Professional Documents
Culture Documents
821-825, 1996
Copyright © 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
S0141-0296(96)00014-4 0141~)296/96 $15.00 + 0.00
ELSEVIER
Winter's classic bracing paper is reviewed and his method for full
bracing is extended to cases when less than full bracing is used.
Full bracing, as redefined in this paper, permits the column or
beam to support load levels corresponding to an unbraced length
of KL where L is the distance between braces and K is an effective
length factor taken as 1.0. Using Winter's rigid bar model, an
approximate buckling load for members with less than full bracing
is developed that compares very favourably with an exact solution.
It is also shown that Winter's model rigid bar model can be
extended to cases where the brace spacing is unequal. Copyright
© 1996 Elsevier Science Ltd.
1. Introduction
P,
Most practical bracing fi)rmulations used in design of col-
umns and beams are based on the principles developed by full bracing
Winter 1. He developed a simple rigid link model with fic- 4¢P
titious hinges at the brace joints to calculate the bracing
stiffness and strength requirements for columns and beams.
The purpose of his dewdopment as directly quoted from
his paper,
-i P ~=
x2EI L~
Q
L2
'to devise a single method which permits: ¢
(1) To calculate a safe lower limit (rather than an
exact value) of the necessary rigidity of the bracing
such that the strength of the braced member will
0 2 L/p,
attain its maximum possible value, and
Figure 1 Effect of brace stiffness
(2) For a bracing of rigidity equal to or larger than so
calculated, to determine a safe, lower limit (rather
than an exact value) of the strength required of ness, L is the distance between braces and P = rtaEl/L 2, the
such bracing.' load corresponding to buckling between the braces. At
brace stiffnesses less than the ideal value,/3i, buckling will
His paper concentrated on requirements for 'full bracing' occur with lateral movement at the brace point. However,
which was defined or inlerpreted as: the buckling load, Pcr increases almost linearly with
increases in brace stiffness until full bracing is reached. For
(1) buckling is forced to occur between braces /3 ->/3~, the column will buckle between the supports. The
(2) the brace is equivalent to an unyielding support exact relationship between Pcr and /3 for 0 -</3 </3i is a
trigonometric function that is too complex for general
In this paper Winter's approach to design of equally design purposes 2. Since the brace stiffness requirements for
spaced beams is reviewed and his methods are extended to full bracing are small, Winter concentrated only on this
cases when less than full bracing is provided and when particular bracing stiffness. To calculate the ideal stiffness,
braces have unequal spacing. Full bracing will also be rede- Winter developed a rigid-link model with a fictitious hinge
fined. at the brace points as illustrated in Figure 2 for the case
of a single brace. Multiple braces will be discussed later.
Displacing the structure with the axial load P by an amount
2. Basic concepts A, cutting the structure at the hinge at mid-height and
The exact relationship between column load and brace stiff- satisfying equilibrium gives PA = L/3A/2. The A cancel and
ness for the case of a uniform straight column with a brace taking P = Pe at full bracing, the/3 = / 3 i = 2Pe/L which cor-
at mid-height 2 is shown in Figure 1. The column buckles responds to the ideal /3~ shown in Figure 1. Winter credits
between braces when/3i --: 2Pe/L where/3 is the brace stiff- Fr. Bleich with introducing the concept of the use of tic-
821
822 Winter's bracing: J. A. Yura
Ao Ao Ao
Ar- 2P - ~ - /3i (3)
1-~ 1-~o~ 1-~
1 ;~2A =Ao 1 !
I
P
0.8
Pe 0.6
P
0.8
0.6
-//
3p/i
P " Pl
0.4 0.4
0.2
°o
(a)
4 8
Ao=0.002Lb
I'=
AT/Ao
~p
16
0.2
f
(b)
,,0.8~ P.
FI~ ( % of P)
Ao = O'O02Lb
3~
P 0.8
P@ Similarly, cuts at C and D give
0.6 c
P.Ac = ~ (2An + 4Ac + 2AD) (6)
0.4
PeAD = ~ (AB + 2Ac + 3AD) (7)
0.2
P
Defining X~ = m c / m B and X2 = AD/AB and solving equa-
2 3 J~i 4 tions (5), (6) and (7) simultaneously gives three solutions:
PL,/Pe
Figure 5 Column with three supports ( 1) X~ = ~/2; X2 = 1 and/3L/Pe = 0.586
(2) X~ -- 0; X2 = -1 and/3L/Pe = 2.000
Pcr='TrZEI/(4L) 2. At low brace stiffness, the column
buckles into a single (lst mode) wave. As the brace stiff- (3) X~ = -~/-2; )(2 = 1 and/3L/Pe= 3.414
ness is increased, the buckled shape changes and additional
brace stiffness becomes less effective. Full bracing occurs Case 3 controls since it requires the largest brace stiffness.
at Pcr=Tr2EI/L2 when /3L/Pe=3.41 which compares The buckled shapes corresponding to the three solutions
exactly with the value gJLvenby Timoshenkoz. are shown in Figure 7. They are similar to the three modes
The rigid-link mode} shown in Figure 6 has three shown in Figure 5 before full bracing is reached. Winter
unknown displacements, A~, Ac and AD at each of the brace only utilized the root that gave the required brace stiffness
locations where fictitiou,; hinges are shown. Taking sum- for full bracing. However, all the roots can be utilized to
mation of moments at point E gives ,cA = (/3/4)(3AB + 2Ac approximate the complete solution for any brace stiffness.
