Advances in Steel Structures - part 3 pdf

xx Contents
Analysis
Determination of Section Properties of Complicated Structural Members
Z.X. Li, J.M. Ko, T.H.T. Chan and Y.Q. Ni
Adaptive Finite Element Buckling Analysis of Folded Plate Structures
C.K. Choi and M.K. Song
Hoop Stress Reduction by Using Reinforced Rivets in Steel Structures
K.T. Chau, S.L. Chan and X.X. Wei
1109
1117
1125
Safety Analysis and Design Consideration for Oil and Gas Pipelines
A.N. Kumar
1133
Prediction of Residual Stresses: Comparison Between Experimental and Numerical
Results
Y. Vincent, J.F. Jullien and V. Cano
1141
Soil Structure Interaction
Composite Foundation of Deep Mixing Piles for Large Steel Oil Tanks on Soft Ground
X. Xie, X. Zhu and Q. Pan
An Analytical Study on Seismic Response of Steel Bridge Piers Considering Soil-
Structure Interaction
A. Kasai and T. Usami
1151
1157
Late Papers
Modelling Hysteresis Loops of Composite Joints Using Neural Networks
J.Y. Wang, Y.L. Wong and S.L. Chan
New Design Methods for Concrete Filled Steel Tubular Columns
Y.C. Wang

Keynote Paper
The Implications of the Information Society on the Practice of and Training for
Steelwork Construction
G. Owens
1167
1175
1187
Index of Contributors
II
Keyword Index 15
Keynote Papers
This Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left Blank
UNBRACED COMPOSITE FRAMES: APPLICATION OF THE
WIND MOMENT METHOD
D A Nethercot 1 and J S Hensman 2
ISchool of Civil Engineering, University of Nottingham, University Park,
Nottingham NG7 2RD, UK
2Caunton Engineering Limited, Moorgreen Industrial Park, Moorgreen,
Nottingham NG16 3QU, UK
ABSTRACT
Proposals are given to extend the simplified design technique known as the Wind Moment Method
to cover a limited range of composite frames. The range represents that of most interest in practice
in the UK. Justification is by comparison with the findings from an extensive numerical study.
KEYWORDS :
Composite Construction, Connections, Frames, Joints, Steel Structures, Structural
Design
INTRODUCTION
The Wind Moment Method (WMM) has long been established as a simple, intuitively based, design
approach for unbraced frames. More recently, it has been the subject of scientific study, designed to
provide a more fundamental understanding of the link between actual frame behaviour and the

inherent design simplifications. This work has, until now, been restricted to bare steel construction.
In a recent study Hensman, (1998), the authors have examined the case for an extension of the
WMM to cover composite steel/concrete frames. Although the approach adopted resembles that
used for bare steelwork, a number of particular features have had to be addressed. This paper
summarizes the main outcomes from that study.
The basis for the extension was numerical modelling, utilising the available body of knowledge on
the performance of composite connections, the previous application of the WMM to bare steelwork
and the capabilities of the ABAQUS package. It was also found necessary to conduct a detailed
examination of the role of column bases - a feature not previously addressed in WMM
D.A. Nethercot and J.S. Hensman
investigations. Several of the findings therefore have relevance to potential improvements in the
WMM for bare steel frames. This paper covers: appraisal of the basic source data, outline of the
numerical studies, presentation of the key findings and an indication of the resulting design
approach. This last item will be presented in a fashion suitable for direct use by designers in a
forthcoming Steel Construction Institute Design Guide.
KEY FEATURES OF THE WIND MOMENT METHOD
The approach was originally devised in the pre-computer era, when overall structural analysis of
unbraced frames represented an extremely challenging and potentially tedious task. It therefore
sought an acceptable simplification so that the labour involved in the structural analysis might be
minimised. This was achieved by recognising that some simplification in the representation of the
actual behaviour would be necessary. Although it is now quite widely accepted that the true
behaviour of all practical forms of beam to column connection in steel and concrete construction
function in a semi-rigid and partial strength fashion- with the ideals of pinned and rigid only
occasionally being approached- early methods of structural analysis could only cater for one or
other of these ideals. Thus the basic WMM uses the principle of superposition to combine the
internal moments and forces obtained from a gravity load analysis that assumes all beams to be
simply supported and a wind load analysis that assumes beam to column connections to be rigid
with points of contraflexure at the mid-span of the beams and the mid-height of the columns as
illustrated in Figure 1. This second assumption permits use of the so-called portal method of frame
analysis.

