Nutzungsbedingungen
Die ETH-Bibliothek ist Anbieterin der digitalisierten Zeitschriften. Sie besitzt keine Urheberrechte an
den Inhalten der Zeitschriften. Die Rechte liegen in der Regel bei den Herausgebern.
Die auf der Plattform e-periodica veröffentlichten Dokumente stehen für nicht-kommerzielle Zwecke in
Lehre und Forschung sowie für die private Nutzung frei zur Verfügung. Einzelne Dateien oder
Ausdrucke aus diesem Angebot können zusammen mit diesen Nutzungsbedingungen und den
korrekten Herkunftsbezeichnungen weitergegeben werden.
Das Veröffentlichen von Bildern in Print- und Online-Publikationen ist nur mit vorheriger Genehmigung
der Rechteinhaber erlaubt. Die systematische Speicherung von Teilen des elektronischen Angebots
auf anderen Servern bedarf ebenfalls des schriftlichen Einverständnisses der Rechteinhaber.
Haftungsausschluss
Alle Angaben erfolgen ohne Gewähr für Vollständigkeit oder Richtigkeit. Es wird keine Haftung
übernommen für Schäden durch die Verwendung von Informationen aus diesem Online-Angebot oder
durch das Fehlen von Informationen. Dies gilt auch für Inhalte Dritter, die über dieses Angebot
zugänglich sind.
http://www.e-periodica.ch
Composite Construction - CE1S
Conventional and Innovative
CONFERENCE REPORT
A
cecm
EKB JP
mm
PG3
nilem
International Conference
Innsbruck, Austria September 16-18,1997
Conference Secretariat
Composite Construction
c/o IABSE Phone +41-1-633 2647
ETH-Hönggerberg Fax +41-1-371 2131
CH-8093 Zurich e-mail secretariat@iabse.ethz.ch
Switzerland web http://www.iabse.ethz.ch
3
Foreword
The 1997 Innsbruck Conference is a joint effort of the main international professional
societies in the field of structural engineering, and organised on their behalf by the
International Association for Bridge and Structural Engineering (IABSE). The enthusiastic
response to the call for papers (430 abstracts submitted) indicates the high level of
interest in the subject.
The present volume contains the 180 papers accepted for publication by the Scientific
Committee. This publication provides an excellent overview of the many aspects of
composite construction, ranging from combinations of traditional building materials to more
unconventional structural materials like fibre-reinforced plastics. The tone for the
Conference is set by the Keynote Lectures presented at the four plenary sessions, included
in these proceedings, while the latest developments are given in the papers and
posters presented in parallel working sessions.
It is the hope of those involved in organising this event that the Innsbruck Conference
will promote increased collaboration among professional engineering societies, and also
lead to a more problem-oriented, less materials-oriented approach to the art of structural
engineering.
Foreword
Plenary Sessions
D. MICHAEL
Concepts of Composite Construction - Mutatis Mutandis
R. BRIDGE, M. PATRICK, J. WEBB
High Strength Materials in Composite Construction
G. HANSWILLE
Outstanding Composite Structures for Buildings
J.-B. SCHLEICH
Slim Floor Construction: Why?
J.-M. CREMER
Variety of Composite Bridge Construction
F.M. MAZZOLANI
Composite Building Structures in Earthquake Engineering
J. MARTINEZ CALZON
Rehabilitation and Repair of Structures Using Composite Systems
U. MEIER
Repair Using Advanced Composites
Working Sessions
Connection between Materials
M.A. ASTIZ
Composite Construction in Cable-Stayed Bridge Towers
D. KRAUS, O. WURZER
Bearing Capacity of Concrete Dowels
N. GATTESCO
Fatigue in Stud Shear Connectors
G. TAPLIN, P. GRUNDY
Incremental Slip of Stud Shear Connectors under Repeated Loading
5
C. KALFAS, P. PAVLIDIS
Load-Slip Curve of Shear Connectors Evaluated by FEM Analysis 151
S. BULLO, R. Dl MARCO
The Load-Bearing Capacity of Steel-HPC Composite Beams 157
J. NATTERER
Concepts and Details of Mixed Timber-Concrete Structures 175
C. AMADIO, M. FRAGIACOMO
Evaluation of Long-Term Effects in the Steel-Concrete Composite Beams 211
M. LYDMAN
Deformations of Composite Precast Concrete Slabs Subject to Creep and Shrinkage 217
J. SUMEC, N. JENDZELOVSKY
Stress-Strain Distribution in the Contact Surface of a Twö-Layered RC Structural Element 247
J. RONDAL, A. MOUTAFIDOU
Study of Shear Bond in Steel Composite Slabs 259
H. BODE, F. MINAS
Composite Slabs with and without End Anchorage under Static and Dynamic Loading 265
B. UY
Slenderness Limits for Thin-Walled Steel Concrete Filled Box Columns 277
Z. MARCINKOWSKI, P. BERKOWSKI
Experimental Verification of Bearing Capacity of Composite Truss Girders 289
A. DALL'ASTA, L. DEZI
Construction Sequence Effects on Externally Prestressed Composite Girders 301
Structural Design
J.T. MOTTRAM
Design Guidance for Joints Using Polymeric Composite Materials 313
M. O'SHEA, R. BRIDGE
The Design for Local Buckling of Concrete Filled Steel Tubes 319
H. KOUKKARI
Design Basics of a Continuous Composite Slab with Unbonded Tendons 325
M.V. LESKELÄ
Composite Decking Unit of Thin-Walled Z Purlins and Thin Concrete Slab 331
P. HASSINEN, L. MARTIKAINEN
Design of Continuous Lightweight Structural Sandwich Panels 337
7
P.J. WRIGHT
-
Slimdek Development of an integrated Floor System 343
N. UCHIDA, H. TOHKI
Design of High-Rise Building using Round Tubular Steel Composite Columns 373
Fire Resistance
L.-H. HAN
New Developments in Fire Resistance of Concrete Filled Steel Tubes in China 391
M. FONTANA, A. FRANGI
Fire Resistance of Timber-Concrete Composite Slabs 403
D. NETHERCOT
Behaviour and Design of Composite Connections 421
F. TSCHEMMERNEGG
Connection of Floor Systems to Columns - Conventional and Advanced 445
G. COUCHMAN, M. LAWSON
The Use of Composite Connections in Practice 451
T. FUKUMOTO, Y. SAWAMOTO
CFT Beam-Column Connection with High Strength Materials 463
A. SARJA
Composite Floors of the Buildings 489
Y. YOSHIDA, T. UEDA
Composite Steel and Concrete Pier Using Durable Precast Form 501
J. FINK
Truss Composite Bridges 507
G. YAN
Composite Arch Bridges Developed in China 513
J. KOZAK, P. DISTLER
Steel-Concrete Structures in 24 Storey Bank Building in Bratislava 519
F. WEARE
Long Span Office Construction Using Composite Cellular Beams 525
B. MONOV, V. SELIVERSTOV
Erection of Composite Bridges with Precast Deck Slabs 531
R. SAUL
Erection Methods for Long Span Steel Composite Bridges 537
9
A. KILPATRICK, T. TAYLOR
Application of Eurocode 4 Design Provisions to High Strength Composite Columns 561
R.P. JOHNSON
Shear Connection for Composite Bridges, and Eurocode 4: Part 2 573
H. POMMER
Composite Bridges in View of Existing Standards and Eurocode 585
J. GLATZL
Composite Bridges for High Performance Lines in Austria -
Trial calculation according to ENV 1994-2 591
Earthquake Resistance
S.P. SCHNEIDER
Detailing Requirements for Concrete-Filled Steel Tubes Connections 609
O. BURSI, M. BALLERINI
Low-Cycle Behaviour and Analysis of Steel-Concrete Composite Substructures 615
T. YAMAMOTO, T. OHTAKI
Seismic Performance of a Composite Frame Structure 633
10
B. TSUJI
Ultimate Strength of Composite Structures Using Superposed Strength Method 639
V. STOIAN, I. OLARIU
Steel-Concrete Seismic Resistant Structure for Multi-Storey Buildings 645
X. GAO, F. ZHOU, X. XU
Dynamic Response of a High-Rise SRC Building Model 651
R.T. LEON
Seismic Design of Composite Semi-Continuous Frames 657
I. OKURA
Stud Arrangement to Reduce Fatigue Cracks and Application of Drilled Holes 663
H. HARTL
Elastic Composite Construction between Timber and Wood-Based Materials 687
P. UNTERHOLZNER, W. STIEBELLEHNER
Renewal of Bridge Superstructures under traffic at the A13 Brenner Motorway 711
Advanced Composites
W. STEINER
Strengthening of Structures with Carbon Fibre Laminates 723
T. TRIANTAFILLOU
Behaviour of Masonry Structures Strengthened with Composites 729
11
G. SCHWEGLER
Earthquake Resistance of Masonry Structures Strengthened with CFRP-Sheets 735
F. SEIBLE
Advanced Composites for Bridge Infrastructure Rehabilitation and Renewal 741
U. MEIER, H. MEIER
Evolution of Stay Cables through the Use of CFPR 747
H. FALKNER, V. HENKE
Potsdamer Platz: Steel Fibre Concrete for Underwater Concrete Slabs 771
Behaviour Modelling
A.N. GUZ
The Fracture Theory of Composite at Bearing Strain in End Faces 783
B. UY
Ductility and Strength of Thin-Walled Concrete Filled Box Columns 801
K. MAEGAWA
Ductility of Steel Tube-Reinforced Concrete Composite Beams 807
B. JONAITIS, V. PAPINIGIS
Crack Formation in Bending Composite Elements 813
M. PAJARI
Interaction between Hollow Core Slabs and Supporting Beams 831
12
P.G.F.J. VAN DER SANDEN, H.H. SNIJDER, J.W.B. STARK, H.W. BENNENK
A New Test for Stud Connectors in Ribbed Slabs 838
M.V. LESKELÄ
Connection Characteristics for Joints between Hollow Core Slabs and Slim Floor Beams 840
L. NASCH
Interlayer Bond Deterioration under Repeated Shear Load 842
E.K.R. WISSER
Connection of New and Old Concrete with Bonded Reinforcement Bars 844
A. PYE, S. LEDBETTER
A Design Method for Glass-Adhesive-Glass Composite Structural Elements 846
O. BERNARD, E. BRÜHWILER
Long Term Behaviour of Composite Concrete Structures 850
H. WENZEL
Health and Safety Monitoring of Composite Structures 852
F. SAJE, J. LOPATIC
Time-dependent Response of Composite Structures 854
H. ABE, A. NAKAJIMA
Study of Behaviour of Concrete Beams Strengthened by Steel Plates against Shearing Force 856
C.D. GOODE
Composite Cylinders Subjected to External Pressure 858
C.D. GOODE
The Safety of Composite Sub-Sea Structures 860
B. DERETIC-STOJANOVIC
Calculation of Stresses for Composite Structures 864
13
K. TAKAMI, S. HAMADA
Stochastic Long-Term Analysis of Composite Girders 866
Structural Design
E. AIRUMYAN, O. BOYKO
Design and Tests of New Steel-Concrete Slabs 868
J. STRASKY, I. HUSTY
Arch Bridge Crossing the Brno-Vienna Expressway 870
Fire Resistance
A. MALHOTRA
Approach to Assessment of Fire Damages: Composite Structures 876
H. SHAKIR-KHALIL
Connection of Steel Beams to Concrete-Filled Tubular Columns 878
H. TAKANO
Column-Pile Joints Made of Steel Pipes Filled with Concrete 880
J. BERTHELLEMY
The Antrenas Tubular Arch Bridge 888
P. VAN BOGAERT
Deformation-Compatibility of Steel Truss and High-Strength Concrete Girders 892
14
Y. INOKUMA
Planning of Steel Truss Web Prestressed Concrete Bridge 894
Earthquake Resistance
F. GASTAL
Collapse and Rehabilitation of Composite Trussed Structures 916
F. MAZZOLANI, E. MELE
Use of Aluminium Alloys in Retrofitting Ancient Suspension Bridges 918
P. MENETREY, E. BRÜHWILER
Shear Strengthening of Existing Reinforced Concrete Slabs: an Experimental Investigation 920
R. SILMAN
Major Repairs to Frank Lloyd Wright's Largest House: Wingspread, Racine, Wl, USA 922
15
Advanced Composites
O. JOH, A. KITANO
Shear Strengthening of RC Columns by Carbon Fiber Sheet 926
Y. XIAO
Seismic Retrofit of Concrete Columns Using Advanced Composite Materials 928
K. ONO, M. MATSUMURA
Strengthening of Reinforced Concrete Bridge Piers by Carbon Fiber Sheet 930
J. NATTERER, J. HAMM
Timber-Glass Composite in Structural Glazing 932
M. WERNLI, F. SEIBLE
Advanced Composite Stay Cables 934
Behaviour Modelling
U.A. MEIERHOFER
Long Term Performance of Timber Concrete Composite Structural Elements 938
M.V. LESKELÄ
Vertical Shear Resistance Models for a Deltabeam 940
(Plenary Session)
J.E. ROBERTS
Composite Construction in California Bridge Seismic Retrofitting 943
Plenary Sessions
Leere Seite
Blank page
Page vide
19
Summary
Composite construction encourages a focus on minutiae which becomes self reinforcing.
Composite construction can be seen as a system concept. If one reviews it on that basis, the huge
scope awaiting exploitation reveals itself.
Introduction
Construction and in particular structural design seem to stimulate the desire to extend the scope
of what can be analysed holistically or in a unified manner, what we might describe as the
analysable entity. In civil and structural engineering, this desire has often been focused on
construction made of a notionally single material, reinforced concrete or stone or steel.
Examples of the bigger or more complex include many of those bridges which are the proud
record of IABSE members. But they also include structures more modest in scale, such as the
shells of Candela or the membrane structures of Frei Otto or the glass structures of Tim
MacFarlane.
Examples of advances in technique include the development of limit state theory, matrix methods,
finite element analysis and Jacques Heyman's analyses of Gothic stone structures.
20 CONCEPTS OF COMPOSITE CONSTRUCTION - MUTATIS MUTANDIS
The Forth Railway Bridge and the Buckminster Fuller "tensegrity" structures both demonstrate
the power of articulation into tension and compression elements. The structural behaviour, the
system points and load paths, can be sensed and read off the completed structure.
In some cases the actual articulation is used to dissect the analysable entity out from its complex
surroundings. A recent case is seismic design work for a Californian building making use of base
isolation elements to separate the upper structure, which is then regarded as a virtual pendulum.
The intellectual restatement of the general process with which I am most familiar is what is called
total architecture or total design, as formulated by Ove Arup. This conception has itself been
interpreted and developed in use, with our growing understanding of the detailed implications of
design. It is always characterised by the creative tensions between synthesis and analysis,
between harmony and invention, between the established and the unknown and hopefrilly by
sufficient eventual reconciliation.
We should pair the topic of the analysable entity with the framework of time which gives us the
order to events that we call history. Civil engineering structures have usually been conceived as
one act, however large and even if their actual construction stretches over decades.
With building structures, however, it is a surprisingly recent phenomenon, probably less than 300
years old for major buildings again to be designed as a unity and then to be completed more or
less without change. Sir Christopher Wren was the first modern designer of a cathedral to see his
design built. Brunelleschi's Duomo was constructed on supports which had been designed and
built more than a century earlier and with little idea of how the space could be spanned. Many
may argue bitterly that we have again regressed and design in parts as the construction proceeds.
• we add elements for changes of use and we add repairing or strengthening elements to
cope with wear and ageing
• we use temporary works and falsework as a crucial element in the process or explore
whether to eliminate them. The design for the Kingsgate Footbridge in Durham is an
example of treating the temporary works by rotating the structures as integral to the total
concept.
Composite construction
There is nothing unusually special about composite construction. The composite concept is very
old.
The origins of the minerals and metals which we use in construction are in the ground in an
unhelpful mix of composites. We variously recover, refine, transform and recombine these
minerals into our composite materials, cements, plasters, concrete, brick, clay tiles, terra cotta,
irons and steels, aluminium alloys and glasses.
The natural organic building materials (straw, reeds, timber and so on) are in themselves,
composite materials. Natural structures offer us fascinating models for structural form and
environmental control through their material arrangements.
D. MICHAEL 21
One of man's earliest deliberately composite materials is probably the sun-dried, straw-bound
brick. Even then, there was some understanding of the criticality of quality control, the history
recording the warning about bricks without straw.
The use of iron cramps in the Acropolis stonework, of Victorian cast and wrought iron and of
timber roof trusses, right through to today's use of steel/concrete and advanced polymers, all
provide more recent examples of the use of the composite technique.
Today, we have started to combine materials to exploit and extend our modern understanding of
composites; steel with concrete, glass and carbon with polymer and so on.
Composite construction thus describes the combination of elements or materials in ways which
can be regarded as delivering a single analysable entity. These have also meant the synthesis of
discrete elements, extending the spatial extent of that analysable entity. The extraordinary
bridges by Maillart show us how we can extend the spatial scope of what we learn as elements
into one whole bridge. The work of Fazlar Khan gave the language for a major evolution in our
structural concepts of the tall building. A less obvious example is the long history of developing
the structural theory of the column.
The creature Pegasus was formed by combining the body of a horse with the wings of a bird.
Initially we comprehend the idea through the properties and qualities which the elements of
composition, the horse and the bird bring to the whole. Later, through use and in language,
narrative and recollection we come to treat the overall idea as a concept in itself and to realise
what is distinct and gives new meaning in the unified concept.
We find this in design. A motor-cycle is more than an engine plus a bike and has become a
distinct single concept. We know that e-mail is more than electrified correspondence.
So it is with composite construction. The most familiar example is reinforced concrete, which so
often we can usefully regard as homogenous, as a single isotropic material. The RC paradigm
has some of the original properties of the separate constituents but more importantly it has its
own qualities not possessed by any one constituent.
The rediscovery today of lime putty mortars adds to our options of cement mortars, from which
we see more clearly that bricks-and-mortar is a wide repertory of composite materials. This
counterpoint of new and old brickwork as distinctive materials is nicely demonstrated in two
buildings designed by Michael Hopkins. Glyndebourne Opera House has lime putty mortar to
eliminate movement joints. The Inland Revenue Centre has cement mortar, so that the brickwork
could be built in the factory and then be transported and erected on site as precast elements.
In the vistaof composite construction, the most interesting issues are those made possible by the
new understandings and new possibilities of the composites, those aspects which are not
properties of the separate parts. These are our contribution to the Pegasus paradigm.
22 CONCEPTS OF COMPOSITE CONSTRUCTION - MUTATIS MUTANDIS
The Challenge of Composites
We are at the stage in the development of composite construction where we can ask some
questions
• How do we apply composite ideology? Have these ideas transformed our thinking?
• How will doubters come to permit the use of apparently untried novelties, such as are the
inevitable progeny of the composite approach? Can we evolve to a concept of controlled
innovation which is customary and reliable?
We now understand established and potential construction materials in fundamentally new ways,
becauseof innovations in knowledge, interpretation and measurement. These have led to
improved knowledge and understanding of
In some cases, this has enabled us to rediscover some traditional materials and techniques,
tailoring combinations of these materials to demand. The modern developments of
ferrous castings use 19th Century craft increasingly combined with the new
understandings gained through computer simulations of a casting's cooling behaviour and
through fracture mechanics. These have been used in a progression of building projects:
Bush Lane House, London; Centre Pompidou, Paris; Alban Gate, London; Bracken
House, London; Ponds Forge International Pool, Sheffield; Menil Gallery, USA; Western
Morning News, Plymouth.
In future, we will be able to tailor materials to meet requirements, whether of the process
or for the final service in place. We can regard our principal structural materials concrete,
steel, masonry, timber and polymers as each referring to families of materials, creating in
our minds a more continuous spectrum.
• Structure, where advanced computer methods make complex analysis freely accessible
through modelling, analysis of elements and inter-action of elements. These allow a
unified view of many structures which hitherto had to be analysed and hence handled in
stages.
• The technological and industrial context, where the construction industry hovers
uncomfortably at the gate still unsure whether to change radically the concept of the
construction process.
We find that we are able to design composite constructions and structures which are significantly
larger and more complex in space, time and material. They are a larger proportion, sometimes
D. MICHAEL 23
almost 100%, of the whole construction. We have substantially extended and redefined the scope
and practical meaning of the analysable entity. Our total engineering is increasingly coterminous
with total architecture.
This review is important because of changes in the context of the development of the engineer's
work
• we should seek to confront with all our energy the emerging social demands of the huge
mega-cities of the 21st Century, built on and in unmeasured terrain and environments, in
desperately poor but irrepressibly optimistic congregations of citizens.
• we will soon be faced with the puzzle of how we decide structure and its form, when
almost any material can be tailored to suit our process or performance requirements and
almost any form can be analysed, the classic problem of rich choice.
Composite materials
The construction sector is characterised by a long list of performance requirements that must all
be met in some measure.
The construction materials industry has produced this huge spectrum of materials on the macro
scale. By altering the combination of different materials, almost the full range of properties can
theoretically be obtained within a composite whole framework. It is becoming possible to create
24 CONCEPTS OF COMPOSITE CONSTRUCTION - MUTATIS MUTANDIS
a range of advanced material solutions tailored to almost any design requirement. Many will
become viable on cost as well. It is usually possible to identify the straightforward material
solution; the challenge for inventive designers is to develop real benefits for their clients and users
through consideration of a many new material developments or new combinations of existing
materials.
Created needs are a common concept in other industries. Personal stereos (Sony Walkman) and
home video games came more from the supply side of a design technology push perspective than
from the demand side of consumer pull.
It maybe difficult for a bridge designer to develop new concepts without considerable support
from clients and investors to reach the required confidence. Nonetheless, such an approach has
been part of our history. It is an essential approach if our industry is to innovate and best make
use of the opportunities that composite materials and systems provide.
For architects, this approach can be realised with relative ease since they are largely relieved of
the realities of delivery carried by the engineers and builders. Richard Rogers' concept for the
Centre Pompidou with Renzo Piano and the Lloyds' Building concentrated on total flexibility of
use within a concept of heavily populated space and its exploitation. The buildings themselves
grew systematically from this idea, adapting to the need for services, fire protection, access and
the like. The central focus, of total physical re-arrangeability for users, was substantially achieved
in the actual constructions.
Technically, the design of the Barcelona Communications Tower was driven by the need to
deliver radio transparency. By recognising advantages of non-metallic tendons at an early stage,
the designers were able to achieve sufficient confidence in the design of this new material to
beneficially exploit the use of these materials in the final structure.
From specific innovative solutions, further innovation can be released to benefit subsequent
projects. The confidence gained from the use of non-metallic tendons at Barcelona was a key
factor in providing justification for the use of composite non-metallic prestressing in a concrete
reservoir in Nottingham. Whilst the prime motivation for use differs (radio transparency or long-
term corrosion resistance) the innovation continues from one application to the next. This is
achieved partly by publication but more usefully through the personal confidence of the
participants. Without practice innovation can rapidly cease. It is therefore strategically important
to society that a culture of innovation be sanctioned.
Composite structures
Cases to illustrate possibilities in composite structures for new designs are the Commerzbank HQ
structure in Frankfurt and the use of the New Austrian Tunnelling Method.
Much work in the USA, Japan and elsewhere has focused on the big issue of remedial work for
old infrastructure, where the compositeness arises through the use of new composite materials
per se and their use in being added to existing materials or structures. A USA survey identified
the following priorities
This potential for enhancing existing structures demonstrates the further extension of composite,
the combination of the old and its repair, a time-dependent case of the composite analysable
entity.
Peter Rice exploited the potential from resolving a structural form into repetitive elements which
would rationalise and economise production as well as expressing a powerful overall design
theme. He sometimes combined this with an attempt to develop composite structures in which
the constituents' weaknesses were deliberately confronted and resolved in the Pegasus mode.
Examples include the IBM travelling pavilion and the Seville pavilion. His critical emphasis was
the pursuit of regularity, repetition and modularity, of standardisation in the best sense, thereby
bringing dreams across to reality.
Industrial issues
Specifiers tend to utilise only tried and tested materials and systems in construction for very
human reasons. Despite many benefits from materials such as polymer composites, their
exploitation to date has been limited to applications where their higher initial materials cost may
be offset against unique benefits and the investment in empirical testing for future applications.
The extensive knowledge that we now possess about materials and systems has created new
standards of admissibility for materials and systems. It allows us to begin to think more clearly
about controlled innovation in the manner of all advanced industries.
At the moment in the UK there are major research initiatives to see how far construction can
learn from manufacturing and other areas of advanced production engineering. One programme
is by the Engineering and Physical Sciences Research Council as its Innovative Manufacturing
Initiative. This has a specific programme for construction, called Construction as a
Manufacturing Process. The Royal Academy of Engineering published the results of its study on
the subject in 1995 in a report called "A Statement on the Construction Industry".
Remember that construction is not a backward form of manufacturing, which should look
uncritically to say the car industry as a model for emulation. Consider some distinctive features
of our sector:
• since construction is mostly fixed to the ground, it needs a mobile industry to reach each
site and this imposes various hazards and uncertainties which make the process as unique
as the delivered product.
26 CONCEPTS OF COMPOSITE CONSTRUCTION - MUTATIS MUTANDIS
• because they enclose activities constructions are bulky in their nature and are likely to
remain so.
• in the building sector, we use thousands of different products and components, often only
a few of each on any given project. This leads to an extraordinary range of production
and assembly methods and consequent complexity of the process and of the product in
contrast to the brilliant simplicity of many of the components. In civil engineering, we
tend to use a smaller range of materials and components, but in larger quantities which
present their own logistical problems. It often pays to study them in analytically more
demanding ways
• since elements have to be long-lasting, durability is tricky as are the associated forms of
finance and procurement. Society does not wish to wait a full life cycle before it claims
the benefit of some new composite material.
• in a long-industrialised and urbanised country such as the UK, over half of construction is
work to existing buildings/structures.
• construction is made from relatively cheap heavy materials compared to many other
manufactured products. Construction materials cost about £0.2/kg and finished
construction around £0.8/kg, compared with cars, where materials typically cost around
£6/kg and the finished product around £12/kg. The methods of recovery, conversion,
manufacturing, handling and transport are consequently distinctive.
The comparison shifts when we consider the cost per cubic metre of created volume in
the products of different industries, like houses, factories, reservoirs, cars, ships, trains or
planes.
Construction is different and distinct from manufacturing, but we can still learn a lot from these
other forms of advanced production, mostly in rethinking the whole project process and
systematically incorporating controlled innovation.
We are now faced with a new synthesis of product and process with the developments in CAD-
CAM, modelling and simulation of supply chain processes, the use of virtual reality and the use of
single project databases as the unifying common information of the design team. These methods
already exist in the aero-space, automotive and process plant industries. We should relish their
impact upon our own sector. They will be crucial as we exploit the full potential of composite
design in construction.
An effective solution is not always achieved. Clinker as an aggregate for 19th Century concrete
was cheap and readily available. It can however support combustion with difficulties for
compartment floors and walls and create incompatibility problems when wet for embedded iron
or steelwork, due to its content of sulphur.
For today's polymer composites, considerable unknowns remain regarding aspects like creep, fire
performance and methods of joining to other components.
The construction industry must attempt to clarify such unknowns if only by appropriate safety
factors to allow for problems in use. Use will then be more extensive and we will find
worthwhile challenge in the unresolved contradictions.
The concept of controlled innovation has served in other industries. It is available for us in the
construction sector to embrace more overtly. It would affect the industry in its organisation as
well as in the components of construction.
The argument would deliver a virtuous circle driven by modern technological and industrial
methods in which
• we redefine a larger and more complex proportion of the designed product as the
analysable entity, inevitably more composite
• we add new materials to old constructions, to enhance their life and performance
• we understand and exploit the significance of greater precision of manufacture and new
organisation of production
• we better define and then simulate both the product and the production processes
• we feel free to examine more materials, more performance attributes, more structural
combinations
In this formulation, the concept of composite construction is the temporarily stable means by
which we can get on with aspects of our work, including the resolution of inherent performance
contradictions of materials or structural forms. The very stability sows the seeds of further
change, through redefinition of the design task and eventually through its demands for ever more
precise and controlled production.
The engine of the creative resolution of these current issues will be the new ideas from our
designers.
28 CONCEPTS OF COMPOSITE CONSTRUCTION - MUTATIS MUTANDIS
Even with all these more powerful techniques, tools and understandings we have not eliminated
the need for good engineers. It is like using a Stradivarius violin. If you give me one, I shall
probably be concerned with not dropping it. If you give one to an orchestra musician, her
performance will sound a bit better. But if you want to hear the difference it makes to use high
class instruments, you need a world-class musician. So it is with engineering: to get the best out
of the new tools, we need excellent engineering designers. Even then the designer is not to be set
merely equal to the world class musician, whose art is a reproductive or at least an interpretive
one. The designer has a clean sheet each time, and is limited only by his own mind and the other
minds he can access.
Summary
High strength steels and high strength concretes have been used successfully in composite
construction in Australia for some years. The use of high strength materials has developed
through both practical considerations and careful research. High strength materials have been
used in beams, columns and slabs to improve economy. It has been found that in addition to
strength and serviceability, stability, local buckling and ductility are also important effects in the
design of composite members incorporating high strength materials.
High strength materials offer new improvements in economy in composite construction. High
strength materials also offer significant benefits in some specific areas of both concrete and steel
construction. In composite construction similar benefits emerge, but the opportunity to combine
materials creatively gives other opportunities and challenges.
Particular examples of the benefits in the use of high strength materials include:
• Use of high strength concrete in columns and core walls. Studies (eg. Sparrow [1], Rose &
Martin [2]) have shown that increase in concrete strength has an economic benefit. Concrete
up to 80 MPa (cylinder strength) is now commonly used in Australian building, while 100
MPa concrete can be supplied in some major cities. Using minimum reinforcement (1%) in
conjunction with the highest strength concrete available consistently produces the most
economic result. For core walls, this is further accentuated by the increase in rentable space
produced by the reduction in structural sizes.
• Use of high strength steel in non-serviceability critical applications including radio masts.
Over the past seven or eight years, Connell Wagner has carried out structural designs for
numerous major multi-storey buildings around Australia. Many of these are in the 40 storey plus
category. In addition, concept and/or advanced designs have been carried out on very tall
buildings such as the 72 and 84 storey schemes for Melbourne Central and options for three
supertall buildings in Brisbane. Some advantages of the use of high strength concrete in high
rise buildings are given in Table 1.
In steel structures, the grades of steel most commonly available have now been raised from
250 MPa and 350 MPa to 300 MPa and 400 MPa respectively providing economies whenever
serviceability requirements are not critical. This is also reflected in composite beams for which
deflection is not generally critical, provided construction deflection of unpropped beams is
controlled by precambering the steel section.
30 HIGH STRENGTH MATERIALS IN COMPOSITE CONSTRUCTION
Requirement Solution
Reduced member sizes, Smaller lighter members
foundation loads
Increased rentable floor space Reduced core wall thickness and column dimensions
1.2 Economics
One of the more interesting aspects of composite construction is in the design of composite
columns. Connell Wagner performed many comparisons of column economy during the late 80's
and early 90's. These were times of high timber formwork costs when steel systems are
particularly attractive. These studies showed the steel tube filled with concrete to be an
economical construction medium, which matched the concrete column, and gave other
advantages. These columns were used very successfully on the Casselden Place project [3] and
several subsequent projects in the early 90's.
10 levels 450x450
8Y32
J
<x)r
1
XX
xy
450x450
8Y20
I
[Ml
410x410
oo
500 dia
6Y20
500 dia
3 UnU
Costings for a series of alternative column configurations to resist the same load are given in
Table 2 (Webb and Peyton [4]). While these costings are not necessarily up to date, they do
illustrate some important points eg. a steel circular tube column filled with high strength
concrete can be cost effective compared to the all steel column.
Another interesting feature is revealed in Figures 1 and 2. For a constant load carrying capacity
for a given concrete strength, the cost of reducing the overall size of a reinforced concrete
column using additional reinforcement is very high and becomes prohibitive for higher strength
concrete (Figure 1). This phenomenon is not nearly so marked in a tube column (60 MPa
concrete) where similar size reductions using increased plate thickness can be achieved with only
modest cost penalty (Figure 2).
1.5 -
Rate
1.4
Vs.
1.3
\/
4%'
1.2
3%
I
1.1
1.0
Figure 1 Relative costs for reinforced concrete Figure 2 Relative costs for concrete-filled
columns with constant load capacity tubes with constant load capacity
and varying percentage reinforcement and varying wall thickness
Tube columns also provide a better environment for high-strength concrete because:
1.3 Constructability
The concrete-filled steel tubular column is seen as an attractive viable system because it provides
the ability to construct a composite building with the benefits normally associated with
traditional steel construction. It is a natural efficient method for using high strength concrete.
Research [6] is now underway using these columns with 100 to 120 MPa concrete as outlined in
Section 2. The economy of this construction technique is driven by using the minimum amount
of reinforcement and steel in the tubes with the highest strength concrete.
32 HIGH STRENGTH MATERIALS IN COMPOSITE CONSTRUCTION
Placing concrete for these columns using a pumping technique has been pioneered on the
Casselden Place project. The concrete is pumped into the steel tubes through a nozzle
connection. In any one lift, concrete is placed from the bottom and pumped up as many as six
stories at a time. The placement method, which eliminates the need to vibrate the concrete, was
validated by a full-scale prototype test [3],
The tube is erected much like a traditional steel column, with floor reinforcing, concreting, etc.
occurring as in conventional steel building practice. The tubes are erected in either two-or three-
storey lengths and connected temporarily by tumbuckles. These facilitate plumbing and
alignment and allow the crane to quickly release the column. This technique can also achieve
excellent tolerances in column plumbing and alignment. The splice is completed using butt-
welding, and the tumbuckles are then removed.
Concrete with strength up to 70 MPa (10,000 psi) was used on the project. The superplasticised
concrete mix contains silica fume, principally to eliminate bleeding, thereby providing
consistency over the height of the placement and eliminating the need to scabble the interface
between placements. The method is extremely efficient, and large numbers of columns can be
filled at a time by a small workforce and with minimal material waste.
The bare steel tube was capable of supporting six floors of construction, giving the contractor
flexibility as to when and where to place the concrete.
High strength concrete is particularly economical when used in circular concrete-filled tubes
subjected mainly to axial compression for which the high compressive strength can be fully
utilised in design. To obtain this economy, it is essential that the highest strength can be
achieved still using conventional materials and manufacturing and placement procedures.
Figure 3 Stress-strain curve for high strength concrete from cylinder tests.
A commercial concrete mix has been developed that meets these objectives and can develop a
compressive strength around 120 MPa. However, the post-ultimate stress-strain response of high
strength concrete can be characterised by a very rapid unloading, even exhibiting the phenomena
of "snap-back". This brittle behaviour cannot be measured using normal testing procedures
including displacement control. A number of alternative techniques can be used. A simple
method is to use the diametric strain to control the displacement of the testing machine as this
was found to increase monotonically with axial displacement. The results of such a test are
shown in Figure 3. The concrete cylinder labelled with an "M" was moist-cured in a lime bath at
R. BRIDGE, M. PATRICK, J. WEBB 33
100% relative humidity and a constant temperature of 20°C. The cylinder labelled with a "D"
was dry-cured by sealing the cylinder in a polythene wrap and storing it at ambient temperatures
to simulate conditions in the concrete-filled tube specimens.
For concrete-filled steel tubes in which the concrete and steel are loaded simultaneously,
enhancement of concrete strength due to confinement by the tube can be obtained for low to
medium strength concretes and this has been recognised in Eurocode 4 [5], The cross-sectional
strength Nuoin axial compression including the beneficial effects of concrete confinement for
concretes with strengths up to a maximum of 50 MPa is given by
where As and Ac are the cross-sectional area of the steel and the concrete, fy and fc are the
characteristic strengths of the steel and concrete, and yc and y, are the partial safety factors which
may be taken to be unity when the strength of the materials has been accurately measured as in
laboratory tests. The increase in concrete strength from confinement (accounted for by the 77/
factor) and the corresponding decrease in steel strength (accounted by the 772 factor) may be
considered if the non-dimensional column slenderness A is less than 0.5 and the eccentricity of
loading does not exceed Dl 10 where D is the external tube diameter.
Recent axial load tests by O'Shea and Bridge [7] on steel tubes filled with high strength
concretes with strengths in excess of 100 MPa have revealed that virtually no enhancement can
be obtained, the concrete behaving as if unconfined up to the maximum strength. Five tubes
designated CS were tested under axial load. Their dimensions and material properties are given
in Table 3 together with the maximum load capacity Nreu obtained in the test.
In Table 4, code predicted strengths Nc,,de (Equation 1) are compared to the actual test strengths
Njest The code values for rp and 772 are calculated using Eurocode 4 [5]. Using these values, the
equivalent factor implied in the code for concrete strength enhancement (a factor of 1.0 being
zero enhancement) can be calculated and is shown in Table 4 under "Concrete Enhancement
EC4", the thicker tubes having more enhancement. The Grimault and Janss [8] effective steel
area was used in the calculations.
An approximate value of the concrete enhancement factor in the tests can be calculated by
subtracting the maximum bare steel strength from the maximum concrete-filled steel tube
capacity and dividing the remainder by the concrete area and the concrete cylinder strength. This
is shown in Table 4 under "Concrete Enhancement Test". The values shown are all slightly less
than unity suggesting that there is little concrete enhancement (or alternatively the concrete
cylinder strengths are inaccurate). As the actual concrete strength was determined from at least
ten cylinder tests, it is more likely that there is little confinement of the concrete in the steel
tubes. This is also supported by the fact that the test values are essentially constant and do not
34 HIGH STRENGTH MATERIALS IN COMPOSITE CONSTRUCTION
increase with tube wall thickness. If no confinement is assumed i.e. letting rj2= 1.0 and r\i 0,
Eurocode 4 [5] provides a good prediction of the section strength as shown in the last column of
Table 4. When confinement is considered, Eurocode 4 [5] is slightly unconservative for axially
loaded thin-walled steel tubes infilled with high strength concrete. Therefore, confinement for
high strength concretes with strengths in the range 100 MPa and above should be ignored.
Two circular tubes designated CL were tested [7], S20CL and S12CL, with only the concrete
loaded. Their dimensions are given in Table 3 and were similar to their companion S20CS and
S12CS tubes. The concrete was loaded axially with the steel unbonded. This was achieved
through greasing the internal tube surface prior to filling with concrete. Strains measured on the
steel tubes using rosettes verified the debonding procedure. Special loading disks were
manufactured to ensure that the axial load was only applied to the concrete. Therefore the steel
only provided lateral confinement to the concrete and did not carry any direct axial load.
4000
S20CS1 OA
3500 / S20E210B
S20E110B
3000
// S20CL10A
0
r
* 1
It can be seen that the maximum load for the concrete loaded only specimen S20CL was higher
than that for specimen S20CS for which the concrete and steel were loaded simultaneously. The
same behaviour was observed for the S12 tubes as indicated in the last column of Table 3.
Therefore, an increased strength can be obtained without the steel being axially loaded. The
R. BRIDGE, M. PATRICK, J. WEBB 35
strength increase was 9.8% for the thicker S20 tubes and 5.3% for the thinner S12 tubes, the
thicker tube providing more confinement and hence higher strength increase as expected. Hence,
confinement of high strength concrete is possible provided the concrete alone is loaded.
The more efficient use of steel tubes filled with unbonded concrete with only the concrete loaded
has been proposed by Orito et. al. [9]. However, detailing, especially at beam to column joints,
needs careful consideration to ensure axial load is not transferred to the steel tube.
With the use of high strength steels, thinner steel sections can be achieved for the same load
carrying capacity. However, the effects of local buckling have to be considered. O'Shea and
Bridge [7] have examined the local buckling of circular steel tubes with or without concrete
infill. The results of axial load tests on bare steel tubes (BS tests) and steel tubes with unbonded
concrete infill with only the steel loaded (BSC tests) are shown in Table 5.
b/tV (fy/250)
having clamped edges with a buckling coefficient k of 9.99. This results in an increase in
column strength as evidenced in the test results plotted in Figure 5 where b is the width of the
tube and t is the plate thickness. Also, current design methods in standards such as AS4100-
1990 [10] can be used to predict the plate strength fu of such tubes provided the correct value of
buckling coefficient k can be included in the design method.
The G550 sheet steels used to make profiled steel sheeting range in thickness from 0.6 to 1.0 mm
with a total zinc coating mass of 200 to 450 g/m2. Steels of such high strength are not normally
used for this purpose in Europe or America where hot-rolled grades of between about 275 and
350 MPa predominate. Some potentially undesirable properties of G550 steels are their low
elongation at fracture and lack of strain-hardening.
(a) Stress-strain curve from coupon testing (b) Composite slab load-deflection curve
Figure 6 Tensile testing of G550 sheet steels
values of the tensile-strength-to-yield-stress ratio (ft/fy) equal to unity are consistently obtained.
The modulus of elasticity is typically about 200 GPa, and this value can be used for design.
Yield stress values for G550 sheet steels are significantly above the minimum specified 550
MPa. Mean strengths tend to increase as the base metal thickness reduces due to the increased
amount of cold reduction. Large variation occurs in the strength and ductility of material taken
from different mills. Post-ultimate ductility measured by the fracture-to-ultimate-load ratio is
comparable to that for mild sheet steels and can reach as low as 0.7.
Tests on the Australian profile Bondek® II have yielded design values r= 1 \5^(tbmfc) in kPa
and ß 0.6, where t^m is the nominal base metal thickness of the sheeting (mm) and fc
is the
characteristic compressive strength of concrete at 28 days (MPa) [15]. The tops of the sheeting
ribs are embossed which contributes significantly to the magnitude and ductility of the
mechanical resistance they develop. Embossing must not cause any local fracturing in the G550
sheet steel, since this can lead to premature fracture of sheets when they act as reinforcement at
the composite stage.
The slabs may be designed as simply-supported for some limit states and continuous for others,
subject to certain conditions being satisfied. Reinforcing steel is placed transversely in the slabs
for shrinkage and temperature control. Longitudinal reinforcement may be placed in the top face
over supports to increase negative moment capacity and control flexural cracking. It may also be
placed in the bottom face, where (because the Australian profiles develop strong, ductile
mechanical resistance) it can act in conjunction with the sheeting and increase the positive
moment capacity. A set of limit-state design rules has recently been prepared to Australian
Standards for composite slabs incorporating Bondek II [15] and are briefly discussed below.
ultimate load due to possible fracture of the reinforcement (100 per cent redistribution).
Premature fracture of low-elongation reinforcement in the bottom-face acting in conjunction with
the steel sheeting may also be a problem and may need to be considered in design.
Partial shear connection strength theory can be used to design the positive moment regions of
composite slabs exhibiting one-way action. Simplified equations have been formulated to
calculate the design positive moment capacity of Bondek II slab cross-sections with either partial
or complete shear connection and including the contribution of bottom-face reinforcement. The
model used in the theory recognises the existence of support friction and the additional
anchorage of the sheeting continuing into adjacent spans. The accuracy of the model has been
validated by testing [17, 18].
In one such test, the theory was used to predict the failure load of a Bondek II slab (see
Figure 6(b)). The test showed that the full moment capacity of the peak moment cross-section
(corresponding to yielding of the steel sheet over its full cross-sectional area) was reached. The
stress-strain curve of the steel (t^m 0.750 mm) in the longitudinal direction took the form
shown in Figure 6(a) with an upper yield point. Tensile coupons from the steel gave varied
results with yield stress fy ranging from 630 to 635 MPa, upper yield stress from 640 to 660
MPa, and elongation after fracture (50 mm gauge length) from 5.6 to 8.5 per cent. Interestingly,
however, uniform elongation (37.5 mm gauge length outside fracture zone) was close to zero for
coupons with a high upper-yield-stress-to-yield-stress ratio, while it reached a maximum of
nearly 5 per cent for other coupons. In the slab test, failure occurred when the steel fractured at a
mid-span deflection of almost span/50, which is acceptable considering that a single line-load
was used in the test and there was a steep moment gradient. Fracture occurred at a major flexural
crack and extended across a whole sheeting pan and through an adjacent rib. The peak of the
curve in Figure 6(b) could be predicted very accurately. More complex situations involving the
effects of partial shear connection and support friction have also been accurately predicted
(within 10 per cent) using the theory [18]. The residual stresses described in Section 3.2.3
created during roll-forming can be ignored in the analysis, as can any initial flexural stresses
arising during construction prior to composite action.
The shear capacity of the positive moment regions of Bondek II slabs has been investigated
experimentally [19], and it has been recommended that for this profile the occurrence of vertical
shear failure (viz. by diagonal splitting or flexure shear) can be ignored under uniform-loading
conditions. This recommendation applies only when the amount of reinforcing steel taken into
account in determining the moment capacity of the slab is less than a certain amount [15].
The main design principles are: (i) crack width is calculated using a simplified equation derived
using the method for calculating crack width defined in BS 8110, Part 2 [20]; (ii) a limiting
width of 0.3 mm specified in BS 8110, Part 2 is adopted; (iii) elastic analysis is used to calculate
design bending moments under both short-term and long-term service loading; (iv) the stress in
the reinforcement is calculated using elastic, cracked-concrete modular ratio theory and is kept
Ft. BRIDGE, M. PATRICK, J. WEBB 39
below the yield stress under short-term service loading and 80 per cent of the yield stress under
long-term service loading; and (v) the nominal negative moment capacity of the support regions
is to exceed the cracking moment by at least 20 per cent so that more than one flexural crack will
form.
Bandbeams are wide, shallow beams that are supported in a parallel arrangement on isolated
columns. Slabs are formed between the bandbeams. Both the slabs and bandbeams are assumed
to exhibit one-way action. While slab depths in steel-frame buildings seldom exceed 200 mm,
slabs spanning between bandbeams can reach 400 mm or more. The span-to-overall-depth ratio
of a slab (L/Dcs) in either of these situations typically varies between 20 and 40, depending on
the support conditions, magnitude of loading and deflection limits.
The design rules described in Section 3.3 have been used to prepare sets of solutions for
examining the effect of base metal thickness, live load, yield stress, moment redistribution and
slab support conditions on the utilisation of sheet steel tensile capacity in Bondek II slabs. For
brevity, only some of the results are presented here. Slab overall depth was considered to vary
from 100 to 400 mm. All solutions satisfy the criteria concerned with design for strength,
deflection control (At0/L < 1/250 and Ainc/L < 1/350), and crack control when appropriate.
Utilisation is examined for the positive moment region using the ratio of design positive bending
moment to design positive moment capacity, i.e. A M*/((j>MU0). The sheeting is assumed to be
the only bottom-face reinforcement, and it is assumed to be fully anchored at the critical cross-
section. Cases for which A > 1 imply that the tensile capacity of the sheeting is fully utilised and
additional bottom-face reinforcement is required for strength.
tta, Q= (a)0.6,10 (b)0 6.3 (c)1.0,10 (d)1.0,3 mm, kPa twQ=(a)06,10(b)06.3(c)! 0,10(d)l 0,3mm,kPa
b
"T 1
r-T—T
J I I
(a) Yield stress varied from 550 to 350 MP a (b) Moment redistribution of 100%
Figure 7 Factors affecting utilisation of G550 sheet steels in Bondek II slabs
The cases shown in Figure 7(a) are for interior spans. The solid and dashed lines are for fy 550
and 350 MPa, respectively. Solutions for different combinations of base metal thickness tym and
live load Q are also labelled. It can be seen that G550 steel is most effectively utilised in
situations corresponding to bandbeam construction when depths are large (>250 mm), live load
is heavy (10 kPa) and base metal thickness is least (0.6 mm). However, the solutions for G350
steel show that its tensile capacity can be exceeded in steel-frame building situations and hence
bottom-face reinforcement would be required to supplement the sheeting.
The cases shown in Figure 7(b) are for end spans using only G550 steel The solid and dashed
lines are for 0% and 100% redistribution respectively from negative to positive moment. The
values of A can be seen to increase substantially with moment redistribution indicating better
utilisation of the high strength G550 steel. (Crack control reinforcement must be provided
separately for this case if required.)
40 HIGH STRENGTH MATERIALS IN COMPOSITE CONSTRUCTION
The effects of partial shear connection on the curves in Figure 7 can be ignored for slabs
incorporating Bondek II on account of the strong mechanical resistance it develops.
References
[1] Sparrow, C.J., "High Strength Concrete in the Melbourne Central Project", CIA Biennial
Conference, Adelaide, 1989, Concrete Institute of Australia, Sydney, 1989.
[2] Rose, M.A., and Martin, O., "Optimisation of Reinforced Concrete Cores in Tall
Buildings", CIA Biennial Conference, Adelaide, 1989, Concrete Institute of Australia,
Sydney, 1989.
[3] Webb, J., "High Strength Concrete - Economics, Design and Ductility ACI Concrete
International, January 1993, pp 27-32.
[4] Webb, J., and Peyton,J.J., "Composite Concrete Filled Steel Tube Columns", 2nd National
Structural Engineering Conference, Institution of Engineers, Adelaide, Oct. 1990,
[5] Eurocode 4, CEN (1992), "prENV 1994-1-1 Eurocode 4, Design of Composite Steel and
Concrete Structures, Part 1.1 - General Rules and Rules for Buildings", European
Committee for Standardisation, 1992, Brussels.
[6] O'Shea, M.D. and Bridge, R.Q., "High Strength Concrete in Thin-walled Circular Steel
Sections", Proceedings, 6th International Symposium on Tubular Structures, Melbourne,
Australia, 14-16 December 1994, pp. 277-284.
[7] O'Shea, M.D. and Bridge, R.Q., "Circular Thin-walled Tubes with High Strength Concrete
Infill", Proceedings, Engineering Foundation Conference, Composite Construction III,
Irsee, Germany, June, 1996.
[8] Grimault, J.P. and Janss, J., "Reduction of the Bearing Capacity of Concrete Filled Hollow
Sections Due to Local Buckling", Preliminary Report, ECCS Colloquium on Stability of
Steel Structures, Liege, 1977, pp. 175-179.
[9] Orito, Y., Sato, T., Tanaka, N. and Watanabe, Y., "Study on the Unbonded Steel Tube
Composite System", Proceedings, Composite Construction in Steel and Concrete, ASCE
Engineering Foundation, Potosi, Missouri, 1987, pp. 786-804.
[10] Standards Australia, "Steel Structures AS4100-1990 ", Standards Australia, Sydney,
Australia.
[11] Bridge, R.Q. and O'Shea, M.D., "Local Buckling of Square Thin-walled Steel Tubes Filled
with Concrete", Proceedings, 5th International Colloquium on Stability of Metal
Structures, Structural Stability Research Council, Chicago, 1996, pp. 63-72
[12] Standards Association of Australia, "Steel Sheet and Strip - Hot-dipped zinc-coated or
aluminium/zinc-coated, AS1397-1993", Sydney, Australia.
[ 13] Standards Association of Australia, "Methods of Test for Elements of Composite
Construction, Method 1: Slip-Block Test", Committee Draft, Doc. No. BD/32/4/96-2,
Sydney, Australia.
[14] Patrick, M. and Bridge, R.Q., "Partial Shear Connection Design of Composite Slabs",
Engineering Structures, Vol. 16, No. 5, 1994, pp. 348-362.
[15] Patrick, M., Goh, C.C. and Proe, D.J., "Rules for Limit-State Design to Australian
Standards of Simply-Supported and Continuous Bondek II Composite Slabs in Steel-Frame
and Masonry Wall Buildings", BHP Research Report, No. BHPR/SM/R/005, 1995
[16] Patrick, M., Akbarshahi, E. and Warner, R.F., "Ductility Limits for the Design of Concrete
Structures containing High-Strength, Low-Elongation Steel Reinforcement", Proceedings,
Concrete 97 Conference, Adelaide, Australia, May 14-16, 1997.
[17] Veljkovic, M., "Influence of Load Arrangement on Composite Slab Behaviour and
Recommendations for Design", Journal Constructional Steel Research, in press.
[18] Patrick, M., "Shear Connection Performance of Profiled Steel Sheeting in Composite
Slabs", Ph. D. Thesis, School of Civil and Mining Engineering, University of Sydney,
1994.
[19] Patrick, M., "Testing and Design of Bondek II Composite Slabs for Vertical Shear", Steel
Construction Journal, Australian Institute of Steel Construction, Vol. 27, No. 2, May,
1993, pp. 2-26.
[20] British Standards Institution "Structural Use of Concrete; Part 2: Code of Practice for
Special Circumstances", BS 8110: Part 2: 1985.
41
Summary
The paper gives an overview of the development of composite structures for buildings in
Germany in recent years. The background and advantage of partially concrete encased
columns and girders is demonstrated with three buildings for the car industry, an office
building and the extension of the airport in Hannover. Furthermore two impressive high rise
buildings in Düsseldorf and Frankfurt will be presented.
1 Introduction
Comparing the development of composite structures for buildings in Europe it is obvious that
in contrast with neighbouring countries, in Germany the technology with partially concrete
encased members has been favoured over the last twenty years. The paper will show the
advantages of this technology for some typical examples of outstanding composite structures
for industrial and office buildings.
Industrial buildings for the car industry require a high flexibility because the technical
equipment must be converted frequently during the design life. The Figures 1 to 3 show three
typical buildings for the car industry built between 1982 and 1992. These buildings
demonstrate the development of composite structures from the conventional type to the modern
technology of concrete encased beams and columns with significant advantages, regarding fire
resistance as well as durability and last but not least economic benefit.
In 1981 the Opel corp. erected a new paint unit in Rüsselsheim, shown in Figure 1 III. The
building measures 405 x 80 metres and has three stories with a total height of 31.5 metres.
Additionally a penthouse with dimensions of 340 x 20 metres with a height of 7 m is located
between gridlines D and E. In the transverse direction the structure consists of sway frames
with rigid connections at the top of each storey level on gridline C and with all the other
42 OUTSTANDING COMPOSITE STRUCTURES FOR BUILDINGS
Only 4 years later the new body unit of Porsche was built in Stuttgart (Fig. 2). The design
philosophy had completely changed. In the meantime a code for composite columns had been
published in Germany and intensive research work on fire resistance of partially concrete
encased beams had been carried out. The building with three stories and an additional
penthouse measures 125.5 x 65 m. The structure is stabilised in the transverse direction by
concrete gable walls and in the longitudinal direction by truss bracings. Main and secondary
beams are continuous and concrete partially encased sections are used; the columns are
nominally pinned at both ends and fully encased composite sections are now used. The
building has an extremely high degree of installation, requiring a lot of web openings in the
main girders with a span of 20 metres. The highly stressed main girders have flanges and webs
with a wide variation of plate thickness (Fig. 4). In comparison to the Opel paint unit the
spacing of the secondary beams is reduced to 2.5 m to avoid propped construction for the
composite slabs with a depth of 240 mm. The structural fire design was based on a model
where the effect of temperature on material characteristics is taken into account by reducing
the material properties and the dimensions of the cross sections. For the required fire
resistance Class R90, additional reinforcement was necessary in the concrete encasement of the
main and secondary beams. This reinforcement was not taken into account for normal
temperature design.
In 1992 the new Opel paint unit in Eisenach was one of the first major investments in the new
states of reunified Germany (Fig. 3). The building is one of the most modern of this type in
Europe and sets new milestones for the car industry 131. The structure with measurements of
240 x 56 metres and up to 30 m high consists of two main production levels and an additional
technical penthouse. The experiences with the body unit in Stuttgart showed that systems with
nominally pinned columns lead to disadvantages during erection, because an enormous number
of temporary bracings is necessary. Therefore a structural system with sway composite frames
in the transverse direction and truss bracings in the longitudinal direction was chosen in
Eisenach.
In order to minimise the time of construction and welding on site, the detailing of the
connection is very important. Figure 3 shows the main frame joint used on gridlines B and C
and the support of the main girders at the corner columns on gridlines A and D.
G. HANSWILLE 43
n v
1
3150
—1 1 1 1—*— "
1
m 2000
—+
20,50
' | i i i i
1 iii i i |i
9,00
v
0,00
"1
J 20,00 m 2000m 20,00m 20,00m j
100,0
^ 112.5 I
j 1,5 112 S
^ 1112.5
404,50m
«
1.5 100.IJ
1112,5 ||
P 87,5
J 112 5
38,50
"31 5sz_
20 5y_
*
mil 111 lap 11111 r*T*11 injPTTT7
90s_
002_
* ;r w.,
'
//'VW//'/, propped slabs
-girder for
suspended
I loQd 4
29,88 9 9
Penthouse
21,75
JU LJ L1
14,90
in
6,85
±0,00
2
26,00 JQ4Q 12,00
t J.
I 0 (p (p Ç) (p
56,26 m
7,50: '5,50
125.50 m
29,88
i —n A fl
i
21,75 / " '1 1 V
i on in II
6,85
n
III
V
0,00 ï II IAA H
§
B, C
I
secondary main beams
700
beams with web S
150
columns openings
gr idlines m
A ,0 '
400 I
400
"n
®
4~
B)
54,00
©
;
D)
's> -
'&S&
18,00 J 16,00 18,00
30,60
_SL
•PENTHOUSE -3
+ 20,00
\/yyzz/*yy///.?\ ^
>8,90
[ 1180 secondary
beam
TT
(p (|) (|) © (p (8)
(p®Jppd6) ^ ^ dp
®~o1
O
cd
®-o-
O
ao
cd
1Z0C^12,0|
mmm
„ 12,0j
^+30,60
„20,00 X X
r* î^s
/\
frame joint
gridlines
giidlines B and C
A and D
The depth of the columns changes at each storey level, and the supporting reactions of the
main girders are introduced into the columns by end plates. The bending moments are
transferred into the columns by reinforcement and contact plates at the top flange and by
contact in combination with welded plates at the bottom flange. In comparison with the paint
unit in Rüsselsheim, the composite columns had the advantage of increasing the horizontal
stiffness of the frame significantly. To avoid the high costs for the formwork of completely
concrete encased sections, partially encased sections are used for the columns. Additionally
this has the advantage that new installations can be fixed at the free steel flanges without
difficulties. In Germany it was the first time that a mixed system with composite columns,
composite girders and steel trusses was designed as a composite sway frame. The design of
this type of framing is not covered by Eurocode 4 and the national German codes. In a first
trial calculation an elastic calculation was carried out with an effective cracked stiffness for the
columns. The effect of cracking of concrete in the main girders was taken into account by
reducing the stiffness at internal supports to the stiffness of the steel section consisting of
structural steel and reinforcement. Effects of creep and shrinkage were taken into account by
use of different modular ratios for permanent actions, shrinkage and hyperstatic effects
developing in time. For the standard frames the second order effects lead to an increase of
bending moments up to 25%, which is significantly lower than the second order effects of the
steel frame of Opel Rüsselsheim. In a second step, a non-linear calculation was carried out for
critical load arrangements, taking into account cracking of concrete and tension stiffening
effects. These calculations have shown that the simplified method gives safe results.
Figure 4 shows the material distribution of the main girders of the body unit of Porsche and
the paint unit of Opel. It is noticeable that the number of changes of cross-sections of the
flanges and the web of the Opel beam is significantly reduced. As explained above, in the
design of the girder of Porsche the concrete encasement and the reinforcement were taken into
account only for fire resistance. For the Opel girder the concrete encasement and the
reinforcement were used to improve the bending resistance for normal temperature design as
well as fire design. For economical reasons, instead of altering the flanges of the structural
steel section, additional reinforcement in the concrete encasement was provided. Table 1 gives
a calculation example for a typical main girder with concrete encasement and additional
reinforcement, and in comparison for a girder with conventional fire protection by encasement
with fire resistance boards, based on average unit prices of the year 1996. The comparison
demonstrates that the concrete encased beam is more economical and has the additional
advantages described above and has as well a significantly higher flexural stiffness.
T:
*
M2 L20 L12
15 15
1 0 \
paint unit Opel 12*25 25 ,12*25nI $
1
'S 1
1h
L L15 -fL35 Li5
18000 9000 uso
>.
—
Fig. 4 Material distribution of the main girders of the body unit of Porsche
and the paint unit of Opel in Rüsselsheim
G. HANSWILLE 47
In contrast to the body unit of Porsche the span of the composite slabs is increased to 3,0
metres for the Opel paint unit. Propping of the sheeting is normally necessary for a slab with
Holorib sheeting and a thickness of 200 mm. The slabs were poured in two layers, in order to
avoid the high costs of propping in industrial buildings with a height between floors of more
than 10 m. The thickness of the first layer resulted from the bending resistance of sheeting
acting as formwork without props. After hardening of the first layer the composite slab was
capable to resist the remaining concrete for the required slab thickness of 200 mm. To achieve
a sufficient longitudinal shear resistance between the two concrete layers, additional shear
reinforcement had to be provided.
