Elementos Finitos Avanzados: Elemento Viga Plana C de Timoshenko
Elementos Finitos Avanzados: Elemento Viga Plana C de Timoshenko
Elementos Finitos Avanzados: Elemento Viga Plana C de Timoshenko
the beam model used in the Introduction to FEM course.) This is because some of the easier-construction
advantages cited for the C 0 element are less noticeable, while no artificial devices to eliminate locking are
needed. The C 1 element is also called the Hermitian beam element because the shape functions are cubic
polynomials specified by Hermite interpolation formulas.
9.2
BEAM MODELS
We consider a two-node,
straight, prismatic C 0 plane beamelement
moving in the (X, Y ) plane, as
1
Z (X ) (X )
Euler-Bernoulli
C
:
depicted in Figure 9.7(a). For simplicity in the following derivation the X axis system is initially
aligned with the longitudinal direction in the reference configuration, with origin at node 1. This
current
configuration
assumption
is relaxed
in the following section, once
invariant strain measured are obtained.
Current
cross section
area A0 and second moment of inertia I0
The reference element length is L 0 . The cross section
5
with respect to the neutral axis are defined by the area integrals
A0 =
motion
d A,
A0
Y, y
A0
Y d A = 0,
I0 =
Y 2 d A,
(9.1)
A0
u (X )
u X (X
As in the previous Chapter the identification superscript (e) will
be )omitted to reduce clutter until
it is necessary to distinguish elements within structural assemblies.
reference
configuration
The element
has the
six degrees of freedom depicted in Figure 9.4. TheseReference
degrees of freedom and
X displacement and nodecross
the associated node forces are collected in the node
force
vectors
section
f X1
u X1
Figure 9.2. Definition of beam kinematics in terms of the three displacement functions
uY 1
f Y 1the
EB model kinematics.
u X (X ), u Y (X ) and (X
figure actually depicts
). The
fby
1(X
1 normality
In the Timoshenko model,
) is not constrained
(see next figure).
f=
u=
,
.
f X2
u X2
uY 2
fY 2
2 through the centroid
f 2
rotation occurs about a neutral axis that passes
of the cross section.
(9.2)
Timoshenko
This
classical
beam
theory with first-order shear deformation
The loads Model.
acting on
themodel
nodescorrects
will be the
assumed
to be
conservative.
effects. In this theory cross sections remain plane and rotate about the same neutral axis as the EB
model, but do not remain normal to the deformed longitudinal axis. The deviation from normality
the beam model used in the Introduction to FEM course.) This is because some of the easier-construction
advantages cited for the C 0 element are less noticeable, while no artificial devices to eliminate locking are
needed. The C 1 element is also called the Hermitian beam element because the shape functions are cubic
polynomials
by Hermite interpolation formulas.
referencespecified
configuration
finite element idealization
of reference configuration
9.3. X -Aligned Reference Configuration
We consider a two-node,
straight, prismatic C 0 plane beam element moving in the (X, Y ) plane, as
1
Euler-Bernoulli
C : 9.7(a). For simplicity in the following derivation the X axis system is initially
depicted in Figure
aligned 1with the longitudinal direction in the reference configuration,
with origin at node 1. This
0 (Timoshenko)
(a) C (BE) model
(b)
C
model
2
assumption is relaxed in the following section, once invariant strain measured are obtained.2
The reference element length is L 0 . The cross section area A0 and second moment of inertia I0
with respect to the neutral axis5 are defined by the area integrals
1
A0 =
d A,
A0
A0
Y d A = 0,
I0 =
Y 2 d A,
(9.1)
A0
uY 2
Y, y In the current configuration those quantities
uY 2 become A, I and L, respectively, but only L is frequently
x previous Chapter the identification superscript (e) will be omitted to reduce clutter until
As inX,the
uX 1
uX 1
uX 2
u X 2 within structural assemblies.
it is necessary
to distinguish elements
2
1 9.4. These degrees of freedom
The1 element has the six degrees 2of freedom depicted in Figure
and
the associated node forces are collected in the node displacement and node force vectors
X1
fY 1
uY 1
f 1
1
f =mean
(9.2)are
= as measure
,
shear
. distortion.2 These angles
model, the difference = is uused
of
f X2
u X2
Superficially
it appears
thattheone
should
the latter
when shear effects are to be considered,
The loads
acting on
nodes
will select
be assumed
to beonly
conservative.
