CIVE96008: Dynamics of Structures Spring 2020/2021

Lecture 2: Free vibrations and impulse loads

February 1
Lecturer: Dr. Christian Málaga-Chuquitaype [email protected]

Disclaimer: Although carefully edited, these notes have not been subjected to the scrutiny reserved for formal
publications. Therefore, they may be distributed outside this class only with the permission of the Lecturer.

Learning outcomes
By the end of this session you should be able to:

• Understand the solution for the displacement response of a system under free vibration conditions

• Characterise the extent to which damping influences the free vibration response

• Understand how the logarithmic decrement may be used to infer damping characteristics

• Understand how the response to some arbitrary transient loading can be found from considering
the free-vibration response of a SDOF

2.1 Free vibrations

A structure undergoes free vibrations when it is brought out of its static equilibrium and can then
oscillate without any external dynamic excitation

Free vibrations can be damped or un-damped as expressed by the following equations:

𝑚𝑢̈ + 𝑘𝑢 = 0

𝑚𝑢̈ + 𝑐 𝑢̇ + 𝑘𝑢 = 0
In the following sections we will solve these equations using different formulations of the harmonic (e.g. amplitude
and phase, exponential functions and trigonometric functions)

2.2 Undamped free vibrations

𝑚𝑢̈ + 𝑘𝑢 = 0 (2.1)

2-1
Lecture 2: February 1 2-2

u(t)

m
fk(t)
k

Figure 2.1: Un-damped single degree-of-freedom system

2.2.1 Formulation A: Amplitude and phase angle

Ansatz:

𝑢(𝑡) = 𝐴𝑐𝑜𝑠(𝜔𝑛 𝑡 − 𝜙) (2.2)

̈ = −𝐴𝜔2𝑛 𝑐𝑜𝑠(𝜔𝑛 𝑡 − 𝜙)
𝑢(𝑡) (2.3)

By substituting Equations 2.2 and 2.3 into 2.1

𝐴(−𝜔2𝑛 𝑚 + 𝑘)𝑐𝑜𝑠(𝜔𝑛 𝑡 − 𝜙) = 0 (2.4)

−𝜔2𝑛 𝑚 + 𝑘 = 0 (2.5)

Natural circular frequency of the structural system 1 :

√
𝜔𝑛 = 𝑘∕𝑚 (2.6)
Relationships:

Angular velocity [rad/s]: √

𝜔𝑛 = 𝑘∕𝑚 (2.7)
Number of revolutions per time [1/s],[Hz]:
𝜔𝑛
𝑓𝑛 = (2.8)
2𝜋
Time required per revolution [s]:
1 2𝜋
𝑇𝑛 = = (2.9)
𝑓𝑛 𝜔𝑛
Transformation of the equation of motion:
̈ + 𝜔2𝑛 𝑢(𝑡) = 0
𝑢(𝑡) (2.10)

1 Please note that from now on we will be calling the natural circular frequency of the structure 𝜔 to differentiate it from the circular
𝑛
frequency of the excitation, 𝜔
Lecture 2: February 1 2-3

The unknowns 𝐴 and 𝜙 in Equation 2.2 can be obtained by considering the static equilibrium disturbed by the
̇
initial displacement 𝐮(𝟎) = 𝑢0 and initial velocity 𝐮(𝟎) = 𝑣0 . Therefore, solving the system of two equations (Eqs.
2.2 and 2.3) and two unknowns we have:

√
( )2
𝑣0 𝑣0
𝐴= 𝑢20 + ; 𝑡𝑎𝑛𝜙 = (2.11)
𝜔𝑛 𝑢0 𝜔𝑛

2.2.2 Formulation B: Trigonometric functions

Ansatz:
𝑢(𝑡) = 𝐴1 𝑐𝑜𝑠(𝜔𝑛 𝑡) + 𝐴2 𝑠𝑖𝑛(𝜔𝑛 𝑡) (2.12)
̈ = −𝐴1 𝜔2𝑛 𝑐𝑜𝑠(𝜔𝑛 𝑡) − 𝐴2 𝜔2𝑛 𝑠𝑖𝑛(𝜔𝑛 𝑡)
𝑢(𝑡) (2.13)

By substituting Equations 2.12 and 2.13 into Equation 2.1:

𝐴1 (−𝜔2𝑛 𝑚 + 𝑘)𝑐𝑜𝑠(𝜔𝑛 𝑡) + 𝐴2 (−𝜔2𝑛 𝑚 + 𝑘)𝑠𝑖𝑛(𝜔𝑡) = 0 (2.14)

