You are on page 1of 60


1.1. Latar Belakang

Critical book report adalah mengkritik ataupun membandingkan dua buku

yang berbeda. Baik perbedaan buku dan pengarang, untuk membahas tentang
kelebihan dan kekurangan dari buku yang ingin dibandingkan. Laporan
pengkritikan buku ini bertujuan menambah wawasan mahasiswa – mahasiswi
Universitas Negeri Medan . Dan buku yang ingin penulis kritik berjudul Steel
Framed Structures.Dalam kedua buku ini sangat lah tepat untuk di critic karena
adanya perselisihan yang ada pada kedua buku dalam masalah teori.
Buku Steel Framed Structures merupakan buku yang mencakup seluruh
pelajaran mengenani berbagai macam Struktur rangka baja yang digunakan dalam
proses penggunaan material dalam hal konstruksi sipil.

1.2. Tujuan Pembahasan

Adapun Tujuan Pembahasan dalam pengkritikan sebuah buku yaitu

- Membantu pembaca mengetahui gambaran dan penilaian
umum dari dua buku secara ringkas

- Menguji kualitas buku dengan membandingkan terhadap karya

dari penulis yang lainya.

1.3. Manfaat Pembahasan

Adapun Manfaat Pembahasan dalam pengkritikan sebuah buku yaitu

- Menambah ilmu pengetahuan dalam pembelajaran Struktur
Rangka Baja
- Mengetahui berbagai jenis Struktur Rangka Baja
- Mengetahui perbedaan buku yang ingin dibandingkan



2.1. Ringkasan Isi Buku pertama

Chapter 1
Frame instability and the plastic design of rigid frames
Idealised approximations to material stress-strain relationships lead to
corresponding idealised limit loads—in particular, the rigid-plastic collapse load
and the least elastic critical load. The real behaviour, allowing for stability and
change of geometry, causes a reduction of carrying capacity below the rigid-
plastic collapse value. The extent of the reduction depends on the slendernesses of
the members and may be related to the value of the elastic critical load.
Justifications for the Merchant-Rankine load, and for the modification suggested
by Wood, are discussed, and applications to unbraced multi- storey frames show
the usefulness of this procedure. An example of the design of a multi- storey
frame involving frame stability effects is given. Special frame stability problems
in single-storey pitched roof frames are discussed, and safeguards are described
and illustrated by reference to two designexamples.


The stability and strength of a framed structure may be explored in relation

to the various approximations that may be made to the real stress-strain behaviour
of the material of which the structure is composed. Mild steel has a stress-strain
curve in tension or compression of the form shown in Fig. 1.1(a), in which an
elastic phase OA up to an upper yield stress σU and a strain εy=σU/E (where E is
the elastic modulus) is followed by plastic deformation at a lower yield stress σ L
up to a strain εs of the order of 10εy. Beyond a strain of εs, strain-hardening occurs,
the strain ultimately becoming non-uniform due to necking, leading to fracture at
an ultimate stress σult some 25–40% above σL and at an elongation of some 30%.
Within the range of structural interest, the idealised elastic- plastic stress-strain

relation in Fig. 1.1(b) is a sufficiently closeapproximation.

With the aid of computers, it is possible to follow analytically the

behaviour of entire structures on the basis of any assumed stress-strain
relationship. However, even if the idealised elastic-plastic stress-strain relation of
Fig. 1.1(b) is used, limits of computer capacity can soon be reached. Further
idealisations of behaviour facilitate design and may be used if their respective
limitations are recognised. Provided the effects of the deformations of the
structure on the equations of equilibrium are neglected, the idealised elastic
behaviour (Fig. 1.1(c)) leads to a linear relationship between intensity of loading
and the deformations and stresses induced in the structure, and is the basis for
many design procedures. However, the analysis is valid only up to the stage at
which yield is reached somewhere in the structure—a point which has no
consistent relationship to the ultimate strength of the structure. For this purpose,
the rigid-plastic stress-strain relationship in Fig. 1.1(d), leading to the concept of
plastic collapse mechanisms, gives for many structures a close estimate of the
actual load at which collapse would occur.


The plastic theory of structures is now well understood by engineers, and

offers a straightforward means of assessing the ultimate load of a continuous
structure. For many single-storey and low-rise frames, plastic design can form the
basis of the design procedure, but it may be necessary to estimate the effect of
instability on the plastic collapse load. The reason why instability affects the
collapse load lies in the effect of deformation on the calculated internal forces—
deformations either within the length of a member, or of the frame as a whole. The
problem is therefore best introduced by considering the effect of deformations on
plastic collapse loads.
The fundamental theorems of plasticity refer strictly to rigid-plastic

materials, that is, to materials with an infinitely high modulus of elasticity (Fig.
1.1(d)). The structure is assumed to have no deformations at the collapse load. In
exploring the effect of the finite deformations induced by elastic behaviour before
collapse occurs, it is instructive first to consider the effect of finite deformations in
a rigid-plastic structure.



It has been seen that, if the material of a structure is assumed to be rigid-

plastic, a drooping load-deflection curve GH is obtained (Fig. 1.7), descending
from the rigid- plastic collapse load factor λP. If ideal elastic behaviour is
assumed, the load-deflection curve OBC rises to the elastic critical value λC. The
actual behaviour OAFD follows the elastic curve up to the load factor λy at which
yield first occurs, rising to a peak at F, before approaching the rigid-plastic
mechanism line GH at large deflections. Merchant (1954, 1958) suggested that it
might be possible to consider the failure load factor λF as some function of the
load factors λy, λP and λC, and also of the purely abstract load factor λG (Fig. 1.7)
obtained at the intersection of the elastic curve OAC with the rigid-plastic
mechanisms line GH. The advantage of such an approach is that these load factors
are much easier to calculate than the failure load factor λF itself.
Merchant (1954) tested, for a large number of theoretical structures, the
following formula, which may be regarded as a generalisation of Rankine’s

formula for struts. The failure load factor λF is approximated to by the Merchant-
Rankine load factor λMR, where


Sinceformanypracticalstructures λCis large compared with λP, the

requirement for such structures is satisfied. If λF/λP is plotted vertically and
λF/λC horizontally, eqn (1.11) with λF=λMR is simply given by the straight line AB
in Fig. 1.11. The points plotted on Fig. 1.11 were obtained theoretically by Salem
(1958) for one- and two-storey frames loaded as shown. Lines corresponding to
various ratios of λP/λC have been drawn, and it is readily seen that the Merchant-
Rankine formula (eqn (1.11)) is most successful when λP/λC is small and the
collapse load is close to the rigid-plastic collapse value. When λP/λC>0·3 the
scattering of the points away from the Merchant-Rankine formula is considerable.
Merchant suggested the formula as a safe (that is, lower) limit for the collapse
load. Its theoretical significance has been discussed by Horne(1963).



Numerical examples based on second-order elastic-plastic analyses led

Merchant (1954) to the conclusion that the Merchant-Rankine load (eqn (1.11))
represented an approximate lower bound to the failure load. He did not claim it as
a strict lower bound (see Fig. 1.11). However, if it is to be used as the basis for
design, one needs some assurances that no serious error on the unsafe side will
arise. Some evidence has been given emphasising the extent to which the
Merchant-Rankine load is not a lower bound (Adam, 1979) and for this reason
Anderson and Lok (1983) examined a number of 4-, 5- and 10-storey unbraced
frames. These frames were realistic in that they were designed economically for
wind and floor loadings specified in codes, with sway deflections at unit factored
wind loading limited to 1/300 of each storey height. Some of their results are
summarised in Table 1.1. Two loading conditions are considered—one with
maximum wind loading and the other with maximum vertical loading. Values are

quoted for the ratio soft helo west elastic critical load factorλC, the Merchant-
Rankine load factorλMR and the modified Merchant-Rankine load factor λWMR to
the accurately calculated ‘second-order’ (i.e. allowing for change of geometry)
elastic-plastic failure load factor λF. It will be seen that, without exception, the
Merchant-Rankine load is a safe estimate of the failure load. Anderson and Lok
found that only when the storey height exceeded the span of the beams, and when
the wind loading was exceptionally low, was the Merchant- Rankine load higher
than the elastic-plastic failure load. Such frames are not realistic— and if they
occurred, would certainly be braced.
Anderson and Lok recommend that the Merchant-Rankine formula should
not be used when the bay width is less than the greatest height.



General Considerations

The assumption has usually been made in the past that overall stability
problems do not affect the design of single-storey frames, the argument being that
the mean axial stresses in the columns are generally small. While this latter
statement is correct, it is also true that such frames may be quite slender in the
plane of bending, and this may bring down the ratio of critical load to plastic
collapse load to an unacceptably lowlevel.

Single-Bay Frames

In single-bay frames the usual deflection limitations when the frame is

subjected to wind loading will usually ensure that overall stability is not a
controlling factor. This is certainly the case when the horizontal deflections at the
tops of the stanchions are limited to height divided by 300, under unfactored
loads. However, if it can be shown that greater deflections would not impair the
strength or efficiency of the structure, or lead to damage to cladding, then this
deflection limit may be allowed to be exceeded, and there will undoubtedly be a

desire to take advantage of this in many single-storey frames. A safeguard against
deflections which could affect strength and safety is therefore needed, and is
provided by the following requirement (see Horne, 1977).

The horizontal deflection δ, in millimetres, at the top of a column due to

horizontal loads applied in the same direction at the top of each column, and equal
to 1% of the vertical load in the column due to factored loads, must not exceed
1·8h, where h is the height of the column in metres. In calculating δ, allowance
may be made for the restraining effect of cladding.

