2 Wave equations and their solution

2.1 Wave equations

e appreciate that wave motions result from the delayed response to an adjacent displacement. But how can we quantify the motion? How can we determine the speed of propagation, the energy contained within a wave, whether the wave shape changes as it propagates, and so on? The answer is that we must first quantify the physics of the wave system, by writing explicit equations that describe how the adjacent disturbance influences a given point and how the response is delayed. For any system, we shall see that it is possible to write a wave equation which embodies the governing physics in the form of a partial differential equation that relates derivatives of the wave displacement (or wave function) with respect to time and position. Although different physical systems have different wave equations, most are similar in form and can be solved using the same techniques. 2.1.1 The Mexican wave equation Let us first consider further the sports-ground phenomenon, in which the displacement at a given point in the crowd depends upon that at an adjacent point, with a delay before it takes effect. Put mathematically,

H. J. Pain, The Physics of Vibrations and Waves, 6th ed., pp.108-112 A. P. French, Vibrations and Waves, pp. 161-167 F. S. Crawford, Waves, pp.48-56

f ( x , t ) = f ( x x , t t )
to which the solutions are

(2.1)

f ( x, t ) = f ( x vt )

(2.2)

where v = x/t. Comparing the differentials of f(x, t) with respect to x and t, we can write the behaviour of our system as a wave equation

df df = v dt dx

(2.3)

Implicit here is that the ripple does not change shape as it propagates around the stadium: the function f(x-vt) takes the same value f() at all points for which x=+vt (where is a constant). v is thus the velocity with which the waveform is moving.

In general, a system exhibiting wave motion may be described by a wave equation relating derivatives (not necessarily the first) with respect to time to derivatives with respect to position. We shall later encounter the electromagnetic wave equation resulting from Maxwells equations and Schrdingers equation for the quantum motion of a free particle. We start, however, by considering the simpler example of a guitar string.

PHYS2023 Wave Physics

11/9/2009

8 Wave equations and their solution 2.1.2 Partial differentiation and partial differential equations With wave motions, as with any other form of dynamics, we are interested in how quantities such as the wave displacement vary with time. The displacement measured at any point in space will therefore be a function of time, and we may indicate this explicitly by writing it as, for example, f(t). This means nothing more than f, of course its just a way of reminding ourselves that it depends upon time. If f(t) were the z-coordinate of a mass m subject to gravity, for example, then Newtons equation for the motion of the mass would be simply

H. J. Pain, The Physics of Vibrations and Waves, pp.107-8 M. L. Boas, Mathematical Methods in the Physical Sciences, Ch 4, 13.

d2 f = mg dt 2

(2.4)

Our waves involve the motions at a large number of different particles at a large number of different points in space. One way of labelling these points (or particles) would be by numbering them and then referring to their displacements as f1, f2 etc. The nth displacement fn(t) would then describe the nth particle at the nth position. The displacement f therefore depends upon both t and n. As the number of points increases, this notation becomes increasingly cluttered, and we choose to identify the particular f by the position of the point or particle, for example

f ( x, t )

(2.5)

The symmetry of this notation suggests that x can be treated in just the same way as t, so to indicate the acceleration of particles at a given point, we specify the (constant) coordinates. Equation (2.4) then becomes

2 f m t 2

= mg x

(2.6)

The full derivative d has been replaced by the partial derivative , to indicate that were considering the dependence of the function f upon only one of its parameters time, t and weve indicated that the other parameter, corresponding to the position coordinate, is held fixed. The function f doesnt need to vary continuously from point to point it could be a series of discrete values but, in the context of waves, it usually turns out to do so. For example, a guitar string might be represented by

f ( x, t ) = a cos kx sin t
2 f t 2 2 = a cos kx0 sin t x0

(2.7)

so that at x = x0, the displacement will be f(x0,t), and the acceleration will be

(2.8)

We have therefore differentiated with respect to t, taking acoskx0 to be a single constant.

