Beruflich Dokumente
Kultur Dokumente
Amit H. Varma
Professor
School of Civil Engineering
Purdue University
Ph. No. (765) 496 3419
Email: ahvarma@purdue.edu
OUTLINE
Definition of stability
Types of instability
Methods of stability analyses
Examples – small deflection analyses
Examples – large deflection analyses
Examples – imperfect systems
Design of steel structures
1
8/25/2014
STABILITY DEFINITION
STABILITY DEFINITION
2
8/25/2014
STABILITY DEFINITION
3
8/25/2014
BUCKLING
P P P
d
4
8/25/2014
TYPES OF INSTABILITY
TYPES OF INSTABILITY
BIFURCATION BUCKLING
Member or structure subjected to loads. As the load is
increased, it reaches a critical value where:
The deformation changes suddenly from state-1 to state-2.
And, the equilibrium load-deformation path bifurcates.
Critical buckling load when the load-deformation path bifurcates
Primary load-deformation path before buckling
Secondary load-deformation path post buckling
Is the post-buckling path stable or unstable?
5
8/25/2014
SYMMETRIC BIFURCATION
ASYMMETRIC BIFURCATION
6
8/25/2014
INSTABILITY FAILURE
INSTABILITY FAILURE
FAILURE OF BEAM-COLUMNS
P M K=0
M
K<0
M
d
P
No bifurcation.
Instability due to material
and geometric nonlinearity
7
8/25/2014
INSTABILITY FAILURE
Snap-through buckling
P
Snap-through
INSTABILITY FAILURE
8
8/25/2014
OUTLINE
Definition of stability
Types of instability
Methods of stability analyses
Examples – small deflection analyses
Examples – large deflection analyses
Examples – imperfect systems
Design of steel structures
9
8/25/2014
STABILITY ANALYSES
Each method has its advantages and disadvantages. In fact,
you can use different methods to answer different questions
The bifurcation approach is appropriate for determining the
critical buckling load for a (perfect) system subjected to loads.
The deformations are usually assumed to be small.
The system must not have any imperfections.
It cannot provide any information regarding the post-buckling load-
deformation path.
The energy approach is the best when establishing the
equilibrium equation and examining its stability
The deformations can be small or large.
The system can have imperfections.
It provides information regarding the post-buckling path if large
deformations are assumed
The major limitation is that it requires the assumption of the
deformation state, and it should include all possible degrees of
freedom.
STABILITY ANALYSIS
The dynamic method is very powerful, but we will not use it in this class
at all.
Remember, it though when you take the course in dynamics or earthquake
engineering
In this class, you will learn that the loads acting on a structure change its
stiffness. This is significant – you have not seen it before.
Ma
4E I 2E I
P qa Ma qa Mb qb
Mb
L L
10
8/25/2014
STABILITY ANALYSIS
FOR ANY KIND OF BUCKLING OR STABILITY ANALYSIS –
NEED TO DRAW THE FREE BODY DIAGRAM OF THE DEFORMED
STRUCTURE.
WRITE THE EQUATION OF STATIC EQUILIBRIUM IN THE
DEFORMED STATE
WRITE THE ENERGY EQUATION IN THE DEFORMED STATE
TOO.
THIS IS CENTRAL TO THE TOPIC OF STABILITY ANALYSIS
NO STABILITY ANALYSIS CAN BE PERFORMED IF THE FREE
BODY DIAGRAM IS IN THE UNDEFORMED STATE
BIFURCATION ANALYSIS
Always a small deflection analysis
To determine Pcr buckling load
Need to assume buckled shape (state 2) to calculate
Example 1 – Rigid bar supported by rotational spring
L
Rotationally restrained at end
L P
kq
q
L cosq
L (1-cosq)
11
8/25/2014
BIFURCATION ANALYSIS
L P
kq L sinq
q
L cosq
L (1-cosq)
BIFURCATION ANALYSIS
k
L
L cosq
L (1-cosq)
12
8/25/2014
BIFURCATION ANALYSIS
Write equations of static equilibrium in deformed state
P
L
L sinq
q
O
k L sinq
L cosq
L (1-cosq)
BIFURCATION ANALYSIS
Example 3 – Three rigid bar system with two rotational springs
P k k P
A D
B C
L L L
P k k P
q1 q2
A L sin q2 D
L sin q1 (q1 – q2) L
L
C
B (q1 – q2)
13
8/25/2014
BIFURCATION ANALYSIS
Write equations of static equilibrium in deformed state
P k k P
q1 q2
A L sin q2 D
L sin q1 (q1 – q2) L
L
C
B (q1 – q2)
k P
P q2
q2(q1 – q2) L sin q2 D
L
A q1 C
L sin q1
L k(2q2-q1)
B q1+(q1-q2)
k(2q1-q2)
BIFURCATION ANALYSIS
Equations of Static Equilibrium
k (2q1 q 2 ) P L q1 0 2k PL k q1 0
k (2q 2 q1 ) P L q 2 0 k 2k PL q 2 0
Therefore either q1 and q2 are equal to zero or the determinant of the
coefficient matrix is equal to zero.