Chapter 2


The ultimate limit state design of plane multi-storeyed sway frames

involves consideration of nonlinear effects induced by changes of geometry and
the influence of member axial forces. In recent years, nonlinear matrix methods of
analysis have been developed and it is now possible to analyse the complete
loading history of such frames up to collapse. This chapter deals with the elastic
linear, nonlinear and instability analysis of frames using matrix methods. Rigorous
formulations of the problem are presented first so that the approximations
frequently incorporated in such analyses can be identified. Approximate methods
for determining elastic critical loads and the magnitude of the nonlinear effects are

The total potential energy V of a structural system (Washizu, 1968) can be
defined by the equation


where Piand qirepresent the external forces and corresponding displacements, and
σiand εirepresent the internal stresses and corresponding strains. V0 is the potential
energy of the system prior to application of external forces. The integrals in eqn
(2.1) represent the work done by the external forces and the strain energy of the
structure, which are equal. Therefore, along any equilibrium path, V is constant
and hence the first and second variations of V along the equilibrium path, denoted
by δV and δ2V, arezero.
Assuming that V can be expressed in terms of a number of prescribed
displacements, qi, δV and δ2V are defined as


Hence from eqn (2.1)


Since the structure as a whole, and individual parts of the structure, are in
equilibrium, eqn (2.3) is valid for all δqi, not just variations along an equilibrium
path. Equation (2.3) provides a basis for linear and nonlinear iterative analysis
while equation (2.4) provides a basis for nonlinear incrementalanalysis.
Variational principles can also be used to investigate the stability of
structures. For equilibrium δV=0. Stability of the system requires that positive
work be done to move the system from the equilibrium state and hence δV=0
corresponds to a minimum value. Conversely, if, as the system moves slightly
from the equilibrium state, energy is given out, which can be manifest only as
kinetic energy, the system is unstable. Hence for stable equilibrium, δV=0 and the
second variation of V for stationary values of the external forces, denoted by
δ2VP, is positive definite. From equation (2.4), δ2VP is given by

Critical conditions occur when δ2VP changes from positive definite to

zero, indicating a possible transition from stable equilibrium to instability.


Nonlinear elastic analysis of frames can be performed either incrementally

or iteratively.Incremental analysis involves the determination of the incremental
or tangent stiffness matrix relating small increments in external forces and
corresponding displacements; this depends on the current geometry and state of
stress. Complete solutions for the entire loading history can then be obtained by
incrementing either forces or displacements (Roberts, 1970; Roberts and Ashwell,
1971). Incrementing displacements has the advantage that solutions do not break
down at horizontal tangents on load-deflection curves, which is one form of
critical load condition.
Iterative solutions are based on the determination of a secant stiffness
matrix, which is derived assuming that the current geometry and state of stress is
known (Majid, 1972). The first cycle then gives new values for the current
geometry and state of stress which are then included in the second cycle, and the
sequence is repeated until the assumed values are consistent with the calculated
In general, incremental analysis is theoretically more sound than iterative
analysis. Incremental analysis follows the complete loading history and is able to
detect bifurcations or branching points along equilibrium paths. This is not true of
iterative solutions which may not converge to the lowest equilibrium path which is
of interest in practice. However, for multi-storey frames, such complexities
seldom exist and either form of solution appears satisfactory. Incremental analysis
is considered first since it illustrates the full interaction between bending and axial
displacements for individual elements and will help to indicate the approximations
often made in iterative solutions.

Incremental Analysis

The incremental stiffness matrix for a finite element of a frame can be

derived as follows (Roberts, 1970). It is assumed that all displacements are small
so that nonlinear strains can be related to the initial geometry; this is a satisfactory
assumption for the majority of practical frames. For an element having an initial
transverse imperfection w0, the nonlinear expression for the axial strain ε is
(Timoshenko and Gere,1961)

As mentioned previously, the most significant nonlinear influence in the
elastic behaviour of frames is the influence of axial forces on the flexural stiffness
of members. Tensile axial forces can be considered as increasing the flexural
stiffness while compressive forces decrease the flexural stiffness. If a set of
compressive member axial forces is increased to the extent that the bending
stiffness of the frame as a whole reduces to zero, the frame becomes unstable.
There are a number of ways in which the elastic instability of frames can be
analysed, many of which reduce to the solution of the same basic set of equations
after simplifying assumptions are made.
Vanishing of the Second Variation of Total PotentialEnergy

The most general approach to the analysis of the elastic instability of

structures is based on the vanishing of the second variation of total potential
energy, defined by eqn (2.5) (Roberts and Azizian, 1983). Assuming that the
nodal displacements qiare linear functions of displacement variables, δ2qivanishes
and critical conditions are defined by the equation δ2VP is identical to the right
hand side of eqns (2.31) and (2.32) and is a complete quadratic form which
changes from positive definite to zero, indicating critical conditions, when the
determinant of [KL]+ [KGA] vanishes. Hence critical conditions for the complete
structure occur when
An alternative way of interpreting eqn (2.48) is that critical conditions
occur when the incremental or tangent stiffness matrix becomes singular.
Although eqn (2.48) is of general applicability, it requires a knowledge of the
axial and flexural deformations at the critical points on the loading path. The
analysis can be simplified if it is assumed that prior to the frame becoming
unstable, onlyaxial

Horne’s Method

Horne (1975) proposed an approximate method for determining the elastic

critical loads of plane multi-storey sway frames, the only analytical requirement
being that of performing a standard linear elastic analysis of the frame. The
method can be illustrated by considering the instability of the column of length a
shown in Fig.2.4(a).
Assuming that the column buckles into a state of neutral equilibrium, i.e.
zero kinetic energy, the potential energy remains constant and the loss of potential
energy of the external forces is equal to the increase in strain energy of the
column. If, due to buckling, the end A of the column sways by with a
corresponding axial shortening δū, the governing energy equation can be written
Two extreme cases are now considered for the buckled shape, as shown in Fig.
2.4(c) and (d). The first, which is a simple rigid body rotation, represents the case
in which the columns of a frame are stiff compared with the beams. The second,
which is a pure sway mode, represents the case in which the beams are stiff
compared with the columns. Assuming simple polynomials to represent the
buckled shape δw, the axial shortening is given by


Although considerable effort has been devoted to determining the elastic
critical loads of plane multi-storey sway frames, elastic critical loads have found
little direct application in practice. Sway limitations (Anderson and Islam, 1979;
Majid and Okdeh, 1982) to prevent serious damage to non-structural cladding
ensure that so-called ‘instability effects’ or second order nonlinear effects are
relatively minor. Apart from this, elastic critical loads do not provide a direct
measure of the magnitude of the second order effects and further complicated
calculations are required to provide information of use to designers.
Roberts (1981) proposed a procedure for estimating the magnitude of the

second-order effects, based on a standard linear elastic analysis. The results can
also be used to estimate, very simply, the elastic critical loads of frames if

A linear elastic analysis of a plane, multi-storey sway frame gives

displacement and member forces at all the joints of the frame. For any member, it
is possible to deduce the displacements w, rotation wx, moments m, shear forces ƒ
and axial forces t at the joints or nodes relative to the local coordinate axes of the
Each member is in equilibrium, only in the absence of the axial forces
which produce the second order effects. It is assumed that the member axial forces
produce an additional transverse deflection δw of the member. The corresponding
axial shortening δū is then given by The equivalent load vectors for each member
of the frame can now be determined and a second linear elastic analysis performed
with the frame loaded by all the member equivalent load vectors to give a first
approximation for the second-order effects. Since this is only a first
approximation, the final solution should ideally be obtained iteratively.
A new set of equivalent loads should be calculated from member axial
forces and displacements w+δw and the process repeated until the calculated
values of w+δw are consistent with those assumed. An approximation to the
iterative procedure can however be made as follows. Consider any of the non-zero
displacements, for example the sway of the ith storey of the frame (see Fig.
2.5). Assuming that δw is proportional to w, the final value of , which is
denoted by , is given approximatelyby
Equation (2.68) was expressed in terms of storey sway for a particular
reason. Provided that the primary loading produces sway in each storey, an
estimate of the sway critical load can be made using the principle (Horne and
Merchant, 1965) that member axial forces at a load factor A have the effect of
increasing the deformations corresponding to the lowest elastic critical load by a
factor 1/(1−λ/λcr). If λ is taken as unity for the primary frameloads


Meaningful comparisons of the various nonlinear methods of analysis
discussed are limited, since detailed solutions seldom appear in the literature and
so-called ‘exact’olutions of nonlinear problems are often questionable. However,
all the methods discussed have one aspect in common. After making certain
simplifying assumptions, they all lead to a prediction of the elastic critical loads of
frames and it is this aspect which will be used, herein, for comparison.
Details of the frames analysed are shown in Fig. 2.6. A single element
was used to represent each member (beam or column) of the frame and the axial
forces in members were assumed either statically determinate or as given by a
preliminary linear elastic analysis. With
the exception of methods incorporating stability functions, improved
accuracy is achieved by using more than one element to represent each member,
due to the approximate displacement functions assumed in deriving the element
stiffness matrices. In applying the method discussed in Section 2.6, the frames
were also loaded with horizontal forces at each storey equal to 10% of the vertical
load applied at that storey, to excite sway deformations.
Results of the analysis discussed in Section 2.6 are presented in Table
2.1, in which βivalues for each storey are given, the maximum from which the
elastic critical load factor is calculated being underlined. Also given in Table 2.1
are critical load factors determined in accordance with Section 2.5.1 (eigenvalue
solution), Section 2.5.2 (stability functions) and Section 2.5.3 (Horne’s method).

Matrix formulations for the elastic linear, nonlinear and instability
analysis of frames have been presented. Nonlinear analysis can be performed
either incrementally using the tangent stiffness matrix or iteratively using the
secant stiffness matrix. Rigorous derivations of both the tangent and secant
stiffness matrices indicate fullinteraction between bending and axial
Critical loading conditions occur when the second variation of the total
potential energy changes from positive definite to zero, indicating a transition
from stable equilibrium to instability, this condition being defined by the

vanishing of the determinant of the incremental or tangent stiffness matrix.
For practical multi-storeyed sway frames, sway limitations to prevent
damage to non- structural cladding generally ensure that nonlinear effects are of
only minor significance. This enables simplifying approximations to be introduced
in matrix analysis, the most significant of which is the assumption that member
axial forces are either statically determinate or as given by a preliminary linear
elastic analysis. This assumption is equivalent to considering only the influence of
axial forces on the flexural behaviour of members and neglects axial shortening
due to flexure. Based on this assumption, approximate methods of estimating
second order effects and elastic critical loads can be established, and several
alternative approaches for determining elastic critical loads reduce to the solution
of the same set of equations.

