Академический Документы
Профессиональный Документы
Культура Документы
14
Analysis of Beams and Rigid Frames
14.1 INTRODUCTION
In Chapter 3 direct stiffness formulation of beams was discussed, which is actually stiffness matrix method for the analysis. In Chapter 5 shape functions were derived for two noded beam element by considering polynomial interpolation function and also using Hermitian functions both ultimately result into same shape functions. Assembling of strain displacement matrix [B-matrix] was presented in Chapter 4. In this chapter assembling of stiffness matrix for a two noded beam element by variational approach is presented. Taking simple problems for hand calculation, formulation of system equations and solution procedure is explained. The members of rigid frames are similar to beam elements, but their orientation in global system are different. Transforming the stiffness equations of beam element to global system is required before assembling global system. This aspect is presented and illustrated with numerical problem taking simple frame analysis. It was found difficult to extend the beam theory discussed above to plate bending. A new beam theory was developed in which lateral deflection w and rotation x were decoupled and treated as independent variables. However it needs inclusion of shear deformation and hence shear strain energy. Hence the beam theory becomes Co-continuity problem. This is known as Timoshenko Beam theory and the element developed on this theory is also presented in this chapter.
x=0 z s=0
= 1
le 2 1 s 2 =0
Fig. 14.1
l q
= 1 2
3 4 = w1 1 w2 2
243
i Rotation at node i.
In Chapter 5 the shape functions for such element have been determined as,
N1 = l
3x 2
2 le
2x3
3 le
N2 = x 3x 2
2 le
2x2 x3 + 2 le le 2x3
3 le
N3 =
and N 4 =
x2 x3 + 2 le le
(14.1)
If we use non-dimensiolising concept as used in isopatametric formulation i.e. varying from 1 to 1, then the shape functions are
N1 = N2 = N3 =
where
2 3 + 3 4 le 1 2 + 3 2 4 l 1 + 2 + 3 2 + 3 3 and N4 = e 2 4 4
(14.3)
2x 1. le
2y x 2
Since
y= N
lq
= N1
N2
N3
N4
lq
244
we get
M = EI = D
LM N MN x
2 2
2 N2 x
2
2 N3 x
2
2 N4 x
2
l q
OP l q PQ
e e
(14.4) (14.5)
where
[D] = EI
B =
LM N NM x
2 2
2 N2 x 2
2 N3 x 2
2 N4 x 2
OP QP
(14.6)
which is stress resultant curvature matrix. Strain Energy From basic solid mechanics we know strain energy dUe in an elemental length dx is given by
dU e =
1 M2 dV 2 EI 1 2
1 2
Ue =
z z
le 0 0
REI y U | | S x V | | T W
2 2
I dx EI
1 2
le
EI
2y x
le 2
dx =
1 = 2
lq
T e
z
0
z
0
le
EI B e
d l qi
le d 2
lq
dx
EI B
lq
since
2x 1 le
l U e = EI e 2
l q
1 2
T e
z
0
le
lq
2
(14.7)
Now,
B =
LM N MN x
2
2 N2 x
2
2 N3 x
2 N4 x
2
OP PQ
since
2x 1 le
Ni Ni = x
Ni = x
2 le
245
2 Ni x
2
2 Ni = x x le 4
2 le
FG IJ H K
2 Ni
d 2
2
F2 Gl H
e
Ni
I J K
B =
(14.8)
(14.9)
z
1
d = 2,
z
1
2 d = 0, and
z
1
3d =
2 3
We can write
1 U e = EI 2
lq
T le e 2
1 4 le
LM12 1 EI M = l q 2 l M MN
T 3 e
LM24 MM MN
24 12le 24
OP PP lq PQ
OP PP l q PQ
1 2
lq
T e
l q
(14.10a)
246
LM12 EI M is equal to l M MN
3 e
OP PP PQ
(14.10b)
Potential Energy Potential energy of an element is equal to strain energy minus the work done by the external forces acting on the element. Thus
1 e = Ue 2
z
le 0
pydx Pm Ym
Mk
F dy I G dx J H K
(14.11)
k
Where pDistributed load per unit length PmConcentrated load at point m MkExternal moment applied at k. The strain energy term Ue has been already derived. The work done by external loads can be assembled as explained below: Due to Uniformly Distributed load P/unit length:
pl = e 2
z
1
LM 2 3 + NM 4
py dx =
p N1
N2
N3
N4
l q
le d 2
le 1 2 + 3 2 4
2 + 3 3 4
le 1 + 2 + 3 2 4
Noting that
z z z
1 1 1
C d = 2C,
z z
1 1
OP l q d QP
e
d =
3 d = 0 and
2 d =
2 , we get 3 ple l 2 22 e 2 8 2 3
py dx =
LM N
F G H
I J K
22
le 2 2 + 2 3
F G H
I OP l q JQ K
LM pl NM 2
2 ple 12
ple 2
2 ple 12
OP lq QP
(14.12)
This equivalent load on the element is shown in the Fig. 14.2. The point loads like Pm and Mk are readily taken care by introducing nodes at the points of application.