as in deep beams whereas the EB model is used for ordinary beams. But here a twist appear
because of finite element considerations. This twist is one that has caused significant confusion
the beam model used in the Introduction to FEM course.) This is because some of the easier-construction
advantages cited for the C 0 element are less noticeable, while no artificial devices to eliminate locking are
needed. The C 1 element is also called the Hermitian beam element because the shape functions are cubic
polynomials specified by Hermite interpolation formulas.
98
We consider a two-node,
straight, prismatic C 0 plane beam element moving in the (X, Y ) plane, as
1
Euler-Bernoulli
C : 9.7(a). For simplicity in the following derivation the X axis system is initially
depicted 1in Figure
linear-displacement-and-rotations
C 0 configuration,
2-node Cwith
(cubic)
element
aligned
the longitudinal
direction in the2-node
reference
with origin at node 1. This
element
for
Timoshenko
beam model:
for Euler-Bernoulli beam model:
assumption is relaxed in the following section,
once invariant strain measured are obtained.
plane sections remain plane but not
plane sections remain plane and
normal
to deformed
axis
section
area A0longitudinal
and second
moment of inertia I0
The
reference
element
length axis
is L 0 . The cross
normal
to deformed
longitudinal
with respect to the neutral axis5 are defined by the area integrals
A0 =
(a)
d A,
A0
A0
Y d A = 0,
I0 =
(b)
Y 2 d A,
(9.1)
A0
In the current configuration those quantities become A, I and L, respectively, but only L is frequently
C 1 element
used in the TL formulation. The material remains linearly elastic with elastic
modulus E relating
with same DOFs
the stress and strain measures defined below.
As in the previous Chapter the identification superscript (e) will be omitted to reduce clutter until
it is necessary to distinguish elements within structural assemblies.
The element has the six degrees of freedom depicted in Figure 9.4. These degrees of freedom and
the associated node forces are collected in the node displacement and node force vectors
f X 1 models based on
u Xtwo-node
1
Figure 9.6. Sketch of the kinematics of
beam finite element
u and (b) Timoshenko
f Y 1beam
theory. These
theory,
(a) Euler-Bernoulli beam
respectively,
in theFEM literature.
CY0 1beams,
models are called C 1 and
f 1
1
f=
(9.2)
u=
,
.
f X2
u X2
uY 2
fY 2
What would be the first reaction of an experienced
but old-fashioned
(i.e, never heard about
f
2
2
FEM) structural engineer on looking at Figure 9.6? The engineer would pronounce the C 0 element
unsuitable for practical use. And indeed the kinematics looks strange. The shear distortion implied
The
loads acting
ontothe
nodesviolate
will bethe
assumed
to be conservative.
by the
drawing
appears
grossly
basic assumptions
of beam behavior. Furthermore, a
huge amount of shear energy would be require to keep the element straight as depicted.
the beam model used in the Introduction to FEM course.) This is because some of the easier-construction
advantages cited for the C 0 element are less noticeable, while no artificial devices to eliminate locking are
needed. The C 1 element is also called the Hermitian beam element because the shape functions are cubic
polynomials specified by Hermite interpolation formulas.
97
9.2
BEAM MODELS
We consider0 a two-node, straight, prismatic C 0 plane beam element moving in the (X, Y ) plane, as
Timoshenko
:
to reference
depicted C
in Figure
9.7(a). For simplicity in the followingnormal
derivation
the X axis system is initially
beam
axis
X
aligned with the longitudinal direction in the reference configuration, with origin at node 1. This
assumption is relaxed in the following section,
once invariant strain measured are obtained.
normal to deformed
d A,
A0
A0
Y d A = 0,
I0 =
Y 2 d A,
(9.1)
A0
_
(X =
X)frequently
In the current configuration those quantities become A, I and L, respectively, //
butXonly
L is
90
used in the TL formulation. The material remains linearly elastic with elastic modulus E relating
the stress and strain measures defined below.
ds
As in the previous Chapter the identification superscript (e) will be omitted to reduce clutter until
_
it is necessary Note:
to distinguish
within structural
assemblies.
in practiceelements
<< ; typically
0.1% or less.
The magnitude
_
of the
issix
grossly
exaggerated
in thedepicted
figure for in
visualization
The element has
degrees
of freedom
Figure 9.4.convenience.
These degrees of freedom and
the associated node forces are collected in the node displacement and node force vectors
section
rotation
Figure 9.5. Definition of total section
and BE
f X 1 rotation in the
u X1
Timoshenko beam model.
deformation
f Y 1 is = , which
mean shear
u Y 1The
For smalldeformations
section.
is constant over the cross
of typical engineering
f 1
1
materials <<u1;=for
values
| .| are O(104 ) radians
f=
, typical
for
example,
u
f
whereas rotations
and X 2may
radians.
be of the order
Xof2 several
uY 2
2
fY 2
f 2
(9.2)
Although the Timoshenko beam model appears to be more complex because of the inclusion of 5
shearThe
deformation,
on thistomodel
are in fact simpler to construct! Here are
loads acting finite
on theelements
nodes willbased
be assumed
be conservative.
the two main reasons:
Elemento9.4Viga
de REFERENCE
Timoshenko
ARBITRARY
CONFIGURATION
915
_
// Y
Timoshenko C0:
Cinemtica II
uY 2
_
X
1(x1 ,y1)
Y, y
X, x
// X
_
Y
_
// X
2(x2 ,y2)
_
// Y
= +
uY 1
2(X2, Y2 )
uX 2
C0
uX
1(X1, Y1 )
Figure 9.8. Plane beam element with arbitrarily oriented reference configuration.
Timoshenko C0:
Cinemtica I
(a)
(b)
L
_
=
Y, y
(X)
2
C(xC ,yC )
xC
y
uYP
yC
P(x,y)
uYC
uY (X) = u YC
uXC
X
Po (X,Y)
Y
1
Lo
C(X+uX ,u Y )
C
uX (X) = u XC
uXP
X, x
Co (X,0)
Co (X)
C0
Figure 9.7. Lagrangian kinematics of the C 0 beam element with X -aligned reference
configuration: (a) plane beam moving as a two-dimensional body; (b) reduction
of motion description to one dimension measured by coordinate X .
p=
L0
BT z d X .
(9.34)
This expression may be evaluated by a one point Gauss integration rule with the sample point at
= 0 (beam midpoint). Let m = (1 + 2 )/2, m = m + , cm = cos m , sm = sin m ,
em = L cos(m )/L 0 1, m = L sin( m )/L 0 , and
12 L 0 m
cm
sm
12 L 0 m
cm sm
1
Bm = B| =0 =
(9.35)
sm
cm 21 L 0 (1 + em ) sm cm 12 L 0 (1 + em )
L0
0
0
1
0
0
1
Timoshenko
C0: m stands for beam midpoint. Then
where subscript
cm
T
p = L 0 Bm z = sm
0
sm
cm
0
1
L
2 0 m
cm
1
2 L 0 (1 + em ) sm
1
0
sm
cm
0
1
L
2 0 m
T (
12 L 0 (1 + em )
1
N
V
M
(9.36)
(9.37)
L0
This is again the sum of the material stifness K M and the geometric stiffness KG .
M
where S is the diagonal constitutive matrix with entries E A0 , G A0 and E I0 . Because h = B u,
the term BT z becomes BT SB u = K M u whence the material matrix is
!
KM =
BT SB d X .
(9.39)
L0
917
918
Chapter 9: THE TL TIMOSHENKO PLANE BEAM ELEMENT
918
This
integral
is evaluated
by the one-pointPLANE
Gauss rule
at ELEMENT
= 0. Denoting again by Bm the matrix
Chapter
9: THE
TL TIMOSHENKO
BEAM
918
(9.35),
we
find
This integral is evaluated by the one-point
Gauss rule at = 0. Denoting again by Bm the matrix
!
(9.35),
we findis evaluated by the one-point
This integral
Gauss rule at = 0. Denoting again by Bm the matrix
K M = ! BmT SBm d X = KaM + KbM + KsM
(9.40)
(9.35), we find
L0
T
a
b
s
K M = ! Bm SBm d X = K M + K M + K M
(9.40)
T
a
b
s
L
0
= axialB(bar),
d X = Kand
K M +stiffness,
KM
(9.40)
M to
Ka , Kb and Ks are K
due
respectively:
m SBmbending,
M +shear
Rigidezwhere
Material:
Chapter 9: THE TL TIMOSHENKO PLANE BEAM ELEMENT
L0
b
s
where KaM , K
M and 2K M are due to axial (bar), bending, and shear2 stiffness, respectively:
cm sm
cm m L 0 /2
cm
cm sm
cm m L 0 /2
c
s
where KaM , KbM and mK
are
due
to
axial
(bar),
bending,
and
shear
stiffness,
respectively:
2
cmcs2mM
cc
smm2 sm
m /2
m s2m
m /2
csm sm
cmmL0msL
c
cmmL0msL
0 /2
0 /2
m
m
2
2
2
2
c 2L /2 m Ls02sm /2
E A0
mmLL00/4
Lm0s2/2
m L
mmLL00/4
m
0 sm2s/2
s
cmmsm
c
/2 cm c
c
c
c
/2
m sLm0/2
mm m
m sLm0/2
KaM =
mm
m cmmc2msm0
2
cmss20msm /2 c
m2 LL20s/4
/2
s0ms2m /2 c
m2 LL20s/4
/2
L
E0A0 cc
m
m
cmmmLs0m/2 mcmLs
m 0 /2 m L
m 0 m /2 cmc
m 0 m /2
mcmms2L
KaM =
2m
2 m
2
2
2
s
s
L
s
/2
c
s
s
L
s
/2
c
2
2
m
m /4
m /4
cmmm0mLL
/2 cm m
/2
ELA0 cmc
mmmL 0 /2 c
cmmmL 0 /2 mcLmm0ssmm /2 cmm2m0mL2L
m Lm0ssmm /2
00
00
2
2
KaM = 0
m Ls20 /2 m L
/2
/4
/4
0 s2m
s
cm
LLm00sL
cmmmLLm00sL
m
c
/2 cm cmmcLms20m/2 m Lc0smsm2msm/2
/2
c
m mm
m0/2
m0/2
L 0 cmc
mmsm
m
(9.41)
2
c L /2 L 2s
2 2
2
2s /2
L0 s0 m/4/2 cm cmm L
s
mmLL0 0sm/4/2
c
m m
ms0mm /2
m
m sm0
m0 /2 m Ls0m
0 0mmL
0
0
0
0
2
cm m L 0 /2 m L 0 sm
/2
m2 L(9.41)
0 0 m2 L0 20 /40 0cm 0m L0/2 m L 0 sm /2
0 /4
0
0
0
0
0
0
(9.41)
E I0
0 0 1 0 0 1
(9.42)
KbM =
0 0 0 0 0 0
LE0I000 00 010 00 00 01
0
b
(9.42)
KM =
0
EL I00 00 00 001 00 00 01
b
(9.42)
KM =
0 0 1 0 0 1
L0 0 0 0 0 0 0
0 1 L00 sm 1
0 m2 10 cm sm
sm2
cm sm
a
/2
a1 L 0 sm /2
0 0 s
2
cm2m sm
ca
c
ca
0m a11L0L00s/2
1 0 cms
0smm2 1
cmsm2sm
m a11LL00s/2
c
cmmsm
m /2
m /2
2 2
2 2
G A0
a L s2 /2 cm a1cL20 /2
0msm
1 L20 /2
cam1 1aLL10L0/4
smm2/2 cm acc
cam1 1aLL10L0/4
cm sm
a
s0m/2
/2 a1 Lcs
a
s0m/2
/2
KsM =
m smm
1c0sm2mmsm
2
2
2
2
2
s
c
s
a
L
s
/2
s
c
s
a
L
s
/2
2 0 /2
2 0 /2
L 0A0
G
m
m
m
/2 cm ma1cL
a1 Lcmm0 ssmm/2 cmc
a1 L
1 Lm
0 ssm
c1am1a0L1mL
c1am1a0L1mL
ac
0 /4
0 /4
m
m
0 /2
0 /2
2m
2 m
KsM =
2
2sm
c
s
c
c
a
L
/2
c
c
c
a
L
/2
2
2
2
2
m
m
m
1
0
m
m
1
0
s
c
s
a
L
s
/2
s
c
a
L
s
/2
m
mm sLm /2
L
G
A
m
m
1
0
m
1
0
m
a
L
s
/2
c
a
L
/2
a
L
/4
a
L
s
/2
c
a
a
L
/4
0
0
m
m
1 0 m
m 1 0
1 0 m
m 1 0
21 2 0
21 2 0
KsM =
2
2
a1 L
s
/2
c
a
L
/2
a
L
/4
a
L
s
/2
c
a
L
/2
a
L
/4
2
2
0
m
m
1
0
1
0
m
m
1
0
c
s
c
c
a
L
/2
c
s
c
c
a
L
/2
1
0
1
0
m m
s
cm smm
a1mL 01sm0/2
smm m
cm sm
a1mL 01s(9.43)
L0
m0/2
m
2 2
2
a1 L 0 sm /2 cm a1 L20 /2
L 0msmsm/2 cm ac12L 0 /2 cam12 L
/40 /2
cm
cam1 L
a10L/40 /2 a1c
a10L
m
in which a1 = 1 + ecmm. sm
(9.43)
a1 L 0 sm /2 cm a1 L 0 /2
a12 L 20 /4
a1 L 0 sm /2 cm a1 L 0 /2
a12 L 20 /4
in which a1 = 1 + em .
(9.43)
9.6.2. Eliminating Shear Locking by RBF
in which a1 = 1 + em .
9.6.2. Eliminating Shear Locking by RBF
How good is the nonlinear material stiffness (9.42)-(9.43)? If evaluated at the reference configura9.6.2.
Eliminating
tion
aligned
XShear
axis,material
cLocking
smby=RBF
em(9.42)-(9.43)?
= m = 0, andIfwe
get
m = 1,stiffness
How
good iswith
the the
nonlinear
evaluated
at the reference configuration
aligned
with
the
X
axis,
c
=
1,
s
=
e
=
=
0,
and
we
get
m
m
m
m
isEthe
How good
material
stiffness
(9.42)-(9.43)?
If evaluated
at the reference
configuraA0nonlinear
0
0
ELA0
0
0
L
0and we get
tion aligned
cm = 1, sm = em = m = 0,
E0A0the XGaxis,
with
A00
0ELA0 G0A0
1 0
1 0
0L 0
G
A
G
A
0
0
0
2
2
L
L
G00A0
G00A0
E A0
1 0
1 0
E A0
1
WV =
L0
and W M
N12 L 0
N1 cos
N1 sin
N1 N2 L 0
N1 N2 L 0
N2 cos
N2 sin
N22 L 0
(9.53)
= 0. Notice that the matrices must be symmetric, since KG derives from a potential. Then
N1 cos
0
N2 cos
N1 sin
0
0
N2 sin
N1 cos
0
0
N2 cos
N1 sin
0
0
N2 sin
Rigidez Geomtrica:
'
L0
(W N N + WV V ) d X = KG N + KGV .
(9.54)
Again the length integral should be done with the one-point Gauss rule at = 0. Denoting again
quantities evaluated at = 0 by an m subscript, one obtains the closed form
0 0
0 0
Nm
sm cm
KG =
2 0 0
0 0
sm cm
0 0
0 0
Vm
cm sm
+
2 0 0
0 0
cm sm
sm
0
cm
0
1
2 L 0 (1 + em ) sm
sm
0
cm
0
1
2 L 0 (1 + em ) sm
cm
0
0
sm
0
0
1
2 L 0 m cm sm
cm
0
0
sm
0
0
1
2 L 0 m cm sm
0
sm
0
cm
1
cm 2 L 0 (1 + em )
0
sm
0
cm
cm 12 L 0 (1 + em )
cm
sm
1
2 L 0 m
.
cm
sm
12 L 0 m
(9.55)
10
10
In the sense that one must use more elements to get equivalent accuracy.
921