Figure 5 shows a multi-storey building for Siemens in Berlin, erected in 1993 /3/, /4/.
Maximum flexibility was required by the client with regard to the usability of different parts as
office, production and storage areas. The building rises five stories on an area of 81 x 81
meters with a column grid of 14.8 x 10.8 m. The slabs with a span of 3.6 m and a depth of 20
cm are composite with Holorib steel sheeting. All beams with a depth between 450 mm and
900 mm are continuous with concrete encasement and additional reinforcement taken into
48 OUTSTANDING COMPOSITE STRUCTURES FOR BUILDINGS
account in the design at normal temperature and for fire resistance. Slabs and beams had to be
designed for traffic loads up to 30 kN/m2. In the areas with a height between floors of 11.5 m
the concrete was poured in two layers to avoid propping of the sheets. In all other areas
propped construction was preferred for the sheeting. The columns are partially concrete
encased I-sections and concrete filled tubes. In the centre of the building, a big composite
grillage with an integrated composite ring beam is located for a domelight. The building of
Siemens is an good example how the architectural requirements showing the steel members
can be combined with high fire resistance and high load bearing capacity of the structure.
79,20
11 x 7.20
t t t
i i I
8
0 1
'-J / /
Sj
mKilMESS. 1
I 1
M
1
•=———
I
i '
i!
i
22,20
i-
>18.00
•13.80
8,40
4,40
,±0,00
,-3,10 "S"
Figure 7 shows the extension of the airport in Hannover under construction in 1997. The two
storey building has a triangular plan with a side length of approximately 144 m. It is stabilised
for horizontal action by concrete cores. Most columns are concrete filled steel tubes. The
beams are concrete encased welded I-sections with an extremely high ratio of reinforcement.
Between the flanges, up to 20 bars with a diameter of 28 mm are provided. As explained
above, this type of beam is economical because the unit prices for reinforcement are
significantly lower than those for structural steel and because the bending resistance at normal
temperatures and not the fire resistance governs the design. For this special type of composite
beam, tests were carried out to ensure that the total reinforcement of the concrete encasement
can be taken into account for the plastic bending resistance, where there is only 50% of full
shear connection. Furthermore the crack formation, the crack width and the deformation
behaviour of the beams were checked because of the special public interest in the building.
The test results have shown, that for beams predominately loaded by point loads the design
rules in Eurocode 4 regarding the minimum degree of shear connection are very conservative.
In addition to the main reinforcement of 28 mm bars, small-diameter bars near the surface had
to be provided, to ensure a sufficient distribution of cracks in the concrete encasement in order
to develop the full plastic bending resistance and to avoid early failure of the main
reinforcement at single big cracks.
section A-A
147,643 m
plan
144,00m
mv
main girder
5 High-rise Buildings
5.1 General
In 1997 two very interesting building projects will be finished in Düsseldorf and Frankfurt.
The Stadttor (Towngate) in Düsseldorf 151 and the Commerzbank tower in Frankfurt 161. Both
high rise buildings use a large variety of different composite members with sophisticated load-
bearing structures.
The building with a rhomboid plan consists of two 16 storey office towers connected by
horizontal truss members extending through the three top floors (Fig. 8). The horizontal
stability is realised by small concrete cores and mainly by three portal frames consisting of
composite truss-columns and the horizontal truss members. The three portal truss-frames form
a z-shaped system in the plan. The trusses are made by concrete filled steel tubes where the
horizontal and diagonal members above the fourth floor are without concrete. Composite slabs
in combination with concrete encased composite main beams and secondary composite beams
without concrete encasement are used for the vertical loads. With regard to the required fire
resistance Class R90 cementious coatings or fire resistance boards are employed for the
secondary beams. The fire resistance of the columns is ensured by concrete and additional
reinforcement without any further protection of the steel tubes.
115,75
The new building of the Commerzbank is the most interesting project in Germany at the
present time. With a overall height of 298.74 m, including antenna, 63 floors and an effective
area of 52700 m2, the building will give room for 2400 employees of the Commerzbank. The
plan of the building has the shape of an equilateral triangle with rounded corners and slightly
curved sides with length of approximately 60 m (see Fig. 9). Three cores at the corners,
ending at different heights, contain stair cases, elevators, adjoining corridors and the
installation for the building. The cores are connected by office areas where the standard floors
consist of two office areas. The third side is kept free over 4 storeys and contains a garden
area. This standard floor plan is turned by 120 degree every four storeys. For the horizontal
loads, a three dimensional structure (consisting of the mega-columns connected by link frames,
the vierendeel frames located in the facade, and the composite slab floors) forms a tube, which
is fixed in the foundation. All other columns inside the cores and the atrium columns are
nominally pinned at both ends and do not contribute to the horizontal stability of the building.
The composite floors consist of Super-Holorib sheeting in combination with light weight
concrete with a density class of 20. In the office area composite beams with welded sections
and large web openings span between the atrium beams to the outside vierendeel steel frames.
Within the cores mixed systems with steel and composite beams are used. The required fire
resistance Class R120 for the beams was achieved by fire resistance boards. The atrium
columns are designed as composite columns with an inner and outer equilateral-triangle steel
section with steel grades S460 and S355. Composite action is achieved by headed stud shear
connectors. Because this type of section is not covered by Eurocode 4, an elasto-plastic design
had to be carried out. The design for fire resistance is based on a reduced effective section
with decreased material properties. The mega columns in the edges of the plan consist of fully
concrete encased steel trusses forming a composite column together with concrete and
additional reinforcement. All other columns are designed as steel members.
References
III Muess, H.: Anwendung der Verbundbauweise am Beispiel der neuen Opel-
Lackiererei in Rüsselsheim, Der Stahlbau, Heft 3, 1982
121 Jost, E., Hanswille, G., Heddrich, R. Muess, H., Williams, D.A.: Die neue Opel
Lackiererei in feuerbeständiger Verbundbauweise, Der Stahlbau, Heft 8, 1992
/4/ Eichhorn, H., Kühn, B., Muess, H.: Der Neubau der Siemens AG Verkehrstechnik in
Berlin Treptow, Der Stahlbau 65, 1996
151 Lange, J.: The Düsseldorfer Stadttor - A 20 storey office Building in Composite
Construction, Engineering Foundation Conferences Composite Construction HI,
Irsee 1996
Summary
Slim floor construction is being used increasingly in Europe. Several construction systems
have been developed and their characteristics are described on behalf of some buildings which
just have been completed.
1. Historical Background
The architects and owner wish for greater flexibility in multi-storey construction has led to
more frequent use of slim floors. The main characteristic of this form of construction is a
shallow floor in which beams and slab elements are integrated within the same depth.
In fact the principle of the slim floor exists at least since the middle of last century. In 1845
timber floors were replaced in Great Britain by stone arches comprising integrated iron
beams. A typical development consisted in the so-called "Prussian Cap Floor", shown in
figure 1.
By the end of the nineteenth century standard rolled sections were used as integrated within
the concrete slabs. However as the span of the concrete was still quite small, the distance
between the steel beams was also quite reduced.
New developments in steel and concrete construction have favoured a real "Renaissance" of
slim floor construction since 1975. According to Dr. Lars Wallin of the Swedish Institute of
Steel Construction [1], "One effective way to reduce the total floor construction height is to
support the floor elements on the bottom flanges of the floor beams. For that purpose a low,
welded floor beam with a wide bottom flange and a narrow top flange has been developed.
The selected beam height should be approximately equal to or somewhat lower than the floor
element thickness. Hence the floor construction height can be reduced quite substantially.
This type of floor beam can also be used with composite action through welded studs and with
continuity to further reduce the beam height."
54 SLIM FLOOR CONSTRUCTION: WHY?
Consequently a particular steel frame system characterized by slim floors and integrated fire
resistance has become the dominant system for many buildings in Nordic Countries.
Corresponding steel beams are shown in figure 2a, b and c. Slightly different integrated floor
beams are now being used in Continental Europe and in Great Britain according to figures 2d
and 2e, and to chapters 2.1 and 2.3 [2].
Fig. 2 Slim floor beams in the Nordic countries (a,b,c) and in Continental Europe and
Great Britain i.e. IFB (d) and SFB (e).
- integrated floor beam IFB with hollow core prestressed units HCU,
- rolled profiles with in-situ normal weight concrete, the S+V system,
- slim floor beam SFB with deep decking and in-situ light weight 7concrete, the "Fast Track
Slim floor",
- rolled profiles with deep decking and in-situ normal weight concrete, the so-called
"Additional Slab".
The slim floor based on the integrated floor beam IFB was developed by PROFIL-ARBED
[2,3,4,5] since 1991. It represents a part of a dry construction system based on steel columns,
one directional steel beams and prestressed hollow core units (see figure 3).
1 Steel column
2 Bolts
3 End plate welded to
beam
4. Steel beam IFB
A3» 5 Hollow core unit
section,
- various steel profiles and steel grades with a yield point of 235 up to 460 N/mm2 are
available,
- the integrated floor beam may be designed as simply supported or as a continuous beam,
- the void between hollow core units and the web of the steel profile is filled with in-situ
concrete,
- it is strongly recommended to supply an in-situ topping with reinforced concrete in order to
improve the transversal load distribution between hollow core elements. This structural
topping is also increasing the shear resistance of hollow core units [6],
- a similar effect is produced by filling the cores at each end of the slab units,
- to ensure sufficient shear resistance to hollow core units during exposure to fire, it is
recommended to provide complementary stirrups [7],
- complementary reinforcing bars, parallel to the steel beam, guarantee its ISO-fire resistance
up to 120 minutes, whereas the lower steel flange may remain unprotected [8].
56 SLIM FLOOR CONSTRUCTION: WHY?
As a consequence the floor slab, supported on the bottom flange of the steel beam, presents
from beneath a straight and clean ceiling surface.
This slim floor based on rolled profiles with various shapes was developed by STAHL &
VERBUNDBAU GmbH [2]. It represents a part of a construction system based on steel or
composite columns, steel beams and in-situ reinforced concrete slab (see figure 5).
- the steel beam is composed of rolled profiles or half sections with or without a steel plate
welded to the free web end,
- various steel profiles and steel grades with a yield point of 235 up to 460 N/mm2 are
available,
- the steel beams may be designed as simply supported or as continuous beams,
- during in-situ concreting of the reinforced slabs, concrete may be supported either by a
timber shutter, prefabricated filigree concrete decks or profiled steel sheets,
- the steel profile, encased in concrete with the exception of the lower steel flange, behaves
as a composite cross-section, also thanks to the shear studs welded on the profile,
- this cross-section may be designed according to ENV1994-1-1 for normal temperature
conditions [9] and following ENV1994-1-2 for fire exposure [8]; the lower steel flange
may remain unprotected up to the ISO requirement of R120,
- this slim floor cast in-situ has an assured monolithic behaviour, useful in certain design
situations when f.i. large openings have been foreseen in the floor or for special loadings
due f.i. to earthquake shocks
This slim floor, similarly as the previous one, presents from beneath a straight and clean
ceiling surface.
One-directional
spaning
Bi-directional
spaning
~/f\- SS sections
Fig. 6 S+V cross sections based either on a timber shutter (a), prefabricated filigree concrete
decks (b) or profiled steel sheets (c)
58 SLIM FLOOR CONSTRUCTION: WHY?
This slim floor based on the SFB beam was developed by BRITISH STEEL [2,3,4]. The
corresponding construction system is based on any steel or composite column type, steel
beams with enlarged lower flange supporting the deep decking, and in-situ reinforced light
weight concrete (see figure 7).
- the steel beam is composed of rolled profiles reinforced on the lower flange by an enlarged
and welded steel plate,
- various steel sections and steel grades up to a yield point of 355 N/mm2 are used,
- reinforcing bars are foreseen inside and on top of the deep decking,
- light weight concrete cast over the whole floor constitutes a concrete lattice in which the
steel beam is fully integrated,
- composite actions is activated by welding shear studs on top of the steel beams and by the
embossments of the deep decking,
- large openings may be realized through the webs of SFB beams, permitting to integrate all
types of pipes beneath the deep decking but still localized within the height of the slim
floor,
- this slim floor system may be designed according to ENV1994-1-1 and ENV1994-1-2
[9,8]; specific design rules have been confirmed by full scale testing in the Netherlands as
well for normal temperature conditions as for fire exposure [10,11,12,13].
This slim floor based on rolled profiles of any shape was developed by HOESCH
SIEGERL AND WERKE GmbH [2]. The corresponding construction system is based on any
steel or composite column type, steel or composite beams fabricated out of rolled profiles
J.-B. SCHLEICH 59
with shear studs and bearing blocks welded on to the upper flanges which are supporting the
deep decking, and in-situ reinforced normal weight concrete (see figure 8).
In fact this "Additional Slab" is not, generally spoken, a slim floor, as the steel beam may be
higher than the deep decking. This floor system is characterized by the following aspects:
- the steel beam is composed of rolled profiles of any shape and height and any yield point
up to 460 N/mm2,
- reinforcing bars are foreseen inside and on top of the deep decking, supported by special
bearing blocks welded on top of the steel beams,
- normal weight concrete cast over the whole floor constitutes a concrete lattice,
- composite action is guaranteed by welding shear studs on top of the steel beams and by the
embossments of the steel decking,
- this floor system may be designed according to ENV 1994-1-1 and ENV 1994-1-2 [9,8];
the ISO-fire requirement R120 may be fulfilled.
Design recommendations may be split in general and specific principles and rules, function of
the various slim floor systems described in chapter 2, whereas general recommendations
should be issued concerning global fire safety considerations.
It may be assumed that these requirements are fulfilled for the bulk of slim floor systems
given before and designed according to ENV 1994-1-1 and ENV 1994-1-2 [9,8]
The following special features shall be considered for normal temperature conditions and for
the fire situation
Shear resistance of hollow core units, HCU, should be established as a function of the
flexibility of the corresponding supports Investigations and research, undertaken in Nordic
countries since 1990, clearly indicate that the shear resistance of HCU elements decreases
when the flexibility of the supporting beams increases [14,15]
Furthermore it is considered that in-situ structural topping may contribute m a significant way
to improve the shear resistance of hollow core units [4,6]
Finally it would be useful to study the "General Shear Design Method" established by
COLLINS & MITCHELL [16] This so-called unified method is presented for the shear
design of both prestressed concrete members and nonprestressed concrete members The
method can treat members subjected to axial tension or axial compression and treats members
with and without web reinforcement In fact the normal cracked concrete contribution V cRd is
added to the reinforcement contribution V sRd and to the effect of prestressing V £d, so that the
shear resistance follows from-
In chapter 2 7 3 2 (11) of the FIP RECOMMENDATIONS [7] it is written that "The shear
capacity of hollow core units during exposure to fire is affected by (a) the increase of the
tensile stresses in the webs due to the temperature gradient etc " and in chapter 4 5, Rule (4)
of ENV 1992-1-2 [18] it is declared that " special consideration should be given when
tensile stresses are caused by non-linear temperature distributions (e g voided slabs, thick
beams, etc A reduction in shear strength should be taken equivalent to these tensile
stresses" Unfortunately no quantified design rule is given permitting to check the shear
resistance in the fire situation, VRd 0,of hollow core units
For that reason several slim floor fire tests were performed in France and in Switzerland
[19,20], clearly testifying the beneficial effect when providing a structural topping and
stirrups (see figure 4) Furthermore a first attempt was done to quantify, in connection to finite
element calculations, the shear resistance of HCU's [21,22] In that respect the procedure
explained in [16] was adapted to the fire situation
Consequently it is recognized, that slim floors based on HCU's constitute a quite complex
construction system with steel, concrete and prestressed components. This situation requires
multi-disciplinary research based on a scientifically sound background and supported by a set
of realistic fire tests.
Developments under way since 1985 in the field of fire safety engineering permit nowadays to
take advantage of the new structural fire design standards of CEN. Indeed global structural
analysis may be performed on the entire structure, the combination rule for actions during fire
shall be considered and the evolution of a natural fire may be assumed This realistic
structural fire design may even be improved by considering the effect of active fire safety and
fire fighting measures. Consequently this produces real safety for people, f.i. by adequate
partitioning, by sufficient escape routes and proper smoke venting [23,24,25].
Our intention is not to discuss principles of architecture, like those concerning the so-called
architecture of freedom. But it is worthwhile to remember MIES VAN DER ROHE who said:
"A clear structure is the backbone in it all and makes the free plan possible".
In order to illustrate architectural advantages of slim floors, the following buildings are
shortly described hereafter.
This ENPC high school building with a total clear surface of 31.200 m2 has been erected in
the "City Descartes" in the East of Paris. The bulk of floors is conceived as IFB slim floors
[26],
- composite columns with partially encased steel profiles (HE300AA), designed according to
ENV 1994-1-2, were erected in elements covering two levels. The grooved concrete and
the steel flanges are remaining visible (see figure 9),
- the integrated floor beams IFB are composed of the half of the rolled profile HP400 on top
of which the steel plate 140x40 mm was welded. Furthermore shear studs were welded to
this upper flange in order to create composite beams (see figure 10),
- a structural reinforced topping of 8 cm thickness was put on top of the hollow core units
HCU, so that the total floor thickness is 25 cm.
62 SLIM FLOOR CONSTRUCTION: WHY?
This AOB office building is composed of two wings with nine levels and a total volume of
61.000 m3. The global fire safety has been designed according to 3.3, so that the steel
structure remains visible in the entire building. Floors are conceived as IFB slim floors
[23,24],
5. Conclusion
Multi-disciplinary and correctly guided research is needed to develop fully transparent design
rules in relation to slim floors. In that case slim floors in general, and those based on hollow
core prestressed units in particular, may be considered as the most promising construction
models in the future.
6. Bibliography
[1] WALLIN Lars; Technical and economic advantages of steel construction, building costs
and overall economy. ECCS publication, Brussels, 1978.
[2] SUTTROP W.; Geschossbau in Stahl, Flachdecken Systeme. Bauberatung Stahl,
Dokumentation 605, Düsseldorf, 1996.
[3] AC1-ECCS; Design guide for slim floors with built-in beams. ECCS publication N° 83.
Brussels, 1995.
[4] BODE H., SEDLACEK G.; Composite Action in Slim Floor Systems. Conference
"COMPOSITE CONSTRUCTION III", Irsee (G.), 1996.
[5] BODE H., SEDLACEK G.; Untersuchung des Tragverhaltens bei Flachdecken-
Systemen. SAES-P261, Schlussbericht, Düsseldorf, 1997.
[6] CEN TC229; PrENl 168, Precast prestressed hollow core elements. CEN Central
Secretariat, Brussels, fourteenth draft 1996.
[7] FIP Recommendations; Precast prestressed hollow core floors. Thomas Telford, ISBN
0 7277 13752, London, 1988.
[8] CEN TC250; ENV 1994-1-2, Eurocode 4 - Design of composite steel and concrete
structures, Part 1.2, Structural Fire Design. CEN Central Secretariat, Brussels, D.A.V.
30.10.1994.
64 SLIM FLOOR CONSTRUCTION: WHY?
[9] CEN TC250; ENV 1994-1-1, Eurococe 4 - Design of composite steel and concrete
structures, Part 1.1, General Rules and Rules for Buildings. CEN Central Secretariat,
Brussels, 1992.
[10] BREKELMANS J.W., DANIELS B.J., VAN HOVE B.M., KOUKKARIH.; Analysis
of the combined vertical and horizontal shear tests on deep deck composite slabs. ECSC
Research 7210-SA/621, TNO Report 95-CON-R1148, Delft, 1995.
[11] BREKELMANS J.W., DANIELS B.J., SCHUURMAN R.G.; Analysis of the vertical
load tests on deep deck composite slabs. ECSC Research 7210-S A/621, TNO Report
96-CON-R1147, Delft, 1996.
[12] FELLINGER J.H., BREKELMANS J.W., VAN DE HAAR P.W., TWILT L.; Fire test
on a two span integrated shallow floor system. ECSC Research 7210-SA/621, TNO
Report 95-CVB-R0708, Test Data, Delft, 1995.
[13] FELLINGER J.H., VAN DE HAAR P.W., TWILT L.; Fire test on a two span integrated
shallow floor system. ECSC Research 7210-SA/621, TNO Report 95-CVB-R0765,
Numerical Simulations, Delft, 1995.
[14] PAJARI M., YANG L.; Shear capacity of hollow core slabs on flexible supports. VTT
Research Notes 1587, Espoo (FI), 1994.
[15] LESKELÄ M.V.; Shear flow calculation for slim-type composite beams supporting
hollow-core slabs. Proceedings of the fourth international conference of ASCCS,
Kosice-Slovak Republic, 1994.
[16] COLLINS M.P., MITCHELL D„ ADEBAR P., VECCHIO F.J.; A general shear design
method. AC1 Structural Journal, Vol 93 N° 1, Detroit, 1996.
[17] DIN 4102, Teil 4; Brandverhalten von Baustoffen und Bauteilen. BeuthVerlag GmbH,
Berlin, 1994.
[18] CEN TC250; ENV 1992-1-2, Eurocode 2 - Design of concrete structures, Part 1.2,
Structural Fire Design. CEN Central Secretariat, Brussels, 1995.
[19] FRECHET O., KRUPPA J.; Essais de résistance au feu des planchers avec dalles
alvéolées et poutres à talon métalliques. CTICM, Rapports d'essais N° 93-G-127,
N° 95-E-467, N° 95-E-533, N° 96-U349, N° 96-U350, Maizières-lès-Metz, 1993-1996.
[20] FONT ANA M., BORGOGNO W.; Versuche zum Tragverhalten von Betonhohlplatten
mit flexibler Auflagerung bei Raumtemperaturen und Normbrandbedingungen. IBK
Bericht Nr. 219, ETHZ, ISBN 3-7643-5467-4, ZÜRICH, Mai 1996.
[21] SCHLEICH J.B., CAJOT L.G.; Calcul de l'effort tranchant résistant de hourdis
précontraints en cas d'incendie. PROFILARBED-Recherches, Luxembourg, 16.04.1996.
[22] DOTREPPE J.C1.; Note d'évaluation du document "Calcul de l'effort tranchant résistant
de hourdis précontraints en cas d'incendie". Université de Liège, Service Ponts et
Charpentes, 04.03.1996.
[23] SCHLEICH J.B.; Brandsichere Stahlbauten-Harmonisierung von Entwurfsmethoden.
ALLIANZ Report 3/96, ISSN 0943-4569, München, 1996.
[24] SCHLEICH J.B.; Der unsichtbare Brandschutz. Verlag Wiederspahn, BAUKULTUR
6.96, ISSN 0722-3099, Wiesbaden, 1996.
[25] KINDMANN R., SCHLEICH J.B., SCHWEPPE H., CAJOT L.G.; Verallgemeinertes
Sicherheitskonzept für die Brandschutzbemessung. PROFILARBED-Recherches,
Luxemburg, März 1997.
[26] PHILIPPON P., VILCOCQ P.; Nouvelle Ecole des Ponts et Chaussées et des Sciences
Géographiques. Les cahiers de l'APK, N° 14, Paris, octobre 1996.
65
Jean-Marie CREMER
Civil Engineer
Engineering Office Greisch
Liège, Belgium
Summary
The construction of composite bridges is rising and numerous publications give an outline of the
success met by this type of bridge. Everywhere in the world but with developments differing
according to the countries, innovations and improvements come up. Girder and truss bridges, arch
and bowstring bridges and, of course, cable-stayed bridges, all types of bridges allow to benefit
from the steel concrete composite technique to design high quality bridges, so much in terms of
economy as of aesthetics and durability.
1. Introduction
For the last few years, the construction of composite bridges has known a growing success
worldwide, with developments of course differing according to the countries.
This wide success can be explained by the improvements and innovations that occurred these last
two decades in many industrialized countries.
The innovations are also due to the abilities of the composite structures to solve, with elegance,
lightness and economy, problems that seemed unadapted to full concrete bridges or too expensive
for steel bridges with an orthotropic slab.
Less spectacular innovations, although their importance will appear as the years go by, are related
to the improvement of the execution, to the care in the design of the details, to the quality of the
materials and, most of all, to the aesthetical quality of the bridges.
66 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
Presently, there is no doubt left that the innovations have first to be searched to increase the overall
quality of the bridges : technical quality and durability, but also aesthetical quality and care of the
environmental integration.
The future of the composite bridges is full of promises; beyond its numerous qualities (lightness,
fast and secure realization, weak sensitivity to differential settlements), they have a henceforth
major advantage, they are durable. The promises are even greater if the effort is made to design
them in accordance with their specificity and not as bad copies of steel or concrete bridges.
The evolution of the calculation methods on a theoretic level (progress in the stability, torsion and
fatigue domains) and the computer means allows to simplify the structure. The number of
stabilizing elements, such as the stiffeners, cross beams and wind-bracings, decreases and they are
sometimes even completely suppressed.
This tendency towards a greater simplicity is even reinforced by economical requirements. Facing
the increasing costs of the manpower and the rather stagnant cost of the steel, the bridge builders
add even more to this simplicity to decrease the working time.
The evolution of the knowledges, the calculation means and the manpower costs cannot explain by
itself the totality of this change. Twenty or twenty-five years ago, the first criterion to choose a
variant for a bridge, either a composite or prestressed concrete one, was its price; other criterions,
such as durability or aesthetics, were often further relegated.
This order in the appreciation criterions has led to quite a few disappointments. Some bridges, after
a few years only, had to be heavily repaired. Other ones, rather inaesthetical, have been severely
criticized among the population. Therefore, the hierarchy of these criterions has evolved. Presently,
we think that price, durability and aesthetics have the same importance.
2. The principles
The initial justification of the composite structure is the idea to use at best the own qualities of the
materials :
the concrete, for its good compressive strength and its low cost per volume unit,
the steel, for its high tensile strength.
The typical example is obviously the cable-stayed bridge. We all know that the pylon and the deck
of modem multi-hangers bridges are mainly compressed, along with weak bending moments, as the
hangers are obviously only tense.
The cross-section of this same box-girder cable-stayed bridge with a central layer of hangers, shows
steel inner tense stmts and a concrete compressed box-girder. These are the leading principle for the
Wandre and Ben-Ahin bridges in Belgium, for the Elorn bridge in France...
The second justification, that explains the quality of the composite beam, comes from an
economical analysis of the various constituents of the resistant structure of a bridge deck. The price
of a full metallic bridge is burdened by the exorbitant cost of the orthotropic slab that, in spite of a
very light weight, asks a very important working time.
J.-M. CREMER 67
On the other hand, in concrete bridges, the webs of the girders are usually too thick on a matter of
resistance, as the minimal thickness is usually bounded to technologic requirements such as the
presence of prestressing cables and the concreting operations.
The third justification is the possibility to use steel for its lightness in areas where the weight has a
baleful influence and to place the concrete where the dead load of the structure, either has little
influence, or is favorable to the stability. Numerous examples have been built on basis of this
principle these last years, the Chevire and del Milenario bridges and, of course, the most famous
one, the Normandy cable-stayed bridge.
Unfortunately, many of those continuous composite structures have insufficient reinforcement steel
and know longevity problems, particularly in countries where de-icing salts are used. The problems
related to the durability of the deck slab are similar to the ones of the full concrete bridges; they
need a special care and a extremely compact concrete.
4. The bridges
The different kinds of bridges will reviewed about and, in each case, the large variety of solutions
able to be brought by the composite construction will be pointed out.
The last two decades have shown a wide simplification of the composite structures, for the full web
girder and truss girder bridges.
The traditional bridge (fig. 1) has two longitudinal girders, bounded to the concrete deck slab by
shear connectors. A limited number of cross beams, welded to the vertical stiffeners, binds the main
girders. This type of bridge can be used both for road and railway traffic and allows spans over
100 m.
68 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
For wide bridges, the present tendency is to keep as much as possible the by-girder system, either
by transversal prestressing in the deck slab or by the use of more cross beams closer to each
together and a cantilever external slab.
By-girder bridges with a lower deck can also be used for both road and railway traffic as this High-
Speed Rail bridge in France or this magnificent small Romanche bridge due to the engineer
Tonello.
For spans no longer than 60 m, a even simpler design than the by-girder system is made of two
small rectangular box-girders with no stiffener, no cross beam, no transverse member, but only a
diaphragm at the bearings on pile. This solution, due to its simplicity and its small surface to be
painted, is very economical to built and will also have low maintenance costs.
The comparison of the lengths of the lines of welding, the transversal butt joints and the surfaces to
be painted easily shows the interest of this design. It is obvious that the most recent methods to
check the stability of the plates have to be used to come up to such a simplicity. Some national
regulations, as the Swiss and Belgian standards, allow the evolution of the calculation methods.
Other ones, more conservative, do not allow it yet. The Bridges Eurocodes are indeed still to come.
As much as possible, the box girders with stiffened bottom are avoided for they are less
economical. Their high torsonial rigidity is an advantage, that makes them necessary as the plane
curvature is high or for long spans.
Box-girder bridges allow to cross long spans. In Spain, the nice bridge of Professor Martinez-
Calcon, with a 180 m long span, has been a world record. I don't think to be mistaken as I affirm
that the Canori bridge in Venezuela, designed by the Leonard office and later presented by
Mr. Saul, holds the present word record with a 213 m long main span. Both these bridges have a
bottom of box-girder that is metallic in the zone with positive bending moments and metallic
strengthened with concrete in the zone with negative bending moments on bearings.
For many years, researches have been undertaken, on the one hand, by the designers of metallic
bridges to get their bridges more economical and, on the other hand, to the concrete supporters to
lighten their structures.
Truss beams bearing a concrete deck slab constitute an elegant alternative to the classical by-girder
system. Requiring higher beams but offering good transparency, these structures ask for exactness
and much sobriety in the design of the details and connections.
As examples, the small Cruchten bridge, built with higher elastic limit Histar steel and the new
Blois bridge, with a truss of variable height, for which the lightness of the structure has
unfortunately to suffer from a lack of delicacy in the design of the bearings.
The most beautiful and impressive examples of composite truss bridges, are railway bridges. The
most famous one is the Nautenbach bridge in Germany, with a 208 m long span. Others examples
exist too in Austria. All these bridges will be presented during this session.
J.-M. CREMER 69
Another bridge, with shorter spans but very elegant, is the Arc bridge in France, presently under
construction. The great Spanish engineer Torroja has already designed quite a few of these
bridges...
Another particular truss bridge is the Dreirosen bridge in Switzerland, with two roadway levels.
This idea is more and more largely used, even for very long cable-stayed bridges.
A transition can be made from the truss girder bridges to the arch bridges with two examples of
original bridges :
the Antrenas bridge in France, half arch, half spatial tubular truss,
another bridge in Czech Republic, with a metallic tubular arch filled with concrete.
During these last two decades, the long arch bridges have been built in concrete, which seems quite
normal for an mainly compressed structure. However, a few particular bridges with interesting
characteristics show the composite structure as the solution of the future.
In Italy, a small steel-concrete composite arch bridge shows how easy can be the construction of
this type of structure.
On a flattened concrete arch, to limit the stresses in the concrete, a light composite deck is
launched. That is the Roche Bernard bridge in France.
In Spain, several metallic arch bridges with a composite deck are really beautiful. The Ricobayo
bridge has to be pointed out.
In Belgium, an arch bridge with 270 m long span (fig. 2) has a particularly small height for the arch
elements (1/100 of the span); this gives a very light and transparent structure. The use of thick
plates and the small cross-section of the box-girders allow to avoid stiffeners and wind-bracing and,
therefore, to realize an economical bridge.
We firmly believe that this type of arch, much lighter and less sensible to the quality of the
foundation soil and to the seismic effects, has a real future towards similar concrete bridges.
70 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
The most marvelous examples come from China, which is the topic of another paper. During these
last years, the Chinese engineers have built several arch bridges whose metallic tubes were filled
with concrete. The longest one, word record with a 420 m long span, is the Wanxian Yangtze
River bridge, presently under construction. In this last case, the metallic tubular arched is coated
with concrete.
These bridges mainly comprise a compressed chord, the arch, and a tense chord, the deck. They
allow numerous variants contravening to the base principles of the composite structures. Most
often, the arch is metallic and the deck is either in concrete or steel-concrete composite.
An example of this kind of bridge with a prestressed concrete slab, the Chanxhe bridge, is under
construction in Belgium. Another one is the Ronquoz bridge in Switzerland.
A concrete arch showed unacceptable fissures, the prestressing of the tense deck was insufficient
and the bearing outfits were too weak. The replacement of the concrete arch by a much lighter
metallic arch was the solution to all these problems.
Bowstring bridges, with an approximately 150 m long span an a steel-concrete composite deck are
very elegant solutions to cross a river or a canal, as the Hermalle (fig. 3) and Milsaucy bridges
in Belgium.
On the Mediterranean High Speed Rail track, the National French Railway Company is presently
building composite bowstring bridges, Donzere and Mornas-Mondragon, particularly well
adapted to the fatigue resistance and dynamic behavior requirements.
Once again, we have to point out the Chinese bridges with cylindrical tubular arches filled with
concrete et spans up to 200 m.
J.-M. CREMER 71
The cable-stayed bridges are without any doubt the most fashionable modem bridges at the present
time. Here too, the composite structures are interesting, both in the overall design and to solve
particular problems.
Perhaps is it useful to remind of the Annacis bridge in Canada (fig. 4), composite cable-stayed
bridge with a 467 m span, absolute word record for cable-stayed bridges for many years. Presently,
the Yangpu bridge in Shanghai (fig. 5) has a 603 m long span with a similar design, two
longitudinal beams supported by two layers of stays et bounded together by cross beams bearing a
reinforced concrete slab. The Normandy bridge, present world record for all categories cable-
stayed bridges, is also a composite structure, as its central span is metallic and the balancing spans
are made of prestressed concrete.
The metallic cable-stayed bridges, less expensive than the suspended ones, are now believed to
reach 1500 m spans between pylons. The experience shows that the full metal deck with orthotropic
slab is mainly dedicated to the very long spans. Up to 700 m or maybe even 800 m, the composite
structures can be reasonably considered as the optimal solution.
The bridges built these last few years show the trend of those stmctures. The Annacis and
Shangai bridges are typical examples : the classical "composite cable-stayed bridges". This
denomination is usually understood as a cable-stayed structure with metallic beams and a
concrete deck slab, as pointed out earlier. The objective is of course to use at best the best
characteristics of each material.
The steel allow the préfabrication in workshop of small beam elements, quite light and very
resistant, with precise dimensions and easy to assemble. This préfabrication occurs as the
foundations and the pylons of the structure are built on site, which reduces the construction time
towards other techniques.
72 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
On the other hand, those classical composite bridges rely, more the other lull metal or full
concrete decks, on the construction method.
Indeed, for a classical composite section, in a very simplified schema of longitudinal behavior,
the shear forces and the bending moments are handled by the metallic beam and the concrete
deck slab takes the big compressive forces due to the inclination of the stays. That is why the
composite bridges built by the cantilever method are designed following the by-girder type with
two layers of hangers.
In that case of two layers of hangers, the stays can be fixed either directly on the longitudinal
beams, where the balance of the horizontal and vertical forces is realized, as for the Annacis and
Yangpu bridges or on very rigid transversal beams that transmit the forces of the stays to the
main beams, as for the new Severn bridge in England.
The construction of the deck on site has to be scheduled very precisely. The metallic beams are
placed in balanced cantilever. For each beam element, a hanger has to be placed to bear it. The
transversal elements, the cross beams, complete the metallic structure.
The construction of the deck slab, either in precast elements or concreted on site, will follow
regularly the assembly of the metal and actually, as close as possible. Let us recall that the
concrete slab handles the major part of the compressive forces in the deck. The slab has therefore
imperiously to be efficiently connected as early as possible to the main beams, which, as said
earlier, are bounded to the stays.
The most delicate things to be treated in this kind of structure are the following ones :
the steel-concrete connection, of course,
the anchorage of the stays and the transmission of the vertical forces to the deck slab,
the effects of shrinkage and creep,
the problems of wind stability of this composite deck with external girders, whose
aerodynamic profile is not really good.
-H-
I I \
The principles that I just described, the classical transversal cross-section and the construction
method, have usually been used to the major existing bridges (listed below), with just a few
small differences
The Baytown bridge in the United States of America (fig 6) has the particularity to be
constituted of two separate decks and two bounded pylons, which insures an obvious transversal
stability.
Three bridges have to be pointed out, as they show noticeable differences towards the classical
schema described earlier
the Rama VIII bridge in Bangkok, with a single Y-inverted pylon, two layers of hangers
and a wide composite girder-box,
the Karnali bridge in Nepal, with a 325 m long main span, a single pylon, two layers of
hangers and a composite truss deck,
the Ting Kau bridge in Hong-Kong (fig 7), with three pylons, two main spans, four
layers of hangers, used for the suspension and for the stabilization of the central pylon
Main Tower
The present tendency to use thinner and thinner transversal cross-sections, needing two or even
four layers of hangers, comes from the following establishment. the bending solicitations in the
deck are proportional to the proper rigidity of the deck if the multiple stays are regularly
disposed on the length of the bndge. A direct consequence of the thinning down of the deck is
obviously the decrease of the surface exposed to the wind
However, the aerolastic stability requires a significant torsional rigidity as the span reaches
500 m, even with a double layer of hangers
If, during these last years, a general tendency has led the designers to thin down the decks thanks
to a better knowledge of their behavior, thick decks have to be designed to solve specific
problems as for bndges for heavily loaded railway tracks or roadways on two levels. The thin
decks are subject to deformations that are incompatible with the railway traffic requirements
That is why truss or high inertia box-girder decks are used
74 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
That is the case for decks with two levels of traffic, as the new Kap Shui Mun bridge in
Hong Kong, the Higashi Kobe bridge in Japan and the new Oresund bridge In Sweden,
presently under construction.
The developments towards high performance, as well for steel as for concrete, will soon allow to
build composite bridges with a main span probably over 800 m long. Anyway, it will still be
necessary to adapt and to improve the aerodynamic characteristics of the decks.
Among those long bridges that impress us by their exceptional dimensions and the achievements
that they required, the last ones show some originality. We believe that the composite
construction offers a wide diversity of solutions, already pointed out in the last presented bridges.
These are smaller bridges, not built with the cantilever method and therefore offering more
liberty of design, as for example the Seyselle bridge in France, the Saint-Maurice bridge in
Switzerland and several bridges in Finland.
Another bridge, soon to be built, the Kortrijk bridge, shows that the combination of two
materials can solve difficult problems with much elegance. This small bridge crossing a river in
the middle of the town, needed a very thin deck for obvious reasons of navigation clearance and
grade profile of the crossing road. Furthermore, for aesthetical reasons and cross-roads
congestion problems, a single central layer of hangers was imperative, providing to solve the
problem of the balancing stays of the pylon. This has been realized with this stand-shaped
metallic pylon, restrained in the prestressed concrete deck.
The Alzette bridge in Luxembourg (fig. 8) is a symmetrical cable-stayed structure with a single
central pylon and one layer of semi-radial hangers. An important characteristic is the horizontal
curvature, rather uncommon for a cable-stayed bridge. The curvature, measured at the longitudinal
axis, has a radius of 1750 m.
The steel concrete composite deck is composed of two steel trapezoidal box girders with bracing
frames and a reinforced concrete deck with a variable thickness. There are no longitudinal stiffener
in the webs and in the bottom of the box-girder.
This composite deck, lighter than a concrete box-girder and more economical than an orthotropic
slab, forced itself for the transversal solicitations on the pylon, due to the plane curvature, required
a light structure.
J.-M. CREMER 75
Another very interesting curved cable-stayed bridge, the Arena viaduct in Spain, has six pylons
and seven cable-stayed spans.
Last example of a particular composite cable-stayed bridge, but it has not been realized. All the
movable bridges were, up to not so long ago, metallic structures. However, some movable
composite or even totally prestressed concrete bridges are built.
It is absolutely obvious that the extra-weight of a movable bridge has a direct financial impact on
the mechanisms and in the electric power consumption. However, for long spans, the wind effects
become preponderant in the design of the mechanisms, as well for the swing bridges as for the
bascule bridges. That is why the importance of the various parameters of the economical balance of
the project evolves and, more and more, the orthotropic slab will also give way to composite
solutions with concrete deck slab.
Some interesting suspended bridges with a small main span have also to be pointed out. The deck,
either composite or in concrete, combined to the metallic suspension cables, allows to build high
quality economical bridges. Two example : the footbridge over the Neckar river in Germany and
the Vranov Lake bridge in Czech Republic (fig. 9).
5. Conclusion
The impressive quantity and the variety of composite bridges having been built during the past two
decades show enough how this kind of bridge has fine prospects before it.
With the increasing length of the spans, the steel continues to be the indispensable material, as the
composite structures, in spite of some reticences, compete more and more the prestressed concrete
bridges for medium and smaller spans.
These composite structures will be all the more competitive as the regulations will allow us to
design simple structures, limiting or avoiding the expensive stiffeners.
We cannot forget that the composite construction requires from the engineer a good knowledge of
both materials, without apriorism but with the constant care to question established ideas.
If, in addition, the engineer disposes of performing calculation means and if the regulations, too
often retrograde, allow him to reap advantage of them, he will be able to design economical,
efficient and high quality bridges.
The harmony of a composite construction is expressed at best with this footbridge in Japan, the
Inachus bridge (fig. 10).
76 VARIETY OF COMPOSITE BRIDGE CONSTRUCTION
References
Summary
This paper describes the situation of composite bridge construction in Austria illustrated by some
representative examples. The replacement of existing structures by composite bridges predominates
over newly constructed bridges. In Vienna, Nordbrücke, one of the most frequented bridges
of Austria, was successfully upgraded. The work performed is described. On the basis of the
dynamic characteristics of the structures, innovative methods of inspection and assessment are
applied in Austria, which are also presented.
1. Introduction
Due to its topography, Austria traditionally is a country of bridge builders. The motorway
network in the Alps, including the three north-south crossings of Brenner Motorway, Tauem
Motorway and Pyhrn Motorway, results in a high percentage of bridges in the overall road
network. Since the beginnings of reinforced-concrete bridge construction, it was tried to combine
steel and concrete in a way to ensure that both materials fulfil the functions best suited to their
properties in a bridge structure. This means that the compressive strength of concrete as well as
the tensile strength and compressive strength of steel are used specifically, while their interaction
is safeguarded by efficient and permanent doweling. In spite of the clear technical advantages of
composite steel bridges, this construction method was seldom applied in Austria, as heavy
-
competition
-
similar to the conditions in our neighbouring countries made co-operation between
steel and concrete construction companies difficult.
The time has not yet come when composite bridges offer advantages in economic terms. However,
they meet today's higher quality standards in modem bridge construction and succeed over
other construction methods, in particular, in case of difficult basic conditions. Composite
construction frequently offers absolutely new design options, such as lower construction depths or
low dead weights for medium-span bridges, and often allows for good solutions under difficult
external conditions. In this field, composite construction can absolutely compete with prestressed
concrete construction, while all-steel construction is only advantageous under extreme basic
conditions in Austria. Based on the development on the European market which shows a strong
78 COMPOSITE BRIDGES IN AUSTRIA
• Composite construction is a simple bridge construction method which does not require
heavy temporary scaffolding during building so that the terrain below the bridge need not
be used.
• Another advantage of mounting is that it takes very little time and, thus, projects can be
completed earlier.
• Composite structures can be modified easily which has frequently resulted in substantial
advantages when bridges had to be reinforced or widened.
• Concrete decks damaged in most cases due to the action of de-icing salt can be easily
repaired, removed or replaced. Likewise, advantages are found when analysing the costs of
composite bridges throughout their life. In particular, steel recycling will contribute to
covering demolition costs.
• Demolition work is less difficult than in case of prestressed concrete structures.
Due to intensified competition resulting from the economic situation in Europe, steel construction
has to attempt to maintain its competitiveness by cutting costs. In this context, potentials are
opened up by the following:
• The advantages of composite girders, mainly as regards shear strength, have to be fully
utilised. Attention, however, has to be paid to carefully detailing the connections in order
to prevent damage which already occurred in the past.
• A more economical design of the reinforced web of main girders seems to be possible. By
slightly raising the amount of material used, it is possible to reduce high labour costs.
• An absolutely essential advantage results from the application of concrete decks in wide-
span composite girder bridges. The simple principle of using the best suited material at the
right place can still be refined, as is shown by some examples.
• The utilisation of even more far-reaching innovations, such as detachable bonds, timed
shifting in mounting and the use of prestressed concrete components, let us expect
numerous innovative solutions for composite bridges.
The great majority of our composite bridges were built during the extension of the high-ranking
road network in the 1960's and 1970's. This development started here, near Innsbruck, when the
Inntal and Brenner Motorway was extended, and finally spread to the east in the course of the
construction of other Alpine crossings and of bridges across the River Danube. Thus, we will
start our presentation of examples in the west, near Innsbruck, moving east towards Vienna.
Along the Brenner Motorway, there are numerous examples for the successful application of
composite bridges. In the course of rehabilitation works carried out after thirty years of extreme
utilisation, composite structures are executed again and again. For example, Steinbruch Bridge, a
prestressed concrete bridge with a clear span of 5 x 20 m located at a hillside slope involving the
risk of slides, was replaced by a composite bridge with a clear span of 100 m.
Miezener Bridge was widened without the main girders having to be reinforced. Just the wind
bracings and one bridge bearing had to be reinforced. The lanes of Gschnitztal Bridge were
W. HUFNAGEL, P. BIBERSCHICK, H. WENZEL 79
widened in the years from 1986 to 1988 since operation demanded an additional lay-by for safety
reasons. As a result, the old reinforced-concrete deck was removed and replaced by a new,
thinner deck with limited transverse prestressing.
In the first place, the higher traffic load which results from the widening of the bridge lane would
normally lead to an overload for the steel load-bearing system. Given the simplified calculation
methods applied at that time, which were on the safe side without exception, and given the
higher traffic loads in relation to the bridge width, which were stipulated at that time, it was
possible to
prove with today's sophisticated calculation methods that the bearing capacity is safe for
the reduced traffic load permitted in relation to the increased bridge width without having to
reinforce the steel load-bearing
system. As these high valley bridges did not have any inspection
wagons, such wagons were installed in the two superstructures before widening was started.
They were screwed to each of the outer main girders by means of one vertical rail and two
horizontal rails using brackets. The vehicles were designed in such a
way that it was possible to use
them also as mounting aids. The removal and the replacement of the deck was performed from
this mounting wagon with the aid of a temporarily mounted rail.
The entire deck was removed and finally produced new section by section. From the block
dowels, the loops were removed and replaced by headed shear connectors. The new reinforced-
concrete deck was fully doweled also in the support zones. At the steel superstructure, only the
lower wind bracings had to be reinforced by welding on cantilever segments. In order to improve
fatigue strength, reinforcing butt straps were installed. Transverse prestressing of the concrete
slab was performed after completion of steel construction work. In parallel to widening the
bridge, its bearings and expansion joints were rehabilitated. Bearings maladjusted in construction
were corrected, thus resolving the problems observed in the behaviour of the supporting
structure. The examples described show that the rehabilitation and adaptation of all-steel bridges
is the most cost-efficient method by far. The costs of rehabilitation amounted to an average of
US $ 850 per square meter.
-
In the second half of the 1960's, „Pilzbrücken" slim concrete structures with joints in every
-
span with a total of approx. 56,000 m2 were constructed in the motorway section Innsbruck
Brenner. The supporting structure of this bridge type is produced by combining span-sized,
-
single-support structures. „Halbpilzbrücken" with a span width of 15 m were constructed for the
lane facing the valley and „Vollpilzbrücken" with a span width of 30 m for the entire cross-
section of the motorway. Due to leakage in the numerous system-inherent pin joints and due to
the action of the de-icing salt, corrosion heavily damaged the reinforcement and the prestressing
anchors which required thorough rehabilitation. Brenner Autobahn AG invited several consulting
engineers to participate in a design competition which had the objective of developing suitable
rehabilitation plans taking into account the constraint that traffic could be restricted to one lane
per direction, but never interrupted or detoured and that 2x2 lanes had to be available during the
summer holidays.
The plan for the rehabilitation of Reichenbichl Bridge specified the complete renewal of the
supporting structure since this seemed to be the only possibility for eliminating all the defects
completely and permanently. This rehabilitation plan is based on two innovative ideas:
• Without affecting traffic, a prestressed crosshead is joined to the existing hollow stanchion
of the old structure. The crosshead essentially transmits its load via friction, supported by
80 COMPOSITE BRIDGES IN AUSTRIA
circumferential prestressing, to the stanchion and does not need any other connecting
elements.
• For the „Vollpilzbrücken", a phased removal plan was developed which was co-ordinated
with the construction stages of the new composite structure. As a result, both the remaining
cross-section of the old supporting structure and first parts of the new structure can take
over traffic functions offering an absolutely safe bearing capacity.
This first rehabilitation project was performed on schedule from autumn 1992 to June 1995. As
the implementation of the plan was successful in all respects, the client decided to have other
„Pilzbrücken" rehabilitated also in accordance with this concept. At present, the design plans are
implemented at Große Larchwiesen Bridge and Weber Bridge, and will also be applied to the rest
of the „Pilzbrücken".
The supporting structure is a composite bridge with an S-shaped layout. For each direction, a
separate composite steel structure was erected. The bigger width of the lane to Salzburg results
from the inclusion of a climbing lane. The two plain main girders with parallel chords per
supporting structure have a construction depth of approx. 3.50 m and are continuously curved. Due
to the layout of rhomb-shaped, horizontal web systems, two parallel torsion boxes are created.
The composite deck made of B 400 steel has a standard thickness of 25 cm and is haunched to
40 cm at the main girders. The separate supporting structures of each direction were mounted by
cantilevering in parallel and at the same time. The composite deck was erected using a formwork
transport wagon with concrete being placed in a staggered way. In the support zones, concrete
was placed after the span areas were finished in order to prevent tensile stress in the composite
deck due to the concreting load. The total weight of the steel structure, which is mainly made up
of special Alfort steel, amounts to approx. 2,000 ton, corresponding to approx. 150 kp/m2.
The Altersberg hillside bridge is also located at Tauem Motorway. This bridge was designed and
erected in the years 1974 to 1975. Its spans are 48 + 6 x 78 + 2 x 82 + 90 + 69 839 m, and its
total width is 25.5 m. There is one common structure consisting of four main girders carrying a
single concrete deck. The plan of the bridge shows a curvature with radii between 2,500 m and
straight line. The main steel structure consists of welded plate girders (height 3,55 m) with
distances of 5.5 + 7.3 + 5.5 m, partially connected with bracings and lattice cross girders. The
steel qualities are St 37 T, St 44 T and ALFORT (permissible stress 288 N/mm2). The concrete
deck is of quality B 400 without prestressing. To reduce the tension stresses at hogging moments
a longitudinal compression force (up to 40,000 kN) was produced by hydraulic jacks installed in
gaps across the deck. To enable this the concrete plate had a slide way on the upper flange of the
main girders. The erection was executed step by step from one side with a derrick crane which
set in 18 m long parts of the main girder and after this was moved forward to take over the next
parts. Before reaching the following pier a special pantented cantilevering „bill" was used to take
over the bearing force by lifting the end of the bridge with hydraulic jacks. This action reduced
the cantilever moment at the previous pier.
W. HUFNAGEL, P. BIBERSCHICK, H. WENZEL 81
The Niederranna Bridge across the River Danube is located at Ebenhoch Provincial Road near
Wesenufer, approximately half-way between Linz and Passau.
The supporting structure is a three-span composite steel bridge with continuous bond. The
haunched, plain main girders have a construction depth of 3.75 m in the slab area and of 6.00 m
in the support zone. The reinforced-concrete deck made of B 400 is doweled to the top chords of
the main girder along the entire length of the bridge and is not prestressed in the longitudinal
direction. A state-of-the-art crack-control reinforcement system prevents the formation of
harmful cracks. The standard thickness of the composite deck is 25 cm which is raised to 40 cm
near the main girders.
Special mention has to be made of the mounting technique. In parallel to the construction of the
substructures and the piers, the prefabricated steel construction was assembled into five large
elements at the right bank. These structural elements which were up to 95 m long and weighed
up to 300 ton were placed by floating cranes within four days. This method did not require any
mounting supports in the Danube and resulted in a substantial reduction of the construction
period. In order to minimise tensile stress in the composite deck, an optimised concreting
sequence was selected. The basic principle was to place the concrete in the support zones after
completing the span areas. To reduce tensile stress in the composite deck even further, it was
raised by approx. 1.30 m at the abutments. The total weight of the supporting steel structure is
slightly below 1,000 ton, corresponding to approx. 230 kp/m2.
The Steyregger Bridge across the River Danube is located at the federal road B3 and provides the
area to the east of Linz with a direct connection to the industrial zone of Linz.
Since the objective was to achieve a carriageway gradient as low as possible above the clearance
defined for navigation, the most economical solution was to construct a cable-stayed bridge
82 COMPOSITE BRIDGES IN AUSTRIA
taking into account the main navigation channel's width of 161.2 m. Comparative calculations
demonstrated that composite construction was more economical than a steel deck solution.
Therefore, a composite cable-stayed bridge was constructed. The bridge beam is guyed via an A-
shaped pylon in the main navigation channel. The steel structure of the longitudinal girder is a
continuous girder grille with four main girders and one load-distributing cross girder per span.
The reinforced-concrete deck made of B 500 is doweled to the main girder's top chords along the
entire length of the bridge and pretensioned in the longitudinal direction. The deck has a thickness
of 20 cm at a main girder distance of 6.4 m and is haunched to 35 cm above the main girders.
The upper bracing between the inner main girders (which was placed slightly lower due to
the formwork wagon) absorbs wind load before the deck hardens. In the cable installation zones,
the bracing was widened to the outer main girders and allows for cable force distribution to all
four main girders. The A-shaped pylon (approx. 44 m high) is based upon the cantilever arms of
the pylon cross girder via pivoting point bearings. In the pylon head, the cables run across a
welded saddle bearing. Each of the two cable trains is made up of 15 locked-wire strand cables,
0 69 mm, with multi-layered round core and three Z wire bearings the outermost of which is
hot-dipped galvanised. They were pre-stretched by the manufacturer.
The mounting of the bridge was started at the Steyregg end span using two temporary frames and
was continued to the Linz abutment by cantilevering. In the second and third span, only the inner
main girders were cantilevered to the pier from the middle of the span on while the outer ones
were constructed after fixing the inner main girders. After the pylon was erected, cantilevering
continued in the navigation channel. Due to the installation of the cable train, the clearance was
bridged without interruption of navigation. The deck was concreted in sections with a length of
approx. 40 m using a formwork wagon travelling in the top wind bracing so that tensile stress in
the deck was minimised. By raising the bridge's ends, in particular the support zones 1 and 4
were relieved. Apart from longitudinal pretensioning by means of tendons (strand cable
St 160/180) with a total force of 9,000 ton in the pillar zones, the deck was prestressed via the
diagonal cables by raising the saddle bearing at the pylon. Due to the optimised concreting and
prestressing sequence, the bonding effect was ensured along the entire bridge. The total weight of
the steel construction, including the pylon, cables and installations, amounts to approx.
3,000 ton, corresponding to 260 kp/m2.
Recently, composite bridges have also been used in railway construction. In the course of the
expansion of the western railway line between Vienna and Salzburg, Eisenbahn-
Hochleistungsstrecken AG plans to construct a by-pass near Melk in order to raise capacity while
providing also a link to the existing train station of Melk. Both the high-capacity section and the
by-pass cross the River Melk and a federal road. The comparison of the prestressed concrete
solution and composite construction included in the specifications of the call for tenders took into
account not only initial investment but also the different costs of maintenance and utilisation.
Based on these conditions, the composite steel solution was selected as being more cost-efficient
in the final analysis.
The two bridges have haunched plain girders. The high-capacity bridge is a four-span structure
(33 m + 48 m + 33 m+ 31m) having a total length of 146.2 m. The bridge of the by-pass follows
the railway track with a radius of 700 m and is a five-span structure (53 m + 53 m + 79 m + 53 m
+ 36 m) having a total length of 276.2 m. The static calculations were performed according to the
new Austrian standards ÖNORM B 4003 and B 4300. As there are no Austrian standards
applicable to composite railway bridges, the assumptions were made in accordance with the
W. HUFNAGEL, P. BIBERSCHICK, H. WENZEL 83
composite road bridge standard ÖNORM B 4502. A comparative calculation according to Eurocode
1994-2 yielded good correspondence with the assumptions made. The steel grade used was
S35510. Since big chords with a thickness of up to 90 mm had to be welded, thermomechani-
cally rolled steel of the grade DIMC-355B was used for improved welding properties. For the
conventionally reinforced composite deck, B 400 concrete with a thickness of 40 cm was used
for the main bridge and B 500 concrete with a thickness of 50 cm for the access bridge. The steel
construction was connected with the concrete deck by means of headed shear connectors. The
approx. 1540 ton bridge structure was manufactured in Vienna. The individual components had a
total weight of approx. 40 ton, a width of up to 5.2 m and a length of 25 m. They were delivered
by special transport to the site and mounted by means of two-engine rubber-mounted cranes.
The Pöchlarn Bridge across the Danube is planned to be constructed in the course of Pöchlarn
B 209 road and will link the Al western motorway to B 3 Danube road. Two lanes and a bicycle
path are to be crossed. The total width of the bridge amounts to 13.45 m. In the course of 1997,
two parallel call for tenders will be issued in which a composite solution and a prestressed
concrete solution will compete directly.
The special feature of this design is the type of erection envisaged. First of all, the steel part of
the cross-section will be advanced using the launching method. The concrete slab will be
produced analogously by inserting the deck which is concreted section by section at the southern
dam and which will only be bonded subsequently. In addition to a slight economic advantage,
this construction method is expected to result in significant improvements of the quality of the
deck since the fresh concrete will not be subjected to the load which cannot be prevented in
conventional construction nor will squeezing due to shrinking occur. Moreover, the deck can be
very easily replaced in the future.
The bonding effect is produced by a welding seam between the top steel chord and a
„Perfobondleiste" which is inserted together with the deck and slightly protrudes over the lower
edge of the deck. This protrusion makes it relatively easy to fix the bonding strip in the form-
84 COMPOSITE BRIDGES IN AUSTRIA
work and also functions as a lateral guide of the deck during insertion. In order to achieve a
guide play, each screwed connection of the upper cross arm is fixed for insertion by means of
two provisional screws with lower diameters in a slightly displaced position. After insertion, the
connections are unscrewed and the top chord is advanced to the bonding strip.
~T~
Fixing the Perfobondleiste Subsequent production
during concreting of bonding
Figure 4: Details
In Vienna, there are several innovative composite bridges, such as Kaisermühlen Bridge:
Already during the dredging of Neue Donau, two piers were constructed. At that time it was
planned to re-erect the emergency bridges used for the traffic across the River Danube following
the collapse of Reichsbrücke as pedestrian bridges at this site. The positions of the piers were
adjusted to these 80 m long one-span girders. After this plan was dropped, the position of the
piers still was a binding condition for the design of the bridge. The plane environment without
any dominance made a significant cable-stayed bridge appear to be a particularly desirable
solution. However, the effective span length ratios were very unfavourable for this approach.
Instead of guying a specific centre span back to the abutments, the two end spans required an
elastic support in this case. As a result, the longitudinal girder of the centre span was stiffened
with a triangular truss structure which is harmoniously integrated into the room created by the
gradient ascending to the centre. Thus, the non-rigid end spans are elastically stabilised against
the stiffened centre of the bridge via bundles of three cables each.
