Mechanical Design 2

MIET2072
Mechanical Design 2
Topic 8
College of Science, Engineering and Health Learning Package

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2

Topic 8: INTRODUCTION TO THE DESIGN OF

STEEL COLUMNS OR COMPRESSION
MEMBERS
Learning Outcomes
Upon successful completion of this topic, you will be able to:

determine the proportions of supporting columns including consideration of

buckling, yield, eccentric load and initial crookedness.

Introduction to the Topic

Pressure vessels are supported in a number of ways, one of which is with lugs
resting on columns.
Columns are a common part of engineering structures, the basic theory of which
you have studied in the prior course Solid Mechanics 3.
Columns or struts i.e. structural members loaded compressively, differ from
tension loaded members in that they have the additional mode of failure of
buckling. You have studied the basics of this in the prior course.
Real column design builds on this basic theory but has to add to it consideration of
imperfections such as the load being applied off centre and the column having
some initial crookedness. Both of these increase the tendency of the column to
bend.

Background Skills and Knowledge

Students will require the following:

Familiarity with basic column theory.

Familiarity with normal stress and bending stress.

Familiarity with the solution of quadratic equations.

Familiarity with the solution of 2nd order differential equations.

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2

Activity 8A - Video of buckling

Please view the short video of buckling to reacquaint yourself with this extra
mode of failure that engineers must be conscious of when designing components
subject to compressive loads.
Topic 8 Buckling video 1
Note: Right click (Google Chrome browser) and choose Save Video As to download video

Activity 8B - AS3990 Mechanical Equipment- Steelwork

In your studies on this topic you will be making frequent reference to some
pages of AS3990 Mechanical Equipment- Steelwork (previously known as
AS1250 SAA Steel Structures Code, published by Standards Australia). Print
out pages 20, 21, 22, 23, 40, 42, 54 and 55 and have them to hand while you are
reading these notes. You will also need to refer briefly to AS1210 Pressure
Vessels.
This Australian Standard can be accessed via the RMIT Library - SAI Global link
to the Australian Standards. 2
NOTE: You will need to login with your student login and password.

Activity 8C - Reading
Read all of Chapter 8 below.

https://www.dlsweb.rmit.edu.au/set/Videos/MIET2349/Topic_8_Buckling.m4v

https://login.ezproxy.lib.rmit.edu.au/login?url=http://www.saiglobal.com/online/autologin.asp

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2

Chapter 8: INTRODUCTION TO THE DESIGN OF

STEEL COLUMNS OR COMPRESSION
MEMBERS
Section 8.1 Concentrically Loaded Columns or Compression
Members
The fundamental theory of columns as developed by Euler has been studied in
Solid Mechanics 3, and can be found in most texts on the Mechanics of Solids. Some
key points of this theory will be revised below then consideration will be given to
the behaviour of real columns whose behaviour deviates from the ideal due to
effects such as unintended inaccuracy in the concentricity of the load, initial
crookedness of the column, inhomogeneity and residual stresses. An equation will
be developed that reasonably predicts the failure load for real columns, which, with
the application of a factor of safety, may be used for the design of concentrically
loaded columns.
Consider now the two columns shown in Figures 8.1a & b, one squat and the other
slender, supported on pinned joints, both subjected to an axial compressive load of
P Newtons.

Fig. 8.1a and b, Axial compressive load on Columns RMIT University, 2013 (Dixon C., Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
If the slender column is given a temporary lateral load that deflects it sideways,
creating a deflection y at height x as shown in Figure 8.1c(ii), then the column
experiences a bending moment P x y at this point and generates within itself, due to
d2y
its elasticity, a resisting or restoring moment EI
.
dx 2

Fig. 8.1c Pin jointed column given temporary sideways deflection

RMIT University, 2013 (Dixon C., Marchiori G.)

If the restoring moment is greater than the bending moment then the column will
return to its original straightness, and it was therefore in stable equilibrium before
the temporary disturbance. If the load P is such that the bending moment is equal
to the restoring moment then the column will remain in the deflected position after
removal of the temporary lateral load. This load is referred to as the critical load
Pcrit. If the load P is > Pcrit then if the column is deflected sideways the bending
moment P x y will be greater than the restoring moment and the column will
collapse by bending. Therefore the situation that existed before the temporary
sideways load was applied was one of unstable equilibrium.
The critical situation is therefore when:
Bending Moment Pcrit y =

d2y
dx 2

d2y
EI 2 the restoring moment
dx

Pcrit
y =0.
EI

The solution to this differential equation is:

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
y

C1 Sin

Pcrit
P
x + C 2 Cos crit x
EI
EI

The boundary conditions are y = 0 at x = 0 and at x = L, the first of which yields

C2=0 and the second gives:

C1 Sin

Pcrit
L=0
EI

This means that either (i) C1=0, which is a trivial solution corresponding to there
being no lateral displacement imposed (i.e. no temporary lateral load applied to test
stability), or (ii) that:

Sin

Pcrit

Pcrit
L=0
EI
Pcrit
L = n where n = 1,2,3 etc.
EI
=

n 2 2 EI
L2

n = 1,2,3 etc.

The buckled shapes at two of the critical loads are shown below in Figure 8.1d.

Fig. 8.1d Buckled shape at two of the critical loads RMIT University, 2013 (Dixon C, Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
Therefore the lowest value of the critical load Pc is for n = 1
2 EI
eqn. C1
L2
So far it has been assumed that buckling took place in the plane of the page, and
this assumption would be met if there were lateral supports for the column normal
to the page. However if there are no such lateral supports and it is only supported
at its ends, it is free to buckle in any direction, and will in fact bend about the
principal centroidal axis having the lowest value of the second moment of area I.

Pc

The analysis above was for a column with pinned ends. Other end supports also
occur such as built-in at one or both ends, one end free, etc. It is found for these
that:
Pc
where: l

=
=

2 EI

(k e L )2

2 EI
l2

eqn. C2

the effective length = keL

and ke, the effective length factor, depends on the method of end support.

Activity 8D
Effective length factor determination
Step 1: View effective length factor for several basic column configurations.
Table 6.2 in the extract at the following e-reserve link shows the effective length
factor for several basic column configurations. The effective length of the column
is equal to the distance between points of inflexion of its buckled shape. Please
refer to Table 6.2 on page 120 of the Steel designers handbook (PDF 88KB). 3
Step 2: Consider pressure vessel support columns
Figures 3.24(a) and (b) in AS1210 show column supports on pressure vessels.
Please go to AS1210 and look at those diagrams. Further column support
arrangements may also be seen in Figs 5.1, 2 and 3 on pages 144 of the Pressure
Vessel Design handbook by Bednar (PDF 39KB). 4

3 Gorenc, B. E. & Tinyou, R., 1981, Compression members, Steel designers handbook, 5th ed., New South Wales
University Press, Kensington, N.S.W., pp. 120. viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/817df72d-47bc-43c5-9cf2-5fe6c4db74c3/1/130823_3_063.pdf>

4 Bednar
https://equella.rmit.edu.au/rmit/file/11437b2d-663d-4213-a8695ffd10b146b8/1/130821_3_036.pdf >

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
Step 3: Consider the effective length factor for pressure vessel support
columns.
With regard to the effective length factor of column supports on pressure
vessels, Bednar makes useful comments in the extract on pages 148-149 of the
Pressure vessel design handbook (PDF 122KB) 5 and concludes with a suggested
value of 1.5.
Step 4: Optional further reading
Detailed consideration of the effect on effective length of adjoining framework
and connection type thereto is beyond the scope of this course. The interested
reader is referred to Appendix E of Australian Standard AS3990 Mechanical
Equipment Steelwork. This appendix applies in particular to columns with
elastically restrained ends. The ratio of the stiffness of columns to the stiffness of
beams is computed at both ends of the column being considered. Further
reading on the effect on effective length of adjoining framework and connection
can be found in the following references.

Trahair, N. S., 1977, The behaviour and design of steel structures, 1st ed.,
Chapman and Hall, London

Gorenc, B. E. & Tinyou, R., 1981, Compression members, Steel designers

handbook, 5th ed., New South Wales University Press, Kensington,
N.S.W., ISBN 0 86840 248 6. For example the extract 6.3.1 Interaction
between the various component members on page 121 of the Steel
designers handbook (PDF 88KB) 6 gives an indication of some of the
issues involved.

5 Bednar H 1986, Pressure vessel design handbook 2nd ed. Van Nostrand Reinhold, New York. p.148- 149,
viewed on 28th August 2013 < https://equella.rmit.edu.au/rmit/file/11437b2d-663d-4213-a8695ffd10b146b8/1/130821_3_036.pdf>

Gorenc, B. E. & Tinyou, R., 1981, Compression members, Steel designers handbook, 5th ed., New South Wales
University Press, Kensington, N.S.W., pp. 121. viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/817df72d-47bc-43c5-9cf2-5fe6c4db74c3/1/130823_3_063.pdf>
6

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
Returning to equation C2:

2 EI
Pc = 2
l

The second moment of area I may be expressed as:

I=Ar2
where A is the cross sectional area, and r is the radius of gyration. Substituting this
into equation C2 gives:
Pc

2 EA

(l r )

The elastic buckling stress, or Euler critical stress, Foc =

elastic critical load is therefore:
Foc

The term l

2E

(l r )

Pc
corresponding to this
A

eqn. C3

is called the slenderness ratio.

For slender columns, i.e. l

is large, the Euler critical stress is less than the yield

r
stress and lateral buckling is therefore of more concern than gross yielding.
However for squat columns, such as is shown in Fig. 8.1a the greater concern is
gross yielding that will occur when fac, the average axial compressive stress, reaches
a value equal to the yield stress FY. These two limits are shown by the Euler curve
and squash limit in Fig. 8.1e.

Fig. 8.1e Column Buckling and Gross yield limits RMIT University, 2013 (Dixon C., Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

MIET2072 2013/Ver. 1

Mechanical Design 2
Another possible mode of failure in columns, or beams on their compressive side, is
that of local buckling in a thin wall or flange such as illustrated in Figure 8.1f.

Fig. 8.1f Local Plate Buckling in a rectangular hollow section having excessive breadth to thickness ratio
RMIT University, 2013 (Dixon C., Marchiori G.)

Detailed treatment of local plate buckling is beyond the scope of this course. The
interested reader is referred to chapter 4 Local Buckling of Thin Plate Elements in
The Behaviour and Design of Steel Structures by Trahair, N. S., 1977, 1st ed.,
Chapman and Hall, London . There it is shown that for a plate of length L, breadth b
and thickness t, simply supported at all four edges and subjected to a uniformly
distributed load P at each end, the critical load for plate buckling Pcp is

Pcp
b

Et 3
2
2
12 1
k
b2

eqn. C4

School of Aerospace, Mechanical and Manufacturing Engineering

10

MIET2072 2013/Ver. 1

Mechanical Design 2
where the buckling co-efficient k depends on the ratio L

and the number of wave

b
nodes or points of inflexion. The lowest value of the buckling co-efficient k for the
simply supported plate is 4, which is appropriate for the high values of most
structural steel members.
If a different method of edge support applies then values for k will be different.

Et 3
The quantity
is called the flexural rigidity of the plate often given the
12 1 2
symbol D , and is analogous to the quantity EI used in beams and columns. This
quantity D has been met in the previous topic of the bending of circular plates.

It can be seen that equation C4 for the buckling of plates is similar to that for the
buckling of columns:

Pc

2 EI

eqn. C2

(k e L )2

Returning to equation C4, the critical value of stress to cause plate buckling (a case
of which is local buckling in the wall of a hollow column) is:

Fol

Pcp
bt

E
12 1 2
2
b
t

( )

eqn. C5

which is similar to the Euler critical stress equation for columns.

Foc

Pc
A

2E

(l r )

eqn. C3

Looking at equation C5 it can be seen that if b

is large then the stress Fol to cause

t
local buckling will be low, and local buckling could be of more concern than yield.
Recall also that Fol is affected, via k, by the method of edge support (as is Foc in
columns affected by the method of end support).

School of Aerospace, Mechanical and Manufacturing Engineering

11

MIET2072 2013/Ver. 1

Mechanical Design 2
At some limiting value of width to thickness ratio
equal to the yield stress FY. For example:
a)

b
the plate buckling stress Fol is
t

For flanges supported along both edges as in box section beams or columns,
k in equation C5 is 4 and if E is 200,000 MPa, is 0.3 and FY is expressed in
MPa then Fol < FY if:

b
t

>

4 2 x 200,000 851

=
2
FY
12 1 - 0.3 FY

b 800
if local buckling is to be
<
t
FY
avoided in as-rolled and stress-relieved flanges which is implicit in part of Rule
4.3.2 in Australian Standard AS3990 Mechanical Equipment Steelwork. For a
steel whose yield strength is 250 MPa this limit corresponds to a width-thickness
b
ratio of approximately 50.
t
(ii) For welded plates which are not stress relieved.
b
560
should be <
t
FY
(i) This may be simplified to the advice to keep

(iii) And for cold form hollow section members:

635
b
should be <
t
FY
(b) For flanges supported along one edge and free along the other, as in an I section,

b
256
should be <
which is implicit in rule 4.3.1 of AS3990. Excess widths
t
FY
beyond the limits of b from the above four equations shall be neglected when
calculating the effective geometrical properties of the section. [See Trahair, N.S.,
The Behaviour and Design of Steel Structures and Australian Standard
AS3990 Mechanical Equipment Steelwork for more detail.]
Many commercial structural sections do keep

b
less than the limits shown above
t

b
in catalogues along with other sectional
t
properties. This being the case yield of the flange rather than local buckling is the
concern. So for a concentrically loaded straight column consisting in essence of
b
plates joined at various edges whose values are not too large, the maximum load
t

as advised and tabulate their values of

School of Aerospace, Mechanical and Manufacturing Engineering

12

MIET2072 2013/Ver. 1

Mechanical Design 2
should be that which causes yield for squat columns ( l
buckling for slender columns ( l

small) and gross lateral

large).

Activity 8E - Torsional buckling

Buckling by twisting about a longitudinal axis
Another type of buckling however is also possible for open section compression
members, which may buckle by twisting about a longitudinal axis as shown in
Fig.3.27 on page 79 of The behaviour and design of steel structures (PDF
104KB) 7 or by combined bending and twisting. For most rolled steel sections the
critical load for this mode of failure is greater than the critical load Pc for lateral
buckling about the weaker axis so the possibility of torsional buckling can be
ignored.

Refer back to Fig. 8.1e where the squash limit (gross yield) and Euler curve (gross
buckling) are shown. Early workers found that for real columns, supposedly
initially straight and concentrically loaded, that were of slender proportions, the
critical buckling stress was less than the Euler equation predicts for elastic
buckling. This was found to be due to initial crookedness, non-concentricity of the
load (i.e. an unintended eccentricity) lack of homogeneity, residual stresses and
inelastic buckling (at intermediate slenderness ratios). Perry, [Ayrton, W.E., and
Perry, J., On Struts in The Engineer, 10th December and 24th December 1886.]
who was one of the significant early workers in this field, analysed theoretically the
combined effects of initial crookedness and unintended initial eccentricity, and
some of their results will be considered later in this chapter.
You have previously studied the effect of eccentricity on column behaviour in Solid
Mechanics 3. This will be revised now and compared with the effect of initial
crookedness. A column pin jointed at each end, with a length L will be considered,
for which the effective length factor ke is 1. For other end support types the effective
length l=keL would take the place of L.

Trahair, N. S., 1977, The behaviour and design of steel structures, 1st ed., Chapman and Hall, London, p. 79,
viewed on 28th August 2013 <https://equella.rmit.edu.au/rmit/file/3995dd92-9269-490d-811b15f2f5860b02/1/130823_3_057.pdf>
7

School of Aerospace, Mechanical and Manufacturing Engineering

13

MIET2072 2013/Ver. 1

Mechanical Design 2

Initial crookedness

Eccentric load

Fig. 8.1g Column with initial crookedness

Fig. 8.1h Column with eccentric load

RMIT University, 2013 (Dixon C., Marchiori G.)

RMIT University, 2013 (Dixon C., Marchiori G.)

Initial shape as shown In Fig.8.1g(i)

x
could be described by y o = aSin
L
Final ordinates of the deflected shape
are given by y, deflection from the initial
shape is y1 as shown in Figure 8.1g(ii)

Deflection from the initial shape is y as

shown in Fig.8.1h(ii):

the bending moment due to P is

the bending moment due to P is

The elastic restoring moment is

d 2 y1
EI
dx 2
For moment equilibrium

The elastic restoring moment is

d2y
EI 2
dx
For moment equilibrium

d 2 y1
EI
= P( yo + y1 )
dx 2

d2y
EI 2 = P(e + y )
dx

M = P( yo + y1 )

d 2 y1 P
P
x
aSin
y1 =
+
2
L
EI
EI
dx

Finding the general solution for y1 and

then considering the boundary
conditions yields
y1 =

x
aSin
1
L

M = P(e + y )

P
d2y P
y= e
+
2
EI
EI
dx

Finding the general solution for y and then

considering the boundary conditions yields

P
P
P1 L
x + Cos
x 1
Sin
y = e tan

EI
EI
EI 2

School of Aerospace, Mechanical and Manufacturing Engineering

14

MIET2072 2013/Ver. 1

Mechanical Design 2
where =

P
P
which is
2
Pc
EI / L
2

Dividing top and bottom of the latter

equation by cross sectional area A.

average compressive stress

Euler critical stress

f ac
Foc

The final ordinates of the deflection

curve
y = y o + y1 = aSin
y=

x
x

+
aSin
L
L 1

1
x
aSin
1
L

The maximum deflection occurs at the

L
middle of the column where x = .
2
Substituting this into the above equation
yields

P L
1
y max = e Sec
EI 2

Bringing L under the

multiplying and dividing
yields

by

P


y max = e Sec
1

2
2

2
EI
L

y max = e Sec
2

Therefore the maximum deviation from

the vertical at the middle of the column
which is

and

Putting =

P
1
Pc

P
gives
Pc

also the maximum moment arm is

y max =

1
a
1

y max = e Sec
1
2

The maximum moment arm, which is at

the centre is y max + e

Moment armmax =

1
a
1

Thus the initial deviation a at the middle

of the
1
1
when the axial compressive load P is
applied. Note that when P approaches
Pc, and hence approaches one, this
amplification factor approaches infinity.

column is amplified by the factor

Moment armmax = eSec

1
and the
Cos
magnitude of Cos is from 0 to 1 it can
be seen that the moment arm e at the top
of this column

Noting that Sec =

has been amplified at the middle of the

column by the factor Sec

when
2
the axial compressive load P is applied.
Note that when P approaches Pc and
hence approaches 1, this
amplification factor approaches infinity.

School of Aerospace, Mechanical and Manufacturing Engineering

15

MIET2072 2013/Ver. 1

Mechanical Design 2
The bending moment at the middle of
the column M max = Py max
M max = Pa

1
1

eqn. C6

The maximum compressive stress is the

sum of the average axial compressive
stress and the bending compressive
stress
i.e. max =

P M max c
+
A
I

P P ac 1
+
A A r 2 1
ac 1
= f ac + f ac 2
r 1

max

eqn. C8

ac
is sometimes called the
r2
imperfection ratio.

The term

f ac
and if it is
Foc
decided that the yield stress FY is the
limiting value of max and giving the
symbol FL to the consequent limit value
of the average compressive stress fac
gives
Recalling that =

FY = FL + FL

ac
r2

1
F
1 L
Foc

eqn. C10

Figure 8.1i shows plots of FL for

ac
different values of 2
r

Refer to Fig. 8.1i

M max = PeSec

eqn. C7

The maximum compressive stress is the

sum of the average axial compressive
stress and the bending compressive
stress
i.e. max =

where c is the distance from the neutral

axis to the outer fibre. Recalling
I = Ar 2
1
Pa
c
P
1

max = +
A
Ar 2
max =

The bending moment at the middle of

the column M max = P x moment arm max

P M max c
+
A
I

where c is the distance from the neutral

axis to the outer fibre. Recalling
I = Ar 2

c
PeSec
P

2
max = +
2
A
Ar
P P ec

max = +
Sec
2
A Ar
2
ec

max = f ac + f ac 2 Sec

2
r

eqn. C9

ec
is sometimes called the
r2
eccentricity ratio.

The term

P P / A f ac
=
=
, and if
Pc Pc / A Foc
it is decided that the yield stress FY is
the limiting value of max , and giving
the symbol FL to the consequent limit
value of the average compressive stress
fac gives
Noting that =

FY = FL + FL

ec

Sec
2
2
r

FL
Foc

eqn. C11

Figure 8.1j shows plots of FL for

ec
different values of 2
r

Refer to Fig. 8.1j

School of Aerospace, Mechanical and Manufacturing Engineering

16

MIET2072 2013/Ver. 1

Mechanical Design 2

Fig. 8.1i Initial crookedness RMIT University, 2013 (Dixon C., Marchiori G.)

Fig. 8.1j Eccentric load RMIT University, 2013 (Dixon C., Marchiori G.)

Comparing the two graphs (Fig.8.1i and 8.1j) and using the effective length l, it can
l
ac ec
be seen that for a given value of slenderness ratio , if 2 = 2 , the values of FL for
r
r
r

School of Aerospace, Mechanical and Manufacturing Engineering

17

MIET2072 2013/Ver. 1

Mechanical Design 2
each are very similar. When
of FL (the value of

l
= 0 both formulas C10 and C11 give the same value
r

P
that will cause yield to start at the outer fibre on the concave
A

l
> 0 , the curves of FL for the non-concentric case are slightly lower
r
than those for the initially crooked case. It is not surprising that these two situations
should produce similar results since the eccentrically loaded column might be
thought of as a column which first has an end moment Pe applied causing an initial
curvature like the initial crookedness, and then the axial load P is applied which
amplifies the deviation from vertical. Indeed comparing equations C8 and C9 it can
be seen that the two differences are that equation C8 has initial crookedness a
1

instead of eccentricity e and an amplification factor

instead of Sec
.
1
2
Perry and Ayrton [Ayrton, W.E., and Perry, J., On Struts , pgs 464 to 465 of The
Engineer, 10th December and 24th December 1886] compared these two
amplification expressions which is done in the table following.

side). When

P
Pc

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1

1
1

1.00
1.11
1.25
1.43
1.67
2.00
2.50
3.33
5.00
10.00

Sec

1.00
1.14
1.31
1.53
1.83
2.25
2.88
3.94
6.06
12.42

is
They observed that the amplification factor for the eccentric case Sec
2

6
slightly larger by a factor of approximately than that for the initially crooked
5
case. This greater (i.e. more detrimental) amplification factor for the eccentric case
is the reason that the limiting FL curves for that case are lower than those for the
initially crooked case.
Based on their observations from the table, Ayrton and Perry suggested that the
combined effects of the unintended eccentricity called here eo and initial
crookedness a could be expressed as a modified initial crookedness which is
given the symbol here of a , such that:

School of Aerospace, Mechanical and Manufacturing Engineering

18

MIET2072 2013/Ver. 1

Mechanical Design 2

5
a + eo
6

eqn. C12

They further suggested that inhomogeneity may be accounted for by a term of the
same kind. Putting a instead of a in equation C10 gives:
FY

FL + FL

a c
r2

1
F
1 L
F

This is in fact a quadratic in FL which can be re-arranged in the more conventional

form:

a c
FL2 FY + 1 + 2 Foc FL + FY Foc = 0
r

{ This will be recognised as the quadratic form:

a FL2 + b FL + c = 0 for which the roots are:

FL =

(b

2a

4a c

The smaller root of this quadratic provides the critical value of FL at which yielding
will occur at the outer fibre on the concave side of the column. The smaller root is:

FL

a c
a c

FY + 1 + 2 Foc
F
F
1
+
+

oc
Y
2

r
r
F F

Y oc
2
2

eqn. C12

School of Aerospace, Mechanical and Manufacturing Engineering

19

MIET2072 2013/Ver. 1

Mechanical Design 2
Important unknown parameters at this stage are the initial crookedness a and
unintended eccentricity eo, which are combined in the term a as in equation C12.
Another problem is that c and r2 are terms referring to a specific cross-sectional
a c
shape. The symbol is given to 2 and called the imperfection factor. Robertson
r
(Robertson 1925, p.55) 8 suggested that it be assumed that varied linearly with the
slenderness ratio l which bypassed the problems just mentioned. He further
r
l
suggested the value of = 0.003 would produce a graph for FL versus slenderness
r
ratio that would pass beneath the great majority of column test results then
available. The equation for FL has therefore become essentially an empirical
equation to fit the data although its origin was in the theoretical analysis just given.
Godfrey (Godfrey 1962, p.97-112) 9 later suggested the imperfection factor be
changed to:

l
0.00003
r

eqn. C13

and this is the value used in AS3990.

Equation C12 thus becomes:

FL

F + (1 + )Foc 2

FY + (1 + )Foc

F
F
Y
oc

2
2

eqn. C14

The graph of this equation is shown in Figure 8.1k, as the PERRY-ROBERTSON

CURVE .

Robertson, A., 1925, The Strength of Struts, I.C.E. Selected Engineering Papers, No. 28, p.55, The Institution of
Civil Engineers, London.,

Godfrey, G.B., 1962, The Allowable Stresses in Axially-Loaded Steel Struts, Structural Engineer Vol. 40., No. 3,
p.97-112,.The Institution of Structural Engineers, United Kingdom.,

School of Aerospace, Mechanical and Manufacturing Engineering

20

MIET2072 2013/Ver. 1

Mechanical Design 2

Fig. 8.1k Column Perry-Robertson limit RMIT University, 2013 (Dixon C., Marchiori G.)

P
This FL then is the value of average axial compressive stress f ac = at which
A
yielding is predicted to occur on the concave side of the column. For a safe design
F
the actual value of this average stress fac should be L where is a factor of

1
safety. In AS3990 = 1.67 =
. The maximum permissible average compressive
0.6
F
stress L is given the symbol Fac. Thus fac is to be Fac where:

Fac

FY + (1 + )Foc 2

1 FY + ( + 1)Foc

F
F

Y oc

2
2

eqn. C15

which is equation 6.1.1 in AS3990.

School of Aerospace, Mechanical and Manufacturing Engineering

21

MIET2072 2013/Ver. 1

Mechanical Design 2
Table 6.1.1 on page 42 of AS3990 conveniently gives values of Fac [MPa] for a range
of yield strengths FY [MPa] and slenderness ratios l/r. A sample of this data is given
below.
FY
l/r

50
60

125
118
109

230

122
114
104

240

130
123

127
118

480

Foc

249
218

235
200

790
548

653
467

70

Fac l/r 70

Fac l/r 75

113

107

181

164

403

351

80

98

92

101

95

148

133

308

273

For example if a concentrically loaded column is made from steel with a yield
strength of 230MPa and is of proportions such that it has a slenderness ratio l/r of
70, then its permitted average axial compressive stress is 109MPa. If it has a
slenderness ratio l/r of 75, then its permitted average axial compressive stress is
104MPa.
The two right hand columns of the table give values of Euler critical stress Foc for
various slenderness ratios. Recall that:
Foc

2E
2
l
r

( )

Activity 8F - Graphical representation of Fac

Graphs of the maximum permissible average axial compressive stress Fac for a
range of yield strengths and slenderness ratios may be viewed at Permissible
stresses in compression members (PDF 143KB) 10.
Note that the vertical axis has a change of scale when

l
= 100 .
r

Trahair, N. S., 1977, Permissible stresses in compression members, Compression members, The behaviour
and design of steel structures, Chapman and Hall, London, pp. 65, viewed on 28th August 2013
<https://equella.rmit.edu.au/rmit/file/e3770396-30b1-4301-8623-b978a9de13d1/1/130823_3_055.pdf>
10

School of Aerospace, Mechanical and Manufacturing Engineering

22

MIET2072 2013/Ver. 1

Mechanical Design 2
Fac is the usual maximum allowed value of the average axial compressive stress in
columns or struts, however extra restrictions are placed in Clauses 6.1.2 & 3 of
AS3990 on this stress for built up columns or struts, or on ones made from angle (to
avoid torsional buckling in the case of the angle).
Returning now to the equation for Fac in AS3990 (equation C15 in these notes). The
designer of a concentrically loaded column will typically know the load P that has
to be carried, the effective length l, for the set up, the yield strength (FY) and
Youngs modulus (E) of the steel proposed for use, and will probably have some
idea of a suitable cross sectional shape (e.g. square hollow section, I section, circular
hollow section, etc.)

Activity 8G Sections used for Columns

Step 1: Typical sections used for columns and struts
Refer to figure 6.3 on page 118 of The Steel Designers Handbook (PDF 80KB) 11
for an overview of typical sections used for columns and struts.
Step 2: Safe axial load for various sections
Refer to table 6.1 on page 119 of The Steel Designers Handbook (PDF 80KB) 12
The table compares the safe axial load for various sections, each having the same
cross sectional area, and operating with the same effective length.

11 Gorenc, B. E. & Tinyou, R., 1981, figure 6.3, Steel designers handbook, 5th ed., New South Wales University
Press, Kensington, N.S.W., p.118, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/d817860a-baa5-40b2-9722-ed3e38630d2e/1/130823_3_064.pdf>

Gorenc, B. E. & Tinyou, R., 1981, Table 6.1, Steel designers handbook, 5th ed., New South Wales University
Press, Kensington, N.S.W., p.119, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/d817860a-baa5-40b2-9722-ed3e38630d2e/1/130823_3_064.pdf>
12

School of Aerospace, Mechanical and Manufacturing Engineering

23

MIET2072 2013/Ver. 1

Mechanical Design 2
Having made some trial decision on shape, what the designer has to find next is the
actual section size that can carry the load P without generating an average axial
compressive stress in excess of Fac.

P
i.e. f ac = must be Fac
A
i.e.

F + (1 + )Foc 2

P 1 FY + ( + 1)Foc

F
F

Y

Y oc

2
2
As

where

( r)

0.00003 l

2E
and Foc
=
2
l
r
Looking at these three equations, it can be seen that the unknown quantities are the
effective cross sectional area As and the appropriate radius of gyration . [If l is
different in one plane of buckling to another, the plane that has the larger value of
slenderness ratio l is the one in which buckling is likely to take place so the r
r
value is the one appropriate to that direction of buckling. Clearly if l is the same for
both planes of buckling the appropriate radius of gyration is the minimum one as it
will produce the larger value of slenderness ratio.]

( )

One possible approach is to guess a size perhaps based on experience, and check
that fac is Fac . If for the size guessed fac is << Fac then the size chosen is wasteful
and a smaller section could be checked for suitability. In view of the fact that
column bases are often in moist environments and therefore prone to corrosion it
could be argued that having extra metal thickness to compensate for such corrosion
over the life of the structure would be prudent. A sound maintenance regime of
course aims to prevent such corrosion.
A second approach, which tries to reduce the number of trial selections, makes use
of charts such as those shown in Activity 8H which follows.
Safe load tables for concentrically loaded columns as are provided by many
manufacturers; please see Appendix 8 for some examples.

School of Aerospace, Mechanical and Manufacturing Engineering

24

MIET2072 2013/Ver. 1

Mechanical Design 2

Activity 8H Design charts to aid column size selection

Step 1: Design chart for selecting size of Australian universal section
compression members
Refer to Fig. 3.15 on page 66 of The behaviour and design of steel structures
(PDF 132KB) 13 to view an example of a design chart for selecting size of
universal section compression members.
Using this chart one could enter the vertical axis with the desired axial working
load P, and the horizontal axis with the effective length and hence find the
smallest universal section size that could carry the load.
Step 2: Determination of approx. values for Fac for columns of various sections
Refer to Figure 6.10 on page 136 of Steel designers handbook (PDF 84KB) 14
In Figure 6.10 one could enter the horizontal axis with the parameter combining
2
effective length and desired working load l , go to the sloping line
P
representing the section type being considered then move across to the vertical
axis to find the approximate maximum permitted average axial compressive
stress Fac* . The approximate required effective cross sectional area can then be
found from:
P
Fac
fac
=
As

As

P
Fac*

13 Trahair, N. S., 1977, The behaviour and design of steel structures, 1st ed., Chapman and Hall, London, p.66,
viewed on 28th August 2013 < https://equella.rmit.edu.au/rmit/file/929639e4-619e-4279-ac449f927f920a5c/1/130823_3_056.pdf>

Gorenc, B. E. & Tinyou, R., 1981, Figure 6.10, Steel designers handbook, 5th ed., New South Wales University
Press, Kensington, N.S.W., p.136, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/2277a2a2-1f72-47c8-ba66-2dcf359af72f/1/130823_3_065.pdf>
14

School of Aerospace, Mechanical and Manufacturing Engineering

25

MIET2072 2013/Ver. 1

Mechanical Design 2
One would then find a section of the desired shape that had an effective area that
predicted as being required. Having found this size one would then determine fac
using its effective area As, and Fac using its effective radius of gyration r, and check
that f ac Fac . Recall also that Clauses 6.1.2 and 6.1.3 list special limiting provisions on
the maximum allowed average compressive stress for built-up columns or struts, and
columns or struts made of angle.
Note that the effective cross sectional area As is calculated from the specified sizes
of the member cross section deducting:
a) All holes other than bolt holes (which are not deducted for members in
compression).
b) Excess width as given in Clause 4.3 of AS3990. This deduction is related to
b
plate with excessive width to thickness ratios being prone to local
t
buckling before it reaches the yield stress and therefore the excess width is
not regarded as useful. Cross sectional areas are often manufactured with
proportions such that this is not a concern. Manufacturers tables usually
indicate the value of the effective and gross cross sectional areas, and may
also indicate where they differ (see Appendix 8).

b
t

Activity 8I Excess width determination

To aid in the determination of excess widths Gorenc and Tinyou have prepared
some helpful diagrams in Fig. 6.11 of their Steel designers handbook(PDF 81KB) 15
The sketches there show excessive section areas for compression members with
large width to thickness ratios. The small calculations shown are for steel with FY of
250 MPa (e.g. for the flange on the I beam in the left most sketch. Clause 4.3.1(a)
says in essence projection b1 greater than 256T1 FY shall be neglected when
calculating the effective properties of the section; 256T1

250 16T1 ).

Gorenc, B. E. & Tinyou, R., 1981, Figure 6.11, Steel designers handbook, 5th ed., New South Wales University
Press, Kensington, N.S.W., p.136, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/2277a2a2-1f72-47c8-ba66-2dcf359af72f/1/130823_3_065.pdf>
15

School of Aerospace, Mechanical and Manufacturing Engineering

26

MIET2072 2013/Ver. 1

Mechanical Design 2
There are two noteworthy effects of the effective cross sectional area being less than
the actual cross sectional area:

P
increases;
firstly the calculated average axial compressive stress f ac =
A
s

secondly, the effective radius of gyration r changes and may decrease,

particularly when the neglected area is far from the centroid, and if r is less
then the effective slenderness ratio l increases and therefore the allowed
r
maximum value of average compressive stress Fac is reduced.

( )

All things considered it would seem better, as regards use of metal, to avoid use of
sections with excessive plate width to thickness ratios.
With regard to the appropriate effective radius of gyration of the chosen size of
section, it must not be so small that the resulting slenderness ratio exceeds the
limits allowed by Clause 4.6 of AS3990.

Section 8.2 Eccentrically Loaded Columns or Compression

Members
It is often not possible for a vertical load to be transmitted to a column such that it
acts axially through the centroid. The column is therefore subjected to an eccentric
load which generates a bending moment. This situation is shown schematically in
Figure 8.2a.

Fig. 8.2a: Column with eccentric load RMIT University, 2013 (Dixon C., Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

27

MIET2072 2013/Ver. 1

Mechanical Design 2
Refer to Clause 6.4 of AS3990 for comments on appropriate choices for the amount
of eccentricity to be included in calculations in certain structural situations.
A column may also be subjected to end couples form some other source than an
eccentric load, and be simultaneously subjected to an axial load. When studying
concentrically loaded columns consideration was given to the effect of small
unintended eccentricity and initial crookedness both of which finally ended up as
part of the imperfection parameter in the expression for the allowed average axial
compressive stress Fac. One of the important observations in that topic was the
amplifying effect the axial load had on the bending moment. It is helpful now to
revisit some of the main results of that study and to look at the amplifying effect
the axial load can produce in some other situations. This is done in the table on the
next page, which summarises results from analyses done by Timoshenko and Gere.
(Timoshenko & Gere 1963) 16
Regarding the amplification, caused by the axial load, of the central deflection,
Timoshenko & Gere also observed (Timoshenko & Gere 1963) that the three
amplification factors (u), (u) and (u) could be approximated with good
accuracy by an amplification factor

1
1

eqn. C16

P
Pc

if the ratio

P
is not large.
Pc

16 Timoshenko, S., Gere, J M., 1963, Beam-columns, Theory of elastic stability, 2nd ed., International student ed.,
McGraw-Hill, Auckland.

School of Aerospace, Mechanical and Manufacturing Engineering

28

MIET2072 2013/Ver. 1

Mechanical Design 2

Central deflection before

axial load applied

Amplification factor for

central deflection caused
by axial load

Initially crooked plus axial

compressive load

Equal and opposite couples at each

end plus axial compressive load

Central concentrated lateral load

plus axial compressive load

Uniformly distributed lateral load

plus axial compressive load

Ml 2
8 EI

Ql 3
48 EI

5ql 4
384 EI

3
Tan u u
u3

12
2 Sec u 2 u 2
4
5u

called ( u )

called ( u )

1
1

P
Pc

Angular deflection at ends

before axial load applied

Amplification factor for

angular deflection at ends
caused by axial load

Central bending moment

before axial load applied

Central bending moment

after axial load applied
Amplification factor for
central bending moment
caused by axial load

2
Sec u 1
u2

where u =
2

called ( u )

P
Pc

1
Pa
1 P

Pc

not applicable

P
Pc

Ml
2 EI

Ql 2
16 EI

ql 3
24 EI

Tan u
u

(u )

(u )

Ql
u

ql
8

M Sec u

Ql Tan u
u
u

ql 2
Sec u 1
8 u2

Sec u

Tan u
u

(u )

School of Aerospace, Mechanical and Manufacturing Engineering

29

MIET2072 2013/Ver. 1

Mechanical Design 2

P
less than 0.6, the error in the approximate expression for the
Pc
three amplification factors (u), (u) and (u) is less than 2%.
They note that for values of

Regarding the amplification, caused by the axial load, of the central bending moment,
Tan u

2
and (u ) = 2 [Sec u 1] may also be
the three amplification factors Sec u,
u

u
1
P
approximately represented by
, if the ratio
is not large as may be seen in the
P
P
c
1
Pc
table below which shows various amplification factors as a function of 2u, where
P
. Data in the last three columns is drawn from Timoshenko and Gere (1963)
u=
2 Pc
It is noteworthy that as P approaches Pcrit the amplification factors approach infinity.
1

P
Pc

2u

2u
=

P
Pc

Sec u

1
2u
1

Tan u
u

(u )
=

2
(Sec u 1)
u2

(u )

(u )

3
[Tan u u ]
u3

12
2 Sec u 2 u 2
5u 4

1.000

1.000

1.000

1.000

1.000

1.000

0.2

0.004

1.004

1.005

1.003

1.004

1.004

1.004

0.4

0.016

1.016

1.020

1.014

1.016

1.016

1.016

0.6

0.036

1.038

1.047

1.031

1.038

1.037

1.037

0.8

0.065

1.069

1.086

1.057

1.073

1.068

1.070

1.0

0.101

1.113

1.139

1.093

1.117

1.111

1.114

1.20

0.146

1.171

1.212

1.140

1.176

1.169

1.173

1.40

0.199

1.248

1.307

1.203

1.255

1.245

1.250

1.60

0.259

1.350

1.435

1.287

1.361

1.346

1.354

1.80

0.328

1.489

1.609

1.400

1.504

1.482

1.494

2.00

0.405

1.681

1.851

1.557

1.704

1.672

1.690

2.20

0.490

1.962

2.205

1.786

1.989

1.949

1.962

2.40

0.584

2.402

2.760

2.143

2.441

2.382

2.400

2.60

0.685

3.174

3.738

2.771

3.240

3.144

3.181

2.80

0.794

4.863

5.883

4.141

4.938

4.808

4.822

2.90

0.852

6.762

8.299

5.681

6.940

6.680

6.790

3.00

0.912

11.350

14.137

9.401

11.670

11.201

11.490

School of Aerospace, Mechanical and Manufacturing Engineering

30

MIET2072 2013/Ver. 1

Mechanical Design 2
The approximate substitution of

1
P
1
Pc

for Sec u and

Tan u
is not as accurate as it is
u

for (u ) [ and (u ) and (u )] . For example when

1
P
is 0.6 the expression
P
Pc
1
Pc
Tan u
by 12%, and underestimates
underestimates Sec u by 13%, overestimates
u
(u) by 1.8%.
It is worth recalling at this point the observation made by Ayrton and Perry
(Ayrton, W.E., and Perry, J., On Struts in The Engineer, 10th December and 24th
December 1886.), that was discussed earlier when considering the effects of initial
crookedness and unintended eccentricity, that:
1
1

P
Pc

Sec
6
2

P
Pc

(i.e.

5
Sec u )
6

The reader is referred to Chapter 1 of Theory of Elastic Stability (Timoshenko &

Gere 1963) for discussion and data on the amplifications that occur when a
compressive axial load is applied in beam-column situations with: couples at each
end that differ in size and sense; several concentrated lateral loads; and variable
distributed load.
Brief consideration will be given here to a column with moments of equal
magnitude at each end and an axial compressive load.
For a column such as is shown again in Figure 8.2b(i), loaded with equal
eccentricities at each end, the magnitude of the moment at each end is Pe and their
sense is opposite (i.e. one clockwise and one anti-clockwise). They produce single
curvature bending.

School of Aerospace, Mechanical and Manufacturing Engineering

31

MIET2072 2013/Ver. 1

Mechanical Design 2

Fig. 8.2b: Effect of equal and opposite sense end moments RMIT University, 2013 (Dixon C., Marchiori G.)

This eccentric loading could also be represented by a couple M of magnitude Pe at

each end, and an axial compressive load as shown in Figure 8.2b(ii).
As has been previously shown, the maximum deflection is at the centre and is
M
[Sec u 1] . The maximum bending moment is at
e[Sec u 1] which could be written
P
the centre and is formed from:
Mmax =

M + P x maximum deflection
M
=
M + P x [Sec u 1]
P
M Sec u

Mmax =
1
& Mmax
M
P
1
Pc
i.e. the axial compressive load has caused amplification of the bending moment by a factor of
1
approximately
P
1
Pc

School of Aerospace, Mechanical and Manufacturing Engineering

32

MIET2072 2013/Ver. 1

Mechanical Design 2
If the moments at each end are of equal magnitude and both of the same sense (e.g
both clockwise) then the column has a double curvature shape as is shown in Fig 8.2c.

Fig. 8.2c: Effect of equal and same sense end moments RMIT University, 2013 (Dixon C., Marchiori G.)

Instead of a single maximum deflection at mid position there are now two; one at
and one at position. Furthermore it will be appreciated that if the moments are of
the same magnitude as was the case in Fig 8.2b then the maximum deflections in Fig
8.2c, will be less, and consequently the moment amplification factor will have a value
1
..
less than
P
1
Pc
The moment amplification factor for beam columns such as shown in Figure 8.2b and
1
8.2c can thus be represented as C m
where Cm is a number 1.
P
1
Pc

School of Aerospace, Mechanical and Manufacturing Engineering

33

MIET2072 2013/Ver. 1

Mechanical Design 2
If the moments at each end are different in magnitude and/or sense then the shape
taken by the beam column is more complicated, and the maximum deflection from the
centre line may not be at mid height, as shown in Activity 8J .

Optional Activity 8J Further reading on moment amplification in Beamcolumns

In-plane bending moment distribution in a beam-column where moments at
each end are different in magnitude and/or sense
Refer to Fig 7.3 on page 239 of the The behaviour and design of steel structures
(PDF 68KB) 17
The extra symbol used here can have any value between 1 (single curvature
bending) and +1 (double curvature bending). So in Figure 8.2b = 1. The end
moments shown in the Figure 7.3 in this reference, where the column has pin
jointed ends, are applied end moments not reactive end moments from built-in
ends and the like (which affect effective length and hence critical load).
For situations where the end moments differ it can be shown (Trahair 1977) that:
(i) Mmax =

M 1 + ( Cosec 2 u + Cos 2 u )

when < Cos 2 u ,

and the point of maximum moment lies in the span
(ii) Mmax =

M
when > Cos 2 u ,

1
The expression in (i) simplifies to M max = M Secu M

P
1

Pc

which is the maximum value Mmax can have.

when = 1 ,

Trahair, N. S., 1977, Beam-columns and frames, The behaviour and design of steel structures, Chapman and
Hall, London, p. 239, viewed on 28th August 2013 < https://equella.rmit.edu.au/rmit/file/4dee5467-64484047-87f6-cfa947033e08/1/130823_3_059.pdf>
17

School of Aerospace, Mechanical and Manufacturing Engineering

34

MIET2072 2013/Ver. 1

Mechanical Design 2
Obviously since the bending moment has been amplified by the presence of the axial
compressive load so too will the resulting bending stress be amplified compared to
what it would be were the axial load not present. Thus:

M max
M

bending max =

f bc

is the bending stress that would exist if the axial compressive load
where
fbc
was not present
fbc

M
where Z is the appropriate section modulus
Z

Therefore when the end moments are equal and opposite:

bending max

P
1
Pc

f
bc

and when the end moments are other than equal and opposite bending max will not be
amplified as much, hence:

bending max

A fraction 1
depending on the
1

relative magnitudex
and sense of the 1 P
Pc

end moments

f
bc

In the case of a beam without an axial compressive load the bending compressive
stress fbc would ordinarily be compared to the maximum permissible stress due to
bending (in a member not subjected to axial force) Fb, and the maximum
permissible compressive stress due to bending (in a member not subjected to axial
force) Fbc.

School of Aerospace, Mechanical and Manufacturing Engineering

35

MIET2072 2013/Ver. 1

Mechanical Design 2
However in a beam-column there is an axial force, the major effects of which are:
(i) to cause the bending moment and consequently the bending stress to be
amplified

(ii) to cause an average axial compressive stress f ac =

which is
Effective Area
superimposed on the bending compressive stress. [Recall that in a concentrically
loaded column this stress would be compared to Fac, the maximum average axial
permissible compressive stress in an axially loaded strut not subjected to bending.]
AS3990 provides equations in Clause 8.3.1 Axial compression and bending which
enable stresses to be checked in the situation where there are the combined
interacting effects of bending about two axes, and an axial load imposed on the
beam-column.
One such equation is this interaction or unity equation:

1
C mx

f ac

0.6 Focx
f ac

+
Fbcx
Fac

1
f C
bcx my
f ac
1

0.6 Focx

+
Fbcy

f
bcy

1 eqn. 8.3.1(a)1 AS3990

f ac
expresses as a fraction the comparison of the average
Fac
axial compressive stress fac to the lesser maximum permissible average axial
compressive stress Fac about either principal axis as found in Section 6 of AS3990
(and discussed in these notes when considering concentrically loaded columns).
The first group of terms

The second and third group of terms deal with bending stress about the x and y
axis respectively.

School of Aerospace, Mechanical and Manufacturing Engineering

36

MIET2072 2013/Ver. 1

Mechanical Design 2
Considering the second group of terms: fbcx is the maximum bending stress about
the x axis, occurring either at or between braced points along the beam-column,
calculated without inclusion as yet of the amplifying effect of the axial compressive
force. The bending stress fbc is calculated using the properties of the section based
on the effective cross sectional area.

1
The factor

f ac
1
0.6 Focx

could be written as

P
1
0.6 Pcx

which makes it

recognisable as the approximate amplification factor hitherto discussed with the

inclusion of a load factor of 0.6 which tends to increase the amplification. [The
authors interpretation of this 0.6 is that when fac = 0.6 Focx, i.e. P = 0.6 Pcx, then P is
at the maximum value allowed for a theoretically perfect column (imperfection
1
, and there is hence no spare capacity
parameter = 0) if the factor of safety is
0.6
in the column to deal with any bending stress. The amplification term as shown
then becomes so that the inclusion of any bending stress fbc would make it
impossible to satisfy the unity equation.]
As will be recalled application of the full amplification factor was appropriate
when there are equal and opposite end moments (and hence single curvature), but
for other situations with end moments this can be moderated by a fraction 1
depending on the relative magnitude and sense of the end moments. Cmx is such a
moderating fraction. The standard gives a variety of values for different
circumstances, and the reader is referred thereto; the range of Cmx and Cmy varies
from 0.4 to 1.0. For the circumstance where the member is in a frame where
sideways is prevented, and there is no lateral (transverse) loading between the
members supports in the plane of bending, then linearised expressions for Cm may
be used. These equation are shown in the following optional activity. They give
values of Cm between 0.4 and 1 . Note that if the column is not braced against
sidesway then Cm should be taken as 0.85 for all the ratios of bending moments.
However for simplicity the conservative approach of putting Cm=1 may be taken;
and this is the approach recommended by Bednar (1986) 18 for pressure vessel
support columns.

18

Bednar H 1986, Pressure vessel design handbook 2nd ed. Van Nostrand Reinhold, New York, p.148.

School of Aerospace, Mechanical and Manufacturing Engineering

37

MIET2072 2013/Ver. 1

Mechanical Design 2

Optional Activity 8KLinearised expression for Cm

Step 1: Variation of Cm with End Moment Condition
Refer to Fig 131.1, Variation of Cm with moment on page 46 of Source book for
the Australian Steel Structures Code (PDF 73KB) 19
Step 2: Values of the co efficient Cm for columns subject to bending with axial
compressive force
Refer to Fig 6.14 on page 141 of the Steel designers handbook (PDF 44KB) 20
Figure 6.14 provides information for obtaining Cm for a variety of loading
configurations.

The terms Fbcx and Fbcy are the maximum permissible compressive bending stresses
for bending about each axis. These are obtained from Section 5 of AS3990. In
particular for use in equation 8.3.1(a)(1) Fbcx and Fbcy are derived using Clauses 5.2,
5.3 and 5.4 (if applicable).
Values for Fbc for some commercial square hollow sections may be found on the last
page of the extract in Appendix 8.
Clause 5.2 is based on considerations of yield, Clause 5.3 on consideration of local
buckling, and Clause 5.4 on considerations of flexural-torsional buckling that may
occur when a beam is bent about its axis of maximum strength.

Lay, M. G., 1982, Fig 131.1 variation of Cm with movement condition, Source book for the Australian Steel
Structures Code, AS 1250, 3rd ed., Australian Institute of Steel Construction, Milsons Point, N.S.W, p. 46.,
viewed on 28th August 2013 < https://equella.rmit.edu.au/rmit/file/b0a21c25-b0b2-4522-9b88569ae6750d09/1/130823_3_053.pdf>
19

Gorenc, B. E. & Tinyou, R., 1981, Figure 6.10, Steel designers handbook, 5th ed., New South Wales University
Press, Kensington, N.S.W., p.141, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/4cdd940b-30a9-4f8f-8328-bdc3f16d21a9/1/130823_3_066.pdf>
20

School of Aerospace, Mechanical and Manufacturing Engineering

38

MIET2072 2013/Ver. 1

Mechanical Design 2
Clause 5.2 limits the tensile or compressive bending stresses Fb to 0.66 FY for all
beams other than solid round and square bars and solid rectangular bars bent
about the axis of minimum strength for which the limit is 0.75 FY.
With respect to the limit of 0.66 FY, it is noteworthy that this is higher than the limit
set as the average axial compressive stress in stocky columns (small l ) of 0.60 FY
r
1
which was putting a factor of safety of
against yielding. In the case of beams
0.6
the limit 0.66 FY set on the outer fibre stress corresponds approximately to a factor
1
against yielding right across the section, i.e. the beam being fully
of safety of
0.6
plastic (i.e. a plastic hinge). For most rolled sections the plastic section modulus S
is 1.1 to 1.2 times the elastic section modulus Z so taking conservatively 1.1 the
fraction 0.66 = 1.1 x 0.6 is arrived at. Similarly the limiting value of 0.75 FY for solid
bars conservatively reflects the ratio of plastic section modulus to elastic section
modulus for these.
The values of Fb may not however be the upper limit permissible, the limit having
to be lowered if local buckling or flexural-torsion buckling are issues, as dealt with
by Clauses 5.3 and 5.4 respectively.
Earlier in this chapter local buckling was discussed and it was shown that plate
elements loaded in compression were liable to buckle locally before yielding if their
b
width to thickness ratios were larger than certain limits. Limits were therefore
t
imposed on this ratio and widths that caused these limits to be exceeded were
regarded as excessive and the excess portion of width was not to be included in the
calculation of effective cross sectional area.
If a flange has a width that is right on the limit then the compressive stress if large
enough would cause yield and exposure to the likelihood of local buckling to occur
simultaneously. Such flanges therefore have the permitted compressive stress
b
limited to 0.60 FY. If the flange ratio is less than the upper limit then the
t
compressive stress may be more than 0.60 FY, the allowed value gradually
b
b
increasing to 0.66 FY as gets less. This 0.66 FY limit is allowed once the ratio
t
t
has fallen to the point where were the factor of safety 1.0 the compressive stress
could theoretically get high enough without local buckling happening to allow the
beam to yield right through [i.e. a plastic hinge]. These limits on compressive
b
stress as affected by ratio, are dealt with in Clause 5.3 in AS3990.
t

School of Aerospace, Mechanical and Manufacturing Engineering

39

MIET2072 2013/Ver. 1

Mechanical Design 2

Optional Activity 8L - Extra reading on limits to permitted bending

compressive stress as affected by b/t ratio.
b
ratio, are
t
illustrated graphically for the case of the flange on an I section (i.e. plate element
supported on one side only) in Fig 45 of the following extract:
Reduction of Permitted Compressive Bending Stress due to local buckling for an
I section (PDF 102KB) 21

These limits on permitted compressive bending stress Fbc as affected by

For flanges in box sections, i.e. in which the plate is supported on both sides, the
scale on the horizontal axis of Fig 45 is altered by a factor of:
560
for welded flanges or plates that have not been stress relieved
256
800
for those that have been stress relieved
256
635
for cold formed rectangular hollow sections
256

Fbc would have to fall below 0.60 FY if the bending stress fbc were calculated based
on the full cross sectional area rather than the effective cross sectional area, once
b
the limit associated with local elastic buckling is exceeded. This falling away is
t
(b t )limit
described by Fbc = 0.60 FY
and is illustrated in Figure 4 in the following
(b t )
extract:
Reduction in Maximum Permissible Compressive Stress in Bending Fbc (PDF
120KB) 22

Lay, M. G.,1982, Comments 41-88, Source book for the Australian Steel Structures Code, AS 1250, 3rd ed.,
Australian Institute of Steel Construction, Milsons Point, N.S.W., p.18, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/9d342f52-3ab6-4eca-b7ea-ba3d314853ff/1/130823_3_052.pdf>
21

Australian Institute of Steel Construction 1974, Flange local buckling, Part 1 - Design if beams, design of
columns (metric units) : Steel design course : a short course on design of steel structures with special reference
to AS 1250 SAA steel structures code 2nd ed., Australian Institute of Steel Construction, Sydney, p. 4-5, viewed
on 28th August 2013 < https://equella.rmit.edu.au/rmit/file/55ba3166-d3f5-449d-b3769a6373d60125/1/130823_3_054.pdf>
22

School of Aerospace, Mechanical and Manufacturing Engineering

40

MIET2072 2013/Ver. 1

Mechanical Design 2
The limit on compressive stresses for rolled sections, as calculated on effective
section properties that may have had to be lowered from 0.66 FY to 0.60 FY to
guard against local buckling, may have an even lower limit imposed on it if
flexural-torsional buckling is a potential mode of failure. Flexural-torsional
buckling of beams (sometimes referred to as lateral buckling of beams) is the
twisting sideways that can occur when a beam is bent about the axis of major
strength.

Activity 8M - Flexural-torsional buckling of beams

View an example of flexural-torsional buckling of beams which is illustrated in
Figure 8 in the following extract:
Torsional flexural buckling of a simply supported beam without immediate
lateral restraints (PDF 53KB) 23

Clause 5.4 of AS3990 deals with this issue. A maximum allowed bending stress is
determined that will, with an appropriate factor of safety, guard against this mode
of failure. The calculation of this permitted stress takes into consideration the ratios:
depth of section (parallel to web)
flange thickness

and

effective length
radius of gyration about the axis of minimum strength

Obviously if a section having equal strength about each axis, such as a square
hollow section, is chosen for use as a beam column then the issue of this mode of
failure would not arise.

Hosking A K & Harris M, 1981, Structural design to AS 1250-1981, Applied mechanical design 2nd ed., H and
H Publishing, Forest Hill, Vic. p.12.7, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/c2da2008-78aa-4360-b1d0-041a6a589255/1/130821_3_034.pdf>
23

School of Aerospace, Mechanical and Manufacturing Engineering

41

MIET2072 2013/Ver. 1

Mechanical Design 2
Returning now to the unity of equation 8.3.1(a)(1), we see that the second and third
group of terms are essentially comparing for each axis of bending, the appropriately
amplified compressive bending stress to the allowed compressive bending stress, said
comparison being expressed as a fraction. Thus for the whole equation it can be seen
that:

appropriately amplified compressive appropriately amplified compressive

average axial

compressive stress
bending stress for bending about x axis bending stress for bending about y axis

maximum permitted average maximum permitted compressive maximum permitted compressive

axial compressive stress bending stress for bending about bending stress for bending about

x axis
y axis
1

If the average axial compressive stress is comparatively low ac < 0.15 then the
Fac

amplification of bending stress that the axial load produces would be small and is
disregarded, giving:

f ac f bcx f bcy
+
+
1
Fac Fbcx Fbcy

eqn. 8.3.1(a)(2)

Equations 8.3.1(a)(1) and (2), as appropriate, are used for checking the member as a
whole, using the maximum values of fbcx and fbcy that occur along it. A further check is
also done at the support, using the values of fbcx and fbcy at the support:
f bcy
f ac
f
+ bcx +
1
0.60 FY Fbcx Fbcy

eqn. 8.3.1(b)

For this equation Fbcx and Fbcy are derived using Clauses 5.2 and 5.3 only (i.e. flexuraltorsional buckling is not limiting the allowed compressive stress at the support ends
of the beam-column).
Unity equations such as the three above are sometimes thought of as:
fraction of column' s strength
fraction of column' s strength fraction of column' s strength

+
+
1
devoted to axial compressive load devoted to bending about x axis devoted to bending about y axis

School of Aerospace, Mechanical and Manufacturing Engineering

42

MIET2072 2013/Ver. 1

Mechanical Design 2
The designer must therefore choose a size and type of steel section whose
properties i.e. yield strength, effective cross sectional area, effective section
modulus, and radii of gyration are such that in combination with the loads and
effective length of the set-up, the stresses generated within the section are such as
to satisfy the interaction unity equations.
As earlier shown in Activity 8H, charts are available to help make trial selections of
column section area to safely carry axial loads, however where the column is
subject to end moments as well that section area would need to be increased. The
process typically involves trial selection aided by past experience and subsequent
checking as described above to make sure the stresses generated within the trial
section are such as to satisfy the interaction unity equations
For the case of the columns to be used to support the lugs on the pressure vessel in
Question 6(b) of Project Part B it is suggested you look at the relative proportions of
the column and lug widths in Figure 3.24(a) in AS1210 and similarly in Fig 8.2d
below. In your response to Question 6(a) the lugs have already been selected to be
of adequate strength, and their associated gusset spacing, shown as gin fig 8.2d,
is hence known. Selecting a trial column breadth B to match gwill provide
column walls aligned for good weight transfer from the lug.

Fig. 8.2d: Column to lug trial proportionality RMIT University, 2013 (Dixon C., Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

43

MIET2072 2013/Ver. 1

Mechanical Design 2
Appendix 8 gives sample commercial catalogue data of square hollow sections that
can be used to select such a trial column.
Safe load tables for axial loads are also given in that sample data, however it must
be remembered that the loading situation for this capped column is not simple
axial, rather it is eccentric loading and must be analysed using the relevant
interaction unity equation. The worst possible position for the vertical load on the
column cap is that shown in Fig 8.2e producing biaxial bending, hence all three
additive terms need to be included in the relevant interaction unity equation.

Fig. 8.2e: Possible eccentricity on column cap under lug RMIT University, 2013 (Dixon C., Marchiori G.)

School of Aerospace, Mechanical and Manufacturing Engineering

44

MIET2072 2013/Ver. 1

Mechanical Design 2

Appendix 8A - Sample Commercial data of square hollow sections

Refer to the sample commercial data of square hollow sections on pages 2, 9, 16
and 24 of the Tubeline: safe load tables from Tubemakers of Australia Limited.
(PDF 404KB). 24 Use it as appropriate in Question 6b of Project Part B.

24 Tubemakers of Australia Limited, Steel Pipe Division, 1984, Tubeline : safe load tables 2nd ed., Milsons Point,
N.S.W. : Steel Pipe Division Tubemakers of Australia, p. 2, 9, 16, 24, viewed on 28th August 2013 <
https://equella.rmit.edu.au/rmit/file/00615f08-73a5-40c6-9afe-53b9d36d622d/1/130826_3_069.pdf>

School of Aerospace, Mechanical and Manufacturing Engineering

45

MIET2072 2013/Ver. 1

Mechanical Design 2

Activity 8N - Reading, calculations and drawing

Step 1: Refer to Project Part B in Assessment
Refer to Project Part B in the Assessment Section. Read Question 6 (b) and (c).
Step 2: Have to hand relevant parts of AS3990
If you have not already done so, print out pages 20, 21, 22, 23, 40, 42, 54 and 55 of
AS3990 Mechanical Equipment Steelwork This Australian Standard can be
accessed via the RMIT Library - SAI Global link to the Australian Standards. 25
NOTE: You will need to login with your student login and password.
Step 3: Plan.
Discuss with your partner the strategy for tackling Question 6 (b) and (c).
The essence of Question 6 (b) is checking that your proposed column satisfies the
appropriate unity inter-active equation. One such equation is 8.3.1(a)( 1) in
AS3990. There are many terms in this equation, which have been discussed in
chapter 8. Some tips are: Values for Fbc for some commercial square hollow
sections may be found on the last page of the extract accessible in Appendix 8.
Fac and Foc may be found in Table 6.1.1. in AS3990. Cmx and Cmy may be taken as
1.0; this is a conservative approach. fac is the axial compressive stress in your
column which you must calculate. fbcx and fbcy are the unamplified bending
compressive stresses induced in your column by the eccentricity of the load,
which you must calculate.
Step 4: If you are the partner responsible for completing Question 6 (b), do so,
consulting with your partner where appropriate.
Step 5: If you are the partner responsible for completing Question 6 (c), do so,
consulting with your partner where appropriate.
Feedback:
Feedback will be provided on your submitted project documentation by the
engineering lecturer/tutor responsible for marking it.

25

https://login.ezproxy.lib.rmit.edu.au/login?url=http://www.saiglobal.com/online/autologin.asp

School of Aerospace, Mechanical and Manufacturing Engineering

46

MIET2072 2013/Ver. 1

Mechanical Design 2

Summary and Outcome Checklist

Tick the box for this statement if you agree with it:
I can determine the proportions of supporting columns including
consideration of buckling, yield, eccentric load and initial crookedness.

Assessment
This topic will be assessed as part of the Project Part B and the end of semester
examination (see: Assessment section of the Course Introduction for more detail).

School of Aerospace, Mechanical and Manufacturing Engineering

47

MIET2072 2013/Ver. 1