Mechanical Vibrations

Modelling real life objects as mathematical objects:

Simply put, the world is too complex! Even when we know all the governing equations that describe some
phenomena, those equations are usually hard to solve entirely (analytically). This is because real life
problems require solving partial differential equations of many continuous and independent variables over all
of space and time! Consider for example a deceptively simple problem of calculating the trajectory of
projectiles. One only needs to apply Newton's Laws of Motion and the Universal Law of Gravitation in order
to get the trajectory of a ball thrown from the surface of the earth. But to do this one has to also make a
number of assumptions like; assume the mass is a point-mass, assume there is no air resistance, ignore the
effect of the balls rolling motion, assume the ground is perfectly flat and that the change in gravity of the
earth along the trajectory is negligible, and that the gravitation of the moon is also negligible (to get a
uniform gravitational field)... you get the point. It's not that the problem is unsolvable without these
assumptions, it's just too hard and the extra accuracy gained by our loyalty to reality is not worth all the
computational cost. So that's why we assume that the mass is a particle at the objects center of mass. We
make assumptions to make our lives easier.
That said, here's what Prof. Stephen Hawking said about 'model-dependent-realism':
"According to model-dependent realism, it is pointless to ask whether a model is real, only whether it
agrees with observation. If there are two models that both agree with observation, like the goldfish's picture
[distorted by the bowl] and ours, then one cannot say that one is more real than the other. One can use
whichever model is more convenient in the situation under consideration. For example, if one were inside the
bowl, the goldfish's picture would be useful, but for those outside, it would be very awkward to describe
events from a distant galaxy in the frame of a bowl on earth...". So, what matters for a model is how
successful it is and not how much 'real' we think it is. With that said , let's now try to model a few real life
examples...
Example: Try and find a mathematical equation to describe the motion of the mass in the following
spring-mass system.
If we try to tackle the problem as it is, we will find that the
external force, the weight, is actually distributed so will
require to integrate the density over volume. Furthermore,
each section of the spring will 'feel' a different weight as the
spring itself has weight. But if we model the system as a
point-mass attached to a massless spring with stiffness k, we
will arrive at a solution much quickly and readily without
sacrificing too much accuracy. So, starting with a governing
equation...
―→
∑ Fexternal ＝ m a

-m ⋅ g - k ⋅ x + c ⋅ x' ＝ m ⋅ x'' where k is the stiffness constant and c is some dampness

scaling factor...

This is a 2nd order, linear ODE which we then solve to find the solution, x(t).

Notice there are 3 major constants attached to the three orders of derivation of x. Mass, stiffness and damping.
This turns out to be an over-arching theme in vibrations. To model complex systems, we need to find the
equivalent mass, stiffness and damping.

Page 1 of 12 11/08/2024
Lumping of Parameters: In every vibration, there's an exchange between kinetic and potential
energies. The potential energy is maximum whenever the spring is compressed or stretched and the kinetic
energy is maximum when the mass zooms by the equilibrium position at maximum speed. Like we said
before, there are three major elements of every vibration system: mass, stiffness and damping. The kinetic
energy of the system is due to the mass and how fast it goes, the potential energy due the stiffness of the
'spring' and how far it's stretched and the damping constant is due to lost energy from friction or viscosity.
When modeling problems, we often 'lump' parameters together to get an equivalent and simplified system.
In mechanical vibrations specifically, we want to simplify complex problems into 3 elements only or at
least combinations of plenty of these three elements. The elements are, a mass equivalent to the total mass
of the system, a stiffness constant equivalent to the total stiffness of the system and a damping constant
equivalent to the 'damping effect' of the whole system.

Equivalent Stiffness: Stiffness gives rise to potential energy. The equivalent stiffness constant is
that which stores the same potential energy as all the springs in a system combined. That is...

1 2 1 2
―keq ⋅ xeq ＝ ∑ ―ki ⋅ xi
2 2
For springs connected in parallel, the
deflection (x) is the same on all springs... 1 2 1 2
―keq ⋅ x ＝ ∑ ―ki ⋅ x >> keq ＝ ∑ ki
2 2

1 1 F F
For springs connected in series, the force ―F ⋅ xeq ＝ ∑ ―F ⋅ xi >> xeq ＝ ∑ xi >> ―― ＝ ∑―
experienced by all springs is the same. 2 2 keq ki

1 1
>> ―― ＝ ∑―
keq ki
Equivalent Mass: Mass gives rise to kinetic energy, so the equivalent mass is defined as...
1 2 1 2
―meq ⋅ veq ＝ ∑ ―mi ⋅ vi ...were vi means each element velocity
2 2
meq ⋅ x'eq 2 ＝ ∑ mi ⋅ x'i ...where xi is the measure of displacement for
each moving members of the system
To solve for meq, we should first relate xeq to all xi through what are called 'constraint' equations...

Equivalent Damping More on this later...

Page 2 of 12 11/08/2024
 Free Vibration of a single degree of freedom

Consider a simple point mass attached to the end of a spring. When the mass is given some push, two forces
are at play, the elastic resistive force of the spring (acting against the direction of acceleration), and the mass's
own inertial force (F=ma) in the direction of acceleration. Both forces are functions of the position of the mass
only.