Analysis of Beams and Rigid Frames Thus work done by external load is assembled. Let it be represented by
247
F
i
= F1 1 + F2 2 + F3 3 + F4 4 =
l q lFq
T e
(14.13)
y
p/unit length
x le z ple 2 ple 2
le pl2 e
12
pl2 e 12
Fig. 14.2
Minimization of Potential Energy It is to be noted that, in solid mechanics, minimization of potential energy of entire structure is to be assembled. In finite element analysis the total potential energy of the system is considered as the summation of total potential energy of the elements. Thus
= e = Ue
l q lFq
T e
1 F 2
l q k l q l q lFq
T T
where { } and {F} are nodal unknown vector and load vector respectively. From the principle of minimization of potential energy we get,
d =0 d
k F=0
or
lq
k =F
lq
(14.15)
In finite element analysis, element stiffness matrix [k]e is assembled and placed in global matrix at appropriate place. When this process is completed for all the elements, we get global stiffness matrix [k]. Similarly global load vector {F} is assembled. The necessary boundary conditions are imposed by (i) Elimination Method if the hand calculations are made or by (ii) Penalty Method if computers are used.
248
The solution of equation 14.14 gives the displacement vector { }. The required stress resultants are determined for each element. Moment at nodes:
M = EI B
= EI
2 le
lq
1 3 le
b1 + 3 gl l q
e
(14.15)
V=
=
dM dM d 2 dM = = dx d dx le d
6 3le 6 3le
2 EI
3 le
l q
k
EI
3 le
12 6le
12 6le
l q
(14.16)
The reactions at supports are nothing but end equilibrium forces. Hence
lRq =
lq
(14.17)
R4
R2
le R1 R3
RR U | | | | |R | | | = EI S V l |R | | | | | |R | | | T W
1 2 3 3 e 4
LM 12 MM MM 6l MM12 MM MN 6l
e e
6l e
2 4l e
12 6l e 12 6l e
6l e
2 2l e
OP R PP | | 2l P | PP | S 6l P | PP | | 4l P | T Q|
6l e
2 e e 2 e
U R pl | |2 | | pl | | | | 12 V S pl | | | |2 | | pl | | | T 12 W |
e 2 e e
2 e
U | | | | V | | | | | W
Note that the above reactions are to be interpreted as per sign conversion i.e. R1, R2 are positive in positive direction of z-axis (i.e. downward) and R2, R4 (the end moments) are positive when they are clockwise (refer Fig. 14.3). Example 14.1: Analyse the beam shown in Fig. 14.4 (a) by finite element method and determine the end reactions. Also determine the deflections at mid spans given E = 2 105 N/mm2 and I = 5 106 mm4
249
5m (a)
1 3
5m
1
2
1
4
2
6
(b)
Fig. 14.4
106 103
= 2 108 kN/m 2
I = 5 106 mm4 = 5 106 m4 EI = 2 108 5 106 = 1000kN-m2 Let the two elements be numbered as shown in Fig. 14.4 (b). The nodal displacement vector is
l q
= 1 2
3 4 5 6
6l1 12 6l1 12 30 12 30 2 2 4l1 6l1 2l1 1000 30 100 30 50 = 3 6l1 12 6l1 12 30 12 30 5 2 2 2l1 6l1 4l1 30 50 30 100
3 4 Global Numbers
k1 =
E1 I1
3 l1
LM 12 MM 6l MN612 l
1 1 2
OP PP PQ
LM MM MN
OP PP PQ
Similarly,
30 12 30 100 30 50 30 12 30 50 30 100
OP 1 PP 2 3 PQ 4
30 100 30 50
12 30 12 30
k2
OP 50 P 30P P 100 Q
30
Global Numbers
3 4 5 6
250
LM 12 MM 30 12 = 8M MM 30 MM N
LM 12 MM 30 12 = 8M MM 30 MM 0 N0
1
30 12 100 30 12 30 +12 50
30 +30
30 50
30 +30 100 +100
12 30 12 30
12 30
30 50
OP P 30 P P 50 P P 30P 100 P Q
6
global numbers
1 2 3 4 5 6
OP 1 PP 2 PP 3 4 PP 5 PQ 6
global numbers
R pl |2 | pl | | F = S 12 | pl |2 | pl | | 12 T
e 2 e e
2 e
U | | | | V | | | | | W
2
Global Numbers
251
Global Numbers
LM 12 MM 30 12 8M MM 30 MM 0 N0
OP R PP | | PP | S PP | | | QP | T
1 2 3 4 5 6
1 = 2 = 3 = 5 = 0
Imposing them by elimination method, we get
i.e.,
i.e.,
LM200 50 OP R U = R25U N 50 100Q S V S 50 V T W T W L4 1OP R U = R25U 400 M N1 2Q S V S 50 V T W T W R U = 1 LM4 1OP R25U S V 400 N1 2Q S 50 V T W T W
4 6
4 6
4 6
252
=
End reactions
2 1 1 1 400 8 1 1 4
LM N
Answer
RR U LM 12 |R | M 30 | |=8 SR V M12 | | M |R | N 30 T W
1 2 3 4
12 30 100 30 50 30 12 30 50 30 100 30
since
1 = 2 = 3 = 0 and 4
For element 2
RR U LM 12 |R | M 30 | |=8 SR V M12 | | M |R | N 30 T W
3 4 5 6
30 12 30 100 30 50 30 12 30 50 30 100
0 | | | | OP R 100 U R60U R70.714U | | S V | . | PP | 2800 | |50| = |53571| S 0 V |60| S49.286V PQ | 225 | | 50 | | 0 | | | T W | | T W | 2800 | T W
Answer
y = N1 =
N2
N3
3
N4
LM 2 3 + MN 4
l q
le 1 2 + 3 2 4
2 + 3 3 4
le 1 + 2 3 2 4
OP l q PQ
0.5 0125le .