When q1 and q2 are not equal to zero – that is when buckling occurs –
the coefficient matrix determinant has to be equal to zero for equil.
Take a look at the matrix equation. It is of the form [A] {x}={0}. It can
also be rewritten as ([K]-l[I]){x}={0}
2k k
1 0 q 0
L L P
0 1 q 0
1
k 2k 2
L L
14
8/25/2014
BIFURCATION ANALYSIS
This is the classical eigenvalue problem. ([K]-l[I]){x}={0}.
We are searching for the eigenvalues (l) of the stiffness matrix [K].
These eigenvalues cause the stiffness matrix to become singular
Singular stiffness matrix means that it has a zero value, which means that
the determinant of the matrix is equal to zero.
2k PL k
0
k 2k PL
(2k PL) 2 k 2 0
(2k PL k ) (2k PL k ) 0
(3k PL) (k PL) 0
3k k
Pcr or
L L
BIFURCATION ANALYSIS
Each eigenvalue or critical buckling load (P cr) corresponds to a buckling shape
that can be determined as follows
Pcr=k/L. Therefore substitute in the equations to determine q1 and q2
k (2q1 q 2 ) P L q1 0 k (2q 2 q1 ) P L q 2 0
Let P Pcr k Let P Pcr k
L L
k (2q1 q 2 ) kq1 0 k (2q 2 q1 ) kq 2 0
kq1 kq 2 0 kq1 kq 2 0
q1 q 2 q1 q 2
All we could find is the relationship between q1 and q2. Not their specific
values. Remember that this is a small deflection analysis. So, the values are
negligible. What we have found is the buckling shape – not its magnitude.
The buckling mode is such that q1=q2 Symmetric buckling mode
P k k P
A q1 q2=q1 D
L
L
B C
15
8/25/2014
BIFURCATION ANALYSIS
Second eigenvalue was Pcr=3k/L. Therefore substitute in the equations to
determine q1 and q2
k (2q1 q 2 ) P L q1 0 k (2q 2 q1 ) P L q 2 0
Let P Pcr 3k Let P Pcr 3k
L L
k (2q1 q 2 ) 3kq1 0 k (2q 2 q1 ) 3kq 2 0
kq1 kq 2 0 kq1 kq 2 0
q1 q 2 q1 q 2
All we could find is the relationship between q1 and q2. Not their specific
values. Remember that this is a small deflection analysis. So, the values are
negligible. What we have found is the buckling shape – not its magnitude.
The buckling mode is such that q1=-q2 Antisymmetric
C
buckling mode
L
P k k q2=-q1 P
A q1 D
L
OUTLINE
Definition of stability
Types of instability
Methods of stability analyses
Bifurcation analysis examples – small deflection analyses
Energy method
Examples – small deflection analyses
Examples – large deflection analyses
Examples – imperfect systems
Design of steel structures
16
8/25/2014
ENERGY METHOD
ENERGY METHD
The energy method is the best for establishing the equilibrium
equation and examining its stability
The deformations can be small or large.
The system can have imperfections.
It provides information regarding the post-buckling path if large
deformations are assumed
The major limitation is that it requires the assumption of the
deformation state, and it should include all possible degrees of
freedom.
