SCI ADVISORY DESK

AD 238: Bearing capacity of connected parts with

slotted holes
The SCI recently carried out an experimental investigation with the aim of devising design
guidance on the bearing capacity of plates with slotted holes. The conclusion was that
plates with slotted holes had a lower bearing capacity than standard clearance holes.
Therefore, it is recommended that when ordinary bolts are used and the plate is loaded
perpendicular to the direction of the slot the bearing capacity Pbs is reduced by a reduction
factor kbs taken as:
-

0.7

for short slots (d + 6 mm)

0.5

for all other lengths up to 3.5d

Bearing capacity covers several different possible failure modes, including possible shearout or tear-out at the end of the plate as well as actual bearing stresses, but the criterion
that is most affected by using slotted holes is the deformation of the connection under
service loads. The reduction factors kbs have been determined so that the deformation of
the plate at working load is limited to 1.5 mm, which is in line with the current BS 5950-1
recommendations for clearance holes.
For slotted holes that are designed to act as part of an expansion joint, it is not sufficient to
apply these reduction factors. For an expansion joint to slide freely under load, the bearing
stresses and deformations permitted by these recommendations (or the existing BS 5950-1
checks for ordinary holes) will be too high to allow efficient operation. A reduced bearing
stress based on the Hertz contact pressure would be more appropriate for such joints.
On the other hand, connections with preloaded HSFG bolts are designed not to slip under
service loading, so deformation under service loading is not considered in their bearing
capacity. Provided that sufficient end distance is available (generally about 3d), much higher
bearing capacities are recommended in BS 5950-1 in its check on capacity after slipping into
bearing. In such cases the proposed bearing capacity reduction factor for slotted holes need
not be applied (though the current reduction factors for HSFG bolts in slotted holes still
apply).
Based on the tests carried out, the report recommends that the bearing capacity
based on:
Pbs = kbs d t pbs
where

d
t

is the bolt diameter

is the plate thickness

pbs

is the bearing strength of the plate

kbs

is a reduction factor

Subject to the limits investigated in the tests:

-

1.4D end distance < 3.0d

edge distance 1.25D
pitch 2.5d

Pbs

is

SCI ADVISORY DESK

and proposes that the reduction factors be increased to 1.0 when end distance is three times
the bolt diameter or greater.
In making proposals to the BSI committee for inclusion in an amendment to BS 5950 : Part 1,
the recommendations have been reviewed and refined. The scope has been extended to
include oversize holes, and the kbs factors have been applied to the current BS 5950-1
capacity, including the effect of end distance. This provides a sliding scale and avoids the
need for a sudden increase at an end distance of 3d. In addition, it provides a more robust
solution for the small number of cases where the experimentally determined value of kbs
would otherwise be less than 0.5. It also provides a simple safe solution to a subsidiary
problem not investigated in the tests, the possible effect of slotted holes on shear-out.
In summary, the recommendation is to determine Pbs using:
Pbs = kbs d t pbs

but

Pbs 0.5 kbs e t pbs

where kbs has the value:

1.0

for normal holes

0.7

for short slotted holes or oversize holes

0.5

for long slotted holes up to 3.5d.

SCI ADVISORY DESK

AD 239: Direct Tension Indicators (load indicating

washers)
Readers who look at the small print may have read note 12 under Table 1 from AD 226 and
wondered what Direct Tension Indicators are. Direct Tension Indicators are specified in
BS 7644: 1993: Part 1 for compressive washers and Part 2 for nut face and bolt face
washers.
The note following Clause 1 of Part 1 reads Compressible washer-type direct tension
indicators are also know as load indicating washers.
The requirements of Part 1 generally follow the advice given over many years by Cooper and
Turner on the use of their Coronet load indicating washers, average indicator gaps being the
same but with the possibility of tightening by bolt rotation added. However, this method of
tightening is not generally the preferred approach and might require a bolt face washer.
Being little used, these washers are not produced as a stock item.
Nut face washers and bolt face washers to Part 2 are very similar, the difference being that
the hole in the bolt face washer is slightly bigger. Nut face washers are marked with an
indented letter M on one face and bolt face washers with the letters BM.
Annex A of Part 1 describes the use of these indicators and refers to Figures 2 & 3 for the
correct assembly for each method of tightening. These figures are reproduced below.

Indicator
washer

Indicator
washer

Before
tightening

Before
tightening
Gap

Throughhardened
washer

a) Under bolt head fitting

Nut face
washer

Gap

a) Under nut fitting

Indicator
washer

Indicator
washer

Before
tightening

Nut face
washer

Throughhardened
washer

Before
tightening
Bolt face
Gap washer

Gap

b) Under nut fitting

b) Under bolt head fitting

Fig. 2:
Bolt tightening by rotation of the nut
(normal method of assembly

Fig. 3: Bolt tightening by rotation of the bolt

(alternative method of assembly)

It should be noted that average indicator gaps are smaller for the cases shown in Figures 2(b)
& 3(b)

SCI ADVISORY DESK

Note: See also AD226

SCI ADVISORY DESK

AD 240: Imposed Load Reduction When Designing

Multi-storey Columns
A recent question concerned the application of load reduction when designing multi-storey
columns. The query came to light because the engineer was using the multi-storey design
example at the end of Chapter 2, Sheet 4, in the Steel Designers Manual (SDM) 5th Edition
and had been told (incorrectly) by his checking engineer that this did not properly apply the
load reduction given in Table 2 of BS 6399: Part 1: 1996 Specified loading for building: code
of practice for dead and imposed loads.
What BS 6399 requires can be illustrated using the simple example in Figure 1.

Roof
Fifth floor
Fourth floor
Third floor

Column No. 1

Second floor
First floor

Column No. 2

Ground Level

Fig. 1: Column in a typical multi-storey building

Consider the column between second and third floor levels (marked No 1 in Figure 1). This
column carries four floors. According to Table 2, a 30% reduction can be applied to the
imposed loadings on all four floors (including the roof) in the design. However, consider also
the column at ground level (marked No. 2 in the Figure); it carries six floors and therefore a
40% reduction can be applied to the imposed loads on all six floors carried by this column.
As the design of the various columns throughout a multi-storey building proceeds
(downwards, from the roof), the percentage reduction to the imposed loading allowed by
Table 2 increases. This means that the design load of a particular column is not simply the
load (after reduction) of the column above plus the reduced loading on the floor above; the
unreduced load must first be totalled and then the appropriate reduction should be applied.
This is the correct way to apply Table 2, and will give a greater reduction than summing
reduced loads.
This might appear to require a lot of design work, especially for hand calculations, but the
multi-storey design example (Sheet 4) in the SDM shows that it can be carried out easily by a
tabular approach. The sheet shows, for example:

SCI ADVISORY DESK

1.

At level 6, for column 1: the column imposed loads (unreduced) for the six floors
are given in the third column of the Table as 40.4, 189.2, 152, 152, 152, 152,
totalling 837.6 kN. Applying the 40% reduction for 6 floors, gives a value of
502.56 kN, which agrees with the value in the seventh column of the Table.

2.

At Level G, for column 2: the column imposed loads (unreduced) for the twelve
floors are given in the fourth column of the table as 72.1, 334.9, 269.2 x 10,
totally 3099 kN. Applying the 50% reduction for over ten floors, gives a value of
1549.5 kN, which agrees with the value in the eighth column of the Table.

Needless to say, the reductions should never be applied to any dead loading.

SCI ADVISORY DESK

AD 241: Transverse Reinforcement in Composite

Beams
Transverse reinforcement in the form of mesh or additional loose bars is required in
composite beam design to transfer the longitudinal shear force from the shear connectors
into the effective width of the slab. Traditionally, light mesh reinforcement has been used
throughout the slab, as a deemed to satisfy approach, but design to BS 5950: Part 3 and
Eurocode 4 can lead to the requirement for additional reinforcement, particularly within the
high shear regions of primary beams.
The question often posed to SCI by designers is: what is the effect of light mesh
reinforcement on the longitudinal shear resistance, and therefore on the bending resistance of
a beam, when this reinforcement is less than that required according to code rules for
transverse reinforcement? There are two important aspects to this question:
1.

Often, composite beams are not designed to their full bending resistance, Mc,
because serviceability limits control the design. If the utilisation factor on bending
resistance at the ultimate limit state is given by the factor M/Mc, it may be shown
that the longitudinal shear force transferred between the beam and the slab
reduces conservatively in proportion to M/Mc.

2.

Failure by concrete splitting along the line of the shear connectors is a ductile
mode of failure, provided the amount of reinforcement is above a certain minimum.
In principle, the effective shear resistance of the shear connectors may be
down-rated according to the amount of transverse reinforcement provided, rather
than having to provide sufficient transverse reinforcement to equate to the design
resistance of the shear connectors.
The minimum amount of transverse reinforcement that is provided should be
sufficient to achieve the minimum degree of shear connection for the beam span
multiplied by the utilisation factor, M/Mc (see below).

These observations are based on full-scale composite beam tests at the University of
Cambridge in which only light mesh reinforcement was provided in the primary beam tests.
In these 9 m span tests, the secondary beam connections also acted as effective transverse
ties to control splitting. In design terms, the minimum longitudinal shear force that should be
transferred by the concrete splitting mode of failure is conservatively given by:
Fq,r Fq (M/Mc) Kmin
where Fq,r is the longitudinal shear force per unit length due to the splitting mode of failure,
as influenced by the transverse reinforcement that is provided (and any anchorage effect of
the steel decking crossing the beams). Fq is the longitudinal shear force per unit length due
to the shear connectors, as required for full shear connection, and M/Mc is the utilisation of
bending resistance at the ultimate limit state. Kmin is the minimum degree of shear
connection required for a symmetrical steel section for a beam span, L, in metres (where Kmin
(L-6)/10 but Kmin 0.4).
If this equation is not satisfied, the ratio M/Mc should be reduced so that the magnitude of
Fq,r that is determined from the transverse reinforcement provided equates to the right hand
side of the equation.

