xx yy 00

xx yy 00

xx

xx xx

N
e e e e
h
e=1
128 NONLINEAR FINITE ELEMENT ANALYSIS

Domain Boundary Triangular element Quadrilateral element

(a) (b)

xx yy 00
!e

xx yy 00
!e

xx x yy y
"e

x y

x y
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 129

xx x yy y n

xx yy 00 n
!e "e

n n
e
n

ij
n

y
ny ^
n

dy nx
dx ds
Dom ain ,

Bou n dar y,

x
130 NONLINEAR FINITE ELEMENT ANALYSIS

n
e e e
h j j
j=1

e e e
j h j

e
i j j ij

e
j j

n e n e n
j j e
xx j yy j 00 j j
!e j=1 j=1 j=1

n
"e
n
e e e e
ij j i i
j=1

e e e e
e i j i j e e
ij xx yy 00 i j
!e

e e
i i
!e
e e
i n
"e

e e
ij ji
e
j

e e e e
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 131

(r!1) (r)

e e (r!1) e (r) e

(r!1) (r!1)

(r!1) (r!1) (r!1)

(r!1)
(r!1)

r (r!1)

e e (r!1) e e e (r!1) e e (r!1) e (r!1)

e n e
e i im e e
ij e m ij
j m=1 j
132 NONLINEAR FINITE ELEMENT ANALYSIS

Suppose that aexx and aeyy have the form

!u !u
aexx = aex (x, y) + aexu u + aexux + aexuy
!x !y
!u !u
aeyy e e e
= ay (x, y) + ayu u + ayux e
+ ayuy (5.4.10)
!x !y

where ax , axu , and so on are functions of only x and y. In addition, we assume that a is
only a function of x and y. Then we have
n
e
!Kim
Tije = Kij
e
+ ue (5.4.11)
!uj m
m

where
n n
e
!Kim !aexx !"ie !"m
e !ae !"ie !"m
e !ae e e
ue = + yy + " " dx dy uem
!uj m !uj !x !x !uj !y !y !uj i m
m m
n
!"je !" e !"ie !"m
e
= aexu "je + aexux + aexuy j
!x !y !x !x
m

!"je !" e !"ie !"m

e
+ aeyu "je + aeyux + aeyuy j dx dy uem
!x !y !y !y

!u !"ie !"je !" e

= aexu "je + aexux + aexuy j
!x !x !x !y

!u !"ie !"je !" e

+ aeyu "je + aeyux + aeyuy j dx dy (5.4.12)
!y !y !x !y

Note that, although [K e ] is symmetric, the symmetry of [T e ] depends on the nature of the
nonlinearity.

e e
ij ij
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 133

m m
ee ee
j j j j
j=1 j=1

n n
e e e e
j j j j
j=1 j=1

e
j
e e
j

e
134 NONLINEAR FINITE ELEMENT ANALYSIS

e e e e
e i j i j e e
ij xx yy 00 i j
!e

!"ie !"ie
!x !1 !#
!"ie !"ie
!y !$
e
i
e
i
e

!x !y e
e !# !#
!x !y
!$ !$

e
!x !y e m ! " i m ! " i
e !# !# i=1 i !# i=1 i !#
!x !y m ! " i m ! " i
!$ !$ i=1 i !$ i=1 i !$
e
e 1 1
! " 1 ! " 2 ! " m
!# !# !# 2 2
! " 1 ! " 2 ! " m
!$ !$ !$
m m

j j
e
j
e e
j j

11 22 12 21

!"ie !"ie !"ie

!x !# !#
!1 "
!"ie !"ie !"ie
!y !$ !$
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 135

" 22 " 12 " 11 " 21

11 12 22 21

e
ij

e e e e
e " i " i " j " j
ij xx 11 12 11 12

!
e e e e
" j " j " j " j
yy 21 22 21 22

e e
00 i j

ij

!

1 1 1 N
ij ij ij J J

! !1 !1 !1 J =1
M N
ij I J I J
I=1 J=1

I J I J

1 1
2 2
136 NONLINEAR FINITE ELEMENT ANALYSIS

e e
e i i
i

i i

!1

e e
i i

e e e

n e
e e im e
i ij m
m=1 j

e
i

x0 y0

x0 x00 x0x x0y y0 y00 y0x y0y

Consider heat transfer in an isotropic medium [axx = ayy in Eq. (5.2.1)] of rectangular shape
a b = 0.18 0.1 m. The conductivity axx = ayy = k is assumed to vary according to the
relation (a = 0)
k = k (1 + #T ) (5.5.13)
where k is the constant thermal conductivity, # the temperature coe!cient of thermal
conductivity, and T the temperature. Suppose that there is no internal heat generation (i.e.
f = 0) and the boundary conditions are
T (0, y) = 500! K, T (a, y) = 300! K (5.5.14a)
!T
= 0 at y = 0, b for any x (5.5.14b)
!y
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 137

Subroutine (in Fortran) ELECOFNT for the calculation of element

matrices e , e , and e of the model problem in Eq. (5.2.1).

SUBROUTINE ELECOEF(NPE,NGP,ITYPE,NONLN)
C ________________________________________________________________
C
C Element calculations based on linear and quadratic rectangular
C elements and isoparametric formulation are carried out for the
C model equation in (4.2.1).
C
C NPE Nodes per element (4: linear; 8: serendipity quadratic,
C 9: complete quadratic)
C NGP Number of Gauss points.
C ITYPE Type of iterative method used:
C ITYPE=1, Direct iteration, ITYPE>1, Newton Raphson iteration
C ________________________________________________________________
C
IMPLICIT REAL*8(A-H,O-Z)
COMMON/STF/ELF(7),ELK(9,9),ELXY(9,2),ELU(9)
COMMON/PST/A10,A1X,A1Y,A20,A2X,A2Y,A00,F0,FX,FY,
* A1U,A1UX,A1UY,A2U,A2UX,A2UY
COMMON/SHP/SF(9),GDSF(2,9)
DIMENSION GAUSPT(5,5),GAUSWT(5,5),TANG(9,9)
C
DATA GAUSPT/5*0.0D0, -0.57735027D0, 0.57735027D0, 3*0.0D0,
2 -0.77459667D0, 0.0D0, 0.77459667D0, 2*0.0D0, -0.86113631D0,
3 -0.33998104D0, 0.33998104D0, 0.86113631D0, 0.0D0, -0.90617984D0,
4 -0.53846931D0,0.0D0,0.53846931D0,0.90617984D0/
C
DATA GAUSWT/2.0D0, 4*0.0D0, 2*1.0D0, 3*0.0D0, 0.55555555D0,
2 0.88888888D0, 0.55555555D0, 2*0.0D0, 0.34785485D0,
3 2*0.65214515D0, 0.34785485D0, 0.0D0, 0.23692688D0,
4 0.47862867D0, 0.56888888D0, 0.47862867D0, 0.23692688D0/
C
C Initialize the arrays
C
DO 100 I = 1,NPE
ELF(I) = 0.0
DO 100 J = 1,NPE
IF(ITYPE.GT.1)THEN
TANG(I,J)=0.0
ENDIF
100 ELK(I,J)= 0.0
C
C Do-loops on numerical (Gauss) integration begin here. Subroutine
C SHPRCT (SHaPe functions for ReCTangular elements) is called here
C
DO 200 NI = 1,NGP
DO 200 NJ = 1,NGP
XI = GAUSPT(NI,NGP)
ETA = GAUSPT(NJ,NGP)
CALL SHPRCT (NPE,XI,ETA,DET,ELXY)
CNST = DET*GAUSWT(NI,NGP)*GAUSWT(NJ,NGP)
X=0.0
Y=0.0
U = 0.0
UX= 0.0
UY= 0.0
138 NONLINEAR FINITE ELEMENT ANALYSIS

Box 5.5.1 (continued)

DO 140 I=1,NPE
U=U+ELU(I)*SF(I)
UX=UX+ELU(I)*GDSF(1,I)
UY=UY+ELU(I)*GDSF(2,I)
X=X+ELXY(I,1)*SF(I)
140 Y=Y+ELXY(I,2)*SF(I)
C
C Define the coefficients of the differential equation
C
FXY=F0+FX*X+FY*Y
AXX=AX0+AXX*X+AXY*Y
AYY=AY0+AYX*X+AYY*Y
IF(NONLN.GT.0)THEN
AXX=AXX+AXU*U+AXUX*UX+AXUY*UY
AYY=AYY+AYU*U+AYUX*UX+AYUY*UY
ENDIF
C
C Define the element source vector and coefficient matrix
C
DO 180 I=1,NPE
ELF(I)=ELF(I)+FXY*SF(I)*CNST
DO 160 J=1,NPE
S00=SF(I)*SF(J)*CNST
S11=GDSF(1,I)*GDSF(1,J)*CNST
S22=GDSF(2,I)*GDSF(2,J)*CNST
ELK(I,J)=ELK(I,J)+A11*S11+A22*S22+A00*S00
C
C Define the part needed to add to [K] to define [T]
C
IF(ITYPE.GT.1)THEN
S10=GDSF(1,I)*SF(J)*CNST
S20=GDSF(2,I)*SF(J)*CNST
S12=GDSF(1,I)*GDSF(2,J)*CNST
S21=GDSF(2,I)*GDSF(1,J)*CNST
TANG(I,J)=TANG(I,J)
* +UX*(A1U*S10+AXUX*SXX+AXUY*S12)
* +UY*(A2U*S20+AYUX*SYX+AYUY*S22)
ENDIF
160 CONTINUE
180 CONTINUE
200 CONTINUE
C
C Compute the residual vector and tangent matrix
C
IF(ITYPE.GT.1)THEN
DO 220 I=1,NPE
DO 220 J=1,NPE
220 ELF(I)=ELF(I)-ELK(I,J)*ELU(J)
DO 240 I=1,NPE
DO 240 J=1,NPE
240 ELK(I,J)=ELK(I,J)+TANG(I,J)
ENDIF
RETURN
END

This is essentially a one-dimensional problem (in the x coordinate), and can be solved as
such (see Table 3.5.1).
Table 5.5.1 shows the linear and nonlinear solutions T (x, y ) for any y . Direct iteration
is used to solve the problem. It took two iterations to converge ( = 0.01). Also, the solution
is independent of the mesh in the y-direction. The present results were found to be identical
to those obtained with the one-dimensional model (for the same tolerance of = 0.01).
 TWO-DIMENSIONAL PROBLEMS IN SINGLE VARIABLE 139

Finite element solutions of a nonlinear heat conduction equation

[ 0 W/(m # K) and !3 (# K!1 )].

x Linear 4 4L* 2 2Q9 8 8L4 4 4Q9

0.0000 500.00 500.00 500.00 500.00 500.00

0.0225 475.00 ! ! 477.31 477.31
0.0450 450.00 454.02 454.03 454.03 454.03
0.0675 425.00 ! ! 430.12 430.12
0.0900 400.00 405.56 405.57 405.57 405.57
0.1125 375.00 ! ! 380.32 380.32
0.1350 350.00 354.33 354.34 354.34 354.34
0.1575 325.00 ! ! 327.58 327.58
0.1800 300.00 300.00 300.00 300.00 300.00

* m nL4, for example, denotes a mesh of the four-node linear (L) elements.

5.1 Consider the nonlinear problem of Example 5.5.1 (heat transfer in two dimensions). Use
the uniform 4 4 nine-node quadratic element mesh to analyze the problem using the
following data and boundary conditions:

a = 0.18 m, b = 0.1 m, f = 0 W/m (a)

k = k (1 + #T ) , k = 25 W/(m ! C) (b)
T (0, y) = 100 ! C , T (a, y) = 50 ! C
!T
k + hc (T ! T" ) = 0 at y = 0, b (c)
!n
Use # = 0.2, T" = 10! C, and hc = 50 W/(m ! C).
5.2 The energy equation for simultaneous conduction and radiation in a participating
medium can be expressed by
!$ [ke (T )$T ] = g

where
16$n T
ke (T ) = k +
3#
Here T is the temperature, g is the internal heat generation, n denotes the refractive
index of the medium, $ is the StefanBoltzman constant, and # is the Roseland mean
extinction coe!cient (see Ozisik [4]). Develop the nite element model of the equation
and determine the tangent coe!cient matrix for a planar (two-dimensional) domain.
5.3 Repeat Problem 5.2 for a radially axisymmetric domain.
5.4 Suppose that the boundary of a typical nite element is subject to both convective and
enclosed radiation heat transfer [cf. Eq. (2.3.36)]:

!u !u
(axx nx + ayy ny ) + hc (u ! uc ) + $(u ! uc ) = qn
!x !y

where $ is the StefanBoltzman constant and is the emissivity of the surface.

Reformulate the nite element equations in (2.3.39) to account for the radiation term.
Hint: $(u ! uc ) = $(u + uc )(u + uc )(u ! uc ) % hr (u)(u ! uc ).
140 NONLINEAR FINITE ELEMENT ANALYSIS

5.5 Compute the tangent coe!cient matrix associated with the nonlinear radiation boundary
condition of Problem 5.4.

1. Reddy, J. N., An Introduction to the Finite Element Method, 2nd edn, McGrawHill,
New York (1993).
2. Reddy, J. N. and Gartling, D. K., The Finite Element Method in Heat Transfer and
Fluid Dynamics, 2nd edn, CRC Press, Boca Raton, FL (2001).
3. Reddy, J. N., Energy Principles and Variational Methods in Applied Mechanics, 2nd
edn, John Wiley, New York (2002).
4. Ozisik, M. N., Finite Di!erence Methods in Heat Transfer, CRC Press, Boca Raton,
FL (1994).
5. Holman, J. P., Heat Transfer, 7th edn, McGrawHill, New York (1990).