The pylons are rigidly connected to the structure; due to this restraint and the widening of the
cross-section in both directions, there was no need for an upper cross arm for stabilisation
purposes. The longitudinal girder is formed by two plain girders placed in an inclined arrangement
to each other so that the width of the pedestrian level of 4 m visually widens to the top. The
footway construction consists of a ribbed concrete slab with trapezoidal sheeting as permanent
formwork.
w
-±z -M
\3 -
The supporting sections of the two end spans were welded together along the bridge axis at the
mounting sites and launched via temporary frames and barges to the piers in the river. The
pylons were already provisionally connected to the supporting structure via joints and launched
together with it. The centre piece of the longitudinal girder was mounted to the stiffening truss in
parallel to the river bank from where it was lifted by a mobile crane to two barges which
transported it exactly below the two cantilevering parts of the end spans. By means of presses
and strands, the centre part was lifted to the correct level of the supporting structure where it was
welded to the two end spans. Finally, the pylons were folded up, rigidly welded to the longitudinal
girders, the cables were partly stayed, the carriageway was concreted and the final cable
tension was adjusted.
For the construction of the 20 m wide Ameis Bridge across the western railway line, a one-span
structure without piers was to be designed in order to ensure a flexible arrangement of the tracks
below.
Trumpet-like widenings reaching far into the bridge had to be accepted so that it was not possible
to apply a trough-bridge design which had been used in the old structure to be replaced. The
already low clearance profile of the railways with 5.50 m above the rail top and the gradient
permitted a maximum construction height from 1.65 to 1.75 m at the span's centre as a function
of the cross-section of the bridge, which was even reduced by more than 50 cm towards the
abutments so that the sickle-shaped girders had a high slendemess ratio of only 1/35. For the
one-span structure, the composite solution was particularly well suited since the concrete was to
be placed exclusively in the compression zone. This design was clearly more economical than a
steel-deck bridge with an orthotropic plate. The 30 cm thick concrete slab, which was required
anyway due to the high compressive stress, also offers the advantage of being able to ensure
transverse distribution alone due to its stiffness, requiring no further bracings. It was possible to
keep the steel part very simple by using nine single I-shaped girders which were only structurally
connected by extreme cross girders. The trumpet-shaped widenings are handled by means of two
additional, slanted girders.
The concrete strength class used was B 500, as planned, and necessary enhancements were
achieved by inserting compressive reinforcements. The increased mass as compared with a pure
steel construction improves the vibration behaviour of the one-span structure. The option of
installing a vibration absorber was envisaged, but based on subsequent measurements carried out
under traffic load it turned out to be unnecessary.
The total hog steel girders was 60 cm, out of which 15 cm were attributable to the impact
of the
of the concrete's shrinking and creeping. Even though individual deformation components
showed slight deviations from the calculated values during construction, the aggregate deformation
was still estimated correctly. The main girders received a field connection approximately in
the centre of the bridge where it was possible to set up a temporary frame between the tracks.
The girder components with a length of up to 32 m were transported to the site by night and
placed using a two-engine rubber-mounted crane. The webs were connected by high-strength,
friction grip bolts, and the chords were welded on site with the lower chords being made of
Alfort steel with a thickness of 90 mm.
86 COMPOSITE BRIDGES IN AUSTRIA
3. Nordbrücke
Nordbrücke which was completed in 1964 has with 2x2 lanes and follows the north-west railway
line which was abandoned due to revised plans for Vienna's railway network.
-
This bridge across the Danube was erected in composite steel construction for the first time in
-
Vienna with a total of four plain main girders made of steel St 52 with a height of 2.60 m and a
span of 83 m. The supporting steel structure was assembled at the right bridge head and launched
across the railway facilities and the River Danube towards the left bank. In the years 1984 and
1985, a ramp which already branches off on Nordbrücke was annexed to the composite structure.
In this process, the existing cross-section was supplemented by another main girder. Taking into
account the creeping of concrete, the achievement of the bonding impact and the connection to
the old concrete carriageway deck were timed in such a way that the 30 cm wide connecting joint
remained virtually stress-free during concreting.
With its current peak traffic load of up to some 108,000 vehicles per day and, above all, with its
maximum load of 2,310 vehicles per lane and hour, this structure is among the bridges with the
highest load in Central Europe. After more than 30 years of bearing high traffic loads, the resulting
state of the bridge required the performance of the first general rehabilitation. The damage to
the deck is largely attributable to de-icing salt whose impact was still unknown at the time when
the bridge was constructed. Other systematic damage was caused by the ageing of the bituminous
water-proofing and by the omission of the insulation below the central strip and the pedestrian
area, which used to be common practice for economical reasons at that time. This damage was
monitored and documented in the on-going inspections of MA 29 and, eventually, recorded in a
summary assessment which formed the basis for the decision in favour of general repair works.
After the removal of the entire bridge equipment, bituminous surfacing and old water-proofing,
the concrete located at the raw structure was repaired and covered by epoxy resin before the new,
double-flamed bituminous water-proofing, a levelling course and 2 x 3.5 cm of mastic asphalt
were applied.
As pedestrian and bicycle traffic was moved to Nordsteg, the narrow pavements on Nordbrücke
were no longer needed. Thus, it was possible to establish one narrow lay-by in addition to the
two lanes per direction, without widening the bridge in total, which would have been impossible
W. HUFNAGEL, P. BIBERSCHICK, H. WENZEL 87
or at least highly expensive for static reasons. This yielded a considerable increase in traffic
safety and a reduced frequency of traffic jams.
For this purpose it was necessary to remove the old footway cantilever and to mount a margin
protection in the form of an 80 cm high safety wall at the outer end of the cantilever arm. For the
reason of weight alone, the structure was made of steel and consists of steel posts placed at intervals
of 4 m to which 8 mm thick wall sheets are screwed. Additionally, a handrail is mounted on
them which supports the safety wall in case of accidents as a continuous tieback. The anchorage
of the wall posts at the top of the cantilever slab was designed for a static impact force of
100 kN. In order to produce the anchorage, first of all, the outermost 50 cm of the cantilever
slabs were removed by means of high-pressure water jets and the reinforcement was uncovered.
Additionally, slots were cut, also by means of high-pressure water, for a strong, upper secondary
reinforcement concentrated near the posts. At the end of the cantilever arm, an edge beam was
concreted which acts as a stiffener and distributes the truck wheel load required by the standards
in such a favourable way that it was not necessary to enhance the reinforcement within the standard
range. The outer formwork of the edge beam is formed by a continuous margin sheet at the
outside of which the posts of the safety wall are mounted.
The mobile scaffolding for concreting the edge beams was conceived in such a way that the loads
were introduced close to the main girder. The formwork was pressed to the existing slab by
means of anchor bars. On the whole, all the re-structuring measures were designed to have no
significant impact on the main system of the bridge. In relation to the overall area of the slab, the
changes and additions to the cross-section are of minor importance. The prestressing of the
tension zones was not affected in any way.
At its 30th anniversary, Nordbrücke is a practically new structure, furnished with lay-bys, a steel
margin protection stiffening the entire concrete deck and new central strips with safety rail
damping elements, all that in maintenance-friendly versions. Moreover, a convenient pedestrian
link was provided by the Nordsteg replacement bridge. The costs of the general repair works
amounted to approx. ATS 200.0 million. Apart from damage due to the action of de-icing salt
and moisture penetration, the 30-year old composite deck does not show any cracks and fatigue
problems in the bonding area and was turned practically new by the repair works. Since
constructional defects were not found in the composite structure either, this bridge is again in a
perfect state so that composite construction is recommendable for bridges of this type both in
technical and economic terms.
For the purpose of „System Identification" a monitoring set-up was created, that enables quick an
efficient recording as well as signal processing and report generation. The basis is the measurement
of acceleration in a well determined layout of relevant locations of a structure. This
provides data for the FFT analysis to generate the desired spectra. In addition data of the actual
displacement of the structure is collected by infrared laser to gather information on the static
behaviour and its relation to the dynamic action.
The collected data are processed to provide an informative report, which shall contain information
on the signals itself in the desired units, the power spectrum of the readings, raw and
smoothened, the drift of the readings and the relevant displacements. In a further step the readings
of the various locations are combined to get an averaged spectrum and the related displacements.
This is the basis for the animation of the modes of the structure and the visualisation of it.
3.4 Conclusion
Due to the fact, that this bridge was monitored during 3 different stages, before, during and after
the rehabilitation, valuable information was gained about the influence of the state of the structure
on the response spectrum. From this basis it is tried to develop further tools to assess the
quality of structures using data from dynamic monitoring.
Summary
Suitable combinations of constructional materials may generate composite actions which are
successfully utilized in seismic resistant building structures. The main behavioural features of such
combinations, usually called composite systems, have been examined in case of new building
construction as well as in case of old building retrofitting.
After an overview of the international situation both in research and codification, some example
of application of composite building structure in seismic areas have been presented.
1. Introduction
The ability of structural typologies to withstand severe actions is particularly proven when the
building constructions are submitted to the violence of an earthquake. The examination of
damages always represents a precious source of information on the ultimate performance of
constructional materials [I].
Referring to the traditional typologies, the old masonry structures are the first to prematurely fail
under the seismic attack, due to their intrinsic features which are very often worsened by the age
and the revage of time. They need to be upgraded by means of more ductile and modern
materials, like concrete and steel, giving rise to different kinds of composite actions.
But unfortunately also many reinforced concrete structures are seriously damaged and sometimes
collapse because of the earthquake, due to bad execution and poor material quality, which
produce a tremendous lowering of ductility.
Looking to steel buildings the past experience show that the cases of global collapse are very rare,
even if the traditional image of steel as the more suitable material in seismic resistant applications
has been seriously undermined after the damages recently occurred during the Northridge (17
January 1994) and Kobe (17 January 1995) earthquakes.
Summing up, from the experimental evidence of the sad after-earthquake scenarios, it is easy to
recognize that all the common constructional materials from the worst to the best used alone in
the traditional typologies can badly perform under severe earthquakes, producing serious damages
up to the collapse.
90 COMPOSITE BUILDING STRUCTURES IN EARTHQUAKE ENGINEERING
In order to increase the reliability of constructional materials, it can be observed that a rational
combination of non-ductile (masonry, concrete) with ductile (timber, steel) materials can produce
a kind of synergic effect which improves the behaviour of the construction under severe actions.
Looking back to the historical development of the seismic resistant structures, we can find that,
after the catastrophic earthquake which destroyed the Calabria region in South of Italy at the end
of 17th century (1783), the government imposed to build the new constructions by using a timber
lattice-work inserted into the masonry walls (Fig. 1). The so-called "casa baraccata" (treillis
house) represented the ancestor of a composite structure (masonry plus timber) conceived for
seismic resistant purposes and its performance was largely appreciated during the subsequent
earthquakes. This system is very similar to the one imposed in Lisbon after the earthquake of
1655.
Only more recently the composite systems have been used for new constructions, but the original
motivation was mainly based on economic aspects rather than on the requirement to improve the
structural performance. Nevertheless the well known composite structures made of steel elements
working together with RC elements demonstrated a good synergic behaviour also under severe
seismic actions. Considering the recent damage to connections of a number of steel structures
during both recent Northridge (Los Angeles) and Hygoken - Nanbu (Kobe) earthquakes and the
numerous failures of new reinforced concrete structures during all the known earthquakes, it
appears that the use of steel-concrete composite systems could mitigate some of the
vulnerabilities of steel and reinforced concrete structures alone.
The technological systems allowing the development of a composite action give rise to composite
constructions, which can act at two different and separate levels:
A - member level: different materials (usually steel and concrete) can form parts of the cross-
section; it comprises beams, columns, slabs, walls;
F.M. MAZZOLANI 91
B- structure level: sub-systems made of different materials can compose the whole structure; it
comprises the possible combinations of frames, bracings, walls and cores, which can be made
of simple (steel or RC) as well as composite elements.
Level A leads to the so-called "composite members" and level B gives rise to the so-called
"mixed" or "hybrid structures", but all together they belong to the family of composite systems.
From the point of view of seismic resistance, composite systems are suitably used both in the
construction of new buildings as well as in the refurbishment of old existing buildings. The main
advantages of a composite system in seismic resistant applications, respect to structural steel or
reinforced concrete alone, can be identified in the following points:
- high stiffness and strength of beams, columns and moment connections;
- satisfactory performance of all members and the whole system under fire conditions, which can
arise after an earthquake;
- high constructability for floor decks, tubolar infilled columns, moment connections;
- increase of ductility for encased beams, encased columns and beam-to-column connections;
- satisfactory damping properties for the whole system.
Due to these synergic properties, it seems logical to utilise the two basic materials (steel and RC)
in tandem and to consider, therefore, both composite and mixed structures as an attractive
solution to seismic design problems.
Many research results [4, 5, 6, 7, 8, 9, 10, 11, 12, 13] have shown the interest of using composite
structures in seismic areas, particularly due to the presence of concrete, which increases the
resistance in the elastic field up to 50%, contemporary increasing the stiffness and largely
preventing local buckling. After complete concrete crushing, the structure behaves always like a
bare steel structure when submitted to very large displacements.
Due to the complexity of the stress state in composite connections many experimental and design-
oriented research project have been developed in USA, Japan and Europe for several types of
connection details, which demonstrate their potential for use in seismic resistant applications [12,
13, 19, 20], In terms of seismic design, composite connections often avoid or minimize the use of
stiffners comparing to structural steel design.
Among the composite beams types, the encased beams (Fig. 3) represent a suitable system which
provides a good performance under cyclic loading due to the presence of the concrete mass which
avoids or at least postpones the local buckling phenomenon in the web of the double T section.
This increases the rotation capacity of the member and therefore the ductility of the whole
92 COMPOSITE BUILDING STRUCTURES IN EARTHQUAKE ENGINEERING
structure. The behaviour of encased beams has been proved by many monotonie and cyclic tests
[6,11].
Vertical Reinforcing Transverse Spot-Weld Composite columns made of round, square or
rectangular hollow steel sections fdled up with
concrete present also the advantage to improve the
local buckling resistance of the steel wall, allowing to
increase the b/t ratio of the section.
Alternative composite solutions for the bracing function can be obtained by means of RC frames
with encased masonry and RC frames with steel braces, the last being particularly suitable in case
of retrofitting (Fig. 5).
Many of the above combinations are just theoretical; in practice only few cases have been
experienced, but we can find some interesting proposals, like the one shown in Fig. 6.
The use of composite steel-RC beams in steel MRF structures provide interesting results when
applied according to the following phases (Fig. 7):
- the steel beam is connected to the column tree by means of a bolted cover plate only in the
web, so the joint behaves as a pin;
- the concrete slab is casted excluding the part corresponding to the joint, so the dead lead of the
floor does not produce bending moments in the columns, except for the affect of eccentricity in
the frame node;
- also the beam flanges are connected by means of cover plates, leading to the complete scheme
of frame.
F.M. MAZZOLANI 93
The simple reparation of a damaged element can be made in different ways. Some examples are
given in Fig. 10, dealing with the main systems using steel as active material for strengthening
masonry walls, RC beams, wooden floors [16 ]. RC can be also used for repairing masonry walls,
steel floors, wooden floors.
The main technological systems used in case of retrofitting of existing structures give rise to
different composite elements belonging to level A, like: masonry - steel; masonry - RC; RC -
steel; timber - steel; timber - RC.
At the upper level, a typical application is the upgrading of RC frames by means of steel braces
[3], This system has been used several times: in Mexico City after the earthquake in 1986; in
Berkeley after the Loma Prieta earthquake in 1989; in Santa Monica after the earthquake of
Northridge in 1994. Some examples of the mentioned applications are given in the Fig. 11. In
addition to the strengthening contribution, it can be observed that in many cases the addition of
steel bracing produces a substancial improvement of the aesthetic aspects of the façades, which,
before the retrofitting, were completely anonimous and sometimes ugly.
F.M. MAZZOLANI 95
1
reinforcement
2 corrugated steel sheets
3 wooden planks
4 wooden joist
5 wooden girder
6 steel I-section
a) b)
0 d)
Fig. 10. Technologial systems based on composite actions, which are commonly used in
consolidation operations: a) composite steel-timber floor; b) steel frame in a masonry
opening; c) steel reinforcement for masonry walls and columns; d) integration of RC
sections with steel plates.
5. Codification
In the industrialized areas of the world the use of new systems, like composite structures, requires
the assessment of specific provisions for seismic applications as it has been extensively done for
steel structures [17], Perhaps the main reason for not using widely the advantages of composite
systems in seismic areas is due to the lack of seismic design codes In fact, it is well known that
the Eurocode 8, now in the conversion phase from ENV to EN, contains the Chapter "Specific
rules for composite buildings" which is just informative, not normative [18],
In addition, from the comparison between Eurocode 4 and Eurocode 8 many incongruicies arise,
which produce some perplexity in the application of composite structures in seismic areas In
particular, EC4 esplicitally excludes the use of sway frames and the design rules are referred only
to braced non-sway frames, stating that the unbraced frames are outside the scope of EC4 in the
design of composite connections. It means that there are strong limitations in the choice of a
solution in the wide range of composite typologies and the use of bracings is always compulsory,
what vanishes the meaning of the behaviour factors given in EC8 for other typologies
The first U.S seismic design provisions for composite constructions (NEHRP) have been
developed by the Building Seismic Safety Council (BSSC) within the National Hazard Mitigation
Program in 1994 [19] The most challenging parts of this code are devoted to establish seismic
force reduction factors and drift amplification factors In estabilishing reduction factors, the
available research data have been integrated by engineering judgment and phisical understanding
96 COMPOSITE BUILDING STRUCTURES IN EARTHQUAKE ENGINEERING
of the behaviour of these systems. The basic composite structural framing systems identified in the
NEHRP code and the corresponding force reduction factors (R) and drift amplification factors
(Cd) are the following:
COMPOSITE SYSTEMS R Cd
Special moment frames 8.00 5.50
Ordinary moment frames 4.50 4.00
Partially restrained frames 6.00 5.50
Eccentrically braced frames 8.00 4.00
Special concentrically braced frames 8.00 4.50
Concentrically braced frames 6.00 5.00
RC shear wall with steel elements 8.00 6.50
Shear wall reinforced by steel plates 8.00 6.50
For these systems specific design requirements are provided, with particular reference to
connections and detailing.
c) d) e)
Fig. 11. Seismic upgrading of RC frames by means of steel bracings: a), b), c) Berkeley;
d) Santa Monica - Los Angeles; e) Mexico City.
F.M. MAZZOLANI 97
In Europe an extensive research work has been recently carried out to focus on the cyclic
behaviour of composite members and connections [4 to 13], These studies have confirmed the
feasibility of composite frames designed to resist seismic actions, but very few applications for
building construction in the European earthquake prone Countries seem to derive from these
theoretical basis. As an example, a new steel-composite structure has been recently built in
Timisoara - Romania [22], according to the Romanian Seismic Code, which is largely ispired to
Eurocode 8.
In Italy the majority of composite (hybrid or mixed) structures have been erected in the area of
Naples city and surroundings. Due to the damages produced by the bradyseism phenomenon, the
old town of Pozzuoli was completely evacuated in the early eigthies and a new town has been
erected for 25.000 people. The pressing need to give hospitality the population in the shortest
period of time oriented the choice of the structural typology on prefabricated solutions for low-
rise buildings of 4 to 6 stories. The mixed system composed by concrete cores and steel skeleton
has been selected in the majority of cases, because of both quick erection and seismic reliability.
This can be considered an interesting example of extensive use of composite structures in low
seismicity areas.
both precasted concrete elements lightened with polystyrol and trapezoidal steel sheetings infilled
with casted concrete. Beside to this current typology, also some special systems, always
composite, have been conceived with innovative solutions; three of them merit to be mentioned
[23].
First, the Law Court Building, composed by three towers, which are equal in plan but have
different height varying from 78 to 177 m (Fig. 13). For each tower the structural system is
composed by reinforced concrete curved walls, which provide strength and stability under
horizontal loads, and a steel skeleton resisting vertical loads only. The floor slab is connected to
the upper flange of beams by means of studs, giving rise to a composite horizontal diaphragm
connecting the steel structure to the reinforced concrete walls.
serviceability conditions and to provide damping and energy absorbtion during an earthquake
(Fig. 16).
Fig. 15. The main building of the new Fire Fig. 16. Special devices to provide base
Department Centre in Naples. isolation in the building of Fig. 15.
References
[1] MAZZOLANI F. M.: Design of steel structures in seismic zones (state-of-the-art lecture),
10th European Conference on Earthquake Engineering, Vienna, August 28 - September 2,
1994, in Proceedings edited by G. Duma, Balkema, Rotterdam, 1995.
[2] MAZZOLANI F. M.: Refurbishment, Arbed - Tecom, Luxemburg, 1990.
[3] MAZZOLANI F. M.: Strengthening options in rehabilitation by means of steelworks,
SSRC 5th International Colloquium on Structural Stability, Rio de Janeiro, August, 1996.
[4] SCHLEICH J. B. &nd PEPIN R.: Seismic resistance of composite structures, EUR 14428
EN Report, 1992.
[5] PLUMIER A. & SCHLEICH J.B.: Seismic resistance of Steel and Composite frame
structures, J. Construct. Steel Research 27, 1993.
[6] BALLIO G.: Test report of the Milan laboratory, EUR 14428 EN Report, 1992.
[7] BURSI O. S. & ZANDONINI R.: A numerical validation study for pseudodynamic
analysis of semi-rigid composite sway frames, in Structural Stability and Design (edited by
S. Kitipornchai, G. J. Hancock, M. A. Bradford), Balkema, Rotterdam, 1995.
[8] BRODERICK B. M. & ELNASHAI A. S Seismic response of composite frames. 1°
Response criteria and input motion, Engineering Structures, vol. 18, n. 9, 1996.
[9] ELNASHAI A. S. & BRODERICK B. M.: Seismic response of composite frames. 11°
Calculation of behaviour factors, Engineering Structures, vol. 18, n. 9, 1996.
[10] PACURAR V.: Ductility of steel-concrete mixed section beams, in Behaviour of Steel
Structures in Seismic Areas (edited by F. M. Mazzolani and V. Gioncu), E & FN SPON, an
Imprint of Chapman & Hall, London, 1995.
[11] PLUMIER A., ABED A. & TILIOUINE B Increase of buckling resistance and ductility
of H-sections by encased concrete, in Behaviour of Steel Structures in Seismic Areas (edited
by F. M. Mazzolani and V. Gioncu), E & FN SPON, an Imprint of Chapman & Hall,
London, 1995.
[12] PRADHAM A. M. & BOUWKAMP J. G.: Structural performance aspects on cyclic
behaviour of the composite beam-column joints, in Behaviour of Steel Structures in Seismic
Areas (edited by F. M. Mazzolani and V. Gioncu), E & FN SPON, an Imprint of Chapman &
Hall, London, 1995.
[13] DUNAI L., OHTANI Y. & FUKUMOTO Y.: Cyclic behaviour of steel-to-concrete end-
plate connections, in Behaviour of Steel Structures in Seismic Areas (edited by F. M.
Mazzolani and V. Gioncu), E & FN SPON, an Imprint of Chapman & Hall, London, 1995.
[14] MAZZOLANI F. M. & PILUSO V.: Theory and Design of Seismic Resistant Steel
Frames, E & FN SPON, an Imprint of Chapman & Hall, London, 1996.
[15] MAZZOLANI F. M.: Strengthening of Structures, lecture 16.1, ESDEP, 1994.
100 COMPOSITE BUILDING STRUCTURES IN EARTHQUAKE ENGINEERING
[16] MAZZOLANI F. M.: The use of steel in refurbishment, lrst World Conference on
Constructional Steel Design, Acapulco, November, 1992.
[17] MAZZOLANI F. M.: The European Recommendations for Steel Structures in Seismic
Areas: Principles and Design, Annual Technical Session of S SRC, Chicago, 1991.
[18] ELNASHAI A. S. & BRODERICK B. M.: Eurocode 8. Requirements for the seismic
design of composite structures, 10th European Conference on Earthquake Engineering,
Vienna, August 28 - September 2, 1994, in Proceedings edited by G. Duma, Balkema,
Rotterdam, 1995.
[19] ASTANEH-ASL A,: Seismic design of composite structures in the United States, in
Behaviour of Steel Structures in Seismic Areas (edited by F. M. Mazzolani and V. Gioncu),
E & FN SPON, an Imprint of Chapman & Hall, London, 1995.
[20] LU L. W., RICLES J. M. & KASAIK Research on seismic behaviour of steel and
composite structures at Lehigh University, in Behaviour of Steel Structures in Seismic Areas
(edited by F. M. Mazzolani and V. Gioncu), E & FN SPON, an Imprint of Chapman & Hall,
London, 1995.
[21] ELNASHAI A. S.: Seismic resistance of composite structures, 10th European Conference
on Earthquake Engineering, Vienna, August 28 - September 2, 1994, in Proceedings edited
by G. Duma, Balkema, Rotterdam, 1995.
[22] STOIAN V. & OLARIUI Models, simulations and condensations in the design of a stee-
concrete composite structure placed in seismic zone, in Behaviour of Steel Structures in
Seismic Areas (edited by F. M. Mazzolani and V. Gioncu), E & FN SPON, an Imprint of
Chapman & Hall, London, 1995.
[23] MAZZOLANI F. M Seismic-resistant solutions in the new Management Centre of
Naples, Fifth World Congress of the Council on Tall Buildings and Hurban Habitat,
Amsterdam, May 1995.
[24] MAZZOLANI F. M.: Seismic resistant system for a composiie steel-concrete building,
IABSE-ECCS Symposium, Luxemburg, September 1985.
[25] MAZZOLANI F. M.: The seismic resistant structures of the new Fire Station of Naples,
Costruzioni Metalliche, n. 6, 1986.
101
Summary
This paper presents a general and conceptual overview of the increasing importance of the
rehabilitation and repair of existing structures that has been necessitated by growing demands
and requirements of use. It calls attention to the positive features that both the direct application
and derivations of composite construction can contribute to these types of activities. Finally,
several unique cases of bridge modifications, such as the enlargement of platforms and the
removal or translation of piers to increase low clearances, are introduced in order to help clarify
the previous statements.
1. General overview
The incredible wealth that contemporary societies have in existing and utilized infrastructures
and buildings, inherited from both historic and recent activity, is being subjected to degradation
and aging with the passage of time, as well as to the loss of efficiency and safety against the
growing demands of all types of activities: loadings, repetitive cycles, dynamic effects, etc.,
higher, in almost all cases, than the values considered during the design and construction of the
work. This can be attested by the successive changes in codes and rules, even in the most recent
periods.
The transformation of construction has been a systematic fact throughout history and there are
many well-known cases in which a work was demolished in order to reuse its materials in the
construction of a new, but conceptually different structure. At times this process was extreme, as
can be seen in religious examples; but there are also cases in which the change was made for
show or functionality or simply for urban renewal, prompted by the desires of a new invader,
monarch or owner.
Today, this re-use of materials is non-existent. Our present construction materials, once
dismantled, are of practically no value. Even to the point that a significant cost of transport to
and storage at a rubbish dump must be taken into careful consideration in any demolition. And
this comes before the negative ecological effect of this dumping is even considered, a problem
which is gradually increasing and, will eventually make it necessary to find appropriate treatment
methods for reducing its negative influence on the society.
For these and other reasons which will be introduced throughout this presentation, the best way
to get the most out of this inheritance is to reuse the structural systems -not the materials- and to
confer upon them new possibilities of resistance and improved durability and functionality that
will adjust them to today's conditions and actively incorporate them in the process, thus avoiding
102 REHABILITATION AND REPAIR OF STRUCTURES USING COMPOSITE SYSTEMS
Rehabilitation conditions can be very wide and assorted, and therefore, although in general the
total cost ot the modification oi a work tends to be cheaper, in many cases it can be equal to or
even higher than the cost of the demolition and replacement of an older structure
A number ot factors such as the historic or artistic value of a building or bridge, its beauty or
unique characteristics ot its site, the îeduction of nuisances that the preservation of the work
would mean for the local residents and users (demolition, transport, noise, etc or even deeper,
possibly ethical factors, which aie of increasing importance due to the growing awareness of the
intrinsic value of all the possible actions, can be determinants in the decision to choose an
integral rehabilitation ovei the leplacement of the work
Furthermore, in the field of îefurbishment or upgrading, higher costs, due to the greater
construction periods and laboi that this solution pioduces. can be a positive component for its
selection in societies with high levels ot unemployment, as it îequires more manpower while
reducing the basic energy costs ot new materials
In another sense, recently, new qualities and possibilities ot a number of materials, techniques,
equipment and processes of all types, which allow toi a substantial improvement in the control,
treatment and processing of methods used to facilitate and favoi îehabilitation have appeared or
been notably improved upon, independent ot the similai advancements that new constructions
techniques also piovide
All of this, together with the previously mentioned, inci easing social sensitivity to these
methods, can lead to the conclusion that piesent conditions favor the development of this type of
construction and that it is likely to become an important source of activity in many countries,
especially when there is an histonc component, that is. an important legacy to preserve in
buildings and bridges
From anothei side, apart from lefuibishment 01 upgiading activities, other unexpected situations
such as earthquakes, impacts, and internal constiuction defects, can require the repairing or
strengthening of structures or membeis These activities must be done independent of economic
criteria and, in general, in shoit terms and with guaianteed results
The following presentation of the diffeient topics or aspects that are a part of the modification of
structures clarifies, in very bioad teims, the importance of correct activity in this problematic
field For the most part, only an index with brief comments about the different possibilities
which can be integrated in the modification of structures, will be shown However, in the cases
in which composite steel and conciete systems aie used, some geneial ideas of the more common
processes will be developed and then completed with a more detailed presentation of some
unique solutions
2.2 Materials
In lines of resistance, durability and construction, aspects of great importance in the selection of
solutions and alternatives, a wide iange of possible materials can be relied upon, which includes.
but is not limited to
• High strength concietes and moitais
• Fluid mortals, without shrinkage oi expansion
• Mortars and other chemical products with high qualities oi inalteiability. strength, adherence,
and fluidity for injection or substitution ot cracked oi damaged areas
• High strength structural steels which are easily weldable but don't require special precautions
because of their low carbon equivalent
• Weathering steel with its high quality coloi and textuie
• Stainless structural steels with a wide variety ot textuies glossy, matte, colored
• Stainless steel re-bars for legions with very restricted covering thicknesses
• Prestressing bars with a wide variety of qualities and diameteis
• External prestressing steels in single, self-protected strands
• Advanced, high strength, composite materials wires, cables, and sheets with unlimited
durability and great lightness that balance then elevated price in risky cases, especially in
areas which are highly susceptible to corrosion
• Elastic bearing systems and shock transmission units (STU) for selected damping effects
Also with large number and variety ofconditions which include, but are not limited to
• Glues and adhesive resins for very strongly bonded connections between concrete and steel,
concrete and composite sheets, steel and composite materials, etc
• Standard and specialized welds with low heat procedures
• Very reliable chemical or mechanical bolts
• Welded or mechanical stud and bolt shear connectors foi standard carbon or stainless steels
• High strength bolts (HST)
General procedures
Regarding the different types ofstructuies and conditions to be consideied for each solution.
there are many different procedures which can be used, such as
• Precambenng by different methods mechanical, thermal (pre-heating of the steel
strengthening elements to be welded), prestressing (external and internal), etc
• Preloading by support settlement
• Load transfer between members by jacks
• Lifting and transversal sliding for widening and pairing
• Launching systems by pulling or pushing
104 REHABILITATION AND REPAIR OF STRUCTURES USING COMPOSITE SYSTEMS
The basic goal of composite structures, is to achieve the maximum collaboration between steel
and concrete and to exploit the best qualities of both in order to obtain a global system which is
superior to the simple addition or juxtaposition of the materials
The development of composite construction, the types and units for connection and the ways to
reach this objective are extremely similar to the basic concepts used for the strengthening,
repairing or rehabilitating of existing structures' full-employment of materials from the existing
structure, or at least those which are still suitable for use. combined with the minimum quantities
of additional materials, in order to achieve the fixed requirements of the final restored structure.
The behavior and all of the resources' formal, geometrical, resistant, links, joints, etc., of
composite construction and protect rehabilitation and repair clearly coincide and. solutions
similar to new. contemporary composite structures, can be obtained directly for the repair and
rehabilitation of existing works allowing foi changes in the dimensions and arrangement of the
materials, but with a nearly identical approach
In fact, there are many, very well known, typical cases of the direct employment of composite
systems for repair and rehabilitation, and to complete this presentation, first, several of these
more systematic cases, will be briefly reviewed. Then, some more unique and special cases
which open up the field and might stimulate in others ideas for the possible uses of new and
favorable shapes in composite systems will follow
3.1 Strengthening of concrete members by attaching thin steel sheets with resins, not only to
their bottom faces but to the upper ones
It is helpful to complete these types of joints with mechanical or chemical bolts at the ends of the
sheets to avoid local and progressive peeling due to: slight impacts, vibrations, curvature, end
defects, etc.
Typical carbon steel sheets can be replaced by stainless steel ones to increase the durability: or
by composites strips of organic materials' carbon fiber: etc. The latter are very easy to use
because of their light weight and adaptability, but can cost moie per kN of introduced tension
3 2 Placement of flanges, bars or prestressing tendons in the bottom of the steel members of
the structure, with the possibility of eventually combining these with the placement of new or the
improvement of existing upper concrete slabs, in order to increase the global capacity
In general, the joining and anchoring of these elements is relatively simple and the only real
difficulty lies in the capacity of the existing shear connections between the tension and
compression reinforcement, which can be problematic, or even a reason to discard certain types
of solutions.
3.3 Total or partial wrapping of conciete columns or piers in order to create a composite
member with greatei axial resistance and. more importantly. with a higher shear capacity to
improve the performance of these members m seismic areas or in areas were they are susceptible
to impacts.
Shear connections by studs or. prefeientially. by bonded resins combined with hoop stresses due
to the welding or pre-heating shrinkage of the partial elements which wrap the concrete shaft, can
be use to obtain the final element or membei
J. MARTINEZ CALZON 105
4. Unique systems
4.1 Widening of bridge platforms
Bridge over Tardera River (Barcelona) 1940. Project by Eduardo Torroja. First composite
bridge in Spain.
3 isostatic spans of 46-54-46 m. Twin, steel funicular truss girders with concrete deck.
Original platform width: 8.50 m. Final weight: 11.50 m. Ah 35%
Procedure
• Strengthening of the bottom chord of the truss girder with welded plates.
• Removal by cutting of the existing cantilevers.
• Placement of precast, composite slabs, fastened to the existing concrete deck slab with HSTs.
• Placement of reinforcement, concreting over the precast slabs, and thickening of the central
part of the deck.
• Finishing.
Final structural behavior
Complete, composite collaboration of the older system and the concrete newly incorporated into
it.
45 75 54 30 45 75
r -r T" g •/.
"T
'
/h I TTT
^
ELEVATION
1 100
T II 50
'"ÖÖ
I
25 3 50 L 3 50 I
25
UU L 3 45 3 45 [ UU
I
fsiDEWALK
SIDEWALK'T 151
O
ROADWAY]I ROADWAY *|
£
O
f
AT ABUTMENTS AT ABUTMENTS
-BB—T-
AT PIERS
CROSS SECTION (EXISTING) CROSS SECTION (WIDENED)
50
125 100 3 50 3 50 1
00 125 1
fl
T T 1 T I n
05
9L..
I
2% 2%
'0 20
'0 24
WIDENED PLATFORM
106 REHABILITATION AND REPAIR OF STRUCTURES USING COMPOSITE SYSTEMS
0.40
a 4.50 AT BRIDGE EDGE
0.I0 .0.30
Î
> 0.041 L
[
0.I5 0.575 1 1.485 11 0.35 10.251 lO.IO
4.56
1.00 2.56 1.00
1 1
1 1
Ï
1
1
1
1 |
--
1
1
1 1
1
1 1
1
1
J
Bridge over Asma River (Tarragona) 1910. Eight single spans of 13.80 m. Reinforced 7i shape
concrete deck platform
Original platform width 6 20 m Final width 10.50 m A,, 69.4%
Procedure
• Removal by cutting of existing cantilevers
• Placement of weathering steel box girders that are attached with fasteners to the concrete n
member.
• Placement of precast composite slabs anchored to the steel girders
• Placement of reinforcement, concreting over the precast slabs and thickening of the central
part of the deck
• Finishing
Final structural behavior
Completely composite structure with collaboration between the new concrete slab, the older rr
shaped slab, the reinforced concrete beam; and the steel box girders.
m
-*
i
)• '"f
075 \
I 1
[
[
075 80 80 80 80
|
} 13 J 13 [
13 13 13 SO 13 13 SO 13
ELEVATION
10 50
1 5 25 425
0 75 2 35 2.25 .0 10 T\35^0
I
65 BEARING EDGE OF
1 H
35 35
1
1 1
I
STEEL BOX SECTION
ttsmrr Jl'
VARIABLE
The solution is based on the incoipoiation of two weathering steel box girders placed parallel to
the deck slab under its overhang Like a stretcher, the system transfers the loads from the existing
piers, to new ones
Originally, several othei solutions were analyzed some with external prestressing. others which
included modifications of the abutments etc but the cost penod. erection and. especially, the
resulting aesthetics, deteimined the final solution chosen
Piocedure
• Amplification of unaffected pieis by connecting hollow steel shafts to them
• Erection of the new. composite pieis. which appeal identical to the previously widened,
unaffected ones
• Transverse drilling of the conciete deck slab o\er the piers to be eliminated in order to
introduce the tying piestiessing bars
• Placement of elastic, neoprene beanngs ovei the new piers and in contact with the bottom of
the conciete deck
• Placement of the steel box gndeis. fust with cianes and then by sliding them over teflon
strips
• Transfei of the loads tiom the pieis to be removed onto the box using groups of hydraulic
jacks The initial position of the conciete deck slab remains unchanged The precambered
steel gliders lose then cuivature and their bottoms îemain parallel and at the level of the deck
slab soffit
• Connection of the steel girders to the deck slab with piestressing bars
• Removal of jacks
• Load testing of the bridge
• Demolition of old pieis
Essential aspects
Neither the traffic on the highway noi on the overpass was disturbed
Final structural behavior
The concrete deck slab and the steel box girder woik together to transfer the loads to the
bearings The elastic beanngs at the new piers lestnct the hogging bending moments of the
concrete slab at these points to the maximum internal capacity of the affected cross sections
J. MARTINEZ CALZON 109
ORIGINAL ELEVATION
FINAL CLEARANCES
Transfer of loads
Final structure
\ IDEAL SOFFIT OF
THE DECK SLAB
BEARING ON THE
WIDEN STEEL
y
SHAFT
\ APPLYING LOADS
BEARING ON OF TRANSFER
NEW PIER
.PIER TO REMOVE
NEW PIER CENTRAL PIER
o-o oQ:;
EXISTING PIER TO REMOVE
WINDOW TO
^ANCHORAGE ACCESS
VO:Öf0'O:/
WIDENED PIER
ELASTIC
BEARING
0 775
f f
0.350
NEW PIER
Villanueva Auditorium at the Prado Museum in Madrid, helow Paintings Velazquez Hall
Conceptual application of load transfer, similar to the process described for bridges above, for a
sensitive site.
A system of parallel beams was placed under the existing longitudinal beam and adjacent to the
columns to be removed, so that once they were eliminated the newly placed beams filled the gap
that was left by the original.
J. MARTINEZ CALZON 111
Procedure
• Placement of pairs of new transverse girders
• Placement of plates and bolts for transfer
• Turning of bolts with gradual transfer of the column's loads to the new girders, which are free
to deflect, without changing the position of the original structural system
• Placement and welding of a pair of ancillary beams, parallel to the existing central one
• Removal of the original central beam
• Removal of the columns
• Closing of the central holes in the transverse grids
• Finishing
Final structural behavioi
Combined work of the grid consisting of the transverse and longitudinal girders
r ' ^
R\ *250 200 20
v*J_
\*2 \ DETAIL
lh
NEW GIRDER D
112 REHABILITATION AND REPAIR OF STRUCTURES USING COMPOSITE SYSTEMS
Final Auditorium
113
Summary
This paper discusses the advantages of using advanced polymer matrix composite materials,
originally developed for high-performance aircraft, for post-strengthening existing structures.
Criteria for evaluating and designing with these materials are suggested. In Switzerland,
retrofitting by externally bonding carbon fibre reinforced plastic (CFRP) laminates has been
shown to be less expensive than the technique of external steel plate bonding, especially if ease of
handling is a dominant cost factor.
1. Introduction
Changing social needs, upgrading of design standards, increased safety requirements and
deterioration result in existing structures that need to be retrofitted or demolished. Many existing
structures are part of the architectural heritage and demolition is not a viable option. Even for
newer structures, rehabilitation is in most cases a much better use of resources than replacement.
Bridges represent a major proportion of engineering structures. They are a significant factor in the
infrastructure and their maintenance has implications on the economic life of a nation through
disruption and traffic delays.
Today in Western Europe and in other parts of the world the strengthening technique for bonding
steel plates is wide spread and is the state of the art. In a non-corrosive environment this
technique shows very good long-term behaviour (Fig. 1). However, weathering tests over
extended periods of time have indicated that long term problems concerning corrosion of the steel
must be expected in outdoor applications. Ladner, Pralong, Weder [1] observed "small traces of
rust" on unprimed as well as primed bonded steel plates even after only 3 years exposure to
weathering. The rust became mere extensive during the course of the test and after IS years
exposure the areas have grown to 10 mm in diameter. These tests indicate a weakness in the
strengthening of structural systems using steel plates [2].
Fig. 1 Creep data for 26 years of a RC girder 190-depth x210 x 2895 mm) post-strengthened
with a steel plate (7x 120 x 2820). The steel plate is bonded to the bottom of the beam using a
filled epoxy resin (Ciba), without mechanicalfasteners. The beam is loaded in 4-point bending
(945 mm shear spans) using 30 kN lead weights at each of the two loading points. The induced,
constant bending moment is 28.35 kN m. The temperature variation during the indoor test is 16
to 24°C, and the range of relative humidity is 40 to 80%. The curves represent
a) compressive strain of the concrete at the top of the girder [0100]
b) deflection at mid-span [mm]
c) tensile strain of the steel plate [0100]
d) tensile strain of the concrete at the steel plate interface above (c) [0100].
Steel plates have other disadvantages. During renovation work, particularly on bridges, generally
only a limited amount of mechanized lifting machines are available. In the interior of box girders,
for example, the heavy strengthening plate has to be carried by hand to the point of installation.
Consequendy, due to handling limitations on site, the steel plates are rarely longer than 6-8 m;
however, if die strengthening work involves greater plate lengths, a butt-joint system must be used.
This type of joint cannot be welded since the welding temperatures would destroy the adhesive
bond, consequently butt-jointed steel plates have to be formed from single shear lap joints. If steel
plates were replaced by high strength fiber composites a relatively thin component could be
delivered to the building site in rolls of lengths in excess of 300 m. Compared to steel plates then-
bonding technique is greatly simplified. Using bonded CFRP laminates [3] the quality assurance can
U. MEIER 115
be demonstrated by infrared inspection in the field, as discussed in a later section; this is not
possible with steel plates.
The partial substitution of steel plates with polymer matrix/fiber composites was discussed in
Europe in the early eighties. One of the most important decisions which had to be addressed at that
time was that regarding the most suitable fiber composite material for this application; Table 1 lists
criteria which specifically relate to the use of composite materials as a post strengthening material
for structures and applies particularly to prestressed laminates. These criteria may not necessarily
satisfy other applications. The ratings in Table 1 are rather crude, however, it is clear that the
tensile strength is a relevant criterion, but the significance of the compressive strength may be
questioned as concrete generally has to be strengthened in the tensile region of the beam. In certain
static systems, however, there may be regions which are normally stressed in tension but which
may also be subjected to compressive stress depending upon the load distribution; in these
situations bonded steel plates are not acceptable as they will peel off. Aramid fiber reinforced
polymers would also fail due to their poor compressive strength. Deuting [4] has shown that
carbon fiber reinforced composites (CFRP) do satisfy the compressive requirement
The modulus of elasticity of the laminate material is of great significance when the laminate is not
prestressed before being bonded, because only stiff laminates are able to relieve the stresses in the
existing internal steel reinforcement. Laminates fabricated from glass fiber reinforced composites
(GFRP) must be 4-10 times thicker than CFRP laminates to achieve the same tensile stiffness. If
such GFRP composites are longer than 6-10 m their handling on the construction site is difficult
The fatigue behavior of the system may be important or insignificant depending on the structure
and the nature of the loading. The bulk density of the material is less important as a criterion since
the density of all the fiber composites considered is low compared to that of steel.
The cost criteria is important If a comparison of the price of fiber composites is made with that of
the standard steel Fe 360 then it would appear, at first glance, that fiber composites are far too
expensive. The price factor based on unit volume of material is 4 to 20 times greater than that of
steel. However, when the cost of upgrading the structure is considered, the material cost amounts
to less than 20% of the total cost of the construction, consequently, when the ease of handling of
the fiber composite system is considered the solution becomes competitive due to its light weight
116 REPAIR USING ADVANCED COMPOSITES
(iii) In applications such as slabs with fire sprinkler systems, the pipe installation would have to be
removed in order to bond heavy steel plates on to the under-strength units. Composites,
however, would be able to compete with steel successfully as the thin strengthening
components would be bonded insitu. Consequently, labor costs would be reduced
substantially and the fire protection system of pipes and outlets would be operational during
rehabilitation work.
(iv) The results shown in Table 1 clearly indicate that carbon fiber reinforced polymer composites
most closely fulfill requirements for the post strengthening of structures. Consequently, from
the early eighties laboratories in Europe have concentrated their research efforts using this
material and all further discussion will be restricted to carbon fiber reinforced polymers.
Typical properties of the composites considered are given in Table 2. The large scale research
project undertaken in 1993 in the USA and in Canada in the area of carbon fiber/polymer
composites and many successful applications in Japan have confirmed the earlier Swiss
decision to use this material in construction.
When a change of temperature takes place the differences in the coefficient of thermal expansion
of concrete and the carbon fiber reinforced epoxy resin composites result in thermal stresses at the
joints between the two components. After 100 frost cycles ranging from + 20 degree C to - 25
degree C, no negative influence on the loading capacity of the three post-strengthened beams was
found [5].
For the post strengthened beams the following failures were observed:
(i) The CFRP composite failed during loading with a sharp explosive snap, the impending failure
was preceded well in advance of the failure by cracking sounds, concrete cracking and large
deflections.
(ii) Classical concrete failure in the compressive zone of the beam.
(iii) Continuous peeling-off of the CFRP laminates due to an uneven concrete surface. For thin
laminates of thickness less than 1 mm and bonded to the concrete surface with the aid of a
vacuum bag, an even bonding surface is required. If the surface is too uneven, the laminate
will slowly peel off during the loading.
(iv) Shearing of the concrete in the tensile zone (it can also be observed as a secondary failure).
(v) Interlaminar shear within the CFRP laminate (observed as secondary failure).
(vi) Failure of the reinforcing steel in the tensile zone (this failure mode was only observed during
fatigue tests).
The following failure modes were not observed but are theoretically possible;
(i) Cohesive failure within the adhesive.
(ii) Adhesive failure at the interface between the CFRP laminate and the adhesive.
(iii) Adhesive failure at the interface between the concrete and the adhesive.
For post strengthening with CFRP composites it is recommended that the design rule for the
CFRP composite is such that it should fail during yielding of the steel reinforcing bars before a
compressive failure of the concrete. Yielding of the steel bars should not occur before reaching
the permitted service loads.
Kaiser [5] investigated a 2 m span beam under fatigue loading. The cross-section was 300 mm
wide and 250 mm deep. The existing steel reinforcement consisted of 2 rebars of 8 mm diameter
in the tension and in the compression zones. This beam was post strengthened with a glass/carbon
fiber hybrid composite having the dimension 0.3 by 200 mm. The fatigue loading was sinusoidal at
a frequency of 4 Hz; the test set up corresponded to a four point flexure test with loading at the
one third points. The calculated stresses in the hybrid laminate and the steel reinforcement are
listed in Table 3. After 480'000 cycles the first fatigue failure occurred in one of the two
reinforcing rods in the tension zone; after 560"000 cycles the second reinforcing rod failed at
another cross-section; after 61'000 cycles a further break was observed in the first reinforced rod
and after 720'000 cycles a second break in the second rod was observed. The first damage to the
composite appeared after 750'000 cycles and it was in the form of fractures of individual rovings
of the laminate; die beam exhibited gaping cracks, which were bridged by the hybrid laminate. The
relatively sharp concrete edges rubbed against the hybrid laminates at every cycle and after
SOSXXX) cycles the composite finally failed, however, the test was executed with unrealistically
high steel stresses. The aim of the test was to gain insight into the failure mechanism after a
complete failure of the steel reinforcement; it was surprising to observe how much the hybrid
laminate could withstand after failure of the reinforcement
118 REPAIR USING ADVANCED COMPOSITES
Deining [4] performed further fatigue tests on a beam with a span of 6 m under realistic loading
conditions. The total load canying capacity of this beam amounted to 610 kN without post
strengthening. When the beam was strengthened by bonding a CFRP composite laminate, with
dimensions of 200 x 1 mm (laminate type T 300, Table 2), its load canying capacity was
increased by 32% to 815 kN. The calculated stresses in the CFRP laminate and the steel
reinforcement are given in Table 4. The beam was subjected to this loading for 10.7 million cycles.
After 10.7 million cycles the tests were continued in an environmental condition where the
temperature was raised from room temperature to 40-degree C and the relative humidity to a
value of 95% r.H. The aim of this test was to verify that the bonded CFRP composite could
withstand very high humidity under fatigue loading. Initially the CFRP composite was soaked
with water to nearly 100% saturation. After a total of 12 million cycles the first steel reinforcement
failed due to fretting fatigue. The joint between the CFRP laminate and the concrete did not
present any severe strain fatigue. In the next phase of the test program, the external loads were
held constant (Table 4) and the stresses in the reinforcing steel and the CFRP laminate decreased.
After 14.09 million cycles the second reinforcing steel rod failed, also due to fretting fatigue. The
cracks which were bridged by the CFRP composite laminate rapidly grew and after failure of the
third reinforcing rod, due to yielding of the remaining steel, the CFRP laminate was sheared from
the concrete.
In most applications in which CFRP laminates are used for strengthening, they are not exposed to
lightning strikes as they are inside a building or box girder which is equivalent to grounded cages.
Composite laminates used in bridge strengthening are positioned on the soffits of the beam and
lightning will have no access to them in this case. If there are situations where lightning may be a
danger, metal grids, which are used with composites in aircraft design, have to be utilized.
In 1994 the EMPA performed a series of bending tests on strengthened beams positioned in a
large horizontal testing oven [8]. The span of the 6 beams tested was 5.2 m and their width and
depth were 400 mm and 300 mm respectively; the volume fraction of the steel rebars was 0.65%.
The beams were loaded by hydraulic activators with the maximum short time load as laid down in
the Swiss code SIA 160 (1989) in four point bending. The oven was heated according to the ISO
Standard 834 with a temperature of 925 K after 1 hour. One beam was not plated and acted as the
control, another beam was post strengthened with steel plates (75 mm wide, 8 mm thick) and in
addition four beams were post strengthened with CFRP laminates (74 mm wide, 1 mm thick).
After 8 minutes duration of the test the steel plate came away from the beam. During the test in
which the beam were post strengthened with CFRP laminates, the fibers started to bum at the
surface of the laminates and their cross sections slowly decreased in value, thus causing a slow
decrease in stiffness. The CFRP composites finally became unbonded from the beam after one
hour. The main reason for the superior behavior of the CFRP composites compared with that of
the steel plates was their low thermal conductivity in the lateral direction.
Since the early seventies it was [1] always recommended that the post strengthening of a structure
should not be more than 50%. Therefore, after an accidental failure of the post strengthening of
the beam, a residual factor of safety of approximately 1.2 would remain and the collapse of the
structure could be avoided.
8. Quality Assurance
In Switzerland the pulse thermography (infrared inspection) [8] is applied for quality assurance of
the bonding of the CFRP laminates to the structural surface. This non-destructive testing method
relies on changes in thermal conductivity caused by flaws or damage. The equipment used for
infrared inspection is currently small and light weight, thus allowing analysts to gather more
sophisticated information regarding the object being tested on line on the construction site. The
technique operates on the principle that an infrared camera is positioned in front of the laminate
which is heated with a flash lamp. The sensors in the infrared camera detect the heat that is
absorbed and then re-radiated from the surface and the digitized information is sent to a video
board in a computer from where an image is constructed on a video screen. This image allows fast
and accurate judgment on the quality of the strengthening work.
9. Applications in Europe
To the best knowledge of the author the Kattenbusch Bridge in Germany is the first place in the
world where fiber reinforced plastic laminates were used to strengthen a bridge. After World War
n numerous prestressed concrete bridges for motor vehicles were built in Germany employing the
method of in-situ span wise construction. These continuous multispan bridges are mostly designed
as box girders. The working joints are at the points of contraflexure where usually all of the
tendons are coupled. Many of the bridges now exhibit cracks at the working joints. Usually, the
120 REPAIR USING ADVANCED COMPOSITES
bottom slab of the box girder is transversely cracked at the joint. This relatively wide crack grows
into the webs with diminishing width. Thereby it crosses the lower tendons and couplings. The
main cause of these cracks is a temperature restraint which was not taken into account during
previous designs [9-11]. In combination with other stresses tensile stresses at the bottom increase
and exceed the concrete tensile strength at the joint. As the reinforcement ratio of the bottom slab
was often low, yielding of the steel occurred and wide cracks formed. Due to increased fatigue
stresses, the durability of the reinforcement and the tendons was no longer assured. Thus the
necessity for repair arose. In the late seventies Rostasy and his co-workers [12] developed a
technique to strengthen such joints with bonded steel plates. The first successful application was
the Sterbecke Bridge near Hägen (Germany) in 1980. In 1986/87 this method was used for the
first time with glass fiber reinforced plastic laminates on the Kattenbusch Bridge. The Kattenbusch
Bridge is designed as a continuous, multispan box girder with a total length of 478 m. It consists
of 9 spans of 45 m and 2 side spans of 36.5 m each. There are 10 working joints. The depth of the
twin box girder is 2.70 m. The bottom slab of the girder is 8,50 m wide. One working joint was
strengthened with 20 glass fiber reinforced laminates. Each plate is 3200 mm long, 150 mm wide
and 30 mm thick. Loading tests performed by Rostasy and co-workers showed a reduction in the
crack width of 50% and a decrease of the stress amplitude due to fatigue of 36%. The static and
the fatigue behavior was at least equal to the steel plate bonding technique. From the corrosion
point of view, the expectations of the glass fiber reinforced plastic laminates are much higher.
The valuable gothic roof structure of the Church of our Ladies in Meissen in Germany was built in
1447. The gothic vault reaches into the A-shaped cross section of the wooden roof truss. The
observed deformations gave evidence that the horizontal tensile members of the "A" did no longer
work satisfactory. The masonry of the nave received due to this insufficient action shearing forces
from the roof. This fact was underlined by the observed cracks between the vault and the walls of
the nave, which opened up to 4 cm. O. Kempe [16] developed prestressed diagonal racetrack
CFRP tensile links to relieve the load on the connections of the wooden tensile members and to
reduce the horizontal forces of the base points of the "A" produced in filament winding
technique. The length of the two types of racetrack links is 1.8 and 3.25 m. The links are
U. MEIER 121
connected to the structure, the prestressing elements and to themselves by bolts. Before the
successful application and certification of the system trough the German building authorities it
was tested in full scale at the EMPA in Dübendorf, Switzerland [17]. In Greece Triantafillou [18]
and Schwegler [19] are using CFRP laminates for the rehabilitation of seismic damaged historical
buildings in the old part of the city of Patras. Schwegler is using the same method for the same
purpose in Zurich, Switzerland [20]. The latest Italian venture is strengthening historical
structures with advanced composites. CFRP laminates are used for the retrofitting of masonry
vaults, slabs and walls [21].
The North Sea Oil Industry upgraded a wind wall to a blast wall on the Mobil operated Beryl B
Platform using high strength, high modulus CFRP laminates. This work was part of the safety
improvement plan that was a direct result of the Piper disaster and the subsequent legislation
requiring safety cases to be prepared for each platform [22].
Several prestressed, adjacent concrete box-beam bridges in the State of Delaware have developed
longitudinal cracking on the bottom soffit of the beams. The cause of the cracking was a lack of
transverse reinforcement on the bottom face of the precast, prestressed beams, built prior to 1973.
Advanced composite materials were used to upgrade such a bridge north of Wilmington. The first
in the US bridge rehabilitation field demonstration of carbon fiber tow sheet from Tonen Corp.
was successfully conducted in October 1994 by Chajes and coworkers from the University of
Delaware [23].
Another important pilot project was the repair of Interstate Highway 95 Bridge over route 702 in
West Palm Beach, Florida by Shahawy, Ballinger and coworkers [24] early 1995. A truck
traveling Eastbound on route 702, hit the outermost bridge girder -a 25.9 m long AASHTO Type
m prestressed girder. The truck hit caused a longitudinal torsion (twist) in the beam that resulted
in two major cracks in the girder. Although the capacity of the girder to carry vehicle loads on the
bridge was not substantially reduced, it was necessary to strengthen the beam against possible
additional truck hits that could subsequendy weaken the girder and the bridge. It was repaired
with Mitsubishi Chemical Corporation's Replaik carbon fiber sheet Repair of the damage and
strengthening of the prestressed concrete girder involved removal of broken and loose concrete,
patching with a repair mortar, installation of the CFRP sheets and protection with a UV barrier
paint. The repair was done over a period of a few nights, with a small work crew, with no impact
on traffic flow on 1-95 and relatively small cost
bridge in Tokyo. The method has also here been proven to provide superior external
reinforcement performance and potential cost advantages compared with conventional
strengthening methods. The Japanese systems are mostly consisting of continuous carbon fiber
tows formed into wide sheets adhered to a backing net and removable backing paper are impregnated
in the field with epoxy resins. The composite of fiber and resin is adhered to the surface of
concrete or masonry to reinforce the structures. Hie method has found widespread field
application. There are over 200 installations to date in Japan alone. The Japanese method is
different to the method mostly used in Europe [25]. There pultruded, cured laminates are adhered
to the concrete surface. As long as the surface is even there should result no difference between
this two methods. If the surface should be uneven, there will be a certain danger that laminates
produced insitu follow the contour of the uneven concrete. Therefore exists a certain risk of
peeling off the laminate after loading of the structure.
Among 1992 and 1995 the annual number of publications has been more than tripled. Before
1992 there were not more than five groups working in this field. Today this number is
approximately fifty. Especially Canadian teams [26-28] started to be very successful based on
their Advanced Composite Materials in Bridges and Structures (ACMBS) Network. The lately
announced new Network of Centers of Excellence (NCE) with a total budget of over twenty
million Canadian $ will even accelerate Canadian research. It includes the project "Advanced
composites and integrated sensing technology for structural rehabilitation". In the USA, Asia and
Europe there exists not yet such a good networking like in Canada. In the domains of seismic
strengthening of columns and masonry walls there are centers of excellence at the University of
California in San Diego [29-30] and at the University of Arizona [31-32]. The Universities of
Delaware [33] and South Florida [34] have a lot of competence in flexural strengthening of
concrete, steel and wooden girders. The most important Asian contributions are from Japan [25]. In
Europe there are research groups in England [35], France [36], Germany [37], Greece [38], Italy
[39], Sweden and Switzerland [13-15].
Today's advanced composite materials could solve many of the worldwide rehabilitation
problems. They are characterized by the following types of improvement: enhanced durability and
service life; superior strength; resistance to corrosion, chemicals, and fatigue; initial and life-cycle
cost efficiencies; ease of application; aesthetic and environmental compatibility and ability for
structural control. There is an increasing demand for rehabilitation systems with the
characterizations given above. Therefore more applied research will be needed. The urgent
necessary maintenance work on infrastructure has dramatical implications on the economic life of
a nation. Hence government agencies and industries will be ready to sponsor "useful" research.
Consequently it will be more and more important that research projects do not end up only with
papers and reports but have full scale demonstration projects. That is the most successful way to
transfer knowledge from the universities to practice.
The future is advanced composites. This materials revolution predicted during the 1960s to take
place by the tum of the century has, as expected, been more an evolution than a revolution.
Confidence is very difficult to build in advanced composites in civil engineering but easily
destroyed. Therefore it is essential to use pilot projects to be capable to learn from mistakes and
to convince the owners of structures and the building authorities of the outstanding opportunities
of composites for structural repair and retrofitting. Up to date there are over 700 field
U. MEIER 123
applications worldwide and there are not yet any failures known. The most important factor to
remember is not the cost per kg of advanced composite materials, but rather the cost effectiveness
of the rehabilitation of a structure, considering the life expectancy and the costs of the
alternatives. Figure 3 gives a prediction of the future demand of CFRP for the external flexural
post-strengthening with pultruded laminates.