l q
= 0.02232 m = 22.32 mm
253
= 0.02790 m = 27.9 mm = 27.9 mm, downward Answer Example 14.2: A beam of length 10 m, fixed at one end and supported by a roller at the other end carries a 20 kN concentrated load at the centre of the span. By taking the modulus of elasticity of material as 200 GPa and moment of inertia as 24 106 m4, determine: 1. Deflection under load 2. Shear force and bending moment at mid span 3. Reactions at supports Solution: The beam is shown in Fig. 14.5 (a). Its finite element idealization is shown in Fig. 14.5 (b). In this problem, E = 200 GPa = 200 109 N/m2 = 200 106 kN/m2
20 kN
5m (a)
1 3
5m
13.75
6.25
Fig. 14.5
254
l q
ki = Ei I i
3 le
= 1 2
3 4 5 6
LM 12 MM 6l MN612 l
e e
12 6le 12 6le
3
OP PP PQ
, Global Numbers
k1 =
4800 125
LM 12 MM 30 12 MN30
4
30 100 30 50
12 30 12 30
OP 50 P 30 P P 100 Q
30 3 4 5 6
1 2 3 4
Similarly,
k2
30 100 30 50
12 30 12 30
OP 50 P 30P P 100 Q
30
4
, Global Numbers
OP PP PP PP QP
, Global Number
1 2 3 4 5 6
lFq
The stiffness equation is The boundary conditions are,
= 0 0 20 0 0 0 k = F
lq lq
1 = 2 = 5 = 0
Analysis of Beams and Rigid Frames Imposing these boundary conditions by elimination method, the stiffness equation reduces to
255
EI 125
By using direct inversion, we get
LM24 MM 0 N30
0 200 50
OP R | 50 P S 100P | QT
30
3 4 6
R | S | T
3 4 6
U 125 LM20000 2500 1 |= V EI 24b20000 2500g + 30b6000g M 1500 | MN 6000 W LM17500 1500 6000OP R20U 125 1 | | = MM 1500 1500 1200PP S 0 V EI 240000 T W N6000 1200 4800 Q | 0 | R182.292U 1 R182.292U R 0.03798 1 | = S 15.625 | = 4800 | 15.625 | = |3.2552 10 V S EI | | | 62.5 V S 0.01302 | | T 62.5 W T W T
OP R20U | | PP S 0 V T W Q |0|
U | V | W
Deflection under the load = 3 = 0.03798 and rotation under the load = 4 = 3.2552 106 radians Shear Force and bending Moment at midspan: Considering element (1), for mid span of beam = 1
LM 0 OP 0 1 M EI P 6 2l 6 4l M= EI M 182.292P l MN 15.625 PQ LM 0 OP 0 1 = 6 10 6 20 M MM182.292PPP = 31.250 KN m Answer 25 N 15.625 Q LM 0 OP 0 1 M EI P = 31.75 kN Answer 12 6 5 12 6 5 V = 125 EI M182.292P MN 15.625 PQ
2 e e e
256
R U SR V T W
R1
2
LM N
LM OP M Q MM N
OP PP R U SV PQ T W
RR U = R13.75U SR V S37.50V T W T W
1 2
R U S V T W
LM N
R6.25U S0V T W
The reactions and shear force and bending moment at midspan are shown in Fig. 14.5 (c). Example 14.3: Derive the expression for consistant load, which varies linearly from p1 at node 1 to p2 at node 2 on a beam element of length le. Solution: The element with given load is shown in Fig. 14.6.
