform

Integration
Using integration property this becomes
c k
mA(s) + A(s) + 2 A(s) = U(s)
s s
Rearranging the terms give
s2
A(s) = GA (s)U(s) where GA (s) =
ms 2 + cs + k
Remember that the transfer function for X (s) is
1
GX (s) = such that
ms 2
+ cs + k
X (s) = GX (s)U(s)
The integration property gives the same result, i.e.
1
X (s) = A(s)
s2
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 65 / 117
 Properties of Laplace Transform

Integration
Rt
Example : Find the Laplace transform of f (t) = 0 cos(ωτ )dτ

From earlier slides we know that

s
L [cos(ωt)] =
s 2 + ω2
Using the integration property we can write
Z t 
1 s 1
L cos(ωτ )dτ = 2 2
= 2
0 s s +ω s + ω2
Note that
Z t   t
1
L cos(ωτ )dτ = L sin(ωτ )
0 ω
 0
1 1
=L sin(ωt) = L [sin(ωt)] X
ω ω
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 66 / 117
 Properties of Laplace Transform

Integration
Rt
Example : Find the Laplace transform of f (t) = 0 sin(ωτ )dτ

From earlier slides we know that

ω
L [sin(ωt)] =
s 2 + ω2
Using the integration property we can write
Z t 
1 ω ω
L sin(ωτ )dτ = 2 2
=
0 s s +ω s(s + ω 2 )
2

Note that
Z t   t  
1 1 1
L sin(ωτ )dτ = L − cos(ωτ ) = L − cos(ωt) +
0 ω 0 ω ω
 
1 1 1 s
= (L [1(t)] − L [cos(ωt)]) = − 2 X
ω ω s s + ω2
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 67 / 117
 Properties of Laplace Transform

Convolution
Convolution of two functions is defined as
Z t
(f ∗ g )(t) = f (τ )g (t − τ )dτ
0
t is not necessarily time, but in our context it is

This operation is commutative, i.e. (f ∗ g )(t) = (g ∗ t)(t) as clearly seen

here
Z t
(g ∗ f )(t) = g (τ )f (t − τ )dτ
0
Substitute υ = t − τ and dυ = −dτ to get
Z 0
(g ∗ f )(t) = g (t − υ)f (υ)(−dυ)
t
Z t
= g (t − υ)f (υ)dυ
0
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 68 / 117
 Properties of Laplace Transform

Convolution

But what is convolution?

Convolution operation calculates the area under f (τ ) weighted by g (−τ )

shifted by t. This area is calculated for all t ∈ [0, ∞) which gives (f ∗ g )(t)

f (t) = max(0, 1 − |t|) , g (t) = e −2t for t ≥ 0

ref : https://stackoverflow.com/a/56132955/6811631

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 69 / 117

 Properties of Laplace Transform

Convolution

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 70 / 117

 Properties of Laplace Transform

Convolution

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 71 / 117

 Properties of Laplace Transform

Convolution

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 72 / 117

 Properties of Laplace Transform

Convolution
Convolution in time domain corresponds to multiplication in the frequency
domain (i.e. s-domain)
Z ∞
L [(f ∗ g )(t)] = (f ∗ g )(t)e −st dt
Z0 ∞ Z t 
= f (τ )g (t − τ )dτ e −st dt
0 0

Note that the limits of the inner

integral is a function of t. The area
spanned by (t, τ ) pair is like →

We can change the order of

integration. We need to update the
limits of integrals properly.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 73 / 117

 Properties of Laplace Transform

Convolution

L [(f ∗ g )(t)]
Z ∞
= (f ∗ g )(t)e −st dt
0
Z ∞ Z t 
= f (τ )g (t − τ )dτ e −st dt
Z0 ∞ Z0 ∞ 
= f (τ )g (t − τ )e dt dτ mult. by e −sτ e sτ
−st
0 τ
Z ∞ Z ∞ 
−sτ −s(t−τ )
= f (τ )e g (t − τ )e dt dτ define t̄ = t − τ
0 τ
Z ∞ Z ∞ 
−sτ −s t̄
= f (τ )e g (t̄)e d t̄ dτ
Z0 ∞ 0

= f (τ )e −sτ [G (s)] dτ = F (s)G (s)

0
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 74 / 117
 Properties of Laplace Transform

Convolution

Multiplication of functions in s-domain corresponds to convolution in time

domain

L [(f ∗ g )(t)] = F (s)G (s)

