The Laplace Equation

ES 201
Advanced Mathematical Methods in
Engineering I
Today we’ll discuss the 2D Laplace
equation
• What is 2D Laplace equation?
• How to derive 2D Laplace equation from physical
phenomena?
• What is 2D Laplace equation in other coordinate
systems?
• What are the initial and boundary conditions needed
to solve 2D Laplace equation?
• How to solve 2D Laplace equation by method of
separation of variables?
 What is 2D Laplace equation?
∂ 2u ∂ 2u 2 independent variables (x & y)
0= +
∂x 2 ∂y 2 Elliptic differential equation

Elliptic partial differential equation

∂ 2u ∂ 2u ∂ 2u  ∂u ∂u 
a ( x, t ) 2 + b ( x, t ) + c(x, t ) 2 = f  x, t , u , , , b 2 < 4ac
∂x ∂x∂t ∂t  ∂x ∂t 
When do we get to solve Laplace equation?
Steady state two-dimensional heat equation/diffusion
temperature distribution, stress, fluid and electric potential, fluid conc’n

2
2 2 2 
∂u ∂ u ∂ u ∂ u ∂ 2u ∂ 2u ∂ 2u ∂ 2u ∂ 2u
=a  2 + 2 + 2  0= + + 0= +
∂t  ∂x ∂y ∂z  2 2 2 2
∂y 2
 ∂x ∂y ∂z ∂x
When interested on TRANSIENT When interested on STEADY STATE 2D body, e.g. a plate
heat conduction behavior (3D body) heat conduction behavior (3D body)
 How to derive the 2D Laplace equation
from physical phenomenon?
Imagine a thin, flat plate of heat-conducting material between two sheets of insulation

Sufficient time has passed; temperature now only depends on spatial coordinates x
and y (steady state condition)

Apply law of conservation of energy to a small

rectangle with sides x and y
y
rate of heat IN = q x (x, y + ∆y / 2)∆y∆z + q y (x + ∆x / 2, y )∆x∆z
(x,y) rate of heat OUT = q x (x + ∆x, y + ∆y / 2)∆y∆z
+ q y (x + ∆x / 2, y + ∆y )∆x∆z
O x
q is heat flux (heat rate/area); z = plate thickness

[qx (x, y + ∆y / 2) − qx (x + ∆x, y + ∆y / 2)]∆y∆z +

Since heat IN = heat OUT:
[q y (x + ∆x / 2, y ) − q y (x + ∆x / 2, y + ∆y )]∆x∆z = 0
 How to derive the 2D Laplace equation
from physical phenomenon?
[qx (x, y + ∆y / 2) − qx (x + ∆x, y + ∆y / 2)]∆y∆z +
Divide both sides by x y z
[q y (x + ∆x / 2, y ) − q y (x + ∆x / 2, y + ∆y )]∆x∆z = 0
[qx (x, y + ∆y / 2) − qx (x + ∆x, y + ∆y / 2)] + [q y (x + ∆x / 2, y ) − q y (x + ∆x / 2, y + ∆y )] = 0
∆x ∆y
∂q x ∂q y ∂  ∂u  ∂  ∂u 
+ =0 limit as x and y → 0  − kx  + − ky =0
∂x ∂y ∂x  ∂x  ∂y  ∂y 
Using Fourier’s first law
If we have an isotropic material (kx = ky = k):

∂ 2u ∂ 2u
2
+ 2
=0 also known as 2D steady state heat equation, since
∂x ∂y

∂u  ∂ 2 u ∂ 2u 
= k 2 + 2  2D heat equation; ut → 0 as t →∞
∂t  ∂x ∂y 

 What is 2D Laplace equation in
other coordinate systems?
∂ 2u 1 ∂u ∂ 2u
In POLAR coordinates: x = r cos θ ; y = r sin θ ; z = z + + 2 =0
2 r ∂r ∂z
∂r
Used for problems possessing axisymmetry

∂ 2u 1 ∂u 1 ∂ 2u
If the solution is independent of z: 2
+ + 2 2
=0
∂r r ∂r r ∂θ

x = r cos φ sin θ ; y = r sin φ sin θ ; z = r cosθ

In SPHERICAL coordinates:
r 2 = x2 + y2 + z2

∂  2 ∂u  1 ∂  ∂u 
If there is no dependence to φ :  r  + sin θ  =0
∂r  ∂r  sin θ ∂θ  ∂θ 

θ is the angle measured down to the point from z axis

φ is the angle between x axis and the projection of the point on the xy plane
 What are the initial and boundary conditions
used to solve the Laplace equation?
NO INITIAL CONDITIONS! since Laplace’s equation involves time-independent
phenomena; only specify boundary conditions

