Acoustics Problem Set MUSI11076: Strings and Bars 1

Acoustics

Problem Set MUSI11076: Strings and Bars

1 Impedance of the ideal string

Consider a string fixed a t x = L a nd d riven b y a n i nput f orce f˜(t) = F 0 ejωt a t x = 0 , w here F 0 i s a
real amplitude and ω is the driving frequency. The input impedance of the string can be defined as the
complex ratio of the driving force to the resulting velocity at the same input point. Therefore in this
case:

f˜(t)
Z̃in = (1)
ṽ(0, t)

In general, Zin is a function of temporal frequency ω.

We know that in the steady state the string will undergo simple harmonic motion, with a complex
displacement ỹ(t) of the form

ỹ(x, t) = Ãej(ωt+γx) + B̃ej(ωt−γx) (2)

where à and B̃ are complex constants to be determined, and as usual γ is the (angular) spatial frequency
(wavenumber).
From conservation of energy, we also know that the driving force must be equal and opposite to the
restoring force in the string at that point. In other words:

∂ ỹ(0, t)
f˜(t) = −T0 (3)
∂x
(a) [3] Using the above information show that the input impedance Zin for this finite ideal string can
be written as
 
T0 ωL
Z̃in = −j cot . (4)
c c

You will first need to find appropriate relationships between à and B̃, which will lead you to
an expression for ỹ(x, t) in terms of only F0 , T0 , L, γ, ω, x and t. From here, substitute your
expression, with appropriate derivation, into the equation for the input impedance.
ωL
(b) [2] Sketch the magnitude of the input impedance |Zin | over the range c ∈ [0, 4π]. At what
frequencies fn are the system’s resonances (for integer n)?

(c) [1] What happens to the input admittance at resonance? Comment on the physical significance
of this finding.

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]

Acoustics Problem Set MUSI11076: Strings and Bars 2

2 A plucked string
In class, we managed to write out the general solution to the fixed-fixed st ring as
∞  
(m) (m)
X
y(x, t) = C+ ejωm t + C− e−jωm t Xm (x) (5)
m=1

where the modal shapes Xm are given by

mπx
Xm (x) = sin (6)
L
(m) (m)
Assume the string to be defined for x ∈ [0, L]. We said that the coefficients C+ , C− can be determined
from the initial conditions. To that end, you first need to prove that the modes are mutually orthogonal.
Hence, define the inner product and norm as, respectively,
Z L p
Inner Prod: ⟨Xm , Xn ⟩ = Xm Xn dx, Norm: ∥Xm ∥ = ⟨Xm , Xm ⟩ (7)
0

a) [5] Start from the ordinary differential equation for Xm , obtained after separation of variables as
seen in class:
2 ′′
−ωm Xm = c2 Xm (8)

Take an inner product of this equation with Xn , and integrate by parts the right-hand side twice,
using appropriate arguments regarding the boundary terms. Show that, necessarily,

⟨Xm , Xn ⟩ = δn,m ∥Xm ∥2 (9)

where δn,m = 0 if n ̸= m, and δn,m = 1 if n = m. Compute the explicit expression for the norm
∥Xm ∥.

b) [2] Initial conditions are now given for y(x, t). Let the initial velocity be

∂y(x, 0)
v0 (x) ≜ =0 (10)
∂t

Use this condition to prove that

∞
(m)
X
y(x, t) = 2C+ cos(ωm t)Xm (x) (11)
m=1

c) [5] Let the initial displacement be

(  
2π(x−L/4)
1 − cos L/2 if L/4 ≤ x ≤ 3L/4
y0 (x) ≜ y(x, 0) = (12)
0 elsewhere

This kind of function is known as raised cosine, and it may approximate a plucked string: plot it
in a graph. Using the orthogonality property (9) into (11), show that

(m) ⟨y0 , Xm ⟩
C+ = (13)
2 ∥Xm ∥2

(m)
Give an analtyic expression for C+ , and compute the values for m ∈ [1, 8], using L = 1.
Comment briefly about your result, in light of the symmetry of the initial condition y0 (x).

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]

Acoustics Problem Set MUSI11076: Strings and Bars 3

3 Modal frequencies of the ideal bar: Clamped–Simply supported

Consider the ideal bar equation

∂2y 4
2∂ y
= −ν (14)
∂t2 ∂x4
defined over the interval x ∈ [0, L], where we have substituted ν = κcL . In this problem, we will
examine the modes of vibration of this system. We assume time harmonic behaviour for the complex
transverse displacement y(x, t) = Y (x)ejωt . Substituting this into the equation above gives the ordinary
differential equation
d4 Y ω2
= Y (15)
dx4 ν2
as we saw in class. The general solution to this equation is
Y (x) = A sin(γx) + B cos(γx) + C sinh(γx) + D cosh(γx) (16)
p
where γ = ω/ν is the (angular) spatial frequency, and where the constants A, B, C and D are as
yet undetermined. We will find the values of these four arbitrary constants by considering a set of four
boundary conditions (two at either end), which determine how the bar is physically supported. Various
combinations of boundary condition are possible, as we discussed in class.
Finding these arbitrary constants will lead us to a quantised (i.e. discrete) set of allowed values of γn ,
indexed by mode number n.