Fig. 3 Prediction of the worldwide demand of CFRP for external flexural post-strengthening
References
1. Ladner, M„ Prolong, J., Weder, Ch., "Geklebte Bewehrung: Bemessung und Erfahrungen", EMPA-Bericht
Nr. 11615, CH-8600 Dabendorf / Schweiz, Eidgenössische Materialprüfungs- und Forschungsanstalt, EMPA,
April 1990
2. Cantieni, R. and Egger, G. EMPA-Reports 150'582 (in German), Swiss Federal Laboratories for Materials
Testing and Research, CH 8600 Dabendorf, Switzerland, 1994.
3. Meier, U., "Bridge Repair with High Performance Composite Materials", Material und Technik, 1987,15,
225-128 (in German and in French).
4. Deuting, M., "Verstärken von Stahlbeton mit gespannten Faserverbundwerkstoffen", EMPA-Bericht Nr. 224,
1993, (Post strengthening of Concrete Structures with Pre-siressed Advanced Composites, published by the
EMPA in German as Research Report No. 224, CH-8600 Duebendorf/Switzeriand.
5. Kaiser, HP., "Strengthening of Reinforced Concrete with Epoxy-Bonded Carbon Fibre Plastics", Doctoral
Thesis, Diss. ETH Nr. 8918,1989 ETH Zürich, CH-8092 Zürich/S witzerland (in German).
6. Triantafillou, T.C., Deskovic, N. and Deuting, M.. "Strengthening of Concrete Structures with Prestressed
Fibre Reinforced Plastic Laminates", ACI Structural Journal, 1992,89,235-244.
7. Meier, U., Deuting, M. Meier, H. and Schwegler G., "Strengthening of structures with advanced composites",
Alternative Materials for the Reinforcement and Pre-stressing of Concrete, Edited by J. L. Clarice, 1993;
8. "Nachträgliche Verstärkung von Bauwerken mit CFK-Lamellen" (in German), SIA / EMPA Dokumentation!
D 0128, Editor U. Meier, Swiss Society of Engineers and Architects (SIA), Zurich, Switzerland, 1995.
9. Rostasy, F. S., Ranisch, E. H. and Aida, W., "Strengthening of prestressed concrete bridges in the region of
working joints with coupled tendons by bonded steel plates", Part 1 (in German), Forschung, Strassenbau und
Strassenbautechnik, Heft 326, Bonn 1980.
10. Rostasy, F. S. and Ranisch E. H., "Strengthening of prestressed concrete bridges in the region of working
joints with coupled tendons by bonded steel plates". Part 2 (in German), BMV-Forschungsbericht Nr. 15,099,
Bonn/Braunschweig, 1981.
11. Rostasy, F. S. and Ranisch E. H., "Strengthening of Reinforced Concrete Structural Members by Means of
Bonded Reinforcement", Betonwerk- und Fertigteiltechnik, 1981, p. 6-11, 82-86.
12. Rostasy, F.S., personal communication.
13. Meier, U. and Deuting, M., "The application of fibre composites in bridge repair", Strasse und Verkehr, 77,
1991, page 775.
14. Meier, U. (1992). "Carbon Fibre-Reinforced Polymers: Modem Materials in Bridge Engineering", Structural
Engineering International, 2,7-12.
124 REPAIR USING ADVANCED COMPOSITES
15. Meier, U. "Strengthening of Structures Using Carbon Fibre/Epoxy Composites", will be published in
Construction and Building Materials.
16. Kempe, O.; "The stabilization of the gothic roof bearing structure of the Church of our Ladies in Meissen
with CFRP-tension", Techtextil-Symposium 1995, Frankfurt, Presentation No. 541.
17. Esslinger, V. and Sauter, P., " Dachstuhl Vorspannung mit CFK-Schlaufen, Frauenkirche Meissen" (in
German), EMPA Report No. 151 '060, March 1994.
18. Triantafillou, T. and Faidis, M., "Advanced Composites for Strengthening of Historic Structures", IABSE
Rome 1993.
19. Schwegler, G„ "Masonry Construction Strengthened with Fiber Composites in Seismically Endangered
Zones", 10th European Conference on Earthquake Engineering, Vienna Austria 1994.
20. Schwegler, G., "Mit Kohlenstoff gegen Erdbeben: Innovative Gebäudesanierungen", Hauszeitschrift
Basler&Partner AG, Zürich, Herbst 1995, Seiten 7 8.
21. Spena, F. R. et al., "L'uso di materiali compositi per il consolidamento delle stnitore", Centro Internazionale
di Studi di Archtettura Andrea Palladio, Vicenza, Italia, 1995.
22. Galbraith, D., "Offshore structures: upgrading of existing installations", Preceding of "Advanced composites
in building, civil engineering & offshore structures, Institution of Civil Engineers, London 1995.
23. Chajes, M. J. et al., "Rehabilitation of Cracked Adjacent Concrete Box Beam Bridges, Proceedings of the
Symposium on Practical Solutions for Bridge Strengthening and Rehabilitation, Sponsored by NSF, Des
Moines, Iowa, April 1993.
24. Ballinger, C., Craig Ballinger & Associates, 314 Ayito Road, S. E., Vienna, VA 22180-5983, USA, personnel
communication.
25. Nanni, A. "Concrete Repair With Externally Bonded ERP Reinforcement: Examples from Japan", Concrete
International, June 1995, p. 23.
26. Erki, M. A. and Heffeman, P. J., "Reinforced Concrete Slabs Externally Strengthened With Fibre-Reinforced
Plastic Materials", Non-metallic Reinforcement for Concrete Structures, edited by L. Taerwe, published in
1995 by E&FN Son, London, p. 509.
27. Erki, M. A. and Heffeman, P. J., "Equivalent Capacity- A Design Concept for FRP Strengthened Reinforced
Concrete Beams", Ibid [26], p. 517.
28. Wight, R. G., Green, M. F. and Erki, M. A., "Post-strengthening concrete beams with prestressed FRP
sheets", Ibid [26], p. 568.
29. Seible, F., Hegemier., G. A. et al.. The U.S.-TCCMAR Full-Scale Five Story Masonry Research Building
Test University of California, San Diego, Reports No. SSRP-94/01 to 05,1994.
30. Laursen, P. T., Seible, F., Hegemier., G. A. and Innamorato, D., "Seismic Retrofit and Repair of Masonry
Walls with Carbon Overlays", Ibid [26], London, p. 616.
31. Saadatmanesh, H. "Wrapping With Composite Materials", Ibid [26], p. 576.
32. Saadatmanesh, H., Ehsani, M. R., and Li, M. H., Strength and Ductility of Concrete Columns Externally
Reinforced with Fiber Composite Straps, ACI Structural Journal, Vol. 91,1994, p. 434.
33. Chajes, M J. et all., "Reinforcement of Concrete Structures Using Externally Bonded Composite Materials",
Ibid [26], p. 501.
34. Sen, R. et all, "Strengthening Steel Composite Bridge Members Using CFRP Laminates", Ibid [26], p. 591.
35. Swamy, R. N. and Mukhopadhyaya, P., "Role and Effectiveness of Non-Metallic Plates in Strengthening and
Upgrading Concrete Structures", Ibid [26], p. 472.
36. Varastehpour, H. and Hamelin, P., "Structural Behaviour of Reinforced Concrete Beams Strengthened by
Epoxy Bonded Plates", Ibid [26], p. 559.
37. Rostasy, F. S. and Ranisch E. H., "Strengthening of R/C- and P/C-Structures with Bonded FRP Plates", Proc.
of the First ACMBS Int. Conf., Sherbrooke, Canada 1992, p. 253.
38. Triantafillou, T. C. and Plevris, N., "Reliability Analysis of Reinforced Concrete Beams Strengthened with
CFRP Laminates", Ibid [26], p. 576.
39. Arduini, M., di Tommaso, A. and Manfroni, O:, "Fracture Mechanisms of Concrete Beams Bonded with
Composite Plates", Ibid [26], p. 483.
125
Working Sessions
Leere Seite
Blank page
Page vide
127
Summary
This paper presents an application of composite construction to solve the problem of anchoring
the cables to concrete towers in cable-stayed bridges. The use of steel saddles made out of
perforated plates is proposed as a compact, economical and fatigue resistant solution to this
problem. Design and analysis of such a solution is presented as applied to a cable-stayed bridge
with a 203 m long central span.
1. Introduction
A well known design problem in cable-stayed bridges consists in anchoring the cables in the
tower. As concrete towers are usually the most economic alternative to transmit predominantly
axial compression loads to the foundations, many problems arise to transmit horizontal cable force
components from front stays to backstays. Moreover, in moderate spans it is difficult to fit both
anchorages inside the tower. Several solutions have been proposed to this problem [1,2] although
many of them are not folly satisfactory.
Basically the available alternatives consist either in overlapping anchorages (which may lead to
awkward arrangements of tubes, anchorages and cables) or designing an internal gallery inside the
tower (which means complicated prestressing arrangements as well as important dimensional
constraints both for the size of the tower and to manage the jacks) or, finally, anchoring the cables
in steel elements which may adopt very different shapes and sizes [2-5],
The solution which is presented here consists in the use of steel saddles which are embedded in
tower concrete and which support both anchorages. Adherence between the saddles and concrete
is obtained by means of perforated plates. This solution allows a very slender design of the towers
as well as a quick positioning of the saddles (thus avoiding usually lengthy positioning of stay
tubes).
The design of the saddles has to take into account both their intrinsic resisting properties and the
force transmission to concrete. The rationale of the design as well as the analysis of such elements
is presented in this paper to show that this is a very valuable alternative for anchoring cable stays
in concrete towers.
128 COMPOSITE CONSTRUCTION IN CABLE-STAYED BRIDGE TOWERS
2. Tower concept
The project for which we have developped this saddle concept is the Papaloapan bridge (Mexico),
which was opened for traffic in 1995. This is a cable-stayed bridge with concrete deck and a 203
m long main span (Fig. 1). Because of the very flat landscape, the overall design concept was
aimed at minimize the transverse dimensions of all the elements of the bridge; the depth of the
deck is only 1.44 m for 23.4 m width in order to raise the bridge profile as little as posible to
reduce the approach embankments. The stays are arranged in two vertical planes, on both sides of
the deck.
Fig. Papaloapan cable-stayed bridge (bridge and tower elevations; shaft cross sectionj
1.
The towers are made of two independent shafts which are only linked together by a triangular
bracing under the deck. The sense of verticality of these shafts is enhanced by the symmetry of the
cable arrangement, the aspect ratio of the shafts (1/19) and their octogonal constant cross-section.
Then, as result of these conceptual ideas, it was very important to maintain symmetry in all
possible aspects of the design and to avoid any interruption in the vertical contour lines of the
tower shafts. Cable overlapping in the tower was precluded and there was no room for an interior
gallery in the shafts. As a consequence, the natural choice seems to be the use of steel connecting
elements between front and back anchorages. The need to maintain vertical contour lines leaves
no other choice than embedding these steel elements inside the shafts.
3. Saddle concept
The steel elements which are designed to connect front and back anchorages will be called saddles
from now on since their concept is somewhat similar to the saddles of suspension bridges. Their
design has to take into account two main constraints: size has to be reduced such as to fit them
inside the tower and fatigue resistance has to be excellent since there will not be any possibility for
inspection as they are embedded in concrete.
Both reasons made us consider the perforated plate as a suitable concept since it was originally
proposed as a good alternative to the stud connector concept for its fatigue strength [6,7],
Nevertheless important differences may be found between the loads which are applied on a
perforated plate shear connector as described previously and our present saddles.
Shear connection is only part of the static problem in the saddles. They also have to transmit
tension forces between both stays and vertical compression forces to the concrete tower. A
typical design (Fig. 2) consists of a vertical 60 mm thick plate with a double array of 100 mm
diameter circular holes, two horizontal 40 mm thick horizontal plates and some additional
stiffening plates. Design changes slightly as a function of the slope of the cables. Both anchorages
are connected to the vertical plate by means of an intermediate tapered butt-welded cast steel
plate whose cross section changes from rectangular to circular in order to be screwed to an
annular connector holding the cable anchorage.
M.A. ASTIZ 129
Among the load cases which have to be considered there are two limit situations which define the
resisting properties of the saddles:
• The symmetric load pattern corresponds to roughly similar cable forces on both sides (Fig.
3). In this case the saddles are supporting tension forces and for such load case the stress
concentrations around the holes may be very significant. The saddles are also transmitting
vertical components to the concrete shaft through the horizontal plates as well as through the
shear connectors.
• The unsymmetric load pattern corresponds to unbalanced cable forces; the limit case would be
that which happens during erection when only one cable may be stressed (Fig. 3). In this case
the saddles are mainly shear connecting devices since they transmit horizontal and vertical
forces to the concrete shaft by means of the circular holes as well as through the horizontal
plates (in the case of vertical components).
6 ô è cb 6 b (6 Of Or Q- aQ
6 6 à 6 6 6 ef er g- g- a q
TTTTn rmfm
SYMMETRIC FORCES UNBACANCED FORCES
Statical analysis of the saddles could be performed by means of a complete 3D finite element
model (Fig. 4). In this model concrete is supposed to be sliding against the steel plate and shear is
only trasmitted through the concrete dowels. Nevertheless, as complicated concrete cracking
patterns make such analysis very unstable, it is difficult to get ultimate strength values. Then such
model is only used for elastic analyses. Consequently the static problem has been divided into
three parts (the shear connection, the steel plates stresses and concrete stresses) as a way to get a
reliable and safe design.
Although previous analytical and experimental studies are available [6,7], this kind of connection
is not yet standard and there are no generally accepted formulas to define its ultimate capacity. If
we forget about the plate itself, shear capacity is a function of concrete shear strength (with many
limitations because of the complicated geometrical configuration) and of transverse reinforcement
as in any standard shear analysis.
This case is somewhat different from the perforated plate with a single array of holes since in that
case shear may be transmitted in a fully three-dimensional pattern (normal and parallel to the
plate). The multiple array of holes forces shear transmission to concrete only in the normal
direction to the plate. The problem is then bi-dimensional and, in some sense, simpler. Several
methods have been considered for the design.
If we consider each external plane of the plate as a shear joint and we apply standard design rules
such as CEB-FIP [8], Eurocode [9] or ACI [10], the contribution of concrete will be computed as
the product of a shear stress and a reference area. Depending on the choice of both factors this
contribution could range between 10 and 30 kN for a 100 mm diameter hole and a C40 concrete.
The contribution of reinforcement bars is also subjected to different interpretations specially with
respect to the coefficient of friction to be applied to the shear transmission plane (it may vary
between 0.7 if we consider the contact between the steel plate and concrete and 1.4 for
monolithical concrete). A safe estimation of this coefficient of friction (0.9) would give a 608 kN
contribution for a 32 mm diameter rebar (fy<)=420MPa). Then transverse reinforcement is by far
the most important factor in defining shear capacity of the connection.
Another limit state which has to be checked corresponds to cracking at inclined angles but this is
a more standard computation and it happens to be not as demanding for transverse reinforcement
as the previous one. The capacity per hole in the same conditions as before would be 677 kN for
the conservative assumption of having cracks at 45°.
Finally the connection has been checked against the dowel action failure. This point might be
controversial since reinforcement is embedded in the 100 mm diameter concrete cylinder which is
monolithical with the whole shaft. This check gives the most conservative estimation of the
connection shear capacity. According to CEB-FIP Model Code [8], the computed capacity would
be 217 kN per hole.
M.A. ASTIZ 131
Andrä's formulas [7] were also used as an estimation of shear capacity but they have to be taken
with care since they have only be shown to be valid for smaller holes and for a three-dimensional
shear transmission pattern. Moreover they give the global capacity of the connection including
plate failure and our design does not maintain the same scale factor for all geometrical
dimensions. As applied here they would predict a 400 kN capacity with a 36 mm rebar per hole
which would could be converted into 316 kN for our 32 mm diameter rebar.
We finally used the dowel action model as the most conservative estimation. Nevertheless this
design process shows how interesting it might be to carry an extensive testing program to define
precisely the shear capacity of this kind of connection.
The second link in the transmission of forces between the cables and the shaft is the plate
assembly. To check its state of stress a finite element model has been analyzed (Fig. 5). The most
interesting feature of the model may be found in the modelling of the holes. These holes were
considered to be filled with concrete and elastically supported at their center. Horizontal plates
were also elastically supported at their base In this way it is possible to obtain a reasonable
reaction distribution among the different holes and plates and a reliable stress distribution in the
steel plates. Spring constants were evaluated from the three-dimensional finite element analysis
which was mentioned earlier (Fig. 4). Resulting value was expressed as 3. IGcA/D (where Gc is the
concrete shear modulus, A is the area of the hole and D its diameter) to emphasize the concrete
dowel effect which is the origin of this stiffness
Stress results were checked according to AASHTO Standard [11] in service load design
(allowable stress is 0.55 fy) and against fatigue (stress variation at any point is limited to a value
ranging from 27 to 110 MPa depending on the detail) According to finite element results,
maximum stress is by far the governing design criterion; fatigue is not conditioning in any case the
design of the saddle (maximum stress variation is only 9 MPa). This result is important since it
shows, the interest in using this type of connection for a fatigue sensitive structure.
With respect to the fatigue strength of this connection it has to be emphasized that the areas of
the plate which show a certain fatigue risk are the stress concentrations around the holes. Since
the holes are machined and stress relieved afterwards it is very unlikely that residual stresses may
be present a these points. Then the fatigue process may be very well controlled through the static
analysis of the plate.
132 COMPOSITE CONSTRUCTION IN CABLE-STAYED BRIDGE TOWERS
Previous analyses lead to reaction forces at all the elastic supports. These reaction forces are input
to a new finite element model of the shaft to check concrete stresses. This new analysis does not
discover any new aspect of the connection since all the results may be obtained by equilibrium
conditions and standard design rules. This analysis shows that tensile stresses develop in the
neighbourhood of the saddles due to strain compatibility. Although these stresses are not
important, transverse prestressing was arranged in the direction of the saddles to reduce cracking
and further increase the fatigue strength of the connection. This transverse prestressing (4 no.36
mm diameter bars per saddle) is in any case much less than what it would be necessary to
prestress a hollow shaft to fully transmit cable forces.
5. Conclusions
The perforated plate concept has been selected as a compact and fatigue resistant device to be
used simultaneously as a link to transmit tension forces and as a shear connector to absorb
unbalanced forces to a concrete structure. Its use in cable stayed bridge towers simplifies design
and erection. They are specially interesting for short and medium span bridges where the size of
the towers does not leave much space for other connecting devices. A design methodology has
been presented but it is too conservative. Extensive experimental testing should be performed to
define precisely the strength of such connection.
References
[1]. GimsingN.J., "Cable Supported Bridges, Concept and Design", John Wiley & Sons, 1983
[2]. Schiaich J., "On the Detailing of Cable-Stayed Bridges", Cable-Stayed Bridges, Recent
Developments and their Future, ed. M. Ito et al., Elsevier Science Publishers, 1991
[3], Haas G., Petersen A. & Ostenfeld K.H., "The Faro Bridges, New Developments in Design
and Construction", Int. Sypm. on Strait Crossings, Stavanger, 1986
[4], Crémer J.M., "Rotation: Original Mode of Erection of the Ben-Ahin Bridge", 13th IABSE
Congress Report, pp.975-980, 1988
[5], Virlogeux M., "The Normandie Bridge, France: A New Record for Cable-Stayed Bridges",
Structural Engineering International 4, pp. 208-213, 1994
[6], Leonhardt F., Andrä W., Andrä H P. & Harre W., Neus, vorteilhaftes Verbundmittel fur
Stahlverbund-Tragwerke mit hoher Dauerfestigkeit, Beton und Stahlbetonbau, 12, pp. 325-331,
1987
[7]. Andrä H P., "Economical Shear Connectors with High Fatigue Strength", IABSE Symposium
Brussels, pp. 167-172, 1990
[8], CEB-FIP Model Code 1990, Thomas Telford, 1993
[9], ENV 1992-1-3, "Design of Concrete Structures", 1991
[10], ACI Committee 318, "Building Code Requirements for Reinforced Concrete (ACI318-89),
1989
[11], AASHTO, "Standard Specifications for Highway Bridges", 1991
133
Dieter Kraus, bom 1941, obtained his Otto Wurzer, born 1964, obtained his
civil engineering degree in 1969 and civil engineenng degree at the Techn
his Dr -Ing. degree in 1975 at the Techn Univ München After one year, as a
Univ München Project engineer in a member of construction company, he
construction company and Professor for joined the Univ der Bundeswehr
concrete Engineenng at the Univ der München as a research Assistant
Bundeswehr München Dieter Kraus passed
away m March 1997
Summary
The bearing and deformation behavior of a new shear connector called Concrete Dowel was
subject of extensive experimental and theoretical investigations introduced at the University of the
German Armed Forces during the last years The main results of these investigations, a new
mechanical model and a draft of a design concept are presented in this paper
1. Introduction
A new shear connector used for composite beams has been developed since 1986 as „Perfobond
Strip" [1] or „Kombi-Dowels" [2] The so called „Concrete Dowels" are built by parts of the
concrete slab interspersing circle or drop shaped holes, which are located in steel strips welded
upright on I-sectioned steel girders (Fig 1) It is also possible to locate these holes directly at the
upper edge of the web in 1-sectioned steel girders (Fig 2)
The previous level of knowledge about the bearing behavior of Concrete Dowels is mainly based
on tests with small holes [1], [2] In these tests shear yielding of the steel stems remaining
between the holes and local damage of the concrete interspersing the holes were observed as
failure criterias As a simple mechanical model, the concrete interspersing the holes may be
considered as to build a dowel loaded to shear and extreme local compression However the
general validity of this model has not been sufficiently proved yet
As well as the strength, the stiffness and the deformation capacity represent further important
properties of a shear connector The characteristic deformation capacity of small Concrete
Dowels (width b, < 43 mm) does not justify the assumption of a ductile shear connector, while
large Concrete Dowels up to a width of 100 mm showed really ductile load-slip-behavior in initial
push-out-tests
Further extensive experimental and theoretical investigations regarding the bearing and
deformation behavior of Concrete Dowels have been introduced at the University of the German
Armed Forces during the last years The states of stress and the failure mechanism causing
concrete damage are further important subjects of these investigations as well as the main
influence parameters The main results of these investigations and a draft of a new design concept
are presented below More detailed informations about these investigations are given in [3]
134 BEARING CAPACITY OF CONCRETE DOWELS
Fig. 1 Composite beam with Perfobond Strip Fig. 2 Composite beam with Kombi Dowels
2. Experimental Investigations
Altogether 42 push-out-tests using large Concrete Dowels (h bj > 70 mm) have been executed
and evaluated according to Eurocode 4 [4]. Figure 3 shows the typical test specimens and the
investigated variant of Concrete Dowels. The influence of the following parameters has been
investigated in these tests: material properties of concrete, dimensions and shape of holes and
stems, tranverse reinforcement, loading of concrete slab.
The typical deformation behavior (load-slip-relation) of Concrete Dowels observed in the tests
can be divided in three characteristic sections named I, II, III (see figure 4 At lower load steps
(section I) only small deformations occur. According to Eurocode 4 [4] 25 load cycles with an
amplitude A P 0,35 have been introduced in this section to remove the adhesion between
steel strip and concrete slab. Longitudinal splitting cracks occur in the concrete slab at a load level
Pcrack s 0,75 Pmax, which cause a sharp increase in deformation with further rising load (section II).
The maximum shear resistance PmJX is reached, when local parts of the slab surface are wedging
off close to the Concrete Dowels. After reaching Pnux shear resistance is decreasing slowly with
further deformation and progressive concrete erosion (section III).
At the end of the tests some of the concrete slabs have been longitudinally cutted to examine the
condition of the concrete dowels (see figure 5). The concrete interspersing the holes seemed to be
local damaged. Wedges of completely compacted concrete, which have been found close to the
contact surface, must be particularly mentioned.
D. KRAUS, O. WURZER 135
steel strip
i|
(thickness s) |jM
j
bo | b0 s 0,8 bj
•
'
i
11
* + A
0 5 I0 slip s [mm]
3. Mechanical Model
The composite force is transmitted from the steel strip to the concrete slab by extreme local
compression (effect of partial area loading), which acts at the contact surfaces of the hole. The
area, where the load spread is taking place in the concrete dowel, may be separated in two main
parts named zone A and zone B (see Fig. 6).
In the load transmission zone A concrete is confined causing triaxial compression. There the
bearing and deformation behavior of the concrete depends mainly on the pore structure of the
cement stone. Above a critical load step crushing of pore sides occurs caused by the triaxial
compression. Afterwards damaged concrete material fills up the pores.
136 BEARING CAPACITY OF CONCRETE DOWELS
A limit state is reached, when the pores in zone A are completely filled with damaged concrete
material and no further volume reduction is possible there. The pulverized material causes a quasi-
hydrostatic pressure on the confining concrete, which may lead to splitting of the concrete slab
and finally to local wedging off of parts of the slab surface close to the dowels.
The validity of the presented mechanical model has been successfully proved using nonlinear finite
element analyses. The main conditions and results of these FE-analyses are given in [5],
4. Significant parameters
The tests resulted a nearly linear relation between the compressive strength frm and the shear
resistance P,
of Concrete Dowels (see figure 8). Figure 8 shows as well the effect of an increase
in transverse reinforcement A,,,, which rises the shear resistance P„ux of Concrete Dowels.
Reinforcing bars positioned inside the holes take most effect there.
Therefore the depth h of the holes and the thickness s of the steel strip must be considered as the
essential dimension parameters. However, in the presented tests a decrease of the ultimate local
compressive stresses fx transmitted in the contact surface has been observed with increasing Al.
As shown in figure 9 as well, the ultimate values fcc of large Concrete Dowels have not been
considerably influenced by the shape of the holes.
In the presented investigations the pore volume of the cement stone was recognized as an
important parameter for the deformation behavior of Concrete Dowels, but no influence of the
compressive strength fcm of concrete was observed in these tests.
The deformation behavior of Concrete Dowels was getting more ductile with increasing
dimensions (see figure 10). Concrete Dowels with dimensions h b, 70 [mm] showed a similar
deformation behavior like stud shear connectors with a diameter 0= 22 mm. But only a small
influence of the shape of the holes was found especially for large Concrete Dowels.
138 BEARING CAPACITY OF CONCRETE DOWELS
The results of the push-out tests were evaluated according to Eurocode 4 [4] The characteristic
deformation capacity 8,* amounts about 8,0 [mm] for the Concrete Dowels with h b, 70 [mm]
and 5uk 10 mm for the one with h b, 100 [mm] Therefore a Concrete Dowel with
dimensions h b, > 70 [mm] may be considered as a ductile shear connector, because it fiillfills
the criterion 8,* > 6,0 [mm] according to [4]
All other parameters (transverse reinforcement, loading of concrete slab) showed only small
influence on the deformation behavior of the Concrete Dowels
5. Design Concept
Based on the mechanical model presented in chapter 3 the design shear resistance Prj of Concrete
Dowels in the ultimate limit state (acc to [4]) can be determined from equation (2) The factor rj
in equation (2) depends on the dimensions and the shape of the Concrete Dowels and also on the
transverse reinforcement For example r| 6 8 was evaluated for drop shaped Concrete Dowels
with h b, 70 [mm], bo/b, 0,8 s 10 [mm], which requires a transverse reinforcement
determined for the tensile force F, 0 5 PR(1
PRd TJ fck • h • s •
— (2)
Yv
where
fdc the characteristic cylinder strength of the concrete
h height of the holes of Concrete Dowels
s thickness of the steel strip
Yv partial safety factor, yv 1,25 according to [4]
T| factor, found by statistical evaluation of test results
References
[1] Andra, HP Economical Shear Connectors with High Fatigue Strength IABSE
Symposium „Mixed Structures including New Materials" IABSE-Reports Vol 60, pp
167-172, Brüssel 1990
[2] Bode, H Gutachterliche Stellungnahme zum Antrag auf bauaufsichtliche Zulassung der
Kombi-Verdubelung (expert-report on the application for general approval of small Kombi-
Dowels) Not published
[3] Wurzer, O Bearing Capacity of Concrete Dowels Research report Universität der
Bundeswehr München, in preparation
[5] Kraus, D Wurzer, O Nonlinear Finite Element Analysis of Concrete Dowels Computers
& Structures Vol June 1997
139
N. GATTESCO
Assistant Professor Natahno Gattesco. born m 1958.
obtained his engineering degree in
Department of Civil Engineering
1983 at the University of Udine
University of Udine, Italy He is assistant professor at the
University of Udine His mam
research interests concern
composite beams under cyclic
loads, nonlinear analysis of
concrete structures, bolted
connections m wood elements and
structural strengthening of
existing masonry structures
Summary
The results of a numerical simulation of the behavior of steel and concrete composite beams
subjected to cyclic loads are here presented Both partial and complete shear connections were
considered The results show that no significant differences in slip occur with partial interaction
Two groups of beam specimens were numerically tested concerning either connectors distributed
according to the longitudinal shear loads or uniformly spaced The second group showed values
of the slip considerably greater
1. Introduction
In steel and concrete composite beams subjected to cyclic loads a particular attention has to be
paid to the connection because it may reach a premature failure due to fatigue If the connection
during loading cycles remains in the elastic range a large number of cycles is needed before it
collapses (high-cycle fatigue) on the contrary, if inelastic deformations of the connection are
involved, the failure may occur after a limited number of cycles (low-cycle fatigue)
Normally the connections are designed to be able to transmit a longitudinal shear force equal to
the difference between the resisting longitudinal forces, in either the steel member or the concrete
slab (the lesser), at two adjacent sections of maximum moment or free end (complete shear
connection) However, sometimes it is useful or necessary to provide a partial shear connection
In fact, when precast elements are used for the concrete slab it is quite difficult to obtain complete
shear connections because for constructional needs only limited space is available to allocate the
connectors Moreover, often the composite system is needed mainly to increase the stiffness of
the structure so that a complete shear connection may cause an excessive resistance
Current codes [1] for composite bridges do not allow the use of partial shear connections and
moreover the connectors have to be spaced according to the longitudinal shear force so as to
enable the use of the constant stress approach for fatigue This approach is based on the
hypothesis of linear elastic behavior of elements (high-cycle fatigue approach)
As a matter of fact, the load-slip relationship of the connector is nonlinear even for very low
values of the load Moreover at the end of the reloading branch of each cycle an increment of slip
As0 is accumulated due to the progressive damage both in the concrete in front of the stud and in
the shank of the stud
On the basis of such considerations the scope of the present research is to simulate the connector
behavior in a beam by means of a numerical procedure The investigation concerns bridge-type
beams subjected, for simplicity, to a cyclic uniformly distributed load and arranged with different
140 FATIGUE IN STUD SHEAR CONNECTORS
degree of connection in order to check if partial shear connection is really unproposable for
bridges
To emphasize the problem a low-frequency heavy load was considered as reference in the
simulations so to involve significant nonlinear displacements.
Moreover such a procedure allows to obtain the slip history of the studs in the beam which is
indispensable for the experimental study of the resistance of studs to fatigue when significant
inelastic deformations are involved. In fact, as stated in [2], in these cases the tests have to be
performed using a strain-control procedure instead of the common stress-control one The slip
history used in that paper [2] was obtained with the same numerical procedure but using a more
rough load-slip relationship for connectors [3]
2. Numerical model.
A numerical approach based on the four noded finite element of Fig was adopted in the
1
simulation [3], Each element has 12 degrees of freedom which become 8 if the uplift of concrete
slab is neglected and the curvature of concrete element and steel beam are the same
The concrete (C) and the steel (S) are assumed as linear elastic materials while for the connection
the following load-slip relationships are assumed [3,4], In particular the monotonie curve is
described by the function
0 a • (1 -e a +y • ,v
(1)
where 0 is the shear force of the connector, s is the slip between the concrete slab and the steel
beam and the coefficients a, ß, y are constants to be experimentally determined.
The unloading curves refer to a local axis with the origin in points (0„ 5;) where the unloading
starts (Fig. 2)
N. GATTESCO 141
z,= 4^)
X
Q, Q,-^-0-e + yz.)
K
(with
S
(2)
h
where s'i, is the slip, referred to the local axis, corresponding to a zero value of the load and it
depends to the shear load Ol
s' u C]-Ql +C2-Q„ (3)
<5)
where p and e are constants and the suffixj means cycle number
j
The reloading curves are referred to the same local axis Each point of the curves is obtained by
an horizontal translation Ar, (Fig 2), starting from the unloading curve, which depends on s'm
according to the relationship
As,=t;-(s'ul-s\), (6)
where s'm represents the slip value at the beginning of the reloading £, can be expressed as a
function of .v'/ï'h in the following way
_£iii
c'
Ç c3-( 1-e + c5 z,+c6, (7)
c3, c4 c5 are constants and c6 has the relationship
(8)
s m
The increment in slip As01 at the end of the reloading curve, which represents the cumulative
damage at each cycle, may be determined in this way
*sJo,=^L7TL-(Xzi!L + ») (9)
si J
where v and p are constants The coefficients c3 and c4 vary with the number of cycles in the
following way
c3=4-(^-^f) 5 ». S
(10)
u,
ci=C4c\ (11)
3. Specimen details.
The study concerns simply supported bridge-type beams with the geometric characteristics
illustrated in Fig 3 The yielding stress of structural steel was fys 355 MPa (ys 1 00), and the
142 FATIGUE IN STUD SHEAR CONNECTORS
compressive strength of concrete was/ck 30 MPa (yc 1.50) The coefficients needed to
describe the load-slip of connectors under repeated loads were derived from experimental results
[5] and they are reported in Table 1 Also the geometric and mechanical characteristics of stud
connectors are indicated in the same table
nui n Ulliu iu
Qk
IIUI II U I HII 11111111111
I I I I I III 1 |i
X400x40
gjj 11111111 111
^1505x10
L
40000/2 1750 3500
11000/2
As aforesaid the beam was subjected to cyclic uniformly distributed loads which vary between
zero and a maximum value It is supposed that the structure be built without the use of props so
that the own weight of the composite beam was supported by the steel member alone The own
weight (tfif-25 kN/m per each single beam), then, was not considered in the analyses
Five different degrees of shear connection were assumed for the beams (N/Nf =06-1 0) and the
connectors were arranged either following the longitudinal shear diagram (triangular distribution)
or equally sspaced (uniform distribution)
The load applied to the beams with complete shear connection was the maximum service load (on
a single beam) for a three lane bridge according to the Italian code (c/^-56 kN/m). The steel
member, the concrete slab and the connection were designed on the basis of the ultimate load
associated to the cited service loads (yG 35, Jq 1 50) 1
The same geometric characteristics for concrete slab and steel beam were assumed for the
specimens (beams) with partial shear connection The load applied was derived from the
equilibrium method [1] on the critical section (maximum moment) The loads evaluated in this
way are reported in Table 2
4. Results
The numerical simulations concern bridge-type beams with five different degrees of connection
and with two types of arrangement of studs along the span (triangular and uniformly distributed)
The analyses were conducted up to 2000 cycles so as to investigate if the solutions tend to an
almost stable value, in terms of slip or shear force, in the connection It has to be noted that the
Eq (9), representing the cumulative damage in the connection, do not include the fast increase in
damage which precede the stud failure [5,6] So that the results which can be obtained with such a
N. GATTESCO 143
Jmm /Is"
N/Nf Number Load t/k s" s" s"
'max °min Jmax °min Jmax
of studs [kN/m] [mm] [mm] [mm] [mm]
Beams with studs arranged as the longitudinal shear
1 00 255 56 00 0311 0 161 0 612 0 420 0 685
090 230 54 12 0 356 0 179 0 643 0 406 0 631
0 80 204 52 16 0417 0 220 0 690 0 398 0 577
0 70 179 49 07 0 483 0 264 0 736 0 384 0 520
0 60 153 44 56 0 545 0 307 0 771 0 363 0 470
Beams with studs equally spaced
1 00 255 56 00 0 990 0 345 1 266 0 462 0 365
0 90 230 54 12 1 160 0 397 1 420 0 482 0 340
0 80 204 52 16 1 407 0 477 1 650 0515 0 312
0 70 179 49 07 1 658 0 562 1 880 0 559 0 297
0 60 153 44 56 1 860 0 627 2 057 0 585 0 285
400 800 1200 1600 2000 400 800 1200 1600 2000
Number of cycles Number of cycles
(a) (b)
Fig. 4 - Maximum slip versus number of cycles: triangular a) and equally spaced studs b).
This results seems to indicate that the partial shear connection is not a solution very much worse
than the full interaction one
The increase in maximum slip is less pronounced for partial shear connections (AWf= 0 6-
5maxAmax 1
41) than for complete shear connection (^maXAmax ' 97) So that the difference
144 FATIGUE IN STUD SHEAR CONNECTORS
between the maximum slip at n 2000 cycles of the beam with degree of connection N/N( =06
and that of the beam with full interaction (N/N{= 0) is quite limited (~1 26)
1
This results seems to indicate that the partial shear connection is not a solution very much worse
than the full interaction one
In Table 2 are also reported the peak values of the slip (.vmax, smm) of the extreme connectors of
the beam both at the first cycle and at the 2000-th cycle In the last column of the table the values
of the ratio between the minimum and the maximum slip at the 2000-th cycle are reported
The curves of the maximum slip versus the number of cycles for beams with equally spaced
connectors are plotted in Fig 4b Also these curves tend to an almost constant value of the slip
after 2000 cycles. The increase in slip with cycles is more limited than in Fig 4a because a greater
redistribution of the shear force along the span is possible (the connectors close to midspan are
initially very little engaged).
The principal results are summarized in Table 2 The values of the slip in these cases are
significant and then considerable inelastic deformations are involved in the connectors at each
cycle so that it is likely to occur the failure of the connectors after a low number of cycles (low-
cycle fatigue) [2]
The loads considered are heavy repeated loads which may be present on the bridge only some
thousands times during the structure life but the scope of the research was to investigate the
effects under such severe loads Actually at these loads all the other loads with higher frequency
has to be added and then to consider the combined effect
5. Conclusions
The study refers to low-frequency heavy loads and considers five different degrees of connection
The results evidence that there is a considerable increase in slip in the first 300-400 cycles and
then it tends to an almost constant value
The changes between different degrees of interaction, in case of triangular distribution of the
studs, are not very large so that it seems reasonable to considéré partial shear connections also for
bridges.
The case with equally spaced connectors indicate very large values of the slip even in the case of
complete shear connection.
This early results are part of a study which aims to increase the knowledge on the problem of
fatigue in the connection of composite systems Moreover this results provide important
information on the slip history of the connectors in the beam which is needed to perform strain-
control experimental tests on single connectors [2]
Acknowledgement. Thanks to prof Ezio Giuliani for the time spent in stimulating discussion.
References
1 Eurocode No. 4 Design of Composite Steel and Concrete Structures Part 1 1 General Rules
and rules for Buildings (March 1992) and Part 2 Bridges (July 1996)
2 Gattesco N Giuliani E Gubana A, "Low-Cycle Fatigue Test on Stud Shear Connectors",
Journal of Structural Division, ASCE, 123 (2), 1997
3 Gattesco N Giuliani E "Analysis of steel and concrete composite beams under repeated
loads", Technical Report of the University of Udine (Italy), IMTA No 7, Dec 1990.
4 Gattesco N., "Load-Slip Relationship of the Connectors Under Repeated Loads", in
preparation
5 Gattesco N., Giuliani E, "Experimental study on stud shear connectors subjected to cyclic
loading", Journal of Constructional Steel Research, 38 (1), pp. 1-21, 1996.
6 Mainstone RJ., Menzies J.B, "Shear connectors in steel-concrete composite beams for
bridges: Part 2, Fatigue tests on beams", Concrete, Vol 1, No 10, Oct. 1967, pp.351-358.
145
Summary
A test to measure the incremental slip of stud shear connectors under repeated loading is
described. Results of both symmetric and unidirectional cyclic load tests are presented, together
with empirical equations for the rate of slip growth as a function of load.
1. Introduction
Incremental slip between the concrete slab and the steel beam in a composite beam constructed
with stud shear connectors has been observed by a number of researchers while undertaking
fatigue testing of shear studs, but relatively few have focussed on the importance of incremental
slip in its own right, and attempted to quantify the slip behaviour as a function of load and
number of cycles (see for example Hallam (1976), Hawkins & Mitchell (1984), Oehlers &
Carroll (1987), Oehlers & Coughlan (1986)). Incremental slip occurs due to crushing of the
compression concrete ahead of the shear stud, and yielding of the shear stud.
Under monotonie loading this slip is not detrimental to the behaviour of a composite beam, but
rather it is beneficial because it provides the ductility in the shear connection which allows loads
to be redistributed at the ultimate limit state. Under repeated loading, however, the slip
accumulates with each cycle of load. If the slip increments reduce with load cycles, the slip will
stabilise, and the beam can be considered to have "shaken down" to a stable equilibrium
condition. If the slip increments are constant or increase with each cycle of load, then
shakedown will not occur, and incremental collapse of the beam will result. A serviceability
failure of the structure is the likely outcome of this latter situation.
2. Test set up
Many authors have discussed the requirements for a test set up that will give a true
representation of the strength of shear studs in composite beams (see for example Viest (1956),
Chapman (1964), Mainstone & Menzies (1967), Davies (1967a and b), Goble (1968), Ollgaard,
Slutter & Driscoll (1971), Hawkins (1973), Johnson & Oehlers (1981), Maeda, Matsui & Hiragi
(1983) and Gattesco & Giuriani (1996)). From this work, the "push-out" test has evolved as the
standard test for the strength of stud shear connectors. The parameters of the test are described
in various codes of practice (see for example Eurocode 4 (ENV 1994-1-1: 1992)). The test has
gained acceptance despite the long recognised fact that prying forces are developed across the
concrete-steel interface, resulting in tension in the shear studs. Therefore, the test can be
considered as a standardised measure of shear stud performance, rather than an accurate
reproduction of the behaviour that occurs in a composite beam. An analogy would be the cube
or cylinder test for concrete, which has developed as a standard test despite the fact that the
loading conditions for concrete in a beam situation are not accurately reproduced in the standard
test
In undertaking this investigation into the incremental slip of shear stud connectors under
repeated load, an early decision had to be made as to whether to undertake the investigation
using a test rig that would reproduce as accurately as possible the conditions in a composite
beam, or to use the "push-out" test, adapted for cyclic loading, and accept the known differences
between the test conditions and the situation that exists in a composite beam. The push-out test
was adopted for this testing programme, because it was considered that, just as it has gained
acceptance with industry as a standardised measure for the static strength of a shear stud, it has
the potential to be accepted as the standard test for the incremental slip behaviour under repeated
loading.
The two concrete slabs used for each push-out test were 450 mm (in the direction of the applied
load) by 500 mm x 90 mm thick. Each slab was connected to the flange of a steel I beam with
four 12.5 mm diameter by 50 mm headed shear studs. The studs were arranged in two rows
with a lateral spacing of 65 mm (5.2 diameters). The longitudinal spacing was 50 mm. To
prevent longitudinal splitting, the slabs were reinforced with two layers of 8 mm wires at 100
mm centres. Maeda, Matsui & Hiragi (1983) have identified the need for each pair of slabs to be
cast horizontally with the shear studs vertical. Various researchers have used different methods
to achieve this, including casting the slab for one side of the push-out specimen, inverting the
specimen when the concrete has hardened sufficiently, and casting the other slab. This has the
disadvantage that each specimen is by necessity cast from a separate concrete mix, and given the
variation that occurs in concrete strength between mixes, the two slabs will very likely have
slightly different properties. Another method that has been adopted is to cast the slabs onto
separate steel plates which, when welded to a third plate, form the flanges and web of an I beam.
In this research program, a modification of this procedure was undertaken whereby the studs
were welded to a 20 mm plate. Each plate was subsequently connected to a flange of a rolled
steel I beam by bolting the plate to the beam flange with 10 high strength 12 mm bolts. This
procedure met the objectives of casting the specimens horizontally with the studs vertical, and
from the same concrete mix, but avoided the need for expensive and time consuming welding as
part of the specimen preparation. The plate was greased prior to casting to remove friction
between the concrete and the steel.
G. TAPLIN, P. GRUNDY 147
A purpose built reaction frame was constructed for these tests (Figure 1). Because the
specimens were tested under cyclic load, the specimens were clamped between steel plates and
the reaction frame acted in both tension and compression. The steel plates provided horizontal
restraint to the slabs on their compression edge by friction between the concrete and the steel.
The specimens were loaded by a 100 kN servo controlled Instron hydraulic actuator. Monotonie
tests were conducted under displacement control (0.6 mm per minute) and the cyclic load tests
were tested under load control, so that incremental slip measurements could be obtained. Slip
between each slab and the steel plate to which it was connected was measured directly via a
linearly variable differential transformer mounted on the steel plate and reacting against a
bracket on the concrete slab. The compliance of the load frame, and any potential slip in the
bolted connection between the steel plate and the steel I beam was therefore excluded from the
slip measurements.
3. Test results
Four specimens were tested under monotonie loading to establish the static strength of the stud
shear connectors, Pu. The results of those tests are presented in Table 1 below.
Test number Ultimate load per stud, Pu (kN) Slip at ultimate load (mm)
2 47.4
7 50.8 6.11
8 46.0 5.11
10 52.4 6.11
average 49.2 S.78
Table 1: Monotonie test results
148 INCREMENTAL SLIP OF STUD SHEAR CONNECTORS UNDER REPEATED LOADING
1
y P/Pu
0 2 4 6
-1-
-0.6 slip (mm)
Fig. 2 Load versus slip plots for symmetric cyclic (test 15 - left) and unidirectional cyclic (test 13
- right) loading
0.80
t slip (mm) P/Pu=0.52j
0.42
0.32
0.25
-0.20 --
-+- -+- —I 0
50 100 150 0
cycle
Fig. 3 Slip per cycle versus cycle plots for symmetric cyclic (test 15 - left) and unidirectional
cyclic (test 13 - right) loading
G. TAPLIN, P. GRUNDY 149
Six specimens were tested under cyclic load, where the loading pattern involved complete
reversal of the load. The time for one load cycle was typically three minutes. The rate of
loading was increased as the load range increased in order to keep the cycle time constant. The
load was applied for thirty cycles at each load level. A typical load-slip plot is presented in
Figure 2 (test 15). From the figure it is clear that the slip incremented with every cycle, even at
the lowest applied load level, however it is not immediately clear from the figure whether or not
shakedown was achieved at the lowest applied load level. In order to obtain a better
appreciation of the rate of increase of slip with load cycles, the data was presented as slip versus
cycle number. Again, the data for test 15 is presented in Figure 3. It is clear from this figure
that at a given load range the slip increases approximately linearly with cycles, and so it was
possible to fît a straight line to the data within any given load range. This was repeated for
every test, at every load range, for load in each direction. The slope of the straight line gives the
slip growth per load cycle. This is presented in Figure 4 on a logarithmic scale.
P/Pu
0.1 --
I
0.01 -
0.001
2
es
0.0001 -
a
0.0001 ± 0.00001
Fig. 4 Slip growth per cycle versus load plots for symmetric cyclic (test 15 - left) and
unidirectional cyclic (test 13 - right) loading
Three specimens were tested under cyclic load, where the load ranged from zero to maximum
load. The time for one load cycle was typically one and a half minutes. A typical load-slip plot
is presented in Figure 2 (test 13), the slip versus cycle number is plotted in Figure 3, and the slip
growth per load cycle is presented in Figure 4 on a logarithmic scale.
From the data presented in Figure 4, best fit lines were calculated to predict the slip growth per
load cycle. This gave the following relationships:
symmetric cyclic load,
log10[slip growth per cycle (mm/cycle)] -4.41 + [peak load (kN)] x 0.0119 Eqn. 1
It is evident form figure 4 that symmetric cyclic loading leads to a faster growth of slip.
4. Conclusions
Incremental slip occurs in stud shear connectors under repeated loading, leading to incremental
collapse of composite beams. Symmetric cyclic loading leads to a faster rate of slip growth than
unidirectional cyclic loading. By using a push-out test adapted for cyclic loading, it is possible
to obtain empirical equations for the rate of slip growth as a function of load. This information
can be used by designers to estimate the rate at which incremental collapse will occur in
composite beams under repeated loading.
5. References
Chapman, J.C. Composite construction in steel & concrete - the behaviour of composite beams,
Structural Engineer Vol 42 No 4 April 1964 pp 115-125
Davies, C. Small scale push-out tests on welded stud shear connectors, Concrete Sept 1967 pp
311-315
Davies, C. Steel-concrete composite beams with flexible connectors: a survey of research,
Concrete Dec 1967 pp 425-430
ENV 1994-1-1:1992 Eurocode 4: Design of composite steel and concrete structures Part 1.1
General rules and rules for buildings
Gattesco, N. & Giuliani, E. Experimental study on stud shear connectors subjected to cyclic
loading, Journal of Constructional Steel Research vol 38 No 1 1996 pp 1-21
Goble, G.G. Shear strength of thin flange composite specimens, Engineering Journal AISC Vol 5
No 2 April 1968 pp 62-65
Hallam, M.W. The behaviour of stud shear connectors under repeated loading, University of
Sydney School of Civil Engineering Research Report R 281 August 1976
Hawkins, N.M. & Mitchell, D. Seismic response of composite shear connections, Journal of
Structural Engineering ASCE Vol 110 No 9 September 1984 pp 2120-2136
Hawkins, N.M. The strength of stud shear connectors, Civil Engineering Transactions IEAust
1973 pp 46-52
Johnson, R.P. & Oehlers, D.J Analysis and design for longitudinal shear in composite T-beams,
Proceedings ICE Pt 2 71 Dec 1981 pp 989-1021
Maeda, Y. Matsui, S. & Hiragi, H. Effects of concrete placing direction on static and fatigue
strengths of stud shear connectors, Technology Reports of the Osaka University Vol 33 No 1733
Oct 1983 pp 397-406
Mainstone, R.J. & Menzies, J.B. Shear connectors in steel-concrete composite beams for
bridges, Part 1: Static and fatigue tests on push out specimens, Concrete Sept 1967 pp 291-302
Oehlers, D.J. & Carroll, M.A. Simulation of composite beams subjected to traffic loads, in
"Composite construction in steel and concrete" C.D. Buckner & I.M. Viest ed 1987 pp 450-459
Oehlers, D.J. & Coughlan, C.G. The shear stiffness of stud shear connections in composite
beams, Journal of Constructional Steel Research 6 (1986) pp 273-284
Ollgaard, J.G. Slutter, R.G. & Fisher, J.W. Shear strength of stud connectors in lightweight and
normal weight concrete, Engineering Journal AISC April 1971 pp 55-64
Thirugnanasundralingam, K. Continuous composite beams under moving loads, PhD Thesis
Monash University May 1991
Viest, I.M. Investigation of stud shear connectors for composite concrete and steel T-beams,
ACI Journal Vol 27 No. 8 April 1956 pp 875-891
151
Summary
A new numerical method for the evaluation of the characteristics describing the mechanical
behaviour of the shear connection in composite structures, is proposed. The method is based on
FEM. and takes into account linear and non linear behaviour of the materials The reliability of
the method is proved by comparison with experimental results.
1. Introduction
From the very beginning of the application of composite structures many investigators have been
concerned with the behaviour of the shear connection between concrete slab and steel beam.
Many different types of shear connectors have been developed and investigated all over the world.
Among them the most preferable are the headed studs and for the determination of their
mechanical behaviour a numerous of papers has been published In these works the experimental
procedure of the push out test is mostly used in order to obtain the full load-slip curve, the failure
load and the up-lift between slab and beam Although this procedure has been established as the
basic method for the design of composites elements, it has some disadvantages as time and money
costs and need of specialized laboratories.
Due to the complexity of the three dimensional stress and strain state, no mathematical modeling
of the push out test has been appeared. So, there are not closed analytical solutions of this
problem and the only available formulas for the calculation of the parameters affected the
behaviour of the shear connection, are based on the statistical evaluation of test results
In our days Finite Elements Method (F.E.M.) can be used to solve such problems. This is the
basic idea for the development of a new numerical method based on F E M. which is described in
the current paper The three parts consisted a composite section are simulated with different types
of standard Finite Elements (F.E which can be found in the library of any F.E. package The
model is then analyzed taking into account linear and non linear behaviour of the materials and
introducing the appropriate yield criteria According to this procedure all the characteristics of the
shear connection are determined.
The results of the proposed method were compared with experimental data in order to prove the
reliability of the method Experimental results obtained from series of push-out tests performed in
152 LOAD-SLIP CURVE OF SHEAR CONNECTORS EVALUATED BY FEM ANALYSIS
the Steel Structures Laboratory of the Democritus University of Thrace (D.U.TH.), according to
the push out test procedure of EC4 [1],
In the developing of the proposed method the basic idea is the modeling of the concrete slab, the
steel beam and the shear connector of a push-out test specimen by different types of finite
elements. In the beginning a simple model had been investigated taking into account the linear
behaviour of the materials [7] and the reliability of that model had been proved by comparison
with experimental results from Nakajiama and Abe[8]. The same model was used for a new series
of push out tests which were performed in the Steel Structures laboratory of D.U.TH. according
to the procedure and full instructions giving in EC4. The experimental results were more accurate
with the F.E. model of the new specimen [6].
In the present paper, a modified F.E. model taking into account the inelastic behaviour of the
materials, is studied. Each of the three parts of the composite section has been modeled by
different elements. The concrete slab is modeled by non linear volume elements, the steel beam by
a rigid bar element and the shear connectors by non linear beam elements. All these elements are
offered as standard ones in all commonly used F.E. packages. In our case COSMOS/M program
is used. The F.E.M. discretization of the push out test, as it is appeared in the F.E.M. program, is
shown in Fig. 1.
The concrete slab is divided into 48 elements along X-direction, 40 elements along Y-direction
and 5 elements along Z-direction. For the steel beam only one element is needed due to the high
stiffness of the used element. Each of the shear connectors, is divided into 4 elements along the Z-
direction. The resulted aspect ratio of the volume elements is 1 1 1.6, which satisfies the limits for
these aspect ratios [3],
Only the half of the push out arrangement is modeled, due to the symmetry and the solution time.
For the application of the support conditions all nodes at XI and X2 surfaces (Fig. 1) are
restricted to move in Y-direction as these surfaces resist in the compression load. All nodes along
the rigid bar element, which model the steel beam, are restricted to move in Z-direction due to the
symmetry. A basic observation refers to the separation of the concrete behind the shear connector
even when the loading is low [4,10], According to this observation, a double grid of nodes
towards the shear connectors (in Z-direction) is created producing volume concrete elements. The
concrete nodes in front of the studs are merged with the nodes of the beam elements representing
the studs (Fig 1 - Area 1 & 2). The concrete nodes of the second grid behind the studs, are not
merged with the nodes of the beam elements due to the detachment of the concrete. This
assumption eliminates additional stiffness actually added in the area behind the studs.
For each material of the composite section, appropriate yield criterion is considered. The Huber-
von Misses model is used to simulate the behaviour of the shear connector material. For this
model the yield criterion can be written in the form
F V3a-ay 0
where •
a : the effective stress
CTy the yield stress
For the concrete two models are introduced, the Drucker-Pracker and the Huber-von Misses
model In Fig 5 the load-slip curves obtained by both models, are compared with experimental
data. As it seems, better results are obtained by the Drucker-Pracker model, for which the
behaviour and the failure load are more close to the experimental data For this reason the last
model is preferred In this case the yield criterion can be written in the form
F 3aoM - ay + k 0
where :
aM : the mean stress
<7y
•
the effective stress
a, k material parameters
•
For the solution of the above F E problem the Newton-Raphson iterative method is used A
solution is reached when the difference between external and internal forces approaches zero. At
each iteration, displacements are modified to minimize this difference The COSMOS/M program
is used in the non linear mode analysis
In Fig. 2 the stress concentration is shown when the loading is applied to the model
Concentration of stresses is occurred at the surfaces XI, X2 (Fig 1) and in the area in front of the
stud connectors Experiment verifies that concentration, by the crushed concrete area
^Load
ÛLCC1
Beam
Concrete
Slab
Shear
Connectors
3. EXPERIMENTAL WORK
All the experimental work has been carried out in the Steel Structures Laboratory of the
Democritus University of Thrace (D.U.TH.). Two categories of tests were performed. The first
category is concerned with the determination of the mechanical behaviour of the materials used in
the preparation of the push-out specimen. Such tests were performed according to EC2 [2] and
EC3 specifications for the materials of the concrete slab and shear connector, respectively. In the
second category a series of push-out tests were prepared and performed according to EC4 [1]
specifications.
For the concrete, cube specimens, 15x15x15 mm, were tested according to EC2 [2], The mean
compressive strength at 28 days was found 21.4 N/mm2 The elastic modules of the concrete is
calculated according to the procedure proposed by EC2 and the strain-stress diagram is shown in
Fig. 4. For the material of the stud connectors, tension test has been performed according to the
procedure proposed by EC3. The idealized stress-strain diagram of the stud connectors is shown
in Fig. 3. No tests for the material of the steel beam have been made, because the beam is
deliberately very rigid in comparison with the other two elements, so a high stiffness element was
introduced for which no additional information is needed.
Each push-out specimen consists of an HE 260 B steel beam of total length 600 mm, two
concrete slabs attached to the flanges of the steel beam with dimensions 500x600x100 mm for
each slab, and stud connectors with a shank of 13 mm in diameter and 75 mm in height. A short
recess of 200 mm wide and 3-4 mm height is formed to the concrete slabs. The natural bond at
the interface between the steel flange and the concrete is prevented by greasing the steel flanges
before casting the slab. During the execution of the test, the slip between the steel beam and the
concrete slabs is measured by four electronic deflectometers, which are placed at the top surface
of the slab in different points to ensure the accuracy of the overall test. All the data from the
gauges and the load cell are registered electronically into the computer for further elaboration.
The mode of failure in all the cases was the shearing off, just above the weld collar of the
connectors. A very small uplift separation was observed.
Fig. 3 Stress-Stain diagram for the Fig. 4 Stress-Stain diagram for the
of the shear stud connector. concrete slab.
C. KALFAS, P. PAVLIDIS 155
Slip (mm)
4. COMPARISON OF RESULTS
The comparison between experimental and numerical results, is shown in a slip-shear force plot
(Fig. 5). Values of the shear force per stud are on the vertical axis and the corresponding average
slip values are on the horizontal one. The upper dashed curve represents the best fitted line to the
measurements of the push-out tests described in the previous paragraph. The other two curves
have been obtained by the F.E.M. analysis of the present model. The continuous line represents
the results of the analysis when the Drucker-Pracker yield criterion is used for the concrete, and
the lower curve has been plotted for the results obtained by using the Huber-Von Misses model.
As it is shown from the diagram the results of the F.E.M. analysis are in a good agreement with
the experimental ones and always on the safety side. The Drucker-Pracker curve approaches the
curve of the experimental data more closely. The deviation is very small when the load is low and
becomes greater for higher values of the load. The maximum deviation between the experimental
and the F.E.M. analysis (Drucker-Pracker model) results is about 14%, which is acceptable for
the case.
The deviation appears to have greater values in the region corresponding to intermediate values of
loading. There is only a slight deviation at the point of failure. The mean value of the failure load
calculated from the test measurements, is 84 KN. The corresponding values taken by the present
method are 80 KN and 77 KN for the Drucker-Pracker and the Huber-Von Misses model,
respectively.
156 LOAD-SLIP CURVE OF SHEAR CONNECTORS EVALUATED BY FEM ANALYSIS
5. CONCLUSIONS
A simple method based on the finite element techniques has been developed to simulate and
analyze the push out arrangement. Appropriate yield criteria are used for the materials. The
numerical results are close to the experimental ones and on the safety side. The method is reliable
for the elastic, which is restricted, and inelastic area.
The advantage of the present method is that the method permits a prediction of the behaviour of
the shear connection and the stud. Since all the numerical results are conservative, but still close
to the experimental ones, the proposed method can be used substituting or together with the
experimental procedure cover cases such test procedures are a great percentage of the overall
cost in money and time. Also, this method is useful in cases of lack of specialized laboratories
needed for the performance of push out tests.