SCI ADVISORY DESK

For example, in the case of 10 m span beam, Kmin = 0.4, and for a typical design
M/Mc = 0.8. Therefore, Fq,r should exceed 0.32 Fq. If Fq,r is inadequate, additional
reinforcement is required.
Assuming that full shear connection is based on the compressive resistance of the slab and
the beam is subject to 2 point loads, it may be shown that the minimum percentage of
reinforcement in the shear plane should be at least 0.2% of the cross-sectional area of the
slab (in this case the concrete topping). It follows that A193 mesh would be sufficient as
transverse reinforcement. Therefore, in practical design cases, additional bar reinforcement
would not be necessary.
This simplified approach gives designers a quick and safe way of determining whether a
beam designed with a small amount of transverse reinforcement is adequate. If not, a more
sophisticated analysis should be performed, based on the partial shear connection principles.

SCI ADVISORY DESK

AD 242: Varying Depth Beams: BS 5950 : Pt 1

Appendix B3
Recently a question arose concerning the use of Appendix B3 in BS 5950 : Part 1 in checking
a beam of varying depth for lateral torsional buckling. On inspection of Appendix B3 this
proves more complex than appears at first sight. Consider the following beam subjected to
any bending moment arrangement with a maximum value of Ma.
In order to check for lateral torsional buckling, Appendix B3 clearly states that pb (bending
strength) must be determined from the cross section properties at the point where maximum
moment occurs in the beam. This may be at the maximum or minimum depth, or some
intermediate point, and N, u, v, x and ry must be taken from this point in order to determine
pb. Once this has been done, the appropriate value of Sx (plastic modulus about the major
axis) must be determined for equation 4.3.7.3, Mb=Sx pb.
In fact, Sx
M
stress
Zx

must be taken from the critical cross section in the beam, i.e. where the bending

has its maximum value. This may not be at the point of maximum moment. To

find the critical cross section, the bending stress must be calculated at several locations to
find the point of maximum value. When the critical cross section has been located, Sx is
determined for this location using the gross section properties at that point. This is what is
meant by the sentence in Appendix B3 which states This value of pb applies throughout the
length between adjacent restraints.

'0rm03s'

It is intended to make this issue clearer in the forthcoming amendment to BS 5950 Part 1.
B

)IC
W fl W

q;bTp

Fig. 1: Beam length L between points of lateral restraint.

_
m=1.0 (see Note 1,Table 18, BS 5950 : Pt 1) M = m Ma=Ma

)c

_______

SCI ADVISORY DESK

AD 243: Splices within unrestrained lengths

Ideally a splice in a compression member, or a laterally unrestrained beam, should be
positioned close to a restraint, or, if the member is continuous, at a point of inflexion of the
buckled shape. If the splice cannot be located at such a point, it may be possible to achieve
the desired effect by adding an additional restraint instead. Nevertheless it is sometimes
unavoidable to locate a splice within the unrestrained length of a member, and the purpose
here is to explain how to design such splices. The procedures given are based on those
expected to be included in the forthcoming amendment to BS 5950: Part 1.
The first thing to note is that these design procedures only affect the splice. The member
itself is not affected, provided that it has been correctly designed in the first place and that
the recommended procedure is used for the splice.
Design requirements
The basic requirements are that the splice must be both:

stiff enough to avoid reducing the buckling resistance of the member below that required,
and also

strong enough to transmit the forces and moments in the member.

Providing a splice that is more flexible than the member itself is considered to be bad
practice, because it would be liable to be both uneconomic and potentially unsafe. In any
case it would be a complex matter to calculate the buckling resistance of a member in which
the splice did not have at least the same stiffness as the member.
It is recommended, if possible, to use a splice with cover plates sized to provide adequate
stiffness, by making the moment of inertia of the splice material at least as great as that of
the member, considering both axes. This is normally relatively easy to achieve by providing
at least as much area in the cover plates as in the relevant element of the member crosssection.
In a splice connection with end plates it is generally more difficult to achieve the same
stiffness as the member. An accurate elastic analysis of the connection should be used to
verify that it is at least as stiff as the member. It is likely that relatively thick end plates will
be needed. Extended end plates may be required if there is a significant moment. Even
where a splice connection is entirely in compression, it is advisable to maintain full continuity
of stiffness through the connection to safeguard the robustness of the structure.
Whichever type of splice is adopted, preloaded HSFG bolts should be used. In the case of
cover-plate splices, they should be designed to prevent slip under ultimate limit state loading,
by using a coefficient of 0.9 rather than 1.1 in the formula for slip resistance. An alternative
would be to use fitted bolts.
The capacity of the splice connection needs to be checked for all the forces and moments in
the member at that point. Besides the internal forces and moments derived from equilibrium
with the applied loads, allowance must be made for the following second order effects, as
outlined below:
a) moments due to strut action;
b) moments due to lateral-torsional buckling, and
c) moments due to amplification.

SCI ADVISORY DESK

For an explanation of these effects see AD 244.
Design procedures:
Note: In addition to the procedures that follow, any localised forces and moments resulting
from the detailed design of the splice, such as those due to the transfer of shear across the
connection, or due to eccentricity between the centroidal axes of the sections either side of
the splice, should also be included.
1) Splice for members subjected to compression only:
Design forces and moments for the splice:
Fc,splice
Mx,splice
My,splice

=
=
=

Fc
[Ms(x-x)]
[Ms(y-y)]

applied force
(a) strut action major axis
(a) strut action minor axis

[Include only one of the terms in square brackets - the one with the most onerous effect.]
where:
Ms
Ms

is
=

a strut action moment, obtained from:

Mmax sin (180 Lz/LE)

[Ms should be calculated for major (Ms(x-x)) and minor Ms(y-y)) axes, using LEx and LEy with
the respective values for Lz];
Lz
LE
Mmax
Mmax
fc
pc
py
S

is
is
is
=
is
is
is
is

the distance along the member to the nearest point of inflexion;

the effective length (i.e. the distance between points of inflexion);
the maximum internal moment due to strut action, given by:
(py/pc 1) fc S
the compressive stress due to axial force;
the compressive strength;
the design strength;
the plastic modulus of the strut for bending about the axis of buckling.

2) Splice for members subjected to major axis moments only.

Design moments for the splice:
Mx,splice
My,splice
where:
Mys

=
=

Mx
Mys

applied major axis moment at the splice

(b) lateral-torsional buckling (LTB)

is

Mys

the additional minor axis moment at a distance Lz along the member

from the nearest point of inflexion, which is obtained from:
My,max sin (180 Lz/LE)

The additional minor-axis moment (an internal second-order moment equivalent to the strut
action moment in a compression member) in a member subject to major axis moment should
be taken as having a maximum value My,max midway between points of inflexion of the
buckled shape (the points between which the effective length LE is measured), given by:
My,max

(py/pb 1) (Mcy/Mcx) mLT Mx,max

where:
Mcx and Mcy

are the major and minor axis moment capacities of the section, for zero
shear;

SCI ADVISORY DESK

Mx,max

is

mLT
pb

is
is

the maximum major axis moment in the length LLT between lateral
restraints;
the equivalent uniform moment factor for lateral-torsional buckling;
the bending strength for resistance to lateral-torsional buckling.

3) Splice for members subjected to combined axial compression and bending:

Design forces and moments for the splice:
Fc,splice
Mx,splice
My,splice

=
=
=

Fc
applied
Mx + [Ms(x-x)] + (Madd,xs)
applied + (a) strut + (c) amplification
My + [Ms(y-y)] + (Mys) + Madd,ys
applied + (a) strut + (b) LTB + (c)
amplification

[Include only the more onerous of the strut action terms in square brackets and the more
onerous of the two terms in round brackets.]
where:
Ms(x-x) and Ms(y-y) are strut action moments, see (1);
My
is
the minor axis moment at the splice;
Mys
is
the additional minor-axis moment due to lateral-torsional buckling,
see (2);
Madd,xs and Madd,ys
are additional moments due to amplification of Mx and My by Fc each at
its appropriate distance Lz along the member from the relevant point of
inflexion, obtained from:
Madd,xs
=
Madd,x,max sin (180 Lz/LEx)
Madd,ys
=
Madd,y,max sin (180 Lz/LEy).
LEx
is
the effective length for flexural buckling about the major axis;
LEy
is
the effective length for flexural buckling about the minor axis;
mx M x
Madd,x,max
=
( pEx / fc 1)
my My
Madd,y,max
=
(pEy / fc 1)
mx

is

my

is

pEx

pEy

the equivalent uniform moment factor for flexural buckling about the
major axis;
the equivalent uniform moment factor for flexural buckling about the
minor axis;
2E
x2
2E
y 2

Note: See also AD244

SCI ADVISORY DESK

AD 244: Second order moments

The aim in this item is to give a simple explanation of the three types of additional secondorder moment encountered in AD 243, namely:
moments due to strut action;
moments due to lateral-torsional buckling;
moments due to amplification.
Moments due to strut action:
It is well known that compression members need to be designed using a compressive
strength pc, which is less than the design strength py of the steel. The value of pc
depends on the design strength py and the slenderness . The formula from which the
tables giving pc are derived are based on the fact that the actual stress in a slender
compression member is higher than the apparent stress fc, and at failure the actual stress at
mid-length reaches the design strength py . The strut action moment accounts for the
difference between the compression strength pc and the design strength py .
The strut action moment arises from the eccentricity of the axial force relative to the
deformed shape of the member as it buckles. This process is initiated by the naturally
occurring imperfections in a real member. The moment varies along the length of a member.
It is zero at pinned ends. Its effect on an un-spliced member is covered by using pc, but in a
spliced member it needs to be accounted for explicitly at the splice, unless this is at a pinned
end or a point of contraflexure.
Moments due to lateral-torsional buckling:
By analogy with a compression member, the effects on a beam subject to lateral-torsional
buckling resulting in a bending strength pb less than py can be simply represented by an
additional moment similar to a strut action moment. The analogy is convenient, but only
approximate. This type of additional moment arises from a moment applied about the major
axis of the section, but needs to be considered as acting about its minor axis. Like a strut
action moment, it varies along the length of the member and is zero at points of
contraflexure.
Moments due to amplification:
In a similar way to strut action, the axial force in a member acts at an eccentricity from the
deformed shape caused by an applied moment, producing an extra moment. The difference
between the applied moment and the final amplified moment is an additional moment due to
flexure, which needs to be accounted for in designing a splice within the length of the
member. Unlike one due to lateral-torsional buckling, an additional moment due to flexure
acts in the same plane as the applied moment causing it. As in the other cases, it varies
along the length of the member and is zero at points of contraflexure.
Note:
In line with the treatment of member buckling adopted in BS 5950 Part1, it is logical to
assume that applying a major axis moment to an axially loaded member leads to only one