Spring-mass system

elastic force inertial force ⎛ 2 ⎞

d
F ＝ -k ⋅ x ((t)) F ＝ m ⋅ a ＝ m ⋅ ⎜―― x ((t))⎟
2
⎜⎝ d t ⎟⎠
So >> -k ⋅ x ((t)) ＝ m ⋅ x′′ ((t))

...and the elastic force is always opposite

(restorative to the motion), hence the minus

k
or x′′ ((t)) + ― x ((t)) ＝ 0
m

The above is only true when there is no external load acting on the spring-mass system, i.e. at the natural
frequency of the spring-mass. It's a 2nd order, linear and homogenous O.D.E with constant coefficients so its
general solution will have the form:

solve , x
rewrite , cos
assume , k > 0 , m > 0 ⎛ ⎞ ⎛ ⎞
k t ⋅ ‾‾
k t ⋅ ‾‾k
x′′ ((t)) + ― x ((t)) ＝ 0 ――――――→ ((_z1 - _z)) ⋅ 1i ⋅ sin ⎜――⎟ + ((_z1 + _z)) ⋅ cos ⎜――⎟
m ⎜ ‾‾ ⎟ ⎜ ‾‾ ⎟
⎝ m ⎠ ⎝ m ⎠
‾‾‾
k
the ― term is the natural resonance frequency of the spring-mass. If we add a damping constant, c, and
m
k
some external force; x′′ ((t)) + c ⋅ x′ ((t)) + ― x ((t)) ＝ F ((t)) now it becomes forced oscillation and it's no longer
m
homogenous O.D.E.

A better way of writing the general solution is as follows:

⎛B⎞
where C 2 ＝ A 2 + B 2 and γ ＝ atan ⎜―⎟
A ⋅ cos ((ω ⋅ t)) + B ⋅ sin ((ω ⋅ t)) ＝ C ⋅ cos ((ω ⋅ t - γ)) ⎝A⎠

Here C is the amplitude of motion (range of motion), ω is the angular frequency and γ is the phase shift of the sine
wave.

Page 3 of 12 11/08/2024
 Example:
Suppose that m=2kg and k=8N/m.
The whole mass and spring setup is sitting on a truck that was traveling at 1 m/s.
The truck crashes and hence stops. The mass was held in place 0.5 meters forward from the rest position. During
the crash the mass gets loose. That is, the mass is now moving forward at 1 m/s, while the other end of the spring
is held in place. The mass therefore starts oscillating. What is the frequency of the resulting oscillation? What is
the amplitude? The units are the mks units (meters-kilograms-seconds).
N m
Given: m ≔ 2 kg k≔8 ― I.C.: x0 ≔ 0.5 m x'0 ≔ 1 ―
m s
k
Assuming an undamped, free oscillation: x′′ ((t)) + ― x ((t)) ＝ 0
m
‾‾‾
k rad ⎛ 2 ⋅ rad ⋅ t ⎞ ⎛
General solution: ω≔ ―= 2 ―― x ((t)) ≔ A ⋅ cos ((ω ⋅ t)) + B ⋅ sin ((ω ⋅ t)) → B ⋅ sin ⎜―――⎟ + A ⋅ cos ⎜
m s ⎝ s ⎠ ⎝
To find A and B:
⎡ x ((0)) ＝ x0 ⎤ solve , A , B ⎡ 0.5 ⋅ m ⎤ ‾‾‾‾‾‾‾
2 2 ⎛B⎞
[[ A B ]] ≔ ⎢ ⎥ ―――― → ⎢ 0.5 ⋅ m ――― ⎥ C ≔ A + B = 0.707 m γ ≔ atan ⎜―⎟ = 0.785
⎣ x′ (
(0 )
) ＝ x'0⎦ ⎣ rad ⎦ ⎝A⎠

2π
t ≔ 0 s , .1 s ‥ 10 s So the specific answer is: x ((t)) ≔ C ⋅ cos ((ω ⋅ t - γ)) T ≔ ―― = 3.142 s
ω
You can see from the plot that
0.75 the period is close to 3s which
0.6
0.45 is also what we get from the
0.3 calculation.
0.15
0
-0.15 0 1 2 3 4 5 6 7 8 9 10 x ((t)) ((m))
-0.3
-0.45
-0.6
-0.75

t ((s))
clear ⎛⎝x , t , T , C , γ , A , B , k , m , x0 , x'0 , ω⎞⎠

Damped Oscillations: Most spring-mass systems don't oscillate forever as in the ideal case. There is friction,
which opposes motion and has a damping effect on the oscillation. It's proportional to the speed with a constant
of proportionality, C. So, now the general equation for spring mass system becomes:
m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ F where c is the damping constant and F is the external
load. For now, let F=0.
m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ 0 This is the general equation for free, damped oscillation at
natural frequency, a 2nd order, linear, homogeneous O.D.E

Plugging in e r ⋅ t to get the x ((t)) ≔ e r ⋅ t → e r ⋅ t

characteristic equation, we get what's
called the characteristic equation of m ⋅ x′′ ((t)) + c ⋅ x′ ((t)) + k ⋅ x ((t)) ＝ 0 → ⎛⎝m ⋅ r 2 + c ⋅ r + k⎞⎠ ⋅ e r ⋅ t ＝ 0
the differential equation.

