This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

8 Double Integrals

In many instances in physics and engineering, we are required to calculate

the integrals of functions over two variables. Consider a function of two

∆ A

∆x

∆

y

x

y

A

Figure 1: Region A divided in small rectangles or area ∆A = ∆x∆y.

variables f(x, y) which is deﬁned within a region A of the x-y plane. We

can divide the region A into a gridwork of tiny pieces all of length ∆x and

height ∆y. The area of each small rectangle is ∆A = ∆x∆y (see Figure 1).

We can formally deﬁne the double integral of f(x, y) over A to be

_

A

f(x, y)dA = lim

∆x,∆y→0

(all ∆A in A)

f(x, y)∆A . (1)

This is in fact a pretty loose deﬁnition (for example, ask yourself what

happens to the range of the sum ‘all ∆A in A’ if we take ∆x → 0 while

leaving ∆y ﬁnite) , however more rigorous deﬁnitions are beyond the scope

of this course, and for most people it is enough to know that such integrals

can be deﬁned and employed without ambiguity.

In order the calculate these things, we need to decide on the order in

which to perform the double integral. First we note that the region A lies

entirely within the two curves which we can call y

u

(x) and y

l

(x), and the

range of x is bounded by the number α

1

and α

2

(see Figure 2).

Then we can say that

_

A

f(x, y)dA =

_

α

2

α

1

_

_

yu(x)

y

l

(x)

f(x, y)dy

_

dx . (2)

We can do the inner integral ﬁrst, this leaves us with a function of x which

we can then integrate over the range [α

1

, α

2

].

1

y

A

α x α

y (x)

l

y (x)

u

2 1

Figure 2: Region A which lies between a lower curve y

l

(x) and an upper

curve y

u

(x) for x within the range [α

1

, α

2

].

Example: Find the integral of the function f(x, y) = 2x

2

y − 1 over the

area bounded by the functions

y

u

(x) = x + 1

y

l

(x) = 0

and lying between the lines x = 0 and x = 2 (see Figure 3).

y

x 2

1

3

Figure 3: Region A which lies between a lower curve y

l

(x) = 0 and an upper

curve y

u

(x) = x + 1 for x within the range [α

1

, α

2

] = [0, 2].

Solution:

I =

_

A

f(x, y)dA

=

_

2

0

__

x+1

0

(2x

2

y −1)dy

_

dx

=

_

2

0

_

x

2

y

2

−y

_

x+1

0

dx

2

=

_

2

0

_

x

2

(x + 1)

2

−(x + 1)

_

dx

=

_

2

0

_

x

4

+ 2x

3

+ x

2

−x −1

_

dx

=

_

x

5

5

+

x

4

2

+

x

3

3

−

x

2

2

−x

_

2

0

=

196

15

4.8.1 Finding the area of a region

In order to ﬁnd the area of a region A we just calculate the double integral

of the function f(x, y) = 1 over A. i.e.

Area =

_

A

dA (3)

This makes sense because in calculating this integral you are just ‘adding

up’ all the small areas dA within the region A, the sum of which has to give

you the total area.

Example: Find the area between the curves

y = x

2

and

y = x

1/2

in the range 0 ≤ x ≤ 1 (see Figure 4)

y = x

y = x

2

1

1

y

x

Figure 4: Region A which lies between a lower curve y

l

(x) = x

2

and an

upper curve y

u

(x) =

√

x for x within the range [α

1

, α

2

] = [0, 1].

Solution:

Area =

_

A

dA

3

=

_

1

0

_

x

1/2

x

2

1.dydx

=

_

1

0

_

x

1/2

−x

2

_

dx

=

_

2

3

x

3/2

−

1

3

x

3

_

1

0

=

1

3

4.8.2 Reversing the order of integration

There is no reason why we have to do the y integration ﬁrst followed by the

integration over the x coordinate. It is sometimes more convenient to do

them the other way around.

y

A

x

β

β

1

2

l

x (y)

x (y)

r

Figure 5: Region A which lies between a left curve x

l

(y) and a right curve

x

r

(y) for y within the range [β

1

, β

2

].

Referring to the Figure 5 above, we note that the region A is bounded by the

curves x = x

l

(y) and x = x

r

(y), and extends over the range β

1

≤ y ≤ β

2

.

