GENERAL

Portal Frame Design Tips Seminar Proceedings

Author:
Affiliation:

Clark Hyland
Steel Construction New Zealand Inc.

Editor:
Affiliation:
Date:
Ref.:

Kevin Cowie
Steel Construction New Zealand Inc.
27th April 2010
GEN7001

Key Words
Portal frame, design tips

Introduction
In October, 2009, Steel Construction New Zealand Inc., (SCNZ) ran technical seminars throughout New Zealand.
One of the topics covered was Portal Frame Design Tips, presented by the Manager of SCNZ, Clark Hyland.
These proceedings outline the main messages delivered on this topic at the seminar series and were edited by
Kevin Cowie.
This paper summarises material predominantly from two Australian Steel Institute (Woolcock et al, 1999; Hogan
et al, 1997), and one Steel Construction Institute (Salter, 2004) publications, contextualised for New Zealand
practice in accordance with the New Zealand Steel Structures Standard NZS 3404 (SNZ, 2007). The use of
these referenced documents in particular are gratefully acknowledged.

Portal Frame Types

Pitched Roof Portal (Fabricated from UBs)

A single-span symmetrical pitched roof portal frame (Figure 1) will typically have:

A span between 15 m and 50 m

An eaves height between 5 and 10 m

A roof pitch between 3o and 5 is commonly adopted

A frame spacing between 8 m and 12 m (the greater spacings being associated with the longer span
portal frames)

Haunches in the rafters at the eaves and apex.

Figure 1: Single-span Symmetrical Portal Frame (Salter, 2004)

Disclaimer: SCNZ and the author(s) of this document make no warrantee, guarantee or representation in connection with this
document and shall not be held liable or responsible for any loss or damage resulting from the use of this document

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Propped Portal Frame

Where the span of a portal frame is large (greater than say 30 m), and there is no need to provide a clear span,
a propped portal frame (Figure 2) can reduce the rafter size and also the horizontal thrust at the base, giving
economies in both steelwork and foundation costs.
This type of frame is sometimes referred to as a "single span propped portal", but it is a two-span portal frame
in terms of structural behaviour.

Figure 2: Propped Portal Frame (Salter, 2004)

Tapered Section or Cellular Beam Portal Frame

In recent years portal frames have been constructed using tapered welded sections and cellular beams. Cellular
beam frames commonly have curved rafters (Figure 3), which are easily achieved using cellular beams or
welded sections. Where splices are required in the rafter (for transport), they should be carefully detailed, to
preserve the architectural features for this form of construction.

Figure 3: Tapered Section or Cellular Beam Portal Frame (Salter, 2004)

Many cellular beam portal frames in the span range of 40 m to 55 m have been constructed in the United
Kingdom; greater spans are possible. Elastic design is used because the sections used cannot develop plastic
hinges at a cross-section, which is an essential criterion for elastic design with moment redistribution.

Purlins
Purlin Deflections

The following deflection limits are recommended for purlins and girts.
Under dead load alone:
Span/360
Under live load alone:
Span/180
Under serviceability wind load alone:
Span/150

Purlin Bolts

The standard bolt is an M12-4.6/S which comes with loose washers. It should be remembered that washers
under both the head and nut are essential. This is because the standard punched holes in purlins are too big for
M12 bolt heads and nuts, even though the height of the hole through lapped purlins is less because of the
lapping.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Frame Analysis
General

NZS 3404 permits a number of types of structural analysis, consisting of first and second order elastic analysis.
First order elastic analysis assumes the frame remains elastic and that its deflections are so small that secondary
effects resulting from the deflections (second order effects) are negligible. First order analysis is generally
carried out using plane frame analysis computer programs. Despite the basic assumption of first order analysis,
second order effects are not negligible. Second order effects are essentially P- effects which arise from the
sway of the frame, or P- effects which arise from the deflections of individual members from the straight
lines joining the members' ends. NZS3404 requires that the bending moments calculated by first order analysis
be modified for second order effects using moment amplification factors. First order elastic analysis of portal
frames in accordance with NZS 3404 utilises a simple procedure that does not account for P- and P- effects.
The use of moment amplification factors can be avoided by using second order elastic analysis. Second-order
elastic analysis essentially involves a number of iterations of first order elastic analysis with the deflected shape
of the previous iteration being used for the second and subsequent iterations until convergence is obtained.
Second order elastic analysis programs are now widely available, and as the moments obtained do not require
amplification and are generally less conservative than amplified first order elastic moments, second order elastic
moments is recommended ahead of first order amplified elastic analysis.
It should be noted that second order analysis should only be performed for load combinations and not for
individual load cases. Second order elastic analysis is performed on load combinations and not on individual
load cases, since the second order analyses using the individual load cases cannot be superimposed. Therefore,
it is necessary to have two separate sets of output for second order elastic analysis: the first for load cases and
load case deflections (as obtained by first order elastic analysis) and the second for member forces and
reactions for load combinations (as obtained by second order elastic analysis).

Elastic Analysis

Although the use of elastic analysis with moment redistribution of portal frames at the ultimate limit state is well
established in New Zealand, it is not widely used internationally. Furthermore, there are situations where elastic
analysis is more appropriate e.g. where:
Tapered or cellular members are used.
Instability of the frame is a controlling factor.
Deflections are critical to the design of the structure

Elastic Analysis with Moment Redistribution

Plastic hinges may form in the members within the structure as their plastic moment capacity is reached as the
structure redistributes moments. It assumes that the members behave elastically up to the full value of the
plastic moment capacity, then plastically (without strain hardening) to allow redistribution of moments around
the frame. Members required to redistribute ultimate limit state moments are required to have sufficient flexural
torsional restraint to ensure development of the plastic section capacity of the section.
This method has several advantages including that it optimises the use of a single hot rolled section in a frame
leading to fabrication simplification.

Frame Design with Haunches

For preliminary computer analysis, selection of the rafter and column sizes is from experience or by guesswork.
The computer model should have at least two nodes near each knee joint to allow for modelling of the rafter
haunches in the final design phase. Nodes at the mid-height of each column and at quarter points of the rafter
can give useful bending moment diagrams in some cases, although this is generally unnecessary when using
modern computer packages.
Haunches dont need to be included in the initial computer run as they do not have much effect on the frame
bending moments. However, significant reductions in deflection can be achieved later in the analysis.
Once the first computer analysis is run, the limit state bending moments in the column and in the rafters should
be checked against the section capacities to check the assumed sizes.
For preliminary design, reducing the column bending moment to the underside of the haunch or reducing the
section capacity to allow for coincident axial forces can be disregarded. The calculated moment at the knee
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

should be checked against the column section capacity Msx. Implicit in this check is that sufficient fly braces
can be provided to ensure that the full section capacity is achieved.
The calculated bending moments in the rafter should be similarly checked against the capacity Msx. except in
the vicinity of the knee joints where haunches will probably be provided to cater for the peak rafter moments in
these areas. Some small margin in flexural capacity should be retained in order to cater for axial forces. The
member sizes assumed should then be adjusted accordingly and the frame analysis re-run.