SCI ADVISORY DESK

additional moment, either Mys about the minor axis due to lateral-torsional buckling, or else
Madd,xs about the major axis due to amplification, but not both (see (3) in AD 243).
Note: See also AD243

SCI ADVISORY DESK

AD 245: Allowance for Bolt Holes

The effect of bolt holes on moment capacity and shear capacity is not as clear in BS 5950:
Part 1: 1990 as is the effect on tension capacity or compression resistance, although the
basic rules on net and gross areas are intended to apply in all cases. The question has also
arisen of the effect of bolt holes at plastic hinge locations. This answer is based on
clarifications expected to be included in the forthcoming amendment to the code.
Tension: A bar in tension will fail either at its yield strength Ys on its gross cross-sectional
area Ag or in rupture at its ultimate tensile strength Us on its net area An at bolt holes. It
should also not suffer permanent deformation under service loading. The resulting design
criteria are:

yielding:

Pt

Ag Ys / m1

rupture:

Pt

An Us / m2

deformation:

Pt

f Pt,ser

f An Ys / ser

and, allowing for the values of the various factors, this leads to the code rule:
Pt

Ae py

=
=
=

Ke An
(Us / py) / 1.2
Ys

in which:
Ae
Ke
py

but
but
but

Ae
Ke
py

Ag
1.2
Us / 1.2

Accordingly, no allowance need be made for bolt holes in a tension member unless
An < Ag / Ke .
Compression: Resistance to buckling of a compression member is based on the gross area
Ag . The capacity of a cross-section can also be based on Ag because reaching the yield
strength in compression on the net area cannot lead to necking, as in tension, and strain
hardening takes over.
Bending moment: The tension flange of a beam is similar to a bar in tension, whereas the
compression flange is similar to a compression member. Thus, bolt holes in the tension
flange may need to be allowed for. However, the code does not currently describe how this
should be done. It is recommended that, as in Eurocode 3, the same rules as for a tension
member should be applied to the tension flange itself. Holes in the tension zone of the web
need not be allowed for unless there are also holes in the tension flange. If there are bolt
holes in the web and the flange at the same point in the length of the member, the rules for a
tension member should be applied to the complete tension zone comprising the tension flange
plus the tension zone of the web.
Shear: Nothing is currently said in the code about the effect of bolt holes on shear, but
current good practice is to follow the procedure given in the Green book, which covers this
issue along with the block shear effect. The forthcoming amendment to the code is
expected to cover these two items separately. The resistance of a web to shear rupture is
about 70% of its resistance to rupture in tension, whereas its resistance to yielding in shear
is about 60% of its resistance to yielding in tension. Accordingly, it is recommended that
no allowance be made for the effect of bolt holes on shear capacity unless 70% of Av,net is
less than 60% of Av / Ke , that is, unless Av,net < 0.85 Av / Ke .

SCI ADVISORY DESK

Tension flanges:
From the above it can be seen that no allowance need be made for bolt holes in a tension
flange provided that An Ag / Ke . If this is not satisfied, either the bolt holes can be
staggered or a reduced moment capacity can be calculated by taking an effective area for the
tension flange of Ke An . For a simple but conservative check, the moment capacity can be
reduced by multiplying it by the ratio Ke An / Ag .
Plastic hinges:
In the case of plastic analysis, in the region of a plastic hinge location it is necessary for the
tension flange to be able to stretch plastically without any other damage.
Punching of bolt holes hardens the steel locally around the holes, making it liable to crack in
such circumstances. Accordingly, at plastic hinge locations, and within a length each side
equal to the depth of the member:
full size punching of holes is not acceptable.
For a plastic hinge to rotate, it is necessary for the flange to yield uniformly over a similar
length, rather than just deforming locally at bolt holes. Accordingly at plastic hinge locations:
drilled holes are acceptable - but only if, for the tension flange: An Ag / Ke
This is not currently stated in the code (although it is implied in Eurocode 3) but it is
expected to be included in the forthcoming amendment.

SCI ADVISORY DESK

AD 246: Standards Organisations

BSI (British Standards Institution)
BSI is an independent body incorporated by Royal Charter. Representative BSI committees,
each with an independent chairman and secretariat, are responsible for issuing and
maintaining British Standards.
BSI is the UK member body of CEN and ISO. Appropriate BSI committees "mirror" the
relevant CEN and ISO committees and nominate the UK delegates to them. BSI shares in
providing the chairmen and technical secretariats of CEN and ISO committees. All CEN
Standards and appropriate ISO Standards are also issued as BSI Standards.
The contents of all BSI Standards, including those produced by CEN or ISO, are copyright.
The documents of all BSI committees are marked "private circulation" and not to be used for
any purpose other than the work of the committee. However, for briefing purposes,
committee members may circulate these documents within the organization they represent.
Contact information
BSI Standards, British Standards House, 389 Chiswick High Road, London W4 4AL.
Tel: 0181 996 9000
Fax: 0181 996 7400
E-mail: info@bsi.org.uk
Web-site:www.bsi.org.uk
CEN (European Committee for Standardization)
CEN is an international association, established under Belgian law by an agreement between
15 national standards bodies. Its aim is to implement standardization on a Europe-wide basis.
It now has 19 members, the national standards bodies of Austria, Belgium, Czech Republic,
Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg,
Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom.
CEN Standards exist in three official versions (English, French and German). Versions in any
other language made by translation under the responsibility of a CEN member into its own
language and notified to the Central Secretariat has the same status as the official versions.
CEN Standards are issued in member countries by the national standards body - BSI in the
UK.
Delegates to CEN meetings are experts chosen to represent their national standards
organization, which is the committee member. National delegations must always speak as
one, but positions must be flexible for the work to result in European standards.
The members of CEN are also members of ISO. In recent years the so-called "Vienna
agreement" has helped avoid duplication of work between CEN and ISO. Many draft
standards are now presented for parallel approval by both organizations.
Contact information
CEN Central Secretariat, Rue de Stassart 36, B-1050 Brussels, Belgium
Tel: 00 32 2 519 6811

SCI ADVISORY DESK

Fax: 00 32 2 519 68 19
E-mail: infodesk@cenorm.be
Web-site: www.cenorm.be
ISO (International Standards Organisation)
ISO and IEC (International Electrotechnical Commission) together form a system for
international standardization as a whole. They develop and maintain international standards
through technical committees. ISO working languages are English, French and Russian. All
ISO Standards are published in English and French (and sometimes in multi-lingual editions
including Russian and other languages). These versions are equivalent, and each is regarded
as an original-language version.
The character of ISO Standards varies. Some are intended for direct adoption as (or in place
of) national standards, whilst others are intended as a basis for drafting national standards.
Some ISO Standards become British Standards through being adopted as CEN Standards.
Others are implemented directly as British Standards, but some are not adopted by BSI for
various reasons. This can be due to their character, or because the relevant BSI committee
does not consider their content to be appropriate, or regards them as irrelevant to the UK.
Alternatively there may be an established BSI or CEN Standard, or work in progress in CEN,
that is regarded as more relevant.
In recent years the so-called "Vienna agreement" has helped avoid duplication of work, and
many drafts are now presented for parallel approval by both CEN and ISO.
Contact information
ISO Central Secretariat, 1 Rue de Varamb, Case postale 56, CH-1211 Genve 20,
Switzerland.
Tel: 00 41 22 749 01 11
Fax: 00 41 22 733 34 30
E-mail: central@isocs.iso.ch
Web-site: www.iso.ch

SCI ADVISORY DESK

AD 247: Use of Composite Construction in an

Aggressive Environment
Composite construction, using hot-rolled steel beams and composite steel deck slabs, is
widely used for medium and high rise buildings in the UK. In internal mild exposure
conditions the G275 galvanising layer on the decking ensures the desired design life can be
achieved. Leaving the upper surfaces of the beam top flanges unpainted, so that the shear
studs can be through-deck welded after laying out the decking, does not present a
corrosion risk in these conditions.
In more aggressive (moderate or severe) environments, protection of the steel decking and
steel beams requires further consideration. The following recommendations identify ways in
which the steel components may be detailed/specified in order to overcome problems of
increased exposure to the underside of the slabs. In some situations, such as car parks, it
will also be necessary to take measures to prevent corrosion of the steel components due to
water percolating through cracks in the slab. This may be achieved by controlling cracking,
or by applying protection to the top surface of the slab.

Decking
G275 galvanising (275 g of zinc coating per square metre) is the standard for decking.
Although thicker coatings (up to G600) can be applied, decking with this degree of
galvanising is difficult to obtain and likely to require a large minimum order. A far better
solution than trying to obtain non-standard galvanising may be to specify coated steel or
additional paint protection. Such an additional layer of protection has the advantage that it
can be regularly inspected and remedial work undertaken. Clearly, manufacturers/specialist
advice should be sought before specifying a particular product.

Beams
Through-deck welding of the shear studs is beneficial because it enables continuous sheets
of decking to be laid on the steel beams prior to fixing the studs. It also enhances the way in
which the decking behaves as transverse reinforcement of the slab. However, in a
potentially severe corrosive environment the need to keep the upper surfaces of the beams
free from paint (to avoid contamination of the stud welds) is not acceptable. This leaves the
designer who wishes to use composite beams and slabs with two options:

Use shear connectors that are attached to the beams without the need for welding. A
number of connectors that use shot-fired pins are available.

Weld the studs to the beams in the fabrication shop, prior to applying the corrosion
protection.

An implication of adopting the latter approach is that the decking can no longer be laid in
sheets that are continuous over two spans. The sheets must be laid in single spans and
butted up to the studs, or have pre-punched holes to allow the decking to be dropped over
the studs. Both these approaches have disadvantages:

Single span decking is less structurally efficient than when it can be made continuous.

Trying to align holes and studs is easier said than done, and the resistance of the shear
studs may be reduced.

SCI ADVISORY DESK

The presence of paint on the shear studs should not compromise their performance, which
does not rely on surface bond.