p2 p1
1 2
le
Fig. 14.6
p = p1 + p2 p1
g lx
e
= p1 + p2 p1
= p1
g b1 + g since lx = 1 + 2 l
e
e
1 1+ + p2 2 2
(14.18)
257
lFq =
e
z z
1
N T pdx NT p le d = 2
N T p1
1 le d + 2 2
N T p2
1 + le d 2 2
3
z
1
R 2 3 + U | | 4 | | | 1 + l | | | 4 2 | 1 l | S 2 + 3 V 2 p 2 d + | | | 4 | | 1 + + l | | | | 4 2| W T
3 2 3 e e 2 1 2 3 e 1
z
1
R 2 3 + U | | 4 | | | 1 + l | | | 4 2 |1+ | S 2 + 3 V 2 p | | | 4 | | 1 + + l | | | | 4 2| W T
2 3 e 2 2 3 e 2
le d 2
Now
z
1
F 2 3 + GH 4
I b1 g d = e2 5 + 3 JK
1 1
Noting that
1 1
and
2 3 + 3 2 2 28 1 d = 2 2 0 + 3 + 0 = 4 3 5 5 1 The first term in p1 is, p1le pl 28 = 1 e 21 16 5 60 Similarly the other terms may be evaluated. Finally we get,
z
1
z z z
1 1
+ 3 + 4 d
c d = 2c,
1 3
z z
1 1
d =
3 d = 0
L O d = M P MN 3 PQ L O d = M P MN 5 PQ
5
=
1
2 3
=
1
2 , 5 we get
b g
R 21 U R 9 U | | | | p l | 3l | p l | 2l | Fq = l 60 S 9 V + 60 S 21 V | | | | |2l| |3l| T W T W
1 2 e
(14.19)
258
Example 14.4: Determine the consistent nodal vector due to loads acting on the beam shown in Fig. 14.7.
100 kN 20 kN/m
4m (a)
6m
F1
F3
F5
F2
F4
(b)
F6
Fig. 14.7
Solution: The beam is idealized with two elements as shown in Fig. 14.7 (b). Due to concentrated load, the nodal vector is directly obtained since there is a node directly under the load. It is given by
Global Numbers
259
Global Numbers
Due to udl,
Answer
The members of a rigid frame differs from the beam in the following two respect: (i) They carry axial loads also and hence their deformation in axial direction also is to be considered. (ii) They are oriented in any direction in the plane. The typical frame element is shown in Fig. 14.8. The element selected is in x y plane. The right hand thumb rule is used for the selection of Cartesian coordinates. Let x y be the local coordinate system. The nodal displacement vector in local coordinate system is
l q
T e
= 1 2
5 3 4 6
(14.20)
lq
T e
= 1 2
3 4 5 6
(14.21)
260
x'
6 4
y y' '2
y '1 y'
2
'3
3 1
x
(a) (b)
Fig. 14.8
1e = 1e cos + 2 e sin = 1e l + 2 e m
e = 1e sin + 2e cos = 1e m + 2e l 2 e = 3e 3
Similarly
e = 4 e l + 5e m 4 5e = 4 e m + 5e l
e = 6e 6
where l, m are direction cosines
l q = L l q
e
(14.22a)
where
LM l MM m 0 =M MM 0 MM 0 N0
m 0 0 0 0 l 0 0 0 0 0 1 0 0 0 0 0 l m 0 0 0 m l 0 0 0 0 0 1
OP PP PP PP PQ
(14.22b)
1 and are like the degrees of freedom of bar element while , 3 , 5 and are like the degrees 4 2 6 of freedom of beam element. Hence the stiffness matrix of frame element in the local coordinate system can be obtained by appropriately placing the two stiffnesses as
261
LM EA MM l MM 0 MM 0 =M MM EA MM l MM 0 MM 0 N
e e
0 12EI
3 le 6 EI 2 le
0 6 EI
2 le 4 EI
EA le 0 0
0 12EI
3 le 6 EI 2 le
le 0 6 EI
2 le 2EI le
0 12EI
3 le 6 EI 2 le
EA le 0 0
0 12EI
3 le 6 EI 2 le
OP P 6 EI P l P P 2EI P l P P 0 P P 6 EI P P l P 4 EI P P l P Q
0
2 e 2 e 2 e e
(14.23)
l q = L l q U = e L l q j = l q L
e e T e
b g
e
T e
l q
e T
k k
lq L l q = l q
L
e e
T e
where
k e= L
(14.24)
[k]e is element stiffness matrix in global coordinate system. Using stiffnesses matrices of all elements, system stiffness matrix [k] can be assembled.
Load Vector
ple 2
2 ple 12
x'
y y' ple 2 x
y pl2 e 12 x
Fig. 14.9
262
Finite Element Analysis Figure 14.9 shows the uniformly distributed load acting on the typical element. Noting that the load
By transforming it to global system, we get, {F}e = [L ]T {F }e . Using such expressions for all elements the load vector of the system due to the distributed loads can be assembled. The load vector due to concentrated loads and external moments can be directly added to global load vector, since at all such loads nodes are selected. The final system equation is [ k ] { } = {F} as usual. After introducing boundary conditional, the system equations are solved to get nodal displacement vector { }. Then the required stress resultants can be assembled. Example 14.5: Assemble element stiffness matrices for the rigid frame shown in the Fig. 14.10 (a). Explain how do you proceed further to solve the problem. Take, E = 200 GN/m2 Io = 40 106 m4 A = 4 103 m2
20 kN/m
lF q = LMM0 N
e
ple 2
2 ple 12
ple 2
2 ple 12
OP QP
(14.25)
30 kN 2 2I 0
3m
I0 1
I0
4 4m (a)
5 64
8 7
2 1 3 (b)
11 12 10
Fig. 14.10
263
Solution: Figure 14.10 (b) shows the positive directions of the 12 nodal displacements. Thus in this problem
lq = l
T
2 ... 12
q
0 12 EI
3 le 6 EI 2 le
LM EA MM l MM 0 MM 0 =M MM EA MM 0l MM MM 0 N
e e
0 12 EI
3 le
0 6EI
2 le
EA le 0 0
6EI
2 le
4 EI le 0 6EI
2 le
0 12 EI
3 le
EA le 0 0
0 12 EI
3 le 6 EI 2 le
6EI
2 le
2 EI le
OP P 6 EI P l P P 2 EI P l P P 0 P P 6 EI P P l P 4 EI P PP l Q
0
2 e e 2 e e
For element No. 1 le = 3, EA = 200 109 4 103 = 800 106 N = 800 103 kN EI = EI0 = 200 109 40 106 = 800 103 N m2 = 800 kN m2
0 0 0 0 333.333 0.444 0.667 0 0.444 0.667 0.667 1333 0 0.667 0.667 . 0 0 333333 0 0 . 0 0.444 0.667 0.444 0.667 0.667 0.667 0 0.667 1333 .
