This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Page i

Bridge Deck Analysis

Page ii

This book is dedicated to Orlaith, Sadhbh and Ailbhe, and to Margaret

Page iii

**Bridge Deck Analysis
**

Eugene J.O’Brien and Damien L.Keogh Department of Civil Engineering, University College Dublin, Ireland

Chapter 4 written in collaboration with the authors by

Barry M.Lehane Department of Civil, Structural and Environmental Engineering, Trinity College Dublin, Ireland

London and New York

Page iv

First published 1999 by E & FN Spon 11 New Fetter Lane, London EC4P 4EE Simultaneously published in the USA and Canada by Routledge 29 West 35th Street, New York, NY 10001 E & FN Spon is an imprint of the Taylor & Francis Group This edition published in the Taylor & Francis e-Library, 2005. To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk. © 1999 Eugene J.O’Brien and Damien L.Keogh Cover photograph: Killarney Road Bridge, courtesy of Roughan and O’Donovan, Consulting Engineers All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility or liability for any errors or omissions that may be made. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data O’Brien, Eugene J., 1958– Bridge deck analysis/Eugene J.O’Brien and Damien L.Keogh. p. cm. Includes index. ISBN 0-419-22500-5 1. Bridges-Floors. 2. Structural analysis (Engineering) I.Keogh, Damien L., 1969–. II. Title. TG325.6.027 1999 624’.253–dc21 98–48511 CIP ISBN 0-203-98414-5 Master e-book ISBN

ISBN 0-419-22500-5 (Print Edition)

5 Impact loading 2.1 Introduction 1.3 Imposed traffic loading 2.1 Introduction 2.4 Thermal loading 2.2 Moment distribution viii x 1 1 1 2 8 26 29 32 34 40 40 42 43 46 51 52 54 67 67 67 .Page v Contents Preface Acknowledgements Chapter 1 Introduction 1.7 Prestress loading Chapter 3 Introduction to bridge analysis 3.3 Cross-sections 1.5 Articulation 1.6 Dynamic effects 2.7 Joints 1.1 Introduction 3.6 Bearings 1.2 Factors affecting structural form 1.2 Dead and superimposed dead loading 2.8 Bridge aesthetics Chapter 2 Bridge loading 2.4 Bridge elevations 1.

4 Planar finite-element analysis of slab decks 5.1 Introduction 7.7 Application of moment distribution to grillages Chapter 4 Integral bridges 4.3 Differential settlement of supports 3.5 Wood and Armer equations Chapter 6 Application of planar grillage and finite-element methods 6.5 Differential temperature effects 3.4 Voided slab bridge decks 6.2 Contraction of bridge deck 4.1 Introduction 5.3 Conventional spring model for deck expansion 4.6 Prestress 3.4 Three-dimensional analysis 7.5 Beam and slab bridges 6.5 Run-on slab 4.3 Grillage analysis of slab decks 5.Page vi 3.6 Cellular bridges 6.3 Effective flange width 7.2 Simple isotropic slabs 6.2 Shear lag and neutral axis location 7.2 Thin-plate theory 5.1 Introduction 4.7 Skew and curved bridge decks Chapter 7 Three-dimensional modelling of bridge decks 7.5 Upstand grillage modelling 75 78 89 104 111 121 121 128 133 137 145 147 151 151 151 169 185 191 200 200 200 203 211 218 228 236 240 240 240 242 244 245 .4 Modelling expansion with an equivalent spring at deck level 4.4 Thermal expansion and contraction 3.1 Introduction 6.6 Time-dependent effects in composite integral bridges Chapter 5 Slab bridge decks—behaviour and modelling 5.3 Edge cantilevers and edge stiffening 6.

7 Prestress loads in three-dimensional models Appendix Reactions and bending moment diagrams due to applied load A 252 260 263 Appendix Stiffness of structural members and associated bending moment diagrams 265 B Appendix Location of centroid of a section C 267 Appendix Derivation of shear area for grillage member representing cell with flange 269 D and web distortion References Index 272 274 .Page vii 7.6 Upstand finite-element modelling 7.

joints and aesthetics not commonly found in bridge analysis books. if an error arose early on in the calculations. fairly complex skew. between them. This is particularly useful for . in turn leading to the calculation of mx . This book fulfils just that role. it is possible to change a dozen variables and a computer program will recalculate stresses and reactions in seconds. Indeed. Now. however. Written by two engineers who have. gave surfaces for various stress and aspect ratios up to a 45° skew. experience of almost all aspects of modern bridge design and analysis. irregularly supported structures is essential. In many aspects. copiously and carefully illustrated. the book contains either a novel approach to design or entirely new methods. it represents years of scholarship and research presented in a lucid and understandable style which should make even the more complex theory understandable to all engineers. many days could be spent in re-analysing. perhaps more so now than in the past. unambiguous English. with sections on bearings. loading (with prestress treated as a special case of loading) and details of a unique graphical approach to moment distribution—a powerful tool in engendering an understanding of fundamental structural behaviour. Written in clear. an understanding of the behaviour of non-symmetrical. eccentrically loaded. translated from the German by the Cement and Concrete Association. my and mxy moments. amongst other techniques. it includes chapters on every aspect of bridge deck analysis that a practising bridge engineer is ever likely to need. The method was tedious. the use of planimeters on the way to calculating volumes under the influence surface.Page viii Preface Twenty-five years ago. for a bridge engineer to understand how a bridge deck responds to various combinations of load and to be able to decide if the ‘answer’ (output) is sensible. To be confident of this. somewhat approximate and could often take weeks. There is still a need. The famous Rusch and Hergenroder influence surface charts. prestressed concrete bridge decks could be analysed with a fair degree of accuracy—but only by using manual methods. Full analysis of a bridge deck involved. It covers construction in some detail.

Other chapters deal comprehensively with integral bridges (with a major geotechnical input from Dr Barry Lehane) and the increasing acceptance of FE methods of analysis. Structural and Environmental Engineering Trinity College Dublin . although the merits of grillage methods are not ignored. Professor S.Page ix checking the output of computer analyses. this must prove the standard work on bridge deck analysis for decades to come. All in all.Perry Civil.H.

The assistance of Chris Davis and Michael Barron of Mott McDonald with Chapter 2 is gratefully acknowledged. and readers are encouraged to bring errors of substance to our attention. Ancon CCL are also acknowledged for providing a number of illustrations. arising from information contained in this publication. A sabbatical stay in Slovenia for the first author made the initial drafting of many chapters possible.Perry and Trinity College Dublin. The authors of STRAP (ATIR software. some of which are novel and have not been exposed to the rigours of time. The stay in Slovenia was greatly enhanced and enriched by Alenka Ž nidarič . where both authors were employed for a time.Page x Acknowledgements We would like to thank Dr A. The publisher and authors disclaim any liability.Ghali most sincerely for major contributions to some of the earlier chapters. Tel Aviv) and NIKE3D (Lawrence Livermore National Laboratories. The material represents the opinions of the authors. USA) are thanked for the use of their programs. including the cover illustration. Sound engineering judgement should be the final arbiter in all stages of the design process.H. Disclaimer This publication presents many advanced techniques. . Despite the best efforts of all concerned. Readers should use their own judgement as to the validity of the information and its applicability to particular situations and check the references before relying on them. Special thanks is due to Joe O’Donovan for providing some of the photographs in the text. The support of Roughan and O’Donovan Consulting Engineers. in whole or in part. He gave most generously of his time with the sole objective of getting it right. and should be treated as such. is much appreciated. This would not have been feasible without the enthusiasm of Aleš nidarič the Ž of Slovenian National Building and Civil Engineering Institute and the support of the University of Ljubljana. The initial writing effort was greatly facilitated for both authors through the support of Professor S. typographical or editorial errors may occur.

There is now a significant move away from bridges that are easy to design towards bridges that will require little maintenance.Page 1 Chapter 1 Introduction 1. in many others.1 which are commonly used in bridge engineering. In this figure. 1.2 Factors affecting structural form In recent years. bridges are carrying significantly more traffic load than originally intended. The bridges that were easy to design were usually determinate. This is often associated with joints that are leaking or with details that have resulted in chloride-contaminated water dripping onto substructures. The new awareness of the need to design durable bridges has led to dramatic changes of attitude towards bridge design.1 Introduction A number of terms are illustrated in Fig. 1. . There may be upstands or downstands at the ends of the cantilever for aesthetic purposes and to support the parapet which is built to retain the vehicles on the bridge. and transversely. However. The deck spans longitudinally. Bridge decks are frequently supported on bearings which transmit the loads to abutments at the ends or to piers or walls elsewhere. it has been established that a significant portion of the world’s bridges are not performing as they should. In some cases. which is perpendicular to it. The main body of the bridge superstructure is known as the deck and can consist of a main part and cantilevers as illustrated. all parts of the bridge over the bearings are referred to as superstructure while the substructure includes all parts below. the problem is one of durability—the widespread use of de-icing salt on roads has resulted in the ingress of chlorides into concrete. Joints may be present to facilitate expansion or contraction of the deck at the ends or in the interior. Problems have also been reported with post-tensioned concrete bridges in which inadequate grouting of the ducts has lead to corrosion of the tendons. which is the direction of span.

The methods of construction in turn are often dictated by the particular conditions on site. is not a very efficient structural form as the second moment of area of a rectangle is relatively small. For example.Page 2 Fig. steel beams carry the self weight of the deck while composite steel and in-situ concrete carry the imposed traffic loading.3 Cross-sections 1. Various alternative structural bridge forms and methods of construction are presented in the following sections. For example. However. illustrated in Fig. such structural forms have many joints which are prone to leakage and also have many bearings which require replacement many times over the lifetime of the bridge.1 Portion of bridge illustrating bridge engineering terms e.2.g. when a bridge is to be located over an inaccessible place. The structural forms of bridges are closely interlinked with the methods of construction. Such a bridge is generally constructed of reinforced concrete (particularly for the shorter spans) or prestressed concrete. The method of construction also influences the distributions of moment and force in a bridge. simply supported spans and cantilevers.3. Due to the inefficiency of this structural form. This immediately limits the structural forms to those that can be constructed in this way. in some bridges. the stresses . 1. such as a railway yard or a deep valley. 1. 1. The move now is towards bridges which are highly indeterminate and which have few joints or bearings. the construction must be carried out without support from below.1 Solid rectangular The solid rectangular section.

2 In-situ solid rectangular section: (a) without cantilevers. However. 1. 1.2(a) and (b). the shuttering costs for a bridge with a flat soffit are relatively low and the reinforcement is generally simple. bridges can be constructed with or without cantilevers. Such construction is clearly more economical when support from below the bridge is readily available.2. With this form of construction. without much reduction in the second moment of area. Comparing bridges of the same width.g. what is often the more important advantage of cantilevers is the aesthetic one. Holes are cast at frequent intervals along the length of such beams to facilitate the threading through of transverse bottom reinforcement. As can be seen in Fig. Solid rectangular sections can be constructed simply from in-situ concrete as illustrated in Fig.Page 3 Fig. 1. e. (b) with cantilevers induced by the self weight of the concrete can become excessive. such as illustrated in Figs. 1. 1.3 Precast and in-situ solid rectangular section . over railway lines or deep waterways.8. which is discussed in Section 1. 1. In-situ reinforced concrete is then poured over the precast beams to form the complete section. When this is not the case. it can be seen that the bridge with cantilevers has less weight. the precast beams must be designed to carry their self weight plus the weight of the Fig. a rectangular section can be constructed using precast pretensioned inverted-T-sections as illustrated in Fig. As a result.2. this form of cross-section is often the most cost-effective for shorter spans (up to about 20 m).3. However.

the preferred solution.4 Voided slab section with cantilevers (initially wet) in-situ concrete. in some cases. As a result. These decks can be constructed from ordinary reinforced concrete or can be post-tensioned. particularly that designed to resist transverse bending. the section tends to be deeper for a given span. 1. are more expensive in terms of shuttering . 1. Including voids in a bridge deck increases the cost for a given structural depth because it adds to the complexity of the reinforcement.3 T-section The T-section is commonly used for spans in the range 20–40 m as an alternative to voidedslab construction. For the span range of 20–30 m. it is common practice in some countries to use in-situ concrete with polystyrene ‘voids’ as illustrated in Fig. In-situ T-section decks. 1.5. This problem is not so much one of steel straps failing as of grooves being cut in the polystyrene by the straps. However. Hence it is. It is common practice to treat voided slabs as solid slabs for the purposes of analysis provided that the void diameter is less than 60% of the total depth. The complete rectangular section is available to carry other loading.3. the T-section is a less efficient structural form as it tends to have more material close to the neutral axis of the bridge than a voided slab. illustrated in Fig. Guidance is given on the analysis of this type of deck in Chapter 6. the voids must be accounted for when considering the design to resist transverse bending. Regardless of the diameter-to-depth ratio. Concerns have been expressed about voided-slab construction over the lack of inspectability of the concrete on the inside of the void and there are many countries where this form is virtually unknown. solid rectangular sections become increasingly less costeffective due to their low second moment of area to weight ratio.Page 4 Fig. 1.4.2 Voided rectangular For spans in excess of about 20 m. However. particularly when the designer wishes to minimise the structural depth. The shuttering costs are also less than for in-situ concrete T-sections which are described below. 1. It is essential in such construction to ensure that sufficient stays are provided to keep the voids in place when the concrete is poured and to prevent uplift due to flotation.3. it reduces considerably the self weight and the area of concrete to be prestressed without significantly affecting the second moment of area.

An in-situ concrete slab. 1. 1. as illustrated in Fig. 1. spans transversely between the beams while acting as flanges to the beams longitudinally. it becomes economical to use ‘cellular’ or ‘box’ sections as illustrated in Fig. supported on permanent shuttering. precast concrete or steel forms of T-section. Over less accessible places.6. These consist of pretensioned prestressed concrete or steel beams placed in position along the length of the span.Page 5 Fig. These have a higher second moment of area .4 Box sections For spans in excess of 40 m.5 In-situ concrete T-sections: (a) single web such as might be used for a pedestrian bridge.3. (b) multiple webs such as would be used for wider decks costs than voided slabs but have a major advantage in that all of the bridge deck is totally inspectable.7. are favoured. 1.

to inspect the inside of the void. (b) composite precast Y-beam and in-situ concrete Fig. However. (b) multi-cellular per unit weight than voided slab or T-sections. they are only considered economical at higher spans as it is only then that the structural depth becomes sufficiently great (about 2 m) for personnel to enter the void to recover the shuttering and.7 Box sections: (a) single cell.Page 6 Fig. 1. 1.6 T-sections: (a) composite steel and concrete. . when the bridge is in service.

1. It was also common practice in the past to build bridges of ‘pseudo-box’ construction as illustrated in Fig.9 Precast M-beam . In the past.6(b)) used more commonly today. The section is more efficient than a T-section as more concrete is located away from the centroid.8 Composite precast and in-situ box section Box sections can be constructed of in-situ or precast concrete or can be composite with a precast pre-tensioned U-section and an in-situ concrete slab as illustrated in Fig.10. The analysis of this type of bridge is similar to that of any T-section bridge. 1.Page 7 Fig. The bottom in-situ concrete was reinforced transversely by threading bars through holes cast in the M-beams. However. M-sections were often placed side by side with the bottom flanges within millimetres of each other. 1. 1. 1. in the past. if water leaks into the voids. Fig.5 Older concepts Many variations of the above structural forms have been used in the past and are evident in existing bridge stocks. it was common practice to construct Tsection decks using precast ‘M-beams’ (Fig.3. For example. corrosion problems can result and.8. 1.9). 1. These were constructed of M-beams with insitu concrete near the bottom to form a void. A disadvantage of the M-section is that it is difficult to compact the concrete properly at the top surface of the wide bottom flange. These have wider bottom flanges than the precast ‘Y-beams’ (Fig.

The structural behaviour of the pseudo-box section is similar to that of a small multi-cellular box section.Page 8 Fig. 1. Thus the transverse deformation is assumed to be as illustrated in Fig. i. The latter ‘shear keys’ are assumed to be capable of transferring shear force but not transverse bending moment as they have no transverse reinforcement. rotation is assumed to occur at the joints between precast units. This consists of precast concrete slab strips joined using longitudinal strips of insitu concrete. 1.11(b). (b) assumed transverse deformation .10 Pseudo-box section due to the nature of this structural form. 1.4 Bridge elevations The cross-sections described above can be used in many different forms of bridge. Fig. 1. However. they are no longer popular due to concerns about the durability of the in-situ joints. assessment and repair is difficult.11 Shear-key deck: (a) section through small portion of deck. Many of the alternative bridge elevations and their methods of construction are described in the following sections. illustrated in Fig. 1.e.11(a). Another form of construction used widely in the past is the ‘shear key’ deck. Shear key decks were popular for railway bridge construction as the railway line could be reopened even before the in-situ concrete was placed.

12.12 Simply supported beam or slab Fig. For example. In addition. have significantly fewer joints and bearings. This form.4. as illustrated in Fig. the bending moment diagrams due to a uniformly distributed loading of intensity ω(kN/m) are illustrated in Fig. the concrete pours are moderately sized. if constructed using in-situ concrete.1 Simply supported beam/slab The simplest form of bridge is the single-span beam or slab which is simply supported at its ends. is widely used when the bridge crosses a minor road or small river. there is less disruption to any traffic that may be below as only one span needs to be closed at any one time. The cross-section is often solid rectangular but can be of any of the forms presented above. The simply supported bridge is relatively simple to analyse and to construct but is disadvantaged by having bearings and joints at both ends. 1. Like single-span bridges.14 Continuous beam or slab . The implication of this is that the bridge deck needs to be correspondingly deeper. Fig. 1. the span is relatively small and multiple spans are infeasible and/or unnecessary. as illustrated in Fig. 1. It can be seen that the maximum moment in the simply supported case is significantly greater (about 25%) than that in the continuous case. A further disadvantage of simply supported beam/slabs in comparison to continuous ones is that the maximum bending moment in the former is significantly greater than that in the latter.Page 9 1.14. Continuous beams/slabs. 1. 1. In such cases.4. 1. there are a great many joints and bearings with the result that a series of simply supported beams/slabs is no longer favoured in practice.13. 1.15. one after the other. It is particularly favoured on poor soils where differential settlements of supports are anticipated. 1. However.2 Series of simply supported beams/slabs When a bridge crossing is too wide for an economical single span.13 Series of simply supported beam/slabs Fig. it is possible to construct what is in effect a series of simply supported bridges. this form is relatively simple to analyse and construct. It also has the advantage that. illustrated in Fig.

17. 1. . 1. The slab at the support in this form of construction is particularly flexible and tends to attract a relatively low bending moment. Further. as the total bridge length becomes large. 1. the in-situ concrete is cast to the full depth of the bridge over all supports to form what is known as a diaphragm beam.3 Continuous beam/slab with full propping during construction As stated above. This is possible due to the fact that members of low structural stiffness (second moment of area) tend to attract low bending moment. In the alternative form of partially continuous bridge. 1. Precast concrete or steel beams are placed initially in a series of simply supported spans.4 Partially continuous beam/slab When support from below during construction is expensive or infeasible. In-situ concrete is then used to make the finished bridge continuous over intermediate joints.16.Page 10 Fig. continuity over intermediate supports is provided only by the slab.4. the concrete can be poured in-situ in one pour. This tends to increase cost as the construction becomes more of a batch process than a continuous one. the amount of concrete that needs to be cast in one pour can become excessive. However. For bridges of moderate total length. it is possible to use precast concrete or steel beams to construct a partially continuous bridge. In the form illustrated in Fig. Thus the in-situ slab alone is required to resist the complete hogging moment at the intermediate supports. continuous beam/slab construction has significant advantages over simply supported spans in that there are fewer joints and bearings and the applied bending moments are less. Two forms of partially continuous bridge are possible. 1. There is concern among some designers about the integrity of such a joint as it must undergo significant rotation during the service life of the bridge. (b) one three-span continuous beam with span lengths l 1.15 Bending moment diagrams due to uniform loading of intensity ω (a) three simply : supported spans of length l. This completely removes the need for any joints.6. illustrated in Fig. Elsewhere the cross-section is similar to that illustrated in Fig.4.

1. 1.16 Partially continuous bridge with full-depth diaphragm at intermediate supports: (a) elevation.Page 11 Fig. (b) plan view from below Fig.17 Partially continuous bridge with continuity provided only by the slab at intermediate supports .

due to creep. The total bending moment diagram will be a combination of that due to self weight and other loading. 1. 1.18 Joint detail at intermediate support of partially-continuous bridge of the type illustrated in Fig. By the time the imposed traffic loading is applied. In partially continuous bridges. self weight continues to cause deformation in the bridge after it has been made continuous. the precast concrete or steel beams carry all the self weight of the bridge which generates a bending moment diagram such as that illustrated in Fig. 1. Fig. 1.17 as the main bridge beams rotate at their ends. Unfortunately.Page 12 Fig.19(a) for a two-span bridge. This introduces a complexity into the analysis compounded by a great difficulty in making accurate predictions of creep effects. (b) bending moment due to loading applied after bridge has been made continuous . 1.18.19(a). 1. At this stage it is resisted by a continuous rather than a simply supported beam/slab and it generates a distribution of bending moment more like that of Fig.19 Typical distribution of bending moment in two-span partially-continuous bridge: (a) bending moment due to self weight.19(b) than Fig. the joint must move longitudinally to accommodate this rotation as illustrated in Fig. 1. the bridge is continuous and the resulting bending moment diagram is as illustrated in Fig. 1.19(b).

Fig. Proprietary post-tensioning couplers. However.20(b). the point where one concrete pour meets the next is designed to transmit bending moment and shear force and is not intended to accommodate movements due to thermal and creep effects. an intermediate joint may become necessary to relieve stresses due to expansion/contraction. The method is also of a continuous rather than a batch form as the precast beams can be constructed at a steady pace. starting even before work has commenced on site.Page 13 The great advantage of partially continuous construction is in the removal of all intermediate joints while satisfying the requirement of construction without support from below. resulting in minimum disruption to any existing traffic passing under the bridge. while intermediate joints have been removed. 1. in-situ construction. Construction on site is fast. Particularly for the form illustrated in Fig. can be a viable option. 1.4. two bearings are necessary at each intermediate support. 1.20(a). This is achieved using temporary formwork supported on the bridge piers as illustrated in Fig. It has been said that joints should be provided every 100 m at least.17. one span at a time. intermediate bearings are still present with their associated maintenance implications.5 Continuous beam/slab—span-by-span construction For construction of particularly long bridges when access from below is expensive or infeasible. In this form of construction. A significant disadvantage is that. The joint may sometimes be located at the quarter-span position as illustrated in Fig. where bending moments and shear forces are relatively small. 1. this figure is constantly being revised upwards as the problems of bridge joints in service receive ever more attention. In particularly long continuous beam/slabs.20 Temporary support system for span-by-span construction: (a) joint over intermediate pier. 1. can be used to achieve continuity of prestressing across construction joints. 1. (b) joint at quarter span .21. such as illustrated in Fig.

22(b)) until such time as they can be permanently posttensioned into place as illustrated in Fig.22. precast beams are not generally available to span the complete length at once. 1. 1. 1. one on each side.4. This form of bridge is generally made of post-tensioned prestressed concrete. The sequence of construction is illustrated in Fig. in anticipation of the need to post-tension future segments at later stages of construction. either alternately on opposing sides or simultaneously in pairs.6 Continuous beam/slab—balanced cantilever construction When the area under a bridge is inaccessible and spans are in excess of about 40 m. Ducts are placed in all segments when they are first cast. 1. either method is only capable of resisting a relatively small out-of-balance moment so it is necessary to have approximately equal lengths of cantilever on each side at all times during construction. At spans of this length. This is prevented from rotation either by a rigid connection between pier and deck or by construction of a temporary prop or props connecting the deck to the foundation as illustrated.22(a)). The segments are supported by a ‘travelling form’ connected to the existing bridge (Fig. However. in the case of .Page 14 Fig. Segments of deck are then added to the base segment.22(c). 1. An intermediate pier is cast first and a small part of the bridge deck (Fig.21 Post-tensioning coupler to transmit prestress forces across a construction joint (photograph courtesy of Ancon CCL) 1. it is often economical to construct bridges by the balanced cantilever method. The crosssection is generally of the box type constructed either of in-situ concrete or precast segments of relatively short length (4–5 m longitudinally). Segments can be cast in-situ or precast.

Page 15 the latter. it does not normally serve any structural purpose. there is typically a ‘shear key’ as illustrated in Fig. 1. . While epoxy resin is commonly used to join segments.22(d) to provide a positive method of transferring shear between segments. Moment is transferred by the concrete in compression and by the post-tensioning tendons.

24. After the casting of the stitching segments and completion of construction.23 Casting of stitching segment .23. the moment due to self weight during construction is such as illustrated in Fig. a ‘stitching segment’ is cast to make the bridge continuous as illustrated in Fig. 1. When cantilevers meet at mid-span. (b) temporary support of segments. 1. 1. illustrated in Fig. such as illustrated in Fig. (c) sectional elevation showing tendon. to resist the sagging moment that will exist in the finished structure due to applied traffic loading. the bridge forms a continuous beam and the imposed service loading generates a distribution of moment. 1. Post-tensioning tendons are placed in the bottom flange and webs by means of ‘blisters’.25(b). 1. This form of bridge is quite inefficient as parts of it must be designed to resist a significant range of moments from large hogging to large Fig. The bending moment in a balanced cantilever bridge is entirely hogging while the bridge remains in the form of a cantilever. (d) precast segment Segments are added on alternate sides until they reach an abutment or another cantilever coming from the other side of the span.25(a).Page 16 Fig.22 Balanced cantilever construction: (a) elevation of base segment and pier. 1. Thus.

1. ‘incremental-launch’ or ‘push-launch’ becomes a viable alternative to balanced cantilever as a method of construction. Nevertheless. In pushlaunch construction. (b) due to imposed loading after completion of construction .25(b). it is frequently the most economical alternative for construction over deep valleys when propping from below is expensive.25 Distributions of bending moment in balanced cantilever bridge: (a) due to self weight during construction. 1. 1.26(b)). This process is continued until the complete bridge has been constructed behind the abutment and pushed into place. The analysis of balanced cantilever bridges is complicated by a creep effect similar to that for partially continuous beams.25(a) towards a form approaching that illustrated in Fig.7 Continuous beam/slab—push-launch construction For spans in excess of about 60 m. 1.26(a). 1. This is caused by a tendency for the distribution of moment due to self weight to change in the long term from the form illustrated in Fig. 1. temporary sliding bearings are used to minimise friction forces. a long segment is cast behind the bridge abutment as illustrated in Fig.4.24 Blisters and tendon in the bottom flange (sectional elevation) sagging. Hydraulic jacks are then used to ‘push’ this segment out into the first span to make way for the casting of another segment behind it (Fig. Fig.Page 17 Fig. 1. When the deck is being pushed over intermediate supports. This results from creep deformations which are still taking place after the bridge has been made continuous.

1. 1.28(b). and are only a viable solution if it can be accommodated. Some engineers design bridges in an arch form for aesthetic reasons but articulate the bridge like a .26 Push-launch construction: (a) casting of the first segment. Parts of the deck must be designed for significant hog moment during construction as illustrated in Fig. This doubling of cantilever length has the effect of quadrupling the moment due to self weight during construction. must be designed for the creep effect and are subject to the associated complexity and uncertainty in design.8 Arch bridges For larger spans (in excess of about 50 m). However. the arch form is particularly effective. (b) pushing of the partially constructed bridge over first span The method has a considerable advantage of access. an arch is still a possibility if it is tied such as illustrated in Fig. arches generate a significant horizontal thrust. A significant disadvantage stems from the distribution of bending moment generated temporarily during construction.4. the horizontal thrust is taken by the tie. 1. These same parts may be subjected to sag moment in the completed bridge as illustrated in Fig. as illustrated in Fig.27(b). like those designed for balanced cantilever construction. Bridges designed for pushlaunch construction. All of the bridge is constructed in the same place which is easily accessible to construction personnel and plant.28(a). 1.Page 18 Fig. In a tied arch.27(a). This can be achieved if the bridge is located on a particularly sound foundation (such as rock). The effect is greater than in balanced cantilever construction as the cantilever length is the complete span length (as opposed to half the span length for the balanced cantilevers). If this is not the case. 1. 1.

28 Arch bridges: (a) conventional form with deck over the arch. (b) tied arch with deck at base of arch .27 Distributions of bending moment in push-launch bridge: (a) due to self weight during construction.Page 19 Fig. (b) due to imposed loading after completion of construction Fig. 1. 1.

as the bridge has no means by which to resist the horizontal thrust. while a continuous beam/slab crossing a 60 m motorway would normally be divided into two or four spans. the structural depth can be very small and large clear spans can readily be accommodated.29 Simply supported beam bridge in the shape of an arch Fig. While traditional masonry arches were designed to be completely in compression. Thus the self weight generates a distribution of stress which is.30 Deflected shape of arch subjected to thermal contraction . This is perfectly feasible but.30. in fact. is the fact that the curved form results in shuttering which is more expensive than would otherwise be the case. Other advantages of arches are that they are aesthetically pleasing in the right environment. movements due to thermal expansion/contraction and creep/shrinkage do generate some stresses but these are not as significant as those in the frame form of construction discussed below. An additional major advantage is that arches require no bearings as it is possible to cast the deck integrally into the substructures. 1. 1. If arches are located over inaccessible areas. modern concrete or steel arches have no such restriction and can be designed to resist bending as well as the axial compression generated by the arch form. other than the problem of accommodating the horizontal thrust. Fig. considerable temporary propping is required to support the structure during construction. 1. As can be seen in Fig. The principal disadvantage of concrete arches. The arch action causes the self weight to generate a compression which has all the advantages of prestress but none of the disadvantages of cost or durability associated with tendons. as illustrated in Fig. it behaves structurally as a simply supported beam. beneficial and assists in the resistance of stresses due to imposed loading. 1. Concrete arches are particularly effective as concrete is very strong in compression.Page 20 simply supported beam. For example.29. an arch can readily span such a distance in one clear span creating an openness under the bridge that would not otherwise be possible.

1. Given the great upsurge of interest in maintenance and Fig. until recently. 1. The effects of deck shortening relative to the supports is to induce bending in the whole frame as illustrated in Fig. 1. there is the usual complexity and uncertainty associated with such calculations.31 Frame/box culvert bridges: (a) box culvert. If some of this shortening is due to creep or shrinkage.Page 21 1. can be analysed using two-dimensional models. The minimal maintenance requirement of frame/box culvert bridges is their greatest advantage. accommodating movements due to temperature changes or creep/shrinkage can be a problem and. unless the transverse width is relatively small.4. (b) three-span frame . the structural behaviour is three-dimensional. 1. However. This is because the maximum bending moment tends to be less. There are no joints or bearings as the deck is integral with the piers and abutments. A further complexity in the analysis of frame bridges is that. Continuous slab bridges on the other hand. it was not considered feasible to design frame bridges of any great length (about 20 m was considered maximum). as can be seen from the examples of Fig.31.33. are more effective at resisting applied vertical loading than simply supported or continuous beams/slabs. such as illustrated in Fig.9 Frame/box culvert (integral bridge) Frame or box bridges.32.

1. (b) continuous beams. If the supports are fully fixed against translation. (c) frames/box culverts Fig. 1.32 Typical distributions of bending moment: (a) simply supported spans. this lack of maintenance has resulted in an explosion in the numbers of bridges of this form.33 Effect of thermal contraction of deck in frame bridge: (a) deflected shape. both relating to longitudinal movements.34. There are two implications for longer frame-type bridges. (b) distribution of bending moment durability in recent years.Page 22 Fig. deck movements in such bridges will generate enormous stresses. 1. If the bridge is supported . It is now considered that bridges of this type of 100 m and longer are possible. This problem has been overcome by allowing the supports to slide as illustrated in Fig. Ever longer spans are being achieved.

Fig. Such a joint is remote from the main bridge structure and. To overcome this.35.34 Sliding support and run-on slab in frame bridge on piles. There are a number of variations of this form of construction which are considered further in Chapter 4. 1. the axes of the piles are orientated so as to provide minimum resistance to longitudinal movement. 1.35 Composite precast and in-situ concrete frame bridge . These can be used in combination with in-situ concrete to form a frame bridge as illustrated in Fig. The second implication of longer frame bridges is that the bridge moves relative to the surrounding ground.Page 23 Fig. 1. a joint is required to facilitate translational movements. Precast pretensioned concrete beams have a good record of durability and do not suffer from the problems associated with grouted post-tensioning tendons.6(b). 1. engineers specify ‘run-on’ slabs as illustrated in the figure which span over loose fill that is intended to allow the abutments to move. if it does leak. will not lead to deterioration of the bridge itself. Crosssections are typically of the form illustrated in Fig. A precast variation of the frame/box culvert bridge has become particularly popular in recent years. The run-on slab can rotate relative to the bridge deck but there is no relative translation. at the ends of the run-on slabs. Thus.

were used. can be placed in position very quickly over a road or railway requiring a minimum closure time. Thus. 1. which provides access. 1. In older bridges of the type. 1. is illustrated in Fig. as illustrated in Fig. some older bridges were constructed of precast concrete with drop-in spans. However. Fig. two ‘halving joints’. A more convenient alternative.10 Beams/slabs with drop-in span For ease of construction and of analysis. 1. The form has the disadvantage of having joints and bearings at the ends of the drop-in span as well as at the extremities of the bridge itself. This bridge is determinate as the central ‘drop-in’ part is simply supported. it can readily be constructed over inaccessible areas. it is still popular in some countries for pedestrian bridges over roads.4.Page 24 1.37(b). (b) alternative detail with access . The joint and bearing detail at the ends of the drop-in span in this form of construction is particularly important. 1. The side spans are simply supported with cantilevers to which point loads from the drop-in span are applied at their ends. A typical example is illustrated in Fig. This detail is particularly problematic as access to inspect or replace the bearings is extremely difficult.37(a). The drop-in span. in particular.36.37 Halving joint at end of drop-in span: (a) traditional detail (no access).36 Beam bridge with drop-in span Fig.

38. 1. halving joints frequently cause difficulty for a number of reasons: • Even for pedestrian bridges in which de-icing salts are not used. • As can be seen in Fig. An analysis complication is introduced by sag in the longer cables which has the effect of making the stiffness of the support provided non-linear. The deck can then be designed as a continuous beam with spring supports. the cross-sections of cable-stayed bridges are often composite with steel beams and concrete slabs. regardless of which alternative is chosen. steel box section decks are used to reduce the bridge self weight. there can be difficulty finding space to provide sufficient reinforcement to resist all of the types of structural action that take place in the halving joint. the joints tend to leak.4. It is also generally necessary to carry out a dynamic analysis for bridges of such slenderness. 1. . which promotes corrosion of the halving joint reinforcement. 1. becomes feasible when the total bridge length is in excess of about 150 m and is particularly economical for lengths in the 200–400 m range.Page 25 Fig. The concept of cable-stayed bridges is simple. The cables are only required to take tension and they provide support to the deck at frequent intervals. 1. The maximum main span achievable is increasing all the time. illustrated in Fig. • There are very high tensile and shear stresses at a point where the structural depth is relatively small. the current limit is of the order of 1000 m. For spans of moderate length.38 Reinforcement detail in halving joint However. for the longest spans.39.11 Cable-stayed bridges Cable-stayed construction.

It lends itself readily to staged construction with the cables being added as required to support successively placed segments of the deck. 1.12 Suspension bridges The very longest bridges in the world. are of the suspension type illustrated in Fig. They are more expensive to construct than cable-stayed bridges as they are not particularly suited to staged construction and the initial placing of the cables in position is onerous.5 Articulation Bridge design is often a compromise between the maintenance implications of providing joints and bearings and the reduction in stresses which results from the accommodation of deck movements. it is a necessary consideration for those which do. 1. shrinkage and thermal effects while keeping the structure stable. For these reasons. wind and accidental impact forces from errant vehicles. 1.39 Cable-stayed bridge The economy of the cable-stayed form stems from its ease of construction over inaccessible places. the problems of creep. 1. Horizontal forces are caused by braking and traction of vehicles. it is sometimes difficult to cater for the horizontal forces generated at the ends of the cables. While this clearly does not apply to bridges without joints or bearings. up to about 2000 m span. The articulation of a bridge is the scheme for accommodating movements due to creep.Page 26 Fig.4. the bridge must have the capacity to resist some relatively small forces while accommodating movements. segments are placed successively on alternate sides of the pylon.40.40 Suspension bridge . cablestayed construction is generally favoured except for the very longest bridges. In suspension bridges. 1. the main cables are in catenary and the deck hangs from them applying a substantially uniform loading. Thus. Further. Fig. As for balanced cantilever bridges. While the present trend is to provide ever fewer joints and bearings. shrinkage and thermal movement are still very real and no one form of construction is the best for all situations.

41(a) and (b) where the arrows indicate the direction in which movements are allowed. 1. guided sliding bearings are provided at C and.41(c). Fig. fixed—no horizontal translation allowed. (b) twospan skewed slab. To make the structure stable in the horizontal plane. 3. In many bridges. 1. For both bridges.41 Plan views showing articulation of typical bridges: (a) simply supported slab. When bridges are not very wide (less than about 5 m). also at E. The bearings usually allow free rotation but may or may not allow horizontal translation. A is a fixed bearing allowing no horizontal movement. At the same time they accommodate longitudinal movements. in the case of the two-span bridge. Two of the simplest forms of articulation are illustrated in Figs. such as those due to temperature changes. it may be possible to articulate ignoring transverse movements such as illustrated in Fig. These bearings are designed to resist horizontal forces such as the impact force due to an excessively high vehicle attempting to pass under the bridge. 2.Page 27 In-situ concrete bridges are generally supported on a finite number of bearings. free sliding—fully free to move horizontally. guided sliding—free to move horizontally in one direction only. They are generally of one of the following three types: 1. Free sliding bearings are provided elsewhere to accommodate transverse movements. (c) two-span bridge of small width . 1. a combination of the three types of bearing is provided.

42(a). 1. as B has 1 2 moved to B'. C' must move a corresponding distance to C″If the strain is the same in AB and . Further.Page 28 When bridges are not straight in plan. considerably shortening its life. Further. This can be seen in the simple example illustrated in Fig. |AC|. the movements would be accommodated by the arrangement of bearings illustrated in Fig. Uplift can also occur due to applied Fig. Uplift can occur at the acute corners of skewed bridges such as B and E in Fig. The orientation of bearings which accommodate this movement is illustrated in Fig. the magnitude of the movement |CC″ is proportional to the radial distance from the fixed point. Similarly. 1. However. 1. shrinkage or thermal movement results in a predominantly longitudinal effect which causes AB to shorten by δ to AB'. if unanticipated net uplift occurs. the net result is a movement along a line joining the fixed point. (d) articulation to accommodate movement . BC shortens by δto BC'. |. 1. 1. the orientation of movements tends to radiate outwards from the fixed bearing. Similarly for the curved bridge illustrated in plan in Fig.42(d). Creep. BC.42(b).42 Plan views showing articulation of crooked and curved bridges: (a) movement of crooked bridge. dust and other contaminants are likely to get into the bearing.42(c). (c) movement of curved bridge.41(b). 1. (b) articulation to accommodate movement. Bearings are generally incapable of resisting an upward ‘uplift’ force. A to C.

