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

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)

6 Bearings 1.1 Introduction 3.2 Moment distribution viii x 1 1 1 2 8 26 29 32 34 40 40 42 43 46 51 52 54 67 67 67 .2 Dead and superimposed dead loading 2.6 Dynamic effects 2.4 Thermal loading 2.4 Bridge elevations 1.1 Introduction 1.Page v Contents Preface Acknowledgements Chapter 1 Introduction 1.7 Prestress loading Chapter 3 Introduction to bridge analysis 3.2 Factors affecting structural form 1.3 Cross-sections 1.8 Bridge aesthetics Chapter 2 Bridge loading 2.3 Imposed traffic loading 2.5 Articulation 1.5 Impact loading 2.7 Joints 1.1 Introduction 2.

3 Conventional spring model for deck expansion 4.Page vi 3.5 Run-on slab 4.1 Introduction 5.4 Planar finite-element analysis of slab decks 5.3 Grillage analysis of slab decks 5.2 Thin-plate theory 5.3 Effective flange width 7.7 Skew and curved bridge decks Chapter 7 Three-dimensional modelling of bridge decks 7.4 Voided slab bridge decks 6.4 Modelling expansion with an equivalent spring at deck level 4.2 Contraction of bridge deck 4.5 Beam and slab bridges 6.6 Time-dependent effects in composite integral bridges Chapter 5 Slab bridge decks—behaviour and modelling 5.5 Differential temperature effects 3.6 Prestress 3.1 Introduction 4.5 Wood and Armer equations Chapter 6 Application of planar grillage and finite-element methods 6.4 Three-dimensional analysis 7.6 Cellular bridges 6.5 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 .2 Shear lag and neutral axis location 7.7 Application of moment distribution to grillages Chapter 4 Integral bridges 4.2 Simple isotropic slabs 6.1 Introduction 7.3 Edge cantilevers and edge stiffening 6.4 Thermal expansion and contraction 3.1 Introduction 6.3 Differential settlement of supports 3.

6 Upstand finite-element modelling 7.Page vii 7.7 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 .

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

Professor S.H.Perry Civil. 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. although the merits of grillage methods are not ignored. All in all. this must prove the standard work on bridge deck analysis for decades to come.Page ix checking the output of computer analyses. Structural and Environmental Engineering Trinity College Dublin .

in whole or in part. and should be treated as such. typographical or editorial errors may occur. Despite the best efforts of all concerned.Page x Acknowledgements We would like to thank Dr A. some of which are novel and have not been exposed to the rigours of time. The support of Roughan and O’Donovan Consulting Engineers.Ghali most sincerely for major contributions to some of the earlier chapters. The authors of STRAP (ATIR software. 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. arising from information contained in this publication. . 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. Special thanks is due to Joe O’Donovan for providing some of the photographs in the text.Perry and Trinity College Dublin. The stay in Slovenia was greatly enhanced and enriched by Alenka Ž nidarič . Disclaimer This publication presents many advanced techniques. and readers are encouraged to bring errors of substance to our attention. The publisher and authors disclaim any liability. Tel Aviv) and NIKE3D (Lawrence Livermore National Laboratories. A sabbatical stay in Slovenia for the first author made the initial drafting of many chapters possible. including the cover illustration. Ancon CCL are also acknowledged for providing a number of illustrations.H. The material represents the opinions of the authors. where both authors were employed for a time. The initial writing effort was greatly facilitated for both authors through the support of Professor S. is much appreciated. He gave most generously of his time with the sole objective of getting it right. USA) are thanked for the use of their programs. The assistance of Chris Davis and Michael Barron of Mott McDonald with Chapter 2 is gratefully acknowledged. Sound engineering judgement should be the final arbiter in all stages of the design process.

which is perpendicular to it. In some cases. it has been established that a significant portion of the world’s bridges are not performing as they should. However.2 Factors affecting structural form In recent years. and transversely.1 which are commonly used in bridge engineering.Page 1 Chapter 1 Introduction 1. The bridges that were easy to design were usually determinate. The new awareness of the need to design durable bridges has led to dramatic changes of attitude towards bridge design. In this figure. in many others. The deck spans longitudinally. Joints may be present to facilitate expansion or contraction of the deck at the ends or in the interior. bridges are carrying significantly more traffic load than originally intended. There is now a significant move away from bridges that are easy to design towards bridges that will require little maintenance. which is the direction of span. the problem is one of durability—the widespread use of de-icing salt on roads has resulted in the ingress of chlorides into concrete. Problems have also been reported with post-tensioned concrete bridges in which inadequate grouting of the ducts has lead to corrosion of the tendons. 1. . This is often associated with joints that are leaking or with details that have resulted in chloride-contaminated water dripping onto substructures. 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. all parts of the bridge over the bearings are referred to as superstructure while the substructure includes all parts below.1 Introduction A number of terms are illustrated in Fig. The main body of the bridge superstructure is known as the deck and can consist of a main part and cantilevers as illustrated. Bridge decks are frequently supported on bearings which transmit the loads to abutments at the ends or to piers or walls elsewhere. 1.

The structural forms of bridges are closely interlinked with the methods of construction. This immediately limits the structural forms to those that can be constructed in this way. is not a very efficient structural form as the second moment of area of a rectangle is relatively small. simply supported spans and cantilevers. such as a railway yard or a deep valley. 1. The method of construction also influences the distributions of moment and force in a bridge. For example. the construction must be carried out without support from below. 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.g. in some bridges. Various alternative structural bridge forms and methods of construction are presented in the following sections.3. Due to the inefficiency of this structural form. For example.1 Portion of bridge illustrating bridge engineering terms e. when a bridge is to be located over an inaccessible place.3 Cross-sections 1.1 Solid rectangular The solid rectangular section. 1. the stresses . illustrated in Fig. steel beams carry the self weight of the deck while composite steel and in-situ concrete carry the imposed traffic loading. Such a bridge is generally constructed of reinforced concrete (particularly for the shorter spans) or prestressed concrete. However.Page 2 Fig.2. The methods of construction in turn are often dictated by the particular conditions on site. 1. The move now is towards bridges which are highly indeterminate and which have few joints or bearings.

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

3 T-section The T-section is commonly used for spans in the range 20–40 m as an alternative to voidedslab construction. Guidance is given on the analysis of this type of deck in Chapter 6. 1. 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. For the span range of 20–30 m. The complete rectangular section is available to carry other loading. are more expensive in terms of shuttering . 1.5.4 Voided slab section with cantilevers (initially wet) in-situ concrete. In-situ T-section decks. 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. 1.Page 4 Fig. However. The shuttering costs are also less than for in-situ concrete T-sections which are described below. 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.3. in some cases. As a result. illustrated in Fig. These decks can be constructed from ordinary reinforced concrete or can be post-tensioned. This problem is not so much one of steel straps failing as of grooves being cut in the polystyrene by the straps. 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. solid rectangular sections become increasingly less costeffective due to their low second moment of area to weight ratio.4. particularly when the designer wishes to minimise the structural depth.3. the preferred solution. 1. However. it is common practice in some countries to use in-situ concrete with polystyrene ‘voids’ as illustrated in Fig. particularly that designed to resist transverse bending. Hence it is. 1. Including voids in a bridge deck increases the cost for a given structural depth because it adds to the complexity of the reinforcement. it reduces considerably the self weight and the area of concrete to be prestressed without significantly affecting the second moment of area. the section tends to be deeper for a given span.2 Voided rectangular For spans in excess of about 20 m. Regardless of the diameter-to-depth ratio. the voids must be accounted for when considering the design to resist transverse bending.

it becomes economical to use ‘cellular’ or ‘box’ sections as illustrated in Fig. as illustrated in Fig. 1. 1.3. Over less accessible places. (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.Page 5 Fig. spans transversely between the beams while acting as flanges to the beams longitudinally. supported on permanent shuttering.5 In-situ concrete T-sections: (a) single web such as might be used for a pedestrian bridge.7. These have a higher second moment of area .6. 1. 1. precast concrete or steel forms of T-section. These consist of pretensioned prestressed concrete or steel beams placed in position along the length of the span.4 Box sections For spans in excess of 40 m. are favoured. An in-situ concrete slab.

when the bridge is in service. However. 1. .7 Box sections: (a) single cell. to inspect the inside of the void.6 T-sections: (a) composite steel and concrete. 1.Page 6 Fig. (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. (b) composite precast Y-beam and in-situ concrete Fig.

10. if water leaks into the voids. These were constructed of M-beams with insitu concrete near the bottom to form a void. The bottom in-situ concrete was reinforced transversely by threading bars through holes cast in the M-beams. M-sections were often placed side by side with the bottom flanges within millimetres of each other. In the past. in the past. For example.9 Precast M-beam . However. The analysis of this type of bridge is similar to that of any T-section bridge. 1.5 Older concepts Many variations of the above structural forms have been used in the past and are evident in existing bridge stocks.6(b)) used more commonly today. 1. 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. 1. Fig. The section is more efficient than a T-section as more concrete is located away from the centroid.Page 7 Fig.9). 1.8. 1.3. It was also common practice in the past to build bridges of ‘pseudo-box’ construction as illustrated in Fig.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. corrosion problems can result and. it was common practice to construct Tsection decks using precast ‘M-beams’ (Fig. These have wider bottom flanges than the precast ‘Y-beams’ (Fig. 1. 1.

Fig. 1. The latter ‘shear keys’ are assumed to be capable of transferring shear force but not transverse bending moment as they have no transverse reinforcement. they are no longer popular due to concerns about the durability of the in-situ joints. This consists of precast concrete slab strips joined using longitudinal strips of insitu concrete.10 Pseudo-box section due to the nature of this structural form.11(a). Shear key decks were popular for railway bridge construction as the railway line could be reopened even before the in-situ concrete was placed.e.11(b). However. 1. 1. Thus the transverse deformation is assumed to be as illustrated in Fig. The structural behaviour of the pseudo-box section is similar to that of a small multi-cellular box section. Many of the alternative bridge elevations and their methods of construction are described in the following sections. illustrated in Fig. rotation is assumed to occur at the joints between precast units. (b) assumed transverse deformation . assessment and repair is difficult. i. 1.Page 8 Fig.4 Bridge elevations The cross-sections described above can be used in many different forms of bridge.11 Shear-key deck: (a) section through small portion of deck. 1. Another form of construction used widely in the past is the ‘shear key’ deck.

The implication of this is that the bridge deck needs to be correspondingly deeper.1 Simply supported beam/slab The simplest form of bridge is the single-span beam or slab which is simply supported at its ends. 1. This form. as illustrated in Fig. For example.2 Series of simply supported beams/slabs When a bridge crossing is too wide for an economical single span. the concrete pours are moderately sized. The cross-section is often solid rectangular but can be of any of the forms presented above. It is particularly favoured on poor soils where differential settlements of supports are anticipated. as illustrated in Fig.12 Simply supported beam or slab Fig. 1. However. Continuous beams/slabs. Like single-span bridges.14 Continuous beam or slab . 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. is widely used when the bridge crosses a minor road or small river. It can be seen that the maximum moment in the simply supported case is significantly greater (about 25%) than that in the continuous case. The simply supported bridge is relatively simple to analyse and to construct but is disadvantaged by having bearings and joints at both ends. 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.4.4.15. In addition.Page 9 1. 1.14.13.12. if constructed using in-situ concrete. 1. illustrated in Fig. one after the other. this form is relatively simple to analyse and construct. 1. 1.13 Series of simply supported beam/slabs Fig. the span is relatively small and multiple spans are infeasible and/or unnecessary. Fig. there is less disruption to any traffic that may be below as only one span needs to be closed at any one time. it is possible to construct what is in effect a series of simply supported bridges. 1. In such cases. It also has the advantage that. 1. the bending moment diagrams due to a uniformly distributed loading of intensity ω(kN/m) are illustrated in Fig. have significantly fewer joints and bearings.

Two forms of partially continuous bridge are possible. In-situ concrete is then used to make the finished bridge continuous over intermediate joints. 1.17. For bridges of moderate total length. Further. continuity over intermediate supports is provided only by the slab. it is possible to use precast concrete or steel beams to construct a partially continuous bridge. In the form illustrated in Fig. the concrete can be poured in-situ in one pour. Thus the in-situ slab alone is required to resist the complete hogging moment at the intermediate supports. as the total bridge length becomes large. The slab at the support in this form of construction is particularly flexible and tends to attract a relatively low bending moment.16. This tends to increase cost as the construction becomes more of a batch process than a continuous one. 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.4 Partially continuous beam/slab When support from below during construction is expensive or infeasible.4. 1. This completely removes the need for any joints.6. 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. 1. . This is possible due to the fact that members of low structural stiffness (second moment of area) tend to attract low bending moment. illustrated in Fig. (b) one three-span continuous beam with span lengths l 1. Precast concrete or steel beams are placed initially in a series of simply supported spans. 1.3 Continuous beam/slab with full propping during construction As stated above. However.4. Elsewhere the cross-section is similar to that illustrated in Fig. the amount of concrete that needs to be cast in one pour can become excessive.15 Bending moment diagrams due to uniform loading of intensity ω (a) three simply : supported spans of length l. 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. In the alternative form of partially continuous bridge.Page 10 Fig.

Page 11 Fig. 1. (b) plan view from below Fig. 1.17 Partially continuous bridge with continuity provided only by the slab at intermediate supports .16 Partially continuous bridge with full-depth diaphragm at intermediate supports: (a) elevation.

Unfortunately. This introduces a complexity into the analysis compounded by a great difficulty in making accurate predictions of creep effects. In partially continuous bridges.19(a) for a two-span bridge.19 Typical distribution of bending moment in two-span partially-continuous bridge: (a) bending moment due to self weight. 1.Page 12 Fig. The total bending moment diagram will be a combination of that due to self weight and other loading. By the time the imposed traffic loading is applied.19(b) than Fig. 1. due to creep. the bridge is continuous and the resulting bending moment diagram is as illustrated in Fig.19(b).17 as the main bridge beams rotate at their ends. 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. 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.18 Joint detail at intermediate support of partially-continuous bridge of the type illustrated in Fig.18.19(a). (b) bending moment due to loading applied after bridge has been made continuous . self weight continues to cause deformation in the bridge after it has been made continuous. 1. Fig. 1. the joint must move longitudinally to accommodate this rotation as illustrated in Fig. 1. 1.

The method is also of a continuous rather than a batch form as the precast beams can be constructed at a steady pace. Fig. can be used to achieve continuity of prestressing across construction joints. This is achieved using temporary formwork supported on the bridge piers as illustrated in Fig.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.20(a). while intermediate joints have been removed. resulting in minimum disruption to any existing traffic passing under the bridge. where bending moments and shear forces are relatively small. such as illustrated in Fig. two bearings are necessary at each intermediate support. It has been said that joints should be provided every 100 m at least. Proprietary post-tensioning couplers. this figure is constantly being revised upwards as the problems of bridge joints in service receive ever more attention. 1.20(b). However. In particularly long continuous beam/slabs. The joint may sometimes be located at the quarter-span position as illustrated in Fig. an intermediate joint may become necessary to relieve stresses due to expansion/contraction. 1. 1. in-situ construction. A significant disadvantage is that. 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. In this form of construction. 1. starting even before work has commenced on site. Particularly for the form illustrated in Fig. intermediate bearings are still present with their associated maintenance implications.20 Temporary support system for span-by-span construction: (a) joint over intermediate pier. can be a viable option. one span at a time.5 Continuous beam/slab—span-by-span construction For construction of particularly long bridges when access from below is expensive or infeasible.17. 1.4.21. (b) joint at quarter span . Construction on site is fast. 1.

Page 14 Fig. The segments are supported by a ‘travelling form’ connected to the existing bridge (Fig. 1.6 Continuous beam/slab—balanced cantilever construction When the area under a bridge is inaccessible and spans are in excess of about 40 m. in anticipation of the need to post-tension future segments at later stages of construction.22(b)) until such time as they can be permanently posttensioned into place as illustrated in Fig. 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).22(a)). An intermediate pier is cast first and a small part of the bridge deck (Fig. precast beams are not generally available to span the complete length at once. 1. it is often economical to construct bridges by the balanced cantilever method. in the case of . Segments of deck are then added to the base segment. At spans of this length. This form of bridge is generally made of post-tensioned prestressed concrete. However. 1. 1. Ducts are placed in all segments when they are first cast. one on each side. Segments can be cast in-situ or precast. either alternately on opposing sides or simultaneously in pairs. 1. 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. The sequence of construction is illustrated in Fig.22(c).4. 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.21 Post-tensioning coupler to transmit prestress forces across a construction joint (photograph courtesy of Ancon CCL) 1.22.

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

23 Casting of stitching segment . After the casting of the stitching segments and completion of construction. 1. the moment due to self weight during construction is such as illustrated in Fig. 1. (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. (c) sectional elevation showing tendon. 1.Page 16 Fig. 1. 1.25(b).24. Post-tensioning tendons are placed in the bottom flange and webs by means of ‘blisters’. illustrated in Fig.23. a ‘stitching segment’ is cast to make the bridge continuous as illustrated in Fig.25(a). 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.22 Balanced cantilever construction: (a) elevation of base segment and pier. such as illustrated in Fig. When cantilevers meet at mid-span. Thus. the bridge forms a continuous beam and the imposed service loading generates a distribution of moment. 1. The bending moment in a balanced cantilever bridge is entirely hogging while the bridge remains in the form of a cantilever. to resist the sagging moment that will exist in the finished structure due to applied traffic loading. (b) temporary support of segments.

4.7 Continuous beam/slab—push-launch construction For spans in excess of about 60 m. a long segment is cast behind the bridge abutment as illustrated in Fig.25(a) towards a form approaching that illustrated in Fig. 1. 1. This results from creep deformations which are still taking place after the bridge has been made continuous. 1.24 Blisters and tendon in the bottom flange (sectional elevation) sagging. This process is continued until the complete bridge has been constructed behind the abutment and pushed into place. 1. temporary sliding bearings are used to minimise friction forces.25 Distributions of bending moment in balanced cantilever bridge: (a) due to self weight during construction. In pushlaunch construction. ‘incremental-launch’ or ‘push-launch’ becomes a viable alternative to balanced cantilever as a method of construction. 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.26(a). it is frequently the most economical alternative for construction over deep valleys when propping from below is expensive.26(b)).25(b). 1. Nevertheless.Page 17 Fig. Fig. 1. 1. When the deck is being pushed over intermediate supports. 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. (b) due to imposed loading after completion of construction . The analysis of balanced cantilever bridges is complicated by a creep effect similar to that for partially continuous beams.

27(b).8 Arch bridges For larger spans (in excess of about 50 m).26 Push-launch construction: (a) casting of the first segment. must be designed for the creep effect and are subject to the associated complexity and uncertainty in design. If this is not the case. A significant disadvantage stems from the distribution of bending moment generated temporarily during construction. an arch is still a possibility if it is tied such as illustrated in Fig. Some engineers design bridges in an arch form for aesthetic reasons but articulate the bridge like a . arches generate a significant horizontal thrust.28(b). Parts of the deck must be designed for significant hog moment during construction as illustrated in Fig. 1. the horizontal thrust is taken by the tie. 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). All of the bridge is constructed in the same place which is easily accessible to construction personnel and plant. This can be achieved if the bridge is located on a particularly sound foundation (such as rock). as illustrated in Fig. In a tied arch.Page 18 Fig. and are only a viable solution if it can be accommodated. This doubling of cantilever length has the effect of quadrupling the moment due to self weight during construction. 1. 1. like those designed for balanced cantilever construction.27(a). 1. Bridges designed for pushlaunch construction. 1.28(a). (b) pushing of the partially constructed bridge over first span The method has a considerable advantage of access. 1.4. However. the arch form is particularly effective. These same parts may be subjected to sag moment in the completed bridge as illustrated in Fig.

(b) tied arch with deck at base of arch . 1.28 Arch bridges: (a) conventional form with deck over the arch. (b) due to imposed loading after completion of construction Fig.Page 19 Fig. 1.27 Distributions of bending moment in push-launch bridge: (a) due to self weight during construction.

1.29 Simply supported beam bridge in the shape of an arch Fig. Other advantages of arches are that they are aesthetically pleasing in the right environment. An additional major advantage is that arches require no bearings as it is possible to cast the deck integrally into the substructures.Page 20 simply supported beam. beneficial and assists in the resistance of stresses due to imposed loading. an arch can readily span such a distance in one clear span creating an openness under the bridge that would not otherwise be possible. other than the problem of accommodating the horizontal thrust. As can be seen in Fig. as illustrated in Fig. 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. 1. 1.30. is the fact that the curved form results in shuttering which is more expensive than would otherwise be the case. considerable temporary propping is required to support the structure during construction. while a continuous beam/slab crossing a 60 m motorway would normally be divided into two or four spans. 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. 1. as the bridge has no means by which to resist the horizontal thrust. If arches are located over inaccessible areas. This is perfectly feasible but. Concrete arches are particularly effective as concrete is very strong in compression. While traditional masonry arches were designed to be completely in compression. it behaves structurally as a simply supported beam. The principal disadvantage of concrete arches. Thus the self weight generates a distribution of stress which is.30 Deflected shape of arch subjected to thermal contraction . Fig.29. in fact. the structural depth can be very small and large clear spans can readily be accommodated. For example. 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.

as can be seen from the examples of Fig.31 Frame/box culvert bridges: (a) box culvert. A further complexity in the analysis of frame bridges is that. Continuous slab bridges on the other hand.31. Given the great upsurge of interest in maintenance and Fig. until recently. the structural behaviour is three-dimensional. are more effective at resisting applied vertical loading than simply supported or continuous beams/slabs. However. 1. there is the usual complexity and uncertainty associated with such calculations. This is because the maximum bending moment tends to be less. it was not considered feasible to design frame bridges of any great length (about 20 m was considered maximum). can be analysed using two-dimensional models. There are no joints or bearings as the deck is integral with the piers and abutments. such as illustrated in Fig.9 Frame/box culvert (integral bridge) Frame or box bridges. 1.Page 21 1. unless the transverse width is relatively small.33. 1.32. If some of this shortening is due to creep or shrinkage. The minimal maintenance requirement of frame/box culvert bridges is their greatest advantage. (b) three-span frame . accommodating movements due to temperature changes or creep/shrinkage can be a problem and.4. 1. The effects of deck shortening relative to the supports is to induce bending in the whole frame as illustrated in Fig.

This problem has been overcome by allowing the supports to slide as illustrated in Fig. both relating to longitudinal movements. 1. There are two implications for longer frame-type bridges. 1.Page 22 Fig. If the supports are fully fixed against translation.34.33 Effect of thermal contraction of deck in frame bridge: (a) deflected shape. It is now considered that bridges of this type of 100 m and longer are possible. deck movements in such bridges will generate enormous stresses. If the bridge is supported . (b) continuous beams. (b) distribution of bending moment durability in recent years.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. Ever longer spans are being achieved. (c) frames/box culverts Fig. 1.

Thus. 1. 1. 1. There are a number of variations of this form of construction which are considered further in Chapter 4. the axes of the piles are orientated so as to provide minimum resistance to longitudinal movement. a joint is required to facilitate translational movements.Page 23 Fig. at the ends of the run-on slabs.34 Sliding support and run-on slab in frame bridge on piles.35. Such a joint is remote from the main bridge structure and. A precast variation of the frame/box culvert bridge has become particularly popular in recent years. if it does leak. Precast pretensioned concrete beams have a good record of durability and do not suffer from the problems associated with grouted post-tensioning tendons. 1. These can be used in combination with in-situ concrete to form a frame bridge as illustrated in Fig.35 Composite precast and in-situ concrete frame bridge . To overcome this. will not lead to deterioration of the bridge itself.6(b). engineers specify ‘run-on’ slabs as illustrated in the figure which span over loose fill that is intended to allow the abutments to move. The second implication of longer frame bridges is that the bridge moves relative to the surrounding ground. The run-on slab can rotate relative to the bridge deck but there is no relative translation. Crosssections are typically of the form illustrated in Fig. Fig.

The side spans are simply supported with cantilevers to which point loads from the drop-in span are applied at their ends. it can readily be constructed over inaccessible areas. The joint and bearing detail at the ends of the drop-in span in this form of construction is particularly important.Page 24 1.36. 1. A typical example is illustrated in Fig. two ‘halving joints’. as illustrated in Fig. However. In older bridges of the type. some older bridges were constructed of precast concrete with drop-in spans. The drop-in span.37(a).37(b). 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. Fig. which provides access.36 Beam bridge with drop-in span Fig.10 Beams/slabs with drop-in span For ease of construction and of analysis. 1. were used.37 Halving joint at end of drop-in span: (a) traditional detail (no access). This bridge is determinate as the central ‘drop-in’ part is simply supported. This detail is particularly problematic as access to inspect or replace the bearings is extremely difficult.4. Thus. it is still popular in some countries for pedestrian bridges over roads. 1. 1. in particular. can be placed in position very quickly over a road or railway requiring a minimum closure time. is illustrated in Fig. A more convenient alternative. (b) alternative detail with access . 1.

• There are very high tensile and shear stresses at a point where the structural depth is relatively small. The maximum main span achievable is increasing all the time. 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.38 Reinforcement detail in halving joint However. The cables are only required to take tension and they provide support to the deck at frequent intervals. the joints tend to leak. The concept of cable-stayed bridges is simple. 1. The deck can then be designed as a continuous beam with spring supports.38. illustrated in Fig. For spans of moderate length. halving joints frequently cause difficulty for a number of reasons: • Even for pedestrian bridges in which de-icing salts are not used. 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 current limit is of the order of 1000 m. the cross-sections of cable-stayed bridges are often composite with steel beams and concrete slabs. steel box section decks are used to reduce the bridge self weight.4. 1. which promotes corrosion of the halving joint reinforcement. • As can be seen in Fig.11 Cable-stayed bridges Cable-stayed construction. . 1. 1.Page 25 Fig. It is also generally necessary to carry out a dynamic analysis for bridges of such slenderness.39. for the longest spans. regardless of which alternative is chosen. 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.