⎡ ‾‾‾‾‾‾‾‾‾‾‾‾ 2 ⎤
((4 ⋅ k ⋅4m
-Page of)) +
12c - c ⎥ 11/08/2024
⎢ ―――――――
⎡ ⎤ solve , r
 ⎡ ‾‾‾‾‾‾‾‾‾‾‾‾ ⎤
-((4 ⋅ k ⋅ m)) + c 2 - c ⎥
⎢ ―――――――
⎡ r1 ⎤ solve , r ⎢ We can see that the -((4 ⋅ k ⋅ m)) + c 2
2 2⋅m ⎥
⎢⎣ r2 ⎥⎦ ≔ m ⋅ r + c ⋅ r + k ＝ 0 ――→ ⎢ ⎥ factor determines whether we get a
- ‾‾‾‾‾‾‾‾‾‾‾‾
-((4 ⋅ k ⋅ m)) + c 2 - c ⎥
⎢ ――――――― real or complex root.
⎢⎣ 2⋅m ⎥⎦
2
Case 1). -((4 ⋅ k ⋅ m)) + c > 0 :2 real and distinct roots.

⎡ ‾‾‾‾‾‾‾‾‾‾‾‾ ⎤ ⎡ ‾‾‾‾‾‾‾‾‾‾‾‾ ⎤
-((4 ⋅ k ⋅ m)) + c 2 - c ⎥
⎢ ――――――― -((4 ⋅ k ⋅ m)) + c 2
⎢ ―――――― c ⎥
- ――
⎡ r1 ⎤ ⎢ 2⋅m ⎥ expand ⎢ 2⋅m 2⋅m ⎥
→
⎢⎣ r2 ⎥⎦ ⎢ ⎥ ――― → ⎢ ⎥
- ‾‾‾‾‾‾‾‾‾‾‾‾ 2
-((4 ⋅ k ⋅ m)) + c - c ⎥
⎢ ―――――――
‾‾‾‾‾‾‾‾‾‾‾‾
-((4 ⋅ k ⋅ m)) + c
⎢ -――――――
2
c ⎥
- ――
⎢⎣ 2⋅m ⎥⎦ ⎢⎣ 2⋅m 2 ⋅ m ⎥⎦

⎡ ‾‾‾‾‾‾‾‾‾‾
⎛k⎞ ⎤ ⎡ ‾‾‾‾‾‾‾‾
k ⎤
⎢ -⎜―⎟ + ((p)) 2 - p ⎥ ⎢ -―+ p 2 - p ⎥ c
⎡ r1 ⎤ ⎢ ⎝m⎠ ⎥→⎢ m ⎥ where p= ――
This is equivalent to: ⎢⎣ r2 ⎥⎦ ≔ ⎢ ‾‾‾‾‾‾‾‾‾‾
⎛k⎞ ⎥ ⎢ ‾‾‾‾‾‾‾‾ ⎥
2⋅m
k
⎢- -⎜―⎟ + ((p)) 2 - p ⎥ ⎢ - -―+ p 2 - p ⎥
⎣ ⎝m⎠ ⎦ ⎣ m ⎦
k ‾‾‾‾‾‾‾‾
k
Now since r1 and r2 are always real, -―+ p 2 < p 2 because k and m are positive, so, ||p|| > -―+ p 2
m m
r1 ⋅ t
and the roots r1 and r2 will always be negative, i.e. the general solution given as C1 ⋅ e + C2 ⋅ e r2 ⋅ t will
be a decaying function with time.
let r1 ≔ -1 r2 ≔ -2
⎡ ( ) ⎤ solve , C1 , C2
(
x ((t)) ≔ C1 ⋅ e r1 ⋅ t + C2 ⋅ e r2 ⋅ t → C1 ⋅ e -t + C2 ⋅ e -(2 ⋅ t)
)
[[ C1 C2 ]] ≔ ⎢ x (0) ＝ 1 ⎥ ――――→ [[ -1 2 ]]
⎣ x′ ((0)) ＝ -3 ⎦

( )
x ((t)) ≔ C1 ⋅ e r1 ⋅ t + C2 ⋅ e r2 ⋅ t → -e -t + 2 ⋅ e -(2 ⋅ t)

In this case, the mass-spring system is

1
said to be 'over-damped' because the
0.9 damping factor is more than the elastic
0.8 force (bouncy factor). The graph always
0.7 decays to zero, which is the rest position
0.6 of the mass. Note that changing the
0.5 initial conditions changes the graph for
0.4 the first few seconds but as t approaches
0.3
x ((t)) ∞ , x always goes to zero. If x' is big
0.2 enough in the negative direction, the
0.1 graph 'overshoots' and starts decaying
0
0 1 2 3 4 5 6 7 8 9 10 back to zero from the negative side.
-0.1
-0.2

t clear ((x , r1 , r2 , C1 , C2 , C))

