Heat Conduction ch3 Solution PDF

PROBLEM 3.
Three sides of a rectangular plate L H are y h1 , T h2 , T

maintained at uniform temperature To while the
L/2
fourth side exchanges heat by convection. The heat
transfer coefficient is non-uniform, being h1 over 2 HBC H
To To
half the surface and h2 over the remaining half. The L
ambient temperature is T . Determine the two- 0 x
dimensional steady state temperature distribution. To
(1) Observations. (i) This is a steady state two-

dimensional conduction problem. (ii) Four boundary conditions are needed. (iii) Three
conditions can be made homogeneous by defining a new temperature variable
 ( x, y )  T ( x, y )  To . (iv) The heat transfer coefficient along boundary (x,H) is not uniform.
This requires special consideration.
(2) Origin and Coordinates. The origin and coordinate axes are selected as shown.
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) no energy generation and (4)
constant thermal conductivity.
(ii) Governing Equations. Introducing the above assumptions into eq. (1.8) gives the
heat equation
 2  2
 0 (a)
 x2  y2
where
 ( x, y )  T ( x, y )  To (b)
(iii) Independent Variable with Two Homogeneous Boundary Conditions. The two
boundaries at x = 0 and x = L have homogeneous conditions. Thus, the x-variable has two
homogeneous boundary conditions.
(iv) Boundary Conditions. The boundary conditions are listed in the following order:
starting with the two boundaries in the variable having two homogeneous conditions and
ending with the non-homogeneous condition. Thus, the first three conditions in this order are
homogeneous and the fourth is non-homogeneous. Therefore, we write
(1)  (0, y )  0 , homogeneous
(2)  ( L, y )  0 , homogeneous
(3)  ( x,0)  0 , homogeneous
The fourth boundary condition is first written in terms of the variable T and then
transformed in terms of the variable  . Using Fourier’s law and Newton’s law
T ( x, H ) h1 T ( x, H )  T , 0  x  L/2
k  
y h2 T ( x, H )  T , L/2  x  L
3-1
PROBLEM 3.1 (continued)
Expressed in terms of  , the above gives the fourth boundary condition

 ( x, H ) h1  ( x, H )  (T  To ), 0  x  L / 2
(4)  k  f ( x) =  non-homogeneous
y h2  ( x, H )  (T  To ), L / 2  x  L
This condition represents non-uniform convection.
(4) Solution.
(i) Assumed Product Solution. Assume a solution in the form
 ( x, y ) = X(x) Y(y) (c)
Substituting into (a), separating variables and setting the resulting equation equal to constant
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy
Assuming that n is multi-valued, the above gives

d2Xn
2
 2n X n  0 (d)
dx
and
d 2Yn
2
 2nYn  0 (e)
dy
(ii) Selecting the Sign of the 2n Terms. Since the x-variable has two homogeneous
boundary conditions, the 2n X n term in (d) takes the positive sign. Therefore, (d) and (e)
become
d2Xn
2
 2n X n  0 (f)
dx
and
d 2Yn
2
 2nYn  0 (g)
dy
For the important case of n  0, equations (f) and (g) become
d2X0
0 (h)
d x2
and
d 2Y0
0 (i)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to eqs. (f)-(i) are
X n ( x )  An sin n x  Bn cos n x (j)
3-2
Yn ( y )  Cn sinh n y  Dn coshn y (k)

X 0 ( x )  A0 x  B0 (l)
Y0 ( y )  C0 y  D0 (m)
Corresponding to each value of n there is a temperature solution  n ( x, y ). Thus

 n ( x, y )  X n ( x )Yn ( y ) (n)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (o)
The complete solution becomes

 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (p)
(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions

(j) and (l) gives
Bn  B0  0 (q)
Boundary condition (2) applied to (j) gives the characteristic equation for n
sin n L  0 (r)
Thus n
n
n  (n = 1,2,3,….) (s)
L
Similarly, boundary condition (2) applied to (l) gives
A0  0
With A0  B0  0, the solution corresponding to n  0 vanishes. Application of boundary

condition (3) to (k) gives
Dn  0
Solution (p) becomes

 ( x, y )  
n 1
a (sin  x )(sinh  y )
n n n (t)
where an  An Cn . The only remaining unknown is an. Application of condition (4) gives

f(x) =  k  (a  n n cos hn H ) sin n x (u)
n 1
where f ( x ) is defined in boundary condition (4).

(v) Orthogonality. To determine a n in equation (u) we apply orthogonality. Note that the
characteristic functions  n ( x )  sin n x in (u) are solutions to equation (f). Comparing (f)
with eq. (3.5a) shows that it is a Sturm-Liouville equation with
3-3
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
Thus eq. (3.6) gives
p ( x )  w( x )  1 and q( x )  0
Since the boundary conditions at x = 0 and x = L are homogeneous, it follows that the
characteristic functions  n ( x )  sin n x are orthogonal with respect to the weighting
function w( x )  1. Multiplying both sides of (u) by w( x ) sin m x dx and integrating from x
= 0 to x = L
L  
  
L
f ( x ) w( x ) sin m x dx =  k  (a n coshn H ) sin n x w( x )(sinm x ) dx
0 0  n1 
Using the definition of f ( x ) in condition (4) and noting that w( x )  1, the above gives
0  ( x, H )  (T  To )sin m x dx  h2 L / 2  ( x, H )  (T  To )sin m x dx 

L/2 L
h1
L   (v)
k 0   ( a n n coshn H ) sinn x  sin m x dx
 n 1 
Evaluating (t) at y = H and substituting into (v)
L / 2     

   
L
h1  ( a n sinh n H ) sinn x  sin m x dx  h2  ( a n sinh n H ) sinn x  sin m x dx 
0  n 1  L/2 
 n 1 
L  

  (a 
L/ 2 L
h1 (T  To )
 0
sin m x dx  h2 (T  To )

L/2
sin m x dx  k
0 n 1
n n cosh n H ) sinn x  sin m x dx

Interchanging the integration and summation signs and applying orthogonality, eq. (3.7),
L/2 L
h1a n sinh n H 0 sin2 n x dx  h2 a n sinh n H L / 2 sin n x dx 
2
L/ 2 L L
h1 (T  To ) 0 sin n x dx  h2 (T  To ) L / 2 sin n x dx  kan n coshn H 0 sin n x dx
2
Evaluating the integrals and solving for a n we obtain
an 
   
2(T  To ) (h1 L / k ) 3  ( 1) n1  (h2 L / k ) 1  ( 1) n1  2( 1) n  (w)
n L(h1  h2 )( L / k ) sinh n H  2(n L) coshn H 
(5) Checking. Dimensional check: (i) The arguments of sin, sinh and cosh are
dimensionless. (ii) The coefficient a n in (t) must have units of temperature. Equation (w)
shows that a n has units of temperature.
Limiting check: (i) If h1  h2  0 , no heat can be exchanged at the (x,H) boundary and
consequently the entire plate should be at a uniform temperature To . Setting h1  h2  0 in
equation (w) gives a n  0. When this is substituted into (t) we obtain
3-4
 ( x, y )  T ( x, y )  To  0
which gives T ( x, y )  To .
(ii) If T  To , no heat can be exchanged at the (x,H) boundary and thus the entire plate
should be at a uniform temperature To . Setting T  To in equation (w) gives a n  0. When
this is substituted into (t) we obtain T ( x, y )  To .
Boundary conditions check: Direct substitution into solution (t) shows that boundary
conditions (1), (2) and (3) are satisfied.
Special case: For the special case of h1  h2  h equation (w) gives
an 

2( Lh / h)(T  To ) 1  ( 1) n 
(n L)(hL / k ) sinh n L  (n L) coshn L
Formal determination of a n for this case gives the same result.
(6) Comments. (i) A non-uniform non-homogeneous boundary condition along a boundary
does not introduce added complications. (ii) The solution is expressed in term of hL/k which
appears in a n . This dimensionless parameter is the Biot number. Equation (s) gives the
dimensionless roots n L . However, the argument k H is determined by multiplying and
dividing by L according to k H = (k L)( H / L) . This introduces an additional dimensionless
parameters H / L .
3-5
PROBLEM 3.2
Heat is generated in a rectangular plate 3L H at a
uniform volumetric rate of q . The surface at (0,y) is y T1 T2 T3
maintained at To . Along surface (3L,y) the plate L L L
h
exchanges heat by convection. The heat transfer
q H T
coefficient is h and the ambient temperature is T . To 
Surface (x,H) is divided into three equal segments 3L
0 x
which are maintained at uniform temperatures T1 ,
T2 , and T3 , respectively. Surface (x,0) is insulated. Determine the two-dimensional steady
state temperature distribution.
(1) Observations. (i) This is a steady state two-dimensional conduction problem. (ii) Four
boundary conditions are needed. (iii) Only one boundary condition is homogeneous. (iv)
The energy generation term makes the differential equation non-homogeneous.
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) uniform energy generation and
(4) constant thermal conductivity.
heat equation for this problem
 2T  2T q 
  0 (a)
 x2  y2 k
(iii) Independent Variable with Two Homogeneous Boundary Conditions. Neither
variable has two homogeneous conditions. Only the x-variable can have two homogeneous
conditions.
(iv) Boundary Conditions.
(1) T (0, y )  To , non-homogeneous
 T (3L, y )
(2)  k  hT (3L, y )  T  , non-homogeneous
x
T ( x,0)
(3)  0 , homogeneous
y
T1 , 0 x L

(4) T ( x, H )  f ( x )  T2 , L  x  2 L , non-homogenous

T3 , 2 L  x  3L
This condition represents a boundary with non-uniform temperature given by f ( x ).
(4) Solution. Equation (a) is non-homogeneous due to the heat generation term. Thus we
can not proceed directly with the application of the method of separation of variables.
Instead, we assume a solution of the form
3-6
T ( x, y )   ( x, y )   ( x ) (b)
Note that  ( x, y ) depends on two variables while  ( x ) depends on a single variable.
Substituting (b) into (a)
 2  2 d 2 q 
   0 (c)
 x2  y2 d x2 k
The next step is to split (c) into two equations, one for  ( x, y ) and the other for  (x ). We
select
 2  2
 0 (d)
 x2  y2
Thus, the second part must be
d 2 q 
 0 (e)
d x2 k
Boundary conditions on  ( x, y ) and  ( x ) are obtained by substituting (b) into the four
boundary conditions on T . In this process the function  ( x, y ) is assigned homogeneous
conditions whenever possible. Using (b) and boundary condition (1)
 (0, y )   (0)  To
Let
 (0, y )  0 (d-1)
Thus
 (0)  To (e-1)
Boundary condition (2) and (b) give
 (3L, y ) d (3L)
k k  h (3L, y )  h (3L)  T 
x dx
Let
 (3L, y )
k  h (3L, y ) (d-2)
x
Thus
d (3L)
k  h (3L)  T  (e-2)
dx
 ( x,0)
0 (d-3)
y
Finally, boundary condition (4) gives
 ( x, H )   ( x )  f ( x )
Or
 ( x, H )  f ( x )   ( x ) = F(x)
3-7
Using the definition of f ( x ) in boundary condition (4) the above gives
T1   ( x ) , 0 x L
 ( x, H )  f ( x )   ( x )  F ( x )  T2   ( x ) , L  x  2 L , non-homogeneous (d-4)
T   ( x ) , 2 L  x  3L
 3
Thus non-homogeneous equation (a) is replaced by two equations: (d), which is a
homogeneous partial differential equation and (e) which is a non-homogeneous ordinary
differential equation. Equation (d) has three homogeneous conditions, (d-1), (d-2) and (d-3)
and one non-homogeneous condition, (d-4). The solution to (e) is
q  2
 ( x)   x ExF (f)
2k
where E and F are constants of integration. Application of boundary conditions (e-1) and (e-
2) gives the two constants
(3q L / k )1  (3hL / 2k )  (h / k )(To  T )
E , F  To (g)
1  (3hL / k )
(i) Assumed Product Solution. To solve partial differential equation (d) we assume a
product solution. Let
 ( x, y )  X ( x )Y ( y ) (h)
Substituting (h) into (d), separating variables and setting the resulting two equations equal to
a constant,  2n , we obtain
d2Xn
2
 2n X n  0 (i)
dx
and
d 2Yn
2
 2nYn  0 (j)
dy
(ii) Selecting the Sign of the  2n Terms. Since the x-variable has two homogeneous
boundary conditions, the plus sign is selected in (i). Thus (i) and (j) become
d2Xn
2
 2n X n  0 (k)
dx
and
d 2Yn
2
 2nYn  0 (l)
dy
For the special case of n  0 the above equations take the form
d 2X0
0 (m)
d x2
and
3-8
d 2Y0
0 (n)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to equations (k)-
(n) are
X n ( x )  An sin n x  Bn cosn x (o)
Yn ( y )  Cn sinh n y  Dn coshn y (p)
X 0 ( x )  A0 x  B0 (q)
and
Y0 ( y )  C0 y  D0 (r)
Corresponding to each value of n there is a solution  n ( x, y ). Thus

 n ( x, y )  X n ( x )Yn ( y ) (s)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (t)
The complete solution to T ( x, y ) becomes

T ( x, y)   ( x)  X 0 ( x)Y0 ( y)  X
n 1
n ( x)Yn ( y ) (u)
(iv) Application of Boundary Conditions. Applying boundary condition (d-1) to

equation (s) and (t) gives
Bn  B0  0 (v)
Condition (d-2) applied to (s) gives the characteristic equation for n

 3n L  (3hL / k ) tan 3n L (w)
Condition (d-2) applied to (t)
A0  0 (x)
Boundary conditions (d-3) and (s) give
Cn  0 (y)
With A0  B0  0, the solution X 0Y0 vanishes. Substituting into (s) and using (v) and (y),
the solution to  ( x, y ) becomes