17
8/25/2014
ENERGY METHOD
Example 1 – Rigid bar supported by rotational spring
Assume small deflection theory
k P Rigid bar subjected to axial force P
L
Rotationally restrained at end
L P
kq
q
L cosq
L (1-cosq)
kq L sinq
q
L cosq
L (1-cosq)
Write the equation representing the total potential energy of system
U We
1
U kq2
2
We P L (1 cosq )
1
k q 2 P L (1 cosq )
2
d
k q P L sin q
dq
d
For equilibrium; 0
dq
Therefore, k q P L sin q 0
For small deflections; kq P Lq 0
k
Therefore, Pcr
L
18
8/25/2014
1
k q 2 P L (1 cosq ) d2
2 When P Pcr 0 Stableequilibrium
d dq 2
k q P L sin q k q P L q d2
dq When P Pcr 0 Unstableequilibrium
d
2
dq 2
k PL d2
dq 2 When P Pcr 0 Not sure
dq 2
Stable
19
8/25/2014
U We
1
U kq2 L P
2
We P L (1 cosq ) kq
q L sinq
1
k q P L (1 cosq )
2
2 L cosq
d L (1-cosq)
k q P L sin q
dq
d
For equilibrium; 0
dq
Therefore, k q P L sin q 0
kq
Therefore, P for equilibrium
L sin q
The post buck lingP q relationship is given above
1 kq
P for equilibrium
L sin q
0.8 P q
Pcr sin q
Load P/Pcr
0.6
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00
20
8/25/2014
d 1 d2 1 d3 1 d4 1 dn
q 0 q q2 q3 q 4 ..... qn
dq q 0 2! dq q 0
2
3! dq q 0
3
4! dq q 0
4
n! dq n q 0
Since the first non-zero term is > 0, the state is stable at P=Pcr and q=0
21
8/25/2014
1
STABLE
STABLE
0.8
Load P/Pcr
0.6
STABLE
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00
k q0 L P
L cos(q0)
k(qq0) L sinq
q
q0
L cosq
L (cosq0-cosq)
22
8/25/2014
k(qq0) L sinq
q
q0
U We
1 L cosq
U k (q q 0 ) 2 L (cosq0-cosq)
2
We P L (cosq 0 cosq )
1
k (q q 0 ) 2 P L (cosq 0 cosq )
2
d
k (q q 0 ) P L sin q
dq
d
For equilibrium; 0
dq
Therefore, k (q q 0 ) P L sin q 0
k (q q 0 )
Therefore, P for equilibrium
L sin q
The equilibrium P q relationship is given above
0.8
Load P/Pcr
0.6
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00 q00.05 q00.1 q00.2 q00.3
23
8/25/2014
2
d
k (q q 0 ) P L sin q
dq
d2
k P L cosq
dq 2
Equilibrium path will be stable
d2
if 0
dq 2
i.e., if k P L cosq 0
k
i.e., if P
L cosq
k (q q 0 ) k
i.e., if
L sin q L cosq
i.e., q q 0 tanq
Which is always true, hence always in STABLE EQUILIBRIUM
24
8/25/2014
k
L
L cosq
L (1-cosq)
U We P
L
1 1
U k ( L sin q ) 2 k L2q 2 L sinq
2 2 q
O
We P L (1 cosq ) k L sinq
1
k L2 q 2 P L (1 cosq )
2 L cosq
d
k L2 q P L sin q
dq
L (1-cosq)
d
For equilibrium; 0
dq
Therefore, k L2 q P L sin q 0
For small deflections; k L2q P Lq 0
Therefore, Pcr k L
25
8/25/2014
L sinq
1
U k ( L sin q ) 2 q
2 O
We P L (1 cosq )
1
k L2 sin 2 q P L (1 cosq ) L cosq
2
d L (1-cosq)
k L2 sin q cosq P L sin q
dq
d
For equilibrium; 0
dq
Therefore, k L2 sin q cosq P L sin q 0
Therefore, P k L cosq for equilibrium
The post buck lingP q relationship is given above
26
8/25/2014
0.8
Load P/Pcr
0.6
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00
27
8/25/2014
d 1 d2 1 d3 1 d4 1 dn
q 0 q q2 q3 q 4 ..... qn
dq q 0 2! dq q 0
2
3! dq q 0
3
4! dq q 0
4
n! dq n q 0
1
k L2 sin 2 q P L (1 cosq ) 0 d4
2 4k L2 cos 2q P L cosq
dq 4
d 1
k L2 sin 2q P L sin q 0 d4
dq 2 4k L2 k L2 3k L2
dq 4
d2
k L2 cos 2q P L cosq 0 d4
dq 2 0
dq 4
d3
2k L2 sin 2q P L sin q 0 UNSTABLE at q 0 when buck lingoccurs
dq 3
Since the first non-zero term is < 0, the state is unstable at P=Pcr and q=0
1 UNSTABLE
UNSTABLE
0.8
Load P/Pcr
0.6
UNSTABLE
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00
28
8/25/2014
k
L cos(q0)
P
L
L sinq
q
L sinq0
O
q0
L cosq
L (cosq0-cosq)
L sinq
q
L sinq0
O
q0
U We
1
U k L2 (sin q sin q 0 ) 2
2 L cosq
1
k L2 (sin q sin q 0 ) 2 P L (cosq 0 cosq )
2
d
k L2 (sin q sin q 0 ) cosq P L sin q
dq
d
For equilibrium; 0
dq
Therefore, k L2 (sin q sin q 0 ) cosq P L sin q 0
sin q 0
Therefore, P k L cosq (1 ) for equilibrium
sin q
The equilibrium P q relationship is given above
29
8/25/2014
0.8
Load P/Pcr
0.6
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
End rotation q
q00 q00.05 q00.1 q00.2 q00.3
30
8/25/2014
31
8/25/2014
32
8/25/2014
dA A; x dA I y ; y dA I x
2 2
A A A
33
8/25/2014
s
Vy
t
Ix y t ds
O
Vx s
I y O
t x t ds
q y
dz 2
E I x f y q y
34
8/25/2014
E I x f y q y
v
fy
1 (v) 2 3/ 2
35
8/25/2014
MZ=MSV + MW
Where,
MSV = G KT f′ and MW = - E Iw f"‘
MSV = Pure or Saint Venant’s torsion moment
KT = J = Torsional constant =
f is the angle of twist of the cross-section. It is a function of z.