1 Sliding bearings Horizontal translational movements can be accommodated using two surfaces which are in contact but which have the capability to slide relative to one another. not only is there a risk of deterioration in the central bearing but. even with no skew and typical span lengths.44. as illustrated in Fig. 1.6 Bearings There are many types of bearings and the choice of which type to use depends on the forces and movements to be accommodated and on the maintenance implications. However. If this occurs. Only a limited number of the more commonly used types are described here. differential thermal effects can cause transverse bending which can result in uplift as illustrated in Fig. Further details of these and others are given by Lee (1994). it is better to provide two bearings only. Such a situation can be prevented by ensuring that the reaction at the central bearing due to permanent loading exceeds the uplift force due to temperature. 1. If this is not possible.43 Uplift of bearings due to traffic loading Fig. as it is not taking any load. 1.43. 1. 1. the two outer bearings must be designed to resist all of the load which renders the central bearing redundant. .Page 29 Fig.44 Uplift of bearing due to transverse bending caused by differential thermal effects loading in right bridges if the span lengths are significantly different.6. 1.

this form is also referred to as the spherical bearing. it is common in such systems for the lubricant to be squeezed out after a number of years. rotation is facilitated through some other mechanism and plane sliding surfaces are used which allow translation only. Thus. 1. In some combinations. guides are used such as illustrated in Fig. They are also used in combination with plane sliding surfaces to provide free sliding . consist of a metal cylinder containing an elastomer to which the force is applied by means of a metal piston.2 Pot bearings Pot bearings. 1. The elastomer effectively acts as a retained fluid and facilitates some rotation while preventing translation. Sliding bearings offer a frictional resistance to movement which is approximately proportional to the vertical force. When translation is to be allowed in one direction only. namely polytetrafluoroethylene (PTFE). Some bearings are lubricated. resulting in a reduced coefficient of friction. In other cases. They are frequently used for motorway bridges of moderate span. the sliding surfaces are spherical and allow rotation. 1. 1.Page 30 Fig. at which time the coefficient returns to the unlubricated value. it has been suggested that they be treated as wearing parts that eventually need to be replaced. Sliding bearings today generally consist of a stainless steel plate sliding on a PTFE-coated surface. such as illustrated in Fig.45.45 Guided sliding bearing (photograph courtesy of Ancon CCL) This is possible due to the availability of a material with a high durability and a very low coefficient of friction. However.46. pot bearings by themselves are commonly used at the point of fixity. Whether or not sliding bearings are lubricated. They can take many forms and are often used in combination with other forms of bearing.6.

Fig.g. e. 1. 1.47 Elastomeric bearing: (a) rotation.45). 1. They are considered to be quite durable except in highly corrosive environments and require little maintenance.47(a)) and translation by a shearing deformation (Fig.3 Elastomeric bearings When the forces to be resisted are not very high.46 Pot bearing bearings. 1. Elastomeric bearings accommodate rotation by deflecting more on one side than the other (Fig. such a combination can also be used to form a guided sliding bearing.6.Page 31 Fig. 1.47(b)). when bearings are provided under each beam in precast construction. By incorporating guides (Fig. 1. They are made from rubber and can be in a single layer (for relatively low loading) or in multiple layers separated by metal plates. (b) translation . elastomeric bearings can be a very economical alternative to sliding or pot bearings.

Page 32 1.49.1 Buried joint For movements of less than 10–20 mm.48.7 Joints While bearings in bridges can frequently be eliminated. and frequently leak. 1. This form has been successfully used for movements of up to 40 mm and is inexpensive to install or replace. can result in a minimum maintenance solution. movements will always occur with the result that joints will always be needed. 1.7. if designed well.48 Buried joint (after Lee (1994)) . Joints are notoriously problematic. 1. However. as illustrated in Fig. However. particularly in road bridges. A typical arrangement is illustrated in Fig. The material used to span the joint is important. joints buried beneath road surfacing are possible and. Fig. Even in integral construction. the movement must be accommodated at the end of the run-on slab.2 Asphaltic plug joint The asphaltic plug joint is similar to the buried joint in that the gap is protected by road surfacing. 1. allowing saltcontaminated water to wash over the substructures. in this case the road surfacing over the joint consists of a specially formulated flexible bitumen. the number of movement joints being used in bridge construction is decreasing with the philosophy that all of the associated maintenance implications should be concentrated into as few joints as possible. for larger gaps. 1. it is difficult to find a suitable material which carries the impact loading due to traffic across the gap while facilitating the necessary movement.7.

Page 33 Fig. Fig. is no longer favoured in many countries. It can accommodate movements of similar magnitude to the asphaltic plug joint but has a reputation for frequent failure and leakage. The nosings today are made up of cementitious or polyurethane binders instead of the epoxy mortars popular in the 1970s which were often found to deteriorate prematurely. 1. 1. illustrated in Fig.3 Nosing joint Very popular in the 1960s and 1970s. 1.7.49 Asphaltic plug joint (after Lee (1994)) 1. the nosing joint.50 Nosing joint (after Lee (1994)) .50.

1. Fig. It is generally agreed that the upstand and parapet are important and that they should be carried through from the bridge to corresponding upstands and parapets in the abutment wing walls as illustrated in Fig. The sun tends to shine directly on upstands while the main deck tends to remain in shadow (Fig. 1. Dublin) . particularly on what constitutes an aesthetically displeasing bridge.51. 1. However.51 Continuity of upstand and parapet (photograph courtesy of Roughan and O’Donovan Consulting Engineers. there is generally some common ground. This effect can be useful. 1. Some aspects of aesthetics are common to most bridges.8 Bridge aesthetics The art of bridge aesthetics is a subjective one with each designer having his/her own strongly held opinions. look better than others and attention to this can substantially improve the appearance of the structure. The depth of the upstand and the main deck relative to the span is a critical issue as will be seen in the following sections. The effect can be emphasised by casting the upstand in a whiter concrete or by casting the outer surface at an angle to the vertical as illustrated in Fig. Further details on these and longer-span bridge aesthetics can be found in the excellent book on the subject by Leonhardt (1984).53.52). The aesthetics of the more common shorter-span bridges are considered in this section. This serves to give a sense of continuity between the bridge and its setting as the eye can follow the line of the bridge from one end to the other. particularly if the designer wishes to draw attention away from an excessively deep main deck.Page 34 1. Certain bridge proportions in particular.

1.Page 35 Fig.52 Shading of main deck relative to upstand (photograph courtesy of Roughan and O’Donovan Consulting Engineers. In this example. the shape of the opening has a significant influence on the aesthetics.8.1 Single-span beam/slab/frame bridges of constant depth For very short-span bridges or culverts.54.53 Section through upstand 1. the shape of the opening is square (span equals height) and the abutment wing walls are large triangular . The abutment wing walls also play an important role as can be seen in the example of Fig. Dublin) Fig. 1. 1.

For such a bridge the main deck can be constructed of the same material (e. Three alternatives are illustrated in Fig.54. if aesthetics are important. 1. 1.54 Square opening with alternative span/upstand and span/main deck depth ratios: (a) 10 and 5 with brick wing walls. (d) 10 and 5 blocks. Ratios of 20 .g. a relatively deep main deck is often recommended such as one-fifth of the span. this clearly is a matter of opinion and also depends on the relative depths of the main deck and the upstand. (c) 20 and 10. A typical solution is illustrated in Fig. For a square opening.54(a) while leaving the main deck and upstand in concrete. it may be better to clad the wing walls in masonry as illustrated in Fig.54(b) with a span/upstand depth ratio of 20 and a span/main deck depth ratio of 5. 1. concrete) as the abutment walls. 1. (b) 20 and 5.Page 36 Fig. However. it may be difficult to get a good finish with in-situ concrete and. However.

1.55 Rectangular opening with small wing walls: (a) slender deck and deep upstand. where people and traffic are close to the structure which is large relative to their size.2 Multiple spans The relative span lengths in multi-span bridges have a significant effect on the appearance.Page 37 Fig. an even more slender deck is favoured. The heavier looking alternative illustrated in Fig. The bridge illustrated is probably typical with a main span/upstand depth . For aesthetic reasons.) A structure with similar proportions looks much better in Fig.55(b) has ratios of 60 and 10. 1. it is common practice in three-span construction to have the centre span greater than the side spans. (b) deep deck and slender upstand and 10 are illustrated in Fig.54(c) for upstand and main deck respectively. This can be convenient as the principal obstruction to be spanned is often in the central part of the bridge. This is illustrated in Fig.56(a). span/upstand depth ratios of 20 and a span/main deck depth ratio of 10 is often recommended. (In this structure.55(a) with a span/upstand depth ratio of 40 and a span/main deck depth ratio of 20. It can be seen that the upstand appears too thin and/or the deck too deep. 1. this proportioning also tends to bring the relative dimensions of the rectangular openings closer. while ratios of 10 and 5 are illustrated in Fig. a parapet wall is integral with the upstand making it look deeper than necessary. 1. typically by 25–35% as illustrated in Fig. 1. 1. as illustrated in the figure. 1. 1. For rectangular openings with less pronounced wing walls.54(d) and (a).57.56(b) as it is smaller and is more likely to be viewed from a distance. a much more slender deck is desirable. For a 2×1 rectangular opening with wing walls of similar size. When the ground level is lower at the centre. Typical ratios are illustrated in Fig. Leonhardt points out that scale is important as well as proportion.8. 1. which has a good aesthetic effect.

Page 38 Fig. (b) curved alignment achieved using two curves of differing radius. 1. 1.56 The influence of scale on appearance: (a) large structure near the viewer looks heavy. (b) small structure remote from the viewer looks better than in(a) Fig. (c) curved haunches .57 Three-span bridge with good proportions Fig. 1.58 Variable depth bridges: (a) straight haunches.

curved decks are strongly favoured over straight ones. However. 1. This greatly complicates the detailing but makes for an efficient light structure and tends to look very well. they are not as aesthetically pleasing as a curved profile. 1. As for single-span bridges. . where the depth is increased at the points of maximum (hogging) moment. both in terms of shuttering and reinforcement details.Page 39 ratio of 40 and a span/deck depth ratio of 20. Straight haunches are considerably cheaper than curved ones. When a road or rail alignment is straight. the upstand is continuous from end to end. An open parapet is also used in the bridge of Fig.58(b) and (c). 1. effectively tying the bridge together. Varying the depth of bridges allows the depth to be increased at points of maximum moment. straight haunches are possible as illustrated in Fig.58(a). illustrated in Figs. When alignments are curved.57 to increase the apparent slenderness of the bridge.

particularly in frame and arch bridges.1 Introduction For bridges. Some of these are treated in greater detail in the following sections as indicated in the third column of the table. For example. Thermal changes can have significant effects. Another source of loading is earth pressure on substructures. Dead and superimposed dead loads consist of permanent gravity forces due to structural elements and other permanent items such as parapets and road surfacing. Three codes of practice are referred to in this chapter. wind and imposed traffic loading. exceptional loads (such as snow) and construction loads. The predominant effect is the vertical gravity loading including the effect of impact. which . Imposed traffic loads consist of those forces induced by road or rail vehicles on the bridge. For example. Both the British standard and the AASHTO treatments of temperature are somewhat tedious in that different load ‘combinations’ must be considered. it is often necessary to consider phenomena which would normally be ignored in buildings. the British Department of Transport standard BD37/88 (1988). the draft Eurocode EC1 (1995) and the American standard AASHTO (1995). This is considered in Chapter 4 in the context of integral bridges. The various types of loading which need to be considered are summarised in Table 2.1. Where footpaths or cycle tracks have been provided. the gravity loading due to pedestrians/cyclists can be significant. horizontal loading due to braking/traction and centrifugal effects in curved bridges must also be considered. Other types of loading which may occur but which are not considered here are the effects of shrinkage and creep. namely. However. effects such as differential settlement of supports frequently need to be considered by bridge designers while generally being ignored by designers of building structures.Page 40 Chapter 2 Bridge loading 2. These and other more common forms of bridge loading are considered in this chapter. An alternative. the AASHTO standard specifies one combination which includes the effects of temperature.

Differential settlement 7.2 2.3 – 2. Prestress Description Gravity loading due to structural parts of bridge Gravity loading due to non-structural parts of bridge Loading due to road or rail vehicles Gravity loading due to non-vehicular traffic Uniform and differential changes in temperature Relative settlement of supporting foundations Impact loading due to collision with errant vehicles Effect of bridge vibration Horizontal loading due to wind on parapets. No geotechnical guidance is given in either BD37/88 or AASHTO on how bridges should be analysed to determine the effect of this phenomenon.4 – 2.e. In pedestrian bridges.2 2. Differential settlement of supports can induce significant bending in continuous beam or slab bridges. vehicles and the bridge itself Effect of prestress on indeterminate bridges Section 2. Dead 2. In practice. Impact 8. Thermal 6. Dynamic effects 9.5 2. in AASHTO and in the draft Eurocode. where the natural frequency of the bridge is at a level which can be excited by traffic or wind.Page 41 Table 2. The draft Eurocode treats temperature in a manner similar to other load types and applies the same method of combining loads as is used throughout EC1. Vibration is generally only significant in particularly slender bridges.6 – 2. excludes some thermal and wind effects but includes a higher traffic loading. it is recommended that a combined model of the bridge structure and the supporting soil be used to determine the stresses induced by settlement. . Similarly high levels of impact loading are in use in many European national standards. The load specified in the UK has increased dramatically in recent years. The loading due to impact from collisions with errant vehicles can be quite significant for some bridge elements. recommends that the process of soil/structure interaction be taken into consideration for accurate analysis of problems of this type.1 Summary of bridge loads Load type 1. For example. this usually only includes pedestrian bridges and long-span road and rail bridges. Imposed traffic 4. Superimposed dead 3. as will be demonstrated in Chapter 3. it should be ensured that the natural frequency of the bridge is not close to that of walking or jogging pedestrians. Wind 10. i.7 must also be considered. Pedestrian and cycle track 5. EC7 (1994). The calculation is complicated by the use of different factors of safety and the specification of different design limits for the different combinations. The draft Eurocode on Geotechnical Design. the service stresses permitted in prestressed concrete bridges are higher for the combinations in BD37/88 which include temperature than for combinations which do not.

Thus dead load should be estimated as accurately as possible rather than simply rounded up. Thus. superimposed dead load tends to be assigned higher factors of safety than dead load. lightweight concrete has been successfully used in order to reduce the dead load. it is probable in many cases that the parapet will need to be replaced during the life of the bridge and the new parapet could easily be heavier than the original one. in indeterminate bridges it is necessary to analyse to determine the effect of prestress so it is often convenient to treat prestress as a form of loading.Page 42 In addition to its ability to induce vibration in bridges. However. This is particularly true of long-span bridges. Prestress is not a load as such but a means by which applied loads are resisted. It is simply calculated as the product of volume and material density. Superimposed dead load is the gravity load of non-structural parts of the bridge. such superimposed dead loading is particularly prone to increases during the bridge lifetime. dead or permanent loading is the gravity loading due to the structure and other items permanently attached to it. In such cases. In some cases. concrete or composite steel beams with concrete slabs are the usual materials. There is clearly always going to be a parapet so it is a permanent source of loading. wind can induce static horizontal forces on bridges. For shorter spans. The most notable item of superimposed dead load is the road pavement or surfacing. a particularly high load factor is applied to road pavement. The critical load case generally occurs when a train of high vehicles are present on the bridge resulting in a large vertical projected area. Such items are long term but might be changed during the lifetime of the structure. For this reason. For prestressed concrete bridges. it is important to remember that an overestimate of the dead load can result in excessive stresses due to prestress. there is a subdivision of this into dead loading and superimposed dead loading. The former is the gravity loading of all structural elements. In BD37/88. The methods used are very similar to those used to determine the effects of temperature changes. steel or aluminium decks can become economically viable due to their high strength-to-weight ratio. An example of superimposed dead load is the weight of the parapet. Bridges are unusual among structures in that a high proportion of the total loading is attributable to dead and superimposed dead load. .2 Dead and superimposed dead loading For general and building structures. Because of such uncertainty. Wind tends not to be critical for typical road bridges that are relatively wide but can be significant in elevated railway viaducts when the vertical projected surface area is large relative to the bridge width. 2. More accurate (and complex) methods are also specified for cases where wind has a significant effect. Both the British and the American standards specify a simple conservative design wind loading intensity which can be safely used in most cases. It is not unusual for road pavements to get progressively thicker over a number of years as each new surfacing is simply laid on top of the one before it. However.

Most codes allow a reduction for long footpaths. WIM technology has resulted in a great increase in the availability of truck weight statistics and codes of practice are being revised to reflect the new data. The AASHTO code also specifies notional lanes of fixed width. Bridge traffic loading is often governed by trucks whose weights are substantially in excess of the legal maximum. sampling was carried out by taking trucks from the traffic stream and weighing them statically on weighbridges. The British Standard on the other hand (for carriageway widths in excess of 5 m) allows the lane width to vary within bands in order to get an integer number of lanes without having any remaining area.g.1 Imposed loading due to road traffic While some truck-weighing campaigns have been carried out in the past. a reduced intensity is allowed by some codes to reflect the reduced probability of both traffic and pedestrian loading reaching extreme values simultaneously. a truck of specified dimensions and axle weights must be considered. is known as the ‘remaining area’. more importantly. In the Eurocode. its importance should not be underestimated. In recent years the situation has improved considerably with the advent of weigh-in-motion (WIM) technology which allows all trucks passing a sensor to be weighed while they travel at full highway speed. While pedestrian/cycle traffic loading on bridges is not difficult to calculate. Alternatively. Bridge codes commonly specify a basic intensity for pedestrian loading (e. A dynamic factor is applied to the truck to allow for the increased stresses which result from the sudden arrival of a speeding vehicle on a bridge. the imposed traffic loading specified by AASHTO is considerably less onerous than that specified by both BD37/88 and the Eurocode. there tends to be a bias as drivers of illegally overloaded trucks quickly learn that weighing is taking place and take steps to avoid that point on the road. The AASHTO code specifies a traffic lane loading which consists of a knife-edge load plus a uniformly distributed lane loading. Normal traffic loading or Highway A (HA) represents an extreme . each 3 m wide. the quantity of data collected is relatively small but. there has been a scarcity of good unbiased data on road traffic loading until recent years. Vehicular and rail traffic are considered in subsections below. 2. Bridge traffic loading is applied to notional lanes which are independent of the actual lanes delineated on the road. In the past.Page 43 2. rail or pedestrian/cycle or indeed any combination of these. In general. There are two problems with this as a means of collecting statistics on truck weights. BD37/88 and the draft Eurocode specify two types of traffic loading.3. The outstanding road width between kerbs. 5 kN/m 2 in the draft Eurocode and the British standard and 4 kN/m 2 in the American code). When a structural element supports both pedestrian and traffic loading.3 Imposed traffic loading Bridge traffic can be vehicular. In the first place. the road width is divided into a number of notional lanes. after removing these lanes. ‘normal’ and ‘abnormal’.

While there are a number of factors which can vary between road classes and between countries. a typical one being full HA in Lanes 1 and 2 combined with 60% of full HA in the other lanes as illustrated in Fig. In addition.Page 44 combination of overloaded trucks of normal dimensions. 19. This consists of an exceptionally heavy vehicle of the type which is only allowed to travel under licence from the road/bridge authority. Particularly on roads with rough surfaces. On the other hand. 2 and 3. (b) British standard HA loading . Illustrated in Fig. This could be a traffic jam involving a convoy of very heavy trucks as would tend to govern for a long bridge. the abnormal load in BD37/88 is Fig. 2. there can be a considerable dynamic component of truck loading which is deemed to be included in the specified normal load. ‘full’ HA lane loading consists of a uniform loading whose intensity varies with the loaded length and a ‘knife edge’ concentrated loading of 120 kN. 2. 400 kN and 200 kN for Lanes 1.6.6. or 29. While there are exceptions. the standard combination is a load intensity of 9 kN/m2 in Lane No. It is scaled in gross ‘units’ of 40 kN so that a minor road bridge can be designed.2. to take 25 units (a 1000 kN vehicle) while a highway bridge can be designed for 45 units (a 1800 kN vehicle). The AASHTO code allows similar reductions in lane loading for multi-lane bridges to account for the reduced probability of extreme loading in many lanes simultaneously.1 ‘Normal’ road traffic loading: (a) Eurocode normal loading. respectively. The possibility of abnormal or Highway B (HB) loading must also be considered in British and Eurocode designs. Different countries have different classes of abnormal vehicle for which bridges must be designed.1(b). 1 and 2.6. for example. In BD37/88. Combinations of normal traffic and an abnormal vehicle must be considered in bridge design. Eurocode normal loading consists of uniform loading and a tandem of four wheels in each lane as illustrated in Fig. The four wheels of the tandems together weigh 600 kN. In the British standard. it could be a chance occurrence of two overloaded moving trucks near the centre of a short bridge at the same time. A large number of alternative abnormal vehicle classifications are specified in the draft Eurocode from which individual countries can select combinations for which roads of specified classes are to be designed. the vehicle is known as the Highway B or HB vehicle. 24. 2.6 m. only one abnormal vehicle is specified but it may have a length of 9.5 kN/m 2 elsewhere. 2.1(a). For bridges with many notional lanes.6. 14. there is uniform loading in the remaining area. a number of possibilities must be considered.

3. On bridges. However. known as Railway Upper (RU). This effect can be minimised by incorporating transition zones at the ends of the bridge with ballast of reducing depth. bridges throughout a rail network are generally designed for the same normal load model.2 British standard abnormal (HB) vehicle consisting of 16 wheel loads of F=2. The other disadvantage to the use of track slabs depends on the method used to maintain and replace ballast. In BD37/88. 2. The weights of railway carriages can be much better controlled than those of road vehicles with the result that different load models are possible depending on the railway line on which the bridge is located. On passenger transit ‘light rail’ systems. the normal load model. A standard light rail load model. If this is done using automatic . This follows from the fact that the train can generally be assumed to remain on the tracks. is specified in the British standard. The static loads specified for the design of railway bridges must be increased to take account of the dynamic effect of carriages arriving suddenly on the bridge.5 kN per unit generally taken to replace the normal loading throughout the length of the vehicle and for a distance of 25 m before and after it. This factor is a function of the permissible train speed and of the natural frequency of the bridge. In addition. Railway tracks on grade are generally laid on ballast. However. Railway Lower (RL). Normal load is placed throughout the remainder of the lane and in the other lanes. the Eurocode provides for an alternative abnormal load model. an additional vertical dynamic load is induced by the change from the relatively ‘soft’ ballast support to the relatively hard track slab. tracks can be laid on a concrete ‘track slab’ or the bridge can be designed to carry ballast and the track laid on this. 2.2 Imposed loading due to rail traffic The modelling of railway loading is considerably less onerous than that of road traffic loading as the transverse location of the load is specified. less onerous load models can be applied. When used. less stringent models have been used for the design of bridges on some light rail networks. there are some aspects of traffic loading that are specific to railway bridges which must be considered. The standard Eurocode normal load model consists of four vertical point loads at 1. is similar in format. There are two disadvantages to the use of track slabs.Page 45 Fig. However.6 m intervals of magnitude 250 kN each and uniform loading of intensity 80 kN/m both before and after them.

Both the draft Eurocode and the British standard specify contour plots of maximum and minimum ambient temperature which can be used to determine the range of temperature for a particular bridge site. high early temperatures can result from the hydration of cement. etc.4 Thermal loading There are two thermal effects which can induce stresses in bridges. this can generate significant axial force. Horizontal loading due to braking and traction is more important in railway bridges than in road bridges as the complete train can brake or accelerate at once. It is important in bridge construction to establish a baseline for the calculation of uniform temperature effects. In concrete bridges. those made from precast concrete or steel will have temperatures closer to ambient during construction. . If the top of a beam heats up relative to the bottom. Another aspect of loading specific to railway bridges is the rocking effect. if it is restrained from doing so. Resulting stresses in the period after construction will tend to be relieved by creep although little reliable guidance is available on how this might be allowed for in design. The second effect is that due to differential changes in temperature. Completion of the structural form could be the process of setting the bearings or the making of a frame bridge integral. beam and slab. It is assumed for design purposes that more than half of the load (about 55%) can be applied to one rail while the remainder (about 45%) is applied to the other. Longitudinal horizontal loading in bridges can affect the design of bearings and can generate bending moment in substructures and throughout frame bridges. it is statistically much less likely.Page 46 equipment.e. i.). a considerable delay can be caused by the need to remove the equipment at the start of the bridge and to reinstall it at the end. bending moment and shear force are generated. In ‘moderate’ climates. The British Standard and the draft Eurocode specify no baseline. the temperature of the bridge at the time of construction. 2. particularly for concrete with high cement contents. metal bridges must be designed for temperatures in the range − °C to 49 °C and concrete 18 bridges for temperatures in the range − °C to 27 °C. The AASHTO code specifies a baseline temperature equal to the mean ambient in the day preceding completion of the bridge. This can generate torsion in the bridge. such as in an arch or a frame bridge. it tends to bend. Different figures are specified for 12 ‘cold’ climates. bending moment and shear. It is possible to control this baseline by specifying the permissible range of temperature in the structure at the time of completion of the structural form. While it is possible in road bridges for all vehicles to brake at once. If restrained. Uniform changes in temperature result from periods of hot or cold weather in which the entire depth of the deck undergoes an increase or decrease in temperature. The American approach is much simpler. Unlike in-situ concrete bridges. The first is a uniform temperature change which results in an axial expansion or contraction. The difference between ambient temperature and the effective temperature within a bridge depends on the thickness of surfacing and on the form of construction (whether solid slab.

integral bridges undergo repeated expansions and contractions due to daily or seasonal temperature fluctuations. There is an ‘equivalent’ axial force and bending moment associated with any distribution of temperature. As for uniform changes in temperature.Page 47 As is discussed in Chapter 4. Transverse temperature differences can occur when one face of a superstructure is subjected to direct sun while the opposite side is in the shade. the baseline temperature is clearly a mean temperature which relates to the density of the adjacent soil.3 is subjected to the differential increase in temperature shown. This is resolved into axial. Methods of analysing to determine the effects of the equivalent loads are described in Chapter 3. The reverse effect tends to take place in the evening when the deck is warm in the middle but is cooling down at the top and bottom surfaces. i.1: Differential temperature I The bridge beam illustrated in Fig. The applied temperature distribution is converted into the equivalent stress distribution of Fig. such as in the morning when the sun shines on the top of the bridge heating it up faster than the interior. bending and residual distributions as will be illustrated in the following examples. In such cases. Example 2. The effects of both uniform and differential temperature changes can be determined using the method of ‘equivalent loads’. The equivalent axial force can readily be calculated as the sum of products of stress and area: . the magnitude of the resulting thermal stresses can be significantly overestimated.4(a) by multiplying by the coefficient of thermal expansion and the modulus of elasticity. bridges are subjected to differential temperature changes on a daily basis. no such distribution is typically specified in codes. However. Cracking of reinforced concrete members reduces the effective cross-sectional area and second moment of area. the implication being that the distributions specified represent the differences between the baseline and the expected extremes.e. one corresponding to the heating-up period and one corresponding to the cooling-down period. This effect can be particularly significant when the depth of the superstructure is great. this causes the backfill behind the abutments to compact to an equilibrium density. It is required to determine the effects of the temperature change if it is simply supported on one fixed and one sliding bearing. the baseline temperature distribution is important. A distribution of stress is calculated corresponding to the specified change in temperature. 2. If cracking is ignored. These distributions can be resolved into axial. The coefficient of thermal expansion is 6 12×10− and the modulus of elasticity is 35000 N/mm2. In addition to uniform changes in temperature. The corresponding forces and moments are then readily calculated. that distribution which exists when the structural material first sets. 2. bending and residual effects as will be illustrated in the following examples. Two distributions of differential temperature are specified in some codes. After some time.

4(b). Thus. this beam is supported on a sliding bearing at one end and is therefore free to expand.81/35000=23×10− . 2. 2. there is in fact no axial stress but a strain of magnitude 6 0. (c) bending component. 2.4 Components of imposed stress distribution: (a) total distribution. (d) residual stress distribution This corresponds to a uniform axial stress of 579600/(600× 1200)=0. (b) axial component.Page 48 Fig. The equivalent bending moment is found by taking moments about the centroid (positive sag): The corresponding extreme fibre stresses are: .3 Beam subject to differential temperature change Fig. However.81 N/mm2 as illustrated in Fig.

5(a).4(a). illustrated in Fig.2 Calculation of force Block a b c d Details 3α (2. the equivalent axial force. 2. 2. it is free to rotate and there is in fact no such stress. The difference between the applied stress distribution and that which results in axial and bending strains is trapped in the section and is known as the residual stress distribution. 2. The coefficient of thermal expansion is α the modulus of elasticity is E. Example 2.11/35 000=±32×10− . 2.5 Beam and slab bridge subject to differential temperature: (a) cross-section. Instead.100α E Total force= 3. (b) imposed distribution of temperature Table 2. bending moment and residual stresses are required due to the differential temperature increases shown in Fig. 2.890α E 0.15)= E Force 1.5(b).4(b) and (c) from 2. As the beam is simply supported.Page 49 as illustrated in Fig.220α E . 2. It is found simply by subtracting Figs.4(d).4(c). and Fig.150α E 0. a strain distribution is generated which varies linearly in the range 6 ±1.2: Differential temperature II For the beam and slab bridge illustrated in Fig.4×0.080α E 1.

(b) corresponding imposed stress distribution By summing moments of area.6 Division of section into blocks: (a) cross-section.3 Calculation of moment Block a b c d Details Moment − 0.062α E Total moment= − 0.2.Page 50 Fig.718α corresponds to E stresses (positive tension) of: Table 2. The bridge is split into two halves.506α E − 0. The total tensile force per half is then found by summing the products of stress and area for each block as shown in Table 2.22α E E/0.60α E. 2. 0.262α E − 0.718α E . the centroid of the bridge is found to be. Similarly moment is calculated as the sum of products of stress.012α E 0. The total force of 3. The temperature distribution is converted into a stress distribution in Fig.064 86 m 4.70= 4. each of area. area and distance from the centroid as outlined in Table 2. 2. 0. below the 2 top fibre.70 m and second moment of area.3 (positive sag).6 and divided into rectangular and triangular blocks. The total moment of − 0.22α corresponds to an axial tension of 3.

2. The simple case illustrated in Fig. Peq . the collision of a vehicle with a bridge is highly non-linear. codes of practice often greatly simplify the procedure by specifying equivalent static forces.7 Resolution of stress distribution into axial.7(b) and (c) from the applied distribution of Fig.2) Fig.5 Impact loading Most bridge analysis is based on static linear elastic principles. (2. travelling at a velocity. This section considers the basis on which these forces are derived. To overcome the resulting complications. v.7(a).Page 51 Fig.7. 2. m. 2. 2. 2. bending and residual components: (a) total distribution. which causes a deflection.8 Impact of undeformable sphere with spring . K. 2. However. 2. (d) residual stress distribution Hence the applied stress distribution can be resolved as illustrated in Fig. (c) bending component. A static force. collides with a spring of stiffness. The residual distribution is found by subtracting the distributions of Figs.1) On impact. this is converted into strain energy in the spring. Δ generates a strain energy of: . An undeformable sphere of mass. The kinetic energy of the sphere is: (2.8 is considered first. (b) axial component.

in the draft Eurocode.3) For a spring of stiffness. An impact force is also specified for a derailed train colliding with a pier.3) gives an alternative expression for P eq: (2. to consider collision of trains with bridge decks overhead.6 Dynamic effects Vibration can be a problem in slender bridges where the natural frequency is at a level which can be excited by wind or traffic. it can be used as a basis for determining equivalent static forces.5) Substituting for E k in this equation gives the equivalent force in terms of mass and velocity: (2. This is frequently done by ‘lumping’ the distributed mass of a bridge at a finite number of nodes. For these reasons. For example. 2. E k. mass has a significant effect on the response of the structure to a given load and computer models must incorporate a representation of mass as well as stiffness. K. Such a possibility can be investigated by means of a dynamic analysis. It follows from these assumptions that a vehicle with kinetic energy. Further. the simply supported beam bridge of Fig. the draft Eurocode specifies that the impact force due to a truck be applied at a specified height above the road surface.6) While this is a very simple case. The mechanics of a collision between a vehicle and a structure are quite complex. It is not necessary.9(a) could be represented by the lumped mass .4) Substituting for Δin equation (2.Page 52 Hence the equivalent static force is: (2. will generate the equivalent force given by equation (2. because only the top of the vehicle is likely to impact on the bridge. In dynamics. A table of design static forces is specified in the draft Eurocode based on the expected masses and velocities of trucks on roads of various class. However.5) on the outer surface of a structural element. On bridge piers. a small difference in the impact location or the impact angle can result in a substantial change in the effect. 2. there is a possibility that trucks passing underneath will collide with the bridge deck. On bridges over road carriageways. a substantial reduction factor applies. the situation is simplified by treating the vehicle as undeformable and the structural element as a spring. Similar equivalent static loadings are specified in the AASHTO standard and in BD37/88. a force P eq generates a deflection: (2.

by wind. it may vibrate at one of these frequencies. (b) typical second mode shape . 2.9(b). 2.Page 53 model of Fig. 2. (b) lumped mass model Fig. If it can be shown that the natural frequencies of the bridge are not close to the frequency of all expected sources of excitation. However.9 Idealisation of beam for dynamic analysis: (a) original beam. Common forms of excitation are truck vibration. In such an analysis. the source of excitation of the bridge is not considered and only the natural frequencies and mode shapes are determined. 2. Even when the frequencies are not close. and jogging or walking pedestrians. The shape of the structure during such vibration is known as the mode shape.10. say. the interaction of the Fig. the equivalent static loads specified in codes of practice take account of this phenomenon and incorporate a ‘dynamic amplification’ factor. as illustrated in Fig. In the simplest form of dynamic analysis. there may be no need for further dynamic analysis. If the excitation frequency is close to one of the natural frequencies of the bridge. Dynamic amplification can be defined as the ratio of the actual stress to that due to the corresponding static load.9 is excited. 2. If the bridge of Fig. All structures have a number of natural frequencies at which they tend to vibrate. as would often be the case for a road bridge excited by traffic.10 Mode shapes of simply supported beam: (a) typical first mode shape. wind. further analysis is required to determine the dynamic amplification in what is known as a ‘forced vibration’ analysis. a suddenly applied load generates significantly more stress than a statically applied one.

Examples of analysis using equivalent prestress loads are given in Chapter 3.Page 54 applied loading and the bridge is taken into account. the draft Eurocode specifies an equivalent static force.11(a). Furthermore.1 Equivalent loads and linear transformation The equivalent loading due to prestress can generally be found by simple equilibrium of forces. equilibrium of vertical forces gives an upward force at B of: As the angle. For example. For such cases. Such a method is only necessary in the case of indeterminate bridges. even for simply supported slab or beam-and-slab bridges.5. the concept of linear transformation is also introduced. In the case of road traffic. Finally. In this section. In such cases. The combined effect of a number of tendons can then be found by simply combining the loadings. the applied loading is a truck or trucks of considerable mass. However. the trucks are moving so the location of their masses are changing with time. this can be approximated as: . the code allows for the carrying out of a dynamic analysis. vibrating on their own tyres and suspensions. 2. θis generally small. methods will be given for the calculation of their magnitudes. In Section 2. Such a loading can readily be specified in a computer model and the maximum distribution of stress determined. for the externally prestressed bridge illustrated in Fig.7. 2. the equivalent force is assumed to increase from zero to its full value over a very short time (measured in milliseconds) and to maintain a constant value for a further short time (of the order of 200 ms). Such an analysis is currently only possible with specialist computer programs which incorporate the complexities of truck rocking and bouncing motions and the variations in truck dynamic characteristics which may be expected in typical traffic.7) It also follows from the small angle that the horizontal force is P cosθ P. For a qualitative understanding of the effects of prestress. 2. (2. it is often necessary to analyse to determine the degree to which prestressing of one member affects others. prestress can be handled using the method of equivalent loads. design for the impact of vehicles colliding with bridges was discussed. Whether the bridge consists of beams or a slab.7 Prestress loading While prestress is not in fact a loading as much as a means of resisting load. as the ≈ forces are eccentric to the centroid at the ends. equivalent loadings can be found for individual tendons. As an alternative. Like temperature. there are concentrated . it is often convenient to treat it as a loading for analysis purposes.

11 Prestressed concrete beam with external post-tensioning: (a) elevation showing tendon. 2. 2. A small segment Fig.Page 55 Fig. It can be shown that the equivalent loading due to prestress is always self-equilibrating.11(b).12 Segment of parabolically profiled tendon: (a) elevation. 2. (b) equivalent loading due to prestress moments there of magnitude (Pcosθ 2 ≈ 2. (b) equivalent loading . A parabolically profiled prestressing tendon generates a uniform loading which again can be quantified using equilibrium of vertical forces. Hence the total equivalent loading due to )e Pe prestress is as illustrated in Fig.

11) The equivalent loads on the segment are illustrated in Fig. At point 1.Page 56 of such a profile is illustrated in Fig.13 is prestressed using a single parabolic tendon set out according to the equation: (2. Similarly the vertical component of force at 2 is: (2. there is an upward vertical component of the prestress force of: (2.12(b). The intensity of uniform loading on this segment is: (2. This force is upwards when the slope is positive. Example 2. It is required to determine the equivalent loading due to prestress.8) As the angles are small: (2. .3: Parabolic profile The beam illustrated in Fig.12) where s is referred to as the sag in the tendon over length l as indicated in the figure.9) where x1 is the X coordinate at point 1. 2. 2.12(a). 2.10) where F 2 is downwards when the slope is positive.

the slope is negative and the force is downwards of magnitude P(−B +e A+4s)/l.13) As θ is small: A For a positive slope. e The slope at B is calculated similarly: .12) gives: (2. the equivalent point load at A would be upwards and of magnitude P(eB− A− e 4s)/l. However. (b) equivalent loading due to prestress Differentiating equation (2.13 Beam with parabolic tendon profile: (a) elevation. in this case.Page 57 Fig. 2.

increases the (sagging) moment due to prestress. A profile such as that illustrated in Fig. As was seen above. they do not significantly affect the distribution of bending moment induced by prestress.14). It is required to determine an amendment to the profile to increase the stress at this point without increasing the prestress force. Hence. in fact.14.13(b). the change only results in adjustments to the equivalent point loads at A and B and to the equivalent loading near B.14(b). which increases the compressive stress at the top fibre. This is because the eccentricity at B has been increased without increasing the tendon sag in the spans.4: Qualitative profile design A prestressed concrete slab bridge is to be reinforced with 10 post-tensioned tendons. unaffected by eccentricity at the ends of the span. This increase in tendon eccentricity. as illustrated in Fig. Such a uniform upward loading in a two-span beam generates sagging moment at the interior support which has the desired effect of increasing the top-fibre stress there. illustrated in Fig. The intensity of uniform loading is given by equation (2. in an indeterminate structure. 2.14(c) where the profile is lowered in AB and BC while maintaining its position at the support points. e A and eB while keeping the sag. does little to increase the compressive stress at the top fibre at that point. This has the effect of increasing the tendon sag which increases the intensity of equivalent uniform loading. In a determinate structure. As these forces are at or near supports.13): (2. A more appropriate revision is illustrated in Fig. e. s.14(a). 2.14) This too is illustrated in the figure. 2. The preliminary profile for the tendons. However. 2. the equivalent point loads are as illustrated in Fig.13(a) can be adjusted by changing the end eccentricities. In the structure of Fig. . Such an adjustment is known as a linear transformation and will have no effect on the intensity of equivalent uniform loading as can be seen from equation (2. 2. unchanged. Example 2. Thus. stress at the top fibre can be increased by moving the prestressing tendon upwards to increase the eccentricity locally. results in insufficient compressive stress in the top fibres of the bridge at B. this force is downwards when positive.11) where the second derivative is found by differentiating equation (2. Pe.3 illustrates the fact that the intensity of equivalent uniform loading due to a parabolic tendon profile is independent of the end eccentricities. the response of a structure to such changes is not so readily predictable. 2. the equivalent uniform loading due to prestress is a function only of the sag and is.Page 58 As B is on the right-hand side. increasing the eccentricity locally at B without changing the sags. Example 2. This phenomenon is particularly useful for understanding the effect of prestressing in continuous beams with profiles that vary parabolically in each span.

2. (b) raising of profile at B by linear transformation. This can be seen in the following example.15(a). The intensities of loading are found from equation (2. It is required to determine the equivalent loading due to prestress assuming that the prestress force is constant throughout the length of the bridge.14 Adjustment of tendon profile: (a) original profile. For the first parabola: . (c) lowering of profile in AB and BC to increase sag Most prestressing tendons are made up of a series of lines and parabolas and the equivalent loading consists of a series of point forces and segments of uniform loading. Example 2. half of which is illustrated in Fig.Page 59 Fig. 2.11).5: Tendon with constant prestress force A three-span bridge is post-tensioned using a five-parabola symmetrical profile.