REFERENCES
1. EC-4, ENV 1994-1, Design of composite steel and concrete structures, 1992.
2. EC-2, ENV 1992-1-1, Design of concrete structures, 1992.
3. Horst, Werkle, 1995, Finite Elemente in der Baustatik, Vieweg Verlag.
4. Jayas, B.S. & Hosain, M.U., 1987, Behaviour of Headed Studs in Composite Beams : Push-
out Tests, Civil Engrs, vol 15, pp 240-253.
5. Kalfas, C. & Pavlidis, P & Tzourmakliotou, D. 1994, Finite element simulation and analysis
Proceedings of the 4th ASCCS International
of steel-concrete composite structures,
Conference, Kocise, Slovakia, pp 548-551
6. Kalfas, C. & Pavlidis, P. & Galoussis, E. & Liolios, A., 1995, A F.E.M. Evaluation of Push
Tests for Shear Connectors, Proceedings of the Eurosteel 1995, May 18-20, Athens, Greece.
7. Kalfas, C. & Pavlidis, P. & Galoussis, 1995, An Approach to simulate the Push out Tests,
Proceedings of the Nordic Stell Construction Conference, June 19-21, Malmo, Sweden.
8. Nakajima, A. & Abe, H., 1994, Experimental Study on Behaviour of Horizontal Shear of
Composite Girder, Proceedings of the 4th ASCCS International Conference, Kocise, Slovakia,
pp 348-351.
9. Narayanan, R., 1988, Steel-Concrete Composite Structures - Stability and Strength, Elsevier
Applied Science Publishers Ltd
10. Yam, C.P., 1981, Design of Composite Steel-Concrete Structures, Surrey University Press.
157
Summary
As far as steel-concrete composite beams are concerned, the use of high performance concrete
(HPC) gives rise to an increased stiffness in the shear connectors and a reduction in ultimate
slipping.
A numerical simulation referring to simply supported beams demonstrates that, at least in the
case of full interaction, the brittle behaviour of the connection does not significantly affect the
load-bearing capacity of the structure because flexural failure of the midspan section due to
rupture of the materials occurs before any shear failure of the connection due to its capacity for
deformation being exceeded.
L Introduction
The mechanical characteristics of high performance concrete (HPC) lead to a variation in the
behaviour of the beam-slab system, relating to two distinct factors:
- a variation in the behaviour of the cross section because of the greater strength and stiffness of
the concrete forming the slab;
- a marked change in the behaviour of the shear connectors between the two elements.
Experimental tests on stud connectors, forming part of a research program being developed at the
University Institute of Architecture in Venice, have shown that, as the strength of the concrete
increases, there is an increase in both the strength and the stiffness of the connection, while there
is a significant reduction in its ductility III.
It is therefore essential to investigate the effects that the brittle behaviour of the connector may
have on the behaviour of the beams - be it in the case of a full connection or of a partial
connection - because if the slip requirement is greater than the slip capacity of the connectors,
then shear failure will occur before the ultimate flexural load is reached.
A parametric investigation was developed, varying both the mechanical characteristics of the
materials and the geometric dimensions of the cross sections, to assess ductility requirements in
different conditions and identify any design rules.
Table 1 Average experimental values for maximum loads and slipping of Nelson connectors
(Rcm=mean cubic compressive strength, shank diameter 19 mm).
3. Numerical Model
The study assessed the load-bearing capacity of simply supported composite beams, with
uniformly distributed load, considering both the behaviour of the HPC and the load-slip
relationships that can reproduce the behaviour of shear connectors in HPC.
Since the problem is far from linear, an incremental procedure till the collapse was used: at each
step, the solution was found using an iterative process on the cross sections discretized in strips
and along the axis of the beam divided into short lengths dx.
The model assumed the linearity of the strains in the steel beam and slab cross sections (indexes
s and c,respectively) (Fig. 1) and any effects of the lifting of the slab were disregarded in view of
their scarce influence on the slipping value emerging from the study 111.
Having assumed the sectional deformations (the strain s0S(x) of the top fiber of the steel part; the
curvature x(x) common to the two elements; and the relative slipping s(x)) as unknown
quantities, the conditions of equilibrium were established. In view of the presence of slipping, the
equilibrium condition for the shear stresses at the beam-slab interface was added to the usual
equilibrium conditions for the translation and rotation of the section.
Taking a secant linear formulation for the constitutive laws into account, the system for finding
the solution took shape as follows:
eos(x)[EA]+x(x)[ES]3p{EA]c=0
sos(x)[ES]+X(x)[El]+^tES]c=-M(x) (1)
%^]sÂËSis=-q(x)
dx dx
where:
S. BULLO, R. PI MARCO 159
- [EA], [ES], [El], [EA]C, [ES]C are the stiffness coefficients of the whole section and of the slab,
depending on the state of deformation across the secant modulus of the materials Ec sec and
Es,seel
- [EA] [ESj are the stiffness coefficients of the steel section, depending on the variation in the
deformation m dx, across the secant modulus relating to said variation Es,sec =dcr/ds);
- q(x) is the shear action per unit of length coming to bear on the connector (q(x)=Rsec s(x); Rsec
is the secant stiffness of the connector per unit of length).
The finite differences method was used, applying backward integration, to solve the differential
equations comprising system (1).
Once the boundary conditions had been established (£os(x=0)=0; x(x=0)=0; s(x=L/2)=0), the
solution was obtained by applying the shooting technique, i.e. having assigned an arbitrary value
to s(x=0), system (1) was then solved for the subsequent sections up to the midspan. The
procedure was iterated, updating s(x=0), until the condition s(x=L/2)=0 was satisfied.
4. Numerical analysis
The analysis was performed in order to emphasize the extent of the maximum slip requirement in
relation to changes in the following parameters:
- the reaction of the connector and slab to changes in the strength of the concrete;
- the span of the composite beam;
- the arrangement of the connectors along the beam.
where: a=1.7, ß=1.15 mm-', sc=7.0 mm for the type 1 curves (studs in ordinary concrete)
a=0.5, (3=1.10 mm*1, sc=3.5 mm for the type 1 curves (studs in HPC) (Fig. 2)
The investigation considered beams characterized by the cross sections illustrated in Fig. 3, with
spans of 15,25, 30 and 40 m, mean cylindrical concrete strengths of 35 and 80 MPa, and
connectors having the constitutive laws of Fig. 2, according to the following table.
CODE NC1-15 NC2-15 NC1-25 HP2-15 HP2-25 NC1-30 NC1-40 HP2-30 HP2-40
fCm (MPa) 35 35 35 80 80 35 35 80 80
P-s Type 1
Type 2 Type 1
Type 2 Type 2 Type 1
Type 1
Type 2 Type 2
Span L (m) 15 15 25 15 25 30 40 30 40
L/H 13.0 13.0 21.5 13.0 21.5 13.3 17.8 13.3 17.8
10+18 cm"
For each of the cases considered, the most significant results are given in Table 3.
CODE NC1-15 NC2-15 NC 1-25 HP2-15 HP2-25 NC 1-30 NC 1-40 HP2-30 HP2-40
Mmax.sez (LN*m) 4830.8 4830.8 4830.8 5113.3 5113.3 30225.0 30225.0 31645.0 31645.0
Mmax/MmaXjsez 0.9933 0.9957 0.995 0.986 0.987 0.987 0.989 0.991 0.988
slipmax (mm) 3.275 2.583 4.095 1.651 1.841 3.196 3.367 2.563 2.717
sl'Pmax/Su 0.468 0.738 0.585 0.470 0.525 0.457 0.481 0.732 0.776
These data show that, assuming a perfect interaction between the two materials (sc=ss), the load-
bearing capacity of the beam (Mmax) is always lower than might be expected on the basis of the
flexural strength of its midspan cross section (Mmax,sez).
In fact, the increase in the load applied, and consequently in the acting moment, coincides with
an increase in the rate of slipping ds/dx=EC-£s (curves a and b, Fig. 5, for the two types of
concrete) and this leads to a reduction in the values of both the maximum resisting moment and
the ultimate moment (curves c and d for the ordinary concretes, and and g for the HPC, Fig. 5).f
The maximum load that the beam can withstand is the load at which the acting moment at the
midspan cross section reaches the same value as the resisting moment, with the corresponding
value of the parameter ds/dx; further loading is impossible because it would induce a
corresponding increment in the acting moment and in ds/dx, and hence a reduction in the
resisting moment.
S. BULLO, R. PI MARCO 161
New equilibrium conditions beyond the maximum load condition can only be achieved by
reducing the actions, and the branch of the loading curve up to failure due to the maximum
strength threshold being reached becomes unstable.
u/^max,sei.
i.v
2x/L
Fig. 5 Shear stresses and slipping along the connection for beams NC1-15 and HP2-15, andfor
stud connector arrangements A and B.
Fig. 5 shows the distributions of the theoretical resisting actions of the connector (which are
uniform in case A, whereas they are twice as great in the supporting area as in the middle of the
beam in case B) and the distributions of the actual stud reactions when the maximum load-
bearing capacity is reached.
From a comparison of the diagrams, it is clear that the design conditions, based on the
assumption f a full plasticization of the connector, coincide substantially with the actual situation
in both types of concrete.
162 THE LOAD-BEARING CAPACITY OF STEEL-HPC COMPOSITE BEAMS
The beam-stud-slab system therefore seems, even in the case of HPC, to allow for a
redistribution of the stresses, having a generally ductile behavior instead of the brittle behavior
detected in the connector alone.
The two distributions consequently prove virtually equivalent in terms of load-bearing capacity,
but the same cannot be said for the slipping requirement. The diagram in Fig. 5 shows that, in the
case of ordinary concrete, the type B arrangement of the connectors leads to a more uniform
distribution of slipping than with the type A arrangement, also reducing the maximum slipping
value; in the case of HPC, on the other hand, the uneven distribution of the slipping
phenomena is accentuated and their maximum value, which is reached nearly the middle of the
beam, is greater than in the case of the uniform (type A) connector arrangement.
6. Conclusions
This study has shows that the behaviour at failure of composite beams made with HPC is
influenced not only by changes in the behaviour of the connection, but also by the interaction of
the latter with the resistant and deformative reaction of the slab.
In the cases considered here, the ultimate load coincided with flexural failure, thanks also to the
contribution of the HPC slab towards reducing the slip requirement.
The study has also demonstrated that the arrangement of the connection is of little significance
for the purposes of flexural failure, whereas the arrangements considered were far from
comparable in terms of any failure occurring due to the ultimate slipping threshold being
exceeded, so special attention must be paid to the identification of the ideal arrangement of the
connectors.
References
1. S. BULLO, R. DI MARCO: "Effects of high-performance concrete on stud shear connector
behavior", Proceedings of Nordic Steel Construction Conference, Malmö, Sweden, June
1995, pp. 577-584.
2. J. M. ARIBERT, K. ABDEL AZIZ: "Calcul des poutres mixtes jusqu'à l'état ultime avec
un effect de soulèvement à l'interface acier-béton", Construction Métallique, n°. 4, 1985,
pp.3-36.
3. CEB: Bulletin d'Information n. 228 "High performance concrete - Recommended
extensions to the Model Code 90", July 1995.
4. J. G. OLLGAARD, R. G. SLUTTER, J. W. FISHER: "Shear strength of stud connectors in
lightweight and normal-density concrete", Engrg. J. Am. Inst. Steel Constr., vol. 8, April
1971, pp. 55-64.
5. J. M. ARIBERT, A. G. LABIB: "Modèle de calcul élasto-plastique de poutres mixtes à
connexion partielle", Construction Métallique n°. 4, 1982, pp. 3-51.
6. J. M. ARIBERT : "Etude critique par voie numérique de la méthode proposée dans
l'Eurocode 4 pour le dimensionnement des poutres mixtes acier-béton à connexion
partielle", Construction Métallique, n.l, 1988, pp. 3-26.
7. R. P. JOHNSON, N. MOLENSTRA: "Partial shear connection in composite beams for
buildings", Proc. Instn. Civ. Engrs., Part 2, vol. 91, Dec. 1991, pp. 679-704.
8. S. BULLO, R. DI MARCO: "Influenza délia limitata capacità di scorrimento délia
connessione sulla portanza di travi miste realizzate con calcestruzzi ad alte prestazioni",
C.T.E., Napoli, 7-9 November 1996, pp. 21-30.
9. D. J. OEHLERS, C. G. COUGHLAN: "The shear stiffness of stud shear connectors in
composite beams", Journal of Construct. Steel Research, n.6,1986, pp.273-284.
10. D. J. OEHLERS, G. SVED: "Composite beams with limited-slip-capacity shear
connectors", Journal of Structural Engineering, Vol. 121, No. 6, June 1995, pp. 932-938.
11. EUROCODE N.4: "Design of composite steel and concrete structures", Part 1-1 : General
rules and rules for buildings, ENV 1994-1-1, October 1992.
163
Fritz Münger, born in 1944, Manfred Wicke, born in 1933, has Norbert Randl, born in 1967,
graduated with a degree in civil his civil engineering and Ph.D. graduated with a degree in civil
engineering from the Swiss Fed. degrees from Vienna Technical engineering from the Innsbruck
Inst, of Technology, Zurich in University. He worked in and later University. Since 1993 he has
1968. For 20 years he has worked headed a design office of a firm, worked in the Institute for
in Switzerland and South Africa as primarily involved in the design of Concrete Structures at Innsbruck
a design engineer in industrial and buildings, bridges and power University.
bridge construction. Since 1989 he plants. Since 1971 he has been a
has been a project manager with full professor for concrete
Hilti Corporate Research. structures at Innsbruck University.
Summary
For the design of concrete composite constructions, the transfer of internal stresses across the
bond interface between new and old concrete is a critical aspect. A design method has been
developed with the aid of specific tests for rough, sand-blasted and smooth surfaces. Test results
known from literature have been taken into account. This new design approach considers
cohesion, friction and the shear resistance of the reinforcement in determining the effective shear
transfer. It has been found that, contrary to the usual design approach, the full yield strength of
the reinforcement cannot be equated to the tension clamping force across the interface.
1. Introduction
Placing new concrete on older concrete is a
routine task in construction. It occurs at every
joint in concrete construction work. For some
time now, placing overlays has gained in
importance as a result of the more frequent
strengthening of existing structures. In such
cases, a loadbearing layer of new concrete
(overlay), is placed on the existing concrete
structure. This overlay is usually cast directly
or placed as shotcrete. It functions to augment
the flexural compression or flexural tension
zones, depending on the placement.
One of the problems encountered is the
transfer of internal stresses acting across the
bond interface between the old and the new Fie. 1: Strengthening of a bridge
concrete. In this respect, knowledge is
required of the resistance of the bond interface to tensile and compressive forces in any direction.
Initially, stresses in the bond interface result from a combination of external loads and internal
forces of constraint. It must be borne in mind that stresses due to shrinkage and temperature
gradients in the new concrete typically reach their maximum at the perimeter of the overlay. The
164 SHEAR TRANSFER IN CONCRETE-CONCRETE COMPOSITE STRUCTURES
combination of external and internal stresses often exceeds the capacity of the initial bond, thus
requiring the designer to allow for a debonded interface. This is particularly true in the case of
bridge overlays which are subject to fatigue stresses resulting from traffic loads. Furthermore,
these stresses are dependent on time, and bond failure can take place years after overlay
placement.
The test results confirm the strong influence 200 Test No. 18 (Water-blasted: 2<j>12~)
of roughness on shear resistance and shear
stiffness. If the load-displacement curves ^160
are regarded in conjunction with the
180
S 140
M
the
£ 120 -f *
three o 100 - friction and surface interlock
measured displacement,
components of cohesion, friction and dowel a 80 -
action can be isolated and determined J 60- —
40 dowel action
quantitatively. They make different 20-
contributions to the overall resistance (figs. 0 -1—I—1 1 1 1 1 1
F—I—I
1
1—1- 4—I—I-
1
Generally, an interface must be assumed to be cracked for design work. Connectors installed
across the interface must be positioned in such a way that the shear force, VRd, between the new
and the old concrete is transferred in the ultimate limit state.
VRd xRdj ' bj ' lj > V.Sd (1)
Vrj Design value of interface shear force resistance tRdj Design value of the transferable interface shear
Vsd Design value of interface shear force due to stress in zone under review as per formula (2)
actions bj Width of interface in zone under review
lj Length of interface in zone under review
The design value of transferable shear stress, xRdj, can be calculated using formula (2) [10].
When doing so, the upper limit is given by the transferable compressive stress in the strut model
for concrete:
T„dbasic value of design shear strength as per [1], A, / b [ amount of reinforcement from connector
section 4.3.2.3 (smaller value of new/old concrete) in zone under review
1^ cohesion factor as per [13], table 1 S 0,6 fj normal stress to external loads acting on the
p. coefficient of friction as per [13], table 1 interface (compression positive)
a coefficient of dowel action as per [13], table 1 design value of yield strength of connectors
ß coefficient as per [13], table 1 design value of cylinder compressive strength of
v coefficient as per [1] formula (4.20) concrete (smaller value of new and old concrete)
mean roughness derived from sandpatching method
k coefficient for tensile force in the connector
as per [13], table 1
(i.e. difference between peaks and valleys 2r)
An evaluation of equations (1) and (3) for S500 grade steel is provided in the diagrams 1 to 3.
7.0 7.0 7.0
C50/60
6.5 — C50/60 6.5 — C50/60 6.5
— C45/55 6.0
— C45/55 6.0 — C45/55 6.0
C40/50 — C40/50 — C40/50
C35/45 5.5 5.5
— C30/37
5.5 — C35/45 — C35/45
C25/30 5.0 — C30/37
/ 5.0 — C30/37 5.0
„
-L 2U Z5
V, /
/~
*/, —
4.5 ""j
4.0 j
3.5 g
3.0
— C25/30
— C20/25 / 4.5
4.0
3.5 &
3.0
— C25/30
— C20/25
s£
S
4.5 "g
4.0
3.5
3.0
J
g
V
f
A 2.5 H 2.5
2.5 i
A
y 2.0
1.5
/ f A 2.0
1.5
2.0
1.5
1.0
0.5
0.0
/ 1.0
0.5
0.0
1.0
0.5
0.0
»rjTt«ooocj*r«oo
Amount of reinforcement
[%] p Amount of reinforcement p[%]
p [%] Amount of reinforcement p[%]
Dig 1: Water blasted surfaces Dia 2: Sand-blasted surfaces Dia 3: Smooth surfaces
F, MÜNGER, M. WICKE, N. RANDL 167
or sand-blasted.
To shorten the anchorage length in concrete overlays, heads or plates can be provided. The
concrete cone aswell as the bearing stresses in the concrete below the anchoring component must
be checked. The methods of calculation are given in [14].
When surfaces are smooth, shear dowels must be anchored at depths of at least 6 times the
diameter in each case, or, better still, to avoid dowel pull-out at large displacements a value of 9
times the diameter is recommended.
12.0
Mattock, Walraven, Daschner, Hilti
Mattock [5]: "rough" ACI (r =3mm );
fy=350 N/mm1, fc=20-45 N/mm2
10.0
Walraven [6]: cracked concrete;
« fy=450 N/mm2, fc=25-32 N/mm2
8.0 Daschner [7]; raked surface;
M fy=540 N/mm2, fc =15 N/mm2
Dasctaner,Mattock,Paulay,Hanson, Hilti
• Daschner [7]: trowel;
fy=450-1200 N/mm2, fc=15-22 N/mm2
• Daschner [7]: un worked surface;
fy=450-l200 N/mm2, fc =10-17 N/mm2
•Mattock [5]: trowel;
fy=350 N/mm2, fc=35 N/mm2
Paulay [4]: trowel;
fy 318 N/mm2, fc=24 N/mm2
• Hanson [3]: trowel;
fy 345 N/mm2, fc=22-29 N/mm2
Hilti [10]: unworked surface;
fy=508-653 N/mm2, fc=33 N/mm2
Hilti [10]: unworked surface;
fy=508-653 N/mm2, fc=40 N/mm2
• Hilti [10]: unworked surface;
fy=508-653 N/mm2, fc=17N/mm2
fi8 .n —Design function (charact.values) [10]:
0 1 4 6 1U 12
smooth; fck=20 N/mm2, fyk=500 N/mm2
p fy [N/mm2]
Fie. 8 Smooth surfaces
6. Summary
Contrary to design methods given in the literature, dowel resistance is considered along with
cohesion and friction when determining shear resistance. With increasing roughness of surfaces,
shear resistance and shear stiffness improve greatly. Furthermore, the distribution of total
resistance shared by the three components changes considerably. The design method makes use
of one single equation for calculating the resistance from the three components. In some cases, it
is sufficient for the concrete overlay to be anchored at its perimeter.
This new design approach is particularly notable for its transparency. It is verified by the
literature as well as by extensive testing conducted at Hilti Corporate Research. Through the use
of design diagrams, the method can be made particularly straightforward for designers.
Literature
1. EC 2; Design of concrete structures; ENV 1992-1-1: 1991;
Part 1. General rules and rules for buildings
2. EC 4; Design of composite steel and concrete structures: ENV 1994-1-1: 1992;
Part 1-1. General rules and rules for buildings
3. Hanson, N.W.; Precast-Prestressed Concrete Bridges-. Horizontal Shear connections. Journal of the Portland
Cement Association, Research and Development Laboratories, V.2, No.2, May 1960, pp.38-58
4. Paulay, T., Park, R. and Phillips, M.H.; Horizontal Construction Joints in Cast in Place Reinforced Concrete,
ACI-Special Publication SP.42 Shear in Reinforced Concrete, 1974, Vol. II, 99. 599-616
5. Mattock, A. H.; Shear Transfer across an Interface between Concretes Cast at different Times,
Structures & Mechanics Report SM76-3, University of Washington, Seatle, 1976,68 pp.
6. Walraven J. C., Reinhardt H. W.; Theory and experiments on the mechanical behaviour of cracks in
reinforced concrete subjected to shear loading. Heron Vol.26/1981.
7. Daschner, F.; Versuche zur notwendigen Schubbewehrung zwischen Betonfertigteilen und Ortbeton.
Deutscher Ausschuss für Stahlbeton, H 372 Berlin, Ernst und Sohn, 1986
8. Tsoukantas S. G., Tassios T. P.; Shear Resistance of Connections between Reinforced Concrete Linear Precast
Elements. ACI Journal, May-June 1989.
9. Menu, C.; Bonding of Old and New Concrete for Monolithic Behaviour
Swiss Federal Institute of Technology, Zürich. Report No. 185,1991.
10. Randl, N.; „Untersuchungen zur Kraftübertragung zwischen Neu- und Altbeton bei unterschiedlichen
Fugenrauhigkeiten"; Dissertation in Vorbereitung, Universität Innsbruck.
11. Hilti, Fastening Technology Manual, Rebar Fastening Guide B 2.2,1994
12. Hilti, Fastening Technology Manual, Adhesive Ankers B 3.2, 1994
13. Hilti, Fastening Technology Manual, Rebar Fastening Guide for Overlays, B 2.3,1997
14. CEB-Guide; On the Design of Fastenings in Concrete, Part 3, Draft March 95
Characteristic Resistance of Fastenings with Cast-in-Place Headed Anchors.
169
Summary
This paper presents information on the load-deformation behaviour and the manufacture of
different types of timber-concrete connection. The failure modes of four different connections
and their influence on the load-deformation behaviour and the failure modes of timber-concrete
composite beams is outlined. The design of timber-concrete connections with dowel-type
fasteners is summarised and the effect of load sharing between fasteners on the load-carrying
capacity of the composite beam is emphasised.
1. Introduction
Timber-concrete-composite structures, used as bending members in floor systems, mostly consist
of timber members in the tensile zone, a thin concrete layer in the compression zone and the
connection between timber and concrete. The main advantages of this type of composite
structure are:
2. Types of Connection
During a co-operative research project carried out at Delft University of Technology in the
Netherlands and Karlsruhe University in Germany (Blaß et al. 1995), timber-concrete-composite
structures with four different types of connection were studied:
170 CONNECTIONS FOR TIMBER-CONCRETE-COMPOSITE STRUCTURES
• especially designed screws driven into the timber without pre-drilling under an angle of ± 45°,
• punched metal plate fasteners,
• grooved holes in timber beams filled with concrete and combined with a dowel, and
• grooved holes in laminated veneer lumber filled with concrete.
2.1 Screws
The screws used in the study were specifically developed for timber-concrete connections. They
are driven into the timber without pre-drilling, resulting in low labour costs for the manufacture
of the connections. Connections with different screw arrangements were investigated by Timmer-
mann and Meierhofer (1993). Instead of placing the screw perpendicular to the joint between
timber and concrete, they placed the screws under an angle of ± 45°, resulting in a truss-like
loading of the connections, where the screws are loaded in tension and compression rather than
in bending. This fact leads to a much stiffer connection compared to an arrangement
perpendicular to the timber surface. The screw has two heads, the lower pressing the concrete
formwork onto the timber beams. The upper head together with a part of the shank is encased in
concrete.
The punched metal plate fasteners, which are very common as connectors in timber trusses, were
first bent about their longitudinal axis to form a right angle. The nails on the plate part later
encased in the concrete layer were cut off in order to prevent voids in the concrete close to the
plate. The other half was then pressed into the timber beam. This type of connection has to be
prefabricated before shipping the beams to the building site.
11111111111111 r
In order to manufacture this type of cleat joint, first grooves with a diameter of 70 mm are routed
30 mm deep into the timber beam surface. Within this indentation, a hole with a diameter of 20
mm is drilled to take the steel dowel. The steel dowels - short pieces of concrete reinforcement
bars - are then driven into the 20 mm holes. Concrete finally covers the upper part of the dowels
and fills the remaining space in the grooves.
5 '
4
1 | I
r 4 TLÎJT
1
À
150 300 300 50 50
This type of timber-concrete connection is used for composite plate structures, where instead of
timber beams laminated veneer lumber (LVL) as sheet material is used in the tensile zone. The
joints are manufactured by grooving circular holes with a diameter of 115 mm and a depth of 15
mm into the LVL surface. The holes are conical in vertical direction in order to prevent a
separation between timber and concrete. The concrete filling the flat holes forms a type of cleat
which is able to transfer shear forces. In order to avoid a brittle failure of the concrete cleats, they
are reinforced. The manufacture of the indentations is carried out with computer-controlled
routing machines resulting in very economical connections.
"2T
Fig. 4 Timber-concrete connection with grooved indentations in LVL
172 CONNECTIONS FOR TIMBER-CONCRETE-COMPOSITE STRUCTURES
Depending on the thickness of the intermediate layer between timber and concrete, the screws
loaded in tension were either pulled out of the timber or failed in tension in the threaded part of
the shank. After reaching the maximum load, the load on the connection decreased with
increasing withdrawal of the screw (Figure 5 top left). The average maximum load for one pair of
screws was about 18 kN at a displacement of about 1 mm.
50 -NAG-2
I^ w
30
D-SCI4-14
/ Z' r NAC -5
Sas
\
1 J
/
/
1
I>sc H*l
1
0 1 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 0 I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Verschiebung v (mm)
Verschiebung v (mm)
B-N+S-42 F-NPL-48
/
50
10
B -N+ >-41
n - 30 F-Nl>1^5
^ 20
1
10
1
0 I 2 3 4 S 6 7 ft 9 10 11 12 13 14 15 16 0 1
2 3 4 5 6 7 ft 9 10 11 12 13 14 15 16
The connections with punched metal plate fasteners failed due to bending and subsequent
withdrawal of the punched out nails out of the timber. In some cases, nails failed in tension at the
metal plate. The mean maximum load for one plate was about 48 kN, the stiffness modulus at
service load level about 50 kN/mm.
Grooved connections exhibited a particular plastic deformation capability. The tests were
stopped after a relative displacement of 15 mm between timber and concrete was reached. The
failure for grooved connections with dowels was caused by dowel bending combined with
concrete cracking in the vicinity of the dowel. The maximum load for one connector unit
(indentation plus dowel) was about 50 kN, the stiffness modulus at service load level about 75
kN/mm. Grooved connections in LVL acted as large concrete connectors. In some cases, the load
was still increasing when a displacement of 15 mm was reached. Failure was caused by reaching
the embedding strength of the laminated veneer lumber. The average maximum load was about
50 kN per indentation, the corresponding stiffness modulus 120 kN/mm.
H.J. BLASS, M. SCHLAGER 173
5400
no,
137
27
600
150 J1 1500 L 1500 L 1500 J 150
4500
Fig. 6 Test specimens for timber-concrete composite beams. Connections are screws, punched
metal plate fasteners, grooved connections with dowels (top), and grooved connections in
LVL (bottom)
The influence of the connection behaviour on the load-deformation behaviour of the composite
beams depends both on the load-carrying behaviour of the connections and on the load level of
the connection loads before the failure of the beam. Since all four types of connection
approximately exhibit an elastic-plastic load deformation behaviour, the load-deformation
behaviour of the tested composite beams essentially depends on the load level of the fasteners. If
the fasteners between timber and concrete govern the design, the highly loaded fasteners close to
the supports will approach their load-carrying capacity, deform plastically and a certain amount
of plastic deformation of the beams is to be expected before failure. If, on the other hand, the
connection loads remain in the elastic range until the beam fails, the beams will basically behave
linearly until failure.
Depending on the number and capacity of the connections, the behaviour of the tested beams
ranged from linear-elastic until failure until elastic-plastic. The beams with the highest
connection capacity (grooved connections in LVL) exhibited a linear load-deformation behaviour
until failure. The brittle failure was caused by the failure of the LVL-layer under combined
bending and tensile stresses.
The composite T-beams eventually all failed due to the combined bending/tensile failure of the
timber beams. The composite beams with screws and those with grooved connections with
dowels showed a pronounced plastic deformation before failure due to the plastic deformation in
the timber-concrete connections. The failure of the connections close to the supports -
withdrawal of the screws or splitting of the timber end cross-section and cracking of concrete,
respectively - before the failure of the beams could be clearly observed in the tests. The beams
with punched metal plate fasteners displayed a slightly curved load-deformation diagram,
indicating the fact that the first connections approached their load-carrying capacity before the
failure of the composite beam.
174 CONNECTIONS FOR TIMBER-CONCRETE-COMPOSITE STRUCTURES
If the withdrawal strength and stiffness of screws or threaded nails is known, Eurocode 5 Part 2
also provides a method to design inclined fasteners for timber-concrete connections. The
analytical model assumes a truss-like behaviour of the components, where for uni-directionally
inclined fasteners the shear force is transferred by tensile forces in the fasteners and compression
forces between timber and concrete, and for two-directionally inclined fasteners by tensile and
compression forces in the fasteners.
If tests to determine the load-carrying capacity and the slip modulus of timber-concrete
connections show a distinct plastic behaviour, a redistribution of loads from highly loaded
fasteners to less loaded fasteners will occur in composite beams, as soon as the most stressed
fasteners deform plastically. If the connection governs the design of the composite beam with a
large number of fasteners, the characteristic load-carrying capacity of the beam therefore depends
on the characteristic strength of a number of connections loaded in parallel, rather than on the
characteristic strength of a single connection. This load-sharing increases the load-carrying
capacity of the composite beam and should either be taken into account during the connection
design or when determining the characteristic strength of timber-concrete connections.
6. References
Bajolet, D., Gehri, E., König, J., Kreuzinger, H., Larsen, HJ., Mäkipuro, R., and Mettem, C.
(1997). PrENV 1995-2. Eurocode 5 - Design of timber structures - Part 2: Bridges.
Blaß, HJ., Ehlbeck, J„ Van der Linden, M.L.R. and Schlager, M. (1995). Trag- und
Verformungsverhalten von Holz-Beton-Verbundkonstruktionen. Forschungsbericht der
Versuchsanstalt für Stahl, Holz und Steine, Abteilung Ingenieurholzbau der Universität
Fridericiana Karlsruhe, 74 pp.
European Committee for Standardization (1993). ENV 1995-1-1. Eurocode 5 - Design of timber
structures - Part 1-1 : General rules and rules for buildings.
Summary
Buildings can be erected using timber in the floor in order to be more economical and lighter than
conventional concrete constructions. For a span of three to six meters, timber floors made of
vertical nailed planks are built. For a span of seven to fifteen meters, this kind of timber floor is
connected to a concrete deck and becomes a composite wood-concrete system. In this case, the
shear strength is taken by a groove in timber filled with concrete. The timber part can be made of
planks, roundwood, glued laminated timber, regarding the expected aspect of the ceiling.
1. Introduction
Given the constant decrease in energy resources and the new consideration of environmentals
parameters, a way for an increasing use of wood in the construction is open. But the wood
construction should not remain sectarian. In order to be economical, it must work together with
other materials, traditionally used in construction. In the history of timber construction, there have
-
always been composite constructions timber frameworks with glue or mortar, walls of stone and
bricks - the most lasting oneswere in timber architecture. Examples from China and Japan to
Frank and Alsacian framework constructions are well known. Essential criteria are a better
behavior of the whole construction during a fire, as well as acoustics and vibration properties.
- -
Today, quality criteria fire, acoustics, vibration are easily fulfilled through new shape
applications, i.e. massive nail laminated floors and wood-concrete composite systems for wide-
span and load supporting structures. Nail-laminated decks and wood-concrete decks including a
load-bearing concrete slab present new advantages, especially for houses, schools and public
buildings. Thanks to these techniques, the steadiness and bending properties of structures with
minor dead loads can be economically fulfilled. Fire resistance times of 30, 60 or 90 minutes, as
well as phonic insulation criteria up to 60 dB for walls and decks can be reached. The use of
timber as construction material is the only way to save the world's forests. Timber use is directly
linked to forest conservation and the planting of new trees.
The material selection is no proof for "good architecture". It is, however, an important contribution
to the environmental conservation, even if it needs more concentration on the planing phase.
176 CONCEPTS AND DETAILS OF MIXED TIMBER-CONCRETE STRUCTURES
anything else. High acoustical demands may be satisfied with the manufacture of special profiles,
as shown in Figure 3 for ceiling elements. Various solutions which do not requise the use of
special tools are available to manufacture noise absorbers. For instance, a few millimeters shift of
one out of two planks may be enough for the purpose aimed at. Architectural needs may also be a
factor which be considered when proposing solutions.
Figure 4 shows threesolutions out of many others which are available for floors made out of
vertical nailed planks; acoustic insulation may be required, so as impact noise reduction or higher
thermal inertia for the system. There are plenty of solutions and we must keep in mind that they
may be directly used with this new support, provided that they are set up with respect to the wood
material.
J. NATTERER 177
Fig. 4 - Examples of vertical nailed planks floor with different cover layers depending on specific
requirements.
In order to use the composite structure with high performances, it is important that concrete and
timber are linked together with a connection as rigid as possible. In this case, the link between the
two components is done with a system of grooves in the wood and post-tensioned dowels (figure
5).
The tests have shown that the link between the two materials does not depend on the rigidity of the
link, but only on the position and the repartition of the grows. With this consideration, the efficacy
of the liaison can be taken from 85 to 90 % of the composite effect.
Calculations for designing simply supported beams (and unidirectional slabs) are based on a
simplified method taking into account the mechanical properties of both materials and the elastic
behavior of the composite structure. Hypotheses have been made about the load distribution
(uniformly distributed live load) and the real behavior of the groove-dowel detail.
The behavior of the groove-dowel detail is based on the push-rod model used in reinforced
concrete beams. The shear forces are transmitted from the concrete to the wood
178 CONCEPTS AND DETAILS OF MIXED TIMBER-CONCRETE STRUCTURES
Fig. 5 - Detail: groove and post-tensioned Figure 6 - Forces acting on the wood inside a
HILTI-dowels groove in a timber-concrete composite
structure.
by compression on the surface of the grooves. Dowels work in traction to take the vertical part of
the inclined compression force on the grooves (figure 6).
Moreover, dowels are post-tensioned after the concrete curing period. The advantage of this post-
tensioning is that the gap - caused by the concrete shrinkage - is drastically reduced.
One can introduce some rebars inside the grooves, perpendicularly to the planks direction. This
solution will improve the transverse distribution of the load on the slab. Apart from this optional
reinforcement, a low diameter reinforcement mesh is advised in concrete to reduce and redistribute
curing cracks over the surface of the concrete.
Because of the parallel system created by the nailed planks (see figures 1 and 2), low quality
timber can be used with reevaluated properties. As a matter of fact, one single low quality plank
has a social behavior once it is connected with other planks and it is no more necessary to design
such elements with the quality of the lowest plank. Hence, designing becomes also more efficient
and less expensive in materials.
Figure 7 shows the groove and dowel distribution for a simply supported beam. They are
concentrated near the supports, under assumption of a distributed load, in order to take internal
shear forces with efficiency. Several constructions have already been erected using this new
connection detail. Vertical nailed planks are often used for the tension part of the
J. NATTERER 179
Fig. 7 - Groove positioning according to the Fig. 8 - Wood elements and dowels for a
shear force diagram. timber-concrete composite slab: School in
Triesenberg (FL).
composite structure in housing (figure 8), but we can also use round posts or glue laminated
elements for longer span or higher load solicitations, like for bridges or industrial buildings.
Today's experience shows that this kind of system is still at a developing stage, but it is enough
advanced to be applied in different ways. Composite timber-concrete systems show a great
flexibility in their application. This technique has both the usual advantage of timber structure and
the advantage of concrete structure. Figure 9 shows the behavior of three different floors submitted
to the same live load, with regards to the dead load and the acoustical insulation.
WOOD DECK
Load [kg/m 2]
1000 "
900 "
800 -
WOOD/CONCRETE DECK 700 "
600 -
500 "
400 "
300 -
A Compression zone
:
4. Conclusion
Wood planks are usually considered as by-products from the sawmill factories. Their uses are not
regarded with interest by engineers who would rather work with more industrialized wood based
products. This paper shows an interesting alternative to use a sawmill product directly as a
building material. It is worth noticing that this technique can be transfered to small and middle-size
industries without too much financial investment.
Vertical nail planks have been already used as massive construction elements for different parts of
several buildings, including multi-story buildings; walls, roofs and floors, with or without a
composite action with concrete. All the advantages of massive timber constructions are conserved
such as fire resistance. Moreover, concrete with its self-weight improves thermal inertia,
acoustical insulation and gives a greater stiffness to the system. It also protects wooden parts from
water damages caused by either plumbing problems or fire extinguishing.
The timber-concrete composite structure shown in this paper is drastically improving the stiffness
of floors and is in accordance with the latest requirements for modem conveniences housing. As
proposed in this paper, this system can be used in a broad variety of constructions, including
bridges and factory buildings with the same philosophy of using timber materials at its best, with
other complementary structural components.
5. References
Hoeft, M. 1994. Zur Berechnung von Veibundträgern mit beliebig gefugtem Querschnitt, Thèse
No 1213, Ecole Polytechnique Fédéral de Lausanne.
Götz, K.-H.; Hoor, D.; Möhler, K.; Natterer, J. 1978. Holzbau Atlas, Institut für internationale
Architektur-Dokumentation GmbH, München.
Götz, K.-H.; Hoor, D.; Möhler, K.; Natterer, J. 1989. Timber Design and Construction
Sourcebook, McGraw-Hill, Inc.
Natterer, J.; Herzog, T.; Volz, M. 1991. Holzbau Atlas Zwei, Institut fur internattionale
Architektur-Dokumentation GmbH, München.
Natterer, J.; Pflug, D.; Hamm, J. 1996 (à paraître). Méthode simplifiée de dimensionnement des
dalles mixtes bois-béton, Ecole Polytechnique Fédéral de Lausanne.
181
Summary
When using carbon fibre reinforced laminates for concrete strengthening, the anchoring stresses
in the end zones cause special problems A special shear test specimen is used to study the shear
stress distribution and the fracture behaviour in displacement controlled tensile tests The
fracture energy and the bond strength are calculated using fracture mechanics This paper
presents the results of these experiments, and compares the bond strengths predicted by
nonlinear fracture mechanics and by the actually applied phenomenological design methods
The evolution in the technology of new materials makes it possible to replace the classical steel
plates for concrete strengthening by new high-grade materials This has led to the idea of
replacing the steel plates by fibre reinforced composite sheets made of unidirectional, continuous
fibres such as glass, carbon and polymers, bonded together with a matrix such as epoxy resin
[1,2] Carbon fibre reinforced epoxy laminates are the most appropriate for strengthening
concrete beams. In the early stage these laminates had to be autoclaved, which was difficult to
execute in practice. Since the availability of so-called prepreg laminates, these difficulties have
disappeared. Prepreg epoxy laminates are preimpregnated with an epoxy resin, which holds the
fibres together. The prepreg sheets are very flexible and can be cut easily by means of scissors
At application, the prepreg sheets are impregnated again with the right mixing ratio of epoxy
resin components and the chemical reaction is started When the first layer has hardened
enough, the second layer can be applied in the same manner Several layers, up to 10, can be
applied. These prepreg sheets are first developed and produced in Japan Recently, new UD
laminates, which are not preimpregnated, are available In Belgium the first application of
CFRP-laminates took place at the beginning of 1996 [3,4],
182 ANCHORING STRESSES BETWEEN CONCRETE AND CFR LAMINATES
The CFRP laminates have a lot of advantages to classical steel plates. The mechanical
properties are superior (table 1).
3. Anchorage of CFRP-laminates
Because of the higher stresses in the CFRP-laminates, the stresses and stress concentrations in
the anchorage zone will increase too Without any precaution the CFRP-laminate might peel
off Extrapolation of the experimental results on the anchorage of steel plates is not allowed
Research concerning the anchoring phenomena has been started at the Reyntjens Laboratory
This investigation deals with the stress distribution and force transfer at the ends of the laminates
using non-linear fracture mechanics The aim is to obtain design rules, based on theoretical and
experimental results, to make a safe and economical design possible.
Preliminary shear experiments have been done at the Reyntjens Laboratory [5], Two concrete
K. BROSENS, D. VAN GEMERT 183
-^_P lOkN
P 20kN
P 30kN
a> P 37,4kN
f/i
S P 43kN
<D
M P 80 kN
The load, the deformation of the gap between the two prisms and the strains of the CFRP
laminates are recorded during testing The deformations of the gap are registered with two
LVDT-transducers placed at two opposite edges The strains are measured with strain gauges
glued on the CFRP-sheets
Two series of test specimens, A and B, were manufactured The compressive strength were
respectively 47 9 N/mm2 and 46 0 N/mm2
The results show that at the end of the CFRP-sheet a shear stress peak occurs at low forces
When a maximum shear stress is reached the concrete starts to crack At increasing load, the
shear stress peak occur further away from the end zone and the maximum shear stress attains
higher values (figure 2) This behaviour seems to be the same as for externally bonded steel
plates The fracture load was compared with the fracture load calculated with the method of
Van Gemert [6,7], and with the theory proposed by Taljsten [8] (table 2)
184 ANCHORING STRESSES BETWEEN CONCRETE AND CFR LAMINATES
The theory used by Taljsten is based on a non-linear fracture energy concept Steel plates to
concrete connections were tested in pure shear, e mode II failure Both symmetrical and
1
nonsymmetrical overlap-joints were considered When a brittle adhesive (G> 1 0 Gpa), such as
most epoxy adhesives, is used, the NLFM theory leads to the following expression
^CFRP *CFRP
P bK
mix \ 1 +a
F CFRP tCFRP
^ _
E concrtt* t concrata
The difficulties of this method are the exact definition of "the fracture energy" and how to
calculate this fracture energy from the measured values of the load, deformation and shear
stresses This method can not be used as a design rule For that purpose a relation between the
concrete properties, which can be easily determined, and the fracture energy is needed
Research should be done in that area, especially concerning the fracture energy in glued
connections between concrete and other materials, like steel or CFRP-sheets
The method of Van Gemert is a design rule The only parameter needed is the tensile strength at
the concrete surface, which can easily be measured by means of the pull-off test
The fracture load found in this way gives the load at cracking of the concrete in the initial force
K. BROSENS, D. VAN GEMERT 185
transfer zone It does not take into account the reserve available after first cracking This
explains the differences with the experiments A triangular shear stress distribution is assumed
on the basis of a large number of experiments (figure 3) With this assumption an anchorage
length is calculated A large safety factor is used to apply this method for design cases
p mix _
b If ^
The bonded length of the CFRP-sheets on the test prisms were varied too The fracture load
increases when the bonded length increases (figure 4)
95
90 4~
* 85
I 80
e 75
§ 70
£ 65 +-
60
125 175 225 275
However the influence of the bonded length will decrease at larger lengths Before cracking, the
shear stresses reach a maximum value at the end of the CFRP-sheet (figure 5) When the shear
stress exceeds the pull-off strength of the concrete surface, the concrete starts to crack [6]
At higher loads, the maximum shear stress shifts to the right (figure 5) and attains higher values
When the direct tensile strength, which is larger than the pull-off strength, is reached, the crack
will extend and the sheet will peel off in a brittle way
186 ANCHORING STRESSES BETWEEN CONCRETE AND CFR LAMINATES
1
CRACK 1
V1" CFRP
/CFRP
p
[Hîw
p —- CONCRETE p
- CONCRETE
Once the tensile force from the CFRP-sheet is transferee! into the concrete, there are nearly no
shear stresses between the remaining sheet and the concrete. This means that when the bonded
length exceeds a critical length, the fracture load remains constant. The determination of this
critical length can be done experimentally.
In figure 4, the fracture load has not yet attained a constant value, which means that the critical
bonded length is at least larger than 275 mm.
4. Conclusions
The use of carbon fibre reinforced materials offers great opportunities in concrete strengthening.
A lot of research has to be done to really understand the behaviour in the concrete-CFRP
connection. The end zones are the most critical points. Because of the higher stresses in the
CFRP laminates, the risk of peeling off increases. Experiments show that the behaviour of
externally bonded CFRP-laminates is similar to externally bonded steel plates. A non linear
fracture mechanics based design is possible, if the fracture energy Gf of the bonded connection is
known. Actually Gf must be determined experimentally, but further research should allow to
calculate Gf from the characteristics of the connection and its constituents.
5. Bibliography
Summary
The performance of composite steel-concrete bridges during their full service life is essentially
conditioned by the durability of the concrete slab. This article presents a brief survey of the
causes of the degradation processes, the influence of crack opening and the various actions
leading to cracking of the slab. In case there is some doubt about the water - tightness of the
membrane a satisfactory performance during a sufficiently long period can only be ensured by
prestressing of the slab, but several parameters have to be assessed carefully and some questions
still remain open.
1. Introduction
Composite steel-concrete constructions are presently widely used, particularly in the field of
continuous bridges where they appear quite competitive. Solutions have been brought to many
structural problems, but less attention has been paid to the performance of these bridges during
their füll service life1, conditioned by the durability of the concrete slab.
The paper presents first a brief survey of the causes of deterioration of the concrete deck. The
influence of the crack opening on the corrosion of reinforcing steels is then analysed. Crack
openings ranging from 0 to 0.3 or 0.4 mm are now considered as acceptable for a satisfactory
performance, but during a period of time which is presently unknown2.
The first essential protection consists in using a waterproof membrane of good quality placed on a
concrete with very low permeability.
If there is some doubt about the quality of the membrane, a satisfactory performance during a
sufficiently long period of time can only be ensured by prestressing of the slab. In order to
determine the amount of prestressing to be introduced, the various actions leading to cracking are
examined and the results of a practical example presented. Several quantities such as the tensile
stresses and the loss of prestress during the service life have to be assessed carefully. The
question whether crack openings under some variable loads can be accepted has not received
sufficient attention.
188 CRACKING AND DURABILITY OF CONCRETE SLABS OF COMPOSITE BRIDGES
The service life of constructions, and more particularly here of bridges, must also be defined. It is
presently agreed that this period of time is situated between 50 and 100 years. A duration of 50
years appears to be non economical, while a duration of 100 years seems too long due to the
evolution of the traffic needs. Therefore 80 years is presently considered as an optimum service
life for most bridges.
In order to determine the durability of a concrete element, and more particularly of the concrete
deck of a composite bridge, it is necessary to define the possible deterioration processes and their
governing factor. The slab contains concrete and reinforcing steel and both materials must be
durable, as the deterioration of one of them leads to the deterioration of the other and to
insufficient serviceability or resistance.
In nearly all chemical and physical processes influencing the durability of concrete structures, two
dominant factors are involved : transport within the pores and cracks, and water. More
particularly, in the case of the concrete deck of a bridge the following factors have to be
considered : freezing and thawing cycles, effects of deicing salts, penetration of chemically
aggressive agents, alkali-silica reaction
The mechanisms of corrosion must also be examined carefully. This is the most critical
degradation process, as it can lead to collapse of the element. Corrosion is mainly due to
carbonation of concrete in relation with penetration of CO2, and to penetration of chloride ions
originating from deicing salts.
Considering all the degradation processes of a concrete deck, it appears that appropriate
durability cannot be ensured easily, as the concrete slab is submitted to severe environmental
conditions. The major factors in connection with durability are the water-tightness of the
membrane and the quality of concrete. In relation with this characteristic the following parameters
must be examined carefully : W/C ratio, type of aggregate, cement type, cement content,
admixtures, handling and placing, curing.
3. Concrete cracking
Steel in concrete is protected against corrosion by passivation due to the alkalinity of concrete. In
such an environment the corrosion rate is insignificantly low. The passivity of steel may be
destroyed by the carbonation of concrete surrounding the reinforcement and by the penetration of
chlorides through the pores. For passive or active reinforcement situated near the top or the
centre of the concrete slab of a composite bridge the penetration of the chloride ions is the
prevailing action.
On the basis of theoretical and experimental studies a model4 has been proposed which allows one
to better understand the corrosion process in a reinforced concrete element and therefore the
influence of the propagation of the chloride ions. Two stages can be distinguished (Fig. 1) :
M. ERNENS, J.-M. CREMER, J.-C. DOTREPPE, A. LOTHAIRE 189
- the initiation period, during which the metal, having been embedded in concrete remains passive
whilst, within the concrete, environmental changes are taking place that ultimately lead to
depassivation ;
- the corrosion period, which begins at the moment of depassivation and involves the propagation
of corrosion at a significant rate until a final state is reached, when the structure is no longer
considered acceptable regarding structural integrity, serviceability or appearance.
The corrosion process may result in a reduction of cross-section of the reinforcement : the load-
bearing capacity of the steel decreases, but ductility and fatigue strength are reduced more
substantially. Splitting of the concrete cover may also occur. Rust has a substantially higher
volume than steel, which causes cracking and spalling. This may lead to sudden failure, if
longitudinal cracking along the bar occurs in the region of the bar anchorages. These unacceptable
damages usually correspond to the service life of the element (time ti in Fig. 1).
There has been a considerable evolution regarding the problem of the influence of the crack width
on the durability of concrete. Twenty years before it was admitted that this parameter had a
significant effect on the corrosion process. The observation of existing constructions and
laboratory tests have shown that there is no direct relation between the crack widths and the
degree of corrosion provided they remain smaller than 0.4 mm.
However the existence of cracks, even with a small width (0.1 mm), does influence significantly
the corrosion process. It has been shown that the diffusion of chloride ions is ten times more rapid
in a cracked than in an uncracked concrete3. This means that the initiation period will be
approximately ten times longer in an uncracked material compared with a cracked one, provided
that in both cases the permeability of the material is low and concrete cover is sufficient.
Therefore in an uncracked concrete, to will be large enough to prevent the steel from reaching the
propagation stage during the service life. If concrete is cracked, t, is small with respect to the
service life, and ti becomes the critical parameter with much more uncertainty regarding structural
safety.
190 CRACKING AND DURABILITY OF CONCRETE SLABS OF COMPOSITE BRIDGES
Cracks can be classified in various ways. We shall consider here the time of appearance. Cracks
due to chemical effects such as corrosion and alkali reaction will not be considered ; only cracks
due to thermal, physical and structural effects are envisaged.
Early cracks (before hardening) are due to plastic settlement and plastic shrinkage. These two
phenomenons may induce important cracking but preventive measures can be adopted in order to
avoid them.
After hardening there exists at least five causes of cracking under service conditions : external
loads, creep, drying shrinkage, external temperature variation (daily and seasonal) and thermal
shrinkage (appearing very early after concreting of the slab).
ABC
I 34m |
__
34m J
A Ö O
25cm
'J,vIyy-li^y. -VI
Cross-section of the beams
variable along the span
2.3m 8m 2.3m
12.6 m
The classical solution consists in designing the slab with passive reinforcement. In this case cracks
will open and it is impossible to limit their width to values <0.1 mm. Usually suitable detailing
regarding the diameter of the bars and their placing in the slab is adopted in order to limit the
crack width to 0.15 or 0.2 mm. As it is now acknowledged that crack widths up to 0.4 mm can
develop in the slab, would this mean that such a careful detailing for the reinforcement is no
longer necessary We think it would not be wise to do so. Nowadays many bridges are subjected
to very heavy traffic loads. Though very few studies have been devoted to this problem, crack
widths may tend to increase, due to progressive deterioration of bond at the boundaries of the
crack.
As already mentioned other detailing characteristics such as waterproof membranes and joints are
essential, but they are not discussed here.
Despite all precautions taken deicing salts will cause chloride ions to penetrate in concrete. As the
slab is cracked the rate of penetration is very high and depassivation will occur quickly. According
to the model of Fig.l the durability is controlled by the corrosion development. In these
circumstances, as soon as water - tightness is no longer ensured, it does not seem possible to
expect a service life of 80 years. Values situated between 20 and 40 years are sometimes
mentioned, but this has to be confirmed by additional research studies.
The only way of improving noticeably the durability of the slab is then to apply prestressing. The
beneficial effect of transversal prestressing is well-known 1-8, but in this article we will consider
more particularly the problem of longitudinal prestressing.
Several factors have to be examined carefully. The first one is the economy of the project as,
regarding the structural behaviour no prestressing. is necessary ; it is introduced only for
durability purposes.
The second one is the difficulty of calculating the stresses induced by prestressing in the slab. The
efficiency of prestressing is reduced by the composite interaction and by the classical time-
dependent losses due to creep, shrinkage and relaxation.
The type of prestressing must also be examined carefully. Several procedures are used8 :
prestressing by jacking supports, prestressing the slab and steel section by a longitudinal cable
situated in the slab, prestressing the slab only before composite action, use of external cables.
These various methods will not be discussed here. Each of them has its advantages and
drawbacks.
We shall focuse here a little bit more on the problem of the amount of prestress to be introduced
in the slab in order to ensure sufficient durability for a service life of approximately 80 years. This
question is a difficult one and so far it has not received enough attention.
The minimum value should be such that under actions existing at any time, i.e. permanent loads
including creep, drying shrinkage and thermal shrinkage no crack would occur. In this situation
cracks will open under variable actions such as traffic loads and variation of external temperature.
In our opinion this type of design can be unsafe, as cracks may be present during rather long
periods. In this case heavy traffic loads can cause fatigue effects leading to a progressive increase
of the crack width. After a certain time some cracks may remain permanently open.
192 CRACKING AND DURABILITY OF CONCRETE SLABS OF COMPOSITE BRIDGES
Design with full prestressing (no crack at any time) seems more appropriate, though it may be
difficult to introduce such a high level of prestress for technical and economical reasons. For the
example described in Fig.2, it has been calculated that a compressive stress of approximately
5.5 N/mm2 should be introduced in the slab in order to fulfil this condition.
5. Conclusions
1. In order to obtain adequate durability of the concrete slab of a composite bridge, the first
requirement is to place a waterproof membrane of good quality, to use a concrete with good
mechanical characteristics and very low permeability, and to provide sufficient concrete cover.
2. The use of passive reinforcement is the classical solution. Despite the recent studies on the
influence of crack widths, adequate detailing regarding the diameter and the placing of the bars
should still be recommended, as the crack width may tend to increase due to heavy traffic loads
inducing fatigue effects.
According to the model described here for the corrosion process, depassivation will arrive
rather quickly and durability will be controlled by the corrosion development. This leads to
uncertainty regarding service life. Additional research studies and observations on existing
bridges should be performed.
3. Prestressing can be used to improve the durability of the slab. The amount of longitudinal
prestress to be introduced has been discussed in this paper. Full prestressing is not economical
but on the other hand, to allow crack development for all variable loads may be unsafe due to
fatigue effects. Again additional research on this matter should be performed.
References
1. JOHNSON R.P. and BUCKBY R.J. Composite structures of steel and concrete - Volume 2 :
Bridges. Second Edition, Collins Professional and Technical Books, London, 1986.
4. SCHIESSL P. Corrosion of steel in concrete. Rilem Report, Chapman and Hall, London, 1988.
5. BARON J. et OLLIVIER J-P. La durabilité des bétons. Presses de 1ENPC, Paris, 1992.
6. FRANSSEN J-M. The computer code SAFIR. Service des Ponts et Charpentes, Université de
Liège, 1995.
7. COPE R.J. Concrete Bridge Engineering : Performance and Advances. Elsevier Applied
Science, London, 1987.
8. LEBET J-P. Composite bridges. Proceedings of an IABSE Short Course on Composite Steel-
Concrete, Brussels, 1990, pp. 147-164.
193
Jean-Marc Ducret, born 1970, Jean-Paul Lebet, born 1950, obtained his
obtained his civil engineering civil engineering degree from the Swiss
degree from the Swiss Federal Federal Institute of Technology, where he
InsUtute of Technology in 1993 also obtained his Ph.D. in 1987. Working at
and is presently completing a ICOM, he has spent 20 years in the field of
doctorate at ICOM. research and testing on composite structures.
Summary
A detailed study of the causes of transverse cracking in concrete slabs of composite bridges has
been carried out in order to understand better the most important parameters which reduce the
effective tensile strength of deck slabs. Site measurements and laboratory tests have enabled the
behaviour of concrete slabs to be followed from the moment they were placed. The results of
measurements have demonstrated the important influence of concrete hydration on the tensile
stress in the slab. A criterion for evaluating the risk of cracking in young concrete has been
established on the basis of subsequent numerical simulations.
1. Introduction
Over the past 15 years, steel-concrete composite bridge construction in Switzerland and its
neighbouring countries has evolved with a view to reducing labour costs at the expense of
increasing the quantity of steel. For example, thicker steel webs are used in order to reduce the
number of stiffeners to a minimum. Deck slabs are now placed without special provisions using
travelling formwork and allowing composite action to be initiated at the time of concrete setting.
In such bridges, transverse cracks often develop shortly after casting the deck slab, in particular
at or near to internal supports. These cracks are normally between 0.1 and 0.2 mm wide, and are
therefore not easily visible unless water has passed through them before the application of a
waterproof membrane, in which case deposits can be visible on the deck soffit. Raising then
lowering supports in order to introduce compression in the slab over supports after casting is
used in some cases and longitudinal post-tensioning is often considered to be too expensive. The
alternative is to increase the quantity of reinforcement over internal supports.
The importance of these transverse cracks is open to discussion. The structural safety is in no
way compromised; even much wider cracks would not lead to significant damage to the main
beams. It is interesting to note that in European countries, cracks of the order of 0.2 mm wide are
allowed even in humid environments with the presence of de-icing salts. This is possible due to
the use of well detailed waterproofing systems which are carefully installed and subsequently
ensure the durability of a structure. However, there is a need to understand better the causes of
these transversal cracks in order to be able to develop methods to reduce them. It is therefore
concluded that there is a need to study the behaviour of concrete deck slabs from the moment the
concrete is placed [1]. Similar research carried out in France also points to this stage of
construction as being of interest [2].
AT +25° Ec 8 kN/mm2
Section 3
r
Section 1 Section 2
T 19.6
19.6 32.4
-34.2
Figure 1 : Stresses in a composite section during concrete hydration : (a) heating period,
(b) cooling period
The stresses measured in steel beams are different before and after concrete hydration,
indicating the presence of resultant stresses in both the beams and the concrete slab (Figures 4
a and b).
The use of optical fibre sensors (OF) and vibrating wire strain gauges has enabled the
expansion and contraction of the concrete to be measured qualitatively and quantitatively
(Figures 5 a and b).
2 90 m 10.00 m
Bridge 1, ß 0 12 Bridge 2, ß 0 12
Bottom flange
Slab
Numerical
simulation
5 20 0
Q.