IW is the warping moment of inertia of the cross-section. This is
a new cross-sectional property you may not have seen before.
MZ = G KT f′ - E Iw f"‘ ……… (3), differential equation of torsion
dz r df
df
r r f
dz
G r f
M SV r dA G f r 2 dA
A A
M SV G K T f
where, K T J r 2 dA
A
36
8/25/2014
SV G r f
( SV )max G t f
sv
Warping deformations
37
8/25/2014
dM Z
mZ
dz
G K T f E I w f iv m Z
G KT m mz z 2
f iv f Z f C 4 C5 z C 6 coshlz C 7 sinh lz
E IW E IW 2 G KT
mZ
f iv l2 f
E IW
The coefficients C1 .... C6 can be obtained using end conditions
38
8/25/2014
39
8/25/2014
Torsional Stresses
Torsional Stresses
40
8/25/2014
41
8/25/2014
E I x v M x (1)
E I y u M y (2)
G K T f E I W f M z (3)
NOTES:
(1) Three uncoupled differential equations
(2) Elastic material – first order force-deformation theory
(3) Small deflections only
(4) Assumes – no influence of one force on other deformations
(5) Equations of equilibrium in the undeformed state.
42
8/25/2014
43
8/25/2014
Displacements of Q are:
uQ = u + a f sin a
vQ = v – a f cos a
where a is the distance from Q to S
But, sin a = (y0-y) / a
cos a = (x0-x) / a
Therefore, displacements of Q are:
uQ = u + f (y0-y)
vQ = v – f (x0 – x)
Displacements of centroid C are:
uc = u + f (y0)
vc = v - f (x0)
44
8/25/2014
M x M BX
z
(M TX M BX ) P(v f x0 )
L
M y M BY (M TY M BY ) P(u f y0 )
z
L
45
8/25/2014
uc
vc
x
x h
MBx P
Rx
Ry
MBY
x uc
vc
x
z z
y
h
46
8/25/2014
Mς Mξ σ+d
σ
a
Mη
σ
z
Mx
Rx
My Ry P
M x M BX
z
(M TX M BX ) P(v f x0 )
L
(M TY M BY ) P(u f y0 )
z
M y M BY
L
M x M BX
z
(M TX M BX ) P v f P x0 M BY z (M TY M BY )
L L
M h M BY
z
(M TY M BY ) P u f P y0 M BX z (M TX M BX )
L L
47
8/25/2014
Twisting component – 1 of 4
v
u
48
8/25/2014
Twisting component – 2 of 4
u v
Twisting component – 2 of 4
49
8/25/2014
Twisting component – 3 of 4
Twisting component – 4 of 4
Wagner’s effect or contribution
– complicated.
Two cross-sections that are dz
apart will warp with respect to
each other.
The stress element dA will
become inclined by angle (a
df/dz) with respect to dz axis.