Haunch Properties

Once the member sizes have been established with more confidence, it is appropriate to model the haunches.
The standard AISC haunch (AISC, 1985) is formed from the same section as the rafter. It is common to model
the haunch with two or three uniform segments of equal length although reference (Hogan et al, 1997)
indicates that there is no benefit in using more than two segments.
The depth of the haunch is calculated at the mid-point of each segment and the section properties can he
calculated accordingly. Some frame analysis programmes can calculate haunch properties automatically.
Alternatively, the properties of standard UB's which are contained in standard software libraries can be used to
model the haunch segments approximately.
A comprehensive AISC publication (Hogan et al, 1997) in 1997 investigated the design of tapered portal frame
haunches fabricated from universal section members. The publication deals with detailing the cost of fabrication,
the calculation of elastic and plastic section properties, computer modelling (including the effect of varying the
number of segments), and section and member design to AS 4100, which is the source document for much of
NZS 3404. It also reviews the testing of haunches in other literature.

Modelling Base Fixity

General

Column bases are usually considered as being nominally pinned at the ultimate limit state. This simplifies the
design.
However a degree of base stiffness may be considered. Stiffness at the base can reduce the deflections and
increase the stability of the frame considerably. However, foundations that are designed to resist moments are
considerably larger than those designed for axial load and shear forces only and consequently, are much more
costly (Salter et al, 2004).

Rigid Base

Where a column is rigidly connected to a suitable foundation, the following may be assumed:

In elastic global analysis, the stiffness of the base should be taken as equal to the stiffness of the
column for all ULS calculations. However, in determining deflections under serviceability loads, the base
may be treated as rigid.
In elastic analysis with moment redistribution, the assumed base stiffness should be consistent with the
assumed base moment capacity, but should not exceed the stiffness of the column.

Nominally Pinned Base

Where a column is nominally pin-connected to a foundation that is designed assuming that the base moment is
zero, the base should be assumed to be pinned when using elastic global analysis to calculate the other
moments and forces in the frame under ULS loading.
The stiffness of the base may be assumed to be equal to the following proportion of the column stiffness:
10% when checking frame stability or determining in-plane effective lengths
20% when calculating deflections under serviceability loads

Nominally Semi-rigid Base

A nominal base stiffness of up to 20% of the stiffness of the column may be assumed in elastic global analysis,
provided that the foundation is designed for the moments and forces obtained from this analysis.

Base Spring Stiffness Modelling

In practice, allowance for base fixity is usually by the use of spring stiffness or dummy members at the column
base.

Ultimate Limit State

At the ultimate limit state:
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

A nominally rigid base can be modelled with a spring stiffness equal to 4Elcolumn/Lcolumn
A nominally pinned base can be modelled with a spring stiffness equal to 0.4EIcolumn/Lcolumn for frame stability
checks.

Serviceability Limit State:

A nominally rigid base can be modelled with full fixity.

A nominally pinned base can be modelled with a spring stiffness equal to 0.8Elcolumn/Lcolumn.

Rafters
Nominal Bending Capacity Mbx in Rafters
Simplified Procedure

NZS 3404 uses a semi-empirical equation to relate the nominal bending capacity Mbx to the elastic buckling
moment Mo and the section strength Msx, which for Universal and Welded Beams and Columns can be taken as
Zexfy. This philosophy uses a set of semi-empirical equations to relate the member strength to the plastic
moment and the elastic flexural torsional buckling moment.
Equation 5.6.1.1(1) of NZS3404 expresses the nominal member bending capacity Mbx as

Mbx

m sMsx

Msx

where m is a moment modification factor to account for the non-uniform distribution of major axis bending
moment, and s is a slenderness reduction factor which depends on Msx and the elastic buckling moment of a
simply supported beam under uniform moment Mo. The standard gives comprehensive values of m which would
be met in practice. The conservative option of taking m equal to unity is also permitted.
For category 2 and 3 members in seismic resisting frames,

1.0 . For category 1 members

oms ,

reflecting the need to maintain stability under over-strength actions.

The slenderness reduction factor is expressed in Clause 5.6.1.1(3) of the standard as

Msx
Moa

0.6

Msx
Moa

Where Moa may be taken as either (i) Mo which is the elastic buckling moment for a beam with a uniform
bending distribution and with ends fully restrained against lateral translation and twist rotation but unrestrained
against minor axis rotation; or (ii) a value determined from an accurate elastic buckling analysis.
The elastic buckling moment Mo may be determined from the accurate expression given in equation 5.6.1.1(4)
as
2

Mo

EI y

Le

GJ

EI w

Le

Where Le is the effective length, and Ely, GJ and EIw are the flexural bending rigidity, the torsional rigidity and
the warping rigidity respectively. Values of the section properties Iy , J and Iw are given in the ASI Design
Capacity Tables for Structural Steel (AISC, 1997). The use of Equation 5.6.1.1(4) requires the effective length
Le, and determination of this is made using clause 5.6.3.

Alternative Procedure

Clause 5.6.4 of NZS3404 allows the designer to use the results of an elastic buckling analysis, although in most
cases this is not practical for design offices and is really a research tool. If an elastic buckling analysis is to be
used, then the elastic buckling moment Mob , which allows for the moment gradient, restraint conditions and
height of loading, is determined either from a computer program or from solutions given in the literature
(Bradford, 1988).

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Having obtained Mob, the value of Moa to be used in clause 5.6.1 or 5.6.2 is calculated from
Moa

Mob
m

Where values of
m

are obtained either from the standard or from an elastic buckling analysis such that

Mos
Moo

The moment Mos is the elastic buckling moment corresponding to Mob for the same beam segment with the
same bending moment distribution, but with
shear centre loading,
ends fully restrained against lateral translation and twist rotation, and
ends unrestrained against minor axis rotation.
The moment Moo is the reference elastic buckling uniform bending moment Mo given by Equation 5.6.1.1(4) with
Le taken as the laterally unsupported length L.
In the event that the whole rafter is designed as a tapered member, an accurate elastic buckling analysis must
be used. This also applies to the haunched segment of a conventional rafter. The values of Mob and Mos for
tapered rafters may be found in Reference (Bradford, 1988).