S 10 0
96
Time [ Hours J
Beam 2 b)
-Results
Numerical 15
simulation Numerical
— 10 simulation
c-i
I 5
I 0
D "5
-10
Time [ Hours ] Time [ Hours
96 192 0 96
Stresses in steel beam bottom flanges : a) Bridge 1, b) Bridge 2
OF 1
OF 2
OF 3
0 96 192 0 96
Figure 5 : Strains in the deck slab : a) Bridge 1, b) Bridge 2
196 EFFECTS OF CONCRETE HYDRATION ON COMPOSITE BRIDGES
The ß values for the two bridges shown in Figure 2 are relatively high and in both cases cracks
between 0.1 and 0.15 mm wide were visible within ten days after concrete placement. This
indicates that the residual tensile stresses due to concrete hydration were very close to the tensile
strength of the concrete.
140 mm
160 \ 10 160 x 10 160 x 10
_ 160 \ 10
__ 200 x 15
_ 300x 20
96 192
Time [ Hours ]
Figure 7 : Laboratory test results : a) Temperature evolution b) Strains in the deck slab.
J.-M. DUCRET, J.-P. LEBET 197
Figure 7 (a) shows the measured temperature evolution, which was similar to that observed on
site. The evolution of strains shown in Figure 7 (b) highlights the importance of the ratio ß with
respect to the effects of hydration. The strain measurements demonstrate that the restraint
provided by the steel beam increases with its rigidity represented by the ratio ß.
Service load tests at 28 days allowed the residual tensile strength of the concrete slab to be
estimated. The three tests have shown that the residual tensile stress in the concrete slab after
hydration is :
- 0.5-0.8 N/mm2 for a section with a ß value of 0.05,
- 1.0-1.4 N/mm2 for a section with a ß value of 0.08,
- 1.4-1.8 N/mm2 for a section with a ß value of 0.11.
A parametric study based on the approach described in Section 3.2 has illustrated the importance
of restraint provided by the steel beams with respect to the resultant tensile stress in the concrete
deck slab. Using INTRON, numerical simulations of bridges having different cross-sectional
dimensions have enabled the influence of ß on the evolution of stresses in the deck slab during
hydration to be quantified, as illustrated in Figure 8.
c
{ß-Es+Ec2)-{ß-Es + Ecl)
ac residual tensile stress in the concrete,
a coefficient of thermal expansion of the concrete,
ß restraint coefficient defined as the cross-sectional area of the steel beams divided by the
cross-sectional area of the concrete slab,
AT maximum difference between ambient and concrete temperature during hydration,
Es elastic modulus of steel,
Eci mean elastic modulus of concrete during the heating period,
Ec2 mean elastic modulus of concrete during the cooling period.
The following default values for parameters may be used in the absence of other information :
- Eci 6 kN/mm2,
198 EFFECTS OF CONCRETE HYDRATION ON COMPOSITE BRIDGES
- Ec2 25 kN/mm2,
- a= 1 • 10"5 K"1,
- AT 25 °C.
Based on the results presented above, a qualitative evaluation of the influence of ß on the effects
of concrete hydration has led to the following observations :
- ß <, 0.05 limited influence of hydration effects on early cracking,
- 0.05 < ß < 0.08 hydration effects reduce the tensile strength fctm, limited risk of early
cracking,
- 0.08 < ß< 0.12 hydration effects reduce the tensile strength, early cracking is
probable, actions for reducing residual tensile stresses should be
considered,
- ß > 0.12 hydration effects significantly reduce the tensile strength, high risk of early
cracking, actions for reducing residual tensile stresses should be adopted.
The residual tensile stress ac can be calculated using the Equation (2). The effective tensile
strength of concrete fct,eff should then be adopted in subsequent calculations, in particular when
considering the stiffness of the composite section, but excepting the determination of minimum
reinforcement for limiting concrete cracking. The value of fct,efr is given by :
The measures mentioned above for limiting the effects of hydration are aimed at reducing the
difference between the temperature of the concrete slab and that of the steel beams. This could be
achieved for example by using a low-heat cement or by cooling the concrete before or during
curing.
5. Conclusions
The effect of concrete hydration in a deck slab which is directly linked to steel beams in a
composite bridge has been studied with the aid of site and laboratory tests. A numerical model
has been validated using the results of theses tests and has subsequently been used to
demonstrate the importance of the restraint coefficient ß with respect to the residual tensile stress
due to concrete hydration in the deck slab.
Criteria based on the restraint coefficient have been established which define the effects of
concrete hydration as a function of the bridge. Furthermore, a simplified method has been
developed for evaluating the residual tensile stress due to concrete hydration in the deck slabs.
6. References
[1] J.-P. Lebet, Comportement des ponts mixtes acier-béton avec interaction partielle de la
connexion et fissuration du béton. Ecole Polytechnique Fédérale de Lausanne, ICOM-
Construction métallique, Thèse EPFL N° 661,1987.
[2] SETRA, Recommandations pour maîtriser la fissuration des dalles. Service d'Etudes
Techniques des Routes et Autoroutes, 1995.
[3] RILEM TC -119 TCE, Thermal Cracking in Concrete at Early Ages. R. Springenschmid,
Munich, 1994.
[4] K. Van Breughel, Numerical Simulation of Hydration and Microstructural Development in
Hardening Cement-Based Materials. HERON, Volume 37, 1992.
[5] M. Emborg, Thermal Stresses in Concrete at Early Ages. Lulea University of Technology,
Division of Structural Engineering, 1989.
[6] J.-M. Ducret, J.-P. Lebet, Modelling and testing the behaviour of a composite bridge.
Structural Assessment, City University, London, July 1-3,1996.
199
Born 1937. Throughout his Born 1959, civil engineenng Born 1957, civil engineenng
earner he has worked on degree 1983 He joined SETRA degree 1979 Expenence in steel
engineering structures For in 1984 Expenence in concrete and composite bndges Has been
about 25 years, specialised and composite bndges, and m involved in the design of
in pathology pathology innovative composite bndges
Summary
Experience has shown that just applying the French regulations was not sufficient to control
transverse cracking in concrete deck slabs of composite bridges. Unacceptably wide cracks were
observed. In 1995 a working group published Recommendations [1] to control this cracking and
improve durability. The Nevers bridge was built between 1992 and 1995. The provisional
version of the Recommendations [1] was taken into account during construction and its cracking
was successfully controlled.
Composite structures have become frequent m France. They account for nearly 20% of the
surface of bridges currently built as opposed to less than 5% in 1980.
Transverse cracking in the concrete slabs of composite bridges is accepted by the French design
regulations. It requires a minimum reinforcement condition and a limit of the tensile stress in
passive steel reinforcements of the slab in the zone of hogging moment.
Experience has shown that just applying the regulations was not sufficient to control transverse
cracking in deck slabs. Excessively wide cracks were observed late in the 1980s in the zones of
hogging moment, and even cracks in the sagging moment zones where the concrete is
theoretically in compression. These unacceptably wide cracks are likely to affect the
serviceability of these structures.
A working group, made up of the Administration and the contractors, analysed the causes of
cracks observed in composite structures and in 1995 published Recommendations [1] to control
this cracking and improve durability.
All the Recommendations will not be covered in detail here but it will be shown how they have
been taken into account in the Nevers bridge.
200 CRACKING CONTROL OF THE NEVERS COMPOSITE BRIDGE
The bridge is situated on the Nevers bypass m the centre of France It carries the National Road
7, the well-known route down to the south of France, over the River Loire east of the town
It is composed of two composite box girders, each one 420 metres long (fig 1 and fig 2)
I1
«n
'
60m
^
j
60 m
^
I.
60m ^
1
60m
-ig-
^ i
70 m ^
'
55 m ^
$ jj^
Fig 1 Longitudinal section
Half cross section on piers Half standard cross section
10,69
20 m standard segment
15 m standard segment
20 m segment on piers
The contracting company used exactly the same general concreting procedure as that proposed in
the contract. To comply with the tight schedule, it chose to concrete two segments per week (40
metres), using a 78 metric ton crane which travelled directly over the bridge webs for handling
purposes (photo 1).
To meet the deadlines, the contractor had to remove the formwork at 8 a.m. for concreting that
had been completed at about 6 p.m. the previous day. The last concrete casting had therefore
been hardening for 14 hours when the formwork was removed. But the minimum off-form
strength requirement to limit deformations had been fixed at 15 MPa.
-
In addition to the foregoing procedures pilgrim's steps concreting, lifting the support on one
abutment - other steps therefore had to be taken so as not to jeopardize the contractor's time
schedule, while ensuring that cracking in the slab was of reasonable intensity and with controlled
crack widths.
Photo 1
Crane on the bridge
202 CRACKING CONTROL OF THE NEVERS COMPOSITE BRIDGE
One solution consisted in selecting a high-strength concrete that gained strength very quickly.
But because of the thermal and endogenous shrinkage which is constrained by the bridge frame,
cracking was liable to occur in the slab. For this reason, the Recommendations [1] advised that
the strength value of the concrete at 28 days should not be too high, which therefore imposed a
strength value at 14 hours as close as possible to 15 MPa.
-
To meet all three conditions at once delayed setting, quick removal of formwork (14 hours) and
-
a minimum off-form concrete strength while limiting the intensity of the thermal and
endogenous shrinkage, it was finally decided to use a concrete containing a not too rapidly-
hardening cement, that would have a strength at 15 hours of 22 MPa under standardized
temperature conditions (20°C).
The selected concrete mix design and the thermal behaviour of a slab segment were modelled by
a finite element programme (L.C.P.C TEXO) to determine the heating conditions strictly
necessary to obtain an off-form strength of 15 MPa. Bearing in mind that the two slabs were to
be concreted in winter, these calculations enabled two important thresholds to be fixed :
- the minimum external temperature T1 requiring heating of the slab
- the minimum external temperature T2 requiring the use of a hot concrete.
The transverse distribution of stresses due to thermal and endogenous shrinkage was also studied
using a finite element programme (L.C.P.C. MEXO). The analysis showed that these phenomena
were liable to generate tensile stresses of around 1.5 MPa in some parts of the cross section.
As the limit conditions had been determined by calculation, the decision to remove the formwork
could not depend solely on the results of the informative samples. For this reason, in order to
make the lapse of time before removal of formwork as short as possible, the strengthening of the
concrete was monitored by a maturity meter. Based on a previous laboratory measurement
characterizing the change in the concrete strength under known conditions, this instrument is able
to predict the resistance of the concrete to compression at any time by continuously measuring
the actual temperatures in the concrete. This maturity meter enabled the formwork to be removed
at the most appropriate time and considerably helped to reconcile the various constraints.
Photo 2
Positioning the temperature probes
of the maturity meter
The slab was heated by forced-air oil heaters placed directly inside the girder under the newly-
concreted segment. The girder structure lends itself to such heating, the ends of the heated zone
simply have to be closed by a tarpaulin. The hot concrete was obtained by heating the mixing
water.
D. POINEAU, J.-M. LACOMBE, J. BERTHELLEMY 203
During construction : Full calculation was made of the concreting phases. This
particularly highlighted the fact that one zone is far more stressed than the rest of the structure
(-5.5 MPa compared with -3.5 MPa in the other spans). This zone is the second central segment
of span 3 at the time of concreting the third segment of span 2 (longest span: 70m, fig. 4).
concreting cracking
a
CO P1
I
P2
^
P3
I
P4
I
P5 P6 C7
These tensile stresses exceed the tensile strength of a green concrete, which can be estimated at
-2 MPa. A phenomenon often observed on composite structure sites is thus found by calculation.
When the concrete of a span is poured, cracking occurs at the end of the hardened concrete
in the previous span.
In service : The normal theoretical stresses in the concrete slab in service were calculated
with a steel/concrete coefficient of equivalence of 18 for permanent loads and 6 for live loads.
This calculation took into account a shortening effect (shrinkage + temperature) of 2.5 1CT4.
Heavy tensile stresses occur on the support and close to the segment end zones previously
mentioned (up to -6 MPa).
However, in order to achieve reasonable crack widths. S.E.T.R.A. applied the rules of the
provisional versions of the Recommendations [1] in compliance with Eurocodes 2 and 4.
Only two types of longitudinal reinforcements were used :
- segments on piers and the 2nd segment of span 3 are 1.35% reinforced (zones where the tensile
204 CRACKING CONTROL OF THE NEVERS COMPOSITE BRIDGE
stress in the slab exceeded either 4 MPa during construction or 5.5 MPa in service),
- the other zones are 1 % reinforced to take into account the effects of green concrete shrinkage.
3.4 Results
Cracking in the slab was recorded for both decks after the loading tests. The zones effectively
cracked were shown to correspond to those foreseen
by the calculations (these zones are the on-pier
segments and the last central segment of each span).
Photo 3
Transverse underlined cracking in slab
This jobsite was thus successful as regards cracking control but it should be possible to do better
by reducing the extent of cracked areas occurring during the construction stage.
The reinforcements used led to an increase in the steel ratio of around 15 kg/m3, which
corresponds to a 0.4% price increase in the contract.
4. Conclusion
The 1995 Recommendations [1] differentiate between two types of provisions those aimed at -
limiting cracking intensity and those aimed at limiting crack widths.
It was possible to take into account most of the provisions aimed at limiting cracking intensity
and virtually all the provisions aimed at limiting crack widths without jeopardizing the project's
cost effectiveness.
The extra cost, which was less than 1%, seems most reasonable bearing in mind the improvement
in durability to be expected as a result of the cracking control.
From this example, it will be seen that applying the Recommendations [1] does not penalize
composite structures to any significant extent.
If the contract were to be drawn up today, the designers would impose removal of formwork after
24 hours minimum, in accordance with the Recommendations [1], which would enable a
concrete with an even lower heat of hydration to be used. If it proved necessary to use a second
travelling formwork, its effect on project time schedules or cost would have to be analysed.
References
Takao Sugo, born 1955 received his civil Yosihiro Murata, born 1961, received
engineering degree at Nosiro Technical B.S at Akita Univ. He has been engaged
High School. He has been mainly in Design of PC Structures.
engaged in the construction of PC Bridges.
Summary
This Bridge Ginzan-Miyuki Bridge) subjected in this study is a new type bridge of PC box girder
bridge used corrugated steel girder at web and external cable, and has been constructed at 1996 as
the largest bridge of this type in Japan. And this bridge is also constructed by incremental
launching method. It is here investigated the Safety at construction stage by this construction
method and the Serviceability for dynamic behavior and characteristics of this bridge based on the
results of dynamic experiments.
1. Introduction
This Bridge is a mountain bridge built as part of road improvement works commonly known as
Matsunoki Road that were carried out by Akita Prefecture (from 1974 to 1996). This is also a five
span continuous girder bridge and bridge length is 210.0m. To satisfy construction conditions
constrained by harsh topography and weather, the bridge was distinctively planned, designed and
built. Namely, the bridge has a composite structure of PC and steel and was built by diagonally
suspended, incremental launching erection method. Herein we will report on the main details of the
bridge's design, construction and dynamic behavior and characteristics.
dictated that the basic erection method would be incremental launching method, it was considered
most effective to reduce the main girder's weight to enable less labor-intensive construction. As a
result of studying the potential alternatives, it was decided that the structure optimally suited to the
conditions and the scale of the bridge in question was a PC box girder used corrugated steel web
(Figure-2), a steel and PC composite structure lighter than a PC box girder and more economical
than a steel girder.
d steel web
t
9^2
9 ~ 12 nn
shape of corrugated steel
lower comcrete slab
T qnn
nge plate
Figure-2 Cross section of box girder and shape of corrugated steel web
W. ISIGURO, T. SUGO, K. UEHIRA, Y. MURATA 207
The main girder is composed of a total of 19 blocks whose standard length is 11.0 m. The main
girder casting yard occupied about the rear 60 m of the A2 abutment, on the rear 30 m of which all
weather sheds were set up. In terms of the formwork facilities for casting the main girder blocks,
the lower concrete slab formwork facilities, where the corrugated steel web was assembled, and
the upper concrete slab formwork facilities were arranged consecutively for the sake of
convenience. In laying out the facilities, care was also taken so that concrete could be casted
simultaneously so as to not disrupt the block casting cycle.
The corrugated steel web was fabricated by welding flange plates to the top and bottom of steel
plate that had been pressed into a prescribed corrugated shape at the factory, and then welding
stud dowels to the flange plates. It took two of these steel panels (5.5 m + 5.5 m) to make up the
standard block length of 11.0 m.
launching direction
The diagonal suspension cables were tensioned on the pylon side. Thereforer, the diagonal
members' tension is propagated as axial tension from the pylon to the main girder and has a major
effect on the stress of the main girder section directly below the pylon. Hence, it was adjusted the
cables' tension in accordance with each step in the erection process. And it was used PC bars in the
slabs of main girder as prestressing on incremental launching condition.
For the external cables arranged inside the box girder, two-span continuous cables were used in
consideration of factors such as workability and prestressing-loss due to angular change. The
external cables use the support cross-beams as their anchoring points and are laid out so they
overlap with a different cable on every span And for the cable's protective sheathing, high-density
polyethylene tubing (PE tubing) was used.
Measurements were mainly taken by effective-stress-meters placed in the concrete section directly
below the pylon where main girder's stress was critical, and tention-meters in the diagonal
suspension cables anchorage zones to investigate the safety of member during construction by
incremental launching method. And it is mainly shown the tention change of diagonal suspension
cables and the stress change of upper slab concrete directly below the pylon in figure-4 and
figure-5.
140
•jior-.ai'-cowr— ».-Mior«-»
Erection* Step o?Lounci5n| ïonîbtK&on
— cow
*
/ o \u
\view I
point
concrete gauge
Valle Mountain
side de
-DaMgrang Vdue
- M«Munng Vdut of Va»ey Sid«
- M« «unrig Valu» of Mouttom Sid«
— o «o
- o » - o
- --NNCiNNW 8 8 S w S 5 5 S 5
of Construction
Erection Stop I«une hang
Figure-5 Stress change of upper slab concrete directly below the pylon
Though the values obtained through measurement exhibit a slight dispersion, overall they
represent values close to the design values calculated by beam theory. The results in these figure
indicates that the actual girder's behavior vis-a-vis diagonal suspended-member tension is
consistent with design theory.
.external cable
ffc
Al
^ I« ;
PI
i
P2
«
«I
^-devioter P3
I I I
P4
; I I I
A2
<—
X
7
'trVuol nertéer upposed nenber
5. Conclusion
Through the construction method, the static and dynamic experiments summarized above, we
verified that it is confirmed the safety of member during construction and it is possible to conduct
eigenvalue analysis using methods for evaluating torsional and shear rigidity verified through static
analysis.
In closing, we hope that this report will be useful in furthering the development of corrugated-steel
web PC bridges.
211
Summary
In the paper, some of results obtained by using the Age-Adjusted Effective Modulus Method to
evaluate creep and shrinkage effects in the steel-concrete composite beams with rigid or deform-
able connections are presented. Aging coefficients for the creep, relaxation and shrinkage problems,
evaluated in absence of connection and rebars, are proposed. General considerations and
numerical comparisons performed in the presence of rebars in the slab, point out how the
proposed approach can properly substitute the more popular Effective Modulus Method.
1. Introduction
It is known as design criteria proposed by current codes for steel-concrete composite beams,
beside the ultimate limit states, require to satisfy the serviceability limit states [1,2]. Indeed,
deflection limits, mainly introduced to reduce concrete cracking and to preserve supported
elements together with serviceability stress limits based on durability criteria, govern more and more
the design of composite beams. A correct evaluation of creep and shrinkage effects on stress and
deflection long-term response is then very important In general, actual codes make reference to
simplified algebraic methods for the practical evaluation of this response; they are the "Effective
Modulus Method" (EM), the "Mean Stress Method" (MS) and, when an accurate evaluation of
viscous effects is required, the "Age-Adjusted Effective Modulus Method" (AAEM).
In the paper, working in the ambit of the AAEM method, as Age-Adjusted Effective Modulus of
the concrete component beam, is assumed the one computed in absence of shear connection and
rebars. With reference to the CEB 90 Model Code [3], simplified expressions of the Aging
coefficient are given. By some examples, the capability to obtain very accurate long-term solutions
for composite beams with rigid or deformable connections also in presence of rebars is
shown. With the proposed approach, besides to have very correct results without remarkable
complications, it is possible to follow a more physical approach respect to the EM method. For
these reasons we think that it can in general substitute adequately the popular EM method, which
can be adopted to analyse composite beams when long-term effects are not important.
212 LONG-TERM EFFECTS IN THE STEEL-CONCRETE COMPOSITE BEAMS
It is known as the AAEM method for homogeneous structures, in the hypothesis of linear visco-
elasticity, allows to achieve an exact solution for a linear combination of pure creep and relaxation
problems. In all other cases, adopting an aging coefficient % evaluated in these hypotheses, an
approximate solution can be reached. In particular for composite beams, the presence of a viscous
material (the concrete) and an elastic material (the steel), involves a migration of stresses from a
point of the structure to another one, with different variation laws respect to an homogeneous
structure. From a theoretical point of view, it is therefore not appropriate to apply the AAEM
method with the same % values adopted for homogeneous structures, since the response is
influenced strongly by the
presence of the steel beam. For this reason, Trost proposed two % coefficients
in the presence of rigid connection: the former one, labelled XN, related to the normal force in
the concrete component beam; the latter one, named %M, related to the bending moment in the same
beam [4], In the hypotheses of strong steel beam respect to the concrete slab and affinity of the
shrinkage law with the creep law, Trost found for creep or relaxation problems simple relations to
evaluate xN in a rigorous way and xM in approximate way. The complexity of the problem due to
the presence of two coefficients % as well as the limits due to Trost's hypotheses have induced the
authors to propose, for the composite beams with rigid or deformable connections the use of one
approximate % value only [5,6]. For three elementary problems of creep, relaxation and shrinkage,
these x values were evaluated in the hypothesis of no connection between concrete and steel beam
and no rebar in the slab, providing an exact value for creep and relaxation problems and a numerical
evaluation for the shrinkage problem. In particular, for the creep problem was shown as in absence
of connection and rebar, introducing the coefficient ß= (Es Js)/(Ec(to) Je + Es Js where EsJs and
Ec(to) Jc are the steel beam and concrete slab bending stiffnesses at initial time to, the stress
evolution in the slab for a creep problem in the time interval [to,t] is the same of a pure relaxation
problem of an homogeneous concrete structure with the fictitious coefficient <|) (t,to)=ß<t>(t,to),
instead of creep coefficient <(>(t,to). Following the Bazant approach [7], it was then possible to
compute the parameter X=XM for this composite beam. In the hypothesis of no connection, few
additional difficulties respect to an homogeneous structure are then introduced to compute the
exact solution (with the EM method, where x=l, or the MS method, where %=0.5, the response is
obviously approximate). Also the proposed method becomes approximate in the presence of an
elastic or rigid connection. In particular for rigid connections, % is not constant in the slab but it is
in any point a linear combination of %M and %N Trost's values.
With reference to the CEB 90 Model code and ACI92 creep model, by performing a numerical
analysis, we have shown as using the only %=%M value of the beam without connection we get in
general an accurate solution since xN value is less influent on the solution of %M, that is practically
independent of the connection stiffness [6]. As a consequence, by assuming in the study of a
generic composite beam X=XM=XN, it is natural to calibrate the % coefficient on the value
determined in absence of connection, where
%N—>xM, since it can be considered the most significant
value for the beam. To use the AAEM method with the same simplicity of EM and MS methods,
we provide simple relations in order to calculate the % coefficient In particular for the pure creep
problem, to evaluate the X=Xc(t00,to)=Xc„value, linked with the long-term solution, we propose an
extension of the approximate expression given by Lacidogna for a homogeneous structure [8]:
C. AMADIO, M. FRAGIACOMO 213
0-5
to
Xc(3-104, to) too.i
n+
where is the initial time load, n is a corrective coefficient calibrated on the fictitious thickness
to
ho=2Ac7u expressed in centimeters (Ac is the area of the concrete slab while u represents his
perimeter in contact with the atmosphere), the relative humidity R.H.(%), the characteristic
strength of concrete fck (MPa) and the coefficient ß. The coefficient n is calculated as summation
of Lacidogna's term nL and the corrective term nc. The values of nL and nc are defined as:
with
1/3
f (ho)
fl 9Rh
f (ho) -0.772 + 2.917 i f
-lO'V (fdc) 0.772 + 0.0114 fck, (3a,be)
b
e(io-3h0)
and
These expressions provide accurate results when 5 < ho< 160 cm, 50% < R.H. < 80% and
3 < to < 200 days and imply 5% maximum error and 1% medium error. In figure la, a comparison
between exact and proposed (dotted line) values is shown.
In the case of an imposed flexural distortion (very important relaxation problem to evaluate the
stress state in a statically indeterminate composite beam with constant mechanics characteristics
subjected to a settlement of the supports), in the hypothesis of no connection, the concrete beam
is subjected to an effect of pure relaxation which does not depend on the presence of the steel
beam. This allows to affirm that the % coefficient determined by means of the proposed approach
is equal to that of an homogeneous structure subjected to a constant strain and vice versa.
Analogous results is reached by working in the hypothesis of rigid connections because even in
this case the strain law adopted for the determination of % coefficient for homogeneous structures
is exactly respected with reference to concrete (ec is constant in time in accordance with the
second theorem of the linear viscoelasticity). The values, denoted here as %t„, can be then
easily found by setting ß=l in the eqn. (3d).
To evaluate slab shrinkage effects it should be observed that the laws of shrinkage evolution in
time proposed by actual codes are not affine with the creep laws in general (if shrinkage is affine
to creep, x, shrinkage values can be calculated as in the case of constant load, by assuming
X, ^Xc)- With reference to the CEB 90 model code where the shrinkage is not affine, analysing
numerically the long-term response we have verified that also in this case the validity of the
proposed method in the presence of shear connection is good. An approximate expression for the
coefficient is proposed here to evaluate the long-term effects. For this problem the
significant parameters are the shrinkage initial time to, the relative humidity R.H., the characteristic
strength of concrete fck and the fictitious thickness of the slab ho. The stiffness of the steel
beam and concrete slab does not influence practically the response. The values, determined
here with reference to a section in which xnhrç(M> i-e- f°r a small stiffness of the steel beam
respect to the concrete slab, with the same units seen above, take the form:
Figure lb shows a comparison between numerical (continuum line) and proposed (dotted line)
values. It is evident as the x»«, values can be very different from when the CEB 90 model is
adopted.
Fig. 1: a) X» values for the imposed load problem, b) X« values for the shrinkage problem.
For rigid connections, the classical assumption that the whole cross section remains plane can be
adopted and the long-term analysis can be performed at the cross section level. Following the
AAEM approach, the stress-strain law for the concrete can be posed in the form:
ec(t)-£n(t)= iS+Gc(to)(i^-i^)=lS+*>• (5)
where e(t) can be considered as an imposed strain linked to the viscosity effects in the interval [to,t],
while the quantities
Ec(to) Ec(to)
„ " l+<|>(t,t0)
Eceff
'
_
Ecadj
1+ x(t t0)(|)(t)t0)
as known, are the "Effective Modulus" and "Age-Adjusted Effective Modulus" respectively, Ec(t),
En(t) are the elastic and inelastic strain at time t and Oc(to), Oc(t) the elastic stress in the concrete
C. AMADIO, M. FRAGIACOMO 215
at times to and t. To evaluate the response at time to where the concrete has modulus Ec(to) and at
infinite time where it has modulus Ecadj evaluated with X=X~ f°r any elementary problem, it
appears natural to introduce the modular ratios no=Es/Ec(to) and n„=Es/Ecadj, where
Es is the steel
Young modulus, and to work with the transformed cross sections at times to, t„. In the presence
of rigid connections and constant cross section, for the elementary cases of constant external
sustained load, relaxation and shrinkage in the slab, with the above quantities, a long-term solution
characterized by the same difficulties of application of the EM method can be reached [5,6].
Then, by applying the principle of effect superposition, it is possible to solve a large number of
actual problems. We underline that the proposed approach, through the evaluation of the imposed
strains e(t), allows a correct interpretation of viscous problem, since it not consists in the only
To evaluate creep and shrinkage effects in a generic composite beam with deformable connections,
it is not possible to work only at the section level but the whole beam has to be considered.
In order to solve this problem, in general it is convenient to use a computer program. In this case,
a remarkable simplification respect to an algorithm that utilizes a step-by-step procedure can be
obtained. To solve for instance a pure creep problem, instead of 30+50 steps, by using the constitutive
equation (5) and the proposed coefficients, only two steps are necessary: one at time t„
and one at time t_. A very easy method to solve, by means of the proposed % values, simply
supported composite beams with deformable connections under sustained loads or shrinkage in the
Fig. 2 shows a comparison for the T]/Tie and T|/r|r ratios obtained with the proposed % method, the
exact solution and the EM, MS method varying the parameter aL= ^KÛ(EJ)r /[(£A)*(£7)aJ
216 LONG-TERM EFFECTS IN THE STEEL-CONCRETE COMPOSITE BEAMS
that characterises the composite beams model with elastic connections introduced in [11]. In these
diagrams, t|, T|e are the long-term and initial deflections, evaluated in the midspan section of a
simple supported composite beam with deformable connections and length L, while T|r is the
long-term deflection computed in the hypothesis of rigid connection.
The quantity (EA)' is equal to (EiAs E/l^KE/l^+E/l^, with E/ls and EcAc axial stiffnesses of
steel and transformed concrete component beams at time to, while (EJ)r and (EJ)a are the
bending stiffnesses of the tranformed cross section determined at time ^ in the hypothesis of
rigid connection or no connection respectively and K the connection stiffness. These ratios are
computed for a sustained uniform load problem (fig. 2a) and a shrinkage slab problem (fig. 2b)
adopting an IPE 300 steel beam and a 80x15 cm concrete slab with f<* 30 MPa, ho 30 cm and
a percentage of symmetric rebar AS=2%. We can see as the proposed solutions is very adequate
independently of the connection stiffness (when ocL>20 the connection can be considered rigid).
Analogous results have been obtained in terms of stresses. In particular, adopting the % proposed
values, computed in absence of connection and rebars, we have seen that the presence of a
symmetric or asymmetric rebar with a percentage AS=0+3% does not change the response
accuracy. Further comparisons with EM solutions have allowed to point out the advantage of the
proposed approach, in particular for the shrinkage problem where significant errors may occur by
applying the EM method. To conclude, we think that when the effects of viscous problems are
important, the proposed approach can be properly adopted by designer to analyse a large number
of practical problems.
REFERENCES
Summary
The time-dependent response of composite concrete slabs made of precast floor plate and cast in
situ concrete are analytically and experimentally studied. Analysis is carried out with an algebraic
age-adjusted effective modulus method and a relaxation procedure. The moisture transport
between concrete components of different ages has been taken into account. Long-term test
results on five statically determinate composite slabs are reported. The agreement between test
results and theory is shown to be good.
1. Introduction
Composite flooring systems are widespread in building construction. Their scope of use covers
cast in-place monolithic slabs, e.g. floor and roof systems for buildings, parking garages and
bridge decks. Composite concrete slab consists of precast concrete floor plates and cast in situ
concrete topping. The floor plate is used as permanent formwork for the cast in-place topping
with which it works structurally after hardening of the in situ concrete and finally forms the
composite concrete structure. These kinds of composite structures are very highly sensitive to
creep and shrinkage properties of concrete. In practice it has also been found that in some cases
the long-term deflections of composite concrete slabs have not been predicted with sufficient
accuracy. In this paper, time-dependent deflections of composite concrete slabs under service
loads are analytically and experimentally studied. Analysis is carried out on the basis of bending
theory with plane cross sections, taking into account creep and shrinkage of concrete and
relaxation of prestressed steel. The amount of prestressing is designed to be sufficient to avoid
cracking of the concrete. The effects of creep and shrinkage are qualified by the age-adjusted
effective modulus method and a relaxation procedure. Eurocode 2 [1] and RILEM model B3 [2]
are considered when evaluating the magnitude of creep and shrinkage of concrete.
2. Theoretical model
When a composite concrete slab is subjected to load, its response is both instantaneous and time-
dependent. Under a sustained load, the stress and strain in a prestressed concrete structure are
subject to change for a long period of time. In this analysis, the creep analysis is simplified by
applying a linear algebraic method called the age-adjusted effective modulus method (AEMM)
and for stresses occurring at different ages, the principle of superposition is assumed. The total
strain of uniaxially loaded concrete may be subdivided as
218 DEFORMATIONS OF COMPOSITE PRECAST CONCRETE SLABS
in which 8e(t) is the instantaneous strain, Ec(t,to) is the creep strain, Esh(t,to) is the shrinkage strain
and &r(t) is the thermal strain. Shrinkage is generally taken to mean drying shrinkage, which is
the observed strain associated with the moisture diffusion out of concrete under drying
conditions. In this research a linear relationship between the change in average longitudinal
moisture strain and the average moisture changes of concrete has been assumed for the
calculations. Drying shrinkage is also partially irreversible. When concrete is resaturated,
swelling of concrete occurs, but the swelling is insufficient to completely compensate for the
shrinkage that occurred on drying. Thus, we can divide shrinkage into reversible and
irrecoverable components, but a unique definition of this phenomenon does not exist. Therefore,
the analysis used generally does not account for resaturation and swelling of concrete during
environmental changes (theory, no resaturation). In composite concrete slabs, the surfaces of the
concrete components of different ages are in close interaction among themselves. Therefore the
transition of the moisture and its effect on the moisture strains of the structure must be taken into
account. The moisture transport between the construction joint surface retards the drying of
precast concrete and speeds up the drying of cast in situ concrete. In this study this has been
taken into consideration by a simplified calculation method (theory including resaturation) in
which a part of the construction joint surface is assumed to be moisture permeable when
evaluating the time function of shrinkage of precast and cast in situ concrete. Furthermore, for a
time period after casting negative shrinkage strain for precast concrete has assumed. The
evaluation of the length of the time period is done by means of diffusion theory.
If the concrete stress ac(to) is applied at time to and remains constant for time period to to t, the
load-dependent strain e<,(t,to) at time t may be expressed as the sum of instantaneous component
Ee(to) and creep component Ec(t,to). Creep coefficient <j)(t,to) is defined as the ratio of creep strain
Ec(t,to) at time t to the instantaneous elastic strain Ee(to) at time to. A stress introduced gradually
at time period to to t produces creep of smaller magnitude compared to a stress of the same
magnitude applied at age to and sustained during the period to to t. Thus, the stress increment
Aoc(t) is treated as if it were introduced with its full magnitude at age to and sustained to age t
but the creep coefficient (J)(t,to) is replaced by a reduced value which equals %(t,to)<Kt,to), where
%(t,to) is a parameter called the aging coefficient [3]. Use of the aging coefficient % simplifies the
analysis of strain caused by a gradually introduced stress increment Aac. The total strain of
concrete due to the applied stress is given by
1
+
e„(U0) ee(t0) + ec(t,t0) oc(t0) + Aoc(t)1 (2)
"c* 0/ c
The first term in Eq. (2) represents the strain in concrete at age t due to a stress ac(to) introduced
at age to and sustained during the period to to t, and the second term the strain at age t due to a
stress increment of magnitude zero at to increasing gradually to a final value Aac(t) at age t.
Test results on five statically determinate composite slabs are reported for a time period of 434
days [4]. The composite slabs were composed of two main components, precast floor plates and
cast in situ concrete. Three different types of precast prestressed concrete floor plates were
tested: floor plates with a depth of 70 mm (FP70) and 140 mm (FP140), and floor plates with a
depth of 70 mm stiffened by cold formed steel section (FPS70), which requires no propping
during construction with spans up to 10 000 mm. The total slab depth in all specimens was 220
mm. The cross-sections of the specimens also include an amount of prestressed steel wires 05
equivalent to 36 wires at the width of 1200 mm. The initial prestress applied was 1320 MPa. In
addition to prestressing steel, specimen type FPS70 also includes a cold formed L-type steel
section as a stiffener of floor plates, which corresponds to the steel area of 2080 mm (t-8 mm,
equal to 20-180 of reinforcing bars). The floor plates were manufactured at a precast plant and
M. LYDMAN 219
they (and companion concrete specimens) were heat-cured after casting (T«=50°C) 15 hours.
FP 70/2
220x600
FP 70 mm
220x600
18
600
|
'A "
|
À
T
A
T R R
A
t- I J*
2x-R
„
*'—,
"
1
AJ
i
1 1
1 1 1
J
X
FP 70 mm 18
600
A
r| +R a
2x-R + q
q
FP 140 220x600 1
1
FP 140 mm 18
r i
!s J ls
600 q
220x1200
FPS 70/1
+ stiffener FP 70 mm 36
I
r |
'a J l"A AJ
1200
FPS 70/2 220x1200 1
r 1
^— a'
Uii iiii
'A
16 16 6b
After manufacturing the composite slabs were simply supported on supports at a spacing of 7200
mm. Specimens FP70/2 and FP140 were loaded with an external sustained uniformly distributed
load q of 5.4 kN/m2 applied by concrete weights, specimens FP70/1 and FP70/2 were propped
during casting in situ concrete and therefore they also had support forces 2x-R as long-term
loads. Specimens FPS70/1 and FPS70/2 were subjected to selfweight only. The sides of all
specimens were covered by a waterproofing layer to ensure that drying rate of concrete (and
notional size) is similar to infinite slab structures. The slab specimens were stored at an average
temperature 19 °C and average value of relative humidity 50%. In addition to this, specimen
FPS70/2 was air-cured at an elevated relative humidity of 80%, which equals out of doors
atmospheric conditions according to EC2 [1], The last measurements considered in this paper
were made 434 days after manufacturing the floor plates. The deflection curves of the slabs were
measured at ten different points along the span of the slabs at regular intervals.
During casting of the slab specimen, companion concrete specimens were taken for material
property tests [4]. Concrete types tested were type HC (heat cured), type HCP (heat cured with
plasticizer) and type NC (normal concrete), which were used, respectively, for precast floor plate
of specimen FP70/1, FP70/2 and FP140, for precast floor plate of specimen FPS/1 and FPS/2
and for the cast in situ part of all specimens. Nine concrete cylinder shrinkage specimens (0xL
100x200 mm) were cast, and the strains were measured over a period of 2lA to 636 days. The
specimens were stored during measurements at a constant temperature of 20°C and a RH of
45%. Nine concrete cylinders (0xL 100x200 mm) were placed inside a creep testing frame and
total strains were measured when specimens were loaded with a sustained stress of 30% of their
compressive strength. The specimens were loaded after 2'A days and 28 days. In Figs. 1 to 6 are
presented measured and calculated compliance functions (i.e. strain caused by a unit uniaxial
constant stress given in 10'3 MPa"1) and shrinkage strains of concrete types HC and NC.
220 DEFORMATIONS OF COMPOSITE PRECAST CONCRETE SLABS
-0,02 0
0,00 0,1
Fig. 1: Measured and calc. compliance function Fig. 2: Measured and calc. shrinkage strain of
ofprecast concrete (type HC, to 2lA d) ofprecast concrete (type HC, to 2XA d)
Time(t-to) [days] Time (t-to) [days]
0,001 0,1 10 1000 0,1 1 10 100 1000
-0,16
-0,14
Precast, to=28 d
£-0.12 - Model EC2
1-0,10 I
eS
o-0,08 -i
:r-o,o6
~-0,04
iff
-0,02
0,00
Fig. 3: Measured and calc. compliance function Fig. 4: Measured and calc. shrinkage strain of
ofprecast concrete (type HC, to 28 d) ofprecast concrete (type HC, to 28 d)
Time(t-to) [days] Time (t-to) [days]
0,1 10 1000 0,01 0,1 1 10 100 1000
-0,7
-0,6
O Cast Insltu, to=28 d
_-0,5 Model B3
"L-0,4
x„„
—0,3
Model EC2
/i
>°
c
£2-0,2 /
M
co
-0,1
0,1
Fig. 5: Measured and calc. compliance function Fig. 6: Measured and calc. shrinkage strain of
of cast in situ concrete (NC, to=28 d) of cast in situ concrete (NC, to 28 d)
From Figs. 1 to 6 it can be seen that the differences between measured and calculated values of
compliance functions and shrinkage strains are within the acceptable limits compared to mean
coefficient of variation of the predicted values reported elsewhere [1,2]. The differences
M. LYDMAN 221
between measured and calculated values of compliance function varied in the range of -18.7% to
+30.4% and for shrinkage strain in the range of -10.1% to +28.8%.
Four concrete cylinders of each concrete type and testing age were tested to obtain their elastic
modulus and cylinder compressive strength development with time. In addition to these, three
cylinders of concrete types HC and NC were tested to determine their splitting tensile strength.
The mean value of the compressive cylinder strength fcm and the secant elastic modulus Ec, at an
age of 28 days were, respectively, 46.5 MPa and 35500 MPa for concrete type HC, 40.4 MPa
and 29500 MPa for concrete type HCP and 29.2 MPa and 27700 MPa for concrete type NC. The
mean splitting tensile strength was 3.4 MPa for HC concrete and 2.3 MPa for NC concrete.
• •
E E
E E
una 1>
O Experiment, FP70/1
o
T5 "TS
<D Experiment, FP70/2 a> O
Theory, incl. resaturation
- - - - Theory, no resaturation
«20
Q.
to
CL
# Experiment, FPS70/1
O Experiment, FPS70/2
Theory, incl. resaturation
Theory, no resaturation
30 15
Fig. 7: The change of midspan deflection of Fig. 8: The change of midspan deflection of
specimens FP7Oil (n) and FP70/2 (m) spec. FPS70/1(») and FPS70I2 (O)
Fig. 9: The change of midspan deflection of Fig. 10: The change of midspan deflection of
specimens FP140 fAj and FP70I2 spec. FP70I1 (a) and FPS70I1 (•)
222 DEFORMATIONS OF COMPOSITE PRECAST CONCRETE SLABS
Fig. 7 shows the change of midspan deflections of specimens FP70/1 and FP70/2. The
deflections at the end of tests were downwards for both specimens despite prestressing (for
FP70/1 and FP70/2 +7.84 mm and +27.48 mm, respectively). From the figure the effect of the
long-term external load q on the deflections of composite concrete slabs can be seen. The
permanent deflections increased about 160% and with instantaneous deflections, the increase
was over 250% in the case of long-term loading. The change of midspan deflections of
specimens FPS70/1 and FPS70/2 is presented in Fig. 8. From Fig. 8 one can see that the
deflections were first downwards but from the time moment of about 50 to 70 days upwards for
both specimens due to the prestressing and the influence of non-prestressed steel (for FPS70/1
and FPS70/2 -13.21 mm and -1.65 mm respectively). Moreover, the elevation of relative
humidity from 50% to 80% reduced powerfully (+11.56 mm) the upward deflections. Fig. 9
-
shows the change of midspan deflection of specimen FP140 including unloading at t 434 days.
Comparison with specimen FP70/2 is also made. From Fig. 9 one can see the influence of the
construction method (propped compared to non-propped construction) on long-term
deformations (for FP140 and FP70/2 +21.09 mm and +27.08 mm respectively). In both
specimens the cross sections of composite slabs are very similar, but the support forces R caused
by propping during casting the in situ concrete, increase the total deflections of the structure of
about 30% and permanent deflections of 61 %. In Fig. 10 is compared the change of deflections
of specimens FP70/1 and FPS70/1, which had no external loading. From Fig. 10 the influence of
construction method on the deformations can also be seen. Reversed support forces R increase
the deflections downwards and the total difference was 21.05 mm between the specimens.
The structural effects of creep and shrinkage in composite prestressed concrete slabs were
investigated both experimentally and analytically. Time-dependent deformations were measured
in five slab specimens and shrinkage and creep strains were measured in the companion
specimens. The experimental measurements were compared with theoretical results for
deflection calculations. The companion tests showed that the differences between measured and
calculated creep and shrinkage strains varied in the range of -19% to + 30%. The magnitude of
the variation is about the same as reported elsewhere [1,2].
The flexural tests showed that the effect of creep and shrinkage dominates the long-term
behaviour of one-way spanning concrete composite slab. The permanent midspan deflections of
composite slabs increased 160% and in the case of total deflections (permanent and
instantaneous) the increase was over 250% in the case of sustained uniform load of 5.4 kN/m2. It
was also shown that the influence of the construction method (i.e. the loading history of the
structure) on long-term deformations is significant. The support forces caused by propping
during casting the in situ concrete increased the total deflections of the composite slab by about
30%. Moreover the elevation of the ambient relative humidity from 50% to 80% prevented
almost totally the development of long-term deflections.
It is also shown that the theory including the moisture transport between the concrete
components of different ages has a good agreement with experimental results in the case of
uncracked composite concrete slabs. If the moisture transport between the concrete components
is omitted from the theory, the accuracy of the obtained results decreases distinctly. Therefore, in
composite concrete structures the moisture transport between the concrete components of
different ages must be taken into account in order to get theoretical results with good accuracy.
REFERENCES
[1] Eurocode 2: Design of Concrete Structures. Part 1. ENV 1992-1 -1:1991, CEN, 1991.
[2] Bazant, Z.P. and Baweja, S., Creep and shrinkage prediction model - Model B3, A draft
RILEM recommendation in Mater, and Struct. 28,1995, 357-365, 415-430,488-495.
[3] Ghali, A., Favre, R., Concrete Structures: Stresses and Deformations, Second Edition,
Chapman & Hall, London, 1994,444 pp.
[4] Lydman, M., Time-Dependent Behaviour of Composite Concrete Slabs, Report,
Laboratory of Structural Engineering, Helsinki University of Technology (to be published).
223
Summary
Recently, a tendency to make slab span more wide in bridge construction is growing in
Japan from rationalization of structural systems and construction cost. Also high durability
of slabs is required. To match the tendency, concrete filled I-beam grid deck has been
planned to expand the maximum available span length by rolling a new large I-beam of
20cm height. Then, the design method and fatigue durability were discussed through many
analysis and fatigue tests. This paper introduces the expanded design method with some
based analytical results and fatigue phenomena and strength of the grid decks obtained by
the fatigue tests by the Wheel Running Machine.
As the slab is a important structural member to directly support wheel loads of traffic, it
has been required to have the enough durability. Many damages however, were reported
in the ordinary RC slabs and steel orthotropic decks. Also, there are some problems on those
ordinary RC slabs and steel decks such as weight or fabrication cost, respectively. Therefore,
some kmds of precast PC slabs or composite decks are focused as innovated deck types.
The concrete-filled I-beam grid deck, which is a kind of steel-concrete composite decks and
has too much construction records over 1,000 bridges in Japan, is considered a useful
composite deck for the large span decks and some revolutions are required.Before concrete
casting, the I-beam grid deck is a semi-prefabricated steel grillage consisting of I-beams and
the transverse distributing bars. Furthermore, galvanized steel plates of 1mm thickness are
welded by a spot-welding method at the bottom surface as the form during concrete casting,
as shown in Fig. 1. At the construction site the panels are placed on the girders, and after
simple adjustment and some arrangement of reinforcements at the jointing parts of the
panels, concrete is cas ted to fill and envelope the whole steel panels. After concrete
hardening, the both concrete and steel members work together with composite action.The
use of those prefabricated panels is making some reduction of field works and construction
period of bridges and is gives improvement of bridge erection accuracy. Furthermore, the
use of I-beams of high-rigidity instead of the main reinforcements in the ordinary RC slab
makes it possible to reduce the slab thickness and the dead load.
224 FATIGUE STRENGTH OF LARGE-SPAN CONCRETE FILLED I-BEAM GRID DECK
In order to examine the influence of the difference in anisotropy, we analyzed the bending
moment varying a from 0.3 to 0.7 by expanding the slab span length. Fig.2 shows the
relationships between the slab span length and the bending moments in both orthogonal
directions. As seen from Fig.2, the bending moments vary by the ratio of bending rigidities.
When the bending moment, Mx of a= 0.4 for the cross section perpendicular to I-beam and
M. TAKAGI, K. OHTA, S. MATSUI, K. MORI 225
the bending moment, My of a= 0.7 for the cross proposed desi(n bending naît
section perpendicular to distributing bar is, the latter
bending moment seems to be too safety. But the ratio
of bending rigidities calculated with the design cross
sectional properties becomes automatically between
0.4 to 0.5. Those tendency was checked for the decks
having longer span length. Therefore, two design
methods can be recommended as follows:
(1) Following the ordinary design method using the
different ratios for the both directions. In this case,
the ratios of 0.4 and 0.7 are available even for longer
span length.
(2) At first, the cross sections of both directions are
design with the bending moment derived using one
ratio, for example of 0.4. Then the cross section
perpendicular to the distributing bars is checked to
have the bending rigidity to fulfill the assumed ratio 2 3 4 5 6 7
Table 1 Design bending moment per unit width (lm) ofslab due to T-badin
unit:kN-m/m(=kN-m/m)
\ Direction of For main member
Type of \ bending moment perpendicular to traffic
slab Kind of bending moment Bending moment Bending moment
Span of Slab \ for principle for distribution
(m) \ member reinforcement
Simply Bending moment 1.2X (0. 12L 0. 9X (0. 10L
supported through span 0<L^ 8 +0. 07) P +0. 04) P
slab
where L:Span length of slab
P:Weight of one wheel of T- 1oading(=98kN)
hUCTk 3
h bl b2
Dimens10ns(mm)
t hi h2 h3 h4 h5
Prope r
Intertia
ties (ne t section)
Area Weight
Punching (cm4) (cm2) (N/m)
h 4
hole 1-105 105 30 35 4 13. 5 10. 5 25 50 30 167 8. 50 71. 3
65 30 319 9. 98 88. 6
In ordinaiy design method, the spacing of I-beams and distributing bars is limited by the
maximum and minimum values. For example, the minimum spacing of I-beams is 10cm
and the maximum one is 25cm. Also the deck thickness is limited not to increase the dead
weight. When a design of cross sections of longer span deck is carried out considering such
limitations, the design becomes impossible and a new I-beam is required. Last year, the
authors have designed the cross section to I-beam for longer span deck until 8m. The new I-
beam was rolled with the cross section as shown in Table2. Table 2 shows the section
properties of the I-beams including old ones.
By a FEM analysis the stress concentration states were compared by changing the hole
shape and welding position to fix the distributing bar. Regarding to the hole shape, type B
decreased the stress concentration. Also the welding at the side seems to be favorable than
the welding from bottom side. Those stress Type of Type of
decreases can he recognized by the increase of
punching hole cross-point welding
fatigue fives at the fatigue tests in the order.
sTypel:
Table 3
Specimen Cross
Spedmens for I-beam fatigue test
A welding from
=bottom
Type2:
1-1
1-2
1-3
(Welding type)
No welding
Type
Type
1
2
punching hole
Type A
Type A
Type B
Type B
A. welding from
:side
Fig.6 is the S-N results and mean S-N curve about the crack initiation at the hole corner. As
seen the figure, when the fatigue data were plotted by the stress range, all data can plotted
around a curve. The curve is the essential S-N curve of I-beam.
M. TAKAGI, K. OHTA, S. MATSUI, K. MORI 227
bottom form plate and the welding point for fixing Number of cycles N
distributing bar as shown in Fig.5 and Table 4.
The specimen with 20cm thick was simply Fig. 6 S-N curvefor I-beam
(crack intnation point)
supported with the span length of 2.2 m. The
direction of I-beams is perpendicular to the direction of wheel running. A given constant
wheel load runs going and returning motion on the span center.
Table 4 Specimens for wheel running test
Specimen Type of Bot torn Cross point welding Type of Test Loading
concrete plate (Welding type) punching hole load number
IS-1 normal X standard* (Type 1) Type A 147kN 500, OOOcycles
IS—2 expansive X standard (Type 1) Type A 147kN 500, OOOcycles
IS-3 normal X al point (left side
1
Type A 147kN 500, OOOcycles
of s lab Type 2, right (+177kN until
side of slab Type 1) slab fracture)
IS-4 normal O standard(Type 1) Type A 147kN 100, OOOcycles
IS-5 normal X standard(Type 2) Type B 147kN 500, OOOcycles
(+177kN until
slab fracture)
* standard.welding the cross point on 4 sides and some innner points of panel
design strength of concrete Ock=29 4MPa, I-beam pi t ch. 18cm (I—150. J IS SS400)
distributing bar pitch upper 25cm, lower 12 5cm(D16 J IS SD345)
The aimed load and number of cycles to give on each specimen were decided basically
147kN and 500,000 cycles. The wheel load of 147kN was determined as the maximum
measured wheel load in Japanese and 500,000 cycles was enough ones over the equivalent
cycles of 147kN wheel load during 50 years at the common urban expressway. For
Specimen IS-4, only 100,000 cycles was given in order to confirm the effectiveness of the
bottom plate. For specimens IS-3 and IS-5, after loading 147kN, 500,000 cycles, the loading
with the increased load of 177kN was continued until to find out a remarkable fatigue
failure.
3.22 Fatigue failure of the basic decks
All the specimens have endured to 500,000 cycles under the load of 147kN In the specimens
IS-3 and IS-5 which were loaded until the occurrence of fatigue failure by increasing the
load, the presumed fatigue failures have occurred at some I-beams. Those failures seemed
to initiated at the punching holes near the edge of wheel where the highest shear force
occurred. The number of cycles when the fatigue failures were found were 180,000 cycles in
IS-3, and 260,000 cycles in IS-5, respectively. Also, the failured I-beams were the welded
ones with distributing bars. From those fatigue failures, it can be said that welding of
distributing bar at the I-beam hole makes weak for fatigue.
After the tests, all concrete was removed at the specimen IS-3 to check the fatigue cracks.
Then, we can compared the difference of welding points that the fatigue cracking on the I-
beam connecting distributing bars by Type 1 is more severe than the one by Type 2.
When comparing the fatigue hves of IS-3 and IS-5, the difference of the web hole shape is
not clear.
228 FATIGUE STRENGTH OF LARGE-SPAN CONCRETE FILLED I-BEAM GRID DECK
Fig. 8 shows the deflection change to number of cycles with the slab center. In the figure, a
clear difference can be seen. Namely, the specimens IS-1, IS-3 and IS-5, which have no
bottom form plate, are developing large deflection and shows an increasing rate. On the
other hand, deflections of the specimen IS-2 using expansive concrete and the specimen IS-4
having a bottom plate are very small and steady. The difference of deflection is about 2/3 at
steady states. The effect of usage of expansive concrete seems to be due to chemical
prestressing between I-beams and rising slab orthotropy. The effect of the bottom plate can
be said due to composite action with the whole concrete deck. By the composite action, the
orthotropy is rising, too.
Generally, as the concrete-filled I-beam grid deck
is attached the bottom plate, it can not be
presumed that such a fatigue failure as observed
at the specimens without the plate will occur.
Furthermore, when expansive concrete is used, a
remarkable enlargement of fatigue strength of the
deck will be expected. "0 100 200 300 400 500
Number of eye I es (x 1,000)
3.2.3 Stress expression of the hole corners
As seen in Fig. 7, the deflection becomes steady Fig. 8 Deflection of the slab(at the slab center)
after some cycles of about 10,000 cycles. The steady state is identified that the bending
rigidities in the orthogonal two ways have dropped to the ones neglecting tension side
concrete. So, through plate analysis of orthotropic plate theory, the bending moment and
shearing force acting on I-beams can be calculated. Then calibrating the flange stresses just
under the web holes and with the relation between stresses at the flange and hole corner in
a composite beam, the stresses at the hole corners of the I-beams in a deck can be expressed
with the section forces of bending moment and shearing force in the deck.
Equation 1 is the final expression for the hole corner where a fatigue crack will initiate. The
numbers before the each term are stress concentration factors depending on hole shape. By
the stress calculated with the equation and S-N curve as shown in Fig.6, fatigue life of crack
initiation can be presumed.
ft v ßo
yw ' A
i,Q
aiv w
IT B X
+ As
1 TL + 1-W 0)
where a, ß :stress concentration coefficient by the punching hole shape,
MBending moment Q' Shear force, If. Inertia of composite section, If. Inertia of I-beam,
In/. Inertia of compression side of I-beam, Inf. Inertia of tension side of I-beam,
Y w Distance between neutral axis and the comer of I-beam in composite section,
y w : Distance between neutral axis and the comer in tension side of I-beam,
As: Area of I-beam(net-section) B: effective width of concrete^: effective thickness of concrete
n: Young's modulus ratio,/;: span of secondary bending moment by shear force
4. Conclusions
The present study is carried out to apply the concrete-filled I-beam grid deck to wider span
length deck than the ordinary one. Therefore, a new large I-Beam is rolled to be applicable
to decks of 8m span length. Then design bending formulae are prepared considering the
orthotropy which is a feature of the grid deck. Furthermore, fatigue tests are carried out to
make clear the fatigue strength of I-beams themselves and the I-beams in a deck. Through
the tests the use of expansive concrete for the deck can be recommended.
References
1) Japan Road Association: Design Manual of Steel Highway Bridges, JRA, 1972.
(in Japanese)
2) Y.Maeda, S.Matsui and KKushida: Fatigue Strength of Concrete-Filled Grillage Decks,
IABSE Colloquium Lausanne 1982, pp.609-616.
229
Summary
Based on the theory of linear viscoelasticity, an incremental relationship is proposed
to account for the global behaviour of composite structure members. This formulation
can be easily implemented in a finite element program. A specific procedure allows
for construction process by successive phases. An example of application is given for a
cable-stayed footbridge made of concrete, wood, ordinary and prestressing steel. The
calculation provides detailed informations about the time behaviour of the structure.
1. Introduction
The long term behaviour of structures depends on time effects such as shrinkage, creep
or relaxation in their constituting materials. In composite members, the theological
properties of the materials causes a progressive redistribution in the stresses throughout
any cross-section. In hyperstatic structures, the internal forces will be longitudinally
redistributed. As a consequence, the serviceability of the structure may be endangered
because of excessive displacements or unexpected cracking of concrete for example.
It is generally assumed that the analysis of time effects during the service life of
structures, should refer to the theory of linear viscoelasticity. Our work is based on an
incremental formulation of the viscoelastic behaviour [1,2]. We show how to express
the global behaviour of a composite member, talcing the time dependent behaviour
of its constituting materials into account. The formulation may be implemented in a
finite element program. This approach is convenient for the analysis of time effects in
composite structures during the service life, beginning with the period of construction.
The efficiency of the method is illustrated by an application to a cable-stayed footbridge.
230 ACCOUNT FOR TIME EFFECTS IN COMPOSITE STRUCTURES
m m
R(t,t0) E(t0) ^2
M=0
where : Ao 0 ;
^a^ l V <0 (1)
/!= o
E and a^ depends on the time t0 at loading for ageing materials such as concrete. For
other materials, these parameters have constant values V ta.
Vt,At: Acr=^^A<x#, where A<Tll Kli(t) E(t) (At—At") — (l — e X"At) cr^t) (2)
i*=o
where Ae* denotes a free strain variation if any (shrinkage, thermal dilatation
represents a set of cumulative variables, whose actual values depend on the stress history
since first loading. The KM(t) parameters are defined as follows :
Ko(t)-ae(t)(l+ °) ;
«/•(«)-(l e X*At)+% V fi > 1 (3)
where
" +—-——1
E(t)
[$„"
L
0
66
for non-ageing material].
J
aM(«)
The Ofi parameters are calibrated by fitting Equation (1) to a reference curve (from
tests or design standard). The m and A^ parameters being given fixed values (satisfying
results are generally obtained for m 4). In the case of an ageing material (concrete),
the calibration procedure is performed for various initial times ta. The ctß(t0) functions
axe then approximated by suitable analytical functions.
The application of the principle of virtual work for any finite time interval [t, t + At]
yields the fundamental relationship of the finite element method :
In this equation, { Ag} denotes the increase in the node displacements, due to creep
and free strain development during the time interval At. [ff(<)] is a fictitious matrix
of stiffness which accounts for the composition of the member and the rheological
properties of its constituting materials. The nodal forces { AF*} and {Fh"(t)}
correspond to a fictitious loading. They account for the free strain and for the previous
states of stress in the different parts of the member. The corresponding expressions are
detailed in Table 1.
term of
{i^"(i)} JlBYiS^'wjdx history {sh"w}=£ JA
free
{AF*} J[BY{AS'}dx strain
Table 1 : Expressions for the fictitious stiffness matrix and loading vectors in Equation (4)
[5] is the matrix derivative of the shape functions. Every z, and J, denote the location
of the centroid and the second moment of inertia of the corresponding area A,, related
to the middle-line of the finite element. The values of F,(t), cr^"(t) and Acr*, related to
any elementary area A, of the composite member, are detailed in Table 2.
kt(t) A<
elastic Et Et Ae* 0
m, mi
viscoelastic F,(t) ^2 F,(t) Ae*
M=0 M=1
According to this global approach, the contribution of every elementary parts of the
composite beam segment are taken into account in one single finite "composite beam"
element.
the connection between parts. Each new stage induces an elastic step followed by one
or several viscoelastic steps. Similarly, the service life of the structure is described by a
number of viscoelastic steps of calculation. Further details may be found in reference [3].