Twist produced by each stress
element about S is equal to
df
dMz 3 a( dA) a
dz
df
dz A
Mz 3 a 2 dA
50
8/25/2014
Twisting component – 4 of 4
Let , a 2 dA K
A
df
Mz 3 K
dz
df
Mz 3 K for small angles
dz
Twisting component – 4 of 4
Let , a 2 dA K
A
df
Mz 3 K
dz
df
Mz 3 K for small angles
dz
x
y
x
y
51
8/25/2014
M x M BX (M TX M BX ) P v f P x0 M BY (M TY M BY )
While, z z
L L
M h M BY
z
(M TY M BY ) P u f P y0 M BX z (M TX M BX )
L L
52
8/25/2014
z
x
53
8/25/2014
Therefore,
( M TX M BX ) P v f P x0 M BY ( M TY M BY )
z z
M x E I x v M BX
L L
z z
Mh E I y u M BY (M BX ) P u f P y0 M BX
MBY
M TX +M (M BY )
MBX
M TY +M
L TY L TX
z
M z G KT f E I w f ( M BX ( M BX M TX ) P y0 ) u
L
z v u
( M BY ( M BY M TY ) P x0 ) v (M TY M BY ) (M TX M BX ) K f
L L L
54
8/25/2014
1 E I x v P v f ( P x0 ) 0
2 E I y u P u f ( P y0 ) 0
3 E I w f (G KT K ) f u ( P y0 ) v ( P x0 ) 0
55
8/25/2014
K f a 2 f dA
A
where,
P M x y Mh x
E Wn f
A Ix Iy
M x P (v f x0 )
Mh P (u f y0 )
P P (v f x0 ) y P (u f y0 ) x
K f E Wn f f a 2 dA
A
A
Ix Iy
P P (v f x0 ) y P (u f y0 ) x
K f E Wn f f a 2 dA
A Ix Iy A
P
A A
Neglecting higher order terms; K f f a 2 dA
a 2 dA x02 y02 x 2 y 2 2 x0 x 2 y0 y dA
A A
a 2 dA x02 y02 dA x 2 dA y 2 dA 2 x0 x dA 2 y0 y dA
A A A A A A
a dA ( x y ) A I x I y
2 2
0
2
0
A
Finally,
P
K f ( x02 y02 ) A I x I y f
A
Ix Iy
K f P ( x02 y02 ) f
A
Ix Iy
Let r02 ( x02 y02 )
A
K f P r02 f
56
8/25/2014
1 E I x v P v f ( P x0 ) 0
2 E I y u P u f ( P y0 ) 0
3 E I w f ( P r02 G KT ) f u ( P y0 ) v (P x0 ) 0
Where
Ix I y
r02 x02 y02
A
3 E I w f iv ( P r02 G KT ) f 0
P P P r02 G KT
Let , Fv2 Fu2 Ff2
E Ix E Iy E Iw
57
8/25/2014
All three equations are similar and of the fourth order. The
solution will be of the form C1 sin lz + C2 cos lz + C3 z + C4
Need four boundary conditions to evaluate the constant C1..C4
For the simply supported case, the boundary conditions are:
u= u”=0; v= v”=0; f= f”=0
Lets solve one differential equation – the solution will be valid for
all three.
58
8/25/2014
Summary Py 2
L2
2 E Iw 1
Pf 2
G KT 2 3
L r0
These are, flexural buckling about the x and y axes and torsional
buckling about the z axis.
As you can see, the three buckling modes are uncoupled. You must
compute all three buckling load values.
The smallest of three buckling loads will govern the buckling of the
column.
59
8/25/2014
E Iy
2
Py 2
( K L)
2
y
2 E I 1
Pf w
G KT 2 3
( K z L )
2
r0
60
8/25/2014
rx
2 E I 1 2 E I A
Pf w
G KT 2 w
G KT rx2 2
( z ) x ( x y)
I
2 2
K L r0 K L
r I
z rx
Y Kx
Y
Kx
rx rx rx
E A (ry / rx ) E (ry / rx )
2 2 2 2
Py 791.02
PY L
2
A Y L
2
L
2
Ky Y K y
rx rx rx
2
Pf E Iw 2 A 1
G K r
rx ( I x I y ) A Y
2 T x 2
PY
K L
z rx
Pf 2 E I w 1
G KT rx2 2
x ( x y) Y
I
2
K L
PY r I
z rx
Pf 578.26
2
0.2333
PY L
rx
61
8/25/2014
1.6
1.4
1.2
Yield load PY
Cannot be exceeded
1
0.8
62
8/25/2014
The first equation for flexural buckling about the x-axis (axis of
non-symmetry) becomes uncoupled.