 ( x, y )   a n (coshn y ) sin n x (z)
n 1
where a n  An Dn . Thus the only remaining unknown is a n . Application of condition (d-4)

gives

F ( x)  a n (coshn H ) sin n x (z-1)
n 1
3-9
where F(x) is defined in (d-4). To determine a n from this result we apply orthogonality.
(v) Orthogonality. The characteristic functions, sin n x, in (z-1) are solutions to

equation (k). Comparing (k) with eq. (3.5a) shows that it is a Sturm-Liouville equation with
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
p( x )  w( x )  1 and q( x )  0
Since boundary conditions (d-1) and (d-2) at x = 0 and x = 3L are homogeneous, it follows
that the characteristic functions sin n x are orthogonal with respect to w = 1. Multiplying
both sides of (aa) by sin m x dx , integrating from x = 0 to x = 3L
  
 
 
3L 3L
 F ( x)w( x) sin 
0
m x dx =
0


 n1
( a n coshn L ) sin n w( x )(sin m x ) dx
x


Interchanging the integration and summation signs, noting that w( x )  1 and applying
orthogonality, the above gives
3L 3L
 0
F ( x ) sin n x dx = a n coshn H 
0
sin2 n x dx
Evaluating the integral on the right-hand-side and solving the above for a n
3L
an 
0
F ( x ) sin n xdx
(z-2)
(1 / 2n ) coshn H 3n L  (1 / 2) sin 6n L
The integral in (z-2) is evaluated using the definition of F(x) in (d-4) and the solution to
 ( x ) given in (f). Thus
 T 
3L L
 0
F ( x ) sin n xdx 
0
1  ( q  / 2k ) x 2  Ex  F ) sin n xdx 
 T  (q / 2k ) x  Ex  F )sin  xdx 

2L
2
2 n
L
 T  (q / 2k ) x  Ex  F )sin  xdx 

3L
2
3 n
2L
where E and F are given in equation (g). Evaluating each of the three integrals in the above
 T 
L
1  (q  / 2k ) x 2  Ex  F sin  n xdx  ( F  T1 )(1 /  n )(cos n L  1) 
 
0
(q  / 2k3n ) 2 n L sin  n L  ( 2n L2  2)(cos n L)  2  ( E /  2n )sin  n L   n L cos  n L 

(z-3)
3-10
 T 
2L
2  (q  / 2k ) x 2  Ex  F sin  n xdx  ( F  T2 )(1 /  n )(cos 2 n L  cos  n L) 

L
 
(q  / 2k3n ) 4 n L sin 2 n L  (42n L2  2)(cos 2 n L)  2 n L sin  n L  (2n L2  2)(cos  n L) 
sin 2 n L  2 n L cos 2 n L  sin  n L   n L cos  n L
( E / 2n ) (z  4)
and
 T  (q / 2k )x 
3L
3
2
 Ex  F sin  n xdx  ( F  T3 )(1 /  n )(cos 3 n L  cos 2 n L) 
 
2L
(q  / 2k3n ) 6 n L sin 3 n L  (92n L2  2)(cos 3 n L)  4 n L sin 2 n L  (42n L2  2)(cos 2 n L) 
( E / 2n )sin 3 n L  3 n L cos 3 n L  sin 2 n L  2 n L cos 2  nL (z  5)
(5) Checking. Dimensional checks: (i) The coefficient a n in (z-2) should have units of
temperature. Since F(x) has units of temperature and n is measured in (1/m), (z-2) has units
of temperature. (ii) Each term in (z-3), (z-4) and (z-5) has units of ( o C  m ) .
Limiting check: For the special case of q   0 and T1  T2  T3  T  To , physical
consideration requires that the plate be at a uniform temperature To . For this case E = 0 and
 ( x )  To . Substitution into (z-2) gives a n  0. Equation (u) gives T ( x, y )  To .
Boundary conditions check: Boundary conditions (1) and (3) are readily shown to be
satisfied.
(6) Comments. (i) The same approach can be used to solve the more general problem where
the heat generation rate is variable, q   q (x ) . This will affect the solution of  ( x ) and the
integral in (z-2). (ii) The solution is expressed in terms of the Biot number, hL/k, which
appears in equation (w).
3-11
PROBLEM 3.3
A rectangular plate L H is maintained at uniform y qo

temperature To along three sides. Half the fourth side is
insulated while the other half is heated at uniform flux L/2
qo. Determine the steady state heat transfer rate To
2 HBC H
To
through surface (0,y). L
0 x
To
(1) Observations. (i) To determine the heat transfer
through surface (0,y) it is necessary to first determine the temperature distribution in the
plate. (ii) This is a steady state two-dimensional conduction problem. (iii) Four boundary
conditions are needed. (iv) Three conditions can be made homogeneous by defining a new
temperature variable  ( x, y )  T ( x, y )  To . (v) The heat flux along boundary (x,H) is non-
uniform. This requires special consideration.
(3) Formulation.
heat equation for this case as
 2  2
 0 (a)
 x2  y2
where
 ( x, y )  T ( x, y )  To (b)
Once the temperature distribution is determined, Fourier’s law give the heat transfer rate
through surface (0,y). Thus
H
T (0, y )
q(0)  k
0
 x
dy (c)
where q (0) is the heat transfer rate through surface (0,y)

(iii) Independent Variable with Two Homogeneous Boundary Conditions. The
two boundaries at x = 0 and x = L have homogeneous conditions. Thus, the x-variable has
two homogeneous boundary conditions.
(iv) Boundary Conditions. The required four boundary conditions are
(1)  (0, y )  0 , homogeneous
(2)  ( L, y )  0 , homogeneous
The fourth boundary condition is first written in terms of the variable T and then
transformed in terms of the variable  . Using Fourier’s law
3-12
T ( x, H ) 0 , 0  x  L/2
k  
y  qo , L/2  x  L
Expressed in terms of  , the above gives the fourth boundary condition

 ( x, H ) 0, 0  x  L/2
(4)  k  f ( x) =  non-homogeneous
y  qo , L/2  x  L
This condition represents a heat flux which varies along the boundary according to f ( x ).
(4) Solution.
 ( x, y ) = X(x) Y(y) (d)
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy
d2Xn
2
 2n X n  0 (e)
dx
and
d 2Yn
 2nYn  0 (f)
d y2
boundary conditions the 2n X n term in (e) takes the positive sign. Therefore, (e) and (f)
become
d2Xn
2
 2n X n  0 (g)
dx
and
d 2Yn
2
 2nYn  0 (h)
dy
For the important case of n  0, equations (g) and (h) become
d2X0
0 (i)
d x2
and
d 2Y0
0 (j)
d y2
3-13
(iii) Solutions to the Ordinary Differential Equations. The solutions to eqs. (g)-(j) are
X n ( x )  An sin n x  Bn cos n x (k)
Yn ( y )  Cn sinh n y  Dn coshn y (l)
X 0 ( x )  A0 x  B0 (m)
Y0 ( y )  C0 y  D0 (n)
Corresponding to each value of  n there is a temperature solution  n ( x, y ). Thus

 n ( x, y )  X n ( x )Yn ( y ) (o)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (p)

 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (q)

(k) and (m) gives
B n  Bo  0 (r)
Boundary condition (2) applied to (k) gives the characteristic equation for n
sin n L  0 (s)
Or
n
n  n = 1,2,3…. (t)
L
Similarly, boundary condition (2) applied to (m) gives
A0  0

condition (3) to (l) gives
Dn  0
Solution (q) becomes

 ( x, y )   a n (sin n x )(sinh n y ) (u)
n 1
where an  An Cn . The only remaining unknown is the set of constants an. Application of
condition (4) gives

f(x) =  k  (a  n n cos hn H ) sin n x (v)
n 1
3-14
(v) Orthogonality. To determine a n in equation (v) we apply orthogonality. Note that the
characteristic functions  n ( x )  sin n x in (v) are solutions to equation (g). Comparing (g)
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
p ( x )  w( x )  1 and q( x )  0
function w( x )  1. Multiplying both sides of (v) by w( x ) sin m x dx and integrating from x
= 0 to x = L
L L   
0
 0
 n 1
 
f ( x ) w( x ) sin m x dx =  k  (a n n coshn H ) sin n x w( x ) sin m x dx

Interchanging the summation with integration and invoking orthogonality gives
L L
0
f ( x ) w( x ) sin n x dx =  k an n coshn H
0
w( x ) sin2 n x
Solving for a n and recalling that w(x) = 1 the above gives

L
0 f ( x) sin n x dx 2 L
an    f ( x) sin n x dx
 kn coshn H  sin2 n x dx kn L coshn H 0
L (w)
0
Using the definition of f ( x ) in condition (4) the integral in (w) is evaluated
L L/2 L
0
f ( x ) sin n x dx  
0
(0) sin n x dx  
L/2
qo sin n x dx  ( q  / n )(cosn L  cosn L / 2)
Substituting into (w) and using (t) to eliminate n L and (cosn L  cosn L / 2) , we obtain
an 
qoL / k
(n ) (coshn H )
2

( 1) n 1  2( 1) n  1  (x)
Substituting (x) into (u) and rearranging the result gives the dimensionless temperature
distribution in the plate

T ( x, y )  To ( 1) n 1  2( 1) n  1
qoL / k
  ( n ) 2 (coshn H )
(sin n x ) sinh n y (y)
n 1
To determine the heat transfer rate through surface (0,y) we substitute (y) into (c)
3-15
H 
(1) n 1  2(1) n  1
q(0)  k (q o L / k ) 
0 n 1 (n ) 2 (cosh  n H )
 n (cos  n 0) sinh  n ydy 
H 
(1) n 1  2(1) n  1

H
 q o L
0 n 1 (n ) 2 (cosh  n H )  sinh
0
n yd ( n y )
Evaluating the integral
H 
(1) n 1  2(1) n  1
q(0)  q o L 
0 n 1 (n ) 2 (cosh  n H )
cosh  n H  1 (z)
(5) Checking. Dimensional check: (i) The arguments of the sin, sinh and cosh are
dimensionless. (ii) Each term in solution (y) is dimensionless.(iii) Equation (z) has the
correct units of heat transfer rate per unit length.
Limiting check: If qo  0 the entire plate should be at a uniform temperature To . Setting
qo  0 in (y) gives T ( x, y )  To .
Boundary conditions check: Conditions (1), (2) and (3) can be shown to be satisfied by
direct substitution.
the heat flux at the boundary is variable, qo  qo (x ). This will affect the integral in (w). (ii)
Equation (t) gives the dimensionless roots n L . However, the argument k H is determined
by multiplying and dividing by L according to k H = (k L)( H / L) . Thus the solution is
characterized by a single dimensionless parameters H / L .
3-16
PROBLEM 3.4
A semi-infinite plate of width 2L is maintained at y To 2 HBC

uniform temperature To along its two semi-infinite
sides. Half the surface (0,y) is insulated while the qo
L 
remaining half is heated at a flux qo. Determine
the two-dimensional steady state temperature L
distribution.
0 To x
(1) Observations. (i) This is a steady state two-
dimensional conduction problem. (ii) Four boundary conditions are needed. (iii) Far away
from the heated end the temperature should be uniform equal to To . (iv) Three conditions
can be made homogeneous by defining a new temperature variable  ( x, y )  T ( x, y )  To .
(v) The heat flux along boundary (0,y) is non-uniform. This requires special consideration.
(3) Formulation.
heat equation for this as
 2  2
 0 (a)
 x2  y2
where
 ( x, y )  T ( x, y )  To (b)

two boundaries at y = 0 and y = 2L have homogeneous conditions. Thus, the y-variable has
(2)  ( x,2 L)  0 , homogeneous
(3)  ( , y )  0 = finite, homogeneous
 (0, y ) 0, 0 x L
(4)  k  f ( y) =  non-homogeneous
x qo , L  x  2L
This condition represents a heat flux which varies along the boundary according to f ( y ).
(4) Solution.
 ( x, y ) = X(x) Y(y) (c)
3-17
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy
d2Xn
2
 2n X n  0 (d)
dx
and
d 2Yn
2
 2nYn  0 (e)
dy
(ii) Selecting the Sign of the 2n Terms. Since the y-variable has two homogeneous
boundary conditions the 2nYn term in (e) takes the positive sign. Therefore, (d) and (e)
become
d2Xn
2
 2n X n  0 (f)
dx
and
d 2Yn
2
 2nYn  0 (g)
dy
d2X0
0 (h)
d x2
and
d 2Y0
0 (i)
d y2
X n ( x )  An exp( n x )  Bn exp( n x ) (j)
Yn ( y )  Cn sin n y  Dn cosn y (k)
X 0 ( x )  A0 x  B0 (l)
Y0 ( y )  C0 y  D0 (m)

 n ( x, y )  X n ( x )Yn ( y ) (n)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (o)
3-18


 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (p)

(k) and (m) gives
D n  Do  0 (q)
sin 2n L  0 (r)
Thus n
n
n  (n = 1,2,3….) (s)
2L
C0  0
With C0  D0  0, the solution corresponding to n  0 vanishes. Application of boundary
condition (3) to (j) gives
Bn  0

 ( x, y )   a n exp( n x )sin n y (t)
n 1
condition (4) gives

f(y) = k  an n sin n y (u)
n 1
where f ( y ) is defined in boundary condition (4).

characteristic functions  n ( y )  sin n y in (u) are solutions to equation (g). Comparing (g)
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p ( y )  w( y )  1 and q( y )  0
Since the boundary conditions at y = 0 and y = 2L are homogeneous, it follows that the
characteristic functions  n ( y )  sin n y are orthogonal with respect to the weighting
function w( y )  1. Multiplying both sides of (u) by w( y ) sin m y dy and integrating from y
= 0 to y = L
3-19
2L 2L  
 f ( y ) w( y ) sin m y dy = k  
 a n n sin n y w( y ) sin m y dy
0 0
 n 1 
2L 2L
 0
f ( y ) w( y ) sin n y dy = k an n 0 w( y ) sin2 n y
Solving for a n and recalling that w(y) = 1 the above gives

2L
 f ( y ) sin n y dy 2L

1
an  0
 f ( y ) sin n y dy
2L
kn L
kn  sin n y dy
2 0
0
Evaluating the integral in the denominator, the above gives

2L

1
an  f ( y ) sin n y dy (v)
kn L 0
Using the definition of f ( y ) in condition (4) the integral in (v) is evaluated