In order to insure a satisfying accuracy, the size of the time intervals must be adjusted
to the rate of creeping of the material. Talcing account of the decreasing rate of creeping
with time, we propose to fix the time intervals as follows :
As -
284
5.21 crrftml ~e± t
^—Profiled sheeting
-Cable connections- S-
j
- Glulam beams
18x120
The time analysis of the structure has been performed according to the global
approach detailed above. The pylon and the deck are represented by finite "composite
B. JURKIEWIEZ, J.-F. DESTREBECQ 233
beam" elements. For the pylon, each finite element accounts for the concrete and the
reinforcing steel. Concerning the deck, the finite elements account for the glulam beams
during the second stage of the construction. Once the concrete slab has been cast (third
stage), the cross-section characteristics and the rheological properties of the concrete,
reinforcement and steel sheeting are included in the same finite element. The cables are
represented by linear finite elements. Their longitudinal modulus is adjusted to account
for the sag caused by the cable self-weight.
The calculation covers the period of construction (20 calculation steps for 110 days),
followed by the first ten years of service life of the structure (30 steps of calculation).
The low number of calculation steps results from the choice of time adjusted steps
(Eq.5). The loading accounts for the dead weight of the structure (live loads not taken
into account). The main results are discussed below.
Fig.2 : Deäection of the deck (left span) Fig.3 : Bending moment in the left span
On Fig.2, it is shown that the maximum deflection of the deck doesn't exceed 30 mm
after 10 years. Anyway, despite the procedure adopted for the tensioning of the cables,
a significant redistribution in the bending moment is observed in the deck (Fig.3).
The cable tensions increase about 15% when the deck receives its final equipment
(Fig.4), but they remain about constant during the service life of the structure. As a
consequence, the tensile stress ranges about 1.7 Mpa to 1.9 Mpa at the uppermost fiber
of the composite slab, in the centre cross-section of the deck (Fig.5).
MTVtC« IK*
200 400 600 800 2ÖÖ 3&T 400
Time [days] Time [days]
A better design of the structure could low this tensile stress in order to avoid the
cracking of the concrete. This could be achieved by adjusting the initial tensions in the
cables in order to decrease the bending moment in the mid-section of the deck, or by
removing the support at the centre pier. The last figure (Fig.6) shows the distribution of
the force of connection between the composite slab and the wooden glulam beams, along
the left span of the deck. The connection force is increasing in time. The highest values
are reached near the abutments and the centre pier.
(1 before tensioning
(2) after tensioning
(3) service life (day 0)
(4) service life (10 y.)
1Ï.5 S 34.5
Distance from left abutment [m]
6. Conclusion
The combination of a global approach of the behaviour of structural members and
an incremental formulation based on the theory of linear viscoelasticity, proves very
efficient for the time analysis of composite structures. This formulation may be easily
implemented in a finite element program. Each finite "composite beam" element
accounts for the constitution of the composite member and for the time dependent
behaviour of its constituting materials. The incremental formulation makes the
computation process quite simple, specially while taking the process of construction
into account. The choice of a time adjusted length for the calculation step, insures a
good accuracy for low calculation cost. The efficiency of the method is illustrated by an
application to the time analysis of a cable-stayed footbridge made of severed materials.
7. References
1. Bazant, Z.P., WU, S.T., "Rate type creep law of aging concrete based on Maxwell
chain", Materials and structures, Vol 7, No 37, pp. 45-60, 1974.
2. Fairbairn., E.M.R., "L'expérience brésilienne dans l'utilisation du modèle
rhéologique de la chaîne de Maxwell pour la résolution des problèmes thermo-mécaniques
des barrages en béton", Annales I.T.B.T.P., No 520, pp. 150-179, 1994.
3. Jurkiewiez, B., Destrebecq, J.-F., Vergne, A., "A step-by-step computation
method to account for time-dependent effects in building structures", in 'Advances in
computational techniques for structural engineering', Civil-Comp Press (Edinburgh),
pp. 189-197, 1996.
4. Eurocode 2, "Design of concrete structures -Part 1 : General rules and rules for
buildings", ENV.1992-1: 1991 E, 253 p., 1991.
235
Antonio J. Reis, born 1949, Civil Luis G. Melo, born in 1968, Civil
Engineer and Ph.D., is Professor of Engineer and M.Sc by the Technical
Bridges and Structural Engineering at Univ. of Lisbon, where hi did a thesis
the Technical Univ. of Lisbon. on composite bridges. He has been also
Technical Director of his own design involved in bridge design at GRID-
of many medium and long span bridges Consulting Engineers
Summary
The structural behaviour of composite bridge decks, made of two plate girders and a
reinforced concrete slab, is considered on the basis of two design examples and some
parametric studies developed with a nonlinear numerical model. The effects of the bracing
system, external prestressing, cracking of the concrete slab and the postbuckling behaviour of
the webs are discussed for ultimate (ULS) and serviceability limit states (SLS).
1. Introduction
In the last few years, composite bridge decks made of two plate girders only with thin webs
and thick flanges (up to 150mm at supports) have been extensively adopted [1,2] for both
highway and railway bridges. The webs have very often transverse stiffeners only and its
postbuckling strength is taken into consideration in design. Class 4 sections, according to the
new eurocodes EC3 and EC4 [3,4] are generally adopted for medium and long span bridges.
Special problems at ULS should be considered, namely bending moment redistributions due
to effects of cracking of the concrete slab or due to web buckling. The first problem is
considered, under different approaches, by present design codes [4 to 8]. The effect of local
plate buckling on cross section properties is considered by most of the codes for the section
analysis only and not for the structural (global) analysis. Also the effects of the bracing
system on the stresses induced on the girders by vertical loads (permanent and live), are
currently neglected. Some results for these effects are presented in this paper, and the
possible advantages of adopting externally anchored prestressing schemes in composite bridge
decks is discussed.
236 COMPOSITE PLATE GIRDER BRIDGES: SAFETY AND SERVICEABILITY
2. Design cases
Two railway composite bridge decks designed by the authors have been selected to highlight
present design criteria at ULS and SLS. The first design case (Fig: 1) consist on a 3 span
deck (main span 55 meters) one track railroad. Transversely, the deck is asymmetric due to
specific site constraints; however, the girders where located to optimize the transverse load
distribution under rail loading.
h —
End span
•+" ~
Central span
H AT SUPPORT JTWTl AT SPAN
Fig. 1 Longitudinal and cross section of a composite railway bridge over river Ave, Portugal
AT SUPPORT AT SPAN
For the parametric studies, the section was
simplified (Fig. 2). The bridge deck was
designed to be erected in 3 stages, first the
end spans plus 1/5 of the mid span at each
side, and finally the central part (33.0m) is
lifted from the river.
The second design case, refers to an
incrementally launched railway deck - the
Northern Approach Viaduct to the Tagus
Suspension Bridge in Lisbon. A composite
bridge deck (Fig.3), almost 1000m long, was
designed to be incrementaly launched from
Fig.2 Simplified section both ends in two parts (Fig. 4).
The superstructure is a continuous deck, with typical 76m spans from the northern abutment
to pier P14 (392.3m), where an expansion joint exists, and from the transition pier to the
suspension bridge to P14 (526.6m). Details of the conceptual design and of the erection
scheme are presented in [9]. For the parametric study a 3 span continous beam model
(60+80+60) was considered.
AT SUPPORT AT SPAN
Y////////Â f/M/M///]
<100/
1200x100 \800x40
4000x25 4000x25
1200x120
u 800x60
A
Fig. 3 Viaduct cross section Fig. 4 Launching phase of the viaduct
A.J. REIS, L.G. MELO 237
The main girders in a composite bridge deck are connected at the lower flange level by a
bracing system (Fig. 6). If a 3D model is adopted for the structural analysis, the part of the
overall bending (even under symmetric loading) taken by the bracing system may be
evaluated. In Fig. 5 one shows the bracing system of the composite bridge deck shown in
Fig. 3 and a typical result of the induced stresses. For the diagonal bracing, horizontal
transverse loads are introduced at the "gussets". These loads produce transverse bending
stresses at each flange; of course the two transverse bending moments are self equilibrated at
the overall section. For the section analysis at ULS of the deck, it is acceptable in design
pratice to assume some redistribution of the stresses at the lower flange.
4. External prestressing
External prestressing in composite plate girder bridges, has been adopted in several design
cases. The main advantage is a reduction of the amount of steelwork required to achieve the
resistant capacity at ULS. External prestressing shall be considered as an external force and
the beneficial effect due to the increase of the prestressing force during the loading stages
may be neglected. That compensates for some reductions of the prestressing effects due to
geometrical nonlinearities. The main disadvantage of the classical external prestressing
scheme, where the cables are attached to the steel structure, are the high compression forces
induced at the girders requiring complex details at the anchorages. This disadvantage is
eliminated when an externally anchored prestressing scheme is possible. The example in Fig.
6 shows an external prestressing scheme solution developed for the railway deck of Fig. 3.
The cables were designed to be stressed at two stages: immediatly after erecting the steel
structure, by incremental launching, and after casting the deck slab. In Fig.7 one shows the
benefitial effect of the external prestressing in reducing the maximum stresses at the flanges.
With this scheme (case B) it would be possible to adopt 80mm thick plates at the flanges at
the support sections instead of 120mm plates required by the conventional composite girder
(case A). A reduction of about 25% in the total amount of the steelwork could be achieved
by the externally anchored prestressing scheme. Of course, part of these savings are canceled
by the cost of the prestressing cables, anchorages and deviators. For the present design case
the conventional solution (without external prestressing) was prefered mainly because the
execution time was shorter.
238 COMPOSITE PLATE GIRDER BRIDGES: SAFETY AND SERVICEABILITY
68 MPo 32 MPo
150 MPo
For the model developed, the nonlinearities taken into consideration are restricted to cracking
effects in the slab and local buckling at the steel section. The slab is considered as an axially
loaded bar element under pure tension or compression. This assumption is acceptable for
medium to large spans, where deep girders are adopted. The constitutive relationship [10] for
the slab under tension is shown in Fig. 8. The tension branch is similar to the one proposed
in Jan. 1997 Draft of Part 2 of EC4 - Annex L, for the "Effects of tension stiffening in
composite bridges ".The finite element considered is a bar element with only two degrees of
freedom at each node (vertical displacement and rotation), and the linear system of equations
is solved by the Gauss Substitution Method. At each step of the loading process, a new
stiffness matrix is evaluated, according to an iterative procedure, based on the Newton-
Rapson technique. This new matrix is updated taking into consideration the effective steel
section in the postbuckling range (based on EC3
formulae), and cracked or uncracked properties of
the concrete slab. The numerical model was
developed for continuous plate girder bridges,
with a composite deck, taking into consideration
the evolution of the static scheeme during
construction.Span by span, incremental launching
or other execution methods, may be considered.
At each step, steel or composite (full interaction)
Fig. 8 Constitutive relationship for section properties are evaluated depending on the
concrete slab under tension phase at which the slab is casted.
allowed but with reduced moments of inertia to account for cracking effects.
Typical results are presented in the following Table for moments of inertia and bending
moments evaluated according to the codes mentioned above. It may be concluded from Table
1, that the influence of the different approaches is much greater for I than for M.
Tab. 1 Comparative results for I and M (for live load) according to several design codes
When evaluating cross section properties, like
I, the modular ratio n=Es/Ec between the
elastic modulus of steel and concrete, time
dependent effects are generally taken into
consideration in the present codes [4 to 9] by
multiplying the modular ratio for short term
actions n0, by a coeficient [ 1 + i]/tp( t, t 0) ]
0 20 40 60 80 100120140 160180200 Here <p is the concrete creep coefficient and v|t
Distance [m]
is dependent on the type of action involved. For
example, i|j =0, 0.55, 1 or 1.6 in EC4 [4]
respectively for short term, permanent,
shrinkage effects and prestressing by imposed
deformations. In the French instructions i|r(p=2
for all permanent actions and in the new
Spanish recommendations a more sophisticated
0 20 40 60 80 100 120 140 160 180 200 age adjusted effective modulus method is
Distance [m]
proposed. The nonlinear model described was
used to develope some parametric studies for
I— As 160 cm2 As=240 cm2
ultimate load of composite class 4 sections, and
Fig. 9 Influence of reinforcement on to investigate bending moment redistributions
the slab behaviour due to cracking and local buckling effects.
By taking the design case of Fig. 3, the influence of the concrete strenght class, thickness of
the slab, amount (density) of reinforcement, bond action of the reinforcement and the
sequence of the erection scheme were investigated. As a general conclusion, for the ULS of
the composite sections, most of the parameters refered to above have very little influence. In
what concerns S LS, namely crack widths, the most relevant parameter is the amount of
reinforcement (Fig. 9) as also the sequence of the erection scheme in what concerns
limitation of crack widths and deformability. In what concerns bending moments, the results
obtained by elastic uncracked models are about +5% at support and -10% at span sections
compared to the ones obtained by the nonlinear model described in section 5. The results
obtained for ULS by the elastic cracked model, as proposed by EC4, are less than 1 %
different for bending moments and +5% for displacements compared to the ones obtained by
the nonlinear model.
240 COMPOSITE PLATE GIRDER BRIDGES: SAFETY AND SERVICEABILITY
Another aspect is related to code requirements concerning the effects of local buckling in
composite bridge decks. These effects are only considered when evaluating cross section
resistance (i.e at section level) and not for the structural (global) analysis. In composite plate
girder bridges with Class 4 sections, the webs are very often in the postbuckling range at
ULS. So, a redistribution of bending moments shall be expected. In Fig. 10 one compares
results obtained by the nonlinear model and from EC4 model, for the support cross section
of the bridge deck shown in Fig. 2.The difference in the effective width of the web is +11%
and at the ultimate moment is only +2%. The force at the slab is 22% lower by EC4 mainly
due to cracking effects. In Fig. 11 the influence of the phase at which composite action is
considered is highlighted for the cross section of Fig. 2, under increasing imposed curvature.
5483 kN 6689 kN
Composite
> 1.0
u
>
'5 0.9
ë
ëo 0.8
Steel Composite
m 0.7
§
8 0.6
W,=-97012it/Vm M„=-99334iWm 0.5
0.0 0.2 0.4 0.6 0.8
a, =-355 MPa a, =-355 MPa Curvature [10~-3/m]
a. 218 MPa a. 198 MPa
Fig. 10 Ultimate moment obtained from Fig. 11 Influence of the composite action
linear (EC4) and nonlinear analyses on the effective width of the web
8. Conclusions
Safety and serviceability problems for composite bridge decks were discussed on the basis of
two design cases. A nonlinear model was adopted for parametric studies and results were
compared with the ones obtained by a draft for EC4 (Part 2).
References
[1] - Virlogeux,M.-Les ponts mixtes associant l'acier et le beton précontraint, Bulletin Pont Metaliques n. 15.
[2] - Reis,A.J.- Recent developments in composite bridges, in steel structures, Eurosteel'95, Balkema, 1995.
[3] - Eurocode 3 - "Design of Steel Structures", Fev. 1992.
[4] - Eurocode 4 - "Design of Composite Steel and Concrete Structures", Jan. 1997.
[5] - BS 5400 (part 5) - Steel, Concrete and Composite Bridges.
[6] - Cire. n°81-63,"Relative au règlement de calcul de pont mixtes acier-béton", 1981.
[7] - SIA - "Construction Metâlliques", Société suisse des ingénieurs et architectes, 1990.
[8] - RPX - "Recomendaciones para el proyeeto de puentes mixtas para carreteras", 1996.
[9] - Reis, A. J. and Melo, L. G. - "The incremental launched railway deck for the Tagus bridge viaduct",
New Technologies in Structural Engineering - rep., Lisbon 1997.
[10] - Ghali, A. and Favre R. - "Concrete structures: stresses and deformations".
241
Roberto Leon received his Ph.D. from Daniel J. Flemming received his M.S.
the U. of Texas at Austin in 1983, in structural engineering from the
and taught at the U of Minnesota University of Minnesota in 1994. He
for 10 years. He joined Georgia Tech is currently studying for a divinity
in 1995. degree.
Summary
This paper reports on the results of one test on a half-scale, two-beam, two-span composite
bridge. The test is unique in that varying amounts of composite action (50% and 80%) were
used, and in that actual moving loads were used to load the structure. The experimental and
theoretical loads compared favorably for the case of 80% interaction when the actual material
properties were used in the calculations. The results also indicate that if 50% or less interaction
is provided, the structure may not be able to carry cyclic loads into the inelastic range.
1. Introduction
While the high-cycle, low-amplitude fatigue behavior of shear studs has been studied
extensively, relatively little is known of the performance of shear studs under reversed cyclic
loads (low-cycle, high amplitude regime). This aspect of shear stud performance is of great
interest in rating steel composite bridges [Galambos et al. 1993] which are often classified as
structurally deficient because of insufficient strength to handle the increase in truck weights that
has taken place since they were put in service. If it can be shown that the shear studs have
sufficient strength and stiffness to allow the structure to shakedown under large overloads, it
may be possible to increase the rating in many of these bridges so that they comply with current
loading criteria. This particular aspect of bridge design is the focus of this paper, in which the
results of a test on a two-span, half-scale composite bridge will be reported [Flemming 1994].
More generally, the development of knowledge regarding the degradation of shear interaction
between steel beams and concrete slabs subjected to cyclic loads is of fundamental importance in
understanding the behavior of older steel beam-concrete slab bridges where the interaction may
come either from partial encasement or friction and adhesion at the members interface. The
behavior of shear studs under cyclic loads is also of great interest in other areas of structural
engineering, such as in improving our understanding of composite beam behavior in moment
frames subjected to seismic forces, where the floor slab acts as diaphragm in transmitting the
inertial forces.
242 SHAKEDOWN PERFORMANCE OF COMPOSITE BEAMS WITH PARTIAL INTERACTION
2. Shakedown of Bridges
A recent study on the behavior of straight, continuous composite bridges has suggested that the
useful life of many deficient short-span structures could be extended significantly if the
structures were allowed to enter into the inelastic range for a low number of cycles [Galambos et
al. 1993]. This work developed model rating systems based on the theory of shakedown [König
1987], which is a well-developed aspect of plastic design of structures. Shakedown is a term
used to describe structural behavior under large cyclic loads. Shakedown implies that after
repeated applications of a prescribed load history, which exceeds the elastic limit but not the
plastic collapse load of the structure, the residual deflections in the structure will stabilize.
Residual deflections are the permanent deformations remaining in the structure after the load has
been removed. Because yielding has occurred, there will be additional forces, known as
residual moments, locked into the structure when the loads are removed. It is important to note
that shakedown implies some damage to the structure, generally in the form of yielding of main
members, and thus may result in a serviceability failure. However, a key feature of shakedown
is that once the deflections stabilize, the structure will respond elastically to any additional
cycles of the prescribed load history.
In the initial work done by the senior author [Galambos et al. 1993] as well as in that by Grundy
and Thiru [Grundy and Thiru, 1995], composite beams with different degrees of interactions
showed a marked tendency to loose strength and stiffness due to slip at the steel-concrete
interface. Slip at the interface could be due, among many other reasons, to damage to the
concrete in bearing, cracking of the concrete, damage to the stud due to cyclic plasticity, and/ or
propagation of low-cycle fatigue cracks. Other sources of strength and stiffness deterioration of
concern in composite beams are local buckling and local bending beneath the shear studs of the
steel beam flange. To address the problem of shear connection deterioration an experimental
program that included the testing of a half-scale, two-span bridge was developed. This
programs is unique in several ways. First, as far as the authors know, this is the first large scale
shakedown test to be carried out on a composite bridge in the world. Second, it is the first
laboratory bridge test in the U.S. to use an actual rolling load with rubber tires, as opposed to
concentrated loads provided by loading jacks, to apply the loads to the structure. Third, it is the
first attempt at using partial composite action in this type of structural system, and the first to try
to quantify slip at the interface for a partial composite bridge beam.
3. Experimental Work
The design of the model bridge required that compromises be made among scaling laws,
structural simplicity, and loading requirements [Flemming 1994], The model bridge was based
on a prototype designed by the current AASHTO LFD Specification. This design was then
scaled by following as much as possible similitude laws, recognizing that it was impossible to
scale all quantities properly. The most important consideration was the use of shear studs
whose behavior closely simulated that of the ones in the prototype. The smallest shear stud
commercially available is 10 mm, while actual shear studs in bridge construction range from 19
mm to 25 mm. Thus a half-scale model was deemed to be the smallest that could be tested.
The one important subject that was not directly addressed in the scaling process was the quantity
and placement of shear studs. The shear stud design for the half scale structure was carried out
R.T. LEON, D.J. FLEMMING 243
using the AASHTO strength criteria alone. Thus fatigue criteria, which will usually govern in
the design of the shear connection, were not considered. Moreover, since the intent of the
research was to study the behavior of the interface, something less than the full interaction as
given by code equations was desirable. Thus, it was decided that 80% of the design requirement
for shear studs (100% interaction two 9 mm studs at 200 mm) would be placed in one span and
50% of the design requirement in the adjacent span. This approach was used to maximize the
amount of information that could be gathered by this test concerning the behavior of the
structure in the shakedown range. Fig. 1 shows a schematic view of the structure and its
loading system. The loading system consisted of the tandem axle taken from a real truck. The
loads for the shakedown test were increased by adding large lead ingots to the tandem axle.
a Pulley system
274311220 mm-| |«- $
for moving load
I
'I
b F mm
[
àf Stop
1
2438 mm
E
101 mm
12 mm bars at 101 mm
-W360 x 33
fc 26.4 Mpa
Two 9 mm studs- -9 mm bars at 190 mm
at 250 mm in East span
at 406 mm in West span
1828 mm All bars are 420 Mpa
Figure 1
- Details of the test specimen.
4. Results
The results will be summarized first with a plot of the total axle load against the maximum
centerline deflection for each span (Figs. 2 and 3). For each beam, the maximum deflection at
each cycle at a particular load are shown, i.e., only the envelope of response is shown. Fig. 2
shows this data for the two beams (labeled North and South) for the West span, which had an
80% interaction. In this plot the incremental collapse limit for the composite section, the
composite yield capacity, and the non-composite yield capacity are also shown for reference. A
cycle of load was defined as one full pass (forward and backwards) of the bogie. Two lines are
shown for each beam to indicate the different directions of travel for the axle, since the direction
of travel did seem to influence the behavior of the specimen.
244 SHAKEDOWN PERFORMANCE OF COMPOSITE BEAMS WITH PARTIAL INTERACTION
The test was begun by applying to two full cycles of load with the axle self-weight only (75.6
kN). The response of the structure to these cycles was purely elastic. The first lead ingots were
then loaded onto the tandem axle to give a total load of 165.1 kN. This value was very close, but
lower, than the calculated first yield. The response of the structure was mostly elastic, but some
strain gages on the West side of the center support showed slight yielding. The load was then
increased to 195.7 kN where the first signs of yield occurred. Four cycles were applied at this
level, until the changes in residual deformation from one cycle to the next were less than 1.3
mm. The latter was the criteria used throughout the test to determine whether shakedown had
been achieved. As can be seen from Fig. 2, the West beam failed gradually, and reached its
nominal shakedown capacity. A hinge formed at the centeriine support first, with clear
indications of aplastic hinge behavior occurring at a load of 241.8 kN. At this level it took 10
cycles to reach shakedown. The loads were increased in five increments up to 285.1 kN, where
it took 14 cycles to shakedown. A slightly increase of the load to 292.7 kN resulted in the
formation of a full plastic hinge at the center of the West span and the collapse of the structure.
Fig. 3 shows similar data to Figs. 2, but for the North Beam on the East span, which had 50%
interaction. The behavior of this span was quite different. After the hinge formed at the center
support at a load of 241.8 kN, successive passes on the East span resulted in the immediate
formation of a plastic hinge and large deflections. With a slight increase of the load to 256.7 kN,
the East span collapsed. The collapse was quite rapid after the second pass, with the deflection
taking only a few seconds to reach the temporary support jacks placed about 150 mm below the
undeformed bridge to prevent a complete collapse of the structure. This span only reached the
plastic moment capacity of the bare steel beam. Much of the composite action was lost in the
East span due to a progressive failure of the shear studs beginning at the East span centeriine
and propagating to the center support.
R.T. LEON, D.J. FLEMMING 245
Fig. 4 shows similar deflection data but plots it versus the cycle number. Fig. 4 also shows the
bounds provided by a simple analysis assuming that the section remained either fully composite
or non-composite throughout the test. It is evident from Fig. 4 that the beam did not behave as
fully composite past the first few cycles. However, it was able to mobilize considerable strength
and ductility such that for purposes of strength calculations the West beam can be assumed to
have acted compositely. From the degree of yielding observed it is clear that substantial strain
hardening occurred and this is probably the best explanation of the strength performance.
Perhaps the most surprising data obtained in this test concerns the slip at critical locations in the
positive moment regions of the East and West beams respectively. The data from eight
different slip and strain measurements indicate that the slip did not stabilize under any given
load cycle. For the East beam, which behaved mostly as non-composite, the slip reached 0.8
mm just before the structure collapsed. For the West beam, which acted compositely, the slips at
the same load level were roughly half those of the East beam. At collapse, the slips in the West
beam were close to 1.1 mm. It is interesting to compare this range to typical monotonie tests on
shear studs, where 0.8 to 1.0 mm of slip is considered to be the service limit and 4 to 5 mm of
slip at ultimate is considered desirable. Although at lower levels of load (less than 241.8 kN) the
slip seemed to be beginning to stabilize after a few cycles of load, near the end of the test the
slips seemed to increase almost constantly with every cycle of load.
5. Conclusions
The experimental and theoretical load calculations for the test compared favorably after the
actual material properties were used in the section capacity calculations. The theoretical static
collapse load limit in the West span was within 2.8% of the experimental load at which failure
was observed. In the East span, however, the actual incremental collapse load limit was just
slightly above the theoretical non-composite incremental collapse load limit. The theoretical
composite static collapse load should have been 35.8% higher than the observed failure load.
This seems to imply that if 50% or less of the required shear connection is provided one should
not expect the structure to carry cyclic loads into the inelastic range. On the other hand if at least
80% of the required connection is provided the structure may achieve its fully composite
capacity assuming elastic-perfectly plastic section capacities. The most important experimental
observation from this test is that the slip did not stabilize in the case of extreme overloads.
6. Acknowledgments
This work was sponsored by the Minnesota Department of Transportation and the Center for
Transportation Studies at the University of Minnesota.
7. References
Summary
The contribution deals with the time-dependent analysis of the state of stress and strain
distributions on the contact surface of a reinforced concrete two layered element (in the sense that
„new material" means topping layer and „old material" means prestressed plate structure). We
suppose that the structural system is linearly viscoelastic quasihomogenic isotropic continuum.
The linear aging model for concrete creep is used. The theoretically derived solution procedures
are examplified by particular numerical examples
L (u) f (1)
in the domain Q Q, cj Q2 (Q <z E3) which is bounded per partes smooth surface
S 5, u S2 (5, o S2 0), u u(P, t) {u(x, y, z, t), v(x, y, z, t), w(x, y, z, t)} is a sufficiently
smooth vector-function, where P e Q, / > 0. The solution of the Eq (1) satisfied the following
conditions
ux^ FX
+o(0),c,1nK onS-(s'uSb), (3)
The basic idea of FEM is to represent the displacement functions u (h,v,w) within the element
by continuum shape functions of the form
where N, (i 1,2,...Z,) are the shape functions associated with the L nodes and
ui (/ \,2,...L) are nodal values of the displacement u(x, y, z, t), etc. The strain rate field within
the element can be defined as [3]
where B is a matrix of gradients of the shape functions and 6 is the nodal velocity vector. The
equilibrium of the stress rates ct is given by
£ j*BT6dV
Ve Ve
Ra, (8)
where R'
is the applied nodal load rate vector. The stress-strain relationship for inelastic rate
processes may be given by the formula
Tl 91(CT,t,T).
We may transform the system of equilibrium equations into the matrix equation
K5=R'+R'1 on Q, (10)
where K ^ J BTDB dV
e Ve
is the assembled stiffness matrix and
is the vector of the creep process. After some rearrangements Eqs (10) and (11) can be written in
he incremental form over a time interval
where
Numerical schemes for the solutions of Eqs (12), (13) are discussed in [4].
J. SUMEC, N. JENDZELOVSKY 249
where
i
-V
-V
n(t,f)=j(.,f) j(l + v)
2(1 + V)
;d + v)
and
J(u') — +Y
E(t') ft
AT1
' (t')(l - e~(1~n'T) E(t') E28J
V4 + 0.85t
A
where are constants (retardation times) and Aj are aging coefficients. For shrinkage of
concrete we can assume the relation [6]
-
edi (t) 0.0008c' (t 7) / (35 +1 7) - (15)
6
in which t is in days and drying is assumed to begin at t0 7 days, c'= lie, is correction factor
[6].
3. Numerical Example
This section deals with the numerical solution of some viscoelastic aging creep problems - two
layered reinforced concrete plate with a prestressed lower layer. The analysed plate is simply
supported in positions a and b (Fig. 2). The length of the system is 6000 mm, width 1190 mm.
The thickness of the lower layer is 70 mm and upper concrete layer 170 mm, respectively. The
cross-section of the given structure is in Fig.l
The plate carries a uniformly distributed load with the intensity q 5,68 kN / m2 (dead load)
and a line load with the intensity p 20,07 kN/m (Fig.2). The triangle represents the course of
the prestressed force in the individual strings. The prestress on 8 wires 5mm is realised.
<|>
250 CONTACT SURFACE OF A TWO-LAYERED RC STRUCTURAL ELEMENT
P 33,4 kN
q 6,76 kN/m
I \1/ \L ,1/ J, ,1/ \L ,1/ \L \L \L J, \L J, J, J/ J, \L J, J, J,
\ c
1
d
7
prestressing forces
[kN]
900 1800
Characteristics of the materials are as follows: a lower concrete layer of type B 30 (E! 32500
MPa) according to the code STN 73 1201 [7] and an upper layer B 20 (E2 27000 MPa),
respectively. The following cases were considered:
1) upper layer shrinking, lower layer non creeping s/nc
2) both layers creeping O c/c
The time intervals taken: 0, 5, 10, 20, 50,100, 300, 500, 1000 and 3000 days. We assume a 28
day time interval for hardening of concrete. The top layer was laid down after 200 days.
Between the layers the conditional technological horizontal defect range of the half-span was
taken into account, too (case B). It means, that the following cases were solved:
A - without defect on section a - b
B - with defect on the section a - d.
Figs 3 and 4 represent the time - dependet vertical deflections in the half-span of the system.
Normal and shear stresses for individual cases (A and B) under line - load (position c - where
max. bending moment was expected) were calculated. Courses of normal stresses o, for cases
A and B are given in Figs 5 and 6.
Fig. 5 Normal stresses ox(c,t) for case A; s/nc, O c/c. The full lines represent values
a^c.t ->oo) anddashedlineo^c.tsO)
252 CONTACT SURFACE OF A TWO-LAYERED RC STRUCTURAL ELEMENT
Fig. 6 Normal stresses o,(c,t) for case B; s/nc, O c/c. The full lines represent values
a x (c, t -» oo) and dashed line a x (c, t 0)
4. Conclusions
- on the interface of the layers cumulation of the shear stresses on the perimeter of the structure
occured
- creep - deflection was 3 time larger than classical elastic deflection, that positively influence
the course of the stresses
- the presence of technologically conditioned defects on the interface of the layers negatively
influenced the stiffness of the system, in spite of the prestress on the lower layer
- a very significant element of the layered system is the technological surfacing, of the interface
of the layers.
Acknowledgement: The writers are grateful for support from the Grant Agency VEGA,
Grant No. 2/1264/94
References
1. Mikchlin, S. G.: Variational Methods in Mathematical Physics (in Russian), Nauka, Moscow
1970.
2. Washizu, K.: Variational Methods in Elasticity and Plasticity. 3rt Ed., Pergamon Press, N.Y.,
Oxford, Tokyo, 1982.
3. Zienkiewicz, O.C.: The Finite Element Method. 3rd Ed. McGraw-Hill, N.Y., 1977
4. Wiliam, K.J.: Numerical solution of inelastic rate processes. Comp. & Struct., £,1978, pp.
511-531.
5. Sumec, J.: Mathematical modelling of materials with time-dependent physical properties (in
Slovak). Internal Research Report HI-3-4/9.4, USTARCH SAV, Bratislava 1983
6. Basant, Z.P.: Theory of creep and shrinkage in concrete structures: a precis of recent
developments. Mech. Today,2, 1975, pp 1-93.
7.STN 73 1201. Design of concrete structures.
253
Summary
The effect of deformability of connection in composite beams was investigated on the basis of an
experimental research. The performance was studied in both ultimate and serviceability
conditions. The differences in performance between bridge and flooring composite beams are
shown. The use of composite construction to improve buckling resistance and to reduce shear lag
effects in box girders is proposed.
1. Introduction
Concrete steel composite structures are very efficient. Concrete is effective in compression and
steel in tension, compressed steel parts are stiffened against buckling by a concrete slab. In old
composite structures, a very stiff connection between steel and concrete was designed. Welded
studs are connectors which are used very often mainly because of the simple production. The
studs are flexible to some extent and therefore only partial interaction between a steel flange and a
concrete slab exists. This implies that slip occurs at the interface, and so causes a discontinuity of
strains. For most connectors used in practice, failure by vertical separation is unlikely and any
uplift would have only negligible effect on the behaviour of the composite structure If the
deformability is taken into account in the design, the steel concrete structures can perform
satisfactorily during their whole service life.
2. Stud resistance
A series of experiments has been carried out at the CTU Prague, in order to get the load slip
diagrams, which made it possible to observe the progressive failure of one stud. They were
arranged so that the results were not influenced by redistribution of the load between the two
studs as it is in the case of a usual push out test. The loading process was controled by slip
between steel and concrete which made it possible to detect the drop of ultimate load after failure.
The mode of failure depends on the length of the stud embeded in concrete slab, on the diameter
254 PERFORMANCE AND STUD FAILURE IN STEEL-CONCRETE COMPOSITE BEAMS
of the shank of the stud and on the quality of concrete It may be simply concluded that the short
studs fail due to failure of concrete adjacent the stud and longer studs, which are well anchored in
concrete usually fail by reaching the ultimate load carrying capacity of the shank in shear just
above the steel beam
The load - slip diagram is affected by the failure mode of the stud at the final stage just before its
failure A typical load-slip diagram of one stud is plotted in Fig. 1 The initial stage of the diagram
shows a development of the shear force without any relevant slip Then the slip starts to develop
and the stiffness of connection is reduced significantly Before failure of the stud a considerable
increase of slip can be seen Results of the experimental study carried out at studs of different
sizes showed that no slip occurs if the shear force is about 25-30% of the ultimate shear force or
even more
Stud 16/60 mm
W
0 0,5 1 1,5 2 2,5 3 3,5 4
Slip [mm]
To model failure of connection between steel and concrete, a simplified method has been
developed [2] The performance of studs is assumed according to a quadrilinear diagram (Fig 2)
The first two lines (ascending part of the diagram) can be determined from the experimental
results directly The descending branch can be taken also from deformation controlled tests and
the last horizontal line represents friction between steel and concrete after failure of studs If no
experiments are available, the descending branch can be assumed according to an empirical
calculation, e g [1]
The behaviour of composite beams depends on the geometry of the cross-section. The beams
used in buildings have rather thick and wide concrete slabs compared with the steel beam
dimensions. The neutral axis usually lies in the concrete slab Thus, a part of concrete section is
under tension, which causes cracking and reduces stiflhess of the slab. The studs are anchored in
the tensile zone, and so their stiflhess is necessarily lower than that of the studs anchored in
concrete under compression. The contribution of the slab to the overall stiffness is high and the
structure is sensitive to cracking, creep and shrinkage effects
J.L. VITEK, V. KRISTEK 255
nShearflcw
v„ v„ Slip
Fig. 2 The cross-section of composite beams studied and the idealized shear - slip diagram
Bridge beams behave differently. They are rather deep and the concrete slab is relatively thin. The
width of the concrete slab is comparable with the dimensions of the steel beam. The neutral axis
lies normally in the steel part of the cross section and thus the concrete slab is fully in compression
zone. The studs are anchored in the compression zone of the slab which increases their load
carrying capacity.
The distributions of the bending moments and axial forces in composite beams with deformable
and stiff connectors are different. Near midspan, where the connection is stiff, the total bending
moment is taken by the axial forces in steel and concrete, which form a couple. On the other hand,
at the end regions of the beam, the connection can be rather weak due to partial failure of
connectors, and then the total bending moment equals roughly the sum of the moments in the
cross-section parts, with a reduced contribution of the couple of axial forces. Typical stress
distributions in the concrete slab and in the steel beam of a bridge composite beam are plotted in
Fig.3a,b together with the shear flow variation (Fig.3c) along the bridge span of the length 30 m.
The stress variations are plotted in the stage when the connection partialy failed and the stress in
the steel beam reached the stress limit. The sudden changes in the stress variations are due to a
simplified shape of the shear force - slip diagram as shown in Fig.2. In a real structure, the
changes would be smooth, but not disappear. In the central part, where no slip occurs, the stress
becomes lower and in the end parts the slip develops. The extreme compression stress in concrete
is in the central part, the extreme tensile stress at the bottom surface of the concrete slab develops
approximately in the middle of the end parts (Fig.3a). The extreme stresses in steel are at the
location where the slip starts to develop (Fig.3b). In the case of stiff connection, the stresses
would correspond to those stresses plotted in Fig. 3 at the central part of the beam and the stresses
in steel (particularly in the top flange) would be significantly underestimated against the real
arrangement with deformable connection. Tensile stresses occur at the bottom surface of the
concrete slab (Fig. 3 a) where compression stresses in concrete would be expected in the case of
stiff connection.
The effect of variable stiffness of connection has been studied on a bridge beam of 35 m span. The
depth of the steel I beam was 2.84 m and the thickness of the 2.125 m wide concrete slab was 0.2
m. The deformability of connection significantly affects the stress distribution and deflections of
composite beams. However, its effect on stress and deflection variations is not proportional with
respect to the deformability of connection. The degree of deformability varies from 0 (stiff
connection) to 1 (no connection) in this study. The bridge beam with deformable connection
exhibits a deflection wv„/ and the bridge beam with stiff connection has a deflection ws„ff. The
256 PERFORMANCE AND STUD FAILURE IN STEEL-CONCRETE COMPOSITE BEAMS
Fig. 3 Stress distributions and shear flow variation along the length of the beam
ratio Wde/ws„jjr in dependence on the deformability of connection is plotted in Fig.4a. The thin line
illustrates the effect of short term loading (age of concrete 28 days). It can be seen that the
increasing connection deformability has initialy low effect on the deflection ratio, in the middle
part of the diagram a substantial increase of deflection takes place and in the right hand side part
of the diagram again very small change is seen. The thick line (10000 days after loading) shows
increase of deflection ratio due to long-term loading. At the beam with stiff connection the effect
of creep of concrete is significant, because of a great contribution of concrete slab to the overall
bending moment, due to action of axial forces. At the beam with very deformable connection, the
contribution of concrete slab to the overall bending is small, because the flexural stiffness of the
concrete slab is much smaller than the stiffness of the steel beam. Therefore the creep effect is
negligible. The development of stress redistribution in steel exhibits a similar trends like
deflections.
The behaviour of the bridge beams was compared with the behaviour of the flooring beam. The
depth of the rolled steel I beam was 312 mm, the concrete slab was 120 mm thick and 2.5 m
wide. The deflection ratio similar to that of the bridge beam in dependence on the deformability of
connection is plotted in Fig. 4b. The thin line (short term loading at the age of concrete 28 days)
shows a similar trend as that of the bridge beam, however, the increase of deflection is higher
If
(bridge beam 1.75, flooring beam 2.5). the long-term load is assumed (thick line - 10000 days),
the increase of deflection is even more significant. The concrete slab represents a significant part
of the cross-section and its stiffness, which is reduced due to creep, influences significantly the
deflection of the whole composite beam. Flooring beams are therefore more sensitive to the creep
than bridge beams. The redistribution of stresses is also more significant at floor composite beams
than at bridge composite beams.
J.L. VITEK, V. KRISTEK 257
Fig. 4 Deflection ratio of the beams with deformable and stiff connection
5. Buckling resistance and controlling the effect of shear lag in box girders
The composite action may also be very effectively applied to increase the buckling resistance in
the bottom flange plates in the hogging moment regions over the internal supports of continuous
girders which are simultaneously very suspectible to the effect of shear lag.
It is well known that the effects of shear lag are particularly severe in local regions immediately
adjanced to applied concentrated loads, in flange plates that are wide in relation to their length
and in flanges that are longitudinally stiffened. In the case of a continuous steel bridge girder,
these three circumstances can coincide to produce an exceptionally high shear lag effect in the
lower flange over an intermediate support. Bridge bearings are often positioned under girder webs
and bear almost directly on to the girder. The points of contraflexure are usually close to the
intermediate support so that the effective span of the girder is short and the resulting span/width
ratio is well within the critical range. Within such a region of hogging moment, the lower steel
flange may be heavily stiffened against buckling under longitudinal compression.
A substantial reduction in the accentuated shear lag effect due to longitudinal stiffening is
possible. The flange plate within the critical region can be stiffened with a layer of concrete that is
made to act compositely with the steel flange which will be easy to place on the lower flange of
the box over the intermediate support.
As an example a multispan, continuous bridge girder is considered and the performance of the
two types of cross-section shown in Figs 5(a) and 5(b) is compared. In the first case, both flanges
of the steel girder are stiffened by longitudinal ribs in the conventional way, with the lower
compression flange stiffened by nine ribs to give a ratio of stiffener/flange plate area of 0.2. At an
internal support, the distance between points of contraflexure is calculated to be 15.2 m so that
the flange span/width ratio is 2.2. The girder is subjected to uniform line loading of 100 kN/m
above each web and the upward reaction force at the support is calculated as 3600 kN; this is
distributed over a bearing length of 100 mm.The distribution of the longitudinal stress across the
width of the bottom flange above the intermediate support is plotted in Fig 5(a). It shows the
expected shear lag effect with the amplified edge stress reaching a maximum value of 147.2
N/mm2 The alternative solution is shown in Fig. 5(b). In this case, the thickness of the lower
flange has been reduced from 12 mm to 8 mm and the lower flange stiffeners have been removed
258 PERFORMANCE AND STUD FAILURE IN STEEL-CONCRETE COMPOSITE BEAMS
(a) i_ 2 1m (b)
300 300 300
HHH
Stilleners 100 x 12 mm Slilfeners 100 x 12 mm
12 mm
Slilfeners 125 x 15 mm
8 mm I 70 mm1 J
j. 690 J. 690 j. steel concrete
Fig. 5 Stress in an typical bridge girder, (a) Stiffened bottom flange, (b) Composite bot. flange
and replaced by a 70 mm thick layer of concrete acting compositely with the steel; other girder
details remain the same The plotted values show a significant reduction in the maximum steel
stress from 147 2 N/mm2 to 108 3N/mm2 This reduction of about 27%, together with the amount
of steel saved, shows the effectiveness of the proposed method
Changing the number, flexibility, spacing and distribution of studs it is possible to control the
distribution of axial stresses in the steel flange and in the concrete layer. It is obvious that as the
stud stiffness increases the steel sheds more of its load into the concrete. On the other hand, to
achieve the buckling resistance the use of a reduced number of studs can be sufficient.
6. Conclusions
1 Studs of different sizes were tested until failure and the shear force - slip diagrams were
investigated The slip has a significant effect on the distribution of internal forces, if the
connection starts to fail. In those regions where slip occurs, the overall bending is transfered
more by local bending of the steel beam and of the concrete slab In the case of the simply
supported beam subjected to a uniformly distributed load, the extreme stress in steel beam need
not be at the midspan
2 The effect of deformability of connection concerning the redistribution of stress and the growth
of deflections, both due to short and long term loadings, is much more pronounced at flooring
composite beams than that at the bridge composite beams
3 Composite construction can be efficiently applied to achieve required buckling resistance and
Acknowledgement
Support provided by the Grant Agency of the Czech Republic for the projects No 103/95/1644 and
No 103/96/0017 is gratefully acknowledged
References
[1] Bazant, Z.P., Vitek, J.L.. Stud Connectors m Comoposite Beams: Simplified Failure Modlemg and Size
Effect m Fracture and Damage of Quasibrittle Structures, Bazant et al. eds E & FN Spon, London, 331-341
[2] Bazant, Z P Vitek, J.L Composite Beams with Deformable Connection Damage and Size Effect. Submitted
for publication, Journal of Struct. Eng ASCE.
[3] Evans, H R Kristek, V and Skaloud, M • Strengthening steel box girder bridges by controlling the effects of
shear lag, Proc.Instn Civ Engrs Structs and Bldgs, Paper 10593, Nov 1994, 104, pp. 483-488
259
Summary
The aim of this paper (which consists a part of the doctoral thesis of the second author) is to study
the collaboration (and especially the role of the embossments to the shear bonding) between the cold-
formed profiled sheeting and the concrete in a composite slab In order to attempt this aim, an
experimental program has been undertaken on shear bond pull-out tests and a numerical model, with
the help of a non-linear 3D finite element program, using large strains, has been developped to
simulate the shear bond between the two materials
1. Composite slabs
1.1 General
A composite slab consists of a cold-formed profiled steel sheet covered with a concrete (containing
or not reinforcement) Such slabs, which incorporate profiled steel sheeting as both permanent
formwork and tensile reinforcement of an in situ concrete slab and through deck welded shear studs
to provide composite beam action, are becoming a very popular form of flooring system in a lot of
countries This success is due to the significant constuction benefits occuring as well as provision of
an elegant structure
There are many types of profiled sheets used for the construction of composite slabs, which vary, for
exemple, in form, rib depth, by the methods of stiffenning the flat elements of the profil etc.
The bond between the concrete slab and the profiled sheet must be able to transmit longitudinal shear
at the steel-concrete interface This connection can be made, for exemple, by embossments, by the
reentrant shapes of the ribs creating a bond by friction or by other ways
260 STUDY OF SHEAR BOND IN STEEL COMPOSITE SLABS
1.2 Types of shear bonding between cold-formed steel sheeting and concrete
Three types of shear bonding between cold-formed steel sheeting and concrete have been observed:
(a) chemical bond (bond resulting from the adherence of the çement paste to the steel sheeting; such
bonds are broken under impact or repeated loading); (b) frictional resistance (resistance to applied
shear forces, proportional to the application of lateral force between the steel sheeting and the
concrete), (c) mechanical resistance (the physical interlocking of concrete and steel sheeting; such
physical interlock occurs at abrupt changes in geometry such as embossments)! 1],
Frictional and mechanical resistances are in reality due to the same phenomenon. They differ only in
scale: frictional resistance is related to microscopic variations in surface geometry i.e. roughness;
mechanical resistance is related to macroscopic geomettic differences i e. embossments! 1].
Learning upon test observations, three failure modes were established as follows: (a) vertical shear
(or punching failure near concentrated or line loads; this is associated with heavy wheel loads or short
span lengths), (b) shear debonding (resulting in the loss of composite action; loss of full composite
action is indicated by the initiation of end slip), (c)bending (due to plastification of sheeting or
crashing of concrete, normally, partial or complete plastification is observed after the loss of full
composite action for long spans)! 1]
2.1 General
The results of a pull-out test are resumed to a (P-u) curve (where P is the shear load and u is the
J. RONDAL, A. MOUTAFIDOU 261
average slip). The curves (P-u) for the Hibond 77 (M.S.M. Lagoratory) are presented in Fig. 3.
3 » WO S70
^
Enon
Fig. 2 Cross section and embossment pattern of Hibond 77 (all dimensions in mm)
1 00E+O4
9 00E+03
8.00E+03
7.00E+03
Z 6 00E+03
Û.
J 5.00E+03
J 4.00E+03
3.00E+03
2.00E+03
1.00E+03 P1HI7710
O.QOE+OO
0.00 0.50 1 00 1 50 2 00 2 50 3 00 3 50 4 00 4 50 5 00 5 50 6 00
Fig 3 Shear-slip (P-u) curves for Hibond 77 from the pull-out tests (M.S.M. Laboratory)
In this section, a finite element simulation of the pull-out test is presented in view to help us to
understand better the shear bond mechanism between the steel sheeting and the concrete. The aim is
to create a model, which -since we have proved its validation after a comparaison of its results with
the experimental ones- can be used in the future, not only for examining the behaviour of the
composite action (especially the shear bond mechanism) between the concrete and an existing at the
262 STUDY OF SHEAR BOND IN STEEL COMPOSITE SLABS
market steel sheeting in a composite slab, but also for optimizing an existing sheeting (by changing its
geometry, its thickness or a parameter of its embossments, such as their geometry, position, height))
or for creating a new one; so, since the model is reliable and accurate, it won't be necessary to
fabricate a new steel sheeting (each time we change a parameter) and to submit it to a pull-out test
for examining its behaviour (gain of money and time).
A non-linear 3D finite element program, using large strains, developped at the M.S.M. Departement,
has been the appropriate tool with which this objective was achieved. In this 3D analysis, the contact
between concrete and steel is assumed without friction <j> 0 (or with a very small Coulomb's
friction coefficient(j) 0.1 for numerical reasons); so, there is only one type of shear debonding
between the cold-formed steel sheeting and the concrete, i.e. the mechanical resistance due to the
embossments. In this way, one can insist more on the important role of the embossments which
consist one of the most important factors determining the behaviour of a composite slab. For our
numerical simulation, we used a profiled sheeting of type Hibond 77 (with thickness 0.88mm).
3.2 Discretization
Of course, the symmetry of the two profiled sheetings used in the test has been taken into account for
the finite element simulation. As we want to treat the case of an infinite number of embossments, only
a small part of the profiled steel sheeeting (that is to say a part of the sheet with one embossment)
has been discretized.
The contact surface (the interface between the concrete and the steel sheeting) is discretized by a type
of 3D mechanical contact element (with 4 or 8 nodes), connected with the upper face of the sheeting.
For the contact, we used a constitutive law for unilateral thermo-mechanical contact (a coupled
Coulomb 3D law) (adequate for a thermo-mechanical analysis of problems involving unilateral
contact between two bodies, Coulomb dry friction is used, the contact condition is enforced via a
penalty method or augmented Lagrangian method).
For these two discretizations, the same mesh was used (only the type of finite element for the steel
sheeting discretization and the number of nodes were different) In spite of the advantages of the
mixed elements of the second case (for our problem, they can give almost the same results as the 3D
shell elements [2], which consists a more logical approach for a sheeting of small thickness), they are
much more rigid from the brick elements of the first case, when, of course, the same mesh is used).
That's the reason, we tried to refine the mesh of the second case (by divising each element to 4!) and
we set about the third case
(3) so, the third discretization consists of 4x88=352 solid 3D mixed elements working on large
elastoplastic deformations (with 8 nodes); for this case, 352 contact elements (with 4 nodes) and 704
triangular segments were used.
In order that the procedure of the loading of the simulation was closer to the one of the pull-out test,
we imposed a uniform displacement in the z direction to all the nodes of the sheeting along the side
z=const.
For the steel sheeting discretization, the following laws were tried:
(i) an elastic constitutive law for solid elements at constant temperature (adequate for a mechanical
analysis of elastic isotropic solid undergoing large strains) and
(ti) an elastoplastic constitutive law for solid elements at constant temperature (adequate for a
mechanical analysis of elastoplastic isotropic element undergone large deformations; mixed or
isotropic hardening is assumed, when the volume elements (bricks with 20 nodes) or the solid 3D
mixed elements (with 8 nodes) are used respectively.
3.2.5 Results
In Fig. 4, the curves (R-u) (with R the reaction to the movement caused by the imposed
displacements) are presented, as this figure shows, we tried each of the cases of the steel sheeting
discretization for both elastic and elastoplastic laws
264 STUDY OF SHEAR BOND IN STEEL COMPOSITE SLABS
4. Conclusions
Since actually the last case of the numerical simulation (with the third try of the steel sheeting
discretization and an elastoplastic law) is not yet completed, general conclusions on the advantages
and disadvantages of the different discretizations can not be formulated
Fig. 4 Shear-slip (R-u) curves for Hibond 77 from the numerical simulation of the pull-out test.
References
[1] Daniels Byron J., "Shear bond pull-out tests for cold-formed-steel composite slabs", Rapport
d'essais, Ecole Polytechnique Fédérale de Lausanne, ICOM 194, Avril 1988.
[2] Kaiping Li, "Contribution to the finite element simulation of three-dimensional sheet metal
forming", Thèse de doctorat, Université de Liège, Faculté des Sciences Appliquées, 1995.
265
Summary
This report deals with strength and behaviour of composite slabs with three typical, but different
profiled steel sheet geometries used with and without end anchorage means (headed studs and
bent rib anchors). A brief survey of slab design up to a modified global plastic analysis is given.
Furthermore an improved partial shear connection method based on the results of small scale slip-
block tests is introduced In addition some results of slab tests with static and dynamic loadings
are presented.
1. Introduction
This paper deals with the behaviour and design of composite slabs. Composite slabs consist of
particular profiled metal deckings with concrete topping. They permit an easy and high speed
construction of steel framed or even concrete buildings, whereby the steel sheet acts during
erection among other things as safe working platform, permanent shuttering and horizontal
bracing. After the concrete has hardened the profiled steel sheet is part or all of the tensile
reinforcement. Their efficiency and advantages in combination with flexibility lead to an increasing
use of composite slabs. Therefore the development of many different shapes all over the world is
going on.
Numerous slab-, pull-out- and slip-block -tests with different types of composite slabs have also
been carried out in the laboratory for structural engineering of Kaiserslautern University,
Germany, in order to determine design values for their load carrying capacity and to investigate
their behaviour. Till now in Germany in most cases Holorib-type metal decking is used. The new
generation of these sheetings has embossments in the top flanges at least. But it shall be pointed
out that trapezoidal composite slabs show also a similar ductile behaviour and high resistance in
bending and longitudinal shear, if end anchorage means are applied. This end anchorage can be
provided by throughwelded headed studs, which are anyhow necessary for the composite beam
action. Objectives of our work and this paper are therefore
266 COMPOSITE SLABS WITH AND WITHOUT END ANCHORAGE
• profiles with re-entrant shape (dovetail ribs, Holorib-type) without embossments, but together
with end anchorage means, to create sufficient composite action,
• the Holorib-type sheeting with embossments (e.g. Coffastra 40, Haircol, Super Holorib,
SupeRip)
• and the trapezoidal profiles with embossments or idendations and with or without end
anchorage means.
Trapezoidal shapes lead to a consumption of steel per m2 flooring smaller than re-entrant profiles.
However the vertical separation of the two composite partners is not efficiently prevented and
therefore the longitudinal shear resistance is only about one tenth of the new generation of
reentrant profiles. The tu - values [1], [3] permit an easy comparison of the different metal decking
types, see table 1.
The load-deflection curves in figure 1 are test results, but they illustrate the typical behaviour of
these three different types. Cross-sections, test set up and descriptions are in agreement with
Eurocode 4 and quite similar for all curves: span length 3.0 m, shear span length Ls L/4 750
mm, depth of composite slab h, 140 mm and nominal sheet thickness tN 0.88 mm.
Re-entrant profiles of the original generation (Holorib, without embossments) show a brittle
failure mode. As soon as the pure shear bond is destroyed (increasing bending moment, cracking
of concrete in tension, shear stresses on both sides of these cracks exceed the pure bond strength,
end slip occurs) the load decreases suddenly and considerably. Due to clamping forces in the
shear span length the re-entrant geometry makes somewhat higher loads possible.
Composite slabs with trapezoidal profiles (e.g. Cofradal 60) show also a brittle failure mode. But
end anchorage means, in this case Cofradal 60 with throughwelded headed studs 0 3/4"
(19 mm), lead to a clear improvement of the slab behaviour up to failure, particulary with regard
H. BODE, F. MINAS 267
to bending resistance and ductility Almost no end slip occurs, and the sheet fails in load bearing
at the headed studs at the panel ends.
The re-entrant shapes with embossments (new generation, e.g. Super Holorib) show however a
very ductile
Load [kN] behaviour, after first
end slip has been
recorded, together
Super Holorib with bent rib anchors
with high load
carrying capacities.
In this case
significant slip occurs
at the steel-concrete
interface before the
maximum load is
reached (longitudinal
shear failure). As an
alternative, the ribs at
0 50 ]00 150 200 the panel ends can be
Midspan-Deflection w [mm] ^ent, t0 provide
Figure J:Load - deflection - curves for different profile shapes
another type of end
anchorage and to
with and without end anchorage
prevent the wet
concrete from flowing through the dovetail ribs.
A statistic evaluation of seven Cofradal 60 slab tests with throughwelded headed studs 0
3/4"
leads to the characteristic shear strength PpbRk 25.1 kN per stud. Eurocode 4 [1], 7.6.1.4
proposes for the load bearing capacity of headed studs corresponding to our test conditions the
conservative value PpbRk 19.6 kN. The evaluation of 23 earlier pull-out tests without concrete
yields a load bearing capacity of such headed studs 0 3/4" being about 15.7 kN. The
strengthening effect of the surrounding concrete results in about twice the value: from five pull-
out tests with concrete a characteristic value of 31.3 kN per stud can be deduced.
A good approximation for the strength of one bent rib anchor is the area of the anchored rib times
the yield strength of the steel sheet For the Holorib 51/0.88 shape this leads to PpbRk Ara ' fyp
(3.6 + 2 ' 5.1) 0.88 ' 32.0 38.9 kN.
•
The m+k - method - slightly modified from the original North Amercian version - is the standard
method for the longitudinal shear verification.
The partial connection method (tu - method) with incomplete interaction and considerable relative
displacements at the steel concrete interface is in accordance with the well known partial
connection design for composite beams with flexible connectors. Therefore it can be used to
design composite slabs with enough ductility even in case of horizontal shear failure. The same
mechanical model is being used for the test evaluation and for design purposes (fig. 2a, b). This
268 COMPOSITE SLABS WITH AND WITHOUT END ANCHORAGE
©
iq} }Lsj
0 £2 Oi 06 as 10
Figure 2a: Determination of the degree of shear Fig. 2b:Design with partial interaction
connection from slab test diagram
Mr. Ms (kNmJ
Mr
method leads to a very good agreement between
."Vr ^Ms(i 0 8)
theoretical solutions and test results. It is easy to
.Ms(n «0 2)
take into account additional (fire) reinforcement
and end anchorage measures. Even the
determination of the ultimate loads of
continuous composite slabs by means of a
modified global plastic analysis is possible [S],
[6], Figure 2 shows this design method applied
to a two span slab indcluding the critical cross
section for ri 0.486 (degree of shear
connection).
Fig. 2c:Design of continuous composite slabs
This shear strength tu consists mainly of the overriding resistance, but also of friction over
supports and clamping forces, if there are any. Line 1 in fig. 3 indicates the overestimation of
tu [kN/m'l longitudinal shear resistance for small Ls/h,
600 lt values. The filled round marks in fig. 3
indicate the t„ - values of 13 full scale slab
tests over the ratio shear span to depth
(Ls/hi). The tests with Super Holorib
sheets were carried out and evaluated as
shown in fig. 2 according to EC 4. Line 1
is a linear approximation throughout these
~
mean value evaluated 13 values. Results of the so called slip-
with friction
Tu -= 470 kN/m»
block tests [8], [13] can be used to
400 -r improve this method. Fig. 4 illustrates the
2,0 4,0 6,0 8,0 L./ht 10,0 test arrangement with a little composite
Figure 3:Relationship between t„ and the Lfht ratio
H. BODE, F. MINAS 269
between concrete and profiled steel sheeting, in order to evaluate the stress range Act in the steel
part.
Now we are carrying out coupon tests with varying stress ranges Act to derive values for further
improved S-N-curves.
6. Conclusions
Composite slabs with ductile horizontal shear failure - such as re-entrant profiles with
embossments or trapezoidal profiles with throughwelded headed studs as end anchorage means -
show a very high load carrying capacity and good servicability properties under static and
dynamic loading. This has been proved by means of composite slab tests. In addition a procedure
has been outlined to improve the partial connection method by means of so called slip-block tests.