OP PP PP PP PQ
264
LM 0 MM1 0 L=M MM 0 MM 0 N0
1 0 0 0 0 1
0 0 0
0 0 0 0
0 0 0 1 0 0 1 0 0 0 0 0
OP 0P 0P P 0P 0P P 1P Q
. 333333 0 0 0.444 0 0.667 0.667 0 0.667 . 0 333333 0 0.667 0.444 0 . 0.667 0 1333
k L
OP PP PP PP PQ
0 0 1 0 0 0
0 0 0 0 0 0 0 1 1 0 0 0
0 0 0 0 0 1
OP PP PP PP QP
4 5 6 , Global Numbers
k L = L
1
d k L i
3
0.667 0.444 0.667 0 0 . 333333 333.333 0 0 0 . 0 1333 0.667 0 0.667 0 0.667 0.444 0 0.667 333333 . 0 0 333.333 0 . 0 0.667 0.667 0 1333
OP PP PP PP PQ
1 2 3 4 5 6
For element No. 2, local and global coordinates are the same, and le = 4m, Ie = 2Io
265
= k
0 0 0 250 0.750 0 0.375 0.750 0.750 2.0 0 10 0.75 . 0 0 250 0 0 0 0.375 0.75 0.375 0.75 0.75 10 0 2.0 0.75 . 0 0.375
OP PP PP PP QP
, Global Numbers
4 5 6 7 8 9
For element No. 3, if we take it as member 43, we get the stiffness matrix identical to element 1. However nodal displacement vector for it is [ 10
11 12 7 8 9 ]. Hence
7 8 9
10
11
12
0 0 0.667 0.444 0.667 10 11 333.333 0 0 0 333.333 0 1333 0.667 0 0.667 12 . 0 0.667 0.444 0 0.667 7 8 0 0 333333 0 333.333 . 0 0.667 0.667 0 1333 9 .
OP PP PP PP PQ
, Global Numbers
{F}T = [0 0 0 30.0 40 26.667 0 40 26.667 0 0 0]. In case of elimination method of imposing boundary conditions, we get reduced load vector corresponding to the directions 4, 5, 6, 7, 8 and 9 as
lFq
266
b 20g 4 2
2
40
20
GH 4 JK 2
40
4 b 20g 12
2
3 26.667
4 b 20g 12
2
26.667
Fig. 14.11
(iv) The system equations may be solved to get the nodal displacement vector. (v) The member forces may be calculated as usual. The tensile force as positive is given by,
P=
EA 4 1 le
g R | b1 + 3 gl | S | | T
e 2 3 5 6
(14.26)
M=
EI
2 le
1 3 le
U | | V | | W
(14.27)
V=
EI
3 le
12 6le
12 6le
R | | S | | T
2 3 5 6
U | | V | | W
(14.28)
(vi) The end reaction vector, +ve in the +ve direction of degrees of freedom is obtained by the expression
lRq = k l q lFq
(14.29)
lq
= 1 2
= u1 v1
3 ... 12
w1 x1 y1 z1 u2 v2 w2 x 2 y 2 z 2
(14.30)
267
y1
y2
y x z
v1 u1 w1 i
x1
v2 u2 w2 j
x2
xm
z1
z2
zm
zm
Fig. 14.12
The 12 12 stiffness matrix in local coordinate system, may be written down as shown in equation 14.31 by carefully noting the contribution of each nodal displacement.