Page 5 of 12 11/08/2024
 k ‾‾‾
k
Case 2). -((4 ⋅ k ⋅ m)) + c 2 < 0 : complex roots. Or, -―+ p 2 <0 or p < ―
m m

⎡ ‾‾‾‾‾‾
k ⎤ k
⎢ ‾‾‾
-1 ⋅ ―- p - p ⎥ substitute , ―
2 ＝ ω0 2 ⎡ ⎤ substitute , ‾‾‾‾‾‾‾‾
⎡ r1 ⎤ ⎢ m 1i ⋅ ‾‾‾‾‾‾‾‾
-p 2
+ ω 2
- p -p 2 + ω0 2 ＝ ω1 ⎡ -p + 1i ⋅ ω ⎤
m ⎥ ――――――→ ⎢ 0 ⎥ ――――――――― 1
⎢⎣ r2 ⎥⎦ ≔ ⎢ ‾‾‾‾‾‾ ⎥ ⎢ ⎛ ‾‾‾‾‾‾‾‾ 2 ⎞ ⎥
→⎢
-p - 1i ⋅ ω
⎥
k 2
-⎝1i ⋅ -p + ω0 ⎠ - p ⎦ ⎣ 1⎦
⎢ - ‾‾‾ ⎣
2
-1 ⋅ ― - p - p ⎥
⎣ m ⎦

where ω0 is the angular frequency if the system were undamped and ω1 is always positive

Now general solution will be: r1 → -p + 1i ⋅ ω1 r2 → -p - 1i ⋅ ω1

expand
( )
collect , e -(p ⋅ t) -(p ⋅ t)
――――→ e ( ) ⋅ ⎛⎝C1 ⋅ e 1i ⋅ ω1 ⋅ t + C2 ⋅ e -⎝1i ⋅ ω1 ⋅ t⎠⎞⎠
r1 ⋅ t r2 ⋅ t ⎛ ⎞
C1 ⋅ e + C2 ⋅ e

⎛ ⎞
rewrite , sincos
C1 ⋅ e 1i ⋅ ω1 ⋅ t + C2 ⋅ e -⎝1i ⋅ ω1 ⋅ t⎠ ――――― → ((-((1i ⋅ C2)) + 1i ⋅ C1)) ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + ((C2 + C1)) ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠
*where C1 and C2 are any complex
Replacing this with new const. A and B and substituting... numbers...

( )
general solution: x ((t)) ＝ e -(p ⋅ t) ⋅ ⎛⎝A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞⎠
float , 3
let... p ≔ .3 ω0 ≔ 2 ω1 ≔ ‾‾‾‾‾‾‾
ω0 2 - p 2 ――― → 1.98 * ω1 is the damped natural frequency of
vibration
-((p ⋅ t)) -0.3 ⋅ t
( )
x (t) ≔ e ⎛ ⎛ ⎞ ⎛ ⎞⎞
⋅ ⎝A ⋅ sin ⎝ω1 ⋅ t⎠ + B ⋅ cos ⎝ω1 ⋅ t⎠⎠ → e ⋅ (A ⋅ sin (1.98 ⋅ t)) + B ⋅ cos ((1.98 ⋅ t))))
( (
solve , A , B
⎡ ( ) ⎤ ⎡ ⎤ float , 3
[[ A B ]] ≔ ⎢ x (0) ＝ 17 ⎥ → ⎢ B ＝ 17
―――― → [[ 1.57 17.0 ]]
⎣ x′ ((0)) ＝ -2 ⎦ ⎣ -0.3 ⋅ B + 1.98 ⋅ A ＝ -2.0 ⎥⎦
⎛B⎞
-((p ⋅ t)) where C ≔ ‾‾‾‾‾‾‾
A 2 + B 2 and γ ≔ atan ⎜―⎟
We can further simplify this as: x ((t)) ≔ e ⋅ ⎛⎝C ⋅ cos ⎛⎝ω1 ⋅ t - γ⎞⎠⎞⎠ ⎝A⎠
t ≔ 0 , .1 ‥ 10
Now, the spring mass is said to be under-damped.
17.5 You can see that the mass oscillates back and forth,
14 overshooting the equilibrium mark multiple times
10.5
before decaying to zero eventually. The two traces in
7
red and green are the envelope equations that bound
3.5 x ((t))
0
the oscillation,i.e. the mass will never go beyond this
0 1 2 3 4 5 6 7 8 9 10
lines. Like before, changing the initial conditions
-3.5
C ⋅ e -p ⋅ t
-7 changes the graph, same way increasing the initial
-10.5 stretch on the spring-mass makes it oscillate longer,
-C ⋅ e -p ⋅ t
-14
and so on. When -((4 ⋅ k ⋅ m)) + c 2 ＝ 0 the system is
-17.5
said to be critically damped and the behavior is
similar to the over-damped case. However, this is
t almost never the case since it's unstable.

