Australian Journal of Structural Engineering

ISSN: 1328-7982 (Print) 2204-2261 (Online) Journal homepage: http://www.tandfonline.com/loi/tsen20

Review of the second-order moment amplification

factors in AS4100 for the system design approach

C. K. Iu

To cite this article: C. K. Iu (2017): Review of the second-order moment amplification factors
in AS4100 for the system design approach, Australian Journal of Structural Engineering, DOI:
10.1080/13287982.2016.1270627

To link to this article: http://dx.doi.org/10.1080/13287982.2016.1270627

Published online: 04 Jan 2017.

Submit your article to this journal

Article views: 9

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tsen20

Download by: [Fudan University] Date: 15 January 2017, At: 13:39

Australian Journal of Structural Engineering, 2017
http://dx.doi.org/10.1080/13287982.2016.1270627

Review of the second-order moment amplification factors in AS4100 for the

system design approach
C. K. Iu
School of Civil Engineering and Built Environment, Queensland University of Technology, Brisbane, Australia

ABSTRACT ARTICLE HISTORY

The first-order elastic analysis cannot capture the second-order effects owing to the change in Received 29 November 2015
structural geometry, such as P-Δ and P-δ effect, which can cause the second-order moment in Accepted 4 December 2016
order to deteriorate the stiffness of a structure. Unfortunately, the first-order elastic analysis is
KEYWORDS
prevalent in the design office and the current practical engineers have not yet well equipped Moment amplification factor;
for manipulating the advanced analysis for design purpose. To this end, AS4100 stipulates the braced frame; sway frame;
design calculation of moment amplification factor for the braced and sway member continuous elastic buckling load; system
in a framed structure in Section 4 of the method of structural analysis as a system approach; design approach
specifically for an axially compressive loaded column in either a braced or sway frame. However,
in this study, it is shown that the moment amplification factors as a system approach is not as
effective and sophisticated as the member design approach. In this regard, this paper reviews
the moment amplification factor for the second-order effects on a rectangular portal frame with
recourse to the first-order elastic analysis, and finally discusses an issue between the second-
order moment amplification factor and the member buckling design, which can facilitate the
design section of the system approach in AS4100 in alignment with the advanced analysis for
the future system design approach.

1. Introduction AISC LFRD (1993), the second-order P-δ and P-Δ

effects can be estimated from the first-order analysis
Because of the restriction of calculation technique in
using the respective B1 and B2 amplification factor,
the past, the simple case of flexural buckling of an iso-
which is detailed in Chen and Lui (1991), to correlate
lated and individual member was only investigated by
the linear moments to the second-order moments based
means of the theoretical formula and experiments in
on the results from Bjorhovde et al. (1978). Duan and
order to investigate the P-δ effect owing to the change in
Chen (1988, 1989) presented the effective length fac-
the geometry of a structure. For example, Timoshenko
tor for the braced and unbraced frames to account for
and Gere (1961) analytically presented a comprehensive
the second-order effect on a framed structure. In the
study of flexural buckling of a column with different
AS4100 (1998), the moment amplification factors δb
ideal boundary conditions, which can be evaluated by
and δs are utilised to project the linear moment onto
the imaginary member length so-called the effective
the second-order moment due to respective P-δ and
length method.
P-Δ effects, which is also comprehensively discussed in
Unfortunately, the overall frame instability prob-
Trahair and Bradford (1998).
lem for a continuous column is ineffectively studied by
According to the Eurocode 3 (2005) recommended
means of the effective length method, because the end
that an out-of-plumb of a column, such as with 1/200
conditions of a column is no longer simply represented
times the column height, is incorporated into the linear
by the ideal conditions, and thereby it is inevitable to
elastic modelling to stimulate the P-Δ effects. Another
take the geometrical change of a whole framed struc-
method is an equivalent notional load method proposed
ture into account. Therefore, the elastic frame buckling,
by Wood, Beaulieu and Adams (1976), which produces a
which is the so-called P-Δ effect, had received consid-
fictitious loads applied to the column continuous in the
erable attention of the scholars.
typical rectangular frame. As a result, a s­ econd-order
In the past decades, the traditional design proce-
P-Δ moment due to the sway effect on a designated
dures to account for the nonlinear flexural and frame
column can be reasonably generated by the first-order
buckling due, respectively, to P-δ and P-Δ effects have
elastic analysis.
recourse to the empirical design specifications. In the

CONTACT  C. K. Iu  [email protected]

© 2017 Engineers Australia
2   C. K. IU

The frame instability issue of a framed structure 2.  Study of the moment amplification factor
owing to the P-Δ effect can also be approximated by the for a simple rectangular steel frame
elastic buckling mode from the buckling analysis (e.g.
Since the linear elastic analysis is reliable and commonly
Hancock, Papangelis and Clarke [1995]), such as the
adopted in the design office, the change in geometry of
elastic buckling load factor λcr. This method gives close
a structure is not well captured by this method; par-
approximations, when the buckling pattern is the same
ticularly restrictive for the whole structural system.
at each storey of a typical rectangular framed structure.
Therefore, the moment amplification factor for the sec-
Merchant (1954) proposed the simple Merchant-
ond-order effects, such as braced effect δb and sway effect
Rankine formula to compute the elastic buckling load
δs, on a continuous framed structure is inevitable when
factor λcr of a rectangular framed structure, which is also
based on the linear analysis. According to the AS4100,
widely adopted in the industry.
the second-order moment amplification factors are
Recently, the second-order inelastic analysis (i.e.
determined by the greater value of Equations (1) and
Iu and Bradford [2012a, 2012b]) is advocated for the
(2) but less than 1.4 as written,
system design approach, whose principle is to replicate
the additional loading distribution on an entire structure cm
owing to the second-order effects (i.e. geometrical 𝛿b = ( ∗/ ) ≥1 (1)
1 − N Nomb
change), such as P-δ and P-Δ effects, as well as material
yielding. This system approach is in contrast to the
empirical member-by-member design approach whose
general principle is to reduce the resistance of a member 1
𝛿s = � ≥1
because of the second-order effects; for example the (2)
�
N∗
∑
Δs
1−
widespread effective length method. On the contrary, the
∑ ∗
h V

second-order effects cannot deteriorate the resistance

where Δs is the relative lateral displacement at the top of
of a structure, but degrading its stiffness that produces
storey height hs under concern; N* and V* are the axial
the additional loading distribution on a whole structure.
load on column and shear force at the base of column;
In this sense, the system design approach (i.e. moment
Nomb is the elastic buckling load of a column as given,
amplification factor) in the AS4100 well aligns with the
principle of the numerical system design approach in 𝜋 2 EI
future. Nomb = ( )c2 (3)
In summary, the aforementioned design meth- ke h
ods for evaluating P-δ and P-Δ effects rely on various
and cm is the factor accounting for the effect of bending
design parameters, such as the effective length and the
moment distribution as well as the boundary condi-
moment amplification factor for an individual mem-
tions of a braced member implicitly by the along-span
ber or a member continuous in the framed structure
moment distribution together with the effective length
respectively, to evaluate the structural stability problem.
factor ke in Equation (3), which is comprehensively dis-
It is commonly foreseen that the emphasis on the mem-
cussed by Chen and Zhou (1987), and written as:
ber design approach (e.g. AS4100; Eurocode 3; LFRD)
will be transformed into the structural system design cm = 0.6 − 0.4𝛽m ≤ 1 (4)
approach, such as Section 4 of methods of structural
A single bay and one storey simple portal frame is selected
analysis in the AS4100, as discussed by Iu et al. (2008).
for this study as indicated in Figure 1, of which the beam
Therefore, in order to facilitate the design procedure of
connection is rigidly connected to the columns. And the
the whole system design approach, the aim of this paper
elasticity of the steel in this frame is 200GPa. This frame
is to review the design procedures used in AS4100 for
is of concern because it is vulnerable to the change in its
the P-δ and P-Δ effects (i.e. braced and sway effects)
structural geometry, and also this frame is so simple that it
on an elastic rectangular framed structure, which is
can generally replicate the overall second-order behaviour
based on a series parametric studies of a simple portal
of the rectangular framed structure, sometimes, even
frame, including the geometric aspect ratio, the stiff-
including multi-storey and multi-bay frames. In view of
ness ratio between beam and column section, the load
this, only simple rectangular frame is included herein.
ratio of both vertical and horizontal loads as well as the
For example, if a simple portal frame in Figure 1 is
boundary conditions. At last, the recommendation of
connected with diagonal bracing as shown in Figure 2,
the design calculation for the braced and sway effects in
the moment amplification factors given in Equation (1)
the AS4100 is made in order to facilitate the Australian
are no longer applicable, since the accurate elastic
design practice.
 AUSTRALIAN JOURNAL OF STRUCTURAL ENGINEERING   3

P P (a)
H

Figure 1. Layout and details of the simple portal frame for this (b)
study.

P P P

H H
Note: End condition
is unavailable in
AS4100, so the
buckling load Nomb of
this column is
inaccurately
evaluated

Figure 2. Limitation to assess the true buckling load of a column.