This is necessary to ensure that the tendon does not generate concentrated forces at these points. 2.15 Tendon profile for Example 2.15(b). the intensities of loading in the second and third parabolas are respectively: and The point load at the end support is the vertical component of the prestress force.Page 60 Fig. 2.16(a) . (b) equivalent loading due to prestress Similarly. 2. Note that in selecting the profile. This is illustrated in Fig. prestress forces are not constant through the length of bridges because of friction losses. Verifying that these forces are in equilibrium can be a useful check on the computations. it has been ensured that the parabolas are tangent to one another at the points where they meet. Differentiating the equation for the parabola gives the slope.2 Prestress losses In practical post-tensioned construction.5: (a) partial elevation showing segments of parabola.7. 2. from which the force is found to be: All of the equivalent loads due to prestress are illustrated in Fig.

Example 2.6: Tendon with varying prestress force The post-tensioning tendon of Example 2.Page 61 Fig.17. the difference between prestress forces at adjacent points is generally not very large. Therefore.9)–(2. However. (b) equivalent loading (Pav=(P1+P2 )/2) where the forces at points 1 and 2 are different. A useful method of checking the equivalent loads is to apply them in the analysis of a determinate beam. The use of equivalent loads which do not satisfy equilibrium can result in significant errors in the calculated distribution of prestress moment. The resulting loading is illustrated in Fig. It will be seen in Example 2. In such a case. 2.5. 2. The bridge is post-tensioned from both ends with the result that the prestressing forces vary symmetrically about the centre.5 is subject to friction losses which result in the prestress forces presented in Fig. 2.16 Equivalent loading due to varying prestress force: (a) segment of beam and tendon. a sensible approach to the derivation of equivalent prestress loading is to start by substituting the average prestress force for P in equations (2. It is required to determine the equivalent loading due to prestress taking account of the loss of force.6 that this equivalent loading satisfies equilibrium of forces and moments.16(b).11). . the moment due to the equivalent loading should be equal to the product of prestress force and eccentricity at all points. The eccentricities given in this figure have been calculated from the equations for each parabola given in Example 2.

5 but using average prestress forces. point loads must be applied at the end of each segment in accordance with Fig. x=0. the equation for the parabola is: At A. 2.16(b).Page 62 Fig.17 Tendon profile showing varying prestress force (in kN) and eccentricity (in m) With reference to Example 2. 2.1322 and the upward force is: . the equivalent intensities of uniform loading are: In addition. In segment AB. the slope is − 0.

2. 2.18 Equivalent loading due to prestress: (a) loading on each segment. 2. . 2.18(b). together with the other equivalent uniform loads. At B.19(b) where s is distance along the centroid. the slope of the profile is: giving a downward force at the right end of magnitude: The corresponding point load components for the other segments of parabola are calculated similarly and are presented.18(a). 2. This clearly affects the eccentricity and hence the moment due to prestress.19(a). In such a beam. A segment of beam with a curved centroid is illustrated in Fig. the prestress forces are resolved parallel and perpendicular to the centroid and the eccentricity is measured in a direction perpendicular to it. It can be verified that the forces and moments on each segment are in equilibrium. (b) total the minus sign indicating that the force is actually downwards. The forces and moments at the ends of each segment are summed and the result is illustrated in Fig. The resulting equivalent loading is illustrated in Fig. the location of this centroid varies along the length of the bridge.7. In nonprismatic bridge decks. 2.3 Non-prismatic bridges The eccentricity of a prestressing tendon is measured relative to the section centroid. in Fig.Page 63 Fig.

It is required to determine the equivalent loading due to prestress. The definition of the parabola is independent of the section geometry. 2.3. With reference to Example 2.Page 64 Fig.12). (b) equivalent loading Example 2. In addition.19 Equivalent loading due to variation in location of centroid: (a)segment of beam and tendon. it is defined by an equation of the same form as equation (2. the centroid changes depth linearly between A and B and between B and C.9: Equivalent loading due to change in geometry The beam illustrated in Fig. AB and BC. The beam is divided into just two segments. there are friction losses of 12% which vary linearly between A and C (friction losses generally do not vary linearly but this is a widely accepted approximation).e.: .20 has a non-prismatic section. 2. It is prestressed with a tendon following a single parabolic profile from A to C. i.

it can be found as the difference between y and the line representing the centroid.20 Elevation of beam and tendon profile If the eccentricity is approximated as the vertical distance. for BC. the derivatives are: . 2.Page 65 Fig. the eccentricity is given by: Differentiating the equation for segment AB gives: Similarly. Hence for segment AB: Similarly for segment BC.

The forces are combined in Fig.97P and 0. 2.21(b). 2. 2. The resulting equivalent loading due to prestress is illustrated for each segment in Fig.91P respectively. for both segments: The average values for prestress force in segments AB and BC are 0. where P is the jacking force.21(a).Page 66 Fig.21 Equivalent loading: (a) loading on each segment. (b) total Differentiating again gives. .

Moment distribution has been selected as there is a physical action corresponding to each stage of the calculation which makes it easier to develop a qualitative understanding of the phenomena. In addition to moment distribution. is performed not by adding numbers in a table but rather by adding bending moment diagrams. a knowledge of such methods is extremely useful for developing a complete understanding of the nature of bridge behaviour under load. It is. not practical in most situations to analyse bridges by hand. In this section. Moment distribution is a convenient hand method that can be used in many cases. moment distribution and the method of equivalent loads. 3.Page 67 Chapter 3 Introduction to bridge analysis 3. the method is illustrated using some simple examples. The process of releasing joints. actions other than forces that can induce stress in a bridge. The method is also useful for checking computer output as it provides approximations of increasing accuracy throughout the analysis process. . of course. the method of equivalent loads is presented as a means of analysing for the effects of ‘indirect actions’. i.2 Moment distribution Moment distribution can be used to check computer output and to develop insight into the behaviour of a great range of bridge types subjected to many different types of action. This may be slower to perform in practice but provides a much clearer explanation of the process and is less prone to error. The method consists of determining loads which have the same effect on the structure as the indirect action.e. However.1 Introduction Two approaches to bridge analysis are presented in this chapter. familiar to most engineers. The approach to moment distribution used in this book is a little different in its presentation to that used traditionally. Analysis for the equivalent loads can be carried out by conventional computer methods or by moment distribution.

Table 3. The bending moment diagram (BMD) The 3 members are isolated by applying fixities at B and C as shown: due to the applied loading on the resulting ‘fixed’ structure is sketched. Appendix A gives the BMDs for members with a range of end conditions. This fixed structure is equivalent to: The resulting bending moment diagram (BMD) is found (with reference to Appendix A): . The fixities are numbered and the direction of each is defined.Page 68 The analysis procedure consists of four steps.1 Moment distribution General Example Step 1: All members of the structure are isolated from one another by applying a number of fixities. These are presented in the left-hand column of Table 3.1 and are illustrated using the example presented in the right-hand column.

boxed below.25l)=6.Page 69 General Example Step 2: The BMDs due Unit rotation at B induces a BMD of (refer to Appendix B): to application of unit displacements at each of the fixities are found.2EI/l. The total discontinuity at B is 3EI/l+4EI/(1. Appendix B gives the BMDs for a wide range of such displacements. i. a BMD with a unit discontinuity at B which results from some applied rotation at B: The corresponding BMD for rotation at C is found similarly: .e. These BMDs are then normalised to give a unit value at each point of moment discontinuity. Dividing the BMD by this gives the normalised version.

the first iteration has resulted in a BMD which is sufficiently accurate for most practical purposes. Example The moment just left of B in the fixed BMD (1) of 0.006wl 2.005wl2 . For this particular example. to get an exact answer.005wl 2. such discontinuities are successively removed by applying ‘rotations’. The resulting BMD is: The discontinuity at C is now (0. Hence. must be repeated until no discontinuity remains. often characterised by discontinuities in the BMD. there is generally a lack of equilibrium of bending moment at the fixing points. The correction of the discontinuity at B had the effect of increasing the discontinuity at C. In this step.125)wl2=0. This is removed by subtracting the normalised BMD corresponding to rotation at C (3). This is performed simply by adding or subtracting the normalised bending moment diagrams.Page 70 General Step 3: In the fixed BMD (Step 1). . the process of adding normalised BMDs. the correction of the discontinuity at C had the effect of reintroducing a discontinuity at B.130wl 2) by 0. factored by the discontinuities. factored by 0.006wl2 (the BMD is subtracted.131–0.125wl 2 is less than that just right of B (0. This is the final solution. Similarly. factored by 1. as adding it would increase the discontinuity). This discontinuity is removed by adding the normalised BMD corresponding to rotation at B (2). The resulting BMD is: Step 4: The process described in Step 3 is repeated until a BMD is arrived at in which equilibrium is satisfied everywhere. scaled in each case by the appropriate discontinuity.

which gives a unit discontinuity of moment at B and C. this time using symmetry. as illustrated in Fig. scaled by 0. The arrows indicate the directions of positive rotation for Step 2. 3. exact.1(d) (boxed). no iteration is required for this example and the BMD of Fig. The resulting BMD is illustrated in Fig.1(a).Page 71 Example 3. As will be demonstrated in this example.1(c) and the normalised version. is illustrated in Fig. (d) normalised BMD . 3. 3. 0.005wl2.1 Moment distribution using symmetry: (a) symmetrical system of fixities. The beam is fixed simultaneously at B and C as illustrated in Fig.1(d). (b) unit rotation simultaneously at B and C.1: Continuous beam using symmetry Concepts of symmetry can be used to great effect when analysing by moment distribution. 3. 3. (c) BMD due to unit rotation. as before.2 is. As there are no further discontinuities. The beam of Table 3.1 is analysed again. 3. Step 3 consists of removing these two discontinuities (simultaneously) by adding the BMD of Fig. it is possible to isolate members from each other by the simultaneous application of a pair or pairs of equal and opposite fixities.005wl 2. The resulting BMD (from ) is illustrated in Fig. Hence. Fig. 3. The discontinuities at B and C in the fixed BMD are.2. in Step 2. 3. two equal and opposite rotations must be applied simultaneously at B and C.1(b). in fact. The fixed bending moment diagram (BMD) (Step 1) is unaffected by the symmetric system of fixities but.

and D as illustrated in Fig. This is corrected by applying the BMD of Fig. 3. For simplicity. Step 1: The members are isolated by applying fixities at A.4(e). this culvert is assumed to be supported at two discrete points under the walls and to have constant flexural rigidity throughout.3 Box culvert example . 3. The fixed BMD is. from . This is corrected by adding the BMD of Fig. 3.4(f). Symmetry is exploited by simultaneously fixing A and B and simultaneously fixing C and D as shown.2 Final BMD for three-span beam Example 3. 3. 3. Step 3: The discontinuity at A and B in the fixed BMD (Fig.4(Pl/8). The resulting BMD is illustrated in Fig.Page 72 Fig. Fig. The discontinuity now present at C and D is 0. Step 2: Applying unit rotation simultaneously at A and B (Fig. When normalised.4(c)) results in the BMD illustrated in Fig.3. 3. 3. 3. factored by Pl/8. 3. B.4(e).4(a).4(b).2: Box culvert The application of moment distribution to a two-dimensional frame type of structure is demonstrated using the box culvert illustrated in Fig.4(f).4(b)) is Pl/8. C.5(b). 3. this becomes the BMD of Fig.5(a). factored by that amount. 3. The normalised BMD due to rotation at C and D is found similarly and is as illustrated in Fig. 3. 3. 3. which gives the BMD of Fig.4(d). as illustrated in Fig.

(d) BMD associated with unit rotation at A and B. (b) fixed BMD. 3. (f) normalised BMD for rotation at C and D Fig. (c) a after second correction at A and B.Page 73 Fig. (b) after correction of discontinuity at C and D. (c) moments required to induce unit rotation at A and B. (e) normalised BMD for rotation at A and B.5 BMD after successive corrections: (a) after correction of discontinuity at A and B.4 Analysis of box culvert (a) system of fixities. 3. (d) after second correction at C and D .

The discontinuity now existing at A and B is considered to be sufficiently small for the purposes of this example and the BMD of Fig. A higher hogging moment (0. 3. is added to give the BMD of Fig.5(c). The interaction of bridges with the surrounding soil is considered further in Chapter 4. 3. This box culvert of Example 3. 3.825(Pl/8)) is found with a corresponding reduced sagging moment.064(Pl/8). A more typical situation would be that of continuous support from granular material throughout the length of the base and side walls. factored by this amount.6 Finite-element model of box culvert and surrounding soil: (a) finite-element mesh.4(e).5(d).4(f). (b) resulting BMD . factored by 0.2 was assumed to be supported at two discrete points.6(a) and the resulting BMD in Fig. 3. 3. Fig. Earth pressure on a structure of this type generates an additional distribution of moment. 3.5(d) is deemed to be the final solution. Finally. 3. Adding the BMD of Fig.6(b).Page 74 Step 4: The correction at C and D has reintroduced a discontinuity at A and B of 0. 3.16(Pl/8). the BMD of Fig. gives the BMD of Fig. A more realistic finite-element (FE) model taking account of these effects and assuming typical soil properties is illustrated in Fig.

(b) imposed support settlement . distributions of bending moment and shear are induced in the deck.7 Three-span beam example: (a) geometry. Clearly soil deforms under the vertical forces applied through bridge piers and abutments. Thus. following example serves to demonstrate the effect of a differential settlement on a continuous beam bridge.9(b). Example 3.7 is subjected to a settlement at B of Δrelative to the other supports. The disadvantage of this is that differential settlement is more often caused by a relatively weak patch of soil under one support rather than by a non-uniform distribution of applied loads. this becomes the BMD of Fig. However. Δ relative to the others and to determine the effects of this on the structure. 3. an alternative approach. a rotation at C results in the normalised BMD of Fig. 3.5EI∆ 2. The resulting BMD is required given that the beam has uniform flexural rigidity. the structure and the surrounding soil may be represented using non-linear computer models. Step 2: Applying a unit rotation as illustrated in Fig. Hence. the beam is fixed as illustrated in Fig. If the deformation is not uniform. the discontinuity at C is of the /l same magnitude. the fixed BMD is as illustrated in Fig.8. The discontinuity of moment at B is 1. The system of fixities cannot be symmetrical as the ‘loading’ is not symmetrical. Similarly.9(a) results in the BMD illustrated in Fig. Step 1: Referring to (4th and 5th BMDs). frequently adopted by bridge engineers. as the effect is often not very significant.3 Differential settlement of supports There is considerable research and development activity currently taking place in the field of soil/structure interaction. EI.3: Differential settlement by moment distribution The continuous beam illustrated in Fig. 3. 3. To accurately analyse for this effect. 3. 3.9(d).9(c). is to assume that a foundation support settles by a specified amount.Page 75 3. 3.8(a). When normalised. By coincidence. 3. Fig. many structural engineers treat the soil as a spring or a series of springs in the numerical model. The .

Page 76 Fig. 3.8 First step in analysis of three-span beam: (a) system of fixities. (b) BMD associated with unit rotation at B. (b) fixed BMD Fig.9 Effect of rotations at points of fixity: (a) unit rotation at B. 3. (c) normalised version of BMD associated with rotation at B. (d) normalised BMD associated with rotation at C .

8(b) is 1.9(c).10(a). Step 4: One further iteration gives the BMD of Fig. 3. /l factored by this amount. 3. /l 3. I.Page 77 Fig.8 EI∆ 2. The moment at the support which settles is proportional to the second moment of area. 3. gives the BMD of Fig. This is important as supports in continuous beams are generally subjected to hogging moment and are often not designed to resist significant sag. Differential settlement has the effect of generating sagging moment at the support which settles. Adding the BMD of Fig. (c) after second corrections at1 and 2 Step 3: The discontinuity at B in Fig.9(d) factored by this amount to give the BMD of Fig. This correction at B has the effect of increasing the discontinuity at C to 1. 3.5 EI∆ 2.10(b). k). It is usual to size a bridge by selecting a depth which is proportional to span length (i.e. l.10(c). This is corrected by adding the BMD of Fig. which are typical of differential settlement: 1. divided by the square of the span length.10(c) which is deemed to be of sufficient accuracy. 3. depth=l/k for some constant. It is interesting to note two additional things about the final BMD illustrated in Fig.10 BMD after successive corrections: (a) after correction at1. As the second moment of area is proportional to the cube of . 3. (b) after correction at 2. 3.

the beam is fixed at both ends as illustrated in Fig. If. Further. However. The implication of this is that. if the specified settlement is deemed to include such time-dependent effects. and its temperature is reduced by Δ then there will be no strain. analysis for the effects of axial expansion/contraction due to temperature changes is considered.11(b). The beam then contracts by α T)l where l is its length. (∆ is coefficient of thermal expansion (strain per unit change in temperature). the differential settlement should be larger as the supports are further apart and soil conditions are more likely to be different.e. This total restraint generates a stress of magnitude Eα T). the distribution of moment due to differential settlement is proportional to the elastic modulus. 3. in practice. This beneficial effect of the creep in concrete is countered by the fact that the magnitude of the differential settlement itself often increases with time due to timedependent behaviour in the supporting soil. for a given settlement Δ the induced moment is more critical for bridges with long spans than . for those with short ones. values are often specified which are independent of span length. 3. moment is proportional to span length. the span/depth ratio is particularly important. a modest increase in slenderness can considerably reduce the moment due to differential settlement. there can be no tendency to crack.11 Extreme restraint conditions For axial temperature: (a) free. (∆ where E is the elastic modulus. However. As there is no stress. (b) fully fixed . 2. A widely accepted approximate way to model the effect of creep is to reduce the elastic modulus. The stress is manifested in a tendency to crack. restrained against contraction. 3. 3. the maximum moment due to differential settlement is roughly proportional to (l/k)3/l2=l/k3. As moment is proportional to this modulus. Fig. namely. there are two thermal effects for which bridge analysis is required. on the other hand. It might be expected that for longer spans. If a beam is on a sliding bearing as illustrated in Fig. axial expansion/contraction and differential changes in temperature through the depth of the bridge deck. There cannot be any strain as the beam is totally T. it is reasonable to anticipate some reduction in moment due to concrete creep. no stresses are generated as no restraint is (∆ offered to the contraction.Page 78 the depth. i. This is particularly significant for concrete bridges where considerable creep occurs. A (negative) strain will occur of magnitude α T) where α the T.11(a) and the temperature is reduced by Δ it will contract freely. In this section. However. Unlike BMDs due to applied forces. it follows that creep has the effect of reducing the moment due to differential settlement over time.4 Thermal expansion and contraction As discussed in Chapter 2.

12(a). 3. 3. (b) deformed shape after expansion of deck Fig. where a beam is partially restrained. it is required to find the bending moment.13 First step in analysis of frame: (a) fixing system. This happens for example in arch bridges where contraction is accommodated through bending in the arch (Fig. shear force and axial force diagrams due to an increase in deck temperature of Δ T. Example 3. (b) fixed axial force diagram . It also happens in frame bridges where the piers offer some resistance to expansion or contraction of the deck.30).12 Frame subjected to axial change in temperature: (a) original geometry.Page 79 The most common case requiring analysis is the one in between the two extreme cases described above. Fig. 1. 3.4: Restrained axial expansion by moment distribution For the bridge illustrated in Fig.

its resistance to bending restrains the expansion a little and generates a small compressive stress in the deck between A and B. 3. (b) associated axial force diagram. Thus. a thermal expansion tends to bend the pier as illustrated in Fig.Page 80 The deck is supported on a bearing at B which prevents relative translation between it and the supporting pier but allows relative rotation. BD. As the pier is fully fixed at its base. (d) free body diagram showing lack of equilibrium of forces at B. (c) associated shear force diagram. bending moment is generated in the pier.12(b). 3. Fig. In addition. (e) normalised free body diagram .14 Effect of translation at fixing point: (a) forces required to induce unit translation.

3. 3. 3.13(a). a force is required to bend BD (. where α the coefficient of thermal expansion and E is the elastic modulus.13(b). The normalised version of Fig. 3. The associated axial force and shear force diagrams are illustrated in Figs. Hence the axial force diagram is as (∆ (∆ illustrated in Fig. 3. such a fixity is not necessary to isolate the members in this case.) There is no bending moment or shear force in the fixed structure. The (∆ is corresponding force is α T)E(area)=6000α T)EI/h2. (c) shear force diagram. Step 2: To apply a unit translation at B requires a force to compress AB of E(area)/(length)=1500EI/h3 as illustrated in Fig. The required external force at B can be seen in the free-body diagram of Fig.14(e). 4th case) of 3EI/h3 giving a total required force at B of 1503EI/h3. corresponding to unit discontinuity of force at B.14(b) and (c). (b) axial force diagram.14(a). Fig. (While a rotational fixity at this point is also possible.15 Results of analysis: (a) free body diagram with restored equilibrium at B. (d) bending moment diagram . the stress in AB is α T)E. 3.14(d). 3.14(d). is illustrated in Fig.Page 81 Step 1: The substructure and superstructure of the bridge are isolated from one another by the imposition of a translational fixity at B as illustrated in Fig. However. 3. The lack of force equilibrium in this diagram corresponds to the moment discontinuity in the BMDs of the usual moment distribution problems. In addition.

the bending moment in BD varies from zero at this point (∆ to a maximum of 12α T)EI/h at D as illustrated in Fig.002×6000α T)EI/h2 to Fig.e.5: Thermal contraction in frame bridge by moment distribution The frame structure illustrated in Fig. in-situ concrete bridges generate significant quantities of heat while setting and consequently have their initial set when the concrete is warm. However.14(e) factored by 6000α T)EI/h2.13(b)) as there is no axial force in BC and no shear force in the pier to correspond to the axial force in AB.13(b).15(b) and (c). The final axial force and shear force diagrams are illustrated in Figs. as for differential settlement. this is the final free body diagram.16 Integral frame of Example 3.5 .15(d).16 is integral having no internal bearings or joints. may be relieved by the effect of creep. thermal contraction or expansion induces bending moment as well as axial force and shear. It is subjected to a uniform reduction in temperature through the depth of the deck (ABC) of 20°C and no change in temperature elsewhere. 3.Page 82 Step 3: There is a lack of force equilibrium in the fixed structure at B (Fig. The relative 2 flexural rigidities are given on the figure and the area of the deck is 500I0/l . Substantial temperature changes occur on a short-term basis during which the effects of creep do not have a significant ameliorating effect. 3. 3. (∆ As there is no further force discontinuity. i. As the shear force across the pin at B is 12α T)EI/h2.15(a).998×6000α T)EI/h 2 and a shear (∆ (∆ force in BC of 0. As a result. This means that such stresses. (∆ There are some points of interest about axial temperature effects apparent from this simple example. Example 3. if sustained in a concrete structure. The resulting distribution of bending 6 moment is required given that the coefficient of thermal expansion is 12×10− . 3. the moments and forces due to changes in temperature are proportional to the elastic modulus. 3. The sustained stresses generated by the subsequent contraction of the concrete as it cools can be relieved substantially by creep. The result is illustrated in Fig. Hence the rise in temperature results in a lot of strain and in very little stress in the deck. 3. The area of the deck is typically numerically much larger than the second moment of area of the pier with the result that the restraint to deck expansion is relatively small. This situation is corrected at B by subtracting the forces illustrated in Fig. 3. 3. Most noteworthy is the effect of the relative values of deck area and pier second moment of area. by adding an axial tension in AB of 0. Fig. It is also of interest to note that.

12EI 0/l .17(a)) requires the moments and forces illustrated in Fig. Hence. However.17(a). As the rotation is applied while fixing against translation. Due to symmetry. there is no tendency for point B to rotate and this point can be considered fixed without applying a fixity.17(b).17(c).18(a) and generates the BMD illustrated in Fig. (c) free body diagrams showing shear and axial forces in fixed frame Step 1: Due to symmetry. as points A and C will tend to rotate as well as translate.2EI0/l and is illustrated in Fig. However. no axial forces are generated in the members.2EI0 /l. 3. 3.17 First step in analysis of frame: (a) system of fixities. one translational and one rotational as illustrated in Fig.18(c). 3. 3. the attempt to contract generates a tensile stress in ABC of 6 6 (12×10− )(20°)E and an axial force reaction at each end of (12×10− )(20°)(500EI0/l2)− 2 0. the fixities at A and C are taken to be equal and opposite as illustrated. (b) fixed axial force diagram. Applying a unit translation in Direction 2 (while preventing rotation) requires the moments and forces illustrated in Fig. there is a lack of force equilibrium at A and C as illustrated in Fig. 3. No distribution of bending moment is present in the fixed structure but the axial force diagram is as illustrated in Fig. 3. there is a shear force just below A and C which is unmatched by an axial force in AB or BC. In the fixed structure. 3. is as illustrated in Fig.19(a) and generates the BMD illustrated in . Step 2: Applying a unit rotation in Direction 1 (Fig. The normalised version is found by dividing by 7. when divided by 7. there is a lack of force equilibrium at A and C which.18(d).Page 83 Fig. 3. two fixities are needed at each. 3.18(b). As there is axial force in ABC but no corresponding shear force in AD or CF. 3.

3.4EI 0/l3. 3.Page 84 Fig. adding it to a BMD of zero. 3.20(c). 3. The discontinuity or lack of equilibrium at A is 538.19(b). (d) normalised shear and axial forces associated with rotation Fig.18 Effect of rotation at A and C: (a) moments and forces required to induce unit rotation. 3. 3. (b) BMD associated with unit rotation.20(c) . (c) normalised BMD associated with rotation. 3. In addition. 3.19(c). The normalised lack of joint equilibrium is illustrated in Fig. Normalising with respect to this value gives Figs. as there is no moment induced in the fixed structure.17(c) is corrected by factoring Fig. In addition.19(d) that there is a shear force just below A which is not matched by the axial force to its right. a distribution of axial force is generated which is illustrated in Fig. the joint forces of Fig.20(a) by 0.12EI0/l 2 and. 3. Step 3: The lack of force equilibrium in the fixed structure illustrated in Fig. It can be seen in Fig.20(a) and (b).

3.21(a) and (b) respectively.17(c).18(c) and (d) by 0.19 Effect of translation at A and C: (a) forces and moments required to induce unit translation. (b) BMD associated with unit translation. Step 4: The removal of the moment discontinuity reintroduces a lack of force equilibrium which is evident in Fig. 3.21(a). (d) free body diagram at joint A showing lack of equilibrium .20(a) and (c) are Fig. 3. The results are illustrated in Fig. 3. 3.00214EI0/l and adding them to Figs. There is a discontinuity in the BMD (or lack of moment equilibrium) at A and C evident in Fig. 3. This is corrected by scaling Figs. Figures 3. (c) axial force diagram associated with unit translation. 3. The results are illustrated in Fig.12EI 0/l2 and added to those of Fig.22.00285EI0/l2.22(b) of magnitude. 0.21. 3.Page 85 must be factored by 0.

21 Effect of correcting for lack of force equilibrium: (a) corrected BMD. (c) free body diagram showing unit discontinuity of forces Fig. (b) normalised axial force diagram. (b) corrected free body diagram .20 Normalised effect of translation at A and C: (a) normalised BMD.Page 86 Fig. 3. 3.

23 Results of analysis for effects of thermal contraction: (a) BMD. (b) free body diagram scaled by this amount and added to Figs. 3. .23(a) is therefore adopted as the final BMD.Page 87 Fig.22(a) and (b). (b) corrected free body diagram Fig. Figure 3. This leads to Fig. 3. 3. 3.18(c) and (d). 3.22 Effect of correcting for discontinuity in BMD: (a) corrected BMD.23 where the lack of force equilibrium is deemed to be sufficiently small. The resulting moment discontinuity is corrected by factoring and adding Figs.

However. To some extent this alters the resistance to contraction or expansion. 3. The equivalent loads for this example are illustrated in Fig. 3. However. thermal movement is resisted by bending in both the piers and the deck.25(a).6: Introduction to equivalent loads method The equivalent loads method will first be applied to the simple problem of the partially restrained beam illustrated in Fig. Stage A—Calculate the equivalent loads and the associated stresses: The loading is found which would generate the same strain in an unrestrained member as the distribution of temperature.5 serves to illustrate the effect of a moment connection between the bridge deck and the piers. While the method may not at first seem to be any simpler to apply than the procedure used above. sectional area and E is the elastic modulus of the beam. F 0.4. a more important effect of the moment connection is the bending moment induced in the deck by thermal movement. An axial expansion can be generated in an unrestrained beam by applying an axial force. this distribution of stress must be subtracted to determine the stresses generated indirectly by the change in temperature. Example 3. In such a case. it is necessary to identify the ‘associated stresses’. will generate both.Page 88 Example 3.24 Beam on rollers with partial (spring) restraint .25(a) and the associated stress distribution in Fig.25(b). Stage B—Analyse for the effects of the equivalent loads: The beam is analysed for the loading illustrated in Fig. This can become a significant factor in bridge deck design. on the other hand.1 Equivalent loads method The method of equivalent loads is a method by which a thermal expansion/contraction problem can be converted into a regular analysis problem. The equivalent loads method consists of three stages as follows. Normally this stage would be done by computer but it Fig. 3.24 which is subjected to an axial increase in temperature of Δ The expansion is partially restrained by a spring of stiffness AE/(2l) where A is crossT. Therefore. The equivalent force. it is particularly useful when a computer is available to carry out the analysis but the program does not cater directly for temperature effects. 3.e. In Stage C. i. temperature on an unrestrained is member generates strain but not stress. even on an unrestrained beam. that distribution of stress which is inadvertently introduced into the structure by the equivalent loads. where: where α the coefficient of thermal expansion. 3. 3.

In this case.25(b) from that of Fig. 3. The distribution of stress due to application of F 0 is an axial tension throughout the beam of magnitude 2F0/(3A) as illustrated in Fig. the bending takes place freely and the beam curves upwards as the top expands relative to the bottom.25(d). (d) stress distribution due to temperature change is trivial for this simple example. (c) equilibrium of forces at spring.25 Analysis by equivalent loads method: (a) equivalent loads. this consists of subtracting the axial stress distribution of Fig. This is the final result and is what one would expect from a thermal expansion in a partially restrained beam. If a linear distribution of this type is applied to a simply supported single-span beam. 3. This corresponds to the case of a beam on rollers subjected to an axial increase in temperature in that strains take . 3. the top tends to increase in temperature faster than the bottom. 3. It is well known that. it is resisted in proportion to their stiffnesses. the beam acts as a spring of stiffness AE/l. (b) associated stress distribution.25(c). 3. strain is generated but also some compressive stress. For this example. Stage C—Subtract the associated stresses: The distribution of associated stresses is subtracted from the stresses generated by the equivalent loads. Hence the force is taken in the ratio 1:2 as illustrated in Fig. when a load is applied to two springs. a differential temperature distribution develops which tends to cause the bridge to bend. The result is an axial compression of F0/(3A) throughout the beam.5 Differential temperature effects When the sun shines on the top of a bridge.Page 89 Fig. 3. Thus.25(d).

The BMD will be determined using the method of equivalent loads. where αis the coefficient of thermal expansion. It is required to determine the BMD due to the temperature change given that the coefficient of thermal expansion is α .26 Beam of Example 3. The centroid of the beam is at mid-height. temperature generates a curvature of: The corresponding equivalent moment is: Fig. The temperature change varies linearly from an increase of 5° at the top to a decrease of 5° at the bottom. the change in is . the free bending is prevented from taking place and the situation is one of stress but no strain. partial restraint against bending is present as will be seen in the following examples. The ratio 1/R is known as the curvature. the elastic modulus is E and the second moment of area is I. In multi-span beams and slabs. R is radius of curvature.Page 90 place but not stress. 3. Stage A—Calculate the equivalent loads and the associated stresses: The temperature change would generate a distribution of strain varying from 5α the top to − at the bottom at 5α of an unrestrained beam.26 is subjected to a change of temperature which is non-uniform through its depth. κIn this case.7: Differential temperature in two-span beam The two-span beam illustrated in Fig. is where ε strain. If such a differential temperature distribution is applied to a beam in which the ends are fixed against rotation. σ stress and y is distance from the centroid.7 and applied distribution of temperature . Consider the familiar flexure formula: where M is moment. 3. Example 3.

(b) BMD due to application of equivalent loads. The equivalent loads and associated BMD are illustrated in Figs.27(a) and (b) respectively. 3.e. Therefore. it is necessary to identify the ‘associated BMD’. 3.28 Stages in equivalent loads method: (a) applied equivalent loads. 3.Page 91 Fig. Fig. even on unrestrained beams. (c) BMD after subtraction of associated BMD . (b) associated BMD Temperature on an unrestrained structure generates strain and curvature but not bending moment or stress. The equivalent moment on the other hand will generate both curvature and bending moment. that distribution of moment which is inadvertently introduced into the structure by the equivalent loading. i.27 Application of equivalent loads method: (a) equivalent loads.

29 Differential temperature example Fig.30: Fig. 3.28(b). 3. a strain but no stress. the axial component will result in a free expansion.28(c). 3. 3. axial strain and bending strain. The depth of the beam is h and the centroid is at mid-depth. Example 3. It is required to determine the BMD due to the temperature increase given that the coefficient of thermal expansion is α .30 Resolution of applied change in strain into axial and bending components .e. 3.8: Differential temperature change in continuous beam The three-span beam illustrated in Fig. 3.Page 92 Stage B—Analyse for the effects of the equivalent loads: Analysis of a symmetrical two-span beam is trivial because.28(a) and the solution can be determined directly from Appendix B.27(b) from Fig. does not rotate. The bending component will result in some moment but not as much as would occur if the beam were totally prevented from bending. i. As the beam is free to expand. The BMD will be determined using the method of equivalent loads. 3. The distribution is then resolved into two components. due to symmetry.28(b) gives the final result illustrated in Fig. as illustrated in Fig. Stage A—Calculate the equivalent loads and the associated stresses: In this example. The temperature distribution is first converted into a strain distribution by multiplying by the coefficient of thermal expansion. the curvature is. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig. B. Hence. 3. 3. from Fig. The BMD due to the applied equivalent loading is as illustrated in Fig. the elastic modulus is E and the second moment of area is I. a. 3. the central support point. it is effectively fixed as illustrated in Fig.29 is subjected to an increase in temperature which varies linearly from a maximum of 20° at the top to 10° at the bottom.30.

31 Application of equivalent loads method: (a) equivalent loads. the equivalent loads and associated BMD are as illustrated in Fig.32 First step in analysis by moment distribution: (a) system of fixities. the equivalent moment becomes: Thus.31.Page 93 Hence. 3. (b) associated BMD Stage B—Analyse for the effects of the equivalent loads: The frame is analysed for the loading of Fig. (c) BMD in fixed structure . Normally this stage would be done by computer but it will be done using moment distribution for this simple example. 3. 3. Fig.31(a). (b) applied loading on fixed structure. Fig. 3.

(c) normalised BMD associated with rotation Fig. As the ‘loading’ is symmetrical.34 Completion of equivalent loads method: (a) BMD due to analysis by moment distribution. Fig.32(c).32(b) and the resulting BMD (Appendix B) in Fig.Page 94 Step 1: The beam is fixed at B and C in order to isolate the three spans.32(a). 3. The applied loading on the fixed structure is illustrated in Fig. the fixities at B and C are equal and opposite as indicated in Fig. (b) BMD after subtraction of associated BMD .33 Effect of rotation at fixing points: (a) moments required to induce unit rotation. 3. (b) BMD associated with unit rotation. 3. 3. 3.

3.35(c). the equi ivalent loading is a force of 580 kN and a moment of 160 kNm of which only the mome is of relevance. The upward reaction from the bearing due to 6 the dead load is 300 kN. Referring to that example.32(c) is corrected by adding Fig.1 (Chapter 2).33(b).34(a) gives the final result illustrated in Fig. This is the BMD due to the differential temperature increase. 3. 3. Example 3. 3.33(c).35(a) and (b) is subjected to the differential increase in temperature shown in Fig. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig. The cross-section and temperature distribution for this examp le are identical to those of Example 2. 3. It is required to determ if there will be uplift at B c due to combined temperature and dead load. 3.9: Bridge diaphragm The bridge diaphragm illustrated in Figs. Fig. Step 4: As no discontinuity now exists. 3. no further iteration is required. (c) applied temperature distribution . 3. 3. The normalised version is illustrated in Fig.Page 95 Step 2: Unit rotation at B and C requires the application of the moments illustrated in Fig.34(b). Step 3: The discontinuity of moment at B and C evident in Fig. /h.35 Bridge diaphragm example: (a) plan of geometry.33(a) and generates the BMD illustrated in Fig.33(c) factored by 5EIα The result is illustrated in Fig.31(b) from Fig. (b) section through diaphragm.34(a). 3. 3. the coefficient of thermal expa is 12×10 − and the modulus of 2 elasticity is 35 000 N/mm . 3.

3. The associated BMD is illustrated in Fig.36(c). 3. there is no uplift of this bearing due to the differential temperature change. 3. Hence (as in Example 3. As the reaction due to dead load exceeds this value. 3. . 3. The reactions at A and C can be found from the free body diagram illustrated in Fig. Subtracting the associated BMD gives the final BMD illustrated in Fig.36(b). the structure is analysed for the loading illustrated in Fig.36(d). By symmetry. 3.36(a). 3.7) the BMD due to applied loading is as illustrated in Fig.37: Hence the reaction at B is 80+80=160 kN.Page 96 Fig. (c) results of analysis. (d) final BMD Fig.36 Analysis to determine effect of imposed differential temperature: (a) equivalent loading. (b) associated BMD. Point B does not rotate and is effectively fixed.37 Free body diagram for diaphragm beam To determine the reaction due to this moment.

67α At the bottom fibre the total stress is E− E=− E.38(d). It was established in Example 2. 3. (d) final BMD Thus.38(b).718α for half of the bridge. .38(c) gives the final distribution of moment due to restrained bending illustrated in Fig.32α 5.38α and E − 5. 3. (c) results of analysis.45α (restraint to expansion induces compression at the extreme fibres).29α and 11. 3. 3.2.077α which E gives stresses (tension positive) of − 5.38(b) from Fig. 3. Hence.10: Differential temperature in bridge of non-rectangular section The beam-and-slab bridge whose section and temperature loading is described in Example 2. B.38 Analysis to determine effect of differential temperature change: (a) equivalent loading. Stage C: Subtracting the associated BMD of Fig. of 1.29α 5. the total E stress at the top fibre is − 5.2 that the residual stresses are − 5.2 consists of two 10 m spans.38α 10. 3. In Example 2.Page 97 Example 3. Stage B: Analysis by computer or by hand gives the BMD illustrated in Fig. 3. (b) associated BMD. It is required to determine the maximum stresses due to the differential temperature change.45α E− E=5. Fig.87α E. Using the method of equivalent loads: E Stage A: The equivalent loads are illustrated in Fig.38(c). a sagging bending moment is induced over the central support.38(a) and the associated BMD in Fig. 11. it was established that the equivalent moment due to the temperature change is − 0.32α at the top and bottom fibres E E respectively.

11: Variable section bridge Figure 3.39 Pedestrian bridge: (a) elevation. Fig.Page 98 Example 3. (d) section C—C . It is required to determine the equivalent loading and the associated stress distributions given a coefficient 6 of thermal expansion. The deck is subjected to the differential decreases in temperature shown in the figure.39(a) shows the elevation of a pedestrian bridge while Figs. 3. α =12×10− /°C and a modulus of elasticity. The bridge is first restrained when its temperature is somewhere between 5°C and 25°C and the minimum temperature attained during its design life is − 15°C. 3. (c) and (d) show sections through it. (c) section B—B and corresponding imposed temperature distribution. E=35×106 kN/m 2. (b) section A—A and corresponding imposed temperature distribution.39(b).

3. 3. 3.033 m below the top fibre for the solid and hollow sections respectively (Figs.Page 99 By summing moments of area it is found that the centroids are 0. Summing products of stress and area in Fig.39(b) gives the equivalent force (positive tension) on the solid section due to the differential temperature distribution: The corresponding equivalent moment (positive sag) is: Fig.39(b)–(d)). (b) equivalent loading.5 m and 1.40 Model of pedestrian bridge: (a) geometry showing differences in level of centroids. (d) associated BMD . (c) associated axial force diagram.

