491422597.

doc
The vertical displacement u(x,t) of a string stretched along the x-axis between x = 0 and x = L is
given by

n
u(x,t) = 
n 1
sin
L
x[ bn cos( n t) + bn* sin( n t)] (3.2-6)

This displacement is composed of an infinite linear combination of simple product solutions

n cn
un(x,t) = sin x[ bn cos( n t) + bn* sin( n t)], where n =
L L

These un(x,t) are called the normal modes of vibration. The intensity of the sound produced
depends on the amplitude bn2   bn*  since
2

bn cos( n t) + bn* sin( n t) = bn2   bn*  sin( n t + ), where  = tan-1( bn / bn )

2 *

cn
The time dependence is simple harmonic with circular or natural frequency ( n = ) that
L
gives the number of oscillation in 2 units of time. The normal mode n = 1 is called the first
harmonic or fundamental. The larger the natural frequency, the higher the pitch of the sound
produced. The “wave velocity” c =  /  can be varied to produce a desired fundamental
frequency. In this expression  is the tension of the string and  is the mass density per unit
length. Normally, the mass density is fixed. Thus, the instrument is tuned by varying the tension
; the larger , the higher the pitch. Clamping down the string to change the effective length L
can also vary the pitch of a stringed instrument. Shortening L makes the notes higher.

Which initial conditions cause the string to vibrate in a normal mode? A string with initial shape
mx
f(x) = sin( ) and zero initial velocity g(x) = 0 will vibrate in normal mode. This can be seen
L
by evaluating the coefficients bn and bn* from the solution


n
u(x,t) = 
n 1
sin
L
x[ bn cos( n t) + bn* sin( n t)]

2 L nx
bn* =
cn 0
g ( x ) sin(
L
)dx = 0

2 L nx 2 L  m x  nx
bn =
L 0
f ( x ) sin(
L
)dx =
L 0
sin
 L 
 sin(
L
)dx = 0 for m  n

For m = n

2 L n x 2 L 1  2nx  
bn =
L 0
sin 2 (
L
)dx =
L 0
1  cos
2  L 
 dx

45
491422597.doc
L
1 1 L  2nx 
bn = L sin   =1
L L 2 n  L 0

The string displacement in this particular case is then

 nx 
u(x,t) = un(x,t) = sin   cos( n t)
 L 

This is the normal mode of vibration.

Damped vibrations of a string


u (x ,t) u (x ,t) B 

A


0 L x 0 x x + x x
Figure 3.2-1 The shape of the string and the forces acting on a small portion of it.

Figure 3.2-1 shows the transverse displacement u(x,t) of a string stretched along the x-axis
between x = 0 and x = L. The string is free to vibrate in a fixed plane through the x-axis. We
assume that there is a resistance force proportional to the velocity on the small portion x of the
u
string as 2kx , k is a positive constant. The force balance in the vertical direction becomes
t

u  2u
  sin  +  sin   2kx = may = x 2
t t

For small angle  and 

u u
sin   tan  = (x, t) and sin   tan  = (x + x, t)
x x

u u u  2u
 (x, t) +  (x + x, t)  2kx = x 2
x x t t

u u
 ( x  x , t )  ( x, t ) u  2u
 x x  2k =
t t 2
x
As x  0

  2u u  2u  2u u 2  u
2

 x
 2k =  + 2k = c , where c2
= 
2
t t 2
t 2
t x 2

46
491422597.doc
The method of separation of variables will be used to solve the one dimensional wave equation
that describes damped vibration of a string stretched along the x-axis between x = 0 and x = L.

 2u u 2  u
2
+ 2k = c 0 < x < L, t > 0 (3.3-1)
t 2 t x 2

The boundary and initial conditions required for the solution of the wave equations are

B.C. : u(0,t) = 0 and u(L,t) = 0, for t  0

u
I.C. : u(x,0) = f(x) and (x,0) = g(x), for 0 x  L
t

f(x) and g(x) are known initial position and initial velocity of the string, respectively.

f(x) and g(x) are known initial position and initial velocity of the string, respectively.

We assume that u(x,t) can be separated into X(x), a function of x alone, and T(t), a function of t
alone. In other words, u(x,t) will have the form

u(x,t) = X(x) T(t) (3.3-2)

If the assumption is valid, substitution of u(x,t) in terms of X(x) and T(t) into the original PDE
will result in two ODEs. One ODE is a function of x only and the other ODE is a function of t
only.

From Eq. (3.3-2)

u dT  2u d 2T
=X  = X
t dt t 2 dt 2

u dX  2u d2X
=T  = T
x dx x 2 dx 2

In terms of X(x) and T(t), Eq. (3.3-1) becomes

d 2T dT 2
2 d X
X + 2k X = c T (3.3-3)
dt 2 dt dx 2

Divide Eq. (3.2-3) by c2XT to obtain

1 d 2T 2k dT 1 d2X
+ = =  2 = constant (3.3-4)
c 2T dt 2 c 2T dt X dx 2

Since the RHS of Eq. (3.3-4) depends on x only, and the LHS depends on t only, they must equal
to a constant  2. The constant must be negative for non-trivial solution.

47
491422597.doc
The boundary conditions on u(x,t) are changed to the boundary condition on X(x)

u(0,t) = 0  X(0) T(t) = 0  X(0) = 0

u(L,t) = 0  X(L) T(t) = 0  X(L) = 0

From Eq. (3.3-4), the ODE with respect to x is

d2X
=  2X  X = C1cos(x) + C2sin(x)
dx 2

The constant C1 and C2 can be determined from the boundary conditions

At x = 0, X = 0 = C1(1) + C2(0)  C1 = 0

At x = L, X = 0 = C2sin(L)

To avoid the trivial solution, C2  0 and

n
sin(L) = 0  L = n  n =
L

From Eq. (3.3-4), the ODE with respect to t is

1 d 2T 2k dT
2 + 2 =  2
c T dt 2
c T dt

This equation is rearranged to

d 2T dT
+ 2k + (c)2T = 0 (3.3-5)
dt 2
dt

Eq. (3.3-5) is a second order ODE with constant coefficients. The solution can be obtained by
assuming a solution of the form

T = et

The first and second derivative are then evaluated

dT d 2T
=  et  = 2 et
dt dt 2

The function T = et and its first and second derivatives are then substituted into the original ODE
(3.3-5) to obtain

(2 + 2k + (c)2) et = 0

48
491422597.doc

The constant  is then the solution of the algebraic equation

2 + 2k + (c)2 = 0

 =  k  { k2  (c)2}1/2

There are three cases to consider depending on the value of the expression {k2  (c)2}

Case 1: k2  (c)2 > 0

nc kL
k > c  k > , or n <
L c

The solution of the ODE (3.3-5) is then

2
 n 
Tn = An e (  k  n ) t + Bn e (  k n ) t , where n = k2   c
 L 

Tn = e  kt {An e n t + Bn e  n t }

The solution can also be expressed in terms of the hyperbolic functions

e nt  e  nt e n t  e  nt

sinh(n t) = , and cosh(n t) =
2 2

Tn = e  kt {an cosh(n t) + bn sinh(n t) }

Case 2: k2  (c)2 = 0

 =  k  { k2  (c)2}1/2 =  k

Tn = An e  kt

The above function only contains one term and one arbitrary constant therefore it is not the
complete solution of a second order ODE. A new trial solution of Eq. (3.3-5) is required.

Let T = V(t)et, where V(t) is a function of t

dT dV
= et + V(t)et
dt dt
2 2 2
d T t d V t dV t dV t d V dV
= e +  e +  e +  2
V(t)e t
= e + 2et +
dt 2
dt 2
dt dt dt 2
dt
2V(t)et

The function T = V(t)et and its first and second derivatives are then substituted into the original
ODE
49
491422597.doc

d 2T dT
+ 2k + (c)2T = 0 (3.3-5)
dt 2
dt

to obtain

d 2V dV dV
{ + 2 + 2V(t) + 2k + 2kV(t) + (c)2V(t)}et = 0
dt 2
dt dt

d 2V dV
+ (2+ 2k) + [2 + 2k + (c)2]V(t) = 0
dt 2
dt

Since [2 + 2k + (c)2] = 0 and (2+ 2k) = 0, ( =  k for double roots),

d 2V
= 0  V(t) = an + bn t
dt 2

n kL
Tn = (an + bn t) e  kt , where k = c = c or n =
L c

Case 3: k2  (c)2 < 0

nc kL
k < c  k < , or n >
L c

2
 n 
k2  (c)2 = i2[(c)2  k2] = i2(n)2, where n = k2   c
 L 

Tn = An e (  k in ) t + Bn e (  k in ) t

Tn = e  kt {An e int + Bn e  in t }

The solution can also be expressed in terms of the trigonometric functions

e int = cos(n t) + isin(n t) , and e  in t = cos(n t)  isin(n t)

Tn = e  kt {an cos(n t) + bn sin(n t)}

kL
We will now consider the case when n  . The string displacement for damped vibration is
c
then


u(x,t) = 
n 1
Xn(x) Tn(t)

50
491422597.doc

u(x,t) = e  kt  sin
n
x {an cosh(n t) + bn sinh(n t) }
1 n 
kL L
c


+ e  kt  sin
n
x {an cos(n t) + bn sin(n t) }
kL
n   L
c

At the initial condition t = 0, u(x,0) = f(x)

f(x) =  sin
n
x {an + 0} +  sin
n
x { an + 0}
1 n 
kL L kL
n   L
c c

n
f(x) = 
1n  
an sin
L
x

2 L nx
an =
L 0
f ( x ) sin(
L
)dx , where n = 1, 2, 3, …

u
At t = 0, (x,0) = g(x)
t

u
= k e  kt { sin
n 
x [an cosh(n t) + bn sinh(n t) ]
t 1 n 
kL L
c


+  sin
n
x [an cos(n t) + bn sin(n t) ]}
kL
n   L
c

+ n e  kt {  sin
n
x [an sinh(n t) + bn cosh(n t) ]
1 n 
kL L
c


+  sin
n
x [ an sin(n t) + bn cos(n t) ]}
kL
n   L
c

At t = 0 g(x) = k{  sin

n
x [an + 0 ] +  sin
n
x [an + 0]}
1 n 
kL L kL
n   L
c c

+ n {  sin
n
x [0 + bn] +  sin
n
x [0 + bn ]}
1 n 
kL L kL
n   L
c c

n
g(x) = 
1n  
(k an + n bn) sin
L
x

2 L nx
k an + n bn =
L 
0
g ( x ) sin(
L
)dx , where n = 1, 2, 3, …

51
491422597.doc

2 L nx
bn can be determined from the above equation since a n =
L 
0
f ( x ) sin(
L
)dx

52