3-1

COMPRESSION MEMBERS

Introduction

 Compression members, often referred as ‘columns’, ‘struts’ or

‘stanchions’, are the second types of axially loaded structural elements,
whose function is to transmit compressive forces between two points.

 Typical uses of compression members are:

Vertical columns supporting floors, bracing in lateral load resisting
frameworks, compression chord of trusses, and struts in lattice girders.

Chord
Strut

Truss

 Main types of compression sections are:

Hot rolled sections - UC

Compounded sections – built-up H, box, or cruciform sections
Open strut sections – angles, channels, tees and hollow sections
 3-2

Cruciform column

Column Behaviour and Modes of Failure

1) Squashing

 Providing the member length is relatively short as compared to its cross-

sectional dimension and the plate element is not too thin (non slender
section), then the member may fail by compressive squash yielding of
material.

Squash load = AgYs

 3-3

2) Overall flexural buckling

 Failure occurs when the column buckles by deflecting excessively in

the lateral direction of column.

 This failure mechanism controls the design of most compression

members.

 Initially, the column deforms only in the

longitudinal direction.

 When the load reaches a critical value, instability

buckling occurs and causes flexural bending and
excessive lateral deflection.
 Failure is defined when the entire member is
unable to remain in its initial position even though
the steel material may still remain linear elastic.

3) Local plate buckling

 Failure occurs by buckling of one or more individual plate elements (e.g.

flange or web) in a shape of multiple sine waves without overall lateral
deflection.

 This can be prevented by placing suitable limits

on plate width-to-thickness ratios; alternatively,
where such limits are exceeded, the design
strength of steel material must then be reduced.

 Proportions of normal hot-rolled UC sections are

such as to preclude local plate buckling in most
instances.

 This has been discussed in previous lectures on

classification of sections.
 3-4

4) Torsional buckling

 Failure occurs by twisting about the longitudinal axis of the member.

 This mode of instability failure is unlikely to take

place in hot-rolled I or H shaped sections.
 This mode may be important for lighter cold-
formed sections, particularly those with
unsymmetrical shapes.

 Not to be covered in this course.

Elastic Overall Flexural Buckling Analysis of Straight Column

Assumptions:
1) Perfectly straight column without any geometric or material
imperfections.
2) The column is simply supported at the ends of the member.
3) Axial load is applied concentrically along the centroidal axis of the
column.
4) Material is linear elastic.
5) Plane sections remain plane.
6) Lateral deflections are small and are due only to bending (i.e., shear
deformation is ignored).
 3-5

P Pcr

Stable Neutral
equilibrium equilibrium


X
Y

Unloaded member Initially loaded Overall flexural

member buckling of member

Consider the neutral equilibrium condition at the onset of buckling

Two equilibrium paths

By equilibrium,

Mint = Mext

EI  Py
 3-6

For small deformation,

d2y
 2
dx

d2y
 EI 2  Py  0
dx

P d2y
Assume k 
2
, then we get 2
 k2y  0
EI dx

General solution: y  A sin kx  B cos kx

As x = 0, y(0) = 0 → B = 0
e
x = L, y(L) = 0 → Asin kL = 0

Nontrivial solution: sin kL = 0

or kL = nπ where n = 1,2,…

n
k
L L/r

n 2 2 EI P n 2 2
 Pcr  (Note k 
2
 P  k EI  2 EI )
2

L2 EI L

As n = 1, Pcr is the smallest and it is generally known as the Euler load

Pe.. Divide the Pe.. by the cross section area to get the Euler stress as
 2 EI  2E
e   which is inversely proportional to slenderness
L2 A (L r) 2

n  L 
When x =L/2, then y  A sin    y = δm =A
L 2

x
 y  δm sin
L
 3-7

Elastic Buckling of Initially Crooked Columns

All practical steel sections are not perfectly straight. Assume that a column
has small initial curvature in the form of a half sine curve
x
y 0   0 sin
L

y = lateral deflection from

initial crooked position

Initially crooked Crooked column Free body

column after loading diagram

By equilibrium,
EIy  P  y  y0   0 , or
x
y   k 2 y  k 2 0 sin where k 2  P EI (1)
L
Complementary solution: y  A sin kx  B cos kx (2)
x x
Particular solution: y  C sin  D cos (3)
L L

Substitute (3) into (1), one gets the following general solution,
0 P P
x
y  A sin kx  B cos kx  e
sin
1 P L
Pe
Boundary conditions:
when y  0   0 , then B  0
when y  L   0 , then A sin kL  0
 3-8

Since sin kL  0 gives the condition of Euler load which is not the case
herein. Therefore, A must be 0, i.e., A  0

0 P P
x
With A  B  0 , y e
sin
1 P L
Pe

P
Pe x
Total lateral deflection  y  y 0  (1  ) 0 sin
1 P L
Pe
1 x
( ) 0 sin
1 P L
Pe
L
Maximum lateral deflection at mid-length when x 
2
1
At x  L , m  ( ) 0
2 1 P
Pe

Dimensionless

load P Pe

m
1

Dimensionless lateral deflection

 3-9

Maximum bending moment at mid-length at column

 1 
Max M  P m     P o
 1  P Pe

Amplification 1st order moment

Factor

Perry-Robertson Formula

Max stress occurs at x  L 2 where the flexural bending is maximum

when  m increases and so do the moment and the stresses.

Assume linear elastic buckling such that the stress in the column
remains linear elastic
b

P M max b P b P 0
 max    
A IY A IY 1  P Pe

PY
Set limiting  max  YS  , and at such value P  Pc
A

Set the limiting axial load Pc at which the member first yields

Ab  Pc 
PY  Pc  0  
IY  1  Pc Pe 
Pc Ab
 Pc  PY   where the imperfection factor   0
1  Pc Pe IY

 Pc2   PY  1    Pe  Pc  Pe PY  0
 3-10

Solving the quadratic equation for Pc

P  1    Pe  P  1    Pe 
2
Pc  Y   Y   PY Pe
2  2 
P
Therefore, the limiting compressive strength pc  c
A

 p y  1    pe   p y  1    pe 
2
pc        p y pe (*)
 2   2 

PY Pe
where py  , pe 
A A

Eq. (*) is the Perry-Robertson formulae which is valid only for linear elastic
buckling situation in which the effect of geometric imperfection is taken
into due consideration.

The Perry-Robertson formulae has been adopted in the HKSC2011 for

defining the compressive strength of columns; but has been rewritten in a
different form as

p y pe
pc 
c  c2  p y pe
p y  1    pe
where c 
2
 3-11

Effects of Residual Stresses in Buckling of Hot-Rolled Sections

 Residual stresses result mainly from nonuniform

cooling of hot-rolled or welded shapes

 Magnitude and distribution of residual stresses

depend considerably on the geometry of the
cross-section.

 Note that flange tips have more surface exposure,

then they tend to cool faster than the flange-web
juncture.

→ contraction of hotter metal at the juncture is

restrained by the cooler material

→ develop tensile stresses at the juncture &

compressive stresses at the flange tips

Effect of Residual stresses on a column flange

Applied axial Initial residual Partially

compressive stress profile on overstressed
stress the flange

 The presence of residual stresses causes partial local yielding that

reduces the bending stiffness (I) of the section and results in premature
buckling.

 Note that effects of residual stresses on buckling about the weak Y-Y
axis is more severe than for strong X-X axis
 3-12

Inelastic Flexural Buckling

 The earlier discussion of the Euler column has been based on the
assumption that the column is made of linear elastic steel material.
Only columns that are very slender (i.e. having a large slenderness
ratio of L/r) experience elastic buckling.

 Columns that are less slender may fail in an inelastic flexural buckling
mode. In this mode of failure, a member experiences inelastic flexural
buckling with a nonlinear distribution of stresses across the critical
cross section. While the steel material of the loaded side of the column
reaches yielding, the unloaded side may still remain linear elastic.

 Since yielding of material occurs in the column, the column is unable

to return to its original position when the load is removed.

Stocky column Slender column Intermediate column

 3-13

Summary of flexural buckling behaviour

 For straight slender columns, elastic flexural buckling occurs at the

critical Euler buckling load.

 2 EI  2E
Pe  or  e 
L2 L r 
2

 For stocky columns, failure arises from gross area yielding.

 failure  p y

 For intermediate length columns, inelastic flexural buckling occurs.

Failure load is deviated from Pe because of
1) Inelastic yielding
2) Material imperfections – initial residual stresses
3) Geometric imperfections – initial crookedness

Experimental data for column strength of different types of steel sections

 3-14

Concept of Effective Length

 Thus far, the compressive strength of a column has been studied based
on a pin-ended support condition, i.e.

pin: no lateral translation

and free to rotate

 Column buckling depends upon the end fixity (restraint) condition.

 For columns with other than pin-ended conditions, i.e.

Critical buckling load can be determined by using the elastic stability

theory in the same way as we did for the Euler column. However, this
approach is quite cumbersome and the associated differential equation
may be difficult to solve.
 3-15

 An alternative practical approach is to use the concept of an effective

length which is defined as “the length of an equivalent pin-ended
column having the same load-carrying capacity as the member under
consideration provided with its actual conditions of support”.

 2 EI
Pcr  where LE = effective length = kL
L2E

 Effective lengths of idealized columns are given in HKSC 2011 Table

8.6.

N.A.
N.A.
 3-16

Some illustrations of Idealized columns

Theoretical k = Theoretical k = Theoretical k =

HKSC recommended k = HKSC recommended k = HKSC recommended k =

Sway Sway permitted

prevented columns

Theoretical k = Theoretical k = Theoretical k =

HKSC recommended k = HKSC recommended k = HKSC recommended k =

 In practice, the end restraint provided may be different about the two
column axes (e.g., held in position in the x-axis, but fixed end sway in
the Y-axis). Hence, the effective lengths in two planes may be
different, i.e.

LEx  LEy and thus λx  λy

 3-17

Column Design Curves – HKSC 2011 Section 8.7.5 and Appendix 8.4

 The compressive strength, pc, is derived based on the Perry-Robertson

formula as
pE p y
pc =
c + c2 - p E p y
p y + ( 1  ) p E
where c =
2
py = design yield strength (for welded sections, py should be
reduced by 20 N/mm2 and pc should be determined on the
basis of this reduced py according to clause 8.7.6)
pE = Euler strength = π2E/(L/r)2 = π2E/(λ)2
λ = slenderness = LE/r in clause 8.7.4 where the effective
length LE (see Table 8.6) and the radius of gyration should
be appropriately taken with respect to a relevant axis of
buckling.
η = imperfection parameter or Perry factor = (λ-λo)/1000 0

λo = the limiting slenderness = 0.2  2

E
py

 = Robertson constant depending on the section type and

buckling axis, should be taken as 2.0 for strut curve (a); 3.5
for strut curve (b); 5.5 for strut curve (c) and 8.0 for strut
curve (d).

 HKSC2011 uses the Robertson constant  to incorporate the effects due

to imperfections (e.g. material imperfection due to residual stresses and
geometric imperfection due to initial crookedness). The choice of the
appropriate strut curve (a0, a, b c or d) is given in Table 8.7 depending
on the type of section and the axis of buckling. In general, the greater the
value of the Robertson constant , the larger the effect of imperfections
and the smaller the corresponding compressive strength value pc.
 3-18

,α
Compressive strength

a
b
c
d
Strut curves for grade S275 steel

Larger reduction due to higher

effects of imperfections
 3-19

 Four tables (i.e. Tables 8.8(a), (b), (c) and (d)) corresponding to the four
strut curves are given in HKSC2011 to find the value of the compressive
strength (pc) as a function of slenderness (λ) and design strength of steel
for sections identified according to Table 8.7.
 3-20

Column Design Procedure

1) Select the steel grade and section. Typically, UC sections made of

steels supplied in accordance with BS EN standards are used.

2) Determine the design strength p y from Table 3.2 and classify the type
of the section from Table 7.1. For slender sections which are liable to
local buckling, the value of design strength p y for a slender column
must be reduced. Clause 7.7 of HKSC2011 provides a simplified
effective stress method to obtain the reduced design strength of a
slender section. An alternative approach detailed in clause 7.5 based
on the effective width method can be used to reduce the effective
section properties of a slender section.

HKSC2011 clause 7.7:

2
 3 
The value of the design strength may be obtained from p yr    p y
 
in which  is the value of b/T, b/t, D/t or d/t that exceeds the limiting
values of  3 given in Table 7.1 for a Class 3 semi-compact section.

3) Estimate the effective length LE using Table 8.6 based on the end
support conditions.

4) Calculate the slenderness λ = LE/r for the relevant buckling axis (x-
axis or y-axis).

5) Select the appropriate strut curve from Table 8.7 based on the type of
the section and the axis of buckling.

6) Read the compressive strength, pc , from the Tables 8.8(a)-(d),

depending on the design strength p y and the slenderness λ. Or
alternatively, calculate the compressive strength pc using the Perry-
Robertson formula given in Appendix 8.4 of HKSC2011. For welded
sections, p y should be reduced by 20 N/mm2 and pc should then be
determined on the basis of the reduced p y .
 3-21

7) Calculate the compression resistance of the column from the following

equations in accordance with clause 8.7.5 of HKSC2011.

(a) for Class 1 plastic, Class 2 compact and Class 3 semi-compact

sections:
Pc = Ag pc
(b) for Class 4 slender sections:
Pc = Aeff pcs
where Aeff = effective cross sectional area from clause 7.6.
pcs = the compressive strength pc for slender sections
using a reduced slenderness of λ(Aeff/ Ag)0.5 where λ is
the slenderness ratio calculated from the radius of
gyration of the gross sectional area and member length.
Ag = the sum of gross sectional area in clause [Link]
pc = the compressive strength in clause 8.7.6
Or alternatively, Pc = Ag pc, in which the reduced design strength pyr
from clause 7.7 is used in place of py.

Example

1) Determine the axial compression resistance of a 8m long simply

supported hot-rolled steel section having section properties as

E = 205,000 N/mm2
Ag = 76.4 cm2
B = 205.8 mm T = 14.2 mm
d = 160.8 mm t = 9.4 mm
Ix = 6125 cm 4 Iy = 2065 cm4
rx = 8.96 cm ry = 5.20 cm

Assume S275 steel, since all plate thickness is less than 16 mm, from
 3-22

Table 3.2, use

py = 275 N/mm2

Check section classification:

outstand flange:
b 205.8 / 2
  7.25  15
T 14.2
for web:
d 160.8
  17.1  40
t 9.4
so, the section is not slender, take full design strength py =275N/mm2

From Table 8.6, the effective length of an idealized simply supported

columns is given as

LE = 1.0 L = 8000 mm (in both x-axis and y-axis)

Calculate the slenderness

8000 8000
X   89.3 ; Y   153.9
89.6 52.0

From Table 8.7, for UC section (rolled H-section) with max thickness 
40mm, then
use Table 8.8 strut curve (c) to get pcy with weak y-y axis of buckling
use Table 8.8 strut curve (b) to get pcx with strong x-x axis of buckling

With p y  275 N mm2 and  y  153.9 , from Table 8.8(c) for strut curve
(c)
pcy  65.7 N / mm2 (controls)
With p y  275 N / mm2 and x  89.3 , from Table 8.8(b) for strut curve
(b)
pcx  162.4 N / mm2

The compression resistance

Pc  Ag pc 
 3-23

2) If the column member is laterally supported at the mid-point against

weak-axis buckling, then determine the compression resistance again.

4000
8000

4000

Strong-axis Weak-axis
buckling buckling

While the effective length of the member for strong x-axis buckling is
kept the same, its effective length for weak y-axis buckling is reduced by
one half to
4000
Y   76.9
52.0

With p y  275 N mm2 and  y  76.9 , from Table 8.8(c) for strut curve (c)
pcy  167.2 N / mm2

When compared to pcx  162.4 N / mm , the compressive resistance of

2

the column is controlled by the x-axis compressive strength as

Pc 

Note that the compressive strength of the column can also be determined
by the Perry-Robertson formula given in the Appendix 8.4 of HKSC2011
 3-24

For the hot-rolled UC (or H-shaped) section buckling about x-axis with an
effective length of 8000 mm

Compression resistance

Pc  Ag pc 