Fig. it is a necessary consideration for those which do. 1. shrinkage and thermal effects while keeping the structure stable. For these reasons.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. Thus.4. up to about 2000 m span.40. shrinkage and thermal movement are still very real and no one form of construction is the best for all situations.Page 26 Fig. cablestayed construction is generally favoured except for the very longest bridges. it is sometimes difficult to cater for the horizontal forces generated at the ends of the cables. 1.12 Suspension bridges The very longest bridges in the world. 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. are of the suspension type illustrated in Fig. It lends itself readily to staged construction with the cables being added as required to support successively placed segments of the deck. In suspension bridges. Horizontal forces are caused by braking and traction of vehicles. wind and accidental impact forces from errant vehicles. While the present trend is to provide ever fewer joints and bearings. the bridge must have the capacity to resist some relatively small forces while accommodating movements. Further. 1. 1. While this clearly does not apply to bridges without joints or bearings.40 Suspension bridge . The articulation of a bridge is the scheme for accommodating movements due to creep. the main cables are in catenary and the deck hangs from them applying a substantially uniform loading.39 Cable-stayed bridge The economy of the cable-stayed form stems from its ease of construction over inaccessible places. 1. As for balanced cantilever bridges. segments are placed successively on alternate sides of the pylon. the problems of creep.

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

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

1. 1. it is better to provide two bearings only. not only is there a risk of deterioration in the central bearing but. as illustrated in Fig.Page 29 Fig. 1. Only a limited number of the more commonly used types are described here.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.6.43 Uplift of bearings due to traffic loading Fig. Further details of these and others are given by Lee (1994).44. If this is not possible.44 Uplift of bearing due to transverse bending caused by differential thermal effects loading in right bridges if the span lengths are significantly different. the two outer bearings must be designed to resist all of the load which renders the central bearing redundant. 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.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. as it is not taking any load.43. However. even with no skew and typical span lengths. If this occurs. differential thermal effects can cause transverse bending which can result in uplift as illustrated in Fig. . 1. 1. 1.

it is common in such systems for the lubricant to be squeezed out after a number of years. They are frequently used for motorway bridges of moderate span. guides are used such as illustrated in Fig. The elastomer effectively acts as a retained fluid and facilitates some rotation while preventing translation. such as illustrated in Fig. 1.Page 30 Fig. the sliding surfaces are spherical and allow rotation. Whether or not sliding bearings are lubricated. resulting in a reduced coefficient of friction. They are also used in combination with plane sliding surfaces to provide free sliding . rotation is facilitated through some other mechanism and plane sliding surfaces are used which allow translation only. However.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.6. Some bearings are lubricated. Sliding bearings offer a frictional resistance to movement which is approximately proportional to the vertical force.46. at which time the coefficient returns to the unlubricated value. this form is also referred to as the spherical bearing. Thus. it has been suggested that they be treated as wearing parts that eventually need to be replaced. They can take many forms and are often used in combination with other forms of bearing. 1. 1. When translation is to be allowed in one direction only. Sliding bearings today generally consist of a stainless steel plate sliding on a PTFE-coated surface. In other cases. consist of a metal cylinder containing an elastomer to which the force is applied by means of a metal piston.45. 1. In some combinations. pot bearings by themselves are commonly used at the point of fixity.2 Pot bearings Pot bearings. namely polytetrafluoroethylene (PTFE).

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

joints buried beneath road surfacing are possible and. 1. However. A typical arrangement is illustrated in Fig. 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. 1. as illustrated in Fig. it is difficult to find a suitable material which carries the impact loading due to traffic across the gap while facilitating the necessary movement.48 Buried joint (after Lee (1994)) . can result in a minimum maintenance solution. 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.Page 32 1.7. 1. This form has been successfully used for movements of up to 40 mm and is inexpensive to install or replace.48. Joints are notoriously problematic. movements will always occur with the result that joints will always be needed. 1.7 Joints While bearings in bridges can frequently be eliminated.2 Asphaltic plug joint The asphaltic plug joint is similar to the buried joint in that the gap is protected by road surfacing. for larger gaps.49.7. if designed well. and frequently leak. Even in integral construction.1 Buried joint For movements of less than 10–20 mm. Fig. particularly in road bridges. However. The material used to span the joint is important. the movement must be accommodated at the end of the run-on slab.

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. It can accommodate movements of similar magnitude to the asphaltic plug joint but has a reputation for frequent failure and leakage. illustrated in Fig.50.Page 33 Fig.7. 1. 1. is no longer favoured in many countries.3 Nosing joint Very popular in the 1960s and 1970s.50 Nosing joint (after Lee (1994)) . the nosing joint. Fig.49 Asphaltic plug joint (after Lee (1994)) 1. 1.

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. Certain bridge proportions in particular. 1. Fig. Some aspects of aesthetics are common to most bridges.52).8 Bridge aesthetics The art of bridge aesthetics is a subjective one with each designer having his/her own strongly held opinions. 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. 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.Page 34 1. The sun tends to shine directly on upstands while the main deck tends to remain in shadow (Fig. Further details on these and longer-span bridge aesthetics can be found in the excellent book on the subject by Leonhardt (1984). 1.53. However.51 Continuity of upstand and parapet (photograph courtesy of Roughan and O’Donovan Consulting Engineers. 1. 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. The aesthetics of the more common shorter-span bridges are considered in this section. This effect can be useful. particularly if the designer wishes to draw attention away from an excessively deep main deck. particularly on what constitutes an aesthetically displeasing bridge.51. there is generally some common ground. look better than others and attention to this can substantially improve the appearance of the structure. Dublin) . 1.

In this example. 1. The abutment wing walls also play an important role as can be seen in the example of Fig. 1. the shape of the opening has a significant influence on the aesthetics. Dublin) Fig.53 Section through upstand 1. 1. the shape of the opening is square (span equals height) and the abutment wing walls are large triangular .1 Single-span beam/slab/frame bridges of constant depth For very short-span bridges or culverts.8.Page 35 Fig.52 Shading of main deck relative to upstand (photograph courtesy of Roughan and O’Donovan Consulting Engineers.54.

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

This can be convenient as the principal obstruction to be spanned is often in the central part of the bridge. It can be seen that the upstand appears too thin and/or the deck too deep.Page 37 Fig. (b) deep deck and slender upstand and 10 are illustrated in Fig. For aesthetic reasons. this proportioning also tends to bring the relative dimensions of the rectangular openings closer.2 Multiple spans The relative span lengths in multi-span bridges have a significant effect on the appearance. where people and traffic are close to the structure which is large relative to their size.56(a). The bridge illustrated is probably typical with a main span/upstand depth . 1. it is common practice in three-span construction to have the centre span greater than the side spans. Leonhardt points out that scale is important as well as proportion.54(c) for upstand and main deck respectively. This is illustrated in Fig. span/upstand depth ratios of 20 and a span/main deck depth ratio of 10 is often recommended. typically by 25–35% as illustrated in Fig. 1. When the ground level is lower at the centre.54(d) and (a). For rectangular openings with less pronounced wing walls. The heavier looking alternative illustrated in Fig. Typical ratios are illustrated in Fig.55(b) has ratios of 60 and 10. while ratios of 10 and 5 are illustrated in Fig. For a 2×1 rectangular opening with wing walls of similar size. 1.8. 1. a much more slender deck is desirable.) A structure with similar proportions looks much better in Fig. 1. which has a good aesthetic effect. a parapet wall is integral with the upstand making it look deeper than necessary. 1. 1.55(a) with a span/upstand depth ratio of 40 and a span/main deck depth ratio of 20.55 Rectangular opening with small wing walls: (a) slender deck and deep upstand.56(b) as it is smaller and is more likely to be viewed from a distance. 1. as illustrated in the figure. an even more slender deck is favoured. (In this structure. 1.57.

57 Three-span bridge with good proportions Fig. (b) curved alignment achieved using two curves of differing radius. 1.58 Variable depth bridges: (a) straight haunches.Page 38 Fig.56 The influence of scale on appearance: (a) large structure near the viewer looks heavy. (c) curved haunches . (b) small structure remote from the viewer looks better than in(a) Fig. 1. 1.

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

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

Differential settlement of supports can induce significant bending in continuous beam or slab bridges. it is recommended that a combined model of the bridge structure and the supporting soil be used to determine the stresses induced by settlement.5 2. Superimposed dead 3.6 – 2. In pedestrian bridges. 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. Similarly high levels of impact loading are in use in many European national standards.3 – 2. Thermal 6. The loading due to impact from collisions with errant vehicles can be quite significant for some bridge elements.2 2.4 – 2. Vibration is generally only significant in particularly slender bridges.e. vehicles and the bridge itself Effect of prestress on indeterminate bridges Section 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. i.Page 41 Table 2. Imposed traffic 4. Impact 8. Wind 10. Pedestrian and cycle track 5. excludes some thermal and wind effects but includes a higher traffic loading. For example. The draft Eurocode on Geotechnical Design. in AASHTO and in the draft Eurocode. The load specified in the UK has increased dramatically in recent years. Differential settlement 7. recommends that the process of soil/structure interaction be taken into consideration for accurate analysis of problems of this type. The calculation is complicated by the use of different factors of safety and the specification of different design limits for the different combinations. Dead 2. EC7 (1994). this usually only includes pedestrian bridges and long-span road and rail bridges. No geotechnical guidance is given in either BD37/88 or AASHTO on how bridges should be analysed to determine the effect of this phenomenon. as will be demonstrated in Chapter 3. Dynamic effects 9. it should be ensured that the natural frequency of the bridge is not close to that of walking or jogging pedestrians. where the natural frequency of the bridge is at a level which can be excited by traffic or wind. In practice. .7 must also be considered. 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.2 2.1 Summary of bridge loads Load type 1.

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

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.3. Bridge traffic loading is applied to notional lanes which are independent of the actual lanes delineated on the road. Alternatively. The AASHTO code specifies a traffic lane loading which consists of a knife-edge load plus a uniformly distributed lane loading.3 Imposed traffic loading Bridge traffic can be vehicular. 2.1 Imposed loading due to road traffic While some truck-weighing campaigns have been carried out in the past. 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. In the Eurocode. more importantly. 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 imposed traffic loading specified by AASHTO is considerably less onerous than that specified by both BD37/88 and the Eurocode. ‘normal’ and ‘abnormal’. rail or pedestrian/cycle or indeed any combination of these. The AASHTO code also specifies notional lanes of fixed width. there has been a scarcity of good unbiased data on road traffic loading until recent years. Bridge traffic loading is often governed by trucks whose weights are substantially in excess of the legal maximum. There are two problems with this as a means of collecting statistics on truck weights. is known as the ‘remaining area’. 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. sampling was carried out by taking trucks from the traffic stream and weighing them statically on weighbridges. its importance should not be underestimated. Normal traffic loading or Highway A (HA) represents an extreme . BD37/88 and the draft Eurocode specify two types of traffic loading. When a structural element supports both pedestrian and traffic loading.g. In the first place. The outstanding road width between kerbs. a reduced intensity is allowed by some codes to reflect the reduced probability of both traffic and pedestrian loading reaching extreme values simultaneously. the quantity of data collected is relatively small but. In the past. 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. Bridge codes commonly specify a basic intensity for pedestrian loading (e. While pedestrian/cycle traffic loading on bridges is not difficult to calculate. Vehicular and rail traffic are considered in subsections below. 5 kN/m 2 in the draft Eurocode and the British standard and 4 kN/m 2 in the American code). In general. each 3 m wide. Most codes allow a reduction for long footpaths. a truck of specified dimensions and axle weights must be considered.Page 43 2. the road width is divided into a number of notional lanes. after removing these lanes.

6. 400 kN and 200 kN for Lanes 1. 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. respectively. 2 and 3. While there are a number of factors which can vary between road classes and between countries. 14. a number of possibilities must be considered. or 29. In the British standard. Illustrated in Fig. 24. 2.1(b). 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.2. Eurocode normal loading consists of uniform loading and a tandem of four wheels in each lane as illustrated in Fig.Page 44 combination of overloaded trucks of normal dimensions. for example. It is scaled in gross ‘units’ of 40 kN so that a minor road bridge can be designed.6. The four wheels of the tandems together weigh 600 kN. 2. Combinations of normal traffic and an abnormal vehicle must be considered in bridge design. the vehicle is known as the Highway B or HB vehicle.5 kN/m 2 elsewhere. In addition.1 ‘Normal’ road traffic loading: (a) Eurocode normal loading. 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.1(a). there is uniform loading in the remaining area.6 m. 2. This consists of an exceptionally heavy vehicle of the type which is only allowed to travel under licence from the road/bridge authority. it could be a chance occurrence of two overloaded moving trucks near the centre of a short bridge at the same time. the standard combination is a load intensity of 9 kN/m2 in Lane No. For bridges with many notional lanes.6. to take 25 units (a 1000 kN vehicle) while a highway bridge can be designed for 45 units (a 1800 kN vehicle). Different countries have different classes of abnormal vehicle for which bridges must be designed. 19. On the other hand. 1 and 2. there can be a considerable dynamic component of truck loading which is deemed to be included in the specified normal load. While there are exceptions. ‘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. (b) British standard HA loading .6. 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 BD37/88. 2. Particularly on roads with rough surfaces. only one abnormal vehicle is specified but it may have a length of 9. The possibility of abnormal or Highway B (HB) loading must also be considered in British and Eurocode designs.

2. bridges throughout a rail network are generally designed for the same normal load model. 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.2 British standard abnormal (HB) vehicle consisting of 16 wheel loads of F=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. known as Railway Upper (RU).3. there are some aspects of traffic loading that are specific to railway bridges which must be considered. the normal load model. This factor is a function of the permissible train speed and of the natural frequency of the bridge. the Eurocode provides for an alternative abnormal load model. Railway tracks on grade are generally laid on ballast. less onerous load models can be applied. an additional vertical dynamic load is induced by the change from the relatively ‘soft’ ballast support to the relatively hard track slab. This follows from the fact that the train can generally be assumed to remain on the tracks. A standard light rail load model. The other disadvantage to the use of track slabs depends on the method used to maintain and replace ballast. There are two disadvantages to the use of track slabs. Normal load is placed throughout the remainder of the lane and in the other lanes. is similar in format. less stringent models have been used for the design of bridges on some light rail networks. The standard Eurocode normal load model consists of four vertical point loads at 1. However.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. Railway Lower (RL). is specified in the British standard. This effect can be minimised by incorporating transition zones at the ends of the bridge with ballast of reducing depth. When used. On bridges.Page 45 Fig. 2. However. In addition. tracks can be laid on a concrete ‘track slab’ or the bridge can be designed to carry ballast and the track laid on this. However. In BD37/88.6 m intervals of magnitude 250 kN each and uniform loading of intensity 80 kN/m both before and after them. If this is done using automatic . 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.

While it is possible in road bridges for all vehicles to brake at once. 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.4 Thermal loading There are two thermal effects which can induce stresses in bridges. In concrete bridges. such as in an arch or a frame bridge. if it is restrained from doing so. this can generate significant axial force. i. 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. beam and slab. it is statistically much less likely. Longitudinal horizontal loading in bridges can affect the design of bearings and can generate bending moment in substructures and throughout frame bridges.Page 46 equipment. Another aspect of loading specific to railway bridges is the rocking effect. Different figures are specified for 12 ‘cold’ climates. 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. high early temperatures can result from the hydration of cement. This can generate torsion in the bridge. It is important in bridge construction to establish a baseline for the calculation of uniform temperature effects. the temperature of the bridge at the time of construction. etc. The British Standard and the draft Eurocode specify no baseline. If the top of a beam heats up relative to the bottom. The American approach is much simpler. bending moment and shear force are generated. 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. 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. The AASHTO code specifies a baseline temperature equal to the mean ambient in the day preceding completion of the bridge. those made from precast concrete or steel will have temperatures closer to ambient during construction. In ‘moderate’ climates. bending moment and shear. it tends to bend. 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.). Unlike in-situ concrete bridges. The first is a uniform temperature change which results in an axial expansion or contraction. particularly for concrete with high cement contents. 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. Completion of the structural form could be the process of setting the bearings or the making of a frame bridge integral. 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. . 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.e. 2. The second effect is that due to differential changes in temperature.

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

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

and Fig.5(a).15)= E Force 1.4(d).5 Beam and slab bridge subject to differential temperature: (a) cross-section.4(b) and (c) from 2.150α E 0.890α E 0. 2.2 Calculation of force Block a b c d Details 3α (2. The coefficient of thermal expansion is α the modulus of elasticity is E. illustrated in Fig.11/35 000=±32×10− .220α E . 2. it is free to rotate and there is in fact no such stress. (b) imposed distribution of temperature Table 2. Instead. 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.4(a). It is found simply by subtracting Figs. 2.080α E 1. bending moment and residual stresses are required due to the differential temperature increases shown in Fig.4×0. Example 2. 2. 2. a strain distribution is generated which varies linearly in the range 6 ±1.Page 49 as illustrated in Fig. 2.4(c).100α E Total force= 3.2: Differential temperature II For the beam and slab bridge illustrated in Fig. the equivalent axial force.5(b). As the beam is simply supported.

0.Page 50 Fig.22α corresponds to an axial tension of 3. each of area.262α E − 0.506α E − 0.6 and divided into rectangular and triangular blocks.22α E E/0. (b) corresponding imposed stress distribution By summing moments of area.012α E 0.718α E .2. 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. The bridge is split into two halves. 2.062α E Total moment= − 0.60α E. the centroid of the bridge is found to be.718α corresponds to E stresses (positive tension) of: Table 2. below the 2 top fibre. The total force of 3.3 Calculation of moment Block a b c d Details Moment − 0. The temperature distribution is converted into a stress distribution in Fig.6 Division of section into blocks: (a) cross-section. area and distance from the centroid as outlined in Table 2. 0. 2.3 (positive sag).70= 4. Similarly moment is calculated as the sum of products of stress. The total moment of − 0.70 m and second moment of area.064 86 m 4.

travelling at a velocity. (d) residual stress distribution Hence the applied stress distribution can be resolved as illustrated in Fig. The kinetic energy of the sphere is: (2. 2.7. However. Δ generates a strain energy of: .2) Fig. To overcome the resulting complications. An undeformable sphere of mass.8 is considered first. the collision of a vehicle with a bridge is highly non-linear. bending and residual components: (a) total distribution.1) On impact. (2. 2. which causes a deflection. 2. collides with a spring of stiffness. The residual distribution is found by subtracting the distributions of Figs. 2. (c) bending component.7 Resolution of stress distribution into axial. (b) axial component. codes of practice often greatly simplify the procedure by specifying equivalent static forces. K. v. 2. 2.7(a). this is converted into strain energy in the spring.8 Impact of undeformable sphere with spring . m. A static force.5 Impact loading Most bridge analysis is based on static linear elastic principles.Page 51 Fig. 2. This section considers the basis on which these forces are derived. The simple case illustrated in Fig. Peq .7(b) and (c) from the applied distribution of Fig.

the situation is simplified by treating the vehicle as undeformable and the structural element as a spring.4) Substituting for Δin equation (2. because only the top of the vehicle is likely to impact on the bridge.3) For a spring of stiffness. will generate the equivalent force given by equation (2. Similar equivalent static loadings are specified in the AASHTO standard and in BD37/88. there is a possibility that trucks passing underneath will collide with the bridge deck. 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.6) While this is a very simple case. the simply supported beam bridge of Fig.5) Substituting for E k in this equation gives the equivalent force in terms of mass and velocity: (2. K. a substantial reduction factor applies. It is not necessary. In dynamics. it can be used as a basis for determining equivalent static forces. in the draft Eurocode. a force P eq generates a deflection: (2.5) on the outer surface of a structural element. The mechanics of a collision between a vehicle and a structure are quite complex. For these reasons. to consider collision of trains with bridge decks overhead.9(a) could be represented by the lumped mass . This is frequently done by ‘lumping’ the distributed mass of a bridge at a finite number of nodes. the draft Eurocode specifies that the impact force due to a truck be applied at a specified height above the road surface. E k. However. For example. a small difference in the impact location or the impact angle can result in a substantial change in the effect. 2. An impact force is also specified for a derailed train colliding with a pier.3) gives an alternative expression for P eq: (2.Page 52 Hence the equivalent static force is: (2.6 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. Further. It follows from these assumptions that a vehicle with kinetic energy. 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. On bridges over road carriageways. On bridge piers. 2. Such a possibility can be investigated by means of a dynamic analysis.

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

vibrating on their own tyres and suspensions. the applied loading is a truck or trucks of considerable mass. For example. it is often convenient to treat it as a loading for analysis purposes. Such a loading can readily be specified in a computer model and the maximum distribution of stress determined. (2. equivalent loadings can be found for individual tendons. Finally.7. it is often necessary to analyse to determine the degree to which prestressing of one member affects others. 2.7 Prestress loading While prestress is not in fact a loading as much as a means of resisting load. even for simply supported slab or beam-and-slab bridges. design for the impact of vehicles colliding with bridges was discussed. As an alternative. Like temperature.Page 54 applied loading and the bridge is taken into account. this can be approximated as: . there are concentrated . 2.7) It also follows from the small angle that the horizontal force is P cosθ P. 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). In such cases. 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.1 Equivalent loads and linear transformation The equivalent loading due to prestress can generally be found by simple equilibrium of forces. 2. The combined effect of a number of tendons can then be found by simply combining the loadings. the draft Eurocode specifies an equivalent static force. For a qualitative understanding of the effects of prestress.5. For such cases. the concept of linear transformation is also introduced. In this section. methods will be given for the calculation of their magnitudes. θis generally small. the code allows for the carrying out of a dynamic analysis. Such a method is only necessary in the case of indeterminate bridges. prestress can be handled using the method of equivalent loads. for the externally prestressed bridge illustrated in Fig. equilibrium of vertical forces gives an upward force at B of: As the angle. as the ≈ forces are eccentric to the centroid at the ends. Furthermore.11(a). In the case of road traffic. the trucks are moving so the location of their masses are changing with time. Whether the bridge consists of beams or a slab. However. Examples of analysis using equivalent prestress loads are given in Chapter 3. In Section 2.

2. 2.11 Prestressed concrete beam with external post-tensioning: (a) elevation showing tendon. (b) equivalent loading .12 Segment of parabolically profiled tendon: (a) elevation. 2. It can be shown that the equivalent loading due to prestress is always self-equilibrating.Page 55 Fig. A parabolically profiled prestressing tendon generates a uniform loading which again can be quantified using equilibrium of vertical forces. (b) equivalent loading due to prestress moments there of magnitude (Pcosθ 2 ≈ 2.11(b). A small segment Fig. Hence the total equivalent loading due to )e Pe prestress is as illustrated in Fig.

2. The intensity of uniform loading on this segment is: (2. At point 1. Example 2.9) where x1 is the X coordinate at point 1. 2.11) The equivalent loads on the segment are illustrated in Fig.12(b).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) where s is referred to as the sag in the tendon over length l as indicated in the figure.12(a). This force is upwards when the slope is positive. .8) As the angles are small: (2. It is required to determine the equivalent loading due to prestress.3: Parabolic profile The beam illustrated in Fig.10) where F 2 is downwards when the slope is positive. 2.Page 56 of such a profile is illustrated in Fig.

13 Beam with parabolic tendon profile: (a) elevation. the slope is negative and the force is downwards of magnitude P(−B +e A+4s)/l. 2.Page 57 Fig. However. in this case. e The slope at B is calculated similarly: .13) As θ is small: A For a positive slope. (b) equivalent loading due to prestress Differentiating equation (2.12) gives: (2. the equivalent point load at A would be upwards and of magnitude P(eB− A− e 4s)/l.

2. they do not significantly affect the distribution of bending moment induced by prestress.4: Qualitative profile design A prestressed concrete slab bridge is to be reinforced with 10 post-tensioned tendons.13): (2.14). 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. Pe. This increase in tendon eccentricity. A more appropriate revision is illustrated in Fig. In the structure of Fig. . the equivalent point loads are as illustrated in Fig. This has the effect of increasing the tendon sag which increases the intensity of equivalent uniform loading.14) This too is illustrated in the figure. in an indeterminate structure. The intensity of uniform loading is given by equation (2. This phenomenon is particularly useful for understanding the effect of prestressing in continuous beams with profiles that vary parabolically in each span.14(a). in fact. Thus. the equivalent uniform loading due to prestress is a function only of the sag and is. illustrated in Fig. Hence.3 illustrates the fact that the intensity of equivalent uniform loading due to a parabolic tendon profile is independent of the end eccentricities. stress at the top fibre can be increased by moving the prestressing tendon upwards to increase the eccentricity locally.13(b).14(b). does little to increase the compressive stress at the top fibre at that point. 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. It is required to determine an amendment to the profile to increase the stress at this point without increasing the prestress force. 2. Example 2. unaffected by eccentricity at the ends of the span. unchanged. increases the (sagging) moment due to prestress. as illustrated in 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. which increases the compressive stress at the top fibre. In a determinate structure. As these forces are at or near supports. As was seen above. A profile such as that illustrated in Fig. The preliminary profile for the tendons. increasing the eccentricity locally at B without changing the sags. This is because the eccentricity at B has been increased without increasing the tendon sag in the spans. 2. the response of a structure to such changes is not so readily predictable. e.14(c) where the profile is lowered in AB and BC while maintaining its position at the support points. the change only results in adjustments to the equivalent point loads at A and B and to the equivalent loading near B.Page 58 As B is on the right-hand side. However. e A and eB while keeping the sag. 2. 2. 2.14.13(a) can be adjusted by changing the end eccentricities. s. Example 2.

This can be seen in the following example.11). 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. For the first parabola: . (b) raising of profile at B by linear transformation. (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. half of which is illustrated in Fig.Page 59 Fig. The intensities of loading are found from equation (2.14 Adjustment of tendon profile: (a) original profile. Example 2. 2.5: Tendon with constant prestress force A three-span bridge is post-tensioned using a five-parabola symmetrical profile.15(a).

from which the force is found to be: All of the equivalent loads due to prestress are illustrated in Fig. Verifying that these forces are in equilibrium can be a useful check on the computations. prestress forces are not constant through the length of bridges because of friction losses. 2. 2. 2. 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. This is illustrated in Fig. Differentiating the equation for the parabola gives the slope. 2.Page 60 Fig.15 Tendon profile for Example 2. (b) equivalent loading due to prestress Similarly. Note that in selecting the profile. it has been ensured that the parabolas are tangent to one another at the points where they meet.7.16(a) .2 Prestress losses In practical post-tensioned construction.5: (a) partial elevation showing segments of parabola. This is necessary to ensure that the tendon does not generate concentrated forces at these points.15(b).

It is required to determine the equivalent loading due to prestress taking account of the loss of force. The resulting loading is illustrated in Fig.9)–(2.5.16(b). A useful method of checking the equivalent loads is to apply them in the analysis of a determinate beam.5 is subject to friction losses which result in the prestress forces presented in Fig.6 that this equivalent loading satisfies equilibrium of forces and moments.Page 61 Fig. . (b) equivalent loading (Pav=(P1+P2 )/2) where the forces at points 1 and 2 are different.16 Equivalent loading due to varying prestress force: (a) segment of beam and tendon. 2. a sensible approach to the derivation of equivalent prestress loading is to start by substituting the average prestress force for P in equations (2. However.6: Tendon with varying prestress force The post-tensioning tendon of Example 2. The bridge is post-tensioned from both ends with the result that the prestressing forces vary symmetrically about the centre. The eccentricities given in this figure have been calculated from the equations for each parabola given in Example 2.17. In such a case. It will be seen in Example 2.11). The use of equivalent loads which do not satisfy equilibrium can result in significant errors in the calculated distribution of prestress moment. Therefore. Example 2. the difference between prestress forces at adjacent points is generally not very large. the moment due to the equivalent loading should be equal to the product of prestress force and eccentricity at all points. 2. 2.

x=0.16(b). In segment AB. the slope is − 0. point loads must be applied at the end of each segment in accordance with Fig. the equation for the parabola is: At A. 2.Page 62 Fig.1322 and the upward force is: .5 but using average prestress forces.17 Tendon profile showing varying prestress force (in kN) and eccentricity (in m) With reference to Example 2. 2. the equivalent intensities of uniform loading are: In addition.

. the prestress forces are resolved parallel and perpendicular to the centroid and the eccentricity is measured in a direction perpendicular to it. A segment of beam with a curved centroid is illustrated in Fig.3 Non-prismatic bridges The eccentricity of a prestressing tendon is measured relative to the section centroid. The resulting equivalent loading is illustrated in Fig. together with the other equivalent uniform loads.7. It can be verified that the forces and moments on each segment are in equilibrium. the location of this centroid varies along the length of the bridge. 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. 2.18(b). In such a beam. The forces and moments at the ends of each segment are summed and the result is illustrated in Fig. 2.Page 63 Fig. 2. At B.19(a). 2.19(b) where s is distance along the centroid.18 Equivalent loading due to prestress: (a) loading on each segment. 2. This clearly affects the eccentricity and hence the moment due to prestress. 2. (b) total the minus sign indicating that the force is actually downwards. In nonprismatic bridge decks.18(a). in Fig.

Page 64 Fig. (b) equivalent loading Example 2.3. 2. 2.19 Equivalent loading due to variation in location of centroid: (a)segment of beam and tendon.9: Equivalent loading due to change in geometry The beam illustrated in Fig. The beam is divided into just two segments.: . It is required to determine the equivalent loading due to prestress. 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). i. It is prestressed with a tendon following a single parabolic profile from A to C. it is defined by an equation of the same form as equation (2.12). the centroid changes depth linearly between A and B and between B and C. The definition of the parabola is independent of the section geometry. AB and BC. In addition.e.20 has a non-prismatic section. With reference to Example 2.

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. 2. for BC. Hence for segment AB: Similarly for segment BC. the derivatives are: . the eccentricity is given by: Differentiating the equation for segment AB gives: Similarly.Page 65 Fig.

Page 66 Fig.97P and 0. The forces are combined in Fig.21(a).21(b).91P respectively. 2. for both segments: The average values for prestress force in segments AB and BC are 0.21 Equivalent loading: (a) loading on each segment. The resulting equivalent loading due to prestress is illustrated for each segment in Fig. . where P is the jacking force. (b) total Differentiating again gives. 2. 2.

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

These are presented in the left-hand column of Table 3.1 and are illustrated using the example presented in the right-hand column.Page 68 The analysis procedure consists of four steps. 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): . Table 3. The fixities are numbered and the direction of each is defined.1 Moment distribution General Example Step 1: All members of the structure are isolated from one another by applying a number of fixities.

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: . Appendix B gives the BMDs for a wide range of such displacements.e. i.25l)=6.2EI/l.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. 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. boxed below. The total discontinuity at B is 3EI/l+4EI/(1.

The resulting BMD is: The discontinuity at C is now (0. For this particular example. In this step. This is performed simply by adding or subtracting the normalised bending moment diagrams. to get an exact answer.006wl2 (the BMD is subtracted. Example The moment just left of B in the fixed BMD (1) of 0. factored by the discontinuities. Hence. often characterised by discontinuities in the BMD. scaled in each case by the appropriate discontinuity. . The correction of the discontinuity at B had the effect of increasing the discontinuity at C. the first iteration has resulted in a BMD which is sufficiently accurate for most practical purposes. This is the final solution.130wl 2) by 0. must be repeated until no discontinuity remains.125)wl2=0. the correction of the discontinuity at C had the effect of reintroducing a discontinuity at B. 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.125wl 2 is less than that just right of B (0.005wl 2. factored by 0. This is removed by subtracting the normalised BMD corresponding to rotation at C (3). there is generally a lack of equilibrium of bending moment at the fixing points. as adding it would increase the discontinuity). such discontinuities are successively removed by applying ‘rotations’.131–0. factored by 1.006wl 2. This discontinuity is removed by adding the normalised BMD corresponding to rotation at B (2). Similarly. the process of adding normalised BMDs.Page 70 General Step 3: In the fixed BMD (Step 1).005wl2 .

The arrows indicate the directions of positive rotation for Step 2. no iteration is required for this example and the BMD of Fig.1(d) (boxed).1(c) and the normalised version.005wl2.2 is. Hence. (c) BMD due to unit rotation. 3. 3. The resulting BMD (from ) is illustrated in Fig. The fixed bending moment diagram (BMD) (Step 1) is unaffected by the symmetric system of fixities but. Fig. As will be demonstrated in this example. scaled by 0. in Step 2. 0. 3. it is possible to isolate members from each other by the simultaneous application of a pair or pairs of equal and opposite fixities.Page 71 Example 3.005wl 2.1(a). 3. The discontinuities at B and C in the fixed BMD are. (d) normalised BMD . as before.2. Step 3 consists of removing these two discontinuities (simultaneously) by adding the BMD of Fig. 3. The beam is fixed simultaneously at B and C as illustrated in Fig. this time using symmetry. as illustrated in Fig. 3. The beam of Table 3.1 is analysed again. two equal and opposite rotations must be applied simultaneously at B and C.1(b). (b) unit rotation simultaneously at B and C. 3. which gives a unit discontinuity of moment at B and C. in fact.1(d). The resulting BMD is illustrated in Fig. As there are no further discontinuities.1: Continuous beam using symmetry Concepts of symmetry can be used to great effect when analysing by moment distribution. is illustrated in Fig. exact. 3.1 Moment distribution using symmetry: (a) symmetrical system of fixities.

3. This is corrected by applying the BMD of Fig. from . 3. this becomes the BMD of Fig. 3. 3.4(e). This is corrected by adding the BMD of Fig. 3.2: Box culvert The application of moment distribution to a two-dimensional frame type of structure is demonstrated using the box culvert illustrated in Fig. 3. 3.3. For simplicity.2 Final BMD for three-span beam Example 3. The normalised BMD due to rotation at C and D is found similarly and is as illustrated in Fig. B. 3.4(f). When normalised.4(f).4(e). 3. 3. factored by that amount. factored by Pl/8. C. Step 2: Applying unit rotation simultaneously at A and B (Fig. Step 3: The discontinuity at A and B in the fixed BMD (Fig. Fig. as illustrated in Fig.5(b). 3. Step 1: The members are isolated by applying fixities at A. Symmetry is exploited by simultaneously fixing A and B and simultaneously fixing C and D as shown.4(c)) results in the BMD illustrated in Fig. The resulting BMD is illustrated in Fig. this culvert is assumed to be supported at two discrete points under the walls and to have constant flexural rigidity throughout.4(Pl/8).4(a).5(a). 3.Page 72 Fig.4(b)) is Pl/8.3 Box culvert example . and D as illustrated in Fig. The fixed BMD is.4(d). 3. 3. The discontinuity now present at C and D is 0.4(b). which gives the BMD of Fig.

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

A more realistic finite-element (FE) model taking account of these effects and assuming typical soil properties is illustrated in Fig.5(d) is deemed to be the final solution. 3. Earth pressure on a structure of this type generates an additional distribution of moment. the BMD of Fig. 3. This box culvert of Example 3.6(a) and the resulting BMD in Fig.4(e).6 Finite-element model of box culvert and surrounding soil: (a) finite-element mesh. A higher hogging moment (0. 3. 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. The interaction of bridges with the surrounding soil is considered further in Chapter 4. factored by this amount. 3.5(c). Finally. Adding the BMD of Fig. 3.2 was assumed to be supported at two discrete points.5(d). gives the BMD of Fig. is added to give the BMD of Fig. factored by 0.825(Pl/8)) is found with a corresponding reduced sagging moment. 3.6(b).Page 74 Step 4: The correction at C and D has reintroduced a discontinuity at A and B of 0.16(Pl/8).064(Pl/8). (b) resulting BMD .4(f). A more typical situation would be that of continuous support from granular material throughout the length of the base and side walls. Fig. 3. 3.

the fixed BMD is as illustrated in Fig. To accurately analyse for this effect. Clearly soil deforms under the vertical forces applied through bridge piers and abutments. is to assume that a foundation support settles by a specified amount. 3. many structural engineers treat the soil as a spring or a series of springs in the numerical model. However.8(a). (b) imposed support settlement . distributions of bending moment and shear are induced in the deck.8.9(a) results in the BMD illustrated in Fig.7 Three-span beam example: (a) geometry. Thus. 3. Example 3. an alternative approach. Similarly. EI. Hence.7 is subjected to a settlement at B of Δrelative to the other supports. Step 1: Referring to (4th and 5th BMDs). the beam is fixed as illustrated in Fig. a rotation at C results in the normalised BMD of Fig. 3. The resulting BMD is required given that the beam has uniform flexural rigidity. 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. When normalised. frequently adopted by bridge engineers. this becomes the BMD of Fig.9(c). If the deformation is not uniform. the structure and the surrounding soil may be represented using non-linear computer models.3 Differential settlement of supports There is considerable research and development activity currently taking place in the field of soil/structure interaction. The system of fixities cannot be symmetrical as the ‘loading’ is not symmetrical.5EI∆ 2.9(d). Fig. 3. 3. 3. The . as the effect is often not very significant. 3. the discontinuity at C is of the /l same magnitude. 3. The discontinuity of moment at B is 1.Page 75 3. Step 2: Applying a unit rotation as illustrated in Fig. By coincidence.3: Differential settlement by moment distribution The continuous beam illustrated in Fig. Δ relative to the others and to determine the effects of this on the structure. following example serves to demonstrate the effect of a differential settlement on a continuous beam bridge.9(b).

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

10(b). l. which are typical of differential settlement: 1.e.9(d) factored by this amount to give the BMD of Fig. 3. (c) after second corrections at1 and 2 Step 3: The discontinuity at B in Fig. This is corrected by adding the BMD of Fig. Differential settlement has the effect of generating sagging moment at the support which settles. Step 4: One further iteration gives the BMD of Fig. /l factored by this amount. (b) after correction at 2. divided by the square of the span length. /l 3. 3.8(b) is 1. I. 3.10(a). The moment at the support which settles is proportional to the second moment of area. Adding the BMD of Fig. As the second moment of area is proportional to the cube of .Page 77 Fig. 3. It is interesting to note two additional things about the final BMD illustrated in Fig. This is important as supports in continuous beams are generally subjected to hogging moment and are often not designed to resist significant sag. 3.10(c) which is deemed to be of sufficient accuracy. gives the BMD of Fig. It is usual to size a bridge by selecting a depth which is proportional to span length (i.8 EI∆ 2.9(c). k).5 EI∆ 2.10 BMD after successive corrections: (a) after correction at1. depth=l/k for some constant.10(c). 3. This correction at B has the effect of increasing the discontinuity at C to 1. 3.

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

30). (b) fixed axial force diagram . where a beam is partially restrained. 1. It also happens in frame bridges where the piers offer some resistance to expansion or contraction of the deck. 3. Fig. 3.Page 79 The most common case requiring analysis is the one in between the two extreme cases described above. 3.13 First step in analysis of frame: (a) fixing system. Example 3.12 Frame subjected to axial change in temperature: (a) original geometry.4: Restrained axial expansion by moment distribution For the bridge illustrated in Fig. (b) deformed shape after expansion of deck Fig. shear force and axial force diagrams due to an increase in deck temperature of Δ T. it is required to find the bending moment. This happens for example in arch bridges where contraction is accommodated through bending in the arch (Fig.12(a).

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

14(d). However. 3.14(d). (d) bending moment diagram . The lack of force equilibrium in this diagram corresponds to the moment discontinuity in the BMDs of the usual moment distribution problems. (While a rotational fixity at this point is also possible. (b) axial force diagram. is illustrated in Fig. 3.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. Fig.14(a). corresponding to unit discontinuity of force at B.13(b). the stress in AB is α T)E. 4th case) of 3EI/h3 giving a total required force at B of 1503EI/h3. 3.) There is no bending moment or shear force in the fixed structure. where α the coefficient of thermal expansion and E is the elastic modulus. The required external force at B can be seen in the free-body diagram of Fig. 3. a force is required to bend BD (. Hence the axial force diagram is as (∆ (∆ illustrated in Fig.14(b) and (c). The normalised version of Fig. 3. (c) shear force diagram.13(a). 3. 3. 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 associated axial force and shear force diagrams are illustrated in Figs.14(e). 3.15 Results of analysis: (a) free body diagram with restored equilibrium at B. In addition. The (∆ is corresponding force is α T)E(area)=6000α T)EI/h2. such a fixity is not necessary to isolate the members in this case.

5: Thermal contraction in frame bridge by moment distribution The frame structure illustrated in Fig. As a result.14(e) factored by 6000α T)EI/h2. as for differential settlement. 3.002×6000α T)EI/h2 to Fig.15(a).13(b). The sustained stresses generated by the subsequent contraction of the concrete as it cools can be relieved substantially by creep.15(d). The final axial force and shear force diagrams are illustrated in Figs. 3.e. 3. However. 3. if sustained in a concrete structure. Most noteworthy is the effect of the relative values of deck area and pier second moment of area. thermal contraction or expansion induces bending moment as well as axial force and shear. by adding an axial tension in AB of 0.16 Integral frame of Example 3. i. 3. As the shear force across the pin at B is 12α T)EI/h2. the moments and forces due to changes in temperature are proportional to the elastic modulus.15(b) and (c). (∆ As there is no further force discontinuity.5 . This means that such stresses. 3. Substantial temperature changes occur on a short-term basis during which the effects of creep do not have a significant ameliorating effect. Example 3.16 is integral having no internal bearings or joints. in-situ concrete bridges generate significant quantities of heat while setting and consequently have their initial set when the concrete is warm. The resulting distribution of bending 6 moment is required given that the coefficient of thermal expansion is 12×10− . Fig. this is the final free body diagram. 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. 3.998×6000α T)EI/h 2 and a shear (∆ (∆ force in BC of 0. may be relieved by the effect of creep. 3. It is also of interest to note that. The relative 2 flexural rigidities are given on the figure and the area of the deck is 500I0/l . 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.Page 82 Step 3: There is a lack of force equilibrium in the fixed structure at B (Fig. (∆ There are some points of interest about axial temperature effects apparent from this simple example. 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. The result is illustrated in Fig.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. Hence the rise in temperature results in a lot of strain and in very little stress in the deck.

Applying a unit translation in Direction 2 (while preventing rotation) requires the moments and forces illustrated in Fig. is as illustrated in Fig. Step 2: Applying a unit rotation in Direction 1 (Fig. 3. Due to symmetry. there is a lack of force equilibrium at A and C which. However. However. As the rotation is applied while fixing against translation.18(b). In the fixed structure. when divided by 7. 3. 3. Hence.18(d). 3. no axial forces are generated in the members.17(a)) requires the moments and forces illustrated in Fig. (b) fixed axial force diagram.12EI 0/l . there is a lack of force equilibrium at A and C as illustrated in Fig. No distribution of bending moment is present in the fixed structure but the axial force diagram is as illustrated in Fig.17(a). 3. As there is axial force in ABC but no corresponding shear force in AD or CF.19(a) and generates the BMD illustrated in . 3. one translational and one rotational as illustrated in Fig. as points A and C will tend to rotate as well as translate.18(c).2EI0 /l. 3.18(a) and generates the BMD illustrated in Fig.Page 83 Fig. 3. The normalised version is found by dividing by 7. (c) free body diagrams showing shear and axial forces in fixed frame Step 1: Due to symmetry.2EI0/l and is illustrated in Fig. 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 0.17(b).17 First step in analysis of frame: (a) system of fixities. there is a shear force just below A and C which is unmatched by an axial force in AB or BC. the fixities at A and C are taken to be equal and opposite as illustrated. 3.17(c). there is no tendency for point B to rotate and this point can be considered fixed without applying a fixity. two fixities are needed at each. 3.

The normalised lack of joint equilibrium is illustrated in Fig. 3. the joint forces of Fig. as there is no moment induced in the fixed structure.17(c) is corrected by factoring Fig. a distribution of axial force is generated which is illustrated in Fig. In addition. In addition.Page 84 Fig. Normalising with respect to this value gives Figs. 3.20(a) and (b). 3.20(c) . 3. 3. (d) normalised shear and axial forces associated with rotation Fig. adding it to a BMD of zero. It can be seen in Fig.19(d) that there is a shear force just below A which is not matched by the axial force to its right.19(c). 3. 3. (b) BMD associated with unit rotation. (c) normalised BMD associated with rotation. The discontinuity or lack of equilibrium at A is 538.20(c).12EI0/l 2 and.4EI 0/l3. 3.18 Effect of rotation at A and C: (a) moments and forces required to induce unit rotation.19(b).20(a) by 0. Step 3: The lack of force equilibrium in the fixed structure illustrated in Fig. 3.

The results are illustrated in Fig.22. 3.21. 3. This is corrected by scaling Figs.Page 85 must be factored by 0.17(c). (c) axial force diagram associated with unit translation. Step 4: The removal of the moment discontinuity reintroduces a lack of force equilibrium which is evident in Fig.00214EI0/l and adding them to Figs. 3. 3. Figures 3. There is a discontinuity in the BMD (or lack of moment equilibrium) at A and C evident in Fig. 3. 3. 3.00285EI0/l2. 3. 0.21(a).22(b) of magnitude.21(a) and (b) respectively.20(a) and (c) are Fig.19 Effect of translation at A and C: (a) forces and moments required to induce unit translation.18(c) and (d) by 0.12EI 0/l2 and added to those of Fig. The results are illustrated in Fig. (d) free body diagram at joint A showing lack of equilibrium . (b) BMD associated with unit translation.

(b) normalised axial force diagram.20 Normalised effect of translation at A and C: (a) normalised BMD. (c) free body diagram showing unit discontinuity of forces Fig.Page 86 Fig. (b) corrected free body diagram . 3.21 Effect of correcting for lack of force equilibrium: (a) corrected BMD. 3.

Page 87 Fig. . Figure 3.23 Results of analysis for effects of thermal contraction: (a) BMD.23(a) is therefore adopted as the final BMD.18(c) and (d). 3. 3.23 where the lack of force equilibrium is deemed to be sufficiently small.22 Effect of correcting for discontinuity in BMD: (a) corrected BMD.22(a) and (b). 3. (b) corrected free body diagram Fig. This leads to Fig. 3. 3. (b) free body diagram scaled by this amount and added to Figs. The resulting moment discontinuity is corrected by factoring and adding Figs.

a more important effect of the moment connection is the bending moment induced in the deck by thermal movement. 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. 3.1 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. sectional area and E is the elastic modulus of the beam. The equivalent loads method consists of three stages as follows.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.25(b). this distribution of stress must be subtracted to determine the stresses generated indirectly by the change in temperature.Page 88 Example 3. However. where: where α the coefficient of thermal expansion.24 Beam on rollers with partial (spring) restraint . In Stage C.25(a) and the associated stress distribution in Fig. Stage B—Analyse for the effects of the equivalent loads: The beam is analysed for the loading illustrated in Fig. thermal movement is resisted by bending in both the piers and the deck. on the other hand. In such a case. However. temperature on an unrestrained is member generates strain but not stress. 3.25(a). will generate both. 3.4. The equivalent force. While the method may not at first seem to be any simpler to apply than the procedure used above. it is necessary to identify the ‘associated stresses’. 3. This can become a significant factor in bridge deck design. i. even on an unrestrained beam. 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. F 0. 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. An axial expansion can be generated in an unrestrained beam by applying an axial force. Normally this stage would be done by computer but it Fig.e. Example 3. To some extent this alters the resistance to contraction or expansion. that distribution of stress which is inadvertently introduced into the structure by the equivalent loads. Therefore.5 serves to illustrate the effect of a moment connection between the bridge deck and the piers. The equivalent loads for this example are illustrated in Fig.

25(d). 3. 3. Thus. a differential temperature distribution develops which tends to cause the bridge to bend.25(d). (b) associated stress distribution. (d) stress distribution due to temperature change is trivial for this simple example. 3. the bending takes place freely and the beam curves upwards as the top expands relative to the bottom. Hence the force is taken in the ratio 1:2 as illustrated in Fig. This is the final result and is what one would expect from a thermal expansion in a partially restrained beam.Page 89 Fig. 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. 3.5 Differential temperature effects When the sun shines on the top of a bridge. The result is an axial compression of F0/(3A) throughout the beam. the top tends to increase in temperature faster than the bottom. In this case. This corresponds to the case of a beam on rollers subjected to an axial increase in temperature in that strains take . when a load is applied to two springs.25(b) from that of Fig. 3. If a linear distribution of this type is applied to a simply supported single-span beam. For this example. strain is generated but also some compressive stress.25 Analysis by equivalent loads method: (a) equivalent loads. 3.25(c). Stage C—Subtract the associated stresses: The distribution of associated stresses is subtracted from the stresses generated by the equivalent loads. It is well known that. this consists of subtracting the axial stress distribution of Fig. it is resisted in proportion to their stiffnesses. (c) equilibrium of forces at spring. the beam acts as a spring of stiffness AE/l.

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

even on unrestrained beams. it is necessary to identify the ‘associated BMD’. (b) associated BMD Temperature on an unrestrained structure generates strain and curvature but not bending moment or stress.27 Application of equivalent loads method: (a) equivalent loads. 3. Therefore.e. i.Page 91 Fig.28 Stages in equivalent loads method: (a) applied equivalent loads. The equivalent moment on the other hand will generate both curvature and bending moment. Fig.27(a) and (b) respectively. that distribution of moment which is inadvertently introduced into the structure by the equivalent loading. The equivalent loads and associated BMD are illustrated in Figs. 3. (c) BMD after subtraction of associated BMD . (b) BMD due to application of equivalent loads. 3.

29 is subjected to an increase in temperature which varies linearly from a maximum of 20° at the top to 10° at the bottom. 3. The temperature distribution is first converted into a strain distribution by multiplying by the coefficient of thermal expansion. 3. Stage A—Calculate the equivalent loads and the associated stresses: In this example. 3.28(b). the elastic modulus is E and the second moment of area is I. the central support point. B. the axial component will result in a free expansion. 3. 3.30: Fig. axial strain and bending strain. a.29 Differential temperature example Fig.28(b) gives the final result illustrated in Fig.8: Differential temperature change in continuous beam The three-span beam illustrated in Fig. The distribution is then resolved into two components. 3. Example 3.27(b) from Fig. as illustrated in Fig.30.30 Resolution of applied change in strain into axial and bending components . As the beam is free to expand. It is required to determine the BMD due to the temperature increase given that the coefficient of thermal expansion is α . a strain but no stress. 3. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig.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.e. The BMD will be determined using the method of equivalent loads. Hence. 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. 3.28(c). 3. i. from Fig. 3. the curvature is. The depth of the beam is h and the centroid is at mid-depth. The BMD due to the applied equivalent loading is as illustrated in Fig. due to symmetry. it is effectively fixed as illustrated in Fig.

32 First step in analysis by moment distribution: (a) system of fixities. (b) applied loading on fixed structure. the equivalent moment becomes: Thus. the equivalent loads and associated BMD are as illustrated in Fig.Page 93 Hence.31(a). 3.31.31 Application of equivalent loads method: (a) equivalent loads. Fig. (c) BMD in fixed structure . Fig. 3. Normally this stage would be done by computer but it will be done using moment distribution for this simple example. 3. 3. (b) associated BMD Stage B—Analyse for the effects of the equivalent loads: The frame is analysed for the loading of Fig.

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

3. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig. 3.33(a) and generates the BMD illustrated in Fig. the coefficient of thermal expa is 12×10 − and the modulus of 2 elasticity is 35 000 N/mm . 3.33(c).35(a) and (b) is subjected to the differential increase in temperature shown in Fig. the equi ivalent loading is a force of 580 kN and a moment of 160 kNm of which only the mome is of relevance. 3. (c) applied temperature distribution .34(a) gives the final result illustrated in Fig. 3. The normalised version is illustrated in Fig.33(c) factored by 5EIα The result is illustrated in Fig. 3. Example 3.31(b) from Fig. 3. 3. The cross-section and temperature distribution for this examp le are identical to those of Example 2.1 (Chapter 2).32(c) is corrected by adding Fig.34(b). This is the BMD due to the differential temperature increase. It is required to determ if there will be uplift at B c due to combined temperature and dead load.9: Bridge diaphragm The bridge diaphragm illustrated in Figs. (b) section through diaphragm. 3. /h.Page 95 Step 2: Unit rotation at B and C requires the application of the moments illustrated in Fig.34(a).35 Bridge diaphragm example: (a) plan of geometry. Step 3: The discontinuity of moment at B and C evident in Fig. 3.33(b). 3.35(c). The upward reaction from the bearing due to 6 the dead load is 300 kN. Fig. Referring to that example. 3. no further iteration is required. Step 4: As no discontinuity now exists.

36(b). The associated BMD is illustrated in Fig. 3. Subtracting the associated BMD gives the final BMD illustrated in Fig. (c) results of analysis.36 Analysis to determine effect of imposed differential temperature: (a) equivalent loading.37 Free body diagram for diaphragm beam To determine the reaction due to this moment.36(a). By symmetry.37: Hence the reaction at B is 80+80=160 kN.36(d).36(c). The reactions at A and C can be found from the free body diagram illustrated in Fig.7) the BMD due to applied loading is as illustrated in Fig.Page 96 Fig. 3. there is no uplift of this bearing due to the differential temperature change. . 3. 3. 3. 3. the structure is analysed for the loading illustrated in Fig. As the reaction due to dead load exceeds this value. (d) final BMD Fig. 3. Hence (as in Example 3. Point B does not rotate and is effectively fixed. (b) associated BMD.

38(b) from Fig.38α 10. (d) final BMD Thus.2.32α at the top and bottom fibres E E respectively.077α which E gives stresses (tension positive) of − 5.718α for half of the bridge. 3. a sagging bending moment is induced over the central support.38(d). 3.38(c). Fig. Hence.32α 5. It was established in Example 2. 3. 3. the total E stress at the top fibre is − 5.10: Differential temperature in bridge of non-rectangular section The beam-and-slab bridge whose section and temperature loading is described in Example 2. (c) results of analysis.38(c) gives the final distribution of moment due to restrained bending illustrated in Fig. Using the method of equivalent loads: E Stage A: The equivalent loads are illustrated in Fig.45α (restraint to expansion induces compression at the extreme fibres). In Example 2. It is required to determine the maximum stresses due to the differential temperature change.Page 97 Example 3.38 Analysis to determine effect of differential temperature change: (a) equivalent loading. 3.38(a) and the associated BMD in Fig.29α 5. 3.38(b).29α and 11. Stage B: Analysis by computer or by hand gives the BMD illustrated in Fig.38α and E − 5. 11. of 1. B. Stage C: Subtracting the associated BMD of Fig. (b) associated BMD.87α E.67α At the bottom fibre the total stress is E− E=− E.2 that the residual stresses are − 5. 3. it was established that the equivalent moment due to the temperature change is − 0.45α E− E=5. .2 consists of two 10 m spans.

The deck is subjected to the differential decreases in temperature shown in the figure. (b) section A—A and corresponding imposed temperature distribution. 3.11: Variable section bridge Figure 3. (d) section C—C . E=35×106 kN/m 2.Page 98 Example 3.39 Pedestrian bridge: (a) elevation. (c) section B—B and corresponding imposed temperature distribution. 3.39(b). 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. α =12×10− /°C and a modulus of elasticity. (c) and (d) show sections through it.39(a) shows the elevation of a pedestrian bridge while Figs. It is required to determine the equivalent loading and the associated stress distributions given a coefficient 6 of thermal expansion. Fig.

39(b)–(d)).40 Model of pedestrian bridge: (a) geometry showing differences in level of centroids. 3.39(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.Page 99 By summing moments of area it is found that the centroids are 0. 3. Summing products of stress and area in Fig. (b) equivalent loading. (c) associated axial force diagram. 3.033 m below the top fibre for the solid and hollow sections respectively (Figs.5 m and 1. (d) associated BMD .

41(a) to allow axial expansion in both the X and Y directions.39(b) and (d). However. 3.Page 100 In the hollow section.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. The bending moment and axial force distributions due to the temperature decreases can be found by analysing for the equivalent loading illustrated in Fig. the particle tends to expand in all three directions. it tends to bend about both axes. If there is restraint to either or both rotations. Example 3. when a differential distribution of temperature is applied through the depth of a bridge slab. 3. the bridge is two-span . 3. 3. respectively.40(c) and (d) from the results. for rotation. 3. The associated axial force and bending moment diagrams are illustrated in Figs. using members with very large stiffnesses can generate numerical instability in a computer model. 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.g. the area is 2. the area is 2. 3. Noting that the axial effects apply to all members while the differential temperature distributions only apply to the deck (abcd). the equivalent loads are illustrated in Fig.40(c) and (d). bending moment results about both axes as will be illustrated in the following example. Note that the short vertical members at b and c could be assumed to have effectively infinite stiffness. 3. an equivalent force of: For the hollow section. Similarly.5. 3.40(a). ten times) generally provides sufficient accuracy without causing such problems.1 Temperature effects in three dimensions When the temperature of a particle of material in a bridge is increased.6 m 2 giving E.40(b).6×1=2. However.40(b) and subtracting the associated distributions of Figs.12: Differential temperature The slab bridge of Fig. Therefore.41 is articulated as shown in Fig.

there are three bearings transversely at the ends so that it is not able to bend freely transversely either.42 corresponding to the different parts of the temperature distribution and the temperature . The specified temperature distributions are different in the cantilevers and the main deck of this bridge. the bridge will tend to act as one unit and bending will take place about the centroid.Page 101 Fig. 3.12: (a) plan showing directions of allowable movement at bearings. The coefficient of thermal expansion is 9×10 − /°C and the modulus of elasticity is 32×106 kN/m 2.41 Slab bridge of Example 3. Further. (d) imposed temperature distribution in cantilever (section 2–2) longitudinally and is therefore not able to bend freely. The deck and cantilevers are subjected to the differential temperature increases illustrated in Figs. 3. However. for longitudinal bending. 3. The location of this centroid is: below the top surface. (b) section A-A.41(c) and (d) respectively. (c) imposed temperature distribution in deck (section 1− 1). The bridge deck is divided into parts as illustrated in Fig. It is required to determine the equivalent loading and the associated 6 BMD due to this temperature change.

bending is about the centroid of the Fig. 3.43 Equivalent loading due to temperature .43. 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. 3.Page 102 Fig.42 Cross-section with associated distribution of imposed stress: (a) deck. In the cantilever region. The transverse direction is different from the longitudinal in that the cross-section is rectangular everywhere. 3.

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

Prestress. if a beam rests on a sliding bearing at one end.46(a). . it can undergo axial changes in temperature without incurring any axial stress.5 is used again here as illustrated in Fig. Example 3. 3. The normalised versions are presented here in Figs. As these applied moments generate distributions of longitudinal and transverse moment. on the other hand. results in changes of both stress and strain.43.13: Frame subject to axial prestress by moment distribution The frame of Fig.5. the problem is completed by analysing the slab (by computer) and subtracting the associated BMDs from the solution. An unrestrained change in temperature results in a change in strain only and no change in stress. It is required to determine the net prestress force in the deck and the resulting BMD.Page 104 In the main deck. The frame is analysed by moment distribution. 3. 3. P. Step 2: The effects of inducing rotations or translations at the fixing points are the same as for Example 3.47(a) and (b) (unit discontinuity in moment) and in Figs. the stress distributions are easily calculated and analysis is not generally required. of magnitude.45. For example. there are two associated BMDs as illustrated in Fig.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.47(c) and (d) (unit discontinuity in force). 3. is subjected to a prestressing force along the centroid of the deck. As for the previous example. prestressing that beam does (as is the objective) induce a distribution of stress. 3. However. However. there are many bridge forms where the effects of prestress are restrained to some degree or other and where analysis is necessary. 3. 3. Step 1: The system of fixities used in Example 3. reproduced here as Fig. However.16. ABC. the differential distribution is applied to a 0.46(b). 3.8 m deep rectangular section giving a moment about the centroid of: As M3 is applied to the outside of the cantilever. there is one important distinction. When the movements due to prestressing are unrestrained. only (M4− 3) needs to be applied at the M deck/cantilever interface as illustrated in Fig. The BMD due to applied ‘loading’ on the fixed structure is zero everywhere as the prestress forces are applied at fixing points.

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

Page 106 Fig.47(a) and (b) by 0.0178Pl and adding to give Figs. It can be seen in the results. (d) internal forces after correction for moment equilibrium Step 3: The translational fixity is released first to apply the prestress force. that equilibrium of forces at A and C is then satisfied. 3. 3. .48(d) is satisfied to a reasonable degree of accuracy.48(c) and (d). 3. (c) BMD after correction for moment equilibrium. 3. 3. (b) internal forces after correction for force equilibrium. This consists simply of factoring Figs.48(a) and (b).47(c) and (d) by P. Step 4: As force equilibrium in Fig. 3.48 Effect of prestress force: (a) BMD after correction for force equilibrium. illustrated in Figs. no further iteration is deemed necessary. The discontinuity of moment which results is removed by factoring Figs.

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

52(c).53. the resulting BMD in Fig. 3. (b) fixed BMD Fig.51 First step in analysis by moment distribution: (a) system of fixities. Step 2: The moments required to induce unit rotation at B and C are illustrated in Fig. (b) BMD associated with unit rotation. The BMD in the fixed structure due to the equivalent loading is as illustrated in Fig.52(c) by Pe/2 and adding. As there is no further discontinuity. this is the final BMD due to prestress.52 Effect of rotation of fixing points: (a) moments required to induce unit rotation.Page 108 Step 1: The beam is fixed as illustrated in Fig.51(b) is removed by factoring Fig.51(b).52(b) and the normalised BMD in Fig. The result is illustrated in Fig. (c) normalised BMD . 3. 3.52(a). 3. 3. Step 3: The discontinuity of bending moment evident in Fig. 3. Fig. 3. 3.51(a). 3. 3.

A parabolic profile generates a uniform loading. the actual profiles are sometimes approximated by ignoring the transition curves over the internal supports as illustrated in Fig. P. For this beam. it is required to find the BMD due to a prestress force. (This was covered in greater detail in Chapter 2. 3. the intensity of which can be determined by considering equilibrium of forces at the ends of the parabola.Page 109 Fig. a tendon below the centroid generates hogging moment. Example 3. In a simply supported beam.53 Final BMD due to eccentric prestress force It is interesting to note from Example 3. 3.14 that the effect of the tendon below the centroid is to generate sagging moment in the central span. 3. the slope is found by differentiating the equation as follows: Fig.) For the parabola in Span AB. For preliminary design purposes.54 Beam with profiled prestressing tendon .15: Profiled tendons In most post-tensioned bridges the tendons are profiled using a combination of straight portions and parabolic curves.54.

(c) BMD after correction for discontinuity in BMD .56 Equivalent loads method: (a) system of fixities for analysis by moment distribution. (b) summary of all equivalent forces on beam Fig.55 Equivalent loading due to profiled tendon: (a) equivalent forces in span AB. (b) equivalent loads and BMDs due to prestress in fixed structure.Page 110 Fig. 3. 3.

The bridge is subjected to uniform vertical loading of intensity. 3. vertical loading induces torsion as will be demonstrated in this example. is 1.56(a) and the associated BMD (Appendix A) is given in Fig. 3. x=0 and the slope becomes − 0.2P/l. w. only those grillages are considered in which there is no such joint displacement. 3. it is not necessary to provide a vertical translational fixity. 3. Fig. which is illustrated in Fig.2P/l. Example 3. Step 1: The symmetrical system of fixities is illustrated in Fig.08P. the vertical components can be found similarly. From Fig.12P.56(b).56(b). wBC=0. equilibrium of vertical forces requires a uniform loading of intensity: In BC. . As there is no further discontinuity. Similarly.14 and illustrated in Fig.56(c). The obvious difference is that typical prestress loading is in the opposite direction to loading due to self weight. the intensity is.57(b). As there is a support there already.Page 111 At A. Thus. In most practical grillages. wCD =wAB=0. Hence. They are both equal to 0. at x=l.1P and the intensity of loading is. EI. 3. Step 1: The two members are isolated from each other by the fixing of point B. GJ. As a result.08. In CD.7 Application of moment distribution to grillages A great many bridges are analysed by computer using the grillage analogy.00833Pl in Fig. 3. Example 3. The result is illustrated in Fig. In this method.52(c) is factored by this amount and added to Fig.55(a) it can be seen that the vertical component of the prestressing force at A is P sin θ≈ tan θ=0. 3. This bridge is long and narrow so it can be idealised by two beam members as illustrated in Fig. Step 2: The BMD associated with simultaneous rotations at B and C is identical to that derived for Example 3. 3. this is the final BMD due to prestress in this beam. 3.52(c).55(b). The torsional rigidity. coincidentally.16:Torsion due to vertical loading When bridges are curved or crooked in plan. described in detail in Chapters 5 and 6. the 1 P 1 slope is 0. the continuous bridge slab is represented by a mesh of discrete beams. Step 3: To remove the moment discontinuity of 0.5 times the flexural rigidity. The beam is analysed for this loading using moment distribution. moment distribution is applicable to the analysis of grillages but is tedious to apply for most examples.57(a).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. In this chapter. the complete equivalent loading due to prestress is as illustrated in Fig. by symmetry. 3.56(b). significant vertical translational displacements occur at the joints. 3.12 and the vertical component of prestress is 0. 3.

57(c) (rotation about two axes). This is done in Fig. (c) plan view of system of fixities. it needs to be resolved into components parallel and perpendicular to AB. The double headed arrows indicate rotational fixities where the positive direction is clockwise when looking in the direction of the arrow. 3. the bending moment diagram is as illustrated (in elevation) in Fig. The internal bending moment at the left end of BC is wl 2/8 as illustrated in Fig. (e) plan view of fixed BMD However.57 Analysis of crooked bridge: (a) plan view of geometry. A unit rotation is first applied . At B. 3. there is a transition between bending moment and torsion in the members. 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.58(b) and it can be seen that there is a discontinuity of moment at B of (1− 2)wl2/8=0. two rotational fixities are required in orthogonal directions as illustrated in Fig.57(e). 3. In plan. However. 3.Page 112 Fig.58(a). the BMD for the two beams are illustrated in Fig. From Appendix A. (b) plan view of idealisation. there is a 1/√ discontinuity in torsion of wl2/(8√ 2). In addition. (d) elevation of applied loading and resulting BMD in AB while fixed.293wl2 /8. 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. 3. In order to compare this moment to that just left of B. 3.57(d). Vertical loading on AB in the fixed structure is applied to a beam which is fixed at one end.

59(b). The discontinuity of moment at the joint can be seen by resolving the internal moments and torsions in Fig.59(d). the discontinuities in the fixed bending moment and torsion diagrams are removed by scaling and adding the diagrams derived in Step 2.60(c). In BC. Unit rotation at the end of AB results in the deformed shape and BMD illustrated in elevation in Fig. The discontinuity of moment at B in the BMD of Fig.59(c) can be seen when the moments are resolved in Fig. 3. The . the normalised version of Fig. Hence the BMD due to unit rotation at B is as illustrated in Fig. 3.59(d).59(e).Page 113 Fig. 3. The discontinuity parallel to Direction 2 at B is GJ/l+3EI/2l=3EI/l. the rotation at the joint must be resolved into components as illustrated in Fig. 3. (b) resolution of end moment in BC parallel and perpendicular to AB in Direction 1 (Fig. 3. Applying a unit rotation in Direction 2 (Fig. To determine the effect on member BC.59(c). 3. 3. 3. 3.59(a). is illustrated in Fig.60(b).60(d). 3.60(a) and it can be seen that it generates no torsion and the BMD illustrated in Fig. 3.57(c)) generates no bending but a torsion of GJ/l in AB. dividing by this value gives the normalised version illustrated in Fig. As there is no resistance to twisting at C.58 Plan views showing internal moment: (a) end moment in BC. it is necessary to resolve the rotation into components parallel and perpendicular to that member as illustrated in Fig. 3. normalised for moment. 3. 3.57(c)). It is (3+3/2)EI/l=9EI/2l. the application of a twist of 1/√ at B does 2 not generate any torsion in BC. Step 3: In the third step.

3.805wl2/8.60(d) to give the diagram illustrated in Fig. Adding the BMD 1/√ of Fig. 3. 3.61(a). The new discontinuity now introduced parallel to Direction 1 is removed by adding a diagram proportional to Fig.293wl2 /8. 3.59 Effect of rotation in direction 1 at B: (a) elevation of AB showing imposed unit rotation and associated BMD.61(b).61(c) and the discontinuity in that is removed by adding a diagram parallel to Fig. (b) resolution of rotation parallel and perpendicular to BC. (c) BMD due to unit rotation.59(e) to give Fig. 3. 3.61(d).58(b) is (1− 2)wl2/8=0.57(c)) has now been removed at B.60(d).Page 114 Fig. 3. The corresponding bending moment and torsion diagrams are illustrated in Figs. 3. 3. (d) free body diagram showing lack of equilibrium of moments at B. This is removed by adding the diagram of Fig. 3.59(e) scaled by minus this value gives the moments and torsions illustrated in Fig. . (e) normalised free body diagram discontinuity in the fixed BMD of Fig.61(e) and (f). 3. 3. While the discontinuity parallel to Direction 1 (Fig. scaled by minus this value to give the diagram illustrated in Fig. there is still a discontinuity parallel to Direction 2 of 0.

Page 115 Fig. (b) BMD due to unit rotation. (d) normalised free body diagram . (c) free body diagram showing lack of equilibrium of moments at B. 3.60 Effect of rotation in direction 2 at B: (a) resolution of rotations parallel and perpendicular to BC.

3. Example 3. (d) after second correction in direction 2.61 Successive corrections to internal moments and torsions: (a) after correction of moments in direction 1. (c) after second correction in direction 1.62 Long skewed bridge: (a) plan view. the process must be continued until no discontinuity remains.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. (c) plan view of idealisation . (b) cross-section.17: Torsion due to skew supports The skewed bridge illustrated in Figs.Page 116 Fig. 3. (f) final torsion diagram Step 4: To get an exact answer. (e) final BMD. 3. (b) after correction of moments in direction 2.

8. (b) simplified system of fixities. The symmetry of the system is exploited recognising that the three fixities at B are identical to the corresponding fixities at E. It is subjected to vertical uniform loading of intensity w.63(a). Fig.0. The circles in this figure indicate translational fixities with a direction of positive upwards. This deck has flexural rigidities of (EI)ABC = (EI) DEF=0. 3.62(c). This is represented diagrammatically in Fig.0. Step 1: To isolate the members from one another.63 First step in analysis by moment distribution: (a) complete system of fixities with two rotations and one translation. 3. total fixity must be imposed at B and E. i.Page 117 illustrated in Fig. (EI)BE=3. (GJ)BE =4. it is necessary to fix against vertical translation and against rotation about both axes. (c) fixed BMD. 3.e. (d) free body diagram showing lack of moment equilibrium in fixed structure .6 and torsional rigidities of (GJ)ABC =(GJ)DEF=2.

Page 118 The system of fixities illustrated is adequate. sin θ )(GJ)BE/l1=8 sin θ 1. As the two bearings are relatively close together (i.63(d). There is no torsion in the bridge in its fixed state. (b) elevation showing moments required to induce rotations in BE. Therefore it is sufficient to fix the bridge as illustrated in Fig.63(c). to rotate member BE in Direction 1 requires it to be rotated and twisted. 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. In the fixed structure. Step 2: The second step consists of applying a rotation at B and E. 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. 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.63(b). 3. 3. The member must also undergo bending in order to rotate at each /l end through cos θThe elevation showing the . 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 . the rotation in Direction 2 will be small. (c) BMD associated with unit rotations at B and E . but unnecessary in practice.e. BE is fixed at each end and the BMD (Appendix A) is as illustrated in Fig. 3. 3. This can be seen by resolving the unit rotation into directions parallel and perpendicular to the member as illustrated in Fig. 12<<l 1). Fig. 3.64(a).64 Effect of rotations at B and E: (a) plan showing resolution of rotations into components parallel and perpendicular to BE. However. as to apply a unit rotation there would require a moment that is very large. By the same token.

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 .65(a). Resolving parallel to the direction of fixity gives the discontinuity of moment corresponding to unit rotation: Normalising with respect to this discontinuity.66(a). 3. 3. 3. last BMD) is illustrated in Fig. 3. 3.65(b) scaled by that amount. /l The internal moments and torsions at B and E are illustrated in Fig. 3. Fig. the discontinuity in moment parallel to Direction 1 in the fixed structure is (wl2/12) cos θThis is removed by adding Fig. 3. the corresponding BMD (Appendix B. Thus.64(b).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.Page 119 required deflected shape is illustrated in Fig. results in the internal moments and torsions illustrated in Fig. 3.64(c). . (b) normalised free body diagram .65(b).63(d). The resulting diagram is illustrated in Fig. Step 3: As stated in Step 1 and illustrated in Fig.

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

4.4 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.3) and (4.4) (4. 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.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 . 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. Such an approach is described later in this chapter.2) where (4.Page 125 Fig. 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. or to a wall translation of Hret /20. Hret/10. these are in keeping with the h general guidelines set out in BA42/96 (1996).

it is necessary to quantify .5 for equation (4.2 Hret . Therefore.7) where α the coefficient of thermal expansion of the deck. For the latter case.1. For most cases. for a bridge deck of length L which experiences an increase in temperature of Δ δ may be calculated T. Example 4.4 indicates that Kp=17. Assume α the deck is 12×10− per °C and for From equation (4.6) (4.Page 126 restraint offered by typical abutments and backfill is relatively small). it is reasonable to assume that horizontal stress acting on both sides of the abutment are given by equation (4. is Implicit in equations (4.3 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. The peak angle of friction of the fill is 45° and its dry density is 1900 kg/m3. av as: (4.6) and (4.3) gives: and . The design extreme event for the determination of maximum abutment pressures is a 40° 6 increase in temperature. av The unit weight of the soil (γ ) is soil Therefore for z<6 m (Hret ): 4. therefore. As Hret=6 m and δ <Hret/20.7): Figure 4.7) 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.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.5) when the depth exceeds approximately 1.

A typical approximate relationship has been proposed by Lehane et al. The stress-strain relationship for soil is non-linear at strains in excess of about 0. e.01. patm is the atmospheric pressure (100 kN/m 2). p' is the mean confining stress less the pore water pressure in the soil. which is related to the void ratio.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. ρ.5 Secant Young’s modulus for granular soil (assuming Gs=2.65) (after Lehane et al. a higher soil stiffness will lead to higher axial forces and bending moments in the deck due to its longitudinal expansion or contraction. Clearly. by the expression: d (4. 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). the level of confining stress and the loading history. This can only be achieved with a knowledge of the appropriate soil stiffness parameters.9) where Gs is the specific gravity of the soil particles (typically 2.8) where E s is the secant Young’s modulus in kN/m2. e is the void ratio of the soil. The degree of compaction of backfill on site is often specified in terms of the dry density.Page 127 the restraint provided by the soil. (1996)) . used as a reference stress and γ the shear strain which is taken to lie within the is 6 range 50×10 − to 0. (1996): (4. The design stiffness used for the calculation of such forces and moments should therefore be a maximum credible value. 4.65) and ρ is the w Fig.

Thus.2 Contraction of bridge deck There is generally a lesser height of soil in front of bridge abutments than behind them.Page 128 density of water. the principal uncertainty relates to the resistance to movement at the bases of the piers and abutments. 4. is plotted in Fig. This means that. in an analysis to determine the effects of elastic shortening. d or Fig.6 Contraction of frame rigidly fixed at supports: (a) geometry. 4. and/or shrinkage. However. Guidance on appropriate values for ρ p' and γ specific cases is given in subsequent for d. creep. The secant Young’s modulus.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. and shear strains (γ Equation (4. the resistance provided by such soil to the contraction of a bridge deck is usually small. sections. 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. These conditions are applicable if the abutment foundations are cast in very dense soil or rock. (b) bending moment diagram from example 3. 4. 4.5 .5 can be used to estimate the secant Young’s modulus for cohesionless soil.2. mean confining stresses (p').5 for a range of in-situ dry densities (ρ). derived using equation (4. 4. As a result. e.8).8) ). specification of the dry density effectively dictates the void ratio. Fig.

the decrease in temperature would generate a large tensile force in the deck and there would be no contraction. However. The soil around the strip foundation can be idealised by a number of linear elastic springs. In that case. embedded to a depth of between 0.5–1. Strip foundations or pile caps are commonly founded at around 0. only 5% of the potential level. some movement of the deck was possible through bending in the abutments.Page 129 The bridge illustrated in Fig. It is this small depth of soil.2.6(a) was considered in Chapter 3 for an axial contraction due to temperature of 20° in the deck (ABC) (Example 3. due to the integral nature of the bridge. 4. 4.10): (4.0 m below the ground level are given in equation (4. Es. 3. ) . bending in the deck also. and a relatively large contraction.6(b). 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). together with sliding resistance at the base of the pad. The axial contraction induced bending in the abutments and. k hori and krot are the stiffnesses per metre length of strip foundation for vertical.0 m below the ground level on the inside of the abutment as illustrated in Fig.7.10) where kvert.2 Contraction of bridge on flexible supports Most bridges are constructed on supports which have some degree of flexibility. that resists bridge contraction.8) assuming a p' value equivalent to the foundation bearing pressure and a shear strain (γof 0. Design spring stiffnesses on the inside of the abutment for a strip foundation of width B.001. Thus. Expressions for the stiffness of such springs have been deduced here from relationships provided by Dobry and Gazetas (1986) for an elastic soil. The complete bending moment diagram is illustrated in Fig. 4. Abutments and piers are generally either supported on foundations bearing directly on the ground below or on pile caps underlain by piles. horizontal and rotational displacement respectively. 4. upper bound estimates of the secant Young’s modulus of elasticity. Conservative.5). 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.5 m and 1. The end result for that example was a relatively small axial tension in the deck. If the ends of the deck were fully prevented from contracting.20(c) shows that only 7% of the potential force is applied to the abutments). the bridge was fully restrained at the base of each abutment and pier. may be calculated using equation (4. However. the resistance of the abutments to movement was considerably less than the axial stiffness of the deck (Fig.

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. (b) deep abutment Example 4.5 m. gives a void ratio of: .7 End of integral bridge showing shallow depth of soil on inside: (a) bank seat. Inverting equation (4.Page 130 Fig.65.8 is subjected to a shrinkage strain of 200×10− . d of 1900 kg/m 3. 4. The degree of compaction has been controlled by specifying a dry density of backfill. The foundation is assumed to be working under a bearing pressure of 300 kN/m2 and the breadth of the strip foundation is 2.2: Contraction for shallow strip foundation 6 The bridge illustrated in Fig. ρ. 4.9) and assuming Gs=2.

8) then gives: Equation (4.Page 131 Fig.2: (a) elevation. 4. (b) detail at abutment Substituting in equation (4.8 Bridge of Example 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. 4.9 Computer model for bridge of Example 4.10 Analysis results: (a) deflected shape. (d) corrected axial force diagram .10(a)–(c).10(d). axial force and bending moment diagrams illustrated in Fig.Page 132 The equivalent loads and the associated axial force diagram are illustrated in Fig. Subtracting the associated axial force diagram gives the actual distribution of axial force generated by the shrinkage. No adjustment is necessary for the deflected shape or bending moment diagram. 4. (c) bending moment diagram. 4. The frame was analysed using a standard analysis package which gave the deflected shape. illustrated in Fig. (b) associated axial force diagram Fig.2: (a) equivalent loading and springs. 4. Fig. (b) axial force diagram from computer analysis. 4.

Out of a total potential 6 shortening of 6 mm (200×10− ×30000 mm) at each end. the stresses generated by an increase in deck temperature. Such a model is imperfect as it does not allow for shear transfer within the soil as there is no interaction between the .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. 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). 2.4 N/mm 2. Nonlinear elastic finite-element analyses by Springman et al. with time.7 mm is predicted to actually occur. The stiffness of the soil is influenced by the shear strain in the backfill. The selection of a suitable soil stiffness value (Es) is essential for appropriate modelling of the backfill.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. will be affected significantly by the properties of the soil behind the abutments. for example. 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. However.3 mm does generate distributions of stress in the frame. stiffness value. Thus. to tend to an equilibrium density compatible with the strain amplitude that it is regularly subjected to. This observation suggests that the use of a constant soil stiffness value with depth (for a given strain) is reasonably realistic. 3. Some notable features have been observed from experimental studies by Springman et al. an average shear strain must be assumed.7). 4. However the restraint which prevents the remaining 0. the bending moment at the ends are more significant at 568 kNm. The conventional spring model represents the backfill soil and soil beneath the abutment by a series of spring supports. and others: 1. Assuming uncracked conditions. this corresponds to a maximum flexural stress of 2. The average shear strain in the backfill must be less than 2δ and could conservatively be assumed as about 2δ /H /3H. Cyclic variations in temperature (and associated expansions and contractions of the deck) cause the backfill to compact and. 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. To adopt a single soil /H.Page 133 Example 4. 5.3 N/mm 2. (1996). (1996) support the validity of this assumption. The axial tension is relatively small at 337 kN corresponding to a stress in the deck of less than 0. 4.

of the backfill behind an abutment of depth H and transverse length. has been specified as 1600 kg/m 3. have the advantage of simplicity and is considered here because it remains a popular approach among bridge engineers.: Inverting equation (4. i. L: (4. It does.65. assuming linear elasticity. The d piles are assumed to provide insignificant lateral restraint to the deck. An approximate expression. 4. 4.11) The application of equation (4.11 is subjected to an increase in temperature of 20 °C.9) and assuming Gs=2. Example 4. The dry density of the backfill.3 . It is assumed that the density of the backfill reaches an equilibrium value 20% in excess of that specified. gives a void ratio of: Fig.Page 134 springs. 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.11 Culvert of Example 4.e.3: Conventional spring model The culvert illustrated in Fig. has been developed for the horizontal spring stiffness per square metre.11) is illustrated in the following example. ρ.

4. the average shear strain in the affected backfill is then: On the basis of Note 2. The bending . the product of the temperature increase. the expansion of the culvert is estimated as its unrestrained value. i.12(b).11): The model for a 1 m strip of the frame is then as illustrated in Fig. Then equation (4.e. 4.8) gives: The horizontal spring stiffness is then given by equation (4. a horizontal stress of p'=50 kN/m2 is assumed. The equivalent loading is: and the associated distribution of axial force is illustrated in Fig. 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.Page 135 To estimate the average shear strain induced in the backfill.12(a).

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

however. 4. 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. 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. i. no passive failure or abutment lifting were allowed) and that no slip between the abutment and the soil occurred (e. 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). described in Section 4.14). (b) unit rotation .g.14 Stiffness components at top of abutment: (a) unit translation. given that relatively small movements are required to reduce pressures to their minimum (active) values on the inner face of the abutment. It was therefore assumed conservatively that the soil had limitless compressive and tensile strength (e. This technique consists of modelling both the abutment and the surrounding soil with an equivalent lateral and rotational spring at deck level. This method does not. Fig.3. 4. the analyses assumed that any soil present on this side did not contribute to the resistance. the forces and moments associated with passive movements which occur as a consequence of deck expansion. provide details concerning the distribution of moment in the abutment or the pressure distribution in the soil. However.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.e. and (ii) a rotation θ with zero horizontal displacement (Fig.4.1 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.g.Page 137 4. 4.

1.14).12) were the most important factors controlling the magnitudes of the lateral force (Fh) and moment (M) at the top of the abutment (Fig. All values in this stiffness matrix can be reduced by 15% if friction between the abutment and soil is considered negligible.05 1.14) .5×106 >0. 4.14) B (m) (Fig.0×104− 2.5–3. (4. The values of Fh and M were also seen to increase systematically as the base width (B) increased and its height (H) reduced.Page 138 It was found that the flexural rigidity of the abutment (EIa ) and the ratio.5− 12 0.13) where f 1 and f 2 are functions of the ratio. Best-fit expressions were obtained for Fh and M for the range of parameter values given in Table 4.1 Range of parameters used in derivation of equation (4. defined as: (4. 4.14) Fig. r.5 H (m) (Fig. 4. 4. H/B which are given by equation (4.14) for r>0. They are given here in matrix form: (4.05 3 m −.13) Parameter Es(kN/m2 ) EI a(kNm /m) 2 3 r=Es/EIa(m− ) Allowable range 10000− 500000 1.15 End part of frame bridge showing locations and directions of fixity Table 4.

the stiffness matrix. When the bridge is embedded in soil and this is taken into account.15) becomes: (4.16) A comparison of equations (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. Ld and Id are the cross-sectional area. span length and second moment of area of the deck respectively. Similarly.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.13) was found to predict values of F h and M to within 10% of the values given by the finite element analyses.18) .Page 139 For the range of parameters listed in Table 4. the terms involving Ia and H are replaced with terms from equation (4. gives: (4. 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.15. equating the K12 (and K21) terms gives: (4. second column) terms in equations (4.17) where Heq and Ieq are the equivalent abutment height and second moment of area respectively. is: (4. [K].15) and (4. in the absence of soil.15 is used. Equating the K22 (second row. equation (4.15) where A d.13) with the result that equation (4.15) and (4. 4. 4.16). 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 (4.16).1. Alternatively.

21) gives a spring stiffness of: (4.Page 140 Equations (4.21) 4.23) Finally.24) These equations can be used to estimate the properties of an equivalent frame for an integral bridge with deep abutments.17) and (4. For values of (H/B) in excess of 10. the parameters f1 and f2 approach their minimum values of 0.40 respectively.18) can be simultaneously satisfied by selecting an equivalent abutment second moment of area equal to: (4.2 Expansion of frames with deep abutments The equivalent single-spring model can be simplified for the case of deep abutments. As a result.20) gives an equivalent height of: (4.19) The equivalent abutment height is then: (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: (4.33 and 0. substituting for f 1 and f2 in equation (4.4. .22) Substituting for f 2 in equation (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.

from equation (4.23): The stiffness of the single spring on each side is given by equation (4. is: Fig.3.16 Computer model for bridge of Example 4. as for Example 4. The dry density of the backfill has been specified as 1600 kg/m 3.4 .11 due to a temperature increase of 20°. The elastic modulus of the soil is found as for Example 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.Page 141 Example 4.16.12) is then: The equivalent height of abutment is then.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.24): The equivalent frame and loading are illustrated in Fig. 4. 4. The magnitude of the equivalent loads.3 to be: and the second moment of area of a 1 m strip of the abutment is: The ratio defined by equation (4. 4.

4. 4. Effective rotational stiffnesses at this lateral movement are about 75% of the purely elastic case. It is important to remember that the distribution of moment in the abutment is not realistic. It is therefore recommended that this equation be used in preliminary analysis and that a finite- . when the soil is treated as an elastic perfectly plastic material.14) imply that an abutment provides a greater resistance to deck expansion if it has a lesser depth of embedment (H).12(b). with a finite strength defined by its friction angle. However. 4.14) would suggest. 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. 87 kNm. 4. . In the example of Fig.3 Expansion of bank seat abutments Equations (4.4. 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. 4. predictions from finite-element analyses are presented of a horizontal force/deflection relationship. linear. the deflections per unit load can be seen to be significantly greater.18(b). 4. is that shallow abutments are more likely to slide than deep ones and will therefore offer less restraint to deck expansion than equations (4.13) and (4. of course.3.17 Bending moment diagram from computer analysis of bridge of Example 4. of course. the magnitude of moment in the deck. The influence of a limited soil strength on the resistance offered by a bank seat is illustrated in Fig. The model was analysed using a standard computer program and the resulting bending moment diagram is illustrated in Fig.13. the true shape of this distribution will be similar to that given in Fig. 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. 4.13) and (4. the function is. In Figure 4.4 The associated axial force diagram is as illustrated in Fig.13). is likely to be more reliable than the value found in Example 4.18. When the soil is linear elastic and infinitely strong. On the other hand.Page 142 Fig.18.17. The reality. Similar results can be shown for moment/rotation functions and for force/rotation and moment/deflection functions.

18 Finite-element analysis results for bank seat abutment (E s= 100000 kN/m 2. The 3 dry density of the backfill has been specified as 1600 kg/m . section through bank seat. soil friction angle. (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.12) is then.018 m4. Example 4.19 due to a temperature increase of 20°C.3 but using the smaller abutment height.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.31. . and the second moment of area of a 1 m strip of the abutment is Ia =0. 4.Page 143 Fig. 4. foundation bearing pressure= 200 kN/m2. The ratio defined by equation (4. The concrete has an 6 elastic modulus of 28×106 kN/m 2 and a coefficient of thermal expansion of 12×10−/°C. The elastic modulus for the soil is found in the same manner as for Example 4. r=0. Ec=30×10 6 kN/m2 .

4.5/3=0.20): The equivalent abutment second moment of area is given by equation (4.19): Finally.14): The equivalent height is then calculated directly from equation (4.Page 144 Fig. the ratio of embedment depth to foundation breadth. from equation (4.5 For this example.83. is 2.19 Bridge of Example 4.21): . f1 and f2 are calculated from equation (4. H/B. the spring stiffness is. The parameters.

5 The equivalent frame and loading are illustrated in Fig.20 Computer model for bridge of Example 4.22. it transfers the relative horizontal movement from the end of the deck to the end of . This is achieved in many cases by the installation of a run-on slab as illustrated in Fig.21 Bending moment diagram from computer analysis of bridge of Example 4. Preventing relative horizontal translation is not so simple. Preventing relative vertical translation significantly improves the rideability for vehicles travelling over the bridge. the bridge still expands and contracts relative to the surrounding soil and the incorporation of a run-on slab does not prevent this. this must be accommodated if premature deterioration of the pavement is to be avoided. as for Examples 4.20.4. 4. Clearly.Page 145 Fig.5 Fig. 4. 4.5 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. 4. It is clear from Fig. 4.3 and 4. 4. is: This model was analysed and the bending moment diagram is illustrated in Fig. On a road bridge. In effect.18 that this result is quite conservative. The effect of such a slab is to allow relative rotation between the deck and the run-on slab while preventing relative translation. The magnitude of the equivalent loads.21. 4. The maximum magnitude of moment in the deck due to the expansion is 114 kNm.

and in The assessment of the required length of the run-on slab relies on observations of measured behaviour and engineering judgement. . This is because existing models which attempt to simulate the soil’s response to a complex history of cyclic straining are very approximate. Analytical prediction of the shapes and magnitudes of settlement troughs is. Such straining is imposed on the backfill by the abutment which moves in response to thermal movements of the deck. The extent of the settlement trough is also controlled by the amount of backfill subjected to cyclic abutment movements and therefore. after many cycles of imposed lateral movement δδ varies . not commonly attempted by bridge designers. however. 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. 4. max between about 10δ 20δ well-compacted fill for both deep abutments and bank seats.22 Run-on slab the run-on slab. 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. 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. Much larger settlements occur in initially loose backfills where considerable volumetric contractions take place before an ‘equilibrium’ density is attained. (1996) that. Settlement troughs arise because of the tendency for cohesionless backfill. for a given movement of the top of the abutment.Page 146 Fig. Run-on slabs are designed to span the settlement troughs that develop behind the abutments of integral bridges. 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. whatever its density. 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. It has been shown max by Springman et al. difficult to use and require measurement of a large range of representative geotechnical parameters from cyclic laboratory tests.

the run-on slab should have a length of at least 2. 4. 4.1(2. the precast beams are simply supported and the self weight of the bridge induces a sagging moment. 4. (b) section A—A . When the in-situ concrete Fig.6H 0. as illustrated in Fig. 4.25 m.5)=5.23 Composite integral bridge made from precast and in-situ concrete: (a) elevation.19) is calculated. 4.2.1H These observations and those taken during centrifuge model tests by Springman et al. When the in-situ concrete is cast. As an example. a maximum trough length of 2.9H Loosely compacted 1. (1996) suggest that the length of the trough (Lt ) is unlikely to exceed the limits given in Table 4.6 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. Hence.4H 2.1H can be assumed from Table 4.2.5 (Fig.24(a).2 Approximate upper limits on expected trough lengths Granular fill Well compacted Deep abutments Bank seats 0. the length of run-on slab required for the bridge of Example 4.23.Page 147 Table 4. As the backfill is loosely compacted (density= 1600 kg/m3) and the abutments are not deep.

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

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

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. Due to creep. such curvature is unrestrained so it results in instantaneous strain and a moment which is the simple product of prestress force and eccentricity. 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 calculated the maximum longterm sagging moment for beams made integral when between 21 and 100 days old. 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. 4. further curvature is resisted and the resulting distribution of moment is as illustrated in Fig. the beams hog upwards as illustrated in Fig. 4. It can result in cracking at the bottom of the deck over the supports as illustrated in Fig. 4.26(a). The long-term result is a distribution of prestress moment such as that illustrated in Fig.Page 150 Fig. 4. 4. 4. For smaller beams. these hogging strains increase with time. When prestress is first applied below the centroid.27(c).27(b).26(b)). 4.28 Detail at support showing points where long-term cracking is likely to occur Fig. Clark and Sugie (1997) carried out a parametric study of the time-dependent effects in composite integral bridges.27(a). particularly at the interface between the precast and insitu concretes. When the bridge is made integral. they suggest designing for a moment of 600 kNm. If some strands are debonded.28. As they are simply supported. In a study of continuous bridges made integral at the interior supports. . the equivalent moments at the ends are less but a further increment of equivalent moment is applied at the points where debonding ceases (Fig.

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

the points a. Such a plate might be constructed of a material where the microstructure is orientated in two mutually perpendicular directions. These types of slab are referred to as geometrically (or technically) orthotropic. δ In this figure the thickness of the plate is taken to be d. such as reinforced concrete slabs with significantly different amounts of reinforcement in the two directions or voided slabs. 5. This type of plate is not typical of that found in bridge decks but is frequently used as an approximation of actual conditions. 5. the cube both moves and distorts. While the theory is strictly only applicable to cases of material orthotropy. A materially (or naturally) orthotropic plate is composed of a homogeneous material which has different elastic properties in two orthogonal directions. b. at which point.2 move to a'. applied.e. Thus. Although this type of material is rarely found in bridge construction. When a load is z. such as timber. will be u plus the change in u over the distance δ i. Hence the length of a'b' projected onto the X axis is: . the theory of materially orthotropic thin plates is developed. The origin of the axis system is at Y mid-depth in the plate.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. but the same geometric properties.2. 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. the x displacement at that point in the X direction.b'. In the following sections. 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. Many bridge slabs possess different second moments of area in two directions.Page 152 materials in which the behaviour in all directions is the same. Considering point b. isotropic plate theory can be used with reasonable accuracy for the analysis of many bridges. a distance δ from a. c' and d' as illustrated in Fig. 5. it is common practise to extend it to include geometric orthotropy.: x. z=0. Figure 5.1 shows a portion of a thin plate in the X− plane. The displacement of point a in the X direction is denoted u.2 Bending of materially orthotropic thin plates Figure 5.3. Considering initially the X− plane. Z c and d shown in Fig.

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

Page 154 Fig.3 Distortion of cube of material in X. if v and w are the displacements in the Y and Z directions respectively. the strain in the X direction is: (5.2) and: (5. 5.1) Similarly. it can be shown that: (5.3) .Z plane By definition.

a number of assumptions are made to simplify the mathematics involved. As can be seen in Fig. the difference between c'a'b' and cab.4) Similarly the shear strains in the X− and Y− planes are respectively: Y Z (5. 5.e. or that w is a function of x and y only. α and βReferring to the figure: .4 illustrates the implications of this assumption. Figure 5. As ∂ x is small. The physical meaning is . this reduces to: u/∂ The other component of strain can be found similarly to be: Hence the shear strain is: (5. .Page 155 The shear strain in the X− plane is defined as the change in the angle.7) This implies that w is independent of z. there are two components. i.3.6) In thin-plate theory.5) (5.: (5. The first of these assumptions is that there is no strain in the Z direction. cab from the original Z 90°.e. i.

In other words. their behaviour is dominated by bending rather than shear deformation. 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. 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. This assumption is again a simplification of the true behaviour. a means for determining shear stresses will be required.Page 156 Fig. concrete bridge slabs do not have great shear strength. but is justified by the fact that. the depth of the slab remains unchanged throughout.5 where it can be seen that the 90° angle of cab is preserved in the distorted c'a'b'. i. Notwithstanding this.8) (5. 5. bridge slabs being relatively thin. .9) The consequences of this are shown in Fig.: (5.4 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. Such a method is presented later in this section.e. and all points deflect vertically by exactly the same amount as the points directly above and below them. and although shear strains are small. 5.

Hence. Substituting this into equation (5. at z=0. there is no displacement in either the X or Y directions at z=0. u and v are both zero. As the origin is located at the centre of the plate and bending is assumed to occur about that point. this implies: (5.10) where C is a constant of integration.10) implies that the constant C is zero giving: (5. 5.5 Segment of plate in X− plane showing assumed lack of shear distortion Z Rearranging equation (5.11) .8) gives: As w is independent of z.Page 157 Fig.

16)–(5.5) gives: (5.12) Substituting equations (5. the equations is are similar.16) (5.21) .15) respectively then gives: (5. Y and XY directions which are given by: (5.14) Similarly equation (5.12) into equations (5.17) (5.13) (5.11) and (5.18) Substituting equations (5.13)–(5.Page 158 By rearranging equation (5. a similar expression can be derived for v: (5.18) into equations (5. but there are now curvatures in the X.15) In the flexural theory of beams. In thin-plate theory.1) and (5.3) respectively gives: (5. the curvature is defined as: where κ the curvature and R is the radius of curvature.20) (5.9).19) (5.

εis related to the stress. Equation (5. Such cases are discussed further in Chapter 7. but some cases do exist where this is not so.2.6 (a) shows a onedimensional bar subjected to a tensile force. it follows that plane sections remain plane. This strain.6(b). From this. expressions were established for the various strains in a thin plate.and three-dimensions: (a) one-dimensional bar. as is generally assumed in beam theory. By defining the X axis as the direction of the applied force.3 Stress in materially orthotropic thin plates In the previous section. (b) three-dimensional body showing the effect of stress in the axial direction on strains in the orthogonal directions . 5. the strain in that direction is given by: (5.22) Fig. 5. . elasticity.19) shows that strain in the X direction is a linear function of z. Expressions are now developed for the corresponding stresses. E. as 2 κ=∂w/∂2 is independent of z. This is generally a reasonable assumption for slab bridge decks.20) shows that the same applies to the strain in the x x Y direction. by: In the three-dimensional case. strains in the other two directions become significant.6 Distortion in one. as is indicated in Fig. Figure 5. 5.Page 159 Examination of equation (5. The only significant strain in this system is in a direction parallel to the axis of the bar. σand modulus of .

Y and Z directions respectively. 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. and ν νand νare the corresponding Poisson’s ratios.22) assumes that the plate is x. Equation (5. ν etc. E y and Ez are the moduli of elasticity in the X. For a thin plate in bending. Consequently the last term of equation (5. y z made of a homogenous material and that the elastic constants (Ex. each other.24) .22) can be ignored.Page 160 where E x.23) (5. the stress in the Z direction is small and the Poisson’s ratio is generally small for bridge deck materials.) are independent of x. as is appropriate for the materially orthotropic (or anisotropic) case.

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)

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

2. can be significant. while numerically small. 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. In the simple flexural theory of beams. However. shear stresses.44) and: (5.7.43) The moment mxy (=myx ) is often referred to as a twisting moment and is distinct from the normal moments mx and my . 5. particularly in concrete slabs which are quite weak in shear. Figure 5. Figure 5. .10(a) shows the direction in which each of these moments acts while Fig. Defining qx and qy as the downward shear forces per unit breadth on the positive X and Y faces respectively then gives: (5. In thin-plate theory.9)) that shear deformations in the plate were negligible. and illustrated in Fig. gives an expression for the shear force Q: (5. Unlike beams.45) It was assumed earlier (equations (5. one for each direction (X and Y). 5. a similar expression can be derived.5 Shear in thin plates Vertical shear forces occur in bridge decks due to the shear stresses.Page 167 where j is known as the torsional constant and is given by: (5. the same phenomenon exists and an expression is found from equilibrium of forces on a segment.e. Taking moments about the left hand end gives: Rearranging and ignoring the term. This is a reasonable assumption as shear deformation is generally small in bridge slabs relative to bending deformation.46) i.8) and (5. 5.10(b) shows the type of deformation associated with each of them.11 shows a segment of a beam of length dx in bending. dQdx which is relatively small. the shear force is the derivative of the moment. there are two shear forces at each point.

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)

5. 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. 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.13 and 5.56) where i is the second moment of area per unit breadth. . for example. as large variations of torque may exist between the longitudinal and transverse beams. This technique is discussed further in the next section. but the torques in grillage members b1 and b2 will not necessarily be equal to the torques in b3 and b4.53) and: (5.2 Grillage member properties—isotropic slabs A grillage member in bending behaves according to the well-known flexure formula: (5. unless myx is particularly large. The shear forces per unit breadth in the slab.54) Equations 5. qx and qy.47 and 5. particularly for orthotropic plates with significantly different flexural stiffnesses in the two directions.55) where M is the moment. this is not normally done as the resulting inaccuracy in the shear forces tends to be small. This may be quite unsatisfactory.47). However. This accounts for the first term of equation (5. m is found: (5.46). By substituting the curvature 1/R with κ rearranging.40) stated that mxy and myx are equal for materially orthotropic plates.3. Examining. 5. E the modulus of elasticity and R the radius of curvature. namely. 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. (Figs. This could be calculated in the grillage by finding the derivative of the torques in b3 and b4 with respect to y.Page 173 Equation (5. At point p. I the second moment of area.14) these are given by: (5.14. 5.48 gave expressions for the shear forces per unit breadth in the slab. but there is no account taken in the grillage analysis of the second term. the shear force Vb1 in Fig. the moment per and unit breadth. are found from the shear forces in the grillage members in a similar manner to the moments. the derivative of myx with respect to y.

As it is the relative values of stiffness that affect the calculated bending moments and shear forces. if this approximation is applied to both mx and my.2 for concrete). is relatively small in bridge slabs (approximately 0. it is common practice to ignore the second term in this equation. v.57) must be equated. A grillage member in torsion behaves according to the well known equation: (5.34) gives an expression for the moment per unit breadth in the X direction in the slab. This can be justified by the fact that Poisson’s ratio is small. for νand νin that equation gives: x y As Poisson’s ratio.58) where is the angle of twist.57) To achieve similitude of moments between a slab and the corresponding grillage. such an adjustment has very little effect on the final results.59) . The moment/curvature relationship then becomes: (5. Venant constant). they are both affected by the same amount.15 shows a portion of a beam of length δ in torsion. G is the shear modulus and J is the torsion constant (St. giving: A further simplification is made by equating the term below the line to unity. Figure 5. For an isotropic slab.56) and (5. there is only one value for E and νSubstituting E for Ex and ν . l is the length of the beam. the stiffness terms of equations (5. T is the torque. Further. 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.Page 174 Equation (5. 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.

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

. The preceding derivation of grillage member torsional properties is applicable to thin plates of rectangular cross-section where equation (5. the stiffness terms of equations (5. 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. 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. Equating this to jgril gives: (5. The torsion constant for the grillage member can alternatively be expressed in terms of the slab second moment of area: (5. Equation (5.43) for the torsional constant is valid. this is carried out automatically by the grillage program. in the slab and torques. 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. in the grillage members.43) gives an expression for the torsion constant of the slab.42) gives an expression for the twisting moment per unit breadth in the bridge slab: (5.63) and (5. This can clearly be achieved by adopting the same shear modulus and torsion constant in the grillage member as is in the slab.66) Although equations (5.63) where j gril is the torsion constant per unit breadth in the grillage member.66) are based on the grillage member having the same shear modulus as the slab.67) Typically. Equation (5.65) and (5. However. t: (5.Page 176 This can be rewritten in terms of torque per unit breadth. Equation (5.64) To achieve similitude of moments. mxy.65) ensures that the grillage members in both directions will have the same torsional constant per unit breadth.58) can be applied. t. it will not generally be necessary to specify Gxy for the grillage model.64) must be equated.65) where d is the slab depth.

66) for isotropic slabs. the second moments of area per unit breadth for the grillage and the slab are equated.69) In the slab. in the Y direction. It is common practise to use the equations developed for materially orthotropic thin plates to represent geometrically orthotropic bridges. equation (5. for both directions. i.36) shows that the moment mxy is arrived at by summing only the shear stresses in the horizontal direction (i. The reason for this lies in the definition of torsion in a beam and of moment mxy in a slab.68) predicts a torsion constant for the beam which is twice that predicted by equation (5. qx as illustrated in Fig. 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). 5.Page 177 For rectangular beams with depth d and a breadth of greater than 10d. . Further. 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.3 Grillage member properties—geometrically orthotropic slabs Equation (5.16 shows a portion of a beam of breadth b and depth d in torsion.3 times the slab depth from the edge so as to coincide with the resultant of the shear stresses.e. in both the horizontal and vertical directions. the torsional constant may be approximated with: (5. reproduced here.e. The vertical shear stresses are accounted for in the grillage in the same manner by the shear forces qy in the transverse beams. Similarly. Figure 5. The torque in the beam results from both of these shear stresses and is given by: (5.34). It has been recommended that the edge grillage members be placed at 0. E. 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. applies to materially orthotropic slabs: However. 5.17. only). most bridges have the same modulus of elasticity. many bridges are geometrically orthotropic. the shear stresses in the vertical direction are accounted for by the shear force per unit breadth.3.68) It can be seen that equation (5.

42). as stated in equation (5. . Then. However.: and: There is no facility in a grillage model to ensure that the two curvatures at a point are equal.40) stated that the two twisting moments at a point in a materially orthotropic slab are equal to each other.e. 5. if the same shear modulus and torsional constant are used in the two directions. i.16 Beam subjected to torsion showing resulting shear stresses Equation (5. curvatures in the orthogonal directions at a point will be approximately equal. If it is assumed that the same conditions hold for geometrically orthotropic slabs. in a fine grillage mesh. it follows from equation (5.41). the two twisting curvatures are the same.Page 178 Fig. Further.

5. 5.71) It can be seen that this equation is consistent with equation (5. G. The shear modulus for a slab made from one material. E.4 Computer implementation of grillages There are many computer programs commercially available which are capable of .70) Hambly (1991) recommends using such a single torsional constant for both orthogonal directions: (5.Page 179 Fig.66) for an isotropic slab. equation (5.70). νIt is generally calculated internally in computer programs using . that the twisting moments are equal: (5. and Poisson’s ratio.67).3.17 Slab with vertical shear stresses and corresponding grillage members with shear forces per unit breadth reproduced and adapted here as equation (5. is a function of the elastic modulus.

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

the effect of curvature in the Y direction was ignored (see equation (5. Equation (5. mx. The magnitude of these discontinuities can be reduced by choosing a finer grillage mesh. this discontinuity will be exaggerated.18 where it can be seen that the torque T in the transverse beam. such as where two longitudinal beams along the edge of a grillage meet only one transverse beam. There is no mathematical or physical principle in the grillage to make this so. significant differences can remain. As a result of this. excessively large discontinuities in moments. torques or shears indicate a grillage mesh which is too coarse. as are the corresponding curvatures in the two directions. having no other transverse beam to balance it. When deriving the properties of a grillage member parallel to the X axis. 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. 5. and requires the addition of more beams.34) gave an expression for moment per unit breadth.Page 181 the node. 5. This expression involved terms accounting for the curvature in the X and Y directions. Equation (5. The required moment is arrived at by averaging the moments on either side of the node. As was mentioned earlier. as other factors may also have an effect. This is illustrated in Fig. Where only three beams meet at a node. A similar simplification was made for my .40) stated that the moments mxy and myx are equal in a slab. Fig.57)). 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. corresponds to the discontinuity between the moments Mb1 and Mb2 in the longitudinal beams.18 Distribution of bending moment in a segment of grillage mesh showing discontinuity in moment (T b3=Mb1− b2 ) M . The opposite of this is not necessarily true. in the slab. However. The same phenomenon causes discontinuities in torques and shears. 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. which should be treated in the same manner.

However. the effective breadth decreases and the calculated concentration of shear adjacent to the support increases. 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.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. It will be seen from the recommendations given here that the traditional need for economy in the . if reasonably accurate results are to be obtained. It follows that. Thus the designer would design for the shear force calculated at a deck depth from the support. except for bridges with high skew. 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.3. Fortunately. whichever direction the beam lies in. If it were assumed that shear enhancement was sufficient to cater for local concentrations of shear near a support. and should be used in the context of good engineering judgement. there are sharp concentrations of shear intensity near each support. such as those relating to voided or skewed bridge decks.Page 182 Equations (5. These should not be viewed as absolute. then grillage member spacing would assume a much reduced importance. are given in Chapter 6.48) provide expressions for the shear forces per unit breadth. the magnitude of these moments is generally relatively small. 5. 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. qx and qy . some general recommendations are valid for most grillage models.3. 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. 5. When bridges are supported at discrete intervals. Some more specific recommendations. which is applicable to such a wide variety of structural forms. mxy or myx. This direct relationship between mesh density and the calculated maximum shear intensity means that. There is no account taken of the derivative of the twisting moments. if the grillage mesh density increases. the grillage member spacing has to be fixed near the support so that it gives the correct result.7 Recommendations for grillage modelling It is difficult to make specific recommendations on the use of a technique such as grillage modelling.47) and (5. The first of these equates the shear force per unit breadth qx to the sum of two derivatives: In the grillage.

grillage members should be located such that nodes coincide with the locations of supports to the bridge slab. Where possible. illustrated in Fig. Lines of strength may consist of concentrations of reinforcement. 5.Page 183 numbers of grillage members no longer applies. Transverse beams should have a spacing which is similar to that of the longitudinal beams. A choice of between one and three times the longitudinal spacing would be reasonable. However. The procedure of moving nodes locally to coincide with supports. should be avoided if possible. 2. Longitudinal grillage members should be provided along lines of strength in the bridge slab. as the computational power available to today’s engineers is well in excess of that available when earlier recommendations were made. Nonetheless. particularly in wide bridge slabs. location of prestressing tendons. 1. there is no advantage in providing excessive numbers of grillage members as the amount of output data will be excessive. Often this spacing will be greater than that of the longitudinal beams. should these exist. as this may result in skewed members which complicate the interpretation of results. (b) non-constant mesh spacing . 5. 4.19 Alternative grillage meshes near point supports: (a) local adjustment to mesh near supports to maintain constant spacing of members elsewhere. 3. no additional accuracy will be achieved. or precast beams in beam-and-slab bridges. Spacing will often be dictated by the location of supports or lines of strength in the bridge slab. A reasonable spacing of longitudinal beams is between one and three times the slab depth. as the magnitude of moment in the transverse beams is generally relatively small.19(a). The transverse grillage members Fig. significantly greater spacings are often possible without great loss of accuracy. There is little point in having longitudinal beams too closely spaced. and. beyond a certain point.

3d from the end. for example. 5.e. one with more longitudinal and transverse beams at a closer spacing.3d from the edge of the slab. in the bridge slab as illustrated in Fig. If the spacing of grillage members is in doubt. however. such as heavily reinforced diaphragms above bridge piers. the use of elastic springs to Fig.17.3d should be ignored. 6. The objective is to locate these beams close to the resultant of the vertical shear stresses. the breadth of slab outside 0.3d from the edge except for the end transverse members . increasing the mesh density tends (up to a point) to increase the accuracy. For bending moment results. Figure 5. It has also been recommended that. 5. Care should be taken. 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. so that the span length between supports in the grillage and the bridge slab are the same. 7. a check can be performed by comparing the output of a grillage with that from a more refined grillage. where d is the slab depth. 5.20 Segment of grillage mesh showing longitudinal members 0. Supports to the grillage should be chosen to closely resemble those of the bridge slab. should they exist. i. The second moments of area of these beams are calculated using the full breadth of slab in the normal way. 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.Page 184 should also be chosen to coincide with lines of transverse strength in the bridge slab. that this recommendation does not result in supports being placed in the wrong locations. .20 illustrates an example where a member is correctly placed more than 0. when determining the torsional constant of these longitudinal grillage members.

Some of these types of model are discussed in Chapter 7. The most common types of element used are quadrilateral in shape although triangular elements are sometimes also necessary. There is a risk that inexperienced users will attempt to analyse complex bridges without understanding the true nature and behaviour of the structure. Finite-element analysis is well known to bridge designers. Finite-element analysis is relatively easy to use and comprehend and.Page 185 simulate deformable bearings or ground conditions as discussed in Chapter 4. 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. Much development has taken place since this pioneering work and many texts now exist which give a comprehensive description of the method (see. 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. . 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 behaviour of which is known. when applied correctly. 8. and often more accurate than. where shear enhancement occurs. Such analyses are only necessary if it is specifically required to model in-plane effects. 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). 5. the grillage method. 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. This said. for example. No particular problem arises from using elements which allow in-plane deformations in addition to out-of-plane bending. 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. When applied to the analysis of slab bridge decks. 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). Some elements do not model in-plane distortion and consequently the nodes have only three degrees of freedom. the FE method involves the modelling of a continuous bridge slab as a finite number of discrete segments of slab or ‘elements’. A useful method of gaining familiarisation with a specific FE program is to begin by analysing simple structures. namely out-ofplane translation.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. 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. and then to progress to more complex structures. grillage analysis is much less reliable. and rotation about both in-plane axes. that of Zienkiewicz and Taylor (1989)). such as axial prestress. Closer to the support. is at least as accurate as. reasonable accuracy can be achieved with most sensible member spacings.

and the problem inherent in grillage modelling of torques per unit breadth not being equal in orthogonal directions does not arise.47) and (5. E x. it is advisable to determine whether or not shear forces are calculated correctly using equations (5. five elastic constants. Some x. in-plane distortion or a combination of both of these.4. and unlike a grillage analysis. If this is the case. although not all programs offer this facility. The twisting moment term can readily be accounted for. mx.1 Similitude between finite-element model and bridge slab The moments per unit breadth.35) give expressions for the moments mx and my in a thin plate. These are plate elements which can model out-of-plane bending. are output directly by FE programs.34) and (5. Isotropic bridge slabs In the case of bridges which are idealised as isotropic plates. These are generally given at the element centres and/or corners. The finite elements will satisfy this equation. ν and ν typically need to be specified. Equation (5. In FE analysis. Many programs provide the ability to determine these values at any arbitrary point using interpolation. Where the twisting moments are significant. When materially orthotropic finite elements are used. Equations (5. It was shown above that a grillage model does not take account of the derivative of the twisting moment. y. although in some programs it may not be. Finally.67). Gxy.48). program from these constants directly according to equation (5.42) gave an expression for the moments mxy and myx in a thin materially orthotropic plate. will account for the effect of curvature in one direction on the stiffness in the other direction.48) give expressions for the shear force per unit breadth in a thin plate. The finite elements will behave according to these equations. shear force per unit breadth can be calculated. E and νThe shear modulus. the second moment of area per unit breadth is given by equation (5. G. a check is useful to ensure that the values given are consistent with those at the neighbouring nodes. If this facility is used.33): .Page 186 5. is determined by the . The material properties of the elements are defined in relation to the material properties of the bridge slab. This is a significant advantage of the FE method over the grillage approach. programs assume a value for Gxy based on the values input for the other four elastic constants. E y. equations (5. the validity of this relationship should be checked for the particular plate under consideration.2 Properties of finite elements The types of finite element considered here are those used for the modelling of slab bridge decks. Each of these expressions involves terms relating to the curvature in both the X and Y directions.47) and (5. These expressions involve derivatives of the direct moment mx (or my ) and the twisting moment myx (or mxy). As the element is of constant depth. only two elastic constants need to be defined for the finite elements. my and mxy . 5.4.

Page 187 In a typical program. mx .72). Equation (5.77) The equivalent element depth can be calculated from equation (5. ix ≠ . 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. 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.e. the user simply specifies the element depth as: (5.35) gives a similar expression for my . i.: (5. for both directions. but only one depth can be specified.73) and (5. In such cases. . In most geometrically orthotropic bridge slabs.73) (5. 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: (5.76) and (5.72) Geometrically orthotropic bridge slabs Geometrically orthotropic bridge decks are frequently modelled using materially orthotropic finite elements. Equation (5. E slab. and .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. but there are two second moments of area per unit breadth. However.75) Substituting this into equations (5. there is only one modulus of elasticity.74) gives: (5.34) gives an expression for the moment.

80) diminishes as the variation in the elastic properties in the two directions increases.74) to give: (5. To determine if the influence of the shear modulus on the analysis is significant.42). Instead of arbitrarily equating the modulus of elasticity of the finite element in the X direction to the corresponding modulus of the slab.79) To be consistent with the equations for and i elem derived above.3. a similar expression can be determined by substituting from equations (5.80) and analysing again using a shear modulus of half this value. This would lead to alternative expressions to the above. 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. Alternatively.78) was derived by assuming an average value of the elastic moduli in the two directions and an average Poisson’s ratio. an arbitrary depth of finite element could be chosen (say. In such cases the shear modulus may need to be reduced.73) and (5. It was reported by Troitsky (1967). Consequently the accuracy of this and equation (5. equation (5. the authors would suggest analysing the orthotropic plate using a value predicted by equation (5.Page 188 For a materially orthotropic slab. has been suggested by Troitsky (1967): (5. These types of model are discussed further in Chapters 6 and 7.76) applies and equation (5. The lower value of 0.80) Equation (5. from the results of analysis and experimentation on steel orthotropic bridge decks. .5 and 0.79) becomes: (5. the moment/curvature relationship for the twisting moment. mxy . that the shear modulus given by the above expression may need to be reduced by a factor of between 0. 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.78) For a geometrically orthotropic slab with a single modulus of elasticity and Poisson’s ratio. Then. An approximate expression for the constant. the orthotropic nature of the plate might be better handled using a combination of elements and beam members or a three-dimensional model. As an alternative. the moduli in the Y direction could be equated.3 was reported to come from an extreme case where the flexural stiffness in the two directions varied by a factor of 20. Gxy. is given by equation (5.

2.21(b). In contrast to grillage modelling. it is often more convenient if a consistent mesh density is used throughout a bridge. More specific recommendations are given below and further guidance. 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.3 apply. Considering. 5. a typical program may be able to deal with elements of the type shown in Fig.22(a) where elements (1) and (2) are connected to each other at point P but are not connected to element (4). 5. for example.4. 5.21(a) but may give an inaccurate representation for the elements shown in Fig. 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. Some elements have mid-side nodes so that it is possible for example to have .3 Recommendations for finite-element analysis There are many commercially available computer programs for FE analysis of bridge decks. it may become necessary to limit the number of elements. but can easily be modified where this is not the case. the finite element response to applied loading is based on an assumed displacement function. 5. Once again these should not be viewed as absolute. Unlike the grillage method.Page 189 The expressions given above relate to bridge slabs with the same modulus of elasticity in both directions. more elements tend to result in greater accuracy although this is by no means guaranteed. quadrilateral elements with nodes at the four corners. 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°. In general. These should tend towards squares in the case of quadrilateral elements and towards equilateral triangles in the case of triangles. This type of model is useful where in-plane effects (such as axial prestress) are to be considered. considerable deviation from these shapes may be permissible and the documentation provided with the program should be consulted for specific recommendations. Obviously. However. Regularly shaped finite elements should be used where possible. Many engineers use denser meshes of elements in those parts of a bridge where bending moment changes rapidly such as near an interior support. and quite often the program will allow the user to define elements which do not conform to this shape. it is difficult to make specific recommendations relating to FE modelling of bridge slabs but some general guidelines are given here. The implementation of the FE model is carried out in a similar manner to a grillage and many of the comments in Section 5. 1. In the absence of information to the contrary. Mesh discontinuities should be avoided. These may occur when attempting to refine the mesh such as in Fig. is given in Chapter 6. applicable to voided and skewed bridge decks. As with grillage modelling. This function may be applicable to elements of a certain shape only. as some programs may not be able to deal with excessive numbers.

(b) good arrangement . (b) potentially problematical shapes Fig.Page 190 Fig. 5.22 Meshes of finite elements at transition between coarse and dense mesh: (a) potentially problematic arrangement. 5.21 Possible shapes of quadrilateral finite elements: (a) generally good shapes.

6. However. The vectors representing the moments are resolved to determine the moments on the face AB. As bending moment is a vector. it is useful to compare the output of a model with the chosen mesh density to that of a model with a greater density. Shear forces near points of support in finite-element models tend to be unrealistically large and should be treated with scepticism. 1997).5 Wood and Armer equations Much of this chapter has been concerned with methods of analysis of slab bridges. The spacing of elements in the longitudinal and transverse directions should be similar. AB. 5. 5. for example.23(b) using double headed arrows to denote bending moment. is introduced where N is normal to the face AB and T is parallel (tangential) to it. Elements should be located so that nodes coincide with the bearing locations. the three components can be combined using vector addition in a manner similar to the concept of Mohr’s circle of stresses. as can to be seen in the figure. Supports to the finite-element model should be chosen to closely resemble those of the bridge slab. mx. If mesh density is in question. A small segment of slab is illustrated in Fig. 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.Page 191 elements (3) and (4) connected to the mid-side node of element (1) at Q. .23(a) and the possibility is considered of failure on a face. The length of the face AB is l and. mxy and myx. All vectors are resolved parallel and perpendicular to AB in Fig. . This is generally easily achieved. The results of such analyses give three components of bending moment at each point. 5. if excessive. This may involve. 5.24. 5. my and mxy . 4. a second axis system. the use of elastic springs to simulate deformable bearings or ground conditions as discussed in Chapter 4. Similar results from both would suggest that the mesh was sufficiently dense. Resultant moments can be calculated at any angle of orientation and can. The twisting moments per unit length. This will be complied with if the first recommendation is adhered to. result in yield of the slab at any such angle. 5. 7. are also illustrated in this figure.22(b) where mid-side nodes are not needed and all elements are connected. For convenience. the projected lengths on the X and Y axes are l sin θ l cos θ and respectively. A mesh is shown in Fig. where the moment is about the axis of the arrow. N–T. 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. This section addresses the design problem of how the engineer should calculate the moment capacity required to resist such moments. at an angle of θ the Y axis. . 3. The direct moment per unit length on AB is denoted mn and the twisting moment per unit length is denoted mnt. 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.

Page 192 Fig. 5.83) and: (5. (b) applied bending and twisting moments Considering components parallel to AB first: (5.23 Segment of slab: (a) geometry.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: (5.84) . 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:

mx=190.89) and (5. my=80 and mxy=20.2: 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 (5.93). 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. It is required to determine economical moment capacities given that providing costs twice that of providing . for example.94) This can be used to find an economical value for in equations (5.96) These are known as the Wood and Armer equations (Wood. their product is positive giving: (5. Hence.90)) that and mxy were of the same sign.93) become: (5. The cost of providing moment capacity in the two coordinate directions may not necessarily be equal as a bridge may. equation (5. . then ρ and equations (5.93) Any value for can be selected by the designer and these equations used to determine the minimum required moment capacities.Page 198 It was established earlier (by comparing equations (5. be prestressed in one direction and reinforced with ordinary reinforcement in the other.92) Similarly. In general.91) becomes: (5.92) =1 and (5. Example 5. 1968).95) (5. If the cost of providing moment capacity is the same in both directions.

i.92) and (5.e. Equations (5.Page 199 As cost is proportional to critical angle is defined by: the constant. ρis 0.93) then give the required moment capacities: .5 and the minimum cost value for the .

In Chapter 7. For this reason. the behaviour of bridge slabs is considered. 6. They can. more complex non-planar methods of analysis are considered.Page 200 Chapter 6 Application of planar grillage and finite-element methods 6. both of these planar methods of analysis are used to model a range of bridge forms. with adaptation. For certain bridges. In this chapter. Planar methods are among the most popular methods currently available for the analysis of slab bridges. grillage and finite-element methods. they can also be considerably more complex and can take much longer to set up. non-planar models are considerably more accurate than planar models. both of which consist of members lying in one plane only. Two methods of analysis are introduced. This will be demonstrated in the following examples.2 Simple isotropic slabs When bridge slabs are truly planar. planar grillage and finite-element models are at present the method of choice of a great many bridge designers for most bridge slabs. However. Further. their basis is well understood and the results are considered to be of acceptable accuracy for most bridges. it is a simple matter to prepare a computer model following the guidelines specified in Chapter 5. .1 Introduction In Chapter 5. be applied to many different types of slab as will be demonstrated.

The end rows of transverse members are taken through the centres of the bearings.3.1: Grillage model of two-span right slab A two-span bridge deck is illustrated in Fig.5 m which gives a ratio of transverse to longitudinal spacing of between 1. 6.2 and 1. Figure 6. The deck is supported on four bearings at either end and on two bearings at the centre as illustrated in the figure.5.1 Plan view of two-span bridge Fig.1: (a) plan. A combination of fixed.3 times the depth from the edge of the slab. The transverse members have been placed at a spacing of 1.8 m depth.1.2 (a) shows a convenient grillage mesh for this bridge deck. 6. (b) section .Page 201 Example 6. free-sliding and guided-sliding bearings is used so that the bridge can expand or contract freely in all directions in plane. The longitudinal members have been placed along the lines of the bearings. 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. with an additional line at the centre of the deck. 6. a row of longitudinal members has been placed at a distance of 0.2 Grillage mesh for bridge of Fig. 6. As recommended in Section 5. It is to be constructed of prestressed concrete and is to have a uniform rectangular cross-section of 0. Fig.

5 as the slab extends 0.3.2 (b) shows a cross-section of the slab with the grillage members superimposed.63 m was used.0491 0. Similarly.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.1 and Fig. when determining the value of the torsion constant of the longitudinal members in rows R1 and R9. 6. 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. R6 R5 Transverse Members End members All intermediate members 0.1. The two rows of . Example 6. 6.Page 202 Table 6.0537 0. in keeping with recommendation number 6 of Section 5.1280 Figure 6.0470 Torsion constant (m4) 0.0964 0.0938 0.0483 0. For the transverse end members.0862 0. The breadths of the elements are chosen such that nodes coincide with the locations of the supports.2. These values are presented for all of the grillage members in Table 6.87− 0.3d=0. R7 R4. It can be seen that this breadth is taken to be from midway between adjacent members on either side.0470 0.3 shows a convenient finite-element mesh.1 Second moment of area (m4) Longitudinal members R1.0938 0.0534 0.0371 0. this is reduced by 0.1 Grillage member properties for Example 6.1.0981 0.2: Finite-element model of two-span right slab A planar finite-element model is required for the bridge deck of Example 6.7.24)=0. This is used to determine the breadth of slab attributable to each longitudinal grillage member. the breadth is 1. 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 (5.2 (b). 6.0640 0.24 m for the calculation of the torsion constant. R9 R2. However.5/2+0. R8 R3. Figure 6.5 m past the centre of the bearing. a reduced breadth of (0.

then the increased stiffness which they provide generally needs to be considered. This is a somewhat arbitrary choice. where the edge cantilever is relatively short or stocky . If they are made integral with the deck. a similar degree of accuracy could be expected.4. In slab bridges.8). In such a case. equal to the average breadth of the elements. such as those illustrated in Figs. It is not necessarily conservative to ignore the additional stiffness provided by them.3 Edge cantilevers and edge stiffening Slab bridge decks often include a portion of reduced depth at their edges known as an edge cantilever. to carry a protective railing. are often included at the edges of the slab. the neutral axis will not remain straight as the upstand tries to bend about its own axis. the appropriate stiffness is determined by first finding the neutral axis location for the complete deck. Only decks where the neutral axis remains substantially straight are considered here. As this is an isotropic bridge slab. may be precast to ensure a good quality of finish.1 elements at each edge of the model could be replaced with one row of 1. the upstand may not be integral with the bridge deck and can simply be considered as an additional load on it. Upstands or downstands. The properties of each part are then calculated about this axis. say. 6. causing the bridge neutral axis to rise.1. the only geometric property which has to be assigned to the elements is their depths. These will be similar to those illustrated in Fig. 6.Page 203 Fig. Cross-sections of typical slab decks with edge cantilevers are illustrated in Fig. This type of construction is chosen partly for its reduced self weight and partly for its slender appearance (see Section 1. Figure 6.2 m which is equal to the breadth of the widest element. The length of the elements along the span of the bridge was chosen as 1. 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.4(c) and (d).8 m which is equal to the actual depth of the bridge slab. finding the location of the neutral axis may not be straightforward. 6.3 Finite-element mesh for bridge of Fig. In some bridge decks. These are frequently important aesthetically and. in the case of concrete bridges. 6. 6.5 shows the cross-section of a deck with a long slender edge cantilever with an upstand at its edge. All of the elements are assigned a depth of 0. or simply for aesthetic reasons. but the extra number of elements in the model chosen is not considered to be excessive. and had the length been taken as.5 m breadth.4. In such cases. 6. Bridge decks of this type are discussed further in Chapter 7. either to stiffen the edge. As for Example 6. E=35×10 6 kN/m2.

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

is placed at a distance of 0.7.3 times the average depth of cantilever. R5 and R6 (and R8 and R9). The spacings of longitudinal grillage members is given in Fig. is located at the centre of the portion of cantilever which it represents. Row R1. Figure 6. R2.Page 205 Fig. but that chosen here seems reasonable.7(a) shows the divisions chosen and the corresponding grillage members.3. Note that row R4 is not exactly at the centre of the portion it represents.6 Bridge deck of Example 6. of grillage members are located to coincide with the supports to the bridge deck. and middle row.7(b). is placed at a distance of 90 mm from the edge of the cantilever. . 6. R3. • The second row of grillage members from the edge. • The third row of members from the edge. (b) plan The cross-section is divided into a number of segments. This is in keeping with recommendation number 6 of Section 5. 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. The location from which this distance is taken is somewhat arbitrary.3×1200=360 mm) from the midpoint of the sloping edge of the main deck.3 times the depth of the deck (0.3 (dimensions in mm): (a) section. This distance corresponds to 0. R7. these members represent a portion of bridge slab of breadth 1000 mm and they are located at the centre of that portion. In each case. • Two rows of grillage members. each of which is represented by a row of grillage members. R4. • The outermost row of grillage members. • The fourth row. are chosen between the supports. 6.

each row of longitudinal grillage members is considered separately.7 (c) illustrates a plan of the grillage mesh with dimensions in mm. 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 . between 1:1 and 1:1.7 Grillage model (dimensions in mm): (a) cross-section showing grillage members and corresponding segments of deck. However. Twenty one rows of transverse members with a spacing of 1000 mm were chosen. For this example.27. (b) schematic of cross-section showing spacing between members.Page 206 Fig. This is a very dense mesh having a spacing less than the slab depth. (c) plan of mesh Figure 6. The second moment of area about the centroid (of the bridge) of each portion of deck is determined. 6. Due to the variation in depth between rows R2 and R3. the transverse members between these rows have been modelled as two separate members with a row of nodes where they join. it gives a good longitudinal to transverse spacing ratio.

019 0.143 0.146 Torsion constant (m 4) 0.010 0.290 0.144 0. running from the edge as far as the row of nodes indicated in Fig. These are labelled Tc in Fig. 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.278 . The depth of these members is taken as the average depth of the cantilever.178 0. are taken about their own centroids as they will bend (transversely) about their own centroids.2. R12 R3. R8. The second group are those in the main portion of the deck and account for all of the other transverse members. R6.021 0. The second moment of area of the transverse members in the cantilever.144 0. The transverse members are divided into two groups.7. 6. R13 R2.002 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). i. The results are presented in Table 6.2 Grillage member properties for Example 6.131 0.7(c).261 0. Tc. R7.71) as this is an orthotropic deck : Table 6. R9 Transverse members Tc—End members Tc—Intermediate members Tm—End members Tm—Intermediate members 0. Tm. 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 6.110 0. are also calculated about their own centroids as it is about these that they will bend. R11 R4. 300 mm.2. The torsion constants for the members are determined in accordance with equation (5.002 0.Page 207 of deck.013 0. These are labelled Tm in the figure.e. R10 R5. 6.3 Second moment of area (m4) Longitudinal members R1.029 0.034 0. The first group are those in the cantilever portion.

The torsion constant per unit breadth of the transverse grillage members.2) is given by: The second moment of area per unit breadth of the transverse members is 0. is given by: Considering next the longitudinal members in row R2 and the transverse members Tc. Tc.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. an approximation is made by taking an average value for the torsion constant per unit breadth of the transverse members. the condition of Section 5.2) is given by: Therefore the torsion constant per unit breadth of the longitudinal members. the torsion constant per unit breadth of the longitudinal members. At this stage. To apply this equation. the X direction is arbitrarily chosen as the longitudinal direction.Page 208 where and are the second moments of area per unit breadth in the X and Y directions respectively. Tc. However. and the transverse members. In doing this.2) is given by: . 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 6.002 m3. R1. is therefore: Considering the longitudinal members in row R3 and the transverse members Tm. the average value is considered acceptable. R2. Hence. as the two distinct values are very close. Considering the longitudinal members in row R1 and the transverse members Tc. 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. Tc. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6.3.

the division of the deck for the finite-element model varies somewhat from that of the grillage. Example 6. the breadth is reduced by 0. 6. 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.3×1. and the transverse members. For the longitudinal members in rows R3 and R11. 6. These values are given in Table 6. have the same second moment of area per unit breadth (with reference to Table 6.Page 209 The second moment of area per unit breadth of the transverse members is 0. R4 to R10.8(b) shows a cross-section through the finite-element model.3 and Fig.6(a) is divided into a number of segments in a similar manner to the grillage model.9=0. and the longitudinal members in rows R1 and R13. The depths of the elements have not been drawn to scale in this figure. . the torsion constant per unit breadth of the longitudinal members. the breadth is reduced by 0.3×0. 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. For the end transverse members.6.3=0.8(c) shows a plan of the finite element model with rows of elements labelled r1 to r14. For the end transverse members. Figure 6.2) which is: Hence. This results in 20 elements in each of the 14 longitudinal rows.09 m. As the nodes form the boundaries of the elements and the location of the supports must coincide with nodes. Tm. is given by: This value is adopted for longitudinal members R4 to R10. Tc. the breadth is reduced by 0. R4 to R10. The cross-section of Fig.27 m. and the transverse members.8(a) shows the division of the deck and Fig.2=0.144 m3 and therefore the torsion constant per unit breadth of the longitudinal members. R3.3×0. Tm. Tm. The length of the elements (in the longitudinal direction) is taken as 1000 mm.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. the need to average two dissimilar values of torsion constant was avoided.2. Figure 6. The other longitudinal members.36 m. is given by: This value is adopted for the longitudinal members in row R3.

r2. (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. 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. In the case of the elements in row r3. r13 and r14). In the Y direction. are determined for each portion of the bridge deck. 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. these are calculated about the centroid of the bridge which was seen in Example 6. In the X direction.8 Finite-element model (dimensions in mm): (a) cross-section showing division of deck into elements. (b) schematic of cross-section showing breadths of elements.3. 6. . In the case of the elements representing the edge cantilevers (rows r1. The second moments of area per unit breadth for each row of elements are given in Table 6. The second moments of area per unit breadth. and .Page 210 Fig. the transverse stiffness is based on the average depth of that portion of cantilever. it is difficult to determine the transverse stiffness as the depth varies significantly.3 to be located 563 mm below the top surface.

is calculated using equation (5.989 Ec 0. then the equivalent depth. r9.4 Finite element row number r1. it is common practice to model the voided slab using the same methods as are used for . 068 Ec 0. r12 r4. The variation of second moment of area in the two directions is allowed for by specifying two different elastic moduli.2 for concrete. These values are also given in Table 6.1440 Ec 0. r10 (m 3) 0. Arbitrarily choosing the elastic modulus in the X direction. terms of the elastic modulus of the concrete. the elastic modulus and the second moments of area per unit breadth. 356 Ec 0. r8.9 shows a cross-section through a typical voided slab bridge deck with tapered edges. to be equal to the elastic modulus of the concrete.732 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 (5.1456 (m3) 0.0490 0.109 1. as the second moments of area vary in two orthogonal directions. r11 r5 r6.204 0 . 6.876 1. delem .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.3.027 Ec 0 . within the formwork before casting the concrete.0036 Ec 0. .80) by substituting values for the Poisson’s ratio. values of were arrived at for each row of elements. In the finite-element program.1456 0. When the void diameter is less than about 60% of the slab depth.0833 Ec 0. The shear modulus. Figure 6.1138 0.Page 211 Table 6. 106 Ec 0.0561 0. r7. 414 Ec The bridge deck is geometrically orthotropic.1440 Ec delem (m) 0. 414 Ec 0.3. it is modelled as materially orthotropic with a single value for element depth. in The elastic moduli in the two directions and the equivalent depths of each row of elements are given in Table 6. r14 r2. usually made from polystyrene.204 1.76)): Equation (5.989 Ec 0. It is common practice to discontinue the voids over the supports which has the effect of creating solid diaphragm beams there.77) then gives an expression for the elastic modulus in the Y direction. Ec : . Assuming a Poisson’s ratio of 0.838 0. These are created by placing void formers.0013 Ec 0.064 Ec 0. Ec .3 Finite-element properties for Example 6. . r13 r3.

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

dv (Fig.55 0.1) does not take into account the spacing of the voids as the authors maintained that this was not a significant factor. to that of solid slab.6 or less. 6.75 0.87 0.11 Cross-section through segment of voided slab bridge Table 6.68 0.77 0.78 0.70 0. Examination of equation (6.56 0. This is quite often a reasonable assumption when considering transverse bending. When the void diameter to slab depth ratio is 0.65 0.79 0.6.66 0. iv-slab.86 0. This equation assumes that the centre of the voids and the deck centroid (for transverse bending) are located at mid-depth.Page 213 transverse second moment of area. Clearly this equation is only applicable to slabs with a sensible void spacing.64 0.76 0. Bakht et al.71 0.72 0.82 0.86 0.9 0.85 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. 6.7 0.6 0.70 0. d. the transverse stiffness can be approximated as being equal to the longitudinal stiffness.69 0.58 0.90 . islab (from Bakht et al. and the diameter of the voids.8 0.84 0.48 0.11): (6.75 0.1).1) shows that the presence of the voids reduces the transverse stiffness by only 12% for a ratio of 0.64 0.74 0.81 0.5 0.62 0.88 0.1981) 0.45 0.82 0.84 0.1) Equation (6.80 0.60 0.4 Ratio of torsional stiffness of voided slab.61 0. (1981) recommend using the method of Elliott which gives this quantity in terms of the depth of the slab. Fig.80 0.90 0.85 0.51 0.89 0.

in most practical cases.5 and 6.5 . Fig. jv-slab.12 Cross-section through bridge of Examples 6.12 shows the cross-section of a prestressed concrete bridge deck which incorporates circular voids along its len gth. It was suggested that the values given in Table 6. Bakht et al.4 are only applicable to internal voids in an infinitely wide slab because those at the edges possess much lower torsional rigidities. (1981) recommend using the method of Ward and Cassell.13 Grillage mesh for bridge of Example 6. However. conclude that. The deck spans 24 m between the centres of supports and is supported on four bearings at either end as illustrated in the figure.5: Grillage model of voided slab bridge Figure 6. Bakht et al. 6. The layout and member properties are required for a grillage model.65) or (5. This gives the values presented here in Table 6.6 Fig. 6. jslab can be determined from equations (5. For a grillage model. 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. j slab.71) and Table 6. Thus the total bridge is 25 m long consisting of 23 m of voided section and two 1 m diaphragms . Example 6. The neutral axis passes through the centroid of the deck which is located at mid-depth as the voids are located there. to that of a solid slab of the same depth.Page 214 For the torsional stiffness of voided slabs per unit depth.4 for the ratio of torsional stiffness of the voided slab jv-slab. The voids stop short at each end forming solid diaphragm beams 1 m wide over the supports.4 can then be used to determine jv-slab.

1. It is not considered appropriate to locate these grillage members at 0.1): Hence. The longitudinal direction is taken to be the X direction. the slab is treated as an orthotropic plate and the properties of the longitudinal and transverse members are determined separately.14. The internal longitudinal grillage members represent the portion of deck illustrated in Fig.13 shows a suitable grillage mesh.: 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. By using this arrangement.3 times the depth of the slab from the edge as this location is within the void. 6.e. 6. the second moment of area is: . for the internal transverse members.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. i. As the void diameters are in excess of 60% of the slab depth.5 m apart. The longitudinal members are located midway between voids. The transverse grillage members are located in 17 rows. the supports coincide with the locations of nodes in the grillage mesh.Page 215 Fig. 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.

67.5 m wide. At the ends.Page 216 For the 1m wide end diaphragms.5 m. the next row of transverse members.3 m wide in order to make up the correct total length. adjacent to the diaphragm.12. will be 1.4. two transverse rows of elements. 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. The transverse rows of elements adjacent to the diaphragms at each end are 1. For convenience. a mesh consisting largely of 1. Each longitudinal row of elements represents a strip of the deck from midway between one void to midway between the next. Both the ratio dv/sv and dv/d are 0.2 m square elements is chosen.71): Example 6.65) 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. Interpolating in the table gives a ratio for the torsion constants per unit breadth of: Taking equation (5. 6. are used to represent the diaphragm. the second moment of area is simply: As the diaphragm is only 1 m wide and the transverse members are spaced at 1. The torsion constant per unit breadth for the diaphragm is given by equation (5.15. The second moment of area per unit .5 and Fig. each 0. as illustrated in Fig.

as the second moments of area (rather than the moduli of elasticity) are different for the longitudinal and transverse directions. de . then equation (5. 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.1) gives: The slab is geometrically orthotropic. the second moment of area per unit breadth is: For the transverse direction.76) implies a depth of element of: Equation (5.2 m wide strip of the deck.6 breadth in the longitudinal direction can be found by considering a 1. Selecting the modulus of elasticity in the X direction. it is necessary to calculate a single equivalent value for slab depth. E x. 6. To model this as a materially orthotropic plate.Page 217 Fig.15 Finite element mesh for bridge of Example 6. equation (6. equal to the modulus for the concrete.77) gives an expression for the elastic modulus in the Y direction: .

16 (a) and (b). In addition to this. the beams generally act alone and must be capable of carrying their self weight. 6. 6. for example. The slab acts to transmit applied loads to the beams by spanning transversely between them. this gives: The diaphragm beams are solid so the corresponding elements are 1. it is important that the slab be idealised correctly in the model as.80): Taking a Poisson’s ratio of 0. The extent of this load sharing is largely dependent on the stiffness of the slab.67). On completion. the most obvious being the casting of an in-situ concrete slab on steel or precast concrete beams as shown in Fig. During construction.Page 218 where E v-slab is the modulus of elasticity of the concrete in the voided slab. the slab provides a means for load sharing between longitudinal beams. This phenomenon is indicated in Fig.17. a precast concrete slab or even a completely in-situ beam and slab as illustrated in Fig. 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. 6. 6. Beam and slab decks may be formed in a number of ways. Many other methods exist. Wide diaphragms also serve to improve the shear . 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 main load-carrying component of a beam and slab deck is the longitudinal spanning beams.2 m thick and have moduli of elasticity in both directions equal to that of the concrete. The shear modulus for the diaphragms is given by equation (5. Therefore they can be analysed by similar methods to those proposed for slab decks in the preceding sections. the structural action of these decks is considered to be two-dimensional.2.5 Beam and slab bridges Beam and slab decks are used for a wide variety of modern bridges. Consequently.16 (c). Transverse diaphragm beams can be used to provide additional load sharing between longitudinal beams. 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. 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. such as steel beams with a composite steel and concrete slab. The shear modulus is calculated from equation (5.

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

each acting about its own axis. 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. 6. . Section 5.18. 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. Solid diaphragm beams. the section properties for beam and slab decks are generally calculated about the centroid of this composite section.7 stated that transverse member spacing should be between one and three times the longitudinal member spacing.3s for L-sections as illustrated in Fig. in beam and slab construction. Unlike slab decks. The properties of the transverse grillage members should be derived from the properties of the relevant diaphragm beam or slab as appropriate. In addition. This approach is justified on the basis that. supports are normally provided directly beneath the beams. The properties of the longitudinal grillage members are determined from the properties of the actual beams and the portion of slab above them. where s is the spacing between beams. Example 6. there will be a much greater variation in the depth of the neutral axis than in slab bridges.3. 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 spacing is also recommended for beam and slab bridges although greater spacings are possible without significant loss of accuracy. transverse members are required to represent the transverse stiffness of the slab.Page 220 Fig. 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. This generally complies with the need to locate beams at the supports as.18 Effective flange width of diaphragm beam: (a) plan at end. The elastic modulus of the precast beams is 34 kN/mm 2 and that of the in-situ slab is 31 kN/mm2. 1 m wide. (b) section through L-beam positioned at the location of the longitudinal beams.or L-section in shape. The slab will act as a flange to such beams making them T. not about the centroid of the whole bridge.7: Grillage model of beam and slab bridge Figure 6. due to the low stiffness of the slab. Hambly (1991) suggests an effective flange breadth of bw+0. Transverse grillage members should clearly be placed at the location of all diaphragm beams.

for a rectangular section according to Ghali and Neville (1997) is: (6.0265 m2 Height of centroid above soffit = 0.19 Beam and slab bridge deck: (a) cross-section.2) . the beam cross-section is approximated as two rectangles as illustrated in Fig.19 (b) shows the exact dimensions of the precast beam. The section properties of the precast beam are generally given by the manufacturer.20. The torsion constant.347 m The torsion constant is generally not given and must be determined by the analyst. in this case. 6. Figure 6. (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). but to factor the stiffness of the slab by this modular ratio.Page 221 Fig. J. the properties are: Area = 0. 6. 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. For the purposes of determining the torsion constant.374 m 2 Second moment of area = 0.

5%). both . Figure 6.20 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. Transverse members are positioned at each end to model the diaphragms. For the interior longitudinal members.Page 222 Fig. A longitudinal grillage member is positioned at the location of each Y-beam.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). 6. ‘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. consideration need not be given to in-plane horizontal movements at this stage. 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. 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. Supports are located at the ends of each longitudinal beam (other than the dummy beams). then these beams should be assigned very small section properties relative to those used elsewhere in the grillage (say. This is a convenient method for applying loads such as those due to parapet railings. 0.21 shows a suitable grillage layout for this bridge deck. This gives a transverse to longitudinal member spacing ratio of 2:1 which is acceptable. Applying this equation to the rectangles of Fig. Some grillage programs allow the definition of ‘dummy’ beams. 6. As the grillage model is planar. Additional transverse beams are located at 2 m centres between these to represent the transverse stiffness of the slab. If this is not the case.

6. The torsion constant of the slab is determined using equation (5. The stiffness of the slab is reduced by factoring it by the modular ratio. Hence: Each edge longitudinal member is similar to the interior members except for a 0. 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. This raises the centroid above that for the interior members. Summing moments of area about the soffit gives: .21 Plan view of grillage mesh taken about the common centroidal axis of the section.65).3 m 2 upstand.Page 223 Fig.2×0.

Page 224 Fig. . 6. the properties are determined in the usual manner. For the second moment of area: The torsion constant is: The slab acts as a flange to the diaphragm beams. For the slab bending about its own axis.22 Section through end diaphragm beam Hence the second moment of area of the edge section is: For the transverse members. 6.22. The recommended flange breadth is the sum of the web breadth plus 0.3 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.

(b) conventional grillage model where slab has excessive transverse span. then the slab can inadvertently be modelled as having an excessively long transverse span. where the transverse grillage members have been subdivided to include much stiffer portions at their ends. 0. If the web width at the top of the longitudinal beams in a beam and slab deck is large relative to their spacing.Page 225 This leaves 0. This would lead to an excessively flexible slab which in turn would lead to the incorrect modelling of load sharing between the U-beams. Figure 6.23 Transverse modelling of decks with wide flanges: (a) in-situ slab on precast concrete Ubeams. 6. (c) improved grillage model . 6.2 m of which is deemed to be bending about its own axis. 6.23(b) shows a grillage model with longitudinal grillage beams for the Ubeams and transverse beams spanning between them representing the slab. One possible solution to this is shown in Fig.23(c).2 Finite-element modelling In finite-element modelling of beam and slab decks. The second moment of area is thus: The torsion constant is calculated allowing for 0. Figure 6. a combined model is generally used which represents the slab with finite elements and the beams with grillage Fig.65): The modulus of elasticity for in-situ concrete is used for the diaphragm beams.5 m of slab to be accounted for in the diaphragm stiffness. It can be seen from this that the span of the slab in the model is too long.5 m of flange from equations (6.5.2) and (5.23(a) shows a deck consisting of a concrete slab on precast concrete U-beams.

Page 226 members. One of two approaches can be taken.19. This is generally straightforward to implement and follows the recommendations made for slab bridge decks. In the first approach. Supports are provided at the ends of each longitudinal grillage member. 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. 6. the second moment of area of the combined section is: Fig.24 Combined finite-element and grillage mesh . They are assigned a modulus of elasticity and a Poisson’s ratio equal to those of the concrete in the slab.7 and Fig. The finite elements continue to the edge of the deck resulting in a row of elements 0.7. The stiffness of the slab which has already been applied through the finite element is subtracted. 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. In the second approach. Example 6. The beams are then modelled by grillage members with the properties of the actual beams excluding the contribution of the slab. From Example 6.16 m which is equal to the depth of the slab. The modulus of elasticity and Poisson’s ratio for the beams are used for these members. The finite elements are assigned a thickness of 0. the slab is modelled using isotropic elements which are assigned a thickness equal to the depth of the actual slab.5 m wide at each side. For the longitudinal grillage members. 6. 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. They are also assigned the elastic properties of the slab. the slab is modelled using orthotropic finite elements with the true transverse and longitudinal properties applied in both directions. Care should be taken when determining the properties of the finite elements representing the slab. Figure 6. Grillage members are used for each of the Y-beams and for each of the end diaphragms.24 shows a suitable finite-element mesh incorporating grillage members longitudinally.8: Finite-element model of beam and slab bridge A finite-element model is required for the beam and slab bridge of Example 6.

the second moment of area for the end diaphragms in the grillage model was calculated as (refer to Fig. As the slab is represented by the elements.22): For the finite-element model.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.7 by adding the individual torsion constants of the Y-beam and slab. the torsion constant to be assigned to the grillage members is simply that of the Ybeam: In Example 6.25).65): 6.2) and (5. Hence. 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 elements are present up to the centre of the diaphragm to represent the transverse stiffness of the slab about its own axis.Page 227 The second moment of area of the 0. such an approach results in a great quantity of reinforcement and has been shown to be .5. 6. However. 6.7. From equations (6.16 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.

27(c).26 shows a number of commonly used cellular deck forms. 1992) allows for the provision of much less reinforcement than would be found by an assumption of bending.27(d). The beams have a considerable lateral stiffness and have the effect of confining the slab. as indicated in Fig. This distortion is caused by the localised bending of the webs and flanges of the individual cells. The result is that load is transferred from the slab to the beams by arching action rather than bending action alone. can also be considered to be of a cellular form. However. 6. 6. In two reported cases (Bakht and Jaeger. the slab depth to beam spacing ratios were 1:12 and 1:13. Clearly the provision of transverse diaphragms along the span of a cellular deck will significantly reduce the degree of transverse distortion. 6.6 Cellular bridges Cellular bridge decks are formed by incorporating large voids within the depth of the slab. The first two of these are longitudinal and transverse bending. Voided slab bridges. with large diameter circular voids. The third form of behaviour is twisting.27(a) and (b). as was discussed in Section 6. is transverse cell distortion. The most common type are box girder decks. as indicated in Fig. The principal factors affecting the distortion are the dimensions of the cells relative to the deck depth. the Ontario Highway Bridge Design Code (OHBDC. To account for observed arching action. 1997). In these cases. the stiffness of the individual webs and flanges. and the extent (if any) of transverse bracing to the cells.25 Detail of section in beam and slab deck quite conservative. 6.5. which characterises cellular structures. with single or multiple rectangular cells. The fourth form. It is the transverse distortional behaviour that makes the analysis of cellular decks different from other forms. There are four principal forms of structural behaviour associated with cellular bridges. 6. Figure 6. The behaviour is similar to that observed in Vierendeel girders. Canadian bridges have been built without any transverse slab reinforcement but using steel straps to guarantee confinement. alternative methods are available for their analysis which are generally more convenient. as illustrated in Figs.Page 228 Fig.4. .

(d) transverse distortion .27 Behaviour of cellular decks: (a) longitudinal bending. 6.Page 229 Fig. (b) transverse bending. 6. (c) twisting.26 Sections through alternative cellular bridge decks 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.6.28(b). where d is their thickness. In this method. the deflection due to flange distortion is: (6.Page 230 6.4) Fig. from equation (6.28 Distortion of single cell with stiff webs: (a) applied loading. 6.3) where I is the second moment of area and E is the modulus of elasticity. Hence. then the distorted shape of the cell is as shown in Fig. 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. 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. 6. (b) distorted shape . then the load acting on each can be taken as P/2. Figure 6. fixed against rotation at both ends and subjected to a vertical force P/2. If the flanges are of equal thickness.3).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. If it is assumed for now that the webs are stiff and that transverse distortion is caused by bending of the flanges only. The method is illustrated below by means of an example. is: (6. The vertical deflection due to the bending of a beam of length l. the deck is idealised as a grillage of beam members in the usual manner.

6) In this example. an expression for the required shear area per unit breadth of a shear flexible grillage member is found: (6. for most structures. subjected to a vertical load per unit breadth of P at its free end is: (6. The second term is the deflection due to shear deformation which. 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. In practice.7) Details of the derivation of this formula are given in Appendix D. 6. Fig.5) where G is the shear modulus and as is the shear area of the section per unit breadth. 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: (6.29 Cross-section through cellular deck showing dimensions of cell . it was assumed that transverse distortion was caused by the distortion of the cell flanges only. is small relative to the deflection due to bending. By equating the shear deformation in a transverse grillage member to the bending deformation of the cell flanges in the bridge.29 shows a single cell of a cellular bridge deck with a constant web thickness but different upper and lower flange thicknesses. Figure 6.Page 231 The total deflection in a cantilever of length l. However. the webs of cellular decks are also flexible and consequently they too contribute to the overall transverse distortion. this may be difficult to carry out accurately in practice due to such factors as cracking in concrete sections. For cellular decks of other shapes.

As for slab bridges.30. where b is the breadth and d the thickness. 6. (6. 6.8) is generally small relative to the second and is often ignored. Such an equation is valid when the shear flows are opposing through the depth of the section as illustrated in Fig. this is not the Fig. 6. As mentioned previously. 6.30 Longitudinal section through deck for transverse bending Fig.8) The first term in equation (6. the torsion constant for a thin rectangular section twisting about its own axis may be approximated by bd3/3. (b) portion of box section with cantilever .31 Shear stresses due to torsion: (a) rectangular section. The torsion constants of the longitudinal and transverse grillage members are based on the portion of section represented by the members. For a portion of box section. 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.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.31(a). For the transverse members. 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.

The two internal members represent the portion of deck from halfway between the first and second webs to the centre. three-cell bridge deck with edge cantilevers. the contribution of the webs is accounted for through the shear forces in the longitudinal beams and should not be accounted for again here. 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. Transverse grillage members are located at the ends and at the central support to represent the transverse diaphragms.31(b) except in the edge cantilevers.29 would give: However.33 shows a convenient grillage mesh. l i.11.9: Shear flexible grillage model of a cellular bridge deck Figure 6. Additional transverse members are placed at 2 m centres giving a longitudinal to transverse member spacing ratio of 1:1. 6. There are 2 m thick solid diaphragms at the end and central supports. A formula suggested by Hambly (1991) halves the constant and removes the web term: (6.32 illustrates a two-span. is an increment of length and di is the thickness of that increment. A grillage model is required. The torsion constant for a thin-walled box section is given by: (6.9) where a is the area enclosed by the centre line of the wall. 6. The two edge members represent the portion of deck from the edge to halfway between the first and second webs (Fig. the centroid can be shown to be located at 0. It is assumed that the deck is continuously supported transversely at each support. one at the centre of each web. Figure 6. 6. Applying equation (6. Four longitudinal members are chosen. By summing moments of area about any point in the section.32).9) to the single cell of Fig.65 m .

6.32 Cellular bridge of Example 6. 6. the second moment of area per unit breadth.9 (dimensions in m): (a) cross-section.Page 234 Fig.8): . For the edge longitudinal members: For the internal longitudinal members: For the transverse members. itrans is given by equation (6. The second moments of area for the longitudinal members about this axis are then determined.33 Plan view of grillage mesh above the soffit. (b) longitudinal section Fig.

giving: The torsion constant per cell is given by equation (6. giving: . Then equation (5. giving: For the longitudinal members. a common approximation for I-sections.10): This gives a torsion constant for the interior longitudinal members of 0.2 is assumed.36 m4. a Poisson’s ratio of 0.Page 235 The breadth of the transverse members is 2 m.7): For concrete. 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 (6.67) gives: which results in a shear area of: The breadth of the transverse members is 2 m. the shear area is taken as the area of the webs.

1 Grillage modelling A suitable grillage model of a skew deck will depend largely on the angle of skew. In highly skewed decks.Page 236 The end and central diaphragm beams are 1. This deck will tend to span in the skew direction so the longitudinal grillage members are aligned in that direction. the shear area can be shown to equal 83.2): The shear area of the transverse diaphragm is taken as the actual shear area as no significant transverse distortion is assumed to take place. Hence: 6.7.3% 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 (6.7 Skew and curved bridge decks Many bridge decks incorporate some degree of skew and others are curved in plan. An exception to this is in concrete decks where the transverse reinforcement is not . Significant skew in bridge decks leads to a non-uniform distribution of reactions between supports. 6.34(b) shows a suitable grillage layout. Care is needed in modelling the support system in such cases as any flexibility will cause a redistribution of reactions. As a result. For a rectangular section. the span length and the width of the deck. An important consideration is to place the grillage members in the directions of principal strength.34(a) shows a long narrow bridge deck with a high degree of skew and Fig. the Wood and Armer equations can dictate a requirement for top reinforcement near supports where hogging would not normally be expected. A high degree of twisting accompanied by large torsional moments (mxy ) are also associated with skew decks. Large reactions at obtuse corners lead to high shear forces which can also be difficult to design for. 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. 6. The transverse grillage members should generally be oriented perpendicular to the longitudinal members. in reinforced concrete.2 m deep by 2 m wide. Figure 6. 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 and 6. (b) grillage layout.35(a) shows a short. 6. 6. The longitudinal members.Page 237 perpendicular to the longitudinal reinforcement. Consequently. it is generally more appropriate to orientate the transverse members parallel to the transverse reinforcement as illustrated in the alternative grillage layout of Fig. Once again.35 Short. This deck will tend to span perpendicular to the supports rather than along the skew direction. 6. highly skewed bridge deck: (a) plan view. but straight beams will be sufficiently accurate if the grillage mesh is fine enough. (c) alternative grillage layout Fig. Curved decks pose no particular problem for grillage modelling.34 Long. Some analysis programs will allow the use of curved beams.35(b) shows a suitable grillage layout. narrow. follow the curved layout closely due to the fineness of the mesh. Care should be taken with the edge grillage members which generally will have to be orientated in the skew direction. wide bridge deck with a small angle of skew and Fig.36 shows a suitable grillage mesh for a curved bridge deck. then this should be allowed for when assigning the properties of the edge beams in the grillage. the transverse grillage members are orientated perpendicular to the longitudinal members. (b) grillage layout .34(c). If significant edge beams or stiffening is provided to the bridge deck.35 will require a greater amount of judgement by the analyst in choosing a suitable grillage layout. In such cases. although straight. The Fig. the longitudinal grillage members are orientated in this direction. Bridge decks which fall between the extremes of Figs. wide bridge deck with small skew: (a) plan view. Figure 6. 6. Figure 6.

This is an advantage that the finite-element method has over the grillage method. they are approximately perpendicular to the longitudinal members.Page 238 Fig. Generally. 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.37 Alternative finite-element meshes: (a) skewed quadrilateral finite elements.36 Grillage layout for curved bridge deck transverse members radiate from the centre of the curve. especially for Fig. (b) alternative triangular elements .7. 6.2 Finite-element modelling Finite-element modelling of skew or curved decks should be carried out according to the recommendations for right decks. 6. In this way. 6.

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

the edges do not receive the same amount of axial stress as those near the centre of the bridge. the bridge deck has a noncontinuous neutral axis as indicated in the figure. Bridge decks with edge cantilevers are considered but it is stipulated that only those with short cantilevers should be analysed by the methods proposed. 7.Page 240 Chapter 7 Three-dimensional modelling of bridge decks 7. longitudinal bending stresses are set up. when the bending moment in a flanged beam varies from one point to another. a common centroid can be found and the entire bridge is often assumed to bend about a neutral axis passing through this point. the . If a load were applied to the deck in this condition. independently of the rest. 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. 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. Figure 7.2 Shear lag and neutral axis location When a bridge deck flexes. In this chapter. 7. The theoretical background is reviewed and a number of solutions are suggested including three-dimensional methods of analysis. each part would bend about its own centroid.1.2(b). When flanges or cantilevers are wide and slender.1 Introduction In Chapter 6. the analysis of bridge decks using planar models is discussed. These are distributed transversely from one part of the deck to adjacent parts by interface shear stresses. This common neutral axis can be seen in Fig. In this condition. As the rejoined bridge bends. If the bridge deck is now rejoined. 7.2(a) shows a bridge deck with the edge cantilevers separated from the main part of the deck. interface stresses are generated as illustrated in Fig. Thus. the problems associated with bridge decks such as those with wide edge cantilevers are discussed.

2(c).2 Transverse variation in neutral axis location: (a) if cantilevers and main deck were free to act independently. due to shear lag. This is because the edges of the cantilevers tend to bend about their own Fig. 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.1 Interface shear stresses in flanged beam subject to bending Fig. (d) actual neutral axis location . 7. (c) variation in longitudinal stress at top of deck. 7. (b) commonly assumed straight neutral axis. 7. as can be seen in Fig.Page 241 remote edges of the cantilevers.

4. 7.2(d). 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. Obviously they are not free to do this.3 Effective flange width In the design of bridge decks. 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). A three-dimensional analysis can automatically account for shear lag as it allows for variations in neutral axis location directly. Such a non-uniform neutral axis is illustrated in Fig. Hambly (1991) presents a chart for the determination of effective flange widths for beams subjected to distributed and concentrated loads.3. analysed with no allowance for shear lag. as it is from these points that longitudinal stresses begin to spread out into the cantilevers. The chart. 7. relates the ratio of effective flange width. as illustrated in Fig. but this tendency causes the overall bridge deck neutral axis to move towards the centroid of the cantilevers at the edges. b. It is possible to overcome this problem by assuming an ‘effective flange width’ for the edge cantilevers. to the ratio of actual flange Fig. There is a strong link between shear lag and neutral axis location. a two-dimensional analysis. reproduced here as Fig. 7.Page 242 centroidal axes. 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. as described in Chapter 6. 7. a two-dimensional model with an effective flange width. is often used which does not take account of shear lag. Hence.3 Actual and calculated distributions of longitudinal bending stress at top of flanged deck . and actual flange width. The effective flange width is also dependent on the form of the applied loading. be . can be used to determine the maximum stress in the cantilever. 7.

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.Page 243 Fig. The cantilevers are 2.4 Effective flange width for different loadings (solid line). 7. (b) showing effective flange width .5 Cross-section of bridge deck of Example 7.5(a) shows the cross-section of a bridge deck with edge cantilevers. Example 7. and common approximations (dashed line) width.1: Effective flange width Fig.1: (a) showing actual cantilever widths.4 m wide and the deck has a single simply supported span of 20 m. b. Figure 7. and length between points of contraflexure. L. 7.

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. Figure 7. 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.4/20=0. Inclusion of voids. the length between points of contraflexure.Page 244 As the span is simply supported.6 Portion of bridge deck modelled with solid brick elements .6 shows such a model of a portion of bridge deck with edge cantilevers. Hence. is equal to the span length in this case. is 2. Unfortunately. When the effects of shear lag are significant. In addition to this. From Fig. L.12. A constant stress is assumed in the modelled portion of the cantilever and that part of it outside the effective flange width is ignored.4 the ratios of be/b are 0. Figure 7. The benefit of this type of model is that it can be used to describe the geometry of highly complex bridge decks very accurately. b/L. a cellular structure or transverse diaphragms pose no particular problems. 7. This results in effective flange widths of 2.93 and 0. the model automatically allows for any variations in the location of the neutral axis and hence allows for shear lag in edge cantilevers. 7.23 m and 1.67 for the uniformly and point loaded cases respectively. 7. the use of such models is currently limited mainly to research and highly specialised Fig. the ratio of the cantilever width to this length. some form of threedimensional model is necessary to achieve an accurate representation of the behaviour of the structure. One such technique is three-dimensional finite-element analysis using solid ‘brick’ type elements.61 m respectively.5(b) shows this effective flange width for one of these load cases.

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

9(a) shows the crosssection of a 24. However.6. If part of the actual bridge deck deforms in-plane. as tends to occur at the ends of edge cantilevers. The in-plane distortion seen at the end of the cantilevers is made up of both in-plane shear distortion and in-plane bending. as illustrated in Fig. 7. the imposition of rotational restraints will prevent this behaviour from occurring in the model which may significantly affect the accuracy of the results. the real problem is the occurrence of local in-plane distortions of the grillage members. or the nodes at the ends of the members can be restrained against in-plane rotation. Figure 7. which are clearly inconsistent with the behaviour of the bridge deck.10(b) shows the corresponding quantities at span. To test the accuracy of both models. Assuming the elaborate model Fig. Both of these measures will have similar effects. Figure 7. It is the in-plane bending component which is not modelled by an upstand grillage with in-plane rotational restraints.8 In-plane distortion of members in upstand grillage model .8 m single-span bridge deck with wide edge cantilevers. 1996). 7. similar to that shown in Fig.Page 246 grillage. 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. 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. Figure 7. The members can be given very large in-plane second moments of area. the three-dimensional nature of the model causes in-plane displacements in the grillage mesh. a three-dimensional finite-element analysis using solid ‘brick’ type elements. was also carried out. 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. Figure 7. Such behaviour in the model can be avoided in one of two ways. and it may even be prudent to adopt both. Restraining in-plane rotations in the model may have adverse effects in some cases. the plane grillage model and the upstand grillage model.8.9(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). Only half of the width is shown and the crosssection is included for reference.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. 7.

Page 247 Fig. the complete removal of the rotational restraints resulted in the behaviour illustrated in Fig. 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.9 In-plane deformation in cantilevers of deck: (a) cross-section. . 7. (b) plan view of deflected shape (half) to be accurate. However. This inaccuracy in the upstand grillage is attributable to the use of inappropriate rotational restraints at the ends of the cantilevers. Clearly this is not a satisfactory approach for many bridge decks. Unfortunately. the benefits of the upstand grillage can be seen at this location.8 which caused inaccuracies elsewhere in the upstand grillage model. which may not be available prior to analysis. 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.

Page 248 Fig. .11(a) shows the cross-section of a bridge deck with edge cantilevers. 7. (b) at span Example 7. The deck is 25 m long with a single.10 Calculated longitudinal bending stress on top surface of deck: (a) at mid-span. An upstand grillage model is required.2: Upstand grillage model Figure 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.

The members representing the edge cantilevers are located at the centroid of the cantilevers which is 0.2 m below the top.7. the second moment of area per unit breadth is calculated . Assuming the main deck slab to be isotropic.6 m from the top. (c) plan view of grillage mesh Figure 7. 7.11(c) shows a plan of the upstand grillage mesh.3 times the depth of the side (at that location) from the edge in accordance with the recommendations of Section 5.11(b) shows the cross-section with a suitable upstand grillage model superimposed.4 m. Seventeen rows of transverse members are provided at a constant spacing of 1. (b) crosssection with grillage members superimposed.5 m. Figure 7.Page 249 Fig. the portion of bridge deck associated with each grillage member is indicated by the broken lines.2 (dimensions in mm): (a) cross-section. The properties of the members in the upstand grillage model are easily determined. The X direction is arbitrarily chosen to be parallel to the span of the bridge.11 Upstand grillage model of Example 7.3. The grillage members representing the cantilevers and the outermost members in the main part of the deck have been positioned at 0. The grillage members representing the main part of the deck are located at the centroid of that part which is at 0. This gives vertical members with a length of 0.

5 m past the centres of the bearings. 7. For the longitudinal members in the main deck. the bridge extends 0. resulting in a second moment of area of: a torsion constant of: and an area of: . the second moment of area is: The torsion constant is: and the area is: At the ends. 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.25 m. giving member breadths of 1.65): 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.Page 250 as for a beam: The torsion constant per unit breadth for longitudinal and transverse members is calculated according to equation (5.11.

80 0.180 0.346 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 (5.1 Upstand grillage member properties for Example 7.0161 0. the second moment of area is: the torsion constant is: Table 7.50 .173 0.0042 0.Page 251 For the edge cantilevers.173 J (m4) 0.3 times the depth (at that location) which gives: The area of the longitudinal cantilever members is given by: For the transverse cantilever members.44 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. other than those at the ends of the deck.32 1.65): 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.0073 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 0.44 1.216 0.0121 0.0066 0.50 1.276 0.0080 0.60 1.256 A (m2) 0.432 0.

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

1998).12 Upstand finite-element model Fig. It can be seen that the upstand FE model predicts an almost identical stress distribution to the elaborate three-dimensional brick FE model. 7.8 m and is supported along the entire width of the main part of the deck at each support location.10. The deck is continuous over two spans of 24. Figure 7. 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). The distributions predicted by the elaborate three-dimensional brick FE model and the plane and upstand grillage models described previously are also shown.Page 253 Fig. 7. 7. Example 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. 7.3: Upstand finite-element model Figure 7.13 Calculated longitudinal bending stresses at span on top surface of deck The bridge deck of Fig.9 was analysed by the authors using an upstand FE model (O’Brien and Keogh. An upstand FE model is required. . 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.

The elements representing the edge cantilevers are located at the centroid of the cantilevers which is 0.4 m. This results in vertical members with a length of 0. This is caused by the inability of the planar model to allow for the rising neutral .6 m from the top of the deck.4 m. maximum compressive stress close to span. The elements in the main part of the deck are given a depth of 1. 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. This stress distribution follows the expected pattern with zero stress at the ends.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. (b) finite-element mesh Figure 7.14(b) shows a three-dimensional view of a suitable upstand finite-element mesh.Page 254 Fig. As the model is symmetrical about the central support.15(b) illustrates the corresponding distribution along a line 0. All of the elements are 1. This model was analysed by the authors under the action of self weight.2 m below the top of the bridge deck.3: (a) cross-section (dimensions in mm). 7. 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.24 m long (in the span direction). Figure 7. 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.14 Upstand finite-element model of Example 7. 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 predict a very similar stress at all locations and the plane FE model is in reasonable agreement. Those representing the main part of the deck are located at the centroid of that part which is 0.2 m wide and 1. zero stress close to span and maximum tensile stress above the central support. Figure 7. only one span is shown in the figure.2 m and those in the edge cantilevers a depth of 0.6 m in from the edge of the cantilever.

Page 255 Fig. 7. as the stiffness of each part of the deck is made up of a combination of both of these. (b) 0. Alternatively this can be viewed as the inability of the planar model to allow for shear lag.3: (a) at centre.15 Longitudinal bending stress at top fibre for bridge of Example 7. This example shows the benefits of three-dimensional modelling over planar modelling for bridge decks of this type.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.6.6 m in from edge of cantilever axis in the edge cantilever. Most FE programs only allow the specification of a depth for the finite elements which does not . 7.

the presence of the voids does not greatly affect the longitudinal second moment of area of the deck.2 m long elements in the longitudinal direction would be appropriate for this model. most computer programs will not allow this. this will result in an overly stiff model. The additional grillage members should have zero area. Modelling of voided slabs by the plane FE method is discussed in Section 6. quite sensibly. 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. 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. but causes problems when dealing with voided slabs. As the voids are generally located close to mid-depth of the slab.4: Upstand FE model of voided slab Figure 7. 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. This is not the case when considering the cross-sectional area which is greatly reduced by the presence of the voids. The deck is simply supported with a 24 m span and is supported continuously across its breadth at each end. Figure 7.16(b) shows the cross-section of a suitable upstand FE model for this bridge deck. This is sufficient when dealing with solid slabs. When considering the longitudinal direction. Clearly a member with negative area has no physical meaning and. Example 7. Therefore. In this case the vertical members are 0.Page 256 allow the independent specification of area and second moment of area. each element represents a portion of deck 1. the equivalent depth of the elements will generally be quite close to (but smaller than) the actual depth of the voided slab. 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.2 m wide with one void. A choice of 20. The X direction is chosen as the longitudinal direction. 1. The second moment of area of this is: and the area is: .35 m long. 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. An upstand FE model is required. A solution to this problem is to reduce the area of the elements. For the elements in the main part of the deck. In theory. a finite element with a depth chosen by considering the second moment of area of the voided slab will have an excessive area.4. They should also have zero in-plane second moment of area as the in-plane behaviour is still modelled by the finite elements.16(a) shows the cross-section of a voided slab bridge deck with wide edge cantilevers.

16 Upstand finite-element model of Example 7. is: This gives a shortfall in second moment of area which has to be made up by additional grillage members. deq . each finite element in the main part of the deck is replaced by four elements and four grillage members as illustrated in Fig.Page 257 Fig. Ieq .879 m and the longitudinal grillage members have second moments of area of 0. These elements have the same equivalent depth of 0.4 (dimensions in mm): (a) cross-section.17. . 7. The second moment of area of these additional members.093 m 4.1): . (b) section through finite-element model Equating this to an equivalent solid element with the same area gives an equivalent element depth. The required transverse second moment of area per unit breadth is given by equation (6. of: The second moment of area of this equivalent solid element. is: To incorporate the additional members in the model. 7.

18 Upstand finite-element model with additional grillage members (half) . 7. (b) corresponding combination of elements and grillage members Hence.Page 258 Fig.17 Replacement of plate element: (a) original element. 7. the required additional second moment of area which is provided by the transverse grillage members is: Fig.

6. each about its own centroid.6 elements to give nodes at 0. (b) section through upstand finite-element model . In such cases where the location of the neutral axis is unclear. Rigid. This approach has the advantage of simplicity as there is a direct correspondence between each member and Fig.2 elements in the cantilever were replaced with four 0. provided care is taken to ensure that good similitude exists between the model and the actual structure. or very stiff vertical grillage members are specified at 0.19(b).6 m intervals.19 Upstand finite-element model of beam and slab bridge: (a) cross-section. Figure 7. The horizontal members at different levels are joined by stiff vertical members. 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.6×0. The final upstand FE model with grillage members shown as dark lines is illustrated in Fig. 7. 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. Only one-half of the model is shown as it is symmetrical. Therefore. 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. The properties of the remaining parts of the deck are then calculated. 7.19(a) shows a beam and slab bridge.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.6 m intervals to join the meshes on the different planes. However.Page 259 The edge cantilevers are modelled as finite elements with a depth of 0. and are represented by grillage members at the levels of those centroids as illustrated in Fig. the originally proposed 1. 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. 7.2×1.18. 7. an upstand FE analysis can be used to represent the behaviour more accurately than the alternative planar models.

combined with the tedium of interpreting the results. the number of elements required to achieve this is very large and this.20 Plate finite-element model of cellular bridge: (a) original bridge. If reinforcement is to resist the stresses in a beam and the adjacent elements. it is usual to uncouple the in-plane and out-of-plane behaviours. 7.7 Prestress loads in three-dimensional models When analysing for the effects of prestress in bridge decks. has the advantage of automatically allowing for transverse cell distortion as discussed in Section 6. (b) finite-element model a part of the structure. This model. Transverse diaphragms could also be incorporated into this model with ease. The in-plane behaviour is governed by the distribution of axial stress in the bridge deck and is often determined by a hand calculation. The stresses determined . Unfortunately. The bridge deck is then analysed to determine the effects of the equivalent loading. However. the interpretation of the output can be tedious.6. The out-of-plane behaviour is affected by the vertical components of tendon force and by the moments induced by tendon eccentricity. 7. The calculated moment for each beam member is only applicable to bending about its own centroid. as well as dealing with a varying neutral axis. Care should be taken with such a model to ensure that sufficient numbers of elements are provided through the depth of the webs. 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. to correctly model longitudinal bending there. 7. often rules out its use. 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. Figure 7.20(a) shows a cellular bridge deck and Fig.20(b) shows a suitable model based on a variation of the upstand FE analogy.Page 260 Fig.

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

(b) upstand finite-element model with vertical member at point of application of prestress. (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. In particular.Page 262 Fig. . upstand FE analyses with equivalent loading calculated in the traditional way (as described in Chapter 2) did not always give accurate results. 7.21 Portion of prestressed concrete deck: (a) original deck.

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 .

1 Evaluation of equation (C.1) where xi and y i are the co-ordinates of point i and n is the number of co-ordinate points.60×106 0 0 0.72×106 6. point n+1 is defined as equal to point 1.Page 267 Appendix C Location of centroid of section The centroid. of any section can be found from the co-ordinates of the perimeter points using the formula: (C. For the purposes of this calculation. The terms of equation (C.6. 6. .48×106 0 0 − 32.1 where Top and Bottom refer to the numerator and denominator respectively of the fraction specified in the equation. For the section of Fig.1) are given in Table C.93×10 9 0 0 − 20. Table C.64×10 9 .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 − 23.40×106 0 0 0.76×10 9 Bottom − 39. 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).

.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.

1 Cell with flange and web distortion: (a) assumed distortion. D. Hence. (b) segment of cell between points of contraflexure . the shear force in the top flange will be: Fig.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.

1 for a segment of cell between points of contraflexure. 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: . 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. This force is illustrated in Fig.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.

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 .7): .

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

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

17. post-tensioning 14 Creep 12. 78. 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. 185 elastomeric 31 pot 30–1 sliding 17. 18. 231. 42.Page 274 Index Page numbers appearing in bold refer to major entries AASHTO 40 Aesthetics 34–9 Aluminium deck 42 Analysis. 28. 260. 13–18 Contraction of integral bridges 128–33 Coupler. 269–71 Cellular section see Cross-section. 147–50 Computer implementation of grillages 179–80 Concrete. 180. introduction to 67–120 Anisotropic 151. 228 Box section 5–7 Bridge bashing see Loading. 169.4–5 voided rectangular 4 . impact Buried joint 32 Cable-stayed 25–6 Cantilever 3 balanced 14–17 Cellular bridge 212. impact Composite 25. 182. 265–6 Blister 16 Box culvert 21–3. 147–50 Cross-section 2–8 box 5–7 older concepts 7− 8 solid rectangular 2–4 T. lightweight 42 Continuous beam/slab 10. 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. 228–36 grillage modelling 230–6 three-dimensional finite element model 260 transverse cell distortion 228. 72–4 Box girder 212. location of 267–8 Collision loads see Loading. 229. 29–30 spherical 30 see also Supports Bending moments due to applied loading 263–4. box Centroid. 82.

95.Page 275 Culvert. 252 Edge stiffening 203–11. box 21–3. 180–2. 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. 178. 211. 260–1 Integral bridge 21–3. 245 Elastomeric bearing 31 Elevations 8–26 Equivalent loading due to prestress 54–66. 75–7 Differential temperature 47–51. 237. 245–7. 184 recommendations for modelling 182–5 shear flexible 212. sources of see Grillage. 40. 260–1 Equivalent loading due to temperature/ thermal effects 47. 218–19. 185. 228 Differential settlement 9. 107 Eurocode 40 Expansion of integral bridge see Integral bridge. 230 similitude with bridge slab 171–3 sources of inaccuracy 180–2 U-beams. 237 Effective flange width 242–4. 90. dry 127 Diaphragm 10. 42 Density. dynamic Geometrically orthotropic 152. 43–5 Inaccuracy. 244–6. modelling of 225 see also Upstand grillage modelling Halving joint 24 Impact loading 41. expansion FEA see Finite element analysis Finite element analysis (FEA) application of planar 200–39 beam and slab bridge 225–7 brick elements 244. 240. 252. 51–2. 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. 54 Imposed traffic loading 40. 177–9. sources of inaccuracy Incremental launch 17 In-plane effects 162. 253. 180. 41. 211. 72–4 Curved bridge 236–9 finite element modelling 238–9 grillage modelling 236–8 Dead loading 40. 88–9. 88 Equivalent loads method 67. 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 . 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. 189. 184. shallow strip 130 Forced vibration 53 Frame bridge 21–3 Frequency see Loading. 187–9. 220. 92. 246. 49.

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

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

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