40(b). the bridge is two-span . bending moment results about both axes as will be illustrated in the following example. If there is restraint to either or both rotations. the equivalent loads are illustrated in Fig. the area is 2. a second moment of area several times as large as the maximum used elsewhere in the model (e. 3. the equivalent force due to the differential temperature distribution is: and the equivalent moment is: The maximum axial decrease in temperature is (25− (−15))=40°C and the corresponding stress is 40α For the solid section of Figs.40(a). 3. Note that the short vertical members at b and c could be assumed to have effectively infinite stiffness. ten times) generally provides sufficient accuracy without causing such problems.41(a) to allow axial expansion in both the X and Y directions. 3. it tends to bend about both axes. 3. The associated axial force and bending moment diagrams are illustrated in Figs. respectively. 3. Similarly.5.40(c) and (d) from the results. the area is 2.12: Differential temperature The slab bridge of Fig. using members with very large stiffnesses can generate numerical instability in a computer model. The bending moment and axial force distributions due to the temperature decreases can be found by analysing for the equivalent loading illustrated in Fig. 3.39(b) and (d).40(b) and subtracting the associated distributions of Figs. However. for rotation.64 m 2 and the equivalent force is: A model which allows for the difference in the level of the centroids is illustrated in Fig. 3. an equivalent force of: For the hollow section.1 Temperature effects in three dimensions When the temperature of a particle of material in a bridge is increased. 3.g. However. the particle tends to expand in all three directions.6 m 2 giving E. Therefore. Noting that the axial effects apply to all members while the differential temperature distributions only apply to the deck (abcd).6×1=2. when a differential distribution of temperature is applied through the depth of a bridge slab.41 is articulated as shown in Fig. 3.40(c) and (d).Page 100 In the hollow section. Example 3.

for longitudinal bending. 3.12: (a) plan showing directions of allowable movement at bearings. However. The coefficient of thermal expansion is 9×10 − /°C and the modulus of elasticity is 32×106 kN/m 2. (c) imposed temperature distribution in deck (section 1− 1). 3. the bridge will tend to act as one unit and bending will take place about the centroid.41 Slab bridge of Example 3. 3.41(c) and (d) respectively.42 corresponding to the different parts of the temperature distribution and the temperature . Further. The specified temperature distributions are different in the cantilevers and the main deck of this bridge. The location of this centroid is: below the top surface.Page 101 Fig. there are three bearings transversely at the ends so that it is not able to bend freely transversely either. The deck and cantilevers are subjected to the differential temperature increases illustrated in Figs. The bridge deck is divided into parts as illustrated in Fig. (b) section A-A. It is required to determine the equivalent loading and the associated 6 BMD due to this temperature change. (d) imposed temperature distribution in cantilever (section 2–2) longitudinally and is therefore not able to bend freely.

43 Equivalent loading due to temperature .43.42 Cross-section with associated distribution of imposed stress: (a) deck. 3. bending is about the centroid of the Fig. 3. Taking moments about the centroid gives a longitudinal bending moment per metre on the main deck of: The corresponding bending moment per metre on the cantilever is: These equivalent longitudinal moments are illustrated in Fig. (b) cantilevers changes are converted into stresses. In the cantilever region. The transverse direction is different from the longitudinal in that the cross-section is rectangular everywhere.Page 102 Fig. 3.

Page 103 Fig.45 Associated BMDs: (a) plan showing section locations.44. (c) section A–A. (b) section B–B. (d) section C–sC .44 Resolution of imposed stress in cantilever into axial and bending components cantilever. 3. The axial expansion is unrestrained while the bending stress distribution generates a moment of: Fig. 3. 3. The applied stress distribution is resolved into axial and bending components as illustrated in Fig.

However.13: Frame subject to axial prestress by moment distribution The frame of Fig. 3. Step 1: The system of fixities used in Example 3. It is required to determine the net prestress force in the deck and the resulting BMD. of magnitude. the problem is completed by analysing the slab (by computer) and subtracting the associated BMDs from the solution. if a beam rests on a sliding bearing at one end. only (M4− 3) needs to be applied at the M deck/cantilever interface as illustrated in Fig. 3.Page 104 In the main deck.46(b).5.6 Prestress The effects of prestress in bridges are similar to the effects of temperature and the same analysis techniques can be used for both. The normalised versions are presented here in Figs. there are many bridge forms where the effects of prestress are restrained to some degree or other and where analysis is necessary. on the other hand. When the movements due to prestressing are unrestrained.47(c) and (d) (unit discontinuity in force). The BMD due to applied ‘loading’ on the fixed structure is zero everywhere as the prestress forces are applied at fixing points. results in changes of both stress and strain. there is one important distinction.16. prestressing that beam does (as is the objective) induce a distribution of stress. However. there are two associated BMDs as illustrated in Fig. . 3. P.43. the differential distribution is applied to a 0. ABC. 3. As for the previous example. 3. 3. Step 2: The effects of inducing rotations or translations at the fixing points are the same as for Example 3. it can undergo axial changes in temperature without incurring any axial stress. For example. An unrestrained change in temperature results in a change in strain only and no change in stress.45. is subjected to a prestressing force along the centroid of the deck.5 is used again here as illustrated in Fig. However. reproduced here as Fig. Prestress.8 m deep rectangular section giving a moment about the centroid of: As M3 is applied to the outside of the cantilever. The frame is analysed by moment distribution. Example 3. 3. As these applied moments generate distributions of longitudinal and transverse moment. 3.47(a) and (b) (unit discontinuity in moment) and in Figs.46(a). the stress distributions are easily calculated and analysis is not generally required.

(c) normalised BMD due to translation. 3. (d) normalised forces due to translation .Page 105 Fig.46 Frame subjected to prestress force: (a) geometry and loading. 3. (b) system of fixities Fig. (b) normalised forces due to rotation.47 Effect of displacements at fixing points: (a) normalised BMD due to rotation.

3. The discontinuity of moment which results is removed by factoring Figs. 3. that equilibrium of forces at A and C is then satisfied. (d) internal forces after correction for moment equilibrium Step 3: The translational fixity is released first to apply the prestress force. Step 4: As force equilibrium in Fig. . It can be seen in the results.47(c) and (d) by P.48(d) is satisfied to a reasonable degree of accuracy. no further iteration is deemed necessary. 3. This consists simply of factoring Figs. 3. (c) BMD after correction for moment equilibrium.Page 106 Fig. 3.0178Pl and adding to give Figs. 3.48(c) and (d). illustrated in Figs.47(a) and (b) by 0.48(a) and (b). (b) internal forces after correction for force equilibrium.48 Effect of prestress force: (a) BMD after correction for force equilibrium.

3.49 is prestressed with a straight tendon at an eccentricity. 3. The analysis to determine the BMD will be carried using moment distribution. (b) axial force diagram due to prestress . The only difference is that.13 serves to illustrate the ‘loss’ of prestress force that occurs in a frame due to the restraint offered by the piers. e. In this example. The axial force diagram is clearly as illustrated in Fig. from the centroid with a prestress force. The method of equivalent loads is applicable to prestress just as it is to temperature. In this example. however. P. To determine the bending moment diagram. It is required to determine the induced distributions of axial force and bending moment. 3. as prestress generates stress as well as strain. a prestressed deck will continue to shorten with time due to creep. this bending moment is independent of the elastic modulus and is therefore unaffected by creep.50 First stage in equivalent loads method: (a) equivalent loads. Interestingly. It is also of importance to note the bending moment that is inadvertently induced by the prestress. it is not appropriate to deduct the associated stresses from the analysis results as was necessary in temperature analysis. the bending stresses induced by this shortening are also relieved by creep with the result that creep has little net effect on the bending moment due to prestress. However.49 Beam subjected to eccentric prestress force Fig.Page 107 Example 3.50(b).14: Analysis for eccentric prestressing The beam illustrated in Fig. Example 3. Fig.50(a). 3. In a concrete frame. This is equivalent to applying a moment alongside the force as illustrated in Fig. 3. is not so straightforward as the beam is not free to lift off the supports at B and C. the prestress force is applied at an eccentricity to the centroid. about 5% of the applied force is lost as shear force in the piers.

51(a). 3. the resulting BMD in Fig.52(b) and the normalised BMD in Fig. Fig.52(a). 3. The BMD in the fixed structure due to the equivalent loading is as illustrated in Fig. The result is illustrated in Fig. 3. 3. 3. this is the final BMD due to prestress. 3. 3. (b) fixed BMD Fig. 3.53.52 Effect of rotation of fixing points: (a) moments required to induce unit rotation.52(c).52(c) by Pe/2 and adding. As there is no further discontinuity. 3.51(b). Step 3: The discontinuity of bending moment evident in Fig. Step 2: The moments required to induce unit rotation at B and C are illustrated in Fig.51 First step in analysis by moment distribution: (a) system of fixities.Page 108 Step 1: The beam is fixed as illustrated in Fig.51(b) is removed by factoring Fig. 3. (b) BMD associated with unit rotation. (c) normalised BMD .

the actual profiles are sometimes approximated by ignoring the transition curves over the internal supports as illustrated in Fig. the slope is found by differentiating the equation as follows: Fig. For this beam. the intensity of which can be determined by considering equilibrium of forces at the ends of the parabola. Example 3.15: Profiled tendons In most post-tensioned bridges the tendons are profiled using a combination of straight portions and parabolic curves.54 Beam with profiled prestressing tendon .) For the parabola in Span AB. 3. a tendon below the centroid generates hogging moment. 3. In a simply supported beam. For preliminary design purposes. 3. (This was covered in greater detail in Chapter 2.14 that the effect of the tendon below the centroid is to generate sagging moment in the central span.Page 109 Fig. P.53 Final BMD due to eccentric prestress force It is interesting to note from Example 3. it is required to find the BMD due to a prestress force.54. A parabolic profile generates a uniform loading.

(b) equivalent loads and BMDs due to prestress in fixed structure. (c) BMD after correction for discontinuity in BMD . (b) summary of all equivalent forces on beam Fig. 3. 3.56 Equivalent loads method: (a) system of fixities for analysis by moment distribution.Page 110 Fig.55 Equivalent loading due to profiled tendon: (a) equivalent forces in span AB.

3. moment distribution is applicable to the analysis of grillages but is tedious to apply for most examples. EI.56(a) and the associated BMD (Appendix A) is given in Fig. As a result. Step 2: The BMD associated with simultaneous rotations at B and C is identical to that derived for Example 3. The bridge is subjected to uniform vertical loading of intensity. Thus. The beam is analysed for this loading using moment distribution. Fig. 3. at x=l.55(a) it can be seen that the vertical component of the prestressing force at A is P sin θ≈ tan θ=0. 3.Page 111 At A. The obvious difference is that typical prestress loading is in the opposite direction to loading due to self weight. In CD. They are both equal to 0. The torsional rigidity.5 times the flexural rigidity. This bridge is long and narrow so it can be idealised by two beam members as illustrated in Fig.56(b). As there is no further discontinuity. . the vertical components can be found similarly. is 1. Step 1: The symmetrical system of fixities is illustrated in Fig.00833Pl in Fig.56(c). 3.12 and the vertical component of prestress is 0. 3. the complete equivalent loading due to prestress is as illustrated in Fig.52(c). the 1 P 1 slope is 0. Similarly. Example 3. 3. it is not necessary to provide a vertical translational fixity. GJ. Step 1: The two members are isolated from each other by the fixing of point B. 3. Hence. In this method. x=0 and the slope becomes − 0.57(b). only those grillages are considered in which there is no such joint displacement. the continuous bridge slab is represented by a mesh of discrete beams.57(a).08. wCD =wAB=0.15 serves to illustrate that the effect of profiled prestressing tendons can be quite similar to the effect of self weight in that it applies a uniform loading throughout the beam.12P. equilibrium of vertical forces requires a uniform loading of intensity: In BC. wBC=0. 3. Example 3.56(b). In most practical grillages.16:Torsion due to vertical loading When bridges are curved or crooked in plan. As there is a support there already. Step 3: To remove the moment discontinuity of 0. 3.55(b).56(b).7 Application of moment distribution to grillages A great many bridges are analysed by computer using the grillage analogy. vertical loading induces torsion as will be demonstrated in this example.2P/l. 3. described in detail in Chapters 5 and 6. From Fig. significant vertical translational displacements occur at the joints. this is the final BMD due to prestress in this beam. 3. In this chapter.2P/l. by symmetry.08P.52(c) is factored by this amount and added to Fig. which is illustrated in Fig. The result is illustrated in Fig. w. the intensity is. coincidentally.1P and the intensity of loading is. 3.14 and illustrated in Fig.

57(e). the bending moment diagram is as illustrated (in elevation) in Fig. However. (c) plan view of system of fixities. This is done in Fig. 3. From Appendix A. 3. The double headed arrows indicate rotational fixities where the positive direction is clockwise when looking in the direction of the arrow. the BMD for the two beams are illustrated in Fig. 3. 3.58(b) and it can be seen that there is a discontinuity of moment at B of (1− 2)wl2/8=0. 3. the BMDs for each of these beams is about the axis of that beam so the discontinuity of moment at B is not apparent from the diagram. two rotational fixities are required in orthogonal directions as illustrated in Fig. The internal bending moment at the left end of BC is wl 2/8 as illustrated in Fig. 3. it needs to be resolved into components parallel and perpendicular to AB. A unit rotation is first applied . there is a 1/√ discontinuity in torsion of wl2/(8√ 2).Page 112 Fig. In addition. At B.57(c) (rotation about two axes).293wl2 /8.58(a). (b) plan view of idealisation.57 Analysis of crooked bridge: (a) plan view of geometry. Vertical loading on AB in the fixed structure is applied to a beam which is fixed at one end. In plan. (d) elevation of applied loading and resulting BMD in AB while fixed. Step 2: The second step in moment distribution is to find the bending moment and torsion diagrams due to unit rotation at each of the points of fixity.57(d). (e) plan view of fixed BMD However. there is a transition between bending moment and torsion in the members. In order to compare this moment to that just left of B.

As there is no resistance to twisting at C. The discontinuity of moment at the joint can be seen by resolving the internal moments and torsions in Fig. 3.59(c). normalised for moment. 3. It is (3+3/2)EI/l=9EI/2l. is illustrated in Fig. the normalised version of Fig. In BC. 3.59(a). 3. The discontinuity parallel to Direction 2 at B is GJ/l+3EI/2l=3EI/l.60(a) and it can be seen that it generates no torsion and the BMD illustrated in Fig.60(d). 3. the application of a twist of 1/√ at B does 2 not generate any torsion in BC. dividing by this value gives the normalised version illustrated in Fig. 3. 3.57(c)) generates no bending but a torsion of GJ/l in AB.59(c) can be seen when the moments are resolved in Fig. The discontinuity of moment at B in the BMD of Fig.58 Plan views showing internal moment: (a) end moment in BC. 3. it is necessary to resolve the rotation into components parallel and perpendicular to that member as illustrated in Fig. Unit rotation at the end of AB results in the deformed shape and BMD illustrated in elevation in Fig.60(c). The .57(c)). 3.59(d). Hence the BMD due to unit rotation at B is as illustrated in Fig. 3. Applying a unit rotation in Direction 2 (Fig. 3.59(d). Step 3: In the third step.59(b).Page 113 Fig. the discontinuities in the fixed bending moment and torsion diagrams are removed by scaling and adding the diagrams derived in Step 2. 3. To determine the effect on member BC. (b) resolution of end moment in BC parallel and perpendicular to AB in Direction 1 (Fig.60(b). 3. 3.59(e). the rotation at the joint must be resolved into components as illustrated in Fig.

61(e) and (f).59(e) scaled by minus this value gives the moments and torsions illustrated in Fig.805wl2/8. 3. (c) BMD due to unit rotation. there is still a discontinuity parallel to Direction 2 of 0.60(d) to give the diagram illustrated in Fig. 3.57(c)) has now been removed at B.60(d). 3.61(d). This is removed by adding the diagram of Fig.61(a). . The corresponding bending moment and torsion diagrams are illustrated in Figs. 3. scaled by minus this value to give the diagram illustrated in Fig. 3. 3. (e) normalised free body diagram discontinuity in the fixed BMD of Fig. (d) free body diagram showing lack of equilibrium of moments at B.61(b).59(e) to give Fig. 3.58(b) is (1− 2)wl2/8=0. Adding the BMD 1/√ of Fig. 3. 3.293wl2 /8.59 Effect of rotation in direction 1 at B: (a) elevation of AB showing imposed unit rotation and associated BMD. 3.Page 114 Fig. (b) resolution of rotation parallel and perpendicular to BC. 3. 3. While the discontinuity parallel to Direction 1 (Fig. The new discontinuity now introduced parallel to Direction 1 is removed by adding a diagram proportional to Fig.61(c) and the discontinuity in that is removed by adding a diagram parallel to Fig.

3. (d) normalised free body diagram . (b) BMD due to unit rotation. (c) free body diagram showing lack of equilibrium of moments at B.Page 115 Fig.60 Effect of rotation in direction 2 at B: (a) resolution of rotations parallel and perpendicular to BC.

17: Torsion due to skew supports The skewed bridge illustrated in Figs. (b) cross-section. (e) final BMD.Page 116 Fig. (b) after correction of moments in direction 2. (f) final torsion diagram Step 4: To get an exact answer.61 Successive corrections to internal moments and torsions: (a) after correction of moments in direction 1. 3. 3.62(a) and (b) is long relative to its width and can be modelled using a single longitudinal member and a pair of outriggers at the ends as Fig. (d) after second correction in direction 2. the process must be continued until no discontinuity remains. (c) after second correction in direction 1. 3.62 Long skewed bridge: (a) plan view. (c) plan view of idealisation . Example 3.

(b) simplified system of fixities.0. 3.62(c).8. Fig. 3. This deck has flexural rigidities of (EI)ABC = (EI) DEF=0.Page 117 illustrated in Fig. total fixity must be imposed at B and E. 3. (c) fixed BMD.63(a).6 and torsional rigidities of (GJ)ABC =(GJ)DEF=2.0. (d) free body diagram showing lack of moment equilibrium in fixed structure .63 First step in analysis by moment distribution: (a) complete system of fixities with two rotations and one translation. The circles in this figure indicate translational fixities with a direction of positive upwards. Step 1: To isolate the members from one another. The symmetry of the system is exploited recognising that the three fixities at B are identical to the corresponding fixities at E. (EI)BE=3. (GJ)BE =4. i. It is subjected to vertical uniform loading of intensity w. it is necessary to fix against vertical translation and against rotation about both axes.e. This is represented diagrammatically in Fig.

3. The required twist of one end relative to the other is 2 sin θ the torsion required to generate such a twist (Appendix B) is (2 . This can be seen by resolving the unit rotation into directions parallel and perpendicular to the member as illustrated in Fig. 3. sin θ )(GJ)BE/l1=8 sin θ 1. (c) BMD associated with unit rotations at B and E .63(d). The member must also undergo bending in order to rotate at each /l end through cos θThe elevation showing the . 3.64 Effect of rotations at B and E: (a) plan showing resolution of rotations into components parallel and perpendicular to BE. but unnecessary in practice. the rotation in Direction 2 will be small. In the fixed structure. 12<<l 1). Applying a unit rotation to ABC and DEF about their own axes requires no moment as no torsional resistance would be offered by the bearings. By the same token. 3. However. The discontinuity is (wl 2/12) cos θ there is a moment of this amount to the right of as B and zero moment/ torsion on the other side of it. the vertical deflection at B and E will be relatively small and can be neglected. The discontinuity of moment parallel to the direction of fixity is found by resolving the moment reaction of wl2/12 parallel to the direction of fixity as illustrated in Fig.63(c). as to apply a unit rotation there would require a moment that is very large.64(a). to rotate member BE in Direction 1 requires it to be rotated and twisted.Page 118 The system of fixities illustrated is adequate. Fig.e. BE is fixed at each end and the BMD (Appendix A) is as illustrated in Fig. Therefore it is sufficient to fix the bridge as illustrated in Fig. 3.63(b). (b) elevation showing moments required to induce rotations in BE. As the two bearings are relatively close together (i. There is no torsion in the bridge in its fixed state. Step 2: The second step consists of applying a rotation at B and E.

3. 3. to apply a unit rotation in the direction of fixity generates this BMD plus a distribution of constant torsion throughout member BE of magnitude 8 sin θ1 . (b) normalised free body diagram . .63(d). Resolving parallel to the direction of fixity gives the discontinuity of moment corresponding to unit rotation: Normalising with respect to this discontinuity.65(a). results in the internal moments and torsions illustrated in Fig. /l The internal moments and torsions at B and E are illustrated in Fig. Fig.64(c). 3.65 Internal moments associated with rotations at B and E: (a) free body diagram with resolution of moments parallel and perpendicular to direction of fixity. 3. Thus.65(b) scaled by that amount. last BMD) is illustrated in Fig.65(b). Step 3: As stated in Step 1 and illustrated in Fig. 3. the corresponding BMD (Appendix B. 3. the discontinuity in moment parallel to Direction 1 in the fixed structure is (wl2/12) cos θThis is removed by adding Fig. The resulting diagram is illustrated in Fig.64(b).66(a).Page 119 required deflected shape is illustrated in Fig. 3. 3.

Page 120

Fig. 3.66 Corrected internal moments: (a) free body diagram showing moments after correction for discontinuity at B and E; (b) plan showing final BMD; (c) plan showing final torsion diagram

There is no discontinuity remaining in this diagram as the torsion and moment components are in equilibrium. Hence, no further distribution of moment is required. The final moment and torsion diagrams are illustrated in Figs. 3.66(b) and (c) respectively. It can be seen in Example 3.17 that the skew supports have the effect of introducing a small hogging moment at the ends of the bridge. If the skew, θwere zero, the bridge would in , effect be simply supported and the moment would be positive everywhere (sagging). On the other hand, if the skew were very large, the end hogging moment would be correspondingly large approaching a maximum of wl2/12. The skew also has the effect of introducing a significant distribution of torsion into the bridge.

Page 121

**Chapter 4 Integral bridges
**

4.1 Introduction

Integral bridges are those where the superstructure and substructures are continuous or integral with each other. While the concept is well established, many bridges built in the 1960s and 1970s were articulated with expansion joints and bearings to separate the superstructure from the substructure and the surrounding soil. In the 1980s and 1990s, many of these required rehabilitation due to serviceability problems associated with the joints. As a result, integral construction has recently received a great deal of attention and this form is likely to become much more widespread in the future. In the UK in particular, designers are now required to consider the use of the integral form for most shorter bridges (up to 60 m span and 30° skew).

4.1.1 Integral construction

There are many variations on the basic integral bridge. In the bridge of Fig. 4.1(a), the deck is composed of separate precast beams in each span. While in the past such a deck might have had a joint over the central support, a more durable form of construction is to make it continuous over the support using in-situ concrete, as illustrated. A bridge is shown in Fig. 4.1(b) in which the deck is continuous over the internal support and integral with the abutments at the ends. Figure 4.1 (c) illustrates another variation; this bridge is integral with both the abutments and the intermediate pier. While there are considerable durability advantages in removing joints and bearings, their removal does affect the bridge behaviour. Specifically, expansion and contraction of the deck is restrained with the result that additional stresses are induced which must be resisted by the bridge structure. The most obvious cause of expansion or contraction in bridges of all forms is temperature change but other

Page 122

Fig. 4.1 Integral bridges: (a) precast beams made integral over the interior support; (b) deck continuous over interior support and integral with abutments; (c) deck integral with abutments and pier

causes exist, such as shrinkage in concrete bridges. In prestressed concrete decks, elastic shortening and creep also occur. A simple integral bridge is illustrated in Fig. 4.2(a). If the bases of the abutments are not free to slide, deck contraction induces the deformed shape illustrated in Fig. 4.2(b) and the bending moment diagram of Fig. 4.2(c). Partial sliding restraint at the bases of the abutments results in the deformed shape of Fig. 4.2(d) and a bending moment diagram which is similar in shape to that of Fig. 4.2(c), but of a different magnitude. Time-dependent contractions in concrete bridge decks induce bending moments in integral bridges. While the magnitude of creep contraction is time dependent, creep also has the effect of relieving the induced bending moments over time. The net effect of this is that moments induced by creep contraction are small. Shrinkage strain increases with time but the resulting moments are also reduced by creep. Elastic shortening occurs in post-tensioned prestressed concrete decks during the application of prestress. If the deck is integral with the supports at the time of stressing, bending moments are induced. On the other hand, many integral bridges are constructed from precast pretensioned beams and the bridge is not made integral until after the pretensioning process is complete. In such cases, no bending moments are induced by the elastic shortening. Temperature changes are another major source of deck expansion and contraction. Temperature can be viewed as having a seasonal and hence long-term component as well as a daily or short-term component. The resistance of an integral bridge to movement of any type depends largely on the form of construction of the substructures. Three alternative forms are illustrated in Fig. 4.3. In each case, a run-on slab is shown behind the abutment. These are commonly placed over the transition zone between the bridge and the

Page 123

Fig. 4.2 Frame bridge subject to contraction: (a) geometry; (b) deformed shape if bases are restrained against sliding; (c) bending moment diagram if bases are restrained against sliding; (d) deformed shape if bases are partially restrained against sliding

adjacent soil which generally consists of granular backfill material. Figures 4.3(a) and (b) show two bridges which are integral with high supporting abutments and piled foundations. In such a case, a reduction in lateral restraint can be achieved by using driven H-section piles with their weaker axes orientated appropriately. An alternative form of integral construction is one in which abutments sit on strip foundations like the small bank seat abutment illustrated in Fig. 4.3(c). Minimising the sliding resistance at the base of these foundations helps to reduce the lateral restraint. Care should be taken in the design to ensure that bank seats have sufficient weight to avoid uplift from applied loads in other spans.

Page 124

Fig. 4.3 Ends of integral bridges: (a) deep vertical abutment; (b) deep inclined abutment; (c) bank seat abutment

**4.1.2 Lateral earth pressures on abutments
**

The lateral earth pressures (σ that the abutments of integral bridges should be designed for h) are those that take place during the maximum expansion of the bridge deck combined with any additional surcharge. The expansion has the effect of pushing the abutment laterally into the backfill. The resulting earth pressures developed on the abutment are dependent on the stiffness and strength of the backfill and on the amount of movement of the abutment. The maximum lateral earth pressure that can be sustained by the backfill is termed the passive pressure (σ ) which, for dry backfill at a depth z and no surcharge at ground level, is hp given by the expression: (4.1) where K p is the coefficient of passive pressure and γ is the unit weight of the backfill. The soil coefficient Kp may be estimated from Fig. 4.4 for a given angle of internal friction of the backfill and a given ratio, where δis the angle of interface friction between the a abutment and backfill. One design approach would be to use equation (4.1) directly to determine the maximum lateral pressure distribution on the abutment. This approach, however, is generally overly conservative as abutment movements are usually significantly less than those required to generate passive pressures. The preferred approach is one

Such an approach is described later in this chapter. these are in keeping with the h general guidelines set out in BA42/96 (1996).5) It will be seen later that the actual thermal expansion in integral bridge decks is closely comparable to that which occurs in a similar unrestrained deck (as the .3) and (4. (4.Page 125 Fig. Hret/10.4 Coefficients of passive earth pressure (horizontal component) for horizontal retained surface (after Caquot and Kersiel (1948)) involving an appropriate soil/structure interaction analysis which takes due account of the stiffness of the soil. or to a wall translation of Hret /20. 4.2) where (4. This method is based on experimental observations which indicate that movements to develop full passive pressures typically correspond to an abutment rotation equal to one-tenth of the retained height of soil. It follows that lateral pressures may be related approximately to the average displacement of the abutment over the retained height (δ ). av Expressions for σ emerging from this rationale are given below. A third (and commonly used) approach relates the pressure distribution on the abutment to the degree of mobilisation of its maximum (or passive) lateral capacity.4) (4.

Assume α the deck is 12×10− per °C and for From equation (4. The design extreme event for the determination of maximum abutment pressures is a 40° 6 increase in temperature. is Implicit in equations (4.2 Hret .6) and (4.5) when the depth exceeds approximately 1.Page 126 restraint offered by typical abutments and backfill is relatively small). For the latter case. Example 4. it is reasonable to assume that horizontal stress acting on both sides of the abutment are given by equation (4.1.1: Determination of design abutment earth pressures A 50 m long integral bridge has deep wall abutments which retain 6 m of well compacted granular fill. for a bridge deck of length L which experiences an increase in temperature of Δ δ may be calculated T.7): Figure 4. For most cases. av as: (4.4 indicates that Kp=17. it is necessary to quantify . As Hret=6 m and δ <Hret/20.5 for equation (4. Therefore.7) is the assumption that a bank seat experiences a lateral translation while a deeper abutment bends and rotates about a point just below the ground level on its inner face. av The unit weight of the soil (γ ) is soil Therefore for z<6 m (Hret ): 4. therefore.6) (4. The peak angle of friction of the fill is 45° and its dry density is 1900 kg/m3.7) where α the coefficient of thermal expansion of the deck.3) gives: and .3 Stiffness of soil The longitudinal expansion of integral bridge decks is resisted not just by the abutment supports but also by the backfill soil behind the abutments and the natural/imported soil beneath them.

9) where Gs is the specific gravity of the soil particles (typically 2. The design stiffness used for the calculation of such forces and moments should therefore be a maximum credible value. ρ.65) (after Lehane et al. e. The stress-strain relationship for soil is non-linear at strains in excess of about 0.65) and ρ is the w Fig. (1996)) . The value of the secant modulus at a given strain for a typical cohesionless soil (such as the granular type generally used for backfill) depends primarily on its density (or void ratio).01. The degree of compaction of backfill on site is often specified in terms of the dry density. e is the void ratio of the soil. (1996): (4. used as a reference stress and γ the shear strain which is taken to lie within the is 6 range 50×10 − to 0.Page 127 the restraint provided by the soil. patm is the atmospheric pressure (100 kN/m 2). Clearly.000 05 6 (50×10− ) and it is therefore common to refer to a secant modulus defined as the ratio of stress to current strain. a higher soil stiffness will lead to higher axial forces and bending moments in the deck due to its longitudinal expansion or contraction. by the expression: d (4. which is related to the void ratio. the level of confining stress and the loading history. A typical approximate relationship has been proposed by Lehane et al.8) where E s is the secant Young’s modulus in kN/m2. 4. p' is the mean confining stress less the pore water pressure in the soil.5 Secant Young’s modulus for granular soil (assuming Gs=2. This can only be achieved with a knowledge of the appropriate soil stiffness parameters.

4. an analysis of this type is often used as a first step to determine a limit on the stresses induced by deck contraction when the supports are partially fixed. However. (b) bending moment diagram from example 3.1 Contraction of bridge fully fixed at the supports The case is first considered of an integral bridge in which no translational movement can occur at the base of the abutments. 4. derived using equation (4. e. The secant Young’s modulus. sections. 4. As a result. These conditions are applicable if the abutment foundations are cast in very dense soil or rock.5 for a range of in-situ dry densities (ρ).8) ). the principal uncertainty relates to the resistance to movement at the bases of the piers and abutments.8). and/or shrinkage. mean confining stresses (p').6 Contraction of frame rigidly fixed at supports: (a) geometry. Fig. the resistance provided by such soil to the contraction of a bridge deck is usually small. 4. 4. Thus. is plotted in Fig.Page 128 density of water. Guidance on appropriate values for ρ p' and γ specific cases is given in subsequent for d. This means that.2. in an analysis to determine the effects of elastic shortening.5 can be used to estimate the secant Young’s modulus for cohesionless soil.2 Contraction of bridge deck There is generally a lesser height of soil in front of bridge abutments than behind them.5 . and shear strains (γ Equation (4. specification of the dry density effectively dictates the void ratio. creep. d or Fig.

10): (4. 4.7. The axial contraction induced bending in the abutments and. may be calculated using equation (4. k hori and krot are the stiffnesses per metre length of strip foundation for vertical.10) where kvert. embedded to a depth of between 0.20(c) shows that only 7% of the potential force is applied to the abutments). 3. Thus. In that case.6(a) was considered in Chapter 3 for an axial contraction due to temperature of 20° in the deck (ABC) (Example 3.5).5 m and 1. there was a much greater tendency for the temperature decrease to cause the abutments to bend than to cause an axial stress in the deck. Quantification of the pile resistance is beyond the scope of this text and interested readers are referred to books such as that of Tomlinson (1994).0 m below the ground level on the inside of the abutment as illustrated in Fig. However. Strip foundations or pile caps are commonly founded at around 0. upper bound estimates of the secant Young’s modulus of elasticity. the decrease in temperature would generate a large tensile force in the deck and there would be no contraction.5–1.8) assuming a p' value equivalent to the foundation bearing pressure and a shear strain (γof 0. If the ends of the deck were fully prevented from contracting. Abutments and piers are generally either supported on foundations bearing directly on the ground below or on pile caps underlain by piles.0 m below the ground level are given in equation (4. The complete bending moment diagram is illustrated in Fig. bending in the deck also. Es. Design spring stiffnesses on the inside of the abutment for a strip foundation of width B. only 5% of the potential level. However.Page 129 The bridge illustrated in Fig. 4. Expressions for the stiffness of such springs have been deduced here from relationships provided by Dobry and Gazetas (1986) for an elastic soil. The soil around the strip foundation can be idealised by a number of linear elastic springs. and a relatively large contraction.2. due to the integral nature of the bridge. Conservative. 4. that resists bridge contraction. 4. together with sliding resistance at the base of the pad. horizontal and rotational displacement respectively. It is this small depth of soil. ) .2 Contraction of bridge on flexible supports Most bridges are constructed on supports which have some degree of flexibility. The end result for that example was a relatively small axial tension in the deck. the bridge was fully restrained at the base of each abutment and pier. some movement of the deck was possible through bending in the abutments. the resistance of the abutments to movement was considerably less than the axial stiffness of the deck (Fig.6(b).001.

It is required to determine the distribution of bending moment and axial force generated in the deck given that the Young’s modulus for the concrete is 30×10 6 kN/m2. 4. The foundation is assumed to be working under a bearing pressure of 300 kN/m2 and the breadth of the strip foundation is 2. ρ. The degree of compaction has been controlled by specifying a dry density of backfill. gives a void ratio of: . d of 1900 kg/m 3. (b) deep abutment Example 4. Inverting equation (4.8 is subjected to a shrinkage strain of 200×10− .7 End of integral bridge showing shallow depth of soil on inside: (a) bank seat.5 m.2: Contraction for shallow strip foundation 6 The bridge illustrated in Fig.Page 130 Fig.9) and assuming Gs=2. 4.65.

Page 131 Fig.8) then gives: Equation (4.2: (a) elevation.8 Bridge of Example 4. (b) detail at abutment Substituting in equation (4. 4. the modulus of elasticity of concrete and the cross-sectional area (per metre run): .10) then gives spring stiffnesses per metre run for the supports of: 6 The equivalent load for a shrinkage strain of 200×10− is the product of the strain.

9.10(d). (b) axial force diagram from computer analysis. Fig. Subtracting the associated axial force diagram gives the actual distribution of axial force generated by the shrinkage.10(a)–(c).10 Analysis results: (a) deflected shape. The frame was analysed using a standard analysis package which gave the deflected shape. 4. 4. (d) corrected axial force diagram . illustrated in Fig. (c) bending moment diagram. 4. 4. No adjustment is necessary for the deflected shape or bending moment diagram.9 Computer model for bridge of Example 4. (b) associated axial force diagram Fig. 4. axial force and bending moment diagrams illustrated in Fig.2: (a) equivalent loading and springs.Page 132 The equivalent loads and the associated axial force diagram are illustrated in Fig.

this corresponds to a maximum flexural stress of 2. 2.3 Conventional spring model for deck expansion Soil generally provides considerably more resistance to deck expansion than contraction as abutments are generally backfilled up to the level of the underside of the run-on slab (Fig. 4. The average shear strain in the backfill must be less than 2δ and could conservatively be assumed as about 2δ /H /3H. the bending moment at the ends are more significant at 568 kNm. to tend to an equilibrium density compatible with the strain amplitude that it is regularly subjected to.7). The horizontal stresses acting on an abutment following cyclic expansions and contractions of the deck remain approximately constant to depths of up to 6 m and typically have magnitudes of between 25 kN/m2 and 50 kN/m 2 (depending on the type of compaction plant used). The conventional spring model represents the backfill soil and soil beneath the abutment by a series of spring supports. Nonlinear elastic finite-element analyses by Springman et al.2 is interesting in that it gives an indication of the magnitude of bending moments and axial forces that can be generated by a restrained shrinkage. Thus. Assuming uncracked conditions. and others: 1. Out of a total potential 6 shortening of 6 mm (200×10− ×30000 mm) at each end. However.3 mm does generate distributions of stress in the frame. 4. This observation suggests that the use of a constant soil stiffness value with depth (for a given strain) is reasonably realistic. The maximum shear strain induced in the backfill as the deck pushes out the abutment a distance δ is approximately 2δ where H is the height of the retained fill.3 N/mm 2. The axial tension is relatively small at 337 kN corresponding to a stress in the deck of less than 0. for example. with time. 5. an average shear strain must be assumed. However the restraint which prevents the remaining 0. (1996) support the validity of this assumption. the stresses generated by an increase in deck temperature. The stiffness of the soil is influenced by the shear strain in the backfill. Some notable features have been observed from experimental studies by Springman et al. 3. There is some evidence to suggest that the granular backfill at this stage will have increased in density by a maximum of about 20% from its as-placed density for loose fills and by a maximum of about 10% for well compacted fills.Page 133 Example 4. Such a model is imperfect as it does not allow for shear transfer within the soil as there is no interaction between the . will be affected significantly by the properties of the soil behind the abutments. The selection of a suitable soil stiffness value (Es) is essential for appropriate modelling of the backfill. To adopt a single soil /H.7 mm is predicted to actually occur. Cyclic variations in temperature (and associated expansions and contractions of the deck) cause the backfill to compact and.4 N/mm 2. stiffness value. (1996).

9) and assuming Gs=2. has been developed for the horizontal spring stiffness per square metre. however. The resulting distribution of bending moment is required given that the culvert is made from concrete with an elastic modulus of 28×106 kN/m 2 and a coefficient of thermal expansion of 6 12×10− per °C.e.11 Culvert of Example 4.: Inverting equation (4. An approximate expression. ρ. i. It does. assuming linear elasticity. 4.65. has been specified as 1600 kg/m 3.11) is illustrated in the following example. It is assumed that the density of the backfill reaches an equilibrium value 20% in excess of that specified.11) The application of equation (4. gives a void ratio of: Fig.3 .Page 134 springs. L: (4. The d piles are assumed to provide insignificant lateral restraint to the deck.3: Conventional spring model The culvert illustrated in Fig.11 is subjected to an increase in temperature of 20 °C. have the advantage of simplicity and is considered here because it remains a popular approach among bridge engineers. of the backfill behind an abutment of depth H and transverse length. The dry density of the backfill. Example 4. 4.

the product of the temperature increase. The equivalent loading is: and the associated distribution of axial force is illustrated in Fig. 4.11): The model for a 1 m strip of the frame is then as illustrated in Fig.e. i. Then equation (4.12(b).12(a). a horizontal stress of p'=50 kN/m2 is assumed. the coefficient of thermal expansion and the distance of the abutment from the stationary point (the centre of the culvert): In accordance with Note 3 above. 4. the average shear strain in the affected backfill is then: On the basis of Note 2.8) gives: The horizontal spring stiffness is then given by equation (4. The bending .Page 135 To estimate the average shear strain induced in the backfill. the expansion of the culvert is estimated as its unrestrained value.

3: (a) springs and equivalent loads.19 mm.12 Computer model for culvert of Example 4.20 mm assumed in the estimation of shear strain. . iteration was not considered necessary. The deflection found from the computer analysis was 1. 4.13 Bending moment diagram for Example 4. 4. this is the final distribution of moment due to the expansion. The moment in the abutments can be seen to change sign through its length due to the flexible nature of the horizontal support. 4. As there was no associated distribution of bending moment.Page 136 Fig. (b) associated axial force diagram Fig. As this is similar in magnitude to the deflection of 1.3 moment diagram was found from a computer analysis and is illustrated in Fig.13.