11/9/2009

PHYS2023 Wave Physics

Wave equations and their solution 9 We can also take a snapshot of the guitar string at time t = t0: the displacement will be f(x,t0). The gradient of the string at time t0 is then found by differentiating with respect to x, this time treating asint as a constant:

f = a sin kx sin t x t0

(2.9)

One of the clearest ways to picture a function of two variables is as a landscape, whereby the height h of the land above sea-level varies with longitude and latitude, or with the x and y coordinates of the grid-reference. Suppose, then, that we wish to know how much higher is a point 10 m to the northeast of us. We could reach that point by walking ~7 m to the east and then the same distance to the north. The height gained in the first leg would be approximately

h h1 = 7 x y
while the second leg makes a similar contribution,

(2.10)

h h2 = 7 y x

(2.11)

where the x axis is taken to run to the east and the y axis to the north. Strictly, the partial derivative for h2 should be evaluated at x+7 rather than at x, and of course the above expressions assume constant gradients, but as the distances involved are reduced, these become ever better approximations. In the limit of moving infinitesimal distances x and y, we find that the change in height is given exactly by

h h h = h1 + h2 = x + y y x y x

(2.12)

For classical wave motions, there is an associated potential energy, related to the displacement, that varies with position. The force on any element is given by the gradient of the potential energy, and hence by the partial derivative of the displacement with respect to position. The force acts upon the elemental mass to cause an acceleration which is, of course, the partial derivative of the displacement with respect to time. Newtons third law hence relates a partial derivative with respect to time to a partial derivative with respect to position giving us the wave equation for the system! Precisely which partial derivatives, and the values of the constants in the equation, depend upon the details of the physical system, as we shall see in the following sections. For simplicity, it is common to drop the brackets and fixed variable, so that (h/x)y becomes simply h/x. That this is a partial derivative, with the other variables are held constant, is then implicit.

PHYS2023 Wave Physics

11/9/2009

10 Wave equations and their solution

2.2 Waves on long strings

y

2 1
W

x0-x x0+x x x0 Fig. 2.1 Forces acting on a guitar string

Figure 2.1 shows a section of a guitar string, of mass per unit length M kg.m-1, which is under a tension W N and whose displacement from its resting position is described by y. We shall assume that the section is distant from the bridge and fret which mark the free length of the string. The short length between x = x0 - x/2 and x = x0 + x/2 has a mass of M x and experiences a force in the positive-y direction given by W(sin2 - sin1) N. Provided that y varies slowly with x, we may make the approximation sin y/x and hence, applying Newtons third law, write

y 2y y M x = W x x t2 x0 + x 2 x0 x 2 x0

(2.13)

where 2y/t2 is the acceleration of the string element. Taking the limit as x tends to zero, we arrive at the wave equation
equation governing transverse waves on a string Validity: shallow waves on an infinitely flexible string with no frictional losses.

2y 2y W = t2 x2

(2.14)

This equation completely governs the motion of the string, and we shall find that provided we know the initial displacement and motion y(x, t = 0) and y (x, t = 0) then the future motion of the string is entirely predictable. 2.2.1 Solving the wave equation Weighty volumes have been written on the solution of partial differential equations, and there are indeed wave systems which require solutions of some complexity. Fortunately, unless we venture into particularly technical or esoteric systems, we shall usually meet only a few types of wave equation, all of which are quite straightforward to solve. Indeed, the number of versions of the wave equation is so small that we shall quickly learn to recognize the type of solution that is required, and solving the wave equation then merely involves the insertion of a general form of that solution and the derivation of values for the free parameters that are involved. Usually, this simple and pragmatic approach will suffice, but with more complicated systems it is worth remembering that the success of a particular trial form of wave solution may not necessarily mean that no others are possible. New phenomena have been discovered, and Nobel prizes won, by those that have checked sufficiently carefully the initial assumptions and looked for missing solutions.

use physics/mechanics to write partial differential wave equation for system

insert generic trial form of solution

find parameter values for which trial form is a solution

11/9/2009

PHYS2023 Wave Physics

Wave equations and their solution 11 2.2.2 Travelling wave solutions We shall first of all look for travelling waves of the form y(x, t) = y(x vt) given in eqn (2.2). Although this is a function of both x and t, it depends upon the single quantity x vt; the wave shape does not change with time except for being translated a distance vt along the x axis. It simplifies matters to invent a quantity u = x - vt and write y(x - vt) y(u). We may then differentiate y(x vt) using the chain rule

H. J. Pain, The Physics of Vibrations and Waves, p.112

df u f (u ) a du a

(2.15)

In order to make the following mathematics less cluttered, we use the notation in which each derivative of a function of a single variable is indicated by a prime symbol, similar to an apostrophe, following the function. Thus, f(u) df(u)/du, f(u) d2f(u)/du2 and so on. The notation is therefore similar to that which indicates differentiation specifically with respect to time, t. Substituting the travelling wave into eqn (2.14), and differentiating as above,