clear ⎛⎝A , B , C , γ , ω1 , ω0 , p , x , t⎞⎠

Page 6 of 12 11/08/2024
-10.5
-14 -C ⋅ e -p ⋅ t
-17.5

clear ⎛⎝A , B , C , γ , ω1 , ω0 , p , x , t⎞⎠

Forced Oscillations:
is when the R.H.S is a non-zero fun_ of time that 'forces' the spring-mass to oscillate at a frequency other
than its natural frequency.
m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ F

This is now a linear, 2nd order, non-homogeneous equation. Suppose our external load is a periodic
fun_ given as: F ＝ F0 ⋅ sin ((ω ⋅ t)) at some known frequency ω .
m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ F0 ⋅ sin ((ω ⋅ t))

Ignoring the damping for a moment and solving for the solution of this nonhomogeneous equation, we see that it
will have a general solution of the form:

k ‾‾‾
-k ⎡ 1i ⋅ ω0 ⎤
r 2 + ―＝ 0 r ＝ ―― ＝⎢ xn ((t)) ≔ A ⋅ cos ⎛⎝ω0⎞⎠ + B ⋅ sin ⎛⎝ω0⎞⎠ → B ⋅ sin ⎛⎝ω0⎞⎠ + A ⋅ cos ⎛⎝ω0⎞⎠
⎥
m m ⎣ -1i ⋅ ω0 ⎦ *where xn denotes the null
F0 ⋅ sin ((ω ⋅ t)) space...
k () ( ) ( )
x′′ + ― x ＝ ――――― try x (t) ≔ C ⋅ sin (ω ⋅ t) → C ⋅ sin (ω ⋅ t)
m m

x′′ ((t)) + ω0 2 x ((t)) → ⎛⎝ω0 2 - ω 2 ⎞⎠ ⋅ C ⋅ sin ((ω ⋅ t))

F0 ⋅ sin ((ω ⋅ t)) solve , C F0
C ≔ ⎛⎝ω0 2 - ω 2 ⎞⎠ ⋅ C ⋅ sin ((ω ⋅ t)) ＝ ――――― ――― → ―――――
m ⎛ω - ω 2 ⎞ ⋅ m
2
⎝ 0 ⎠
F0 ⋅ sin ((ω ⋅ t))
So, the particular solution: xp ≔ C ⋅ sin ((ω ⋅ t)) → ―――――
m ⋅ ⎛⎝ω0 2 - ω 2 ⎞⎠

So, general solution: This is only true

F0 ⋅ sin ((ω ⋅ t))
xp + xn ((t)) → ――――― + B ⋅ sin ⎛⎝ω0⎞⎠ + A ⋅ cos ⎛⎝ω0⎞⎠ when ω0 ≠ ω !
m ⋅ ⎛⎝ω0 2 - ω 2 ⎞⎠
But now, suppose that our external clear ((A , B , C))
periodic load is oscillating at the
natural frequency, ω0 , the null try x ((t)) ≔ A ⋅ t ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + B ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ → A ⋅ t ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + B ⋅ t ⋅ cos ⎛⎝
space will still be the same, but an
simplify
Xp that's any multiple of sine will x′′ ((t)) + ω0 2 x ((t)) ――― → -⎛⎝2 ⋅ ω0 ⋅ ⎛⎝B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ - A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠⎞⎠⎞⎠
not work because that is included
in the null space.
F0 ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ F0 F0
-⎛⎝2 ⋅ ω0 ⋅ ⎛⎝B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ - A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠⎞⎠⎞⎠ ＝ ――――― B ≔ ―――→ ――― A ≔ 0
m 2 ⋅ ω0 ⋅ m 2 ⋅ ω0 ⋅ m
*Treating sin() and cos() are
F0 ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠
independent basis, so now Xp xp ≔ A ⋅ t ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + B ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ → ――――――
becomes... 2 ⋅ ω0 ⋅ m

and the general solution: F0 ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠

xp + xn ((t)) → ―――――― + B ⋅ sin ⎛⎝ω0⎞⎠ + A ⋅ cos ⎛⎝ω0⎞⎠ This is only true
2 ⋅ ω0 ⋅ m when ω0 ＝ ω !

clear ((A , B))

Page 7 of 12 11/08/2024
 clear ((A , B))
Lets plot each case to better understand what's going on...
let ... ω ≔ π ω0 ≔ 4 m ≔ 0.5 F0 ≔ 10
F0 ⋅ sin ((ω ⋅ t)) float , 3
x1 ((t)) ≔ ――――― + B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ ――― → B ⋅ sin ((4.0 ⋅ t)) + 3.26 ⋅ sin ((3.14 ⋅ t)) + A ⋅ cos ((4.0 ⋅ t))
m ⋅ ⎛⎝ω0 2 - ω 2 ⎞⎠

⎡ ( ) ⎤ ⎡ ⎤ solve , A , B
[[ A B ]] ≔ ⎢ x1 (0) ＝ 0 ⎥ → ⎢ A＝0
―――― → [[ 0.0 -2.5591 ]]
⎣ x1′ ((0)) ＝ 0 ⎦ ⎣ 4.0 ⋅ B + 10.2364 ＝ 0.0 ⎥⎦