Effective Length and Moment Modification Factors for Bending Capacity

General
If the simplified design procedure in Clause 5.6.1 of NZS 3404 is used, then the effective length Le of the rafter
must be determined in accordance with Clause 5.6.3. The effective length depends on the spacing and stiffness
of the purlins and fly braces, and the degree of twist and lateral rotational restraint for a chosen segment as
follows:
Whether the connection between the purlins and rafter is rigid, semi-rigid or pinned.
The flexural rigidity of the purlins, in that regard NZS 3404 classifies purlins qualitatively as flexible or stiff.
No numerical yardstick is given.
The load height in that regard NZS 3404 allows, for example, for the destabilising effect of loads applied at
or above the shear centre in a beam subjected to downward loads.
Whether the top or bottom flange is the critical flange. For a portal frame, the compression flange is the
critical flange.
The degree of lateral rotational restraint provided at the ends of a segment by adjoining segments.

Top Flange in Compression

Under gravity loads, the top flange is mostly in compression, except near the knees. Purlins provide lateral
restraint to the top flange, but full twist restraint to the rafter from the purlins cannot always be relied upon
because standard oversized slotted holes are often used in purlins. However even where the holes in the purlins
are slotted, the bolts are tightened and so the purlin to rafter connection using a standard purlin cleat and two
bolts can be regarded as a partial twist restraint connection in terms of Figure 5.4.2.2 in NZS 3404.
Fortunately, the standard permits partial twist restraint at the critical flange (in association with lateral restraint)
to be classified as full restraint of the cross-section as shown in Figure 5.4.2.1(b). Therefore, for each segment
between purlins when the top flange is in compression, both ends are fully restrained (FF) and the twist
restraint factor kt is 1.0.
Although gravity loads are applied through the purlins at the top flange, the load height factor kt of 1.4 in Table
5.6.3(2) in NZS 3404 does not apply because the load is not free to move sideways as the member buckles. In
other words, the load is applied at a point of lateral restraint and kt should be taken as 1.0.
The degree of lateral rotational restraint provided at each end of the segment by adjoining segments depends
on whether the adjoining segments are fully restrained laterally or not, as described in Clause 5.4.3.4 of NZS
3404. (A fully restrained segment in accordance with Clause 5.3.2 is essentially one with Mb not less than Ms
which means its m s value is greater than unity.) The standard permits full lateral rotational end restraint or
none. No intermediate option is provided. While segments between purlins under downward loading are short
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

and are likely to be fully restrained laterally, full restraint in accordance with Clause 5.3.2 cannot be guaranteed.
It follows that lateral rotational restraint should strictly speaking be disregarded. There is, however, a high
degree of lateral rotational restraining which would allow kr to be taken safely as 0.85.
In summary, the effective length Le is given by ktklkrL as L e

1.0 x 1.0 x 0.85S p

0.85S p

The spacing between purlins is short in comparison with the length of the rafter (Figure 4), so the moment
modification factor m should usually be taken as 1.0.

Bottom Flange in Compression

With Fly Bracing Under Uplift
Under uplift, most of the bottom flange of a portal frame rafter is in compression. In such cases, the rafter is
attached to the purlins at the tension flange level, and the compression flange of the rafter is unrestrained. In
order to achieve increased member capacity, it is customary to restrain the bottom flange of the rafter laterally
by providing fly bracing using small angle section members joining the bottom flange to the purlins.
With the bottom flange in compression, NZS 3404 classifies a fly brace restraint as a full or partial crosssectional restraint depending on whether the purlins are flexible or stiff. No numerical criterion is given for
assessing the flexibility or stiffness of purlins. Therefore if partial cross-sectional restraint is assumed
conservatively at each end of the segment (PP), the twist restraint factor kt will be greater than 1.0 in
accordance with Table 5.6.3(1) of NZS 3404. However, unless fly braces are closely spaced or the rafter has an
unusually high flange to web thickness ratio, kt will normally be close to 1.0. Considering that the partial
restraint assumption is probably conservative, a kt value of 1.0 is recommended for simplicity.
It may appear that there should be a useful reduction in effective length because the wind loads act at the more
favourable tension flange level. However, the benefit of this is not significant as most of the bending moment
within a segment is due to end moments, and the segment should not be likened to a simply supported beam
under uniformly distributed load applied at the tension flange level. Moreover, the reduction in effective lengths
of a simply supported beam under such loads is limited in some cases and NZS 3404 offers no concession for
bottom flange loading. For this reason, kl should be taken as 1.0.
For a segment between fly braces and with the bottom flange in compression, the lateral rotational restraint
provided at the ends of the segment by adjoining segments should strictly speaking be disregarded because it is
unlikely that the adjoining segments are fully restrained laterally in accordance with Clause 5.4.3.4 of NZS 3404.
There is, however, a degree of lateral rotational restraint which would allow kr to be taken as 0.85.
In summary, the effective length Le for segments between fly braces for uplift conditions is given by ktkekrL as
L e 1.0 x 1.0 x 0.85Sf 0.85Sf .
The moment modification factor m for segments between fly braces will usually be greater than 1.0. For
segments which have a reversal of moment, part of the segment will have its compression flange restrained by
purlins but this benefit should be ignored.

Without Fly Bracing under Uplift

Although some fly bracing is recommended, it is interesting to consider the rafter behaviour under uplift where
there is no fly bracing at all. In this case, the full portal span should be taken as the effective length, and m
should be based on the bending moment distribution across the rafter span. Even though the validity of this
approach for a kinked member is doubtful, the large effective length should equate to such a low capacity that
some fly bracing will be necessary.
Designers often feel that the lateral restraint offered by purlins to the tension flange under uplift conditions
should also increase the lateral buckling capacity. However, theoretical and experimental studies (Dux et al,
1986; Wong-Chung, 1987) of the bracing of beams have confirmed that translational restraint alone acting at
the level of the tension flange, such as that provided by purlins, is virtually ineffective. These studies show that
if the lateral restraint is combined with some twist restraint, the buckling capacity is increased. It is possible to
design the purlin-rafter connection for some rotational capacity by providing two or four friction bolts to the
cleat, or by using wider cleat plates with more bolts. There may be architectural advantages in avoiding fly
bracing, such as when a ceiling is required above the bottom flange level.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Investigations have been carried out (Wong-Chung, 1987) into the effectiveness of standard purlin connections
in providing rotational restraint to the rafters. The results revealed in part that the requirement for rotational
stiffness is a function of the initial geometric imperfections in the rafter. That is, for very crooked rafters, greater
stiffness in the brace is required. The theoretical and experimental studies have so far indicated that ordinary or
standard purlin connections are effective to some degree, provided that the bolts are properly tightened.