y ym xm ym y
zm
(b)
(a)
zm
Fig. 14.13
(a) Orientation of beam element (b) Orientation of principle axes about xm-axis
268
LM EA MM l MM 0 MM 0 MM MM 0 MM 0 MM MM 0 = M EA MM l MM MM 0 MM 0 MM MM 0 MM 0 MM MM 0 N
e e
0 12 EI z
3 le
0 0 12 EI y
3 le
0 0 0 GI x l 0 0 0 0 0 GI x le 0 0
0 0 6 EI y
2 le
0 6 EI z
2 le
EA le 0 0 0 0 0
0 12 EI z
3 le
0 0 12 EI y
3 le
0 0 0 GI x le 0 0 0 0 0 GI x le 0 0
0 0 6 EI y
2 le
0 6 EI z
2 le
0 0 0 6 EI z
2 le
0 0 0 4 EI z le 0 6 EI z
2 le
0 0 0 6EI z l2 0 12 EI y
3 le
0 0 0 2 EI z le 0 6 EI z
2 le
0 6 EI y le2 0 0 0 12 EI y
3 le
0 4 EI y le 0 0 0 6 EI y
2 le
0 6 EI y
2 le
0 2 EI y le 0 0 0 6 EI y
2 le
0 0 0 12 EI y
3 le
0 12 EI z
3 le
EA le 0 0 0 0 0
0 0 0 6 EI z
2 le
0 0 0 2 EI z le
0 0 0 6EI z
2 le
0 0 0 4 EI z le
0 6 EI y le2 0
0 2 EI y le 0
0 6EI y
2 le
0 4 EI y le 0
OP PP PP PP PP PP PP PP PP PP PP PP PP PP PP PP PP PP PQ
(14.31) If the member axes do not coincide with global axes, we need transformation matrix. Figure 14.13 (a) shows arbitrary orientation of the member axis in global system and Fig. 14.13 (b) shows the orientation of principal axes of the member. Then the transformation matrix T works out to be
LM L 0 =M MM 0 NM 0
0 L 0 0
0 0 L 0
0 0 0 L
OP PP PP Q
(14.32)
where L is a 12 12 matrix and each sub matrix on right hand side is a 3 3 matrix. [L ] matrix is given by
LM MM M C C =M MM MM C C MN
x x
Cx y cos C z cos
2 2 Cx + C y y
Cy
2 Cx + Cz2 cos
sin Cz cos
2 Cx
Cz2
2 Cx + Cz2 sin
(14.33)
269
Cx =
x j xi le
, Cy =
2
y j yi le
2 j i
and Cz =
z j zi le
2 i
le =
dx
xi
i + d y y i + dz z i
j
For the derivation of above transformation matrix reorders may refer to Krishnamurthy[1]
Fig. 14.14
xz = G xz
and the shear force as
(14.34)
Q = xz dz b
where xz is the shear strain at distance z from neutral axis at section x.
(14.35)
However the use of general equations 14.34 and 14.35 complicates the problem and a simplified approach is possible. In order to account for non-uniform stress distribution at a cross section while still retaining one dimensional approach, the equations 14.34 and 14.35 are modified using a shear correction factor as follows:
270
xz = G x
and
Q = xz A = AG x
(14.36) (14.37)
The shear correction factor ' ' is a function of the cross sectional shape and Poisson ratio . The term ' A' is the shear area of the section associated with shear and may be denoted as As. Thus
As = A where < 1
(14.38)
Values of for various cross sectional shapes are given in the solid mechanics books by DYM C.L. & Shames I.H. [2] and Ugural A.C. and Fenster S.K. [3]. The value of for a rectangular section is 5/6. Hence,
Q G As
(14.39)
Timoshenko beam theory averages the effect of shear strain over the cross section i.e. it takes,
w x
(14.40)
where is the angle through which the face of the cross section rotates after deformation as shown in Fig. 14.15.
w x
Before deformation
After deformation
Fig. 14.15
M = EI
d dx
M2 EI dx = 2 EI 2
z
le 0
F d I G dx J H K
271
z z
1 xy 2
dV
1 dV = G As 2
1 G 2 2
z
le 0
F w I G x J H K F w I G x J H K
dx
z
le 0
F d I G dx J H K
1 dx + GAs 2
z
le 0
dx
(14.41)
= N1 1 + N 2 2
RwU = LM N S V N 0 TW
0 N1
N2 0
Rw U 0 O | | S V PQ |w | N | | | | T W
1 1 2 2 2
(14.42)
where N1 and N2 are interpolation functions i.e. N1 = 1 Using isoparametric concept, we have
x=
x x and N2 = le le
N x
i
(14.43)
Now,
d = dx
i=1
Ni i x
w = x
(14.44)
Ni i
i =1
i =1
Ni wi x
(14.45)
Strain vector
lq = R U S V TW
272
where is curvature =
d dx dN1 dx N1 0 dN 2 dx dN2 dx x
L R U = MM 0 l q = S V T W MM dN N dx
OP PP = l1 LM0 1 PQ N
e
le x 1
OP xQ
=
Now M = EI For rectangular section I =
D11 =
RM U = D lq SQ V T W
1 bh 3 12
E 3 bh 12
and
Q = As
xy = As
h = A
LE RM U = MM12 bh l q = S V T Q W MM 0 N
D = Ebh h 2 12 0
b g O 0 P P R U = Ebh LMMh0 S V 12 N bh P T W EP 2 Q
E bh E = 21+ 2
0 6
OP R U TW QP S V
(14.48)
= e
z
le 0
LM NM
OP 6 Q P
0
D B dx
z
le
1 0 1 dx le x 1 x
OP Q
2 e 0
z
le
0 6
h2 dx 6x
OP PQ
273
Ebh
2 12 le 0
z
le
LM 6 MM6 bl xg 6 MM 6 x N
e
OP 6 bl x g h + 6xbl x gP h + 6 bl x g PP dx 6 bl x g 6x 6 h + 6x h + 6xbl x g 6x PQ
6 le x
e 2
6
e
6x
LM Ebh M = 12 l M MM N
2 e
z
le 0
0 h 0 h 2
2
1 le x 0 0 2 2 le x le x 0 h + 6 1 le x 0 0 2 x x le x 0 h
OP PP PQ
LM MM MM N
i.e.,
be
(14.49)
where [ ] be and [ ] se are the contributions of bending and shear to the total stiffness. The integrations can be performed to get [ ] be and [ ] se as shown below:
se
se