2L L 2L
0
f ( y ) sin n y dy   0
(0) sin n y dy   L
qo sin n y dy  ( q  / n )(cos2n L  cosn L)
Substituting into (v) and using (s) to eliminate n L and (cosn L  cosn L / 2) , we obtain
an 
2qo L / k
(n ) 2
(1) n 1
 2(1) n  1  (w)
Substituting (w) into (t) and rearranging the result gives


T ( x, y )  To ( 1) n 1  2( 1) n  1 n x
qoL / k
2  (n ) 2
e sin n y
n 1
(5) Checking. Dimensional check: According to (t) the coefficient a n should have units of
temperature. The term qoL / k in (w) has units of temperature.
Limiting check: If qo  0 , the entire plate should be at a uniform temperature To . Setting
qo  0 in (w) gives a n  0. When this is substituted into (t) gives   0 , or T ( x, y )  To .
(6) Comments. The same approach can be used to solve the more general problem where
the heat flux at the boundary is variable, qo  qo ( y ). This will affect the integral in (v).
3-20
PROBLEM 3.5
Consider two-dimensional steady state conduction y h, T
in a rectangular plate L  H . The sides (0,y) and
H /2 q2
(x,H) exchange heat by convection. The heat h
transfer coefficient is h and ambient temperature is T 2 HBC
H /2 q1
T . Side (x,0) is insulated. Half the surface at x = L
L
is cooled at a flux q1 while the other half is heated 0 x
at a flux q2. Determine the heat transfer along
(0,y).
(1) Observations. (i) This is a steady state two-dimensional conduction problem. (ii) Heat
transfer rate along surface (0,y) is determined using temperature distribution and Fourier’s
law. (iii) Four boundary conditions are needed. (iv) Two conditions can be made
homogeneous by defining a new temperature variable  ( x, y )  T ( x, y )  T . (v) The heat
flux along boundary (L,y) is not uniform. This requires special consideration.
(3) Formulation.
(ii) Governing Equations. To determine the rate of heat transfer we must first
determine the temperature distribution in the plate. Introducing the above assumptions into
eq. (1.8) gives the heat equation for this case
 2  2
 0 (a)
 x2  y2
where
 ( x, y )  T ( x, y )  T (b)
two boundaries at y = 0 and y = H have homogeneous conditions. Thus, the y-variable has
(iv) Boundary Conditions. The boundary conditions are
 ( x,0)
(1) 0, homogeneous
y
 ( x, H )
(2)  k  h ( x , H ) , homogeneous
y
 (0, y )
(3)  k  h (0, y ) , homogeneous
x
 ( L , y )  q , 0  y  H /2
(4)  k  f(y) =  1 non-homogeneous
x  q2 , H /2  y  H
This condition represents a heat flux, which varies along the boundary according to f ( y ).
3-21
(4) Solution.
 ( x, y ) = X(x) Y(y) (c)
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy
d2Xn
2
 2n X n  0 (d)
dx
and
d 2Yn
2
 2nYn  0 (e)
dy
boundary conditions the 2nYn term in (e) takes the positive sign. Therefore, (d) and (e)
become
d2Xn
2
 2n X n  0 (f)
dx
and
d 2Yn
2
 2nYn  0 (g)
dy
d2X0
0 (h)
d x2
and
d 2Y0
0 (i)
d y2
X n ( x )  An sinh n x  Bn coshn x (j)
X 0 ( x )  A0 x  B0 (l)
Y0 ( y )  C0 y  D0 (m)
 n ( x, y )  X n ( x )Yn ( y ) (n)
3-22
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (o)

 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (p)

(k) and (m) gives
C n  C0  0 (q)
Boundary condition (2) applied to (k) gives the characteristic equation for n
n H tan n H  (hH / k ) (r)
D0  0
With C0  D0  0, the solution corresponding to n  0 vanishes. Application of boundary
condition (3) to (j) gives
Bn  ( n H )(k / hH ) An

 ( x, y )  
n 1
a sinh  x  ( H )(k / hH ) cosh  xcos 
n n n n ny (s)
where a n  An Dn . The only remaining unknown is the set of constants an. Application of
condition (4) gives

f(y) =  k 
n 1
a  cosh  L  ( H )(k / hH ) sinh  Lcos 
n n n n n ny (t)
where f ( y ) is defined in boundary condition (4).

(v) Orthogonality. To determine a n in equation (t) we apply orthogonality. Note that
the characteristic functions  n ( y )  cosn y in (t) are solutions to equation (g). Comparing
(g) with eq. (3.5a) shows that it is a Sturm-Liouville equation with
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p ( y )  w( y )  1 and q( y )  0
Since the boundary conditions at y = 0 and y = H are homogeneous, it follows that the
characteristic functions  n ( y )  cosn y are orthogonal with respect to the weighting
function w(y) = 1. Multiplying both sides of (t) by w( y ) cosm ydy and integrating from
y  0 to y  H
3-23
  
 
H H
0
f ( y ) w( y ) cos m ydy  k
0

 n 1
ann cosh n L  (n H )(k / hH ) sinh n Lcos n y w( y ) cos m y dy

Interchanging the summation with integration, recalling that w( y )  1 and invoking

orthogonality, eq. (3.7), gives
f ( y ) cos  n ydy  kan  n cosh  n L n H  (k / hH ) sinh  n L

H H
 0
0
cos2  n y dy
Solving for a n the above gives

H
an 
 
0
f ( y ) cos  n y dy
k n cosh  n L  ( n H )(k / hH ) sinh  n L

H
0
cos2  n y dy
Evaluating the integral in the denominator, the above becomes

H
an 
4  0
f ( y ) cos  n y dy
(u)
k cosh  n L  ( n H )(k / hH ) sinh  n L(2 n H  sin 2 n H )
Using the definition of f ( y ) in condition (4) the integral in (u) is evaluated

H H /2 H
0 f ( y ) sin n y dy  q1 0 cosn y dy  q2 H / 2 cosn y dy 
( q1 / n ) sin n H / 2  ( q2 / n )(sinn H  sin n H / 2)
Substituting into (u)
 4(q1H / k ) sin  n H / 2  (q 2 H / k )(sin  n H  sin  n H / 2)

an  (v)
( n H )cosh  n L  ( n H )(k / hH ) sinh  n L(2 n H  sin 2 n H )
Substituting (v) into (s) and rearranging the result gives


sin(n H / 2)  (q2 / q1)(sin n H  sin n H / 2)sinh n x  (n H )(k / hH ) coshn x cos  y
T ( x, y )  T
q1H / k
 4 n 1
(n H )cosh n L  (n H )(k / hH ) sinh n L (2n H  sin 2n H )
n
(w)
Having determined the temperature distribution in the plate, the heat transfer rate through
surface (0,y) can be determined using Fourier’s law as follows
H T (0, y )
q (0)  k o x
dy
where q (0) is heat transfer rate through (0,y) per unit plate depth. Substituting (s) into the
above
3-24

a
H

q (0)  k a n  n cos  n ydy  k
0
n 1
n sin  n H
(5) Checking. Dimensional check: (i) The arguments of sin, cos, sinh and cosh are
dimensionless. (ii) The Biot number, hH/k, is dimensionless. (iii) According to (s), the
coefficient a n should have units of temperature. The term q  / n k in (v) gives a n the
correct units.
Limiting check: If q1  q2  0 , the entire plate should be at a uniform temperature T .
Setting q1  q2  0 in (w) gives T ( x, y)  T .
(6) Comments. (i) If the direction of q1 and/or q 2 is reversed, solution (w) is still
applicable. However, the signs of q1 and q 2 must be changed accordingly. The same
approach can be used to solve the more general problem where the heat flux at the boundary
is variable, qo  qo ( y ). (ii) Equation (r) gives the dimensionless roots  k H in terms of the
Biot number hH/k. However, the argument  k L is determined by multiplying and dividing
by L. This introduces an additional parameter H / L. Thus examining solution (w) we note
that the temperature distribution depends on three parameters: hH/k, H/L and q2 / q1 .
3-25
PROBLEM 3.6
Radiation strikes one side of a rectangular element y T1

L  H and is partially absorbed. The absorbed energy
results in a non-uniform volumetric energy generation
given by q  Ao e   y H qo
L
q  Ao e   y 0 x
T2
where Ao and  are constant. The surface which is radiation
exposed to radiation, (x,0), is maintained at T2 while the
opposite surface (x,H) is at T1. The element is insulated along surface (0,y) and cooled
along (L,y) at a uniform flux qo . Determine the steady state temperature of the insulated
surface.
(1) Observations. (i) This is a steady state two-dimensional conduction problem. (ii) Four
boundary conditions are needed. (iii) Only one boundary condition is homogeneous. (iv)
The energy generation term makes the differential equation non-homogeneous. (v) Energy
generation varies in the y-direction.
(3) Formulation.
` (i) Assumptions. (1) Two-dimensional, (2) steady and (3) constant thermal
conductivity.
(ii) Governing Equations. Introducing the above assumptions into eq. (1.8) gives
 2T  2T q 
  0 (a)
 x2  y2 k
where
q  Ao e   y (b)
 2T  2T Ao   y
  e 0 (c)
 x2  y2 k
variable has two homogeneous conditions. Only the y-variable can have two homogeneous
conditions.
(iv) Boundary Conditions.
(1) T ( x,0)  T2 , non-homogeneous
(2) T ( x, H )  T1 , non-homogeneous
 T (0, y )
x
T ( L, y )
(4)  k  qo , non-homogeneous
x
3-26
(4) Solution. Equation (c) is non-homogeneous due to the heat generation term. Thus we
can not proceed directly with the application of the method of separation of variables.
Noting that the heat generation term is a function of y, we assume a solution of the form
T ( x, y )   ( x, y )   ( y ) (d)
Note that  ( x, y ) depends on two variables while  ( y ) depends on a single variable.
Substituting (d) into (c)
 2  2 d 2 Ao   y
   e 0 (e)
 x2  y2 d y2 k
The next step is to split (c) into two equations, one for  ( x, y ) and the other for  ( y ). We
select
 2  2
 0 (f)
 x2  y2
Thus, the second part must be
d 2 Ao   y
 e 0 (g)
d y2 k
Boundary conditions on  ( x, y ) and  ( y ) are obtained by substituting (d) into the four
boundary conditions. In this process the function  ( x, y ) is assigned homogeneous
conditions whenever possible. Using (d) and boundary condition (1)
 ( x,0)   (0)  T2
Let
 (0, y )  0 (f-1)
Thus
 (0)  T2 (g-1)
 ( x, H )   ( H )  T1
Let
 ( x, H )  0 (f-2)
Thus
 ( H )  T1 (g-2)
 (0, y )
0 (f-3)
x
 ( L, y )
k  qo (f-4)
x
Thus, non-homogeneous equation (c) is replaced by two equations: (f), which is a
homogeneous partial differential equation and (g) which is a non-homogeneous ordinary
3-27
differential equation. Equation (f) has three homogeneous conditions, (f-1), (f-2) and (f-3)
and one non-homogeneous condition, (f-4). The solution to (g) is
Ao
 ( y)   e  y  E y  F (h)
 k
2
where E and F are constants of integration. Application of boundary conditions (g-1) and
(g-2) gives the two constants
 A  A
E1  (1 / H ) (T1  T2 )  2o (e   H  1) and F  T2  2o (i)
  k   k
Substituting (i) into (h)
Ao   y  A  A
 ( x)   e  (T1  T2 )  2o (e   H  1)( y / H ) + T2  2o (j)
 k
2
  k   k
(i) Assumed Product Solution. To solve partial differential equation (f) we assume a
product solution. Let
 ( x, y )  X ( x )Y ( y ) (k)
Substituting (k) into (f), separating variables and setting the resulting two equations equal to
a constant,  2n , we obtain
d2Xn
2
 2n X n  0 (l)
dx
and
d 2Yn
2
 2nYn  0 (m)
dy
(ii) Selecting the Sign of the  2n Terms. Since the y-variable has two homogeneous
boundary conditions, the plus sign is selected in (m). Thus (l) and (m) become
d2Xn
2
 2n X n  0 (n)
dx
and
d 2Yn
2
 2nYn  0 (o)
dy
For the special case of n  0 the above equations take the form
d 2X0
0 (p)
d x2
and
d 2Y0
0 (q)
d y2
3-28
(iii) Solutions to the Ordinary Differential Equations. The solutions to equations (n)-
(q) are
X n ( x )  An sinh n x  Bn coshn x (r)
Yn ( y )  Cn sin n y  Dn cosn y (s)
X 0 ( x )  A0 x  B0 (t)
and
Y0 ( y )  C0 y  D0 (u)
Corresponding to each value of n there is a solution  n ( x, y ). Thus

 n ( x, y )  X n ( x )Yn ( y ) (v)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (w)
The complete solution to T ( x, y ) becomes

T ( x, y )   ( x)  X 0 ( x)Y0 ( y )  X
n 1
n ( x)Yn ( y ) (x)
(iv) Application of Boundary Conditions. Applying boundary condition (f-1) to

equation (s) and (u) gives
Dn  D0  0
Condition (f-2) applied to (s) gives the characteristic equation for n

sin n H  0
Thus
n  n / H , (n  1, 2, 3....) (y)
Condition (f-2) applied to (u)
C0  0
Boundary condition (f-3) and (r) give
An  0
With C0  D0  0, the solution X 0Y0 vanishes. Substituting into (x) and using (r) and (s),
the solution to  ( x, y ) becomes

 ( x, y )   a n (coshn x ) sin n y (z)
n 1
where a n  Bn Cn . Thus the only remaining unknown is a n . Application of boundary

condition (f-4) gives

qo  k a n ( n sinh n L) sin n y (z-1)
n 1
To determine a n from this result we apply orthogonality.
3-29
(v) Orthogonality. The characteristic functions, sin n y, in (z-1) are solutions to
equation (o). Comparing (o) with eq. (3.5a) shows that it is a Sturm-Liouville equation with
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p ( y )  w( y )  1 and q( y )  0
Since boundary conditions (f-1) and (f-2) at y = 0 and y = H are homogeneous, it follows
that the characteristic functions sin n y are orthogonal with respect to w(y) = 1. Multiplying
both sides of (z-1) by w(y) sin m y dy and integrating from y = 0 to y = H
H
 