Once it became possible to conduct full range analyses of steel frames allowing for material and
geometrical non-linear effects and including realistic models of joint behaviour, studies were
undertaken to assess the actual performance of frames designed according to the WMM principles.
The findings permitted observations to be made of the two key behavioural measures:
9 That the load factor at ultimate was satisfactory
9 That drift limits at serviceability were achieved.
This second point is of importance because, when estimating sway deflections at working load, the
WMM normally involves taking the results of an analysis that assumes rigid connections and then
applying a suitable scaling factor. Important contributions in the area of bare steel construction are
those of Ackroyd and Gerstle, (1982), Ackroyd, (1987), and Anderson and his co-workers at
Warwick, Reading, (1989), Kavianpour, (1990), Anderson, Reading and Kavianpour,(1991)
NUMERICAL APPROACH
All the numerical work was undertaken using the ABAQUS package. Whilst this contained
sufficient functionality to cover many of the necessary behavioural features, three items required
particular attention:
9 Representation of the composite beams
9 Representation of composite beam to column connections
9 Inclusion of column base effects
4,4' 4'4, ,1,4, 4'4' 4'4' 4'4' 4'4' 4'4'4'
~4' 4'4' 4'4' 4'4' 4'4' 4'4' 4'4'4'
(a)
t
7" r 77
t t
Unbraced Composite Frames: Application of the Wind Moment Method
Figure 1 Superposition of gravity and lateral load analyses
For the first of these the approach previously utilised by Ye, Nethercot and Li, (1996), that is based
on moment curvature relationships developed by Li, Nethercot and Choo, (1993), was employed.
Since composite endplates were assumed for the beam to column connections, the work of Ahmed
and Nethercot,(1997), in predicting moment-rotation response under hogging moment was directly

employed.
Data on the performance of composite beam to column connections under sagging i.e. opening,
moments was, however, almost non-existent. Previous experience with the Wind Moment Method
had, however, suggested that reversal in the sign of the rotation at any connection might be a rather
unusual event. An approximate model for composite connection behaviour under sagging moments
was therefore devised by examining test data for such connections when subject to cyclic loading.
All previous studies of the WMM have assumed rigid i.e. fully fixed column bases. Enquiries
among practitioners had, however, already revealed that such an option was not attractive. In
addition, there was a widely held belief that all practical forms of "pin" column bases were capable
of supplying quite significant amounts of rotational restraint. Accordingly, all relevant information
on column base effects - particularly previous experimental studies - was carefully reviewed in an
attempt to identify suitable minimum restraint levels likely to be supplied by notionally pinned
bases, Hensman and Nethercot, (2000a). The findings were then incorporated in the full parametric
study. This point is regarded as particularly important as attempts to justify the WMM approach
using truly pinned column bases, Hensman,(1998), had shown that it was almost impossible to
satisfy realistic drift limitations due to the greatly enhanced overall frame flexibility resulting from
the loss of column base restraint (as compared with the usual WMM assumption of fixed bases). It
is believed that the exercise should be repeated- since bare steel columns were assumed
throughout, it would merely be a case of conducting appropriate analyses on bare steel frames - as a
way of similarly relaxing an unattractive restriction in the application of the WMM to bare steel
construction.
6 D.A. Nethercot and J.S. Hensman
Because of concem over the adequacy of the modelling of composite beam to column connections
under sagging moments, particular attention was paid in an initial study, Hensman, J S (1998), to
the occurrence (or not) of reversal in the sign of the connection rotations. Initial studies using the
sub-frame of Figure 2, that was specially configured to represent a typical intermediate floor in a
more extensive structure, showed that for realistic arrangements of frame layout, member sizes and
levels of gravity and wind loading reversal of rotations, even at the potentially most vulnerable
windward connections was extremely unlikely. It was therefore concluded that the full parametric
study need not concern itself with further refinement of this feature.