le very large), h the results obtained by this formulation are not correct. The shear term, which should tend to zero in such cases, do not tend to zero. This is called shear locking. The shear stiffness is increasingly constrained. This is called spurious constraint. There are two popular remedies for the elimination of this type of errors: (i) Reduced Integration Technique (ii) Using field consistency element
The above formulation gives good results for moderately thick beams. For thin beams ( (i) Reduced Integration Technique: If shear stiffness in equation 14.49 is integrated with one point Gaussian technique, we get
0 1 0 1
OP PP PQ
le 2 le 2 2 le 6 l e 2 2 le 3
(14.50)
1 le 2
OP PP PP PP PP PQ
(14.51)
274
se
LM 1 MM l Gh M 2 M = l M 1 MM MM l N2
e e e
le 2 2 le 4 l e 2 2 le 4
1 le 2 1 le 2
le 2 2 le 4 l e 2 2 le 4
OP PP PP PP PP PQ
(14.52)
since and
x=
f d = 2 f = 0
bg
1 1+ , dx = d 2 2
1 3
2 se
LM 1 MM l Gh M 2 M = l M 1 MM MM l N2
e e e
le 2 2 le 3 le 2 2 le 6
1 le 2
1 le 2
le 2 2 le 6 l e 2 2 le 3
OP PP PP PP PP PQ
(14.53)
Now consider the analysis of a cantilever beam with single element (refer Fig. 14.16). The stiffness matrix of the beam is
k = k
= k
be
+ k
se
P M b=1
Fig. 14.16
275
Eh 3 Eh 3 and = 12 le 12 l
LM MM MM 2l MM MM MM 2l N
i.e.,
l 2
l 2
l2 4
l 2 l2 4
l 2
OP R 0 U R 0 U P| | | | | | | | l P | | | | 0 + P 0 4 P | | | | l PS V=S V P |w | | P | 2 PP | | | | | | | | l + P | | | M | | | | | 4 QP T W T W
l 2
2 2 2 2
LM MM l MN 2
Solving the above equation, we get
OP PP R U = R M U Sw V S P V l + P T W T W 4 Q
l 2
2 2 2
w2 =
F l + 1I P + 1 M G 4 J 2 H K
2
(14.54)
and
l P+ M 2 = 2
In case of thin beam, . Hence equation 14.54 reduces to
w2 = 1 lP + M 2 2
(14.55)
and 2 remains same as equation 14.55. Thus the beam deformation is solely due to bending as shown by equation 14.55 and 14.56. Hence for thin beams one point integration gives correct results. If two point quadrature is used for integration, after applying the boundary conditions we get the equations as
LM N
OP Q
(14.56)
LM MM l MN 2
OP PP R U = R M U Sw V S P V l + P T W T W 3 Q
l 2
3 2 2
276
l2 lM 3 w2 = P+ 2 l l2 2 + + 12 12
F G H
I J K
F G H
I J K
and
l P 2 2 = l2 + 12 M +
In case of thin beams . Hence we get
w2 =
4P + 6
M l
(14.57)
and
2 =
6 l P + 2M l
2
(14.58)
The above two equations show that the free end deformation depends on the coefficient correspongind to shear deformation, which is not true in case of thin beams. Hence two point integration (equations 14.57 and 14.58) lead to erroneous results. Thus the reduced integration (1 point Gaussian integration instead of 2 point Gaussian integration) gives better results than the exact integration. The reduced integration technique is used to get good results for thin beams. (ii) Field Consistant Element Formulation: Let us first see how the element formulation is field inconsistent in the case of very thin beams. In the formulation we have taken,
w = N1
and
N2 N2
= N1
N1 =
Rw U = N w + N w Sw V T W R U = N + N S V T W
1 1 1 2 2 1 2 1 1 2
where
1 1+ and N 2 = 2 2
2x 1 le
w=
1 1+ w + w2 w w1 + 2 = a1 + a2 w1 + w2 = 1 2 2 2 2
277
a1 =
w1 + w2 w2 w1 and a2 = 2 2
= b1 + b2
b1 =
1 + 2 1 and b2 = 2 2 2
=
and shear strain
d d d 2 = = b2 dx d dx l
dw d dw 2 = b1 + b2 = b1 + b2 a2 dx d dx l
Ub =
and strain energy due to shear is
Us =
EI 2
F d I G dx J H K
2
dx =
EI 2
F2b I Gl J H K
2 2
dx
(14.59)
GAs 2
bg
dx =
GAs 2
F b 2 a + b I G l J H K
1 2
dx
(14.60)
As the thickness approaches zero, the shear strain energy should vanish and bending strain energy should exist. i.e.,
b1
2 a2 + b2 0 l 2 a2 0 l
(14.61) (14.62)
i.e., and
b1
b2 0
The terms corresponding to condition 14.61 correspond to both the strain fields flexure and shear. Hence it is called field consistant term. The constraint corresponding to equation 14.62 contains the term corresponding only to flexure field. If b2 0 , the strain energy due to bending tends to zero, which should not happen. Hence this constraint requirement in the limiting case is spurious and it is this requirement which causes shear locking. To get rid of this situation in the limiting case, the function smoothening is required i.e. alter the terms to over come this situation. This is achieved by making b2 = 0 in the shear strain field i.e. by taking
= b1 =
Thus,
= b1 =
1 + 2 1 = 1 2
LM N
1 2
OP R U = Q S V T W
1 2
N1
N2
R U S V T W
1 2
278
LM 1 1 OP . This smoothening is N2 2 Q required not only in the shear strain field but even in the shear stress field also.