F0 ⋅ sin ((ω ⋅ t)) float , 3

x1 ((t)) ≔ ――――― + B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ ――― → -2.56 ⋅ sin ((4.0 ⋅ t)) + 3.26 ⋅ sin ((3.14 ⋅ t))
m ⋅ ⎛⎝ω0 2 - ω 2 ⎞⎠
clear ((A , B))
10
F0 ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ floa
8 x2 ((t)) ≔ ―――――― + B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠ ――
2 ⋅ ω0 ⋅ m
6

4
⎡ ( ) ⎤ solve , A , B
2
[[ A B ]] ≔ ⎢ x2 (0) ＝ 0 ⎥ ―――― → [[ 0.0 -0.625 ]]
⎣ x2′ ((0)) ＝ 0 ⎦
0
x1 ((t))
-2
F0 ⋅ t ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠
-4 x2 ((t)) ≔ ―――――― + B ⋅ sin ⎛⎝ω0 ⋅ t⎞⎠ + A ⋅ cos ⎛⎝ω0 ⋅ t⎞⎠
2 ⋅ ω0 ⋅ m
-6
0 2 4 6 8 10 12 14 16 18 20
-8

-10 27

22.5

t 18

13.5

Undamped forced oscillation at ω ≠ ω0 9

4.5

From the plot above, you can see the oscillation

0
0 1 2 3 4 5 6 7 8 9 10
x2 ((t))
-4.5
itself is sort of periodic at T roughly 7 sec. This is
-9
due to the different frequencies, natural frequency
-13.5
ω0 and forced frequency ω . There's both
-18
constructive and destructive interference. However,
-22.5
from the plot to the right, we see that the amplitude
of the oscillation increases non-stop with time.
Mathematically, this is due to the 't' in the first term t
of x2(t). But more importantly, the physics
interpretation of this is that the external load Undamped forced oscillation at ω ＝ ω0
resonates with the material undergoing the
oscillation. This is just a fancy way of saying that
the forced oscillation is at the natural frequency of
the material.

clear ⎛⎝x1 , x2 , A , B , ω , ω0 , F0 , m⎞⎠

Page 8 of 12 11/08/2024
 clear ⎛⎝x1 , x2 , A , B , ω , ω0 , F0 , m⎞⎠

Practical Resonance: (Because real life is almost never ideal!)

In real life, there's always some sort of damping that brings the oscillation to a stop. To see the effect of damping
on forced oscillations, solve for the general solution of the linear, 2nd order, nonhomogeneous O.D.E for
mechanical vibration:
m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ F0 ⋅ cos ((ω ⋅ t)) This is the same as for free, damped oscillations
c ‾‾‾
k
that we saw before with p ＝ ――, ω0 ＝ ―
First, for the null-space: m ⋅ x′′ + c ⋅ x′ + k ⋅ x ＝ 0 2m m
and ω ＝ ω - p‾‾‾‾‾‾‾
2 2
1 0

⎡ C1 ⋅ e r1 ⋅ t
+ C2 ⋅ e r2 ⋅ t ⎤
If it's over-damped ( p 2 > ω0 2 )
⎢ -((p ⋅ t)) ⎥
xn ＝ ⎢ e ⋅ ⎛⎝A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞⎠ ⎥ If it's under-damped ( p 2 < ω0 2 )
⎢⎣ C1 ⋅ t ⋅ e -p ⋅ t + C2 ⋅ e -p ⋅ t ⎥⎦ If it's critically-damped ( p 2 ＝ ω0 2 ), with r1 and r2
being the roots of the characteristic equation.

Now, to find the XP for the case with ω ≠ ω0 plugin... xp ((t)) ≔ A ⋅ cos ((ω ⋅ t)) + B ⋅ sin ((ω ⋅ t)) → B ⋅ sin ((ω ⋅ t)) + A ⋅ cos ((ω
F0
xp′′ ((t)) + 2 p ⋅ xp′ ((t)) + ω0 2 xp ((t)) ＝ ― cos ((ω ⋅ t)) rearranging the equation...
m

