Professional Documents
Culture Documents
AISC5 Useful Stability Concepts PDF
AISC5 Useful Stability Concepts PDF
Joseph A. Yura
Joseph A. Yura, PE, is Emeritus Professor in Civil Engineering at the University of Texas at Austin. He earned a
BSCE degree from Duke, MS from Cornell and a PhD from Lehigh in 1965. He has received numerous teaching
awards during his 40 years of teaching at the U. of Texas. His research contributions are in the areas of stability,
connections, composite construction, offshore structures and elastomeric bearings. He served 32 years as a member
of the AISC Specification Committee. In 2000 he was elected into the National Academy of Engineering for his
work in stability and bracing of steel structures.
ABSTRACT
Five important stability concepts are reviewed: loss of stiffness as the buckling load is approached, inelastic column
buckling, effect of end connection details in built-up columns, lean-on bracing systems and column bracing
fundamentals. The importance of viewing instability as a stiffness issue rather than a strength problem is
emphasized. Physical models and structural failures are used throughout the lecture to illustrate the concepts. Some
unusual applications of the stability concepts are presented.
INTRODUCTION
A theoretical evaluation of stability requires a second order structural analysis, i.e. equilibrium is established for the
deflected position of the structure, not a typical first order analysis. Since most formal engineering education focuses
on the use of first order analysis, many practitioners have had little structural training in stability issues. Usually,
formulas in design specifications for column, beam and frame buckling and for bracing requirements provide the
only means of checking stability. The simplicity and format (stress rather than load, for example) of these formulas
can mask important stability principles.
Two and a half centuries ago, Euler developed his now famous elastic column buckling formula
2 EI
Pcr = (1)
L2
At that time the modulus of elasticity, E, and the moment of inertia, I, as defined presently were unknown quantities.
Euler determined the elastic rigidity term EI from the measured end deflection at the end of a cantilever beam with
an applied load W using the known relationship = WL3/3(EI). The column was assumed initially straight in the
derivation. The use of the Euler formula implies that the initially straight member will remain straight until the Euler
load is reached, thus providing little warning of an impending failure. At the Euler load the column bends without
any increase in load (small deflection theory); the column has lost its stiffness.
Pcr
L/r=160
where o is the initial out-of-straightness, T is 0.4 Fy=50 ksi
the total deflection when a compressive load P is
applied and Pcr is the buckling load. A plot of Eq.
0.2
(2) is shown in Fig. 1. As the applied load P
approaches the buckling load, the slope of the P -
T relationship (stiffness) approaches zero. Since 0
all members in a structure have some initial out- 0 2 4 6 8 10
of-straightness, flexural displacements will get
very large near the theoretical buckling load thus T/ o
providing ample warning of impending disaster. Figure 1. Load Deflection Response for an Imperfect
Column
Quebec Bridge Failure on August 29, 1907 At the time of its construction, the Quebec railroad bridge over the
St. Lawrence River was the longest span double cantilever bridge in the world. The bridge just prior to collapse is
shown in Fig.2. The bridge collapsed as shown in Fig. 3 due to buckling of the compression chords of the truss. The
photos shown were scanned from the official report of the collapse (Canada, 1919).
Chief Engineer Deans insisted that chords 7-L and 8-L had
already been bent when they left the shop. McLure insisted
that they only began to show deflection after being installed on
the bridge. The debate over chords 7-L and 8-L occupied the
greater part of August. Meanwhile work continued, and the
stresses on the lower chords grew.
Hartford Coliseum Collapse on January 18, 1978. The Hartford Coliseum roof structure was a long span space
truss constructed mainly with double angle members. The 30 million dollar structure totally collapsed on January 18,
1978, at 4 a.m., just six hours after the completion of a U. Conn basketball game with five thousand attendees. No
one was injured. There was an ice-snow storm underway at the time of the collapse, but the snow loads were well
within the design load limits. The collapse occurred about three years after construction.
From the Report of the Committee to Investigate the Coliseum Roof Failure, July 13, 1978:
LZA concluded that the initiating cause of the collapse of the Coliseum's space truss roof was a design
deficiency related directly to inadequate bracing of certain top chord compression members of the space truss.
The space truss was assembled on the ground and then lifted into place. During construction the following defection
deficiency was noted:
When F-B-Y designed the space truss, it anticipated deflection at the center of the space truss of 13.07 in.
under full dead and live (snow) load. The actual deflection in March and April, 1975 without any snow load
approximated that number according to measurements.
The dead load deflection was expected to be 9 inches. Since some of the truss compression chords were approaching
their buckling load, they had reduced stiffness resulting in greater truss deflection
Marcy Pedestrian Bridge Collapse on October 10, 2002. During the construction of a 171ft span steel trapezoidal
box girder pedestrian bridge in Marcy, NY, global lateral-torsional buckling occurred when the concrete deck pour
was about 60% completed (Weidlinger, 2003). When the steel alone had been erected, one top flange was about 2
inches higher than the other at midspan. The bridge had been previously assembled in the shop to check the camber
and field splices. In the shop the heights of the two flanges were within a few mm of each other. The cause of the
difference in the field was not resolved, but most likely the bridge was already starting to twist under its own dead
weight.
The following is taken from one of the witness accounts:
The crane was lifting buckets full of concrete onto the bridge deck and they had completed pouring the north
half of the deck when we got there. I noticed that then the concrete finishing machine changed directions, that
there was an unusual amount of bounce. I looked at Brian and he looked at me because we both noticed the
bounce at the same time. Brian, Ted and I were joking about me riding my bike across the bridge and I
commented that there better just be one bike. We never imagined that it was going to collapse, but did notice the
unusual amount of bounce. Brian and I were on the deck about ten minutes and then drove back .. to our office
which is about mile away
The bridge collapsed as they reached their office.
These three examples all deliver the same lesson As instability is approached, the structural system is losing its
stiffness, which translates into large displacements for minor changes in load.
where c = ( L r ) Fy / E .
Two pinned-end column models with the material and cross sections shown in Fig. 7, were developed to study
bracing requirements in the inelastic range (Gill, 1999). The models worked very well for that purpose but they also
provided a clear evaluation of the zero stiffness principle. A composite cross section B was fabricated by epoxing
two low strength steel sections (40 ksi) to a high strength steel core A (proportional limit = 70 ksi). The high
strength steel core was also tested separately. Column tests were conducted on columns with 19 in. and 9.5 in.
lengths. The 19 in. long columns buckled at stresses less than 40 ksi and composite column B carried 40% greater
load than column A as expected since its cross section was 38% larger (I is also 38% larger). When the 9.5 in. long
columns were tested, both sections A and B buckled at the same load. The stress on the composite column at
buckling was greater than 40 ksi so the low strength steel sections yielded and were ineffective in resisting buckling.
The test showed that our concept of inelastic buckling for steel columns is correct. The lesson from the inelastic
model test has practical importance. If a column in a structure must be reinforced to carry additional load, the
column strength may be affected by the stresses already applied to the column before the reinforcing is added. If
buckling is expected in the elastic range of behavior, the column strength can be established from the cross section
properties of the reinforced column. However, if the added stress (added load/ total reinforced cross section) when
Stress(ksi)
80
1/2
60 A
Low Str. epoxy
40 Steel
20
B
0
0 0.002 0.004 0.006 0.008 3/16 3/16
Strain
column cross section
Because of the wide variation in expected out-of-straightness and residual stresses, column design equations in the
inelastic range are determined from curve fits through the test results. However, most design is based on elastic
structural analysis. In order to account for the reduction in stiffness due to inelastic column behavior, the concept of
a Stiffness Reduction Factor (SRF), , was developed (Yura, 1971) to more accurately evaluate frame or system
stability. In other words, the column rigidity in the inelastic range of applied stress is EI. is derived as the ratio of
the inelastic column capacity for a given L/r to the elastic buckling strength for that same L/r. Two common values
of to be used with factored loads are given by Eqs. (3) and (4).
( Pu / Py )
For ( Pu / Py ) > 0.35, = 6.97( Pu / Py ) log (3)
0.90
or
for ( Pu / Py ) > 0.50, = ( 4 Pu / Py )(1 ( Pu / Py )) (4)
where Py is the column squash load, (FyAg), and Pu is the required column strength. SRF = 1.0 when the Pu/Py ratio
is smaller than the limits shown. Eq.(3) is developed from the AISC inelastic column formula, including LRFD
factors, whereas Eq.(4) is based on the column formula used in the 9th Ed. AISC-ASD Specification with the factor
of safety removed. Both give reasonable estimates of the SRF. The use of will be illustrated in the next section.