Chapter 3

Methods are described for the design of multi-storey steel frames to

specified limits on horizontal sway deflection. Approximate methods for
rectangular frames require only simple calculations, and their use is illustrated by
a worked example. More general approaches are also given. These necessitate
iterative calculation and take the form of specialised computer programs. Accurate
allowance can then be made for secondary effects which are of particular
significance in the design of very slender unbraced structures.

Design Studies

The frames examined were rectangular in elevation, of four, seven and ten
storeys in height, and from two to four or five bays in width. Two ratios of bay
width to storey height r were considered, namely 1·33 and 2·0, although within a
particular frame these two dimensions were constant. All bases were fixed.
The unfactored loads are given in Table 3.1 together with the maximum

and minimum basic wind speeds. For simplicity, the resulting horizontal wind
pressure was taken as uniform over the height of the frame, although designers
often use a reduced pressure on the lower storeys. On the other hand, no
allowance was made for eccentricity of vertical loading arising from fabrication
and erection tolerances, and no account was taken of any reduction in live loading
permitted for the design of columns. In the studies, the maximum value of floor
loading was combined with minimum values of wind loading, and vice-versa. A
number of other load combinations were examined also.
The design strength of structural steel, py, was taken as 240 N/mm2,
corresponding to the grade commonly used in medium-rise unbraced frames.
Sway due to unfactored horizontal wind load was to be restricted to 1/300 of each
storey height for the bare frame, in accordance with recent recommendations (BSI,
1977; ECCS, 1978).
Minimum sections were determined by designing against failure by
beam-type plastic hinge mechanisms or by squashing, using partial safety factors
γf of 1·4 and 1·6 on dead and imposed load, respectively (BSI, 1977). These
sections were then increased, as appropriate, to satisfy the restriction on sway at
working load. The method of Anderson and Islam (1979a) described below was
used. Column sections were made continuous over at least two storeys, but beam
sections were changed at each floor level if required. The designs were then
subjected to a second-order elasto-plastic computer analysis (Majid and Anderson,
1968), with γf values of 1·4, 1·2 and 1·2 applied to dead, imposed and wind loads,
respectively (BSI, 1977). If the factored load level was achieved before collapse
occurred, then ultimate strength under combined loading was not the governing
criterion for that particular frame.
The results are summarised in Table 3.2. These are applicable to frames
whose steel design strength is in the region of 240 N/mm2, such as British Grade
43 and European Fe 360 material. The designer needs to determine the ratio of the
sum of the column axial forces P to the corresponding total column wind shear F
in each storey. P and F are calculated using the factored combined loads. The ratio
is then averaged over all the storeys of the frame. Ultimate strength under

combined load is not likely to be critical, provided the limits on P/F are not

Consider the six-storey two-bay frame shown in Fig. 3.1 which is
subjected to the unfactored loadings shown in Table 3.3. The dynamic wind
pressure has been calculated from a basic wind speed of 44 m/s. With frames
spaced longitudinally at 4·5 m centres, the factored combined loads are as given in
Fig. 3.1, the corresponding values of P/F being stated alongside each storey. The
average value is 31·5 andr=L/h=1·6. It is clear from Table 3.2 that ultimate
strength will not be critical for design. The appropriate procedure is therefore to
calculate first the sections required to sustain the factored values applicable to
dead plus imposed vertical load, and then to increase the sections as necessary in
order to limit sway at working load. A final check analysis can then be undertaken
to confirm adequate ultimate strength under combinedloading.


When it is expected that the serviceability limit on sway will be dominant,

the method due to Anderson and Islam (1979a, b) enables suitable section
properties to be calculated directly for rectangular frames. The design equations
are based on threeassumptions:

- Vertical loads have a negligible effect on horizontaldisplacements.

- point of contraflexure exists at the mid-height of each column (except in

the bottom storey) and at the mid-length of eachbeam.

- The total horizontal shear is divided between the bays in proportion to

their relative widths.

These assumptions render a frame statically determinate, except in the

bottom storey, and enable each storey to be considered in isolation. Expressions

relating the sway deflection over a storey height to the inertias of the
corresponding columns and surrounding beams can then be derived.

Intermediate Storey
Figure 3.2 shows an intermediate storey of height h2 subject to
horizontal load. By treating each bay individually, it can be shown that
assumptions (ii) and (iii) result in zero axial load in the internal columns. It is also
implied by the assumptions that the column inertias are related asfollows

Bottom Two Storeys of a Fixed Base Frame

A separate analysis is required for the bottom two storeys because it is

grossly inaccurate to assume a point of contraflexure at mid-height of a ground
storey column. Design equations have been derived for pinned base frames
(Anderson and Islam, 1979a, b), but such bases result in very high inertias for the
bottom storey members, in comparison with elsewhere in the frame. Fixed bases
are preferable for multi-storey frames, unless the need to minimise stress on the
soil is of over-riding importance. The equations given below apply to fixed
The subassemblage is shown in Fig. 3.4. It is assumed that the fixity of the
base and the avoidance of reverse column taper result in sections being governed
by the sway ∆ of the storey next to the bottom. The column section is therefore
continuous over the bottom two storeys. Anderson and Islam also derived
equations applicable to fixed base frames when sway of the bottom storey
governed design. This case can arise when the height of the bottom storey is much
greater than that of the storey above. These special equations are avoided by
checking bottom storey sway using the analysis method described later, and
modifying sections ifnecessary.
By comparing Figs. 3.3 and 3.4 it can be seen that eqn (3.6) still applies
for the rotation θB. θG is obtained by deriving thesway


The method is demonstrated by designing the six storey frame,
discussed earlier, to a limiting sway index of 1/300 under unfactored wind
loading. With a longitudinal spacing of 4·5 m, the resulting loads are as shown in
Fig. 3.5(a). E is taken as 205 kN/mm2. Column sections will only be changed
every second storey, to reduce fabrication costs. As a result, it is only necessary to
design three storeys from theframe.

The sections shown in Fig. 3.5(a) are the minimum ones which withstand
dead plus imposed vertical loading only, using γf=1·4 and 1·6, respectively. They
are obtained by simple plastic theory, taking py=240 N/mm2, and selecting from
the range of British Universal sections.
Design for sway is commenced at the fifth storey, using eqn (3.16) to
calculate an inertia for the internal column of 5672 cm4, as shown in Table 3.4. As
can be seen from the sections in the Appendix, the nearest Universal Column
(UC) is 203×203×60 kg/m (I=6088 cm4). This is adopted, as shown in Fig. 3.5(b).
The required inertia for the external column is half that for the internal member.
The lack of a UC with a suitable property necessitates the provision of a
203×203× 46 kg/m section (I=4564 cm4). The beam inertias are now calculated
from eqns (3.10) and (3.8), taking the effective value of I3,2 as the value actually
provided for the internal column (6088 cm4). The required beam inertias are given
in Table 3.4, and the chosen sections in Fig. 3.5(b). The minimum beam section
305×127×37 kg/m is retained for the lower beam as its inertia is 99% of that
A similar procedure is followed for the third storey, except that it is worth
considering two alternatives for the external column. In the first case a
203×203×71 kg/m UC is chosen (I=7647 cm4). Beam desing is then based on an
effective I3,2 of 14307 cm4 corresponding to the actual section of the internal
column. For the alternative, a 203×203× 60 kg/m UC (I=6088 cm4) is proposed
for the external column. As this is less than twice the inertia of 14307 cm4
provided for the internal column, an effective value of I3,2=2×6088=12176 cm4
must be used for beam design. When the required beam inertias in Table 3.4 are
compared with the list of available Universal Beams (UB), it is found that the
same sections are required in both cases. Hence it is more economical to choose

the lighter section for the external columns, as shown in Fig.3.5(b).
The results for the second storey come from the use of eqns (3.22),
(3.21), (3.19) and (3.18). After column sections are chosen, beam design is based
on an effective I3,2 of 17510 cm4.


Provided secondary effects are not significant, the above method

successfully provides section properties needed to satisfy limits on sway. Rarely,
if ever, will these properties correspond exactly to available sections, but
additional column stiffness can be offset by reduced beam stiffness, and vice-
versa. The use of eqns (3.1)–(3.3), though, precludes a similar trade-off between,
for example, internal and external columns. A designer may wish, therefore, to use
an analysis to modify slightly the sections, to achieve greater economy. Suitable
methods are due to Wood and Roberts (1975) and Moy (1974). The former
method has been included in ECCS recommendations (1978) and is described
below. Such analyses can also be used to check deflections in the top or bottom
storeys if, exceptionally, these could be critical.

Substitute Frame

To use the analysis, each storey of the actual frame must be replaced by an
equivalent structure having the form of Fig. 3.6. This is done by first transforming
the actual frame into a substitute beam-column structure, as shown in Fig. 3.8 for
the six storey frame. The basis of the substitute frame (Wood, 1974) is that:

- for horizontal loading on the real frame, the rotations of all joints at any
one level are approximately equal,and each beam restrains a column at

Ultimate Strength

The calculation of ultimate strength is outside the scope of this chapter.

However, it should be noted that when the design of Fig. 3.9 is subjected to the
factored loading (Fig. 3.1), it is found to possess adequate strength. This confirms

the prediction made in Section 3.2.3, that this criterion would not be critical

If it is expected that for a particular frame ultimate strength will be

dominant, then the structure should be designed first to this criterion. The method
of Wood and Roberts then provides a useful check for the sway at working load. If
in fact some deflections are found to be excessive, Moy (1974) has shown that the
most economical procedure is usually to increase the stiffness of the beams.


The simplified methods described above do not allow for the reduction in
frame stiffness due to compressive forces, nor for the effects of axial shortening
and unsymmetrical loading on sway.
The reduction in stiffness can easily be included by using additional
horizontal shears (Vogel, 1983). If Pudenotes the total vertical loading carried by
the columns at storey level u, then the total shear at this level should be increased
by . Moy (1977) has given an estimate of the sway due to differential axial
shortening. This was derived for frames subject to uniform horizontal loading,
with the floors assumed to be rigid and column cross-sectional areas varying
linearly from the top level to the bottom. Anderson and Islam (1979b) used this
approach on a 15-storey frame and found the accuracy to be reasonable. However,
when sway due to differentialaxial shortening or unsymmetrical vertical loading is
likely to be significant, then simplified methods become appropriate only to the
initial design stage. The final design should be obtained using a standard elastic
analysis program to examine trial sections, or from specialised programs, such as
those of Anderson and Salter (1975) or Majid and Okdeh (1982).

