MCEN3005 Lecture 2 Notes
MCEN3005 Lecture 2 Notes
MCEN3005 Lecture 2 Notes
Lecture 2
TORSIONAL OSCILLATIONS AND
SYSTEMS WITH TWO DEGREES OF FREEDOM
Main Concepts
If the rotor is displaced by angle θ from its equilibrium position, then the restoring torque will
be:
T = −kθ θ …(2.1)
− kθ θ = Iθ
kθ
i.e. θ + θ =0 …(2.2)
I
kθ
θ (t ) = A sin t + φ
I
1 kθ
and the natural frequency is: fN = …(2.3)
2π I
We note that the differential equation and hence the natural frequency is of the same form as it
was for the translational case, equations (1.15) and (1.16). The analogy between the two is
direct.
T Gθ
=
J l
T GJ
Therefore the torsional stiffness, kθ = = …(2.4)
θ l
πd 4
where J, the polar second moment of area, for a circle about its centre is J = .
32
1
Any text on Mechanics of Materials. For example: Gere, J.M. and Timoshenko, S.P. Mechanics of Materials, 4th
Ed. 1999, Pub. Stanley Thornes.
T
d3
d2
d1
l1 l2 l3
Figure 2.2
liT 32liT
θ = θ 1 + θ 2 + θ 3 + + θ n where θ i = =
J i G πd i4 G
n
32T li
=
πG
∑d
i =1
4
i
T πG
∴ kθ = = …(2.5)
θ n
l
32∑ i4
i =1 d i
Note that in reality, there is no such thing as a perfectly rigid 'built-in end' or 'encastré'. The
support, which also supports the torque, will rotate a little and hence lower the torsional stiffness
compared to that calculated in (2.5).
TL πd 4
θ= where J = .
GJ 32
T dx 32T dx
dθ = = .
G J x πG d x 4
d1 d
d0
Now d x = d 0 + 2 x tan α
L
32T dx
∴θ =
πG ∫ (d
0 + 2 x tan α )
4
0
dx x
L
Figure 2.3
L
− 1 32T 1
∴θ =
6 tan α πG (d 0 + 2 x tan α )3 0
d1 − d 0
Putting tan α = ,
2L
− 2l 32T 1 1
θ= − 3 where d 0 + 2 L tan α = d1
6(d1 − d 0 ) πG (d 0 + 2 L tan α ) d 0
3
T 3πG d 03 d 13
Hence the torsional stiffness is: kθ = = …(2.6)
θ (
32 L d 12 + d 1 d 0 + d 02 )
Some torsional systems consist of a shaft fixed at both ends, carrying an inertia. What is the
stiffness of the shaft in this case? We will also require this result later on for another analysis.
x y
l
Figure 2.4
GJθ GJθ 1 1
T = Tx + Ty = kθxθ + kθyθ = + = GJθ +
x y x y
T 1 1
∴ kθ = = GJ + …(2.7)
θ x y
In order to plot the stiffness as a function of the location x of the applied torque, it is better to
change the form a little:
1 1 1 1 l
y = l − x and hence + = + = . Dividing numerator and denominator by l2:
x y x l − x xl − x 2
1 1 1 1
+ = ⋅
x y l x x 2
−
l l
GJ 1 GJ x
∴ kθ = ⋅ = Ψ …(2.8)
l
2
l x x l
−
l l
25
20
Relative torsional stiffness
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L
We note that the minimum stiffness occurs when the torque is applied at the centre of the span
4GJ
where kθ = .
l
In situations where the inertia of the shaft is significant, an improved model can be developed
that takes into account the shaft's inertia.
The approach is identical to that of section 1.11 where the translational system was considered.
It is recommended that you attempt the analysis by analogy with section 1.11 before reading the
following.
ω
Mode shape
ω Ω
x dx
Figure 2.6
l
I
IS
Note that I'll use ω here for rotational velocity rather than frequency. For frequency I'll use
2πf N .
The system is shown in Figure 2.6 at the instant it passes through its static equilibrium position.
This ensures that there is no strain energy stored in the shaft. Consideration of the kinetic energy
will then reflect the total system energy. Assign Is to the inertia of the 'heavy' shaft. Also,
assume that the distribution of angular velocity along the shaft is linear as shown in the diagram.
This is equivalent to assuming the 'mode shape'.
1 2
Kinetic energy of the disk only = IΩ
2
1 dx
Kinetic energy of the element dx of the shaft = I s ω 2
2 l
ω Ω xΩ
2
1 I 2
Total maximum kinetic energy = I + s Ω …(2.9)
2 3
Now consider the instant when the mass is stationary at the point of maximum deflection, Θ.
The kinetic energy will be zero and the strain energy will then reflect the total system energy.
Assuming simple harmonic motion, θ = Θ sin( 2πf N t ) , the maximum amplitude will be Θ and the
maximum velocity will be Ω = θ = 2πf Θ . N
1
Maximum strain energy = kθ Θ 2 …(2.10)
2
1 1 I
kθ Θ 2 = I + s (2πf N Θ) 2 …(2.11)
2 2 3
giving
1 kθ 1 kθ
fN = = …(2.12)
2π I 2π I eq
I+ s
3
Is
or I eq = I +
3
Hence, the natural frequency can be estimated by adding one third of the shaft inertia to the
disk's inertia and then proceeding as for the massless shaft case.
Many rotating machines consist of a shaft with inertias at each end. For example, an electric
motor driven pump has the motor rotor at one end and the pump impellor at the other. A
turbocharger has a shaft connecting a turbine rotor and a compressor impellor. Such machines
have inherent natural vibration modes that must be known at the design stage so that the
operating torques do not excite them during operation. We can build on the knowledge gained
for a single degree of freedom system to model such a two-rotor system.
Figure 2.7
f N A = f NB
1 GJ 1 GJ
=
2π I A a 2π I Bb
IB
a= l …(2.13)
IA + IB
1 GJ
Now since f N = f N A = f N B = , substituting (2.13) we obtain:
2π I Aa
1 GJ I A + I B
fN = …(2.14)
2π l I AIB
GJ
or, alternatively, as = kθ ,
l
1 I +I 1 1 1
=fN =
kθ A B kθ + …(2.15)
2π I AIB 2π I A IB
Note that kθ is the torsional stiffness of the entire shaft between rotors A and B. Consequently,
(2.15) is valid regardless whether or not the connecting shaft is uniform.
If the shaft is uniform and I A = I B , it is evident from (2.13) that the node occurs in the centre of
the shaft. When the rotors have unequal moments of inertia, the node occurs nearer the larger
inertia. When the shaft is not uniform, but I A = I B still applies, the node divides the shaft into
two portions of equal stiffness. This follows from:
1 kθ A 1 kθ B
fN = = , ∴ kθ A = kθ B …(2.16)
2π IA 2π IB
The analogous development for a translational system is identical, generating the following
expression.
k
1 m A + mB
mA mB fN = k
2π m A mB
This section will not be examined but shows how to account for the inertia of the shaft. Follow
the analysis and understand the principles involved.
Again considering the shaft to consist of two 1dof systems vibrating at the same frequency;
ka
ωN = l
A
Ia
IA +
3 a b
GJ Gπd 4 d
where ka = = Node
a 32a Mode
Ia =
πd 2 ( )
d
aρ 2 =
2
πad 4 ρ IA Shape
4 2 32 IB
Figure 2.8
Gπd 4
so that ωN = …(2.17)
A
πad 4 ρ
32a I A +
96
Gπd 4
Similarly ωN = …(2.18)
B
πbd 4 ρ
32b I B +
96
πad 4 ρ πbd 4 ρ
a I A + = b I B + …(2.19)
96 96
Also b = l − a , which when substituted into (2.19) and solving for a gives:
πd 4 l 2 ρ
I Bl +
a= 96 …(2.20)
πd 4 lρ
IA + IB +
48
πd 4 l 2 ρ
I Al +
Similarly b= 96 …(2.21)
πd 4 lρ
IA + IB +
48
Hence the natural frequency can be calculated from (2.17) where a is given by (2.20) or from
(2.18) where b is given by (2.21). The results will be identical.
Review Questions
3. Are the torsional stiffnesses of the three sections of shaft in Figure 2.2
in series or parallel?
4. Compare the mass of a shaft (φ25 x 700 long) with a solid disk (φ250
x 70 long). Now compare their moments of inertia about their axes.
Can you see why the moment of inertia of a shaft rarely needs to be
included when calculating natural frequencies of shaft/rotor systems?
5. What is a node?