Along any section of the boundary, we can have:

1. Dirichlet condition: u is given
∂u
2. Neumann condition: is given; where n is the unit normal direction
∂n
∂u
3. Robin condition: u + α is given
∂n
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
∂ 2u ∂ 2u
Find the solution of the 2D Laplace equation 2
+ 2
=0
∂x ∂y
w/c satisfies the boundary conditions: u (0, y ) = 0, 0 < y < b
u (a, y ) = f ( y ), 0 < y < b
u (x,0 ) = u ( x, b ) = 0, 0 < x < a

u ( x, y ) = X ( x )Y ( y ) X "Y + XY " = 0
X" Y" both sides must be
=−
X Y constant, but the sign of that
constant is not obvious

The boundary conditions can be written as:

X (0) = 0; Y (0) = Y (b ) = 0
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
X" Y" X "+ m 2 X = 0 S.L.
Choosing a negative constant –m2: = − = −m 2
X Y Y "− m 2Y = 0 problems

X (x ) = A cos mx + B sin mx X (0) = A cos 0 + B sin 0 = 0 → A = 0

Y ( y ) = C cosh my + D sinh my Y (0) = C cosh 0 + D sinh 0 = 0 → C = 0
Y (b ) = D sinh mb = 0 → D = 0 since sinh mb ≠ 0
Trivial solution since Y(y) = 0 and u(x,y) = 0

Choosing a zero constant: X " = 0 X (x ) = E + Fx Y (0 ) = G + H (0 ) = 0 → G = 0

Y"= 0 Y ( y ) = G + Hy Y (b ) = H = 0
Trivial solution since Y(y) = 0 and u(x,y) = 0
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
X" Y" X "−k 2 X = 0 S.L.
Choosing a positive constant k2: = − = k2
X Y Y "+ k 2Y = 0 problems

X (x ) = I cosh kx + J sinh kx Y (0) = K cos 0 + L sin 0 = 0 → K = 0

Y ( y ) = K cos ky + L sin ky π 2π 3π nπ
Y (b ) = L sin kb = 0 → for L ≠ 0, k n = , , ,...,
b b b b
 nπ   nπ   nπ 
Yn ( y ) = Ln sin  y X n (x ) = I n cosh x  + J n sinh  x
 b   b   b 
  nπ   nπ   nπ 
u n (x, y ) = X n (x )Yn ( y ) =  Pn cosh x  + Qn sinh x  sin  y
  b   b   b 
Pn = I n Ln ; Qn = J n Ln

∞
  nπ   nπ   nπ 
u ( x, y ) = ∑  Pn cosh x  + Qn sinh  x  sin  y
n =1   b   b   b 
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
∞
  nπ   nπ   nπ 
u ( x, y ) = ∑  Pn cosh x  + Qn sinh x  sin  y
n =1   b   b   b 
∞
 nπ 
Using the boundary condition: u(0,y) = 0: u (0, y ) = ∑ [Pn cosh 0 + Qn sinh 0]sin  y
n =1  b 
∞
 nπ 
u (0, y ) = 0 = ∑ Pn sin  y , 0 < y < b Pn = 0 for n = 1,2,3,...
n =1  b 
∞
  nπ    nπ 
u ( x, y ) = ∑ Qn sinh  x  sin  y
n =1   b   b  Fourier half-range sine series
∞
  nπ   nπ 
Using the boundary condition: u(a,y) = f(y) f ( y ) = ∑ Qn sinh  a  sin  y
n =1   b   b 
0< y<b

 nπ  2 b  nπ  2 b  nπ 
 ∫0
Qn sinh  a  = ∫ f ( y )sin  y dy Qn = f ( y )sin  y dy
 b  b 0  b   nπ  b 
b sinh a
 b 
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
Let’s consider a specific f(y), say f(y) = 100 ( a constant)
2 b  nπ  − 200  nπ 
b
200(1 − cos nπ )
 nπ  ∫0
Qn = 100 sin  y  dy = cos y =
 b   nπ   b 0  nπ 
b sinh a nπ sinh a n π sinh  a
 b   b   b 
∞
200 1 − cos nπ  nπ   nπ 
u ( x, y ) = ∑
π n=1  nπ 
sinh 
 b
x  sin 
  b
y

1 − cos nπ = 1 − (− 1)n
n sinh  a
 b 

400 ∞
sinh[(2m − 1)πx b]
u ( x, y ) = ∑
π m=1 (2m − 1)sinh[(2m − 1)πa b]
sin[(2m − 1)πy b]
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
400 ∞
sinh[(2m − 1)πx b ]
u ( x, y ) = ∑
π m=1 (2m − 1)sinh[(2m − 1)πa b ]
sin[(2m − 1)πy b]

It is useful to plot u = constant curves for your two dimensional results

If this is heat conduction, we will plot isothermal curves in the rectangle

Notice that u values within the rectangle are

between the minimum and maximum values of u
on the boundary (0 and 100)

MAXIMUM principle of potential theory; if

the Laplace equation holds on a domain D, then
the maximum and minimum values of u occur on
the boundary of D, not on the interior.
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 1
Let’s assume now b = 10a (high-aspect ratio case)

Over 2a < y < 8a, the We can approximate this behavior by

problem is essentially one neglecting uyy in the Laplace equation:
dimensional
∂ 2u u x =0 = 0, u x = a = 100
2
≈0
∂x

x
u ( x, y ) ≈ 100 isotherms of w/c
a are the lines x =
constant

excellent approximation of
Over 2a < y < 8a, u 400 ∞
sinh[(2m − 1)πx b]
varies with x but hardly
u ( x, y ) =
π
∑ (2m − 1)sinh[(2m − 1)πa b] sin[(2m − 1)πy b]
m =1
with y
except near the ends
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 2
∂ 2u ∂ 2u
Find the solution of the 2D Laplace equation 2
+ 2
=0
∂x ∂y
w/c satisfies the boundary conditions: u (0, y ) = 20, 0 < y < ∞
u (5, y ) = 50, 0 < y < ∞
u (x,0 ) = f ( x ), 0 < x < 5
u (x, y ) bounded as y → ∞

Semi-infinite strip; why get interested on semi-infinite

strips?
Temperature distribution on a slender fin (ABCE)
attached to a thick base
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 2
X" Y" X "+ m 2 X = 0 S.L.
Choosing a negative constant –m2: = − = −m 2
X Y Y "− m 2Y = 0 problems

X (x ) = A cos mx + B sin mx Since we have the semi-infinite boundary condition in

Y ( y ) = C cosh my + D sinh my the y direction, it is better to re-express Y(y) as

X (x ) = A cos mx + B sin mx
Y ( y ) = Ce my
+ De − my (
u ( x, y ) = ( A cos mx + B sin mx ) Ce my + De − my )

Choosing a zero constant: X " = 0 X (x ) = E + Fx u (x, y ) = (E + Fx )(G + Hy )

Y"= 0 Y ( y ) = G + Hy

X" Y" X "−k 2 X = 0 S.L.

Choosing a positive constant k 2: = − = k2
X Y Y "+ k 2Y = 0 problems
X (x ) = Ie kx + Je −kx
Y ( y ) = K cos ky + L sin ky ( )
u (x, y ) = Ie kx + Je − kx (K cos ky + L sin ky )
 How to solve the 2D Laplace equation by
method of separation of variables? – Example 2
Apply the boundary condition: u (x, y ) bounded as y→∞

( )
Case of positive constant: u (x, y ) = Ie kx + Je − kx (K cos ky + L sin ky )

In order for u to be bounded, K = L = 0; Y(y) = 0; u(x,y) = 0; trivial solution

Case of zero constant: u (x, y ) = (E + Fx )(G + Hy )

In order for u to be bounded, H = 0; NOT TRIVIAL SOLUTION

(
Case of negative constant: u ( x, y ) = ( A cos mx + B sin mx ) Ce my + De − my )
In order for u to be bounded, C = 0; NOT TRIVIAL SOLUTION

u (x, y ) = (M + Nx ) + (P cos mx + B sin mx )e − my

Apply the boundary condition: u (0, y ) = 20, 0 < y < ∞

u (0, y ) = 20 = M + Pe − my M = 20; P = 0 u (x, y ) = 20 + Nx + (B sin mx )e − my

 How to solve the 2D Laplace equation by
method of separation of variables? – Example 2
Apply the boundary condition: u (5, y ) = 50, 0 < y < ∞

u (5, y ) = 20 + 5 N + (B sin 5m )e − my = 50 20 + 5 N = 50 → N = 6 B sin 5m = 0

π 2π nπ
For a non-trivial solution where B ≠ 0: sin 5m = 0 if m= , , ...,
5 5 5
∞ nπ
 nπ  − y
u (x, y ) = 20 + 6 x + ∑ Bn sin x e 5
n =1  5 

To compute Bn, we use the remaining boundary condition: u (x,0) = f ( x ), 0 < x < 5
∞
∞
 nπ   nπ 
f ( x ) = 20 + 6 x + ∑ Bn sin  x f ( x ) − 20 − 6 x = ∑ Bn sin x
n =1  5  n =1  5 
Fourier half-range sine series

2 5  nπ 
Bn =
5 ∫0
[ f ( x ) − 20 − 6 x ] sin  x  dx
 5 