(a) [2] Consider now an ideal bar that is clamped at its left end, and simply supported at its right
end. These boundary conditions can be summarised as
dY d2 Y
Y (0) = 0 =0 Y (L) = 0 =0 (17)
dx x=0 dx2 x=L

Find an implicit equation from which the allowed spatial frequencies γ, may be determined. This
equation should involve only the tangent (tan), and hyperbolic tangent (tanh) functions.
You should begin by applying the relevant boundary conditions, before simplifying your resulting
expressions until you reach a compact formulation, as noted.
(b) [2] On a single diagram, sketch and carefully label each term in the implicit equation from (a).
Your horizontal axis should be the quantity γL, with units of radians, over the range 0 : 3π.
Identify and label the discrete values of γL at which a solution to the implicit equation occurs,
which you should label γn L in terms of increasing n. These are, of course, the ‘allowed’ spatial
frequencies for the given boundary conditions.
You may use Matlab to evaluate and explain the problem numerically, if you wish (if you do so,
please comment on how you did this, and include a print out of your code).
(c) [1] What values do the allowed solutions (i.e. the terms in your implicit equation from (a))
approach as γn L becomes large?
(d) [1] Using your answer to (c), find an approximate expression for the allowed spatial frequencies,
γn , for large n.
(e) [2] Find an expression for the allowed temporal frequencies fn of the bar (in Hz), for large n.
(f ) [2] Construct a general expression for the time harmonic mode shapes of the bar that involves at
most one arbitrary constant.
Your expression should be of the form Yn (x) = ..., i.e. you are looking for the dependance of the
bar shape with position x along the bar, and wavenumber γn . The nth expression hence describes
the spatial distribution of the nth mode.
(g) [2] Consider the behaviour of your expression from part (f ) for large values of γn x. What is the
approximate shape of the bar under this condition? Comment on the form of your answer.

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]

Acoustics Problem Set MUSI11076: Strings and Bars 4

4 Dispersion in 1D
a) [2] Can you write down a dispersion relation, for a 1D string-like system, for which the group
velocity vg is a constant for all frequency, but for which the phase velocity vϕ is not a constant?
Justify your answer carefully, including diagrams where appropriate.

b) [2] Deduce a hypothetical PDE, in time t and space x, which possess the dispersion relation that
you specified in part a).

5 Modal density and sampling density for the ideal bar

Consider the ideal bar equation
∂2y 4
2∂ y
= −ν . (18)
∂t2 ∂x4
If the bar is constrained to be of length L, under simply supported conditions at both ends, we know
that the angular temporal mode frequencies ωn are found to be

νπ 2 n2
ωn = n = 1, 2, . . . (19)
L2

Furthermore, examining a test solution y(x, t) = ejωt+jγx , corresponding to a sinusoidal travelling wave
of angular temporal frequency ω, and angular spatial frequency γ, we find the characteristic equation
to be
ω = νγ 2 (20)

a) [1] Find an expression for the modal frequencies fn in units of Hz.

b) [1] Find an approximate expression for the number N of modes which lie in the range between 0
Hz and some arbitrary frequency fmax Hz.

c) [2] Supposing that ν = 10, L = 0.5, (typical parameters for a musical bar) and fmax = 20000,
approximately how many modes are present? This corresponds to the range of human hearing,
and gives a nice estimate of how much computational work will be required in order to compute
a solution numerically!

d) [2] Rewrite equation 20 in terms of frequency f (in Hz) and wavelength λ (in metres).

e) [2] What is the minimum wavelength λmin over the range of frequencies between 0 Hz and some
arbitrary frequency fmax Hz? Comment on the relationship between λmin and fmax implied by
your result.

f ) [2] Suppose we now have a bar of arbitrary length L, and we wish to represent the spatial solution
on a grid of equally spaced points (this is the beginning of simulation design!). From Nyquist’s
sampling theorem, we know that to adequately render the solution up to a given frequency fmax ,
we must have at least two grid points along the bar for the shortest wavelength we need to render.
What is the number N of grid points required for a bar of length L?

