You are on page 1of 7

ME 313: Fluids

Day 8

The atmosphere

Don’t do this: What is the atmospheric pressure at an altitude of 15 km?

P − P0 = − ρ g( z − z 0 ) ≅ − ( 1 kg m3 )( 9.8 m s 2 ) ( 15000 m) ⇒ P ≈ P0 − 150,000 Pa ≈ − 50,000 Pa

The problem here, of course, is that the density of air is not constant over an
altitude interval of 15 km.

Instead we have to go back to the hydrostatic equation:


r
∇ P = ρg

Let’s again assume that the gravitational field is uniform and vertical, but let’s
use the ideal gas law for density instead:

dP P
= − ρg = − g
dz RT

Then

dP g
= − dz
P RT

Here’s the rub: does the temperature vary with altitude? If the temperature is
constant, then it’s easy to integrate this equation. But if T is some strange
function of altitude, we have to characterize it as a function of altitude and then
do the integration of the RHS appropriately.

>> The U.S. Standard Atmosphere

Because the assumptions we make about atmospheric properties have such a


huge impact on engineering design, there is a standards committee that decides
issues like this. The last time they met was in 1976, when they formulated the
so-called 1976 U.S. Standard Atmosphere. This was just a minor revision to
previous atmosphere definitions.

The basic idea is that the atmosphere is modeled as a set of layers, within each
of which the temperature is either constant or of a fixed lapse rate; that is, the
slope of temperature versus altitude is constant within a layer.
The lowest layer, that is, the one in which we live, is called the troposphere. It
extends from zero altitude to 36089 feet [11 km]. The layer above it is called the
stratosphere. It extends from 36089 feet to 65617 feet [20 km]. The line dividing
the two, that is, the altitude 36089 feet, is called the tropopause. The end of the
stratosphere at 65617 feet, is called the stratopause. Various other layers are
named, but basically the definition of the atmosphere goes up over 200,000 feet.
In every layer the lapse rate is constant.

Within the troposphere, the temperature drops from 59 F = 518.67 R = 15 C to -


69.7 F = 389.97 R.

Within the stratosphere, the temperature is a constant, balmy 389.97 R.

So let’s analyze how pressure varies with altitude. We will have two cases:

1) constant lapse rate s=dT/dz (s is negative in the troposphere and positive in


the ionosphere)
2) zero lapse rate (s=0)

In each case we have to integrate

dP P
= − ρg= − g
dz RT

Case 1: s ≠ 0 ⇒ T = T0 + sz
dP P dP − gdz
= − g⇒ =
dz R( T0 + sz ) P Rs( T0 s + z )
P g  T0 s + z 
⇒ ln = − ln  
P0 Rs  T0 s 
g

 sz  Rs
⇒ P = P0  1 + 
 T0 
Case 2: s = 0 ⇒ T = T0
dP P dP − gdz
= − g⇒ =
dz RT0 P RT0
P g( z − z 0 )
⇒ ln = −
P0 RT0
 g( z − z 0 ) 
⇒ P = P0 exp − 
 RT0 
Acceleration and D’Alembert’s Principle

(hydrostatics in the presence of accelerations in addition to gravity.)

The way to think about accelerated frames of reference is that the acceleration
acts just like a gravitational field in the opposite direction.

In physics this principle is called the principle of equivalence and is the


observation that led Einstein to propose the general theory of relativity. [There
will be no questions about general relativity on your mid-term].

The principle is actually somewhat older and engineers refer to it as


D’Alembert’s principle. In equation form, it is very simple:
r r r r
∑ Fext = ma ⇒ ∑ Fext − ma = 0

Since I know that the sum of forces on a body is equal to its mass times its
r
acceleration, it follows that if I include the product ma as a force on the right-
hand side of the equation, then I can pretend that the body is in equilibrium and
not accelerating.

Let’s do the simplest example of this. Suppose I have a tank full of water on the
back of a pickup truck.

Now the acceleration of gravity is 9.8 m/s2. Initially the surface of the water is
parallel to the ground. Suppose the truck starts to accelerate forward (i.e. in the
+x direction) at 1 m/s2. What happens to the water surface? I hope you all have
the intuition that it shifts to the back of the tank. In other words, D’Alembert’s
principle tells us that since the tank is accelerating in the plus x direction, we can
treat the situation as being equivalent to an additional force of gravity, with that
force being in the minus x direction.

These forces (actually acceleration fields) add vectorially. That is, the result of
summing a vector in the -z direction and a vector in the -x direction is the vector
that points at the angle shown.

Now what is the upshot of this?


r
First, the hydrostatic equation tells us that ∇ P = ρ g . This means that the
gradient of pressure is in the direction of the gravitational field. This means that
surfaces of constant pressure are orthogonal to the gravitational field, something
that was obvious for all the cases we’ve examined so far. But now it just takes a
little thought.

So if the gravitational field points in this direction, what is orientation of this


surface? Specifically, what angle does the surface take with respect to the
horizontal?

Clearly, the answer is that the angle is tan − 1 ( − a x g ) , which equals about 5.8
degrees in this case.

Second, the rate at which pressure increases with depth is proportional to the
gravitational field, according to ∆ P = ρ g∆ z . But now we have to reinterpret this
equation for our pseudo-gravitational field. First off our surfaces of constant
pressure are parallel to this free surface, that is, orthogonal to the resultant
vector. So we have to interpret “depth” as being parallel to the resultant. Next,
the g in this equation is the magnitude of the resultant vector, that is,
g ′= g 2 + a x2 . In this case that result is not much different from 9.8, about 9.85
m/s2.

If I had to give you a recipe, it would be this:

1) Draw vectors with gravity downward and pseudo-gravity in the opposite of the
acceleration direction.
2) Compute the resultant vector’s magnitude and direction.
3) Free surfaces are normal to the resultant.
4) Pressure increases in the direction of the resultant.
5) Use the resultant magnitude in the equation ∆ P = ρ g∆ z to compute the
pressure increase.

Now I have to give you an important caution: this recipe only applies to uniform
acceleration, that is, acceleration which is constant in time and space. It should
be intuitively obvious that transient accelerations will be considerably more
complicated. The water in the tank in the back of the pickup will slosh around
quite a bit if the truck starts up suddenly.

More to the point, the situation is somewhat more complicated for rotating
systems. We’ll discuss two situations of common interest: the free vortex and
the forced vortex. Let’s discuss the forced vortex first.
Here the fluid rotates more or less as a solid body, which is why we can still
discuss this as a fluid static problem. Formally, we posit that the velocity
r
distribution in the fluid is V = Ω rθ$ . The velocity is in the tangential direction and
is proportional to the distance from the z axis. We specify an angular velocity
omega. This is the situation that you would find water in a rotating bucket, at
least if you let it equilibrate.

So what is the direction of acceleration of the fluid? You will recall from your
freshman physics course that the centripetal acceleration is in the radial
direction, and points inward with a magnitude V 2 r = Ω 2 r .

What does D’Alembert’s principle tell us here? We replace the acceleration with
a pseudo-gravity in the radial direction, but pointing outward, in the opposite
direction.

The problem, from our naïve standpoint, is that the magnitude of that pseudo-
gravity varies with radial position. It’s not a constant like we worked with in the
truck problem.

So rather than the free surface lying at a fixed angle, the angle varies with
position. In fact, the tangent of the angle is tan θ = Ω 2 r g .

This tells us the shape of the curve of the surface, of course: [Who remembers
what kind of function has a derivative which is linear? In other words, what
function of x has derivative kx?]

The surface is a parabola, or more precisely, a paraboloid of revolution.

Now how does the pressure vary with depth? We no longer have a simple
interpretation of depth, because our acceleration field varies with position. But I’ll
sketch out the surfaces of constant pressure; they’re just paraboloids of the
same shape as before. The pressure increases most rapidly normal to these
lines.

To find the pressure field, you have to solve the differential equation

∇ P = ρ g = ρ ( − gzˆ+ Ω 2 rrˆ) .
r

The gradient in cylindrical coordinates is

∂Φ 1 ∂Φ ˆ ∂Φ ˆ
∇Φ = rˆ+ Φ + k
∂r r ∂ϕ ∂z

Motivated by our observation that the surface is a paraboloid, we try a solution of


the form

P = Ar 2 + Bz + C

Then

∂P 1 ∂P ˆ ∂P
∇P= rˆ+ Φ + zˆ
∂r r ∂ϕ ∂z
= 2 Arrˆ + Bzˆ

Equating terms from above, we find

Ω 2
A= ρ
2
B= − ρg
Ω 2
P= ρ r − ρ gz + C
2

The pressure difference between any two locations z1,r1 and z2,r2 is

Ω 2

( P − P) = ρ
2 1
2
(r
2
2
− r12 ) − ρ g ( z 2 − z1 )

Another interesting situation is the free vortex. In a free vortex the velocity
distribution is V = C r . This is the situation of fluid rotating with no external
torque, like water going down a drain. Now technically this is not a fluid statics
problem: unlike the case with the solid vortex, there is not frame on reference in
which the fluid is not moving. Try this example yourself; the resulting pressure
difference is
C2  1 
( P2 − P1 ) = ρ 2  r 2 − r12  − ρ g( z2 − z1 )
2 1

So this is about all I’m going to say about rotational motion.

You might also like