collect , cos , sin

xp′′ ((t)) + 2 p ⋅ xp′ ((t)) + ω0 2 xp ((t)) ――――― → ⎛⎝-((2 ⋅ ω ⋅ A ⋅ p)) + ⎛⎝ω0 2 - ω 2 ⎞⎠ ⋅ B⎞⎠ ⋅ sin ((ω ⋅ t)) + ⎛⎝2 ⋅ ω ⋅ B ⋅ p + ⎛⎝ω0 2 - ω
F0
⎛-((2 ⋅ ω ⋅ A ⋅ p)) + ⎛ω 2 - ω 2 ⎞ ⋅ B⎞ ⋅ sin ((ω ⋅ t)) + ⎛2 ⋅ ω ⋅ B ⋅ p + ⎛ω 2 - ω 2 ⎞ ⋅ A⎞ ⋅ cos ((ω ⋅ t)) ＝ ― cos ((ω ⋅ t))
⎝ ⎝ 0 ⎠ ⎠ ⎝ ⎝ 0 ⎠ ⎠
m
solve , A , B
⎡ -((2 ⋅ ω ⋅ A ⋅ p)) + ⎛ω 2 - ω 2 ⎞ ⋅ B ＝ 0 ⎤
⎝ 0 ⎠ ⎡ F0 ⋅ ⎛⎝ω0 - ω⎞⎠ ⋅ ⎛⎝ω0 + ω⎞⎠
[[ A B ]] ≔ ⎢ ⎥ simplify
⎢ 2 ⋅ ω ⋅ B ⋅ p + ⎛ω 2 - ω 2 ⎞ ⋅ A ＝ ― F 0 ⎥ ―――― → ⎢ ―――――――――――― ――――
⎝ ⎠ ⎢
⎣ m ⋅ ⎛4 ⋅ ω 2 ⋅ p 2 + ω 4 - 2 ⋅ ω 2 ⋅ ω 2 + ω 4 ⎞ m ⋅ ⎛4 ⋅ ω 2 ⋅
⎢⎣ 0
m ⎥⎦ ⎝ 0 0 ⎠ ⎝
F0 ⋅ ⎛⎝ω0 - ω⎞⎠ ⋅ ⎛⎝ω0 + ω⎞⎠ 2 ⋅ F0 ⋅ ω ⋅ p
A → ―――――――――――― B → ――――――――――――
m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠ m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠

simplify ‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾
F0 2
C ≔ ‾‾‾‾‾‾‾
A 2 + B 2 ――― → ―――――――――――――
m 2 ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠

2 ⋅ F0 ⋅ ω ⋅ p ⋅ sin ((ω ⋅ t)) F0 ⋅ ⎛⎝ω0 - ω⎞⎠ ⋅ ⎛⎝ω0 + ω⎞⎠ ⋅ cos ((ω ⋅ t))
xp ((t)) ≔ A ⋅ cos ((ω ⋅ t)) + B ⋅ sin ((ω ⋅ t)) → ―――――――――――― + ――――――――――――
m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠ m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4
‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾
F0 2
xp ＝ C ⋅ cos ((ω ⋅ t - γ)) → function ＝ cos ((ω ⋅ t - γ)) ⋅ ――――――――――――― *where γ such that
m 2 ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠ tan ((γ)) ＝ B ÷ A

Now lets see what happens at resonance (when ω ＝ ω0 )

assume , ω ＝ ω0 assume , ω ＝ ω0 F0 assume , ω ＝ ω0 ||F0||
A ――――― →0 B ――――― → ―――― C ――――― → ――――――
2⋅ω⋅m⋅p
2 ⋅ ‾‾‾‾‾‾‾‾‾
ω2 ⋅ m2 ⋅ p2

B assume , ω ＝ ω0 Page 9 of 12 11/08/2024

tan ((γ)) ＝ ―――――― → undefined
 B assume , ω ＝ ω0
tan ((γ)) ＝ ―――――― → undefined π
A because A=0 and so tan() blows up to ∞ . that happens at γ ＝ ―
2

Now lets make some plots and try to understand things visually. (The best way to learn!)
Suppose:
c ‾‾‾
k
k≔1 m ≔ 1 F0 ≔ 1 c ≔ .7 ω ≔ 1.1 p ≔ ―― = 0.35 ω0 ≔ ―= 1 ω1 ≔ ‾‾‾‾‾‾‾
ω0 2 - p 2 = 0.937
2m m

F0 ⋅ ⎛⎝ω0 - ω⎞⎠ ⋅ ⎛⎝ω0 + ω⎞⎠ 2 ⋅ F0 ⋅ ω ⋅ p

A ≔ ―――――――――――― = -0.33 B ≔ ―――――――――――― = 1.209
m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠ m ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠

xp ((t)) ≔ A ⋅ cos ((ω ⋅ t)) + B ⋅ sin ((ω ⋅ t)) To find the null space, compare p 2 > ω0 2 = 0 So, we have a case of
p 2 and ω0 2 p 2 < ω0 2 = 1 underdamped
oscillation!
clear.sym ((A , B))
( )
xn ((t)) ≔ e -(p ⋅ t) ⋅ ⎛⎝A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞⎠ x ((t)) ≔ xp ((t)) + xn ((t))

To find arbitrary constants A and B, we need some I.C.

solve , A , B solve , A , B
⎡ ( ) ⎤ float , 4 ⎡ ( ) ⎤ float , 3
⎡ _A _B ⎤ ≔ ⎢ x (0) ＝ 0 ⎥ ―――― → [[ -1.296 0.3297 ]] ⎡ _A _B ⎤ ≔ ⎢ x (0) ＝ 1 ⎥ ―――― → [[ -0.923 1.33 ]]
⎣ 0 0⎦ ⎣ x′ ((0)) ＝ 0 ⎦ ⎣ 1 1⎦ ⎣ x′ ((0)) ＝ 0 ⎦

solve , A , B solve , A , B
⎡ ( ) ⎤ float , 3 ⎡ ( ) ⎤ float , 3
⎡ _A _B ⎤ ≔ ⎢ x (0) ＝ 1 ⎥ ―――― → [[ 0.145 1.33 ]] ⎡ _A _B ⎤ ≔ ⎢ x (0) ＝ 1 ⎥ ―――― → [[ -1.99 1.33 ]]
⎣ 2 2⎦ ⎣ x′ ((0)) ＝ 1 ⎦ ⎣ 3 3⎦ ⎣ x′ ((0)) ＝ -1 ⎦