SCI ADVISORY DESK

AD 248: Use of High Strength Friction Grip

Connections
Recently we have received several questions concerning where and when to use HSFG
connections. Many designers are familiar with the use of M20 and M24 8.8 bolts in
clearance holes, but do not regularly specify HSFG connections. An ordinary bolt assembly,
comprising, for example an M20 bolt in an M22 clearance hole, transmits applied forces by
shear in the bolt and bearing between the shank of the bolt and the connected ply. This
means that a small amount of slip takes place as the connection is loaded and the clearances
are taken up. Some small deformation as the bolt and plies reach equilibrium in bearing will
also occur.
HSFG connections transfer force quite differently. Although HSFG connections still use bolts
in clearance holes, load is transferred by frictional resistance between the contact surfaces of
the plies. The frictional resistance is achieved by preloading the bolt in considerable tension,
which clamps the plies together. The preload in the bolt is achieved by applying a controlled
torque, or by the so-called part-turn method (where the nut is rotated further, following
tightening to snug tight). Alternatively, torque may be applied and measured by loadindicating washers, which deform under a pre-determined load. Other fasteners may be used
where a sacrificial length of the shank breaks off at a predetermined load. The frictional
resistance depends on the applied load, and the condition of the contact surfaces. Most
commonly, the contact surfaces are shot blasted and left unpainted, although other surface
finishes may be accommodated. The contact surfaces are frequently called the faying
surfaces.
Compared with the use of bolts in clearance holes, friction grip assemblies are generally
considered to be more expensive, due to the protection of the contact surfaces (usually
masked during painting), the equipment and time taken during installation, and the
subsequent application of protection systems to any unpainted steelwork which remains
exposed. The primary advantage of friction grip assemblies is that slip is prevented.
Secondary advantages include the pre-tension in the bolts, which prevents the bolt loosening
under vibration. Clause 6.1.7 of BS5950-1:2000 reminds designers that special consideration
is required when connections are subject to vibration, load reversal, or fatigue.
There are generally four reasons for using HSFG connections:
1.

Fatigue
In structures where fatigue is a design consideration, the use of preloaded assemblies
is recommended. The high level of preload means that the tension in the bolt does not
vary significantly, provided the plies are stiff and truly in contact at the bolt.

2.

Vibration
Severe, continued vibration has been known to lead to ordinary (non-preloaded) bolt
assemblies becoming loose. Preloaded HSFG bolts may be used when vibration is a
concern. Other effective solutions exist to prevent the nuts from coming loose,
including the use of lock nuts, nuts with nylon inserts, or split pins. Bolts may be
avoided altogether by designing a welded detail. It should be noted that nuts working
loose due to vibration is not normally a consideration in orthodox building structures,
and in most cases no special measures are required.

SCI ADVISORY DESK

3.

Load Reversal
When load reversal occurs (excluding reversal solely from wind loading) the possibility
of slip in the connections (as the clearance holes allow movement to and fro) should
be prevented. Significant reversal may take place in structures supporting moving
loads such as cranes. Preloaded HSFG bolts are recommended in these situations.
Alternative solutions exist to prevent slip in the connections, including close
tolerance bolts and injection bolts. Close tolerance bolts (also called fitted bolts)
are used in connections where the normal clearance is reduced to an absolute
minimum, so the bolts are in bearing immediately load is applied. This generally
involves drilling the holes undersize and reaming through both components
simultaneously to ensure an exact match of holes in each ply. Injection bolts permit
the injection of resin into the cavity left by bolts in clearance holes. At present,
injection bolts are not commonly used in the UK.
It may be possible to re-arrange the connection, such that the bolts are used in
tension, or to substitute a welded detail

4.

Dimensional Stability
Even if load reversal is not possible, it may be important to eliminate slip in
connections. Examples include splices in moment-resisting members, where slip in a
cover plate type splice may lead to additional deflections in the member and an
unsightly kink. In some situations, accumulated slip in a number of connections may
lead to additional deflection of the bolted assembly. Unexpected additional deflection
due to accumulated connection slip may lead, for example, to loss of fall on roof
trusses with shallow slopes. Whereas shallow trusses and similar assemblies of
numerous bolted components can be subject to this effect, in most sloping roof
trusses, the pitch is sufficient for the slip in the connections to be neglected. To
prevent the additional deflections from accumulated connection slippage, shallow
slope trusses and the vertical bracing connections in tall, multi-storey buildings are
often made with preloaded HSFG bolts.

These four situations demonstrate when and why a designer would specify a preloaded
(HSFG) bolt. However, for most building structures the use of ordinary bolts (non-preloaded)
in clearance holes is perfectly acceptable, even in moment connections, and the small
amount of slip associated with clearance holes has no practical consequences for the
structure.

SCI ADVISORY DESK

AD 249: Design of Members Subjected to Torsion

We have recently received a number of questions in relation to structural steel members
subjected to torsion. From these calls it is evident that there is some confusion over the
difference in behaviour of open and closed cross-sections. In response to these queries, it
therefore seems timely to present a qualitative background to the theory of torsion.
Torsional loading can arise within members in two ways: an externally applied torque; or
when the applied load acts eccentrically to the shear centre of the cross-section. In both
cases, the member will twist about its longitudinal axis, which passes through the shear
centre of the cross-section.
Categories of cross-sections
Torsional loading has a significant influence on the initial choice of section for maximum
structural efficiency. I-shaped sections are particularly poor in resisting torsion while hollow
sections can be very effective. A distinction is normally made between these two types of
sections, by calling I- and channel sections (which are poor in resisting torsion), Open
Sections (Figure 1); while rectangular and circular hollow sections (which are more effective
in resisting torsion), are referred to as Closed Sections (Figure 1).

Open sections

Closed sections

Fig. 1: Open and closed structural sections

Location of shear centre and its significance

The position of the shear centre is of particular importance in design, since when a load is
applied to a member, a torque will develop if the applied load does not act through the shear
centre of the cross-section. In these circumstances, the torque is simply equal to the applied
load multiplied by its eccentricity from the shear centre. A cross-section having two axes of
symmetry has its shear centre located at the centre of gravity of the cross-section (see
Figure 2(a)). With one axis of symmetry, the shear centre lies on that axis, but will in general
not coincide with the centre of gravity (see Figure 2(b)). However, for a section having skewsymmetry, the shear centre and the centre of gravity do coincide (see Figure 2(c)). For angle
sections, the shear centre is located at the intersection of the two legs (see Figure 2(d)).

SCI ADVISORY DESK

s
s

(a)

(b)

(c)

(d)

Fig. 2: Location of shear centre s and centre of gravity c

Reference will be made below to warping, which is best described by considering the
rectangular hollow section shown in Figure 3. In the initial conditions, the ends of the hollow
section are rectangular and plane. Suppose that a slit is made along one side of this section
(thereby transforming it from a closed section to an open section), and one end is twisted
relative to the other. As can be seen in Figure 3, in response to this applied torque, the ends
of the hollow section remain rectangular, but are no longer plane. This distortion of the crosssection is called warping, and is particularly pronounced in I-beams.

s
df

Warping
displacement, w

y
(a) Cross-section of split
rectangular hollow section

y
(b) Warping displacements

Fig. 3: Warping of a split (open) rectangular hollow section

The total resistance of a structural member to torsional loading may be considered to be the
sum of two components namely, uniform torsion and warping torsion. In some cases, only
uniform torsion occurs. Whereas, when both uniform torsion and warping torsion are included
in the torsional resistance, the member is in a state of non-uniform torsion. A diagrammatic
representation of uniform and non-uniform torsion, on a member composed of an I-section, is
shown in Figure 4.

SCI ADVISORY DESK

Constant torque
ends free to warp

Varying torque

End warping prevented

(a) Uniform torsion

(b) Non-uniform torsion

(c) Non-uniform torsion

Fig. 4: Uniform and non-uniform torsion of a member composed of an I-section (viewed on plan)

Uniform torsion
When a member is subjected to uniform torsion (sometimes referred to as pure or St Venant
torsion), the rate of change of the angle of twist is constant along the member, and the
longitudinal warping deflexions are also constant along the member (see Figure 4(a)). In this
case, the torque acting at any cross-section is resisted by a single set of shear stresses
distributed around the cross-section (see Figure 5). The ratio of the torque T to the twist
rotation , per unit length, is defined as the torsional rigidity GJ of the member; where G is
the shear modulus and J is the torsion constant (sometimes called the St Venant torsion
constant).

Constant shear
flow t

(a)

t
t

(b)

Fig. 5: Shear stresses due to uniform torsion of (a) closed sections; and (b) open sections

Non-uniform torsion
When a member is subjected to non-uniform torsion, the rate of change of the angle of twist
varies along the member (see Figure 4(b) and (c)). In this case, the warping deflexions vary
along the member and, to resist the applied torque, an additional set of shear stresses act in
conjunction with those due to uniform torsion. The stiffness of the member associated with

SCI ADVISORY DESK

these additional shear stresses is proportional to the warping rigidity EH; where E is the
modulus of elasticity and H is the warping constant.
For a member composed of an I-section, the action of warping resistance can be visualised as
follows: the torque T is resisted by a moment comprising of forces equal to the shear forces
in each flange, which are separated by the lever arm, df equal to the depth between the
centroids of the two flanges. If each flange is treated as a beam, the bending moments
produced by these forces lead to the warping normal stresses, as shown in Figure 6(b).
For an I- or H-section, this approach provides a reasonable approximation, but will generally
over-estimate the warping normal stress whilst under-estimating the warping shear stress
(since the approach ignores the shear stresses from uniform torsion). However, it cannot be
readily applied to channel sections; in such circumstances, more rigorous methods of analysis
need to be adopted.

Warping normal stresses fw

z

Mf

Bimoment d w M f
Warping shear stresses w

c,s

df

df
Mf

y
(a) Rotation of cross-section

Vf

Flange shears Vf

Vf
(b) Bimoment and warping stresses

Fig. 6: Warping stresses in an open section member composed of an I-beam

Effect of cross-section on torsional behaviour

Before examining how different types of sections perform in resisting torsion, it is useful to
first introduce the non-dimensional torsion parameter for the member:
K =

( 2 EH / GJL2 ) .

For sections that have a very high torsional rigidity GJ compared to their warping rigidity EH,
K becomes small; in these circumstances the member will effectively be in a state of uniform
torsion (as indicated in Figure 7). Closed sections, whose torsional rigidities are very large,
behave in this way, as do sections whose warping rigidities are negligible, such as angle and
T-sections.
Conversely, for sections whose warping rigidity EH is very high compared to their torsional
rigidity GJ, K becomes very large, and the member is in the limiting state of warping torsion
(as indicated in Figure 7). Very thin-walled open sections, such as light gauge cold-formed

SCI ADVISORY DESK

(Warping torque/total torque) at mid-span

sections, whose torsional rigidities are very small, behave in this way. Between these
extremes, the applied torque is resisted by a combination of the uniform and warping torsion
components, and the member is in the general state of non-uniform torsion. This occurs for
intermediate values of the parameter K, as shown in Figure 7, which are appropriate for most
open sections such as hot-rolled I- or channel sections.