 
H
 qw( y) sin 
0
o m y dy =k
0
 (a n n sinhn L) sin n y w( y )(sinm y ) dy
 n1 
Interchanging integration the integration and summation signs, noting that w( y )  1 and
applying orthogonality, the above gives
H H
 0
q o sin  n ydy =  a n  n sinh  n H  0
sin 2  n y dy
Evaluating the integrals, solving the above for a n and using (y) gives
an 

2( qoH / k ) ( 1) n  1  (z-2)
(n ) 2 sinh n L
The solution to the temperature distribution is obtained by substituting (j), (z) and (z-2) into
(d)
A  A  A
T ( x, y )   2o e   y  (T1  T2 )  2o (e   H  1)( y / H ) + T2  2o +
 k   k   k

(1)  1
n
2( qoH / k )  (n ) 2
sinh n L
(coshn x ) sin n y (z-3)
n 1
Expressed in dimensionless form the becomes
T ( x, y )  T2  y  (T1  T2 )  H 
 1  e    ( e  1) ( y / H ) +
( Ao /  2 k )  Ao /  k
2

q  H 

( 1)  1
n

2 o 2 
Ao /  n 1 (n ) 2 sinh n L
(coshn x ) sin n y (z-4)
(5) Checking. Dimensional checks: (i) The arguments of sin, sinh and cosh are
dimensionless. (ii). Each term in (z-4) is dimensionless.
Limiting check: For the special case of qo  0 temperature distribution should be one-
dimensional. Setting qo  0 in (z-3) or (z-4) gives the one-dimensional solution to the
problem. If we further assume that there is no energy generation, (z-3) gives
3-30
T ( y )  T2  (T1  T2 )( y / H )
This is recognized as the linear one-dimensional solution in a plate with two boundaries at
specified temperatures.
(6) Comments. (i) Selecting  in the assumed solution (d) to be a function of y rather than
x is motivated by the fact that the energy generation term is a function of y. (ii) Equation (y)
gives the dimensionless roots k H . However, the argument k L is determined by
multiplying and dividing by L. This introduces H / L as a parameter, Thus examining
solution (z-4) we note that the temperature distribution depends on four parameters:
T1  T2 qoH H
, ,  H and
Ao /  2 k Ao /  2
L
3-31
PROBLEM 3.7
An electric strip heater of width 2b dissipates energy
at a flux q o . The heater is mounted symmetrically on y
one side of a plate of height H and width 2L. Heat is b b
exchanged by convection along surfaces (0,y) and h h
(2L,y). The ambient temperature is T and the heat T H T

transfer coefficient is h. Surfaces (x,0) and (x,H) are L L
insulated. Determine the temperature of the plate- 0 x
heater interface for two-dimensional steady state
conduction.
(1) Observations. (i) To determine the plate-heater interface temperature it is necessary to
first determine the temperature distribution in the plate. (ii) This is a steady state two-
dimensional conduction problem. (iii) Four boundary conditions are needed. (iv)
Temperature distribution is symmetrical about the vertical centerline. This makes it possible
to analyze half the plate. (v) Only the x-variable can have two homogeneous conditions. (vi)
Defining a new temperature variable  ( x, y )  T ( x, y )  T makes the convection condition
at (L,y) homogeneous. (vii) The heat flux along (x,H) is non-uniform. This requires special
consideration.
y
(2) Origin and Coordinates. Because half the plate is to be
considered, the origin is selected at the centerline and the b
coordinate axes are directed as shown. qo
H Th

(3) Formulation. 2HBC
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) no L
energy generation and (4) constant thermal conductivity. 0 x
(ii) Governing Equations. Introducing the above
assumptions into eq. (1.8) gives the heat equation
 2  2
 0 (a)
 x2  y2
where
 ( x, y )  T ( x, y )  T (b)
two boundaries at x = 0 and x = L have homogeneous conditions. Thus, the x-variable has
 (0, y )
(1)  0, homogeneous
x
 ( L, y )
(2)   h ( L, y ), homogeneous
x
 ( x ,0)
y
3-32
 ( x, H )   q  , 0 xb
(4)  k  f(x) =  o
y 0, b x L
This condition represents a heat flux which varies along the boundary according to f ( x ).
(4) Solution.
 ( x, y ) = X(x) Y(y) (c)
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy
d2Xn
2
 2n X n  0 (d)
dx
and
d 2Yn
2
 2nYn  0 (e)
dy
boundary conditions the 2n X n term in (d) takes the positive sign. Therefore, (d) and (e)
become
d2Xn
2
 2n X n  0 (f)
dx
and
d 2Yn
2
 2nYn  0 (g)
dy
d2X0
0 (h)
d x2
and
d 2Y0
0 (i)
d y2
X n ( x )  An sin n x  Bn cos n x (j)
Yn ( y )  Cn sinh n y  Dn coshn y (k)
3-33
X 0 ( x )  A0 x  B0 (l)
Y0 ( y )  C0 y  D0 (m)
 n ( x, y )  X n ( x )Yn ( y ) (n)
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (o)

 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (p)

(j) and (l) gives
An  A0  0 (q)
n L tan n L  hL / k  Bi (r)
Similarly, boundary condition (2) applied to (l) gives
B0  0

condition (3) to (k) gives
Cn  0 (s)

 ( x, y )   a n (cosn x )(coshn y ) (t)
n 1
where a n  Bn Dn . The only remaining unknown is the set of constants an. Application of
condition (4) gives

f(x) =  k  (a  n n sinh n H ) cosn x (u)
n 1

characteristic functions  n ( x )  cosn x in (u) are solutions to equation (f). Comparing (f)
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
p ( x )  w( x )  1 and q( x )  0
3-34
characteristic functions  n ( x )  cosn x are orthogonal with respect to the weighting
function w( x )  1. Multiplying both sides of (u) by w( x ) sin m x dx and integrating from x
= 0 to x = L
L  
 
L
0
f ( x ) w( x ) cosm x dx =  k
0
 (a n n sinh n H ) cosn x w( x ) cosm x dx
 n 1 
L L
0
f ( x ) w( x ) cosn x dx =  k a n n sinh n H 0
w( x ) cos2 n x
Solving for a n , recalling that w(x) = 1 and evaluating the integral on the right-hand-side, the
above gives
L
an 
4 0 f ( x) cosn x dx (v)
 k sinh n H ( 2n L  sin 2n L)
Using the definition of f ( x ) in condition (4) the integral in (v) is evaluated
L b L
 0
f ( x ) cosn x dx  
0
(  qo ) cosn x dx  b
(0)n x dx  ( qo / n ) sin n b
Substituting into (v)

4 qo sin n b
an  (w)
kn sinh n H ( 2n L  sin 2n L)
The solution to the temperature distribution is obtained by substituting (b) and (v0 into (t)

T ( x, y )  T sin n b
( qoL / k )
4   L(sinh n H )(2n L  sin 2n L)
(cosn x ) coshn y (x)
n 1 n
To determine the plate-heater interface temperature set y = H in (x)


T ( x, H )  T (cothn H )(sin n b)
( qoL / k )
4
 L ( 2  L  sin 2  L )
cosn x 0 xb (y)
n 1 n n n
dimensionless. (ii) The term qoL / k in (x) has units of temperature. Thus both sides of (x)
are dimensionless.
Limiting check: If qo  0 , the entire plate should be at a uniform temperature T .
Multiplying (x) through by ( qoL / k ) and setting qo  0 gives T ( x, y )  To .
3-35
the heat flux at the boundary is variable, qo  qo (x ). This will only affect the integral in (v).
(ii) Taking advantage of symmetry simplifies the problem. (iii) Equation (r) gives the
dimensionless roots n L . However, the argument k H is determined by multiplying and
dividing by L according to k H = (k L)( H / L) . This introduces an additional dimensionless
parameters H / L . Similarly, the argument k b introduces the parameter b / L. Thus, three
parameters characterize the problem: Bi, b / L, and H / L.
3-36
PROBLEM 3.8
The cross section of a long prism is a right angle isosceles
triangle of side L. One side is heated with uniform flux q o
while the second side is maintained at zero temperature. The qo L Tc
hypotenuse is insulated. Determine the temperature Tc at the
mid-point of the hypotenuse.
(1) Observations. (i) This is a steady state two-dimensional T 0

conduction problem. (ii) The geometry does not lend itself to y
qo
analytic solution by the method of separation of variables.
However, due to symmetry the temperature distribution in the
right-angled isosceles triangle is the same as that in a square
with symmetrical boundary conditions about the diagonal as Tc T 0
L
shown. (iii) To determine the temperature at the mid-point of qo 
the hypotenuse (center of square) it is necessary to determine
the temperature distribution in the square. (iv) There are two x L
0
non-homogeneous boundary conditions. Since no direction has
two homogeneous conditions a modified procedure for T 0
applying the separation of variables method is needed.
(2) Origin and Coordinates. The origin and coordinate axes for the square geometry are
selected as shown.
(3) Formulation.
 2T  2T
 0 (a)
 x2  y2
(iii) Independent Variable with Two Homogeneous Boundary Conditions. No
variable has two homogeneous conditions. Thus the method of separation of variables can
not be applied directly to obtain a solution. A modified procedure will be used in which a
two-dimensional problem with two homogeneous conditions in the x-variable will represent
one part of the solution. A second part which is one-dimensional will satisfy the non-
homogeneous conditions in the x-variable.
(iv) Boundary Conditions. Consider the square formed by the triangle and its mirror
image. The four boundary conditions are
(1) T (0, y )  0 , homogeneous
T ( L, y )
(2) k  qo , non-homogeneous
x
(3) T ( x,0)  0 , homogeneous
T ( x, L)
(4) k  qo , non-homogeneous
y
3-37
(4) Solution. Since there are two non-homogenous boundary conditions we assume a
solution of the form
T ( x, y )   ( x, y )   ( x ) (b)
 2  2 d 2
  0 (c)
 x2  y2  x2
Let
 2  2
 0 (d)
 x2  y2
Thus
d 2
=0 (e)
d x2
Boundary conditions on (d) and (e) are obtained by substituting (b) into the four conditions.
Substituting (b) in condition (1)
 (0, y )   (0)  0
Let
 (0, y )  0 (d-1)
Thus
 ( 0)  0 (e-1)
Condition (2) gives
 ( L, y ) d ( L)
  qo
x dx
Let
 ( L, y )
0 (d-2)
x
Thus
d ( L )
 qo (e-2)
dx
Condition (3) gives
 ( x,0)   ( x )  0
Or
 ( x,0)   ( x ) (d-3)
Boundary condition (4) gives
 ( x, L)
k  qo (d-4)
y
Thus partial differential equation (c) has two homogeneous boundary conditions in the x-
variable, (d-1) and (d-2). The solution to (e) with boundary conditions (e-1) and (e-2) is
q
 ( x)  o x (f)
k
We proceed now to determine the solution to (d) using separation of variables.
3-38

 ( x, y ) = X(x) Y(y) (g)
Substituting into (d), separating variables and setting the resulting equation equal to constant
1 d2X 1 d 2Y
2
  2
  2n (h)
X dx Y dy
d2Xn
2
 2n X n  0 (i)
dx
and
d 2Yn
2
 2nYn  0 (j)
dy
boundary conditions the 2n X n term in (i) takes the positive sign. Therefore, (i) and (j)
become
d2Xn
2
 2n X n  0 (k)
dx
and
d 2Yn
 2nYn  0 (l)
d y2
For the important case of n  0, equations (k) and (l) become
d2X0
0 (m)
d x2
and
d 2Y0
0 (n)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to eqs. (k)-(n) are
X n ( x )  An sin n x  Bn cosn x (o)
Yn ( y )  Cn sin n y  Dn cosn y (p)
X 0 ( x )  A0 x  B0 (q)
Y0 ( y )  C0 y  D0 (r)
Corresponding to each value of n there is a temperature solution  ( x, y ). Thus

 ( x, y )  X n ( x )Yn ( y ) (s)
3-39
and
 0 ( x, y )  X 0 ( x )Y0 ( y ) (t)

 ( x, y )  X 0 ( x )Y0 ( y )   X n ( x )Yn ( y ) (u)
n 1
(iv) Application of Boundary Conditions. Boundary condition (d-1) applied to

solutions (o) and (q) gives
Bn  Bo  0 (v)
Boundary condition (d-2) applied to (o) gives the characteristic equation for n
cosn L  0 (w)
Or
( 2n  1)
n L  (n = 1,2,3….) (x)
2
Similarly, boundary condition (d-2) applied to (q) gives
A0  0
With A0  B0  0, the solution corresponding to n  0 vanishes. Thus (u) becomes

 ( x, y )   ( an sinh n y  bn coshn y ) sinn x (y)
n 1
where an  AnCn and bn  An Dn . Condition (d-3) applied to (y) and using (f) gives

 qox  k  b sin  x
n n (z-1)
n 1
Similarly, condition (d-4) and (y) give

qo  k  (a cosh L  b sinh  L) sin  x
n1
n n n n n n (z-2)
(v) Orthogonality. To determine a n and bn in equations (z-1) and (z-2) we apply

orthogonality to each equation. Note that the characteristic functions n ( x )  sin n x in (z)
are solutions to equation (k). Comparing (k) with eq. (3.5a) shows that it is a Sturm-Liouville
equation with
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
p ( x )  w( x )  1 and q( x )  0
function w( x )  1. Multiplying both sides of (z-1) by w( x ) sin m x dx and integrating from
x = 0 to x = L gives
3-40
L  
 
L

 qo x w( x ) sin m x dx = k
0 0
 bn sin n x w( x ) sin m x dx
 n 1 
Interchanging the summation with integration and noting that w(x) = 1, orthogonality gives
L L

 qo x sin n x dx = k bn
0
0
sin 2  n x dx
Evaluation the integrals, using equation (x) and solving for bn gives
8( qoL / k ) ( 1) n
bn  (z-3)
2 ( 2n  1) 2
Similarly, multiplying both sides of (z-2) by w( x ) sin m x dx and integrating from x = 0 to
x = L gives
L  

n  (an coshn L  bn sinh n L) sin n x w( x ) sin m x dx
L

qo w( x ) sin m x dx = k
0
0
 n 1 
Interchanging the summation with integration and noting that w(x) = 1, orthogonality gives
L L

qo sin n x dx = kn ( an coshn L  bn sinh n L)
0
0 sin2 n x dx
Evaluation the integrals using equation (x) and (z-3) and solving for a n gives
8( qoL / k ) 1  ( 1) n sinh n L

an  (z-4)
2 ( 2n  1) 2 coshn L
The solution to the temperature distribution is obtained by substituting (f), (y), (z-3) and (z-
4) into (b) and rearranging the result