N2 =
The greatest advantage of Timoshenko beam element is C1 continuity problem is reduced to Co continuity problem. The extension of this concept by Mindlin to plate bending problems, (to be discussed in next chapter) is a great achievement in the plate analysis.
QUESTIONS
1. Derive the stiffness matrix for a beam element. 2. Analyse the beam shown in Fig. 14.17 using FEM technique. Determine the rotations at the supports. Given E = 200GPa and I = 4 106 mm4.
20 kN/m
A
6m
B
4m
Fig. 14.17
(Ans. A = 0.3, B = 0.375; C = 01625 ) . 3. Assemble the stiffness matrix for a plane beam element oriented at angle to the x-axis. Explain its use in FEA. 4. Assemble element stiffness matrix for the member of plane frame shown in Fig. 14.18, if it is oriented at angle 30o to the x-axis. Take E = 200GPa, I = 4 10-6 m4 and A = 4 103 m2.
y
5m
30
Fig. 14.18
LM150.025 MM M (Ans.) M MM NM
. . 86.558 012 150.015 086.558 012 50.070 0.208 86.558 50.070 0.208 . 0.208 0.8 012 0.4 . SYM 150.015 86.558 012 50.070 0.208 0.8
OP PP PP PP QP
279
5. Explain Timoshenko Beam Theory. Discuss its advantages and disadvantages. 6. Explain the term Timoshenko Beam Theory and briefly explain the stiffness formulation for such element. 7. Explain the term shear locking as used in Timoshenko Beam Theory. How this problem is overcome? 8. Taking the example of a single element cantilever beam, show that single point Gaussian integration overcomes the problem of shear locking. 9. Explain what is meant by field consistant formulation to overcome shear locking in Timoshenko Beam Theory. Use the example of a single element cantilever beam.
References
1. Krishnamoorthy C.S., Finite Element AnalysisTheory and Programming, Tata McGraw-Hill Publishing Co. Ltd., New Delhi. 2. DYM C.L. and Shames I.H., Solid Mechanics, International Students Edition, McGraw-Hill Kogakusha Ltd., 1973. 3. Ugural A.C. and Fenster S.K., Advanced Strength and Applied Elasticity, American Elsevier Publishing Co., New York, 1975.
280
15
Bending of Thin Plates
15.1 INTRODUCTION
Plate is a flat surface having considerably large dimensions as compared to its thickness. Slabs in civil engineering structures, bearing plates under columns, many parts of mechanical components are the common examples of plates. In this chapter, we are considering bending of such plates under lateral loads. The bending properties of a plate depend greatly on its thickness. Hence in classical theory we have the following groups (i) thin plates with small deflections (ii) thin plates with large deflections, and (iii) thick plates In thin plates with small deflections theory, the following assumption are made (a) There is no deformation in the middle plane of the plate. This plane remains neutral during bending. (b) Points of the plate lying initially on a normal to the middle surface of the plate remain on the normal to the same surface even after bending. (c) The normal stresses in the direction transverse to the plate are negligible. This theory is satisfactory for plates with ratio of thickness to span exceeding deflection to thickness less then
1 and the ratio of maximum 10
1 . Many engineering problems lie in the above category and satisfactory 5 results are obtained by classical theories of thin plates. Stresses in the middle plane are negligible, if the deflections are small in comparison with thickness. If the deflections are large, the in plane stresses developed in the so called neutral surfaces are to be considered. This gives rise to theory of thin plates with large deflections, in which geometric non-linearity is incorporated.
1 th the assumption (a) and (b) listed under theory of 10 thin plates will not hold good. Such plates need three dimensional analysis. Theory developed for the analysis of such plates may be called as thick plate theory. In this chapter, analysis of thin plates with small deflections with finite element approach is explained. Since many stress-strain-displacement relation in the theory of thin plates with small deflection are used, the limitations of classical theory remains in this finite element analysis also. The advantage of finite element analysis is that it can handle the structures with different end conditions and shapes easily.
If the plate has thickness to span ratio less than