Sie sind auf Seite 1von 10

16

FEM Analysis Of
Plane Beam
Structure

161

Lecture 16: FEM ANALYSIS OF PLANE BEAM STRUCTURE

TABLE OF CONTENTS
Page

16.1
16.2
16.3

A Plane Beam Structure . . . . . . . . . . . . . . . . .


Finite Element Solution
. . . . . . . . . . . . . . . .
Analytical Solution By Discontinuity Functions
. . . . . . . .

162

163
163
168

16.2

FINITE ELEMENT SOLUTION

This is a supplementary Lecture on the Finite Element Method (FEM), designed to illustrate some
additional concepts: Rotational DOFs, Distributed Load Lumping, and FEM Convergence.
These fall beyond the basic exposition of the DSM in Lectures 1315.
Note: the material presented in this Lecture is not used in any HWs or Midterm Exams. However, it
will be of use in the Airplane Shaker Lab formalized in Fall 2012. Certain measured frequencies
will be compared agaisnt a FEM beam model of the fuselage.
16.1. A Plane Beam Structure
In this Lecture we will analyze the plane beam shown in Figure 16.1, with a two-element FEM
discretization. The FEM displacement results will be compared with the exact analytical solution
obtained by discontinuity functions (D.F.). Then the number of elements will be doubled to 4, 8,
etc., using Mathematica. As the number of elements is increased, the FEM solution will be visually
shown to rapidly converge to the analytical solution.
The beam span is 2L. It has uniform cross section of elastic modulus E and moment of inertia I
about z. The beam is fixed (clamped) at A and simply supported at B. It is loaded by a downward
uniform distributed force of magnitude w0 acting over the right half span L x 2L. The
problem is statically indeterminate.

;
;
;
y

EI constant
over entire beam

w0

;;
B

x
L

L
2L

Numerical properties for examples:


L = 36, w 0 = 16, EI = 1000000.

Figure 16.1. Plane beam problem.

16.2. Finite Element Solution


To illustrate the use of finite elements for beam structures we will discretize the problem of Figure
16.1 using two plane-beam finite elements as illustrated in Figure 16.2.
The equations of a generic plane beam element are worked out in detail in Chapter 12 of the
Introduction to Finite Elements course. For this course the development in that Chapter is overkill;
furthermore it requires graduate level mathematics. Only selected results from are reproduced here
for convenience.
This is posted as a PDF file in
http://caswww.colorado.edu/courses.d/IFEM.d/IFEM.Ch12.d/IFEM.Ch12.index.html
click on Chapter 12 link (first one).

163

Lecture 16: FEM ANALYSIS OF PLANE BEAM STRUCTURE

(a)

E I same for both elements

x
(1)

(2)

(1)

(2)

L =L

L =L

(b)
v1

v2

(1)

;
;
;

v3
3
3

(2)

2
w0

(c)

transformed to node forces


-w0 L /2

-w0 L /2

(2)

(1)
2

;
3

Figure 16.2. Analysis of the problem of Figure 16.1 by finite elements:(a) a two-element
FEM discretization, (b) the six degrees of freedom of the free-free
FEM model, (c) support conditions and applied forces.

Beam finite elements are obtained by subdividing beam members longitudinally. The simplest
plane beam element, depicted in Figure 3, has two end nodes, i and j, four node displacements
grouped in vector u(e) and four node forces, grouped in vector f(e) :
v (e)
u(e) =

i
i(e)
(e)
v
j
(e)
j

f(e)

f (e)
i
m i(e)

=
f (e)
j
m (e)
j

(16.1)

The stiffness relations of a generic prismatic plane beam element of length = L (e) , elastic modulus
E and moment of inertia I are

v (e) f i(e)
12
6 12 6
i
(e)
m (e)
E I 6
4 2 6 2 2

i = i

(16.2)

(e)
3 12 6 12 6 v j f j
6
2 2 6 4 2
(e)
m (e)
j
j
To form the master stiffness equations using the augment and add technique, write the previous
stiffness equations for elements (1) and (2) including all six degrees of freedom shown in Figure
16.2(b).
164

16.2

FINITE ELEMENT SOLUTION

y,v
j

E = E (e) , I = I (e)

vj
vi

x, u

i
x
= L(e)

Figure 16.3. The two-node plane beam element with four degrees of freedom.