E I x v P v 0 (1)
Equations (2) and (3) are still
E I x v P v 0
iv
coupled in terms of u and f.
v iv Fv 2 v 0
P 2 E I y u P u f ( P y0 ) 0
where, Fv 2
E Ix 3 E I w f ( P r02 G KT ) f u ( P y0 ) 0
v C1 sin Fv z C2 cos Fv z C3 z C4
Boundary conditions
These equations will be satisfied by
sin Fv L 0
the solutions of the form
2 E Ix u=C2 sin (z/L) and f=C3 sin (z/L)
Px
( K x Lx ) 2
Buckling mod v C1 sin Fv z
63
8/25/2014
E I y u iv P u f ( P y0 ) 0
E I w f iv ( P r02 G KT ) f u ( P y0 ) 0
z z
Let , u C2 sin ; f C3 sin
L L
Therefore, substituting these in equations 2 and 3
z z z
4 2 2
2 E Iy 2 E Iw 1
Let , Py 2
and Pf 2
G KT 2
L L r0
Py P C2 P y0 C3 0
Pf P r02C3 P y0 C2 0
Py P P y0 C2
2 0
P y0 ( P C
f P ) r0 3
Py P P y0
0
P y0 ( Pf P) r02
64
8/25/2014
The critical buckling load will the lowest of Px and the two roots
shown on the previous slide.
If the flexural torsional buckling load govern, then the buckling
mode will be C2 sin (z/L) x C3 sin (z/L)
This buckling mode will include both flexural and torsional
deformations – hence flexural-torsional buckling mode.
65
8/25/2014
E I x v P v f ( P x0 ) 0 (1)
E I y u P u f ( P y0 ) 0 (2)
E I w f ( P r0 G KT ) f u ( P y0 ) v (P x0 ) 0
2
(3)
For simply supported boundary conditions: (u, u”, v, v”, f, f”=0), the
solutions to the differential equations can be assumed to be:
u = C1sin (z/L)
v = C2 sin (z/L)
f = C3 sin (z/L)
These solutions will satisfy the boundary conditions noted above
2
z z z
E I x C1 sin P C1 sin P x0 C3 sin 0
L L L L
z z z
2
2 z
E Ix P 0 P x0 C1 sin
L L
0
z
2
0 E Iy P P y0 C2 sin 0
L L
0
2 z
P x0 P y0 E I w ( P r0 G K T )
2 C cos
L L 3
L
66
8/25/2014
L L
3
where,
2 E Iw 1
2 2
Px EI x Py EI y Pf G KT 2
L L L2 r0
67
8/25/2014
A long topic
68
8/25/2014
L
Solutionbecomes
Fv2do z
v A sin(Fv z) Bcos(Fv z) sin
2 L
Fv
2
L
Geometric Imperfection
Solve for A and B P
Boundaryconditionsv(0) v(L) 0 PE z
v d sin
P o L
v(0) B 0 1
PE
v(L) A sin Fv L 0
Total Deflection
A0
P
Solutionbecomes
PE z z
Fv2do z v vo d sin do sin
v sin P o L L
1
2 L PE
Fv
2
L P
Fv2 P z 1 z
do E 1do sin do sin
2 P 1 P L P L
d 1
L z PE o z PE PE
v sin sin
Fv2 L 1 P L z
1 AFdo sin
2
PE L
L AF = amplification factor
69
8/25/2014
Geometric Imperfection
1
AF amplification factor
P
1
PE
M x P(v v o )
z
M x AF (Pdo sin )
L
i.e., M x AF (moment due to initialcrook ed)
12
Increases exponentially
10
Limit AF for design
Amplification Factor AF
0
0 0.2 0.4 0.6 0.8 1
P/PE
70
8/25/2014
71
8/25/2014
72
8/25/2014
73
8/25/2014
ES1 E t S2 0
y
where, S1 ( y y y1 ) dA
y y1
y y1
and S 2 ( y y1 y) dA
( d y )
M Pv
y y y1
M dU ( y y y1) dA d L ( y y1 y) dA
y y1 ( d y )
M Pv v ( EI1 E t I2 )
y
where, I1 ( y y y1 ) 2 dA
y y1
y y1
and I 2 ( y y1 y) 2 dA
( d y )
74
8/25/2014
M Pv v ( EI1 E t I2 )
Pv ( EI1 E t I2 )v 0
P
v v 0
EI1 E t I2
v Fv2v 0
P P
where, Fv2
EI1 E t I2 EIx
I1 I
and E E Et 2
Ix Ix
2 EI x E is the reduced or double modulus
PR 2 PR is the reduced modulus buckling load
(KL)
For 50 years, engineers were faced with the dilemma that the
reduced modulus theory is correct, but the experimental data
was closer to the tangent modulus theory. How to resolve?
Shanley eventually resolved this dilemma in 1947. He
conducted very careful experiments on small aluminum
columns.
He found that lateral deflection started very near the theoretical
tangent modulus load and the load capacity increased with
increasing lateral deflections.
The column axial load capacity never reached the calculated
reduced or double modulus load.