This scientific work was financially supported by DFG (German Research Foundation). This
support is gratefully acknowledged.
References
[1] Eurocode 4: Design of Composite Steel and Concrete Structures - Part 1-1 General Rules
:
Specialty Conference on Cold-Formed Steel Structures St. Louis, Missouri, USA, 1988.
[10] Crisinel, M.; Fidler, M.J.; Daniels, B.: Flexural Tests on Composite Floors with Profiled
Steel Sheeting. École Polytechnique Fédéral de Lausanne, March 1986.
[11] Veljkovic, M.: Behaviour and Resistance of Composite Slabs. Experiments and FEM-
Analysis. Ph.D Thesis, Lulea University, 1996.
271
Summary
This paper deals with a test of a two-span continuous composite girder comprising sudden
changes of cross-sections and mixed cross-sections (class 1 at mid-span and class 4 on
internal support). Mainly the behaviour at serviceability limit state and ultimate limit state is
presented. Also comparisons are made with different types of global analyses and with
numerical simulation using a sophisticated software.
1. Introduction
In the frame of new concepts developed in Europe, particularly during the drafting of
Eurocode 4 - Part 2 for composite bridges [1], some discrepancies with the French regulation
[2] have appeared leading SETRA to order a two-span continuous composite beam test to
INSA in Rennes. The loading procedure of this test has included several phases related to :
- the behaviour at serviceability limit state (non accumulation of deformations due to
variable loads, cracking monitoring on the intermediate support) ;
- the resistance at ultimate limit state (maximum loads and possible redistribution of
bending moments).
In parallel with this experimental approach, a specialized software has been developed on the
basis of the finite element method specially adapted to the verification of composite bridge
beams. This software is only used here for the simulation of the present test ; but in the future,
it will be applied to the calibration of the various methods of analysis proposed in [1 ] covering
a wider range of span lengths, different ratios between consecutive spans...
2. Test presentation
As shown in fig. 1, the tested composite beam comprises two continuous spans of 7.5 m and
is subject to two independent concentrated loads P and Q each located at mid-span. This beam
is class 1 at mid-span and class 4 on support A2 (due to the class 4 web). The bottom flange
has a variable thickness (10 mm at mid-span, 15 mm on support A2) in order to simulate a
composite bridge beam where the maximum resistance bending moments are reached almost
simultaneously in sagging and hogging zones at ultimate limit state. Between the two loads,
the web is heavily transversely stiffened to avoid a premature shear buckling.
The mechanical properties of the structural steel, reinforcement and concrete have been
measured on several specimens leading to the characteristic values given in table 1.
272 TEST AND ANALYSES OF A BRIDGE CONTINUOUS COMPOSITE BEAM
Using these properties together with partial safety factors equal to 1, the mechanical
properties of the composite cross-sections are as follows :
108 mm4 =1166 kN m in sagging moment regions
i! 12.3 x Mpi Rd ;
1
1/ JM. je
|
'
1
I
i
* 1 1
k Ir
;
;
Support A3
Load bearing stiffener
Intermediate stiffener
i5~op^
180
a 25
thickness 6
11 80
L_
IM if
[
section on întermediatei Section at
support mid-span
The loading procedure comprises 123 phases. The main phases are summarized in table 2.
J.-M. ARIBERT, J. RAOUL, O. TERPEREAU 273
Figure 2 shows the variation of the beam deflection at the loading point P. Figure 3 shows the
variations of the bending moments at mid-span (loading point P) and over the intermediate
support (point A2). These moments have been derived from measurements of the bearing
reactions at supports Al and A3. They have also been checked against strain gauges measures
on the steel flanges near support A2.
1300 mm
Load P(kN)
1000 A
630 mm
800 "^230 mm ^
600
400
200 Deflection
0 (mm)
Support A2
0 50 100 150
(a) The first yielding appears during phase 40 for P 500 kN and Q 400 kN. A series of 5
cycles has been carried out around this loading (see table 2) and no cumulation of plastic
deformation has been observed. But for phases 61 to 66 (P varying from 400 to 550), there is
a more significant residual deflection (about 6% due essentially to the cracking of the slab
obtained after the first cycle). This result tends to confirm the choice of a limitation of stresses
of 1.0 fy in the structural steel proposed in EC4-2 [1].
b) The crack widths have been carefully monitored up to phase 40 (S.L.S.). The reinforcement
was designed according to EC4-1 [3] for a crack width of 0.3 mm. The distribution of the
crack widths on each side of the intermediate support is shown in fig. 4.
274 TEST AND ANALYSES OF A BRIDGE CONTINUOUS COMPOSITE BEAM
c) In fig. 3 related to the phases 93 to 122 (see table 2), the calculated bending moments at
serviceability limit state have been plotted :
value p|jexp) 890 kN.The high value of P^exp^ is likely to have been allowed on account of
redistribution of moments from mid-span to support A2. The points corresponding to the
above analyses are plotted in fig. 3. In the present investigation, the « uncracked » analysis
with a 10 % redistribution appears particularly appropriate whereas the « cracked » one tends
to underestimate the bending moment on support A2.
(b) The shear plastic or buckling resistance has been verified in accordance with EC4.1[3]
giving :
That leads to a maximum ratio Vsd / VR(j of 0.57 close to 0.50. Consequently, the reduction
of the calculated ultimate load Pu due to the vertical shear force is only 0.5 %.
(c) The connection was designed to resist the maximum possible shear flow at ultimate limit
state (32 and 38 headed studs were welded between Al and P and between P and A2
respectively, each having a shear resistance of about 100 kN).
(d) No lateral-torsional buckling has been observed during the test. But the calculation
according to Annex B of EC4.1 (clause B.1.2) [3] would give 0.92 and a reduction
factor
xlt °f about 0.6 what seems to a paradox.
be The very good behaviour of the tested
beam can only be explained by the presence of the high density of vertical stiffeners welded to
both flanges and to the web. By spreading and adding their stiffness to the flexural stiffness
k2 of the web (such a method is not in EC4.1), Xlt becomes 0.37 < 0.40, what leads to
Xlt
4. Numerical simulation
A numerical model based on the finite element method has been developed at INSA in Rennes
using specialized beam elements (for steel and concrete), shear connector elements and
buckling elements [4]. This model has been generalized for composite bridge girders by
introducing other possibilities such as gradual or sudden changes of cross-sections, local
buckling of web in class 4 (by means of the concept of effective depth), tension stiffening in
slab after the stage of stabilized crack formation, and creep of concrete allowing to take into
account the sequence of construction (this last aspect is not concerned in the paper) [5]. The
effect of tension stiffening has been formulated in accordance with Annex L of EC4.2 [1]
expressing the average tensile force Ns of the slab versus the average strain esm as shown in
fig. 5 (with factor ß 0.40).
N,
Load P+Q (kN)
1600
0 50 100 150
The composite beam tested above has been discretized using 186 finite beam elements and 68
shear connector elements. The simulated curve of the load versus the deflection (at the loading
point P) is compared with the experimental one in fig. 6, showing a good agreement provided
that a light additional deflection due to the vertical shear force in the web is taken into account
for the elastic phase. Moreover, the simulated variations of the bending moments at mid-span
and over support A2 are compared with the experimental ones (already given above) in fig. 7
276 TEST AND ANALYSES OF A BRIDGE CONTINUOUS COMPOSITE BEAM
and 8. But it is worth pointing out that the tension stiffening effect is only included in the
simulation of fig. 7 and not in that of fig. 8. The better agreement in fig. 8 may appear a
paradox though both simulations give the same ultimate load Pu. As explanation of such a
paradox, it is possible to advance the gradual vanishing of tension stiffening during the test
under the repeated cyclic loadings around the serviceability limit state. Also, some uncertainty
of the real effective width of the slab in hogging moment zone may affect the results of the
numerical simulation.
5. Conclusion
The tested beam, representative of a composite bridge beam, confirms that the proposed
methods to verify the serviceability limit state and the ultimate limit state in EC4-2 are safe
and in rather good agreement with the experience. The software used for the numerical
simulation gives results very close to the measured ones. It will be applied in the future to
calibrate the analyses proposed in EC4-2 and to precisely define their scope.
References
Summary
This paper is concerned with slenderness limits of the thin steel plate used in the construction of
concrete filled steel box columns. Requirements for limits on plate slenderness are outlined for
construction, service and ultimate loads. Construction loading with regard to axial forces and
hydrostatic pressure of wet concrete is considered and the time effects of concrete on the steel
plate is studied for service loads. Finally inelastic and post-local buckling under ultimate loads is
addressed and a set of experiments is used to calibrate a series of numerical models.
1. Introduction
The initial development of concrete filled steel columns saw the use of hot rolled steel sections for
these members, (Bridge1,Shakir-Khalil and Zegiche2). Hot rolled steel sections such as
rectangular, square or circular sections used in the past were typically of plate thickness suitable
to avoid local buckling when the columns were unfilled and hollow. The prevention of local
buckling is important as it allows the columns to be designed to take full account of the steel
strength. The relatively thick steel plate of hot rolled steel sections subsequently meant that for
building construction, concrete filled steel columns were not widely used because of the
exorbitant cost of using a steel section essentially as reinforcement. Their major use has been in
highly seismic environments such as Japan where the thick steel sections prolonged local buckling
and provided confinement to the concrete core. Furthermore the limitations of hot rolled steel
section sizes manufactured as hollow steel columns has limited the geometric size of columns
used which meant that for very tall buildings the sizes were unsuitable.
In Australia, a reassessment of concrete filled steel columns has been recently undertaken and
consultants and contractors have developed techniques which allow these columns to be
extremely competitive in terms of construction. Consulting Engineers, Ove Arup and Partners
and Connell Wagner have utilised very thin walled fabricated steel box or tubular columns
respectively to act as erection columns, (Watson and O'Brien3 and Bridge and Webb4). These
columns typically resist construction loads for several levels prior to the concrete being pumped
inside. The use of these columns in Australia has been mainly in tall buildings where thin steel
plate is utilised. This paper highlights the importance of each of the construction, service and
ultimate loading stages for the choice of the slenderness limit of the steel plate used in these
columns.
278 THIN-WALLED STEEL CONCRETE FILLED BOX COLUMNS
During construction, a concrete filled steel column is subjected to both axial force from the
constructed levels in addition to the hydrostatic pressure from the wet concrete which is pumped
inside. The economy of concrete filled steel box columns is influenced by the amount of steel and
therefore the steel plate thickness used. Generally box columns are designed with very thin steel
plate thus generating large slendemess limits. These steel boxes must be designed against yielding
and buckling under the imposed axial forces. Furthermore the effects of wet concrete inducing
hydrostatic pressure on the walls of the column needs to be considered so that the final
constructed geometry of the columns is unaltered by excessive lateral deflections.
Uy and Das5 studied the effects of the wet concrete in the construction stage of a concrete filled
steel box column using a folded plate finite element model. The presence of the axial load from
the constructed floors and the hydrostatic pressure from the wet concrete was incorporated in the
analysis. During this stage the main concern is to ensure that the deformations caused by the
imposed loads are minimised. The results of this analysis are summarised in the curves of Fig. 1
which shows the effects of the slendemess limit (b/t) and the number of levels being pumped, Ns
on the maximum deflection of the column centreline.
20 --
10 --
0--
1 3 5 7 9
Fig. 1. Allowable Slendemess Limits for Wet Concrete Pumping (L=4,000 mm; b=l,000 mm)
A further study was carried out by Uy and Das6 on the effects of intermediate bracing of concrete
filled thin walled steel box columns. This study was undertaken to determine a strategy during
construction to minimise lateral deflections without having to increase the steel plate thickness to
excessive values which would render the method uneconomical. The study considered various
bracing strategies while adopting slendemess limits which would be appropriate for ultimate
strength and service loading. The results of a typical analysis with a column of 1,000 mm width
and 4,000 mm height is shown in Fig. 2. The figure shows a deflection profile for a particular
pumping strategy for a braced and unbraced column, where H is the height above the base of the
column and 5 is the lateral deflection of the steel plate. It is worth comparing the results of Fig. 1
with those presented in Fig. 2 for a column with a plate slendemess of 40. It is shown that in
order to satisfy a maximum deflection of 5 mm, only two levels can be pumped without the
presence of bracing. However the concrete can be pumped to five levels when three intermediate
braces are used. This strategy is desirable for the construction of tall buildings as it greatly speeds
the rate of construction and minimises the amount of steel plate necessary.
B. UY 279
S (mm)
Uy and Das7 have developed an age adjusted effective modulus method which allows for
incremental loading and creep and shrinkage of the concrete for columns in a tall building. The
effects of creep and shrinkage determined from experimental research by Terrey et al8 and
Morino et. al9 have been used in the determination of the stress redistribution and the total axial
shortening of these columns. The analysis was carried out to consider a typical concrete filled
steel column for a 60 storey building using a fairly large section.
Total
Elastic
—I 1
t (days)
The results of the stresses and strains on a lower storey column were calculated as these would be
the most heavily loaded columns in a building. The effects of creep and shrinkage of concrete in a
concrete filled steel column will cause a redistribution of stress to the steel plate. This results in
the steel stress increasing by 45 % from 180 MPa to 260 MPa as illustrated in Fig. 3. The effects
of creep and shrinkage of the concrete on the slenderness limits are therefore significant and this
illustrates that the onset of local buckling may occur prior to ultimate loads being reached.
280 THIN-WALLED STEEL CONCRETE FILLED BOX COLUMNS
As part of a series of ultimate strength experiments on concrete filled steel box columns, several
columns were tested purely to determine the local and post-local buckling behaviour of the steel
plate by applying load to the steel box only. Table 1 outlines the dimensions and properties of
these experiments. These experiments were useful in the determination of the local buckling
stress in addition to the extension of a post-local buckling model originally developed by Uy and
Bradford10 for profiled steel sheeting. Table 1 also illustrates the yield stress ay and residual stress
ctr determined from tensile coupon tests and strain gauge measurements. The maximum load
achieved was recorded and the load at which buckling first occurred N0iis given. The
maximum load given by gross yielding is calculated as Ny. From these results the ratio of the
local buckling stress to yield stress is determined and the effective width of the steel plates is also
calculated. These will be used in comparison with the numerical models in the next section. A
typical failure of these specimens is illustrated in Fig. 4 for specimen NS5 after inelastic local
buckling has occurred showing the local buckle at mid height.
A finite strip method developed previously by Uy and Bradford10 has been recently augmented by
Uy11 to consider the local buckling of welded steel box columns incorporating residual stresses.
Results of this analysis are included in Fig. 5 which highlights the importance of the residual
stresses in the elastic range. The box column experiments are also compared with the finite strip
analysis in Fig. 5 and these suggest that the residual stresses of these columns are quite
substantial. It should be noted that the local buckling stress is difficult to determine from tests
and requires further investigation. The effective widths of the box columns are compared with the
semi-empirical model of Uy and Bradford10 and AS 410012 using a rational local buckling
coefficient. The results illustrate the influence of residual stresses as the box column experiments
allow less redistribution than profiled steel sheets. It is suggested that the AS410012 approach
should be used as it is shown to be conservative.
—cji=0.0
—— Cj!=0.1Cy
01=0.2<Ty
—*— Ot=0.3oy
Box Column Experiments
Fig. 5 Effect of Residual Stresses on Local Buckling Stress Oy=300 MPa; Pure Compression)
p=o0/a,
5. Conclusions
This paper has presented research on the construction, service and ultimate load behaviour of the
steel box in a concrete filled box column and has shown the importance of considering each of
these loading stages for the steel plate slenderness selection. Further research is necessary and
experimental work is currently being conducted into the construction, service and ultimate load
behaviour of these members which will provide further data on the effects of the concrete on the
steel plate.
6. Acknowledgements
The author would like to thank Messrs I.Bridge, S. Das, I.Laird, M. Lyon, I. Pratsas and R. Webb
of the Department of Civil and Mining Engineering for fabrication, construction and conducting
the experiments outlined in this paper. The financial support of the Australian Research Council
is acknowledged. Furthermore the author would like to thank BHP Slab and Plate Products
Division, Port Kembla for the supply of the steel plate used for the fabricated steel columns.
7. References
1. Bridge, R.Q. Concrete filled steel tubular columns, Civ. Engg. Trans., Instit of Engrs, Aust.,
18(2), 1976, pp.127-133.
2. Shakir-Khalil, H. and Zegiche, J. Experimental behaviour of concrete-filled rolled rectangular
hollow -section columns, The Struct. Engr, 67 (19), 1989, pp. 346-353.
3. Watson KB and O'Brien LJ. Tubular composite columns and their development in Australia.
The Institution of Engineers Australia, Struct. Engg. Conf., Adelaide, 1990, pp. 186-90.
4. Bridge, R.Q. and Webb, J. Thin walled circular concrete filled steel tubular columns, Comp.
Constr. in Steel and Concrete II, Proc. of an Engg Found. Conf., 1992, pp. 634-649.
5. Uy, B., and Das, S. Wet concrete loading of thin-walled steel box columns during construction
of a tall building, Submitted for publication.
6. Uy, B„ and Das, S. Bracing of thin walled steel box columns during pumping of wet concrete
in tall buildings, Submitted for publication.
7. Uy, B„ and Das, S. Time effects in concrete filled steel box columns in tall buildings, The
Struct. Des. of Tall Bldgs, 6, (1), 1997, pp. 1-22.
8. Terrey, P.J., Bradford, M.A., and Gilbert, R.I. Creep and shrinkage in concrete filled steel
tubes, Tubular Structures VI, Proc. of the Sixth Int. Symp.on Tubular Structs, 1994, pp. 293-298.
9. Morino, S. Kawaguchi, J. and Cao, Z.S. Creep behavior of concrete-filled steel tubular
members, Comp. Constr. in Steel and Concrete HI, Proc. of an Engg Found. Conf., 1996 (In
press).
10. Uy, B. and Bradford, M.A. Local buckling of thin steel plates in composite construction:
Experimental and theoretical study. Structs and Bldgs,Inst, of Civ Engrs, 110, 1995, pp. 426-440
11. Uy, B. Local and post-local buckling of concrete filled mild and high strength steel welded
box columns, Submitted for publication.
12. Standards Australia. AS 4100 Steel Structures, Australian Standard. Sydney, Australia, 1990.
283
Summary
This paper outlines the formulation, calibration and verification of a three-dimensional, cyclic,
distributed plasticity finite element model for composite frames composed of square or
rectangular concrete-filled steel tube (CFT) beam-columns and steel wide-flange beams. The CFT
model accounts for slip between the steel and concrete, and a brief study is presented which
demonstrates that slip often has little effect on the global response of composite frames, although
it may affect straining at beam-to-column connections.
1. Introduction
Research to date on CFTs has included tests on circular and rectangular columns, beams, beam-
columns, and composite frames and trusses (Gourley et al. 1995), push-out tests to determine the
bond strength at the steel-concrete interface, and connection studies to determine moment-
rotation relationships and slip characteristics at connections (Hajjar et al. 1997a). However,
relatively few computational models have been developed for analysis of composite frames
composed of steel wide-flange girders framing into CFT beam-columns with fully-restrained or
partially-restrained connections (termed "composite CFT frames"). This paper outlines the
formulation of a finite element model for CFT beam-columns. This computational model is
capable of accounting comprehensively for material nonlinear behavior at the stress-strain level,
and for all significant geometrically nonlinear behavior. The resulting formulation is able to
analyze complete three-dimensional composite CFT frames. This formulation has been verified
against a large number of experimental results, and it may be used to assess the static and cyclic
seismic behavior of individual CFT beam-columns, composite CFT subassemblages, or complete
composite CFT frames. It is also suitable for conducting parametric studies to determine the
significant factors which affect CFT behavior, including cross section geometry, constitutive
behavior, slip characteristics, and end restraint, with the goal of developing improved design code
recommendations.
284 NONLINEAR ANALYSIS OF COMPOSITE CONCRETE-FILLED STEEL TUBE FRAMES
The current formulation models slip between the steel tube and concrete core of the CFT element
along the element centroidal axis. The finite element model for slip follows the work of Amadio
and Fragiacomo (1993), who developed a two-dimensional composite beam finite element which
accounted for slip between an elastic steel girder and the concrete slab for monotonie loading.
For the current CFT finite element formulation, three extra translational DOFs are added at each
element end to permit the steel tube and concrete core to translate separately for a CFT which is
arbitrarily oriented in space during a geometrically nonlinear analysis. The transverse
displacements of the steel and concrete are constrained to be identical using penalty functions,
thus allowing slip only along the longitudinal axis of the CFT, due either to flexural or axial
loading. The separate axial DOFs are coupled through a layer of nonlinear slip springs along the
material interface. Slip resistance is provided by friction and adhesion at this interface. The
resulting computational model uses a bilinear load-slip relationship, with an initial stiffness (kslip)
up to the bond strength of the interface (fi,onij), followed by a zero stiffness. This relation is
representative of load-slip data seen in experimental studies of CFTs subjected to slip.
The current CFT finite element formulation extends the work of Amadio and Fragiacomo (1993)
by including the significant geometrically and materially nonlinear effects which may be expected
in CFT members subjected to cyclic loading. The formulation includes low and high-order
geometrically nonlinear stiffness matrices to account for P-A and P-S effects, and an updated
Lagrangian incremental/iterative formulation to evaluate the stiffness matrices during the analysis.
Lateral-torsional buckling, flexural-torsional buckling, and the bowing effect are not modeled.
Additional details of the CFT finite element formulation can be found in Hajjar et al. (1997a).
Nonlinear material constitutive models are used to describe the stress-strain behavior of individual
steel and concrete fibers at the member end cross sections. Both constitutive formulations are
intended to model the full range of material stress-strain behavior commonly seen in CFT
members, including capability to model repeated strain reversals under cyclic seismic loading.
The steel constitutive model used in the current formulation for structural steel wide-flange beams
and for steel tubing is adapted from Shen et al. (1995). This bounding surface formulation
P.H. SCHILLER, J.F. HAJJAR, A. MOLODAN 285
permits accurate prediction of the material tangent modulus during cyclic loading The
formulation was modified for modeling cold-formed steel tubing by including the variation in
material properties within the tube cross section, and accounting for the gradual rounding of the
stress-strain curve due to plastic straining during cold-working.
The concrete constitutive model is adapted from Ameur-Moussa and Buyukozturk (1990),
supplemented by an elastic tensile branch prior to cracking The compressive cyclic model is a
combined plasticity-damage model with a vanished elastic zone. The scalar damage parameter
accounts for micro-cracking in compression The compressive model exhibits all important
concrete behavior including stiffness degradation, hysteretic loading-unloading loops, and a
descending post-failure branch The tensile model includes tensile cracking and crack opening
and closing upon repeated loading The slope of the post-failure branch of the stress-strain curve
was adjusted to match experimental moment-curvature-thrust results from monotonically loaded
short CFT specimens (Tomn and Sakino 1979) This variation m the descending branch of the
compressive stress-strain curve accounts for the added ductility of the concrete due to
confinement of the concrete core by the rectangular steel tube Calibration and verification of the
steel and concrete constitutive models are described m Hajjar et al (1997b)
The current formulation has been verified against both monotonie and cyclic CFT experiments.
The verification is presented in Hajjar et al 1997a, b) Comparisons of two cyclic experiments
with computational results are presented below together with the expenment schematics Figure
1 shows
a planar CFT specimen under constant axial load (P/P0= 0 2, where P0 is the CFT axial
strength) and cyclic shear loading (Sakino and Tomn 1981). Figure 2 shows a three-dimensional
cruciform subassembly composed of steel wide-flange beams framing rigidly into a rectangular
CFT column, with constant axial (P/P0= 0.15) and out-of-plane beam loading, and alternating
antisymmetric in-plane beam loading to simulate seismic loads (Monno et al. 1993).
286 NONLINEAR ANALYSIS OF COMPOSITE CONCRETE-FILLED STEEL TUBE FRAMES
Tube: 100x100x4.21
fy 295 MPa
fc 24.0 MPa
CFT Beam-Column
Qf: 4Q
Steel Beams
Only limited studies have been conducted on the effect that slip has on the load-deformation
behavior and ultimate strength of CFT members (e.g., Dunberry et al. 1987; Shakir-Khalil 1994).
The AISC LRFD Specification (1993) limits the ultimate flexural capacity of CFTs to the plastic
moment capacity of the steel section alone, partially due to the lack of data on slip. This
limitation reduces the economy of CFTs as the primary lateral load resisting elements in unbraced
frames. The formulation in this work may be used to provide supplemental data for determining
the effect of slip on CFT behavior.
A preliminary study of the effect of slip on CFT flexural capacity and moment-curvature behavior
was conducted for three CFT cross sections with different tube width-to-thickness ratios by
varying both fbond and kshp from the calibrated parameters. Reducing k,by
two orders of
magnitude slightly reduced the rigidity of these members, but had no perceptible effect on the
ultimate moment capacity. Reducingto 0.1 MPa [e.g., a value in the range of that
P.H. SCHILLER. J.F. HAJJAR, A. MOLODAN 287
recommended by All (1980)] reduced the computed moment capacity of the CFTs by 3% to 5%,
suggesting that bond strength may have a small effect on the moment capacity of CFTs.
A second study was conducted to determine the effect of slip on frame behavior. Figure 3 shows
a 4-story 4-bay frame subjected to factored gravity and wind loading. The loads are increased
proportionally until the collapse of the frame is detected. Four cases were investigated to
determine the effect of boundary conditions and slip parameters on the behavior of the frame. In
Case I, the steel girder is assumed to engage the concrete directly through the connection, and no
slip is permitted at the joints. In the other cases, the steel girder is assumed to engage only the
tube, and load is transferred to the concrete through the slip interface. No perceptible difference
in the global load-deformation response of the structure was observed between the four cases. An
investigation of the interface stresses for Cases n, III and IV indicates that the values of both fi,ond
and klhp affect whether bond strength is surpassed in a CFT frame. Bond loss is not detected for
Cases II and HI, although values close to 0.6 MPa are seen in Case n. However, Case m has
interface stress values which are approximately 35% of the values for Case n, because for Case
in the transfer of load at the connections occurs over a larger length of the CFT due to the lower
stiffness. For Case IV, the bond strength is breached in several locations (first breach occurs at
19% of the design loading) resulting in localized slip in the connection regions that is
approximately five times that seen in Case II (although the slip magnitudes remain small).
19 9_k£Jj
V/2
L L L L V/2 Beams
W410x60, /, 248 2 MPa
I er 1Sections:
® V/2
L L ®y 2 V/2
<s>
y ®TS 177 8x177 8x6.35
37.9 kl^ ;
6. Conclusions
The computational formulation presented in this paper is suitable for conducting parametric
studies of individual CFT members and complete composite CFT frames. This formulation may
also be used to determine the effect of slip stiffness and bond loss on rigidity, ultimate strength,
ductility, and monotonie or cyclic behavior of composite CFT frames. A preliminary study
demonstrates that the ultimate moment capacity of flexural specimens shows only slight change by
lowering the bond strength. Variations in slip stiffness and bond strength are also seen to have
virtually no effect on the global response of a CFT composite unbraced frame, although the slip
parameters do have a small effect on the CFT behavior specifically in the connection region.
288 NONLINEAR ANALYSIS OF COMPOSITE CONCRETE-FILLED STEEL TUBE FRAMES
Acknowledgments
The authors would like to thank Prof. S. Morino, Mie Univ., Prof. T. Usami, Nagoya Univ., Prof.
O. Buyukozturk, MIT, and K. Fetterer, Univ.of Minnesota, for providing information relevant to
this research. This research was supported by the National Science Foundation (Grant No. CMS-
9410473), and by the Univ. of Minnesota Dept. of Civil Engineering. The authors gratefully
acknowledge this support. Any opinions, findings, and conclusions or recommendations
expressed in this material are those of the authors and do not necessarily reflect the views of NSF.
References
American Institute of Steel Construction, Inc. (AISC) (1993). "Load and Resistance Factor
Design Specification for Structural Steel Buildings." Second Edition, AISC, Chicago, IL.
Ameur-Moussa, R. and Buyukozturk, O. (1990). "A Bounding Surface Model for Concrete."
Nue. Engrg. Des., 121, 113-125.
Architectural Institute of Japan 1980). AIJ Standard for Structural Calculation of Mixed
Tubular Steel-Concrete Composite Structures. AU, Tokyo, Japan.
BS 5400 (1979). Steel, Concrete, and Composite Bridges: Part 5: Code of Practice for Design
of Composite Bridges. London, British Standards Institution.
Dunberry, E., LeBlanc, D., and Redwood, R. G. (1987). "Cross-Section Strength of Concrete-
Filled HSS Columns at Simple Beam Connections." Can. J. Civil Engrg., 14, 408-417.
Gourley, B. C., Hajjar, J. F., and Schiller, P. H. (1995). "A Synopsis of Studies of the Monotonie
and Cyclic Behavior of Concrete-Filled Steel Tube Beam-Columns." Rep. ST-93-5.2, Dept.
of Civil Engrg., U. of MN, Minneapolis, MN.
Hajjar, J. F., Schiller, P. H„ and Molodan, A. (1997a). "A Distributed Plasticity Model for
Concrete-Filled Steel Tube Beam-Columns with Interlayer Slip." Engrg. Struc., sub. for publ.
Hajjar, J. F., Molodan, A., and Schiller, P. H. (1997b). "A Distributed Plasticity Model for Cyclic
Analysis of CFT Beam-Columns and Composite Frames." Engrg. Struc., Spec. Vol., at press.
Morino, S., Kawaguchi, J„ Yasuzaki, C., and Kanazawa, S. (1993). "Behavior of Concrete-Filled
Steel Tubular Three-Dimensional Subassemblages." Comp. Const, in Stl. and Cone. II,
Easterling, W. S. and Roddis, W. M. K. (eds.), Engrg. Found., ASCE, NY, 726-741.
Sakino, K. and Tomii, M. (1981). "Hysteretic Behavior of Concrete Filled Square Steel Tubular
Beam-Columns Failed in Flexure." Trans., Japan Cone. Inst., 3, 439-446.
Shakir-Khalil, H. (1994). "Beam Connection to Concrete-Filled Tubes." Tubular Structures VI,
Grundy, P., Holgate, A., and Wong, W. (eds.), A. A. Balkema, Rotterdam, 357-364.
Shen, C., Mamaghani, I. H. P., Mizuno, E., and Usami, T. (1995). "Cyclic Behavior of Structural
Steels. H: Theory." J. Eng. Mech., ASCE, 121(11), 1165-1172.
Tomii, M. and Sakino, K. (1979). "Elasto-Plastic Behavior of Concrete-Filled Square Steel
Tubular Beam-Columns." Trans., Arch. Inst. Japan, 280, 111-120.
White, D. W. (1985). "Material and Geometric Nonlinear Analysis of Local Planar Behavior in
Steel Frames Using Interactive Computer Graphics." M.S. th., Sch. of Civil and Envim.
Engrg., Cornell Univ., Ithaca, NY, August.
289
Summary
A frequently applied roof form covering industrial shops is a structure which consists of thin-
walled reinforced concrete precast slabs laid on steel truss girders. The structural continuity is
assured by means of reinforced concrete curbs cast-in-place on the truss top flanges. A statical
analysis of such a structure considered as a composite one had been performed. This model
represents quite well the real work of the structure. The calculation results have been positively
veryfied by means of experimental research.
1. Introduction
A blocked, five bay production shop of a copper ore enrichment plant has its dimensions in plan
225.6 x 135.0 m. The cross-section of the production building is shown in Fig. 1, while the
general view of the researched truss girders is presented in Fig. 4. The bays have their heights
respectively 18, 25, 34, 25 and 18 m. Each bay is fitted out with two overhead cranes. The
heaviest overhead crane has its bearing capacity of 160 T. The main load carrying structure
consists of steel and reinforced concrete frames spaced every 6.0 m. The necessary rigidity in the
plan of frames is provided by the middle bay, constructed in the form of reinforced concrete,
monolithic bunkers for crumbled ore. Arrangement of longitudinal and transversal walls of these
bunkers can be considered as a solidly reinforced concrete shaft.
Simply-supported steel truss girders of 30, 30, 15, 30 and 30 m span and laid on them
prefabricated thin-walled reinforced concrete slabs, of 6 m span, are used as structural elements of
the roof. These slabs have been manufactured in Poland for 45 years as typical elements of
industrial shop roofs. Partial corrosion of top flanges of roof girders has been observed after
22 years of exploitation. The object should be still exploited for at least 30 years. There exists
consequently a necessity to qualify the bearing capacity of roof girders.
Bearing capacity had been qualified by means of: 1) a standard calculation procedure as for flat
lattice elements, taking into consideration reduction of top flange cross-section because of
corrosion; 2) taking into consideration a composite model of a truss girder top flange. The
calculated bearing capacity of truss girders is greater than the acting load.
290 VERIFICATION OF BEARING CAPACITY OF COMPOSITE TRUSS GIRDERS
Fig. 3 Composite section of a truss girder top flange: (1) - steel top flange of a girder,
(2) - batten plates connecting branches of a girder flange, (3) - R/C curb,
(4) - precast thin-walled reinforced concrete slabs 30x150x587 cm
Connectors between steel and concrete are not necessary, if tangential stress in connection plane
t
is less than 0.6 MPa. In our case tangential stress caused by:
- temperature difference,
- shearing forces and bending at intervals between nodes,
- concrete shrinkage
is x < 0.8 MPa. It signifies, that batten plates (2) - connectors between steel and concrete are not
loaded. Bearing capacity Npl of a composite section, according to PN-91/B-03302, is given by:
Npl Fs'Rs + 0.85FbRb + FaRa,
where: Fs, Fb, Fa - cross-section areas of steel, concrete and reinforcement; Rs, Rb, Ra - strength
of steel, concrete and reinforcement. This equation was used while determining the state of stress
and strain in the examined composite section, as part of a truss girder.
4 4
Phases of truss girder work are illustrated in Fig. 5, that presents the relation between load and
displacement: P P(f). A diagram presenting P P(o) would be very similar. Introduction
of an equivalent stiffiiess of truss girder would be useful for interpretation of information
presented in Fig. 5. In phase I, that is to the load level corresponding to load Pm existing while
assembling the structure, and the top girder works as a "pure" steel truss. The equivalent stiffiiess
(EJ)i is proportional to tg ai. During phase II, it is above the load level Pm existing while
assembling, the girder works as a composite steel-concrete element. The equivalent stiffiiess
292 VERIFICATION OF BEARING CAPACITY OF COMPOSITE TRUSS GIRDERS
(EJ)2 of the composite section is proportional to tg CC2. It is worth stating that (EJ)2 > (EJ)i. In
the examined problem (EJ)2 1.16 (EJ)i. It indicates that the girder equivalent stiffhess after
assembling increased about 16%.
To verify experimentally the presented calculation model, the deflection measurement of 72 truss
girders were carried out. Deflection measurements were realized for real state of load Pa- The
obtained mean deflection value of truss middle joint (point G in Fig. 5) precisely corresponds to
the expected result delivered for a composite structure working in phase H.
Fig. 6 Histogram and diagram of normal distribution of the geodetic measurement of roof
girders ' displacement
Z. MARCINKOWSKI, P. BERKOWSKI 293
- phase I - the load level corresponding to load Pm existing while assembling - as a "pure"
steel truss;
- phase II - above the load level Pm - as a steel-concrete composite structure.
Measurement of 72 truss girder deflection was carried out. The synthesis of measurement results
is presented in Fig. 6 as a histogram.
Tab. 1 Specification offorce and stress in a composite steel-concrete top flange of a roof
girder
Member Member Assembling Total Equivalent Load in phase II
number area loadPm loadP area (composite section)
-Fig. 3 force stress force stress of composite force stress
FS NSI in steel Ns in steel section NsII Nbll in steel in
concrete
Ea 2.05-105 MPa
Taking into consideration the co-operation between concrete and steel in phase II, that is above
the level of assembling load Pm - the stress in steel section, e. c. for the member 13, are reduced
from 150.4 to 113.0 MPa, that is nearly 25%. These reduced stresses are "consumed" by the
concrete part of the top flange, where the stress in concrete attains a maximum value of 6.2 MPa.
These reduced stresses in steel section in the composite structural model are accompanied by the
adequate lower displacement. Deflection charts for considered load levels are shown in Fig. 7.
294 VERIFICATION OF BEARING CAPACITY OF COMPOSITE TRUSS GIRDERS
Taking into consideration the composite structural model, a perfect consistency between the
calculation results and geodetic measurements was obtained.
Joint number
Fig. 7 Calculated and measured displacements of roofgirders for considered load levels
In Fig. 7 the points G and D(Pa) coincide, and in Fig. 5, under the real load level Pa, the mean
value of the displacement - point G - is obtained.
A classical structural solution of precast thin-walled reinforced concrete slabs supported on upper
flange of steel truss girders is the placing of a concrete curb in phase I of construction work. In
this phase, i. e. till the load reaches the value of the so-called "assembling" load Pm, the girder
works as a "pure" steel truss. In phase II, the concrete curb after hardening and mechanical
connection to the girders' top flange, starts work in carrying the load. Girder top flange then
works as a composite section, delivering a 16% capacity reserve. The "two-phase" calculation
model was positively verified by the geodetic measurements of girder displacement.
The presented example of experimental verification of steel roof girders justifies the formulation
of a postulate to introduce the notion of "experimental bearing capacity" into the building codes,
so that the research results on deflections and stresses would be equally legalized with the results
of static and strength calculations, especially in the case of making a diagnosis of existing structure
technical status.
295
Summary
This paper reviews the results of a series of tests that focus on the behavior and strength of
composite truss girders. Three laboratory size floors were constructed and loaded to failure Top
chord supported open web trusses were used as secondary members Three unique methods of
generating composite action in truss girders are examined
1. Introduction
As the use of composite long-span trusses has increased, interest in developing better and more
efficient applications of this type of system has also increased Historically, composite floor design
has been applied only to the trusses directly supporting the composite slab The girders supporting
these trusses are usually either non-composite hot-rolled universal beams or non-composite truss
girders. As a result, composite truss research has focused only on those tmsses which directly
support the deck sheeting and concrete
With the increased use of composite tmsses, it is of interest to study the behavior of supporting
composite trass girders A primary consideration in creating a floor system with composite trass
girders is the method used to frame the trasses into the trass girders so that the shear studs can be
properly attached to the truss girders Three separate methods of framing were selected for study
Each of these methods is presented and the test results reviewed in this paper
2. Research Program
A series of three composite floor systems were recently constructed and tested at Virginia
Polytechnic Institute and State University. Destructive testing was performed to determine the
strength of the girders and observe their behavior within floor systems These systems included the
influence of trass connection behavior on the girders and thus the girder behavior in an actual
structure was better approximated
296 BEHAVIOR OF COMPOSITE TRUSS GIRDERS
Each test setup used three full-size girders, spanning 9.1m and spaced at 2.1 m on center. The
setups consisted of an interior truss girder (IG) and two exterior truss girders (EGL, EGR). For
the third setup only, a universal beam (EB) was substituted for one of the exterior truss girders to
compare the composite action of the two girder types. Secondary trusses, which would generally
span from 9 m to 20 m in an actual structure, were shortened for testing to about 2 m and placed
between the girders at the third points. Steel sheeting was placed perpendicular to the secondary
trusses and parallel to the truss girders. Figure 1 shows the typical framing plan.
Headed shear studs were used as the shear connectors; the secondary trusses all had double rows
of studs placed in their strong (favorable) positions (Easterling et al 1993). One layer of woven
wire fabric was used throughout the entire slab of each setup for temperature and shrinkage
effects. Transverse reinforcement was designed separately for each girder in each test setup.
The concrete slabs were all constructed of 51 mm steel deck with 76 mm of concrete cover, for a
total depth of 127 mm. The girder ends were simply supported by rollers resting on load cells,
which were used to measure the girder end reactions.
The systems were loaded with two hydraulic rams which were anchored in load frames and
positioned over the secondary trusses. Each ram applied load to a spreader beam which divided
the load into two concentrated loads on each secondary truss line. These loads were transmitted
by the trusses to the girder third points; the girders then carried the loads to their end supports.
The concentrated load points, as well as the rams, could be moved to create different loading
situations.
The first setup was a flush framed configuration. The second setup involved the use of stub
girders, while the third employed concrete haunches over the girders, as shown in Figure 2. The
systems were all tested to observe their composite behavior. Several load applications were
performed on each floor. The first application of each test was to load the exterior girders to
their predicted service loads and then unload. The load points over the trusses were then
repositioned to principally load the interior girder. The second application loaded the interior
girder to its predicted service load. The third application loaded the interior girder to 1.67 x
service loading, which is the load at which the girder bottom chord would theoretically yield.
S.L. SHOWALTER, W.S. EASTERLING, M. RAMBO-RODDENBERRY, T.M. MURRAY 297
The load points were then shifted back to their original positions so that the exterior girders
could be loaded to 1.67 x service loading. Further cycles would then be performed with the load
points directly over individual girders, as necessary, to load them to failure. During these tests,
measurements were taken of strains at various locations along the girders, strains on the surface
of the concrete slab, vertical and horizontal displacements of the girders, and slips between the
concrete slab and the girder top chords. Detailed descriptions of each of these tests are found in
the project reports (Kigudde et al 1996; Showalter et al 1997a, 1997b).
For this setup, the secondary trusses were connected to the truss girders with flush-framed,
bolted connections. This allowed the top chords of the trusses to be positioned at the same
elevation as the top chords of the girders. The steel sheeting could then be placed directly on the
top chords of both the girders and the trusses. Shear studs were spaced along the girder lengths
to provide sufficient shear resistance to develop the truss girder bottom chord nominal yield
stress values. Three-bolt (on EGL) and four-bolt (on EGR) single plate (fin plate) framing
connections were used for evaluation purposes.
Eleven different loadings were placed on the system. Symmetric loading patterns were first used
to load the girders to their predicted service and first yield strengths. A series of tests were then
conducted using unsymmetric loading. Additional symmetrical loads were placed on the system
to establish non-linear behavior in the girders, followed by tests to failure of the individual
girders.
The exterior girder, EGL, carried a maximum total load (structure dead load + applied load) of
783 kN. The bottom chord was beginning to yield when the girder failed abruptly as a
compression web member buckled during a subsequent load application. EGR carried a
maximum total load of 776 kN and later failed when a web member buckled. The bottom chord
had not reached yield when failure occurred. IG carried a maximum total load of 1535 kN, with
definite yielding of the bottom chord. The girder failed completely when compression web
members buckled.
The second setup utilized a stub girder arrangement, similar to the concept described by Chien
and Ritchie (1984), to form the composite truss girders. The secondary truss top chords were
seated directly on the girder top chords at the girder span third points. A series of three shallow
universal steel beams, or stubs, each approximately 3 m in length, were welded along the length
of each girder. The stubs were separated by 200 mm gaps where the secondary truss top chord
seats framed in. These stubs filled the 127 mm gaps between the girder top chords and the truss
top chords to provide a supporting surface for the steel sheeting. A single row of shear studs was
placed directly over the girder centerlines along the girder lengths.
A total of seven loadings were placed on the system. The girders were loaded to their predicted
service and first yield strengths, and then the individual girders were tested to failure. Failure in
the exterior girder, EGL, occurred in part because of a loss of shear connection at one end of the
slab when longitudinal cracking occurred directly along the line of shear studs. Yielding of the
bottom chord occurred beneath the third point near the end where the cracks formed. The
maximum total load was 670 kN. Similarly, the shear connection on EGR failed when severe
298 BEHAVIOR OF COMPOSITE TRUSS GIRDERS
longitudinal cracking occurred along the line of studs on the girder at the opposite end of the
setup. It carried a maximum total load of 636 kN. The interior girder, IG, carried a maximum
load of 1346 kN and failed as a result of concrete crushing beneath the load points directly over
the girder as well as by yielding of the bottom chord at the span third points.
Failure was also caused in part by the formation of plastic hinges in the top chords where the
cross-sections were reduced because of the gaps between the stub beams. This introduced higher
stresses in the bottom chords at the third points than at midspan, where the girders showed no
signs of yielding. The rotation of the slab, occurring as the interior girder deflected, caused
several interesting results. Transverse reinforcement had been designed over the girders to resist
longitudinal shear. This reinforcement was not sufficient, however, to carry transverse tensile
forces in the slab as it rotated. Because of this, the slab cracked longitudinally along the row of
studs, causing a loss of shear connection. In addition, as the slab rotated, the secondary trusses
rotated as well. The truss seats, being held in place on the girders with welds, caused significant
local buckling of the top chord horizontal legs, especially on the exterior girders. This effect
would almost certainly be magnified by more typical length trusses.
The third setup used concrete haunches over the girders. As a result, the concrete over the
girders was twice the thickness of the normal slab and much longer studs were required. These
studs were placed in double rows along the lengths of the girders. The haunches were formed to
encase the truss seats. This meant that the seats were cast into the haunch concrete and could be
considered additional shear connectors.
This setup also included a universal beam in place of one of the exterior truss girders. The two
secondary trusses framing into the beam used four-bolt single plate connections to provide a
flush frame on that side of the floor, eliminating the need for a haunch over the beam. A single
row of studs were welded directly onto the top flange over the beam centerline. The beam was
designed to carry a comparable moment to that of the opposite exterior truss girder (Load 1993).
The specimen was subjected to a total of six loadings. The girders were loaded to their predicted
service and first yield strengths and then the individual girders were loaded to failure. As in the
first setup, no problems were encountered with the strengths of the shear studs on the truss
girders. Failure of the beam was a result of a loss of shear connection. It carried a maximum
load of 617 kN. The truss girders failed due to yielding of the bottom chords at the span third
points. EGL carried a maximum load of 796 kN while IG carried 1845 kN.
The ultimate flexural strength of each girder was calculated for the controlling design limit state
of bottom chord yielding using measured cross-section and material properties. The ratio of the
experimental ultimate moment to calculated ultimate moment, M,/Mc, for each test is shown in
Table 1 and varied from 0.75 to 1.27.
S.L. SHOWALTER, W.S. EASTERLING, M. RAMBO-RODDENBERRY, T.M. MURRAY 299
The values of experimental moment of inertia for the specimens were determined from the slope
of load vs. deflection curves for the initial loading up to service load. The calculated values were
produced by a method incorporating adjustment factors for slip and for the span/depth ratios of
the truss girders. The ratios of experimental moment of inertia to calculated effective moment of
inertia indicate that the stiffnesses of the flush framed girders and stub girders were accurately
predicted. The haunched girders, on the other hand, were much stiffer than predicted.
The stub girder configuration involved the least labor in terms of shop fabrication of the trusses
and truss girders and field erection, but had the poorest performance in terms of strength. The
flush framed setup displayed better performance, but it also had greater fabrication demands due
to the flush framed connections. The haunched girder setup had the best strength performance,
but the field erection was more labor intensive and the slab the most heavily reinforced.
4. Acknowledgments
The research reported on in this paper was sponsored by Nucor Research and Development and
Vulcraft, both divisions of Nucor Corporation. Additionally, thanks go to David Samuelson and
Wayne Studebaker of Nucor Research and Development for the technical assistance they have
provided throughout the tests.
5. Notation
6. References
Load and Resistance Factor Design Specification for Structural Steel Buildings. (1993).
American Institute of Steel Construction. Chicago, Illinois.
Summary
The authors examine composite steel-concrete beams prestressed by external cable, in order to analyze
the influence of the most common construction and prestressing sequences, by evaluating both the initial
linear behavior under service load and the nonlinear behavior under increasing load up to failure. The
analysis is based on a model, described in previous works, which considers the cable slipping at saddle
points and nonlinear constitutive laws for materials composing the structure. In this paper the analysis is
extended for considering different construction sequences in a unitary manner and some numerical
results on simply supported and continuous two-span girders are reported.
1. Introduction
External prestressing, often used in brigde construction and strengthening, usually employes cables
disposed parallel to the steel beam web and anchored at ends. Cable profiles are defined by saddle points
at which they can slip with negligible friction.
Generally, stretched cables permit increasing the global load-carrying capacity, however some other
advantages derive from their presence. At service conditions maximum deflection can be controlled and
notably reduced. In continuous beams concrete deck cracking at intermediate supports can be prevented
under a wide range of external loads. The cables undergo small stress increments under service load
thanks to the redistribution of traction due to slipping and this determines a further ductility resource at
failure.
Substantially different construction and prestressing sequences are adopted in practice [1,2], leading to
different collapse modalities and different stress and strain distributions in materials (concrete, beam
steel, reinforcement steel and high-strength cable steel). Such differences have not been sufficiently
investigated in the past and this paper intends to show the different behaviors under increasing external
loads up to failure for the following three cases.
Case A. The concrete deck is cast on the steel beam supported by a large number of additional supports
(propped beam). After the concrete hardening the additional supports are removed and prestressing is
applied.
Case B. The concrete deck is cast on the steel beam supported by the final constraints only (unpropped
beam). After concrete hardening prestressing is applied.
Case C. Prestressing is applied to the steel beam only, before the concrete deck cast.
In all the previous cases the concrete deck can be separately prestressed before the connection with the
lower steel beam, obtaining a further increment of compressive stress to prevent cracking.
302 EFFECTS ON EXTERNALLY PRESTRESSED COMPOSITE GIRDERS
In previous works [3, 4], the authors presented an analytical model and a numerical solving procedure
which consider the nonlinear consitutive behavior of materials and the slipping of the cable at saddle
points.
In this paper analytical and numerical aspects are briefly recalled and attention is focused on construction
sequences, by defining a unitary approach and by showing some results of practical interest.
Et uo-yvo (i)
The cable strain derives from the deformation of the whole beam. Denoting by the vector Qt with
components (x, ,y, the initial positions of the intermediate saddle points (i=l..Z)-l) and anchorages
(i=0, i=£>), and denoting by A, (<p) the difference of the values assumed by a generic function cp
between the saddle points i and r'-l, the cable strain assumes the following expression
(2)
i=i
1=0
where A ^|ß, - Q,_t | is the cable length in the reference configuration; a, and ß, are geometrical
terms related to the cable profile. The expression is linearized coherently with the assumption of small
displacements and rotations. The nonlinear elastic constitutive laws of beam materials and cable can be
posed in the form:
ct F(e-E0) (3)
The nonlinear analysis prevents superimposing stresses at every construction step, nevertheless the
analysis remains sufficiently simple by identifying the correct residual strain to consider at each
construction stage. In the sequel the configuration at which the stress is null at all the points of the steel
beam, which is present at each construction stage, is assumed as the reference configuration. So that, in
all the anlyses the following condition is considered
e0 0 in steel beam.
The residual strains in concrete deck, reinforcements and cable must be determined in each stage of the
different construction sequences.
The case of propped beam is the simplest situation because no preventive analysis is required for
evaluating the nonlinear behavior under increasing loads. In the case of cast in situ concrete deck, the
A. DALL'ASTA, L. DEZI 303
residual strain is null in the whole beam while the residual strain in the cable e0p permits controlling the
prestressing traction force. The effects due to a preventive prestressing applied to the upper concrete
deck before the connection with the steel beam can be analyzed by introducing a residual negative strain
£0pc furnishing the suitable compressive stress in the concrete slab prestressed separately. In conclusion,
the following residual strains must be adopted in the analysis:
The case of unpropped beam prestressed after the deck cast requires an initial analysis describing the
behavior in the initial stage where the sole steel beam undergoes weight loads of both steel beam and
concrete deck. This preventive analysis also permits evaluating the axis displacements t70 and v0
required for determining the residual strain to be considered in reinforced concrete deck for the analysis
of the second stage where the whole beam (deck, steel beam and cable) is present and where prestressing
can be introduced by means of an adequate value £0p of residual strain in the cable. If the concrete deck
is prestressed separately before connection a further residual strain £0pc must be adopted, as already
discussed in case A. In conclusion, in the analysisof the second stage the following residual strain fields
have to be considered.
The case of unpropped beams prestressed before the concrete deck cast, differs from the previous case
only because in the preventive analysis describing the structural behavior at the first construction stage,
the cable is present and the prestressing force is applied by means of e0p. Once the displacement fields
i70 and v0 existing in the first stage are determined, the residual strain field to be considered for the
upper deck in the second construction stage is furnished by an expression formally equal to that adopted
in case B, even if the axis displacements m0 and v0 are different. Separately prestressing of the concrete
deck can be taken into account as in previous cases.
3. Applications
The applications consider composite beams with depth H=1870mm and with the cross-section described
in detail in Fig.l.
The concrete has a cylindrical compressive strength fc=33 MPa, tensile strength fct=2.5MPa and
analyses were performed by using the constitutive law proposed in EC2 [5] (see Fig.l).Steel behavior
was been modeled by means of a bilinear curve defined by elastic modulus, yield stress, failure stress
and failure strain (see Fig.l). Beam and reinforcement steel have an elastic modulus Es=210 GPa while
cable steel elastic modulus is Es=190 GPa. The yielding stress of beam, reinforcement and cable steel are
respectively fy=280 MPa, fiy=440 MPa, fty=1680 MPa.
The following applications analyze the influences of the three construction sequences by comparing the
deformation and stress progress under increasing loads up to failure.
The results refer to a simply supported beam with span L=30 H, prestressed by a tendon anchored at the
centroids of the end cross section and draped along a profile defined by two intermediate saddle points.
In case A and B the tendon is anchored at the centroid of the composite section while in case C it is
anchored at the centroid of the steel beam. The initial prestressing force is equal to 5000 kN and stress is
equal to 1000 MPa in all the cases.
Fig.2 shows the load-midspan deflection curves for the three cases A, B, and C, while Fig.3 describe the
maximum stress progress respectively in concrete and steel. It can be noted that the ultimate loads are
almost the same in all the cases thanks to stress redistribution due to plasticity of materials while more
large differences can be
observed on displacements and
stress distributions (Fig.2). In
case B the deformation due to
prestressing, applied to the
complete section,
approximatively balances the
deformation due to dead loads,
applied to the steel beam only,
while in case A midspan
displacements are usually
positive under service loads.
Case C differs from case B only
because prestressing-i:ts on the
more deformable steel beam,
o to ooo -0 10 -0 20 -0 30 -0 40 -0 50 -0 60
even if
large differences do not
occur because the cable
Midspan deflection (m)
eccentricity is smaller in case C.
Fig.2 Midspan displacement versus loadfor sequences A, B, C.
02
e -04
-08
In the three cases the stiffness progress is similar: it remains approximatively constant for a large load
range and drastically reduces when the lower steel flange yields (Fig.3). In the considered cases, concrete
reaches its maximum stress for a slightly larger load (Fig.3). Cable does not fail thanks to stress
redistribution due to slipping even if it equally undergoes notably stress increment.
In the case A and B the prestressing force level is conditioned by upper deck cracking under dead loads
while no problem exists in case C where concrete is surely compressed under dead loads, so that this
latter seems to be the most effective construction sequence under the point of view of cracking
prevention. Other phenomena not considered here, such as shrinkage, may contribute to increasing deck
cracking.
A two-span continuous beam with span L=35H and a cable draped by three saddle points, one located at
the intermediate support and two at the span sections with maximum moment under uniform load is
considered. The lower steel
flange has been enlarged in the
negative moment region
(thickness 120mm) so that the
midspan section and support
section approximatively reaches
ultimate strains at the same load
level. The prestressing force is
equal to 7000 kN and the cable
is anchored at composite cross-
section centroids in case A and
B and at steel beam centroids in
case C. Fig.4 shows the curve
load-deflection at the section
where saddle points are located,
Fig.5 reports the progress of
concrete and steel maximum
0 10 0 00 -0.10 -0 20 -0.30 -0 40 -0 50 -0 60 -0 70 -0 80 -0 90 stress in the spans, Fig.6 shows
the progress of maximum stress
Deflection at saddle point (m)
in concrete, reinforcement and
Fig.4 Midspan displacement versus loadfor sequences A, B, C. the beam lower flange at the
support section.
0
-0.1
V '
caseC
1 1
0
V
case B
-0.2
\
£
è
-03
-0.4
"0.5
c&se A \V
-0.6 V\
-07 A
-08
N
N!A
-09
-1
0 20 40 60 80
Lo»d<kN/n,>
\
100 120 140 160
0 20 40 60 80 100
Load (kN/m)
120 140 160
This structural system exibhits a more complex behavior: in this case larger differences occur between
displacements deriving from prestressing on the whole beam (cases A and B) and prestressing acting on
the sole steel beam (case C); case C shows a lower global stiffness for a large load fields, mainly caused
by concrete traction failure; steel however yields for higher loads in case C and the global stiffness
achieves similar value for all the cases in the last phase. The support section attains yielding for a lower
load level with respect to span section and, even in this case, plasticity of materials leads to small
differences in the ultimate carrying load capacity in the three cases. The smallest deflections at low load
level are observed for cases A and C while the largest deflections occur in the case B.
In the considered beam, yielding of steel precedes the concrete failure at spans (Fig.5). The stress
progress in concrete are almost similar in case B and C while a higher compressive level can be observed
in case A. Prestressing however has little effectiveness on concrete at span section. A noticeable
reduction of tensile stress exists in the three cases and yielding occurs at higher load level in the steel
beam of case C thanks to the prestressing modalities.
In continuous beams prestressing permits a sensible reduction of cracking in the region with negative
moment, even if the usual prestressing advantages are partially limited by the positioning of saddle points
parallel to the web that does not permit obtaining large eccentricity.
In particular, it is evident from Fig.6 that in cases A and B cracking is prevented for a large range on
external loads while, contrary to the previous case of simply supported beams, no advantages can be
obtained in case C. The stress progress in reinforcement rapidly increases after the deck craking in cases
A and B while in case C reinforcement yields for a much lower load level (Fig.6). Prestressing shows a
little effectiveness in case C while a significative reduction of compressive stress occurs in case A and B.
O / fc o / fry 0/
Fig.6 Section at support: progress of maximum stress in concrete, reiforcements and beam steel.
4. References
1 Saadatmanesh H., Albrecht P., Ayyub B.M. (1989). Guidelines for flexural design of prestressed
/
composite beams. Struct. Engrg., ASCE, 115(11), 2944-2961.
2 Ayyub B.M., Sohn Y.G., Saadatmanesh H, (1992), Prestressed composite girders. II: Analytical study
of negative moment. J. Struct. Engrg., ASCE, 118(10), 2763-2783.
3 Dall'Asta A., Dezi L. (1993). Nonlinear analysis of beams prestressed by unbonded cables. J. Engrg.
Mech., ASCE, 116(11), 2931 -2951.
4 Dall'Asta A., Dezi L. (1997). Non-linear behavior of externally prestressed composite beams: analytical
model. Submitted for publication.
5 "Eurocode No.2: Design of concrete structures" (1991). Eurocode 2 Editorial Group.
307
Summary
In this paper the method proposed by RPX-95[ 1 ] for the analysis of the behaviour of composite
cross-sections subject to normal forces is briefly explained. This procedure allows the modelling
of the behaviour in Ultimate Limit State, as well as in previous strain situations (moment-
curvature diagram). An experimental program designed to test this approach, and presently under
way, is also presented, as well as some of the results obtained so far.
1. Introduction
Eurocode 4 [2] as well as other standards [3,4] propose different approaches for the analysis of
the bearing capacity of composite cross-sections subject to normal forces (Elastic, Corrected
Elastic, Plastic, Elasto-Plastic). The Elasto-Plastic method, which models the behaviour in the
most realistic manner, is formulated for slender compressed plates and stiffened compressed
plates in terms of stresses. This fact does not permit the determination of the corresponding
strain, and therefore does not allow to evaluate the ductility available.
The recently published RPX-95 'Recommendations for the Design of Composite Highway
Bridges' [1] includes a different elasto-plastic approach for the evaluation of the Ultimate
Bearing Capacity (Ultimate Limit State) and for the evaluation of intermediate strain states
(Moment-Curvature diagram). The main aspects of this procedure are discussed in what follows.
This procedure allows to determine the deformational state of the cross section for every load
level.
In order to test this procedure, an experimental program, which is briefly presented in this article,
is being carried out. Some of the results obtained so far are also presented.
308 BEHAVIOUR OF COMPOSITE BOX GIRDER SECTIONS AT ULTIMATE LIMIT STATE
2. Sectional Behaviour
The study of composite cross-sections subject to normal forces requires the use of models which
adequately represent the behaviour of slender compressed plates, stiffened compressed plates,
slender plates stiffened by conection to a concr