Element (1), nodes 1-2, = L:

12
6L
EI
12

3
L 6L

0
0

6L
4L 2
6L
2L 2
0
0

12
6L
12
6L
0
0

6L
2L 2
6L
4L 2
0
0

0
0
0
0
0
0

v1(1) f 1(1)
0
1(1) m (1)
0
1

(1)

v (1)
f2
0 2
= (1)

m
0
2(1)

2
(1)

f (1)
0
v3
3
(1)
0
3
m (1)
3

(16.3)

0
0
6L
4L 2
6L
2L 2

0
0
12
6L
12
6L

v1(2) f 1(2)
0
1(2) m (2)
0
1
(2)

(2)

6L v2 f 2

(2) = (2)
2L 2

m
2 2
6L v (2) f (2)
3
3
4L 2
(2)
m (2)

(16.4)

Element (2), nodes 2-3, = L:

0 0
0
0
0 0
EI
0 0 12

L 3 0 0 6L

0 0 12
0 0 6L

Assemble by adding the matrices and suppressing element superscripts:

12
6L
12 6L
0
0
v1
f1
0
0 1 m 1
4L 2 6L 2L 2
6L

E I 12 6L
24
0
12 6L v2 f 2
=

2L 2
0
8L 2 6L 2L 2 2 m 2
L 3 6L

0
0
12 6L
12 6L
v3
f3
3
m3
0
0
6L
2L 2 6L 4L 2

(16.5)

The applied uniform distributed load over element (2) is converted to the two node forces 12 w0 L
on nodes 2 and 3, as shown in Figure 16.2(c).
There is a more refined way of converting distributed forces to node forces called consistent node force computation,
which is discussed in the IFEM Chapter referenced in the previous footnote. Consistent forces give generally better
results. The simple force lumping used here is easier to interpret physically and does not require knowledge of energy
methods.

165

Lecture 16: FEM ANALYSIS OF PLANE BEAM STRUCTURE

ClearAll[EI,L,w0];
L=36; w0=16; EI=1000000;
K=(EI/L^3){{12,6*L,-12,6*L,0,0},
{6*L,4*L^2,-6*L,2*L^2,0,0},
{-12,-6*L,24,0,-12,6*L},
{0,0,0,8*L^2,-6*L,2*L^2},
{0,0,-12,-6*L,12,-6*L},
{0,0,6*L,2*L^2,-6*L,4*L^2}};
Kred=(EI/L^3)*{{24,0,6*L},{0,8*L^2,2*L^2},{6*L,2*L^2,4*L^2}};
fred={-(1/2)*w0*L,0,0};
ured=LinearSolve[Kred,fred];
Print["Unknown displacements ured=",N[ured]];
u={0,0,ured[[1]],ured[[2]],0,ured[[3]]};
Print["Complete displacement solution u=",N[u]];
Print["Deflection vC at midspan= ", N[u[[3]]] ];
f=K.u; Print["Node forces including reactions=",N[f]];
Print["Node forces including reactions=",N[f]];

Unknown displacements ured={-0.979776, -0.011664, 0.046656}


Complete displacement solution u={0, 0, -0.979776, -0.011664, 0, 0.046656}
Deflection vC at midspan= -0.979776
Node forces including reactions={198., 3888., -288., 0, 90., 0}

Figure 16.4. Upper box: Mathematica script for FEM analysis


of two-element beam model. Lower Box: results.

Apply the known forces and displacements:

12
6L
12 6L
2
6L
2L 2
6L
4L

E I 12 6L
24
0

2
3
0
8L 2
2L
L 6L

0
0
12 6L
0
0
6L
2L 2

0
0
12
6L
12
6L


0
0
f1
0 0 m1


6L v2 12 w0 L

=
2L 2 2
0


6L
0
f3
4L 2
0
3

(16.6)

The applied force f 3 = 12 w0 L becomes part of the reaction taken by the right-end support, since
v3 = 0. (In FEM, displacement BCs take precedence over force BCs.)
Reduce by removing rows and columns 1, 2 and 5, which pertain to the known node displacements:


  1
0
6L
v2
2 w0 L
E I 24
(16.7)
2 =
0
0 8L 2 2L 2
L 3 6L 2L 2 4L 2
0

The master and reduced equations were implemented in Mathematica using the script shown on
the top box of Figure 16.4. Executing this script under Mathematica produces the output shown in
the lower box.
166

16.2

FINITE ELEMENT SOLUTION

(* Define singularity functions <x-a>^n for n=0,1,2,3,4 *)


SF0[x_,a_]:=If[x>a,1,0,0];
SF1[x_,a_]:=If[x>a,x-a,0,0];
SF2[x_,a_]:=If[x>a,(x-a)^2,0,0];
SF3[x_,a_]:=If[x>a,(x-a)^3,0,0];
SF4[x_,a_]:=If[x>a,(x-a)^4,0,0];
(* Express load q(x) in terms of SF, and integrate four times *)
ClearAll[L,EI,w0,x,C1,C2,C3,C4];
L=36; w0=16; EI=1000000;
q[x_]:=
-w0*SF0[x,L];
Vx[x_]:=
-w0*SF1[x,L]
+ C1;
Mx[x_]:=
-w0*SF2[x,L]/2 + C1*x+C2;
EIthetax[x_]:= -w0*SF3[x,L]/6 + C1*x^2/2+C2*x+C3;
EIvx[x_]:=
-w0*SF4[x,L]/24+ C1*x^3/6+C2*x^2/2+C3*x+C4;
(* Set the boundary conditions and solve for integration constants *)
BC= {Mx[2*L]==0,EIthetax[0]==0,EIvx[0]==0,EIvx[2*L]==0};
Print["Boundary conditions are: ",BC]
solBC=Solve[BC,{C1,C2,C3,C4}]; sol=Simplify[solBC[[1]] ];
Print["Constants of integration: ",sol];
(* Replace C1,C2,C3,C4 into the indefinite integrals & print results *)
V[x_]:=
Simplify[ Vx[x]/.sol ];
M[x_]:=
Simplify[ Mx[x]/.sol ];
theta[x_]:= Simplify[ (EIthetax[x]/.sol)/EI ];
v[x_]:=
Simplify[ (EIvx[x]/.sol)/EI ];
Print["Moment MC at midspan= ",M[L]," = ",
N[M[L]] ];
Print["Deflection vC at midspan= ",v[L]," = ", N[v[L]] ];
(* Plot q,V,M,theta and v over beam *)
Plot[q[x],{x,0,2*L},AxesLabel->{"x","q"},PlotLabel->"Load"];
Plot[V[x],{x,0,2*L},AxesLabel->{"x","V"},PlotLabel->"Shear force"];
Plot[M[x],{x,0,2*L},AxesLabel->{"x","M"},PlotLabel->"Bending moment"];
Plot[theta[x],{x,0,2*L},AxesLabel->{"x","theta"},PlotLabel->"Rotation"];
Plot[v[x],{x,0,2*L},AxesLabel->{"x","v"},PlotLabel->"Deflection"];

Boundary conditions are: {-10368 + 72 C1 + C2 == 0, C3 == 0, C4 == 0,


-1119744 + 62208 C1 + 2592 C2 + 72 C3 + C4 == 0}
Constants of integration: {C1 -> 207, C2 -> -4536, C3 -> 0, C4 -> 0}
Moment MC at midspan= 2916 = 2916.
41553
Deflection vC at midspan= -(-----) = -1.3297
31250

Figure 16.5. Upper box: Mathematica script for exact analysis of beam structure using
Discontinuity Functions and 4th order method. Lower box: results.

The beam deflections given by the FEM model are compared in Figure 16.6 against the exact
167

Lecture 16: FEM ANALYSIS OF PLANE BEAM STRUCTURE


0

Deflection v(x)

0.2
0.4

FEM, 2 elements

0.6
0.8
1

Exact

1.2
1.4
0

10

20

30

40

50

60

70

Figure 16.6. Deflection v(x) of 2-element FEM model versus exact solution.

solution obtained by discontinuity functions below.


Note that the center deflection given by the FEM model is v2 = v(L) = 0.979776, whereas the
analytical solution gives v(L) = 1.3297. The difference is approximately 30%. FEM analysis
generally produces only approximations to the analytical solution of the mathematical model.
The approximation can be improved by using more elements over the beam span. This will be
demonstrated in class from a laptop running Mathematica.
16.3. Analytical Solution By Discontinuity Functions
Recall that the discontinuity function, also known as Singularity Functions or SFs, x an for
n 0 is defined as

(x a)n if x > a
n
x a =
(16.8)
0
if x a
Here  . . .  are the so-called MacAuley brackets. The SF for n = 1, which is not included in the
definition (16.8), is Diracs delta function at x = a, denoted as x a1 in Vables book. The SFs
for n = 0 and n = 1 receive the names unit step function (or Heaviside function) and unit ramp
function, respectively, in electrical engineering.
We start by writing the applied load q(x) in terms of the SF x L0 , which is a unit step function:
q(x) = w0 x L0

(16.9)

V (x) = w0 x L1 + C1

(16.10)

M(x) = 12 w0 x L2 + C1 x + C2

(16.11)

E I (x) = E I v (x) = 16 w0 x L3 + 12 C1 x 2 + C2 x + C3

(16.12)

1
w0 x L4 + 16 C1 x 3 + 12 C2 x 2 + C3 x + C4
E I v(x) = 24

(16.13)

Integrate four times in x:

One notable exception are pin-jointed truss models in 2D or 3D. The FEM solution is then exact and cannot be improved
by using more elements per member.

168

16.3

ANALYTICAL SOLUTION BY DISCONTINUITY FUNCTIONS

Here (x) is the rotation of the beam cross section, which is same as the slope v (x) = dv(x)/d x
in the beam theory used in this course.
The four boundary conditions are
M(2L) = 0,

(0) = v (0) = 0,

v(0) = 0,

v(2L) = 0

(16.14)

Applying these conditions to expressions (11), (12) and (13) gives four linear equations from which
the four constants of integration C1 through C4 can be obtained. Once this is done, substituting
these constants into (10) to (13) gives the expressions for the shear force V (x), bending moment
M(x), cross section rotation (x) = v (x) and cross section deflection v(x) as functions of x.
These functions may be plotted over 0 x 2L.
To carry out the analysis for the specific values w0 = 16, L = 36, E I = 106 we use the Mathematica
program listed in the top box of Figure 16.5.
Executing this script under Mathematica produces the print output shown in the lower box of Figure
16.5, and also the plots of q(x), V (x), M(x), (x) = v (x), and v(x) over the beam span, shown
in Figure 16.7.

169

Lecture 16: FEM ANALYSIS OF PLANE BEAM STRUCTURE


q
Load
x
10

20

30

40

50

60

70

50

60

70

-2.5
-5
-7.5
-10
-12.5
-15
V
Shear force
200
100
x
10

20

30

40

-100
-200
-300

M
Bending moment
4000
2000
x
10

20

30

40

50

60

70

50

60

70

50

60

70

-2000
-4000
theta
Rotation
0.08
0.06
0.04
0.02
x
10

20

10

20

30

40

-0.02
-0.04
v
Deflection
x
30

40

-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4

Figure 16.7. Plots of q(x), V (x), M(x), (x) = v (x), and v(x) over the beam span.
These exact expressions are obtained through Discontinuity Functions (D.F.).

1610

Das könnte Ihnen auch gefallen