Shanley developed a column model to explain the observed
phenomenon
75
8/25/2014
76
8/25/2014
77
8/25/2014
78
8/25/2014
T T=PT/A
79
8/25/2014
T T
T f ( y h / 2) E T
M x ( T f ( y h / 2)E T ) ydA
A
M x T y dA E T f y 2 dA (fh / 2)E T ) y dA
A A A
M x 0 E T f Ix 0
M x E T Ixv
The equation Mx- PTv=0 becomes -ETIxv” - PTv=0
Solution is PT= ETIx/L
2 2
E 1.980E-03
2.178E-03
20
22
10099.8
10098.8
10099.5
10097.6
70.59690679
67.3048795
n1 2.376E-03 24 10094.2 10088.7 64.4113691
0.002 2.575E-03 26 10075.1 10054.2 61.77857434
1 nE 2.775E-03 28 10005.7 9934.0 59.17430952
0.2 0.2 2.979E-03 30 9779.8 9563.7 56.09208286
3.198E-03 32 9142.0 8602.6 51.5097656
E 3.458E-03
3.829E-03
34
36
7697.4
5394.2
6713.6
4251.9
44.14566415
34.1419685
E 4.483E-03 38 3056.9 2218.6 24.00464013
ET 5.826E-03 40 1488.8 1037.0 15.9961201
0.002
n1
8.771E-03 42 679.2 468.1 10.48827475
1 nE 1.529E-02 44 306.9 212.4 6.902516144
0.2 0.2
2.949E-02
5.967E-02
46
48
140.8
66.3
98.5
46.9
4.596633406
3.105440361
1.221E-01 50 32.1 23.0 2.129145204
80
8/25/2014
Ramberg-Osgood Stress-Strain
Stress-tangent modulus relationship
60
12000
50
10000
30 6000
4000
20
2000
10
0
0 10 20 30 40 50
0
0.000 0.010 0.020 0.030 0.040 0.050 Stress (ksi)
10 99.84137641
50 L2
12 91.1422898
14 84.3813604 PT 2 E T Ix 2 ET
16 78.93150275 T
(KL / r)
2 2
18 74.41710153
40 A AL
20 70.59690679
22 67.3048795 2 ET
(KL / r)cr
30
24 64.4113691
26 61.77857434 T
28 59.17430952 20
30 56.09208286
32 51.5097656
34 44.14566415 10
36 34.1419685
38 24.00464013
40 15.9961201 0
42 10.48827475
0 30 60 90 120 150
44 6.902516144
46 4.596633406 KL/r
48 3.105440361
50 2.129145204
81
8/25/2014
E
0 2 b / 2 2
0.5 y y x d dx 0.5 y y x d dx
b / 2 b 0 b
2d y b 2d y b 2
2
0.5 y d b 2 0.5 y d b 2
b 8 b 8
0
82
8/25/2014
0.5 y
2 EIx 2 EIy
Px and Py y
L2 L2 ab ab
Yieldingoccurswhen
x
0.5 y i.e., P 0.5PY
Inelasticbuck lingwill occur after 0.5 y y
Y Y
2Y/b
2
Y Y ab Y (1 2a )
b
83
8/25/2014
1 PTx
PTx Px 2 1 Pcr PTx
2 PY
2
PTx Px 1 PTx Px 1 E r
2 1 Let, 2 2 x
PY PY 2 PY PY lx Y K x Lx
PTx 1 1 PTx
2 1
PY l2x 2 PY
P
21 Tx
PY
l2x PTx
PY
84
8/25/2014
85
8/25/2014
Centroidal axis
86
8/25/2014
Section Dimension
b 12 fiber no. Afib xfib yfib r-fib r-fib Ixfib Iyfib
d 4 1 2.4 -5.7 0 -22.5 -7.759E-04 3.2 78.05
y 50 2 2.4 -5.1 0 -17.5 -6.034E-04 3.2 62.50
3 2.4 -4.5 0 -12.5 -4.310E-04 3.2 48.67
No. of fibers 20 4 2.4 -3.9 0 -7.5 -2.586E-04 3.2 36.58
5 2.4 -3.3 0 -2.5 -8.621E-05 3.2 26.21
6 2.4 -2.7 0 2.5 8.621E-05 3.2 17.57
A 48 7 2.4 -2.1 0 7.5 2.586E-04 3.2 10.66
Ix 64 8 2.4 -1.5 0 12.5 4.310E-04 3.2 5.47
