Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Abstract
The paper presents the background to the development of the provisions of PD 6695-2:2008
Recommendations for the design of bridges to BS EN 1993. That Published Document was
prepared with the objectives of providing information on topics not covered by BS EN 1993-2
and offering guidance where it was considered further explanation of the Eurocode provisions
was desirable for their correct and consistent application. It explains that the main sources of
this material were BS 5400-3 and Designers’ Guide to EN 1993-2, Eurocode 3: Design of
steel structures. Part 2: Steel bridges.
Introduction
The objective of this paper is to give the background to the development of the provisions of
PD 6695-2:2008 Recommendations for the design of bridges to BS EN 1993. The Published
Document was prepared by B525/10 Working Group 3 and is referred to in the National
Annex to BS EN 1993-2 as a source of NCCI. The PD was written with two primary
objectives in mind:
(i) Provision of information on topics not covered by EN 1993.
(ii) Provision of guidance where it was considered further explanation of the Eurocode
provisions was desirable for their correct and consistent application.
The first objective was the subject of debate during drafting because the principle-based
approach used in the Eurocodes, together with the wide range of analysis techniques
permitted, combine to ensure that it is usually possible to design all elements of a bridge
utilising the Eurocode methodologies without further information, if a sufficiently powerful
analysis model is used. However, the drafters of the PD considered it undesirable to require
an increase in the level of complexity of analysis over and above that used in previous
practice, although the flexibility to permit such analysis was considered to be beneficial. The
material included to meet (i) is therefore usually in the form of design rules that can be
applied by hand methods of calculation with a similar level of complexity as required by
previous practice to BS5400. Much of the PD draws on recommendations in BS 5400-3,
where they are non contradictory, as they are perceived by many to be more user friendly than
Eurocodes. Clarification of Eurocode provisions also draws on guidance in Reference 2.
The paper deals with each principal PD clause in turn and provides explanation for the
recommendations given. References to clauses in EN 1993-2 have been abbreviated below.
For example, 3-2/6.3.4.2(5) is a reference to clause 6.3.4.2(5) of EN 1993-2. The PD clause
C Hendy, D Iles, S Chakrabarti page 2
numbers are given in parenthesis in each heading. It should be noted that the clause
numbering in the PD does not follow that in EN 1993-2.
The PD requires that bolt slip should be taken into account at the connections of bracing
systems because a sudden loss of stiffness arising from bolt slip leads to an increase in
deflection of the main member and an increased force on the bracing member, which could
lead to overall failure. Ideally, connections of bracing members should be designed as non-
slip at ULS (Category C to EN 1993-1-8) to avoid the need to evaluate the effects of slip.
The PD permits bolt slip at main beam splices to be ignored in global analysis. It has been
UK practice to design splice bolts to slip at ULS (Category B to EN 1993-1-8) without
consideration of slip in global analysis. This is justifiable as, although slip could alter the
moment distribution in the beam, splices are usually positioned near to points of contraflexure
and therefore slip will not shed significant moment to either adjacent hog or sag zones. Also,
the loading that gives maximum moment at the splice will not be fully coexistent with that for
either the maximum hogging moment or maximum sagging moment in adjacent regions.
Corresponding recommendations are given in clause 19 of the PD for the design of the
connections.
The PD also clarifies how to treat forces from imperfections in bracing systems comprising
both torsional restraints and plan bracing. Where the restraint forces are to be transmitted to
end supports by a system of plan bracing, the plan bracing system should be designed to resist
the more onerous of the forces FEd from each restraint within a length equal to the half
wavelength of buckling and the forces generated by an overall flange bow in each flange
according to clause 5.3.3 of EN 1993-1-1. In the latter case, for a very stiff bracing system
N Ed
with zero first order transverse deflection, each flange applies a total force of m
62.5
uniformly distributed to the plan bracing, where m is the reduction factor for the number of
interconnected beams in BS EN 1993-1-1 clause 5.3.3(1).