Use of Linear Programming

A flow diagram for the procedure due to Anderson and Salter is shown
in Fig. 3.10. Due to the non-linear relationship between deflection and stiffness, a
number of iterations will usually be necessary before the deflection limits are

satisfied. The method is based on standard routines for elastic analysis and linear
programming, and therefore can be developed easily. The procedure is applicable
to a wide variety of frames, and secondary effects are readily included.

Minimum Cost Design

The previous method aims to minimise cost by generating a light
design, but no account is taken of the variations in price per tonne that exist in the
supplier’s section catalogue, nor of the restricted number of sections available.
The program developed by Majid and Okdeh (1982) overcomes these limitations
by costing alternative selections of available sections using a supplier’s price list.
This method also adopts an iterative procedure to avoid direct solution of the
structure’s stiffness equations. The general procedure is explained with reference
to the single storey fixed base frame of Fig. 3.11. For simplicity, axial
deformations will be ignored, and the shear in each column will therefore be F/2,
with equal rotations θ at B andC.

This chapter has described two approximate methods to design
medium-rise unbraced rectangular frames in which sway deflections control the
choice of sections. The method of Anderson and Islam has the advantage that it
actually generates a design without the need for trial analyses. Provided secondary
effects, particularly differential axial shortening, are not significant, the design
will be satisfactory. The choice of sections can be refined by using the analysis
due to Wood and Roberts. Both methods can be used by hand, or programmed for
a micro-computer. The second method enables account to be taken of cladding
If sway due to differential axial shortening is significant, as in very
slender frames, then specialised computer methods are more appropriate. Two
such methods have been described.

Chapter 4

Methods of determining the buckling load factor for regular interbraced

sets of columns and beams are developed and detailed. The equations which allow
the critical load factor for any specific case to be evaluated are set out and
summary charts prepared using such values. Such charts link buckling load to
brace stiffness and number of columns, also allowing for brace eccentricity in
some cases.
Basic in-plane buckling with linear elastic support is detailed and
expanded to cases of flexural torsional buckling involving eccentricity of linear
braces and the addition of rotational and torsional support. Beam systems are
detailed for the cases of uniform moment and for uniform spread load applied at
flanges. The charts included only present a sample of what might be assembled.
The appendices contain the detailed composition of the 2×2 and 4×4 determinants
from which the charts are assembled. The evaluation of these determinants by
desk computer isstraightforward.

The deformed shapes illustrated in Fig. 4.3 are two of the four basic
shape functions which can be used to describe the deformed shape of a general
uniform section beam- column interbrace length. As mentioned, the basic solution
form of eqn (4.8) implies a linear combination of sine, cosine and linear terms in
z, involving four constant scale factors for the components. In the case of a beam-
column, it makes good physical sense to use the two end displacements y and the
two end slopes y′ (Fig. 4.3) as those parameters. This follows normal structural
practice. Since the same forms arise for either x- or y-direction displacements in
the case of the doubly symmetric member, the symbol u(z) will be used to
describe the general shape function forms within amember.

Basic Function Components

Two function forms are required to describe the type of shape function
shown in Fig. 4.3:
Q(z)whereQ(l)=1, Q(0)=Q′(0)=Q′(l)=0
R(z)whereR′(l)=1, R′(0)=R(0)=R(l)=0
Obviously Q and R are linear combinations of the components of the solution to
eqn (4.1) as shown in eqn (4.8), the combinations being arranged so that their end
values are zero or unity. The formulae for the basic s and c functions, for instance,
are derived directly from these using the end curvatures and slopes ofR.
Similar forms may be developed for the flexural-torsional buckling problems
discussed in latersections.

Non-uniform Section, Bracing and Compression

Consider a column pinned at each end and braced at intervals along the
length by elastic linear braces, as illustrated in Fig. 4.5. In such a case the buckling
displacements are assumed to take place in the x−z plane. If the braces are
irregularly spaced, the buckling load will continue to rise with the brace stiffness,
though not necessarily steadily. The mode of buckling with very weak braces
would be close to the half sine wave form into which the unbraced column would
buckle. As the braces are stiffened the effective length decreases, until at some
stage the mode associated with the lowest buckling load will involve two
(distorted) half waves. The practical way to determine the buckling loads is to find
the load factors at which the stiffness matrix of the structure, assembled using s
and c functions, has a zero determinant. This is further illustrated by examining
the simpler, but nonsymmetrical, case shown in Fig. 4.6. The initially straight
uniform column would have a buckling load of π2EI/l2, i.e. PE, if unbraced and
other natural modes of the unbraced column would occur at 4PE, involving two
half sine waves with the length, 9PE involving three half sine waves, etc. The
brace shown would interfere with the pure first mode and would cause the first
critical load to rise above PE continually as the brace stiffness increases. The same
would occur with the second mode, but the three half sine wave mode form
naturally has a node at the brace point and will not cause the brace to be strained.

This buckling load, 9PE, effectively represents an upper limit to the column axial
compression capacity. A plot of buckling load versus brace stiffness would take
the form shown in Fig. 4.7. The three regimes correspond to the first, second and
third modes of buckling, the lower envelope being the effective critical value for
the associated brace stiffness. Specific nondimensional design charts can be
prepared and some are presented later in this chapter.
The compression chords of trusses, such as that shown in Fig. 4.8,
constitute a particular case within this general class of irregular compression
members. The panel lengths are usually equal, the braces equally spaced, but the
compressive force in each chord section is different, varying in a stepped
parabolic manner within a simply supported truss. Braces (purlins) are not always
attached to each panel point of the chord. This class of structure is discussed in
detail by Medland (1977), where the critical load levels are related to brace
stiffness and spacing in a series of nondimensional charts, such as Fig. 4.9. This
figure relates the maximum panel ρ value within the chord at buckling (ρcr) to the
brace stiffness factor ƒ. This nondimensional factor is the actual linear brace
stiffness divided by 12EI/l3, the shear stiffness of a panel chord. In Fig. 4.9 the
four curves relate to cases where only two brace lines are attached (at the 1/3
points) but the number of load points is varied. In this nondimensional form the
three distinct segments of the plots corresponding to the single, double and finally
triple ‘half sine wave’ modes of buckling are apparent and the ρcr value rises with
the number of load points. This rise is a result of a lesser proportion of the chord
length being at peak compression. It is also apparent that the critical load factor
continues to rise if the number of load points is greater than two, in this two brace
case. This is consequence of the non-uniformity in the axial compression within
the chord. No finite brace stiffness will force the chord into a mode which does
not involve displacement at some brace.
The same general analysis applied to a structure comprising a number of
trusses in parallel, their compression chords interbraced by equally spaced lines of
braces as shown schematically in Fig. 4.10, indicates that the curves shown in Fig.
4.9 remain relevant if the nondimensional brace stiffness factor ƒ is divided by the
empirically derived factor

In eqn (4.15), n is the number of bays (number of columns −1). The one set of
curves covers all n. A very similar factor applies to chords under uniform
compression throughout their length.
Figure 4.11 further illustrates the type of design chart which can be prepared from
such analyses. In this case it is assumed thatevery
second load point of the chord is braced. As the number of braces (and panels) is
increased the curves become smoother and lower, crowding towards a ρcrvalue of
1.0 for large brace stiffnesses. The broken line indicates the lower bound curve for
the special case of column sets under uniform compression. This curve can be
seen to be a not too conservative lower bound for the stepped parabolic cases, and
the results of a more complete study allow such lower bound design curves to be
calculated efficiently.

Uniform Section, Bracing and Compression

Figure 4.12 illustrates a highly regular set of parallel, pin-ended, uniformly

compressed columns interbraced at equal intervals by equal stiffness braces. As
stated above, this case provides a conservative estimate for some less regular
cases. The interbraced column length (element) of column number m between
brace lines n and n+1 is designated element n,m. There are N internal brace lines
and N+1 columnelements.
The displacements within column element n,m are governed by the
basic differential equation (eqn (4.1)), arranged here in the general
for which a basic form of solution is given in eqn (4.14). It is reiterated that the
functions Q and R are such that, at an end, one of them will have a value of 1.0,
the other zero. The same applies to their derivatives.

Single Column Example

Consider initially the single column shown in Fig. 4.13. The linear springs
brace each point to a solid foundation on each side. When buckling occurs under
the uniform axial compression, the displacement, slope, moment and shear must
be continuous at each brace point. Using the displacement form of eqn (4.14), and
capitalising on the special forms of Q(z) and R(z), the continuities of one brace
point provide a general relationship. At node n,m joining elements n−1, m and
n,m the four continuities in the above order are written
un−1,m(l)=un,m(0)=Dn,m (say)

The form of eqns (4.22) simply recognises the recurrent form (physically
and, hence, mathematically) of the structure and buckling modes. Equations (4.22)
state that the distribution of brace point displacements Dmwithin the length of the
column is sinusoidal, while the distribution of slopes Bmat those points is
cosinusoidal. This implies pinned support at nodes 0 and N+1. The number of half
sine waves within the length is obviously j. The detail of the approach is treated in
Segedin and Medland (1978) and Medland (1979), and draws on a recurrence
approach put forward for a vibration problem by Miles(1956).
Equations (4.23) and (4.24) constitute a pair of homogeneous linear
equations in Bmand Dm, the slope and displacement amplitude factors. In the
manner of all buckling problems, no unique solution is available. Apart from the
trivial Bm=Dm=0 solution, a range of proportions Bm: Dmare able to be found and
each corresponds to an axial force (represented in t through λ) which causes the
2×2 determinant of the coefficients in eqns (4.23) and (4.24) to become zero.
While the form of the development of the two basic equations may appear
complicated, the evaluation of the 2×2 determinant at a set β and θjand increasing
trial values of t is almost trivial once programmed into a desk computer or

Multiple Column Case
The effect on the above development of there being M columns in
parallel, as shown in Fig. 4.12, is confined to the fact that each spring at a node is
extended by the relative displacements of the columns. As a result, the third
component of eqn (4.20) becomes