PARAMETRIC STUDY
Figure 3 illustrates the basic frame layouts considered and Tables 1 and 2 list the range of variables
considered within the numerical study. Although this was based on the equivalent set of restrictions
given in Anderson, Reading and Kavianpour (1991) it has been adapted somewhat, both to
recognise important differences between bare steel and composite construction e.g. the likely use of
longer span beams, and to reflect certain preferences from the industry and recent changes in the
UK design environment e.g. issue of a new Code for wind loading. A more detailed explanation of
the arrangement of the study, including justification for decisions on joint types, load combinations
etc., is available, Hensman and Nethercot (2000b). Full details of the 300 cases investigated
covering 45 different frame arrangements, including summary results for each, are available in
reference 1. In all cases the approach adopted was to first design the frame using the proposed
WMM technique and then to conduct a full range computer analysis to check its condition at the
SLS and ULS stages.
MAIN FINDINGS
Undoubtedly the most significant overall outcome of the parametric study was the finding that
every frame design using the proposed WMM approach was essentially satisfactory in terms of
providing an adequate margin of safety against ULS load combinations. This was despite the fact
that the actual distributions of intemal forces and moments within the frames often differed
significantly from those presumed by the WMM analyses. Only in an extremely small number of
cases was any degree of column overstress observed (and then less than 4%) - a comforting feature
given that actual end restraint moments obtained from the rigorous analyses were often significantly
higher than the assumed 10% of the WMM. The actual values of up to 30% in certain cases might
suggest that where gravity loads are high beam sections could be reduced by assuming a larger-say
20% - end restraint moment. Before so doing, however, it would be important to check the effect
on overall lateral frame stiffness as it might well prove difficult to satisfy drift limitations with this
inherently more flexible system.
For all cases of frames designed for maximum gravity load and minimum wind load the SLS
conditions were met. However, if higher wind loads were introduced, particularly for frames with
short bay widths, some difficulty in ensuring adequate serviceability performance might well be
experienced.

Unbraced Composite Frames." Application of the Wind Moment Method 7
A general discussion on the findings from the numerical study in terms of possible future
modifications to the WMM and links between flame features and observed behaviour is available in
Hensman and Nethercot (2000b).
Figure 2: Typical subframe arrangement used for preliminary study
(Beam spans vary between 6m and 12m)
8 D.A. Nethercot and J.S. Hensman
Figure 3 9 Schematic diagram of alternative flame layouts used in parametric study
Unbraced Composite Frames: Application of the Wind Moment Method
TABLE 1
RANGE OF VARIABLES CONSIDERED WITHIN THE PARAMETRIC STUDY
Minimum Maximum
Number of storeys 2 4
Number of bays 2 4"1
Bay width (m) 6.0 12.0
Bottom storey height (m) 4.5 6.0
Storey height elsewhere (m) 3.5 5.0
Dead load on floors (kN/m 2) 3.50 5.00
Imposed load on floors (kN/m 2) 4.00 7.50
Dead load on roof (kN/m 2) 3.75 3.75
Imposed load on roof (kN/m 2) 1.50 1.50
Wind loads (kN)
10 .2 40 *2
*' frames can have more than 4 bays, but a core of 4 bays is the maximum
that can be considered to resist the applied wind load.
,2 Wind loads = concentrated point load on plane frame at each floor level
TABLE 1
RELATIVE DIMENSIONS CONSIDERED WITHIN THE PARAMETRIC STUDY
Bay width: storey height
(bottom storey)

Bay width: storey height
(above bottom storey)
Greatest bay width:
Smallest bay width
Minimum Maximum
1.33 2.67
1.33 3.43
1 1.5
RECOMMENDED DESIGN APPROACH
The basic design approach is outlined in the chart of Figure 4. This presents all the relevant steps,
including those intended to identify arrangements for which the WMM is not suitable. Some key
details for certain of the steps in the actual design procedure are discussed below.
Once an initial frame arrangement has been decided upon, global analyses for the three load
combinations:
9 1.4DE + 1.6IL + Notional Horizontal Forces
9 1.2(DL+IL+WL)
9 1.4 (DE+WE)
should be undertaken. Notional horizontal forces should be taken as 0.5% of the factored dead +
imposed load as specified by BS5950: Part 1. Pattern loading should be considered; it may well be