To avoid the need for computer analysis, it was decided to provide empirical rules in the PD
that enable calculation of the slenderness for lateral torsional buckling without explicit
calculation of M cr . The rules have been adapted from those in BS 5400-3 and take advantage
of the fact that for a Class 1 or 2 cross section
fy
LT LT
π2E
C Hendy, D Iles, S Chakrabarti page 4
There is a semi-empirical approach to the determination of slenderness based on the use of the
general method in 3-2/6.3.4.2. See further discussion below in relation to clause 9 of the PD.
The general method cannot be used for cases where only torsional bracing is provided (e.g.
paired beams during construction of the deck slab); in such cases, either the rules in PD clause
8 should be followed or Mcr determined from elastic critical buckling analysis.
The simplified method in 3-2/6.3.4.2 is intended for use for beams where one flange is held in
position laterally. The method is based on representing lateral torsional buckling (actually
lateral distortional buckling, since one flange is assumed to be held in position) by lateral
buckling of the compression flange. The method is primarily intended for U-frame type
bridges but can be used for other flexible bracing systems as well. It can also be applied to
lengths of girder compression flange between rigid restraints, as found in hogging zones in
steel and concrete composite construction. For this method, the St Venant torsional stiffness
of the beam is ignored. This simplification may be significantly conservative for shallow
rolled steel sections but is generally not significant for most fabricated bridge girders.
The PD includes guidance on two aspects not covered by EN 1993-2. First, the expressions
for U-frame stiffness in 3-2/Annex D do not contain a contribution from the flexibility of the
joints between U-frame members. The PD offers suggested values of joint flexibility, in the
absences of specific calculation. The values have been imported from BS 5400-3 (and are
based on research by British Railways in the 1960s) and are acknowledged to be fairly
conservative. Second, the expression for Ncr = mNE in 3-2/6.3.4.2(6) requires there to be rigid
bracings at supports. In conventional U-frame decks, the end U-frames are normally not
sufficiently stiff to be classed as rigid and hence an alternative expression for Ncr is required.
The PD therefore provides a modified expression for m in this case. The expression has been
C Hendy, D Iles, S Chakrabarti page 5
derived from coefficients provided in BS 5400-3 (which were derived from consideration of
beam on elastic foundation theory) which fulfilled the same purpose.
The remainder of clause 11 treats the additional forces generated in U-frame bridges
(including the flanges) by local loading on the cross girders. Loading on a transverse member
will cause that transverse member to deflect and rotate at its connection to the vertical
stiffener. The stiffener will therefore try to deflect inwards. If all cross girders are not loaded
similarly, the tendency is to produce differential deflections at the tops of the stiffeners but
this differential deflection is resisted by the flanges in transverse bending. A transverse force
is therefore generated at the top of the stiffener and a moment My is produced in the flange.
Simplified expressions for the force, Fc, and moment, My, are provided in the PD. They have
been imported from BS 5400-3 and further background on their origins can be found in
reference 3.
The method provided in the PD is simple to carry out but can give conservative results.
Second order analysis carried out on a suitable 3-D model will produce more accurate results.
Further guidance and background on this approach is given in reference 2.
C Hendy, D Iles, S Chakrabarti page 6
The method in clause 12.2 is similar to that proposed in EN 1993-1-7 but ensures
compatibility with EN 1993-1-5 which EN 1993-1-7 does not achieve. EN 1993-2 does not
reference EN 1993-1-7.
The UK‟s decision on the use of 3-1-5/4 and 3-1-5/10 is discussed in the paper on the NA to
EN 1993-1-5.
Generally the effective section method will be the more economic and thus the use of clause
12.2 will generally be the more economic when transverse load is present.
