The Divergence Theorem

1. Describe the boundary of each of the following solids. (Your description should be thorough enough
that somebody reading it would have enough information to find the surface area of the boundary).

(a) The solid x2 + 4y 2 + 9z 2 ≤ 36.

Solution. The boundary is just the ellipsoid x2 + 4y 2 + 9z 2 = 36.

(b) The solid x2 + y 2 ≤ z ≤ 9.

Solution. The boundary has two pieces:
z

y
x

One (the red one) is the part of the paraboloid z = x2 + y 2 lying below the plane z = 9, and the
other is the part of the plane z = 9 satisfying x2 + y 2 ≤ 9.

(c) The solid consisting of all points (x, y, z) inside both the sphere x2 + y 2 + z 2 = 4 and the cylinder
x2 + y 2 = 3.
Solution. The boundary of this solid has 3 pieces, as we can see from the picture:
z

y
x

The cylinder and the sphere intersect when z 2 = 1, or z = ±1. Therefore, the top (red) piece is
the part of the sphere x2 + y 2 + z 2 = 4 above the plane z = 1. The bottom (blue) piece is the
part of the sphere x2 + y 2 + z 2 = 4 below the plane z = −1. The middle (yellow) piece is the part
of the cylinder x2 + y 2 = 3 between the planes z = −1 and z = 1.

2. Let F~ (x, y, z) = hx2 , 2y, ez i. Let S be the surface Z

ofZthe cube whose vertices are (±1, ±1, ±1), oriented
with outward normals. Evaluate the flux integral F~ · dS.
~
S

Solution. Although it is possible to compute the flux integral directly, we would have to parameterize
all 6 sides of the cube, compute a flux integral through each, and add up the answers. That doesn’t
seem like much fun.

1
 Instead, we’ll use the
Z ZDivergence ZTheorem.
ZZ Since S is oriented with outward normals, the Divergence
Theorem says that ~ ~
F · dS = ~
div F dV , where E is the solid interior of S. We just calculate
S E
div F~ : it is div F~ = 2x + 2 + ez .
ZZZ
So, our goal now is: “Compute the triple integral (2x + 2 + ez ) dV where E is the interior of
E
the cube with vertices (±1, ±1, ±1).” Like any other triple integral, we do this by converting it to an
iterated integral:
ZZZ Z 1Z 1Z 1
(2x + 2 + ez ) dV = (2x + 2 + ez ) dx dy dz
E −1 −1 −1

4
= 16 − + 4e
e

3. Let F~ (x, y, z) = hx3 , z 2 , 3y 2 zi. Let S be the surface 2 2

Z Z z = x + y , z ≤ 4 together with the surface
z = 8 − (x2 + y 2 ), z ≥ 4. Evaluate the flux integral F~ · dS
~ if S is oriented with outward normals.
S

Solution. Here is a picture of the surface.

y
x

ZZ
Since S is oriented with outward normals, the Divergence Theorem tells us that F~ · dS
~ =
ZZZ S

div F~ dV , where E is the solid interior of S. In this case, div F~ = 3x2 + 3y 2 .

E
ZZZ
So, our goal now is: “Compute the triple integral (3x2 + 3y 2 ) dV where E is the solid enclosed
E
by z = x2 + y 2 and z = 8 − (x2 − y 2 ).”

It’s probably easiest to do this as an iterated integral in cylindrical coordinates. In fact, we’ve already
done a triple integral with this solid as the region of integration before, in #1 on the worksheet “Triple
Integrals in Cylindrical or Spherical Coordinates”. Using the work we did there, we can rewrite the

2
 triple integral as an iterated integral
ZZZ Z 2π Z 2 Z 8−r 2
2 2
3(x + y ) dV = 3r2 · r dz dr dθ
E 0 0 r2
Z 2π Z 2  z=8−r2 
= 3r3 z  2 dr dθ

0 0 z=r
Z 2π Z 2
= (24r3 − 6r5 ) dr dθ
0 0
Z 2π  r=2 
4 6
= 6r − r  dθ
0 r=0
Z 2π
= 32 dθ
0
= 64π

4. True or false: If F~ is a vector field whose

Z Z divergence is 0 and S is any surface, then the Divergence
Theorem implies that the flux integral F~ · dS
~ is equal to 0.
S

Solution. False. The statement is only true if S is the boundary of some solid. Many surfaces do
not fall into this category. For instance, the disk x2 + y 2 ≤ 1 in the plane z = 0 is not the boundary
of any solid.

5. Let S1 be the surface consisting of the top and the four sides (but not the bottom) of the cube whose
vertices are (±1, ±1, ±1),Zoriented
Z the same way as in #2. Let F~ (x, y, z) = hx2 , 2y, ez i, as in #2.
Evaluate the flux integral F~ · dS.
~ (Hint: Use #2.)
S1

Solution. Let S2 be the bottom of the cube, oriented with normals pointing downward (so that the
bottom side of S2 is the “positive” side). Then, S1 and S2 together form the surface S in #2. So, by
#2, we know that ZZ ZZ
F~ · dS
~+ F~ · dS~ = 16 − 4 + 4e (1)
S1 S2 e
ZZ ZZ
If we can evaluate F~ · dS,
~ then this equation will tell us what F~ · dS
~ is as well. Evaluating
ZZ S2 ZZ S1