A uniform, doubly symmetric cross-section member under axial
compression can buckle in three distinct forms. The lowest buckling load would
normally involve bending about the minor axis in a simple Euler manner, at a load
dictated by the section properties and end support conditions. If this mode is
prevented, or considerably restrained by outside influence, both Euler buckling
about the major axis and pure twist buckling about the longitudinal axis through
the shear centre of the cross-section are possible. If the section has only one
degree of symmetry (e.g. a channel section, an equal leg angle or an I-section with
unequal flange widths) the shear centre will not coincide with the centroid and the
buckling mode will be either pure displacement in the plane of the axis of
symmetry or combined ‘lateral’ displacement and twist. If no symmetry exists, the
buckling mode contains components of both lateral displacements as well as twist,
as illustrated in Fig. 4.15. Adoubly symmetric member on which the axial
compression stress is not symmetrically disposed over the cross-section will
behave in a similar manner. These latter cases are generally referred to as
undergoing flexural-torsional buckling. Given simple support conditions at each
end against each form of displacement, the buckling component forms will all
vary sinusoidally within the length, each having its own amplitude dependent on
its inherent stiffness in thatmode.
In the following, such combined modes occur not as a result of the
section properties (which are doubly symmetric) but because the elastic
constraints (braces) are attached eccentrically to the compression member. As
mentioned earlier, such eccentric attachments precipitate flexural torsional
buckling but often provide some restraint to those movements in compensation.
In members which are subjected to torsion, warping occurs. The torsional shears

cause the basic ‘plane-sections-remain-plane’ assumption to be violated in all but
circular cross-section members. If this distortion is resisted by a stiffening
attachment or, more subtly, by there being non-uniform torsion or torsional
resistance within the member, additional local stresses are engendered. Resistance
to warping considerably enhances the torsional stiffness of the member,
particularly for I and channel shapes, but the resulting stresses can cause local
problems. Figure 4.16(a) illustrates the distortions involved in the free twisting of
an I-section member, while Fig. 4.16(b) shows the extra distortion and
consequently torsional resistance and energy absorption which accompanies the
twist where warping is resisted. The figure illustrates how the warping effect in a
flanged member is dominated by the flanges moving out of plane with one

Interbraced Columns in Flexural-Torsional Buckling

In most interbraced structures of the type discussed in Section 4.3.4, the

bracing members are not simply tension-compression members but, like girts and
purlins, have considerable bending resistance themselves. The connection with the
braced member is often eccentric to the shear centre of that member and is also
capable of transferring rotational forces in the plane of the structure. In such cases
the brace mayprovide torsional, rotational and warping support as well as the basic
lateral, and at the same time causes the buckling mode of the braced member to
involve those four components. Figure 4.17 illustrates a typical connection within
a grid and the associated support forms.
Since out-of-plane displacement is assumed to be zero in the buckling modes, the
four further continuity equations of twist , rate of change of twist , torsional
moment and warping moment must be satisfied. As an example, the torsional
moment continuity across a brace point at which the effective stiffness resisting
torsion is KT (moment per unit twist), the linear in-plane spring stiffness is KL and
the eccentricity of that brace from the shear centre is e, can be written
The coefficient of KT contains the direct and carryover twist effects, while
the other two involve the contributions due to the net extension of the linear
spring, attached eccentrically, by the amount e. The sign convention for

displacements and forces is shown in Fig. 4.18. The form selected for is
the same as that for un,min eqn (4.14). The with different coefficients and using α
instead of λ. The full set of continuity equations across the general brace point can
be written. These are presented in detail by Medland (1979). The factor α
represents the axial load and is definedby
Upon substitution of eqns (4.22) and (4.27), the set of four equations
which are the four degrees of freedom equivalent of the pair of equations (4.28)
and (4.29) may be written and their 4×4 determinant evaluated at increasing levels
of axial force until it becomes zero at the fact that column members exist which
have very similar areas and minor axis I-values but very different warping
constants (for example) is the reason for the lack offurther nondimensionalisation.
Figure 4.19 illustrates the effects of linear brace eccentricity on critical ρ value for
two columns of similar area and flange width but different torsional properties.
Figures 4.20 and 4.21, respectively, indicate the effect of adding torsional and
rotational bracing to a basic eccentric linear braced system. As values of the
nondimensional torsional and rotational brace stiffnesses are raised, the critical ρ
values associated with a given linear support stiffness rise, compensating in some
measure for the eccentricity of the linear braces. The broken curve on each figure
is the ρcr: β plot for a zero eccentricity four linear brace datum case. For realistic
cases, in-plane rotational braces are, not surprisingly, relatively inefficient in
compensating for what is basically a torsional effect. The scale of the
nondimensional rotational stiffness factor γ is small compared to β and η. Specific
cases selected to illustrate the sensitivities are presented in Medland(1979).


Flooring systems comprising parallel beams linked together by lateral
members have the same general features as the column sets discussed previously.
The lateral bracing members will normally be eccentric and may provide
rotational, torsional and lateral bracing. The beams, however, will always buckle
in a flexural-torsional mode. The governing differential equations for an interbrace
beam length under uniform moment M are

The nondimensional eccentricity factor e* is again used for any attached bracing
andplotsofρcr(=Mcr/M0)against brace stiffness and eccentricity can be assembled
sin Fig.

The effect of adding a

torsional brace at the shear centre of the section is illustrated in Fig. 4.23,
while the effect of rotational bracing with eccentric linear is shown in Fig. 4.24.
To determineany specific value on such charts involves the evaluation of the
determinant of a 4×4 system of linear equations in the same way as the flexural-
torsional braced column cases were handled. The detail of the component factors
is contained in Medland (1980). The detailed set of equations whose determinant
is zero at a buckling load is shown in Appendix 3, with explanatorynotes.
For any buckling involving torsion, the geometry of the cross-section has a
marked effect. The a/l ratio is a means of categorising this effect. Figure 4.25
indicates this sensitivity by comparing the critical uniform moment values of
torsionally constrained members at different a/l ratios.
The case where the loading of the beam is due to a uniform spread load
is more common than that of a uniform moment. Such loading is


While the elastic critical loads, the determination of which has been the
subject of Sections 4.1 to 4.5, are important parameters for the designer, a means
of estimating the strength required of the braces must also be found. To determine
the critical loads, only the brace stiffness needs to be considered. Strength is
required if the braced members deform. Some bracing systems are primarily
designed to carry loading to a foundation and will be designed accordingly. If a
brace is placed basically to prevent buckling, it theoretically needs no strength
until buckling occurs. In practice the compression element being braced (strut,
compression flange, etc.) is not perfectly straight and is subjected to secondary
loading which pushes it off line. This results in the braces being strained when the
strut isloaded.

In this section the compression members are assumed to have a specific
crookedness before axial loading is applied. In general, such an initial shape
between the ends of the member can be expressed as a Fourier sine series which,
in the case of a pin-ended column, can be regarded as a series containing the
successive buckling mode shape functions, say
Upon application of a compressive force P, the jth component of the
series is magnified by division by a factor homogeneous because no rotational
springs are attached. The linear springs are stretched by the amount (u−u I) at each
brace point and the uI component makes that equation nonhomogeneous. A
specific ‘magnified’ initial displacement set is defined by the solution of these two
simultaneous equations under any applied axial force. If that force reaches any one
of the basic critical loads of the original perfectly straight system the
displacements become infinite. Obviously the applied axial force must remain
below the lowest critical load of the system.
A detailed derivation of the relationship between the forces in the braces
and the spring stiffnesses at a given axial force level for the type of structure
shown in Fig. 4.23 is presented by Medland and Segedin (1979). Figure 4.27
illustrates the type of non dimensi on al chart which can be prepared from suchan
analysis.Thefactor represents the brace force as a percentage of the axial column
force P, divided by the number of columns in parallel M. The symbol is the
actual brace stiffness K, divided by the nondimensionalising factor 12EI/l 3 and
further divided by 2(1−cos θi) which incorporates the form of the


A coordinated approach to the problem of elastic lateral and flexural-
torsional buckling in multiply braced column and beam members has been
summarised. In very regular systems (e.g. uniform axial compression or moment)
the calculation of the buckling load factor is reduced to the evaluation of 2×2 or
4×4 stiffness determinants at increasing load factors until the determinant
becomes zero. This may have to be repeated for two or three trial modes

Relationships between brace stiffness and column properties for any
number of equally spaced brace lines and columns (or beams) are summarised in
these determinants. In less regular cases (e.g. stepped parabolic axial compression)
a larger, but sparse, matrix determinant is involved. For some of these cases it has
been established that the corresponding uniform case provides a safe and not too
conservative bound. For beams under uniform spread load a mixed analysis
involving a beam-column finite element and some recurrence capitalisation is
used. Eccentricity of load and bracing is covered in the analyses. By assuming
initially deformed members, formulae and charts for brace strength requirements
are putforward.

2.2. Ringkasan Isi Buku Kedua

Before starting to design a structure it is important to clarify what
purpose it is to serve. This may seem so obvious that it need not be stated, but
consider for example a building, e.g. a factory, a house, hotel, office block etc.
These are among the most common structures that a structural engineer will be
required to design. Basically a building is a box-like structure, which encloses
Why enclose the space? To protect people or goods? From what?
Burglary? Heat? Cold? Rain? Sun? Wind? In some situations it may be an
advantage to let the sun shine in the windows in winter and the wind blow
through in summer (Figure1.1). These considerations will affect the design.

Figure 1.1 Design to use sun, wind and convection

How much space needs to be enclosed, and in what layout? Should it be
all on ground level for easy access? Or is space at a premium, in which case
multi-storey may be justified (Figure1.2). How should the various parts of a
building be laid out for maximum convenience? Does the owner want to make
a bold statement or blend in with the surroundings?

The site must be assessed: what sort of material will the structure be built
on? What local government regulations may affect the design? Are cyclones,
earthquakes or snow loads likely? Is the environment corrosive?