14 Stiffness components at top of abutment: (a) unit translation. The purpose of the analyses was to provide credible upper bound estimates of soil resistance. base sliding or slip on the abutment stem were not permitted). i. It was therefore assumed conservatively that the soil had limitless compressive and tensile strength (e. Fig.4. no passive failure or abutment lifting were allowed) and that no slip between the abutment and the soil occurred (e.e.4 Modelling expansion with an equivalent spring at deck level An alternative to the conventional spring model is presented here which has a number of advantages over the traditional approach. This method does not. He conducted a series of finite-element analyses which involved the application at the top of the abutment of (i) a horizontal displacement δ with zero rotation. provide details concerning the distribution of moment in the abutment or the pressure distribution in the soil.14).g. The approach used to derive the spring constants represented the soil as a complete mesh of finite elements rather than a series of springs and is therefore considered theoretically more sound than the conventional spring model. and (ii) a rotation θ with zero horizontal displacement (Fig. 4.Page 137 4. the forces and moments associated with passive movements which occur as a consequence of deck expansion.1 Development of general expression Lehane (1999) determined the forces and moments associated with lateral displacement and rotation of the top of an abutment with retained backfill. However. given that relatively small movements are required to reduce pressures to their minimum (active) values on the inner face of the abutment. (b) unit rotation . however. This technique consists of modelling both the abutment and the surrounding soil with an equivalent lateral and rotational spring at deck level.3. 4. 4. described in Section 4. the analyses assumed that any soil present on this side did not contribute to the resistance.g.

14) B (m) (Fig.05 1.14) Fig.12) were the most important factors controlling the magnitudes of the lateral force (Fh) and moment (M) at the top of the abutment (Fig. (4.14) .05 3 m −. The values of Fh and M were also seen to increase systematically as the base width (B) increased and its height (H) reduced.15 End part of frame bridge showing locations and directions of fixity Table 4. 4. r.13) where f 1 and f 2 are functions of the ratio.14) for r>0. 4.1.5 H (m) (Fig. H/B which are given by equation (4.14). All values in this stiffness matrix can be reduced by 15% if friction between the abutment and soil is considered negligible. defined as: (4.5− 12 0. 4.5×106 >0.0×104− 2.1 Range of parameters used in derivation of equation (4. They are given here in matrix form: (4.Page 138 It was found that the flexural rigidity of the abutment (EIa ) and the ratio.5–3. Best-fit expressions were obtained for Fh and M for the range of parameter values given in Table 4.13) Parameter Es(kN/m2 ) EI a(kNm /m) 2 3 r=Es/EIa(m− ) Allowable range 10000− 500000 1. 4.

15) and (4.1. Ld and Id are the cross-sectional area.Page 139 For the range of parameters listed in Table 4. equating the K12 (and K21) terms gives: (4.17) where Heq and Ieq are the equivalent abutment height and second moment of area respectively. This could readily be achieved in computer analysis programs by allowing the appropriate stiffness terms to be changed in the program to those given in equation (4. When the bridge is embedded in soil and this is taken into account. the terms involving Ia and H are replaced with terms from equation (4.16) A comparison of equations (4. the stiffness matrix. [K]. is: (4. gives: (4. When a frame bridge with an abutment height of H is fixed rigidly at the supports and the system of fixities illustrated in Fig.13) was found to predict values of F h and M to within 10% of the values given by the finite element analyses. Similarly. second column) terms in equations (4.16).15) and (4.18) . in the absence of soil.13) with the result that equation (4. Alternatively.15 is used.16). 4. 4. equation (4.15) becomes: (4.15) where A d. Equating the K22 (second row. it is possible to allow for soil in a conventional structural analysis program through the use of an equivalent abutment second moment of area and height and the addition of a horizontal (translational) spring at X. span length and second moment of area of the deck respectively.15.16) shows that the influence of soil can be taken into account by analysing a model of a form similar to that illustrated in Fig.

. the parameters f1 and f2 approach their minimum values of 0.22) Substituting for f 2 in equation (4.20) To make the first terms (K11) equal requires a further adjustment which can be achieved by the addition of a linear horizontal spring at X of stiffness: (4. substituting for f 1 and f2 in equation (4.23) Finally.20) gives an equivalent height of: (4.33 and 0.18) can be simultaneously satisfied by selecting an equivalent abutment second moment of area equal to: (4.40 respectively.2 Expansion of frames with deep abutments The equivalent single-spring model can be simplified for the case of deep abutments. As a result.17) and (4.21) 4.24) These equations can be used to estimate the properties of an equivalent frame for an integral bridge with deep abutments.19) The equivalent abutment height is then: (4.21) gives a spring stiffness of: (4. For values of (H/B) in excess of 10.4.Page 140 Equations (4. the equivalent abutment second moment of area can be set equal to the actual second moment of area without great loss of accuracy: (4.

12) is then: The equivalent height of abutment is then.4 . is: Fig.24): The equivalent frame and loading are illustrated in Fig.Page 141 Example 4. as for Example 4.4: Equivalent single-spring model for frame with deep abutments The equivalent single-spring model is used to determine the maximum moment in the culvert illustrated in Fig.16. 4.3 to be: and the second moment of area of a 1 m strip of the abutment is: The ratio defined by equation (4. The magnitude of the equivalent loads. The elastic modulus of the soil is found as for Example 4. 4.16 Computer model for bridge of Example 4. from equation (4.3.11 due to a temperature increase of 20°.23): The stiffness of the single spring on each side is given by equation (4. 4. The concrete has an elastic 6 modulus of 28×106 kN/m 2 and a coefficient of thermal expansion of 12×10− per °C. The dry density of the backfill has been specified as 1600 kg/m 3.

4. 4.Page 142 Fig. The model was analysed using a standard computer program and the resulting bending moment diagram is illustrated in Fig. is likely to be more reliable than the value found in Example 4.18.3 Expansion of bank seat abutments Equations (4. with a finite strength defined by its friction angle. the deflections per unit load can be seen to be significantly greater. 4.18. It is important to remember that the distribution of moment in the abutment is not realistic. In the example of Fig. is that shallow abutments are more likely to slide than deep ones and will therefore offer less restraint to deck expansion than equations (4. the magnitude of moment in the deck. the true shape of this distribution will be similar to that given in Fig. 4. .4 The associated axial force diagram is as illustrated in Fig. On the other hand.14) would suggest.13). of course. 87 kNm. This implication arises because of the assumption that the soil is an elastic material with infinite strength and that no sliding along the abutment base can take place. Similar results can be shown for moment/rotation functions and for force/rotation and moment/deflection functions.17.13) and (4. it can be seen that the effective lateral stiffness for a movement at the top of the abutment of 10 mm is only about half that of the purely elastic case. linear.13. In Figure 4.14) imply that an abutment provides a greater resistance to deck expansion if it has a lesser depth of embedment (H). When the soil is linear elastic and infinitely strong. It is therefore recommended that this equation be used in preliminary analysis and that a finite- . 4.17 Bending moment diagram from computer analysis of bridge of Example 4. predictions from finite-element analyses are presented of a horizontal force/deflection relationship. the function is. Effective rotational stiffnesses at this lateral movement are about 75% of the purely elastic case. when the soil is treated as an elastic perfectly plastic material. of course.4. The influence of a limited soil strength on the resistance offered by a bank seat is illustrated in Fig. 4.12(b). It is not possible to generalise the observations made from calculations such as those summarised in this figure other than to say that the restraint provided by bank seats will be less than that predicted by equation (4. The reality. 4.3.13) and (4. However.18(b).

section through bank seat. The ratio defined by equation (4.3 but using the smaller abutment height. 4. The elastic modulus for the soil is found in the same manner as for Example 4. foundation bearing pressure= 200 kN/m2. soil friction angle. r=0.18 Finite-element analysis results for bank seat abutment (E s= 100000 kN/m 2. Example 4.5: Equivalent single-spring model of bank seat The equivalent single-spring model is used to determine the maximum moment in the culvert illustrated in Fig.19 due to a temperature increase of 20°C.Page 143 Fig.31. The 3 dry density of the backfill has been specified as 1600 kg/m .12) is then. (b) horizontal force/displacement relationship ): (a) element soil/structure analysis incorporating a realistic constitutive model for the soil is performed if the effects of deck expansion have a significant influence on the final bridge design. 4. Ec=30×10 6 kN/m2 . and the second moment of area of a 1 m strip of the abutment is Ia =0.018 m4. . The concrete has an 6 elastic modulus of 28×106 kN/m 2 and a coefficient of thermal expansion of 12×10−/°C.

19 Bridge of Example 4. H/B. f1 and f2 are calculated from equation (4. the ratio of embedment depth to foundation breadth.83. is 2.Page 144 Fig.21): . from equation (4.20): The equivalent abutment second moment of area is given by equation (4. the spring stiffness is. 4.5 For this example.19): Finally.14): The equivalent height is then calculated directly from equation (4.5/3=0. The parameters.

It is clear from Fig.22. The effect of such a slab is to allow relative rotation between the deck and the run-on slab while preventing relative translation.4. is: This model was analysed and the bending moment diagram is illustrated in Fig. This is achieved in many cases by the installation of a run-on slab as illustrated in Fig. 4. Preventing relative vertical translation significantly improves the rideability for vehicles travelling over the bridge.20 Computer model for bridge of Example 4. this must be accommodated if premature deterioration of the pavement is to be avoided. 4. In effect.21 Bending moment diagram from computer analysis of bridge of Example 4. it transfers the relative horizontal movement from the end of the deck to the end of . The maximum magnitude of moment in the deck due to the expansion is 114 kNm.5 Run-on slab It has been seen in this chapter that soil provides some restraint against deck movement in integral bridges but that most of the movement still takes place.3 and 4. Preventing relative horizontal translation is not so simple.Page 145 Fig. 4.18 that this result is quite conservative. 4. 4. The magnitude of the equivalent loads. the bridge still expands and contracts relative to the surrounding soil and the incorporation of a run-on slab does not prevent this. 4. 4.20.5 Fig. On a road bridge. as for Examples 4.21. Clearly.5 The equivalent frame and loading are illustrated in Fig.

This is because existing models which attempt to simulate the soil’s response to a complex history of cyclic straining are very approximate. This approach is widely adopted as the failure of a joint at the end of a run-on slab is a minor maintenance problem whereas a leaking joint at the end of a deck can result in deterioration of the bridge itself. difficult to use and require measurement of a large range of representative geotechnical parameters from cyclic laboratory tests. 4. It has been shown max by Springman et al. could be assumed to vary approximately with the height of the retained fill (H). to contract and increase in density in response to cyclic straining. whatever its density. Much larger settlements occur in initially loose backfills where considerable volumetric contractions take place before an ‘equilibrium’ density is attained. The extent of the settlement trough is also controlled by the amount of backfill subjected to cyclic abutment movements and therefore. max between about 10δ 20δ well-compacted fill for both deep abutments and bank seats. and in The assessment of the required length of the run-on slab relies on observations of measured behaviour and engineering judgement. An asphaltic plug joint positioned at the juncture between the run-on slab and the bridge approach road is commonly used to facilitate horizontal movements. Run-on slabs are designed to span the settlement troughs that develop behind the abutments of integral bridges.22 Run-on slab the run-on slab. Settlement troughs arise because of the tendency for cohesionless backfill. . (1996) that. for a given movement of the top of the abutment. after many cycles of imposed lateral movement δδ varies . Settlement profiles may be approximated as having a triangular shape varying from a maximum settlement (δ ) at the abutment to zero at a distance Lt from it. however.Page 146 Fig. Such straining is imposed on the backfill by the abutment which moves in response to thermal movements of the deck. Analytical prediction of the shapes and magnitudes of settlement troughs is. Both analytical and model test studies have shown that the surface settlement trough tends to an equilibrium profile after a large number of cyclic abutment movements of the same magnitude. not commonly attempted by bridge designers.

When the in-situ concrete is cast. (1996) suggest that the length of the trough (Lt ) is unlikely to exceed the limits given in Table 4. 4. the length of run-on slab required for the bridge of Example 4.1H These observations and those taken during centrifuge model tests by Springman et al.19) is calculated.23.2. 4. 4. as illustrated in Fig. When the in-situ concrete Fig.1H can be assumed from Table 4. Hence. (b) section A—A . As an example. 4. the run-on slab should have a length of at least 2.5)=5.Page 147 Table 4.2 Approximate upper limits on expected trough lengths Granular fill Well compacted Deep abutments Bank seats 0.6H 0. the precast beams are simply supported and the self weight of the bridge induces a sagging moment. 4.9H Loosely compacted 1.25 m.2.1(2. As the backfill is loosely compacted (density= 1600 kg/m3) and the abutments are not deep.23 Composite integral bridge made from precast and in-situ concrete: (a) elevation.5 (Fig.6 Time-dependent effects in composite integral bridges Many integral bridges are constructed using a combination of precast prestressed beams and in-situ concrete such as illustrated in Fig. a maximum trough length of 2.24(a).4H 2.

24(c)). 4. 4.24 Bending moment diagrams due to short-term loading: (a) due to self weight. the hogging prestress moment combines with hogging due to applied loading. making it very difficult to prevent tension in the beams. this can be quite uneconomical in its use of prestressing strand. The problem can be countered by the debonding of strands near the ends to prevent the prestress force from acting there. However. This can be quite difficult at points such as A in the figure as these same pretensioned beams must be designed to resist substantial sagging moment near mid-span.25 Detail near support of composite integral bridge . 4.Page 148 subsequently sets. 4. Further. current UK practice is to design to ensure no tensile stress whatsoever in the prestressed beams.25 and it is often necessary to provide great quantities of closely spaced bars to prevent excessive cracking. Fig. Near the supports. the bridge acts as a frame and imposed traffic loading generates sagging near the centres of the spans and hogging over the supports (Fig. The resultant prestress force is therefore designed to be below the centroid near mid-span (Fig. 4.24(b)). The net result is substantial sagging near the centres of the spans and some hogging over the supports (Fig. (b) due to imposed traffic loading.25) to ensure a hogging prestress moment. 4. Non-prestressed reinforcement is generally provided at the top of the deck over the supports to resist the hogging moment as illustrated in Fig. (c) due to self weight plus traffic loading ((a) plus (b)) Fig.

26 Equivalent loading due to a prestress force. at a mean eccentricity.e. (c) total bending moment diagram due to prestress . 4. 4. the distributions of bending moment change due to creep in the prestressed beams.27 Effects of prestress on composite integral frame: (a) equivalent prestress loading and bending moment diagram at time of transfer of prestress. the period immediately following the construction of the bridge.Page 149 All of the above effects occur in the short term. e: (a) complete prestress force applied at ends. i. P. (b) debonding near ends of beam Fig. (b) equivalent prestress loading and bending moment diagram due to creep strains after frame is made integral. In the long term. The equivalent loading due to prestressing strands below the beams’ centroid is illustrated in Fig.

4. further curvature is resisted and the resulting distribution of moment is as illustrated in Fig. 4. As they are simply supported. If some strands are debonded.Page 150 Fig. the equivalent moments at the ends are less but a further increment of equivalent moment is applied at the points where debonding ceases (Fig.28 Detail at support showing points where long-term cracking is likely to occur Fig. Due to creep. 4. They propose the assumption of a sagging moment of 750 kNm (per beam) for spans in the 20–36 m range where the beams are 1100 mm deep or greater. they suggest designing for a moment of 600 kNm. such curvature is unrestrained so it results in instantaneous strain and a moment which is the simple product of prestress force and eccentricity. 4. For smaller beams. They suggest that there is little point in trying to determine the distribution of bending moment that develops in the long term as there are few creep/shrinkage computer models that give consistently reliable results. It can result in cracking at the bottom of the deck over the supports as illustrated in Fig. . The long-term result is a distribution of prestress moment such as that illustrated in Fig.27(b). 4. This phenomenon is particularly significant if the bridge is made integral when the precast concrete is young as this causes most of the creep strain to occur when it is in the integral form.27(a). particularly at the interface between the precast and insitu concretes. 4. they calculated the maximum longterm sagging moment for beams made integral when between 21 and 100 days old.26(b)). 4. When prestress is first applied below the centroid. the beams hog upwards as illustrated in Fig. Clark and Sugie (1997) carried out a parametric study of the time-dependent effects in composite integral bridges. When the bridge is made integral.28. In a study of continuous bridges made integral at the interior supports.27(c). these hogging strains increase with time.26(a).

Orthotropy represents the most general material behaviour usually considered for bridge decks. 5.2 Thin-plate theory Slabs used in the construction of bridge decks are generally thin relative to their span lengths. To understand the basis of such programs and their limitations. it is necessary to first consider the theory of bending of plates. Thick plates correspond to deep beams and are not considered here.1 Introduction The development of a closed-form solution for bridge slabs under the action of applied load is achievable for a limited number of cases.Page 151 Chapter 5 Slab bridge decks—behaviour and modelling 5. Fortunately. A further subset of orthotropic materials are isotropic . 5. slabs can readily be idealised using one of a number of well-proven methods and analysed using structural analysis programs. but is generally impractical.1 Orthotropic and isotropic plates A material in which the behaviour in each direction is independent of the others is referred to as anisotropic. Such slabs can be assumed to behave like thin plates which can be thought of as the twodimensional equivalent of beams. except that bending takes place in two mutually perpendicular directions in the plane of the plate. Thin plates get their strength from bending.2. in a similar way to beams. A subset of anisotropic materials are orthotropic materials in which the behaviour varies in mutually perpendicular directions (X and Y) only.

5. c' and d' as illustrated in Fig.: x. Although this type of material is rarely found in bridge construction.3.2 shows a small segment of plate with dimensions δ y and a cube of material in that segment a distance z above the origin which x×δ has a height of. it is common practise to extend it to include geometric orthotropy. Considering point b. the points a. a distance δ from a. b. such as timber. but the same geometric properties. isotropic plate theory can be used with reasonable accuracy for the analysis of many bridges. Hence the length of a'b' projected onto the X axis is: .2 Bending of materially orthotropic thin plates Figure 5. the cube both moves and distorts. 5.e. such as reinforced concrete slabs with significantly different amounts of reinforcement in the two directions or voided slabs. Many bridge slabs possess different second moments of area in two directions. Considering initially the X− plane. In the following sections. applied. Z c and d shown in Fig. equations are derived assuming the plate to have a uniform depth but they are subsequently extended to decks which have different second moments of area in orthogonal directions.2. A materially (or naturally) orthotropic plate is composed of a homogeneous material which has different elastic properties in two orthogonal directions. the theory of materially orthotropic thin plates is developed. Thus. at which point. The displacement of point a in the X direction is denoted u.1 shows a portion of a thin plate in the X− plane.b'. While the theory is strictly only applicable to cases of material orthotropy. Figure 5. When a load is z. This type of plate is not typical of that found in bridge decks but is frequently used as an approximation of actual conditions. the x displacement at that point in the X direction. will be u plus the change in u over the distance δ i. Such a plate might be constructed of a material where the microstructure is orientated in two mutually perpendicular directions. This implies that the plate has a uniform thickness and hence the same second moment of area in both directions but different moduli of elasticity.Page 152 materials in which the behaviour in all directions is the same.2 move to a'. These types of slab are referred to as geometrically (or technically) orthotropic. z=0. 5. The origin of the axis system is at Y mid-depth in the plate. δ In this figure the thickness of the plate is taken to be d.

5.2 Segment of thin plate and elemental cube of material .Page 153 Fig.1 Portion of thin plate and co-ordinate axis system Fig. 5.

Page 154 Fig.3) . the strain in the X direction is: (5. it can be shown that: (5. if v and w are the displacements in the Y and Z directions respectively.2) and: (5.3 Distortion of cube of material in X. 5.Z plane By definition.1) Similarly.

the difference between c'a'b' and cab.3. i. . a number of assumptions are made to simplify the mathematics involved.: (5. i. α and βReferring to the figure: . 5. As can be seen in Fig. The first of these assumptions is that there is no strain in the Z direction.Page 155 The shear strain in the X− plane is defined as the change in the angle.5) (5. The physical meaning is .e. this reduces to: u/∂ The other component of strain can be found similarly to be: Hence the shear strain is: (5.4) Similarly the shear strains in the X− and Y− planes are respectively: Y Z (5.4 illustrates the implications of this assumption. Figure 5. or that w is a function of x and y only.e.7) This implies that w is independent of z.6) In thin-plate theory. cab from the original Z 90°. there are two components. As ∂ x is small.

This assumption is again a simplification of the true behaviour.9) The consequences of this are shown in Fig.8) (5. In other words.5 where it can be seen that the 90° angle of cab is preserved in the distorted c'a'b'. 5. the depth of the slab remains unchanged throughout.Page 156 Fig. but is justified by the fact that. i. . Clearly this is a simplification but the strains in the Z direction are generally so small that they have negligible effect on the overall behaviour of the bridge slab.4 Segment of plate showing uniformity of distortion in Z direction that there is no compression or extension of the bridge slab in a direction perpendicular to its plane. The second assumption which is made is that the deflection of the plate is caused by bending alone and that shear distortion makes no significant contribution.e. and although shear strains are small. a means for determining shear stresses will be required. Notwithstanding this. Such a method is presented later in this section. concrete bridge slabs do not have great shear strength. and all points deflect vertically by exactly the same amount as the points directly above and below them. 5. bridge slabs being relatively thin.: (5. their behaviour is dominated by bending rather than shear deformation.

8) gives: As w is independent of z. at z=0. Substituting this into equation (5. there is no displacement in either the X or Y directions at z=0.5 Segment of plate in X− plane showing assumed lack of shear distortion Z Rearranging equation (5. this implies: (5.Page 157 Fig.10) implies that the constant C is zero giving: (5. As the origin is located at the centre of the plate and bending is assumed to occur about that point.11) . Hence. 5. u and v are both zero.10) where C is a constant of integration.

a similar expression can be derived for v: (5.20) (5. the equations is are similar.11) and (5. In thin-plate theory.12) into equations (5.1) and (5.15) respectively then gives: (5.9).5) gives: (5.21) .19) (5.16) (5.14) Similarly equation (5.17) (5. but there are now curvatures in the X.12) Substituting equations (5.13)–(5. the curvature is defined as: where κ the curvature and R is the radius of curvature.Page 158 By rearranging equation (5.16)–(5.18) Substituting equations (5.18) into equations (5.13) (5. Y and XY directions which are given by: (5.3) respectively gives: (5.15) In the flexural theory of beams.

Expressions are now developed for the corresponding stresses. (b) three-dimensional body showing the effect of stress in the axial direction on strains in the orthogonal directions . as is indicated in Fig.and three-dimensions: (a) one-dimensional bar. as 2 κ=∂w/∂2 is independent of z.Page 159 Examination of equation (5. From this. This is generally a reasonable assumption for slab bridge decks. 5. 5. expressions were established for the various strains in a thin plate. strains in the other two directions become significant.6(b).19) shows that strain in the X direction is a linear function of z.6 (a) shows a onedimensional bar subjected to a tensile force. The only significant strain in this system is in a direction parallel to the axis of the bar. By defining the X axis as the direction of the applied force. it follows that plane sections remain plane. . elasticity. Equation (5. Such cases are discussed further in Chapter 7.22) Fig.6 Distortion in one.3 Stress in materially orthotropic thin plates In the previous section. σand modulus of . E. the strain in that direction is given by: (5.20) shows that the same applies to the strain in the x x Y direction. by: In the three-dimensional case. This strain. 5. εis related to the stress. as is generally assumed in beam theory. but some cases do exist where this is not so. Figure 5.2.

An expression for strain in the X direction for the case of an orthotropic material with the elastic constants varying in the X and Y directions is then given by: and likewise the strain in the Y direction is given by: In matrix format this becomes: and by rearranging and inverting the matrix we get: which yields expressions for the stresses as follows: (5. y z made of a homogenous material and that the elastic constants (Ex. ν etc. E y and Ez are the moduli of elasticity in the X. the stress in the Z direction is small and the Poisson’s ratio is generally small for bridge deck materials.Page 160 where E x.22) assumes that the plate is x. each other. and ν νand νare the corresponding Poisson’s ratios. Equation (5. Y and Z directions respectively.24) . Consequently the last term of equation (5. as is appropriate for the materially orthotropic (or anisotropic) case.22) can be ignored.23) (5.) are independent of x. For a thin plate in bending.

Page 161 The shear modulus, Gxy, is defined as the ratio of shear stress, gives: to shear strain, γ, which xy

(5.25) Substituting equations (5.19)–(5.21) into equations (5.23)–(5.25) respectively gives expressions for the stresses in terms of curvature: (5.26)

(5.27)

(5.28)

**5.2.4 Moments in materially orthotropic thin plates
**

Figure 5.7 shows a small cube taken from a thin plate with the associated normal stresses σ, x σ and σand shear stresses. It is well established that, to satisfy y z

Fig. 5.7 Elemental cube of material showing normal and shear stresses

Page 162 equilibrium, pairs of shear stresses must be equal as follows: (5.29)

Considering the normal stresses first, Fig. 5.8(a) shows a vertical line of cubes (such as that of Fig. 5.7) through the depth of the plate in the X− plane. Each of these cubes is subjected to a Z normal stress in the X direction as indicated in the figure. When there are no in-plane forces in a bridge deck, the sum of the forces in these cubes is zero. As each cube is of the same surface area, it follows that:

However, there is a bending moment caused by these stresses. The term mx is used to represent the moment per unit breadth due to the σ stresses, summed through the depth of the x deck. Figure 5.8(b) shows the depths of the cubes δ and their distances from the origin, z1, z2, z z3, etc. Each cube has a width perpendicular to the page of δ (not shown in the figure). The y forces F 1, F2, F 3, etc., due to each of the stresses are also shown. The ith cube contributes a component of hogging bending moment of magnitude (σδ y)zi. Taking sagging moment as xi zδ positive and summing over the depth of the plate gives: (5.30)

Substituting equation (5.26) into equation (5.30) gives:

which gives: (5.31)

Page 163

Fig. 5.8 Vertical line of elemental cubes through the depth of a plate: (a) stresses on each cube; (b) forces on the cubes and distances from the origin

Page 164 Applying a similar method it can be shown that the stress σcauses a moment per unit breadth y my which is given by: (5.32) The second moment of area per unit breadth of the plate, i is defined by: (5.33) Therefore equations (5.31) and (5.32) can be rewritten in terms of the second moment of area as follows: (5.34)

(5.35) It is important to remember that mx is the moment per unit breadth on a face perpendicular to the X axis and not about the X axis, i.e. in a reinforced concrete deck it is the moment which would be resisted by reinforcement parallel to the X axis. Likewise, my is the moment per unit breadth on a face perpendicular to the Y axis. Referring to Fig. 5.7, it can be seen that the shear stresses result in forces parallel to the Y axis which will also cause a moment. The moment per unit breadth due to is termed mxy. Figure 5.9 shows a number of cubes through the depth of the plate in the Y− plane. The shear Z force on the face of each cube is given by:

and the moment per unit breadth due to this force is given by:

Taking anti-clockwise as positive on the +X face, the total moment per unit breadth due to is given by: (5.36) Substituting equation (5.28) into equation (5.36) gives:

Page 165

Fig. 5.9 Stack of elemental cubes in the Y− plane showing shear stresses Z

which gives: (5.37)

Similarly the moment per unit length, myx, caused by

(on the Y face) can be shown to be: (5.38)

(5.39) However, as indicated in equation (5.29), equilibrium requires comparison of equations (5.36) and (5.38) yields: and to be equal and

(5.40) It follows from the definition of curvature (equation (5.18)) that the two twisting curvatures are the same: (5.41) so there is no contradiction between equations (5.37) and (5.39). These equations can be rewritten as: (5.42)

10 Bending and twisting moments in a plate: (a) segment of plate and directions of moments. 5. (b) associated distortions .Page 166 Fig.

10(b) shows the type of deformation associated with each of them. dQdx which is relatively small. while numerically small.10(a) shows the direction in which each of these moments acts while Fig.45) It was assumed earlier (equations (5. particularly in concrete slabs which are quite weak in shear. 5. In thin-plate theory. . a similar expression can be derived. shear stresses.44) and: (5. Figure 5. there are two shear forces at each point. 5.8) and (5. Unlike beams. 5.46) i.Page 167 where j is known as the torsional constant and is given by: (5. can be significant. This is a reasonable assumption as shear deformation is generally small in bridge slabs relative to bending deformation.11 shows a segment of a beam of length dx in bending.2. the shear force is the derivative of the moment.7. and illustrated in Fig.5 Shear in thin plates Vertical shear forces occur in bridge decks due to the shear stresses. In the simple flexural theory of beams.9)) that shear deformations in the plate were negligible. Taking moments about the left hand end gives: Rearranging and ignoring the term. Defining qx and qy as the downward shear forces per unit breadth on the positive X and Y faces respectively then gives: (5.e. gives an expression for the shear force Q: (5. the same phenomenon exists and an expression is found from equilibrium of forces on a segment. one for each direction (X and Y).43) The moment mxy (=myx ) is often referred to as a twisting moment and is distinct from the normal moments mx and my . Figure 5. However. The moment and shear force at the left end are M and Q respectively and at the right end are M+dM and Q+dQ respectively.

Page 168

Fig. 5.11 Equilibrium of small segment of beam

A small element from the plate of base dimensions dx×dy is shown in Fig. 5.12, with varying bending moment and shear force. The terms qx and qy refer to shear forces per unit breadth while mx, my and mxy refer to moments per unit breadth. This is different from the beam example above where Q and M referred to total shear force and total moment. Taking moments about the line a–b (Fig. 5.12) gives:

where F z is the body force acting on the segment of slab (for example, gravity). Dividing across by dx dy gives:

where f z is the body force per unit area. The second and third terms of this equation represent very small quantities and can be ignored giving: (5.47) By taking moments about the line b–c (Fig. 5.12), an equation for qy can be derived in a similar manner: (5.48) It can be seen that the expressions for the shear forces per unit breadth (equations (5.47) and (5.48)) are of a similar form to that for a beam (equation (5.46)) except for the addition of the last term involving the derivative of mxy or myx.

Page 169

Fig. 5.12 Equilibrium of small segment of slab

**5.3 Grillage analysis of slab decks
**

The idea of grillage analysis has been around for some time but the method only became practical with the increased availability of computers in the 1960s. Although computational power has increased many-fold since then, the method is still widely used for bridge deck analysis. Some of the benefits that have been quoted are that grillage analysis is inexpensive and easy to use and comprehend. These benefits traditionally favoured the method over finiteelement analysis which was typically only used for the most complex problems. In today’s environment of inexpensive, high-powered computers coupled with elaborate analysis programs and user-friendly graphical interfaces, the finite-element method has begun to replace the grillage method in many instances, even for more straightforward bridge decks. That said, the grillage method has proved to be a versatile tool for the analysis of many bridges and benefits from numerous favourable comparisons with experiments such as those of West (1973). The plane grillage method involves the modelling of a bridge slab as a skeletal structure made up of a mesh of beams lying in one plane. Fig. 5.13(a) shows a simple slab bridge deck supported on a number of discrete bearings at each end and Fig. 5.13(b) shows an equivalent grillage mesh. Each grillage member represents a portion of the slab, with the longitudinal beams representing the longitudinal

Page 170

Fig. 5.13 Grillage idealisation of a slab: (a) original slab; (b) corresponding grillage mesh

Page 171 stiffness of that part of the slab and the transverse grillage members representing the transverse stiffness. In this way, the total stiffness of any portion of the slab is represented by two grillage members. The grillage mesh and individual beam properties are chosen with reference to the part of the slab which they represent. The aim is that deflections, moments and shears be identical in both the slab and the grillage model. As the grillage is only an approximation, this will never be achieved exactly. Clearly different levels of accuracy are acceptable for different applications. For example, a crude representation might be sufficient at the preliminary design stages.

**5.3.1 Similitude between grillage and bridge slab
**

It is necessary to achieve correspondence or similitude between the grillage model and the corresponding bridge slab. A point p is illustrated in Fig. 5.13 corresponding to the junction of longitudinal beams b1 and b2 and transverse beams b3 and b4. Figure 5.14 shows an enlarged view of the junction along with the forces and moments acting on beams b1 and b3 in the grillage. The forces and moments have not been shown on beams b2 and b4 for clarity. The moments at the ends of beams b1 and b2 adjacent to p in the grillage give a measure of the moment mx in the slab while the moments at the ends of beams b3 and

Fig. 5.14 Segment of grillage mesh showing forces and moments on members b1 and b3

Page 172 b4 give a measure of the moment my . The moments in the grillage members are total moments while those which are required in the slab are moments per unit breadth. Therefore, it is necessary to divide the grillage member moments by the breadth of slab represented by each. This breadth is indicated in Fig. 5.13 as sx and sy for the longitudinal and transverse beams respectively. Unfortunately, in the grillage, the moments at the ends of beams b1 and b2 adjacent to p are generally not equal, nor are those in beams b3 and b4. For a fine grillage mesh, the difference is generally small, and it is sufficiently accurate to take the average moment at the ends of the beams meeting at the junction. The magnitude of this difference is often used as a check on the accuracy of the grillage, but it should be borne in mind that a small inequality does not necessarily mean an accurate grillage, as other factors may be involved. The moments per unit breadth in the slab at point p are therefore obtained from the grillage using the following equations, with reference to Figs. 5.13 and 5.14:

or: (5.49) Similarly: (5.50) The moments at any other point in the slab can be found in a similar way. If the point is not at the intersection of longitudinal and transverse grillage members, it is necessary to interpolate between adjacent beams. Care should be taken while doing this, especially if a coarse grillage mesh is used. Some computer programs carry out this interpolation automatically, in which case it is necessary to confirm that the program has interpolated the results in a sensible manner. It is often more convenient to start by considering the locations at which moments will be required and to formulate the grillage mesh in such a way as to avoid the need for interpolation between beams. The twisting moments per unit breadth in the slab, mxy and myx, are found from the torques in the grillage members in a similar manner. These moments at point p (again with reference to Figs. 5.13 and 5.14) are given by: (5.51) and: (5.52)

At point p. as large variations of torque may exist between the longitudinal and transverse beams. Therefore the twisting moment in the slab is arrived at by averaging the torques per unit breadth in all four beams meeting at the point p.53) and: (5.14.46). E the modulus of elasticity and R the radius of curvature. 5. the shear force Vb1 in Fig. 5. unless myx is particularly large.47). for example. This accounts for the first term of equation (5. are found from the shear forces in the grillage members in a similar manner to the moments. By substituting the curvature 1/R with κ rearranging. The situation can be improved by choosing torsion constants for the longitudinal and transverse beams which promote similar levels of torque per unit breadth in both. Examining.47 and 5. qx and qy.14) these are given by: (5. . The shear forces per unit breadth in the slab. 5.56) where i is the second moment of area per unit breadth.48 gave expressions for the shear forces per unit breadth in the slab. This technique is discussed further in the next section. namely. I the second moment of area.3. particularly for orthotropic plates with significantly different flexural stiffnesses in the two directions. this is not normally done as the resulting inaccuracy in the shear forces tends to be small.Page 173 Equation (5. This may be quite unsatisfactory. However.40) stated that mxy and myx are equal for materially orthotropic plates. the derivative of myx with respect to y. (Figs. the moment per and unit breadth.13 and 5.55) where M is the moment. it can be seen that this shear force will be equal to the derivative of the moment Mb1 with respect to x as this beam will comply with equation (5. but there is no account taken in the grillage analysis of the second term. m is found: (5. This could be calculated in the grillage by finding the derivative of the torques in b3 and b4 with respect to y. but the torques in grillage members b1 and b2 will not necessarily be equal to the torques in b3 and b4.2 Grillage member properties—isotropic slabs A grillage member in bending behaves according to the well-known flexure formula: (5.54) Equations 5.

58) where is the angle of twist. The displacement in the Z direction is given by w and the angle x of twist over the length δ is given by: x Hence: (5. As it is the relative values of stiffness that affect the calculated bending moments and shear forces. Further. This can clearly be achieved by adopting the same elastic modulus and second moment of area per unit breadth in the grillage as that of the slab. G is the shear modulus and J is the torsion constant (St.34) gives an expression for the moment per unit breadth in the X direction in the slab. v. the stiffness terms of equations (5. they are both affected by the same amount. giving: A further simplification is made by equating the term below the line to unity.57) must be equated. The moment/curvature relationship then becomes: (5.59) . For an isotropic slab. Venant constant). there is only one value for E and νSubstituting E for Ex and ν . is relatively small in bridge slabs (approximately 0. l is the length of the beam. for νand νin that equation gives: x y As Poisson’s ratio.15 shows a portion of a beam of length δ in torsion. it is common practice to ignore the second term in this equation.57) To achieve similitude of moments between a slab and the corresponding grillage.56) and (5. Figure 5. T is the torque. if this approximation is applied to both mx and my. A grillage member in torsion behaves according to the well known equation: (5.Page 174 Equation (5. This can be justified by the fact that Poisson’s ratio is small.2 for concrete). such an adjustment has very little effect on the final results.

15 gives: (5.59) gives: (5.60) Applying equation (5. 5.18) into equation (5.61) Substituting equation (5.Page 175 Fig.60) into equation (5. 5.58) to the beam of Fig.61) gives: (5.15 Segment of beam subjected to torsion Substituting equation (5.62) .

Equating this to jgril gives: (5.67) Typically. The torsion constant for the grillage member can alternatively be expressed in terms of the slab second moment of area: (5. However.58) can be applied. it will not generally be necessary to specify Gxy for the grillage model. in the slab and torques. This can clearly be achieved by adopting the same shear modulus and torsion constant in the grillage member as is in the slab. the stiffness terms of equations (5. The preceding derivation of grillage member torsional properties is applicable to thin plates of rectangular cross-section where equation (5. they will not necessarily have the same total torsional constant as they may represent different breadths of slab if the grillage member spacing in the longitudinal and transverse directions differ. .42) gives an expression for the twisting moment per unit breadth in the bridge slab: (5. Equation (5.65) ensures that the grillage members in both directions will have the same torsional constant per unit breadth. t: (5.65) and (5.64) must be equated. Torsion in beams is complicated by torsional warping (in all but circular sections) and formulas have been developed to determine an equivalent torsional constant for non-rectangular sections such that equation (5. Equation (5. Equation (5.64) To achieve similitude of moments.63) and (5.66) Although equations (5.Page 176 This can be rewritten in terms of torque per unit breadth.63) where j gril is the torsion constant per unit breadth in the grillage member.43) for the torsional constant is valid. The behaviour of a grillage member is essentially one dimensional and consequently its shear modulus can be derived from the elastic modulus and Poisson’s ratio directly using the wellknown relationship: (5.43) gives an expression for the torsion constant of the slab. this is carried out automatically by the grillage program.66) are based on the grillage member having the same shear modulus as the slab. mxy. t. in the grillage members.65) where d is the slab depth.

e. Further. most bridges have the same modulus of elasticity. equation (5. many bridges are geometrically orthotropic.17.3. Similarly.Page 177 For rectangular beams with depth d and a breadth of greater than 10d.66) for isotropic slabs. in both the horizontal and vertical directions.e. the second moments of area per unit breadth for the grillage and the slab are equated. Figure 5.68) predicts a torsion constant for the beam which is twice that predicted by equation (5.34).16 shows a portion of a beam of breadth b and depth d in torsion. i. It has been recommended that the edge grillage members be placed at 0. applies to materially orthotropic slabs: However. in the Y direction.3 Grillage member properties—geometrically orthotropic slabs Equation (5. E. The torque in the beam results from both of these shear stresses and is given by: (5.69) In the slab. they have different second moments of area per unit breadth in the orthogonal directions. The shear stresses set up in the beam are shown. . 5. The reason for this lies in the definition of torsion in a beam and of moment mxy in a slab. the shear stresses in the vertical direction are accounted for by the shear force per unit breadth. It is common practise to use the equations developed for materially orthotropic thin plates to represent geometrically orthotropic bridges. qx as illustrated in Fig. for both directions.3 times the slab depth from the edge so as to coincide with the resultant of the shear stresses.68) It can be seen that equation (5. The vertical shear stresses are accounted for in the grillage in the same manner by the shear forces qy in the transverse beams. 5.36) shows that the moment mxy is arrived at by summing only the shear stresses in the horizontal direction (i. the torsional constant may be approximated with: (5. Consequently the torsion constant for a grillage member representing a portion of an isotropic slab is only half that of a regular beam (or a grillage member representing a regular beam). only). reproduced here. This is achieved in a grillage by basing the second moment of area per unit breadth of the grillage members in the X direction on that of the slab in that direction. In the slab.

Then. However.41).40) stated that the two twisting moments at a point in a materially orthotropic slab are equal to each other.16 Beam subjected to torsion showing resulting shear stresses Equation (5. 5. Further. as stated in equation (5. i.e. If it is assumed that the same conditions hold for geometrically orthotropic slabs.Page 178 Fig. curvatures in the orthogonal directions at a point will be approximately equal.42).: and: There is no facility in a grillage model to ensure that the two curvatures at a point are equal. in a fine grillage mesh. if the same shear modulus and torsional constant are used in the two directions. it follows from equation (5. . the two twisting curvatures are the same.

The shear modulus for a slab made from one material. 5.70) Hambly (1991) recommends using such a single torsional constant for both orthogonal directions: (5.66) for an isotropic slab. that the twisting moments are equal: (5.3. 5. G. equation (5.71) It can be seen that this equation is consistent with equation (5. and Poisson’s ratio.4 Computer implementation of grillages There are many computer programs commercially available which are capable of . νIt is generally calculated internally in computer programs using .67).Page 179 Fig.70). is a function of the elastic modulus.17 Slab with vertical shear stresses and corresponding grillage members with shear forces per unit breadth reproduced and adapted here as equation (5. E.

Some grillage programs allow. or require. There is no facility for the nodes to deflect in either of the inplane directions or to rotate about an axis perpendicular to the plane. some grillage programs do allow for shear deformation. or may allow one or more of these degrees of freedom. These programs are generally based on the same theory. This may be used to define the bridge self weight. In such cases. according to the principle of superposition. This inhibits the calculation of in-plane effects such as axial thermal expansion or contraction or in-plane prestressing. Such effects are normally determined separately (often by hand due to their simplicity) and added to results from the grillage. Consequently. that of the stiffness method. These nodal supports may be rigid.5 Sources of inaccuracy in grillage models It should always be borne in mind that the grillage analogy is only an approximation of the real bridge slab. Grillage programs model the supports to the bridge slab as restraints at various nodes.Page 180 analysing grillages. The nodes are therefore said to have three degrees of freedom.3. While shear deformation is generally not very significant in typical bridges. 5. the product of which gives the shear area. the definition of a cross-sectional area for the beams. allowing no displacement or rotation in either of the two directions. Some programs which allow the modelling of shear deformation will only give results of shear stresses when this option is invoked. but when used correctly it will accurately predict the true behaviour. a number of which are described here. it should improve the accuracy of the results if it is allowed for in the computer model. two rotations and one translation. It therefore makes sense. Where the grillage is formulated without regard to the nature of the bridge slab. However. when formulating the grillage. These facilities may be used to model the soil/structure interaction as discussed in Chapter 4. Even though the thin plate behaviour considered in Section 5. The points at which these beams are connected are referred to as nodes. This is generally achieved by defining a cross-sectional area and a shear factor. It has been pointed out that the moments in two longitudinal or two transverse grillage members meeting end to end at a node will not necessarily be equal. this approximation may be quite inaccurate. The discontinuity between moments will be balanced by a discontinuity of torques in the beams in the opposite direction to preserve moment equilibrium at . and the imposition of specific support settlements.2 assumed that there was no shear deformation. some inherent inaccuracies exist in the grillage. Most grillage programs will allow the use of spring supports. Some programs also use the cross-sectional area definition to model shear deformation. care should be taken to ensure that the self weight is not applied twice by applying it to both the longitudinal and transverse beams. even if due care is taken. to locate nodes at the centres of the bearings or supports. with some variations from program to program. inplane axial forces are not modelled by the grillage. Each node has the capability to deflect vertically out-of-plane or to rotate about each axis of the plane. The computer implementation of a plane grillage consists of defining a mesh of interconnected beams lying in one plane.

As a result of this. Equation (5. the curvatures in the grillage members in one direction do not effect the moments in the beams in the other direction in the same manner as they do in the bridge slab. in the slab. 5. The magnitude of these discontinuities can be reduced by choosing a finer grillage mesh. Fig. As was mentioned earlier. having no other transverse beam to balance it. mx. 5.40) stated that the moments mxy and myx are equal in a slab.57)). The required moment is arrived at by averaging the moments on either side of the node. Where only three beams meet at a node.Page 181 the node. A similar simplification was made for my . The opposite of this is not necessarily true. excessively large discontinuities in moments. The same phenomenon causes discontinuities in torques and shears.18 where it can be seen that the torque T in the transverse beam. this discontinuity will be exaggerated. and requires the addition of more beams. as are the corresponding curvatures in the two directions.34) gave an expression for moment per unit breadth. such as where two longitudinal beams along the edge of a grillage meet only one transverse beam. Torsions per unit breadth of similar magnitude in both directions in a grillage can be promoted by choosing the same torsional constant per unit breadth for the longitudinal and transverse beams. the effect of curvature in the Y direction was ignored (see equation (5. This potential inconsistency is reduced by the low Poisson’s ratio of bridge slab materials which limits the influence of curvatures in one direction on moments in the orthogonal direction. as other factors may also have an effect. There is no mathematical or physical principle in the grillage to make this so. Equation (5. This expression involved terms accounting for the curvature in the X and Y directions. torques or shears indicate a grillage mesh which is too coarse.18 Distribution of bending moment in a segment of grillage mesh showing discontinuity in moment (T b3=Mb1− b2 ) M . This is illustrated in Fig. significant differences can remain. which should be treated in the same manner. When deriving the properties of a grillage member parallel to the X axis. However. corresponds to the discontinuity between the moments Mb1 and Mb2 in the longitudinal beams.

whichever direction the beam lies in. The first of these equates the shear force per unit breadth qx to the sum of two derivatives: In the grillage. there are sharp concentrations of shear intensity near each support.7 Recommendations for grillage modelling It is difficult to make specific recommendations on the use of a technique such as grillage modelling. if reasonably accurate results are to be obtained. 5. There is no account taken of the derivative of the twisting moments. if the grillage mesh density increases. which is applicable to such a wide variety of structural forms.3. such as those relating to voided or skewed bridge decks. and should be used in the context of good engineering judgement. the shear force in a longitudinal or transverse beam will simply be the derivative of the moment in that beam with respect to X or Y. When bridges are supported at discrete intervals. If it were assumed that shear enhancement was sufficient to cater for local concentrations of shear near a support. Greater shear forces at points closer to the support would be ignored on the basis that load would be carried by direct compression rather than shear mechanisms. the magnitude of these moments is generally relatively small. Some more specific recommendations. then grillage member spacing would assume a much reduced importance. except for bridges with high skew.47) and (5. mxy or myx. These should not be viewed as absolute.6 Shear force near point supports There is a particular problem in using grillage models to determine the intensity of shear force (shear force per unit breadth) near a discrete bearing. Thus the designer would design for the shear force calculated at a deck depth from the support.3.48) provide expressions for the shear forces per unit breadth. This direct relationship between mesh density and the calculated maximum shear intensity means that. the grillage member spacing has to be fixed near the support so that it gives the correct result. are given in Chapter 6. However. O’Brien (1997) found that the grillage member spacing had a much reduced influence on the results for shear at distances of more than a deck depth from the support. It will be seen from the recommendations given here that the traditional need for economy in the .Page 182 Equations (5. Fortunately. the effective breadth decreases and the calculated concentration of shear adjacent to the support increases. qx and qy . 5. It follows that. Each grillage member represents a strip of slab with the result that a point support at a node in a grillage model has an effective finite breadth. some general recommendations are valid for most grillage models.

Page 183 numbers of grillage members no longer applies. illustrated in Fig. There is little point in having longitudinal beams too closely spaced. as this may result in skewed members which complicate the interpretation of results.19 Alternative grillage meshes near point supports: (a) local adjustment to mesh near supports to maintain constant spacing of members elsewhere. The transverse grillage members Fig. Often this spacing will be greater than that of the longitudinal beams. should these exist. beyond a certain point. 5. (b) non-constant mesh spacing . or precast beams in beam-and-slab bridges. 4. as the magnitude of moment in the transverse beams is generally relatively small. there is no advantage in providing excessive numbers of grillage members as the amount of output data will be excessive. significantly greater spacings are often possible without great loss of accuracy. 2. 5.19(a). Lines of strength may consist of concentrations of reinforcement. should be avoided if possible. and. Longitudinal grillage members should be provided along lines of strength in the bridge slab. 1. Spacing will often be dictated by the location of supports or lines of strength in the bridge slab. Where possible. A choice of between one and three times the longitudinal spacing would be reasonable. as the computational power available to today’s engineers is well in excess of that available when earlier recommendations were made. The procedure of moving nodes locally to coincide with supports. However. particularly in wide bridge slabs. grillage members should be located such that nodes coincide with the locations of supports to the bridge slab. Transverse beams should have a spacing which is similar to that of the longitudinal beams. 3. no additional accuracy will be achieved. Nonetheless. location of prestressing tendons. A reasonable spacing of longitudinal beams is between one and three times the slab depth.

For bending moment results.20 illustrates an example where a member is correctly placed more than 0. The validity of this recommendation has been confirmed by the authors through comparisons of grillage analysis results with those of elaborate three-dimensional finite-element models.20 Segment of grillage mesh showing longitudinal members 0. where d is the slab depth. Care should be taken. It has also been recommended that.e. the breadth of slab outside 0. so that the span length between supports in the grillage and the bridge slab are the same. Supports to the grillage should be chosen to closely resemble those of the bridge slab. 7.3d from the end. i. . such as heavily reinforced diaphragms above bridge piers. 6.3d from the edge except for the end transverse members . 5. in the bridge slab as illustrated in Fig. one with more longitudinal and transverse beams at a closer spacing.Page 184 should also be chosen to coincide with lines of transverse strength in the bridge slab. for example. a check can be performed by comparing the output of a grillage with that from a more refined grillage. The second moments of area of these beams are calculated using the full breadth of slab in the normal way. that this recommendation does not result in supports being placed in the wrong locations. If the spacing of grillage members is in doubt. 5. when determining the torsional constant of these longitudinal grillage members.3d should be ignored. should they exist. however. Figure 5.17. This may involve. It has been recommended by Hambly (1991) that the row of longitudinal beams at each edge of the grillage should be located in a distance of 0. 5. increasing the mesh density tends (up to a point) to increase the accuracy. the use of elastic springs to Fig. The objective is to locate these beams close to the resultant of the vertical shear stresses.3d from the edge of the slab.

for example. namely out-ofplane translation. such as axial prestress. but the support arrangement chosen for the model must be such that the model is restrained from free body motion in either of the in-plane directions or rotation in that plane. the FE method involves the modelling of a continuous bridge slab as a finite number of discrete segments of slab or ‘elements’. and rotation about both in-plane axes. Some elements do not model in-plane distortion and consequently the nodes have only three degrees of freedom. Finite-element analysis is relatively easy to use and comprehend and. and often more accurate than. Finite-element analysis is well known to bridge designers. and then to progress to more complex structures. This said. the behaviour of which is known. No particular problem arises from using elements which allow in-plane deformations in addition to out-of-plane bending. Sometimes it is more convenient to carry out an FE analysis with out-of-plane deformation only and to add the in-plane effect of prestress afterwards (which may often be determined by hand). The most common types of element used are quadrilateral in shape although triangular elements are sometimes also necessary. Some of these types of model are discussed in Chapter 7. Originally it was used for in-plane analysis of structures but it was soon extended to the problem of plate bending by Zienkiewicz and Cheung (1964). reasonable accuracy can be achieved with most sensible member spacings. Much development has taken place since this pioneering work and many texts now exist which give a comprehensive description of the method (see.Page 185 simulate deformable bearings or ground conditions as discussed in Chapter 4.4 Planar finite-element analysis of slab decks The finite-element (FE) method was pioneered in the mid 1950s for use mainly in the aeronautical industry. when applied correctly. 5. the grillage method. where shear enhancement occurs. Such analyses are only necessary if it is specifically required to model in-plane effects. All of the elements generally lie in the one plane and are interconnected at a finite number of points known as nodes. Beyond a deck depth from the face of the support. Finiteelement models in which the elements are not all located in the one plane can be used to model bridge decks which exhibit significant three-dimensional behaviour. There is a risk that inexperienced users will attempt to analyse complex bridges without understanding the true nature and behaviour of the structure. some of whom consider it to be the most general and accurate method available for bridge deck analysis while others view it with a degree of scepticism. that of Zienkiewicz and Taylor (1989)). Closer to the support. 8. A useful method of gaining familiarisation with a specific FE program is to begin by analysing simple structures. grillage analysis is much less reliable. . the scepticism expressed by some bridge designers is quite often well founded as the perceived accuracy of the method often overshadows the importance of using it correctly. When applied to the analysis of slab bridge decks. The authors have used the method extensively for the analysis of bridge decks and have found it to be an excellent analysis tool in many cases. is at least as accurate as.

67).4. 5. It was shown above that a grillage model does not take account of the derivative of the twisting moment. If this facility is used. The twisting moment term can readily be accounted for. a check is useful to ensure that the values given are consistent with those at the neighbouring nodes.47) and (5. it is advisable to determine whether or not shear forces are calculated correctly using equations (5. only two elastic constants need to be defined for the finite elements. although not all programs offer this facility. the validity of this relationship should be checked for the particular plate under consideration. ν and ν typically need to be specified. mx. in-plane distortion or a combination of both of these. The finite elements will satisfy this equation. These are generally given at the element centres and/or corners. Finally.4. Each of these expressions involves terms relating to the curvature in both the X and Y directions. The finite elements will behave according to these equations. G. These expressions involve derivatives of the direct moment mx (or my ) and the twisting moment myx (or mxy). The material properties of the elements are defined in relation to the material properties of the bridge slab. E and νThe shear modulus. These are plate elements which can model out-of-plane bending. Many programs provide the ability to determine these values at any arbitrary point using interpolation.2 Properties of finite elements The types of finite element considered here are those used for the modelling of slab bridge decks.47) and (5. is determined by the . five elastic constants. Equation (5. my and mxy . E y. E x. Gxy. As the element is of constant depth.35) give expressions for the moments mx and my in a thin plate.42) gave an expression for the moments mxy and myx in a thin materially orthotropic plate.33): . and the problem inherent in grillage modelling of torques per unit breadth not being equal in orthogonal directions does not arise. If this is the case. y. Where the twisting moments are significant. will account for the effect of curvature in one direction on the stiffness in the other direction. Equations (5. and unlike a grillage analysis.Page 186 5. Some x. program from these constants directly according to equation (5. When materially orthotropic finite elements are used. although in some programs it may not be.34) and (5. the second moment of area per unit breadth is given by equation (5. programs assume a value for Gxy based on the values input for the other four elastic constants. In FE analysis. equations (5. This is a significant advantage of the FE method over the grillage approach.48). are output directly by FE programs.48) give expressions for the shear force per unit breadth in a thin plate. shear force per unit breadth can be calculated.1 Similitude between finite-element model and bridge slab The moments per unit breadth. Isotropic bridge slabs In the case of bridges which are idealised as isotropic plates.

Page 187 In a typical program. mx .76) and (5.74) The modulus of elasticity of the element in the X direction may be chosen arbitrarily to be equal to the modulus of elasticity of the bridge slab. In most geometrically orthotropic bridge slabs. but only one depth can be specified. Equation (5. which will be satisfied by a materially orthotropic finite element: where and ielem are the element elastic modulus and second moment of area per unit breadth respectively. there is only one modulus of elasticity.72). i. In such cases. the user simply specifies the element depth as: (5.75) Substituting this into equations (5.72) Geometrically orthotropic bridge slabs Geometrically orthotropic bridge decks are frequently modelled using materially orthotropic finite elements. This problem can be iy overcome by determining an equivalent plate depth and altering the moduli of elasticity of the element to allow for the differences in second moments of area. but there are two second moments of area per unit breadth. and .35) gives a similar expression for my .e. similitude between the finite element and the bridge slab can be achieved by keeping the products of elastic modulus and second moment of area equal: (5.74) gives: (5.73) (5.73) and (5. . for both directions.: (5. ix ≠ .77) The equivalent element depth can be calculated from equation (5. However.34) gives an expression for the moment. Equation (5. E slab.

equation (5. This would lead to alternative expressions to the above.5 and 0. As an alternative. the moment/curvature relationship for the twisting moment. It was reported by Troitsky (1967). from the results of analysis and experimentation on steel orthotropic bridge decks.78) For a geometrically orthotropic slab with a single modulus of elasticity and Poisson’s ratio.80) and analysing again using a shear modulus of half this value.79) To be consistent with the equations for and i elem derived above. Gxy. a depth that would result in a second moment of area equal to the average of the second moments of area of the bridge slab in the two directions) and expressions determined for the corresponding values of the moduli of elasticity of the element. the modulus of elasticity of the element in the X direction is taken to be equal to the modulus of elasticity of the bridge slab. These types of model are discussed further in Chapters 6 and 7. Consequently the accuracy of this and equation (5.80) Equation (5.73) and (5. the authors would suggest analysing the orthotropic plate using a value predicted by equation (5.79) becomes: (5. a similar expression can be determined by substituting from equations (5. .42).3 was reported to come from an extreme case where the flexural stiffness in the two directions varied by a factor of 20. the orthotropic nature of the plate might be better handled using a combination of elements and beam members or a three-dimensional model. is given by equation (5. an arbitrary depth of finite element could be chosen (say.76) applies and equation (5. mxy . Instead of arbitrarily equating the modulus of elasticity of the finite element in the X direction to the corresponding modulus of the slab.Page 188 For a materially orthotropic slab.74) to give: (5. In such cases the shear modulus may need to be reduced.78) was derived by assuming an average value of the elastic moduli in the two directions and an average Poisson’s ratio. The lower value of 0. Then. Alternatively. To determine if the influence of the shear modulus on the analysis is significant. the moduli in the Y direction could be equated.3.80) diminishes as the variation in the elastic properties in the two directions increases. has been suggested by Troitsky (1967): (5. An approximate expression for the constant. that the shear modulus given by the above expression may need to be reduced by a factor of between 0.

Once again these should not be viewed as absolute. it may become necessary to limit the number of elements. applicable to voided and skewed bridge decks. In contrast to grillage modelling. 5.3 apply. as some programs may not be able to deal with excessive numbers. two rules commonly applied to quadrilateral elements are that the ratios of the perpendicular lengths of the sides should not exceed about 2:1 and that no two sides should have an internal angle greater than about 135°. This type of model is useful where in-plane effects (such as axial prestress) are to be considered. These should tend towards squares in the case of quadrilateral elements and towards equilateral triangles in the case of triangles. Many engineers use denser meshes of elements in those parts of a bridge where bending moment changes rapidly such as near an interior support. One variation between the two methods is that the FE model may allow for in-plane deformations and consequently the nodes will often have five or six degrees of freedom. but can easily be modified where this is not the case. However. and quite often the program will allow the user to define elements which do not conform to this shape.4. it is difficult to make specific recommendations relating to FE modelling of bridge slabs but some general guidelines are given here. More specific recommendations are given below and further guidance.22(a) where elements (1) and (2) are connected to each other at point P but are not connected to element (4). Some elements have mid-side nodes so that it is possible for example to have . 5. The implementation of the FE model is carried out in a similar manner to a grillage and many of the comments in Section 5. the finite element response to applied loading is based on an assumed displacement function. more elements tend to result in greater accuracy although this is by no means guaranteed. In general. it is often more convenient if a consistent mesh density is used throughout a bridge. 2. As with grillage modelling. 5. for example. is given in Chapter 6. In the absence of information to the contrary. Quite often the same program can be used for grillage and FE analysis which saves the user having to become familiar with two separate programs.21(b).21(a) but may give an inaccurate representation for the elements shown in Fig. considerable deviation from these shapes may be permissible and the documentation provided with the program should be consulted for specific recommendations. Regularly shaped finite elements should be used where possible. Mesh discontinuities should be avoided. Considering. These may occur when attempting to refine the mesh such as in Fig. This function may be applicable to elements of a certain shape only. quadrilateral elements with nodes at the four corners. Obviously.3 Recommendations for finite-element analysis There are many commercially available computer programs for FE analysis of bridge decks. a typical program may be able to deal with elements of the type shown in Fig. Unlike the grillage method. 1. 5.Page 189 The expressions given above relate to bridge slabs with the same modulus of elasticity in both directions.

5.21 Possible shapes of quadrilateral finite elements: (a) generally good shapes. (b) potentially problematical shapes Fig. (b) good arrangement . 5.Page 190 Fig.22 Meshes of finite elements at transition between coarse and dense mesh: (a) potentially problematic arrangement.

is introduced where N is normal to the face AB and T is parallel (tangential) to it. The length of the face AB is l and. a second axis system. for example. . the projected lengths on the X and Y axes are l sin θ l cos θ and respectively. This is generally easily achieved. mxy and myx. This may involve.Page 191 elements (3) and (4) connected to the mid-side node of element (1) at Q. A small segment of slab is illustrated in Fig. 4.23(a) and the possibility is considered of failure on a face. result in yield of the slab at any such angle. N–T. Supports to the finite-element model should be chosen to closely resemble those of the bridge slab. 5. Resultant moments can be calculated at any angle of orientation and can.24. For convenience. . The twisting moments per unit length. as can to be seen in the figure. All vectors are resolved parallel and perpendicular to AB in Fig. 6. my and mxy . Similar results from both would suggest that the mesh was sufficiently dense. 5. This section addresses the design problem of how the engineer should calculate the moment capacity required to resist such moments. There is little point in using too many elements as an excessive number slows the running of the program and may not result in significantly greater accuracy. 5. Elements should be located so that nodes coincide with the bearing locations. mx. 5. However. A mesh is shown in Fig. are also illustrated in this figure. the use of elastic springs to simulate deformable bearings or ground conditions as discussed in Chapter 4.22(b) where mid-side nodes are not needed and all elements are connected. The vectors representing the moments are resolved to determine the moments on the face AB. results at more than a deck depth away from the support have been found in many cases to be reasonably accurate (O’Brien et al. . 5. The results of such analyses give three components of bending moment at each point. This will be complied with if the first recommendation is adhered to. 7. The moment per unit length on the X face is mx so the moment on BC is mxl cos θThe corresponding moment on AC is my l sin θThese moments are illustrated in Fig. 1997). where the moment is about the axis of the arrow.5 Wood and Armer equations Much of this chapter has been concerned with methods of analysis of slab bridges. As bending moment is a vector. The spacing of elements in the longitudinal and transverse directions should be similar. AB. the three components can be combined using vector addition in a manner similar to the concept of Mohr’s circle of stresses. The direct moment per unit length on AB is denoted mn and the twisting moment per unit length is denoted mnt. If mesh density is in question. Shear forces near points of support in finite-element models tend to be unrealistically large and should be treated with scepticism. 3. 5. at an angle of θ the Y axis. it is useful to compare the output of a model with the chosen mesh density to that of a model with a greater density. if excessive.23(b) using double headed arrows to denote bending moment.

82) The components of moment on a face perpendicular to AB are considered in Fig.25 where resolution of components gives: (5.81) Considering components perpendicular to AB gives: (5.83) and: (5. (b) applied bending and twisting moments Considering components parallel to AB first: (5.23 Segment of slab: (a) geometry.84) . 5.Page 192 Fig. 5.

Page 193

Fig. 5.24 Resolution of moments on a segment of slab parallel and perpendicular to AB

Fig. 5.25 Resolution of moments on a face perpendicular to AB

A comparison of equations (5.82) and (5.84) verifies that mnt and mtn are equal. Equations (5.81)–(5.84) can be used to resolve all components of moment on a small segment of plate into a new axis system as illustrated in Fig. 5.26. In an orthotropic steel plate, moment capacity is generally provided in the two orthogonal directions. In a concrete slab, ordinary or prestressing reinforcement is provided in two directions, which are not necessarily orthogonal. In this section, only orthogonal systems of reinforcement are considered; similar equations for non-orthogonal systems are given by Clark (1983). Furthermore, only the case in which mn is positive is considered here. The case when mn is negative is also treated by Clark.

Page 194

Fig. 5.26 Transformation of applied moments to an alternative co-ordinate system: (a) moments in X-Ysystem; (b) moments in N-T system

An orthogonal system of reinforcement provides moment capacity in two perpendicular directions which are taken here to be parallel to the co-ordinate axes. Hence, the moment capacities per unit length can be expressed as and as illustrated in Fig. 5.27. This figure is different from Fig. 5.26(a) in that there are no twisting moment terms; no capacity to resist twisting moment is assumed to be provided. Equation (5.81) gives the moment on a face at an angle θ the Y axis. A corresponding equation can readily be derived for the moment to capacity. Leaving out the mxy term in equation (5.81) leads to: (5.85)

Page 195

Fig. 5.27 Segment of slab illustrating the moment capacities provided

While no capacity to resist twisting moment is explicitly provided, capacity can be shown to exist on face AB (Fig. 5.23(a)) by considering equation (5.82) which gives: (5.86) Similarly, from equation (5.83): (5.87) To prevent failure on face AB of Fig. 5.23, the moment capacity must exceed the applied moment. As only the case for which mn is positive is being considered, this becomes:

Substituting from equations (5.81) and (5.85) gives:

Dividing the equation by cos 2 θ gives:

This can be expressed as:

where (5.88) and

Page 196 The function, f(k), is the excess moment capacity for the angle θi.e. the amount by which the , moment capacity exceeds the applied moment for that angle. To prevent failure of the slab, it is clearly necessary that this function exceeds zero for all values of θThe most critical angle . will be that for which f(k) is a minimum. This minimum value is found by differentiating the function and equating to zero, that is:

As k=tan θdifferentiating with respect to θ , gives:

which is never zero. Hence the minimum value for f (k) occurs when: (5.89)

where is a critical value for k. For this to be a minimum excess moment capacity rather than a maximum, the second derivative of f (k) must be positive, i.e.: (5.90)

Taking equations (5.89) and (5.90) together, it can be seen that and mxy must be of the same sign. This fact will be shown to be of significance later in the derivation.

Page 197

Example 5.1: Moment capacity check At a point in a bridge slab, the moments per unit length due to applied loads have been found to be, mx=190, my=80 and mxy=20. It is required to determine if it is sufficient to provide moment capacities of, and Equation (5.89) is used to determine the angle for which the excess moment capacity is minimum:

i.e. the critical angle is 29.7°. The minimum excess capacity is then found by substitution in equation (5.88):

As the excess capacity is negative, the slab will fail for this value of θ . When new bridges are being designed, the moment capacities are not generally known in advance and the problem is one of selecting sufficiently large values for and . It can be seen from equation (5.89) that effectively dictates the value for for a particular set of moments, i.e. choosing amounts to choosing . Thus the designer’s problem can be viewed as one of choosing a suitable value for provided that equation (5.89) is satisfied, i.e. choosing such that: (5.91) It is, of course, also necessary to have a positive excess moment capacity. The minimum required excess moment capacity is:

Substituting from equation (5.91), this becomes:

92) =1 and (5.90)) that and mxy were of the same sign. Hence.93). for example. the cost of providing moment capacity at a point may be taken to be proportional to: The value for which results in minimum cost is found by differentiating: (5. then ρ and equations (5. In general.94) This can be used to find an economical value for in equations (5.91) becomes: (5. be prestressed in one direction and reinforced with ordinary reinforcement in the other. equation (5.Page 198 It was established earlier (by comparing equations (5. The cost of providing moment capacity in the two coordinate directions may not necessarily be equal as a bridge may.96) These are known as the Wood and Armer equations (Wood. Example 5. 1968). .92) Similarly.89) and (5.93) become: (5.2: Wood and Armer equations II At a point in a bridge slab the moments per unit length due to applied loads have been found to be.92) and (5. my=80 and mxy=20.93) Any value for can be selected by the designer and these equations used to determine the minimum required moment capacities. mx=190. If the cost of providing moment capacity is the same in both directions. their product is positive giving: (5. It is required to determine economical moment capacities given that providing costs twice that of providing .95) (5.

Page 199 As cost is proportional to critical angle is defined by: the constant. Equations (5.92) and (5.e. ρis 0. i.93) then give the required moment capacities: .5 and the minimum cost value for the .

the behaviour of bridge slabs is considered. with adaptation.Page 200 Chapter 6 Application of planar grillage and finite-element methods 6. be applied to many different types of slab as will be demonstrated. For this reason. both of which consist of members lying in one plane only. their basis is well understood and the results are considered to be of acceptable accuracy for most bridges. more complex non-planar methods of analysis are considered. planar grillage and finite-element models are at present the method of choice of a great many bridge designers for most bridge slabs. For certain bridges. grillage and finite-element methods. In Chapter 7. This will be demonstrated in the following examples. Planar methods are among the most popular methods currently available for the analysis of slab bridges. both of these planar methods of analysis are used to model a range of bridge forms. . they can also be considerably more complex and can take much longer to set up. In this chapter. Two methods of analysis are introduced.1 Introduction In Chapter 5. Further. non-planar models are considerably more accurate than planar models. They can. it is a simple matter to prepare a computer model following the guidelines specified in Chapter 5. However. 6.2 Simple isotropic slabs When bridge slabs are truly planar.

1 Plan view of two-span bridge Fig. free-sliding and guided-sliding bearings is used so that the bridge can expand or contract freely in all directions in plane. The end rows of transverse members are taken through the centres of the bearings. As recommended in Section 5.8 m depth. 6. 6. 6. (b) section .2 (a) shows a convenient grillage mesh for this bridge deck.5. It is to be constructed of prestressed concrete and is to have a uniform rectangular cross-section of 0. The longitudinal members have been placed along the lines of the bearings.1: (a) plan.2 and 1.5 m which gives a ratio of transverse to longitudinal spacing of between 1.2 Grillage mesh for bridge of Fig. Figure 6. Fig. It is required to design a grillage mesh to accurately represent the deck given that the concrete has a modulus of elasticity of 35×106 kN/m 2.3. with an additional line at the centre of the deck.1: Grillage model of two-span right slab A two-span bridge deck is illustrated in Fig.3 times the depth from the edge of the slab. 6. A combination of fixed.1. a row of longitudinal members has been placed at a distance of 0. The transverse members have been placed at a spacing of 1.Page 201 Example 6. The deck is supported on four bearings at either end and on two bearings at the centre as illustrated in the figure.

0938 0.0483 0.5 m past the centre of the bearing.87− 0. Example 6. It can be seen that this breadth is taken to be from midway between adjacent members on either side.5 as the slab extends 0. However. Figure 6. 6. R8 R3.1 Grillage member properties for Example 6.2.3. R7 R4.3 shows a convenient finite-element mesh.0470 0. The breadths of the elements are chosen such that nodes coincide with the locations of the supports.66): The second moments of area and torsion constants of the grillage members are then determined by multiplying these values by the relevant breadth of each member as given in Fig. the breadth is 1.0964 0.1 and Fig. The bridge slab is assumed to be isotropic and the second moments of area per unit breadth are taken to be equal to those of the slab: The torsion constants per unit breadth are calculated according to equation (5.7. R9 R2. The two rows of . 6. R6 R5 Transverse Members End members All intermediate members 0.0491 0. 6.63 m was used.0534 0. a reduced breadth of (0.2 (b) shows a cross-section of the slab with the grillage members superimposed.0470 Torsion constant (m4) 0.0981 0. in keeping with recommendation number 6 of Section 5.5/2+0. The longitudinal members have been grouped by row as R1 to R9 and the transverse members have been grouped as end members and all intermediate members as illustrated in Fig. when determining the value of the torsion constant of the longitudinal members in rows R1 and R9.1280 Figure 6.0862 0.1. These values are presented for all of the grillage members in Table 6.1.Page 202 Table 6.24)=0.3d=0.0640 0.0537 0. Similarly. this is reduced by 0.2 (b). This is used to determine the breadth of slab attributable to each longitudinal grillage member.24 m for the calculation of the torsion constant.1 Second moment of area (m4) Longitudinal members R1. For the transverse end members.2: Finite-element model of two-span right slab A planar finite-element model is required for the bridge deck of Example 6.0938 0.0371 0.

These are frequently important aesthetically and.3 Finite-element mesh for bridge of Fig. or simply for aesthetic reasons.4(c) and (d). the elastic modulus is taken to be that of the slab. The effect of an edge cantilever or an integral upstand/downstand is to change the stiffness of the bridge deck. 6.5 m breadth. the upstand may not be integral with the bridge deck and can simply be considered as an additional load on it. and had the length been taken as. causing the bridge neutral axis to rise. in the case of concrete bridges. 6. where the edge cantilever is relatively short or stocky .2 m which is equal to the breadth of the widest element. In such cases. These will be similar to those illustrated in Fig. but the extra number of elements in the model chosen is not considered to be excessive. If they are made integral with the deck.8 m which is equal to the actual depth of the bridge slab. then the increased stiffness which they provide generally needs to be considered. are often included at the edges of the slab. say.3 Edge cantilevers and edge stiffening Slab bridge decks often include a portion of reduced depth at their edges known as an edge cantilever. The length of the elements along the span of the bridge was chosen as 1. E=35×10 6 kN/m2. In some bridge decks. Cross-sections of typical slab decks with edge cantilevers are illustrated in Fig. equal to the average breadth of the elements. finding the location of the neutral axis may not be straightforward. Only decks where the neutral axis remains substantially straight are considered here. In slab bridges.8). As this is an isotropic bridge slab.1 elements at each edge of the model could be replaced with one row of 1. This type of construction is chosen partly for its reduced self weight and partly for its slender appearance (see Section 1.4. to carry a protective railing. As for Example 6. In such a case. 6. may be precast to ensure a good quality of finish. Upstands or downstands. the only geometric property which has to be assigned to the elements is their depths. Figure 6.5 shows the cross-section of a deck with a long slender edge cantilever with an upstand at its edge. 6. 6. This is a somewhat arbitrary choice. The properties of each part are then calculated about this axis. a similar degree of accuracy could be expected. It is not necessarily conservative to ignore the additional stiffness provided by them. such as those illustrated in Figs. either to stiffen the edge.Page 203 Fig. 6. the appropriate stiffness is determined by first finding the neutral axis location for the complete deck. All of the elements are assigned a depth of 0. the neutral axis will not remain straight as the upstand tries to bend about its own axis.1.4. Bridge decks of this type are discussed further in Chapter 7.

The bridge deck.Page 204 Fig.4 Typical cross-sections of slab decks showing cantilevers and upstands Fig. In this case. 6. 6. The neutral axis is then taken to be straight across the complete deck and to pass through its centroid. The first task is to determine the location of the deck neutral axis which is taken to be straight and to pass through the centroid. This can be determined by hand or by using one of many computer programs available for such purposes.6. .5 Cross-section of slab deck with slender cantilever and upstand or where the upstand is not excessively stiff. which has a constant cross-section through its length. It is required to design a suitable mesh of grillage members to model the structure. Example 6. the neutral axis is found to be 563 mm below the top of the bridge deck. 6. spans 20 m and is simply supported on three bearings at each end as indicated in the figure.3: Grillage analysis of slab with edge cantilever The cross-section of a prestressed concrete bridge slab with edge cantilevers is illustrated in Fig. Details of a general approach to this calculation are given in Appendix C.

The location from which this distance is taken is somewhat arbitrary.6 Bridge deck of Example 6. In each case. and middle row.3 times the average depth of cantilever. is located at the centre of the portion of cantilever which it represents. • The fourth row. The reasons for this particular arrangement are as follows: • Each edge cantilever is modelled with two separate rows of members so that the reduced depth towards the edge can be allowed for.3 times the depth of the deck (0. R5 and R6 (and R8 and R9).7(b). R4.7(a) shows the divisions chosen and the corresponding grillage members. (b) plan The cross-section is divided into a number of segments. This distance corresponds to 0. • The outermost row of grillage members.3×1200=360 mm) from the midpoint of the sloping edge of the main deck. Note that row R4 is not exactly at the centre of the portion it represents.Page 205 Fig. The spacings of longitudinal grillage members is given in Fig. R7. Row R1. 6. but that chosen here seems reasonable. . This is in keeping with recommendation number 6 of Section 5. R2. • The second row of grillage members from the edge. is placed at a distance of 0. is placed at a distance of 90 mm from the edge of the cantilever.3 (dimensions in mm): (a) section. are chosen between the supports.3. Figure 6. these members represent a portion of bridge slab of breadth 1000 mm and they are located at the centre of that portion. • Two rows of grillage members. each of which is represented by a row of grillage members. 6. • The third row of members from the edge. R3.7. of grillage members are located to coincide with the supports to the bridge deck.

7 Grillage model (dimensions in mm): (a) cross-section showing grillage members and corresponding segments of deck. For this example. However. 6.27. (c) plan of mesh Figure 6. The second moment of area relative to the centroid of the bridge is always greater than (or equal to) that relative to the centroid of the individual portion . Due to the variation in depth between rows R2 and R3.Page 206 Fig. (b) schematic of cross-section showing spacing between members. it gives a good longitudinal to transverse spacing ratio. The second moment of area about the centroid (of the bridge) of each portion of deck is determined. between 1:1 and 1:1.7 (c) illustrates a plan of the grillage mesh with dimensions in mm. each row of longitudinal grillage members is considered separately. Twenty one rows of transverse members with a spacing of 1000 mm were chosen. the transverse members between these rows have been modelled as two separate members with a row of nodes where they join. This is a very dense mesh having a spacing less than the slab depth.

002 0.3 Second moment of area (m4) Longitudinal members R1. 300 mm. The second moment of area per unit breadth of these members is therefore: The second moment of area of the transverse grillage members in the main part of the deck. 6. These are labelled Tc in Fig.010 0. R6. Tc.290 0. running from the edge as far as the row of nodes indicated in Fig.7(c). R7.2. R12 R3.2 Grillage member properties for Example 6. The depth of these members is taken as the average depth of the cantilever.261 0.131 0. The second moment of area per unit breadth of these members is therefore: The second moment of area of the transverse members is then found by multiplying these values by the breadth of the members (which for this example is 1 m). 6.019 0. The first group are those in the cantilever portion.002 0.71) as this is an orthotropic deck : Table 6. The second moment of area of the transverse members in the cantilever.013 0.e. are also calculated about their own centroids as it is about these that they will bend.7.144 0.2. R11 R4. The results are presented in Table 6. The transverse members are divided into two groups. R13 R2.278 .110 0.178 0. R8. i. R9 Transverse members Tc—End members Tc—Intermediate members Tm—End members Tm—Intermediate members 0. Tm. R10 R5.021 0.034 0.Page 207 of deck. are taken about their own centroids as they will bend (transversely) about their own centroids.143 0.029 0. The second group are those in the main portion of the deck and account for all of the other transverse members.144 0.146 Torsion constant (m 4) 0. For example the second moment of area of row R7 is given by: All of the longitudinal grillage member second moments of area are presented in Table 6. The torsion constants for the members are determined in accordance with equation (5. These are labelled Tm in the figure.

Considering the longitudinal members in row R1 and the transverse members Tc. R2.2) is given by: . Tc. is given by: Considering next the longitudinal members in row R2 and the transverse members Tc. and the transverse members. In doing this.Page 208 where and are the second moments of area per unit breadth in the X and Y directions respectively.002 m3. as the two distinct values are very close. the torsion constant per unit breadth of the longitudinal members. To apply this equation. Hence. the X direction is arbitrarily chosen as the longitudinal direction. Tc. However. R1. the average value is considered acceptable.3. the condition of Section 5. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6. Tc.2) is given by: The second moment of area per unit breadth of the transverse members is 0. an approximation is made by taking an average value for the torsion constant per unit breadth of the transverse members.2) is given by: Therefore the torsion constant per unit breadth of the longitudinal members. is therefore: Considering the longitudinal members in row R3 and the transverse members Tm. The torsion constant per unit breadth of the transverse grillage members.3 is not satisfied which required that the torques per unit breadth in the grillage members in the longitudinal and transverse directions be of the same magnitude. At this stage. and the transverse members. is given by: This gives a value for the torsion constant per unit breadth for each of the longitudinal members R1 and R2 but there are two distinct values for the transverse members Tc. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6.