The P concept provides the potential for design flexibility when dealing with sway instability. Since all columns
must deflect the same amount in the sway mode of buckling, designs need not be confined to supporting the load
specific to each particular column. Lateral stiffness can be concentrated with just a few columns or spread evenly
among all the columns (make all columns in the story the same size). The concept works equally for beam systems.
The concept also shows that lateral loading does not have a significant effect on buckling because the sum of the
vertical forces P does not change with lateral loading. For stability of structural systems, Eq (2) is altered as
follows:
0
T = (5)
P
1 u
Pcr
so that Pcr Pu. The Pcr term can be evaluated by a number of techniques. For example, the alignment chart can
be used to determine the contribution of each column, including effects, both in the evaluation of the K-factor and
in the contribution itself. In the AISC-LRFD Specification, the individual contribution can be taken as
2 EI
Pn = 0.90(0.877 )
I
= 226000 (6)
(KL )2
(KL )2
Pcr can also be evaluated by using buckling computer programs, 2nd order analysis, or story stiffness methods
presented in a recent seminar (AISC-SSRC, 2003).
The P concept (or lean-on concept as it is sometimes called) is a very powerful design tool. It has been used to
design jacket legs in offshore platforms where a tubular column is inserted inside a round pile. In seismic design,
round tubular sleeves have been placed around compression diagonal braces. The diagonal supporting the entire load
will attempt to buckle as it yields but it will be confined by the exterior sleeve. Since the sleeve has no axial force, it
is elastic and its contribution to the system will be given by the elastic Euler load.
When two members are interconnected with fasteners to create a built-up column such as a double angle member,
the cross section is treated as a single member for design purposes. In the case of composite beams, the components
of the built up section, the slab and the steel section, are interconnected with shear studs that are spaced uniformly
along the length of a simply CASE A CASE B CASE C
supported span. Plastic concepts No Connectors Fully Connected Connectors at Ends
are used to determine the number
of stud connectors so relative P/2 P/2 P P
movement (slip) will occur
between the slab and the steel dd d d
d d
section. The maximum slip will
occur at the ends of the span. The
slip does not affect the ultimate L
beam strength but does affect the
deflection. However, for columns K = 1.0 K = 1.0 K = 0.5
slip between components will
have a major effect on strength
since column stability and P/2 P/2 P
P
strength depend on the deflection
response, i.e. stiffness of the Pcr = 2E bd3
Pcr = 4 2E bd3
Pcr = 4 2E bd3
member. 6L2 6 L2 6 L2
The model in Fig. 10 shows two columns with the same cross section. Both are constructed from two similar
rectangular sections separated by spacer bars that are pin-connected. The spacers merely force the two components
to buckle in the same direction. The only difference between the two columns is their end connections. The column
on the left has two shear plates that prevent slip at the ends, the other column has no end connections. As illustrated
in Fig. 10, the model follows the theory exactly. The column with the end plates carries four times the load of the
pinned column and buckles in an S-shape. In both cases the column ends are free to rotate.
Further evidence (Sherman and Yura, 1998) of the importance of preventing slip at the ends is given in Fig. 11,
which show the results of tests on 2L5x3 x double angles with long legs back-to-back separated by a gusset
plate connection with two A325 bolts at each end. There are varying numbers of intermediate bolts evenly spaced
with a small spacer plate or hardened washers maintaining the inch separation. The columns were 12.5 ft. long
Tight - Tight
50
Tight - Snug Washers
Tight,Lubed - Snug
40 Snug - Snug
10
0
0 1 2 5
Number of Intermediate Bolts
Winter (1960) established that braces for columns and beams were required to have a minimum stiffness and a
minimum strength. A violation of either of these two requirements will result in a reduction in the buckling strength
of the main member. Brace stiffness and strength provisions have been adopted in the 3rd Edition of the AISC-
LRFD Specification. Strength requirements are directly related to the magnitude of the initial out-of-straightness or
plumbness. Stiffness requirements are generally a function of load, not E, and are valid in the elastic and inelastic
regions of member behavior (Gil 1999, Ales 1993). Prior to the adoption of bracing requirements in the AISC
Specification, engineers were on their own, usually adhering to past practice associated with only a strength
requirement such as a brace force equal to two percent of the column load.
The bracing model shown in Fig. 12 demonstrates the importance of both brace stiffness and brace strength on
column strength. The objective of the demonstration is to provide sufficient strength and stiffness to a brace located
at mid-height so that the column can support a load corresponding to an unbraced length of one-half the total length;
i.e. the column buckles into an S-shape. When the horizontal member is attached at mid-height, a sliding support
can vary the stiffness of the brace. Figures 12(a) and (b) show a brace with inadequate stiffness and Fig. 12(c) one
with adequate stiffness. In both these instances, the brace strength was sufficient and did not control. Otherwise,
(b)
(c)
(a)
(a) (b)
Stability Bracing
Approximate Method
As an alternative to the more rigorous provisions for stability bracing in the LRFD Specification Section
C3, the historically approximate and generally conservative procedure of designing the bracing element for
a required strength equal to two percent of the factored compressive force in the braced member will
normally suffice.
This statement is contrary to published theory and physical experiments.
SUMMARY
It is important to take the time to establish clearly why a structure is deflecting more than expected. Lack of
stability means lack of stiffness. In many instances, disasters could have been avoided if excessive deflections were
taken more seriously. Bracing details are important. A very small brace permits a main member to support a load
much larger than for an unbraced member. Larger factors-of-safety should be used for bracing members because
they are more important than the main member itself. The P concept suggests that a few stiff members (lightly
loaded elastic members) can effectively stabilize a system with highly stressed columns. The importance of
preventing slip at the ends of built-up members has been demonstrated.
REFERENCES
AISC-SSRC (2003), Lecture 4-Frame Stability, from seminar entitled Basic Design for Stability
Ales, J.M. and Yura, J.A.(1993), "Bracing Design for Inelastic Structures," Proceedings, SSRC Conference "Is Your
Structure Suitably Braced?," Milwaukee, WI, April 6-7.
Canada Department of Railways and Canals ( 1919), The Quebec Bridge Over the St. Lawrence River Near the City
of Quebec on the Line of the Canadian National Railways, Report No.1, Ottawa
Committee to Investigate the Hartford Coliseum Roof Collapse (1978), Report to Court of Common Council, City of
Hartford, Connecticut, July 13,1978, 73p.
Gil, H. and Yura, J. A.(1999), Bracing Requirements of Inelastic Columns, Journal of Constructional Steel
Research, Vol. 51,No. 1, July, pp.1-19.
Hall, D.H. (1981), Proposed Steel Column Strength Criteria, ASCE J of the Structural Division, Vol. 107, No.
ST4, April, pp 649-670
Johnston, B. G. (1971), Spaced Steel Columns, ASCE J. of the Structural Division, Vol. 97, No. ST5, May, p 1465
McQuire, W.(1968), Steel Structures, Prentice-Hall International, Inc., London
Sherman, D. R. and Yura, J. A.(1998), Bolted Double-Angle Compression Members, Journal of Constructional
Steel Research, Vol. 46, n1-3, pp. 470-471, Proc, 2nd World Conference on Steel in Construction (CD-ROM),
Paper197, San Sebastian, Spain, May11-13
Tarkov, J. (1986), Quebec Bridge: A Disaster in the Making, Invention & Technology, American Heritage, Vol. 1,
No. 4, pp 10-17
Weidlinger Associates (2003), Miscellaneous Forensic Evaluation of the Marcy Pedestrian Bridge, Final Report &
Appendices, Aug, 61 p
Winter, G.(1960), Lateral Bracing of Columns and Beams, ASCE Transactions, Vol. 125, pp 809-825
Yura, J.A.(1971), The Effective Length of Columns in Unbraced Frames, Engineering Journal, AISC, Vol. 8, No.
2, pp. 37-42, April 1971. Discussions: Vol. 9, No. 1, pp. 40-48, Jan. 1972 and. Vol. 9, No. 4, pp. 167-168, October
1972.
11
Copyright (c) 2011 by 11
The American Institute of Steel Construction