With Fly Bracing under Downward Load

The effect of the bottom flange near the columns being in compression due to gravity loads or other loading
should be considered even though most of the bottom flange of the rafter is in tension. A fly brace is
recommended near each knee and near the ridge to restrain the inside corners of the frame at kinks. A stiffener
between column flanges as indicated in Figure 4 effectively extends the bottom flange of the haunch to the
outside column flange which is restrained by girts. This effectively provides some restraint to the inside of the
knee. However, a fly brace near the knee is still recommended. With fly braces at least at the knees and the
ridge, the effective length will be 0.85 times the spacing between fly braces.
An alternative approach is to consider the rafter segment between the column and point of contraflexure if
accurately known, or nearest purlin beyond the inflection point. The inflection point is considered to be
unrestrained in determining the effective length. This approach is described in an example by Clifton,
Goodfellow and Carson (1989)
The value of the moment modification factor m for the segment should be determined using one of the three
methods in NZS 3404, but using a specifically calculated m in Clause 5.6.1.I(a)(iii) is likely to be most
appropriate if there is no intermediate fly brace between the knee and ridge. It is recommended that any
haunch should be ignored in determining the design bending capacity Mbx of the segment, but the applied
bending moments should be reduced by factoring the moment at any haunch section by the ratio of the elastic
section modulus of the unhaunched section to the corresponding elastic modulus of the haunched section.
Alternatively if each end of the haunch happens to be fly braced as in the design example, the haunch may be
treated as a tapered segment in accordance with clause 5.6.1.1.1 of NZS 3404.

Figure 4: Effective Length Factors for Bending in Rafters and Columns (Woolcock et al, 1999)

Major Axis Compression Capacity Nc

In NZS3404, the nominal member capacity Nc for buckling in plane about the major axis is required in the
combined actions rules for determining the in-plane member capacity in Clause 8.4.2.2. It is obtained from
Clause 6.3.3 as

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Nc

c k f A n fy

where An is the net rafter cross-sectional area, which is generally the gross area for portal frame members (see
Clause 6.2.2 of NZS3404). The member slenderness reduction factor c is given in tabular form in the standard
for values of the modified slenderness ratio

nx

L e / rx

k f fy 250 where Le is the effective length equal to

keL based on the actual rafter length L from the centre of the column to the apex.
Two effective lengths need to be used under Clause 8.4.2.2 of NZS 3404. For combined actions, the effective
length factor ke should be taken as 1.0. The rafter also needs to be checked under axial load alone using
effective lengths determined from the frame elastic buckling load factor. c . This factor can be obtained either
by using the approximate method in clause 4.7.2.1 or 4.7.2.2 of NZS 3404, or by using commercially available
computer packages. The check under axial load alone is unlikely to be critical for portal frames without cranes
because they are principally flexural frames with low axial loads in all members.
The form factor kf which accounts for local plate buckling are given in the steel producers section handbooks.

Minor Axis Compression Capacity Ncy

The nominal member capacity Ncy for buckling about the y axis is required in the combined action rules of NZS
3404 for determining the out-of-plane capacity in Clause 8.4.4.1. It is obtained by taking the effective length Le

as the distance between purlins, since the purlins are restrained longitudinally by roof sheeting acting as a rigid
diaphragm spanning between the roof bracing nodes. The theoretical effective length of an axially loaded
member (rafter or column) with discrete lateral but not twist-rotational restraints attached to one of the flanges
may be greater than the distance between the restraints. Unfortunately, there is no simple method of
determining the effective length of such a member. In the case of a rafter restrained by purlins, some degree of
twist-rotational restraint would also exist. The combined full lateral and partial twist-rotational restraint provided
by the purlins to the outside flange should be effective in enforcing the rafter to buckle in flexure between the
purlins. The capacity Ncy is obtained using the minor axis modified slenderness ratio in clause 6.3.3 of NZS 3404.
ny

L e ry

k f fy 250

Combined Actions for Rafters

The effect of axial tensile or compressive forces in rafters combined with bending should be included in the
design as described in clause 8.4.4.1 and 8.4.4.2 of NZS 3404.

Flexural Torsional Buckling Restraints

Fly Braces

As discussed previously, fly braces are diagonal members bracing the bottom flange of rafters back to purlins, or
the inside flange of columns back to girts to stabilise the inside flange when in compression. Fly braces can take
many forms, with the most common being a single angle each side of the bottom flange, as shown in Figure 5.
The design bracing force is determined from Clause 5.4.3 of NZS 3404, which gives criteria for the strength of
braces to prevent lateral displacement of the braced compression flange. For each intermediate brace, the
design force is 2.5% of the maximum compression force in the braced flange of the segments on each side of
the brace. In this case, a segment is the length of the member between fly braces. Sharing between multiple
intermediate braces is not permitted but each bracing force is related to the local maximum flange compression
force rather than to the maximum flange compression force in the whole rafter or column. It should be noted
that NZS 3404 permits restraints to be grouped when they are more closely spaced than is required for full
lateral support, the actual arrangement of restraints being equivalent to a set of restraints which will ensure full
lateral support.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

Figure 5: Double Fly-brace (Woolcock et al, 1999)

Under these conditions, the capacity of single bolted fly brace angles will be close to their concentric capacity
based on minor axis (y-y) buckling. For this case, even the smallest angle, a 25x25x3, has the capacity in
compression to sustain the force calculated. However, it is not really practical to use a bolt smaller than an M12,
and a 25x25 angle is too small for an M12 bolt whose washer diameter is 24 mm. The smallest angle which can
accommodate an M12 bolt is a 40x40x3 angle. It seems unnecessary to use fly braces on both sides of the
rafter when a small angle on one side is quite adequate.
In some cases, there may be practical or aesthetic objections to fly braces because of the presence of a ceiling
above the bottom flange of the rafter. This could occur in a supermarket for example. In this case, a wider
purlin cleat and four high strength bolts, and a web stiffener on one or both sides to prevent cross-sectional
distortion, as shown in Figure 6 could be used to brace the bottom flange. The bolt shear forces in the friction
type joint can be calculated for the combined case of purlin uplift and moment due to the lateral bracing force at
the bottom flange level. The disadvantage of this approach lies in the non-standard purlin cleats and nonstandard holing of purlins.