We can calculate the integral of the function f(x, y) over the region A as

_

A

f(x, y)dA =

_

β

2

β

1

_

_

xr(y)

x

l

(y)

f(x, y)dx

_

dy . (4)

As with the previous deﬁnition (2), we do the inner integral ﬁrst. This

leaves us with a function in y which we integrate over the range [β

1

, β

2

].

Example:

Integrate the function f(x, y) = xy

2

over the interior of the semicircle

x

2

+ y

2

≤ a

2

, x ≥ 0 .

(see Figure 6).

4

x

y

a

a

a

Figure 6: Region A which lies between a left curve x

l

(y) = 0 and a right

curve x

r

(y) =

√

a

2

−y

2

for y within the range [β

1

, β

2

] = [−a, a].

Solution: The region is bounded by the curves

x

r

(y) =

_

a

2

−y

2

and

x

l

(y) = 0 .

The integral is then easily calculated

I =

_

+a

−a

_

√

a

2

−y

2

0

f(x, y)dxdy

=

_

+a

−a

_

1

2

x

2

y

2

_

√

a

2

−y

2

0

dy

=

_

+a

−a

1

2

y

2

(a

2

−y

2

)dy

=

_

1

6

a

2

y

3

−

1

10

y

5

_

+a

−a

=

2

15

a

5

.

The same answer can be obtained by doing the y integration ﬁrst, as in the

previous section.

4.8.3 Curvilinear coordinates

Many double integrals are made much easier if we use a non-Cartesian co-

ordinate system, such as plane polar coordinates. In this system we replace

x and y by the variables r and θ, by making the substitution

x = r cos θ (5)

y = r sin θ . (6)

5

If we examine a small element of area in polar coordinates, we see that (cf.

Figure 7):

x

y

∆r

∆θ r

∆A

Figure 7: Region A divided in small curvilinear elements or areas ∆A =

r∆θ∆r.

∆A (= ∆x∆y) = r∆θ∆r . (7)

As we take the limit ∆A → 0, we can replace our ‘small elements’ by

diﬀerentials, and so

dA = rdθdr . (8)

Hence the double integral of a function f over a region A is

_

A

f(x, y)dA =

_

A

f(r, θ)rdrdθ . (9)

As before, we must do both integrals in the double integral in order.

In order to do this, we write the region we want to integrate over as being

bounded by the curves r

l

(θ) and r

u

(θ), and lying within the range θ

0

≤ θ ≤

θ

1

. Thus,

_

A

f(x, y)dA =

_

θ

1

θ

0

_

_

ru(θ)

r

l

(θ)

f(r, θ)rdr

_

dθ (10)

Example:

Integrate the function f(x, y) = xy over the region A, which is the quarter

circle (see Figure 9):

x

2

+ y

2

≤ a

2

, x ≥ 0, y ≥ 0 .

6

x

y

l

u

r ( )

r ( )

θ

θ

θ

1

θ

0

Figure 8: Region A which lies between a lower radius r

l

(θ) and an upper

radius r

u

(θ) (a curvilinear curve) for an angle θ within the range [θ

0

, θ1].

a

a

x

y

Figure 9: Region A which lies between a lower radius r

l

(θ) = 0 and an

upper radius r

u

(θ) = a (a circular curve) for an angle θ within the range

[θ

1

, θ

2

] = [0, π/2].

Solution:

The region A is bounded by the curves

r

u

(θ) = a

r

l

(θ) = 0

and lies within the range 0 ≤ θ ≤ π/2. We also place the function f into

polar coordinates:

f(r, θ) = xy = (r cos θ)(r sin θ)

= r

2

cos θ sin θ .

The integral is then

I =

_

A

f(r, θ)rdrdθ

=

_ π

2

0

_

a

0

r

2

cos θ sin θrdrdθ

7

=

_ π

2

0

__

a

0

r

3

dr

_

cos θ sin θdθ

=

_

r

4

4

_

a

0

_ π

2

0

cos θ sin θdθ

=

_

r

4

4

_

a

0

_ π

2

0

sin(2θ)

2

dθ

=

_

a

4

8

__

−cos(2θ)

2

_π

2

0

=

_

a

4

8

_

_

1

2

+

1

2

_

=

_

a

4

8

_

.