1  ( 1) n sinh n L 

T ( x, y ) x 8 1
   sinh n y  ( 1) n coshn y  sin n x (z-5)
( qoL / k ) L  2 n 1 ( 2n  1)
2
 coshn L 
Evaluating (z-5) at the center, x = L/2 and y = L/2, gives Tc

1  ( 1) n sinh n L 

Tc 1 8 1
  2 2
sinh n L / 2  ( 1) n coshn L / 2 sin n L / 2
( qoL / k ) 2  n 1 ( 2n  1) 
coshn L 
(z-6)
dimensionless. (ii) Each term in (z-5) and (z-6) is dimensionless.
Limiting check: If qo  0 , the entire plate should be at zero temperature. Setting qo  0 in
(z-5) gives T ( x, y )  0.
3-41
the heat flux at the side of the right-angled isosceles triangle is non-uniform as long as the
corresponding square problem is assigned the same non-uniform flux on two of its adjacent
sides such that there is symmetry with respect to the diagonal. (ii) The method used to solve
this problem is limited to a right-angled isosceles triangle which forms a square with its
mirror image. It does not work for a right-angled triangle which forms a rectangle with its
mirror image because of asymmetry. (iii) An alternate approach to solving the square
problem is to use the method of superposition by decomposing the problem into two simpler
problems each having a single non-homogeneous condition as illustrated be below. Each
problem is solved using separation of variables directly.
y y y
qo qo
L T 0 L T 0 L T 0
Tc Tc1 qo Tc 2
qo
T ( x, y )
L
= L + T2 ( x, y )
L
x T1 ( x, y )
0 x 0 x 0
T 0 T 0 T 0
3-42
3-43
PROBLEM 3.9
The cross section of a long prism is a right angle isosceles

triangle of side L. One side exchanges heat by convection with h
the surroundings while the second side is maintained at zero T L Tc
temperature. The heat transfer coefficient is h and ambient
L
temperature is T . The hypotenuse is insulated. Determine the
temperature Tc at the mid-point of the hypotenuse. T 0
(1) Observations. (i) This is a steady state two-dimensional y

conduction problem. (ii) The geometry does not lend itself to
h, T
analytic solution by the method of separation of variables.
However, due to symmetry the temperature distribution in the
Tc T 0
right-angled isosceles triangle is the same as that in a square with h
L
boundary conditions symmetrical about the diagonal as shown. T
(iii) To determine the temperature at the mid-point of the
L
hypotenuse (center of square) it is necessary to determine the x 0
temperature distribution in the square. (vi) There are two non- T 0
homogeneous boundary conditions. Since no direction has two
homogeneous conditions the method of superposition should be used.
(2) Origin and Coordinates. The origin and coordinate axes for the square geometry are
selected as shown.
(3) Formulation.
(i) Assumptions. (1) two-dimensional, (2) steady, (3) no energy generation, (4) constant
conductivity and (5) uniform heat transfer coefficient h and ambient temperature T .
 2T  2T
 0 (a)
 x2  y2
(iii) Independent Variable with Two Homogeneous Boundary Conditions. No
variable has two homogeneous conditions. Thus the method of separation of variables can
not be applied directly to obtain a solution. Since there are two non-homogenous boundary
conditions, the problem will be decomposed into two simpler problems each having a single
non-homogeneous condition.
(iv) Boundary Conditions. Consider the square formed by the triangle and its mirror
image. The four boundary conditions are
(1) T (0, y )  0 , homogeneous
T ( L, y )
(2)  k  hT ( L, y )  T  , non-homogeneous
x
(3) T ( x,0)  0 , homogeneous
T ( x, L)
(4)  k  hT ( x, L)  T  , non-homogeneous
y
3-43
(4) Solution. Since there are two non-homogenous boundary conditions, the problem will be
decomposed into two simpler problems each having a single non-homogeneous condition.
Let
T ( x, y )  T1 ( x, y )  T2 ( x, y ) (b)
where T1 ( x, y ) and T2 ( x, y ) are the solutions to two problems each governed by (a) with
boundary conditions as shown.
y y y
h, T h, T h,0
L T 0 L T 0 L T 0
h Tc h Tc1 Tc 2
h
T
T ( x, y )
L
= 0 L + T
T2 ( x , y )
L
x T1 ( x , y )
0 x 0 x 0
T 0 T 0 T 0
 2T1  2T1  2T2  2T2
   0 (c)
 x2  y2  x2  y2
Let
 2T1  2T1
 0 (d)
 x2  y2
Thus
 2T2  2T2
 0 (e)
 x2  y2
The temperature at the center is given by
Tc  Tc1  Tc 2 (f)
We recognize that the center temperature is the same for each problem. Therefore
Tc  2 Tc1 (g)
Thus it is only necessary to solve the temperature distribution in one problem. Working with
the first problem, T1 ( x, y ), the boundary conditions are
(1) T1 (0, y )  0 , homogeneous
T1 ( L, y )
(2)  k  hT1 ( L, y ) , homogeneous
x
(3) T1 ( x,0)  0 , homogeneous
T1 ( x, L)
(4)  k  hT1 ( x, L)  T  , non-homogeneous
y
Thus the x-variable has two homogeneous boundary conditions.
3-44

T1 ( x, y ) = X(x) Y(y) (h)
Substituting into (d), separating variables and setting the resulting equation equal to constant
1 d2X 1 d 2Y
    2n
X d x2 Y d y2
d2Xn
 2n X n  0 (i)
d x2
and
d 2Yn
2
 2nYn  0 (j)
dy
boundary conditions the 2n X n term in (i) takes the positive sign. Therefore, (d) and (e)
become
d2Xn
2
 2n X n  0 (k)
dx
and
d 2Yn
2
 2nYn  0 (l)
dy
For the important case of n  0, equations (k) and (l) become
d2X0
0 (m)
d x2
and
d 2Y0
0 (n)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to eqs. (k)-(n) are
X n ( x )  An sin n x  Bn cos n x (o)
Yn ( y )  Cn sinh n y  Dn coshn y (p)
X 0 ( x )  A0 x  B0 (q)
Y0 ( y )  C0 y  D0 (r)
Corresponding to each value of n there is a temperature solution T1n ( x, y ). Thus
3-45
T1n ( x, y )  X n ( x )Yn ( y ) (s)
and
T10 ( x, y )  X 0 ( x )Y0 ( y ) (t)

T1 ( x, y )  X 0 ( x )Y0 ( y )  X n ( x )Yn ( y ) (u)
n 1
(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions (o)
and (q) gives
Bn  B0  0 (v)
Boundary condition (2) applied to (o) gives the characteristic equation for n
 n L  (hL / k ) tan n L (w)
Similarly, boundary condition (2) applied to (q) gives
A0  0

condition (3) to (p) gives
Dn  0
Solution (u) becomes

T1 ( x, y )  a n (sinh n y )(sin n x ) (x)
n 1
condition (4) gives
 
k  (a  n n cos hn L) sin n x  h  (a n sinh n L) sin n x  hT
n 1 n 1
Rearranging the above


hT   a h sinh  L  k cosh Lsin  x
n1
n n n n n (y)
(v) Orthogonality. To determine a n in equation (y) we apply orthogonality. Note that the
characteristic functions  n ( x )  sin n x in (y) are solutions to equation (k). Comparing (k)
a1 ( x)  a2 ( x)  0 and a3 ( x )  1
p ( x )  w( x )  1 and q( x )  0
3-46
function w( x )  1. Multiplying both sides of (y) by w( x ) sin m x dx and integrating from x
= 0 to x = L
L 
 a (h sinh  L  k coshn L) sin n x w( x ) sin m xdx

L
 0
hT w( x ) sin m xdx 
0
n 1
n n n
Interchanging the summation with integration and invoking orthogonality give

L L
hT 
0
w( x ) sin n xdx  a n ( h sinh n L  kn coshn L) 
0
w( x ) sin2 n xdx
Solving for a n , recalling that w(x) = 1 and evaluating the integrals in the above gives a n
1  cosn L
a n  2T (z)
sinh n L  (k / hL)(n L) coshn L)n L  (cosn L) sin n L
Substituting into (x)

(1  cos n L)(sin n x ) sinh n y

T1 ( x, y )
2 (z-1)
T n 1
Evaluating (s-1) at the center, x = L/2 and y = L/2 and substituting into (g) gives Tc

(1  cosn L)(sin n L / 2) sinh n L / 2

Tc
4 (z-2)
T n 1
dimensionless. (ii) Each term in solution (z-1) is dimensionless.
Limiting check: If T  0 , the entire plate should be at zero temperature. Setting T  0 in
(z-1) gives T1 ( x, y)  0.
Boundary conditions check: Conditions (1) and (3) can be readily shown to be satisfied by
direct substitution in equation (z-1).
(6) Comments. (i) The solution is expressed in terms of the Biot number hL/k. This is
characteristic of all problems involving surface convection, (ii) Equation (z-2) gives the
temperature at the center of a square with convection along two adjacent sides while the
other two sides are at zero temperature. If three sides exchange heat by convection, equation
(z-2) still applies. However, the constant 4 is replaced by 6. This results from decomposing
the problem into three simpler problems whose solutions are given by (z-1). (iii) The method
used to solve this problem is limited to a right-angled triangle which forms a square with its
mirror image. It does not work with a triangle which forms a rectangle with its mirror image
because of asymmetry.
3-47
PROBLEM 3.10
By neglecting lateral temperature variation in the y h, T 2 HBC

analysis of fins, two-dimensional conduction is
H 
modeled as a one-dimensional problem. To examine 0
this approximation, consider a semi-infinite plate of x
H
thickness 2H. The base is maintained at uniform
To h, T
temperature To . The plate exchanges heat by
convection at its semi-infinite surfaces. The heat
transfer coefficient is h and the ambient temperature is T . Determine the heat transfer
rate at the base.
(1) Observations. (i) This is a steady state two-dimensional conduction problem. (ii) At
x   the plate reaches ambient temperature. (iii) All four boundary conditions are non-
homogenous. However, by defining a temperature variable   T  T , three of the four
conditions become homogeneous. (iv) Temperature distribution is symmetrical about y  0
and thus the boundary (x,0) is insulated. Therefore, consideration is given to half the plate.
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) uniform heat transfer coefficient
and ambient temperature, (4) constant conductivity and (5) no energy generation.
 2  2
 0 (a)
 x2  y2
where  is defined as
 ( x, y )  T ( x, y )  T
(iii) Independent Variable with Two Homogeneous Boundary Conditions. In terms
of  , the y-variable has two homogeneous conditions.
(iv) Boundary Conditions. We consider the upper half of the plate. The boundary
conditions for this half are
 ( x,0)
y
 ( x, H )
(2)  k  h ( x, H ), homogeneous
y
(3)  ( , y ) = 0, homogeneous
(4)  (0, y )  To  T , non-homogeneous
(4) Solution.
(i) Assumed Product Solution. Let
(ii)
3-48
 ( x, y )  X ( x )Y ( y )
1 d2X 1 d 2Y
2
  2
  2n
X dx Y dy

d2Xn
2
 2n X n  0 (b)
dx
and
d 2Yn
 2nYn  0 (c)
d y2
conditions, the 2n Yn term in (c) takes the positive sign. Thus
d2Xn
2
 2n X n  0 (d)
dx
and
d 2Yn
2
 2nYn  0 (e)
dy
For the special case of n  0, equations (d) and (e) become
d2X0
0 (f)
d x2
and
d 2 Y0
0 (g)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to eqs. (d)-(g) are
X n ( x )  An exp( n x )  Bn exp( n x ) (h)

Yn ( y )  Cn sin n y  Dn cosn y (i)
X 0 ( x )  A0 x  B0 (j)
Y0 ( y )  C0 y  D0 (k)

 n ( x, y )  X n ( x )Yn ( y ) (l)
and
3-49
 0 ( x, y )  X 0 ( x )Y0 ( y ) (m)

 ( x, y )  X 0 ( x )Y0 ( y )  
n 1
X n ( x )Yn ( y ) (n)
(iv) Application of Boundary Conditions. Boundary condition (1) applied to (i) and (k)
gives
C n  C0  0 (o)
Applying boundary condition (2) to (i) gives the characteristic equation for n
n H tan n L  hH / k  Bi (p)
where Bi  hH / k is the Biot number. Boundary conditions and (2) applied to (k) gives
D0  0
Thus, the solution corresponding to n  0 vanishes. Boundary condition (3) gives

Bn  0
The temperature solution (n) becomes


 ( x, y )  a
n 1
n exp( n x ) cosn y (q)
where a n  An Dn . Finally, we apply boundary condition (4) to (q)


To  T  a
n 1
n cosn y (r)
(v) Orthogonality. To determine a n in equation (r) we apply orthogonality. Note that

the characteristic functions  n ( y )  cosn y in (r) are solutions to equation (e). Comparing (e)
with equation (3.5a) shows that it is a Sturm-Liouville equation with
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p ( y )  w( y )  1 and q( y )  0
characteristic functions  n  cosn y are orthogonal with respect to w(y) = 1. Multiplying
both sides of (r) by w(y) cosm y dy and integrating from y = 0 to y = H, we obtain
H  
0 
H
(To  T ) 0 w( y ) cosm ydy   a n cosn y  w( y ) cosm ydy
 n1 
3-50
Interchanging the summation with integration, setting w(y) = 1 and applying orthogonality,
the above gives
H H
(To  T )

0
cos n ydy  a n

0
cos2 n ydy
Evaluating the integrals and solving for a n

2(To  T ) sin n H
an  (s)
n H  (sin n H ) cos n H
Substituting (s) into (r)

T ( x, y )  T sin n H
To  T
2 
n 1
n H  (sin n H ) cosn H
e n x cosn y (t)
(5) Checking. Dimensional check: (ii) The arguments of sin, cos and the exponential are
dimensionless. (ii) Each term in solution (t) is dimensionless.
Limiting check: If To  T , no heat transfer takes place within the plate and consequently the
entire plate should be at T . Setting To  T in (t) gives T ( x, y )  T .
Boundary conditions check: Conditions (1) and (3) can be readily shown to be satisfied by
direct substitution into solution (q).
(6) Comments. (i) By introducing the definition   T  T it was possible to transform
three of the four boundary conditions from non-homogeneous to homogeneous. (ii) The
solution is in terms of the Biot number hH/k which appears in the characteristic equation (p).
This parameter appears in problems with surface convection. (iii) In the fin model,
temperature variation in the y-direction is neglected and a one-dimensional solution is
obtained. This approximation is valid for small Biot number compared to unity. That is for
Bi  hH / k  1 (u)
Thus the two-dimensional solution (t) should reduce to the one-dimensional fin solution for
small Biot number. Numerical solution to the characteristic equation (p) shows that for
Bi  1 the first root is 1 H  1 . Thus for Bi  1
1 H  1 , sin 1L  1L, cos1H  cos1 y  1 (v)
When this is substituted into (p) we obtain
1 H  hH / k (w)
Substituting (v) and (w) into (t) and taking the first term of the series gives
T1 ( x)  T  (To  T ) e  hk / H x
(x)
This result is identical to the one-dimensional temperature solution for a semi-infinite fin
with surface convection and a specified temperature at the base.
3-51
PROBLEM 3.11
A very long plate of thickness 2H leaves a furnace at 2 HBC

y To
temperature T f . The plate is cooled as it moves with
velocity U through a liquid tank. Because of the high H U 
heat transfer coefficient, the surface of the plate is 0
x
maintained at temperature To . Determine the steady state H
two-dimensional temperature distribution in the plate. Tf
To
(1) Observations. (i) This is a two-dimensional steady
state conduction problem. (ii) Temperature distribution is symmetrical about the x-axis. (iii)
At x   the plate reaches a uniform temperature To . (iv) All four boundary conditions are
non-homogenous. However, by defining a new variable,   T  To , three of the four
boundary conditions become homogeneous. (v) Temperature distribution is symmetrical
about the x-axis. Thus no heat crosses the boundary (x,0).
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) constant properties, (4) uniform
velocity and (5) no energy generation.
(ii) Governing Equations. Introducing the definition  ( x, y )  T ( x, y )  To , eq. (1.7)
gives
 2  2 
  2 0 (a)
x 2
y 2
x
where  is defined as
 c pU
 (b)
2k
(iii) Independent Variable with Two Homogeneous Boundary Conditions. The y-
direction has two homogeneous conditions in the  variable.
(iv) Boundary Conditions. Because of symmetry we consider half the plate. The boundary
conditions for the upper half are
 ( x,0)
y
(2)  ( x, H )  0 , homogeneous
(3)  ( , y ) = 0, homogeneous
(4)  (0, y)  T f  To , non-homogeneous
(4) Solution.
(i) Assumed Product Solution.
3-52
Assume a product solution   X ( x)Y ( y ). Substituting into (a), separating variables and
setting the resulting equation equal to constant gives
d2Xn d Xn
2
 2  2n X n  0 (c)
dx dx
and
d 2Yn
2
 2nYn  0 (d)
dy
conditions, the correct signs in (c) and (d) are
d2Xn d Xn
2
 2  2n X n  0 (e)
dx d x
and
d 2Yn
2
 2nYn  0 (f)
dy
For n  0, equations (e) and (f) become
d 2X0 dX 0
2
 2 0 (g)
dx dx
and
d 2 Y0
0 (h)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to equations (e)-
(h) are
X n ( x )  [ An exp(  x   2  2n x )  Bn exp(  x   2  2n x )] (i)
Yn ( y )  Cn sin n y  Dn cosn y (j)

X 0 ( x )  A0 exp( 2  x )  B0 (k)
and
Y0 ( y )  C0 y  D0 (l)

 ( x, y )  X 0 ( x )Y0 ( y )   X n ( x )Yn ( y ) (m)
n 1
(iv) Application of Boundary Conditions. Condition (1) applied to (j) and (l) gives
C n  C0  0
Boundary condition (2) applied to (j) gives the characteristic equation
3-53
cosn H  0
This gives
2n  1
n H   (n = 1, 2, 3….) (n)
2
Boundary condition (2) applied to (l) gives
D0  0
Condition (3) gives
An  0
With C0  D0  0, it follows that X 0Y0  0. Thus equation (m) becomes
exp(  

 ( x, y )  a
n 1
n  2  2n ) x cos  n y (o)
where a n  Bn Dn . Finally, we apply boundary condition (4) to determine a n


T f  To  a
n 1
n cos  n y (p)
(v) Orthogonality. Note that the characteristic functions  n  cos  n y are solutions to
equation (f). Comparing (f) with equation (3.5a) shows that it is a Sturm-Liouville equation
with
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p ( y )  w( y )  1 and q( y )  0
characteristic functions  n  cos  n y are orthogonal with respect to w(y) = 1. Multiplying
both sides of (p) by w( y ) cos  m ydy and integrating from y = 0 to y = H gives
H H  
(T f  To )

0
w( y ) cos m ydy 
 
0
 a n cos n y  w( y ) cos m ydy
 n1 
Interchanging summation with integration, noting that w(y) = 1 and invoking orthogonality,
the above becomes
H H
(T f  To )
0
cos n ydy  an
0
cos2 n ydy
Evaluating the integrals and solving for a n

sin  n H
a n  2(T f  T ) (q)
n H
Substituting into (o)
3-54
 

T ( x, y )  To sin  n H
T f  To
2 
n 1
n H
exp(    2   2n ) x cos  n y (r)
(5) Checking. Dimensional check: (i) The arguments of sin and cos are dimensionless. (ii)
Each term in solution (r) is dimensionless. (iii) Solution (r) requires that units of  be the
same as that of n . According to (d),  n is measured in (1/m). From the definition of  in
(b) we have
 ( kg / m 3 )c p ( J/kg o C)U ( m/s)
  (1 / m)
k (W/m-o C)
Limiting check: If T f  To , no heat transfer takes place in the plate and consequently the
temperature remains uniform. Setting T f  To in (r) gives T ( x, y )  To .
Boundary conditions check: conditions (1), (2) and (3) are readily shown to be satisfied by
direct substitution in (o).
(6) Comments. (i) For a stationary plate ( U    0 ), the solution to this special case is
obtained by setting   0 in (r)

T ( x, y )  To sin  n H
T f  To
2 
n 1
n H
exp ( n x) cos  n y
(ii) The solution depends on two parameters. The first parameter is the Biot number hH / k
which appears in the characteristic equation for n . The second is  H   c pUH / k which
appears in the exponent of the exponential in equation (r) when the exponent is multiplied
and divided by H.
3-55
PROBLEM 3.12
Liquid metal flows through a cylindrical channel of width H
To Ti
and inner radius R. It enters at Ti and is cooled by convection 0
along the outer surface to To . The ambient temperature x
is T and heat transfer coefficient is h.. The inner surface is y H
insulated. Assume that the liquid metal flows with a uniform R
velocity U and neglect curvature effect (H/R << 1), determine
the steady state temperature of the insulated surface. U
(1) Observations. (i) This is a two-dimensional steady state
conduction problem. (ii) To determine the temperature of the
insulated surface it is necessary to determine the temperature
distribution in the moving liquid metal. (iii) The velocity is h, T
uniform and inlet and outlet temperatures are specified. (iv)
Three boundary conditions are non-homogenous. However, by defining a new variable,
  T  T , the y-direction will have two homogeneous conditions.
(3) Formulation. y
h, T
(i) Assumptions. (1) Steady T U H
i x To
state, (2) Two-dimensional, (3) 0
constant properties, (4) uniform
L  2 R
velocity, (5) negligible curvature
effect (H/R<<1) and (6) no energy generation.
(ii) Governing Equations. Introducing the definition  ( x, y )  T ( x, y )  T , eq. (1.7)
gives
 2  2 
  2 0 (a)
x 2
y 2
x
where  is defined as
 c pU
 (b)
2k
(iii) Independent Variable with Two Homogeneous Boundary Conditions. The y-
direction has two homogeneous conditions in the  variable.
(iv) Boundary Conditions. The boundary conditions for the upper half are
 ( x,0)
y
 ( x, H )
(2)  k  h ( x, H ) , homogeneous
y
(3)  (0, y )  Ti  T , non-homogeneous
(4)  ( L, y )  To  T , non-homogeneous
where L is the length of channel given by
3-56
L  2R (c)
(4) Solution.
(i) Assumed Product Solution. Assume a product solution   X ( x )Y ( y ). Substituting
into (a), separating variables and setting the resulting equation equal to constant gives
d2Xn d Xn
2
 2  2n X n  0 (d)
dx dx
and
d 2Yn
 2nYn  0 (e)
d y2
conditions, the correct signs in (d) and (e) are
d2Xn d Xn
2
 2  2n X n  0 (f)
dx dx
and
d 2Yn
2
 2nYn  0 (g)
dy
For n  0, equations (f) and (g) become
d 2X0 dX 0
2
 2 0 (h)
dx dx
and
d 2 Y0
0 (i)
d y2
(iii) Solutions to the Ordinary Differential Equations. The solutions to equations (f)-
(i) are
X n ( x)  An exp(  x   2  2n x)  Bn exp(  x   2  2n x) (j)

X 0 ( x )  A0 exp( 2  x )  B0 (l)
and
Y0 ( y )  C0 y  D0 (m)
The complete solution is

 ( x, y )  X 0 ( x )Y0 ( y )   X n ( x )Yn ( y ) (n)
n 1
(iv) Application of Boundary Conditions. Condition (1) applied to (k) and (m) gives
3-57
C n  C0  0
Boundary condition (2) applied to (k) gives
n H tan n H  hH / k  Bi (o)
where Bi is the Biot number. This is the characteristic equation whose roots give n .
Boundary condition (2) applied to (m) gives
D0  0
With C0  D0  0, it follows that X 0Y0  0. Thus equation (n) becomes


 (  (    2  2n ) x 
 ( x, y )    an e
n 1 
 2  2n ) x
 bn e  cosn y (p)
where an  An Dn and bn  Bn Dn . Application of condition (3) and (4) to (p) gives


Ti  T   (a
n 1
n  bn ) cosn y (q)
and

To  T   a e
n 1
n
(    2  2n ) L
 bn e
(    2  2n ) L 
 cosn y (r)
(v) Orthogonality. Note that the characteristic functions  n  cosn y in (q) and (r) are
solutions to equation (g). Comparing (g) with equation (3.5a) shows that it is a Sturm-
Liouville equation with
a1 ( y )  a2 ( y )  0 and a 3 ( y )  1
p( y )  w( y )  1 and q( y )  0
characteristic functions n  sin n y are orthogonal with respect to w(y) = 1. Multiplying
both sides of (q) by w( y ) sin m ydy and integrating from y = 0 to y = H gives
H  
0 
H
(Ti  T ) 0 w( y ) cosm ydy   (an  bn ) cosn y  w( y ) cosm ydy
 n 1 
the above becomes
H H
(Ti  T ) 0
cosn ydy  ( an  bn )  0
cos2 n ydy
Evaluating the integral and solving for (an  bn )
3-58
2(Ti  T ) sin n H
( an  bn )  (s)
n H  (sin n H ) cosn H
Similarly, multiplying both sides of (r) by w( y ) sin m ydy and integrating from y = 0 to y =
H gives
H 
 (  (    2  2n ) L 
 
H

 2  2n ) L
(To  T ) w( y ) cosm ydy  a n e  bn e w( y ) cos m ydy (t)
2
n 1  
0 0
the above becomes
H
 (  (    2  2n ) L 
H
 
 2  2n ) L
(To  T ) cos n ydy  an e  bn e  cos2 n ydy
0   0
Evaluating the integrals and rearranging

(    2   2n ) L (    2   2n ) L 2(To  T ) sin n H
an e  bn e  (u)
n H  (sin n H ) cos n H
Equation (s) and (u) are solved for the coefficients a n and bn
 (To  T ) /(Ti  T )  e (    n ) L

2 2
an  2 sin n H
  (v)
(Ti  T )  n H  (sin n H ) cosn H  (    2  2n ) L (    2  2n ) L
e e
and
 
 (To  T ) /(Ti  T )  e (    n ) L 
2 2
bn  2 sin n H
 1   (w)
Ti  T  n H  (sin n H ) cosn H  (    2  2n ) L (    2  2n ) L

 e  e 
Rewriting solution (r) in dimensionless form gives

T ( x, y )  T  an  2  2n ) x 
  T  T (    2  2n ) x bn ( 
Ti  T
 e 
Ti  T
e  cosn y (x)
n 1 i  
where a n /(Ti  T ) and bn /(Ti  T ) are given in (v) and (w). The temperature of the
insulated surface is obtained by evaluating (x) at y = 0

T ( x,0)  T   2  2n ) x 
  T  T
an (    2  2n ) x bn ( 
 e  e  (y)
Ti  T n 1 i  Ti  T 
(5) Checking. Dimensional check: (i) The arguments of the sin, cos and exponentials are
dimensionless. (ii) Solution (x) requires that units of  be the same as that of n .
According to (g),  n is measured in (1/m). From the definition of  in (b) we have
 ( kg / m 3 )c p ( J/kg o C)U ( m/s)
  (1 / m)
k (W/m-o C)
3-59
Limiting check: If Ti  To  T , no heat transfer takes place and consequently the temperature
in the entire region remains uniform. Setting Ti  To  T in (v) and (w) gives an  bn  0.
When this is substituted into (p) gives T ( x, y )  T .
Boundary conditions check: Solution (p) can be readily shown to satisfy condition (1).
(6) Comments. (i) For a stationary material ( U    0 ), the solution to this special case is
obtained by setting   0 in (p)

 ( x, y )  a e
n 1
n
n x

 bn e  n x cosn y
where a n and bn are obtained by setting   0 in (v) and (w).

(ii) Examination of equations (o), (v), (w) and (x) shows that the solution depends on four
parameters. The first parameter is the Biot number hH / k which appears in the characteristic
equation for n . The second is (To  T ) /(Ti  T ) shown in (v) and (w). The third parameter
is  H   c pUH / k and the fourth is L/H which appears in n L  n H ( L / H ) .
3-60
PROBLEM 3.13
A solid cylinder of radius ro and length L is
maintained at T1 at one end. The other end is r T
3
maintained at T2 and the cylindrical surface at T3 . r
o
Determine the steady state two-dimensional 0 z
temperature distribution in the cylinder.
T L 2 T
(1) Observations. (i) This is a steady state two- 1
dimensional conduction problem. (ii) There are three non-homogeneous boundary
conditions. Defining a new temperature variable,   T  T3 , makes the boundary condition
at the cylindrical surface homogeneous. This gives the radial direction two homogeneous
conditions. (iii) A cylindrical coordinate system should be used.
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady, (3) constant conductivity and (4) no
energy generation.
(ii) Governing Equations. Defining   T  T3 and applying the above assumptions to
eq. (1.11) gives
 2 1    2
  0 (a)
 r2 r  r  z2
(iii) Independent Variable with Two Homogeneous Boundary Conditions. The r-

variable has two homogeneous conditions.
(iv) Boundary Conditions. The boundary conditions on  ( r , z ) are
  (0, z )
(1)  0, or  (0, z )  finite, homogeneous
r
(2)  ( ro , z ) = 0, homogeneous
(3)  (r ,0)  (T1  T3 ), non-homogeneous
(4)  (r , L)  (T2  T3 ), non-homogeneous

(4) Solution.
(i) Assumed Product Solution. To solve equation (a) we assume a product solution. Let
 ( r, z )  R( r ) Z ( z ) (b)
Substituting (b) into (a), separating variables and setting the resulting two equations equal to
a constant,  2k , we obtain
d 2Zk
2
 2k Z k  0 (c)
dz
and
3-61
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (d)
dr dr
(ii) Selecting the Sign of the 2k Terms. Since the r-variable has two homogeneous
boundary conditions, the plus sign is selected in (d). Thus (c) and (d) become
d 2Zk
2
 2k Z k  0 (e)
dz
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (f)
dr dr
For the special case of k  0 the above equations take the form
d 2Z0
0 (g)
d z2
and
d 2 R0 dR0
r  0 (h)
dr2 dr
(iii) Solutions to the Ordinary Differential Equations. The solution to (e) is

Z k ( z )  Ak sinh k z  Bk coshk z (i)
Equation (f) is a Bessel differential equation. Comparing (f) with eq. (2.26) shows that
A  B  n  0 , C = 1 and D  k . Thus the solution to (f) is given by eq. (2.27)
Rk (r )  C k J 0 ( k r )  Dk Y0 ( k r ) (j)
The solutions to (g) and (h) are
Z 0 ( z )  A0 z  B0 (k)
and
R0 ( r )  C0 ln r  D0 (l)
The complete solution to  ( r , z ) becomes

 ( r, z )  R0 ( r ) Z 0 ( z )   Rk ( r ) Z k ( z ) (m)
k 1
(iv) Application of Boundary Conditions. Applying boundary condition (1) to equation

(j) and (l) gives
Dk  C0  0
Condition (2) applied to (j) gives the characteristic equation for k

J 0 (k ro )  0 (n)
3-62
Condition (2) applied to (l) gives
D0  0
With C0  D0  0 the solution R0 Z 0 vanishes and the temperature solution (m) becomes

 ( r, z )   ak sinh k z  bk coshk z  J 0 ( k r ) (o)
k 1
where ak  Ak Ck and bk  Bk Ck . Application of boundary conditions (3) and (4) to (o) gives

T1  T3  
k 1
bk J 0 (k r ) (p)
and

T2  T3  a
k 1
k sinh k L  bk coshk L J 0 ( k r ) (q)
(v) Orthogonality. Note that the characteristic functions J 0 (k r ) in (p) and (q) are
solutions to equation (f). Comparing (f) with eq. (3.5a) shows that it is a Sturm-Liouville
equation with
a1 (r )  1/ r, a2 (r)  0 and a3 ( r )  1
p( r )  w( r )  r and q( r )  0
Since the boundary conditions at r = 0 and r  ro are homogeneous, it follows that the
characteristic functions J 0 (k r ) are orthogonal with respect to w(r) = r. Multiplying both
sides of (p) by J 0 ( i r ) w( r )dr and integrating from r = 0 to r  ro
ro  
ro 
(T1  T3 ) 
0
J 0 (i r ) w( r )dr   
0
 bk J 0 (k r )J 0 (i r ) w( r )dr
 k 1 
Interchanging summation with integration, noting that w(r) = r and invoking orthogonality,
eq. (3.7), the above gives
ro ro
(T1  T3 )  0
J 0 (k r )rdr  bk  0
J 02 (k r )rdr
Evaluating the integrals using Appendix B and Table 3.1 gives

2(T1  T3 )
bk  (r)
k ro J 1 ( k ro )
Similarly, multiplying both sides of (q) by J 0 (i r ) w( r )dr and integrating from r = 0 to
r  ro
ro  
ro 
(T2  T3 ) 
0
J 0 (i r ) w( r )dr   
0
 ak sinh k L  bk coshk LJ 0 (k r )J 0 (i r ) w( r )dr
 k 1 
3-63
ro ro
(T2  T3 ) 
0
J 0 (k r ) w( r )dr  (ak sinh k L  bk coshk L) 
0
J 02 (k r )rdr
Evaluating the integrals using Appendix B and Table 3.1

2(T1  T3 )
( ak sinh k L  bk cosh k L)  (s)
k ro J 1 ( k ro )
Solving for a k and using (r) to eliminate bk
ak 
2
(T2  T3 )  (T1  T3 ) coshk L (t)
(sinh k L)k ro J 1 ( k ro )
Substituting (r) and (t) into (o) gives the temperature solution in the cylinder

T (r , z )  T3 
 (T2  T3 )  sinh  k z 
 J ( r )

1
2   cosh  k L  cosh  k z  0 k (u)
(T1  T3 )  r
k 1 k o  (T1  T3 )
  sinh  k L  J 1 ( k ro )

(5) Checking. Dimensional checks: (i) The arguments of sinh, cosh, J 0 and J 1 are
dimensionless. (ii) Each term in (u) is dimensionless.
Limiting check: If T1  T2  T3 , no heat transfer can take place and the entire cylinder should
be at uniform temperature. Setting T1  T2  T3 in (u) gives T ( r, z )  T .
Boundary conditions check: Setting r = 0 and r  ro in solution (u) shows that boundary
conditions (1) and (2) are satisfied.
(6) Comments. (i) An alternate approach is to assume a solution of the form
T ( r, z )   ( r, z )   ( z )
The function  (z ) can be assigned two non-homogeneous boundary conditions in the z-
direction leaving  ( r, z ) with two homogeneous conditions in the z-direction and one in the
r-direction. Another approach is to apply the method of superposition by decomposing the
problem into two simpler problems each having one non-homogeneous boundary condition.
(ii) The solution depends on two dimensionless parameter: (T2  T3 ) /(T1  T3 ) and L / ro .
3-64
PROBLEM 3.14
A solid cylinder of radius ro and length L is r T2

2 HBC
maintained at T1 at one end and is cooled

with uniform flux q o at the other end. The 0
ro
z
cylindrical surface is maintained at a uniform qo
temperature T2 . Determine the steady state T1 L
two-dimensional temperature distribution in
the cylinder.
(1) Observations. (i) This is a steady state two-dimensional conduction problem. (ii) There
are three non-homogeneous boundary conditions. Defining a new temperature variable,
  T  T2 makes the boundary condition at the cylindrical surface homogeneous. This gives
the radial direction two homogeneous conditions. (iii) Use a cylindrical coordinate system.
(3) Formulation.
energy generation.
(ii) Governing Equations. Defining   T  T2 and applying the above assumptions to
eq. (1.11) gives
 2 1    2
  0 (a)
 r2 r  r  z2

  (0, z )
r
(3)  (r,0)  (T1  T2 ), non-homogeneous
 (r , L)
(4)  k  q o non-homogeneous
z
(4) Solution.
 ( r, z )  R( r ) Z ( z ) (b)
d 2Zk
2
 2k Z k  0 (c)
dz
3-65
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (d)
dr dr
d 2Zk
2
 2k Z k  0 (e)
dz
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (f)
dr dr
d 2Z0
0 (g)
d z2
and
d 2 R0 dR0
r  0 (h)
dr2 dr

Rk (r )  C k J 0 ( k r )  Dk Y0 ( k r ) (j)
Z 0 ( z )  A0 z  B0 (k)
and
R0 ( r )  C0 ln r  D0 (l)
The complete solution to  ( r , z ) becomes

 ( r, z )  R0 ( r ) Z 0 ( z )   Rk ( r ) Z k ( z ) (m)
k 1

(j) and (l) gives
Dk  C0  0

J 0 ( k ro )  0 (n)
3-66
D0  0
With C0  D0  0 the solution R0 Z 0 vanishes and the temperature solution (m) becomes

 ( r, z )   ak sinh k z  bk coshk z  J 0 ( k r ) (o)
k 1
where ak  Ak Ck and bk  Bk Ck . Application of boundary conditions (3) and (4) to (o) gives

T1  T2  b k J 0 ( k r ) (p)
k 1
and

qo  k   a
k 1
k k coshk L  bk sinh k L J 0 ( k r ) (q)
(v) Orthogonality. Note that the characteristic functions J 0 (k r ) in (p) and (q) are
solutions to equation (f). Comparing (f) with eq. (3.5a) shows that it is a Sturm-Liouville
equation with
a1 (r )  1/ r, a2 (r)  0 and a3 ( r )  1
p( r )  w( r )  r and q( r )  0
ro ro   
(T1  T2 ) 0
J 0 (i r ) w(r )dr   
0
 bk J 0 ( k r )J 0 (i r ) w(r )dr
 k 1 
ro ro
(T1  T2 )  0
J 0 ( k r )rdr  bk 0
J 02 ( k r )rdr
Evaluating the integrals using Appendix B and Table 3.1 the above gives bk
2(T1  T2 )
bk  (r)
k ro J 1 ( k ro )
Similarly, multiplying both sides of (q) by J 0 (i r ) w( r )dr and integrating from r = 0 to
r  ro
ro  
ro 
qo 
0
J 0 (i r ) w( r )dr  k    a
0
k 1
k k coshk L  bk sinh k LJ 0 (k r )J 0 (i r ) w( r )dr

3-67
ro ro

qo J 0 (k r )rdr  kk (ak cosh k L  bk sinh k L)
0
 0
J 02 (k r )rdr
Evaluating the integrals using Appendix B and Table 3.1 the above gives
( qoro / k )
(ak cosh k L  bk sinh k L)  2 (s)
(k ro ) 2 J 1 (k ro )
Solving for a k and using (r) to eliminate bk
2  ( qro / k ) 
ak    (T1  T2 ) sinh k L (t)
(cosh k L)( k ro ) J 1 ( k ro )  k ro 
substituting (s) and (t) into (o) and rearranging the result

T (r , z )  T2 2   (q ro / k )  sinh  k z 
 J ( r )
T2  T1
 
k 1

( k ro ) 
  k ro (T2  T1 )
 sinh  k L
 cosh  k L
 cosh  k z  0 k
 J 1 ( k ro )

(u)
(5) Checking. Dimensional check: (i) The arguments of sinh, cosh, J 0 and J 1 are
dimensionless. (ii) Each term in (u) is dimensionless.
Limiting check: If T1  T2 , and qo  0 no heat transfer can take place and the entire cylinder
should be at uniform temperature. Setting T1  T2 in (u) gives T ( r , z )  T2 .
Boundary conditions check: Setting r = 0 and r  ro in solution (u) shows that boundary
(6) Comments. (i) One of parameters which appears in the solution is qoro / k (T2  T1 ) . In
addition, roots of (n) give k ro . However, the argument k L  k ro ( L / ro ) introduces a third
parameter which is ( L / ro ) . (ii) An alternate approach is to assume a solution of the form
T ( r, z )   ( r, z )   ( z )
The function  (z ) can be assigned two non-homogeneous boundary conditions in the z-
direction leaving  ( r, z ) with two homogeneous conditions in the z-direction and one in the
r-direction. Another approach is to apply the method of superposition by decomposing the
problem into two simpler problems each having one non-homogeneous boundary condition.
3-68
PROBLEM 3.15
The volumetric heat generation rate in a solid

cylinder of radius ro and length L varies along r To 2 HBC
the radius according to ro
q   qo r z q
0 qo
where qo is constant. One end is insulated while
L
the other end is cooled with uniform flux qo.
The cylindrical surface is maintained at uniform temperature To . Determine the steady state
temperature of the insulated surface.
are two non-homogeneous boundary conditions. (iii) Energy generation makes the heat
equation non-homogeneous. (iv) A cylindrical coordinate system should be used.
(3) Formulation.
(i) Assumptions. (1) Two-dimensional, (2) steady and (3) constant conductivity.
(ii) Governing Equations. Applying the above assumptions to eq. (1.11) gives
 2T 1  T  2T q
   0
 r2 r  r  z2 k
Energy generation q is given by
q  qo r
where qo is constant. Substituting into the above heat equation gives
 2T 1  T  2T qo
   r0 (a)
 r2 r  r  z2 k