√
−𝜔2𝑛 𝑚 + 𝑘 = 0 ; 𝜔𝑛 = 𝑘∕𝑚 (2.15)

The unknowns 𝐴1 and 𝐴2 in Equation 2.12 can be obtained by considering the static equilibrium disturbed by
̇
the initial displacement 𝐮(𝟎) = 𝑢0 and initial velocity 𝐮(𝟎) = 𝑣0 . Therefore, solving the system of two equations
and two unknowns we have:

𝑣0
𝐴1 = 𝑢0 ; 𝐴2 = (2.16)
𝜔𝑛

2.2.3 Formulation C: Exponential functions

Ansatz:
𝑢(𝑡) = 𝑒𝜆𝑡 (2.17)
2 𝜆𝑡
̈ =𝜆 𝑒
𝑢(𝑡) (2.18)

By substituting Equations 2.17 and 2.18 into Equation 2.1:

( )
𝑚𝜆2 + 𝑘 𝑒𝜆𝑡 = 0 (2.19)

𝑚𝜆2 + 𝑘 = 0 (2.20)
𝑘
𝜆2 = − (2.21)
𝑚
√
𝑘
𝜆 = ±𝑖 = ±𝑖𝜔𝑛 (2.22)
𝑚

The complete solution of the Ordinary Differential Equation is:

𝑢(𝑡) = 𝐶1 𝑒𝑖𝜔𝑛 𝑡 + 𝐶2 𝑒−𝑖𝜔𝑛 𝑡 (2.23)

Lecture 2: February 1 2-4

by means of Euler’s formulas (see Lecture 01):

𝑒𝑖𝛼 = 𝑐𝑜𝑠(𝛼) + 𝑖𝑠𝑖𝑛(𝛼) (2.24)

𝑒−𝑖𝛼 = 𝑐𝑜𝑠(𝛼) − 𝑖𝑠𝑖𝑛(𝛼) (2.25)

Equation 2.23 can be transformed as follows:

𝑢(𝑡) = (𝐶1 + 𝐶2 )𝑐𝑜𝑠(𝜔𝑡) + 𝑖(𝐶1 − 𝐶2 )𝑠𝑖𝑛(𝜔𝑡) (2.26)

Which corresponds to Equation 2.12!

𝑢(𝑡) = 𝐴1 𝑐𝑜𝑠(𝜔𝑛 𝑡) + 𝐴2 𝑠𝑖𝑛(𝜔𝑛 𝑡) (2.27)

2.3 Damped free vibrations

In reality vibrations subside because damping exists. In the previous lecture we have seen that damping arises
due to a number of mechanisms (e.g.friction between structural and non-structural elements, intrinsic viscous
damping, inelastic behaviour, elastic hysteresis, micro-cracks, etc). We have also discussed that it is virtually
impossible to model damping exactly because all these different mechanisms act in different ways and locations
throughout the structure and are dependent on various different parameters. From a mathematical point of view,
viscous damping is easy to treat (i.e. by means of a damping constant: c [N.s/m]). So what we normally do is
lump all damping together into an equivalent viscous damper.

u(t)
c
fc(t)
m

fk(t)
k

Figure 2.2: Damped single degree-of-freedom system

𝑚𝑢̈ + 𝑐 𝑢̇ + 𝑘𝑢 = 0 (2.28)

2.3.1 Formulation C: Exponential functions

Ansatz:
𝑢(𝑡) = 𝑒𝜆𝑡 (2.29)
Lecture 2: February 1 2-5

̇ = 𝜆𝑒𝜆𝑡
𝑢(𝑡) (2.30)

̈ = 𝜆2 𝑒𝜆𝑡
𝑢(𝑡) (2.31)

By substituting Equations 2.29 - 2.31 into Equation 2.28:

(𝜆2 𝑚 + 𝜆𝑐 + 𝑘)𝑒𝜆𝑡 = 0 (2.32)

𝜆2 𝑚 + 𝜆𝑐 + 𝑘 = 0 (2.33)

𝑐 1 √ 2
𝜆=− ± 𝑐 − 4𝑘𝑚 (2.34)
2𝑚 2𝑚

Critical Damping when 𝑐 2 − 4𝑘𝑚 = 0 √

𝑐𝑐𝑟 = 2 𝑘𝑚 (2.35)

Damping ratio:

𝑐 𝑐 𝑐
𝜉= = √ = (2.36)
𝑐𝑐𝑟 2 𝑘𝑚 2𝜔 𝑛𝑚

Transformation of the equation of motion:

̈ + 𝑐 𝑢(𝑡)
𝑚𝑢(𝑡) ̇ + 𝑘𝑢(𝑡) = 0 (2.37)

𝑐 𝑘
̈ +
𝑢(𝑡) ̇ + 𝑢(𝑡) = 0
𝑢(𝑡) (2.38)
𝑚 𝑚

𝑢(𝑡) ̇ + 𝜔2𝑛 𝑢(𝑡) = 0

̈ + 2𝜉𝜔𝑛 𝑢(𝑡) (2.39)

Types of vibration:

𝑐
• Under-damped free vibrations: 𝜉 = <1
𝑐𝑐𝑟
𝑐
• Critically damped free vibrations: 𝜉 = =1
𝑐𝑐𝑟
𝑐
• Over-damped free vibrations: 𝜉 = >1
𝑐𝑐𝑟
Lecture 2: February 1 2-6

2.3.1.1 Under-damped free vibrations

𝑐 𝑘
By substituting 𝜉 = and 𝜔2𝑛 = in Equation 2.34:
2𝜔𝑛 𝑚 𝑚
√
𝜆 = −𝜉𝜔𝑛 ±𝜔2𝑛 𝜉 2 − 𝜔2𝑛
√
= −𝜉𝜔𝑛 ± 𝜔𝑛 𝜉 2 − 1
√
= −𝜉𝜔𝑛 ± 𝑖𝜔𝑛 1 − 𝜉 2 (2.40)

Damped circular frequency: √

𝜔𝑑 = 𝜔𝑛 1 − 𝜉 2 (2.41)

𝜆 = −𝜉𝜔𝑛 ± 𝑖𝜔𝑑 (2.42)

The complete solution of the ODE is:

𝑢(𝑡) = 𝐶1 𝑒(−𝜉𝜔𝑛 +𝑖𝜔𝑑 )𝑡 + 𝐶2 𝑒(−𝜉𝜔𝑛 −𝑖𝜔𝑑 )𝑡 (2.43)

𝑢(𝑡) = 𝑒−𝜉𝜔𝑛 𝑡 (𝐶1 𝑒𝑖𝜔𝑑 𝑡 + 𝐶2 𝑒−𝑖𝜔𝑑 𝑡 ) (2.44)

𝑢(𝑡) = 𝑒−𝜉𝜔𝑛 𝑡 (𝐴1 𝑐𝑜𝑠(𝜔𝑑 𝑡) + 𝐴2 𝑠𝑖𝑛(𝜔𝑑 𝑡)) (2.45)

The determination of the unknowns 𝐴1 and 𝐴2 is carried out as usual by means of the initial conditions for
displacement (𝑢(0) = 𝑢0 ) and velocity (𝑢̇ = 𝑣0 ):

𝐴1 = 𝑢0 (2.46)

𝑣0 + 𝜉𝜔𝑛 𝑢0
𝐴2 = (2.47)
𝜔𝑑

0.08 0.08
v0 = 0.4 m/s v0 = 0.4 m/s
0.06 0.06
v0 = -0.4 m/s v0 = -0.4 m/s
0.04 0.04
Displacement [m]
Displacement [m]

0.02 0.02
0 0
-0.02 -0.02
-0.04 -0.04
-0.06 -0.06
-0.08 -0.08
0 1 2 3 0 1 2 3
Time [s] Time [s]
(a) Under-damped free vibrations, 𝜉 = 0.05 (b) Over-damped free vibrations, 𝜉 = 1.10

Figure 2.3: Types of free vibration for a SDOF with 𝑇 = 1 s and 𝑢(0) = 0.03m
Lecture 2: February 1 2-7

2.3.2 Formulation A: Amplitude and phase angle

Equation 2.45 can be rewritten in terms of the motion’s amplitude and phase angle:

𝑢(𝑡) = 𝐴𝑒−𝜉𝜔𝑛 𝑡 𝑐𝑜𝑠(𝜔𝑑 𝑡 − 𝜙) (2.48)

with:

√
( )2
𝑣0 + 𝜉𝜔𝑛 𝑢0
𝐴= 𝑢20 + (2.49)
𝜔𝑑

𝑣0 + 𝜉𝜔𝑛 𝑢0
𝑡𝑎𝑛𝜙 = (2.50)
𝜔𝑑 𝑢0

• The motion is a sinusoidal vibration with circular frequency 𝜔𝑑 and decreasing amplitude 𝐴𝑒−𝜉𝜔𝑛 𝑡 .
𝑇𝑛
• Note that the period of the damped vibration is longer (i.e. the vibration is slower) 𝑇𝑑 = √ .
1 − 𝜉2
• The envelope of the vibration is represented by 𝐴𝑒−𝜉𝜔𝑛 𝑡 .

1
0.9
0.8
0.7
Tn/Td

0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Damping ratio, ξ

Figure 2.4: Relationship between period ratio(𝑇𝑛 ∕𝑇𝑑 ) and damping ratio (𝜉 )
Lecture 2: February 1 2-8

2.4 The logarithmic decrement

0.08
0.06 u0
u1
0.04

Displacement [m] 0.02

0
-0.02
-0.04
-0.06
-0.08
0 1 2 3 4
Time [s]

Figure 2.5: Free vibration. Amplitude decrement

Amplitude of two consecutive cycles

𝑢0 𝐴𝑒−𝜉𝜔𝑛 𝑡 𝑐𝑜𝑠(𝜔𝑑 𝑡 − 𝜙)
= ( ) (2.51)
𝑢1 𝐴𝑒−𝜉𝜔𝑛 (𝑡+𝑇𝑑 ) 𝑐𝑜𝑠 𝜔𝑑 (𝑡 + 𝑇𝑑 ) − 𝜙

with

𝑒−𝜉𝜔𝑛 (𝑡+𝑇𝑑 ) = 𝑒−𝜉𝜔𝑛 𝑡 𝑒−𝜉𝜔𝑛 𝑇𝑑 (2.52)

( )
𝑐𝑜𝑠 𝜔𝑑 (𝑡 + 𝑇𝑑 ) − 𝜙 = 𝑐𝑜𝑠(𝜔𝑑 𝑡 + 𝜔𝑑 𝑇𝑑 − 𝜙) = 𝑐𝑜𝑠(𝜔𝑑 𝑡 − 𝜙) (2.53)

Replacing Equations 2.52 and 2.53 in Equation 2.51:

𝑢0 1
= −𝜉𝜔 𝑇 = 𝑒𝜉𝜔𝑛 𝑇𝑑 (2.54)
𝑢1 𝑒 𝑛 𝑑

Logarithmic decrement: ( )
𝑢0 2𝜋𝜉
𝛿 = 𝑙𝑛 = 𝜉𝜔𝑛 𝑇𝑑 = √ ≈ 2𝜋𝜉 (2.55)
𝑢1 1 − 𝜉2

If 𝜉 is small, likewise, if 𝜉 is small, the damping ratio becomes:

𝛿
𝜉≈ (2.56)
2𝜋
Lecture 2: February 1 2-9

10
9
Exact Approx.
8

Logarithmic decrement, δ
7
6
5
4
3
2
1
0
0 0.2 0.4 0.6 0.8 1
Damping ratio, ξ

Figure 2.6: Relationship between exact damping equation and its approximation

2.5 Transient loads

2.5.1 Impulse

An impulsive load is a load which is applied during a short period of time (Figure 2.7). The corresponding mag-
nitude of this type of load (𝐹̂ ) is defined as the product of the force and the time of its duration (mathematically
defined as the integral of some forcing function over a period of time in [Ns]:

𝑡+Δ𝑡
𝐹̂ = 𝐹 𝑑𝑡 (2.57)
∫𝑡

Force

Time

Figure 2.7: Impulse

It is evident from Figure 2.1 that the same impulse magnitude can be achieved by having a large-amplitude load
over a short period of time or a smaller load over a longer period.

If we treat the force 𝐹 as the net sum of forces acting on a body, then by means of Newton’s Second Law we
have:
Lecture 2: February 1 2-10

𝑑𝑣
𝐹 = 𝑚𝑎 = 𝑚 (2.58)
𝑑𝑡

by expressing the derivative of velocity as a differential and reordering:

𝐹 Δ𝑡 = 𝑚Δ𝑣 (2.59)

Therefore, the impulse magnitude 𝐹̂ = 𝐹 Δ𝑡 is equivalent to the mass multiplied by the change in velocity 𝑚Δ𝑣.

2.5.2 SDOF response to an impulse

If we apply an impulsive load, as described above, to the mass of an elastic SDOF system, we will be inducing a
change in its velocity that is equivalent to (from Equation 2.3):

𝐹 Δ𝑡 𝐹̂
Δ𝑣 = = (2.60)
𝑚 𝑚

This is analogous to impose the following initial conditions:

𝐹̂
𝑢0 = 0 ; 𝑣0 = 𝑢̇0 = (2.61)
𝑚

• Therefore, the response for any time 𝑡 of an un-damped SDOF to an impulsive load of magnitude 𝐹̂ will be
(From Equations 3.12 and 3.16):

𝑣0 𝐹̂
𝑢(𝑡) = sin(𝜔𝑛 𝑡) = sin(𝜔𝑛 𝑡) (2.62)
𝜔𝑛 𝑚𝜔𝑛

• Similarly, the response for any time 𝑡 of an damped SDOF to an impulse 𝐹̂ will be (From Equations 3.45 to
3.47):
( )
−𝜉𝜔𝑛 𝑡 𝑣0
𝑢(𝑡) = 𝑒 sin(𝜔𝑑 𝑡) (2.63)
𝜔𝑑
( )
−𝜉𝜔𝑛 𝑡 𝐹̂
𝑢(𝑡) = 𝑒 sin(𝜔𝑑 𝑡) (2.64)
𝑚𝜔𝑑
( √ )
𝐹̂
𝑢(𝑡) = √ 𝑒−𝜉𝜔𝑛 𝑡 sin(𝜔𝑛 1 − 𝜉 2 𝑡) (2.65)
𝑚𝜔𝑛 1 − 𝜉 2

From an engineering point of view "large" force over a "short" time seem imperfect appreciations of a given
phenomena. But we only need to express 𝐹̂ in terms of the original derivatives of Newton’s Second Law :
𝐹̂ = 𝐹 𝑑𝜏 .
Lecture 2: February 1 2-11

Therefore, the response of a SDOF to an Impulse applied during any time 𝑑𝜏 (from Equations 2.62 and
2.65) can be expressed as:

un-damped SDOF:
𝐹 (𝜏)
𝑑𝑢(𝑡) = sin(𝜔𝑛 (𝑡 − 𝜏))𝑑𝜏 (2.66)
𝑚𝜔𝑛
damped SDOF
𝐹 (𝜏) ( √ )
𝑑𝑢(𝑡) = √ 𝑒−𝜉𝜔𝑛 (𝑡−𝜏) sin( 1 − 𝜉 2 𝜔𝑛 (𝑡 − 𝜏) 𝑑𝜏 (2.67)
𝑚𝜔𝑛 1 − 𝜉 2

2.5.3 Response of a SDOF to an arbitrary loading

Given a load 𝐹 (𝑡) of arbitrary form, it is possible to subdivide the force 𝐹 (𝑡) into a series of impulses acting on the
structure at 𝜏 with a duration 𝑑𝜏 as presented in Figure 2.8

u(t)
c
fc(t)
F(t)
m

fk(t)
k

(a) Damped SDOF (b) Arbitrary load

Figure 2.8: Un-damped forced harmonic vibrations for SDOF systems

Integrating the effects of each one of those differential impulses varying 𝜏 , we obtain:

un-damped SDOF:
𝑡
1
𝑡 ( )
𝑢(𝑡) = 𝑑𝑢(𝑡) = 𝐹 (𝜏) sin 𝜔𝑛 (𝑡 − 𝜏) 𝑑𝜏 (2.68)
∫0 𝑚𝜔𝑛 ∫0

damped SDOF:
𝑡 𝑡 (√ )
1
𝑢(𝑡) = 𝑑𝑢(𝑡) = √ 𝐹 (𝜏)𝑒−𝜉𝜔𝑛 (𝑡−𝜏) sin 1 − 𝜉 2 𝜔𝑛 (𝑡 − 𝜏) 𝑑𝜏 (2.69)
∫0 𝑚𝜔𝑛 1 − 𝜉 2 ∫0

These integrals are know as the Convolution Integrals or Duhamel’s Integrals. In most cases it is much
more effective to numerically evaluate them.

The name Convolution Integrals expresses the fact that the total response of the system at time 𝑡 is the sum of
Lecture 2: February 1 2-12

the responses to all impulses such that

𝑡
𝑢(𝑡) = 𝑝(𝜏)ℎ(𝑡 − 𝜏)𝑑𝜏 (2.70)
∫0

where ℎ(𝑡 − 𝜏) is the unit impulse-response function.

Example:

Consider the case of a constant force of magnitude 𝐹0 applied suddenly to an un-damped elastic oscillator at
time 𝑡 = 0. If the oscillator was at rest at the time of the application of the load. What is the response?