The reduced stress method provided in clause 12.3 has some limitations, notably that the
minimum load amplifier cr (in 3-1-5/10) can legitimately be less than 1.0 due to the post-
buckling reserve of plates, whereupon the amplification factor, ( 1 /(1 1 / cr ) ) in PD clause
12.3 will become negative and the expressions become invalid. They will in any case become
very conservative as cr approaches 1.0. This method is therefore not recommended unless
the criteria for the use of 3-1-5/4 cannot be met.
Beams with vertical curvature develop out of plane bending moments in the flanges as shown
in Figure 1. For I beams, this flange transverse bending is sometimes referred to as “flange
curling”. PD 6695-2 provides methods for calculating the stresses from curvature (similar to
the rules in BS 5400-3) and then for combining them with other effects and verifying the
section; reference is made back to section 12 for the latter purpose. For beams with
longitudinal flange stiffeners, the main out of plane bending effect in the flanges is
longitudinally between transverse restraints. Two options are provided in the PD to account
for this. Either the effects from curvature can be represented by a transverse load or the
curvature can be modelled as in increased imperfection in the stiffener which is included via
the term . The expressions originate from reference 2.
C Hendy, D Iles, S Chakrabarti page 7
Ff
PT Ff
a) Radial force from curved flange
c b
To rectify the latter omission, the PD imports the list of effects listed for transverse stiffeners
in BS 5400-3. It should be noted that this may not be an exhaustive list for all situations,
which was the main reason for not including a list in EN 1993-1-5. For the strength and
stiffness design, it had been intended to include formulae in the PD for calculating the
deflections and stresses incorporating second order effects; paragraph a) in clause 15 of the
PD refers to this intention. These formulae however were inadvertently omitted from the
published version of the PD. They can be found in reference 2.
As with clause 15, for strength and stiffness design, it had been intended to include formulae
for calculating the deflections and stresses incorporating second order effects but they also
have inadvertently been omitted from the published version of the PD. They can be found in
reference 2.
The specification of the Von Mises yield criterion in PD clause 17.4.2 as the means of
verifying spliced plates and cover plates is also made for compatibility with existing practice
to BS 5400-3. The PD recommedation is the same as the yield criterion is referred to in 3-1-
1/6.2.1 as a general verification where no other rule is givenI.
For flat stiffeners only, it is possible to determine a simple shape limitation by following the
h fy
EN 1993-1-5 rules, so this has been provided, namely that s 12.9 .
t s 235
It is likely that FE models using 2D shell elements will be used by designers to determine
stresses for use in verifications. However, elastic FE analysis merges the longitudinal stresses
due to axial force, bending (as modified by shear lag), torsional warping and distortional
warping. When evolving a design, knowledge of the relative magnitudes of the separate
components is useful because certain effects may be neglected at the ultimate limit state due
to the occurrence of plasticity (effects such as torsional warping and much of the shear lag
effects). The algebraic expressions provided in the PD allow these effects to be separated and
are imported directly from BS 5400-3. They also provide a useful check on the output from
FE analysis. A further alternative would be the use of non-linear FE analysis with modelled
imperfections. This could automatically account for the effects of plasticity but is a much
more involved and laborious process.
No attempt has been made to make the rules on diaphragms compatible with the Eurocode
buckling curves and analysis methods; it was simply too great a task. Instead, the method has
been described as “a self-contained procedure for hand methods”. This has been justified on
the basis that diaphragms are not explicitly covered by EN 1993.
Acknowledgements
The work described in this paper was carried out by the authors with the support of and
review by Working Group WG3 of the BSI Committee B525/10. Thanks are expressed to all
contributors.
References
[1] BS 5400-2000 Design of steel bridges. British Standards Institution, London.
[2] Hendy C.R. and Murphy C.J. (2007) Designers’ Guide to EN1993-2, Eurocode 3:
Design of steel structures. Part 2: Steel bridges. Thomas Telford, London. ISBN
9780727731609
[3] Brown, C.W and Iles, D.C. Commentary on BS 5400-3: 2000 - Code of Practice for the
design of Steel Bridges (P295), SCI, 2000, England.