+ AD); similarly, FD = (/3/4)(AB + 2Ac + 3AD). Cutting the The approach is illustrated in Figure 8 (the exact solution
structure at /3 and summing moments gives from Figure 5 is shown as the solid line). The solid line
can be approximated as follows. With no bracing the col-
P umn can support a load of Pe/16, so Pcr/Pe= 0.625 (point
A~
o
ec'V~
L
FA~FA13A
B 17LX *
L
~
707&
c, ~V~-I
L
,<
D, '%'VM 707&
E~
L
F
P f-" 'G
Figure 6 Winter's model -- three supports Figure 7 Three roots - Winter's model
824 Winter's bracing: J. A. Yura
~P
P
p
e
0.8
0.6
0.4
;,.+"7
,j,/"
BI
i i ....
sf'e""
A
7. I' L
Try7
0.4L
3_
pa
-I,v@
,L
0.2 a2
-(9
e p,~
1 2
pLo/P,
a
t
P
Figure 8 Unequal brace spacing (a) Ibl
Figure9 Winter model approximation
1). The brace stiffness required for the first mode shape is
obtained from Case 1, i.e. [3L]P e = 0.586 at Pcr/Pe= 1.0
(point 2). A straight line is constructed, shown dashed,
between points 1 and 2. At zero brace stiffness the second For this case Pe = 7flEI/(O.6L) 2 which would cause buck-
mode starts at Per = 7flEI/(2L) 2 or P,/Pe = 0.25 (point 3) ling in the longest segment. Substituting Pe for P in equ-
and the Case 2 stiffness is f3L/Pe = 2.000 (point 4). A ation (9) gives the ideal /3i= 4.167[~EI/(O.6L) 2]
straight line connects points 3 and 4. For the third mode L = 114.2 EI/L 3. Results for other values of a~/L are given
and no bracing, Per = w2EI/(4L/3) 2 and PJPe=0.5625 in Table 1. The Winter results are exactly the same as those
(point 5) is connected to the Case 3 stiffness, presented by Plaut for the brace stiffness required for col-
f~Z/Pe --- 3.414. The lower bound of the dashed lines defines umn load to reach a level corresponding to Euler buckling
the response which is a good approximation of the exact of the longest unbraced segment. It may be surprising that
theory. Winter's solution is exact for this complex problem so
The rigid link model which has been used to determine some further comments are needed.
the full bracing stiffness requirement can also be used to The Winter solution is exact only for the load level corre-
construct the complete P-/3 relationship for any value of sponding to an assumed Euler buckling load controlled by
brace stiffness. the longest segment. Of course this is exactly the maximum
load that would usually be assumed in design. The column
4. Unequal bracing spacing can support loads higher than this segmental Euler load
because the shorter segment can provide some rotational
The examples and illustrations used by Winter ~ were all restraint to the longer segment but this is usually ignored
associated with members that had equally spaced braces. in design. If load levels higher than the Euler segment load
But this was done for simplicity. Lesser known are the fol- are desired, greater bracing will be required as derived
lowing two sentences in his paper. by Plaut 5.
'By the same simple means of writing equations about If the /3i as derived above is used, there will be move-
the fictitious hinges at the braces it is also possible to ment at the brace point at the Euler load, as shown by the
analyze the required bracing for columns with unequally buckled shape in Figure 10. The Winter model does not
spaced supports and/or with cross-sections which differ give results that correspond to a column load with an
in the individual spans. In this case it is merely necessary immovable support in this case. Winter's 'full bracing'
to introduce for each portion between braces its appro- comments should more correctly be defined as the bracing
priate P e depending on length and cross-section of that to reach a load corresponding to an effective length factor
portion.' of 1.0. If the support was immovable, the column could
support a load 20% higher than the Euler load 5 but the
Plaut 5 has solved the problem of a column with a single bracing requirement would be substantially increased.
brace at any location, as shown in Figure 9a. The lengths
are chosen to be consistent with the Plaut presentation. The
column is divided into two segments, a~ and a2, with L
defined as the total length. For a typical brace location, Table I Comparison of Winter and Plaut
a~/L = 0.4, the Winter model is shown in Figure 9b. Equi-
librium gives the external reaction at A and C as 0.6/3A and Winter Plaut
0.4/3A, respectively. Cutting the structure at the hinge gives al/L /3L3/EI k.
1 Buckled shape, ..xx××~' derived by the Winter approach are only equivalent to
E \/
/Y
immovable supports when the problem is symmetric and
X
¢- X
X
the unbraced segments want to buckle simultaneously.
.o
"G 0.5 ×
X
However, Winter's solution is valid for all cases in which
X K = 1.0 is used in design, even when the bracing spacing is
(D X
X not uniform. Use of the Winter solution will enable critical
D X
0 column segments to reach the Euler load.
0 2'0 4'0 60 8'0 100
References
0.6Lm -~! P = 27.55
()<[---- 1 Winter, G. 'Lateral bracing of columns and beams', Proc. ASCE
EIIL2= 1
L = 100 cm
~ 1.14 kNIcrff-"
1958, 84 (ST2), 1561-1-1561-22
2 Timoshenko, S. and Gere, J. Theory of elastic stability, McGraw-
Hill, New York, 1961
3 Green, G., Winter, G. and Cuykendall, T., 'Light gage steel columns
Figure 10 Buckled shape in wall-braced panels', Cornell Engineering Experiment Station, Bull-
etin No. 35, Part 2, Ithaca, NY, 1947
5. Conclusions 4 Zuk, W. 'Lateral bracing forces on beams and columns', J. Engng
Mech., ASCE 1956, 82 (EM3), 1032-1-1032-16
Winter's method was shown to provide a very general sol- 5 Plaut, R. H. 'Requirements for lateral bracing of columns with two
ution. It is important to note that the bracing requirements spans', J. Struct. Engng, ASCE 1993, 119 (10), 2913-2931