( )
float , 3 -0.35 ⋅ t
xn1 ((t)) ≔ e -(p ⋅ t) ⋅ ⎛_A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + _B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞ ――― →e ⋅ ((-1.3 ⋅ sin ((0.937 ⋅ t)) + 0.33 ⋅ cos ((0.937 ⋅ t))))
⎝ 0 0 ⎠
( )
float , 3 -0.35 ⋅ t
xn2 ((t)) ≔ e -(p ⋅ t) ⋅ ⎛_A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + _B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞ ――― →e ⋅ ((-0.923 ⋅ sin ((0.937 ⋅ t)) + 1.33 ⋅ cos ((0.937 ⋅ t))))
⎝ 1 1 ⎠
( )
float , 3 -0.35 ⋅ t
xn3 ((t)) ≔ e -(p ⋅ t) ⋅ ⎛_A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + _B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞ ――― →e ⋅ ((0.145 ⋅ sin ((0.937 ⋅ t)) + 1.33 ⋅ cos ((0.937 ⋅ t))))
⎝ 2 2 ⎠
( )
float , 3 -0.35 ⋅ t
xn4 ((t)) ≔ e -(p ⋅ t) ⋅ ⎛_A ⋅ sin ⎛⎝ω1 ⋅ t⎞⎠ + _B ⋅ cos ⎛⎝ω1 ⋅ t⎞⎠⎞ ――― →e ⋅ ((-1.99 ⋅ sin ((0.937 ⋅ t)) + 1.33 ⋅ cos ((0.937 ⋅ t))))
⎝ 3 3 ⎠
What I did here is not as complex as it looks. I just made up 4 different I.C. to get 4 different null spaces which I
then added to the Xp to get 4 different general solutions for each initial condition. Each of these general solution is
represented by a different color in the plot below. Note how they all seem to converge to the same sine wave as time
progresses...

x1 ((t)) ≔ xp ((t)) + xn1 ((t)) x2 ((t)) ≔ xp ((t)) + xn2 ((t)) x3 ((t)) ≔ xp ((t)) + xn3 ((t)) x4 ((t)) ≔ xp ((t)) + xn4 ((t))

t ≔ 0 , .1 ‥ 20
Page 10 of 12 11/08/2024
 t ≔ 0 , .1 ‥ 20

1.8
1.5
1.2
0.9
0.6
x1 ((t))
0.3
0 x2 ((t))
0 2 4 6 8 10 12 14 16 18 20
-0.3
-0.6 x3 ((t))
-0.9
-1.2 x4 ((t))
-1.5

t
Important Observations!
Ÿ The null space decays to zero with time (because of damping) but the particular solution oscillates at the
forced frequency ω with an amplitude of C ＝ A 2 + B 2 . So as t increases, Xp becomes dominant and the
null space vanishes. For this reason. the null space is termed the transient solution whereas the particular
solution is termed the steady-periodic solution.
Ÿ The initial condition only affects the start region of the graph and as t approaches ∞ , the effects of the
I.C. vanish. This is because the effect of I.C. (arbitrary constants) is only present in the transient solution,
Xtr, and not present in the steady-periodic Xsp solution.
clear ((ω))
s
‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾ s≔ p≔― p = 0.48
F0 2 50
C ((ω)) ≔ ―――――――――――――
m 2 ⋅ ⎛⎝4 ⋅ ω 2 ⋅ p 2 + ω0 4 - 2 ⋅ ω 2 ⋅ ω0 2 + ω 4 ⎞⎠
⎡ 0 ⎤
⎢ ⎥
solve , ω ‾‾‾
337 ⎥
ωcr ≔ C′ ((ω)) ＝ 0 ――― ⎢
→ ―― ωCr ≔ ωcr = 0.734 CCr ≔ C ⎛⎝ωCr⎞⎠ = 1.187
⎢ 25 ⎥ 1
⎢ ⎥
ω ≔ 0 , 0.1 ‥ 5 ⎢ - ‾‾‾
337 ⎥
―――
⎢⎣ 25 ⎥⎦
0.734
1
1.5 Ÿ Critical amplitude decreases as
1.35
1.187
damping (p) increases and vice
1.2 versa. And as damping decreases,
1.05 ωcr approaches ω0 as can be
0.9
seen in the plot below.
0.75
0.6 C ((ω))
0.45
Ÿ Finally, as ω approaches ∞ , C
0.3 approaches zero, meaning that for
0.15 large frequencies, the interference
0 is mostly destructive. Play with
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
plot below!
ω

Page 11 of 12 11/08/2024
 0.3 approaches zero, meaning that for
0.15 large frequencies, the interference
0 is mostly destructive. Play with
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
plot below!
ω

Page 12 of 12 11/08/2024