Tm. The cross-section of Fig. For the end transverse members.2.144 m3 and therefore the torsion constant per unit breadth of the longitudinal members.2) which is: Hence. and the longitudinal members in rows R1 and R13.4: Finite-element analysis of slab with edge cantilever It is required to prepare a finite-element model for the bridge deck of Example 6.6(a) is divided into a number of segments in a similar manner to the grillage model.8(a) shows the division of the deck and Fig. Tm.3×0. The average of the two values is taken for the transverse members Tm: The torsion constant for each grillage member is then arrived at by multiplying the torsion constant per unit breadth by the breadth of slab represented by that member. The length of the elements (in the longitudinal direction) is taken as 1000 mm.9=0. These values are given in Table 6.3=0.8(b) shows a cross-section through the finite-element model. 6. As the nodes form the boundaries of the elements and the location of the supports must coincide with nodes. R4 to R10. R3. The other longitudinal members. is given by: This value is adopted for the longitudinal members in row R3. is given by: This value is adopted for longitudinal members R4 to R10. Tm.3 and Fig. the breadth is reduced by 0.6. the breadth is reduced by 0. have the same second moment of area per unit breadth (with reference to Table 6. 6. and the transverse members. For the longitudinal members in rows R3 and R11.3×0. Figure 6. This results in 20 elements in each of the 14 longitudinal rows. the breadth is reduced by 0.3×1. . The depths of the elements have not been drawn to scale in this figure.36 m. For the end transverse members. the division of the deck for the finite-element model varies somewhat from that of the grillage. and the transverse members.Page 209 The second moment of area per unit breadth of the transverse members is 0. the torsion constant per unit breadth of the longitudinal members. Figure 6. 6.2=0. It can be seen that by splitting the transverse members running between rows R2 and R3 (and R11 and R12) into two separate transverse members. Example 6. the need to average two dissimilar values of torsion constant was avoided. Tc. R4 to R10.27 m.09 m.8(c) shows a plan of the finite element model with rows of elements labelled r1 to r14.

r2. . In the case of the elements in row r3.3. the second moment of area per unit breadth of each portion is determined about its own centroid as it is about this that transverse bending occurs. In the X direction. (b) schematic of cross-section showing breadths of elements.Page 210 Fig. A depth of 1000 mm is chosen as this seems to be a reasonable compromise and it is felt that the problem does not warrant an in-depth analysis.8 Finite-element model (dimensions in mm): (a) cross-section showing division of deck into elements. it is difficult to determine the transverse stiffness as the depth varies significantly. and . The second moments of area per unit breadth for each row of elements are given in Table 6. (c) plan of element mesh The X axis is again chosen to be in the longitudinal direction and the Y axis to be perpendicular to this. In the case of the elements representing the edge cantilevers (rows r1. r13 and r14). The second moments of area per unit breadth.3 to be located 563 mm below the top surface. these are calculated about the centroid of the bridge which was seen in Example 6. 6. In the Y direction. are determined for each portion of the bridge deck. the transverse stiffness is based on the average depth of that portion of cantilever.

0036 Ec 0.204 1. 356 Ec 0. to be used for the finite elements is found by equating the second moments of area of the element and the slab (equation (5. values of were arrived at for each row of elements. Ec . When the void diameter is less than about 60% of the slab depth.1440 Ec delem (m) 0.1440 Ec 0. r9. Arbitrarily choosing the elastic modulus in the X direction.109 1. is calculated using equation (5.0833 Ec 0.9 shows a cross-section through a typical voided slab bridge deck with tapered edges.989 Ec 0. These are created by placing void formers. r10 (m 3) 0.204 0 . Assuming a Poisson’s ratio of 0. 068 Ec 0.4 Finite element row number r1.2 for concrete. Ec : . It is common practice to discontinue the voids over the supports which has the effect of creating solid diaphragm beams there. as the second moments of area vary in two orthogonal directions.77) then gives an expression for the elastic modulus in the Y direction.0561 0.876 1. The shear modulus.064 Ec 0.1138 0.027 Ec 0 . then the equivalent depth. r8.1456 (m3) 0.4 Voided slab bridge decks Longitudinal voids are often incorporated into concrete slab bridge decks to reduce their self weight while maintaining a relatively large second moment of area.Page 211 Table 6. r11 r5 r6.3 Finite-element properties for Example 6.76)): Equation (5. These values are also given in Table 6.3.80) by substituting values for the Poisson’s ratio. .0013 Ec 0.989 Ec 0. the elastic modulus and the second moments of area per unit breadth.732 Ec 0 .3. In the finite-element program. it is modelled as materially orthotropic with a single value for element depth. to be equal to the elastic modulus of the concrete.0490 0. 106 Ec 0. r7. terms of the elastic modulus of the concrete. The variation of second moment of area in the two directions is allowed for by specifying two different elastic moduli. usually made from polystyrene. it is common practice to model the voided slab using the same methods as are used for . . r13 r3. 414 Ec 0.1456 0. Figure 6. within the formwork before casting the concrete. r14 r2. r12 r4. 6. in The elastic moduli in the two directions and the equivalent depths of each row of elements are given in Table 6. delem .838 0. 414 Ec The bridge deck is geometrically orthotropic.

On the other hand. Even if the voids are large. . Bakht et al. such slabs can be analysed using the same techniques as those used for solid slab decks but with modified member properties. the behaviour becomes more ‘cellular’. 6. The first step in the modelling of a voided slab deck is to determine the location of the neutral axis. (1981) reviewed many methods of analysing voided slab bridges. 6. the properties of each part of the deck are then calculated relative to the neutral axis of the complete deck. is straightforward. a voided slab deck is less likely to distort than the box girder section of Fig. For the Fig. The stiffness of the voided portion is simply subtracted from the stiffness of the solid slab. when the void diameter exceeds about 60%. For planar grillage or finite-element models.10 which can be modelled using a variation of the conventional grillage or FE methods known as ‘shear flexible’ grillage or FE. Determination of the transverse second moment of area and the torsional rigidity are not so simple. This is generally taken to be at a constant depth transversely and to pass through the centroid of the deck. regardless of the size of the voids. without specific guidance.10 Characteristic behaviour of cellular bridge deck: (a) original geometry. 6.10 and.Page 212 Fig. such a shear flexible model would be difficult to implement.9 Cross-section through voided slab bridge solid slab decks. Determination of the longitudinal second moment of area per unit breadth of a voided slab. Cellular decks are characterised by the distortional behaviour illustrated in Fig. then the position of the centroid may not be at mid-depth and should be calculated in the usual way. (b) deformed shape showing characteristic cell distortion . If the bridge deck has edge cantilevers or if the voids are not located at the centre of the deck. They propose that. 6.

88 0.11): (6.1).84 0. A slab where the voids were spaced three to four times the slab depth apart would have a transverse rigidity in excess of that predicted by equation (6.1) does not take into account the spacing of the voids as the authors maintained that this was not a significant factor. Clearly this equation is only applicable to slabs with a sensible void spacing. 6. This equation assumes that the centre of the voids and the deck centroid (for transverse bending) are located at mid-depth.6 or less.51 0.80 0.75 0.58 0. to that of solid slab. dv (Fig.55 0.4 Ratio of torsional stiffness of voided slab.90 0.74 0. and the diameter of the voids.60 0.64 0.45 0.85 0. islab (from Bakht et al.76 0.86 0. 6.7 0.77 0. the transverse stiffness can be approximated as being equal to the longitudinal stiffness.85 0.78 0.84 0. Fig.87 0. Bakht et al.70 0.64 0.68 0.69 0.82 0.71 0.72 0. When the void diameter to slab depth ratio is 0. Examination of equation (6.56 0.61 0. iv-slab.80 0.6 0.81 0. d.70 0.1) shows that the presence of the voids reduces the transverse stiffness by only 12% for a ratio of 0.Page 213 transverse second moment of area.1981) 0.5 0.9 0.89 0.1) Equation (6.65 0.48 0.79 0.75 0.62 0.66 0. (1981) recommend using the method of Elliott which gives this quantity in terms of the depth of the slab.90 .86 0.82 0.8 0. This is quite often a reasonable assumption when considering transverse bending.11 Cross-section through segment of voided slab bridge Table 6.6.

j slab. For a grillage model.12 Cross-section through bridge of Examples 6. Fig. 6.6 Fig. Bakht et al. Thus the total bridge is 25 m long consisting of 23 m of voided section and two 1 m diaphragms .12 shows the cross-section of a prestressed concrete bridge deck which incorporates circular voids along its len gth. jv-slab. reduction of the torsional rigidity for the edge voids is not warranted as voided slab bridge decks are usually tapered at their edges or have substantial edge beams.Page 214 For the torsional stiffness of voided slabs per unit depth. (1981) recommend using the method of Ward and Cassell. to that of a solid slab of the same depth. The voids stop short at each end forming solid diaphragm beams 1 m wide over the supports. The deck spans 24 m between the centres of supports and is supported on four bearings at either end as illustrated in the figure. jslab can be determined from equations (5. Example 6.71) and Table 6.5 and 6. 6. It was suggested that the values given in Table 6.5 . However. in most practical cases.13 Grillage mesh for bridge of Example 6.5: Grillage model of voided slab bridge Figure 6.4 for the ratio of torsional stiffness of the voided slab jv-slab.4 can then be used to determine jv-slab.4 are only applicable to internal voids in an infinitely wide slab because those at the edges possess much lower torsional rigidities. The neutral axis passes through the centroid of the deck which is located at mid-depth as the voids are located there. This gives the values presented here in Table 6. The layout and member properties are required for a grillage model.65) or (5. Bakht et al. conclude that.

Page 215 Fig. 6. By using this arrangement. 1. It is not considered appropriate to locate these grillage members at 0.14.5 m apart. with the exception of the outer row on each side where they are located midway between the edge of the outermost void and the edge of the deck. the supports coincide with the locations of nodes in the grillage mesh.1): Hence. The longitudinal direction is taken to be the X direction. i.13 shows a suitable grillage mesh. As the void diameters are in excess of 60% of the slab depth. the second moment of area is: .e.: The edge longitudinal grillage member represents a portion of deck equal to exactly half that of the internal members with the result that its second moment of area is given by: The second moments of area of the internal transverse members are determined using equation (6. the slab is treated as an orthotropic plate and the properties of the longitudinal and transverse members are determined separately. The longitudinal members are located midway between voids. The transverse grillage members are located in 17 rows. 6.14 Segment of voided slab Figure 6. The second moment of area of this member is found by subtracting the second moment of area of the circle from that of the rectangle. for the internal transverse members.3 times the depth of the slab from the edge as this location is within the void. The internal longitudinal grillage members represent the portion of deck illustrated in Fig.

5 and Fig. each 0. will be 1. 6.15.4.5 m wide. as illustrated in Fig. adjacent to the diaphragm. Both the ratio dv/sv and dv/d are 0. 6.67. the next row of transverse members.65) to calculate the torsion constant per unit breadth for a solid slab then gives: The torsion constants for both the longitudinal and transverse members in the voided slab are then found by multiplying this value by their respective breadths.75 m wide and will have a second moment of area of: The torsion constant for the grillage members is found from Table 6.6: Finite-element model of voided slab bridge A finite-element model is required for the 25 m long voided slab deck of Example 6. Each longitudinal row of elements represents a strip of the deck from midway between one void to midway between the next. The second moment of area per unit .5 m. Interpolating in the table gives a ratio for the torsion constants per unit breadth of: Taking equation (5. are used to represent the diaphragm. two transverse rows of elements. At the ends. The transverse rows of elements adjacent to the diaphragms at each end are 1.Page 216 For the 1m wide end diaphragms. The torsion constant per unit breadth for the diaphragm is given by equation (5. For convenience.12.3 m wide in order to make up the correct total length. the second moment of area is simply: As the diaphragm is only 1 m wide and the transverse members are spaced at 1.71): Example 6. a mesh consisting largely of 1.2 m square elements is chosen.

E x.15 Finite element mesh for bridge of Example 6.2 m wide strip of the deck.77) gives an expression for the elastic modulus in the Y direction: . The total second moment of area of this strip is again calculated by subtracting the second moment of area of the void from that of the equivalent rectangular section: Hence. Selecting the modulus of elasticity in the X direction. equal to the modulus for the concrete. it is necessary to calculate a single equivalent value for slab depth. de .1) gives: The slab is geometrically orthotropic.6 breadth in the longitudinal direction can be found by considering a 1. the second moment of area per unit breadth is: For the transverse direction. To model this as a materially orthotropic plate.76) implies a depth of element of: Equation (5. equation (6. then equation (5. 6. as the second moments of area (rather than the moduli of elasticity) are different for the longitudinal and transverse directions.Page 217 Fig.

a precast concrete slab or even a completely in-situ beam and slab as illustrated in Fig. for example. The shear modulus for the diaphragms is given by equation (5. it is important that the slab be idealised correctly in the model as. 6. Transverse diaphragm beams can be used to provide additional load sharing between longitudinal beams. Load sharing between the beams may be provided by a top slab or by a combination of a top slab and a number of transverse diaphragm beams. the slab provides a means for load sharing between longitudinal beams. 6. the beams generally act alone and must be capable of carrying their self weight. During construction. The slab acts to transmit applied loads to the beams by spanning transversely between them. Beam and slab decks may be formed in a number of ways. Therefore they can be analysed by similar methods to those proposed for slab decks in the preceding sections. Wide diaphragms also serve to improve the shear . The main load-carrying component of a beam and slab deck is the longitudinal spanning beams.80): Taking a Poisson’s ratio of 0.2. 6.Page 218 where E v-slab is the modulus of elasticity of the concrete in the voided slab. They differ from slab bridge decks in that a large portion of their stiffness is concentrated in discrete beams which run in the longitudinal direction.17. Consequently.16 (a) and (b). Many other methods exist. the weight of the slab and any construction loads present. an overly stiff slab may lead to a prediction of load sharing between adjacent beams which does not occur in reality.2 m thick and have moduli of elasticity in both directions equal to that of the concrete. In addition to this. the most obvious being the casting of an in-situ concrete slab on steel or precast concrete beams as shown in Fig. this gives: The diaphragm beams are solid so the corresponding elements are 1. Beam and slab bridges are generally suitable for similar span lengths as slab bridges but are often chosen in preference because of their ability to be easily erected over inaccessible areas such as deep valleys or live roads or railways.16 (c). the structural action of these decks is considered to be two-dimensional. The extent of this load sharing is largely dependent on the stiffness of the slab. This phenomenon is indicated in Fig. On completion. such as steel beams with a composite steel and concrete slab. The shear modulus is calculated from equation (5. 6.5 Beam and slab bridges Beam and slab decks are used for a wide variety of modern bridges.67).

6.16 Forms of beam and slab construction: (a) in-situ slab on steel beams. (b) in-situ slab on precast concrete beams. a diaphragm beam is constructed over intermediate supports to provide additional continuity. The obvious exception is that grillage beams should normally be . continuity between adjacent spans may be provided by the slab alone.5. (c) in-situ beam and slab Fig. 6.Page 219 Fig. In precast concrete beam construction.1 Grillage modelling Grillage modelling of beam and slab decks generally follows the same procedures as for slab decks.17 Load sharing in beam and slab decks: (a) thin slab—little load sharing. 6. but quite often. (b) thick slab— increased load sharing capacity by extending the portion of the bridge near a support which is solid.

Unlike slab decks.Page 220 Fig. The slab will act as a flange to such beams making them T. 1 m wide. The properties of the transverse grillage members should be derived from the properties of the relevant diaphragm beam or slab as appropriate.or L-section in shape.3s for L-sections as illustrated in Fig. This generally complies with the need to locate beams at the supports as.3. there will be a much greater variation in the depth of the neutral axis than in slab bridges. due to the low stiffness of the slab. where s is the spacing between beams. Each precast beam is supported on a bearing at each end and the deck has a single span of 20 m (centre to centre of bearings). 6.7 stated that transverse member spacing should be between one and three times the longitudinal member spacing. the section properties for beam and slab decks are generally calculated about the centroid of this composite section. It is possible to use one grillage member to represent two or more actual beams but this complicates the calculation of properties and interpretation of the results with little saving in analysis time in most cases. For slab decks.19 shows the cross-section of a beam and slab bridge deck consisting of a cast in-situ slab on precast concrete Y-beams. . not about the centroid of the whole bridge. supports are normally provided directly beneath the beams. 6. The properties of the longitudinal grillage members are determined from the properties of the actual beams and the portion of slab above them. Section 5. The elastic modulus of the precast beams is 34 kN/mm 2 and that of the in-situ slab is 31 kN/mm2.18 Effective flange width of diaphragm beam: (a) plan at end. Example 6. each acting about its own axis. Hambly (1991) suggests an effective flange breadth of bw+0. This spacing is also recommended for beam and slab bridges although greater spacings are possible without significant loss of accuracy. Solid diaphragm beams. transverse members are required to represent the transverse stiffness of the slab. in beam and slab construction. are provided at each end and no additional transverse beams are located between these. A grillage model of the beam and slab deck is required. This approach is justified on the basis that. In addition.7: Grillage model of beam and slab bridge Figure 6. (b) section through L-beam positioned at the location of the longitudinal beams. Transverse grillage members should clearly be placed at the location of all diaphragm beams.18.

the beam cross-section is approximated as two rectangles as illustrated in Fig. 6. Figure 6.20. 6. for a rectangular section according to Ghali and Neville (1997) is: (6.374 m 2 Second moment of area = 0.Page 221 Fig.19 (b) shows the exact dimensions of the precast beam.2) .0265 m2 Height of centroid above soffit = 0. J. The torsion constant. The section properties of the precast beam are generally given by the manufacturer. (b) detailed dimensions of Y-beam The modular ratio for the in-situ and precast concrete is: The procedure adopted is to assign a modulus of elasticity of 34 kN/mm 2 to all of the grillage members (except for the end diaphragms). the properties are: Area = 0. in this case. For the purposes of determining the torsion constant.347 m The torsion constant is generally not given and must be determined by the analyst.19 Beam and slab bridge deck: (a) cross-section. but to factor the stiffness of the slab by this modular ratio. The torsion constant of a cross-section made up of rectangles is commonly estimated by calculating the torsion constants of the individual rectangles and summing.

‘Dummy’ longitudinal members with nominal stiffness are provided at the edges and transverse members are continued past the ends of the edge Y-beams to connect to them. Transverse members are positioned at each end to model the diaphragms.5%). then these beams should be assigned very small section properties relative to those used elsewhere in the grillage (say. 6. both . consideration need not be given to in-plane horizontal movements at this stage.20 Equivalent section made up of rectangles for determination of torsion constant where b is the length of the longer side and a is the length of the shorter side. For the interior longitudinal members. Supports are located at the ends of each longitudinal beam (other than the dummy beams).Page 222 Fig. A finite-difference technique was used to determine the constant in this case and a value was found as follows: The simplified method can be seen to be accurate to within 7% for this section. This is a convenient method for applying loads such as those due to parapet railings. Some grillage programs allow the definition of ‘dummy’ beams. This gives a transverse to longitudinal member spacing ratio of 2:1 which is acceptable. 0.21 shows a suitable grillage layout for this bridge deck. 6. Applying this equation to the rectangles of Fig. Figure 6. As the grillage model is planar. If this is not the case. the second moment of area is the sum of the second moment of area of the Y-beam plus the 1 m width of slab above it.20 gives a torsion constant for the Y-beam of: The constant can be found more exactly by applying Prandtl’s membrane analogy as described by Timoshenko and Goodier (1970). Additional transverse beams are located at 2 m centres between these to represent the transverse stiffness of the slab. A longitudinal grillage member is positioned at the location of each Y-beam.

This raises the centroid above that for the interior members. The torsion constant of the slab is determined using equation (5.Page 223 Fig. Hence: Each edge longitudinal member is similar to the interior members except for a 0.2×0. 6.3 m 2 upstand.21 Plan view of grillage mesh taken about the common centroidal axis of the section.65). The stiffness of the slab is reduced by factoring it by the modular ratio. Summing moments of area about the soffit gives: . Hence the equivalent area of the combined section is: The section centroid is found by summing moments of area about the soffit: where yb is the distance of the centroid above the soffit. Hence: The second moment of area of the combined section is: The torsion constant is taken as the sum of the torsion constants of the Y-beam and the slab.

6.22 Section through end diaphragm beam Hence the second moment of area of the edge section is: For the transverse members. For the second moment of area: The torsion constant is: The slab acts as a flange to the diaphragm beams. the properties are determined in the usual manner. The recommended flange breadth is the sum of the web breadth plus 0. For the slab bending about its own axis.3 times the beam spacing: Hence the centroid is: above the soffit. the row of transverse members adjacent to the diaphragm accounts for the slab up to 1 m from the centre of the diaphragm as illustrated in Fig. 6. .Page 224 Fig.22.

a combined model is generally used which represents the slab with finite elements and the beams with grillage Fig. where the transverse grillage members have been subdivided to include much stiffer portions at their ends. 0.23 Transverse modelling of decks with wide flanges: (a) in-situ slab on precast concrete Ubeams. It can be seen from this that the span of the slab in the model is too long.5 m of slab to be accounted for in the diaphragm stiffness. then the slab can inadvertently be modelled as having an excessively long transverse span. Figure 6.65): The modulus of elasticity for in-situ concrete is used for the diaphragm beams. 6.Page 225 This leaves 0.5 m of flange from equations (6.23(c).23(a) shows a deck consisting of a concrete slab on precast concrete U-beams.2 m of which is deemed to be bending about its own axis. One possible solution to this is shown in Fig. (c) improved grillage model . 6. The second moment of area is thus: The torsion constant is calculated allowing for 0. Figure 6.2) and (5. If the web width at the top of the longitudinal beams in a beam and slab deck is large relative to their spacing.23(b) shows a grillage model with longitudinal grillage beams for the Ubeams and transverse beams spanning between them representing the slab. 6.5. (b) conventional grillage model where slab has excessive transverse span.2 Finite-element modelling In finite-element modelling of beam and slab decks. This would lead to an excessively flexible slab which in turn would lead to the incorrect modelling of load sharing between the U-beams.

6.24 shows a suitable finite-element mesh incorporating grillage members longitudinally. Grillage members are used for each of the Y-beams and for each of the end diaphragms.24 Combined finite-element and grillage mesh . They are also assigned the elastic properties of the slab. They are assigned a modulus of elasticity and a Poisson’s ratio equal to those of the concrete in the slab. In the first approach.7.19. The finite elements are assigned a thickness of 0. One of two approaches can be taken. Care should be taken when determining the properties of the finite elements representing the slab. For the longitudinal grillage members. This is generally straightforward to implement and follows the recommendations made for slab bridge decks.7 and Fig. the properties of the combined Y-beam and the 1 m width of slab above it are determined relative to the centroidal axis of the combined section. From Example 6. 6.Page 226 members. the slab is modelled using orthotropic finite elements with the true transverse and longitudinal properties applied in both directions. the slab is modelled using isotropic elements which are assigned a thickness equal to the depth of the actual slab. the second moment of area of the combined section is: Fig.16 m which is equal to the depth of the slab. The beams are then modelled by grillage members with the properties of the actual beams excluding the contribution of the slab. The longitudinal grillage members are then assigned the stiffnesses of the combined beam and associated portion of slab minus those already provided through the finite elements. Figure 6. Supports are provided at the ends of each longitudinal grillage member. The modulus of elasticity and Poisson’s ratio for the beams are used for these members.8: Finite-element model of beam and slab bridge A finite-element model is required for the beam and slab bridge of Example 6. An element length of 1 m in the longitudinal direction results in a maximum element aspect ratio of 1:2 which is considered to be acceptable. Example 6. In the second approach. The finite elements continue to the edge of the deck resulting in a row of elements 0. The stiffness of the slab which has already been applied through the finite element is subtracted.5 m wide at each side.

2) and (5.Page 227 The second moment of area of the 0.16 m thick finite elements is then subtracted to give the second moment of area to be used for the grillage member: The torsion constant for the combined section was arrived at in Example 6.25). 6. the torsion constant to be assigned to the grillage members is simply that of the Ybeam: In Example 6.3 Transverse behaviour of beam and slab bridges The top slab in a beam and slab bridge is often designed transversely as a one-way spanning slab supported by the longitudinal beams (Fig. 6.22): For the finite-element model. However.7 by adding the individual torsion constants of the Y-beam and slab. From equations (6.7. Hence.65): 6.5. the stiffness of the slab bending about its own axis is not required and a small component of stiffness inadvertently contributed by the elements must be subtracted: The torsion constant is that of a rectangular section less the portion inadvertently added through the elements. the second moment of area for the end diaphragms in the grillage model was calculated as (refer to Fig. the elements are present up to the centre of the diaphragm to represent the transverse stiffness of the slab about its own axis. such an approach results in a great quantity of reinforcement and has been shown to be . As the slab is represented by the elements.

as indicated in Fig.26 shows a number of commonly used cellular deck forms.4. This distortion is caused by the localised bending of the webs and flanges of the individual cells. 6. The result is that load is transferred from the slab to the beams by arching action rather than bending action alone. The most common type are box girder decks. The principal factors affecting the distortion are the dimensions of the cells relative to the deck depth.27(a) and (b). In these cases. the Ontario Highway Bridge Design Code (OHBDC. It is the transverse distortional behaviour that makes the analysis of cellular decks different from other forms. To account for observed arching action. The behaviour is similar to that observed in Vierendeel girders. The third form of behaviour is twisting. 1992) allows for the provision of much less reinforcement than would be found by an assumption of bending. . which characterises cellular structures. can also be considered to be of a cellular form.27(d). 6.Page 228 Fig. as illustrated in Figs. The beams have a considerable lateral stiffness and have the effect of confining the slab. The fourth form. and the extent (if any) of transverse bracing to the cells. the stiffness of the individual webs and flanges. is transverse cell distortion. as indicated in Fig. with single or multiple rectangular cells.25 Detail of section in beam and slab deck quite conservative. 6.5. as was discussed in Section 6. alternative methods are available for their analysis which are generally more convenient. However. 6. 6. Clearly the provision of transverse diaphragms along the span of a cellular deck will significantly reduce the degree of transverse distortion.27(c). In two reported cases (Bakht and Jaeger. the slab depth to beam spacing ratios were 1:12 and 1:13. Voided slab bridges. There are four principal forms of structural behaviour associated with cellular bridges. The first two of these are longitudinal and transverse bending. Canadian bridges have been built without any transverse slab reinforcement but using steel straps to guarantee confinement. with large diameter circular voids. 1997).6 Cellular bridges Cellular bridge decks are formed by incorporating large voids within the depth of the slab. Figure 6.

27 Behaviour of cellular decks: (a) longitudinal bending. (b) transverse bending. (c) twisting. (d) transverse distortion . 6. 6.26 Sections through alternative cellular bridge decks Fig.Page 229 Fig.

1 Grillage modelling Grillage modelling of cellular bridge decks can be achieved by use of what is commonly referred to as a ‘shear flexible’ grillage. Hence. then the load acting on each can be taken as P/2. 6. fixed against rotation at both ends and subjected to a vertical force P/2. from equation (6. If it is assumed for now that the webs are stiff and that transverse distortion is caused by bending of the flanges only. the deck is idealised as a grillage of beam members in the usual manner.4) Fig. (b) distorted shape .28 Distortion of single cell with stiff webs: (a) applied loading.28(b). except that the transverse members are given a reduced shear area such that they experience a shear distortion equal to the actual transverse distortion of the cells in the bridge deck. In this method. then the distorted shape of the cell is as shown in Fig. Clearly such a method requires a grillage program which models shear deformation as well as bending and which allows for the specification of a shear (or ‘reduced’) area for the members independently of the other section properties. Figure 6. is: (6.6.3). The method is illustrated below by means of an example. The vertical deflection due to the bending of a beam of length l. 6. where d is their thickness. If the flanges are of equal thickness.3) where I is the second moment of area and E is the modulus of elasticity. the deflection due to flange distortion is: (6.Page 230 6. The flanges of a cell will act as beams transversely with a second moment of area per unit breadth equal to d3/12.28(a) shows a single cell of width l of a cellular bridge deck under the action (transversely) of a vertical load P.

The second term is the deflection due to shear deformation which. the webs of cellular decks are also flexible and consequently they too contribute to the overall transverse distortion. For cellular decks of other shapes. an expression for the required shear area per unit breadth of a shear flexible grillage member is found: (6. 6. Fig. In practice. Assuming points of contraflexure at mid-height and equating the deflection of this cell to the shear deformation of a grillage member gives a more exact and general expression for shear area per unit breadth: (6.7) Details of the derivation of this formula are given in Appendix D. it has been suggested by others that a plane frame analysis be carried out to determine the equivalent shear area of the transverse grillage members.29 shows a single cell of a cellular bridge deck with a constant web thickness but different upper and lower flange thicknesses. it was assumed that transverse distortion was caused by the distortion of the cell flanges only. this may be difficult to carry out accurately in practice due to such factors as cracking in concrete sections.29 Cross-section through cellular deck showing dimensions of cell . is small relative to the deflection due to bending.Page 231 The total deflection in a cantilever of length l. However.6) In this example. Figure 6. for most structures.5) where G is the shear modulus and as is the shear area of the section per unit breadth. By equating the shear deformation in a transverse grillage member to the bending deformation of the cell flanges in the bridge. subjected to a vertical load per unit breadth of P at its free end is: (6.

As mentioned previously. this is not the Fig.31(a). Such an equation is valid when the shear flows are opposing through the depth of the section as illustrated in Fig. 6. (b) portion of box section with cantilever .30. For a portion of box section.Page 232 The second moments of area of the longitudinal members in a shear flexible grillage are determined in the same way as for slab decks. 6. 6. the second moment of area of the top and bottom flanges is calculated about an axis at the bridge mid-depth as illustrated in Fig. the neutral axis of the bridge deck is first determined and the second moment of area of the portion of deck represented by each longitudinal grillage member is determined about that axis.8) The first term in equation (6.30 Longitudinal section through deck for transverse bending Fig. The torsion constants of the longitudinal and transverse grillage members are based on the portion of section represented by the members.8) is generally small relative to the second and is often ignored.31 Shear stresses due to torsion: (a) rectangular section. As for slab bridges. (6. For the transverse members. where b is the breadth and d the thickness. the torsion constant for a thin rectangular section twisting about its own axis may be approximated by bd3/3. 6.

The two edge members represent the portion of deck from the edge to halfway between the first and second webs (Fig. Figure 6. the centroid can be shown to be located at 0. Additional transverse members are placed at 2 m centres giving a longitudinal to transverse member spacing ratio of 1:1. A grillage model is required. By summing moments of area about any point in the section. is an increment of length and di is the thickness of that increment.9) where a is the area enclosed by the centre line of the wall. 6. Four longitudinal members are chosen. Applying equation (6. The first step in determining the grillage member properties is to find the neutral axis of the deck which is assumed to pass through the centroid.10) Example 6.Page 233 case as illustrated in Fig.31(b) except in the edge cantilevers.65 m . There are 2 m thick solid diaphragms at the end and central supports. The torsion constant for a thin-walled box section is given by: (6.9: Shear flexible grillage model of a cellular bridge deck Figure 6. the contribution of the webs is accounted for through the shear forces in the longitudinal beams and should not be accounted for again here. one at the centre of each web. 6. The two internal members represent the portion of deck from halfway between the first and second webs to the centre.32). l i.11. Transverse grillage members are located at the ends and at the central support to represent the transverse diaphragms.9) to the single cell of Fig.33 shows a convenient grillage mesh.29 would give: However. 6. three-cell bridge deck with edge cantilevers. A formula suggested by Hambly (1991) halves the constant and removes the web term: (6.32 illustrates a two-span. It is assumed that the deck is continuously supported transversely at each support.

Page 234 Fig. For the edge longitudinal members: For the internal longitudinal members: For the transverse members.32 Cellular bridge of Example 6.33 Plan view of grillage mesh above the soffit. itrans is given by equation (6. The second moments of area for the longitudinal members about this axis are then determined. the second moment of area per unit breadth.8): . 6. 6. (b) longitudinal section Fig.9 (dimensions in m): (a) cross-section.

10): This gives a torsion constant for the interior longitudinal members of 0. Then equation (5.2 is assumed. giving: . a common approximation for I-sections. the shear area is taken as the area of the webs. a Poisson’s ratio of 0. The edge members only represent half a cell and the contribution of the cantilever is added: The torsion constant per unit breadth for the transverse members is taken to be equal to that of the longitudinal members: The shear area per unit breadth of the transverse grillage members is given by equation (6.7): For concrete.36 m4. giving: The torsion constant per cell is given by equation (6. giving: For the longitudinal members.Page 235 The breadth of the transverse members is 2 m.67) gives: which results in a shear area of: The breadth of the transverse members is 2 m.

A high degree of twisting accompanied by large torsional moments (mxy ) are also associated with skew decks. As a result. The greatest reactions will tend to occur at obtuse corners in skew decks and the smallest reactions at acute corners. uplift can occur at acute corners which is generally to be avoided. 6.34(b) shows a suitable grillage layout.34(a) shows a long narrow bridge deck with a high degree of skew and Fig. This deck will tend to span in the skew direction so the longitudinal grillage members are aligned in that direction.7 Skew and curved bridge decks Many bridge decks incorporate some degree of skew and others are curved in plan. the Wood and Armer equations can dictate a requirement for top reinforcement near supports where hogging would not normally be expected. in reinforced concrete. Large reactions at obtuse corners lead to high shear forces which can also be difficult to design for. In highly skewed decks. Significant skew in bridge decks leads to a non-uniform distribution of reactions between supports.Page 236 The end and central diaphragm beams are 1. Care is needed in modelling the support system in such cases as any flexibility will cause a redistribution of reactions. Hence: 6.3% of the actual area. The second moment of area of the grillage members representing these is therefore: The torsion constant for the diaphragms is determined using equation (6. the shear area can be shown to equal 83. Figure 6. The transverse grillage members should generally be oriented perpendicular to the longitudinal members.2 m deep by 2 m wide. An important consideration is to place the grillage members in the directions of principal strength. the span length and the width of the deck.7. For a rectangular section.1 Grillage modelling A suitable grillage model of a skew deck will depend largely on the angle of skew.2): The shear area of the transverse diaphragm is taken as the actual shear area as no significant transverse distortion is assumed to take place. A grillage or finite-element model can be formulated for such decks based on the recommendations given in earlier sections along with some additional considerations given here. An exception to this is in concrete decks where the transverse reinforcement is not . 6.

wide bridge deck with a small angle of skew and Fig. In such cases. highly skewed bridge deck: (a) plan view. Care should be taken with the edge grillage members which generally will have to be orientated in the skew direction. This deck will tend to span perpendicular to the supports rather than along the skew direction. Figure 6. follow the curved layout closely due to the fineness of the mesh.34 Long. Once again.34(c). (b) grillage layout . The longitudinal members.35(b) shows a suitable grillage layout. wide bridge deck with small skew: (a) plan view.35(a) shows a short. Some analysis programs will allow the use of curved beams. it is generally more appropriate to orientate the transverse members parallel to the transverse reinforcement as illustrated in the alternative grillage layout of Fig. narrow. then this should be allowed for when assigning the properties of the edge beams in the grillage. the longitudinal grillage members are orientated in this direction.Page 237 perpendicular to the longitudinal reinforcement. 6. the transverse grillage members are orientated perpendicular to the longitudinal members. The Fig.35 Short. Figure 6. Bridge decks which fall between the extremes of Figs. 6. but straight beams will be sufficiently accurate if the grillage mesh is fine enough. 6.34 and 6. Consequently. although straight. Curved decks pose no particular problem for grillage modelling.35 will require a greater amount of judgement by the analyst in choosing a suitable grillage layout.36 shows a suitable grillage mesh for a curved bridge deck. 6. 6. (b) grillage layout. If significant edge beams or stiffening is provided to the bridge deck. (c) alternative grillage layout Fig.

no special consideration need be given to directions of strength as the elements are two-dimensional and will model the twodimensional behaviour of the skew or curved slab. 6. Generally.7.36 Grillage layout for curved bridge deck transverse members radiate from the centre of the curve.Page 238 Fig. 6. they are approximately perpendicular to the longitudinal members. This is an advantage that the finite-element method has over the grillage method. especially for Fig.37 Alternative finite-element meshes: (a) skewed quadrilateral finite elements. (b) alternative triangular elements . In this way.2 Finite-element modelling Finite-element modelling of skew or curved decks should be carried out according to the recommendations for right decks. 6.

highly skewed quadrilaterals may result in round-off errors due to calculations involving small angles. triangular elements.37(b).Page 239 inexperienced users who might not have the expertise to formulate a suitable grillage model. Skewed quadrilateral elements. In such cases. 6. may be more effective. However.37(a). 6. can give results which are just as accurate as those for rectangular elements and they are very easy to implement. . as illustrated in Fig. as illustrated in Fig.

the analysis of bridge decks using planar models is discussed. 7. When flanges or cantilevers are wide and slender. Bridge decks with edge cantilevers are considered but it is stipulated that only those with short cantilevers should be analysed by the methods proposed. longitudinal bending stresses are set up. These are distributed transversely from one part of the deck to adjacent parts by interface shear stresses. independently of the rest. the bridge deck has a noncontinuous neutral axis as indicated in the figure. This phenomenon is known as ‘shear lag’ as it is associated with interface shear and is characterised by the lagging behind of axial stresses at the edges of cantilevers. Thus. interface stresses are generated as illustrated in Fig.1 Introduction In Chapter 6.Page 240 Chapter 7 Three-dimensional modelling of bridge decks 7. Figure 7. when the bending moment in a flanged beam varies from one point to another. As the rejoined bridge bends. If a load were applied to the deck in this condition. In this chapter.2 Shear lag and neutral axis location When a bridge deck flexes. a common centroid can be found and the entire bridge is often assumed to bend about a neutral axis passing through this point. the . The extent of the reduction of stress is dependent on both the geometric shape of the bridge deck and the nature of the applied loading. each part would bend about its own centroid. 7. the edges do not receive the same amount of axial stress as those near the centre of the bridge.2(b). In this condition.1.2(a) shows a bridge deck with the edge cantilevers separated from the main part of the deck. If the bridge deck is now rejoined. 7. The theoretical background is reviewed and a number of solutions are suggested including three-dimensional methods of analysis. This common neutral axis can be seen in Fig. the problems associated with bridge decks such as those with wide edge cantilevers are discussed.

1 Interface shear stresses in flanged beam subject to bending Fig. (c) variation in longitudinal stress at top of deck. 7. due to shear lag. The effect of bending is not felt to the same extent in the edges of the cantilevers as it is elsewhere. do not experience the same amount of axial stress as the main part of the deck. This is because the edges of the cantilevers tend to bend about their own Fig.2(c). as can be seen in Fig. (d) actual neutral axis location . 7. 7.2 Transverse variation in neutral axis location: (a) if cantilevers and main deck were free to act independently.Page 241 remote edges of the cantilevers. (b) commonly assumed straight neutral axis.

It could be said that the variation in the neutral axis location in a bridge deck is caused by shear lag or that shear lag is caused by the tendency of each part of the bridge deck to bend about its own neutral axis. as illustrated in Fig. 7. can be used to determine the maximum stress in the cantilever.Page 242 centroidal axes. 7. Such a non-uniform neutral axis is illustrated in Fig. 7. as described in Chapter 6.3 Actual and calculated distributions of longitudinal bending stress at top of flanged deck . 7. 7. but this tendency causes the overall bridge deck neutral axis to move towards the centroid of the cantilevers at the edges. be . reproduced here as Fig. Obviously they are not free to do this. There is a strong link between shear lag and neutral axis location. Hence. The method uses a notional width of cantilever in the grillage or finite element model which has a uniform stress distribution equal in magnitude to the maximum stress in the actual cantilever. a two-dimensional model with an effective flange width. is often used which does not take account of shear lag. b.2(d). a two-dimensional analysis. The chart. The correct effective flange width to be used for the cantilever is largely dependent on the ratio of the actual cantilever width to the length between points of zero moment (points of contraflexure).3. as it is from these points that longitudinal stresses begin to spread out into the cantilevers.3 Effective flange width In the design of bridge decks. and actual flange width. to the ratio of actual flange Fig. analysed with no allowance for shear lag. Hambly (1991) presents a chart for the determination of effective flange widths for beams subjected to distributed and concentrated loads. A three-dimensional analysis can automatically account for shear lag as it allows for variations in neutral axis location directly. The effective flange width is also dependent on the form of the applied loading. relates the ratio of effective flange width. It is possible to overcome this problem by assuming an ‘effective flange width’ for the edge cantilevers.4.

(b) showing effective flange width .4 Effective flange width for different loadings (solid line).5(a) shows the cross-section of a bridge deck with edge cantilevers. Also shown in the figure are the popular approximations for this relationship: and: which can be seen to be reasonably accurate for relatively wide flanges. and length between points of contraflexure. b.1: (a) showing actual cantilever widths. The cantilevers are 2.1: Effective flange width Fig. 7. L.Page 243 Fig.4 m wide and the deck has a single simply supported span of 20 m. 7.5 Cross-section of bridge deck of Example 7. and common approximations (dashed line) width. Example 7. Figure 7.

