Finite Element Method:

One-Dimensional Heat
Conduction I

Engineering 471
Heat Transfer
 Computational Methods in Heat
Transfer

The ability to compute many

complex phenomena has
significantly increased in the
last few years.

http://www-heat.uta.edu
Computational Methods in Heat
Transfer
Computational Methods in Heat
Transfer
The most frequently
used computational
method used to solve
multi-dimensional
heat conduction
problems is known as
the “Finite Element
Method”.

During the next two lectures, we will develop the finite

element equations for one-dimensional heat conduction.
 Governing Differential Equation
(Derivation)

q conv = hP(T − T∞ )dx æ dq cond ö

ç q cond + dx ÷A
(Rate at which heat
is convected from
è dx ø
the surface) (Rate at which heat is conducted
out of the cross section)

q gen Adx
q cond A (Rate at which heat is internally
generated within the differential
volume)
(Rate at which heat
is conducted into P => perimeter
the cross section) h=> convection coefficient
dx A=> cross sectional area
 Governing Differential Equation
(Derivation Continued)
q conv = hP(T − T∞ )dx
æ dq cond ö 1st Law of Thermodynamics
ç q cond + dx ÷A
è dx ø
The governing differential
equation is obtained by
applying the 1st Law of
Thermodynamics
q gen Adx
E cv = E in − E out
q cond A
dx

∂T ∂q cond
CA dx = q cond A + q gen Adx - q cond A − dxA − hP(T − T∞ )dx
∂t ∂x
 Governing Differential Equation
(Derivation Continued)

∂T ∂q cond
CA dx = q cond A + q gen Adx - q cond A − dxA − hP(T − T∞ )dx
∂t ∂x

∂T ∂q cond
CA dx = q gen Adx − dxA − hP(T − T∞ )dx
∂t ∂x

Fourier’s Law of Heat Conduction

∂T
q cond = −k
∂x
Governing
∂T ∂ 2T
CA = Aq gen + kA 2 − hP(T − T∞ ) Differential
∂t ∂x Equation
 Boundary Conditions
n̂
x

x=L

T(L ) = TR
î or
n̂ k
∂T
î ⋅ n̂ + h (T − T∞ ) = 0
∂x
x=0
T(0 ) = TL
or
Each end can have either a
∂T
k î ⋅ n̂ + h (T − T∞ ) = 0 specified temperature or
∂x convection boundary condition.
 Galerkin Finite Element Process
(Assumed Solution)

Assume an approximate solution for the temperature distribution

T(x, t ) ≈ θ(x, t )
When the approximate solution is substituted into the
governing differential equation, there will be an error

∂θ ∂ 2θ
E(x, t ) = CA − Aq gen − kA 2 + hP(θ − T∞ )
∂t ∂x
 Galerkin Finite Element Process
(Assumed Solution)

There will also be an error in satisfying the

boundary conditions.

∂θ(0, t ) ∂θ(L, t )
e(t ) = − kA + h (θ(0 ) − T∞ ) + kA + h (θ(L ) − T∞ )
∂x ∂x

Left End Convection Right End Convection

We are going to restrict our choice of approximate

solutions to those that uniquely satisfy any specified
temperature boundary conditions.
 Galerkin Finite Element Process
(Error Minimization)

Error from Differential Equation

∂θ ∂ 2θ
E(x, t ) = CA − Aq gen − kA 2 + hP(θ − T∞ )
∂t ∂x
Error from Boundary Conditions
∂θ(0, t ) ∂θ(L, t )
e(t ) = − kA + h (θ(0 ) − T∞ ) + kA + h (θ(L ) − T∞ )
∂x ∂x

In the Galerkin Process, the approximate solution is

restricted to those functions that are normal to the error.
 Galerkin Finite Element Process
(Integral Equation)
L

ò θEdx + θ(0)e(0) + θ(L )e(L ) = 0

0
After making the substitutions for the error
equations, this integral equation becomes

L L L L
∂θ ∂ θ 2

ò0 θCA ∂t dx − ò0 q gen Aθ ⋅ dx − ò0 θKA ∂x 2 dx + ò0 θhP (θ − T∞ )dx



∂θ (0 )
− θ (0 )kA + θ (0 )hA (θ (0 ) − T∞ ) Governing
∂x Integral
∂θ (L ) Equation
+ θ (L )kA − θ (L )hA (θ (L ) − T∞ ) = 0
∂x
 Galerkin Finite Element Process
(Similification of Integral Eq.)

Consider the third term in the governing integral equation

L Integration by parts
∂ θ 2

ò0 θKA ∂x 2 dx vdu = d(uv) - udv

L
∂θ ö
L
∂θ ∂θ ∂θ ∂ 2θ
æ u = ; du = 2 dx
= ò d ç θkA ÷ − ò kA dx ∂x ∂x
0 è
∂x ø 0 ∂x ∂x ∂θ
v = θKA; dv = kA dx
∂θ (L ) ∂θ (0 ) ∂x
= θ (L )kA − θ (0 )kA
∂x ∂x
L
∂θ ∂θ
− ò kA dx
0
∂x ∂x
 Galerkin Finite Element Process
(Similification Integral Eq.)
L L L L
∂θ ∂ θ 2

ò θCA dx − ò q gen Aθ ⋅ dx − ò θKA 2 dx + ò θhP (θ − T∞ )dx

0
∂t 0 0
∂x 0

∂θ (0 )
− θ (0 )kA + θ (0 )hA (θ (0 ) − T∞ ) Governing
∂x Integral
∂θ (L )
+ θ (L )kA − θ (L )hA (θ (L ) − T∞ ) = 0 Equation
∂x
L
∂ 2θ
ò0 θKA ∂x 2 dx
∂θ (L ) ∂θ (0 )
L
∂θ ∂θ
= θ (L )kA − θ (0 )kA − ò kA dx
∂x ∂x 0
∂x ∂x
 Galerkin Finite Element Process
(Simplified Integral Equation)

Governing Integral Equation

L L L L
∂θ ∂θ ∂θ
ò0 θCA ∂t dx − ò0 q gen Aθ ⋅ dx − ò0 ∂x KA ∂x dx + ò0 θhP (θ − T∞ )dx


+ θ (0 )hA (θ (0 ) − T∞ ) − θ (L )hA (θ (L ) − T∞ ) = 0

We seek a solution Θ(x,t) to this integral equation. Θ(x,t) will

be an approximation to the solution of the governing
differential equation and boundary conditions.
Governing
∂T ∂T2
CA = Aq gen + kA 2 − hP(T − T∞ ) Differential

∂t ∂x Equation
∂T
k î ⋅ n̂ + h (T − T∞ ) = 0 Convection Boundary
∂x Condition