Architects rather than engineers are usually responsible for the problem
formulation and conceptual design stages of buildings other than purely
functional industrial buildings. However structural engineers are responsible for
these stages in the case of other industrial
buildings. Engineers sometimes accuse architects of designing weird
structures that are not sensible from a structural point of view, while architects in
return accuse structural engineers of being concerned only with structural issues
and ignoring aesthetics and comfort of occupiers. If the two professions
understand each other’s points of view it makes for more efficient, harmonious

Figure1.2 Low industrial building and

high rise hotel

The following decisions need to be made:

1. Who is responsible for which decisions?

2. What is the basis for payment for work done?
3. What materials should be used for economy, strength, appearance, thermal
and sound insulation, fire protection, durability? The architect may

have definite ideas about what materials will harmonise with the
environment, but it is the engineer who must assess their functional
4. What loads will the structure be subjected to? Heavy floor loads?
Cyclones? Snow?Earthquakes? Dynamic loads from vibrating machinery?
These questions are firmly in the engineer’s territory.

Besides buildings, other types of structure are required for various

purposes, for example to hold something vertically above the ground, such as
power lines, microwave dishes, wind turbines or header tanks. Bridges must span
horizontally between supports. Marine structures such as jetties and oil platforms
have to resist current and wave forces. Then there are moving steel structures
including ships, trucks and railway rolling stock, all of which are subjected to
dynamic loads.
Once the designer has a clear idea of the purpose of the structure, he
or she can start to propose conceptual designs. These will usually be based on
some existing structure, modified to suit the particular application. So the more
you notice structures around you in everyday life the better equipped you will be
to generate a range of possible conceptual designs from which the most
appropriate can be selected.


Steel is roughly three times more dense than concrete, but for a given
load-carrying capacity, it is roughly 1/3 as heavy, 1/10 the volume and 4 times as
expensive. Therefore concrete is usually preferred for structures in which the
dead load (the load due to the weight of the structure itself) does not
dominate, for example walls, floor slabs on the ground and suspended slabs
with a short span. Concrete is also preferred where heat and sound insulation are
required. Steel is generally preferable to concrete for long span roofs and bridges,
tall towers and moving structures where weight is a penalty. In extreme cases
where weight is to be minimised, the designer may consider aluminium,

magnesium alloy or FRP (fibre reinforced plastics, e.g. fibreglass and carbon
fibre). However these materials are much more expensive again. The designer
must make a rational choice between the available materials, usually but not
always on the basis of cost.
Although this book is about steel structures, steel is often used with
concrete, not only in the form of reinforcing rods, but also in composite
construction where steel beams support concrete slabs and are connected by shear
studs so steel and concrete behave as a single structural unit (Figs.1.4, 1.5). Thus
the study of steel structures cannot be entirely separated from concrete structures

Having decided on the overall form of the structure (e.g. single level
industrial building, high rise apartment block, truss bridge, etc.) and its location
(e.g. exposed coast, central business district, shielded from wind to some extent
by other buildings, etc.), we can then start to estimate what loads will act on the
structure. The former SAA Loading code AS 1170 has now been replaced by
AS/NZS 1170, which refers to loads as “structural design actions.” The main
categories of loading are dead, live, wind, earthquake and snow loads. These will
be discussed in more detail in Chapter 2. A brief overview is given below.

Once we know the shape and size of the structure and the loads that may
act on it, we can then analyse the effects of these loads to find the maximum load
effects (action effects), i.e. axial force, shear force, bending moment and
sometimes torque on each member. Basic analysis of statically determinate
structures can be done using the methods of engineering statics, but statically
indeterminate structures require more advanced methods. Before desktop
computers and structural analysis software became generally available, methods
such as moment distribution were necessary. These are laborious and no longer
necessary, since computer software can now do the job much more quickly and
efficiently. An introduction to one package, Spacegass, is provided in this book.

However it is crucial that the designer understands the concepts and can
distinguish a reasonable output from a ridiculous output, which indicates a
mistake in data input.


After the analysis has been done, we can do the detailed design –
deciding what cross section each member should have in order to be able to
withstand the design axial forces, shear forces and bending moments. The
principles of solid mechanics or stress analysis are used in this stage. As
mentioned above, dead loads will depend on the trial sections initially assumed,
and if the actual member sections differ significantly from those originally
assumed it will be necessary to adjust the dead load and repeat the analysis and
member sizing steps.
We also have to design connections: a structure is only as strong as its
weakest link and there is no point having a lot of strong beams and columns etc
that are not joined together properly.
Finally, we must document our design, i.e. provide enough information so
someone can build it. In the past, engineers generally provided dimensioned
sketches from which draftsmen prepared the final drawings. But increasingly
engineers are expected to be able to prepare their own CAD drawings.

Chapter 2

To design effectively it is necessary to know something about the
properties of the material. The main properties of steel, which are of importance
to the structural designer, are summarised in this chapter.


Steel is the strongest, stiffest and densest of the common building
materials. Spring steels can have ultimate tensile strengths of 2000 MPa or more,
but normal structural steels have tensile and compressive yield strengths in the
range 250-500 MPa, about 8 times higher than the compressive strength and over
100 times the tensile strength of normal concrete. Tempered structural aluminium
alloys have yield strengths around 250 MPa, similar to the lowest grades of
structural steel.

Although yield strength is an important characteristic in determining

the load carrying capacity of a structural element, the elastic modulus or
Young’s modulus E, a measure of the stiffness or stress per unit strain of a
material, is also important when buckling is a factor, since buckling load is a
function of E, not of strength. E is about 200 GPa for carbon steels, including all
structural steels except stainless steels, which are about 5% lower. This is about
3 times that of Aluminium and 5-8 times that of concrete. Thus increasing the
yield strength or grade of a structural steel will not increase its buckling capacity.

The specific gravity of steel is 7.8, i.e. its mass is about 7.8 tonnes/m3,
about three times that of concrete and aluminium. This gives it a strength to
weight ratio higher than concrete but lower than structural aluminium.

Structural steels are ductile at normal temperatures under normal
conditions. This property has two important implications for design. First, high
local stresses due to concentrated loads or stress raisers (e.g. holes, cracks, sudden
changes of cross section) are not usually a major problem as they are with high
strength steels, because ductile steels can yield locally and relive these high
stresses. Some design procedures rely on this ductile behaviour. Secondly, ductile
materials have high “toughness,” meaning that they can absorb energy by plastic
deformation so as not to fail in a sudden catastrophic manner, for example
during an earthquake. So it is important to ensure that ductile behaviour is
The factors affecting brittle fracture strength are as follows:
(1) Steel composition, including grain size of microscopic steel structures, and
the steel temperature history.
(2) Temperature of the steel in service.
(3) Plate thickness of the steel.
(4) Steel strain history (cold working, fatigue etc.)
(5) Rate of strain in service (speed of loading).
(6) Internal stress due to welding contraction.

In general slow cooling of the steel causes grain growth and a reduction in
the steel toughness, increasing the possibility of brittle fracture. Residual
stresses, resulting from the manufacturing process, reduce the fracture
strength, whilst service temperatures influence whether the steel will fail in brittle
or ductile manner.

Metallurgy and transition temperature

Every steel undergoes a transition from ductile behaviour (high energy
absorption, i.e. toughness) to brittle behaviour (low energy absorption) as its
temperatures falls, but this transition occurs at different temperatures for different
steels, as shown in Fig.2.1 below. For low temperature applications L0
(guaranteed “notch ductile” down to 0 C) or L15 (ductile down to -15 C) should

be specified.

Stress effects
Ductile steel normally fails by shearing or slipping along planes in the
metal lattice. Tensile stress in one direction implies shear stress on planes
inclined to the direction of the applied stress, as shown in Fig.2.2, and this can be
seen in the necking that occurs in the familiar tensile test specimen just prior to
failure. However if equal tensile stress is applied in all three principal directions
the Mohr’s circle becomes a dot on the tension axis and there is no shear stress to
produce slipping. But there is a lot of strain energy bound up in the material, so
it will reach a point where it is ready to fail suddenly. Thus sudden brittle fracture
of steel is most likely to occur where there is triaxial tensile stress. This in turn is
most likely to occur in heavily welded, wide, thick sections where the last part of
a weld to cool will be unable to contract as it cools because it is restrained in all
directions by the solid metal around it. It is therefore in a state of residual triaxial
tensile stress and will tend to pull apart, starting at any defect or crack

Case study: King’s St Bridge, Melbourne

The failure of King’s St Bridge in Melbourne in 1962 provided a good
example of brittle fracture. One cold morning a truck was driving across the bridge
when one of the main girders suddenly cracked (Fig.2.3). Nobody was injured
but the subsequent enquiry revealed that some of the above factors had
combined to cause the failure.
1. A higher yield strength steel than normal was used, and this steel was
less ductile and had a higher brittle to ductile transition temperature
than the lower strength steels the designers were accustomed to.
2. Thick (50 mm) cover plates were welded to the bottom flanges of the
bridge girders to increase their capacity in areas of high bending
3. These cover plates were correctly tapered to minimise the sudden
change of cross
section at their ends (Fig.2.2), but the welding sequence was wrong in

some cases: the ends were welded last, and this caused residual triaxial
tensile stresses at these critical points where stresses were high and the
abrupt change of section existed.

Steelwork can be designed to avoid brittle fracture by ensuring that welded

joints impart low restraint to plate elements, since high restraint could initiate
failure. Also stress concentrations, typically caused by notches, sharp re-entrant
angles, abrupt changes in shape or holes should be avoided.

The properties of steel are more predictable than those of concrete,
allowing a greater degree of sophistication in design. However there is still some
random variation in properties, as shown in Fig.2.4.
Although steel is usually assumed to be a homogeneous, isotropic
material this is not strictly true, as all steel includes microscopic impurities,
which tend to be preferentially oriented in the direction of mill rolling. This
results in lower toughness perpendicular to the plane of rolling (Fig.2.5).
Some impurities also tend to stay near the centre of the rolled item due to
their preferential solubility in the liquid metal during solidification, i.e. near the
centre of rolled plate, and at the junction of flange and web in rolled sections.
The steel microstructure is also affected by the rate of cooling: faster cooling will
result in smaller crystal grain sizes, generally resulting in some increase in
strength and toughness. (Economical Structural Steel Work [3])
As a result, AS 4100 [4] Table 2.1 allows slightly higher yield stresses
than those implied by the steel grade for thin plates and sections, and slightly
lower yield stresses for thick plates and sections. For example the yield stress for
Grade 300 flats and sections less than 11 mm thick is 320 MPa, for thicknesses
from 11 to 17 mm it is 300 MPa and for thicknesses over 17 mm it is 280 MPa.