Figure 6: Alternative Torsional Restraint (Woolcock et al, 1999)

Purlins as Braces

Where the top flange is in compression, it was assumed previously in the rafter design section that the purlins
provided adequate restraint to the top flange. NZS 3404 permits restraints to be grouped when they are more
closely spaced than is required for full lateral support, the actual arrangement of restraints being equivalent to a
set of restraints which will ensure full lateral support.
In summary, where the top flange is in compression, it is recommended that the restraint spacing necessary to
provide the required member capacity be determined. If the required restraint spacing is much greater than the
purlin spacing, then some of the purlins can be ignored as restraints, and two or three purlins near the notional
brace point could be considered as sharing the required bracing force at that point.

Central Props
General

In large span industrial buildings, a central prop is often used to reduce the rafter span and to limit rafter and
external column sizes. An efficient central prop is a square hollow section (SHS) as central props are typically
long and can buckle about both axes. Other sections such as UBs, UCs, WBs or WCs can also be used
effectively, particularly if the lateral stiffness requirements of the portal frame are a problem. The columns can
be detailed with flexible or rigid connections to the rafter. In both cases, there is a need to determine the
effective lengths both in-plane and out-of-place in order to calculate the compression capacity under axial load
alone. In the case of a rigid top connection, there will be in-plane bending moments generated in the column,
and these moments will need to be amplified if a first order elastic analysis has been carried out. If a flexible
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

10

connection between the column and rafter is detailed, it would be prudent to check the central column for both
pinned and rigid top connections as there will be some in-plane moments generated with most practical flexible
connections.

Figure 7: Effective Length of Central Prop (Woolcock et al, 1999)

There can be some uncertainty about how to calculate the effective length for determining the nominal capacity
Ncx in the plane of the portal frame (see Figure 7). The uncertainty arises partly because the top of the rafter is
attached to the apex of a portal frame which can sway sideways. This is dealt with in the following sections.

Effective Lengths of Props for Axial Compression

Top Connection Pinned

If the top of the central column is connected to the portal frame by a flexible connection such as a cleat
perpendicular to the plane of the frame, it would be reasonable to regard this connection as pinned. In this
case, the central column does not interact in flexure with the frame, but the frame must have a certain
minimum stiffness to effectively brace the top of the columns as shown in Figure 7. For a pinned base column
the minimum spring stiffness to ensure that its effective length Le is equal to and not greater than the length L
of the column is 2EI c / L3 .
In practical frames, the side-sway stiffness of the rigid frame with its relatively stiff side columns and rafter is
usually quite sufficient to brace the top of a slender central column. Designers can readily determine the
sideways stiffness by analysing a special load case with a single horizontal load at the apex of the frame.

Top Connection Rigid

If the top connection is rigid, then there should logically be some reduction in effective length of the central
column. However, in accordance with NZS 3404, it is not possible to determine directly the effective length of
individual members in non-rectangular frames. The standard in Clause 4.9 requires a rational buckling analysis
of the whole frame to determine the frame elastic bulking load factor c. The only practical way of determining
c is by means of a frame analysis. These programs also convert the c value for each load combination into
effective lengths for each member by use of Equation 4.5.

Combined Actions with First Order Elastic Analysis

If the top connection is rigid, the frame elastic buckling load factor c for each load combination is used in
Clause 4.4.3.3.2(b) to determine the amplification factor c which is applied to any bending moments from a
first order elastic analysis. The capacity of the central column is then checked under Clause 8.4.2.2 of NZS 3404

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

11

using an effective length factor ke of 1.0 for combined actions, and also an effective length factor calculated
from c for axial load alone.
If the top and bottom connections are assumed to be pinned, there will be no moments from the frame analysis
but a nominal eccentricity in each direction is recommended. The effective length factor ke will then be 1.0 for
both combined actions and for axial load alone if the minimum spring stiffness in Section 4.6.2.1 is provided.

Combined Actions with Second Order Elastic Analysis

Ironically, if a designer has access to programs such to determine c for amplifying first order moments, then it
is likely that the designer also has access to the second order elastic analysis option of these programs. In this
case, a designer would ideally use the second order elastic analysis as this obviates the need to amplify the
moments. The capacity of the central column is then checked as described in the previous section.

Frame Deflections
General

Portal frames are generally designed on the basis of strength first, and are checked for the serviceability
(deflection) limit state according to some arbitrary criteria. Deflection limits can govern the design of portal
frames, and it is therefore important that any deflection limits be realistic.
The selection of deflection criteria for industrial steel frames is a subjective matter. In general, standards are
not prepared to give specific recommendations, probably because deflection limits have not been adequately
researched. The Australian steel code AS4100 states that the responsibility for selecting deflection limits rests
with the designer, but still gives some recommendations. For a metal clad building without gantry cranes and
without internal partitions against external walls, the standard suggests a limit on the horizontal deflection of the
eave as column height/150 under serviceability wind loads. This limit reduces to column height/240 when the
building has masonry walls. The limits suggested in Appendix B of AS 4100 are based on the work in (Woolcock
et al, 1986).

Problems of Excessive Deflection

The potential problems of excessive deflections in industrial buildings include:

Damage to cladding and fixings thereby affecting the hold down capacity of fixings and water
tightness.
Ponding of water on low pitched roofs and possible leakage because of ponding or insufficient pitch.
Visually objectionable sag in rafters or suspended ceilings whose ceiling hangers are difficult to adjust
for sag, e.g. heavy acoustic ceilings.
Visually objectionable sag in the ridgeline because of the deflection of the apexes of internal rafters
relative to the end wall apexes. The end wall rafters do not sag because they are supported by end
wall columns.
The results of the survey were reported in (Woolcock et al, 1986). It is interesting to note that in many answers,
there was no clear consensus of opinion among engineers. What is regarded as acceptable to one engineer is
not necessarily acceptable to another. The results of the survey were rationalised, and deflection limits were
proposed. These are summarised in Figure 8, Figure 9 and Figure 10. It is emphasised that these limits should
be used for guidance rather than as mandatory limits. Further research is required to establish deflection limits
with more confidence.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

12

Figure 8 Notation for Deflection Limits (Woolcock et al, 1999)

Figure 9: Recommended Rafter Deflection Limits

(Woolcock et al, 1999)

Figure 10: Recommended Lateral Deflection Limits

(Woolcock et al, 1999)

Notes:
The wind load deflection limits apply to serviceability wind loads.
L is the rafter span measured between column centrelines.
Precamber or pre-set may be used to ensure that the deflected position of the rafter under dead load corresponds to
the undeflected design profile, or is within the above limits of the undeflected design proflle. Even so, pre-set may be
advisable for internal rafters to avoid visual sag in the ridge line.
For low roof pitches, the check for ponding is really a check to ensure that the slope of the roof sheeting is nowhere
less than the minimum slope reconnnended by the manufacturer. The slope of the rafter in its deflected state can be
determined from the joint rotations output from a plane frame analysis program. The slope of the roofing should also
be checked mid-way between rafters near the eaves where purlins are more closely spaced and where the fascia
purlin may be significantly stiffer than the other purlins.
Where ceilings are present, more stringent limits will probably be necessary.