Iy 576.00 9 2.4 -0.9 0 17.5 6.034E-04 3.2 2.02
10 2.4 -0.3 0 22.5 7.759E-04 3.2 0.29
11 2.4 0.3 0 22.5 7.759E-04 3.2 0.29
12 2.4 0.9 0 17.5 6.034E-04 3.2 2.02
13 2.4 1.5 0 12.5 4.310E-04 3.2 5.47
14 2.4 2.1 0 7.5 2.586E-04 3.2 10.66
15 2.4 2.7 0 2.5 8.621E-05 3.2 17.57
16 2.4 3.3 0 -2.5 -8.621E-05 3.2 26.21
17 2.4 3.9 0 -7.5 -2.586E-04 3.2 36.58
18 2.4 4.5 0 -12.5 -4.310E-04 3.2 48.67
19 2.4 5.1 0 -17.5 -6.034E-04 3.2 62.50
20 2.4 5.7 0 -22.5 -7.759E-04 3.2 78.05
Strain Increment
Fiber no. tot fib Efib Tx-fib Ty-fib Pfib
-0.0003 1 -1.076E-03 -31.2 29000 92800 2.26E+06 -74.88
2 -9.034E-04 -26.2 29000 92800 1.81E+06 -62.88
3 -7.310E-04 -21.2 29000 92800 1.41E+06 -50.88
4 -5.586E-04 -16.2 29000 92800 1.06E+06 -38.88
5 -3.862E-04 -11.2 29000 92800 7.60E+05 -26.88
6 -2.138E-04 -6.2 29000 92800 5.09E+05 -14.88
7 -4.138E-05 -1.2 29000 92800 3.09E+05 -2.88
8 1.310E-04 3.8 29000 92800 1.59E+05 9.12
9 3.034E-04 8.8 29000 92800 5.85E+04 21.12
10 4.759E-04 13.8 29000 92800 8.35E+03 33.12
11 4.759E-04 13.8 29000 92800 8.35E+03 33.12
12 3.034E-04 8.8 29000 92800 5.85E+04 21.12
13 1.310E-04 3.8 29000 92800 1.59E+05 9.12
14 -4.138E-05 -1.2 29000 92800 3.09E+05 -2.88
15 -2.138E-04 -6.2 29000 92800 5.09E+05 -14.88
16 -3.862E-04 -11.2 29000 92800 7.60E+05 -26.88
17 -5.586E-04 -16.2 29000 92800 1.06E+06 -38.88
18 -7.310E-04 -21.2 29000 92800 1.41E+06 -50.88
19 -9.034E-04 -26.2 29000 92800 1.81E+06 -62.88
20 -1.076E-03 -31.2 29000 92800 2.26E+06 -74.88
87
8/25/2014
1
( T/ Y)
0 .8
0 .6
0 .4
0 .2
0
0 20 40 60 80 100 120 140 160 180 200
KL/r ratio
(KL /r)x (KL /r)y
88
8/25/2014
1.2 1.2
1 1.0
Normalized column capacity
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0.0
0.0 0.5 1.0 1.5 2.0
Num- x Num- y Analytic al-x
Elastic
Lambda
AISC-Design Analytic al-y
89
8/25/2014
Therefore,
z z
1 E I x v P v f P x0 M BY ( M TY M BY ) M BX ( M TX M BX )
L L
z z
2 E I y u P u f P y0 M BX (M ( M TX
TY +M
MBXBY )) M BY ((M TY+M
M TX BX))
MBY
L L
z
3 E I w f (G KT K ) f u ( M BX (M BX M TX ) P y0 )
L
z v u
v ( M BY ( M BY M TY ) P x0 ) (M TY M BY ) (M TX M BX ) 0
L L L
90
8/25/2014
91
8/25/2014
Mo
But , y neglecting higher order terms
Ix
Mo
K y ( xo x) 2 ( yo y ) 2 dA
A
I x
Mo
K y x x 2 2 xx0 yo2 y 2 2 yy0 dA
2
o
Ix A
M 2
K o xo y dA y x y dA x0 2xy dA yo y dA 2 yo y dA
2 2 2 2
Ix A A A A A
Ix A
y x y dA
A
2 2
K Mo 2 yo
Ix
y x y 2 dA
2
K M ox where, x A
2 yo
Ix
x is a new sec tional property
92
8/25/2014
93
8/25/2014
l2 L
L L
4 4
2 2
l2 2 l1 2
L L
M o2 2 G KT 2
l2 2
2
E I y Iw L E I w L2
2 G KT 2 2 E I y 2 E Iw
Mo (E 2
I y Iw ) 2
E I w L2
Mo
L2
L2
G KT
L
94