Calculating this double integral using Cartesian coordinates is a little more

complicated, as we see below:

First of all, we note that the region is bounded by the curves

x

r

(y) =

_

a

2

−y

2

and

x

l

(y) = 0 .

The integral is then expressed as follows

I =

_

a

0

_

√

a

2

−y

2

0

f(x, y)dxdy

=

_

a

0

_

1

2

x

2

y

_

√

a

2

−y

2

0

dy

=

_

a

0

1

2

y(a

2

−y

2

)dy

=

_

1

4

a

2

y

2

−

1

8

y

4

_

a

0

=

1

8

a

4

,

which is consistent with the previous calculation performed in curvilinear

coordinates. In other cases, it is more apparent that the choice of curvilinear

coordinates simpliﬁes the derivation.

8

4.9 Surface Integrals

A general surface in three dimensions can be represented by the equation

(see Figure )

Φ(x, y, z) = C (11)

area

Φ =

x

z

y

Const.

S ∆

surface S

Like line integrals, there are two types of surface integrals, which we will

refer to as scalar and vector type integrals. We approach both in more or

less the same way, which is to divide the surface up into a large number

of small rectangles of equal area ∆S, and then sum over these as the area

approaches zero. By projecting onto one of the coordinate planes, we can

convert an integral over a surface into a double integral, something which

we already know how to calculate.

4.9.1 Scalar-type surface integrals

Suppose that T(x, y, z) is some scalar function which is deﬁned on a surface

S in 3D space. We then divide the surface up into a large number of small

squares, each of area ∆S. We then take the limit, in which we get an inﬁnite

number of these small squares and at the same time they become inﬁnitely

small, i.e. ∆S →dS. The surface integral of T over S can then be deﬁned

as _

S

T(x, y, z)dS = lim

∆S→0

(all ∆S in S)

T(x, y, z)∆S (12)

Clearly the evaluation of these integrals is made more diﬃcult because each

inﬁnitesimal element dS is aligned at a diﬀerent angle. That is, each element

has a diﬀerent direction associated with it. This direction is the normal to

the surface ˆ n , which may well change direction as we move from one point

on the surface to another.

In order to calculate the integral, we must project onto one of the coor-

dinate planes. The x−y plane is usually the most convenient, if we do this,

9

then we can see from the ﬁgure following that the projected area dxdy can

be written

dA = dxdy = ˆ n ·

ˆ

k dS , (13)

where

ˆ

k is the coordinate vector pointing in the z direction. We can re-

arrange this equation to get

dS =

dA

ˆ n ·

ˆ

k

=

dxdy

ˆ n ·

ˆ

k

. (14)

dA = dx dy

n

k

z

x

y

k n

dS=

dA

n

dS= dA

dA = dx dy

k

Figure 10: Left: Projection of an inﬁnitesimal element dS (of arbitrary

orientation) onto a ﬂat inﬁnitesimal element dA = ˆ n ·

ˆ

k dS in the x − y

plane. Here, ˆ n denotes the unit normal to the surface S and

ˆ

k is the unit

vector oriented along the z-axis. Right: Idem when dS is parallel to the

x −y plane (here, ˆ n ·

ˆ

k = ˆ n · ˆ n = 1).

Sometimes we ﬁnd that it is not convenient to project onto the x-y plane.

For example we might want to integrate over the surface Φ(x, y, z) = x+y =

1 which is always perpendicular to the x-y plane. In this case we have to

project onto one of the other coordinate planes, either the x-z plane or the

z-y plane. It is a simple matter then to obtain the relations

dS =

dx dy

ˆ n ·

ˆ

k

=

dy dz

ˆ n ·

ˆ

i

=

dz dx

ˆ n ·

ˆ

j

(15)

where, as usual,

ˆ

i ,

ˆ

j , and

ˆ

k are the coordinate vectors in the x, y and z

directions respectively.