Differential Deflections

Generally, where a rafter and post frame has been used, it will be braced and will therefore be much stiffer than
the adjacent portal frames. In practice this is also true with a portal frame gable wall because it will be stiffened
by the cladding. Differential deflection between the gable frame and penultimate frame can therefore be
relatively large, and may be of particular concern if there are cranes, masonry construction, or sensitive cladding
attached to the frame.
Ways of reducing differential deflections include:
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

13

Bracing in the roof between the gable frame and the adjacent frame will reduce the deflection of the
adjacent portal frame to some extent, but this is normally not quantifiable without a 3-D analysis of the
whole structure.
A penultimate frame can be provided of greater stiffness than the other frames to reduce the differential
deflection due to eaves spread and wind loading. This is not usually a sensible option in terms of fabrication
efficiency.
The portal frames should be pre-set carefully to ensure that all dead load deflections result in frames that
line up with the gable frame under dead load only, thus reducing to some extent the differential deflection
due to eaves spread.

Column Bases
Base Plates

Mild steel Grade 4.6 bolts are preferred because they can be adjusted by bending on site particularly if there is a
sleeve or pocket around the holding down bolt for this purpose. Mild steel bolts can also be tack welded into a
cage, whereas Grade 8.8 bolts should not be tack welded because welding can have an adverse effect on steel
properties in the vicinity of the weld. Regardless of the steel grade, it is recommended that holding down bolts
be hot dip galvanised.

Holding Down Bolt Design Criteria

There are many considerations in the design of holding down bolts (Trahair et al, 1998), the most important
being as follows:
The bolts themselves should have sufficient capacity in combined tension and shear.
The grouting or bedding under the base plate should have sufficient capacity in compression to cater
for applied compression and bending moment at the base of the column.
The concrete or the grout filling the space around the bolts and sleeves should have sufficient strength
in bearing to transmit the shear force in the bolt.
If the bolts do not have a suitable head or other anchor at the head to prevent pullout or bearing
failure under the head, the bolts must be sufficiently long or must be suitably cogged or hooked to
satisfy the anchorage requirements for plain deformed bars (as appropriate) in the concrete standard
NZS 3101 (SNZ, 2006).
If the bolts have a suitable head or anchor, the embedment must be sufficient to prevent the bolts
pulling out a cone of concrete (cone failure).
If there is insufficient edge distance, the bolts must be lapped or anchored with reinforcing bars in
accordance with the concrete standard.
Account should be taken of fabrication and erection tolerances when detailing and installing holding
down bolts.
The likelihood of corrosion must be considered carefully. Hot dip galvanizing is recommended.
A minimum of four bolts rather than two bolts is favoured by riggers to assist in supporting columns
during erection.

Base Moments for Foundation Design

It should be noted that, as far as the base moment (and associated forces) for foundation design is concerned,
the following applies:
Where partial base fixity is used to reduce the moments for which frame members have to be designed
(compared to those obtained assuming pinned bases) the base moments should be taken into account in
designing the foundations. This applies for both elastic analysis and elastic analysis with redistribution of
moments.
Where a nominal 10% base stiffness is used only in assessing effective lengths (or elastic critical load
factors) or in determining whether an unbraced frame is 'sway-sensitive' or 'non-sway' it is not necessary to
take account of the base fixity moment in foundation design.

Roof & Wall Bracing

General

Roof and wall bracing often consist of panels of double diagonals which are so slender as to have negligible
capacity in compression. Such members include pre-tensioned rods, slender tubes and angles. In the design of
double diagonal tension bracing, one of each pair of diagonals is assumed to act in tension as shown in Figure
11, depending on the direction of wind loading, and the other diagonal is usually ignored. In addition to tension

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

14

forces, roof bracing diagonals have to carry their own weight whether by cable action in the case of rods, or by
beam action in the case of tubes and angles.

Figure 11: Roof and Wall Bracing (Woolcock et al, 1999)

As common as tension bracing is, there is not a widely accepted method of design which accounts for tension
and self weight. This problem was investigated in References (Kitipornchai et al, 1985; Woolcock et al, 1985).

Temporary Bracing

Portal frames can collapse during construction if adequate care is not taken to use permanent or temporary
bracing to withstand wind gusts. The procedure to be used varies from building to building depending on the
type and location of the permanent roof and wall bracing bays and whether the end wall frame is a braced
frame or a portal frame.

Roof Plane Bracing

Roof plane bracing is placed in the plane of the roof. The primary functions of the roof plane bracing are:
To transmit horizontal wind forces from the gable posts to the vertical bracing in the walls.
To provide stability during erection.
To provide a stiff anchorage for the purlins that are used to restrain the rafters.