M v 2 y (u ) = W y (u )
which is valid for any wave shape y(x vt) provided that

(2.16)

v2 = W M
and hence

(2.17)

v = W M , corresponding to forward and backward

travelling waves. The wave speed thus increases with the tension, which determines the coupling of a point to its neighbours, and decreases with the string density, which accounts for the inertia in response. Our wave equation has the important property of linearity that if y1(x,t) and y2(x,t) are solutions, then so is the superposition y = a y1(x,t) + b y2(x,t), where a and b are arbitrary constants. We shall consider linearity at length later on in the course. In this example, we have seen that there are two classes of solution, corresponding to the forward and backward travelling waves. If we define v to be positive, our general solution may therefore be written

y ( x, t ) = y + ( x vt ) + y ( x + vt )

(2.18)

where y+(x,t) is the forward travelling component and y-(x,t) is the backward travelling part. We stress that our analysis so far holds for any wave shapes y+ and y-. Having found our general solution, then, how do we determine the specific solution for a given system? Suppose that we know the initial waveshape and how it is changing at t = 0. This gives us two equations:

y ( x,0) = y + ( x ) + y ( x )

(2.19)

( ) ( ) y (x,0) = y + x vt + y x + vt t t t

(2.20)

PHYS2023 Wave Physics

11/9/2009

12 Wave equations and their solution where application of the chain rule to equation (2.20) gives

y (x,0) = v y + (u + ) + v y (u ) t u + u
where u+ x + vt, u- x vt. At t = 0, then,

(2.21)

y (x,0) = v{y (x ) y + (x )} t

(2.22)

For each point x along the initial waveform, we therefore have two equations (2.18) and (2.22) to define the two unknowns y+ and y-; our initial conditions therefore suffice to provide a unique solution.
y a

t=0

0 L x x0 Fig. 2.2 The plucked guitar string before being released.

y a

t=t

0 L x x0-vt x0 x0+vt Fig. 2.3 The plucked guitar string, a time t after being released.

2.2.3 Example: the plucked guitar string Figure 2.2 shows a guitar string, such as that considered above, which is fixed by the bridge and fret at x = 0 and x = L. The action of plucking the string involves pulling it at a single point so as to displace it through a distance a; the string is then released at time t = 0. The initial displacement y(x,0) is therefore the triangular shape shown in figure 2.2; the initial velocity y/t(x,0) everywhere is zero. With reference to equations (2.18) and (2.22), we see that the forward and backward travelling components are identical and each equal to half of the initial displacement. The motion of the string may now be derived by adding these two components in their respective positions at a later time t, when they will have moved through a distance vt, as shown in figure 2.3. We see that the overall displacement shows a flattening which starts from the point where the string was plucked and which propagates out with the wave speed v. Beyond the flattened region, the displacement remains unchanged a reminder that information about the release of the string cannot propagate faster than the wave can travel. The analysis above works well for the initial motion of the string around where it was plucked, but the eagle-eyed will have spotted that it is not a complete treatment when the wave propagates to the fixed ends of the string. Strictly, the initial condition expressed in equation (2.19) applied only in the range 0xL; beyond this, the string is undefined, and we can only say that y(0) = y(L) = 0. This additional condition turns out to be satisfied provided that y+(u) = -y-(2L-u), and in the example above the two component waves are periodic sawteeth. This is an example of the effect of boundary conditions, which we shall examine in detail in chapter 10.

2.3 Further examples of wave equation derivation

Just as we began our analysis of the motion of the guitar string by considering the physical mechanisms that governed it, so we can analyse other physical systems by beginning with the principles behind their dynamics. As with the guitar string, this allows us to derive a partial differential equation that describes the wave propagation and embodies all the relevant physics. What remains, as before, is the purely mathematical solution of the wave equation.

11/9/2009

PHYS2023 Wave Physics

Wave equations and their solution 13 In the following examples, as with the guitar string, we first determine how the disturbance or displacement at any point is affected by that at adjacent points, and how the physical properties of the system determine how quickly it can respond. 2.3.1 Waves along a coaxial cable Coaxial cables, used for the distribution of electrical signals and varying from microphone cables to television aerial leads, typically comprise a thin copper Q I(x) V(x) b conductor surrounded by a braided copper sleeve with an insulating spacer to a keep them from touching. Like any pair of isolated conductors, these act as a -Q I(x) capacitor that is, an electric field will exist between them whenever one conductor is given an electrical charge with respect to the other. They also, when connected at the end of the cable by further circuitry, exhibit x+x x x inductance, so that the current through the conductors takes a finite time to Fig. 2.4 Voltages and currents along a respond to a change in the voltage applied. These physical phenomena, section of coaxial cable. described by the laws of Gauss and Faraday, are sufficient to determine how the waves corresponding to electrical signals propagate along coaxial cables. We consider a section of coaxial cable, of length x, aligned along the xaxis as shown in figure 2.4, with an inner conductor of radius a and outer conductor of inside radius b. The voltage between the two conductors at any point along the length of the cable is written as V(x); equal but opposite charges of Q(x) are assumed to occur on the two conductors; and the current I(x) is assumed to flow along the outer conductor and return with equal magnitude along the inner. The principle of charge conservation firstly allows us to relate changes in the stored charge to gradients in current: the difference between the current flowing into the section and that leaving it must appear as a change in the charge stored:

d Q( x ) = I ( x ) I ( x + x ) dt

(2.23)

and we may write the charge in terms of the voltage and capacitance of the element, which in turn we write in terms of a capacitance per unit length C:

Q( x ) = (C x ) V ( x )

(2.24)

Inserting equation (2.24) into equation (2.23) and taking the limit as x 0 thus gives

B A

V I ( x ) I ( x + x ) I = = t x x

(x)
I(x) D

(2.25)

We now consider a rectangular loop, such as ABCD in figure 2.5, formed by x+x x x the inner conductor, a parallel section of the outer conductor, and two radii at x and x + x, through which the current is presumed to induce a magnetic flux Fig. 2.5 Magnetic flux resulting from the current in a coaxial cable. (x). Faradays law of induction relates this to the voltages along the radii:

d ( x ) = V ( x ) V ( x + x ) dt

(2.26)

PHYS2023 Wave Physics

11/9/2009

14 Wave equations and their solution and, if we define a self-inductance per unit length L such that

( x ) = (L x ) I ( x )
then we may now combine equations (2.26) and (2.27) to give

(2.27)

I V ( x ) V ( x + x ) V = = x x t

(2.28)

The manipulations of equations (2.26) to (2.28) thus resemble those of equations (2.23) to (2.25) and, overall, we have two equations relating derivatives of the current and the voltage, which we now combine to give a single differential equation for one variable. Differentiating equation (2.25) with respect to time we obtain

2V 2I = x t t 2 2I 2V = 2 t x x

(2.29)

while differentiating equation (2.28) with respect to position gives

(2.30)

allowing elimination of 2I/xt to give

2V 1 2V = LC x 2 t 2

(2.31)

This expression, which is the wave equation for our coaxial cable, is identical in form to the wave equation (2.14) for guitar strings, and the form of its solutions will therefore be mathematically identical. 2.3.2 Waves along air-spaced coaxial cables The capacitance and inductance per unit length are quite straightforward to obtain if the space between the inner and outer conductors is a uniform insulating medium. The simplest case of this is a vacuum, from which the more practical example of an air-spaced cable differs only slightly. We again consider sections of the cable, but now take slices which are thin enough for the electric and magnetic fields to be approximately uniform. To determine first the capacitance per unit length, we consider the slice shown in figure 2.6. The charge -Q on the inner conductor results in a radial electric field E(r), which we integrate over the surface of the dotted surface that defines a concentric cylindrical region of space of radius r. Application of Gauss law then gives

E(r) -Q

x+x x x Fig. 2.6 Radial electric field within a coaxial cable.

E (r )2r x =

(2.32)

where 0 is the permittivity of free space. It follows that the electric field at a radius r is given by

11/9/2009

PHYS2023 Wave Physics

Wave equations and their solution 15

E (r ) =

Q 1 2 x 0 r

(2.33)

The voltage V(x) between the conductors is hence

V ( x ) = E (r ) dr =

Q b 1 dr r 2 x 0 a a Q = [ln r ]ba = Q ln b 2 x 0 2 x 0 a
b

(2.34)

so that

Q =

2 0 V ( x ) x ln b a 2 0 ln b a

(2.35)

Comparison with equation (2.24) hence yields the capacitance per unit length

C=

(2.36)

To determine the inductance per unit length, we consider the slice shown in figure 2.7. The current I induces an azimuthal magnetic field B(r), which is obtained by considering a circle of radius r around the central conductor. According to Ampres law,

B(r) I

B(r )ds =
B (r ) =

(2.37)
x+x x x Fig. 2.7 Azimuthal magnetic field within a coaxial cable.

where 0 is the permeability of free space. It follows that

0 I 2r

(2.38)

and hence the flux through the radial section ABCD in figure 2.5 will be

I b 0 I b 1 dr = x 0 ln ( x ) = x B(r )dr = x 2 a r 2 a a
b

(2.39)

Comparison with equation (2.27) thus yields the inductance per unit length

L=

0 b ln 2 a

(2.40)

The factor 1/LC in the wave equation (2.31) is thus given by

PHYS2023 Wave Physics

11/9/2009

16 Wave equations and their solution

ln b a 1 2 = LC 0 ln b a 2 0 = 1

0 0

(2.41)

= c2
As we shall see shortly, this means that waves travel along air-spaced coaxial cables with the speed of light. 2.3.3 Ocean waves Another fine example of the simple physics behind everyday wave propagation concerns the motion within and on the surface of a body of water. These waves can vary from rapid ripples on the surface of a pond to the spectacular breakers when ocean swell reaches the shore, and in general require a fully three-dimensional nonlinear treatment for their analysis. Here, however, we restrict ourselves to shallow waves whose energy lies principally in the horizontal motion of the water. These include shallow ocean waves, or swell, which are generated initially by the action of wind over the water surface, and can travel for many hundreds of kilometres. We shall here consider only the one-dimensional motion of the oceans, and assume translational symmetry along the horizontal y-axis; this is equivalent to considering waves with roughly linear wavefronts and, as we shall see in later chapters, is not a significant limitation. We therefore consider a surface whose height h above a given datum varies only along the x-axis, as shown in figure 2.8. We then consider thin vertical slices of the ocean, as might be identified by dropping sheets of a thin plastic film from the surface to the sea bed. We arrange these sheets to be uniformly spaced with a separation dx when the sea is calm and the water surface is level, so that the volume of water contained between each pair of sheets is the same. When the surface is perturbed by a wave motion, we find that the variation in water height within any individual slice is small, and that we may approximate the slice to one of constant height, thus overall forming the stepped profile shown in figure 2.8. We assume that the sheets remain flat as the wave progresses (this proves to be an alternative definition of the shallow water regime). As no water can flow through the sheets, the variation in height h(x) must be accompanied by a horizontal motion of the sheets, so that the volume of water remains the same. That is, if the two sheets around x are displaced by 1 and 2 and we consider a column of water of width dy in the y-direction, then

J. Billingham and A. C. King, Wave Motion, pp.109-123 I. G. Main, Vibrations and Waves in Physics, Ch. 13

h(x)

v1

v2

x-x x+x x x Fig. 2.8 Horizontal motion within shallow water waves.

h( x )(x + 2 1 )y = const
Differentiating with respect to time, we hence obtain

(2.42)

d (x + 2 1 )y h + hy 2 t dt

d 2 =0 dt

(2.43)

11/9/2009

PHYS2023 Wave Physics

Wave equations and their solution 17 We now neglect 1,2 in comparison with x and variations of h in comparison with the average h0, divide by x and cancel the common factor y, to give

d 2 d 2 h dt dt = h0 t x

(2.44)

The derivatives d1,2/dt are simply the horizontal velocities of the dividing sheets, shown in figure 2.8 as v1,2. Taking the limit as the slice thickness x 0, we thus obtain

(v v ) h v = h0 2 1 h0 t x x

(2.45)

We thus have a differential equation relating the temporal derivative of one characteristic of the wave the water height to the spatial derivative of another: the horizontal velocity. To derive a second differential equation relating these two properties, we consider the difference in hydrostatic pressure resulting from the variation in height of the water column. Consider the column of water around the sheet h(x) whose velocity we represent by v1, as shown in figure 2.9. At any height z, there will be a difference in hydrostatic pressure across the column due to the difference in height of water above that level:

h1

h2

P1(z) v1 P2(z)

P1 ( z ) P2 ( z ) = (h1 h2 ) g

(2.46)

where is the density of the water and g the acceleration due to gravity. As a result, a net horizontal force acts upon the column of water, resulting in its acceleration. For a layer of depth h, we may therefore write

x-x x x Fig. 2.9 Hydrostatic pressure variations accompanying shallow water waves.

(h1 h2 ) g h y = x y h v
t v (h1 h2 ) h = g g t x x

(2.47)

Cancelling the common factors, dividing by x and taking the limit x 0, we thus reach our second differential equation: (2.48)

We now combine the two relations between h and v (we may choose to eliminate either) to give the equation for shallow ocean waves:

2h 2v 2h = h = gh 0 0 xt t 2 x 2

(2.49)

We shall see that shallow ocean waves thus travel with a speed that increases with the square-root of the water depth,

V = g h0

(2.50)

PHYS2023 Wave Physics

11/9/2009