Closed
sections

Hot-rolled
sections

Very thin-walled
channel sections

Angle
sections
1.0

Uniform
torsion

Warping
torsion
Non-uniform
torsion

0.8
0.6
0.4
0.2
0

0.1

1
10
Torsion parameter, K

Fig. 7: Effect of cross-section on torsional behaviour

Whether a member is in a state of uniform or non-uniform torsion also depends on the

loading arrangement and the warping restraints. If the torsion resisted is constant along the
member and warping is unrestrained (as shown in Figure 4(a)), then the member will be in
uniform torsion, even if the torsional rigidity is very small. If, however, the torsion resisted
varies along the length of the member (Figure 4(b)), or if the warping displacements are
restrained in any way (Figure 4(c)), then the rate of change of the angle of twist rotation will
vary, and the member will be in a state of non-uniform torsion.
As a result of applying a torque to a member, the torsional stresses induced within the
section, which should be considered in design, are:
a) Shear stresses due to uniform torsion.
b) Shear stresses due to warping torsion.
c) Bending stresses due to warping.
Each of the above stresses is associated with the angle of twist , or its derivatives. Hence,
if is determined for different positions along the member length, the corresponding stresses
can be evaluated at each position.
Torsion of closed sections
As discussed above, the torsional rigidity GJ of a closed section is very large compared with
its warping rigidity EH (see Figure 7). Therefore, a member composed of a closed section
may be considered to be subject only to uniform torsion. For a rectangular or circular hollow
section, the uniform torsion will result in a uniform shear stress developing within the walls
of the cross-section (see Figure 5(a)). In these circumstances, the problem is statically

SCI ADVISORY DESK

determinate, and the shear stress , as well as the angle of twist , may be determined from
simple statics.
Torsion of open sections
As discussed above, for members composed of open sections such as hot-rolled I- or channel
sections, the section may be considered to be subject to the general state of non-uniform
torsion (see Figure 7). In these circumstances, the applied torque is resisted by a combination
of uniform torsion and warping torsion components.

Uniform torsion
If a torque is applied to the ends of a member, in such a way that the ends are free to warp,
then the member will only develop uniform torsion (Figure 4(a)). The resulting shear stresses
will vary linearly across the thickness of the element (see Figure 5(b)): they are maximum at
the element surfaces, with two equal values, but opposite in direction. These stresses are a
function of the rate of change of the angle of twist, and are greatest in the thickest element
of the cross-section i.e., typically the flanges in an I-beam. (At junctions between the web
and the flanges, the local shear stresses may exceed the stresses in the thickest element of
the cross-section; for rolled sections, this effect may be neglected by the designer, as
allowance for the root fillet radii are made in determining the torsional constant J in section
property tables).

Warping torsion
When a uniform torque is applied to a member restrained against warping, the section itself
will be subject to non-uniform torsion with the rate of change of the angle of twist varying
along the length of the member (see Figure 4(c)). The rotation of the section with respect to
a restrained end will be accompanied by bending of the flanges in their own plane
(sometimes referred to as the Bimoment). The warping normal and warping shear stresses
developed by this condition are shown in Figure 6. Warping stresses are also generated in
members of open section when the applied torque varies along the length, even if the ends
are free to warp (Figure 4(b)).
End conditions
As discussed above, the end conditions will also greatly influence the torsional stresses along
the member. Note that end conditions for torsion calculations may be quite different from
those for bending e.g., a beam may be supported at both ends, but torsionally restrained at
only one end: the torsional equivalent of a cantilever. Torsional fixity must be provided by at
least one point along the length of the member, otherwise it will simply physically twist when
the torque is applied. Also, warping fixity cannot be provided without also providing torsional
fixity. As a result, there are three possible boundary conditions which may sensibly
considered in torsion calculations:
i.

Torsion fixed, warping fixed

This condition is satisfied when, at the ends of the member, both twisting about the
longitudinal axis and warping of the cross-section are prevented (sometimes referred to
as a Fixed torsional end condition). Effective warping fixity is practically almost
impossible to achieve in most structures. A connexion providing fixity in both directions
is not sufficient, it is also necessary to restrain the two flanges either side of the web.
Details such as those shown in Figure 8(a) need to be provided to achieve this type of
boundary condition. It should be noted, however, that the provision of warping fixity
does not produce such a large reduction in torsion stresses as is obtained from bending

SCI ADVISORY DESK

fixity. Therefore, it is more practical to assume warping free connexions (see (ii)
below), even when fixity is provided in terms of bending.
ii.

Torsion fixed, warping free

This is satisfied when the cross-section at the ends of the member is prevented from
twisting, but is allowed to warp freely (sometimes referred to as a Pinned torsional end
condition). Such a condition may be readily achieved by providing the relatively simple
standard connexions, such as shown in Figure 8(b).

iii.

Torsion free, warping free

This condition is achieved when the end is free to warp and twist (sometimes referred
to as a Free torsional end condition): the unsupported end of a cantilever illustrates this
case.

Plate

(a) Torsion fixed, warping fixed

Channel

(b) Torsion fixed, warping free

Fig. 8: Practical end conditions

The above gives a basic overview of the considerations that should be made in the design
process for structural steel sections, which are to be subjected to torsion. Design equations
for estimating the stresses due to torsion, in combination with bending, may be found in SCI
publication 057 entitled Design of Members Subject to Combined Bending and Torsion.

SCI ADVISORY DESK

AD 250: Choice of Steel Sub-Grade

During a recent seminar on the Amendments to BS 5950-1: 1990, a straw poll of the (large)
audience indicated that only 5% of the structural engineers present specified the steel
sub-grade. A further 2% of the audience indicated that they checked that the steelwork
contractor chose the correct sub-grade. This left over 90% of the audience who neither
specified the sub-grade themselves, nor ensured that others were specifying the correct
sub-grade.
The implication of this poll is that there is potential for an inappropriate sub-grade of material
used in structures, which could lead to an unacceptably high risk of brittle fracture.
This advisory note serves as a reminder that an appropriate sub-grade must be specified. It
should be noted that the requirement to specify an appropriate sub-grade is not new, being
covered in BS 5950-1:1990.
Brittle fracture becomes increasingly likely:

At lower temperatures.

As the stress level increases.

As the strain rate increases.

The more stress raisers that are present (which disrupt the uniform state of stress).

The thicker the element.

If the circumstances make brittle fracture possible, a tougher steel must be specified.
Tougher steels are generally more expensive.
BS 5950-1:2000 requires the choice of a K factor from Table 3, which covers the
circumstances in which the steel is being used. Table 3 addresses stress level, type of
detail and strain conditions which combine to indicate a K factor between 0.25 and 4. The
lower the K factor, the tougher the steel will need to be.
In most circumstances, the K factor will be at least as low as 1, since this is the K factor for
welded generally, with a stress higher than 30% of the normal yield strength.
The requirement in Clause 2.4.4 is that:
t < Kt1
where

t
t1

is the thickness of each element (commonly the flange)

is the limiting thickness at the minimum service temperature from Table
4 or 5 of BS 5950-1:2000

Knowing the element thickness and knowing the K factor, Table 4 (plates, flats and rolled
sections), or Table 5 (hollow sections), may be used to choose an appropriate sub-grade,
such that t1 > t/K.

Who should specify the Steel Sub-grade?

The designer of the structure knows the information needed to specify the steel sub-grade
and should take responsibility for ensuring an appropriate steel is used. Unless steelwork
contractors are designing the structure, they will need the sub-grade at tender stage to
prepare a quotation. To leave the contractor to choose the sub-grade is not satisfactory, as

SCI ADVISORY DESK

it is not necessarily clear from arrangement drawings if steel is exposed, or if it is highly
stressed.
Note that changes in the construction details could affect the choice of sub-grade. For
example, a welded beam to column connection might be stiffened or unstiffened. The K
factor is higher if the column flange is stiffened. Thus if a heavier, unstiffened column section
is used to replace a stiffened column, the K factor would reduce, and a tougher steel may be
required because of the revised detail and the increased flange thickness.

Steel Used in External Conditions

Specifiers should note that, generally, it will be necessary to specify at least a J0 sub-grade
for steel used in external conditions. This represents a change from BS 5950-1:1990 (as
amended in 1992), which allowed less tough steels to be used in exposed conditions.

Why no history of failures to date?

If, according to the straw poll, there is every possibility of a less capable steel being used
than the design standard would indicate, why is there no extensive catalogue of failure in
service? Clearly, even though the risk is increased, this does not mean that failure occurs.
One further contribution is the steel itself which, when of UK origin, has typically been
tougher than specified. This, however, does not represent an excuse to avoid proper
specification, particularly as more steel is sourced from foreign manufacturers.

SCI ADVISORY DESK

AD 251: Equivalent Uniform Moment Factor, m

BS5950-1:2000 sees the demise of the n factor method when calculating buckling resistance
Mb. In Section 4, the m factor method is the only approach available, and will become a
familiar friend.
The equivalent uniform moment factor, m, allows for the fact that, for two identical beams
with the same maximum applied bending moment, the buckling resistance is greater for a
varying bending moment diagram than for a uniform bending moment diagram.
In the two examples shown below, with the same maximum applied moment, the lateral
displacement of the compression flange is more pronounced when the member is subject to a
uniform bending moment. Thus, the member will buckle at a lower maximum moment if the
moment is uniform. Alternatively, the buckling resistance of a member subject to a varying
bending moment is larger than when the same member is subject to a uniform bending
moment.
In BS5950-1:1990, the m factor method could only be used if there were no intermediate
loads. In BS5950-1:2000 the method has wide application. Table 18 contains members
subject to end moments only, some standard cases of intermediate loads and a formula to
establish mLT for a general loadcase.
BS5950-1:2000 distinguishes between the equivalent uniform moment factor for lateral
torsional buckling, mLT, and the equivalent uniform moment factors for flexural buckling, mx,
my, and myx. Lateral torsional and flexural buckling are different (one does not have the
torsional component) and thus different Tables for m are required. Table 26 covers m values
for flexural buckling, presenting cases for end moments only, standard cases of intermediate
load, and a general case.