Rafter Bracing Forces

In addition to the longitudinal wind forces, the bracing system could also be considered as resisting
accumulated, coincidental purlin or fly brace forces. When the top flange is in compression, the purlins act as
braces whereas fly braces restrain the bottom flange when it is in compression. However, it is unclear whether
the bracing forces should be accumulated. Purlins and fly braces together could be considered as providing a
rotational restraint system in accordance with Clause 5.4.3.2 of NZS 3404. In this case, it would not be
necessary to treat the compression flanges of rafters as parallel restrained members in accordance with Clause
5.4.3.3, and therefore it would not be necessary to accumulate the forces. On the other hand, purlins and fly
braces could be considered as providing restraint against lateral deflection of the compression flange (Clause
5.4.3.1) and in this case the bracing forces would be accumulated.
It is interesting to compare roof trusses as far as accumulation of bracing forces is concerned. The bottom
compression chord of a series of large span roof trusses under net uplift is usually braced back to the end
bracing bays by a system of struts or ties. In this case, the bracing forces should be accumulated and then
combined with forces due to longitudinal wind. When the top chord is in compression, it is usually regarded as
being braced by purlins back to the end bracing bays. Logically, the top chord bracing forces should also be
accumulated, but as the compression in the top chord is generally due to gravity loads, there are no other
longitudinal forces in combination and so the loads on the end bracing bays are not likely to be critical.
It could be similarly argued that the top or bottom flange bracing forces of UB or WB rafters, whichever flange is
in compression, should also be accumulated. However, even if the lateral restraint argument (as opposed to the
rotational restraint argument) is accepted, the accumulated bracing forces are usually a small part of the total
longitudinal force for portal frame buildings. It is therefore considered reasonable for UB or WB rafters, to ignore
accumulated bracing actions in the design of the roof and wall bracing bays.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

15

A summary of the advantages and disadvantages of various options for bracing layouts is as follows as shown in
Figure 12.

Option I: Two End Bays Braced

This is the simplest and most direct option. Intermediate eaves and ridge struts are sometimes used as shown
dashed. However, purlins are usually sufficient to brace internal rafters so that no intermediate struts are
required.
Longitudinal wind loads, as a combination of pressure on the windward wall, suction on the leeward wall and
friction, could be shared between braced bays if purlins have the capacity to transfer some compression load
from one end to the other. However, it is recommended that the bracing at each end be designed to resist loads
from external pressure and internal suction on the adjacent end wall (plus half of the frictional drag forces if
applicable). This keeps the purlin design simple as purlins can then be designed without considering combined
actions. Diagonals are crossed which means that CHS sections, which are efficient as long ties under self weight,
cannot easily be used. This option also excludes the use of the top flange as a bracing plane with angle
diagonals crossed back to back unless higher purlin cleats are used. End bay bracing can have detailing
difficulties at the end wall rafter.

Option II: Double Diagonal Bracing Over Two Bays at Each End
Diagonals intersect at rafters and therefore tubes can be used as diagonals without difficulty if they are
not crossed.
The number of diagonals is the same as for Option I but more struts are required.

Option III: Second Bay from Each End Braced

This option can overcome any detailing difficulties associated with end bay bracing but extra struts are
required to transfer the end wall wind loads to the braced bays unless the purlins can act as struts.

Option IV: One Bay Braced

Struts in the unbraced bays are required to transfer end wall wind loads to the braced bay which is
expensive unless the purlins can act as struts.

Option V: Single Diagonal Tension Bracing at Each End

Unstable during erection.
The windward braced bay takes all of the longitudinal wind loads.
Purlins are usually sufficient to brace internal rafters as in Option I. Leeward end wall forces are
transmitted to the active braced bay at the windward end by purlins in tension.
Tubes can be used for diagonals without difficulty as they are not crossed.
Single diagonal rods with turnbuckles should not be used as there is nothing to tension against.
Temporary diagonals may be necessary to create a double diagonal bracing system for erection
purposes in which case there is little advantage in a single diagonal system.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

16

Figure 12: Bracing Layout Options (Woolcock et al, 1999)

Bracing using Circular Hollow Sections

In the United Kingdom circular hollow section bracing members are generally used in the roof and are designed
to resist both tension and compression. Many arrangements are possible, depending on the spacing of the
frames and the positions of the gable posts (Salter et al, 2004).

Tension Rod Bracing

Rods cater for the lower end of the range of tensile forces, and are very common in light industrial buildings.
Rods differ from tubes and angles in that they must be pre-tensioned to reduce their self weight sag. However,
there are certain aspects of rod pre-tensioning which are not widely understood. The aspects which need to be
considered are as follows:
Steel Advisor GEN7001
Steel Construction New Zealand Inc. 2010

17

The minimum level of pretension force needed to reduce the sag sufficiently to avoid undue axial slack
in the rod.
The level of pretension used in practice. The effect of pretension on the tensile capacity.
The effect of pretension on the end connections, and on the adjacent struts in the roof bracing system,
when wind loads are applied.
Pre-tension actions should be 10% to 15% of the yield capacity. While these levels of pretension may be
adequate, it is not practical to measure or control the pre-stress level in practice. To answer the questions above
properly, it is necessary to examine the behaviour of pre-tensioned rods in some detail.
Long rods behave like cables whose self-weight is carried by tension alone; the tension being inversely
proportional to the sag. For small sags in roof bracing situations, the tensile stress fat versus sag yc relationship
has been shown [13] to be independent of the rod diameter and is given by

fat

9.62 x 10

L2
MPa
yc

in which L is the length of the rod and both y, and L are in mm. This relationship is presented graphically in
Figure 13. Using this equation, it can be demonstrated that as a rod is tensioned, very little force is required to
reduce the sag until the sag gets to about span/100. The rod then begins to stiffen suddenly and behave as a
straight tension member. This is shown graphically in Figure 14. Therefore, the maximum sag of a rod to avoid
undue axial slack should be about span/100. Surprisingly, a stress of only 20 MPa is required to reduce the sag
of a 20 metre cable to the L/100 deflection. However, typical stress levels in practice could be much higher.
In experiments at the University of Queensland (Woolcock et al, 1985), six different laboratory technicians were
asked to tighten rods ranging in diameter from 12 mm to 24 mm with spans up to 13 metres long. They were
told to tighten the nuts as if they were working on site. Once tightened at one end, the force in the rod was
measured with a calibrated proving ring connected to the other end. The experiments revealed that the average
level of pretension force was well in excess of the value of 10% to 15% suggested in (Gorenc et al, 1996). In
fact, it was found that 16 mm diameter rods were tensioned close to their design capacity, while 20 mm rods
were tensioned to between 40% and 55% of their design tensile capacity. Because of these unexpectedly high
pretension forces, excessive sag is not a problem, even for a 20 metre span.
The presence of pre-tension does not affect the ultimate tensile capacity of the rod itself. However, there are a
few other factors that need to be considered in the design of roof bracing rods. In some cases of overtensioning, the active tension diagonal may yield under the serviceability wind load, although yielding will relieve
the pretension in the system to some extent. Fortunately, the fracture capacity of the threaded section typically
exceeds the yield capacity of the rod itself. This means that the main body of the rod will generally yield before
failure of the turnbuckle section. Because of the pretension, the rod connections should be designed so that
their ultimate or fracture capacity is equal to or greater than the ultimate or fracture capacity of the rods. This is
particularly important because oversized rods are often used. For example, a 20 mm diameter rod may be used
because of its robustness where only a 16 mm diameter rod is required. This philosophy for the end connection
design of rods is covered in Clause 9.1.4(b)(iii) of NZS3404.