6 Piano strings: Dispersion and tuning

In the lecture and tutorial we discussed the importance of stiffness in real s trings. The effect of stiffness,
as described by the ideal bar equation, can be combined with the effect o f t ension, a s d escribed i n the
1D wave equation, to give a so-called ‘stiff s tring’ e quation. T his e quation m ay b e w ritten as

∂2y 2
2∂ y
4
2∂ y
= c − ν (21)
∂t2 ∂x2 ∂x4

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]

Acoustics Problem Set MUSI11076: Strings and Bars 5
p p
where c = T /ρ and ν = QAκ2 /ρ (i.e. these constants are defined here as they appear separately
in the wave and bar equations; but note that ρ is the linear density measured in units of kgm−1 , which
is why the extra A term appears in ν, in order to convert this to volume density). This equation is
second order in time, and fourth order in space (due to the presence of stiffness).

a) [1] Using a harmonic trial solution of the form y(x, t) = ejωt ejγx , show that the dispersion relation
that relates the angular frequency ω and wavenumber γ (i.e. seek ω =⇒ ω(γ)) may be written
as
r  νγ 2
ω(γ) = ±cγ 1 + . (22)
c

b) [2] Find expressions for the phase velocity vϕ (ω) and group velocity vg (γ), in terms of ω, care-
fully showing your working. Carefully show your working, and comment on the nature of the
relationships you find.

c) [2] Under simply supported boundary conditions over x ∈ [0, L], the allowed wave numbers are
constrained to a discrete set γn = nπ
L . Find a compact expression for the allowed modal (time)
(stiff-string)
frequencies fn in terms of the constants L, T , ρ, Q, r, π, and n (which is just the set
of positive integers). Assume that the string is circular in cross section, with a radius r. Your
equation should be of the form

fn(stiff-string) = fn(ideal-string) · g (23)

(ideal-string)
where fn describes the modal frequencies of an ideal string (which you should include
explicitly), and g is a (real valued) term that parameterises the difference between the mode
frequencies of an ideal string, and those of a stiff string.

d) [2] For a given string geometry and density, discuss the roles played by T and Q in determining
the degree to which the natural frequencies of a stiff string differ from those of a non-stiff (ideal)
string. Is this what you would expect?

e) [2] In a real piano string, some typical values for the string parameters are given by T = 250N,
ρ = 10−2 kgm−1 , and Q = 2x1011 kgm−2 . A typical string radius (assuming a uniform circular
cross-section) is r = 0.0004m. Assume that the string is simply supported at each end, and has
a length of 1m.
By how many cents does the 10th mode of this stiff piano string deviate from the frequency to be
expected from an ideal string without stiffness (assuming other string parameters are unchanged)?
Is this likely to be audible? Be careful to show all your working.

f ) [1] What about if the string length is doubled, the linear density ρ is doubled, and the string
radius is doubled (as might be the case for a longer, heavier and thicker string)? How many cents
of deviation are there now, at the 10th harmonic?

7 Energy storage and dissipation within the boundary of an ideal

bar
Consider the power balance for a semi-infinite bar (x ∈ [0, ∞]), which can be written as

∂y(0, t) ∂ 3 y(0, t) ∂ 2 y(0, t) ∂ 2 y(0, t)

 
d
(Ek + Ep ) = (QAκ2 ) − . (24)
dt ∂t ∂x3 ∂t∂x ∂x2

In class we discussed the three most well known boundary conditions that lead to lossless conditions
d
(i.e. zero power), such that dt (Ek + Ep ) = 0 (by implication Ek + Ep = E where E is a constant).
These were “simply-supported”, “clamped”, and “free”.

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]

Acoustics Problem Set MUSI11076: Strings and Bars 6

A number of other types of boundary are also possible, such as where the boundary itself either stores
or dissipates energy.
For the following boundary conditions (defined at x = 0), show that the combined energy of the bar
and the boundary (at x = 0) remains constant in time (i.e lossless). We can write this total energy as
E + Eb , where the Eb term describes the energy stored within the boundary. State clearly the form of
Eb in each case.
∂ 2 y(0,t) ∂ 3 y(0,t)
(a) [2] ∂x2
= 0 and ∂x3
+ αy(0, t) = 0 for some constant α ≥ 0.
∂ 2 y(0,t)
(b) [2] y(0, t) = 0 and ∂x2
− α ∂y(0,t)
∂x = 0 for some constant α ≥ 0.
∂ 2 y(0,t) ∂ 3 y(0,t) 2 y(0,t)
(c) [2] ∂x2
= 0 and ∂x3
+ α∂ ∂t2
= 0 for some constant α ≥ 0.

In addition, show that for the following boundary conditions the boundary at x = 0 dissipates energy
d
over time; in other words dt (Ek + Ep ) = B ≤ 0. Show clearly the form of B.

∂ 2 y(0,t) ∂ 3 y(0,t)
(d) [2] ∂x2
= 0 and ∂x3
+ α ∂y(0,t)
∂t = 0 for some constant α ≥ 0.

Semester 1 [UoE DRPS: MUSI10118 & MUSI11076]