Академический Документы
Профессиональный Документы
Культура Документы
p2 EI
p
kL
p
e
; k ; cos cos 0; vL=2 ! 1
L2
L
2
2
0
(2.30)
ML2 PeL2 p2
P=PE e
8EI
8EI
8
v1 L=2
Thus, the deflection is infinite when P equals the elastic buckling load, as
expected.
The center deection of a beam with a moment M at each end is equal to
(from linear structural analysis):
P PE
When
!
1 cos kL
2
vL=2 e
cos kL
2
The maximum deflection occurs at the center at z L=2 and it is equal to:
P
k2
EI
EIv00 Pe v
v00 k2 v k2 e
48
49
(2.32)
5qL
, and therefore the magnification factor MF
The first-order deflection is 384EI
for this case is equal to
!
384
1
kL2
MF 4 4
1
(2.33)
8
5k L cos kL
2
EIviv Pv00 q
A distributed load is not specically an imperfection; however, the transverse loading on the column creates deection due to bending and is thus
similar to the imperfection cases described in the previous sections. The differential equation for this case is equation 2.9 with the foundation modulus
a 0:
If we divide the deflection of the column (equation 2.30) by the linear beam
deflection above we obtain a magnification factor, MF, that defines the
effect of the reduction of stiffness, and thus the increase of deflection, due to
the axial force:
2
q3
1 cos p2 PPE
8
6
7
q 5
MF 2
(2.31)
4
p P=PE
cos p2 PPE
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Magnification Factor
Distributed load
Initial crookedness
10
P Pv P
vA
1
sy
A S
A
S
It is evident from comparing the curves in Figure 2.11 that they are
essentially on top of each other, and thus the simpler equation 2.34 can be
used for all three cases. This fact has been used for many years in design
standards all over the world.
Historically, the idea of assuming an initial out-of-straightness or an accidental load eccentricity was used to arrive at formulas for the design of
columns. They are named, variously, the Rankine formula, or the PerryRobertson formula. Since such formulas were used extensively throughout
the last 150 years, it is useful to give a brief derivation at this point.
The common feature of these formulas is the assumption that the maximum strength of the column is reached when the sum of the axial normal
stress and the exural normal stress equals the yield stress of the material.
This is a very impractical assumption, of course, as we will show in Chapter
3, but since the initial deection is calibrated so that the formula predicts the
actual strength obtained by column tests, a useful and simple method of
design is achieved. The derivation of the Rankine formula is given rst:
plotted in Figure 2.11 against P/PE. Also plotted is the magnification factor
for the case of initial curvature (equations 2.24):
1
MF
(2.34)
1 P=PE
50
P/PE
51
P
sy
sy
A 1 cv2 L 2 1 a L 2
r
r
L
(2.35)
(2.36)
2
1
sy sE
sy sE 1 h
2
L
Empirically in the Australian code, h 0:003 .
r
1
scr sy sE 1 h
2
vo
vo
vo sE
(2.37)
scr 18 scr
18
L 2 15 ksi
1
FS
33
1 18;000
r
scr
P is the axial load, A is the area and S is the elastic section modulus of the
cross-section. sy is the yield stress and v is the
initial deflection at the center
Acv
cv
cv L 2
of the column. Note that Av
2.7
4EIB
uA auA av0 0
LB
MAB
IC
LB
IB
MAB /2
LC
B
and a 3EI
LB when the far end is pinned. The moment at the top end of the
B
The symbol a is a spring constant that is a 4EI
LB when the far end is fixed,
MAB
52
53
PL2C
EIC
(2.39)
EIC
L2C
Pcr 20:19
Equation 2.39 is the buckling equation for a column with a pinned end at
one end and an elastic spring at the other end. When IB a g 0, the
end restraint vanishes and we have a pinned-end column; that is, sin kLC
0 ! Pcr p2 EIC =L2C . When the top end is fixed, the following holds:
IB a g 1
kLC
kLC
tan kLC
kLC 2
g 1
kLC
a kLC
g kLC
PLC a kLC 2 g
a LC
g
EI
C
s
tan kLC
v0 0
av0 0 EIC v00 0 0
vLC 0
v00 LC 0
From the equilibrium condition MAB MAC 0, we then get the fourth
boundary condition. The four boundary conditions are summarized next:
column equals
80
80
100
100
P
T
E Iv''(0)
Bv'(L)
E Iv''(L)
Tv'(0)
v (L) = 0
Bv'(L) E Iv''(L) = 0
Boundary conditions:
v(0) = 0
Tv'(0) EIv''(0) = 0
40
60
= LC / EIC
40
60
= LC / EIC
20
20
END-RESTRAINED COLUMNS
In this section we consider the general case of prismatic columns that are
restrained by elastic springs at their ends. By considering restrained ends,
we can develop a feel for the impact of end restraint on the buckling load
of the column. This situation is similar to a column restrained by beams
of nite stiffness, which is discussed in depth in Chapter 5. We start the
discussion with a compression member whose ends do not translate with
respect to each other (often called a non-sway case) and that have elastic
springs at each end. The column and its end boundary conditions are shown
in Figure 2.14.
Substitution of the four boundary conditions into equation 2.12 results in
four homogeneous simultaneous equations. The determinant of the coefcients of the constants A, B, C, D is equal to
1
0
0
1
1
L
sin kL
cos kL
0
2
0 aT
a
k
EIk
T
0 aB aB k cos kL EIk2 sin kL aB k sin kL EIk2 cos kL
2.8
55
constant. A little restraint goes a long way, but complete restraint is not
worth attaining. This principle is a general characteristic of buckling solutions discussed in more detail in Chapter 5.
0.5
0.6
0.7
0.8
0.9
1.0
1.0
1.2
1.4
1.6
1.8
2.0
2.2
54
P / PE
Effective Length K
RT
aT L
EI
aB L
RB
EI
(2.40)
(2.41)
(2.43)
2EIgB
2IgB =LgB
! RB
LgB
IC =LC
aB
(2.42)
2EIgT
2IgT =LgT
! RT
LgT
IC =LC
aT
Solving equation 2.41 numerically for the smallest kL gives the critical
buckling load. The limiting cases of this equation are the cases of both ends
pinned (aT aB 0 ! RT RB 0), and of both ends fixed (aT aB
1 ! RT RB 1). After some algebraic and trigonometric manipulations it can be demonstrated the eigenfunction for the pinned end column is
equal to sin kL 0 and for the fixed end column it equals sin kL
2 0. These
are indeed the same functions as are shown in Table 2.1 for Cases I and III.
Thus, equation 2.41 encloses all the intermediate conditions between the
totally pinned ends and the totally fixed ends. The critical load thus varies
from Pcr PE to 4PE, and the effective length varies from K 1:0 to 0.5.
The buckling condition of equation 2.41 is directly applicable for the situation where the elastic rotational spring constants aT and aB are known.
Following, we consider the specialization of the expression for the case of a
planar rigid frame. Such an application is within the everyday task of structural design engineers. An example is illustrated in Figure 2.15. We assume
that the far ends of the top and bottom beams have the same slope as the
near ends. This is not the correct situation for this given problem, but it is
the assumption that governs the effective length determination in the AISC
Specication (AISC 2005).
The top and bottom spring constants are:
P
EI
Remembering that k
constant ratios
56