variable has two homogeneous conditions. However, the problem can be decomposed into
two problems one of which has two homogeneous conditions in the r-direction.
(iv) Boundary Conditions. The boundary conditions are
T (0, z )
r
(2) T ( ro , z )  To , non-homogeneous
T ( r,0)
z
T (r , L)
(5)  k  qo non-homogeneous
z
3-69
(4) Solution. Equation (a) is non-homogeneous due to the heat generation term. Thus we
can not proceed directly with the application of the separation of variables method. Instead,
we assume a solution of the form
T ( r, z )   ( r, z )   ( r ) (b)
Note that  ( r, z ) depends on two variables while  (r ) depends on a single variable r. the
function  (r ) is introduced to satisfy the non-homogeneous part of equation (a) which also
depends on the variable r. Substituting (b) into (a)
 2 1   2 d 2 1 d qo
     r0 (c)
 r2 r  r  z2 d r2 r d r k
The next step is to split (c) into two equations, one for  ( r, z ) and one for  (r ) . We let
 2 1   2
  0 (d)
 r2 r  r  z2
Thus
d 2 1 d qo
  r0 (e)
d r2 r d r k
Boundary conditions on (d) and (e) are obtained by substituting (b) into the four boundary
conditions on T. In this process the function  ( r, z ) is assigned homogeneous conditions
whenever possible. Using (b) and condition (1)
 (0, z ) d (0)
 0
r dr
Let
 (0, z )
0 (d-1)
r
Thus
d (0)
0 (e-1)
dr
 ( ro , z )   ( ro )  To
Let
 ( ro , z )  0 (d-2)
Thus
 ( ro )  To (e-2)
 ( r,0)
0 (d-3)
z
 (r , L)
k  qo (d-4)
z
3-70
Thus non-homogeneous equation (a) is replaced by two equations: (d) which is a

homogeneous partial differential equation and (e) which is a non-homogeneous ordinary
differential equation. Equation (d) has three homogeneous conditions, (d-1), (d-2) and (d-3)
and one non-homogeneous condition (d-4). The solution to (e) is
qo 3
 (r)   r  E ln r  F (f)
9k
where E and F are constants of integration. Boundary conditions (e-1) and (e-2) give E and
F. The above solution becomes
qo 3 3
 (r)  ( ro  r )  To (g)
9k
(i) Assumed Product Solution. To solve equation (d) we assume a product solution. Let
 ( r, z )  R( r ) Z ( z ) (h)
Substituting (h) into (d), separating variables and setting the resulting two equations equal to
d 2Zk
2
 2k Z k  0 (i)
dz
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (j)
dr dr
boundary conditions, the plus sign is selected in (j). Thus (i) and (j) become
d 2Zk
 2k Z k  0 (k)
d z2
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (l)
dr dr
d 2Z0
0 (m)
d z2
and
d 2 R0 dR0
r 2
 0 (n)
dr dr
(iii) Solutions to the Ordinary Differential Equations. The solution to (k) is
3-71
Z k ( z )  Ak sinh k z  Bk coshk z (o)
Equation (l) is a Bessel differential equation. Comparing (l) with eq. (2.26) shows that
A  B  n  0 , C = 1 and D  k . Thus the solution to (l) is given by eq. (2.27)
Rk (r )  Ck J 0 (k r )  Dk Y0 (k r ) (p)
The solutions to (m) and (n) are
Z 0 ( z )  A0 z  B0 (q)
and
R0 ( r )  C0 ln r  D0 (r)
The complete solution to  (r , z ) becomes

 (r , z )  R0 (r ) Z 0 ( z )   R (r ) Z ( z )
k 1
k k (s)
(iv) Application of Boundary Conditions. Applying boundary condition (d-1) to

equation (p) and (r) gives
Dk  C0  0
Condition (2) applied to (p) gives the characteristic equation for k

J 0 ( k ro )  0 (t)
Condition (2) applied to (r) gives
D0  0
With C0  D0  0 the solution R0 Z 0 vanishes. Boundary condition (d-3) gives

Ak  0
Substituting into (s) gives

 ( r, z )   (b cosh  z) J
k k 0 ( k r ) (u)
k 1
where bk  Bk Ck . Application of boundary conditions (4) to (u) gives


qo  k  b (sinh  L) J
k 1
k k 0 ( k r ) (v)
(v) Orthogonality. Note that the characteristic functions J 0 (k r ) in (v) are solutions to
equation (l). Comparing (l) with eq. (3.5a) shows that it is a Sturm-Liouville equation with
a1 (r )  1/ r, a2 (r)  0 and a3 ( r )  1
p( r )  w( r )  r and q( r )  0
3-72
sides of (v) by J 0 ( i r ) w( r )dr and integrating from r = 0 to r  ro
ro ro   
qo 
0
J 0 (i r ) w( r )dr  k  
0
 bk (sinh k L) J 0 (k r )J 0 (i r ) w( r )dr
 k 1 
ro ro
qo 
0
J 0 ( k r ) rdr  k bk (sinh k L) 0
J 02 ( k r ) rdr
 2( qoro / k )
bk  (w)
( k ro ) (sinh k L) J 1 (k ro )
2
Substituting (w) into (u) gives the solution  ( r, z )


 ( 1r )
coshk z J 0 (k r )
 (r , z )  2(qo ro / k ) (x)
2
sinh k L J1 (k ro )
k 1 k o
The complete solution to the temperature distribution in the cylinder is obtained by

substituting (g) and (x) into (b) and rearranging the result

T (r , z )  To 1
qo ro3
 1  (r / ro ) 3 
9

2qo
qoro2
  ( r )
1
2
cosh  k z J 0 ( k r )
sinh  k L J 1 ( k ro )
(y)
k 1 k o
k
dimensionless. (ii) Each term in (y) is dimensionless.
Limiting check: If T1  T2 and qo  qo  0 no heat transfer can take place and the entire
cylinder should be at uniform a temperature To . Setting qo  qo  0 in (y) gives T ( r , z )  To .
Boundary conditions check: Setting r = 0 and r  ro in solution (y) shows that boundary
(6) Comments. (i) The key to solving this problem is to recognize that the non-
homogeneous term in differential equation (a) is a function of r. This suggests that the
function  in the assumed solution (b) should also be a function of r. (ii) Two parameters
q  L
characterize the solution: o 2 and .
q oro ro
3-73
PROBLEM 3.16
Heat is added at a uniform flux q o over a r

circular area of radius b at one end of a solid To 2 HBC
cylinder of radius ro and length L. The ro
remaining surface of the heated end and the 0
z qo
opposite surface are insulated. The cylindrical
surface is maintained at uniform temperature To . L
Determine the steady state two-dimensional
temperature distribution.
are two non-homogeneous boundary conditions. Defining a new temperature variable,
  T  To , makes the boundary condition at the cylindrical surface homogeneous. This
gives the radial direction two homogeneous conditions. (iii) Part of boundary (r,L) is heated
while the rest is insulated. Thus the heat flux along this boundary is non-uniform. (iv) A
cylindrical coordinate system should be used.
(3) Formulation.
energy generation.
(ii) Governing Equations. Defining   T  To and applying the above assumptions to
eq. (1.11) gives
 2 1    2
  0 (a)
 r2 r  r  z2
  (0, z )
r
  ( r,0)
z
  ( r, L) qo , 0rb
(4) k  f (r)   non-homogeneous
z 0 b  r  ro
This represents a specified heat flux which varies with r according to f(r) which is defined in
condition (4).
(4) Solution.
3-74
 ( r, z )  R( r ) Z ( z ) (b)
d 2Zk
 2k Z k  0 (c)
d z2
and
d 2 Rk dR
r2
2
 r k  2k r 2 Rk  0 (d)
dr dr
d 2Zk
 2k Z k  0 (e)
d z2
and
d 2 Rk dR
r2 2
 r k  2k r 2 Rk  0 (f)
dr dr
d 2Z0
0 (g)
d z2
and
d 2 R0 dR0
r 2
 0 (h)
dr dr

Rk (r )  C k J 0 ( k r )  Dk Y0 ( k r ) (j)
Z 0 ( z )  A0 z  B0 (k)
and
R0 ( r )  C0 ln r  D0 (l)
3-75

 ( r , z )  R0 ( r ) Z 0 ( z )   R (r ) Z
k 1
k k ( z) (m)

(j) and (l) gives
Dk  C0  0

J 0 ( k ro )  0 (n)
D0  0
With C0  D0  0 the solution R0 Z 0 vanishes. Boundary condition (3) gives

Ak  0
Substituting into (m) gives

 ( r, z )   (b cosh  z) J
k 1
k k 0 ( k r ) (o)
where bk  Bk Ck . Application of boundary conditions (4) to (u) gives


f (r)  k   b (sinh  L) J
k 1
k k k 0 ( k r ) (p)
where f(r) is defined in boundary condition (4).

(v) Orthogonality. To determine bk in (p) we apply orthogonality. Note that the
characteristic functions J 0 (k r ) in (p) are solutions to equation (f). Comparing (f) with
eq. (3.5a) shows that it is a Sturm-Liouville equation with
a1 (r )  1/ r, a2 (r)  0 and a3 ( r )  1
p( r )  w( r )  r and q( r )  0
ro ro   
0
f ( r ) J 0 (i r ) w( r )dr  k  
0
 k bk (sinh k L) J 0 (k r )J 0 (i r ) w( r )dr
 k 1 
3-76
ro ro
0
f ( r ) J 0 ( k r ) rdr  k k bk (sinh k L) 
0
J 02 ( k r ) rdr
Using the definition of f(r), the above becomes

b ro ro
qo 
0
J 0 ( k r ) rdr   b
(0) J 0 ( k r ) rdr  k k bk (sinh k L)  0
J 02 ( k r ) rdr
Noting that the second integral vanishes, the above gives

b ro
qo 0
J 0 ( k r ) rdr  k k bk (sinh k L) 0
J 02 ( k r ) rdr
2( qoro / k ) J 1 ( k b)
bk  (q)
( k ro ) (sinh k L) J 12 (k ro )
2
Substituting (q) into (o) and rearranging gives the solution to the temperature distribution in
the cylinder

T (r , z )  To J 1 ( k b )
qo ro
2 
k 1 ( k ro ) (sinh  k L) J 1 ( k ro )
2
(cosh k z ) J 0 ( k r ) (r)
k
dimensionless. (ii) Each term in (r) is dimensionless.
Limiting check: If qo  0 , no heat transfer can take place and the entire cylinder should be at
uniform temperature To . Setting qo  0 in (r) gives T ( r , z )  To .
Boundary conditions check: Setting r = 0 and r  ro in solution (r) shows that boundary
(6) Comments. Equation (r) shows that the temperature solution depends on two
dimensionless parameters L / ro and b / ro .
3-77
3-78

Heat Conduction ch3 Solution PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Heat Conduction ch3 Solution PDF

Uploaded by

Copyright:

Available Formats

PROBLEM 3.

Three sides of a rectangular plate L H are y h1 , T h2 , T

(1) Observations. (i) This is a steady state two-

Expressed in terms of  , the above gives the fourth boundary condition

Assuming that n is multi-valued, the above gives

Yn ( y )  Cn sinh n y  Dn coshn y (k)

Corresponding to each value of n there is a temperature solution  n ( x, y ). Thus

(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions

With A0  B0  0, the solution corresponding to n  0 vanishes. Application of boundary

where f ( x ) is defined in boundary condition (4).

0  ( x, H )  (T  To )sin m x dx  h2 L / 2  ( x, H )  (T  To )sin m x dx 

Evaluating the integrals and solving for a n we obtain

Using the definition of f ( x ) in boundary condition (4) the above gives

Corresponding to each value of n there is a solution  n ( x, y ). Thus

(iv) Application of Boundary Conditions. Applying boundary condition (d-1) to

Condition (d-2) applied to (s) gives the characteristic equation for n

where a n  An Dn . Thus the only remaining unknown is a n . Application of condition (d-4)

(v) Orthogonality. The characteristic functions, sin n x, in (z-1) are solutions to

 T  (q / 2k ) x  Ex  F )sin  xdx 

 T  (q / 2k ) x  Ex  F )sin  xdx 

1  (q  / 2k ) x 2  Ex  F sin  n xdx  ( F  T1 )(1 /  n )(cos n L  1) 

(q  / 2k3n ) 2 n L sin  n L  ( 2n L2  2)(cos n L)  2  ( E /  2n )sin  n L   n L cos  n L 

2  (q  / 2k ) x 2  Ex  F sin  n xdx  ( F  T2 )(1 /  n )(cos 2 n L  cos  n L) 

A rectangular plate L H is maintained at uniform y qo

where q (0) is the heat transfer rate through surface (0,y)

Expressed in terms of  , the above gives the fourth boundary condition

Assuming that n is multi-valued, the above gives

Corresponding to each value of  n there is a temperature solution  n ( x, y ). Thus

(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions

With A0  B0  0, the solution corresponding to n  0 vanishes. Application of boundary

where f ( x ) is defined in boundary condition (4).

Solving for a n and recalling that w(x) = 1 the above gives

Using the definition of f ( x ) in condition (4) the integral in (w) is evaluated

Evaluating the integral

A semi-infinite plate of width 2L is maintained at y To 2 HBC

(iii) Independent Variable with Two Homogeneous Boundary Conditions. The

Corresponding to each value of n there is a temperature solution  n ( x, y ). Thus

The complete solution becomes

(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions

where f ( y ) is defined in boundary condition (4).

Solving for a n and recalling that w(y) = 1 the above gives

Evaluating the integral in the denominator, the above gives

Using the definition of f ( y ) in condition (4) the integral in (v) is evaluated

Substituting (w) into (t) and rearranging the result gives

(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions

where f ( y ) is defined in boundary condition (4).

Interchanging the summation with integration, recalling that w( y )  1 and invoking

f ( y ) cos  n ydy  kan  n cosh  n L n H  (k / hH ) sinh  n L

Solving for a n the above gives

k n cosh  n L  ( n H )(k / hH ) sinh  n L

Evaluating the integral in the denominator, the above becomes

Using the definition of f ( y ) in condition (4) the integral in (u) is evaluated

 4(q1H / k ) sin  n H / 2  (q 2 H / k )(sin  n H  sin  n H / 2)

Substituting (v) into (s) and rearranging the result gives

Radiation strikes one side of a rectangular element y T1

Corresponding to each value of n there is a solution  n ( x, y ). Thus

(iv) Application of Boundary Conditions. Applying boundary condition (f-1) to

Condition (f-2) applied to (s) gives the characteristic equation for n

where a n  Bn Cn . Thus the only remaining unknown is a n . Application of boundary

To determine a n from this result we apply orthogonality.

Expressed in dimensionless form the becomes

Assuming that n is multi-valued, the above gives

(iv) Application of Boundary Conditions. Boundary condition (1) applied to solutions