Fig. 1: Member with a uniform applied moment

SCI ADVISORY DESK

Fig. 2: Member with a varying applied moment

m Factor for General Case

To use the general case formula, one needs to know the bending moment at quarter points,
and the maximum moment on the segment. Software may already provide this information
(or may be configured to provide this information). No doubt new software to BS5950-1:
2000 will automatically determine this information to calculate m.
Flexural buckling m factors when used with Sway-Sensitive frames
Clause 4.8.3.3.4, in the section covering member buckling checks, contains important
information about the use of flexural buckling m factors mx, my, and myx, when designing
sway-sensitive frames.
If using Annex E (effective length factors greater than 1.0), the minimum value of the factors
mx, my, and myx is 0.85. If amplifying the sway moments, the factors mx, my, and myx can
only be applied to the non-sway component of the total bending moment. Neither of these
restrictions applies to mLT.
Designers should note the potential for misapplication of the m factors, particularly if
applying them to sway sensitive frames. The following table summarises the position.
Factor

mLT

mx, my, myx

Association

Lateral-torsional buckling

Flexural buckling

Use with

MLT

Mx, My

Determined from:

Table 18

Table 26

Restrictions

In sway-sensitive frames:
i) mimimum 0.85 if using Annex E
ii) only applied to non-sway moments if
amplifying sway effects

SCI ADVISORY DESK

AD 252: Axially Loaded Columns - Base Plate

Design - Effective Area Method BS 5950-1: 2000
The empirical method for determining the size of base plates in BS 5950-1:1990 has been
replaced by the effective area method in BS 5950-1:2000. Doubt has been raised as to the
theoretical validity of the empirical method when used with deep UBs or with bases that
have very small outstand dimensions a and b (see Figure 1). The effective area method offers
more economy than the empirical method, while still producing safe designs when compared
to test results.
The effective area method for base plate design may initially seem to be more complex than
the empirical method given in BS 5950-1:1990. However, the approach is much more reliable
and can be used for all column sections. The forthcoming revision of the SCI and BCSA green
book Joints in Steel Construction: Simple Connections will include design procedures for base
plates using the effective area method and will include standard base plates and their
capacities for I, H and hollow section columns.
We have received several questions concerning the effective area method for base plate
design, and in particular what to do when the effective area does not fit on the base plate
selected.
The basic design procedure is set out below.
1. Calculate required area = axial load / 0.6 fcu.
Where:
fcu is the cube strength of either the concrete or the grout, which ever is weaker.
2. Calculate outstand c (see Figure 1) by equating required area to actual area expressed
as a function of c. The expression for the actual effective area of an I or H section
may be approximated to 4c2 + (column perimeter) c + column area.

Effective area
c

a
c

Fig. 1

Overlap area

Fig. 2

Effective area

c
c

a
c

Fig. 3

3. Check that there is no overlap of effective area between flanges (see Figure 2). This
will occur if 2c > the distance between the inner faces of the flanges. If an overlap
exists, modify the expression for effective area and recalculate c.
4. Check the effective area fits on the size of base plate selected (see Figure 3). If
effective area does not fit on the base plate, modify the expression for effective area
to allow for the limitations of the plate size and recalculate c, or select a larger base

SCI ADVISORY DESK

plate. For the case shown in Figure 3, the modified expression for the effective area
will be,
4c2 + (column perimeter) c + column area 2 (B+2c) (c a).
5.

If c has been recalculated, step 3 will need to be repeated.

6. Calculate required plate thickness tp using expression below (given in Clause

4.13.2.2):
0.5
3 w

t p = c
p
yp
Where:
w

= 0.6fcu

pyp

is the plate design strength

The expression for the plate thickness can be derived from equating the moment
produced by the uniform load w to the elastic moment capacity of the base plate (both
per unit length).
Moment from uniform load on cantilever
w c2 / 2
w c2 / 2
Rearranging gives
tp

Elastic moment capacity of plate

= py Z (per unit length)
= py tp2 / 6
= c(3 w /py)0.5

When the outstand of the effective area is equal either side of the flange (as in Figure 1
and Figure 4), the cantilevers are balanced and there is no resultant moment induced in
the flange. However, if the cantilevers do not balance either side of the flange, as would
be the case in Figure 3, then theoretically to satisfy equilibrium there is a resultant
moment induced in the flange (see Figure 5 and Figure 6).

c
a

M=(c2-a2)/2

M= a2/2
M= wc2/2

Fig. 4

M= wa2/2

Fig. 5

M= c2/2

M= wc2/2

Fig. 6

It is important to remember that the method given in BS 5950-1:2000 is only a design

model and does not exactly represent what will happen in practice. The remainder of
the plate (not only the effective area) does exist and does carry load. With this in
mind, the moment induced in the column flange due to unbalanced cantilevers does
not need to be explicitly considered in the design of either the column or the base
plate.

SCI ADVISORY DESK

AD 253: Design Considerations for the Vibration

of Floors Part 1 (Amended)
Introduction
Designers are more frequently facing the need to consider, in detail, the vibration response of
floors. This Advisory Desk article has been written as a summary of the key issues.
A structure that is composed of any material will vibrate if subjected to cyclic or sudden
loading. In most cases, the vibrations are imperceptible, and so can be neglected in building
design. However, in some circumstances, the response of the structure to a common cyclic
load (e.g. walking activities, etc.) is sufficient to produce a response that is perceptible to the
occupants of the building. The most common example of such small, but perceptible,
vibrations occurs in floor structures. This is not a new phenomenon, but it is more noticeable
within the working environment of modern offices.

What disturbs building occupants?

For structures that are subjected to static loading, it is normal for the engineer to calculate
the vertical deflection of the floor. This is to avoid undesirable deflections and to limit the
possibility of damage to brittle finishes that will occur in serviceability conditions. However,
for floors that are subject to cyclic or sudden loading, human perception of motion is usually
related to acceleration levels rather than displacement. In most practical building structures, if
the magnitude of these acceleration levels is not limited, the typical reaction of the building
occupants varies between irritation and a feeling of insecurity. This is based on the
instinctive human perception that motion in a solid building structure indicates structural
inadequacy or failure.
The working environment also affects the perception of motion. For busy environments,
where the occupant is surrounded by the activity that is producing the vibrations, the
perception of motion is reduced. In contrast, for quiet environments (such as laboratories and
residential dwellings), where the source of the vibration is unseen, the perception of motion
is significantly heightened.

A simple model of a vibrating system

A simple model that illustrates the vibration behaviour of a structure is shown in Fig. 1 (a).
The bending stiffness is modelled as a spring of stiffness k, and the floor mass is modelled by
a point of mass m. All practical structures will have some damping, conveniently modelled as
a viscous (or oil-pot) damper. Damping refers to the loss in mechanical energy within a
mechanical system. Practical floor structures possess a low level of natural damping
(normally in the order of 1%), which does not affect the natural frequency of the system.
However, the magnitude of the damping is very important, as this determines the maximum
possible magnification to the acceleration response of the floor (see below).

SCI ADVISORY DESK

Viscous
damper, c

Stiffness, k

Displacement,

Period, T

Time, t

Mass, m
Displacement, v(t)

External
force, p(t)

(a)

(b)

Fig. 1(a): Idealised single-degree-of-freedom system,

basic components

Fig. 1(b):
Idealised single-degree-of-freedom system,
undampened free vibration response

Higher damping depends on the energy dissipation through non-structural components such
as partitions, which are largely dependant on frictional forces. Intuitively, it might be
expected that the presence of the false flooring would have an effect on the damping.
However, experimental studies have shown that this is not the case. The reason is that
deflections of modern false floors do not produce sufficient movement to develop significant
friction forces.
A disturbance to the floor by a suddenly applied load will cause the system to vibrate as
shown in Fig. 1 (b). If no other external forces are applied to the system, the damping will
cause the displacements to die away with time.

Walking activities
Walking produces a cyclic loading that is repeated at regular intervals called periods, which
are inversely proportional to the pace frequency. A typical example of a load-time function
that is produced by walking activities is shown in Fig. 2 (a).
1.8
1.6

Force / static weight

1.4
1.2
1
0.8
0.6
0.4
0.2
0
0

0.25

0.5

0.75

1.25

1.5

Time (s)

Fig. 2(a): Loading caused by an individual walking at 2.0Hz, load-time function

In general, a repeated force can be represented by a combination of sinusoidal forces, whose

frequencies are multiples (or harmonics) of the pace frequency, as shown in Fig. 2 (b). The
magnitude of the force for each of these harmonic components is taken as a proportion of
the static weight of the person multiplied by a Fourier coefficient n . The values of the
Fourier coefficients have been established experimentally for different activities and different
activity frequencies. For walking activities, Fig. 3 shows how the magnitude of the first
three harmonic components vary with frequency.

SCI ADVISORY DESK

1.8
1.6

Force / static weight

1.4
1.2
1
0.8
0.6
1st Harmonic component

0.4

2nd Harmonic component

0.2

3rd Harmonic component

0
0

0.25

0.5

0.75

1.25

1.5

Time (s)

Fig. 2(b): Loading caused by an individual walking at 2.0 Hz, decomposition of the load-time function into
the first three harmonic load components

The trend exhibited by Fig. 3 is that the magnitude of the Fourier coefficient is lower for
higher harmonics. For example, the average Fourier coefficient for the first three harmonics
of walking is typically taken as 1 = 0.4, 2 = 0.1 and 3 = 0.1. In buildings, it is
considered that a walking frequency of between 1.7 and 2.4 Hz can realistically occur.
(However, a higher frequency range may be found for other activities e.g., dancing, aerobics,
etc.).
0.6
Harmonic, n
1
r
2

Fourier coefficient, n

0.5

0.4

0.3

0.2

0.1

0
0

10

Harmonic component frequency (Hz)

Fig. 3: Fourier coefficients for walking activities

Floor response
When cyclic forces (such as those from walking activities) are applied to a structure, it will
begin to vibrate. If the cyclic forces are applied continuously to the simple model in Figure
1(a), the motion of the structure will reach a steady-state i.e., vibration of a constant
amplitude and frequency will be achieved (see Fig. 1 (b)). The magnitude of the peak
acceleration response for a long continued excitation is given by:
applied force
Dynamic magnification factor
a peak =
mass

SCI ADVISORY DESK

As can be seen from the above expression, the peak acceleration response decreases as the
mass of the floor increases and increases as the dynamic magnification factor increases.
Mathematically, the dynamic magnification factor is given by the following identity:
D=

(1 )

2 2

+ (2 )2

where is the ratio of the activity frequency to the natural frequency of the structure and
is the damping ratio.
As can be seen from the second equation, the dynamic magnification factor is related to the
level of damping that the floor possesses, and the ratio (e.g., from walking, etc.) This
variation is shown graphically in Fig. 4 below.
10

Dynamic magnification factor D

= 0.015

8
7
6
5

= 0.1

= 0.5

= 1.0

2
1
0
0

Frequency ratio (activity frequency / frequency of structure)

Fig. 4: Variation of dynamic magnification factor with damping and frequency.

Fig. 4 shows different types of response depending on the frequency ratio . When the
cyclic force from the activity is applied at a frequency much lower than the natural frequency
of the system ( << 1), the response is quasi-static. In these circumstances, the steadystate response is governed by the stiffness of the structure, resulting in an amplitude close to
the static deflection. Conversely, when the dynamic force is applied at a frequency much
higher than the natural frequency of the system ( >> 1), the steady-state response is
governed by the mass (inertia) of the structure. In this case, the amplitude is less than the
static deflection. Neither of these ranges of are of great practical importance.
The range of of greatest importance is where the dynamic force is applied at a frequency
close to the natural frequency of a structure. In any structure which is lightly damped (as is
found in most practical floor systems), very large responses will occur. The condition when
the frequency ratio is unity is called resonance (i.e., the frequency of the applied load equals
the natural frequency of the structure; = 1). At resonance, very large dynamic
magnification factors are possible and, for undamped systems (i.e. = 0), the steady-state

SCI ADVISORY DESK

response tends towards infinity. A more general result may be obtained from the second
equation, which shows that for resonance ( = 1) the dynamic magnification factor is
inversely proportional to the damping ratio, and: (Fig. 4).
D =1 =

1
2

Since in many practical structural systems the natural damping is of the order of 1%, if
precautions against resonance are not made, magnification factors of up to 50 may result.
Given that the force in the structure is proportional to the displacement, the dynamic
magnification factor also applies to structural forces. In spite of this, for office floors, the
magnifying effects to this load are normally neglected when assessing ultimate limit state
criteria because the static weight of an individual walker is so small. This is certainly not the
case when large groups of people (crowds) take part in synchronised activities (i.e. dancing,
aerobics, etc.). In these circumstances, the magnification on the static load of the crowd may
cause extreme loading to the floor, and should be considered as an additional imposed
loading case for design at the ultimate limit state. In the UK, guidance is given in BS 63991:1996 for structures that fall within this special category.
The above gives a basic overview of dynamics, and what considerations should be made in
the design process for floors, which are to be subjected to occupant-induced vibrations.
Design equations for estimating the fundamental frequency, and response of a floor, may be
found in SCI publication 076. In addition, further design recommendations can be found in a
recently completed SCI technical report RT 852 entitled Design Guide for Vibrations of Long
Span Composite Floors, which also compares the performance of present design guide
predictions with measured results. As well as acting in an advisory rle, for the past three
years, the SCI has also offered a consultancy service for designers. Our experts have
produced all the necessary calculations for clients, on a variety of different types of floor. In
addition to the initial design stage, we have evaluated existing floors and, where necessary,
offered advice on remedial measures. We can also offer an in-situ testing service on floors
where it is perceived that there may be a potential vibration problem.
Note: See also AD254
Note: This Note has been amended following the correction to the formula for D given in
New Steel Construction, issue January/February 2002.

SCI ADVISORY DESK

AD 254: Design Considerations for the Vibration of

Floors Part 2
Introduction
AD 253 Design considerations for the vibration of floors- Part 1 warned that where vibrations
disturb the occupants of a building it is usually excessive accelerations that cause the
disturbance. It also warned that having the frequency of the floor above a certain threshold
frequency, such as 3Hz, does not necessarily guarantee that the floor will prove satisfactory
in service. In addition, AD 253 described the forcing function produced by walking activities,
and summarised the dynamic behaviour of floor structures. This Advisory Desk article
follows on from AD 253 by giving information on acceptable levels of acceleration and
explaining how the expected acceleration may be calculated.

Acceptability criteria
The evaluation of the exposure of humans to vibrations within buildings is covered by
BS 6472: 1992 Guide to evaluation of human exposure to vibration in buildings (1Hz to
80Hz), which is written to cover many vibration environments in buildings. This Standard
presents acceleration limits for vibrations as a function of the exposure time and the
frequency. It contains graphs of acceleration against frequency for both lateral and vertical
vibration. These graphs are called base curves. The base curve for vertical vibration is
in Figure 1. Note that the acceleration is given as the rms (root mean square) value.

rms acceleration (m/s)

1.000

0.100

0.010

0.001
1

10

Frequency (Hz)

100

Fig. 1: Building vibration z-axis base curve for root-mean-square (rms) acceleration according to BS
6472:1992

The accelerations acceptable for different uses of buildings are described using the base
curve. Multiplying factors are used to increase the acceleration level of the base curve

SCI ADVISORY DESK

according to the intended use of the building. These multiplying factors are referred to as
response factors within the SCI guide P 076, Design Guide on the Vibration of Floors.
Suitable design values for different office accommodation will also be given in a future
Advisory Desk article.
The maximum calculated value of acceleration is found as follows. The base curve
acceleration, abase , is found from Figure 1 for the fundamental frequency of the floor. This
acceleration, abase , is multiplied by the response factor, R, appropriate to the use of the floor
to give a limiting acceleration abase R .
For example, for a floor with the fundamental frequency equal to 5Hz, the base curve
acceleration is 0.005 m/sec2 rms. If the R value appropriate for the use of the building is 8,
then acceptable acceleration is 0.005 8 = 0.04 m/sec2 rms.

Calculation of the accelerations

Resonance can occur even if the fundamental frequency of the floor is above a given
minimum design value. This is because vibrations arise from components of the walking
activity. These components of walking activity occur because the force versus time graph of
walking is made up of many different sine curves, as shown in AD 253. The largest response
will generally occur when the lowest whole number multiple (harmonic) of the activity
frequency is equal to the fundamental frequency of the floor (i.e., resonance). For example,
consider a floor with a fundamental frequency f 0 = 4.6 Hz. The frequency range for walking
activities is 1.7 Hz to 2.4 Hz, so the pace rate that would cause the highest floor response
would be from the second harmonic of the walking activity (2 2.3 Hz = 4.6 Hz). If the
floor design was changed so that the fundamental frequency was raised to 6Hz, the highest
floor response would be from the third harmonic of the walking activity (3 2.0 Hz =
6.0 Hz.
The peak acceleration response is calculated from the following equation:
a peak =

P0 n 1
M 2

where P0
n
M
and

is
is
is
is

(1)

the
the
the
the

weight (=mass gravity) of the person in Newtons (N),

Fourier coefficient of the nth harmonic component of the walking activity,
effective vibrating mass (kg)
damping ratio.

(The above equation can be derived from AD 253 from the equation for a peak , using D =1 as
the value of the Dynamic magnification factor.)
The response assumed in developing the above equation is purely sinusoidal, whereas the
limits to vibrations in BS 6472 are given in terms of the root-mean-square (rms) acceleration
arms . The rms acceleration is found from this sinusoidal response by dividing a peak by 2 .
The value of

a peak

should be compared with acceleration from the curve in Figure 1, abase ,

2
(for the fundamental frequency of the floor) multiplied by the response factor, R, appropriate

SCI ADVISORY DESK

to the use of the floor. If

a peak
2

is greater than abase R , the floor may be unsuitable for

the intended use.

The mass of a person is commonly taken as 76 kg.
The appropriate Fourier coefficient of the walking activity, n , is the value that coincides
with the fundamental frequency of the floor, which may be calculated by the methods given
below. Simple recommendations for design values of n will be given in a future Advisory
Desk article. The highest values of n occur at the first harmonic of walking activity, as
shown in AD 253 Figure 3. Therefore, minimum frequency limits are normally given in
design guides and codes of practice to ensure that the fundamental frequency of the
structure does not coincide with the first harmonic of walking activities. By ensuring that the
fundamental frequency of the floor is above a the first harmonic of walking activities, the
magnitude of the floor acceleration is reduced because the higher harmonic components of
the walking activity have lower values of n . However, as can be seen from Equation (1),
the acceleration response from the floor may still be very high if the mass is too low. A
practical example of a floor with a low mass would be a steel framed structure with a timber
floor.
The response calculation in Equation (1) uses only one harmonic of the walking activity,
which is the harmonic that coincides with the fundamental frequency of the floor. This is
normally sufficient for checking the response to walking activity. However, for more
vigorous activities such as dancing, aerobics etc, it is recommended that more harmonics are
considered.

Floor mass for vibration and the effective vibrating mass, M

For offices in the UK, 10% of the design imposed loading is normally considered as
permanent, and represents the sensible loading on a furnished floor. In the Eurocodes, the
proportion of the imposed loading considered is much higher, and corresponds to the
permanent loads plus a "frequent variable action" combination factor, 1, times the imposed
load. For offices, the UK National Application Document (NAD) for Eurocode 3 Part 1,
DD ENV 1993-1-1, gives 1 = 0.6 , which implies that a very large proportion of the imposed
load is composed of office furniture and stored materials such as books and paper.
The effective vibrating mass M may be calculated using the SCI publication P 076. This takes
M as the floor mass per unit area (comprising the self weight, and other permanent loads,
plus a proportion of the imposed load) multiplied by one-quarter of the effective vibrating
area. P 076 gives a method to calculate the effective length, Leff , of the effective vibrating
area and the effective width, S , of the effective vibrating area. Therefore, M is calculated
from:
M =

mLeff S
S
4

(2)

where m is mass per unit area (kg/m2) of the floor plus any loading that is considered to be
permanent.

SCI ADVISORY DESK

Fundamental frequency
When individual structural components are inter-connected to form a complete floor system
and this floor system vibrates, the whole floor structure moves up and down in a particular
form, known as a mode shape. Although, each floor frequency has a particular mode shape
associated with it, it is generally the lowest (1st mode) or fundamental frequency that is of
particular interest in design (see Figure 2), because the largest acceleration response is
normally found when this mode is excited to resonance. The fundamental frequency of the
floor is lower than the frequency of any of the individual structural components.

Fig. 2: Beam mode shapes (first three frequencies)

In conventional composite floor systems, the fundamental frequency may be estimated by

using engineering judgement on the likely mode shape, considering how the supports and
boundary conditions will affect the behaviour of the individual structural components. For
example, on a simple floor comprising a slab which is continuous over a number of secondary
beams and these secondary beams are supported by stiff primary beams, there are two
possible mode shapes that may be sensibly considered:
1.

Secondary (floor) beam mode

The primary beams form nodal lines (i.e. they have zero deflexion) about which, the
secondary beams vibrate as simply-supported members (see Figure 3(a)). In this case, the
slab flexibility is affected by the approximately equal deflexions of the supports. As a result
of this, the slab frequency is assessed on the basis that fixed-ended boundary conditions
exist.
2.

Primary (main) beam mode

The primary beams vibrate about the columns as simply-supported members (see Figure
3(b)). Using a similar reasoning as above, due to the approximately equal deflexions at their
supports, the secondary beams (as well as the slab) are assessed on the basis that fixedended boundary conditions exist.

SCI ADVISORY DESK

(a)

Fig. 3:

(b)

Typical fundamental mode shapes for composite floor systems (a) governed by secondary
beam flexibility (b) governed by primary beam flexibility)

The lowest frequency value determined by consideration of these two cases defines the
fundamental frequency of the floor f 0 (and its corresponding mode shape). As steel-concrete
composite construction is essentially an overlay of one-way spanning elements, the
frequency of the whole floor system can be calculated for each mode shape, by summing the
deflexion calculated from each of the above components, and placing this value within
Equation (7). Alternatively, it can sometimes be convenient to use these component
frequencies directly, to evaluate the fundamental frequency of the floor by Dunkerlys
approximation shown in Equation (3) below. These two methods give identical results.
1
f0

1
f12

1
f2

1
f32

(3)

where f1 , f2 and f3 are the component frequencies (Hz) of the composite slab, secondary
beams and primary beams respectively, with their appropriate boundary conditions, as
defined above. The frequency of the components may be calculated using the methods
given below.

Natural frequency of components

The natural frequency of a simple spring-mass system as shown in AD 253 Figure 1(a) is
given by:
f =

1
1
=
T
2

k
m

(4)

where f is the natural frequency (Hz), T is the period for one complete cycle of movement
(seconds), k is the stiffness, and m is the mass.
The mass used is the mass relevant to vibration calculations. It is normally taken in the UK
as the mass from permanent loads plus 10% of the mass of the live load, but the Eurocodes
may suggest adopt a higher load, as explained above in Floor mass for vibration and the
effective vibrating mass, M.
The natural frequency of a beam of uniform section is given by:

SCI ADVISORY DESK

f = CB

EI

(5)

mL4

where EI is the flexural rigidity of the member (Nm), m is the mass per unit length (kg/m), L
is the span of the member (m), and C B is the frequency factor representing the beam
support and/or loading conditions.
Some standard values of C B for elements with different boundary conditions are as follows:
pinned/pinned (simply-supported)
fixed/pinned (propped cantilever)
fixed both ends
fixed/free (cantilever)

/2
2.45
3.57
0.56

For uniform beams which are continuous over supports (e.g., secondary beams connected
either side of a primary beam web), the natural frequency will increase when one span is
stiffer (shorter) than the main span. However, when the spans are equal, the natural
frequency will be the same as for a simply-supported beam (i.e., CB = / 2 ). A graphical
representation of this effect on C B for 2- and 3-span uniform beams is shown in Figure 4
below.

CB

Three-span
L

3.0

2.0
Two-span

R
1.0

0.2

0.4

0.6

0.8

1.0

Span ratio, R /L
Fig. 4: Frequency factor CB for uniform continuous beams

The natural frequency, f, of a simply supported beam may be determined conveniently from
the maximum deflexion when the beam is loaded with mass m appropriate for the
calculation of frequency. For a simply-supported beam, with a uniformly distributed load, this
is:

5wL4
5mgL4
=
384 EI
384 EI

where g is the acceleration due to gravity (i.e., 9.81 m/s).

(6)

SCI ADVISORY DESK

Rearranging Equation (6), and substituting the value of m and C B = / 2 into Equation (5),
gives the well known natural frequency expression that is often used in design:
f =

17.8

18

(7)

where is the maximum deflexion in millimetres due to the self weight and other permanent
loads.
Note that pre-cambering will not affect the natural frequency of the beam because it does
not affect the stiffness of the element.
The above gives a general over-view of the prediction of the acceleration and fundamental
frequency of floors. These calculations are most commonly applied to steel-concrete
composite floors, but the general principles apply to other types of construction, such as
floors of pre-cast concrete or timber, and floors in reinforced concrete, provided the
appropriate damping factors are used.
Note: See also AD253

SCI ADVISORY DESK

AD 255: Design of Asymmetric Cellular Beams:

Application limits of SCI Publication P-100
Introduction
Cellular beams have become the vogue for many long span applications both in the UK, and
increasingly in France. Often cellular beams are designed in buildings with strict height
limitations such as in renovation applications, or in the French experience, to compete
directly with concrete flat slab construction. To satisfy both serviceability requirements and
the need to install reasonably large service ducts, cellular beams are increasingly designed
with highly asymmetric sections and large openings off-centre with respect of the beam
depth. The size of the bottom Tee can be increased disproportionately to the top Tee to
achieve the required natural frequency limit.
Consequently, in practice, there is a
considerable reserve in bending and shear resistance for these long span beams.
From the structural design point of view, these new designs of cellular beams have moved
considerably from those envisaged in the SCI publication P 100. SCI is working with
European partners in an ECSC project to prepare application rules for composite beams with
rectangular and circular openings consistent with Eurocode 4. SCI P 100 is used as the basis
of design, but it is also necessary to extend the application to cover a wider range of design
cases, one of which is the use of highly asymmetric composite cellular beams. Furthermore,
both Eurocode 3 and BS 5950 Part 1: 2000 now increase the design shear resistance that
can be developed in the Tees by at least 10% (due to the increase of the 0.9 factor on shear
area to 1.0), and so it is necessary to have more accurate methods of analysis consistent
with the higher stresses permitted by these new Codes.

Design Model
Asymmetry in the cross-section of cellular beams can arise from both the relative area of the
Tees and from their position in the beam depth. Internal forces in the web post depend on
the rate of increase of axial forces in the Tees and on the differential shear forces resisted by
the Tees. The equilibrium of the top and bottom Tees in composite beams is illustrated in
Figure 1. Ideally, the shear forces resisted by the Tees (including the concrete slab) should
be similar. For severely asymmetric sections, redistribution of shear forces between the Tees
causes an in-plane moment in the web post, and the ability to resist this moment depends on
the width of the web post.
Equilibrium is established in the Tees where the in-plane moment acting on the web post is
determined according to the following equation:
Mh
where Vb
Vt
Vc
Vh
so
e
xt

= (Vb Vt Vc) so/2 Vh e + C (Ds,eff + xt)/2

is the shear force in the bottom Tee
is the shear force in the top Tee
is the shear force in the concrete
is the horizontal shear force in the web post (=T)
is the centre to centre spacing of the openings
is the eccentricity of the centre-line of the opening above the midheight of the beam
is the distance of the centroid of the top Tee from the top of the
beam

SCI ADVISORY DESK

Ds,eff
C
T

is the effective depth of the slab

is the increase in compression force in the slab due to the shear
connectors in length, so.
is the increase in tensile force in the bottom Tee in length, so.

As all parameters, except Vt and Vb are known, the relative magnitude of the shear forces in
the Tees can be selected to minimise Mh. If there is a reserve in shear resistance, Mh can be
set to zero. If not, this moment can be resisted by the web-post, based on its elastic
bending resistance at its narrowest section. The effect of horizontal shear may be taken into
account by retaining a factor of 0.9 on the shear area of the web, as in BS 5950 Part 1:
1990 (BS 5950 Part 1: 2000 increased this factor from 0.9 to 1.0 for plates).
For web-post buckling, the effect of in-plane moment moves the point of zero bending
towards the weaker Tee. The moment acting on a critical plane of 0.45 do above the centreline of the openings is determined according to:

and

Mwb
Mwb
do

= 0.45 do Vh Mh for the top Tee

= 0.45 do Vh + Mh for the bottom Tee
is the diameter of the opening

Web-post buckling is checked using Mwb as in SCI P 100. Therefore, the effect of Mh is
beneficial when the bottom Tee is heavier than the top Tee, which is normally the case.

Simplified Rules
Clearly, it is difficult to make a hand calculation for these effects, and so in design, it is best
to minimise Mh. This can be achieved by one of three methods:

design for a Utilisation Factor in vertical shear of less than 0.7, which allows the weaker
Tee to resist most of the applied shear force

keep within a reasonable degree of asymmetry of the Tees, typically less than 2:1 in
terms of web area of the Tees

increase the width of the web post to more than half of the opening diameter so that the
web post can resist the moment Mh easily.

The following table has been determined to assist the designer in choosing the maximum
degree of asymmetry in web area for scheme design, based on a cell spacing to diameter
ratio of not less than 1.5. In all cases, the minimum depth of Tee is 0.1 beam depth.
Utilisation Factor for Vertical Shear

1.0

0.9

0.8

0.7

0.6

Maximum Ratio of Web Areas of Tees

1.2

1.5

2.0

2.5

3.0

Table 1. Maximum Asymmetry of Tees in terms of Web Area

Design outside these limits can be verified by the forthcoming issue of Version 4 of the
Cellbeam software which SCI has prepared for Westok Structural Services.

Future Testing and Development

SCI is working closely with RWTH Aachen (Germany), Arbed Recherches (Luxembourg),
CTICM (France), and the University of Lulea (Sweden) to devise a range of full-scale

SCI ADVISORY DESK

structural tests on composite beams with web openings of various types to provide
performance data for:

highly asymmetric beams

slender (Class 4 webs)

elongated openings

practical stiffening arrangements

fire engineering.

This work is needed in order to develop application rules in the wider European context,
where experience is less than in the UK.

Vc

Vc
Vt

Vt

Mh
0.5d o

Vb

T - C

Mwp= 0.45d oV h -M h

0.5d o
V

C= NPd

Vh
Vh = T
Mwp= 0.45d oV h +M h
Vb
T

Fig. 1: Behaviour of composite asymmetric cellular beams.