ME 2002: Advanced Engineering Mathematics II

Spring: 2023
Instructor: Dr. I-Fan Lin

Topic IV
Partial Di↵erential Equations and Boundary Value Problems

Partial Di↵erential Equation

Type of PDE
¿ Unsteady Heat Equation

1-D unsteady heat equation

@u @2 u k
= ↵ 2, ↵= : thermal di↵usively
@t @x ⇢C p
n.b. Above PDE needs initial condition, i.e. u(x, 0), and boundary conditions at end points x = 0 and
x = L.

e.g. IC: u(x, 0) = f (x), for 0 < x < L

8
>
<u(0, t) = u1
>
BCs: >>
:u(L, t) = u2
, for all t > 0 ( Dirichlet Problem)
8
>
>
> @u
>
>
> (0, t) = u1
< @x
BCs: >>
>
>
> @u
, for all t > 0 (Neumann Problem)
>
: (L, t) = u2
@x

* x12 2 2) an
n.b. For multi-dimension heat equation =
=
+ +

¡ Wave Equation

-
(1) vibrating string

r
@2 y @2 y
force balance: 2 = a2 2 . where a =
Th Th:horizontal component of tensile force
@t @x ⇢
a:wave
8
>
>
> y(x, ⑧o) = f (x), 0<x<L propagation speed
>
<
ICs: >
> @y
>
>
: (x, 0) = g(x), 0 < x < L
@t
8
>
<y(0, t) = 0
>
BCs: >> , for all t > 0
:y(L, t) = 0

(2) vibrating membrane

c 2023 I-Fan Lin. All Rights Reserved. 1

 ! -3
@2 z
=a
2
2 @ z
+
@2 z
= a2 r2 z
use cylindrical coordinate ( r, 0,3)
@t2 @x2 @y2
8
>
3(4,0,A)
>
>
> z(x, y, 0) = f (x, y)
<
ICs: >
>
> @z (x, y, 0) = g(x, y)
!?
>
:
@t
BC: z(x, y, t) = 0 along R
= +tr)a

¬ Laplace Equation (Steady Heat Equation)

in 2D >O
!
@2 u @2 u

24 4 0(82u d
@u
=↵ +
@t @x2 @y2 + =
=

in 3D
@2 u @2 u @2 u
r2 u = + + =0
@x2 @y2 @z2
8
>
>
>
>u(x, y, z) = f (x, y, z) on surface ⌃ (Dirichlet Problem)
<
BCs: >
> @u
>
>
: (x, y, z) = g(x, y, z) on ⌃
@n
(Neumann Problem)

Laplace Operator in Cylindrical & Spherical Coordinates

r2 = r · r
@ @ @
r=i +j +k
@x @y @z

9
@2 @2 @2
r·r= + + 23
@x2 @y2 @z2

-
In cylindrical coordinates
@ 1 @ @
r = er + e✓ + ez
@r r @✓ @z 7

! ! -
@ 1 @ @ @ 1 @ @ r
Then, r · r = er + e✓ + ez · er + e✓ + ez
@r r @✓ @z @r r @✓ @z
L

-++
x

c 2023 I-Fan Lin. All Rights Reserved. 2

 er',
e.g. er E 20
=

ea
a
>

cer er

finally 2 +Fort at
=

600
-CO -

-- er er ↑

20 20
u

~ "Co ed

2
X

Similarly in spherical coordinates (r, ', ✓)

@ 1 @ 1 @
r = er + e' + e✓
@r r @' r sin ' @✓
u(0,1) 0
=

@2 2 @ 1 @2 cot ' @ 1 @2
r·r= + + + + -y
@r2 r @r r2 @'2 r2 @' r2 sin ' @✓2 8
UIL,A) =

8
->
thin homogeneous bar
X
q
constant density
L L
Unsteady Heat Equations X
UCx, Al:temperature distribution
Method of Separation of Variables ( Fourier method)
• 1-D heat equation

@u @2 u
=↵ 2 (1)
@t @x

BC: u(0, t) = u(L, t) = 0, 8t > 0

IC: u(x, 0) = f (x), for 0 < x < L

&>initial temperature

Solution procedure
8. Then the solution into (1) is
Let u(x, t) = X(x)T (x).
ET= "T

E= 42
->
constant
=
= -

(space & time)

>Sturm -Liouville problem

Tc
=>

c 2023 I-Fan Lin. All Rights Reserved. 3

 Now consider X 00 + X = 0 with BCs

u(0, t) = 0 =) X(0)T (t) = 0 *(0)

= > =
0

u(L, t) = 0 =) X(L)T (t) = 0 X(t) =>

0
=

any time A

solution: X(x) = a1 cos( x) + a2 sin( x)

#

10) 0 G, 0
=
=
=>

:: E(x) azsin(xx)
=

4) 0 azsin(XL)
=
= >

0
=
needs an FO
:

N NT
=

for n 1, 2, 3.---
=

1:
eigenvalues
i.e.x
1.n 1,2,3.
=

=
---

and
*(x) azsin( xx)
=

ansin(*x),
=

n 1, 2,
3,
=

---

⇣ n⇡ ⌘
n.b. sin x : eigensolutions (or eigenfunctions)
L

Turn to solution T (t)

1
X ⇣ n⇡x ⌘ n2 ⇡2
↵t
The general solution is u(x, t) = cn sin e L2

n=1
L

What is cn ?

Recall: Fourier Series representation of a function f (x)

Z
2 L ⇣ n⇡ ⌘
The Fourier coefficient cn = f (⇠) sin ⇠ d⇠
L 0 L

2X
1 Z L ⇣ n⇡ ⌘ ! ⇣ n⇡x ⌘ n2 ⇡2
↵t
The final solution is u(x, t) = f (⇠) sin ⇠ d⇠ sin e L2
L n=1 0 L L

c 2023 I-Fan Lin. All Rights Reserved. 4