ME 563 - Intermediate Fluid Dynamics - Su: Lecture 29 - Waves: More Basics

ME 563 - Intermediate Fluid Dynamics - Su
Lecture 29 - Waves: more basics
Reading: Acheson, §3.1-3.3.
In the last lecture, we started to consider the problem of two-dimensional surface waves on deep
water, for which the fluid motion is driven by gravity. We defined the free-surface position as
y = η(x, t),
with the implicit assumption that y = 0 is the position of the undisturbed free surface. Thus η(x, t)
can be interpreted as the free surface displacement due to the wave motion; we will also refer to
η(x, t) as te ‘disturbances.’ The following relation pertains between the components of the fluid
velocity, u and v, on the surface, and the free-surface position y = η(x, t):
∂η ∂η
+u =v on the surface. (1)
∂t ∂x
Imposing the condition that the disturbances, η, the velocity components, u and v, and their
derivatives be small, we linearized (1) to get
∂φ ∂η
v(x, 0, t) = (x, 0, t) = , (2)
∂y ∂t
(φ is the velocity potential) which says that the ‘speed’ at which the free surface moves in the
y-direction is equal to the v-component of the fluid velocity at y = 0.
1 Bernoulli’s equation for unsteady irrotational flow

We can get another free-surface condition relating the disturbances η to the velocity field by looking
at the following form of Euler’s equation for the velocity of an ideal fluid –

∂u p 1 2
+ (∇ × u) × u = −∇ + |u| + χ , (3)
∂t ρ 2
where χ is the velocity potential. (Recall that g = −∇χ, and since g = −gêy for the present
problem, we have χ = gy.) since the flow in this problem is irrotational, we have also
ω =∇×u=0
u = ∇φ
which, inserted in (3), gives

∂ p 1 2
∇φ = −∇ + |u| + χ ,
∂t ρ 2
which can be written, interchanging the spatial and time derivative on the left side,

∂φ p 1 2
∇ = −∇ + |u| + χ . (4)
∂t ρ 2
We can get rid of the gradient operator ∇ in (4) by integrating. Remember that the gradient can
be interpreted
dφ
∇φ · ds = ds = dφs
ds
1
where ds is an arbitrary vector with infinitesimal magnitude ds. To integrate (4), then, we first
take the dot product of both sides with some vector ds. The integral of both sides can then be
written

∂φ p 1 2
d = −d + |u| + χ , (5)
∂t ρ 2
which gives

∂φ p 1 2
=− + |u| + χ + G(t), (6)
∂t ρ 2
where the arbitrary ‘constant’ of integration G is allowed to be a function of t because the integration
is over spatial variables only. Rearranging this, we get
∂φ p 1 2
+ + |u| + χ = G(t), (7)
∂t ρ 2
which is Bernoulli’s equation for unsteady, irrotational flow. By (7), the quantity
∂φ p 1 2
+ + |u| + χ (8)
∂t ρ 2
is constant in space.
Now let’s look specifically at our problem. At the surface, y = η(x, t), the pressure is equal to
the atmospheric pressure, p0 , which can be treated as constant. Then (7) gives us
∂φ 1 2
+ u + v 2 + gη = G(t), (9)
∂t 2
where we’ve incorporated the pressure term into G, and used χ = gy. Now, we also note that G is
arbitrary, because we can write the velocity potential as
φ = φs (x, y, t) + φt (t), (10)
where now φt (t) is arbitrary, because we are only interested in the velocity potential through the
expression
u = ∇φ = ∇φs , (11)
and φt (t) thus has no effect on the resulting u. In (9), then, we are free to choose G(t) = 0, giving
us
∂φ 1 2
+ u + v 2 + gη = 0. (12)
∂t 2
We can now apply our small-amplitude approximations to this. Specifically, we are treating the
disturbances, η, and the velocity components, u and v, as small. The condition that u and v are
small means that the spatial derivatives of φ are small, and we assume here that we can extend
this to say that the time derivative of φ is also small. Linearizing (12) gives us
∂φ
(x, 0, t) + gη(x, t) = 0, (13)
∂t
where this condition is imposed at y = 0 rather than y = η by the same arguments we used in the
last lecture to derive (2).
2
2 Dispersion relation
We are going to look for a solution to (2) and (13) of the form
η(x, t) = A cos(kx − ωt). (14)
Look at (13) first. Plugging in (14), we get

∂φ
(x, 0, t) = −gA cos(kx − ωt),
∂t
and integrating over t, we get
gA
φ(x, 0, t) = sin(kx − ωt).
ω
We have to be a little careful in interpreting this because φ in this expression is explicitly evaluated
at a fixed y. What we can hypothesize is that φ has the form
φ(x, y, t) = f (y) sin(kx − ωt). (15)
We can make the same hypothesis by considering (2).

The potential φ has to satisfy Laplace’s equation, i.e.
∂2φ ∂2φ
∇2 φ = + 2,
∂x2 ∂y
so
−f (y)k2 sin(kx − ωt) + f 00 (y) sin(kx − ωt) = 0
which gives
f 00 − k2 f = 0.
The general solution of this is (as seen in any textbook on differential equations)
f = Ceky + De−ky . (16)
We can simplify this by considering what happens at y → −∞ (remember that we’re considering
the problem of surface waves on deep water). Look at the u-component of velocity, which is (using
(15))
∂φ
u= = f (y)k cos(kx − ωt) = Ceky + De−ky k cos(kx − ωt).
∂x
In order for this to be finite as y → −∞, we must have D = 0. Thus (16) becomes
f = Ceky ,
and we can write (15) as
φ(x, y, t) = Ceky sin(kx − ωt). (17)
Now we want to get some idea of how the constants in (17) relate to each other. Inserting (17)
into (2), we get
Ckek(0) sin(kx − ωt) = Aω sin(kx − ωt)
3
or
Ck = Aω, (18)
and inserting (17) into (13) we get
−Cek(0) ω cos(kx − ωt) + gA cos(kx − ωt) = 0
or
−Cω + gA = 0. (19)
Equation (18) tells us C = Aω/k, so
Aω ky
φ(x, y, t) = e sin(kx − ωt). (20)
k
Also, (18) and (19) together give us
ω 2 = gk (21)
which we anticipated earlier. This is the dispersion relation that tells us how ω and k relate, i.e.
(remembering that the wave speed is given by c = ω/k and the wavelength is 2π/k) how the wave
speed varies with the wavelength.
3 Particle paths
From our result (20) for the velocity potential, the velocity components are
u = Aωeky cos(kx − ωt)

v = Aωeky sin(kx − ωt).
We will assume that the instantaneous position of any particle, which we’ll denote by (x0 , y 0 ), is
very close to its average position, (x, y). We can then write
∂x0
u= ≈ Aωeky cos(kx − ωt)
∂t
∂y 0
v= ≈ Aωeky sin(kx − ωt)
∂t
which integrates to
x0 = −Aeky sin(kx − ωt)

y 0 = eky cos(kx − ωt)
These trajectories describe circles (Fig. 1).

Two observations. The radii of the trajectories is Aeky , which decreases exponentially as we go
deeper (more negative y), just like the fluid velocities do. The other observation is that near the
peak of a disturbance, there is forward motion of the water, and near the troughs, there is rearward
motion; you can see this very readily by watching a floating object (say, yourself) as it encounters
a series of wavefronts.
4
Figure 1: Particle paths (figure from Faber, Fluid dynamics for physicists).

ME 563 - Intermediate Fluid Dynamics - Su: Lecture 29 - Waves: More Basics

Uploaded by

Copyright:

Available Formats

ME 563 - Intermediate Fluid Dynamics - Su: Lecture 29 - Waves: More Basics

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ME 563 - Intermediate Fluid Dynamics - Su: Lecture 29 - Waves: More Basics

Uploaded by

Copyright:

Available Formats

ME 563 - Intermediate Fluid Dynamics - Su

Lecture 29 - Waves: more basics

Reading: Acheson, §3.1-3.3.

1 Bernoulli’s equation for unsteady irrotational flow

which, inserted in (3), gives

φ = φs (x, y, t) + φt (t), (10)

η(x, t) = A cos(kx − ωt). (14)

Look at (13) first. Plugging in (14), we get

φ(x, y, t) = f (y) sin(kx − ωt). (15)

We can make the same hypothesis by considering (2).

−f (y)k2 sin(kx − ωt) + f 00 (y) sin(kx − ωt) = 0

f = Ceky + De−ky . (16)

and we can write (15) as

φ(x, y, t) = Ceky sin(kx − ωt). (17)

Ckek(0) sin(kx − ωt) = Aω sin(kx − ωt)

and inserting (17) into (13) we get

−Cek(0) ω cos(kx − ωt) + gA cos(kx − ωt) = 0

Equation (18) tells us C = Aω/k, so

u = Aωeky cos(kx − ωt)

x0 = −Aeky sin(kx − ωt)

These trajectories describe circles (Fig. 1).

You might also like