Professional Documents
Culture Documents
1025
__________________________________________________________________________
STEEL CONSTRUCTION:
APPLIED STABILITY
1. INTRODUCTION
Real structural members do not behave exactly as elastic bifurcation theory predicts.
Firstly, the material is not infinitely elastic; as a result elasto-plastic behaviour and
inelastic buckling occur. Secondly, the structural members are affected by several kinds of
imperfections (mainly of geometric and/or mechanical nature), which may severely
weaken the carrying resistance. The rapid change in deformation with increase in applied
load (characteristic of the buckling phenomenon) also gives rise to second order effects
(1)
At this stage, any slight disturbance results in a lateral displacement giving rise to an
additional bending moment. The member collapses with the deformation confined to a
plastic hinge at some point along its length (Figure 2a); the stress distribution is birectangular (Figure 2b), plastic deformation occurring in tension as well as compression.
1 =
(2)
intersect at
is plotted
Compared to the ideal elastic-plastic behaviour, three main differences are evident:
a) at high strains, the material may undergo strain-hardening.
b) the strain is limited to a maximum value, at which point the material breaks.
c) there may be no clearly defined yield point.
As buckling does not require large material ductility, point (b) is usually irrelevant.
Strain-hardening does not affect the first yield but contributes to an increase in the
collapse load. This effect is obviously more marked at low slenderness values, where
plastic deformation is predominant, and should raise the corresponding portion of the
column strength curve (Figure 3a). In practice, the beneficial effect due to strain-hardening
is neglected.
Different steel alloys exhibit different stress-strain behaviours; high strength steels, in
particular, do not usually have a clearly defined yield point and can be classified as
follows:
1. Those having stress-strain curves with an elastic range restricted to the limit of
proportionally p, with the yield plateau at = fy > p (Figure 5a).
2. Those having a similar elastic range but no yield plateau, strain hardening being
experienced throughout the entire inelastic range. In this case the yield stress, fy, is
generally taken as the 0.2% proof-stress (Figure 5b). This lack of a clearly defined
yield point has the following results:
A very stocky member fails by axial plastic squashing without lateral deflection.
In the range of intermediate slenderness ratios ( p), the stiffness of the material
in resisting stress > p decreases progressively as the stress increases beyond p;
inelastic buckling can occur at a load below both squash load and the elastic
critical load.
The strut resistance curve becomes that shown in Figure 6 (strain-hardening has been
disregarded).
(3)
where v(x) is the additional buckling deflection associated with the axial load N. Because
the first buckling mode of a pin-ended strut is a sine half-wave, the initial curvature is
similarly chosen; it is then easy to show that the amplitude of the total deflection vt at the
critical section (at x = 0,5 L in this case) is as follows:
vt = vo/(1 - N/Ncr)
(4)
where Ncr = 2EIz/L2 is the critical column buckling load. The axial load, therefore,
magnifies the initial out-of-straightness and the first-order bending moment Nvo by an
amplification factor 1/(1 - N/Ncr):
M = Nvt = Nvo/(1 - N/Ncr)
(5)
The distribution of bending moment along the member increases with the lateral
deflection. Bending moment and stresses are the greatest at the critical sections; on the
concave side of the strut, compressive stresses due to axial load and bending moment
superimpose and the strut experiences the greatest stress. First yield will occur at this point
as the axial load is steadily increased; the value of N corresponding to first yielding,
termed Ny, constitutes the limit of validity C of the elastic response O'AB (Figure 8).
Any further increase of N above Ny results in a spreading of yielding, both along the strut
and into the cross-section (Figure 9). This results in a further degradation of the strut
stiffness, such that a maximum load NK is attained, at which the strut fails (point D in
Figure 8). The ultimate load carrying resistance is hence a function of two sources of
flexural stiffness deterioration: the axial load (instability) and the spread of yielding
(plasticity). The stress distributions across the strut cross-section at stages C and D are
shown in Figure 10. Usually the concave side of the strut does not yield in tension at
collapse.
c) The post-collapse behaviour is a slowly descending curve for slender struts with any
tolerable initial out-of-straightness.
If the collapse criterion is chosen as the first yield load Ny instead of the maximum load
NK, the strength curve will be lower but of similar shape (curve b - Figure 12).
3.1.2 Eccentricity of loading
An end compressive load, N, applied with an eccentricity, vo, to an initially straight pinended strut (Figure 7b) will induce a first-order bending moment in the strut which will
begin to deflect laterally in a similar manner to the initially crooked strut analysed earlier.
(6)
The sole deviation from Equation (3) is that the initial imperfection vo is not a function of
x. From the integration of Equation (6) it can be seen that the first-order bending moment
M0 = N vo is amplified by a factor =
The amplification factors associated with the initial sine crookedness and the loading
eccentricity are close to each other for the range of N/Ncr values encountered in practice.
Therefore, the load-deflection response for an eccentrically compressed initially straight
strut is similar to that plotted in Figure 8, except that the response curve starts at the origin
of the axes.
Unlike the initial curvature, which is strongly dependent on the strut length, the loading
eccentricity is more related to the section size. In addition, first-order bending is constant
over the entire member length so that bending effects are likely to make the ultimate
strength of very stocky members lower than the plastic squash load, giving, therefore, the
form of strength curve (c) plotted in Figure 12.
achieved in the fibre(s) with peak compressive residual stress r,c at a load Ny for which
p + r,c = fy .
Up to this loading level, the behaviour is elastic; the uniformly distributed applied stress
is plotted against
average axial strain (shortening/length) in Figure 14. This graph is similar to the -
curve for steel with no clearly defined yield point but with a yield plateau (Figure 5a). It
behaves as if the compressed stub column having residual stresses was made with a
fictitious steel having a stress-strain relationship and a limit of proportionality p which is
The collapse load of a stub column is, therefore, not affected by residual stresses, and still
equals the plastic squash load. Similarly, very slender struts, whose critical buckling stress
does not exceed p, buckle elastically and are not affected by residual stresses. The effect
of residual stresses is most marked in the range of intermediate slenderness; in this case
premature yielding reduces the bending stiffness and the struts buckle inelastically at a
load below both the elastic critical buckling load and the plastic squash load. The
corresponding resistance (strength) curve is of the form shown in Figure 6; it must,
however, be kept in mind that the coordinates represent the average applied stress and
axial strain respectively, i.e.
is substituted for p.
The greatest loss of strength resulting from the effect of residual stresses is again at 1.
, so that some
4. CONCLUDING SUMMARY
5. ADDITIONAL READING
1. Ballio G. and Mazzolani F., "Theory and Design of Steel Structures", Chapman
and Hall, London, 1983.
2. Dowling P.J., Knowles P. and Owens, G.W., "Structural Steel Design",
Butterworths, London, 1988.
3. Galambos T.V. (editor), "SSRC Guide to Stability Design Criteria for Metal
Structures", 4th Edition, John Wiley and Sons, New York, 1988.
4. McGuire W., "Steel Structures", Prentice-Hall Inc., Englewood Cliffs, N-J., 1968.
5. Picard A. and Beaulieu D., "Calcul des charpentes en acier", Institut Canadien de
la Construction en Acier, 1991.