We still need a way to calculate the normal ˆ n at each point on the

surface. The way we will do this is by making use of the formula for the

10

normal which involves the gradient. First suppose that we decide to project

onto the x-y plane. Then we can represent the surface S as

z = f(x, y) (16)

That is,

Φ(x, y, z) = z −f(x, y) = C = 0 . (17)

The unit normal to the surface is

ˆ n =

∇Φ

|∇Φ|

=

∂Φ

∂x

ˆ

i +

∂Φ

∂y

ˆ

j +

∂Φ

∂z

ˆ

k

_

_

∂Φ

∂x

_

2

+

_

∂Φ

∂y

_

2

+

_

∂Φ

∂z

_

2

=

1

_

_

∂f

∂x

_

2

+

_

∂f

∂y

_

2

+ 1

_

−

∂f

∂x

ˆ

i −

∂f

∂y

ˆ

j +

ˆ

k

_

. (18)

So,

ˆ n ·

ˆ

k =

1

_

_

∂f

∂x

_

2

+

_

∂f

∂y

_

2

+ 1

(19)

Therefore the surface integral is

_

S

T(x, y, z) dS =

_

A

T(x, y, f(x, y)) (

¸

¸

¸

_

_

∂f

∂x

_

2

+

_

∂f

∂y

_

2

+ 1) dxdy (20)

where A is the ‘projected’ region of S onto the x-y plane (see Figure 11).

11

z

x

y

S

A

Figure 11: Projection of a curved surface S onto a ﬂat area A in the x −y

plane.

Example: Evaluate the integral of the function T(x, y, z) = xyz over the

surface S, which is deﬁned by the equations (see also Figure 12)

x + y + z = 1

x ≥ 0 , y ≥ 0 , z ≥ 0 .

S

1

1

1

x

y

z

Figure 12: Flat surface S deﬁned as the set

_

(x, y, z) , x +y + z ≤ 1 , x ≥

0 , y ≥ 0 , z ≥ 0

_

.

Solution:

We can re-express the surface as

z = f(x, y) (21)

= 1 −x −y .

Then,

T(x, y, z) = xyz

12

= xy(1 −x −y)

= xy −x

2

y −y

2

x .

Projecting onto the x-y plane,

dS =

dxdy

ˆ n ·

ˆ

k

=

¸

¸

¸

_

1 +

_

∂f

∂x

_

2

+

_

∂f

∂y

_

2

dxdy

=

√

3 dxdy

The integral is now

I =

_

S

T dS =

_

A

(xy −x

2

y −y

2

x)

√

3 dxdy

where the region A is shown on Figure 13.

0 1

1

x

y

A

Figure 13: Area A deﬁned as the set

_

(x, y) , x + y ≤ 1 , x ≥ 0 , y ≥ 0

_

.

A is the projection of S onto the x −y plane.

This is a double integral, which is the kind of thing we tackled in the

previous section. It is evaluated:

I =

_

1

0

dx

_

1−x

0

dy

√

3(xy −x

2

y −y

2

x)

=

_

1

0

_

xy

2

2

−

x

2

y

2

2

−

y

3

x

3

_

1−x

0

√

3dx

=

_

1

0

_

−

5

6

(1 −x)

3

_

√

3 dx

=

5

√

3

24

13

4.9.2 Finding the area of a surface

The area of a given surface in 3D can be calculated by integrating the scalar

function T(x, y, z) = 1 over that surface, i.e.

Area =

_

S

dS . (22)

Example:

Calculate the area of the surface S given in the previous example (see Figure

12).

Solution:

Projecting onto the x-y plane, we ﬁnd that the area element is

dS =

√

3 dxdy ,

so that the total area of the surface is

Area =

_

S

dS

=

_

A

√

3 dxdy

=

√

3

_

1

0

dx

_

1−x

0

(1) dy

=

√

3

_

1

0

[y]

1−x

0

dy

=

√

3

_

1

0

(1 −x) dx

=

√

3

_

−

1

2

(1 −x)

2

_

1

0

=

√

3

2

.

4.10 Vector-type surface integrals

Recall that we have divided our surface up into tiny elements of area ∆S,

and that each of these elements has a direction associated with it, a direc-

tion which is speciﬁed by the normal vector ˆ n . We can use this to deﬁne

a new sort of area element,

∆S = ˆ n ∆S (23)

which is a vector quantity, and represents both the area of a surface element

and its orientation (see ﬁgure 14).