Normal structural steels corrode quickly unless protected. Corrosion
protection for structural steelwork in buildings forms a special study area. If the
structural steelwork of a building includes exposed surfaces (to a corrosive
environment) or ledges and crevices between abutting plates or sections that
may retain moisture, then corrosion becomes an issue and a protection system is
then essential. This usually involves consultation with specialists in this area.
The choice of a protection system depends on the degree of corrosiveness
of the environment. The cost of protection varies and is dependent on the
significance of the structure, its ease of access for maintenance as well as the
permissible frequency of maintenance without inconvenience to the user.
Depending on the degree of corrosiveness of the environment, steel may need:

Epoxy paint
ROZC (red oxide zinc chromate) paint
Cold galvanising (i.e. a paint containing zinc, which acts as a sacrificial
coating, i.e. it corrodes more readily than steel)
Hot dip galvanising (each component must be dipped in a bath of
molten zinc after fabrication and before assembly)
Cathodic protection, where a negative electrical potential is maintained
in the steel, i.e. an oversupply of electrons that stops the steel losing

electrons and forming Fe ++ or Fe

+++ and hence an oxide.

Sacrificial anodes, usually of zinc, attached to the structure, which lose

electrons more readily than the steel and so keep the steel supplied with electrons
and inhibit oxide formation.


Before any detailed sizing of structural elements can start, it is necessary
to start to estimate the loads that will act on a structure. Once the designer and
the client have agreed on the purpose, size and shape of a proposed structure and
what materials it is to be made of, the process of load estimation can begin.
Loads will always include the self-weight of the structure, called the “dead
load.” In addition there may be “live” loads due to people, traffic, furniture, etc.,
that may or may not be present at any given time, and also loads due to wind,
snow, earthquakes etc. The required sizes of the members will depend on the
weight of the structure but will also contribute to the weight. So load estimation
and member sizing are to some extent an iterative process in which each affects
the other. As the designer gains experience with a particular type of structure it
becomes easier to predict approximate loads and member sizes, thereby reducing
the time taken in trial and error. However the inexperienced designer can save
time by intelligent use of some short cuts. For example the design of structures

carrying heavy dead loads such as concrete slabs or machinery may be dominated
by dead load. In this case it may be best to size the slabs or machinery first so the
dead loads acting on the supporting structure can be estimated. On the other hand
many steel- framed industrial buildings in warm climates where snow does
not fall can be designed mainly on the basis of wind loads, since dead and live
loads may be small enough in relation to the wind load to ignore for preliminary
design purposes. The wind load can be estimated from the dimensions of the
structure and its location. Members can then be sized to withstand wind loads and
then checked to make sure they can withstand combinations of dead, live and
wind load. Where snowfall is significant, snow loads may be dominant.
Earthquake loads are only likely to be significant for structures supporting a lot
of mass, so again the mass should be estimated before the structural elements are


Dead load is the weight of material forming a permanent part of the
structure, and in Australian codes it is given the symbol G. Dead load estimation
is generally straightforward but may be tedious. The best way to learn how to
estimate G is by examples.

Example: Concrete slab on columns

Probably the simplest form of structure – at least for load estimation - is a
concrete slab supported directly on a grid of columns, as shown in Fig.3.1.

Suppose the concrete (including reinforcing steel) weighs 25 kN/m3, the

slab is 200 mm (0.2 m) thick, and the columns are spaced 4 m apart in both
directions. We want to know how much dead load each column must support.
First, we work out the area load, i.e. the dead weight G of one square

metre of concrete slab. Each square m contains 0.2 m3 of concrete, so it will

weigh 0.2x25 = 5 kN/m2 or G = 5 kPa.

Next, we multiply the area load by the tributary area, i.e. the area of slab
supported by one column. We assume that each piece of slab is supported by the
column closest to it. So we can draw imaginary lines half way between each row

of columns in each direction. Each internal column (i.e. those that are not at the

edge of the slab) supports a tributary area of 16 m2, so the total dead load of the
slab on each column is 16x5 = 80 kN.
Assuming there is no overhang at the edges, edge columns will support a
little over half as much tributary area because the slab will presumably come to

the outer edge of the columns, so the actual tributary area will be 2.1x4 = 8.4 m2

and the load will be 42 kN. Corner columns will support 2.1x2.1 = 4.42 m2 and a
load of 22.1 kN.
To find the load acting on a cross-section at the bottom of each column
where G is maximum, we must also consider the self-weight of the column.
Suppose columns are 150UC30 sections (i.e steel universal columns with a mass
of 30 kg/m, 4m high between the floor and the suspended slab. The weight of one
column will therefore be 30x9.8/1000x4 = 1.2kN approximately. Thus the total
load on a cross section of an internal column at the bottom will be 80 + 1.2 = 81.2
If there are two or more levels, as in a multi-level car park or an office
building, the load on each ground floor column would have to be multiplied by
the number of floors. Thus if our car park has 3 levels, a bottom level internal
column would carry a total dead load G = 3x81.2
= 243.6 kN.

Concrete slab on steel beams and column

A more common form of construction is to support the slab on beams,
which are in turn supported on columns as shown in Figs.3.2 and 3.3 below.
Because the beams are deeper and stronger than the slab, they can span further so
the columns can be further apart, giving more clear floor space.
To calculate the dead load on the beams and columns, we now add
another step in the calculation. Assuming we still have a 200mm thick slab, the
area load due to the slab is still the same, i.e. 5 kPa.

Assume columns are still of 200x200mm section, at 4 m spacing in one
direction. But we now make the slab span 4m between beams, and the beams span
8m between columns. So we have only half as many columns. But we now want to
know the load on a beam. We could work out the total load on one 8m span of
beam. But it is normal to work out a line load, i.e. the load per m along the
beam. The tributary area for each internal beam in this case is a strip 4m wide,
as shown in the diagram above. So the line load on the beam due to the slab only

is 5 kN/m2 x 4 m = 20 kN/m. Note the units

We must also take into account the self-weight of the beam.
Suppose the beams are 610UB101 steel universal beams weighing
approximately 1 kN/m. The total line load G on the internal beams is now 20 + 1
= 21 kN/m. This will be the same on each floor because each beam supports only
one floor. The lower columns take the load from upper floors but the beams do
not. A line load diagram for an internal beam is shown in Fig.3.4 below. Note
that we specify the span (8 m), spacing (4 m), load type (G) and load magnitude
(21 kN/m).

Unlike car parks, most buildings have walls, and we can estimate their
dead weight in the same way as we did with slabs, columns and beams.
Sometimes walls are structural, i.e. they are designed to support load. Other walls
may be just partitions, which contribute dead weight but not strength. These non-
structural partition walls are common because it is very useful to be able to knock
out walls and change the floor plan of a building without having to worry about it
falling down.
Suppose a wall is 100 mm thick and is made of reinforced concrete

weighing 25 kN/m3. The weight will be 25 x 0.1 = 2.5 kN/m2 of wall area. If

it is 4 m high, it will weigh 4m 2.5 kN/m2 = 10 kN/m of wall sitting on the

floor, i.e. the line load it will impose on a floor will be 10 kN/m. The SAA
Loading Code AS 1170 Part 1, Appendix A, contains data on typical weights of
building materials and construction. For example a concrete hollow block
masonry wall 150 mm thick, made with standard aggregate, weighs about 1.73

kN/m2 of wall area. A 2.4 m high wall of this type of blocks will impose a line
load of 1.73 2.4 = 4.15 kN/m.

Light steel construction
Although the dead weight of steel and timber roofs and floors is much less
than that of concrete slabs, it must still be allowed for. The principles are still the
same: sheeting is supported on horizontal “beam” elements, i.e. members designed
to withstand bending “beams,” i.e. flexural members, running at right angles to
each other. These have special names, which are shown in the diagrams below.

Roof construction
Corrugated metal (steel or aluminium) roof sheeting is normally supported
on relatively light steel or timber members called purlins which run horizontally,
i.e. at right angles to the corrugations which run down the slope. In domestic
construction, tiled roofing is common. Tiles require support at each edge of each
tile, so they are supported on light timber or steel members called battens, which
serve the same purpose as purlins but are at much closer spacing, usually 0.3m.
The purlins or battens are in turn supported on rafters or trusses. Rafters
are heavier, more widely spaced steel or timber beams running at right angles to
the purlins or battens, as shown in Fig.3.5, and spanning between walls or
Purlins usually span about 5 to 8 m and are usually spaced about 0.9 to 1.5
m apart. This spacing is dictated partly by the distance the sheeting can span
between purlins, and partly by the fact that it is easier to erect a building if the
purlins are close enough to be able to step from one to another before the
sheeting is in place.


Although analysis is an essential stage in design, some textbooks on the
design of steel structures pay little attention to this stage. Textbooks on structural
analysis, on the other hand, can be rather divorced from the practicalities of
design. Section 4 of AS 4100[1] is entitled “methods of structural analysis,” and it
contains some guidelines and rules, but on its own it is not enough to guide the
designer through the analysis process. AS 4100 Supplement – 1999, the Steel
Structures Commentary [2], contains further explanation and lists a large number
of references on analysis. The present chapter does not attempt to duplicate
existing references but provides some brief explanatory material including
diagrams and examples where it is felt that these will help to clarify the provisions
of AS 4100[1].


AS 4100[1] Clause 4.1.1 refers to three methods of analysis: elastic,
plastic and “advanced.” Most analysis is now done using commercial software
packages such as Spacegass[3], Microstran[4] and Multiframe[5], using elastic
methods, and this chapter will focus mainly on this approach, with a brief section
on plastic analysis. Neither the code nor the commentary give any guidance as to
how “advanced” analysis may be carried out.
Clause 4.1.2 of AS 4100[1] defines braced members, in which transverse
displacement of one end of the member relative to the other is prevented, and
sway members, in which such displacement is allowed. Examples of braced
members include all the members in a braced frame Fig.4.1 (b) and the
horizontal members in a rectangular sway frame Fig.4.1 (a). Examples of
sway members include the vertical members in a rectangular sway frame Fig.4.1
(a) and the columns and rafters in a pitched roof portal frame.

Clause 4.2 of AS 4100[1] distinguishes between “rigid,” “semi-rigid” and
“simple” construction. Most designers assume either rigid construction in which
the angles between members do not change, or simple construction in which
connections are assumed not to develop bending moments. Examples are shown in
Semi-rigid construction, in which connections provide some flexural
restraint but may not maintain original angles, may exist in reality but is more
difficult to analyse and this analysis is usually avoided in practice.


Clause 4.3 deals with some assumptions that can be made to simplify
structural analysis in some situations. For example the analysis of regular shaped
structures with a large number of members can be simplified by treating sub-
structures in isolation from the rest of the structure.

Treating sub-structures in isolation from the rest of the structure

For example the regular three-dimensional structure shown in Fig.4.3(a)
below has 3 storeys, 4 bays in the X direction and 1 bay in the Z direction.
According to Clause 4.3.1(a), this could be treated as a series of two-dimensional
frames of 3 storeys and 4 bays in the x-y plane, and a series of two-dimensional
frames of 3 storeys and 1 bay in the y-z plane as long as loads and stiffness do not
vary markedly from one bay to another Clause 4.3.1(b) deals with vertical loads
on braced multi-storey building frames, in which a floor level plus the columns
above and below can be treated as a sub-structure in isolation from the rest of
the structure, as shown in Fig.4.3(b) below.

The effect of finite width of members
Structural analysis of skeletal structures treats structural members as "line
elements," ignoring the fact that they actually have a finite width. Because of this
finite width, it is usually impracticable to make connections at the centroid of
each member. For example floor beams in simple construction multi-storey
buildings must be joined to the sides of columns, and the weight supported by the
beam does not act through the centroid of the column. The column must therefore
carry not only axial force but bending moment also, as shown in Fig.4.3(c) below
(unless there is another beam on the other side that exerts an equal and
opposite moment on the column).They must therefore be designed as beam-
columns, i.e. members under combined axial and bending loads, using Section 8
of AS 4100.
Clause 4.3.2 of AS 4100 defines the span length as the centre to centre
distance between supports, not the actual length of the beam, as shown in
Fig.4.3(c) below. Clause 4.3.4 of AS4100 specifies the minimum eccentricity e of
the load R from a simply supported beam acting on a column, as shown in
Fig.4.3(c): “A beam reaction or a similar load on a column shall be taken as
acting at a minimum distance of 100mm from the face of the column towards the
span or at the centre of the bearing, whichever gives the greater eccentricity.”

Illustrative Example: Bending moments in columns due to eccentric vertical

In the detail shown in Fig.4.3(e) below, the beam is simply supported on
the angle, which is bolted to the face of the column. It is not clear exactly where
the end reaction acts, so in accordance with Clause 4.3.4, it is taken as either the
middle of the support (75 mm from column face) or 100 mm, whichever is
greater, i.e. 100 mm.
To calculate the bending moment caused in the column by the eccentricity
in the diagram above,

1. Calculate the total load on the beam as 30 kN/m x 6 m = 180 kN

(ignoring the fact that 6 m is the distance between centres of columns and

the actual beam length is a bit less. We will also assume the 30 kN/m
includes the self-weight of the beam.
2. End reaction to support beam =180/2 = 90 kN.
3. Assume that this reaction force acts at a distance of 100 mm from the face
of the column, i.e. in this case 200 mm from the centreline of the column,
since the column section is approximately 200 mm deep. Thus it will
exert a moment of 90 kN x 0.2 m = 18 kNm.
This moment must be balanced by moments in the columns above
and below the
4. Next the distribution factors between connecting members are calculated.
Because a joint must be in equilibrium, the sum of the bending moments in
the members connected at any joint must be zero (taking clockwise as
positive and anticlockwise as negative). The 18 kNm moment exerted by
the beam must be balanced by moments at the ends of the
columns above and below, where they connect to the beam. For example
if the storey height below is 4 m and that above is 3 m, the moments
are distributed in the ratio 3
below to 4 above, i.e. the column below takes 18×3/7 = 7.7 kNm at the
connection and the column below takes 18×4/7 = 10.3 kNm. These
moments in the columns are assumed to decrease linearly to zero at the
floor levels above and below, as shown in Fig.4.3(d). If
there were another beam to the left of the column, the moment in it at this
connection would also have to be taken into account.

Most analysis of steel structures is done using elastic theory, although in
practice some local yielding and plastic behavior is acceptable. Methods of
analysis vary from approximate analysis using simplifying assumptions,
through to highly sophisticated finite element analysis. Manual methods of
analysis have now been largely replaced by faster and more accurate computer

methods, but the designer should be aware of the existence of the older manual
methods such as moment amplification and moment distribution.

First and second order elastic analysis

Clauses 4.4.1 and 4.4.2 of AS 4100 deal with first order analysis, in
which deflections in members are not taken into account in calculating moments
and forces, and second order analysis in which they are. Second order effects are
illustrated in the examples below.

Illustrative Example 1:
A 6 m high signpost of 150x150x6 mm SHS section in Grade 350 steel
carries a 50kN vertical load at an eccentricity of 0.5m, as shown in Fig.4.4(a).
First order analysis predicts a uniform bending moment of 25 kNm, which
corresponds to a maximum stress of 247 MPa, well below yield. The deflection at
the top due to this uniform bending moment would be approximately 200mm.
However this deflection increases the moment arm of the eccentric load
about the column base from 500 to 700 mm, thereby increasing the deflection,
which in turn increases the moment arm, and so on. Second order or non-
linear elastic analysis predicts 297 mm deflection at the top and a maximum
bending moment of 39.84 kNm at the bottom, as shown below. This is a 59%
increase, enough to cause the column to yield. Although this is an extreme
example it illustrates the importance of second order effects.

Illustrative Example 2:
The moment amplification effect is further illustrated in Fig.4.4(b) below.
A two storey frame carries a wind load from the left and gravity loads on the two
beams. First order analysis predicts a lateral movement of 215 mm at the top floor
due to the wind load, but this movement creates an eccentricity which increases
the bending moments in the columns and increases the lateral movement.
AS4100 provides equations for calculating moment amplification effects
in braced members in Clause and for sway members in Clause
These equations tend to give conservative predictions for normal cases and do not

accurately predict extreme cases such as those shown above. They are intended
for use as a part of a manual analysis process, but most analysis is now done
using software packages which make the moment amplification equations


3.1. Data Atau Identitas Buku :

Judul buku : Steel Framed Structures Stability and Strength
Pengarang : R.Narayanan, dkk
Penerbit : Taylor and Francis
Tahun Terbit : 2005
Nomor ISBN : 0-203-97408-5
Jumlah Halaman Buku : 7 6 halaman

Data atau Identitas Buku Pembandingnya Yaitu :

Judul buku : Steel Structures Design Manual To AS 4100
Pengarang : Iyad Hassan Al-jamel & Brian kirke
Tahun Terbit, Cetakannya : 2004 cetakan ke - 1
Jumlah Halaman Buku : 229 halaman


a) Keterkaitan antar Bab

Buku yang diringkas mempunyai persamaan seperti menjelaskan tentang

teori struktur baja, batang tarikdan material baja. Sehingga Kedua buku ini
memiliki keterkaitan antar bab pada setiap materi yang menjeaskan tentang
struktur baja, batang tarik maupun material baja.

b) Kemutakhiran isi buku

Kemutakhiran dalam kedua buku ini adalahterdapatnya teori yang

memperjelas dan juga adanya penjelasan manfaat yang terdapat di baja,dan juga
adanya banyak grafik atau pun tabel-tebel yang ada pada kedua buku.


a) Keterkaitan antar Bab

Kedua buku menggunakan bahasa inggris jadi membuat pembaca malas

untuk membaca dan memahami materi tentang struktur baja. Penjelasan pada
kedua buku singkat sehingga tidak maksimal untuk para pembaca memahaminya.

b) Kemutakhiran buku

Kemuktakhran yang terkait dalam isi buku dalam sisi negatifnya adalah
kurangnya gambar-gambar yang guna memperjelas ataupun bagian dari
lapangan,sehingga para pembacasusah mengerti dalam segi perancangan yang

Impilikasi terhadap :

a) Teori

Dalam kedua buku terdapat banyak penjelasan teori mengenai baja tarik,
batang tarik dan sebagainya sehingga kita bisa mengetahui analisis untuk
perencanaan struktur.

b) Program Pembangunan di Indonesia

Dalam program pembangunan menggunakan struktur rangka baja bisa di

dapatkan di indonesia, seperti jembatan sungai ular, jembatan kereta api di
Tembung dan pengerjaan struktur rangka atap pada rumah bangunan permanen
yang ada di Medan.

c) Analisis Mahasiswa

Berdasarkan pada hasil kritikan dua buku diatas bahwa implikasi terhadap
Analisis Mahasiswa berisi analisa-analisa terhadap pemahaman dalam
mengerjakan suatu struktur bangunan. Sehingga Mahasiswa dapat pembelajaran
mengenai proses analisis struktur rangka baja dengan metode sederhana maupun

6.1. Kesimpulan
Dalam kedua buku ini kita dapat mengetahui tentang adanya struktur
rangka baja ,dengan setiap metode-metode perhitungannya,bahwa dalam struktur
baja ini kita harus mengetahui dulu yang paling utamanya ialah baja tersebut.
Struktur rangka baja merupakan salah satu pembelajaran yang sangat perlu
diaplikasikan dalam dunia perancangan konstruksi bangunan. Bukan hanya beton,
kayu, dan bambu. Material baja juga diperlukan dalam hal melaksanakan
pekerjaan konstruksi sipil.

6.2 Saran
Saran dalam kedua buku adalah sebaiknya buku dapat di perjelas dalam
penjelasan mengenai teori ataupun masalah perhitungan yang ada pada buku, dan
juga dapat memperjelas gambar-gambar yang ada pada buku, masalah perhitungan
sangat membingungkan bagi sang pembaca akibat teori yang kurang jelas pada
buku, sehingga para pembaca sulit untuk memahami isi buku.


R.Narayanan,SteelFramed Structures stability and strength,penerbit

taylor&francis e-Library, 2005

Iyad Hassan Al-Jamel& Brian Kirke,Steel Structures Design Manual To AS

4100,june 2004