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

18

Figure 13: Effect of Axial Stress on Cable and Rod Deflections (Woolcock et al, 1999)

Pre-tensioning could also result in overloading of the struts in the roof bracing system, especially if rods larger
than that required are used. A check should therefore be made in the design of the struts to cater for forces in
the diagonals due to combined pretension and wind load as shown in the design example.

Figure 14: Effective Axial Stiffness of Cables and Rods (Woolcock et al, 1999)

Tubes and Angles in Tension

In contrast to rods, tubes and angles are not easily pre-tensioned and must be sized as beams to limit self
weight sag. The uncertainties for designers, as far as tube and angle section members are concerned, are firstly
the effect of self weight bending on tensile capacity, and secondly deflection limits. Some engineers combine
self weight bending actions with axial tensile actions, while many engineers intuitively ignore the bending
actions.
It can be shown theoretically (Woolcock, 1985) that self weight bending has a marginal effect on the ultimate
fracture capacity of a tube or angle. This is because the sag and self-weight bending moments reduce as the

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

19

tension increases. It can therefore be concluded that self-weight bending actions need not be considered in
combination with axial tension.
As proposed for rods, a maximum sag of span/100 is suggested to avoid undue slack. However, it is advisable to
limit deflections to span/150 to avoid lack of fit without propping during erection, and for aesthetic reasons.
Note that even with a span/150 deflection, there is occasionally concern expressed during construction as the
sag can be quite evident if one sights along the member. The sag is not generally obvious from floor level.
Of course, the designer has the option of suspending the diagonals from the purlins, but very flexible diagonals
(other than rods) can be difficult to erect before the purlins are in place because of lack of fit. If the purlins are
erected first, the stability of the portal frames without bracing may be inadequate and lifting the diagonals into
place will be more difficult because of obstruction from the purlins. Furthermore, the extra labour necessary to
drill and suspend may cost more than the material saved. The effect of purlin uplift loads on the capacity of
diagonals should also be taken into account. With all these factors considered, suspending very flexible
diagonals from purlins is not recommended.

Side-wall Bracing
General

The primary functions of vertical bracing in the side walls of buildings are:
To transmit the horizontal loads, acting on the end of the building, to the ground.
To provide a rigid framework to which side rails may be attached so that they can in turn provide stability to
the columns.
To

provide temporary stability during erection, the bracing system will usually take the form of:
Circular hollow sections in a V pattern.
Tension only cross-braced rods.
Circular hollow sections in a K pattern.
Crossed flats (within a cavity wall).
Crossed hot rolled angles.

The bracing may be located at:

One or both ends of the building, depending on the length of the structure.
At the centre of the building (but this is rarely done due to the need to begin erection from one braced bay
at, or close to, the end of the building).
In each portion between expansion joints (where these occur).
Where the side wall bracing is not in the same bay as the plan bracing in the roof, an eaves strut is required to
transmit the forces from the plan bracing into the wall bracing.

Side-wall Bracing Using Circular Hollow Sections

Circular hollow sections are very efficient in compression, which eliminates the need for cross bracing. Where
the height to eaves is approximately equal to the spacing of the frames, a single bracing member at each
location is economic. Where the eaves height is large in relation to the frame spacing, a K brace may be used.
An eaves strut may be required in the end bays, depending on the configuration of the plan bracing. In all
cases, it is good practice to provide an eaves tie along the length of the building.

References
AISC, Design Capacity Tables for Structural Sections-Volume 1: Open Sections, 2nd edition & Addendum No.1,
Australian Institute of Steel Construction, Sydney, 1997
AISC, Standardized Structural Connections, 3rd Edition, Australian Institute of Steel Construction, Sydney, 1985
Bradford, M.A., Lateral Stability of Tapered Beam-Columns with Elastic Restraints. The Structural Engineer,
66(22), 376-384, 1988
Clifton, G. C., Goodfellow, B., Carson, W., Notes Prepared for a Seminar on Economical Single Storey Design and
Construction, HERA Report R4-52, New Zealand Heavy Engineering Research Association, Manukau City, 1989

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

20

Dux, P.F., Kitipornchai, S., Buckling of Braced Beams, Steel Construction, Journal of the Australian Institute of
Steel Construction, AISC, 20(1), 1-20, Sydney, 1986
Gorenc, B.E., Tinyou, R., Syam, A.A., Steel Designers Handbook. NSW University Press, Sydney, 1996
Hogan, T.J., Syam, A.A., Design of Tapered Haunched Universal Section Members in Portal Frame Rafters, Steel
Construction, Journal of the Australian Institute of Steel Construction, AISC, 31(3), 1-28, Sydney, 1997
Kitipornchai, S., Woolcock, S.T., Design of Diagonal Roof Bracing Rods and Tubes. Journal of Structural
Engineering, ASCE, 115(5), 1068-1094, 1985
Salter, P.R., Malik, A.S., King, C.M. Design of Single-span Steel Portal Frames to BS 5950-1:2000, Steel
Construction Institute, Silwood Park, 2004
SNZ, Concrete Structures Standard, NZS 3101:2006, Standards New Zealand, Wellington, 2006
SNZ, Steel Structures Standard (Incorporating Amendments 1 and 2), NZS 3404:1997, Standards New Zealand,
Wellington, 2007
Trahair, N.S., Bradford, M.A., The Behaviour and Design of Steel Structures to AS4100, 3rd Edition, E&FN Spon,
London, 1998
Wong-Chung, A.D., Theoretical and Experimental Studies of the Geometric and Material Nonlinear Behaviour of
Partially Braced and Unbraced beams, PhD Thesis, The University of Queensland, 1987
Woolcock, S.T., Kitipornchai, S., Bradford, M.A., Design of Portal Frame Buildings, AISC, Sydney, 1999
Woolcock, S.T., Kitipornchai, S., Deflection Limits for Portal Frames. Steel Construction, Journal of the
Australian Institute of Steel Construction, AISC, 20(3), 2-10, 1986
Woolcock, S.T., Kitipornchai, S., Tension Members and Self Weight. Steel Construction, Journal of the Australian
Institute of Steel Construction, AISC, 1(1), 2-16, 1985

Steel Advisor GEN7001

Steel Construction New Zealand Inc. 2010

21