This is clearly observable on block diagrams and their corresponding

overall transfer functions

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 75 / 117

 Properties of Laplace Transform

Convolution
Example : Find the ramp response of a first-order system with a pole at a

The transfer function of the system is

K
F (s) =
s −a
The corresponding time-domain function is f (t) = Ke at

The input to the system is u(t) = t and its Laplace transform reads
1
U(s) =
s2
The response of the system is
1 K
L [(u ∗ f )(t)] = U(s)F (s) =
s2 s − a
Note the order of functions in the convolution!
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 76 / 117
 Properties of Laplace Transform

Convolution
Example : (cont’d)
Z t
(u ∗ f )(t) = u(τ )f (t − τ )dτ
0
Z t 
= (τ ) Ke a(t−τ ) dτ
0
Z t
= Ke at
τ e −aτ dτ
0
Use integration by parts with u = τ and dv = e −aτ dτ
τ −aτ t
 Z t 
at 1 −aτ
· · · = Ke − e + e dτ
a 0 0 a
1 −aτ t
 
at τ −aτ
= Ke − e − 2e
a a 0
K at

= 2 e − at − 1
a
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 77 / 117
 Properties of Laplace Transform

Convolution

Example : (cont’d)

Find the inverse Laplace transform of U(s)F (s) and compare the results.

Pair #19 in Ogata Table A-1 writes

 
1 −at 1
L 2 (at − 1 + e ) = 2
a s (s + a)

Substitution a → −a gives
 
1 at 1
L 2 (−at − 1 + e ) = 2
a s (s − a)

which checks with our conclusion X.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 78 / 117
 Properties of Laplace Transform

Convolution

from sympy import *

# function definition for the convolution operation

def convolve(f, g, t, lower_limit=-oo, upper_limit=+oo):
tau = symbols('tau')
return integrate(f.subs(t, tau) * g.subs(t, tau - t),
(tau, lower_limit, upper_limit))

a, t, K = symbols('a t K', real=True)

u = t # Input to the systems
f = K * exp(a*t) # Process

res = convolve(u, f, t, 0, t)
pprint(res)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 79 / 117

 Properties of Laplace Transform

Time Product

Multiplication in time domain corresponds to convolution in frequency

domain

L [f (t) · g (t)] = (F ∗ G )(s)

Z σc +j∞
1
= F (ξ)G (s − ξ)dξ
2πj σc −j∞

where σc > max(R(sp )) ∀sp s.t. lims→sp F (s) or lims→sp G (s) is undefined.

Note the similarity of this with convolution in time-domain! (slide #74)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 80 / 117

 Properties of Laplace Transform

Time Product

Z ∞
L [f (t)g (t)] = f (t)g (t)e −st dt → subs. L−1 [F (s)] for f (t)
0
" #
Z ∞ Z σc +j∞
1
= F (ξ)e sξ dξ g (t)e −st dt → flip integrals.
0 2πj σc −j∞
Z σc +j∞ Z ∞ 
1 −(s−ξ)t
= F (ξ) g (t)e dt dξ → note shift in time
2πj σc −j∞ 0
Z σc +j∞
1
= F (ξ)G (s − ξ)dξ → note the conv. pattern
2πj σc −j∞
1
= (F ∗ G )(s)
2πj

(for time delay property see slide #42)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 81 / 117

 Properties of Laplace Transform

Multiplication by Time
Multiplication of f (t) by t corresponds to differentiation of F (s) w.r.t. s

d
L [t · f (t)] = − F (s)
ds
Consider L [t · f (t)] as follows
Z ∞
d d
F (s) = f (t)e −st dt
ds ds 0
Z ∞
= −t · f (t)e −st dt
0
Z ∞
=− [t · f (t)]e −st dt
0
= −L [t · f (t)]

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 82 / 117

 Properties of Laplace Transform

Multiplication by Time

Example : Find the Laplace transform of t sin(ωt)

We know from earlier slides that

ω
L [sin(ωt)] =
s2 + ω2
Using the multiplication by time property we get
d
L [t sin(ωt)] = − L [sin(ωt)]
ds  
d ω
=−
ds s 2 + ω 2
ω
= −2s 2
(s + ω 2 )2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 83 / 117

 Properties of Laplace Transform

Multiplication by Time
Example : Find the Laplace transform of mẍ(t) + b ẋ(t) + kx(t) = u(t) in
terms of A(s) = L [a(t)] where a(t) = ẍ(t)

d n
Assuming dt n x(t = 0) = 0 for ∀n ∈ N (i.e. zero initial conds.) and

ẍ(t) = α for t > 0 and c ∈ R (i.e. const. acc.), we can write

ẋ(t) = t · α
1
x(t) = t 2 · α
2
Hence
α
L [ẍ(t)] = L [α · 1(t)] =
s
d α
L [ẋ(t)] = L [t · α · 1(t)] = −α L [1(t)] = 2
ds s
d2
 
1 2 α
L [x(t)] = L t · α · 1(t) = +α 2 L [1(t)] = 3
2 ds s
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 84 / 117
 Properties of Laplace Transform

Multiplication by Time

Example : (cont’d)

Taking Laplace transform of both sides of the equation-of-motion we get

mẌ (s) + b Ẋ (s) + kX (s) = U(s)

1 1 1
(mα) + 2 (kα) + 3 (cα) = U(s)
s s s
Compare this against the equation in slide 65 (copied below)

c k
mA(s) + A(s) + 2 A(s) = U(s)
s s
Note A(s) = L [α · 1(t)] and L [1(t)] = 1s . Hence two equations match X

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 85 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Calculating inverse Laplace transform is not easy. Hence we prefer referring

to tables if possible. However, a transfer function can be as complex as it
could get, while we can tabulate only a limited number of pairs.

Expressing a complex s-domain function in terms of summation of simpler

fractions, and then using the tables can remedy this problem

N(s)
F (s) =
D(s)
(s − z1 )(s − z2 ) · · · (s − zm )
=K
(s − p1 )(s − p2 ) · · · (s − pn )
 
N1 (s) N2 (s) Nn (s)
=K + + ··· +
D1 (s) D2 (s) Dn (s)

where zi ∈ C are zeros, pj ∈ C are poles of F (s) and m ≤ n

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 86 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

N(s)
F (s) =
D(s)
(s − z1 )(s − z1 ) · · · (s − zm )
=K
(s − p1 )(s − p1 ) · · · (s − pn )
Qm
(s − zi )
= K ni=1
Q
i=1 (s − pi )

Note that m ≤ n must hold for physical systems

zi and pj can be real, complex or pure-complex quantities

Since the coefficients of N(s) and D(s) are real valued, if zi or pj is

complex-valued, then its conjugate, z̄i and p̄j resp., must exist as well

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 87 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Poles of a given system might be distinct (simple, 1st order poles) or

repeated (higher-order poles)

If all the poles are simple, then F (s) can be written as summation of
simple fractions as follows
a1 a2 an
F (s) = + + ··· +
s − p1 s − p2 s − pn

where ai for i = {1, 2, · · · , n} is constant and called the residue at pole pn

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 88 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Residue at pi can be calculated simply by multiplying F (s) by (s − pi ) and

evaluating the resultant expression at s = pi

ai = F (s)(s − pi )
i s=p
 
s − pi s − pi s − pi s − pi
= a1 + a2 + · · · + ai + · · · + an
s − p1 s − p2 s − pi s − pn s=pi
 
0 0 0
= a1 + a2 + · · · + ai + · · · + an = ai
si − p1 si − p2 s i − pn

This operation covers up all the denominators except for the i th one,
hence it is called the cover-up method

Note that this method works only for simple poles.

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 89 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

When multiplied by (s − pi ), all terms drop out except for the residue ai

ai = (s − pi )F (s)
s=pi

Since all coefficients of F (s) are real-valued (or equivalently f (t) is

real-valued), iff ai is complex, then āi is the residue at s = p̄i

āi = (s − p̄i )F (s)

s=p̄i

Calculating one suffices since the conjugate is automatically known

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 90 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Once the partial fraction expansion is determined, the corresponding
time-domain function, f (t), can be obtained easily since all the fractions
are in the form
ai
s − pi
We know from earlier discussions that
 
−1 K
L = Ke −at
s +a
Using the superposition property of Laplace transform we can write
 
−1 −1 a1 a2 an
L [F (s)] = L + + ··· +
s − p1 s − p2 s − pn
f (t) = a1 e p1 t + a2 e p2 t + · · · + an e pn t for t≥0
We can drop the condition t ≥ 0 using the function 1(t) as follows
f (t) = a1 e p1 t 1(t) + a2 e p2 t 1(t) + · · · + an e pn t 1(t)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 91 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Example : Calculate the inverse Laplace transform of
(s + 2)(s + 4)
F (s) =
s(s + 1)(s + 3)
Poles are p1 = 0, p2 = −1 and p3 = −3

Since all poles are simple we can write

a1 a2 a3
F (s) = + +
s s +1 s +3
Using the cover-up method we can determined ai as follows

a1 = (s − p1 )F (s)
s=p1
(s + 2)(s + 4) 8
= =
(s + 1)(s + 3) s=0 3
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 92 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

(s + 2)(s + 4) a1 a2 a3
F (s) = = + +
s(s + 1)(s + 3) s s +1 s +3

a2 = (s − p2 )F (s) a3 = (s − p3 )F (s)
s=p2 s=p3
(s + 2)(s + 4) 3 (s + 2)(s + 4) 1
= =− = =−
s(s + 3) s=−1 2 s(s + 1) s=−3 6

Knowing the residues, we can write f (t) as

8 3 1
f (t) = 1(t) − e −t 1(t) − e −3t 1(t)
3 2 6

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 93 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Example : Calculate the inverse Laplace transform of
s 3 + 5s 2 + 9s + 7
F (s) =
(s + 1)(s + 2)
Note (m = 3) ≥ (n = 2), hence we need to rewrite F (s) as
s +3
F (s) = s + 2 +
(s + 1)(s + 2)
Poles of the system are p1 = −1, p2 = −2 and all are simple. We apply
partial fraction expansion on the fractional part of F (s) (call Ff (s))

s +3
Ff (s) =
(s + 1)(s + 2)
a1 a2
= +
s +1 s +2
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 94 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)
s +3 a1 a2
Ff (s) = = +
(s + 1)(s + 2) s +1 s +2
Residuals are calculated as

a1 = (s − p1 )Ff (s) a2 = (s − p2 )Ff (s)

s=p1 s=p2
s +3 s +3
= =2 = = −1
s +2 s=−1 s +1 s=−2

Knowing the residuals we can write f (t) as follows

 
−1 2 1
f (t) = L s +2+ −
s +1 s +2
d
= δ(t) + 2δ(t) + 2e −t − e −2t for t ≥ 0−
dt
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 95 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

In the case of complex conjugate poles, expansion takes the form
a1 a2
F (s) = +
s − p1 s − p̄1
In such a case, a1 and a2 are also conjugate pairs, i.e. a1 = ā2
f (t) then takes the form
f (t) = a1 e p1 t + ā1 e p̄1 t
Assuming p1 = σ + jω and a1 = α + jβ this becomes
f (t) = (α + jβ)e (σ+jω)t + (α − jβ)e (σ−jω)t
= (α + jβ)e σt e jωt + (α − jβ)e σt e −jωt
= αe σt (e jωt + e −jωt ) + jβe σt (e jωt − e −jωt )
= 2e σt (α cos(ωt) − β sin(ωt))
= 2e σt Re (α + jβ)e jωt

= 2Re(a1 e p1 t )

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 96 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Example : Find the inverse Laplace transform of
2s + 12
F (s) =
s2 + 2s + 5
Note that the denominator can be factored as

s 2 + 2s + 5 = (s + 1 + j2)(s + 1 − j2)

Hence p1 = −(1 + j2) and p2 = p̄1 = −(1 − j2). Residuals are calculated
as
2s + 12 5
a1 = (s − p1 )F (s) = =1+j
s=p1 s + 1 − j2 s=−(1+j2) 2
2s + 12 5
a2 = (s − p2 )F (s) = =1−j
s=p2 s + 1 + j2 s=−(1−j2) 2

Note that a1 = ā2 X

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 97 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

Knowing the residuals we can write F (s) as

a1 a2 1 + j 52 1 − j 52
F (s) = + = +
s − p1 s − p2 s + 1 + j2 s + 1 − j2
Then f (t) is simply obtained as
   
5 −(1+j2)t 5
f (t) = 1 + j e + 1−j e −(1−j2)t
2 2
From the previous slides we know this simplifies to Re(ae pt ), hence
  
5 −(1+j2)t
f (t) = 2Re 1+j e
2
= 5e −t sin(2t) + 2e −t cos(2t) for t ≥ 0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 98 / 117

 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d) The strategy followed above is tedious and it is easy to
make mistakes during calculations. For this, we can instead use the
following approach

When F (s) involves conjugate poles, it is more convenient to bring the

expression into sum of damped sine and cosine

In the Ogata Table A-1 we have

ω
L e −at sin(ωt) =
 
(s + a)2 + ω 2
s +a
L e −at cos(ωt) =
 
(s + a)2 + ω 2
F (s) can written as sum of fractions similar to these two forms as
2s + 12 2(s + 1) 5·2
F (s) = = +
s2 + 2s + 5 2
(s + 1) + 2 2 (s + 1)2 + 22
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 99 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

Having F (s) written as a sum of damped sine and cosine and using the
superposition property, we can write f (t) as follows
   
s +1 2
f (t) = 2 · L−1 + 5 · L −1
(s + 1)2 + 22 (s + 1)2 + 22

Here a = −1 and ω = 2 which yields to

f (t) = 2e −t cos(2t) + 5e −t sin(2t) for t≥0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 100 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : Find the Laplace transform of
1
F (s) =
s(s 2 + s + 1)
The denominator has one real and a pair of conjugate roots which makes
this question slightly different than the previous one

We can apply partial fraction expansion as follows

a1 a2 s + a3
F (s) = + 2
s s +s +1
Using the cover-up method, we can find the coefficients as

a1 = (s)F (s) =1
s=0
We then plug-in a1 = 1 and equate the numerators in the expanded and
original expressions
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 101 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

1
F (s) = (original expression)
s(s 2
+ s + 1)
1 a2 s + a3
= + 2 note a1 = 1
s s +s +1
(1)(s 2 + s + 1) s(a2 s + a3 )
= 2
+
s(s + s + 1) s(s 2 + s + 1)
Collecting the terms in the numerators we get the equation
1 = (1 + a2 )s 2 + (1 + a3 )s + 1
→ a2 = −1 → a3 = −1
Then F (s) becomes
1 s +1
F (s) = − 2
s s +s +1
Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 102 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d) We would like to rewrite F (s) as sum of damped sine
and cosine
1 s +1
F (s) = − 2
s s +s +1
1 (s + 1/2) + 1/2
= −
s (s + 1/2)2 + 3/4
√
1 (s + 1/2) 1 3/2
= − 2
−√
s 1
(s + /2) + /4 3 3 (s + /2)2 + 3/4
1
√
Note that a = − 1/2 and ω = 3/2
     √ 
−1 1 −1 (s + 1/2) 1 −1 3/2
f (t) = L −L −√ L
s (s + 1/2)2 + 3/4 3 (s + 1/2)2 + 3/4
√ ! √ !
−t 3 1 −t 3
= 1 − e /2 cos t − √ e /2 sin t for t ≥ 0
2 3 2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 103 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Summary :

If all the poles are real and simple (i.e. 1st order) then we can write F (s) as
n
X ai
F (s) =
s − pi
i=1

where n is the order of the denominator and ai , pi ∈ R

Inverse Laplace transform is then obtained as

n  
X
−1 ai
f (t) = L
s − pi
i=1
n
X
= ai e pi t for t ≥ 0
i=1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 104 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Summary : (cont’d)

If the denominator has a quadratic term with complex roots, then we try
to rewrite F (s) as a sum of damped sine and cosine as
s +a ω
F (s) = Kc + Ks
(s + a)2 + ω 2 (s + a)2 + ω 2

where Kc , Ks ∈ R

Inverse Laplace transform is then obtained as

   
−1 s +a −1 ω
f (t) = Kc L + Ks L
(s + a)2 + ω 2 (s + a)2 + ω 2
= Kc e −at cos(ωt) + Ks e −at sin(ωt) for t ≥ 0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 105 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

The cover-up method works for if F (s) has simple poles. In case one or
more poles are repeated the partial fraction expansion takes the following
form
a1 a2 an
F (s) = + 2
+ ··· +
s − p1 (s − p1 ) (s − p1 )n

Highest-order residual can be obtained simply using the cover-up method

directly as follows

an = (s − p1 )n F (s)
s=p1
 
= a1 (s − p1 )n−1 + a2 (s − p1 )n−2 + · · · + an
s=p1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 106 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

To determine the (n − 1)st residual we first multiply both sides by
(s − p1 )n , then differentiate once w.r.t s and eventually evaluate the
resultant expression at p1 to get an−1

d 
(s − p1 )n F (s)
ds
d 
= a1 (s − p1 )n−1 + a2 (s − p1 )n−2 + · · · + an−1 (s − p1 ) + an
ds
= a1 (n − 1)(s − p1 )n−2 + a2 (n − 2)(s − p1 )n−3 + · · · + an−1

Evaluating both sides at s = p1 gives

d 
(s − p1 )n F (s) = an−1
ds s=p1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 107 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Let’s derive the expression for the (n − 2)nd residual

d2  n

(s − p1 ) F (s)
ds 2
d2  
= 2 a1 (s − p1 )n−1 + a2 (s − p1 )n−2 + · · · + an−1 (s − p1 ) + an
ds
= a1 (n − 1)(n − 2)(s − p1 )n−3 + a2 (n − 2)(n − 3)(s − p1 )n−4 +
· · · + (2) · (1)an−2

Evaluating both sides at s = p1 gives

d2  n

(s − p1 ) F (s) = 2! an−2
ds 2 s=p1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 108 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

... and for the (n − 3)rd residual

d3  n

(s − p1 ) F (s)
ds 3
d3  
= 3 a1 (s − p1 )n−1 + a2 (s − p1 )n−2 + · · · + an−1 (s − p1 ) + an
ds
= a1 (n − 1)(n − 2)(n − 3)(s − p1 )n−4 +
a2 (n − 2)(n − 3)(n − 4)(s − p1 )n−5 +
· · · + (3) · (2) · (1) an−3

Evaluating both sides at s = p1 gives

d3  n

(s − p1 ) F (s) = 3! an−3
ds 3 s=p1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 109 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

This can be generalized to obtain all residuals as below

1 
an = (s − p1 )n F (s)
0! s=p1
1 d  
an−1 = (s − p1 )n F (s)
1! ds s=p1
1 d 2  
an−2 = (s − p1 )n F (s)
2! ds 2 s=p1
..
.
1 d (n−1)  n

a1 = (s − p1 ) F (s)
(n − 1)! ds n−1 s=p1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 110 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : Find the inverse Laplace transform of
s 2 + 2s + 3
F (s) =
(s + 1)3

Since p1 = −1 is a 3rd order pole, F (s) can be written as summation of

the 3 fractions
a1 a2 a3
F (s) = + 2
+
s + 1 (s + 1) (s + 1)3
Using the cover-up method we determine a3 as follows

a3 = (s + 1)3 F (s)
s=−1

= s 2 + 2s + 3


=2
s=−1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 111 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

Residual for lesser-order terms can be calculated as follows

1 d 
a2 = (s + 1)3 F (s)
1! ds s=−1
d 2 
= s + 2s + 3 = (2s + 2) =0
ds s=−1 s=−1
1 d2  3

a1 = (s + 1) F (s)
2! ds 2 s=−1
1 d2  2  1
= 2
s + 2s + 3 = (2) =1
2 ds s=−1 2 s=−1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 112 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

Inverse Laplace transform of F (s) read as

     
−1 −1 1 −1 0 −1 2
L [F (s)] = L +L +L
s +1 (s + 1)2 (s + 1)3
= e −t + 0 + t 2 e −t
= (1 + t 2 )e −t for t≥0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 113 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : Find the inverse Laplace transform of

s +3
F (s) =
(s + 1)(s + 2)2

We have a combination of simple and repeated poles.

a1 a2 a3
F (s) = + +
s + 1 s + 2 (s + 2)2

Using the cover-up method a1 can easily be determined

 
  s +3
a1 = (s + 1)F (s) = =2
s=−1 (s + 2)2 s=−1

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 114 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

Residuals for the repeated pole are found as

 
a3 = (s + 2)2 F (s)
s=−2
 
s +3
= = −1
s + 1 s=−2
d 
a2 = (s + 2)2 F (s)
ds s=−2
 
d s +3
=
ds s + 1 s=−2
 
1 s +3
= − = −2
s + 1 (s + 1)2 s=−2

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 115 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

Example : (cont’d)

a1 a2 a3
F (s) = + +
s + 1 s + 2 (s + 2)2

a1 = 2 a2 = −2 a3 = −1

Finally f (t) reads

     
−1 −1 2 −1 2 −1 1
L [F (s)] = L −L −L
s +1 s +2 (s + 2)2
= 2e −t − 2e −2t − te −2t for t≥0

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 116 / 117
 Partial Fraction Expansion

Partial Fraction Expansion

from sympy import *

s = symbols('s', real=True)
F = (s+3) / (s+1) / (s+2)**2

res = apart(F)
pprint(res)

Dr. Tolga Özaslan (AYBU) Laplace Transform Fall 2024 117 / 117