(c)
buckling load Nomb in the case of ke in Equation (3) not
being available from Clause 4.6.3.2 or Figure 4.6.3.3
in AS4100, for which Equation (1) leads to inaccurate
buckling load reliant on the estimated end condition
as shown in Figure 2; it is especially important, when
braced effect always governs the design of a braced
frame in Figure 2. It also reveals that the empirical mem-
ber-by-member design approach is restrictive to evaluate
the second-order behaviour of a general structure, such
as the deformed shape of a column in a non-rectangular
frame indicated by a red line in Figure 2. Therefore, the
frame with bracing is out of the scope of this study. (d)
The parameters chosen for the moment amplifica-
tion factors of the braced and sway effects in AS4100
consists of the aspect ratio h/L, stiffness ratio between
the beam and column section, load ratio P/H as well as
the boundary conditions of pinned supports. Therefore,
the objective of this study is to investigate the influence
of the braced factor δb and sway factor δs in Section 4 of
the AS4100 by these parameters, and subsequently the
design procedure of the moment amplification factor for
the second-order effect is reviewed.

2.1.  Aspect ratio for the braced and sway effect on

a simple portal frame Figure 3.  Various aspect ratios with different column heights.
(a) Aspect ratio L/h = 0.2, (b) Aspect ratio L/h = 0.6, (c) Aspect
In this parametric study of aspect ratio, the sections of
ratio L/h = 0.8, (d) Aspect ratio L/h = 2.
beam and column are H340A and H300B, respectively.
And the vertical P and horizontal load H are 2800 and
35kN, respectively. spectrum of the study is representative in terms of its
In this study, the aspect ratios L/h of 0.2, 0.6, 0.8 and geometry, in each catalogue of aspect ratio the column
2 are of concern as illustrated in Figure 3 such that the height varies for a particular geometry of a frame. For
4   C. K. IU

3.5

3
db
b (aspect ratio 0.2)

Moment amplification factor

2.5
dss (aspect ratio 0.2)

dbb (aspect ratio 0.6)

2
dss (aspect ratio 0.6)
1.5
dbb (aspect ratio 0.8)

ds (aspect ratio 0.8)

1 s

db (aspect ratio 2)
b
0.5
ds (aspect ratio 2)
s

0
0 5 10 15 20 25 30 35
Elastic buckling load factor c

Figure 4. Moment amplification of different aspect ratio against λc.

example, when column height is 4 m in the catalogue Table 1. Beam and column sections of concern.
of 0.2 aspect ratio, the geometry of the frame is given as Beam section Column section
h = 4 m and L = 0.8 m as shown in Figure 1. Hence, lower 610UB125 310UC158
value of the aspect ratio means the simple portal frame 530UB92.4 250UC89.5
460UB82.1 200UC59.5
is more slender. And also the column height is greater in 360UB56.7 150UC37.2
the corresponding catalogue of the aspect ratio, which 250UB29.8 100UC14.8
150UB14.0
means the frame is scaled up in dimension. The moment
amplification factors against the column heights with
different aspect ratios are indicated in Figure 3. 4.0
From Figure 3, it can be found that the sway effect 3.5
of the frame with various aspect ratios (i.e. 0.2; 0.6; Braced factor with cm

3.0 150UC37.2
0.8; 2) always governs the moment amplification fac-
Moment amplification factor

Braced factor without cm

tor. Moreover, in each aspect ratio catalogue, when the 2.5

column height increases, both braced δb and sway fac- 2.0

200UC59.5
tors δs increase, but the sway factor δs increases dras- 1.5

tically, which implies that the bay length L increases 1.0 250UC89.5 310UC158
with column height that cannot provide adequate
0.5
lateral stiffening. Hence, under the circumstance of
higher aspect ratio with greater column height, the 0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
slenderness of the frame is significant as shown in Elastic buckling load factor c
Figure 3(d).
Figure 3 can be summarised and transformed into Figure 5. Moment amplification factor for braced effect with λc.
Figure 4 in the coordinate of the elastic buckling load
factor λc. It is found that when λc is less than 5, the frame 5.0
is sway-sensitive and the sway factor δs is very signifi- 4.5 Braced factor b

cant. Thus, it is identified that the second-order analysis

Moment amplification factor

4.0 Sway factor s

should be implemented for those frames. Moreover, the 3.5

braced factors δb of different aspect ratios are always 3.0

2.5
close to unity for the most of the range of λc. It is inter-
2.0
esting to highlight that even when the λc is more than 10, 1.5
which equivalent to the non-sway frame, the sway factor 1.0

δs is still more critical. On contrary, the sway factor δs 0.5

can govern the second-order effect due to the geomet- 0.0

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
rical change in a whole rectangular steel frame under Elastic buckling load factor c
all circumstances of this parametric study. Therefore,
the sway factor δs is more critical under the parameter Figure 6.  Moment amplification of different stiffness ratio
of aspect ratio. against λc.
 AUSTRALIAN JOURNAL OF STRUCTURAL ENGINEERING   5

2.2.  Stiffness ratio for the braced and sway effect 2

P/H=800
on a simple portal frame 1.8

1.6

Moment amplification factor

The stiffness ratio between the beam and column section 1.4

is also a factor to influence the braced and sway effect 1.2

P/H=80
in the rectangular framed structure. In this parametric 1
P/H=800

study of stiffness ratio, the dimension of the simple por- 0.8

tal frame is given as h = 5 m and L = 4 m as shown in 0.6 Braced factor
b

Figure 1. And the vertical P and horizontal load H are 0.4

2800kN and 35kN, respectively. 0.2 Sway factor s

Further, the stiffness between the column and beam 0

0 5 10 15 20 25 30
sections can contribute to the lateral stability by the Elastic buckling load factor c

rigid beam to column connections. Hence, this is not

only the issue of the member sections, but also the Figure 7. Strong beam combination as 610UB125 & 310UC158.
member length that can provide coupling shear at both
beam-column connections for the lateral stability of 4

the frame. Further, higher value of the stiffness ratio 3.5

P/H=400
Braced factor
means greater lateral stiffness contributed from the b

Moment amplification factor

3
columns. Sway factor s
2.5
In this study, two batch of beam and column sec-
2
tions are selected for investigation, which are summa-
rised in Table 1, such that the combination between 1.5
P/H=800 P/H=40
beam and column section are chosen for a wide spec- 1

trum in terms of the stiffness contribution to the lat- 0.5

eral stability effect, where the stiffness ratio defined as 0

∑ � Ic �� ∑ � Ib � 0 5 10 15 20

h L
. Elastic buckling load factor c

In regard to the braced effect, the braced factor δb

against the elastic buckling load factor λc is plotted in Figure 8. Weak beam combination as 150UB14 & 310UC158.
Figure 5. It is noted that the first batch of column sec-
tion of 100UC14.8 is excluded from Figure 5, because 5
P/H=80
this column section is too slender such that the braced 4.5
Braced factor b
Moment amplification factor

factor δb is negative, and thereby Equation (1) is not 4

3.5
applicable. Furthermore, the braced factor δb fluctu- 3
Sway factor s

ates with λc because of the cm factor for the bending 2.5

moment distribution, while the braced factor δb with 2

cm = 1 is steady as illustrated in Figure 5. It indicates 1.5

P/H=80
that the section between beam and column can adjust 1

the bending moment distribution on the column. For 0.5

0
example, stronger beam section can withstand more 0 5 10 15 20

bending moment at the top of a column, so the trend Elastic buckling load factor c

of each segment in Figure 5 is declining with increasing

Figure 9. Weak column combination as 250UB29.8 & 150UC37.2.
λc and further declines with increasing the beam sec-
tion. The braced δb factor is always not critical, because 3.5
the double curvature of bending moment distribution P/H=160
3 Braced factor
on the column can mitigate its critical buckling effect.
Moment amplification factor

Hence, the cm for the braced factor δb is always equal 2.5

Sway factor s

to unity. 2

In summary, the moment amplification factors for

1.5
braced and sway effects of different beam to column P/H=160
combination are plotted against the elastic buckling load 1

factor λc in Figure 6. It can be perceived that the sway 0.5

factor δs is slightly greater than the braced factor δb for 0

the corresponding column and beam section combina- 0 5 10 15 20

Elastic buckling load factor

tion. In this sense, the stiffness ratio is not sensitive to c

the difference between the sway and braced effect for the Figure 10.  Strong column combination as 250UB29.8 &
second-order effect. 200UC59.5.
6   C. K. IU

2.3.  Load ratio for the braced and sway effect on a 7.0
L/h=2
simple portal frame

Moment amplification factor

6.0 braced factor b

The load ratio between horizontal load H and vertical 5.0

load P is also a major factor for the braced and sway 4.0
sway factor s

effect in the rectangular framed structure; especially 3.0

the vertical load effect, since it causes the geometrical
2.0
change in the structure and creates the second-order L/h=0.8

moment because of itself. Thus, the load ratios P/H are 1.0

selected as 1, 5, 20, 40, 80, 160, 400 and 800, whereas 0.0
H is set at 35kN for all cases. In this parametric study 0.0 5.0 10.0 15.0

Elastic buckling load factor

of load ratio, the dimension of the simple portal frame c

is given as h = 5 m and L = 4 m as shown in Figure 1. Figure 11. Moment amplification factor of the load ratio of 80.
And four combinations of beam and column sections
are focused. They are; (1) 610UB125 & 310UC158; (2)
150UB14 & 310UC158; (3) 250UB29.8 & 150UC37.2;
(4) 250UC29.8 & 200UC59.5, which are arranged from
the strong beam section to weak column section, and
vice versa.
According to Figures 7 and 8, the column sections
are same, so the buckling loads of the frame due to
P-δ effect are same, and thereby it is not sensitive.
However, in regard to the sway effect, the buckling
load of the frame in the strong beam combination as
illustrated in Figure 7 is greater than that in the weak
beam combination in Figure 8, while the load ratio
Figure 12. Moment amplification factor of the load ratio of 160.
at the corresponding λc in Figure 7 is greater, which
means larger load can be sustained. It shows that the
beam section can take part in the lateral stability for 2.5

L/h=10 braced factor b

the sway effect. 2.0
Moment amplification factor

According to Figures 9 and 10, the beam sections are sway factor s

same, but the column section is various. Apparently, 1.5 L/h=2

the buckling load of the frame due to the sway effect is

less, when the column section is less as shown in Figure 1.0
L/h=50
9. However, for the braced effect, the buckling load is 0.5
insensitive, which is similar to the study of the beam
combination. It means the braced effect does not govern 0.0
0.0 5.0 10.0 15.0 20.0
the buckling load of the frame.
Elastic buckling load factor
The load ratio is heavily dependent of the elastic c

buckling load of the column, but insensitive to the Figure 13. Moment amplification factor of the load ratio of 400.
section combination according to Figures 7–10, when
they are all consistent in the similar fashion. Further,
the sway factor δs is always greater than the braced 12.0

factor δb for all range of elastic buckling load factor λc. 10.0
P/H=20
Moment amplification factor

braced factor b

It heralds when the frame is braced, such as λc > 10, the

8.0
sway factor δs still govern the moment amplification sway factor s

factor. In contrast, the braced factor δb is insensitive and 6.0

always close to unity for all ranges, because the factor 4.0
cm is insignificant, while the beneficial effect against P/H=100

the second-order effect from the double curvature of 2.0

the fixed ended portal frame as shown in Figure 1. 0.0

Therefore, the next parametric study is the boundary 0.0 5.0 10.0 15.0 20.0

Elastic buckling load factor

condition of pinned end that changes the bending c

moment distribution on the columns to be single Figure 14. Moment amplification factor of the aspect ratio of
curvature. 0.8.
 AUSTRALIAN JOURNAL OF STRUCTURAL ENGINEERING   7

2.4.  Boundary conditions for the braced and sway 4.5

P/H=100
effect on a simple portal frame 4.0
braced factor b
3.5

Moment amplification factor

The simple portal frame as shown in Figure 1 is replaced 3.0
P/H=800
sway factor
with the pinned connections at both column bases, which
s

2.5
withstands larger lateral displacement, and thereby is 2.0
prone to the sway effect. However, the moment distribu- 1.5
tion of single curvature on the columns can also exacer- 1.0
bate the braced effect, which means more critical to the
0.5
compression buckling. The load ratio (i.e. P/H = 80, 160
0.0
& 400) and aspect ratio (i.e. L/h = 0.8 & 2.5) are under 0.0 5.0 10.0 15.0 20.0

investigation of this parametric study. Since the stiffness Elastic buckling load factor c

ratio between beam and column section is insensitive, so

Figure 15. Moment amplification factor of the aspect ratio of
the beam and column sections are fixed to be 250UB31.4
2.5.
and 200UC46.2, respectively. Under this circumstance,
the cm factor is equal to 0.6 for all cases.
In order to study the aspect ratio of a portal frame,
the load ratio of 80, 160 and 400 are selected, on which
the length of bay L is fixed (i.e. 4 m) but varies with its
height h; for example, the range of aspect ratio is 50,
20, 10, 5, 2.5, 2, 0.8. Figures 11–13 plot the moment
amplification factors against λc for the load ratio of 80,
160 and 400, respectively.
It can be seen that the sway factor δs is also always
more critical than the braced factor δb in general and
similar to the findings from the study of aspect ratio
in Section 2.1, except for the case of load ratio 400 and
the vertical load P is 14000kN. From Figure 13, the Figure 16. Compression load of an axially loaded member.
sway factor δs reaches 2, when the aspect ratio is 10 (i.e.
h = 0.4 m), at which λc is 2. On the other hand, the braced
sway effect is the load ratio of 20 (i.e. p = 700kN), whereas
factor δb attains 1.27 when λc is 9.45, whereas the sway
the most critical braced effect is the load ratio of 100 (i.e.
factor is 1.1 (aspect ratio L/h = 50) when λc is 11.5, which
p = 3500kN), when excluding lateral displacement for
leads to the braced factor δb greater than the sway factor
sustaining more compression load. In short, when the
δs. The anomaly in Figure 13 can be attributed to the
condition of the pinned support condition leads to the
Equation (1) that is derived under the boundary condi-
significant sway effect, so the maximum vertical load P
tion of being fixed at both ends. Theoretically, the pinned
withstood by the frame with sway effect is less than that
boundary condition at the column bases of the frame
of the frame with braced effect. Figure 15 illustrates that
aggravates the sway effect, so the sway factor δs is very
the sway and braced effect is very consistent. In contrast,
critical even for the aspect ratio of 10 (i.e. short frame)
the sway factor δs is more significant and critical when
when subjected to high vertical load P. Unfortunately,
the most critical load ratio is 100, whereas the braced
the Equation (1) of the braced factor δb excludes the
factor δb reaches its optimal value of the load ratio 800,
effect of lateral displacement such that the pinned simple
which is consistent with the findings in Section 2.3 that
portal frame with a low aspect ratio (i.e. more slender)
the frame with sway effect can sustain the same or less
becomes more critical to the second-order effect, when
vertical load P compared with that of the braced effect.
it merely accounts for member buckling effect (i.e. P-δ
In summary, the sway factor δs is more critical to the
effect).
braced factor δb when the simple portal frame is at the
In regard to the study of the load ratios, two particular
pinned supports.
aspect ratios of 0.8 (i.e. h = 5 m) and 2.5 (i.e. h = 1.6 m)
are chosen, but the load ratio varies (i.e. 800; 400; 100;
3.  Discussion between the moment
80; 40; 20; 10; 5; 1). The moment amplification factor
amplification factor and the effective length
against the elastic buckling load factor λc is plotted in
method
Figures 14 and 15 for the aspect ratios of 0.8 and 2.5,
respectively, in which the vertical load P varies with the After the studies of the moment amplification factors
constant horizontal load H (i.e. H = 35kN). for both braced and sway effects, the objective of this
Similarly, the sway factor δs is more critical to the section is to discuss the different approaches of AS4100
braced factor δb throughout the whole range of λc in to consider these effects, including the moment ampli-
general as shown in Figure 14, in which the most critical fication factor in Section 4 and effective length method
8   C. K. IU

in Sections 6 and 8. In principle, the design aim of the Further, the braced factor δb in Equation (1) multi-
moment amplification factor in Section 4 of the method plies with the first-order moment M* to become sec-
of structural analysis in AS4100 is same as that of the ond-order moment, which must be less than or equal to
effective length method used in Sections 6 and 8 of the the section moment capacity Ms as shown in Equation
relevant axial compression design in AS4100. These (6). Hence, the bending moment M* on the beam-col-
sections can account for the second-order effect, par- umn member under the interaction between axial and
ticularly P-δ effect. However, the principle of the design bending effects is given as:
approach adopted in Section 4 is not same as those in
Sections 6 & 8 of the axial compression and interac- 𝛿b M ∗ ≤ 𝜙Ms
tion, respectively, to which the effective length method (6)
( )
N∗
cm M ∗ ≤ 𝜙Ms 1 −
is resorted. The principle of Section 4 of the AS4100 Nomb
is to determine the second-order moment generated
directly from the change in structural geometry, such where ϕ is the capacity factor; cmM* is the equivalent
as the moment amplification factor, which is well aligned moment distribution accounting for the moment distri-
with the principle of the second-order analysis, such as bution along a member and also the change in structural
Iu and Bradford (2012a, 2012b). geometry along its span.
In order to demonstrate the principle of the effec- The system approach of the moment amplification
tive length method, which is constructive to discuss factor from Equations (5) and (6) is similar to its counter-
the system design approach in Section 4, its principle part in Figure 16 by means of the effective length method
is to reduce the member capacity, thanks to second-or- of the member approach. Therefore, both approaches
der effect that the Euler elastic buckling load Nomb in have the similarity to accommodate the same effect. In
Equation (3) is commonly adopted for member design this regard, there is the recommendation for the system
approach as graphically and conceptually depicted by the design approach in the following Section.
dash line in Figure 16 against the member slenderness
ratio λ. The axial squash load Ns (i.e. section capacity) 4.  Recommendation for the system design
and inelastic axial compression buckling curves Nc with approach
imperfection are also plotted in Figure 16.
According to the principle of the effective length According to the relations of axial load as Equation (5)
method, when the member slenderness ratio λ is small, and bending moment as Equation (6), there is the simi-
the inelastic buckling load Nc of the axial compression larity between two different approaches of the moment
member is equal to its section capacity Ns as shown in amplification factor in Section 4 and the effective length
Figure 16. As the slenderness ratio λ increases, the axi- method in Sections 6 and 8 of the AS4100, which is well
ally loaded member is then critical to the compression developed and sophisticated when including most of the
buckling, and hence the axial compressive load Nc of the practical imperfection. Unfortunately, despite taking the
member decrease drastically. Further Nc of the mem- change in structural geometry (i.e. P-δ and P-Δ effects)
ber approaches to its elastic buckling load Nomb when λ into account, the moment amplification factors, such as
increases. In Figure 16, the derivation between elastic Equations (1) and (2), are not effective and adequate to
buckling load Nomb and inelastic buckling load Nc with accommodate the practical considerations and inelas-
imperfections is because Nc comprises the inelasticity, ticity by the AS4100.
initial geometrical imperfection, residual stress and local Because of this, it is indispensable to facilitate the sec-
plate buckling that can be calibrated in accordance with ond-order moment amplification factor to include the
the Perry–Robertson formula. inelasticity and other practical considerations based on
On the other hand, from the system approach of the the system design approach instead of the member-based
second-order moment amplification factor, Equation (1) design approach, such that the moment amplification
can be rearranged as: factor in Section 4 of the AS4100 is self-sufficient for
the design of an inelastic compression buckling mem-
ber with imperfections, which is continuous in a struc-
( )
𝛿b − cm
N∗ ≤ Nomb = 𝛼Nomb (5) tural system instead of being isolated. In particular, this
𝛿b
approach of the moment amplification factor is therefore
in which α is the reduction factor (i.e. 0 ≤ α < 1) simi- in alignment with the advanced analysis for the future
lar to the slenderness reduction factor αc in Figure 16, design.
but not identical. Considering the change in structural For example, the braced factor δb can be modified
geometry, the axial compression load N* must be less or because of the similarity so as to include the inelastic-
equal to αNomb, which is relevant to the axial compres- ity and other practical imperfections as elaborated in
sion load Nc in Section 6 of the AS4100 but not account- Section 3 of this paper. In the meantime, the effective
ing for the change in geometry. length method should then transform into the moment
 AUSTRALIAN JOURNAL OF STRUCTURAL ENGINEERING   9

amplification factor, of which the design of an isolated framed structure. Under these circumstances, the braced
member with both ideal support conditions under pure factor δb is not applicable. In addition, (3) (conservative)
axial compression should be relied on the interaction as mentioned before in Section 3 of this paper, the aim
capacity check between the linear elastic axial load and of the braced factor δb is same as that of the effective
second-order moment due to the braced effect as: length method to account for the P-δ effect. It heralds
that when the second-order moment is obtained from
N∗ 𝛿 M∗ the braced factor δb for the interaction capacity check
+ b ≤1 (7)
𝜙Ns 𝜙Ms of a braced member, the P-δ effect is taken into account
twice by both bending (moment amplification factor)
where δbM* is the second-order moment created from and axial compression effects (effective length method).
the P-δ effect, such as Pδ, in which δ is the along-span This duplication leads to the conservative design and is
deflection of a member. In Equation (7), there is no unfortunately same for the sway effect; it is particularly
effective length method applied instead of the moment true, when both approaches have the same aim but dif-
amplification factor. It means all second-order effects ferent in principle as discussed in Section 3.
on a pure axial compression member, including P-δ and
• Recommendations of the design rule for braced
P-Δ effects, should be transformed from the member
and sway effects
resistance as the denominator of Equation (7) (i.e. effec-
As pointed out previously, the braced factor δb is
tive length method) into the bending moment distri-
ineffective because of the reasons (1) & (2) in the above
bution as the numerator of Equation (7) (i.e. moment
paragraph, so it is highly recommended that the factor
amplification factor), for which the design calculations
δb should be revised more adequate and sophisticated
must be noticeably simplified and in alignment with the
for the system design approach as discussed in Section
system design approach. It heralds that the second-or-
4 of this paper, such as accommodating the practical
der moment amplification factor should supersede the
considerations (e.g. imperfection, residual stress) and
effective length method for the future system design
inelasticity.
approach.
Alternatively, the braced factor δb should be removed
in Section 4 in AS4100 because of the reasons (1), (2)
5.  Discussions and conclusions and (3), since the effective length method in Sections 6
& 8 is adequate for the braced effect.
• Moment amplification factor valid within the range
According to all parametric studies including aspect • Duplication of the second-order effects by both
ratio, stiffness ratio, load ratio and boundary conditions, moment amplification and effective length method
the moment amplification factor, which is greater than In regard to the reason (3) in this section, it is sug-
1.4, are very significant, while λc is less than 5. It implies gested that the moment amplification factor should not
that the frame is too sensitive to the second-order effect be applied for the braced or sway member particularly
and Equations (1) and (2) is no longer valid. Therefore, subjected to the interaction between axial and bend-
the moment amplification factors in AS4100 are justified ing effect, which leads to the conservative design as
under all circumstances. explained in Section 4 of this paper. Otherwise, the
Equation (7) should be relied on for the system design
• Revision of the braced factor in Section 4 in AS4100 approach. Further study on this issue is required.
According to the findings from different parametric
studies, including aspect ratio, stiffness ratio, load ratio
Disclosure statement
and boundary conditions, (1) (not critical) the braced
factor δb is always ineffective and not critical under some No potential conflict of interest was reported by the author.
particular circumstances compared to the sway factor
δs; for example, the single- and multi-storey single-bay Notes on contributor
portal frame. Moreover, (2) (not applicable) the braced
C. K. Iu, PhD, has much research experience in the area
factor δb of Equation (1) is basically derived from the of advanced computational technique for the general
stability functions of a member, such as Chen and Zhou engineering applications in various countries, including
(1987), of its ends being fixed in position. Therefore, China, United States and Australia; in particular, the future
the Equation (1) can merely validate for a framed struc- structural engineering discipline is at the computer era
ture, whose lateral displacements at each storey are fully
restrained, or for a framed structure with a strong diag- References
onal bracing against the lateral displacements at each
AISC-LRFD. 1993. Load and Resistance Factor Design,
storey. For the latter case, the effective length of a braced
Manual of Steel Construction. Vol. 1 and 2, 2nd ed. Chicago,
member being continuous in the braced frame with IL: American Institute of Steel Construction, AISC.
diagonal bracing member is unavailable, while Figure AS4100. 1998. Australian Standard – Steel Structures. 2nd ed.
4.6.3.3 of the AS4100 is only applicable for a rectangular Standards Association of Australia.
10   C. K. IU

Bjorhovde, R., T. V. Galambos, and M. K. Ravindra. 1978. Iu, C. K., and M. A. Bradford. 2012a. “Higher-order Non-linear
“LRFD Criteria for Steel Beam-columns.” Journal of the Analysis of Steel Structures Part I: Elastic Second-order
Structural Division, ASCE 108 (9): 1371–1387. Formulation.” Advanced Steel Construction 8 (2): 168–182.
Chen, W. F., and E. M. Lui. 1991. Stability Design of Steel Iu, C. K., and M. A. Bradford. 2012b. “Higher-order Non-
Frames. Boca Raton, FL: CRC Press. linear Analysis of Steel Structures Part II: Refined Plastic
Chen, W. F., and S. P. Zhou. 1987. “Cm Factor in Load Hinge Formulation.” Advanced Steel Construction 8 (2):
and Resistance Factor Design.” Journal of Structural 183–198.
Engineering 113 (8): 1738–1754. Iu, C. K., W. F. Chen, S. L. Chan, and T. W. Ma. 2008. “Direct
Duan, L., and W. F. Chen. 1988. “Effective Length Factor Second-order Elastic Analysis for Steel Frame Design.”
for Columns in Braced Frames.” Journal of Structural KSCE Journal of Civil Engineering 12: 379–389.
Engineering 114 (10): 2357–2370. Merchant, W. 1954. “The Failure Load of Rigid Joined
Duan, L., and W. F. Chen. 1989. “Effective Length Factor Frameworks as Influenced by Stability.” The Structural
for Columns in Unbraced Frames.” Journal of Structural Engineer 32: 185–190.
Engineering 115 (1): 149–165. Timoshenko, S. P., and J. M. Gere. 1961. Theory of Elastic
Eurocode 3. 2005. Design of Steel Structures – Part 1-1: Stability. New York: McGraw-Hill.
General Rules and Rules for Buildings. Brussels: European Trahair, N. S., and M. A. Bradford. 1998. The Behaviour and
Committee for Standardization. Design of Steel Structures to AS4100. New York: CRC Press.
Hancock, G. J., J. P. Papangelis, and M. J. Clarke. 1995. PRFSA Wood, B. R., D. Beaulieu, and P. F. Adams. 1976. “Column
User’s Manual. Sydney, Australia: Centre for Advanced Design by P-delta Method.” Journal of the Structural
Structural Engineering, University of Sydney. Division, ASCE 102 (2): 411–427.