F~ · dS
~ should be a lot easier than evaluating F~ · dS,
~ since S2 consists of one face of the cube,
S2 S1
while S1 consists of five faces.
ZZ
To evaluate F~ · dS,
~ we’ll use the definition of the flux integral. That is, we start by parameterizing
S2
the surface S2 . In this case, S2 is the portion of the plane z = −1 with −1 ≤ x ≤ 1 and −1 ≤ y ≤ 1.
So, we can parameterize it by x = u, y = v, z = −1, or ~r(u, v) = hu, v, −1i, with −1 ≤ u ≤ 1 and
−1 ≤ v ≤ 1. (The region R in the uv-plane described by the inequalities −1 ≤ u ≤ 1, −1 ≤ v ≤ 1 is a
square.) We should check whether this gives the right orientation:

~ru = h1, 0, 0i
~rv = h0, 1, 0i
~ru × ~rv = h0, 0, 1i

3
 This points upward, but we wanted to orient S2 with normals pointing downward, so we have the
wrong orientation. Therefore,
ZZ ZZ
F~ · dS
~ = − F~ (~r(u, v)) · (~ru × ~rv ) dA
S2 R
ZZ  
1
= − u2 , 2v, · h0, 0, 1i dA
R e
ZZ
1
= − dA
R e
ZZ
1
= − 1 dA
e R
1
= − (area of R) by the worksheet “Double Integrals”, #2(a)
e
4
= −
e
ZZ
Plugging this into (1), we find that F~ · dS
~ = 16 + 4e.
S
D 2
E
6. Let F~ be the vector field F~ (x, y, z) = z 3 sin ey , z 3 ex sin z , y 2 + z , and let S be the bottom half of the
ZZ
sphere x2 + y 2 + z 2 = 4, oriented with normals pointing upward. Find F~ · dS.
~
S

Solution. Since the vector field F~ has quite a complicated definition, we can guess that doing the flux
integral directly would be horrible. Since this problem is on a Divergence Theorem worksheet, we can
guess that we should use the Divergence Theorem.(1)
However, we can’t directly apply the Divergence Theorem because the given surface S is not the
boundary of a solid. So, we’ll have to be a bit more clever and use the strategy of #5. That is, we
0
Z Z surface ZSZ which, together
want to think of another Z S, encloses a solid E. Then, the ZDivergence
Z Zwith Z
Theorem will relate ~ ~
F · dS, ~ ~
F · dS, and ~
div F dV . So, if we can compute F~ · dS
~
ZZZ S S0 ZZ E S0

and div F~ dV directly, we’ll be able to find F~ · dS.

~
E S

There are lots of surfaces S which, together with S, enclose a solid E. Let’s take S 0 to be the disk
0

x2 + y 2 ≤ 4 in the plane z = 0, oriented with its normals pointing upward (so that the top side of S 0 is
the “positive” side).(2) Then, S and S 0 together enclose the bottom half of the ball x2 + y 2 + z 2 ≤ 4,
and we’ll call this solid E.
(1) A better way to tell that the Divergence Theorem is a good approach is to think through our options. We’ve already decided
we don’t want to compute the flux integral directly, so that means we need to use an integral theorem. The ones we know
that tell us something about flux integrals are Stokes’ Theorem and the Divergence Theorem. Stokes’ Theorem only applies
~ is not the curl of anything
to flux integrals in which the vector field being integrated is the curl of something. In this case, F
because div F~ 6= 0, and #4 on the worksheet “Curl and Divergence” says that the divergence of a vector field that is the curl
of something must be 0.
(2) Here, we just make an arbitrary choice of which way to orient. It doesn’t make any difference, as long as we are consistent

throughout the rest of the problem.

4
 z

y
x

The boundary of E, oriented outwards, is S oriented with downward normals (opposite of what we’re
asked for) and S 0 oriented with upward normals (the way we decided to do it). So, the Divergence
Theorem says ZZ ZZ ZZZ
F~ · dS
~− F~ · dS
~= div F~ dV (2)
S0 S E

(In this equation, we’re thinking of S as oriented with upwardZ Z

normals
Z and S 0 asZoriented
ZZ with upward
normals.) Let’s first calculate the right side. div F~ = 1, so div F~ dV = 1 dV . We know
E E
that this is equal to the volume of E (see #1(a) on the worksheet “Triple Integrals”), which is 16 3 π.
ZZ
Next, let’s calculate F~ · dS.
~ To do this, we need to parameterize S 0 . Since S 0 is the part of the
S0
plane z = 0 satisfying x2 + y 2 ≤ 4, we can just parameterize it as ~r(u, v) = hu, v, 0i (and the restriction
x2 + y 2 ≤ 4 translates to u2 + v 2 ≤ 4). Next, we check whether we have the right orientation:

~ru = h1, 0, 0i
~rv = h0, 1, 0i
~ru × ~rv = h0, 0, 1i

This always points upward, which is what we want. So, if we let R be the disk u2 + v 2 ≤ 4 in the
uv-plane,
ZZ ZZ
F~ · dS
~ = F~ (~r(u, v)) · (~ru × ~rv )
S0
Z ZR
= h0, 0, v 2 i · h0, 0, 1i dA (3)
R
ZZ
= v 2 dA
R

Since R is a disk, this double integral is probably a bit easier in polar coordinates: it is
ZZ Z 2π Z 2
F~ · dS
~ = (r sin θ)2 · r dr dθ
S0 0 0
= 4π

ZZ ZZ
Plugging this all into (2), 4π − ~ = 16 π, so
F~ · dS ~ = −4π .
F~ · dS
S 3 S 3
ZZ
Notice that, when we calculated the flux integral F~ · dS,
~ F~ (~r(u, v)) became very simple in (3): it
S0
was just h0, 0, v 2 i. In fact, we could tell as soon as we chose the surface S 0 that this term would become
very simple: ~r(u, v) parameterizes S 0 , and the surface S 0 that we chose had z = 0. When z = 0, the
vector field F~ becomes very simple. This is one reason the choice of S 0 that we made was a good one.