23 m and 1. Unfortunately. From Fig.12. 7. the length between points of contraflexure. the use of such models is currently limited mainly to research and highly specialised Fig. Hence. a cellular structure or transverse diaphragms pose no particular problems. Inclusion of voids. In addition to this. is equal to the span length in this case. L. some form of threedimensional model is necessary to achieve an accurate representation of the behaviour of the structure. b/L. A constant stress is assumed in the modelled portion of the cantilever and that part of it outside the effective flange width is ignored.67 for the uniformly and point loaded cases respectively.61 m respectively. the ratio of the cantilever width to this length. When the effects of shear lag are significant.4 Three-dimensional analysis The use of two-dimensional analysis methods with effective flange widths is approximate at best and does not address the issue of upstands which are often provided at the edges of bridge cantilevers. This results in effective flange widths of 2.6 shows such a model of a portion of bridge deck with edge cantilevers.4 the ratios of be/b are 0.6 Portion of bridge deck modelled with solid brick elements .4/20=0. The benefit of this type of model is that it can be used to describe the geometry of highly complex bridge decks very accurately. Figure 7.5(b) shows this effective flange width for one of these load cases. is 2.Page 244 As the span is simply supported. 7. Figure 7. the model automatically allows for any variations in the location of the neutral axis and hence allows for shear lag in edge cantilevers. This example highlights the limitations of the effective flange width method as the nature of the loading causes a substantial variation in the effective flange width. 7.93 and 0. One such technique is three-dimensional finite-element analysis using solid ‘brick’ type elements.

5 Upstand grillage modelling In Chapter 6. such as upstand grillage modelling. The plane grillage meshes are then connected using rigid vertical members. There is also no need to assume an effective flange width to allow for shear lag effects. be it straight or varying. Figure 7. 7.7 shows an upstand grillage model for a bridge deck with edge cantilevers. is required. the edge cantilevers are modelled with grillage members which are located at the centroid of the cantilevers while the main part of the deck is modelled with grillage members located at the centroid of that part. The authors have used this type of model extensively to develop and test a number of simplified three-dimensional models which are suitable for everyday bridge design. particularly for post-processing of the large quantities of output data generated. The upstand grillage analogy is a direct extension of the planar grillage analogy. The properties of each part of the deck are determined relative to its own centroid.Page 245 applications due to excessive run times and computer storage requirements and due to a shortage of user-friendly software. it will automatically determine the location of the neutral axis. When this is not the case. 7. will be accounted for automatically. a three-dimensional technique. difficulties arise when in-plane effects are considered. but involves the modelling of each part of the bridge deck as a separate plane grillage located at the centroid of the portion of bridge deck which it represents. Although the upstand grillage seems to be a relatively simple and powerful model. there is no need to make an assumption as to the location of the overall bridge neutral axis. It is only suitable for bridge decks where the neutral axis remains substantially straight across the deck and is coincident with the centroidal axis of the bridge. for each load case considered.7 Upstand grillage model . shear lag where it exists. That type of analysis is referred to as planar grillage as all of the grillage members are located in one plane. Consequently. Unlike the plane Fig. In this. Consequently. Some of these simplified models are discussed in the following sections. grillage modelling is applied to bridge decks including those with edge cantilevers. As the model is three-dimensional.

If part of the actual bridge deck deforms in-plane. or the nodes at the ends of the members can be restrained against in-plane rotation. the real problem is the occurrence of local in-plane distortions of the grillage members.6. which are clearly inconsistent with the behaviour of the bridge deck. the plane grillage model and the upstand grillage model. a three-dimensional finite-element analysis using solid ‘brick’ type elements. Figure 7. as illustrated in Fig. as tends to occur at the ends of edge cantilevers. Such behaviour in the model can be avoided in one of two ways.10(b) shows the corresponding quantities at span.9(a) shows the crosssection of a 24. It is the in-plane bending component which is not modelled by an upstand grillage with in-plane rotational restraints. 7.8 m single-span bridge deck with wide edge cantilevers. Figure 7.8 In-plane distortion of members in upstand grillage model . However.8. Restraining in-plane rotations in the model may have adverse effects in some cases.Page 246 grillage. To test the accuracy of both models. the three-dimensional nature of the model causes in-plane displacements in the grillage mesh. This bridge was analysed under the action of a constant longitudinal bending moment using a planar grillage model and an upstand grillage model (Keogh and O’Brien. 7. 7. Assuming the elaborate model Fig. 1996). Figure 7. and it may even be prudent to adopt both. was also carried out.9(b) shows an exaggerated plan view of the deflected shape of the three-dimensional brick finite-element model (only one-half of the model is shown as it is symmetrical). the imposition of rotational restraints will prevent this behaviour from occurring in the model which may significantly affect the accuracy of the results. Figure 7. The members can be given very large in-plane second moments of area.10(a) shows the longitudinal bending stress predicted along the top of this bridge deck at mid-span by the three-dimensional brick finite-element model. Both of these measures will have similar effects. The upstand grillage predicts almost the same stress as the elaborate three-dimensional brick finite-element model at mid-span while the plane grillage predicts a higher stress in the cantilever and a lower stress elsewhere. The in-plane distortion seen at the end of the cantilevers is made up of both in-plane shear distortion and in-plane bending. Only half of the width is shown and the crosssection is included for reference. This results in a requirement to specify the cross-sectional areas of the grillage members as well as the second moments of area (about both axes) and the torsion constants. similar to that shown in Fig.

7. this is not the case at span where the upstand grillage in fact makes a poorer prediction of stress in the cantilever than the plane grillage.9 In-plane deformation in cantilevers of deck: (a) cross-section. 7. the benefits of the upstand grillage can be seen at this location.8 which caused inaccuracies elsewhere in the upstand grillage model. Clearly this is not a satisfactory approach for many bridge decks. the complete removal of the rotational restraints resulted in the behaviour illustrated in Fig. This inaccuracy in the upstand grillage is attributable to the use of inappropriate rotational restraints at the ends of the cantilevers. (b) plan view of deflected shape (half) to be accurate. which may not be available prior to analysis. One solution is to remove the rotational restraints only where the in-plane bending actually occurs but this method requires a degree of knowledge regarding the behaviour of the deck. . However. Unfortunately.Page 247 Fig.

(b) at span Example 7.11(a) shows the cross-section of a bridge deck with edge cantilevers. The deck is 25 m long with a single. An upstand grillage model is required.10 Calculated longitudinal bending stress on top surface of deck: (a) at mid-span.2: Upstand grillage model Figure 7. 7. simply supported span between bearings of 24 m and is supported along the entire width of the main part of the deck at each end. .Page 248 Fig.

the portion of bridge deck associated with each grillage member is indicated by the broken lines. This gives vertical members with a length of 0. Seventeen rows of transverse members are provided at a constant spacing of 1. the second moment of area per unit breadth is calculated . Assuming the main deck slab to be isotropic.4 m. The properties of the members in the upstand grillage model are easily determined. Figure 7.3.11(c) shows a plan of the upstand grillage mesh. (b) crosssection with grillage members superimposed.2 (dimensions in mm): (a) cross-section.6 m from the top.5 m. (c) plan view of grillage mesh Figure 7.11(b) shows the cross-section with a suitable upstand grillage model superimposed.11 Upstand grillage model of Example 7. 7.Page 249 Fig.2 m below the top. The grillage members representing the main part of the deck are located at the centroid of that part which is at 0.3 times the depth of the side (at that location) from the edge in accordance with the recommendations of Section 5.7. The members representing the edge cantilevers are located at the centroid of the cantilevers which is 0. The X direction is arbitrarily chosen to be parallel to the span of the bridge. The grillage members representing the cantilevers and the outermost members in the main part of the deck have been positioned at 0.

the second moment of area is: The torsion constant is: and the area is: At the ends.11. the bridge extends 0.5 m past the centres of the bearings.25 m. this gives a second moment of area of: and a torsion constant of: except for the edge member in the main deck where the torsion constant is: The area of the longitudinal members is also required and is given by: For the transverse members other than those at the ends of the deck. 7.65): The second moments of area and torsion constants for the grillage members are then determined by multiplying these values by the relevant breadths of the members shown in Fig. resulting in a second moment of area of: a torsion constant of: and an area of: . giving member breadths of 1.Page 250 as for a beam: The torsion constant per unit breadth for longitudinal and transverse members is calculated according to equation (5. For the longitudinal members in the main deck.

173 0.44 0.0080 0.0161 0. the second moment of area is: the torsion constant is: Table 7.256 A (m2) 0.Page 251 For the edge cantilevers.60 1.346 0.276 0. For the longitudinal cantilever members this gives a second moment of area of: The torsion constant is based on the breadth excluding the portion outside 0.2 I (m4) Longitudinal members Cantilever Main deck (interior) Main deck (edge) Transverse members Cantilever (interior) Main deck (interior) Cantilever (ends) Main deck (ends) 0.80 0.0042 0. the bridge slab is again assumed to be isotropic and the second moment of area per unit breadth is calculated according to the simple beam formula: and the torsion constant per unit breadth is calculated according to equation (5.32 1.180 0.0121 0. other than those at the ends of the deck.50 .173 J (m4) 0.432 0.44 1.3 times the depth (at that location) which gives: The area of the longitudinal cantilever members is given by: For the transverse cantilever members.0066 0.50 1.216 0.0073 0.1 Upstand grillage member properties for Example 7.65): The second moments of area and torsion constants for the grillage members are once again determined by multiplying these values by the relevant breadths of the members.

In a series of tests. it is important to realise that the moments are not comparable to those in a planar grillage. the authors have found the upstand FE method to be very suitable for modelling bridge decks with wide edge cantilevers. The cantilevers are idealised as finite elements located at the level of the centroids of the actual cantilevers while the main part of the deck is idealised using finite elements located at the centroid of that part.1. The vertical members are given very large properties so that they will not bend or deform. In the authors’ experience. The longitudinal member at the bottom is sufficient and the specification of members at one level only simplifies the determination of their properties and the interpretation of results. The grillage member properties are given in Table 7. the second moment of area could be 22 m 4 (100×0. It benefits from being three-dimensional while being relatively simple to use. The finite-element meshes on each plane are connected by rigid vertical grillage members. a second moment of area and torsion constant of between 100 and 1000 times the largest values in the model is usually appropriate. the member properties are less. Some programs may have the facility to assign ‘rigid’ properties to members. Thus. This is largely due to the well proven ability of finite-elements to model in-plane behaviour. although not essential. The upstand FE model consists of a number of planes of plate finite elements connected together by rigid vertical members. that vertical beam members are used rather than vertical elements. The values chosen are dependent on the computer and software used as excessively large values may result in round-off errors. It is generally more convenient. A useful way of achieving this is to increase the member properties in successive runs until just before the program becomes unstable due to round-off errors. 7. This approach may need to be verified for particular computers and software. it does not suffer from the problems of modelling in-plane behaviour associated with upstand grillages. the member breadths are less than those of the internal members. If this is available then it should be used for the vertical members. Figure 7. However. so once again.432).6 Upstand finite-element modelling Upstand finite-element (FE) modelling is an extension of plane FE modelling in the same way that upstand grillage modelling is an extension of plane grillage modelling. When interpreting the results of an upstand grillage model.1(b)) so that the transverse members on the cantilever side can be given the properties of the cantilever and those on the other side can be given the properties of the main part of the deck. A row of nodes is located at the junction of the edge cantilever and the main part of the deck (Fig. as bending in the upstand model is not about the bridge neutral axis. for this example.Page 252 and the area is: At the ends.12 shows an upstand FE model for a bridge deck with edge cantilevers.216) and the torsion constant 43 m4 (100×0. Most significantly. It is of importance that no longitudinal grillage member be located at the top of the vertical members. . the difference is accounted for by the presence of axial forces which the bridge must be designed to resist. 7.

13 Calculated longitudinal bending stresses at span on top surface of deck The bridge deck of Fig. Similar results were found at midspan and for all other cases considered.14(a) shows the cross-section of a bridge deck with wide edge cantilevers.10. An upstand FE model is required. 1998).8 m and is supported along the entire width of the main part of the deck at each support location. 7.Page 253 Fig. 7. The distributions predicted by the elaborate three-dimensional brick FE model and the plane and upstand grillage models described previously are also shown. Example 7.3: Upstand finite-element model Figure 7.13 shows the longitudinal stresses predicted along the top surface of the deck at of the span in the same format as that used in Fig. . The deck is continuous over two spans of 24. 7. It can be seen that the upstand FE model predicts an almost identical stress distribution to the elaborate three-dimensional brick FE model.9 was analysed by the authors using an upstand FE model (O’Brien and Keogh.12 Upstand finite-element model Fig. The interpretation of results from upstand FE models is not comparable to those from planar FE models (as is the case for upstand and planar grillage models). 7. Figure 7.

only one span is shown in the figure.4 m.2 m below the top of the bridge deck. The three-dimensional brick FE and upstand FE models predict a very similar stress at all locations and the plane FE model is in reasonable agreement. This results in vertical members with a length of 0. Those representing the main part of the deck are located at the centroid of that part which is 0. All of the elements are 1.24 m long (in the span direction). This stress distribution follows the expected pattern with zero stress at the ends.4 m. The elements representing the edge cantilevers are located at the centroid of the cantilevers which is 0. 7. zero stress close to span and maximum tensile stress above the central support. The plane FE model predicts a significantly greater stress at both the mid-span and central support locations. The three-dimensional brick FE and upstand FE models once again predict very similar stress at all locations but the plane FE model is in poor agreement with these. maximum compressive stress close to span. A plane FE model (in accordance with the recommendations of Chapter 5) and a three-dimensional FE model using solid ‘brick’ type elements were also analysed. This model was analysed by the authors under the action of self weight. (b) finite-element mesh Figure 7.6 m from the top of the deck.15(b) illustrates the corresponding distribution along a line 0.3: (a) cross-section (dimensions in mm). Figure 7.2 m wide and 1.14(b) shows a three-dimensional view of a suitable upstand finite-element mesh.15(a) shows the longitudinal stress distribution at the top of the bridge deck along the centreline of the deck as predicted by each of the models. The elements in the main part of the deck are given a depth of 1.2 m and those in the edge cantilevers a depth of 0. This is caused by the inability of the planar model to allow for the rising neutral .Page 254 Fig. Figure 7.14 Upstand finite-element model of Example 7. The main part of the deck and the edge cantilevers are both taken to be isotropic and consequently the only properties associated with the elements (other than their material properties) are their depths. As the model is symmetrical about the central support.6 m in from the edge of the cantilever.

Page 255 Fig. Alternatively this can be viewed as the inability of the planar model to allow for shear lag. (b) 0.6 m in from edge of cantilever axis in the edge cantilever. This example shows the benefits of three-dimensional modelling over planar modelling for bridge decks of this type. Most FE programs only allow the specification of a depth for the finite elements which does not . as the stiffness of each part of the deck is made up of a combination of both of these. 7. 7.15 Longitudinal bending stress at top fibre for bridge of Example 7.3: (a) at centre.1 Upstand finite-element modelling of voided slab bridge decks The three-dimensional nature of upstand FE modelling requires the specification of the correct area for the elements as well as the correct second moment of area.6.

A choice of 20. but causes problems when dealing with voided slabs.35 m long. In other words. this could be done by incorporating additional grillage members into the model with a negative area and zero second moment of area. When considering the longitudinal direction. The deck is simply supported with a 24 m span and is supported continuously across its breadth at each end. The length of the rigid vertical members is equal to the distance between the centroid of the cantilevers and that of the main part of the deck. this will result in an overly stiff model.Page 256 allow the independent specification of area and second moment of area. the depth of the finite elements is determined by equating the second moment of area of the voided slab to that of an equivalent depth of solid slab. As the voids are generally located close to mid-depth of the slab. Modelling of voided slabs by the plane FE method is discussed in Section 6. A solution to this problem is to reduce the area of the elements. quite sensibly.2 m long elements in the longitudinal direction would be appropriate for this model. most computer programs will not allow this. Figure 7. A more feasible alternative is to choose the depth of the finite elements so that they have the correct area and then to add additional grillage members to make up the shortfall in second moment of area. a finite element with a depth chosen by considering the second moment of area of the voided slab will have an excessive area. As stiffness in the upstand FE model is made up of a combination of both the second moment of area and the cross-sectional area of the elements. They should also have zero in-plane second moment of area as the in-plane behaviour is still modelled by the finite elements.4. An upstand FE model is required. the equivalent depth of the elements will generally be quite close to (but smaller than) the actual depth of the voided slab. In this case the vertical members are 0. For the elements in the main part of the deck. In theory.16(b) shows the cross-section of a suitable upstand FE model for this bridge deck. The additional grillage members should have zero area. 1.16(a) shows the cross-section of a voided slab bridge deck with wide edge cantilevers. This is not the case when considering the cross-sectional area which is greatly reduced by the presence of the voids. This is sufficient when dealing with solid slabs. Therefore. The X direction is chosen as the longitudinal direction. Example 7. the presence of the voids does not greatly affect the longitudinal second moment of area of the deck. The second moment of area of this is: and the area is: .2 m wide with one void. each element represents a portion of deck 1.4: Upstand FE model of voided slab Figure 7. Clearly a member with negative area has no physical meaning and.

4 (dimensions in mm): (a) cross-section.879 m and the longitudinal grillage members have second moments of area of 0. (b) section through finite-element model Equating this to an equivalent solid element with the same area gives an equivalent element depth. 7. each finite element in the main part of the deck is replaced by four elements and four grillage members as illustrated in Fig.1): . These elements have the same equivalent depth of 0. of: The second moment of area of this equivalent solid element. deq .093 m 4.Page 257 Fig. is: To incorporate the additional members in the model. The required transverse second moment of area per unit breadth is given by equation (6.17.16 Upstand finite-element model of Example 7. Ieq . The second moment of area of these additional members. 7. is: This gives a shortfall in second moment of area which has to be made up by additional grillage members. .

18 Upstand finite-element model with additional grillage members (half) . 7.17 Replacement of plate element: (a) original element. (b) corresponding combination of elements and grillage members Hence. the required additional second moment of area which is provided by the transverse grillage members is: Fig. 7.Page 258 Fig.

19 Upstand finite-element model of beam and slab bridge: (a) cross-section. The final upstand FE model with grillage members shown as dark lines is illustrated in Fig. provided care is taken to ensure that good similitude exists between the model and the actual structure. Rigid.19(a) shows a beam and slab bridge.2×1.6 m intervals. The horizontal members at different levels are joined by stiff vertical members. the originally proposed 1. The slab can be represented in the model using finite elements located at its centroid of equal depth to it.5 m which is equal to the actual depth of the cantilever. Each beam in this bridge will act compositely with the slab above it and they are normally assumed to bend about their own centroid rather than that of the bridge as a whole. The properties of the remaining parts of the deck are then calculated. (b) section through upstand finite-element model . 7.19(b).6.6×0. this is clearly an approximation as the exact location of the neutral axis will depend on the flange widths and the relative stiffnesses of the members. each about its own centroid. However.6 m intervals to join the meshes on the different planes.2 Upstand finite-element modelling of other bridge types It is possible to extend the principles of upstand FE analysis to types of bridge other than solid and voided slabs. Figure 7.2 elements in the cantilever were replaced with four 0. and are represented by grillage members at the levels of those centroids as illustrated in Fig.Page 259 The edge cantilevers are modelled as finite elements with a depth of 0. 7. an upstand FE analysis can be used to represent the behaviour more accurately than the alternative planar models. Only one-half of the model is shown as it is symmetrical. In such cases where the location of the neutral axis is unclear. or very stiff vertical grillage members are specified at 0.6 elements to give nodes at 0. 7. The elements used for this example only had nodes at the corners with the result that they could only be joined to the vertical members at their corners. This approach has the advantage of simplicity as there is a direct correspondence between each member and Fig. 7.18. Therefore.

Unfortunately.Page 260 Fig. it is usual to uncouple the in-plane and out-of-plane behaviours. to correctly model longitudinal bending there. The calculated moment for each beam member is only applicable to bending about its own centroid. This model.7 Prestress loads in three-dimensional models When analysing for the effects of prestress in bridge decks. The stresses determined . Figure 7. the interpretation of the output can be tedious. has the advantage of automatically allowing for transverse cell distortion as discussed in Section 6. then the total moment will have to be calculated taking account of the axial forces in the beam and the elements and the distance between them. Care should be taken with such a model to ensure that sufficient numbers of elements are provided through the depth of the webs. 7. the number of elements required to achieve this is very large and this. 7. If reinforcement is to resist the stresses in a beam and the adjacent elements.20(b) shows a suitable model based on a variation of the upstand FE analogy. (b) finite-element model a part of the structure. Transverse diaphragms could also be incorporated into this model with ease. These effects are generally dealt with by calculating the equivalent loading due to prestress (Chapter 2) which is often based on an assumed neutral axis location. The in-plane behaviour is governed by the distribution of axial stress in the bridge deck and is often determined by a hand calculation.6. However. 7. combined with the tedium of interpreting the results. often rules out its use. The bridge deck is then analysed to determine the effects of the equivalent loading. as well as dealing with a varying neutral axis. The out-of-plane behaviour is affected by the vertical components of tendon force and by the moments induced by tendon eccentricity.20(a) shows a cellular bridge deck and Fig.20 Plate finite-element model of cellular bridge: (a) original bridge.

The location of the neutral axis is indicated in the figure. The sources of error in a traditional planar model. below mid-depth of the main part of the deck. but it is unknown at this stage.Page 261 from this analysis are combined with the in-plane axial stresses to obtain the overall effect of the applied prestressing forces. In the three-dimensional approach. Figure 7. the location which is applicable to. the prestress forces are applied directly to the model at the correct vertical location by means of stiff vertical grillage members. h. such as the upstand grillage or upstand FE methods. the inability of the planar model to allow for the variation in neutral axis location may cause inaccuracies in the calculated response to equivalent loading. The additional moment is the product of the prestress force and the distance. as the neutral axis location is load dependent. as discussed in previous sections.21 (b) shows the equivalent portion of an upstand FE model. This prestress force has an unknown eccentricity. This method is often simpler to implement as there is no need to uncouple the in-plane and out-of-plane behaviours. Firstly. The equivalence of Figs. self weight may not be applicable to prestressing. It should be mentioned that. there is an additional error. which is also indicated in the figure. There is no uncertainty concerning the location of the neutral axis about which eccentricity of prestress must be calculated. Figure 7.21(a) shows a portion of a bridge deck with an edge cantilever. When using a three-dimensional model. The deck is subjected to a prestress force.21(a) and (c) can be seen by considering the applied moment. are two fold. the independence of the prestress loading from the neutral axis location is retained but the necessity for a large number of vertical members is avoided. . Figure 7. 7. with the equivalent loading calculated in the normal manner. the prestress force can alternatively be applied at the level of the elements along with an additional moment to allow for the difference in level between the true point of application and the element. say. P. The model is subjected to an axial force which generates a moment of: To avoid the necessity of adding a large number of vertical grillage members to the model. the equivalent loading due to prestress can be applied in a three-dimensional manner. the applied moment is: which is equal to the applied moment of the former. In this way.21 (c) shows this alternative model. The eccentricity of this force is once again e but a knowledge of the magnitude of the eccentricity is not necessary. The prestress force is applied directly to the model through a rigid vertical member of length h. It follows that the calculation of moments due to cable eccentricity are not dependent on any assumed neutral axis location. In the latter. at a distance. Many of the complications involved in determining equivalent loads due to prestress can be avoided in this way. However. e. as the magnitude of the equivalent loading is itself dependent on the eccentricity of prestress and is therefore affected by the neutral axis location. h. because they can be related directly to the design without the need to distinguish between primary and secondary effects. There are also advantages to be gained in the interpretation of results.

In particular. upstand FE analyses with equivalent loading calculated in the traditional way (as described in Chapter 2) did not always give accurate results. 7.Page 262 Fig.21 Portion of prestressed concrete deck: (a) original deck. (c) alternative upstand finite-element model The authors have found this direct method of representing the effects of prestress to be the most accurate of many methods tested when compared to results from elaborate threedimensional finite-element analyses with brick type elements. (b) upstand finite-element model with vertical member at point of application of prestress. .

Page 263 Appendix A Reactions and bending moment diagrams due to applied load .

Page 264 .

Page 265 Appendix B Stiffness of structural members and associated bending moment diagrams .

Page 266 .

72×106 6.48×106 0 0 − 32.1) where xi and y i are the co-ordinates of point i and n is the number of co-ordinate points. Table C. point n+1 is defined as equal to point 1. 6. the co-ordinates are taken from the figure starting at the top left corner and specifying only half the section (which will have the same centroid as the full section). The terms of equation (C. of any section can be found from the co-ordinates of the perimeter points using the formula: (C.1 where Top and Bottom refer to the numerator and denominator respectively of the fraction specified in the equation.6.1) are given in Table C.Page 267 Appendix C Location of centroid of section The centroid.1 Evaluation of equation (C.64×10 9 . For the section of Fig.93×10 9 0 0 − 20.60×106 0 0 0. .19×10 9 2.1) xi 0 5500 5500 1500 1200 0 0 yi 1200 1200 0 0 800 1000 1200 (xi −i+1 ) x − 5500 0 4000 300 1200 0 0 4320000 1440000 0 640000 2440000 3640000 4320000 yi +y i+1 2400 1200 0 800 1800 2200 2400 Sum= Top − 23. For the purposes of this calculation.40×106 0 0 0.76×10 9 Bottom − 39.

.Page 268 The y coordinate of the centroid is then: The same answer can be found by dividing the section into rectangles and triangles and summing moments of area about any common point.

the shear force in the top flange will be: Fig.1 Cell with flange and web distortion: (a) assumed distortion. D.Page 269 Appendix D Derivation of shear area for grillage member representing cell with flange and web distortion The transverse shear force half way across the cell will be distributed between the flanges in proportion to their stiffness. (b) segment of cell between points of contraflexure . Hence.

D. Hence the total moment at the top of the web is: The rotation of the web due to this moment is: where h is the bridge depth (centre to centre of flanges) and i w is the web second moment of area per unit breadth.1 for a segment of cell between points of contraflexure.Page 270 where V is the total shear force and i t and ib are the second moments of area per unit breadth of the top and bottom flanges respectively. The total deflection in the top flange results from this rotation plus bending in the flange itself: Similarly the deflection in the bottom flange can be shown to be: The mean deflection is: Equating this to the shear deformation in a grillage member gives: . This force is illustrated in Fig.

7): . this becomes equation (6.Page 271 If the second moments of area per unit breadth are expressed in terms of the flange and web depths .

and Jaeger. (ed. Eng. in Advances in Computational Methods for Simulation. (1997) Serviceability limit state aspects of continuous bridges using precast concrete beams. Rev. Lee.. Deutsche Verlags-Anstalt.M.. Cheung. Civil Eng. London. Gauthier -Villars. 8.L. Rotterdam. London. B. B. I. B.C. Struct. Aesthetics and Design. EC1 (1995) Eurocode 1: Basis of Design and Actions on Structures. (1983) Concrete Bridge Design to BS5400. L. London. Jaeger. Edinburgh.A. Amsterdam.J. (1991) Bridge Deck Behaviour. 2nd edn. E. Construction Press. L. L.Page 272 References AASHTO (1995) AASHTO LFRD Bridge Design Specifications. E. ASCE J.J. Caquot. UK.G. E&FN Spon. 357–66. B. Can. Part 3: Traffic Loads on Bridges. Balkema. (1948) Tables for the Calculation of Passive Pressure. BA42/96 (1996) Departmental Advice Note BA 42/96. 4th edn.V. D. 112(2).S. 2nd edn. Topping).A. R. Brussels. and O’Brien. June 1999.. European Committee for Standardisation. EC7 (1994) Eurocode 7: Geotechnical Design. E. European prestandard ENV 1997–1:1994.L. Bakht.G. (translated from French by M. pp. 185–90. European Committee for Standardisation. Lehane. Clark. 109–35.H. Bec). D. London. L. Active Pressure and Bearing Capacity of Foundations. (1996) Recommendations on the use of a 3-D grillage model for bridge deck analysis. Stuttgart. A. and Neville. A. 8(4). Washington. F. Department of Transport. (1996) Soil-structure interaction analysis for integral bridges. Design of Integral Bridges.J. Dobry. Harlow. (1997) Evaluation by proof testing of a T-beam bridge without drawings. and Sugie. G. J. Struct. B. Ghali. Eng... 75(19). (1994) Bridge Bearings and Expansion Joints. Brussels. Eng. (1986) Dynamic response of arbitrarily shaped foundations. Paris. Lehane. (1984) Bridges.A. Part 1: General Rules. Department of Transport. Civil-Comp Press. Leonhardt. M. and Kersiel. Geotech. Keogh. Hambly. 376–91.. in Proceedings of 12th European Conference on Soil Mechanics and Geotechnical Engineering. European Prestandard ENV 1991–3:1995. Keogh. J. E&FN Spon. Loads for Highway Bridges. . A. (1981) The state of the art in analysis of cellular and voided slab bridges. SI Units. Eng.A.. D. BD37/88 (1988) Departmental Standard BD 37/88. 75(11). (1999) Predicting the restraint to integral bridge deck expansion. Bakht. and O’Brien. 339–44. A. Struct. 201–10. American Association of State Highway and Transportation Officials.. Clark. and Gazetas. London. E&FN Spon. (1997) Structural Analysis: A Unified Classical and Matrix Approach. and Mufti.

March 1997.L. (1968) The reinforcement of slabs in accordance with a pre-determined field of moments.P. Downsview. and Taylor.W. Trinity College. O. Proc. (1973) C&CA/CIRIA Recommendations on the Use of Grillage Analysis for Slab and Pseudo-slab Bridge Decks. and Ng. R. Troitsky.K. West. Springman. McGraw-Hill. J. 28. Wood. S. S.R. and Goodier. D. Cement and Concrete Association. James F. and Keogh. S.C. M.M. M. McGraw-Hill.S. MSc Thesis. Concrete. (1989) The Finite Element Method. Ohio. (1998) Upstand finite element analysis of slab bridges. E. London. (1967) Orthotropic Bridges: Theory and Design.G. Civil Eng. OHBDC (1992) Ontario Highway Bridge Design Code.C. C.J. Canada. Inst.L. O’Brien. R.Page 273 O’Brien. 671–83. D.W. University of Dublin. Ontario. and Cheung. O’Brien. (1964) The finite element method for analysis of elastic isotropic and orthotropic slabs.N. S. February. .J. 1. and Keogh. London. Ministry of Transportation of Ontario.J. Lincoln Arc Welding Foundation. Y. Timoshenko. South Africa. (1994) Pile Design and Construction Practice. Computers and Structures. (1997) The Analysis of Shear Forces in Slab Bridge Decks. (1996) Cyclic Loading of Sand Behind Integral Bridge Abutments. Johannesburg. London. New York.. UK Highways Agency. O’Brien.H.. Dublin.W. Tomlinson. New York. 69. 471–88. 4th edn. Cleveland. R. (1970) Theory of Elasticity. A. Norrish.. pp 233–7.M. TRL Report 146. Zienkiewicz. 4th edn. FIP Symposium. in The Concrete Way to Development. Zienkiewicz. 69–76. E. Vol. 3rd edn.L.G. (1997) The calculation of shear force in prestressed concrete bridge slabs. O. E&FN Spon.

29–30 spherical 30 see also Supports Bending moments due to applied loading 263–4. 169.Page 274 Index Page numbers appearing in bold refer to major entries AASHTO 40 Aesthetics 34–9 Aluminium deck 42 Analysis. introduction to 67–120 Anisotropic 151. 182. 228 Box section 5–7 Bridge bashing see Loading. 18. box Centroid. 185 elastomeric 31 pot 30–1 sliding 17. 265–6 Blister 16 Box culvert 21–3. 28. 269–71 Cellular section see Cross-section. post-tensioning 14 Creep 12. impact Buried joint 32 Cable-stayed 25–6 Cantilever 3 balanced 14–17 Cellular bridge 212. 78. impact Composite 25. 42. 160 Application of planar grillage and finite element methods 200–39 Arch 18–20 Articulation 26–9 Asphaltic plug joint 32–3 Balanced cantilever 14–17 BD37/88 40 Beam and slab bridge 183. 147–50 Cross-section 2–8 box 5–7 older concepts 7− 8 solid rectangular 2–4 T. lightweight 42 Continuous beam/slab 10. 180.4–5 voided rectangular 4 . 229. 260. 231. 228–36 grillage modelling 230–6 three-dimensional finite element model 260 transverse cell distortion 228. location of 267–8 Collision loads see Loading. 147–50 Computer implementation of grillages 179–80 Concrete. 82. 72–4 Box girder 212. 218–28 arching action of slab 228 finite element modelling 225–7 grillage modelling 219–25 transverse behaviour 227–8 upstand finite element model 259 Bearing 29– 31. 13–18 Contraction of integral bridges 128–33 Coupler. 17.

40. dry 127 Diaphragm 10.Page 275 Culvert. 244–6. 260–1 Equivalent loading due to temperature/ thermal effects 47. 89–104 Downstand 203 Drop-in span 24–5 Dry density 127 Durability 1 Dynamic amplification 53 Dynamic effects 52–4 Earth pressure 124–6 Edge cantilever 203–11. 184 recommendations for modelling 182–5 shear flexible 212. 211. 237. 92. 178. 88 Equivalent loads method 67. 252. 95. 228 Differential settlement 9. 262 mesh 189–91 properties of elements 186–9 recommendations for modelling 189–91 similitude with bridge slab 186 slab bridges 185–91 see also Upstand finite element modelling Foundation. expansion FEA see Finite element analysis Finite element analysis (FEA) application of planar 200–39 beam and slab bridge 225–7 brick elements 244. 260–1 Integral bridge 21–3. box 21–3. 220. modelling of 225 see also Upstand grillage modelling Halving joint 24 Impact loading 41. 237 Effective flange width 242–4. 245–7. 51–2. 43–5 Inaccuracy. 217 Grillage accuracy 171 analysis of slabs 169–85 application of moment distribution 111–20 application of planar 200–39 beam and slab bridges 219–25 cellular bridges 230–6 computer implementation 179–80 member properties 173–9 mesh 169–71. 75–7 Differential temperature 47–51. 218–19. 107 Eurocode 40 Expansion of integral bridge see Integral bridge. 180. 230 similitude with bridge slab 171–3 sources of inaccuracy 180–2 U-beams. 246. 252 Edge stiffening 203–11. 54 Imposed traffic loading 40. sources of see Grillage. 41. dynamic Geometrically orthotropic 152. 177–9. 42 Density. 240. 180–2. 90. 121–50 contraction 128–33 cracking over supports 147–50 expansion 137–45 bank seats 142–5 deep abutments 140–2 time-dependent effects 147–50 Interface shear stress 240 Inverted T 3 . sources of inaccuracy Incremental launch 17 In-plane effects 162. 211. 49. 245 Elastomeric bearing 31 Elevations 8–26 Equivalent loading due to prestress 54–66. 253. 185. 88–9. 72–4 Curved bridge 236–9 finite element modelling 238–9 grillage modelling 236–8 Dead loading 40. 189. 187–9. shallow strip 130 Forced vibration 53 Frame bridge 21–3 Frequency see Loading. 184.

mesh. 173. 186–8. 178–9. 41. 217 M-beam 7 Mesh see Finite element analysis. 51–2. 186. 177– 8. 240–2. 43–5 wind 42 Losses. 193 geometrically 152. 211. 252. 42 Partially continuous beam/slab 10–13 Passive earth pressure 124 Pavement 42 Pier 184 Poisson’s ratio 160. prestress 60–3. 211. 43 prestress 42. 173. 188. 211. 217 Parapet 34. 40. 176. 179. 177–9. notional 43 Launch. 186. 176. 200–3 Joint 13. 173–7. 107 Orthotropic 151–2. 52–4 equivalent due to prestress 54–66. 176. 182. 212. 174. 49. 46 impact 41. shear 15 Lane.Page 276 Isotropic 151–2. 54–66. 188 Pot bearing 30–1 Prandtl’s membrane analogy 222 Precast beam 183 Pressure. 42 dynamic 41. 160. 43 dead 40. 188. 188. 236 Movement. 54 normal traffic 43–4 pedestrian 40. 32–3 asphaltic plug 32– 3 buried 32 construction 13 halving 24 nosing 33 Key. 173. 46–51 differential 47. dynamic Neutral axis 203–4. 232. 260–1 equivalent due to thermal effects 47. 259–61 Nosing joint 33 Notional lane 43 . earth 124–6 passive 124 Prestress loading 54–66 loading in three-dimensional models 260–2 losses 60–3. 244–5. 187–9. mesh Modulus secant 127 shear 161. 78–89 traffic 40. 186– 8. 179. 217 materially 152–67. 191–9. 172–3. 178. 220. 89–104 uniform 46. 231 Moment capacity see Wood and Armer equations distribution 67–120 in orthotropic plates 161–7 twisting see 166–7. 45. 181. 177–8. 107 parabolic profile 56–8 qualitative profile design 58–9 tendon 183 Materially orthotropic 152–67. 42 thermal 40. Grillage. accommodation of 26 Natural frequency see Loading. 104–11 rail traffic 45–6 road traffic 43–5 superimposed dead 40. 186. incremental 17 Lightweight concrete 42 Linear transformation 54–8 Loading 40–66 abnormal traffic 44–5 cycle track 40. 88 HA 43–4 HB 44–5 horizontal 40.

from grillage 173 key 15 key deck 8 lag 240–2. 188. 161 strength of concrete 156 in thin plates 167–9 Simply supported beam/slab 9. 232–3 see also Prandtl’s membrane analogy moment distribution 111–20 Traffic loading 40. Shear. 183–5. 24 Skew deck 116–20. 236 U-beam. differential 9. 172–3. 176. 186. 184. 42 Supports 180. behaviour and modelling 151– 99 run-on 23. 125. 244–5 modulus 161. 269–71 close to point support 182. 41. 185 flexible grillage 212. linear 54–8 Twisting moment 166–7. 145–7 Sliding bearing 17. 176. stitching 16 Series of simply supported beams/slabs 9 Settlement. 173–9. grillage modelling of 225 . 191 Steel deck 42 Stiffness of structural members 265–6 Stitching segment 16 Stress in orthotropic plates 159–61 Strip foundation 130 Structural form. 230–1. 182. 231 strain 155–6. close to point support Suspension bridge 26 Symmetry 71 T-section 4–5 Temperature. 252. 209. 122. solid rectangular Remaining area 43 Rigid vertical members 245. 122. 29–30 Soil stiffness 126–8 Soil/structure interaction 41. 220 see also Bearing. 189. 74. 181.Page 277 Pseudo-box construction 7 Push-launch construction 17–18 Rail traffic loading 45–6 Reactions due to applied loading 263–4 Recommendations for finite element analysis 189–91 for grillage modelling 182–5 Rectangular section see Cross-section. thermal Thin plate theory 151–69 Three-dimensional modelling of bridge decks 240–62 Torsion constant 167. 259. 261 Road traffic loading 43–5 Run-on slab 23. 191. 178–9. 184. 186. 180. 89–104 Temperature loading see Loading. 75–7 Settlement trough 146 Shallow strip foundation 130 Shear area 180. 43–5 Transformation. 236–9 finite element modelling 238–9 grillage modelling 236–8 Slab bridge decks. thermal Terms 1 Thermal loading see Loading. 137 stiffness (for soil) 130 supports 180. 230 force. 145– 7 Secant modulus 127 Section see Cross-section Segment. 188. 180 Span-by-span construction 13 Span/depth ratios 36 Spring model (of soil) 133–6. factors affecting 1–2 Superimposed dead loading 40. 246 enhancement 182. differential 47–51. 221–2. 230–1. 178. 40. 191 distortion/deformation 156–7. 191–9.

261 of other bridge types 259–60 of voided slabs 255–9 Upstand grillage modelling 245–52. dynamic Voided slab 4. 253. 211–18. 228 torsional stiffness 214 WIM 43 Wing wall 35 Wood and Armer equations 191–9. 244 Upstand finite element modelling 252–60. 152. 261 Vibration see Loading.Page 278 Uplift 28 Upstand 34. 236 Y-beam 7 . 203.

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd