MEEN 672

An Introduction to
The Finite Element Method
Fall Semester 2015
J. N. Reddy
e-mail: [email protected]

Read: Chapter 7
Computer
Implementation of
1-D Problems and the use
of Program FEM1D

LOGIAL UNITS OF A FEA PROGRAM

PREPROCESSOR

Read
geometry and material data
finite element mesh data
boundary (and initial) conditions
PROCESSOR
Generate finite element mesh
Calculate element matrices
Assemble element equations
Apply boundary conditions
Solve the equations
Check error tolerance and maximum
number of allowable iterations
JN Reddy

POSTPROCESSOR
Compute the solution
and its derivatives at
desired points of the
domain
Print/plot the results

Computer Implementation: 12

FLOW CHART OF PROGRAM FEM1D

PREPROCESSOR

ECHODATA
and
MESH1D

ASSEMBLE

PROCESSOR

BOUNDARY
EQNSOLVR
or
EIGNSLVR

POSTPROCESSOR
JN Reddy

COEFFCNT
or
TRANSFRM

SHAPE1D

POSTPROC
or
REACTION
Computer Implementation:
13

Flow Chart of a Typical PROCESSOR Unit

Initialize global K and F
DO N = 1 to NEM

Transfer global information

(material properties, geometry,
and solution) to element
Call ELMATRCS to calculate K(N)
and F(N), and assemble to form
global K and F
Call BOUNDARY to impose
boundary conditions and call
SOLVER to solve the equations
JN Reddy

Print
solution

STOP

Computer Implementation:
14

FINITE ELEMENT PROGRAM FEM1D-1

Variables used in the program

NPE
- nodes per element, n
ELX(i ) - Global coordinate of the ith node of element e, xie
ELK(i, j) - Element coefficient, K ije
ELF(i ) - Element coefficient, fie
AX 0, AX 1 Coefficients in the definition of a( x ) : a( x ) = AX 0 + AX 1 * x
SFL(i ) Element shape (or approximation) funtion, yie
DSFL(i ) Derivative of the ith shape function with respect to
dyi
the local (normalized) coordinate x :
dx
GDSFL(i ) Derivative of the ith shape function with respect
dyi dyi dx
1 dyi
to the global coordinate x :
J
=

dx dx dx
dx
15

FINITE ELEMENT PROGRAM FEM1D-2

Numerical Integration
n

x = x y ( x ) = ELX ( j) * SFL( j),

j =1

e
j

e
j

j =1

xb

dyi dy j
1 dyi 1 dy j
x yi ( x )y j ( x ) + dx dx dx = 1 yi (x )y j (x ) + J e dx J e dx J e dx
a
NGP

Fij ( xI )wI =
I =1

NGP

F (GAUSPT( I , NGP )) * GAUSWT ( J , NGP )

I =1

ij

1 dyi 1 dy j

Fij ( x ) = yi ( x )y j ( x ) +
Je ,
J e dx J e dx

dyi dx dyi 1 dyi

dx
=
=
= 0.5 he
, dx = J e dx or J e =
dx dx dx J e dx
dx
GAUSPT ( I , j) IthGauss point, xI ,for the j-point Gauss rule
GAUSWT ( I , j) IthGauss weight, wI ,for the j-point Gauss rule
16

IMPOSOITION OF BOUNDARY CONDITIONS

NSPV Number of specified primary variables
of the problem.
ISPV(I,J) Array containing the information about
the global node number and the local
degree of freedom that is specified.
ISPV(I,1) For the Ith boundary condition, the global
node number at which the BC is specified.
ISPV(I,2) For the Ith boundary condition, the local
degree of freedom that is specified.
VSPV(I) The specified value of the deg. of freedom.
Similar meaning for NSSV, ISSV, and VSSV for
specified secondary variables
17

IMPOSOITION OF BOUNDARY CONDITIONS

NDF= number of primary degrees of freedom at a node

(U K +1 , U K +2 , U K +3 )

I
K = ( I - 1) * NDF

NDF = 3
6

(U16 , U17 , U18 )

18

Table 7.3.2: Description of the input variables

to FEM1D (revised from the book)-1
_______________________________________________________________
* Data Card 1: TITLE
* Data Card 2: MODEL, NTYPE, ITEM

MODEL=1, NTYPE=0: A problem of MODEL EQUATION 1

MODEL=1, NTYPE=1: A circular DISK (PLANE STRESS)
MODEL=1, NTYPE>1: A circular DISK (PLANE STRAIN)
MODEL=2, NTYPE=0: A Timoshenko BEAM (RIE#) problem
MODEL=2, NTYPE=1: A Timoshenko PLATE (RIE) problem
MODEL=2, NTYPE=2: A Timoshenko BEAM (CIE##) problem
MODEL=2, NTYPE>2: A Timoshenko PLATE (CIE) problem
MODEL=3, NTYPE=0: A Euler-Bernoulli BEAM problem
MODEL=3, NTYPE>0: A Euler-Bernoulli Circular plate
MODEL=4, NTYPE=0: A plane TRUSS problem
MODEL=4, NTYPE=1: A Euler-Bernoulli FRAME problem
MODEL=4, NTYPE=2: A Timoshenko (CIE) FRAME problem
ITEM=0, Steady-state solution
ITEM=1, Transient analysis of PARABOLIC equations
ITEM=2, Transient analysis of HYPERBOLIC equations
ITEM=3, Eigenvalue analysis

* Data Card 3: IELEM, NEM

IELEM=0, Hermite cubic finite element

IELEM=1, Linear Lagrange finite element
IELEM=2, Quadratic Lagrange finite element

19

Table 7.3.2: Description of the input variables

to the program FEM1D -2
* Data Card 4: ICONT, NPRNT
ICONT=1, Data (AX,BX,CX,FX and mesh) is continuous
ICONT=0, Data is element dependent
NPRNT=0, Not print element or global matrices
but postprocess the solution and print
NPRNT=1, Print Element 1 coefficient matrices only
but postprocess the solution and print
NPRNT=2, Print Element 1 and global matrices but
NOT postprocess the solution
NPRNT>2, Not print element or global matrices and
NOT postprocess the solution
SKIP Cards 5 through 15 for TRUSS/FRAME problems (MODEL = 4), and
read Cards 5 through 15 only if MODEL.NE.4._____________________
* Data Card 5: DX(I)
Array of element lengths. DX(1) denotes the global coordinate of
Node 1 of the mesh; DX(I) (I=2,NEM+1) denotes the length of the
(I-1)st element. Here NEM denotes the number of elements in the
mesh.
20

Table 7.3.2: Description of the input variables

to the program FEM1D -3
Cards 6 through 9 define the coefficients in the model equations. All
coefficients are expressed in terms of GLOBAL coordinate x. See
Table 7.3 for the meaning of the coefficients, especially for
deformation of circular plates and Timoshenko elements.

* Data Card 6:

AX0, AX1, AX2

* Data Card 7:

BX0, BX1

* Data Card 8:

CX0, CX1

SKIP Card 9 for eigenvalue problems (i.e. ITEM=3)

* Data Card 9:

FX0, FX1, FX2

* Data Card 10:

NNM

21

Table 7.3.2: Description of the input variables

to the program FEM1D -4
SKIP Cards 11 through 16 if data is continuous (ICONT.NE.0). Cards 11-14
are read for each element (i.e., NEM times). All coefficients are expressed
in terms of the GLOBAL coordinates.
* Data Card 11: DCAX(n,i), i=1,2,3
* Data Card 12: DCBX(n,i), i=1,2
* Data Card 13: DCCX(n,i), i=1,2
* Data Card 14: DCFX(n,i), i=1,2,3
READ Cards 15 through 20 for TRUSS/FRAME problems (MODEL = 4).
SKIP Cards 15 through 20 if MODEL.NE.4 ___________________
* Data Card 15: NNM
SKIP Cards 16 through 18 for TRUSS problems (NTYPE = 0). Cards
16 through 18 are read for each element, i.e., NEM times _________
22

Table 7.3.2: Description of the input variables

to the program FEM1D -5
* Data Card 16: PR, SE, SL, SA, SI, CS, SN
PR - Poisson's ratio of the material#
SE - Young's modulus of the material
SL - Length of the element
SA - Cross-sectional area of the element
SI
- Moment of inertia of the element
CS - Cosine of the angle of orientation of the element
SN - Sine of the angle of orientation of the element; the
angle is measured counter-clock-wise from x axis
* Data Card 17: HF, VF, PF, XB, CNT, SNT
HF - Intensity of the horizontal distributed force
VF - Intensity of the transversely distributed force
PF
- Point load on the element
XB - Distance from node 1, along the length of the
element, to the point of load application, PF
CNT - Cosine of the angle of orientation of the load PF
SNT - Sine of the angle of orientation of the load PF; the
angle is measured counter-clock-wise from x axis
23

Table 7.3.2: Description of the input variables

to the program FEM1D -6
* Data Card 18: NOD
READ Cards 19 and 20 only for TRUSS problems (NTYPE = 0). Cards
19 and 20 are read for each element; i.e. NEM times ______________
* Data Card 19: SE, SL, SA, CS, SN
SE - Young's modulus of the material
SL - Length of the element
SA - Cross-ectional area of the element
CS - Cosine of the angle of orientation of the element
SN - Sine of the angle of orientation of the element
Angle is measured counter-clock-wise from x axis
HF - Intensity of the horizontal distributed force
* Data Card 20: NOD
* Data Card 21: NCON

24

Table 7.3.2: Description of the input variables

to the program FEM1D -7
SKIP Card 22 if no constraint conditions are specified (NCON = 0). Repeat
Card 22 NCON times:__________________________________________
* Data Card 22: ICON (i), VCON(i)
* Data Card 23: NSPV
SKIP Card 24 if no primary variables is specified (NSPV=0). Repeat
Card 24 NSPV times ______________________________________
* Data Card 24: ISPV(I,1), ISPV(I,2), VSPV(I) (I = 1 to NSPV)
ISPV(I,1) - Node number at which the PV is specified
ISPV(I,2) - Specified local primary DOF at the node
VSPV(I) - Specified value of the primary variable (PV)
(will not read for eigenvalue problems)
SKIP Card 25 for eigenvalue problems (i.e. when ITEM=3) __________
* Data Card 25: NSSV
25

Table 7.3.2: Description of the input variables

to the program FEM1D -8
SKIP Card 26 if no secondary variable is specified (NSSV=0). Repeat
Card 26 NSSV times __________________________________________
* Data Card 26:
ISSV(I,1) ISSV(I,2) VSSV(I) -

ISSV(I,1), ISSV(I,2), VSSV(I) (I = 1 to NSSV)

Node number at which the SV is specified
Specified local secondary DOF at the node
Specified value of the secondary variable (PV)

* Data Card 27: NNBC

SKIP Card 28 if no mixed boundary condition is specified (NNBC=0).
The mixed boundary condition is assumed to be of the form: SV+VNBC
*(PV-UREF) = 0. Repeat Card 28 NNBC times ____________________
* Data Card 28: INBC(I,1), INBC(I,2), VNBC(I), UREF(I) (I = 1 to NNBC)
INBC(I,1) - Node number at which the mixed B.C. is specified
INBC(I,2) - Local DOF of the PV and SV at the node
VNBC(I) - Value of the coefficient of the PV in the B.C.
UREF(I) - Reference value of the PV
26

Table 7.3.2: Description of the input variables

to the program FEM1D -9
* Data Card 29: NMPC
SKIP Card 30 if no multipoint constraints are specified (NMPC=0).
Repeat Card 30 NMPC times ________________________
* Data Card 30: I=1, NMPC
(IMC1(I,J),J=1,2),(IMC2(I,J),J=1,2),(VMPC(I,J),J=1,4)
SKIP Card 31 if ITEM=0 (read for time-dependent and eigenvalue problems)
* Data Card 31: CT0, CT1
CT0 - Constant part of CT = CT0 + CT1*X
CT1 - Linear part of CT = CT0 + CT1*X
SKIP remaining cards if steady-state or eigenvalue analysis is to be
performed (ITEM=0 or ITEM=3) __________________
* Data Card 32: DT, ALFA, GAMA
DT - Time increment (uniform)
ALFA - Parameter in the time approximation scheme
BETA - Parameter in the time approximation scheme
27

Table 7.3.2: Description of the input variables

to the program FEM1D -10
* Data Card 33: INCOND, NTIME, INTVL
INCOND- Indicator for initial conditions:
INCOND=0, Homogeneous (zero) initial conditions
INCOND>0, Nonhomogeneous initial conditions
NTIME - Number of time steps for which solution is sought
INTVL - Time step intervals at which solution is to printed
SKIP Cards 34 and 35 if initial conditions are zero (INCOND=0) ___
* Data Card 34: GU0(I)
{GU0} - Array of initial values of the primary variables
SKIP Card 34 for parabolic equations (ITEM=1) ________________
* Data Card 35: GU1(I)
{GU1} - Array of initial values of the first time-derivatives of the
primary variables
28

TO RUN THE EXECUTABLE PROGRAM

FEM1DF15.EXE
Executable Computer Programs from the book, An
Introduction to the Finite Element Method by J. N. Reddy,
3rd ed., McGraw--Hill, 2006.
Notes:
Programs FEM1DF15 and FEM2D are a revised versions of the
programs from the second edition of the book. The revisions are
minor. Both FEM1DF15 and FEM2D were compiled using the
Microsoft Fortran compiler, with a fixed array dimensions.
Hence, only a limited size problem can be analyzed using the
compiled versions of the programs. The programs were compiled
with a maximum number of degrees of freedom of 2,500. If your
computer has the storage and you have the source programs, you
may recompile the program after changing the DIMENSION
statements in the programs.
29

TO RUN THE EXECUTABLE PROGRAM

FEM1DF15.EXE
To run a program on a PC, the files should be downloaded to your PC
into a folder (say, FEM_Reddy). The user is required to prepare a data
file for each problem he or she wants to solve, using the instructions in
the book (use Table 7.3.2 for FEM1DF15 and Table 13.4.1 for FEM2D).
Most errors are mistakes made in the preparation of the data files.
Therefore, you must check the data files when you see 'run-time error'
message or the program is not executed (by returning just the echo of the
input data file). All files should be in the same folder where
FEM1DF15.EXE and FEM2D.EXE are placed.
To run the program:
Double click on the executable file (FEM1DF15.EXE or FEM2D.EXE).
A window (called Command Prompt window) will pop open (with
white letter and black background). It will read
File name missing or blank please enter file name
UNIT 5?
30

TO RUN THE EXECUTABLE PROGRAM

FEM1D.EXE
Type the input file name (with its extension) and press Enter. For
example, if the file name you prepared is labeled as Prob1.inp, you
may type it as prob1.inp (not case sensitive). The Command Prompt
window will now display
File name missing or blank please enter file name
UNIT 6?
Type the output file name (with its extension) and press Enter. For
example, if you want the computer to return the output to file
Prob1.out, you may type it as prob1.out. Note that you do not prepare
this file; it will be created by the computer. The Command Prompt
window will disappear, indicating that it has taken the data file and
executed the program. You will find the file prob1.out in the same
folder where you are running the program. Depending on the results
you see, you may have to correct the data file.
31

EXAMPLE PROBLEMS FOR PROGRAM FEM1DF15

Example 1: One-Dimensional Heat transfer in a Rod
d 2q
- 2 + c q = 0, 0 < x < L
dx

bP
= 400
Ak
L = 0.05, c = 400,
= 100, k = 50
(all with proper units)
Insulated

c = m2 =

0 = 300
(a)
L

x
Convection heat transfer
through surface

x = x0 = 0.0
(b)

DX(2)

DX(3)

U1 = (0) = 300

DX(4)

DX( 1 ) = x 0 = 0 .0
DX(I) = h = 0 .0125
I = 2 , 3, 4 , 5

DX(5)

Q24 =

d
=0
dx x=L
32

One-Dimensional Heat transfer in a Rod

Example 1: Heat transfer in a rod (4 linear elements)
1 0 0
MODEL, NTYPE, ITEM
1 4
IELEM, NEM
1 1
ICONT, NPRNT
0.0 0.0125 0.0125 0.0125 0.0125
DX(1)=X0; DX(2), etc.
1.0 0.0 0.0
AX0, AX1, AX2
0.0 0.0
BX0, BX1
400.0 0.0
CX0, CX1
0.0 0.0 0.0
FX0, FX1, FX2
1
NSPV
1 1 300.0
ISPV(1,1), ISPV(1,2), VSPV(1)
0
NSSV
0
NNBC
0
NMPC
33

Example 2: 1-D Heat transfer in a composite wall

Material 1, k1

Material 2, k2

Material 3, k3

k1 = 70 W/(m C)
k2 = 40 W/(m C),

T= 50C

T0= 200C

k3 = 20 W/(m C)
h1 = 2 cm

Surface area, h2 = 2.5 cm

A (=1) h3 = 4 cm
h1
U1 = 200

h3

h2

T = 50 C

= 10 W/(m2 C)

Q2(3) + b (U 4 T ) = 0

34

1-D Heat transfer in a composite wall

Example 2: Heat transfer in a composite wall
1 0 0
MODEL, NTYPE, ITEM
1 3
IELEM, NEM
0 1
ICONT, NPRNT
0.0 0.02 0.025 0.04
DX(I)
70.0
0.0
0.0
0.0

0.0 0.0
0.0
0.0
0.0 0.0

AX0, AX1, AX2

BX0, BX1
CX0, CX1
FX0,FX1,FX2

Data for
Element 1

40.0 0.0 0.0

0.0 0.0
0.0 0.0
0.0 0.0 0.0

AX0, AX1, AX2

BX0, BX1
CX0, CX1
FX0,FX1,FX2

Data for
Element 2

35

1-D Heat transfer in a composite wall

20.0
0.0
0.0
0.0

0.0
0.0
0.0
0.0

0.0
0.0

1
1 1 200.0
0
1
4 1 10.0 50.0
0

AX0, AX1, AX2

Data for
BX0, BX1
Element 3
CX0, CX1
FX0,FX1,FX2
NSPV
ISPV(1,1), ISPV(1,2), VSPV(1)
NSSV
NNBC
INBC(1,1),INBC(1,2),VNBC(1),UREF(1)
NMPC

VNBC(1) = b ,

UREF(1) = T
36

Example 3: Axial deformation of a composite bar

Rigid plate
d = 4 in.

d = 2 in.
d = 2 in.
P

Al
P
8 in

Al
12 in

P = 100 kips = 105 lb

Steel (St) , Es = 30 106 psi
Aluminum (Al), Ea = 10 106 psi

St
4 in

Linear elastic spring

k = 1010 lb/in

Q2(3) + kU 4 = 0

U1 = 0
Q2(1) + Q1(2) = 2P

Q2(2) + Q1(3) = 0
37

Example 3: Axial deformation of a composite bar

Example 3: Axial deformation of a composite bar
1 0 0
MODEL, NTYPE, ITEM
1 3
IELEM, NEM
0 2
ICONT, NPRNT
0.0 12.0 8.0 4.0

DX(I)

12.56637E07 0.0 0.0 AX0, AX1, AX2

0.0 0.0
BX0, BX1
0.0 0.0
CX0, CX1
0.0 0.0 0.0
FX0, FX1, FX2
3.141593E07 0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0 0.0

AX0, AX1, AX2

BX0, BX1
CX0, CX1
FX0, FX1, FX2

Element 1

Element 2

38

Example 3: Axial deformation of a composite bar

9.424778E07 0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0 0.0

AX0, AX1, AX2

BX0, BX1
CX0, CX1
FX0, FX1, FX2

1
1 1 0.0
1
2 1 2.0E05
1
4 1 1.0E10 0.0
0

NSPV
ISPV(1,1),ISPV(1,2),VSPV(1)
NSSV
ISSV(1,1),ISSV(1,2),VSSV(1)
NNBC
INBC(1,1),INBC(1,2),VNBC(1),UREF(1)
NMPC

Element 3

39

Example 4: Bending of a Beam

z

U1 = 0
U2 = 0

1,000 N/m 2,500 N

EI = 2106 N-m2
1,250 N-m
c
Linear spring,

k = 10-4 EI (N/m)
5m
x 5m
2.5 m
2

U3 = 0
Q4(1) + Q2(2) = 1,250

Q3(2) + kU 5 = 2,500
Q4(2) = 0

40

Example 4: Bending of a beam

Example 4: Clamped and Spring-supported Beam (EBT)
3 0 0
MODEL, NTYPE, ITEM
0 2
IELEM, NEM
0 1
ICONT, NPRNT
0.0 5.0 5.0
DX(I)

0.0
2.0E6
0.0
1.0E3
0.0
2.0E6
0.0
0.0

0.0
0.0
0.0
0.0
0.0 0.0
0.0
0.0
0.0
0.0
0.0 0.0

AX0, AX1, AX2

BX0, BX1
CX0, CX1
FX0, FX1, FX2
AX0, AX1, AX2
BX0, BX1
CX0, CX1
FX0, FX1, FX2

Data for
Element 1
Data for
Element 2

41

Example 4: Bending of a beam

3
1
1
2
2
2
3
1
3
0

1 0.0
2 0.0
1 0.0

NSPV
ISPV(1,1), ISPV(1,2), VSPV(1)
ISPV(2,1), ISPV(2,2), VSPV(2)
ISPV(3,1), ISPV(3,2), VSPV(3)

2 1250.0
1 -2500.0

NSSV
ISSV(1,1), ISSV(1,2), VSSV(1)
ISSV(2,1), ISSV(2,2), VSSV(2)

1 2.0E02 0.0

NNBC (with transverse spring)

INBC(1,1),INBC(1,2),VNBC(1),UREF(1)
NMPC

42

SUMMARY
We have discussed the following topics in this presentation:
Review of FE Models of Chapters 3 6
Numerical integration in 1-D
Logical units of a FEA program
Flow chart of a typical processor unit
Element calculations
Computer program FEM1D
Input data to FEM1D
Example problems for FEM1D
Summary

Computer Implementation: 43