14

∆ S = n ∆ S

Figure 14: An oriented surface S can be approximated by a large number

of tiny ﬂat elements ∆S = ˆ n ∆S of unit normals ˆ n .

The reason this new quantity is useful is that if we ‘dot’ product it with a

vector ﬁeld, then it measures how much of that ﬁeld goes through the small

surface element ∆S. That is, if the vector ﬁeld is totally parallel to the

surface then it doesn’t really ‘go through’ the surface element at all, it just

glides across it. In this case F · ∆S is zero (see Figure 15 left). However

if the surface element is oriented so that it is perpendicular to the vector

ﬁeld, then F · ∆S is as large as it can possibly get (see Figure 15 right).

∆S

∆ S

Figure 15: The ﬂux F · ∆S of a vector ﬁeld F across the surface element

∆S is null on the left picture and is maximised on the right picture.

The quantity

F · ∆S (24)

is known as the ﬂux of the vector ﬁeld F across the surface element ∆S.

If we add up the totality of all these ﬂuxes then we can form an integral,

often known as the ﬂux integral:

_

S

F · dS = lim

∆S→0

(all ∆S in S)

F · ∆S , (25)

where we have replaced the area vector ∆S by its inﬁnitesimal counterpart

dS in the limit as ∆S →0.

15

In practise this integral can be evaluated by re-writing it as

_

S

F · dS =

_

S

F · ˆ n dS . (26)

Note that the inﬁnitesimal vector element dS has been replaced by the

scalar quantity dS on the right-hand side. In fact the quantity F · ˆ n is also

a scalar, and so this integral is now a scalar-type surface integral, which we

already know how to calculate.

Example: Calculate the vector-type surface integral

_

S

F · dS ,

where F = z

2

ˆ

k and the surface S is deﬁned by the equations (see Figure

12)

x + y + z = 1 , x, y, z ≥ 0 .

Solution: The surface is described by

Φ(x, y, z) = x + y + z = 1

and so the unit normal to the surface is

ˆ n =

∇Φ

|∇Φ|

=

ˆ

i +

ˆ

j +

ˆ

k

√

1

2

+ 1

2

+ 1

2

=

1

√

3

(

ˆ

i +

ˆ

j +

ˆ

k ) .

So

F · ˆ n = (z

2

ˆ

k ) · (

1

√

3

ˆ

i +

1

√

3

ˆ

j +

1

√

3

ˆ

k )

=

z

2

√

3

=

1

√

3

(1 −x −y)

2

.

Therefore we now have to evaluate a scalar integral,

_

S

F · dS =

_

S

1

√

3

(1 −x −y)

2

dS .

Projecting onto the x-y plane,

dS =

dx dy

ˆ n ·

ˆ

k

=

√

3dx dy

16

The rest is just evaluating the double integral associated with the projection

A of S onto the x −y plane (see Figure 13)

_

S

F · dS =

_

A

1

√

3

(1 −x −y)

2

√

3 dxdy

=

_

1

0

_

_

1−x

0

(1 −x −y)

2

dy

_

dx

=

_

1

0

_

(1 −x)

2

y +

1

3

y

3

−y

2

(1 −x)

_

1−x

0

dx

=

_

1

0

1

3

(1 −x)

3

dx

=

1

12

.

17

method double integral instructional

method double integral instructional

- 13 07 Stokes Thm
- Conventional Formulation in SI Units
- Geometria Diferencial - Gardner - Differential Geometry & Re
- UNIT2_CG
- Mtl Mec Aae Spe Pptsourcesoferror
- Improper Integral1
- Fyc13 Improper Integrals
- Solidworks 2008 - Surface Modeling
- Surface Modeling
- 3 Integrals
- Integration Summary
- AML710 Computer Aided Design Minor II Semester
- 4013-s05
- Integralcalculus
- Detynieck_AGOP_01
- Mathematics and Statistics
- ezil
- Mathematics 12B
- Symbolic Integration
- Rattazzi-PathIntegral
- Math 152 Calculus II Chapter 4
- math.pdf
- Surfacing Best Practices
- Introduction to Functional Analysis_lec08
- Samvida 1 2011 Rule Book
- MAT495_CHAPTER_6
- Chapter 3
- R26_ECE_ingles.pdf
- GATE 2015
- nagaraj

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd