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 1.4 Bridge elevations 1.1 Introduction 1.7 Prestress loading Chapter 3 Introduction to bridge analysis 3.5 Articulation 1.1 Introduction 2.2 Dead and superimposed dead loading 2.7 Joints 1.4 Thermal loading 2.3 Imposed traffic loading 2.3 Cross-sections 1.Page v Contents Preface Acknowledgements Chapter 1 Introduction 1.2 Moment distribution viii x 1 1 1 2 8 26 29 32 34 40 40 42 43 46 51 52 54 67 67 67 .5 Impact loading 2.8 Bridge aesthetics Chapter 2 Bridge loading 2.6 Dynamic effects 2.1 Introduction 3.2 Factors affecting structural form 1.

1 Introduction 4.3 Edge cantilevers and edge stiffening 6.5 Differential temperature effects 3.4 Three-dimensional analysis 7.7 Skew and curved bridge decks Chapter 7 Three-dimensional modelling of bridge decks 7.Page vi 3.6 Cellular bridges 6.7 Application of moment distribution to grillages Chapter 4 Integral bridges 4.5 Run-on slab 4.3 Effective flange width 7.3 Conventional spring model for deck expansion 4.4 Modelling expansion with an equivalent spring at deck level 4.5 Beam and slab bridges 6.2 Shear lag and neutral axis location 7.4 Voided slab bridge decks 6.4 Planar finite-element analysis of slab decks 5.4 Thermal expansion and contraction 3.2 Contraction of bridge deck 4.2 Simple isotropic slabs 6.6 Time-dependent effects in composite integral bridges Chapter 5 Slab bridge decks—behaviour and modelling 5.3 Grillage analysis of slab decks 5.1 Introduction 5.1 Introduction 6.5 Wood and Armer equations Chapter 6 Application of planar grillage and finite-element methods 6.3 Differential settlement of supports 3.6 Prestress 3.5 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 .1 Introduction 7.2 Thin-plate theory 5.

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 .6 Upstand finite-element modelling 7.

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

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

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

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

1. However.2. For example. the stresses . Various alternative structural bridge forms and methods of construction are presented in the following sections. is not a very efficient structural form as the second moment of area of a rectangle is relatively small. For example. Such a bridge is generally constructed of reinforced concrete (particularly for the shorter spans) or prestressed concrete. The structural forms of bridges are closely interlinked with the methods of construction. illustrated in Fig.Page 2 Fig. in some bridges. The method of construction also influences the distributions of moment and force in a bridge. The move now is towards bridges which are highly indeterminate and which have few joints or bearings. 1. such structural forms have many joints which are prone to leakage and also have many bearings which require replacement many times over the lifetime of the bridge.1 Solid rectangular The solid rectangular section. when a bridge is to be located over an inaccessible place.3 Cross-sections 1. This immediately limits the structural forms to those that can be constructed in this way. simply supported spans and cantilevers.1 Portion of bridge illustrating bridge engineering terms e. Due to the inefficiency of this structural form. the construction must be carried out without support from below. such as a railway yard or a deep valley. steel beams carry the self weight of the deck while composite steel and in-situ concrete carry the imposed traffic loading. 1.3. The methods of construction in turn are often dictated by the particular conditions on site.g.

When this is not the case.Page 3 Fig.2 In-situ solid rectangular section: (a) without cantilevers. 1. 1. e. As a result. without much reduction in the second moment of area. a rectangular section can be constructed using precast pretensioned inverted-T-sections as illustrated in Fig.3 Precast and in-situ solid rectangular section . this form of cross-section is often the most cost-effective for shorter spans (up to about 20 m). Such construction is clearly more economical when support from below the bridge is readily available. Comparing bridges of the same width. which is discussed in Section 1. Holes are cast at frequent intervals along the length of such beams to facilitate the threading through of transverse bottom reinforcement. such as illustrated in Figs. bridges can be constructed with or without cantilevers. 1. the precast beams must be designed to carry their self weight plus the weight of the Fig. However. However.3.2. 1.2. 1. In-situ reinforced concrete is then poured over the precast beams to form the complete section. what is often the more important advantage of cantilevers is the aesthetic one.g. it can be seen that the bridge with cantilevers has less weight. 1. over railway lines or deep waterways. (b) with cantilevers induced by the self weight of the concrete can become excessive. With this form of construction. the shuttering costs for a bridge with a flat soffit are relatively low and the reinforcement is generally simple. As can be seen in Fig.2(a) and (b). Solid rectangular sections can be constructed simply from in-situ concrete as illustrated in Fig.8.

4. 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.4 Voided slab section with cantilevers (initially wet) in-situ concrete. For the span range of 20–30 m. Guidance is given on the analysis of this type of deck in Chapter 6. particularly that designed to resist transverse bending. As a result.3. the preferred solution. are more expensive in terms of shuttering . 1. These decks can be constructed from ordinary reinforced concrete or can be post-tensioned. it is common practice in some countries to use in-situ concrete with polystyrene ‘voids’ as illustrated in Fig. illustrated in Fig.Page 4 Fig. the voids must be accounted for when considering the design to resist transverse bending. However. Hence it is.2 Voided rectangular For spans in excess of about 20 m. Including voids in a bridge deck increases the cost for a given structural depth because it adds to the complexity of the reinforcement. 1. Regardless of the diameter-to-depth ratio. 1. The shuttering costs are also less than for in-situ concrete T-sections which are described below.3. it reduces considerably the self weight and the area of concrete to be prestressed without significantly affecting the second moment of area. 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. solid rectangular sections become increasingly less costeffective due to their low second moment of area to weight ratio.3 T-section The T-section is commonly used for spans in the range 20–40 m as an alternative to voidedslab construction. in some cases. However. particularly when the designer wishes to minimise the structural depth. the section tends to be deeper for a given span. The complete rectangular section is available to carry other loading. 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. 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. In-situ T-section decks.5. 1.

Over less accessible places.6. An in-situ concrete slab. These consist of pretensioned prestressed concrete or steel beams placed in position along the length of the span. are favoured. supported on permanent shuttering. 1.4 Box sections For spans in excess of 40 m. it becomes economical to use ‘cellular’ or ‘box’ sections as illustrated in Fig. 1.Page 5 Fig. 1. (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. precast concrete or steel forms of T-section.5 In-situ concrete T-sections: (a) single web such as might be used for a pedestrian bridge. These have a higher second moment of area .7. as illustrated in Fig. spans transversely between the beams while acting as flanges to the beams longitudinally.3. 1.

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

1.6(b)) used more commonly today. It was also common practice in the past to build bridges of ‘pseudo-box’ construction as illustrated in Fig. These have wider bottom flanges than the precast ‘Y-beams’ (Fig. In the past. M-sections were often placed side by side with the bottom flanges within millimetres of each other. 1.8 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. The section is more efficient than a T-section as more concrete is located away from the centroid. 1. in the past. 1. For example. 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. The bottom in-situ concrete was reinforced transversely by threading bars through holes cast in the M-beams.9). Fig.10. The analysis of this type of bridge is similar to that of any T-section bridge.8. 1. These were constructed of M-beams with insitu concrete near the bottom to form a void.9 Precast M-beam . if water leaks into the voids.5 Older concepts Many variations of the above structural forms have been used in the past and are evident in existing bridge stocks. it was common practice to construct Tsection decks using precast ‘M-beams’ (Fig. corrosion problems can result and.Page 7 Fig.3. However. 1.

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

However. there is less disruption to any traffic that may be below as only one span needs to be closed at any one time.15.14. Fig. 1. have significantly fewer joints and bearings. In addition. illustrated in Fig. 1. 1. 1. It also has the advantage that. In such cases. the bending moment diagrams due to a uniformly distributed loading of intensity ω(kN/m) are illustrated in Fig. as illustrated in Fig. one after the other.13.4.2 Series of simply supported beams/slabs When a bridge crossing is too wide for an economical single span. 1.14 Continuous beam or slab . it is possible to construct what is in effect a series of simply supported bridges.12 Simply supported beam or slab Fig. 1. It is particularly favoured on poor soils where differential settlements of supports are anticipated. The cross-section is often solid rectangular but can be of any of the forms presented above. if constructed using in-situ concrete. is widely used when the bridge crosses a minor road or small river. as illustrated in Fig. The implication of this is that the bridge deck needs to be correspondingly deeper.4. 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. this form is relatively simple to analyse and construct.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. there are a great many joints and bearings with the result that a series of simply supported beams/slabs is no longer favoured in practice.13 Series of simply supported beam/slabs Fig. the concrete pours are moderately sized.Page 9 1. 1. It can be seen that the maximum moment in the simply supported case is significantly greater (about 25%) than that in the continuous case. This form. Continuous beams/slabs. the span is relatively small and multiple spans are infeasible and/or unnecessary. For example. The simply supported bridge is relatively simple to analyse and to construct but is disadvantaged by having bearings and joints at both ends.12. Like single-span bridges.

4. In the form illustrated in Fig. as the total bridge length becomes large. 1. continuity over intermediate supports is provided only by the slab. the amount of concrete that needs to be cast in one pour can become excessive. Further. The slab at the support in this form of construction is particularly flexible and tends to attract a relatively low bending moment. In-situ concrete is then used to make the finished bridge continuous over intermediate joints.3 Continuous beam/slab with full propping during construction As stated above. 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.Page 10 Fig. it is possible to use precast concrete or steel beams to construct a partially continuous bridge. Thus the in-situ slab alone is required to resist the complete hogging moment at the intermediate supports. This completely removes the need for any joints.17. 1. Elsewhere the cross-section is similar to that illustrated in Fig.15 Bending moment diagrams due to uniform loading of intensity ω (a) three simply : supported spans of length l. 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.4. Precast concrete or steel beams are placed initially in a series of simply supported spans.4 Partially continuous beam/slab When support from below during construction is expensive or infeasible. For bridges of moderate total length. 1. the concrete can be poured in-situ in one pour. 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. In the alternative form of partially continuous bridge.6.16. This tends to increase cost as the construction becomes more of a batch process than a continuous one. 1. 1. Two forms of partially continuous bridge are possible. illustrated in Fig. However. (b) one three-span continuous beam with span lengths l 1. . This is possible due to the fact that members of low structural stiffness (second moment of area) tend to attract low bending moment.

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

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

can be used to achieve continuity of prestressing across construction joints. one span at a time. such as illustrated in Fig. It has been said that joints should be provided every 100 m at least.20(b). 1. In this form of construction.17. two bearings are necessary at each intermediate support.4. Fig.20 Temporary support system for span-by-span construction: (a) joint over intermediate pier. The method is also of a continuous rather than a batch form as the precast beams can be constructed at a steady pace. This is achieved using temporary formwork supported on the bridge piers as illustrated in Fig. Proprietary post-tensioning couplers. A significant disadvantage is that. where bending moments and shear forces are relatively small. while intermediate joints have been removed. in-situ construction. (b) joint at quarter span . 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 particularly long continuous beam/slabs. The joint may sometimes be located at the quarter-span position as illustrated in Fig.20(a). an intermediate joint may become necessary to relieve stresses due to expansion/contraction. Particularly for the form illustrated in Fig. can be a viable option.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. 1. 1. intermediate bearings are still present with their associated maintenance implications. this figure is constantly being revised upwards as the problems of bridge joints in service receive ever more attention. starting even before work has commenced on site. Construction on site is fast.5 Continuous beam/slab—span-by-span construction For construction of particularly long bridges when access from below is expensive or infeasible. 1. resulting in minimum disruption to any existing traffic passing under the bridge. 1. However.21. 1.

Segments of deck are then added to the base segment.22(c).4.21 Post-tensioning coupler to transmit prestress forces across a construction joint (photograph courtesy of Ancon CCL) 1. At spans of this length.22(a)). The sequence of construction is illustrated in Fig.22(b)) until such time as they can be permanently posttensioned into place as illustrated in Fig. An intermediate pier is cast first and a small part of the bridge deck (Fig. 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. The segments are supported by a ‘travelling form’ connected to the existing bridge (Fig. However. 1. 1.Page 14 Fig. in the case of . 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.6 Continuous beam/slab—balanced cantilever construction When the area under a bridge is inaccessible and spans are in excess of about 40 m. Segments can be cast in-situ or precast. one on each side. precast beams are not generally available to span the complete length at once. 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). either alternately on opposing sides or simultaneously in pairs. 1. it is often economical to construct bridges by the balanced cantilever method. This form of bridge is generally made of post-tensioned prestressed concrete. in anticipation of the need to post-tension future segments at later stages of construction. Ducts are placed in all segments when they are first cast.22. 1. 1.

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

Thus.24. When cantilevers meet at mid-span.Page 16 Fig. 1. a ‘stitching segment’ is cast to make the bridge continuous as illustrated in Fig. (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. Post-tensioning tendons are placed in the bottom flange and webs by means of ‘blisters’. to resist the sagging moment that will exist in the finished structure due to applied traffic loading. the bridge forms a continuous beam and the imposed service loading generates a distribution of moment.23. 1.22 Balanced cantilever construction: (a) elevation of base segment and pier. 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. (c) sectional elevation showing tendon. 1.23 Casting of stitching segment . 1. such as illustrated in Fig.25(a). After the casting of the stitching segments and completion of construction. (b) temporary support of segments. the moment due to self weight during construction is such as illustrated in Fig. The bending moment in a balanced cantilever bridge is entirely hogging while the bridge remains in the form of a cantilever. illustrated in Fig. 1.25(b). 1.

This process is continued until the complete bridge has been constructed behind the abutment and pushed into place. 1. 1.7 Continuous beam/slab—push-launch construction For spans in excess of about 60 m. 1. Nevertheless. This results from creep deformations which are still taking place after the bridge has been made continuous.26(a).Page 17 Fig.25(b). The analysis of balanced cantilever bridges is complicated by a creep effect similar to that for partially continuous beams. Fig.26(b)).25(a) towards a form approaching that illustrated in Fig. a long segment is cast behind the bridge abutment as illustrated in Fig. In pushlaunch construction.25 Distributions of bending moment in balanced cantilever bridge: (a) due to self weight during construction. 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. 1.4. 1. ‘incremental-launch’ or ‘push-launch’ becomes a viable alternative to balanced cantilever as a method of construction. it is frequently the most economical alternative for construction over deep valleys when propping from below is expensive. When the deck is being pushed over intermediate supports. 1. 1. This is caused by a tendency for the distribution of moment due to self weight to change in the long term from the form illustrated in Fig. temporary sliding bearings are used to minimise friction forces.24 Blisters and tendon in the bottom flange (sectional elevation) sagging. (b) due to imposed loading after completion of construction .

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). This doubling of cantilever length has the effect of quadrupling the moment due to self weight during construction. and are only a viable solution if it can be accommodated.26 Push-launch construction: (a) casting of the first segment. These same parts may be subjected to sag moment in the completed bridge as illustrated in Fig. All of the bridge is constructed in the same place which is easily accessible to construction personnel and plant. as illustrated in Fig. 1. arches generate a significant horizontal thrust. Bridges designed for pushlaunch construction. the arch form is particularly effective. 1.28(b). This can be achieved if the bridge is located on a particularly sound foundation (such as rock). the horizontal thrust is taken by the tie. Some engineers design bridges in an arch form for aesthetic reasons but articulate the bridge like a .8 Arch bridges For larger spans (in excess of about 50 m). (b) pushing of the partially constructed bridge over first span The method has a considerable advantage of access. In a tied arch. like those designed for balanced cantilever construction. 1.28(a). Parts of the deck must be designed for significant hog moment during construction as illustrated in Fig. 1. 1. If this is not the case. an arch is still a possibility if it is tied such as illustrated in Fig.4.Page 18 Fig. 1.27(b).27(a). must be designed for the creep effect and are subject to the associated complexity and uncertainty in design. A significant disadvantage stems from the distribution of bending moment generated temporarily during construction. However.

(b) due to imposed loading after completion of construction Fig. (b) tied arch with deck at base of arch . 1. 1.28 Arch bridges: (a) conventional form with deck over the arch.Page 19 Fig.27 Distributions of bending moment in push-launch bridge: (a) due to self weight during construction.

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

can be analysed using two-dimensional models. there is the usual complexity and uncertainty associated with such calculations. A further complexity in the analysis of frame bridges is that. 1.9 Frame/box culvert (integral bridge) Frame or box bridges. The effects of deck shortening relative to the supports is to induce bending in the whole frame as illustrated in Fig.33.Page 21 1. 1. it was not considered feasible to design frame bridges of any great length (about 20 m was considered maximum).31. This is because the maximum bending moment tends to be less. However. are more effective at resisting applied vertical loading than simply supported or continuous beams/slabs. unless the transverse width is relatively small. The minimal maintenance requirement of frame/box culvert bridges is their greatest advantage. 1. 1. until recently. such as illustrated in Fig. There are no joints or bearings as the deck is integral with the piers and abutments. as can be seen from the examples of Fig. Continuous slab bridges on the other hand. (b) three-span frame .4.31 Frame/box culvert bridges: (a) box culvert.32. If some of this shortening is due to creep or shrinkage. the structural behaviour is three-dimensional. accommodating movements due to temperature changes or creep/shrinkage can be a problem and. Given the great upsurge of interest in maintenance and Fig.

1.33 Effect of thermal contraction of deck in frame bridge: (a) deflected shape. If the supports are fully fixed against translation.32 Typical distributions of bending moment: (a) simply supported spans. deck movements in such bridges will generate enormous stresses. this lack of maintenance has resulted in an explosion in the numbers of bridges of this form. (c) frames/box culverts Fig. 1. 1.34. It is now considered that bridges of this type of 100 m and longer are possible. There are two implications for longer frame-type bridges.Page 22 Fig. both relating to longitudinal movements. Ever longer spans are being achieved. (b) continuous beams. (b) distribution of bending moment durability in recent years. This problem has been overcome by allowing the supports to slide as illustrated in Fig. If the bridge is supported .

The run-on slab can rotate relative to the bridge deck but there is no relative translation. 1. There are a number of variations of this form of construction which are considered further in Chapter 4. To overcome this. These can be used in combination with in-situ concrete to form a frame bridge as illustrated in Fig. Thus. A precast variation of the frame/box culvert bridge has become particularly popular in recent years.34 Sliding support and run-on slab in frame bridge on piles.35.Page 23 Fig. 1. if it does leak. 1. Such a joint is remote from the main bridge structure and. Precast pretensioned concrete beams have a good record of durability and do not suffer from the problems associated with grouted post-tensioning tendons.35 Composite precast and in-situ concrete frame bridge . a joint is required to facilitate translational movements. Crosssections are typically of the form illustrated in Fig. 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 axes of the piles are orientated so as to provide minimum resistance to longitudinal movement. 1.6(b). will not lead to deterioration of the bridge itself. The second implication of longer frame bridges is that the bridge moves relative to the surrounding ground. at the ends of the run-on slabs. Fig.

1. 1. The drop-in span. 1. This bridge is determinate as the central ‘drop-in’ part is simply supported.37(b). it can readily be constructed over inaccessible areas. which provides access. The side spans are simply supported with cantilevers to which point loads from the drop-in span are applied at their ends. were used. Fig. A typical example is illustrated in Fig. 1.Page 24 1. two ‘halving joints’. This detail is particularly problematic as access to inspect or replace the bearings is extremely difficult.4. Thus. A more convenient alternative. some older bridges were constructed of precast concrete with drop-in spans.36.37 Halving joint at end of drop-in span: (a) traditional detail (no access). in particular. 1.10 Beams/slabs with drop-in span For ease of construction and of analysis. it is still popular in some countries for pedestrian bridges over roads. as illustrated in Fig. However. 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. The joint and bearing detail at the ends of the drop-in span in this form of construction is particularly important.37(a). is illustrated in Fig. In older bridges of the type. can be placed in position very quickly over a road or railway requiring a minimum closure time. (b) alternative detail with access .36 Beam bridge with drop-in span Fig.

1. The maximum main span achievable is increasing all the time. For spans of moderate length. regardless of which alternative is chosen. 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. steel box section decks are used to reduce the bridge self weight. The concept of cable-stayed bridges is simple. 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. which promotes corrosion of the halving joint reinforcement. The deck can then be designed as a continuous beam with spring supports. • There are very high tensile and shear stresses at a point where the structural depth is relatively small. It is also generally necessary to carry out a dynamic analysis for bridges of such slenderness. 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. . 1. 1. The cables are only required to take tension and they provide support to the deck at frequent intervals.39. illustrated in Fig. 1.38 Reinforcement detail in halving joint However. • As can be seen in Fig. the joints tend to leak. for the longest spans.Page 25 Fig. halving joints frequently cause difficulty for a number of reasons: • Even for pedestrian bridges in which de-icing salts are not used.4. the current limit is of the order of 1000 m.11 Cable-stayed bridges Cable-stayed construction. the cross-sections of cable-stayed bridges are often composite with steel beams and concrete slabs.

40 Suspension bridge . The articulation of a bridge is the scheme for accommodating movements due to creep. are of the suspension type illustrated in Fig. While the present trend is to provide ever fewer joints and bearings. it is sometimes difficult to cater for the horizontal forces generated at the ends of the cables. 1. cablestayed construction is generally favoured except for the very longest bridges.4. 1. shrinkage and thermal effects while keeping the structure stable. For these reasons. 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. it is a necessary consideration for those which do.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.12 Suspension bridges The very longest bridges in the world. wind and accidental impact forces from errant vehicles. the problems of creep. the bridge must have the capacity to resist some relatively small forces while accommodating movements.39 Cable-stayed bridge The economy of the cable-stayed form stems from its ease of construction over inaccessible places. 1. Thus. Further. Horizontal forces are caused by braking and traction of vehicles. up to about 2000 m span. As for balanced cantilever bridges. In suspension bridges. shrinkage and thermal movement are still very real and no one form of construction is the best for all situations. segments are placed successively on alternate sides of the pylon. It lends itself readily to staged construction with the cables being added as required to support successively placed segments of the deck. While this clearly does not apply to bridges without joints or bearings.40. 1.Page 26 Fig. 1. Fig. the main cables are in catenary and the deck hangs from them applying a substantially uniform loading.

Two of the simplest forms of articulation are illustrated in Figs.41 Plan views showing articulation of typical bridges: (a) simply supported slab. The bearings usually allow free rotation but may or may not allow horizontal translation. in the case of the two-span bridge. 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. Free sliding bearings are provided elsewhere to accommodate transverse movements. 1.41(c). guided sliding—free to move horizontally in one direction only. (c) two-span bridge of small width . such as those due to temperature changes.Page 27 In-situ concrete bridges are generally supported on a finite number of bearings. it may be possible to articulate ignoring transverse movements such as illustrated in Fig. For both bridges. Fig. To make the structure stable in the horizontal plane. guided sliding bearings are provided at C and. 2. (b) twospan skewed slab. 1. At the same time they accommodate longitudinal movements. 3. a combination of the three types of bearing is provided. free sliding—fully free to move horizontally. They are generally of one of the following three types: 1. A is a fixed bearing allowing no horizontal movement. also at E. fixed—no horizontal translation allowed.41(a) and (b) where the arrows indicate the direction in which movements are allowed. When bridges are not very wide (less than about 5 m). 1. In many bridges.

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

Such a situation can be prevented by ensuring that the reaction at the central bearing due to permanent loading exceeds the uplift force due to temperature. 1.43. 1. 1. the two outer bearings must be designed to resist all of the load which renders the central bearing redundant. If this is not possible. 1.44 Uplift of bearing due to transverse bending caused by differential thermal effects loading in right bridges if the span lengths are significantly different. 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. 1.6 Bearings There are many types of bearings and the choice of which type to use depends on the forces and movements to be accommodated and on the maintenance implications. as it is not taking any load. . Only a limited number of the more commonly used types are described here. 1.44.43 Uplift of bearings due to traffic loading Fig.6. If this occurs.Page 29 Fig. even with no skew and typical span lengths. Further details of these and others are given by Lee (1994). differential thermal effects can cause transverse bending which can result in uplift as illustrated in Fig.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. However.

1. such as illustrated in Fig. In some combinations. 1. guides are used such as illustrated in Fig.46.Page 30 Fig. The elastomer effectively acts as a retained fluid and facilitates some rotation while preventing translation. it is common in such systems for the lubricant to be squeezed out after a number of years.2 Pot bearings Pot bearings. consist of a metal cylinder containing an elastomer to which the force is applied by means of a metal piston. rotation is facilitated through some other mechanism and plane sliding surfaces are used which allow translation only. They can take many forms and are often used in combination with other forms of bearing. the sliding surfaces are spherical and allow rotation. pot bearings by themselves are commonly used at the point of fixity. However. In other cases. Sliding bearings offer a frictional resistance to movement which is approximately proportional to the vertical force. namely polytetrafluoroethylene (PTFE). 1.45. They are also used in combination with plane sliding surfaces to provide free sliding .6. 1. this form is also referred to as the spherical bearing. When translation is to be allowed in one direction only. They are frequently used for motorway bridges of moderate span. Whether or not sliding bearings are lubricated. Some bearings are lubricated. it has been suggested that they be treated as wearing parts that eventually need to be replaced.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. Thus. Sliding bearings today generally consist of a stainless steel plate sliding on a PTFE-coated surface. at which time the coefficient returns to the unlubricated value. resulting in a reduced coefficient of friction.

3 Elastomeric bearings When the forces to be resisted are not very high.47(b)). They are considered to be quite durable except in highly corrosive environments and require little maintenance.47(a)) and translation by a shearing deformation (Fig. 1.45). Elastomeric bearings accommodate rotation by deflecting more on one side than the other (Fig. when bearings are provided under each beam in precast construction.g. e. 1. By incorporating guides (Fig.46 Pot bearing bearings. 1.Page 31 Fig.47 Elastomeric bearing: (a) rotation. 1. 1. 1.6. such a combination can also be used to form a guided sliding bearing. (b) translation . Fig. elastomeric bearings can be a very economical alternative to sliding or pot bearings. They are made from rubber and can be in a single layer (for relatively low loading) or in multiple layers separated by metal plates.

particularly in road bridges. 1. This form has been successfully used for movements of up to 40 mm and is inexpensive to install or replace. A typical arrangement is illustrated in Fig. as illustrated in Fig. 1. allowing saltcontaminated water to wash over the substructures. 1.7. Joints are notoriously problematic. The material used to span the joint is important.7. can result in a minimum maintenance solution. However.1 Buried joint For movements of less than 10–20 mm.48. in this case the road surfacing over the joint consists of a specially formulated flexible bitumen. 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)) . and frequently leak.49.Page 32 1. if designed well. movements will always occur with the result that joints will always be needed. However. for larger gaps. 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. the movement must be accommodated at the end of the run-on slab.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. joints buried beneath road surfacing are possible and. 1. Even in integral construction. 1.

illustrated in Fig. is no longer favoured in many countries.49 Asphaltic plug joint (after Lee (1994)) 1. 1.50 Nosing joint (after Lee (1994)) .7. It can accommodate movements of similar magnitude to the asphaltic plug joint but has a reputation for frequent failure and leakage. 1.3 Nosing joint Very popular in the 1960s and 1970s. 1. Fig.50. the nosing joint. 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.Page 33 Fig.

1. 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. Some aspects of aesthetics are common to most bridges. Certain bridge proportions in particular. 1. 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.51 Continuity of upstand and parapet (photograph courtesy of Roughan and O’Donovan Consulting Engineers. The aesthetics of the more common shorter-span bridges are considered in this section. 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. 1.Page 34 1. there is generally some common ground. This effect can be useful. Fig.8 Bridge aesthetics The art of bridge aesthetics is a subjective one with each designer having his/her own strongly held opinions. However. 1. particularly if the designer wishes to draw attention away from an excessively deep main deck. Further details on these and longer-span bridge aesthetics can be found in the excellent book on the subject by Leonhardt (1984). look better than others and attention to this can substantially improve the appearance of the structure.53. particularly on what constitutes an aesthetically displeasing bridge. Dublin) . It is generally agreed that the upstand and parapet are important and that they should be carried through from the bridge to corresponding upstands and parapets in the abutment wing walls as illustrated in Fig. The sun tends to shine directly on upstands while the main deck tends to remain in shadow (Fig.51.52).

In this example. The abutment wing walls also play an important role as can be seen in the example of Fig.Page 35 Fig. 1.52 Shading of main deck relative to upstand (photograph courtesy of Roughan and O’Donovan Consulting Engineers. the shape of the opening is square (span equals height) and the abutment wing walls are large triangular . 1. 1. the shape of the opening has a significant influence on the aesthetics.8.1 Single-span beam/slab/frame bridges of constant depth For very short-span bridges or culverts. Dublin) Fig.53 Section through upstand 1.54.

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

56(a). The heavier looking alternative illustrated in Fig. span/upstand depth ratios of 20 and a span/main deck depth ratio of 10 is often recommended. 1. where people and traffic are close to the structure which is large relative to their size.Page 37 Fig. The bridge illustrated is probably typical with a main span/upstand depth . 1.55(a) with a span/upstand depth ratio of 40 and a span/main deck depth ratio of 20. This is illustrated in Fig. 1. this proportioning also tends to bring the relative dimensions of the rectangular openings closer.54(c) for upstand and main deck respectively. 1. (b) deep deck and slender upstand and 10 are illustrated in Fig. an even more slender deck is favoured.) A structure with similar proportions looks much better in Fig. while ratios of 10 and 5 are illustrated in Fig. 1.54(d) and (a).56(b) as it is smaller and is more likely to be viewed from a distance. as illustrated in the figure. (In this structure.55(b) has ratios of 60 and 10. 1.57. a much more slender deck is desirable. It can be seen that the upstand appears too thin and/or the deck too deep. which has a good aesthetic effect. Typical ratios are illustrated in Fig. For a 2×1 rectangular opening with wing walls of similar size. typically by 25–35% as illustrated in Fig. When the ground level is lower at the centre. For aesthetic reasons. This can be convenient as the principal obstruction to be spanned is often in the central part of the bridge. 1. 1.55 Rectangular opening with small wing walls: (a) slender deck and deep upstand. it is common practice in three-span construction to have the centre span greater than the side spans.8.2 Multiple spans The relative span lengths in multi-span bridges have a significant effect on the appearance. a parapet wall is integral with the upstand making it look deeper than necessary. Leonhardt points out that scale is important as well as proportion. 1. For rectangular openings with less pronounced wing walls.

1.56 The influence of scale on appearance: (a) large structure near the viewer looks heavy. 1.57 Three-span bridge with good proportions Fig. (b) small structure remote from the viewer looks better than in(a) Fig. (b) curved alignment achieved using two curves of differing radius. 1.58 Variable depth bridges: (a) straight haunches. (c) curved haunches .Page 38 Fig.

1. An open parapet is also used in the bridge of Fig. where the depth is increased at the points of maximum (hogging) moment.57 to increase the apparent slenderness of the bridge.Page 39 ratio of 40 and a span/deck depth ratio of 20. both in terms of shuttering and reinforcement details. curved decks are strongly favoured over straight ones. However. When alignments are curved. they are not as aesthetically pleasing as a curved profile. effectively tying the bridge together. the upstand is continuous from end to end. This greatly complicates the detailing but makes for an efficient light structure and tends to look very well. Straight haunches are considerably cheaper than curved ones. 1. When a road or rail alignment is straight. . Varying the depth of bridges allows the depth to be increased at points of maximum moment. 1.58(a). straight haunches are possible as illustrated in Fig. As for single-span bridges.58(b) and (c). illustrated in Figs.

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

Superimposed dead 3. Imposed traffic 4. it is recommended that a combined model of the bridge structure and the supporting soil be used to determine the stresses induced by settlement. The calculation is complicated by the use of different factors of safety and the specification of different design limits for the different combinations. Pedestrian and cycle track 5.3 – 2. For example. No geotechnical guidance is given in either BD37/88 or AASHTO on how bridges should be analysed to determine the effect of this phenomenon. Differential settlement 7. in AASHTO and in the draft Eurocode. In practice. i.7 must also be considered. this usually only includes pedestrian bridges and long-span road and rail bridges. The draft Eurocode on Geotechnical Design. excludes some thermal and wind effects but includes a higher traffic loading.Page 41 Table 2. . where the natural frequency of the bridge is at a level which can be excited by traffic or wind. 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 EC1.5 2.2 2. Differential settlement of supports can induce significant bending in continuous beam or slab bridges.4 – 2. In pedestrian bridges. EC7 (1994). The loading due to impact from collisions with errant vehicles can be quite significant for some bridge elements.2 2. it should be ensured that the natural frequency of the bridge is not close to that of walking or jogging pedestrians. recommends that the process of soil/structure interaction be taken into consideration for accurate analysis of problems of this type. as will be demonstrated in Chapter 3. Thermal 6.e. vehicles and the bridge itself Effect of prestress on indeterminate bridges Section 2. Wind 10. Dynamic effects 9. Dead 2. Vibration is generally only significant in particularly slender bridges. 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.1 Summary of bridge loads Load type 1.6 – 2. Impact 8. 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. The load specified in the UK has increased dramatically in recent years.

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

There are two problems with this as a means of collecting statistics on truck weights.3. Most codes allow a reduction for long footpaths. Normal traffic loading or Highway A (HA) represents an extreme . a reduced intensity is allowed by some codes to reflect the reduced probability of both traffic and pedestrian loading reaching extreme values simultaneously. is known as the ‘remaining area’. In the Eurocode. after removing these lanes. The AASHTO code specifies a traffic lane loading which consists of a knife-edge load plus a uniformly distributed lane loading. Bridge traffic loading is applied to notional lanes which are independent of the actual lanes delineated on the road. In the first place.3 Imposed traffic loading Bridge traffic can be vehicular. 2. the road width is divided into a number of notional lanes. 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. there has been a scarcity of good unbiased data on road traffic loading until recent years. its importance should not be underestimated. 5 kN/m 2 in the draft Eurocode and the British standard and 4 kN/m 2 in the American code). The outstanding road width between kerbs. a truck of specified dimensions and axle weights must be considered. each 3 m wide. more importantly. sampling was carried out by taking trucks from the traffic stream and weighing them statically on weighbridges.Page 43 2. Alternatively. ‘normal’ and ‘abnormal’. Bridge traffic loading is often governed by trucks whose weights are substantially in excess of the legal maximum. 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. When a structural element supports both pedestrian and traffic loading.g. 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. Bridge codes commonly specify a basic intensity for pedestrian loading (e.1 Imposed loading due to road traffic While some truck-weighing campaigns have been carried out in the past. 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. 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. In general. While pedestrian/cycle traffic loading on bridges is not difficult to calculate. the imposed traffic loading specified by AASHTO is considerably less onerous than that specified by both BD37/88 and the Eurocode. The AASHTO code also specifies notional lanes of fixed width. BD37/88 and the draft Eurocode specify two types of traffic loading. Vehicular and rail traffic are considered in subsections below. the quantity of data collected is relatively small but. rail or pedestrian/cycle or indeed any combination of these.

This could be a traffic jam involving a convoy of very heavy trucks as would tend to govern for a long bridge. For bridges with many notional lanes.Page 44 combination of overloaded trucks of normal dimensions. ‘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. While there are exceptions. it could be a chance occurrence of two overloaded moving trucks near the centre of a short bridge at the same time. 14. the vehicle is known as the Highway B or HB vehicle.6. It is scaled in gross ‘units’ of 40 kN so that a minor road bridge can be designed. Illustrated in Fig. there can be a considerable dynamic component of truck loading which is deemed to be included in the specified normal load.1(a). In addition.1 ‘Normal’ road traffic loading: (a) Eurocode normal loading. the standard combination is a load intensity of 9 kN/m2 in Lane No. 2. 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. or 29. 2 and 3.1(b).6. the abnormal load in BD37/88 is Fig. 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. 2. In the British standard. In BD37/88. 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. there is uniform loading in the remaining area. 2. 19.6. On the other hand. only one abnormal vehicle is specified but it may have a length of 9. (b) British standard HA loading . Combinations of normal traffic and an abnormal vehicle must be considered in bridge design. a number of possibilities must be considered. The four wheels of the tandems together weigh 600 kN. 2.6 m. 1 and 2.2. 400 kN and 200 kN for Lanes 1. respectively. While there are a number of factors which can vary between road classes and between countries. Particularly on roads with rough surfaces.5 kN/m 2 elsewhere. 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. This consists of an exceptionally heavy vehicle of the type which is only allowed to travel under licence from the road/bridge authority. Eurocode normal loading consists of uniform loading and a tandem of four wheels in each lane as illustrated in Fig. for example.6. 24. The possibility of abnormal or Highway B (HB) loading must also be considered in British and Eurocode designs.

The other disadvantage to the use of track slabs depends on the method used to maintain and replace ballast. 2. an additional vertical dynamic load is induced by the change from the relatively ‘soft’ ballast support to the relatively hard track slab. Railway Lower (RL). However. In addition. On passenger transit ‘light rail’ systems. If this is done using automatic . However. On bridges. 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. Railway tracks on grade are generally laid on ballast. is similar in format.6 m intervals of magnitude 250 kN each and uniform loading of intensity 80 kN/m both before and after them. This follows from the fact that the train can generally be assumed to remain on the tracks. known as Railway Upper (RU).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. In BD37/88. A standard light rail load model. the normal load model. There are two disadvantages to the use of track slabs.3. 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.Page 45 Fig. less onerous load models can be applied. there are some aspects of traffic loading that are specific to railway bridges which must be considered. 2. is specified in the British standard. 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. bridges throughout a rail network are generally designed for the same normal load model. the Eurocode provides for an alternative abnormal load model. This factor is a function of the permissible train speed and of the natural frequency of the bridge. This effect can be minimised by incorporating transition zones at the ends of the bridge with ballast of reducing depth. When used.2 British standard abnormal (HB) vehicle consisting of 16 wheel loads of F=2. 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 1.2 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. Normal load is placed throughout the remainder of the lane and in the other lanes.

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

such as in the morning when the sun shines on the top of the bridge heating it up faster than the interior. the baseline temperature is clearly a mean temperature which relates to the density of the adjacent soil. 2. A distribution of stress is calculated corresponding to the specified change in temperature.1: Differential temperature I The bridge beam illustrated in Fig. This is resolved into axial.Page 47 As is discussed in Chapter 4. bridges are subjected to differential temperature changes on a daily basis. bending and residual effects as will be illustrated in the following examples. The applied temperature distribution is converted into the equivalent stress distribution of Fig. Cracking of reinforced concrete members reduces the effective cross-sectional area and second moment of area. i. However. bending and residual distributions as will be illustrated in the following examples. It is required to determine the effects of the temperature change if it is simply supported on one fixed and one sliding bearing. the baseline temperature distribution is important. the implication being that the distributions specified represent the differences between the baseline and the expected extremes. These distributions can be resolved into axial. one corresponding to the heating-up period and one corresponding to the cooling-down period. 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. There is an ‘equivalent’ axial force and bending moment associated with any distribution of temperature. no such distribution is typically specified in codes. Two distributions of differential temperature are specified in some codes. integral bridges undergo repeated expansions and contractions due to daily or seasonal temperature fluctuations. the magnitude of the resulting thermal stresses can be significantly overestimated. that distribution which exists when the structural material first sets. The coefficient of thermal expansion is 6 12×10− and the modulus of elasticity is 35000 N/mm2. 2. this causes the backfill behind the abutments to compact to an equilibrium density. The corresponding forces and moments are then readily calculated. The equivalent axial force can readily be calculated as the sum of products of stress and area: . As for uniform changes in temperature. In such cases. The effects of both uniform and differential temperature changes can be determined using the method of ‘equivalent loads’. Methods of analysing to determine the effects of the equivalent loads are described in Chapter 3.4(a) by multiplying by the coefficient of thermal expansion and the modulus of elasticity.3 is subjected to the differential increase in temperature shown. Example 2. In addition to uniform changes in temperature.e. Transverse temperature differences can occur when one face of a superstructure is subjected to direct sun while the opposite side is in the shade. After some time. This effect can be particularly significant when the depth of the superstructure is great. If cracking is ignored.

(b) axial component. this beam is supported on a sliding bearing at one end and is therefore free to expand. Thus.4(b). (d) residual stress distribution This corresponds to a uniform axial stress of 579600/(600× 1200)=0. The equivalent bending moment is found by taking moments about the centroid (positive sag): The corresponding extreme fibre stresses are: .81 N/mm2 as illustrated in Fig. 2.Page 48 Fig. there is in fact no axial stress but a strain of magnitude 6 0.4 Components of imposed stress distribution: (a) total distribution. 2. However.3 Beam subject to differential temperature change Fig. (c) bending component. 2.81/35000=23×10− .

2.100α E Total force= 3.5 Beam and slab bridge subject to differential temperature: (a) cross-section.220α E . It is found simply by subtracting Figs. The coefficient of thermal expansion is α the modulus of elasticity is E.2 Calculation of force Block a b c d Details 3α (2.4×0. 2. bending moment and residual stresses are required due to the differential temperature increases shown in Fig.150α E 0. 2.080α E 1. the equivalent axial force.5(a). illustrated in Fig. 2. and Fig.5(b). a strain distribution is generated which varies linearly in the range 6 ±1. Example 2.4(d).4(c).4(b) and (c) from 2. it is free to rotate and there is in fact no such stress. 2.Page 49 as illustrated in Fig.4(a). 2. The difference between the applied stress distribution and that which results in axial and bending strains is trapped in the section and is known as the residual stress distribution.2: Differential temperature II For the beam and slab bridge illustrated in Fig.890α E 0.11/35 000=±32×10− .15)= E Force 1. As the beam is simply supported. Instead. (b) imposed distribution of temperature Table 2.

70= 4.3 (positive sag).22α E E/0.2. 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. 0.Page 50 Fig. The bridge is split into two halves. each of area. (b) corresponding imposed stress distribution By summing moments of area.718α E .506α E − 0.262α E − 0.012α E 0.064 86 m 4.60α E.3 Calculation of moment Block a b c d Details Moment − 0. 2.062α E Total moment= − 0.22α corresponds to an axial tension of 3.70 m and second moment of area. The total force of 3. 2.6 and divided into rectangular and triangular blocks. the centroid of the bridge is found to be. The temperature distribution is converted into a stress distribution in Fig. below the 2 top fibre.718α corresponds to E stresses (positive tension) of: Table 2. The total moment of − 0. 0. Similarly moment is calculated as the sum of products of stress. area and distance from the centroid as outlined in Table 2.6 Division of section into blocks: (a) cross-section.

this is converted into strain energy in the spring. collides with a spring of stiffness.Page 51 Fig.7(a).8 is considered first. which causes a deflection. A static force. (b) axial component. (d) residual stress distribution Hence the applied stress distribution can be resolved as illustrated in Fig. However.7(b) and (c) from the applied distribution of Fig. bending and residual components: (a) total distribution. codes of practice often greatly simplify the procedure by specifying equivalent static forces. 2. The simple case illustrated in Fig. 2.5 Impact loading Most bridge analysis is based on static linear elastic principles. Δ generates a strain energy of: . (c) bending component. the collision of a vehicle with a bridge is highly non-linear.2) Fig.1) On impact. 2. This section considers the basis on which these forces are derived.7 Resolution of stress distribution into axial. travelling at a velocity. An undeformable sphere of mass.8 Impact of undeformable sphere with spring . K. The kinetic energy of the sphere is: (2. v. 2. The residual distribution is found by subtracting the distributions of Figs. 2. 2. Peq .7. To overcome the resulting complications. (2. m. 2.

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

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

2. Like temperature. In Section 2. θis generally small. In the case of road traffic. Furthermore. For a qualitative understanding of the effects of prestress. design for the impact of vehicles colliding with bridges was discussed. the draft Eurocode specifies an equivalent static force.1 Equivalent loads and linear transformation The equivalent loading due to prestress can generally be found by simple equilibrium of forces. there are concentrated . Whether the bridge consists of beams or a slab. Finally. equivalent loadings can be found for individual tendons. the applied loading is a truck or trucks of considerable mass. In such cases. it is often necessary to analyse to determine the degree to which prestressing of one member affects others. even for simply supported slab or beam-and-slab bridges. As an alternative. Such a method is only necessary in the case of indeterminate bridges. the code allows for the carrying out of a dynamic analysis. prestress can be handled using the method of equivalent loads.5.7) It also follows from the small angle that the horizontal force is P cosθ P.7 Prestress loading While prestress is not in fact a loading as much as a means of resisting load. methods will be given for the calculation of their magnitudes. the trucks are moving so the location of their masses are changing with time. In this section. it is often convenient to treat it as a loading for analysis purposes. 2.7. For example. equilibrium of vertical forces gives an upward force at B of: As the angle. 2. 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. (2. for the externally prestressed bridge illustrated in Fig. 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). vibrating on their own tyres and suspensions. The combined effect of a number of tendons can then be found by simply combining the loadings.11(a). However. as the ≈ forces are eccentric to the centroid at the ends.Page 54 applied loading and the bridge is taken into account. Such a loading can readily be specified in a computer model and the maximum distribution of stress determined. For such cases. Examples of analysis using equivalent prestress loads are given in Chapter 3. this can be approximated as: . the concept of linear transformation is also introduced.

Hence the total equivalent loading due to )e Pe prestress is as illustrated in Fig. A small segment Fig. 2. 2.11 Prestressed concrete beam with external post-tensioning: (a) elevation showing tendon. (b) equivalent loading due to prestress moments there of magnitude (Pcosθ 2 ≈ 2. It can be shown that the equivalent loading due to prestress is always self-equilibrating.11(b). A parabolically profiled prestressing tendon generates a uniform loading which again can be quantified using equilibrium of vertical forces. (b) equivalent loading .Page 55 Fig.12 Segment of parabolically profiled tendon: (a) elevation. 2.

12(a). The intensity of uniform loading on this segment is: (2.12(b).11) The equivalent loads on the segment are illustrated in Fig.12) where s is referred to as the sag in the tendon over length l as indicated in the figure. Example 2. . Similarly the vertical component of force at 2 is: (2. 2.10) where F 2 is downwards when the slope is positive. At point 1. This force is upwards when the slope is positive. there is an upward vertical component of the prestress force of: (2.8) As the angles are small: (2. 2.9) where x1 is the X coordinate at point 1.3: Parabolic profile The beam illustrated in Fig. 2.13 is prestressed using a single parabolic tendon set out according to the equation: (2. It is required to determine the equivalent loading due to prestress.Page 56 of such a profile is illustrated in Fig.

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

unaffected by eccentricity at the ends of the span. It is required to determine an amendment to the profile to increase the stress at this point without increasing the prestress force. results in insufficient compressive stress in the top fibres of the bridge at B.3 illustrates the fact that the intensity of equivalent uniform loading due to a parabolic tendon profile is independent of the end eccentricities. this force is downwards when positive. the equivalent point loads are as illustrated in Fig.11) where the second derivative is found by differentiating equation (2.14(c) where the profile is lowered in AB and BC while maintaining its position at the support points.14) This too is illustrated in the figure. 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. 2.4: Qualitative profile design A prestressed concrete slab bridge is to be reinforced with 10 post-tensioned tendons. 2. .13(a) can be adjusted by changing the end eccentricities. in fact. In a determinate structure. the equivalent uniform loading due to prestress is a function only of the sag and is. Example 2. increases the (sagging) moment due to prestress. the change only results in adjustments to the equivalent point loads at A and B and to the equivalent loading near B. This increase in tendon eccentricity. As these forces are at or near supports. e A and eB while keeping the sag.14(a). Example 2. Such an adjustment is known as a linear transformation and will have no effect on the intensity of equivalent uniform loading as can be seen from equation (2. 2. illustrated in Fig. Hence. as illustrated in Fig. 2. This is because the eccentricity at B has been increased without increasing the tendon sag in the spans. As was seen above. does little to increase the compressive stress at the top fibre at that point. which increases the compressive stress at the top fibre. s. in an indeterminate structure. However. e.14.Page 58 As B is on the right-hand side. unchanged. Thus. This phenomenon is particularly useful for understanding the effect of prestressing in continuous beams with profiles that vary parabolically in each span.14(b). 2. 2. they do not significantly affect the distribution of bending moment induced by prestress. stress at the top fibre can be increased by moving the prestressing tendon upwards to increase the eccentricity locally.14). A profile such as that illustrated in Fig. increasing the eccentricity locally at B without changing the sags. the response of a structure to such changes is not so readily predictable. A more appropriate revision is illustrated in Fig. Pe.13): (2. This has the effect of increasing the tendon sag which increases the intensity of equivalent uniform loading. The intensity of uniform loading is given by equation (2. In the structure of Fig.13(b). The preliminary profile for the tendons.

The intensities of loading are found from equation (2.Page 59 Fig. 2.15(a). Example 2.5: Tendon with constant prestress force A three-span bridge is post-tensioned using a five-parabola symmetrical profile. For the first parabola: . (b) raising of profile at B by linear transformation.14 Adjustment of tendon profile: (a) original profile. It is required to determine the equivalent loading due to prestress assuming that the prestress force is constant throughout the length of the bridge. This can be seen in the following example.11). 2. (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.

This is necessary to ensure that the tendon does not generate concentrated forces at these points. prestress forces are not constant through the length of bridges because of friction losses. 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.15(b). 2. Note that in selecting the profile. Differentiating the equation for the parabola gives the slope. it has been ensured that the parabolas are tangent to one another at the points where they meet. (b) equivalent loading due to prestress Similarly.Page 60 Fig. 2.7.5: (a) partial elevation showing segments of parabola. 2. This is illustrated in Fig. Verifying that these forces are in equilibrium can be a useful check on the computations. from which the force is found to be: All of the equivalent loads due to prestress are illustrated in Fig. 2.15 Tendon profile for Example 2.16(a) .2 Prestress losses In practical post-tensioned construction.

2. It is required to determine the equivalent loading due to prestress taking account of the loss of force.16 Equivalent loading due to varying prestress force: (a) segment of beam and tendon. the moment due to the equivalent loading should be equal to the product of prestress force and eccentricity at all points.6: Tendon with varying prestress force The post-tensioning tendon of Example 2.6 that this equivalent loading satisfies equilibrium of forces and moments.5 is subject to friction losses which result in the prestress forces presented in Fig.16(b). The eccentricities given in this figure have been calculated from the equations for each parabola given in Example 2. In such a case. The resulting loading is illustrated in Fig.Page 61 Fig.11). It will be seen in Example 2. Example 2.5. 2. the difference between prestress forces at adjacent points is generally not very large. Therefore. a sensible approach to the derivation of equivalent prestress loading is to start by substituting the average prestress force for P in equations (2. The use of equivalent loads which do not satisfy equilibrium can result in significant errors in the calculated distribution of prestress moment.9)–(2. The bridge is post-tensioned from both ends with the result that the prestressing forces vary symmetrically about the centre. However. A useful method of checking the equivalent loads is to apply them in the analysis of a determinate beam. 2.17. (b) equivalent loading (Pav=(P1+P2 )/2) where the forces at points 1 and 2 are different. .

2.Page 62 Fig. 2. point loads must be applied at the end of each segment in accordance with Fig.17 Tendon profile showing varying prestress force (in kN) and eccentricity (in m) With reference to Example 2.5 but using average prestress forces. In segment AB. x=0.1322 and the upward force is: . the equivalent intensities of uniform loading are: In addition. the slope is − 0.16(b). the equation for the parabola is: At A.

Page 63 Fig. 2. It can be verified that the forces and moments on each segment are in equilibrium. (b) total the minus sign indicating that the force is actually downwards.7. The forces and moments at the ends of each segment are summed and the result is illustrated in Fig.18(b). together with the other equivalent uniform loads. 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. The resulting equivalent loading is illustrated in Fig. .18 Equivalent loading due to prestress: (a) loading on each segment.3 Non-prismatic bridges The eccentricity of a prestressing tendon is measured relative to the section centroid.19(b) where s is distance along the centroid. the slope of the profile is: giving a downward force at the right end of magnitude: The corresponding point load components for the other segments of parabola are calculated similarly and are presented. In nonprismatic bridge decks.19(a). 2. 2. the location of this centroid varies along the length of the bridge. This clearly affects the eccentricity and hence the moment due to prestress. 2. 2. 2. In such a beam.18(a). At B. in Fig.

AB and BC.19 Equivalent loading due to variation in location of centroid: (a)segment of beam and tendon. The beam is divided into just two segments.12). i.9: Equivalent loading due to change in geometry The beam illustrated in Fig. 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). In addition. The definition of the parabola is independent of the section geometry. (b) equivalent loading Example 2.e. it is defined by an equation of the same form as equation (2. the centroid changes depth linearly between A and B and between B and C.Page 64 Fig. 2.20 has a non-prismatic section. It is prestressed with a tendon following a single parabolic profile from A to C. 2.3. It is required to determine the equivalent loading due to prestress.: . 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. the eccentricity is given by: Differentiating the equation for segment AB gives: Similarly. the derivatives are: . Hence for segment AB: Similarly for segment BC. 2. for BC.Page 65 Fig.

for both segments: The average values for prestress force in segments AB and BC are 0. 2.91P respectively.97P and 0.21(b). The resulting equivalent loading due to prestress is illustrated for each segment in Fig. 2. The forces are combined in Fig. (b) total Differentiating again gives.21 Equivalent loading: (a) loading on each segment. where P is the jacking force.Page 66 Fig. .21(a). 2.

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

Appendix A gives the BMDs for members with a range of end conditions. 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. Table 3. The fixities are numbered and the direction of each is defined. 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.1 Moment distribution General Example Step 1: All members of the structure are isolated from one another by applying a number of fixities. This fixed structure is equivalent to: The resulting bending moment diagram (BMD) is found (with reference to Appendix A): .

boxed below. i. Dividing the BMD by this gives the normalised version. Appendix B gives the BMDs for a wide range of such displacements. These BMDs are then normalised to give a unit value at each point of moment discontinuity. The total discontinuity at B is 3EI/l+4EI/(1.e.25l)=6.2EI/l. 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: .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.

125)wl2=0. factored by 1. 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. This is performed simply by adding or subtracting the normalised bending moment diagrams. This is the final solution.Page 70 General Step 3: In the fixed BMD (Step 1). Hence.006wl2 (the BMD is subtracted. the correction of the discontinuity at C had the effect of reintroducing a discontinuity at B. as adding it would increase the discontinuity).125wl 2 is less than that just right of B (0. Example The moment just left of B in the fixed BMD (1) of 0. there is generally a lack of equilibrium of bending moment at the fixing points. factored by the discontinuities. often characterised by discontinuities in the BMD. the process of adding normalised BMDs. This discontinuity is removed by adding the normalised BMD corresponding to rotation at B (2). The resulting BMD is: The discontinuity at C is now (0.006wl 2.005wl2 . scaled in each case by the appropriate discontinuity. The correction of the discontinuity at B had the effect of increasing the discontinuity at C. such discontinuities are successively removed by applying ‘rotations’.131–0. . For this particular example. to get an exact answer. In this step.005wl 2.130wl 2) by 0. Similarly. the first iteration has resulted in a BMD which is sufficiently accurate for most practical purposes. must be repeated until no discontinuity remains. This is removed by subtracting the normalised BMD corresponding to rotation at C (3). factored by 0.

as illustrated in Fig. The fixed bending moment diagram (BMD) (Step 1) is unaffected by the symmetric system of fixities but. As will be demonstrated in this example. as before. The resulting BMD (from ) is illustrated in Fig.1: Continuous beam using symmetry Concepts of symmetry can be used to great effect when analysing by moment distribution. 3.005wl 2. (c) BMD due to unit rotation. this time using symmetry. The resulting BMD is illustrated in Fig. is illustrated in Fig.1 is analysed again.2 is. scaled by 0.1(d) (boxed).1(c) and the normalised version. Fig. (b) unit rotation simultaneously at B and C. The beam of Table 3.1(a). exact. The beam is fixed simultaneously at B and C as illustrated in Fig. The arrows indicate the directions of positive rotation for Step 2.Page 71 Example 3. two equal and opposite rotations must be applied simultaneously at B and C. no iteration is required for this example and the BMD of Fig. (d) normalised BMD .005wl2. Hence. 3. in Step 2. 3. 3. 3. 3. in fact. which gives a unit discontinuity of moment at B and C.1 Moment distribution using symmetry: (a) symmetrical system of fixities. 3.1(b). Step 3 consists of removing these two discontinuities (simultaneously) by adding the BMD of Fig. As there are no further discontinuities.1(d).2. The discontinuities at B and C in the fixed BMD are. 3. 0. it is possible to isolate members from each other by the simultaneous application of a pair or pairs of equal and opposite fixities.

For simplicity. Step 1: The members are isolated by applying fixities at A. as illustrated in Fig. from . 3.4(f).4(f). 3.4(Pl/8). C. The fixed BMD is.5(b). 3.3 Box culvert example . Step 2: Applying unit rotation simultaneously at A and B (Fig. 3. The normalised BMD due to rotation at C and D is found similarly and is as illustrated in Fig. The resulting BMD is illustrated in Fig.4(e). When normalised. Symmetry is exploited by simultaneously fixing A and B and simultaneously fixing C and D as shown.4(d). which gives the BMD of Fig. and D as illustrated in Fig.4(b)) is Pl/8. 3. 3. Step 3: The discontinuity at A and B in the fixed BMD (Fig.3. 3.2 Final BMD for three-span beam Example 3. factored by Pl/8.Page 72 Fig. The discontinuity now present at C and D is 0. 3. 3.4(a). 3. factored by that amount. 3. B.2: Box culvert The application of moment distribution to a two-dimensional frame type of structure is demonstrated using the box culvert illustrated in Fig. 3. This is corrected by adding the BMD of Fig.4(e). this culvert is assumed to be supported at two discrete points under the walls and to have constant flexural rigidity throughout. Fig. this becomes the BMD of Fig. 3.4(b).5(a).4(c)) results in the BMD illustrated in Fig. This is corrected by applying the BMD of Fig. 3.

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

gives the BMD of Fig.16(Pl/8). A more realistic finite-element (FE) model taking account of these effects and assuming typical soil properties is illustrated in Fig.6 Finite-element model of box culvert and surrounding soil: (a) finite-element mesh. 3. The interaction of bridges with the surrounding soil is considered further in Chapter 4. the BMD of Fig.4(e). 3. factored by 0. factored by this amount. Adding the BMD of Fig.064(Pl/8). Finally. Earth pressure on a structure of this type generates an additional distribution of moment. This box culvert of Example 3.6(a) and the resulting BMD in Fig. 3. 3. 3.5(d) is deemed to be the final solution. Fig. A higher hogging moment (0.5(d).5(c). 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.6(b). is added to give the BMD of Fig.Page 74 Step 4: The correction at C and D has reintroduced a discontinuity at A and B of 0. 3.4(f). (b) resulting BMD .825(Pl/8)) is found with a corresponding reduced sagging moment. A more typical situation would be that of continuous support from granular material throughout the length of the base and side walls.2 was assumed to be supported at two discrete points. 3. 3.

a rotation at C results in the normalised BMD of Fig. (b) imposed support settlement . is to assume that a foundation support settles by a specified amount. the structure and the surrounding soil may be represented using non-linear computer models. 3. Step 2: Applying a unit rotation as illustrated in Fig. following example serves to demonstrate the effect of a differential settlement on a continuous beam bridge. By coincidence.3: Differential settlement by moment distribution The continuous beam illustrated in Fig. Hence. many structural engineers treat the soil as a spring or a series of springs in the numerical model. The system of fixities cannot be symmetrical as the ‘loading’ is not symmetrical. EI. The . Clearly soil deforms under the vertical forces applied through bridge piers and abutments. the fixed BMD is as illustrated in Fig. Δ relative to the others and to determine the effects of this on the structure. 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.8(a). 3.9(c). Fig. the discontinuity at C is of the /l same magnitude. Example 3. Thus. When normalised.Page 75 3.5EI∆ 2. 3.7 is subjected to a settlement at B of Δrelative to the other supports. Similarly.8. 3.7 Three-span beam example: (a) geometry. If the deformation is not uniform. The resulting BMD is required given that the beam has uniform flexural rigidity. 3. frequently adopted by bridge engineers. However. To accurately analyse for this effect. this becomes the BMD of Fig.9(d). an alternative approach. 3. 3. 3.3 Differential settlement of supports There is considerable research and development activity currently taking place in the field of soil/structure interaction. The discontinuity of moment at B is 1. as the effect is often not very significant.9(a) results in the BMD illustrated in Fig. Step 1: Referring to (4th and 5th BMDs). distributions of bending moment and shear are induced in the deck.9(b). the beam is fixed as illustrated in Fig.

(b) fixed BMD Fig.8 First step in analysis of three-span beam: (a) system of fixities.Page 76 Fig. (d) normalised BMD associated with rotation at C . 3. 3.9 Effect of rotations at points of fixity: (a) unit rotation at B. (c) normalised version of BMD associated with rotation at B. (b) BMD associated with unit rotation at B.

Page 77 Fig. 3.10(c). which are typical of differential settlement: 1. 3. 3.9(d) factored by this amount to give the BMD of Fig. Adding the BMD of Fig. I. It is usual to size a bridge by selecting a depth which is proportional to span length (i. Differential settlement has the effect of generating sagging moment at the support which settles. (c) after second corrections at1 and 2 Step 3: The discontinuity at B in Fig. This correction at B has the effect of increasing the discontinuity at C to 1. This is important as supports in continuous beams are generally subjected to hogging moment and are often not designed to resist significant sag. 3. depth=l/k for some constant. /l 3.e. 3.8 EI∆ 2. 3. l.9(c). (b) after correction at 2. The moment at the support which settles is proportional to the second moment of area. This is corrected by adding the BMD of Fig.10(a). divided by the square of the span length. Step 4: One further iteration gives the BMD of Fig. k).10 BMD after successive corrections: (a) after correction at1.10(c) which is deemed to be of sufficient accuracy.10(b). /l factored by this amount. It is interesting to note two additional things about the final BMD illustrated in Fig. gives the BMD of Fig.8(b) is 1.5 EI∆ 2. 3. As the second moment of area is proportional to the cube of .

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

Fig.12(a). (b) fixed axial force diagram . where a beam is partially restrained. 1.30). 3. This happens for example in arch bridges where contraction is accommodated through bending in the arch (Fig. (b) deformed shape after expansion of deck Fig. Example 3. It also happens in frame bridges where the piers offer some resistance to expansion or contraction of the deck. shear force and axial force diagrams due to an increase in deck temperature of Δ T.12 Frame subjected to axial change in temperature: (a) original geometry.Page 79 The most common case requiring analysis is the one in between the two extreme cases described above. it is required to find the bending moment.13 First step in analysis of frame: (a) fixing system.4: Restrained axial expansion by moment distribution For the bridge illustrated in Fig. 3. 3.

BD. (c) associated shear force diagram. 3. bending moment is generated in the pier. As the pier is fully fixed at its base. (b) associated axial force diagram. In addition. a thermal expansion tends to bend the pier as illustrated in Fig. (e) normalised free body diagram . (d) free body diagram showing lack of equilibrium of forces at B.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. Thus. 3.12(b).14 Effect of translation at fixing point: (a) forces required to induce unit translation. its resistance to bending restrains the expansion a little and generates a small compressive stress in the deck between A and B. Fig.

(While a rotational fixity at this point is also possible. 3. The required external force at B can be seen in the free-body diagram of Fig. 4th case) of 3EI/h3 giving a total required force at B of 1503EI/h3. 3.14(e). Fig.13(a). 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. However. (b) axial force diagram. 3.15 Results of analysis: (a) free body diagram with restored equilibrium at B. The normalised version of Fig. The (∆ is corresponding force is α T)E(area)=6000α T)EI/h2. (d) bending moment diagram . 3.14(a). 3. Hence the axial force diagram is as (∆ (∆ illustrated in Fig. 3.) There is no bending moment or shear force in the fixed structure. the stress in AB is α T)E.14(b) and (c). corresponding to unit discontinuity of force at B.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. 3.13(b). a force is required to bend BD (.14(d). 3. The associated axial force and shear force diagrams are illustrated in Figs. is illustrated in Fig. In addition. The lack of force equilibrium in this diagram corresponds to the moment discontinuity in the BMDs of the usual moment distribution problems. such a fixity is not necessary to isolate the members in this case. where α the coefficient of thermal expansion and E is the elastic modulus.14(d). (c) shear force diagram.

15(b) and (c). 3.13(b)) as there is no axial force in BC and no shear force in the pier to correspond to the axial force in AB. 3. As the shear force across the pin at B is 12α T)EI/h2. 3. The area of the deck is typically numerically much larger than the second moment of area of the pier with the result that the restraint to deck expansion is relatively small. (∆ As there is no further force discontinuity.14(e) factored by 6000α T)EI/h2. may be relieved by the effect of creep. The result is illustrated in Fig. Fig. by adding an axial tension in AB of 0. i. The relative 2 flexural rigidities are given on the figure and the area of the deck is 500I0/l . 3. This situation is corrected at B by subtracting the forces illustrated in Fig. Hence the rise in temperature results in a lot of strain and in very little stress in the deck.15(a). The sustained stresses generated by the subsequent contraction of the concrete as it cools can be relieved substantially by creep.002×6000α T)EI/h2 to Fig. However. This means that such stresses. Example 3.e.5: Thermal contraction in frame bridge by moment distribution The frame structure illustrated in Fig. this is the final free body diagram. 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 resulting distribution of bending 6 moment is required given that the coefficient of thermal expansion is 12×10− .16 Integral frame of Example 3. 3.16 is integral having no internal bearings or joints. if sustained in a concrete structure. Most noteworthy is the effect of the relative values of deck area and pier second moment of area. 3.Page 82 Step 3: There is a lack of force equilibrium in the fixed structure at B (Fig. the moments and forces due to changes in temperature are proportional to the elastic modulus. thermal contraction or expansion induces bending moment as well as axial force and shear. as for differential settlement.5 .13(b).998×6000α T)EI/h 2 and a shear (∆ (∆ force in BC of 0. 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. (∆ There are some points of interest about axial temperature effects apparent from this simple example. It is also of interest to note that. in-situ concrete bridges generate significant quantities of heat while setting and consequently have their initial set when the concrete is warm. 3.15(d). As a result. 3. Substantial temperature changes occur on a short-term basis during which the effects of creep do not have a significant ameliorating effect. The final axial force and shear force diagrams are illustrated in Figs.

when divided by 7.17(a)) requires the moments and forces illustrated in Fig. As there is axial force in ABC but no corresponding shear force in AD or CF. Due to symmetry. 3. 3. However. 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. 3.17(c).18(c). In the fixed structure. the fixities at A and C are taken to be equal and opposite as illustrated. as points A and C will tend to rotate as well as translate. No distribution of bending moment is present in the fixed structure but the axial force diagram is as illustrated in Fig. However. 3.17 First step in analysis of frame: (a) system of fixities.18(d). one translational and one rotational as illustrated in Fig.2EI0/l and is illustrated in Fig.12EI 0/l . As the rotation is applied while fixing against translation. there is a lack of force equilibrium at A and C as illustrated in Fig. 3. 3. 3.17(b).Page 83 Fig. The normalised version is found by dividing by 7. 3.19(a) and generates the BMD illustrated in . there is no tendency for point B to rotate and this point can be considered fixed without applying a fixity. Hence. 3. Step 2: Applying a unit rotation in Direction 1 (Fig. 3. no axial forces are generated in the members. two fixities are needed at each. is as illustrated in Fig. there is a lack of force equilibrium at A and C which. there is a shear force just below A and C which is unmatched by an axial force in AB or BC. (c) free body diagrams showing shear and axial forces in fixed frame Step 1: Due to symmetry. (b) fixed axial force diagram. Applying a unit translation in Direction 2 (while preventing rotation) requires the moments and forces illustrated in Fig.18(b).18(a) and generates the BMD illustrated in Fig.17(a).2EI0 /l.

3.Page 84 Fig. adding it to a BMD of zero. In addition.12EI0/l 2 and. 3. the joint forces of Fig. It can be seen in Fig.18 Effect of rotation at A and C: (a) moments and forces required to induce unit rotation. 3. Normalising with respect to this value gives Figs.19(b). 3. The normalised lack of joint equilibrium is illustrated in Fig. a distribution of axial force is generated which is illustrated in Fig. 3. Step 3: The lack of force equilibrium in the fixed structure illustrated in Fig. (b) BMD associated with unit rotation. 3.20(c).19(c). 3. 3. (d) normalised shear and axial forces associated with rotation Fig.19(d) that there is a shear force just below A which is not matched by the axial force to its right. as there is no moment induced in the fixed structure.17(c) is corrected by factoring Fig.20(c) .4EI 0/l3. The discontinuity or lack of equilibrium at A is 538. In addition. 3. (c) normalised BMD associated with rotation.20(a) by 0.20(a) and (b).

3.20(a) and (c) are Fig. (c) axial force diagram associated with unit translation.18(c) and (d) by 0. (b) BMD associated with unit translation. The results are illustrated in Fig. Step 4: The removal of the moment discontinuity reintroduces a lack of force equilibrium which is evident in Fig. 3. 3. 3. There is a discontinuity in the BMD (or lack of moment equilibrium) at A and C evident in Fig. The results are illustrated in Fig.21. 3.21(a). This is corrected by scaling Figs. 3.00214EI0/l and adding them to Figs.21(a) and (b) respectively.17(c). (d) free body diagram at joint A showing lack of equilibrium .19 Effect of translation at A and C: (a) forces and moments required to induce unit translation.22.00285EI0/l2. 3.Page 85 must be factored by 0. 0. 3.22(b) of magnitude. Figures 3.12EI 0/l2 and added to those of Fig.

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

(b) corrected free body diagram Fig. 3. This leads to Fig.23(a) is therefore adopted as the final BMD. (b) free body diagram scaled by this amount and added to Figs.18(c) and (d).23 Results of analysis for effects of thermal contraction: (a) BMD. 3.22(a) and (b).22 Effect of correcting for discontinuity in BMD: (a) corrected BMD. 3.Page 87 Fig.23 where the lack of force equilibrium is deemed to be sufficiently small. 3. . Figure 3. 3. The resulting moment discontinuity is corrected by factoring and adding Figs.

While the method may not at first seem to be any simpler to apply than the procedure used above. Stage B—Analyse for the effects of the equivalent loads: The beam is analysed for the loading illustrated in Fig. will generate both. 3. The equivalent loads for this example are illustrated in Fig. even on an unrestrained beam.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.e. The equivalent loads method consists of three stages as follows.25(a). This can become a significant factor in bridge deck design. where: where α the coefficient of thermal expansion. a more important effect of the moment connection is the bending moment induced in the deck by thermal movement. 3. An axial expansion can be generated in an unrestrained beam by applying an axial force.24 Beam on rollers with partial (spring) restraint .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. on the other hand.5 serves to illustrate the effect of a moment connection between the bridge deck and the piers. However.25(b). sectional area and E is the elastic modulus of the beam. 3. that distribution of stress which is inadvertently introduced into the structure by the equivalent loads.4. 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. it is necessary to identify the ‘associated stresses’. Example 3. this distribution of stress must be subtracted to determine the stresses generated indirectly by the change in temperature. F 0. To some extent this alters the resistance to contraction or expansion. The equivalent force. temperature on an unrestrained is member generates strain but not stress. In Stage C. However. In such a case. Therefore. i.25(a) and the associated stress distribution in Fig. Normally this stage would be done by computer but it Fig.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. 3.Page 88 Example 3. thermal movement is resisted by bending in both the piers and the deck. 3. 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.

This is the final result and is what one would expect from a thermal expansion in a partially restrained beam.25(d). It is well known that. Hence the force is taken in the ratio 1:2 as illustrated in Fig. The result is an axial compression of F0/(3A) throughout the beam. 3. it is resisted in proportion to their stiffnesses. For this example. 3. when a load is applied to two springs. 3. If a linear distribution of this type is applied to a simply supported single-span beam. the top tends to increase in temperature faster than the bottom. (d) stress distribution due to temperature change is trivial for this simple example. the bending takes place freely and the beam curves upwards as the top expands relative to the bottom. Stage C—Subtract the associated stresses: The distribution of associated stresses is subtracted from the stresses generated by the equivalent loads. 3. Thus. 3. the beam acts as a spring of stiffness AE/l. a differential temperature distribution develops which tends to cause the bridge to bend.25(c).5 Differential temperature effects When the sun shines on the top of a bridge. In this case.25(b) from that of Fig. (c) equilibrium of forces at spring. 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.25 Analysis by equivalent loads method: (a) equivalent loads. (b) associated stress distribution.Page 89 Fig. This corresponds to the case of a beam on rollers subjected to an axial increase in temperature in that strains take . 3.25(d). strain is generated but also some compressive stress. this consists of subtracting the axial stress distribution of Fig.

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

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

a. 3. The bending component will result in some moment but not as much as would occur if the beam were totally prevented from bending. 3. The temperature distribution is first converted into a strain distribution by multiplying by the coefficient of thermal expansion. due to symmetry. i. 3. 3. Hence.29 is subjected to an increase in temperature which varies linearly from a maximum of 20° at the top to 10° at the bottom.30: Fig. as illustrated in Fig. 3. it is effectively fixed as illustrated in Fig.28(b) gives the final result illustrated in Fig.30. 3.28(b).28(c). the central support point.27(b) from Fig. a strain but no stress.8: Differential temperature change in continuous beam The three-span beam illustrated in Fig. does not rotate. The BMD due to the applied equivalent loading is as illustrated in Fig. from Fig.29 Differential temperature example Fig.28(a) and the solution can be determined directly from Appendix B. 3. B. the elastic modulus is E and the second moment of area is I. Stage A—Calculate the equivalent loads and the associated stresses: In this example. The depth of the beam is h and the centroid is at mid-depth. As the beam is free to expand.Page 92 Stage B—Analyse for the effects of the equivalent loads: Analysis of a symmetrical two-span beam is trivial because.30 Resolution of applied change in strain into axial and bending components . It is required to determine the BMD due to the temperature increase given that the coefficient of thermal expansion is α . 3. The distribution is then resolved into two components.e. 3. 3. the axial component will result in a free expansion. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig. the curvature is. axial strain and bending strain. Example 3. The BMD will be determined using the method of equivalent loads.

Normally this stage would be done by computer but it will be done using moment distribution for this simple example. the equivalent loads and associated BMD are as illustrated in Fig. 3. 3. Fig. (c) BMD in fixed structure . (b) applied loading on fixed structure.31 Application of equivalent loads method: (a) equivalent loads. the equivalent moment becomes: Thus.Page 93 Hence.31(a).31. Fig. 3.32 First step in analysis by moment distribution: (a) system of fixities. 3. (b) associated BMD Stage B—Analyse for the effects of the equivalent loads: The frame is analysed for the loading of Fig.

34 Completion of equivalent loads method: (a) BMD due to analysis by moment distribution.33 Effect of rotation at fixing points: (a) moments required to induce unit rotation. (b) BMD associated with unit rotation.32(a). the fixities at B and C are equal and opposite as indicated in Fig. (b) BMD after subtraction of associated BMD .32(c).Page 94 Step 1: The beam is fixed at B and C in order to isolate the three spans. 3. 3. Fig. 3. 3. (c) normalised BMD associated with rotation Fig. The applied loading on the fixed structure is illustrated in Fig. As the ‘loading’ is symmetrical. 3.32(b) and the resulting BMD (Appendix B) in Fig.

Fig. The cross-section and temperature distribution for this examp le are identical to those of Example 2. Step 3: The discontinuity of moment at B and C evident 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. 3.33(b).35 Bridge diaphragm example: (a) plan of geometry. 3. (c) applied temperature distribution . The upward reaction from the bearing due to 6 the dead load is 300 kN.9: Bridge diaphragm The bridge diaphragm illustrated in Figs. 3. This is the BMD due to the differential temperature increase.35(a) and (b) is subjected to the differential increase in temperature shown in Fig. 3.Page 95 Step 2: Unit rotation at B and C requires the application of the moments illustrated in Fig.35(c). 3. no further iteration is required. Example 3. 3. 3.33(a) and generates the BMD illustrated in Fig.32(c) is corrected by adding Fig. /h. It is required to determ if there will be uplift at B c due to combined temperature and dead load. Step 4: As no discontinuity now exists. 3.33(c). 3. The normalised version is illustrated in Fig.34(a) gives the final result illustrated in Fig.34(a). (b) section through diaphragm. 3. Stage C—Subtract the associated stresses: Subtracting the associated BMD of Fig.34(b). Referring to that example. 3. the coefficient of thermal expa is 12×10 − and the modulus of 2 elasticity is 35 000 N/mm .33(c) factored by 5EIα The result is illustrated in Fig.1 (Chapter 2). 3.31(b) from Fig.

Point B does not rotate and is effectively fixed.36(d). (c) results of analysis. there is no uplift of this bearing due to the differential temperature change. 3. (b) associated BMD. 3.36(a).36(c). 3.Page 96 Fig. 3. Subtracting the associated BMD gives the final BMD illustrated in Fig. (d) final BMD Fig.36 Analysis to determine effect of imposed differential temperature: (a) equivalent loading. the structure is analysed for the loading illustrated in Fig. 3. As the reaction due to dead load exceeds this value. The associated BMD is illustrated in Fig.37: Hence the reaction at B is 80+80=160 kN. . 3.36(b). By symmetry. 3. The reactions at A and C can be found from the free body diagram illustrated in Fig. Hence (as in Example 3.37 Free body diagram for diaphragm beam To determine the reaction due to this moment.7) the BMD due to applied loading is as illustrated in Fig.

38(c). 11.38(c) gives the final distribution of moment due to restrained bending illustrated in Fig. (d) final BMD Thus. 3.2.718α for half of the bridge.32α at the top and bottom fibres E E respectively. Hence. Fig. a sagging bending moment is induced over the central support. . 3.Page 97 Example 3. It was established in Example 2. (c) results of analysis.87α E.29α and 11.38α and E − 5. (b) associated BMD.38(b) from Fig.077α which E gives stresses (tension positive) of − 5. 3.38(b). Using the method of equivalent loads: E Stage A: The equivalent loads are illustrated in Fig. It is required to determine the maximum stresses due to the differential temperature change.38(d). B. 3. Stage C: Subtracting the associated BMD of Fig.38α 10.67α At the bottom fibre the total stress is E− E=− E. 3. it was established that the equivalent moment due to the temperature change is − 0. Stage B: Analysis by computer or by hand gives the BMD illustrated in Fig. the total E stress at the top fibre is − 5.2 that the residual stresses are − 5.38 Analysis to determine effect of differential temperature change: (a) equivalent loading. of 1.32α 5.45α (restraint to expansion induces compression at the extreme fibres).2 consists of two 10 m spans.29α 5.10: Differential temperature in bridge of non-rectangular section The beam-and-slab bridge whose section and temperature loading is described in Example 2.38(a) and the associated BMD in Fig. In Example 2. 3.45α E− E=5. 3.

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

3.40 Model of pedestrian bridge: (a) geometry showing differences in level of centroids.033 m below the top fibre for the solid and hollow sections respectively (Figs. (b) equivalent loading. Summing products of stress and area in Fig. (c) associated axial force diagram.39(b)–(d)).5 m and 1.Page 99 By summing moments of area it is found that the centroids are 0.39(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. (d) associated BMD . 3. 3.

If there is restraint to either or both rotations. 3. 3. 3.40(b).Page 100 In the hollow section. The associated axial force and bending moment diagrams are illustrated in Figs. Similarly. 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. a second moment of area several times as large as the maximum used elsewhere in the model (e. the equivalent loads are 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. 3.40(c) and (d) from the results.40(b) and subtracting the associated distributions of Figs. 3.41 is articulated as shown in Fig.41(a) to allow axial expansion in both the X and Y directions.12: Differential temperature The slab bridge of Fig. it tends to bend about both axes. Therefore. ten times) generally provides sufficient accuracy without causing such problems. an equivalent force of: For the hollow section. 3.64 m 2 and the equivalent force is: A model which allows for the difference in the level of the centroids is illustrated in Fig. Example 3.5. However. respectively. using members with very large stiffnesses can generate numerical instability in a computer model. the particle tends to expand in all three directions. bending moment results about both axes as will be illustrated in the following example. the area is 2. 3.39(b) and (d).1 Temperature effects in three dimensions When the temperature of a particle of material in a bridge is increased.g. the area is 2. However. 3.40(a). when a differential distribution of temperature is applied through the depth of a bridge slab. the bridge is two-span .6 m 2 giving E.40(c) and (d). for rotation. Noting that the axial effects apply to all members while the differential temperature distributions only apply to the deck (abcd). 3. Note that the short vertical members at b and c could be assumed to have effectively infinite stiffness.6×1=2.

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

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

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

The BMD due to applied ‘loading’ on the fixed structure is zero everywhere as the prestress forces are applied at fixing points.8 m deep rectangular section giving a moment about the centroid of: As M3 is applied to the outside of the cantilever. results in changes of both stress and strain. 3. 3. As these applied moments generate distributions of longitudinal and transverse moment. ABC. there is one important distinction.47(a) and (b) (unit discontinuity in moment) and in Figs.13: Frame subject to axial prestress by moment distribution The frame of Fig. P.5 is used again here as illustrated in Fig.47(c) and (d) (unit discontinuity in force). 3.45. 3. . if a beam rests on a sliding bearing at one end. Example 3. When the movements due to prestressing are unrestrained.46(b). of magnitude. Prestress.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. However. the differential distribution is applied to a 0. Step 2: The effects of inducing rotations or translations at the fixing points are the same as for Example 3. it can undergo axial changes in temperature without incurring any axial stress.16. on the other hand. The frame is analysed by moment distribution.5. the problem is completed by analysing the slab (by computer) and subtracting the associated BMDs from the solution. The normalised versions are presented here in Figs. 3. only (M4− 3) needs to be applied at the M deck/cantilever interface as illustrated in Fig. the stress distributions are easily calculated and analysis is not generally required. 3. reproduced here as Fig.43. Step 1: The system of fixities used in Example 3. However. However. prestressing that beam does (as is the objective) induce a distribution of stress. there are two associated BMDs as illustrated in Fig. 3.46(a). It is required to determine the net prestress force in the deck and the resulting BMD.Page 104 In the main deck. 3. As for the previous example. there are many bridge forms where the effects of prestress are restrained to some degree or other and where analysis is necessary. An unrestrained change in temperature results in a change in strain only and no change in stress. is subjected to a prestressing force along the centroid of the deck. For example.

(c) normalised BMD due to translation. 3.Page 105 Fig. (b) normalised forces due to rotation. (d) normalised forces due to translation . 3.46 Frame subjected to prestress force: (a) geometry and loading.47 Effect of displacements at fixing points: (a) normalised BMD due to rotation. (b) system of fixities Fig.

(b) internal forces after correction for force equilibrium. 3. no further iteration is deemed necessary. It can be seen in the results. illustrated in Figs.47(c) and (d) by P.48(d) is satisfied to a reasonable degree of accuracy.48 Effect of prestress force: (a) BMD after correction for force equilibrium. Step 4: As force equilibrium in Fig.48(a) and (b). (d) internal forces after correction for moment equilibrium Step 3: The translational fixity is released first to apply the prestress force.0178Pl and adding to give Figs. 3. . 3. 3. The discontinuity of moment which results is removed by factoring Figs. 3. that equilibrium of forces at A and C is then satisfied.48(c) and (d).Page 106 Fig. This consists simply of factoring Figs. (c) BMD after correction for moment equilibrium.47(a) and (b) by 0. 3.

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

3. 3. Fig. (b) BMD associated with unit rotation. The result is illustrated in Fig.52(c) by Pe/2 and adding. the resulting BMD in Fig.52 Effect of rotation of fixing points: (a) moments required to induce unit rotation. 3. 3.51(b) is removed by factoring Fig. 3.53.51(b). this is the final BMD due to prestress.52(a). Step 2: The moments required to induce unit rotation at B and C are illustrated in Fig. (c) normalised BMD . Step 3: The discontinuity of bending moment evident in Fig. (b) fixed BMD Fig. The BMD in the fixed structure due to the equivalent loading is as illustrated in Fig. As there is no further discontinuity.Page 108 Step 1: The beam is fixed as illustrated in Fig. 3.52(b) and the normalised BMD in Fig. 3.51 First step in analysis by moment distribution: (a) system of fixities. 3.51(a).52(c). 3. 3.

P.54.53 Final BMD due to eccentric prestress force It is interesting to note from Example 3. Example 3. (This was covered in greater detail in Chapter 2. For preliminary design purposes.15: Profiled tendons In most post-tensioned bridges the tendons are profiled using a combination of straight portions and parabolic curves. For this beam. the actual profiles are sometimes approximated by ignoring the transition curves over the internal supports as illustrated in Fig.) For the parabola in Span AB. the intensity of which can be determined by considering equilibrium of forces at the ends of the parabola. the slope is found by differentiating the equation as follows: Fig. A parabolic profile generates a uniform loading. 3.54 Beam with profiled prestressing tendon .14 that the effect of the tendon below the centroid is to generate sagging moment in the central span. In a simply supported beam. 3.Page 109 Fig. 3. it is required to find the BMD due to a prestress force. a tendon below the centroid generates hogging moment.

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

3.00833Pl in Fig.Page 111 At A. They are both equal to 0.52(c). vertical loading induces torsion as will be demonstrated in this example. which is illustrated in Fig.12 and the vertical component of prestress is 0. x=0 and the slope becomes − 0. the complete equivalent loading due to prestress is as illustrated in Fig. In most practical grillages. Step 1: The two members are isolated from each other by the fixing of point B. As there is no further discontinuity. Hence. w.56(b). 3.55(b). significant vertical translational displacements occur at the joints. . The bridge is subjected to uniform vertical loading of intensity.56(c). Thus. 3. 3. wBC=0. EI. As a result. this is the final BMD due to prestress in this beam.2P/l. at x=l.16:Torsion due to vertical loading When bridges are curved or crooked in plan. 3. 3. GJ. 3. equilibrium of vertical forces requires a uniform loading of intensity: In BC. In this chapter. is 1. The torsional rigidity.56(b). 3.57(a). only those grillages are considered in which there is no such joint displacement. 3. The result is illustrated in Fig. by symmetry.7 Application of moment distribution to grillages A great many bridges are analysed by computer using the grillage analogy. 3. Step 1: The symmetrical system of fixities is illustrated in Fig. the continuous bridge slab is represented by a mesh of discrete beams. This bridge is long and narrow so it can be idealised by two beam members as illustrated in Fig. 3. the vertical components can be found similarly.57(b). wCD =wAB=0. As there is a support there already. Example 3. The beam is analysed for this loading using moment distribution. the 1 P 1 slope is 0.56(b). moment distribution is applicable to the analysis of grillages but is tedious to apply for most examples.08. it is not necessary to provide a vertical translational fixity.1P and the intensity of loading is. described in detail in Chapters 5 and 6.2P/l.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.14 and illustrated in Fig. 3.56(a) and the associated BMD (Appendix A) is given in Fig.5 times the flexural rigidity.08P. Fig. In CD. Example 3.55(a) it can be seen that the vertical component of the prestressing force at A is P sin θ≈ tan θ=0.12P. Step 3: To remove the moment discontinuity of 0.52(c) is factored by this amount and added to Fig. Step 2: The BMD associated with simultaneous rotations at B and C is identical to that derived for Example 3. coincidentally. From Fig. In this method. The obvious difference is that typical prestress loading is in the opposite direction to loading due to self weight. Similarly. the intensity is.

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. However. 3. (d) elevation of applied loading and resulting BMD in AB while fixed. At B. two rotational fixities are required in orthogonal directions as illustrated in Fig. (c) plan view of system of fixities. A unit rotation is first applied . it needs to be resolved into components parallel and perpendicular to AB. Vertical loading on AB in the fixed structure is applied to a beam which is fixed at one end. The internal bending moment at the left end of BC is wl 2/8 as illustrated in Fig. In addition.58(b) and it can be seen that there is a discontinuity of moment at B of (1− 2)wl2/8=0.57(d). there is a transition between bending moment and torsion in the members.Page 112 Fig. In order to compare this moment to that just left of B. (b) plan view of idealisation. 3.293wl2 /8.58(a). 3. (e) plan view of fixed BMD However.57 Analysis of crooked bridge: (a) plan view of geometry. 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. 3.57(e). The double headed arrows indicate rotational fixities where the positive direction is clockwise when looking in the direction of the arrow. From Appendix A. the BMD for the two beams are illustrated in Fig. there is a 1/√ discontinuity in torsion of wl2/(8√ 2). the bending moment diagram is as illustrated (in elevation) in Fig. 3. In plan. 3. This is done in Fig.57(c) (rotation about two axes).

Hence the BMD due to unit rotation at B is as illustrated in Fig.60(c). the normalised version of Fig.58 Plan views showing internal moment: (a) end moment in BC.60(d). 3.59(c) can be seen when the moments are resolved in Fig. it is necessary to resolve the rotation into components parallel and perpendicular to that member as illustrated in Fig. The discontinuity parallel to Direction 2 at B is GJ/l+3EI/2l=3EI/l. To determine the effect on member BC. 3. The discontinuity of moment at the joint can be seen by resolving the internal moments and torsions in Fig. 3.57(c)) generates no bending but a torsion of GJ/l in AB. (b) resolution of end moment in BC parallel and perpendicular to AB in Direction 1 (Fig. As there is no resistance to twisting at C.59(d).57(c)). The discontinuity of moment at B in the BMD of Fig. The . 3.59(d). 3. 3. 3. 3. 3. 3.60(a) and it can be seen that it generates no torsion and the BMD illustrated in Fig. the application of a twist of 1/√ at B does 2 not generate any torsion in BC.59(c).59(e). Unit rotation at the end of AB results in the deformed shape and BMD illustrated in elevation in Fig. the discontinuities in the fixed bending moment and torsion diagrams are removed by scaling and adding the diagrams derived in Step 2. 3. 3. Step 3: In the third step. is illustrated in Fig. 3. normalised for moment. Applying a unit rotation in Direction 2 (Fig. dividing by this value gives the normalised version illustrated in Fig. the rotation at the joint must be resolved into components as illustrated in Fig. In BC.59(a).60(b).Page 113 Fig. It is (3+3/2)EI/l=9EI/2l.59(b). 3.

Page 114 Fig. 3. 3. 3. The corresponding bending moment and torsion diagrams are illustrated in Figs.59(e) scaled by minus this value gives the moments and torsions illustrated in Fig. 3. (e) normalised free body diagram discontinuity in the fixed BMD of Fig. Adding the BMD 1/√ of Fig. While the discontinuity parallel to Direction 1 (Fig. This is removed by adding the diagram of Fig. . there is still a discontinuity parallel to Direction 2 of 0.293wl2 /8.57(c)) has now been removed at B.805wl2/8.60(d) to give the diagram illustrated in Fig.59 Effect of rotation in direction 1 at B: (a) elevation of AB showing imposed unit rotation and associated BMD.58(b) is (1− 2)wl2/8=0.61(b). 3. 3. 3. 3.61(a). (c) BMD due to unit rotation.61(c) and the discontinuity in that is removed by adding a diagram parallel to Fig. 3. 3. 3.61(d). (b) resolution of rotation parallel and perpendicular to BC.59(e) to give Fig. (d) free body diagram showing lack of equilibrium of moments at B.60(d).61(e) and (f). The new discontinuity now introduced parallel to Direction 1 is removed by adding a diagram proportional to Fig. scaled by minus this value to give the diagram illustrated in Fig. 3.

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

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

3. it is necessary to fix against vertical translation and against rotation about both axes.e. (b) simplified system of fixities. This deck has flexural rigidities of (EI)ABC = (EI) DEF=0.63(a). i. 3. (EI)BE=3. (GJ)BE =4.8.0. The circles in this figure indicate translational fixities with a direction of positive upwards. total fixity must be imposed at B and E.Page 117 illustrated in Fig. The symmetry of the system is exploited recognising that the three fixities at B are identical to the corresponding fixities at E.63 First step in analysis by moment distribution: (a) complete system of fixities with two rotations and one translation.62(c). This is represented diagrammatically in Fig. 3. Fig.6 and torsional rigidities of (GJ)ABC =(GJ)DEF=2. It is subjected to vertical uniform loading of intensity w. (d) free body diagram showing lack of moment equilibrium in fixed structure . (c) fixed BMD. Step 1: To isolate the members from one another.0.

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

Step 3: As stated in Step 1 and illustrated in Fig. the discontinuity in moment parallel to Direction 1 in the fixed structure is (wl2/12) cos θThis is removed by adding Fig. the corresponding BMD (Appendix B. 3. The resulting diagram is illustrated in Fig. 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 . /l The internal moments and torsions at B and E are illustrated in Fig.65(b). results in the internal moments and torsions illustrated in Fig. (b) normalised free body diagram . 3. 3.65 Internal moments associated with rotations at B and E: (a) free body diagram with resolution of moments parallel and perpendicular to direction of fixity. 3.66(a). .64(b).63(d). Fig. 3. 3.Page 119 required deflected shape is illustrated in Fig.65(b) scaled by that amount. 3. 3.65(a).64(c). last BMD) is illustrated in Fig. Resolving parallel to the direction of fixity gives the discontinuity of moment corresponding to unit rotation: Normalising with respect to this discontinuity. Thus.

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. It follows that lateral pressures may be related approximately to the average displacement of the abutment over the retained height (δ ).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. 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. or to a wall translation of Hret /20. these are in keeping with the h general guidelines set out in BA42/96 (1996). av Expressions for σ emerging from this rationale are given below.3) and (4. This method is based on experimental observations which indicate that movements to develop full passive pressures typically correspond to an abutment rotation equal to one-tenth of the retained height of soil. Such an approach is described later in this chapter.Page 125 Fig. 4. (4.2) where (4. Hret/10.5) 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 .

2 Hret .3) gives: and . For the latter case.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. it is reasonable to assume that horizontal stress acting on both sides of the abutment are given by equation (4. Therefore. As Hret=6 m and δ <Hret/20.1. therefore.7): Figure 4. is Implicit in equations (4. av The unit weight of the soil (γ ) is soil Therefore for z<6 m (Hret ): 4.4 indicates that Kp=17.6) and (4.7) where α the coefficient of thermal expansion of the deck.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 for equation (4. Assume α the deck is 12×10− per °C and for From equation (4. av as: (4. it is necessary to quantify .Page 126 restraint offered by typical abutments and backfill is relatively small).5) when the depth exceeds approximately 1. Example 4. For most cases. The design extreme event for the determination of maximum abutment pressures is a 40° 6 increase in temperature.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. for a bridge deck of length L which experiences an increase in temperature of Δ δ may be calculated T.6) (4.

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

4.2 Contraction of bridge deck There is generally a lesser height of soil in front of bridge abutments than behind them. derived using equation (4. mean confining stresses (p'). The secant Young’s modulus.2. sections. the principal uncertainty relates to the resistance to movement at the bases of the piers and abutments. the resistance provided by such soil to the contraction of a bridge deck is usually small.5 can be used to estimate the secant Young’s modulus for cohesionless soil. 4. specification of the dry density effectively dictates the void ratio. Fig. These conditions are applicable if the abutment foundations are cast in very dense soil or rock. creep.Page 128 density of water. 4. and shear strains (γ Equation (4. However.1 Contraction of bridge fully fixed at the supports The case is first considered of an integral bridge in which no translational movement can occur at the base of the abutments. 4. Thus.5 . e. 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.5 for a range of in-situ dry densities (ρ). Guidance on appropriate values for ρ p' and γ specific cases is given in subsequent for d. This means that. As a result. and/or shrinkage. 4. d or Fig.6 Contraction of frame rigidly fixed at supports: (a) geometry.8) ).8). is plotted in Fig. (b) bending moment diagram from example 3. in an analysis to determine the effects of elastic shortening.

Expressions for the stiffness of such springs have been deduced here from relationships provided by Dobry and Gazetas (1986) for an elastic soil.0 m below the ground level on the inside of the abutment as illustrated in Fig. Design spring stiffnesses on the inside of the abutment for a strip foundation of width B. Strip foundations or pile caps are commonly founded at around 0. Es.5–1.7. The axial contraction induced bending in the abutments and.2 Contraction of bridge on flexible supports Most bridges are constructed on supports which have some degree of flexibility. ) . bending in the deck also. that resists bridge contraction. horizontal and rotational displacement respectively. It is this small depth of soil.20(c) shows that only 7% of the potential force is applied to the abutments). The soil around the strip foundation can be idealised by a number of linear elastic springs. Conservative.6(b).001.0 m below the ground level are given in equation (4.10) where kvert. In that case. If the ends of the deck were fully prevented from contracting.5). However. 4. some movement of the deck was possible through bending in the abutments. 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). 4. together with sliding resistance at the base of the pad. Thus. due to the integral nature of the bridge. upper bound estimates of the secant Young’s modulus of elasticity.5 m and 1. The complete bending moment diagram is illustrated in Fig. Abutments and piers are generally either supported on foundations bearing directly on the ground below or on pile caps underlain by piles. the bridge was fully restrained at the base of each abutment and pier.8) assuming a p' value equivalent to the foundation bearing pressure and a shear strain (γof 0. 4. the resistance of the abutments to movement was considerably less than the axial stiffness of the deck (Fig. embedded to a depth of between 0. may be calculated using equation (4. k hori and krot are the stiffnesses per metre length of strip foundation for vertical. 4. The end result for that example was a relatively small axial tension in the deck. only 5% of the potential level. the decrease in temperature would generate a large tensile force in the deck and there would be no contraction. and a relatively large contraction.2. 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.Page 129 The bridge illustrated in Fig. However. 3.10): (4.6(a) was considered in Chapter 3 for an axial contraction due to temperature of 20° in the deck (ABC) (Example 3.

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

the modulus of elasticity of concrete and the cross-sectional area (per metre run): .8) then gives: Equation (4.8 Bridge of Example 4.10) 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. (b) detail at abutment Substituting in equation (4. 4.Page 131 Fig.2: (a) elevation.

axial force and bending moment diagrams illustrated in Fig.Page 132 The equivalent loads and the associated axial force diagram are illustrated in Fig. (b) axial force diagram from computer analysis. (d) corrected axial force diagram . 4. (c) bending moment diagram.9. illustrated in Fig. 4. 4.10 Analysis results: (a) deflected shape. Subtracting the associated axial force diagram gives the actual distribution of axial force generated by the shrinkage.2: (a) equivalent loading and springs.9 Computer model for bridge of Example 4. 4.10(d). The frame was analysed using a standard analysis package which gave the deflected shape. 4. (b) associated axial force diagram Fig. Fig.10(a)–(c). No adjustment is necessary for the deflected shape or bending moment diagram.

The average shear strain in the backfill must be less than 2δ and could conservatively be assumed as about 2δ /H /3H. 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. an average shear strain must be assumed. 4. 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). with time. The conventional spring model represents the backfill soil and soil beneath the abutment by a series of spring supports. Out of a total potential 6 shortening of 6 mm (200×10− ×30000 mm) at each end. 2. stiffness value. the bending moment at the ends are more significant at 568 kNm. However. Some notable features have been observed from experimental studies by Springman et al. (1996) support the validity of this assumption. To adopt a single soil /H. The selection of a suitable soil stiffness value (Es) is essential for appropriate modelling of the backfill. Cyclic variations in temperature (and associated expansions and contractions of the deck) cause the backfill to compact and. Assuming uncracked conditions. Such a model is imperfect as it does not allow for shear transfer within the soil as there is no interaction between the . 3. will be affected significantly by the properties of the soil behind the abutments. 5. 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. The stiffness of the soil is influenced by the shear strain in the backfill. Thus.Page 133 Example 4.3 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. to tend to an equilibrium density compatible with the strain amplitude that it is regularly subjected to. for example.3 mm does generate distributions of stress in the frame.3 N/mm 2. the stresses generated by an increase in deck temperature. (1996). However the restraint which prevents the remaining 0.7). Nonlinear elastic finite-element analyses by Springman et al. The axial tension is relatively small at 337 kN corresponding to a stress in the deck of less than 0. this corresponds to a maximum flexural stress of 2. 4. This observation suggests that the use of a constant soil stiffness value with depth (for a given strain) is reasonably realistic.7 mm is predicted to actually occur. and others: 1.4 N/mm 2.2 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.

11 Culvert of Example 4.11) is illustrated in the following example. i.3: Conventional spring model The culvert illustrated in Fig. assuming linear elasticity.: Inverting equation (4.11 is subjected to an increase in temperature of 20 °C. 4.65. has been specified as 1600 kg/m 3. An approximate expression.11) The application of equation (4. It is assumed that the density of the backfill reaches an equilibrium value 20% in excess of that specified.9) and assuming Gs=2.3 . The d piles are assumed to provide insignificant lateral restraint to the deck. ρ. gives a void ratio of: Fig. 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. The dry density of the backfill. Example 4. It does. have the advantage of simplicity and is considered here because it remains a popular approach among bridge engineers. L: (4. however.Page 134 springs.e. 4. has been developed for the horizontal spring stiffness per square metre. of the backfill behind an abutment of depth H and transverse length.

the expansion of the culvert is estimated as its unrestrained value.e. the product of the temperature increase. a horizontal stress of p'=50 kN/m2 is assumed. the average shear strain in the affected backfill is then: On the basis of Note 2.Page 135 To estimate the average shear strain induced in the backfill.8) gives: The horizontal spring stiffness is then given by equation (4. 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. i.12(b).11): The model for a 1 m strip of the frame is then as illustrated in Fig. Then equation (4. 4.12(a). 4. The bending .

.3: (a) springs and equivalent loads. The deflection found from the computer analysis was 1.3 moment diagram was found from a computer analysis and is illustrated in Fig.12 Computer model for culvert of Example 4.13 Bending moment diagram for Example 4. As this is similar in magnitude to the deflection of 1.13. 4.Page 136 Fig. As there was no associated distribution of bending moment. 4.20 mm assumed in the estimation of shear strain. The moment in the abutments can be seen to change sign through its length due to the flexible nature of the horizontal support. 4.19 mm. iteration was not considered necessary. this is the final distribution of moment due to the expansion. (b) associated axial force diagram Fig.

(b) unit rotation . However. It was therefore assumed conservatively that the soil had limitless compressive and tensile strength (e.Page 137 4.3. This method does not. 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.g. given that relatively small movements are required to reduce pressures to their minimum (active) values on the inner face of the abutment. the forces and moments associated with passive movements which occur as a consequence of deck expansion. the analyses assumed that any soil present on this side did not contribute to the resistance. and (ii) a rotation θ with zero horizontal displacement (Fig.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.4.14).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. 4. described in Section 4. provide details concerning the distribution of moment in the abutment or the pressure distribution in the soil. base sliding or slip on the abutment stem were not permitted). This technique consists of modelling both the abutment and the surrounding soil with an equivalent lateral and rotational spring at deck level. no passive failure or abutment lifting were allowed) and that no slip between the abutment and the soil occurred (e. Fig. i. 4.14 Stiffness components at top of abutment: (a) unit translation. The purpose of the analyses was to provide credible upper bound estimates of soil resistance.g.e. 4. however. 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.

1.0×104− 2.12) were the most important factors controlling the magnitudes of the lateral force (Fh) and moment (M) at the top of the abutment (Fig. Best-fit expressions were obtained for Fh and M for the range of parameter values given in Table 4.13) where f 1 and f 2 are functions of the ratio.14) B (m) (Fig.5×106 >0. r.14) . 4. The values of Fh and M were also seen to increase systematically as the base width (B) increased and its height (H) reduced. 4. All values in this stiffness matrix can be reduced by 15% if friction between the abutment and soil is considered negligible. They are given here in matrix form: (4. 4.14) Fig. defined as: (4.15 End part of frame bridge showing locations and directions of fixity Table 4. 4.05 3 m −.05 1.Page 138 It was found that the flexural rigidity of the abutment (EIa ) and the ratio.14) for r>0.14).1 Range of parameters used in derivation of equation (4.5 H (m) (Fig. H/B which are given by equation (4.5–3.5− 12 0. (4.13) Parameter Es(kN/m2 ) EI a(kNm /m) 2 3 r=Es/EIa(m− ) Allowable range 10000− 500000 1.

13) was found to predict values of F h and M to within 10% of the values given by the finite element analyses. When a frame bridge with an abutment height of H is fixed rigidly at the supports and the system of fixities illustrated in Fig. 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) and (4. in the absence of soil. Ld and Id are the cross-sectional area. Equating the K22 (second row.15) becomes: (4.15) and (4.Page 139 For the range of parameters listed in Table 4. [K]. equating the K12 (and K21) terms gives: (4. span length and second moment of area of the deck respectively. equation (4. 4.13) with the result that equation (4. When the bridge is embedded in soil and this is taken into account.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.16).17) where Heq and Ieq are the equivalent abutment height and second moment of area respectively. the stiffness matrix.16). Alternatively. Similarly. 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. second column) terms in equations (4. gives: (4.15 is used.15.15) where A d.1.16) A comparison of equations (4.18) . 4. is: (4. the terms involving Ia and H are replaced with terms from equation (4.

20) gives an equivalent height of: (4.40 respectively.23) Finally.19) The equivalent abutment height is then: (4.18) can be simultaneously satisfied by selecting an equivalent abutment second moment of area equal to: (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. As a result.21) gives a spring stiffness of: (4.2 Expansion of frames with deep abutments The equivalent single-spring model can be simplified for the case of deep abutments.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: (4.22) Substituting for f 2 in equation (4.24) These equations can be used to estimate the properties of an equivalent frame for an integral bridge with deep abutments.17) and (4. . substituting for f 1 and f2 in equation (4.33 and 0.21) 4. the parameters f1 and f2 approach their minimum values of 0. For values of (H/B) in excess of 10.Page 140 Equations (4.

Page 141 Example 4.3 to be: and the second moment of area of a 1 m strip of the abutment is: The ratio defined by equation (4. The elastic modulus of the soil is found as for Example 4.23): The stiffness of the single spring on each side is given by equation (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. 4.12) is then: The equivalent height of abutment is then.4 . The magnitude of the equivalent loads.16 Computer model for bridge of Example 4. is: Fig.11 due to a temperature increase of 20°. from equation (4.16. as for Example 4.24): The equivalent frame and loading are illustrated in Fig.3. 4. 4.4: Equivalent single-spring model for frame with deep abutments The equivalent single-spring model is used to determine the maximum moment in the culvert illustrated in Fig. The dry density of the backfill has been specified as 1600 kg/m 3.

13) and (4.3. 4.14) imply that an abutment provides a greater resistance to deck expansion if it has a lesser depth of embedment (H).4. the deflections per unit load can be seen to be significantly greater. with a finite strength defined by its friction angle. the magnitude of moment in the deck. It is therefore recommended that this equation be used in preliminary analysis and that a finite- . In the example of Fig. linear. the function is.13). 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. of course. . 4. The influence of a limited soil strength on the resistance offered by a bank seat is illustrated in Fig. The reality.13) and (4.12(b).4 The associated axial force diagram is as illustrated in Fig.18.18(b).3 Expansion of bank seat abutments Equations (4. In Figure 4. Effective rotational stiffnesses at this lateral movement are about 75% of the purely elastic case. 4. 4. It is important to remember that the distribution of moment in the abutment is not realistic. On the other hand. 4. is likely to be more reliable than the value found in Example 4. 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. 87 kNm. 4.18. of course. When the soil is linear elastic and infinitely strong.17. predictions from finite-element analyses are presented of a horizontal force/deflection relationship. The model was analysed using a standard computer program and the resulting bending moment diagram is illustrated in Fig. 4.13.14) would suggest. Similar results can be shown for moment/rotation functions and for force/rotation and moment/deflection functions. is that shallow abutments are more likely to slide than deep ones and will therefore offer less restraint to deck expansion than equations (4.17 Bending moment diagram from computer analysis of bridge of Example 4. 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. when the soil is treated as an elastic perfectly plastic material. However.Page 142 Fig.

soil friction angle.3 but using the smaller abutment height.12) is then. The 3 dry density of the backfill has been specified as 1600 kg/m . 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. and the second moment of area of a 1 m strip of the abutment is Ia =0.31.Page 143 Fig. foundation bearing pressure= 200 kN/m2.19 due to a temperature increase of 20°C. The ratio defined by equation (4. (b) horizontal force/displacement relationship ): (a) element soil/structure analysis incorporating a realistic constitutive model for the soil is performed if the effects of deck expansion have a significant influence on the final bridge design. 4. r=0.18 Finite-element analysis results for bank seat abutment (E s= 100000 kN/m 2. Example 4. 4.018 m4. . section through bank seat.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. Ec=30×10 6 kN/m2 .

5 For this example.83. H/B.5/3=0. the spring stiffness is. f1 and f2 are calculated from equation (4.20): The equivalent abutment second moment of area is given by equation (4.Page 144 Fig. 4. from equation (4. is 2.19 Bridge of Example 4.19): Finally.21): .14): The equivalent height is then calculated directly from equation (4. The parameters. the ratio of embedment depth to foundation breadth.

4. 4. It is clear from Fig. the bridge still expands and contracts relative to the surrounding soil and the incorporation of a run-on slab does not prevent this. The magnitude of the equivalent loads. it transfers the relative horizontal movement from the end of the deck to the end of .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.21 Bending moment diagram from computer analysis of bridge of Example 4. this must be accommodated if premature deterioration of the pavement is to be avoided. 4. 4.20 Computer model for bridge of Example 4. The maximum magnitude of moment in the deck due to the expansion is 114 kNm. 4.5 Fig. Clearly. 4.4.5 The equivalent frame and loading are illustrated in Fig.20.3 and 4. On a road bridge. 4. This is achieved in many cases by the installation of a run-on slab as illustrated in Fig.22. Preventing relative horizontal translation is not so simple.Page 145 Fig. The effect of such a slab is to allow relative rotation between the deck and the run-on slab while preventing relative translation. is: This model was analysed and the bending moment diagram is illustrated in Fig. Preventing relative vertical translation significantly improves the rideability for vehicles travelling over the bridge.18 that this result is quite conservative. In effect.21. as for Examples 4.

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

2. When the in-situ concrete is cast. When the in-situ concrete Fig.1H can be assumed from Table 4.5)=5.Page 147 Table 4. the length of run-on slab required for the bridge of Example 4. 4. as illustrated in Fig. 4.6H 0. (1996) suggest that the length of the trough (Lt ) is unlikely to exceed the limits given in Table 4. a maximum trough length of 2. As an example.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.2.2 Approximate upper limits on expected trough lengths Granular fill Well compacted Deep abutments Bank seats 0. 4.23.1(2.9H Loosely compacted 1. As the backfill is loosely compacted (density= 1600 kg/m3) and the abutments are not deep.1H These observations and those taken during centrifuge model tests by Springman et al. (b) section A—A .25 m.23 Composite integral bridge made from precast and in-situ concrete: (a) elevation.19) is calculated. 4. 4.24(a). the precast beams are simply supported and the self weight of the bridge induces a sagging moment. the run-on slab should have a length of at least 2. Hence.5 (Fig.4H 2.

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

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

these hogging strains increase with time.27(b). 4. When the bridge is made integral. further curvature is resisted and the resulting distribution of moment is as illustrated in Fig. the equivalent moments at the ends are less but a further increment of equivalent moment is applied at the points where debonding ceases (Fig. 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. the beams hog upwards as illustrated in Fig. 4. In a study of continuous bridges made integral at the interior supports.28 Detail at support showing points where long-term cracking is likely to occur Fig. particularly at the interface between the precast and insitu concretes.27(a). such curvature is unrestrained so it results in instantaneous strain and a moment which is the simple product of prestress force and eccentricity. If some strands are debonded. 4. . For smaller beams.28. 4.27(c). Due to creep. 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. 4.26(a). It can result in cracking at the bottom of the deck over the supports as illustrated in Fig. The long-term result is a distribution of prestress moment such as that illustrated in Fig. Clark and Sugie (1997) carried out a parametric study of the time-dependent effects in composite integral bridges. When prestress is first applied below the centroid.Page 150 Fig. they suggest designing for a moment of 600 kNm. 4. As they are simply supported. 4. This phenomenon is particularly significant if the bridge is made integral when the precast concrete is young as this causes most of the creep strain to occur when it is in the integral form. they calculated the maximum longterm sagging moment for beams made integral when between 21 and 100 days old.26(b)).

in a similar way to beams. 5. A subset of anisotropic materials are orthotropic materials in which the behaviour varies in mutually perpendicular directions (X and Y) only. Such slabs can be assumed to behave like thin plates which can be thought of as the twodimensional equivalent of beams. 5.1 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. slabs can readily be idealised using one of a number of well-proven methods and analysed using structural analysis programs. Fortunately. but is generally impractical. A further subset of orthotropic materials are isotropic . Orthotropy represents the most general material behaviour usually considered for bridge decks.1 Orthotropic and isotropic plates A material in which the behaviour in each direction is independent of the others is referred to as anisotropic. Thin plates get their strength from bending. Thick plates correspond to deep beams and are not considered here.2 Thin-plate theory Slabs used in the construction of bridge decks are generally thin relative to their span lengths.Page 151 Chapter 5 Slab bridge decks—behaviour and modelling 5. it is necessary to first consider the theory of bending of plates.2. except that bending takes place in two mutually perpendicular directions in the plane of the plate. To understand the basis of such programs and their limitations.

such as timber. applied. b.b'.e. the theory of materially orthotropic thin plates is developed. Figure 5.2. Although this type of material is rarely found in bridge construction. Z c and d shown in Fig.Page 152 materials in which the behaviour in all directions is the same. In the following sections. Hence the length of a'b' projected onto the X axis is: . 5. it is common practise to extend it to include geometric orthotropy. but the same geometric properties. Thus. c' and d' as illustrated in Fig. isotropic plate theory can be used with reasonable accuracy for the analysis of many bridges.2 Bending of materially orthotropic thin plates Figure 5. Such a plate might be constructed of a material where the microstructure is orientated in two mutually perpendicular directions. Considering initially the X− plane.: x. the points a. the cube both moves and distorts. at which point. 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. The displacement of point a in the X direction is denoted u. Considering point b. 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.3. These types of slab are referred to as geometrically (or technically) orthotropic.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.2 move to a'.1 shows a portion of a thin plate in the X− plane. δ In this figure the thickness of the plate is taken to be d. The origin of the axis system is at Y mid-depth in the plate. 5. 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. will be u plus the change in u over the distance δ i. 5. This type of plate is not typical of that found in bridge decks but is frequently used as an approximation of actual conditions. When a load is z. such as reinforced concrete slabs with significantly different amounts of reinforcement in the two directions or voided slabs. a distance δ from a. the x displacement at that point in the X direction. Many bridge slabs possess different second moments of area in two directions. z=0.

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

if v and w are the displacements in the Y and Z directions respectively. the strain in the X direction is: (5.1) Similarly.3) . it can be shown that: (5.Page 154 Fig.2) and: (5.Z plane By definition. 5.3 Distortion of cube of material in X.

7) This implies that w is independent of z. this reduces to: u/∂ The other component of strain can be found similarly to be: Hence the shear strain is: (5.: (5.Page 155 The shear strain in the X− plane is defined as the change in the angle.6) In thin-plate theory. the difference between c'a'b' and cab. cab from the original Z 90°.5) (5. i. As can be seen in Fig. i. a number of assumptions are made to simplify the mathematics involved.3. The first of these assumptions is that there is no strain in the Z direction.e. α and βReferring to the figure: . or that w is a function of x and y only. Figure 5. As ∂ x is small. The physical meaning is . .4 illustrates the implications of this assumption.e. 5. there are two components.4) Similarly the shear strains in the X− and Y− planes are respectively: Y Z (5.

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

11) . this implies: (5. Hence. 5. there is no displacement in either the X or Y directions at z=0. u and v are both zero.10) where C is a constant of integration. As the origin is located at the centre of the plate and bending is assumed to occur about that point. Substituting this into equation (5. at z=0.8) gives: As w is independent of z.5 Segment of plate in X− plane showing assumed lack of shear distortion Z Rearranging equation (5.Page 157 Fig.10) implies that the constant C is zero giving: (5.

17) (5.15) In the flexural theory of beams.12) into equations (5.16) (5.18) into equations (5. a similar expression can be derived for v: (5.16)–(5.12) Substituting equations (5.19) (5.11) and (5.1) and (5. In thin-plate theory.21) .13)–(5.13) (5.Page 158 By rearranging equation (5.15) respectively then gives: (5.3) respectively gives: (5. Y and XY directions which are given by: (5.20) (5. the equations is are similar.9). the curvature is defined as: where κ the curvature and R is the radius of curvature.18) Substituting equations (5.5) gives: (5. but there are now curvatures in the X.14) Similarly equation (5.

E.and three-dimensions: (a) one-dimensional bar. strains in the other two directions become significant. This is generally a reasonable assumption for slab bridge decks. The only significant strain in this system is in a direction parallel to the axis of the bar. From this.20) shows that the same applies to the strain in the x x Y direction. This strain.Page 159 Examination of equation (5. 5. Such cases are discussed further in Chapter 7. it follows that plane sections remain plane.3 Stress in materially orthotropic thin plates In the previous section. as is indicated in Fig. (b) three-dimensional body showing the effect of stress in the axial direction on strains in the orthogonal directions . as is generally assumed in beam theory.6(b). elasticity. . as 2 κ=∂w/∂2 is independent of z.22) Fig.6 (a) shows a onedimensional bar subjected to a tensile force. 5. By defining the X axis as the direction of the applied force. the strain in that direction is given by: (5.6 Distortion in one. Figure 5. εis related to the stress. but some cases do exist where this is not so. Expressions are now developed for the corresponding stresses. by: In the three-dimensional case. σand modulus of . expressions were established for the various strains in a thin plate.19) shows that strain in the X direction is a linear function of z. Equation (5. 5.2.

ν etc. y z made of a homogenous material and that the elastic constants (Ex. Y and Z directions respectively.23) (5. the stress in the Z direction is small and the Poisson’s ratio is generally small for bridge deck materials. and ν νand νare the corresponding Poisson’s ratios. E y and Ez are the moduli of elasticity in the X. Consequently the last term of equation (5.22) can be ignored.24) . For a thin plate in bending.22) assumes that the plate is x.Page 160 where E x.) are independent of x. 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. as is appropriate for the materially orthotropic (or anisotropic) case. each other. Equation (5.

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)

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

one for each direction (X and Y). .2.e.7. However. the same phenomenon exists and an expression is found from equilibrium of forces on a segment.10(b) shows the type of deformation associated with each of them. Unlike beams. This is a reasonable assumption as shear deformation is generally small in bridge slabs relative to bending deformation.9)) that shear deformations in the plate were negligible. there are two shear forces at each point. can be significant. the shear force is the derivative of the moment. In thin-plate theory.10(a) shows the direction in which each of these moments acts while Fig. shear stresses. 5.8) and (5. Defining qx and qy as the downward shear forces per unit breadth on the positive X and Y faces respectively then gives: (5. and illustrated in Fig. 5. gives an expression for the shear force Q: (5. Taking moments about the left hand end gives: Rearranging and ignoring the term.46) i.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: (5.44) and: (5. 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.43) The moment mxy (=myx ) is often referred to as a twisting moment and is distinct from the normal moments mx and my . Figure 5. In the simple flexural theory of beams. Figure 5. 5. a similar expression can be derived. dQdx which is relatively small. particularly in concrete slabs which are quite weak in shear. while numerically small.11 shows a segment of a beam of length dx in bending.45) It was assumed earlier (equations (5.

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)

but there is no account taken in the grillage analysis of the second term.Page 173 Equation (5. This accounts for the first term of equation (5. Examining. This may be quite unsatisfactory. the shear force Vb1 in Fig.14) these are given by: (5. but the torques in grillage members b1 and b2 will not necessarily be equal to the torques in b3 and b4.2 Grillage member properties—isotropic slabs A grillage member in bending behaves according to the well-known flexure formula: (5. The shear forces per unit breadth in the slab. I the second moment of area.40) stated that mxy and myx are equal for materially orthotropic plates. unless myx is particularly large. qx and qy.46). the moment per and unit breadth. namely. 5. are found from the shear forces in the grillage members in a similar manner to the moments.56) where i is the second moment of area per unit breadth. This technique is discussed further in the next section.14. 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. (Figs. However. By substituting the curvature 1/R with κ rearranging. the derivative of myx with respect to y.47).54) Equations 5. At point p. for example. This could be calculated in the grillage by finding the derivative of the torques in b3 and b4 with respect to y. 5.55) where M is the moment. 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. 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. particularly for orthotropic plates with significantly different flexural stiffnesses in the two directions. this is not normally done as the resulting inaccuracy in the shear forces tends to be small.3.47 and 5. m is found: (5. . E the modulus of elasticity and R the radius of curvature.53) and: (5. as large variations of torque may exist between the longitudinal and transverse beams.48 gave expressions for the shear forces per unit breadth in the slab.13 and 5.

A grillage member in torsion behaves according to the well known 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.57) must be equated.59) . The moment/curvature relationship then becomes: (5.57) To achieve similitude of moments between a slab and the corresponding grillage. if this approximation is applied to both mx and my. G is the shear modulus and J is the torsion constant (St. there is only one value for E and νSubstituting E for Ex and ν . v. for νand νin that equation gives: x y As Poisson’s ratio. Venant constant). Further.58) where is the angle of twist.15 shows a portion of a beam of length δ in torsion. This can be justified by the fact that Poisson’s ratio is small. giving: A further simplification is made by equating the term below the line to unity. such an adjustment has very little effect on the final results. Figure 5. they are both affected by the same amount. the stiffness terms of equations (5. it is common practice to ignore the second term in this equation.Page 174 Equation (5.2 for concrete). As it is the relative values of stiffness that affect the calculated bending moments and shear forces. For an isotropic slab.56) and (5. is relatively small in bridge slabs (approximately 0. l is the length of the beam. T is the torque. 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: (5.34) gives an expression for the moment per unit breadth in the X direction in the slab.

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

43) gives an expression for the torsion constant of the slab. 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. this is carried out automatically by the grillage program. Equation (5. Equation (5. However. in the grillage members. 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. Equation (5.63) and (5. in the slab and torques.63) where j gril is the torsion constant per unit breadth in the grillage member.43) for the torsional constant is valid. t.42) gives an expression for the twisting moment per unit breadth in the bridge slab: (5.65) where d is the slab depth. This can clearly be achieved by adopting the same shear modulus and torsion constant in the grillage member as is in the slab.64) must be equated.67) Typically.Page 176 This can be rewritten in terms of torque per unit breadth.66) Although equations (5. The preceding derivation of grillage member torsional properties is applicable to thin plates of rectangular cross-section where equation (5.65) and (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.64) To achieve similitude of moments. . it will not generally be necessary to specify Gxy for the grillage model.58) can be applied.66) are based on the grillage member having the same shear modulus as the slab. t: (5. mxy. The torsion constant for the grillage member can alternatively be expressed in terms of the slab second moment of area: (5. Equating this to jgril gives: (5.65) ensures that the grillage members in both directions will have the same torsional constant per unit breadth.

only). reproduced here.68) It can be seen that equation (5. 5.Page 177 For rectangular beams with depth d and a breadth of greater than 10d. in the Y direction. The torque in the beam results from both of these shear stresses and is given by: (5. most bridges have the same modulus of elasticity. The vertical shear stresses are accounted for in the grillage in the same manner by the shear forces qy in the transverse beams.34). It has been recommended that the edge grillage members be placed at 0. the shear stresses in the vertical direction are accounted for by the shear force per unit breadth. The reason for this lies in the definition of torsion in a beam and of moment mxy in a slab. many bridges are geometrically orthotropic. Similarly. 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.3. they have different second moments of area per unit breadth in the orthogonal directions. 5.36) shows that the moment mxy is arrived at by summing only the shear stresses in the horizontal direction (i. In the slab.e. the torsional constant may be approximated with: (5.3 Grillage member properties—geometrically orthotropic slabs Equation (5.e. for both directions.17.3 times the slab depth from the edge so as to coincide with the resultant of the shear stresses. in both the horizontal and vertical directions. qx as illustrated in Fig. It is common practise to use the equations developed for materially orthotropic thin plates to represent geometrically orthotropic bridges.69) In the slab.16 shows a portion of a beam of breadth b and depth d in torsion.68) predicts a torsion constant for the beam which is twice that predicted by equation (5. applies to materially orthotropic slabs: However. 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). i.66) for isotropic slabs. Figure 5. The shear stresses set up in the beam are shown. E. the second moments of area per unit breadth for the grillage and the slab are equated. . Further. equation (5.

Page 178 Fig. in a fine grillage mesh.e.41). if the same shear modulus and torsional constant are used in the two directions. If it is assumed that the same conditions hold for geometrically orthotropic slabs. 5.42). the two twisting curvatures are the same.16 Beam subjected to torsion showing resulting shear stresses Equation (5. i. curvatures in the orthogonal directions at a point will be approximately equal.40) stated that the two twisting moments at a point in a materially orthotropic slab are equal to each other. . as stated in equation (5. Then.: and: There is no facility in a grillage model to ensure that the two curvatures at a point are equal. Further. However. it follows from equation (5.

3. is a function of the elastic modulus.Page 179 Fig.70). equation (5.67).70) Hambly (1991) recommends using such a single torsional constant for both orthogonal directions: (5. 5. that the twisting moments are equal: (5. νIt is generally calculated internally in computer programs using .71) It can be seen that this equation is consistent with equation (5. E. 5. and Poisson’s ratio.66) for an isotropic slab. G.17 Slab with vertical shear stresses and corresponding grillage members with shear forces per unit breadth reproduced and adapted here as equation (5. The shear modulus for a slab made from one material.4 Computer implementation of grillages There are many computer programs commercially available which are capable of .

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

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

5. 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. some general recommendations are valid for most grillage models. mxy or myx. which is applicable to such a wide variety of structural forms. if reasonably accurate results are to be obtained.7 Recommendations for grillage modelling It is difficult to make specific recommendations on the use of a technique such as grillage modelling. qx and qy .48) provide expressions for the shear forces per unit breadth. Fortunately. When bridges are supported at discrete intervals. There is no account taken of the derivative of the twisting moments.Page 182 Equations (5. then grillage member spacing would assume a much reduced importance. are given in Chapter 6. and should be used in the context of good engineering judgement. These should not be viewed as absolute. 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. there are sharp concentrations of shear intensity near each support.3. Some more specific recommendations. However. the magnitude of these moments is generally relatively small. If it were assumed that shear enhancement was sufficient to cater for local concentrations of shear near a support. the effective breadth decreases and the calculated concentration of shear adjacent to the support increases.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. if the grillage mesh density increases. The first of these equates the shear force per unit breadth qx to the sum of two derivatives: In the grillage.47) and (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. such as those relating to voided or skewed bridge decks. 5. whichever direction the beam lies in. except for bridges with high skew. It follows that. 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.3. Thus the designer would design for the shear force calculated at a deck depth from the support. It will be seen from the recommendations given here that the traditional need for economy in the . This direct relationship between mesh density and the calculated maximum shear intensity means that. the grillage member spacing has to be fixed near the support so that it gives the correct result.

Where possible. 5. Transverse beams should have a spacing which is similar to that of the longitudinal beams. Spacing will often be dictated by the location of supports or lines of strength in the bridge slab. 4. Lines of strength may consist of concentrations of reinforcement. as the computational power available to today’s engineers is well in excess of that available when earlier recommendations were made. However. should these exist. as the magnitude of moment in the transverse beams is generally relatively small. significantly greater spacings are often possible without great loss of accuracy. illustrated in Fig. A choice of between one and three times the longitudinal spacing would be reasonable. as this may result in skewed members which complicate the interpretation of results. 2. Nonetheless. no additional accuracy will be achieved. grillage members should be located such that nodes coincide with the locations of supports to the bridge slab. or precast beams in beam-and-slab bridges. location of prestressing tendons. Often this spacing will be greater than that of the longitudinal beams.19 Alternative grillage meshes near point supports: (a) local adjustment to mesh near supports to maintain constant spacing of members elsewhere. (b) non-constant mesh spacing . The procedure of moving nodes locally to coincide with supports. beyond a certain point.Page 183 numbers of grillage members no longer applies. should be avoided if possible. 1. and. there is no advantage in providing excessive numbers of grillage members as the amount of output data will be excessive. The transverse grillage members Fig. There is little point in having longitudinal beams too closely spaced. Longitudinal grillage members should be provided along lines of strength in the bridge slab. particularly in wide bridge slabs. A reasonable spacing of longitudinal beams is between one and three times the slab depth. 3.19(a). 5.

. The objective is to locate these beams close to the resultant of the vertical shear stresses. Figure 5. Care should be taken. the use of elastic springs to Fig. one with more longitudinal and transverse beams at a closer spacing. 5. the breadth of slab outside 0. This may involve. for example. should they exist. If the spacing of grillage members is in doubt. in the bridge slab as illustrated in Fig. where d is the slab depth.3d should be ignored. such as heavily reinforced diaphragms above bridge piers. For bending moment results. 7.17. 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 0.20 illustrates an example where a member is correctly placed more than 0.Page 184 should also be chosen to coincide with lines of transverse strength in the bridge slab. increasing the mesh density tends (up to a point) to increase the accuracy. i. 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. The second moments of area of these beams are calculated using the full breadth of slab in the normal way. when determining the torsional constant of these longitudinal grillage members. so that the span length between supports in the grillage and the bridge slab are the same.e.3d from the edge except for the end transverse members . 6. a check can be performed by comparing the output of a grillage with that from a more refined grillage. that this recommendation does not result in supports being placed in the wrong locations. 5. however.3d from the edge of the slab.20 Segment of grillage mesh showing longitudinal members 0.3d from the end. 5. Supports to the grillage should be chosen to closely resemble those of the bridge slab. It has also been recommended that.

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

G.47) and (5.48) give expressions for the shear force per unit breadth in a thin plate. E and νThe shear modulus. This is a significant advantage of the FE method over the grillage approach. are output directly by FE programs. is determined by the . Gxy. although not all programs offer this facility. the second moment of area per unit breadth is given by equation (5.34) and (5. Where the twisting moments are significant.2 Properties of finite elements The types of finite element considered here are those used for the modelling of slab bridge decks. only two elastic constants need to be defined for the finite elements. Equations (5.Page 186 5. Each of these expressions involves terms relating to the curvature in both the X and Y directions. Some x. These expressions involve derivatives of the direct moment mx (or my ) and the twisting moment myx (or mxy). in-plane distortion or a combination of both of these.33): . Finally. mx. although in some programs it may not be. As the element is of constant depth. If this facility is used.1 Similitude between finite-element model and bridge slab The moments per unit breadth. Many programs provide the ability to determine these values at any arbitrary point using interpolation. E x. The twisting moment term can readily be accounted for. my and mxy . 5. the validity of this relationship should be checked for the particular plate under consideration. In FE analysis. shear force per unit breadth can be calculated. program from these constants directly according to equation (5. equations (5. These are generally given at the element centres and/or corners.42) gave an expression for the moments mxy and myx in a thin materially orthotropic plate. When materially orthotropic finite elements are used.48). a check is useful to ensure that the values given are consistent with those at the neighbouring nodes.47) and (5. The finite elements will behave according to these equations. five elastic constants. Isotropic bridge slabs In the case of bridges which are idealised as isotropic plates. y. These are plate elements which can model out-of-plane bending. The material properties of the elements are defined in relation to the material properties of the bridge slab. It was shown above that a grillage model does not take account of the derivative of the twisting moment. and unlike a grillage analysis. ν and ν typically need to be specified. The finite elements will satisfy this equation. E y.4. it is advisable to determine whether or not shear forces are calculated correctly using equations (5. If this is the case.67).4. will account for the effect of curvature in one direction on the stiffness in the other direction.35) give expressions for the moments mx and my in a thin plate. Equation (5. and the problem inherent in grillage modelling of torques per unit breadth not being equal in orthogonal directions does not arise. programs assume a value for Gxy based on the values input for the other four elastic constants.

: (5. Equation (5.Page 187 In a typical program.35) gives a similar expression for my .73) and (5. and . ix ≠ .77) The equivalent element depth can be calculated from equation (5.74) The modulus of elasticity of the element in the X direction may be chosen arbitrarily to be equal to the modulus of elasticity of the bridge slab.74) gives: (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.e. 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. but there are two second moments of area per unit breadth. 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. However. there is only one modulus of elasticity. In such cases. Equation (5. the user simply specifies the element depth as: (5.72). for both directions.75) Substituting this into equations (5. E slab. i.73) (5.34) gives an expression for the moment. mx . In most geometrically orthotropic bridge slabs. .72) Geometrically orthotropic bridge slabs Geometrically orthotropic bridge decks are frequently modelled using materially orthotropic finite elements.76) and (5. but only one depth can be specified.

5 and 0.74) to give: (5. has been suggested by Troitsky (1967): (5. that the shear modulus given by the above expression may need to be reduced by a factor of between 0.78) was derived by assuming an average value of the elastic moduli in the two directions and an average Poisson’s ratio.73) and (5.3 was reported to come from an extreme case where the flexural stiffness in the two directions varied by a factor of 20. To determine if the influence of the shear modulus on the analysis is significant. mxy . the orthotropic nature of the plate might be better handled using a combination of elements and beam members or a three-dimensional model.80) diminishes as the variation in the elastic properties in the two directions increases. The lower value of 0. Alternatively. It was reported by Troitsky (1967). An approximate expression for the constant.79) To be consistent with the equations for and i elem derived above. .76) applies and equation (5. equation (5. Instead of arbitrarily equating the modulus of elasticity of the finite element in the X direction to the corresponding modulus of the slab. is given by equation (5.80) Equation (5.78) For a geometrically orthotropic slab with a single modulus of elasticity and Poisson’s ratio.Page 188 For a materially orthotropic slab. 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. a similar expression can be determined by substituting from equations (5.3.42). Consequently the accuracy of this and equation (5.80) and analysing again using a shear modulus of half this value. This would lead to alternative expressions to the above. These types of model are discussed further in Chapters 6 and 7.79) becomes: (5. Gxy. the moment/curvature relationship for the twisting moment. As an alternative. Then. an arbitrary depth of finite element could be chosen (say. from the results of analysis and experimentation on steel orthotropic bridge decks. 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. the moduli in the Y direction could be equated. the authors would suggest analysing the orthotropic plate using a value predicted by equation (5. In such cases the shear modulus may need to be reduced.

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

22 Meshes of finite elements at transition between coarse and dense mesh: (a) potentially problematic arrangement.21 Possible shapes of quadrilateral finite elements: (a) generally good shapes. (b) good arrangement . (b) potentially problematical shapes Fig.Page 190 Fig. 5. 5.

5. at an angle of θ the Y axis. mxy and myx. The spacing of elements in the longitudinal and transverse directions should be similar. 6. 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. for example. the use of elastic springs to simulate deformable bearings or ground conditions as discussed in Chapter 4. if excessive. . my and mxy .22(b) where mid-side nodes are not needed and all elements are connected. A small segment of slab is illustrated in Fig. All vectors are resolved parallel and perpendicular to AB in Fig. This section addresses the design problem of how the engineer should calculate the moment capacity required to resist such moments. it is useful to compare the output of a model with the chosen mesh density to that of a model with a greater density. For convenience. This may involve. 5. This is generally easily achieved. This will be complied with if the first recommendation is adhered to. the three components can be combined using vector addition in a manner similar to the concept of Mohr’s circle of stresses. are also illustrated in this figure.5 Wood and Armer equations Much of this chapter has been concerned with methods of analysis of slab bridges. If mesh density is in question. However. where the moment is about the axis of the arrow. as can to be seen in the figure. As bending moment is a vector. The twisting moments per unit length. Resultant moments can be calculated at any angle of orientation and can. 1997). is introduced where N is normal to the face AB and T is parallel (tangential) to it. The results of such analyses give three components of bending moment at each point. . Similar results from both would suggest that the mesh was sufficiently dense.23(b) using double headed arrows to denote bending moment. 7. 5. The direct moment per unit length on AB is denoted mn and the twisting moment per unit length is denoted mnt. AB. 4. the projected lengths on the X and Y axes are l sin θ l cos θ and respectively. 3. N–T.23(a) and the possibility is considered of failure on a face. . The length of the face AB is l and. The vectors representing the moments are resolved to determine the moments on the face AB. 5. Shear forces near points of support in finite-element models tend to be unrealistically large and should be treated with scepticism. mx. Elements should be located so that nodes coincide with the bearing locations.Page 191 elements (3) and (4) connected to the mid-side node of element (1) at Q.24. Supports to the finite-element model should be chosen to closely resemble those of the bridge slab. 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. result in yield of the slab at any such angle. A mesh is shown in Fig. 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. a second axis system. 5. 5.

25 where resolution of components gives: (5.23 Segment of slab: (a) geometry.84) .83) and: (5.82) The components of moment on a face perpendicular to AB are considered in Fig. 5. (b) applied bending and twisting moments Considering components parallel to AB first: (5.Page 192 Fig.81) Considering components perpendicular to AB gives: (5. 5.

Page 193

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

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

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

Page 194

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

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

Page 195

Fig. 5.27 Segment of slab illustrating the moment capacities provided

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

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

Dividing the equation by cos 2 θ gives:

This can be expressed as:

where (5.88) and

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

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

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

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

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

Page 197

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

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

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

Substituting from equation (5.91), this becomes:

92) Similarly.92) and (5.95) (5. It is required to determine economical moment capacities given that providing costs twice that of providing . my=80 and mxy=20. Example 5. for example.90)) that and mxy were of the same sign. 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. 1968).Page 198 It was established earlier (by comparing equations (5. In general.96) These are known as the Wood and Armer equations (Wood.93).93) Any value for can be selected by the designer and these equations used to determine the minimum required moment capacities. then ρ and equations (5. Hence. The cost of providing moment capacity in the two coordinate directions may not necessarily be equal as a bridge may.89) and (5. If the cost of providing moment capacity is the same in both directions.93) become: (5. equation (5. be prestressed in one direction and reinforced with ordinary reinforcement in the other.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. their product is positive giving: (5.94) This can be used to find an economical value for in equations (5.91) becomes: (5.92) =1 and (5. mx=190. .

Equations (5.93) then give the required moment capacities: .e.92) and (5. ρis 0.5 and the minimum cost value for the . i.Page 199 As cost is proportional to critical angle is defined by: the constant.

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

1: (a) plan. It is to be constructed of prestressed concrete and is to have a uniform rectangular cross-section of 0. with an additional line at the centre of the deck. (b) section .8 m depth.1 Plan view of two-span bridge Fig. A combination of fixed.2 and 1.5 m which gives a ratio of transverse to longitudinal spacing of between 1.Page 201 Example 6. Figure 6.3 times the depth from the edge of the slab.2 Grillage mesh for bridge of Fig. The deck is supported on four bearings at either end and on two bearings at the centre as illustrated in the figure. 6. 6. 6. a row of longitudinal members has been placed at a distance of 0.2 (a) shows a convenient grillage mesh for this bridge deck. 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 2.5. As recommended in Section 5.1: Grillage model of two-span right slab A two-span bridge deck is illustrated in Fig. The longitudinal members have been placed along the lines of the bearings. The transverse members have been placed at a spacing of 1. Fig.3.1. 6. The end rows of transverse members are taken through the centres of the bearings. free-sliding and guided-sliding bearings is used so that the bridge can expand or contract freely in all directions in plane.

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 (5.63 m was used.3.3d=0.5 as the slab extends 0. 6. The breadths of the elements are chosen such that nodes coincide with the locations of the supports.1 Second moment of area (m4) Longitudinal members R1. the breadth is 1.0491 0.0470 0.1 and Fig.0981 0.2: Finite-element model of two-span right slab A planar finite-element model is required for the bridge deck of Example 6.1280 Figure 6. in keeping with recommendation number 6 of Section 5. this is reduced by 0.0483 0.3 shows a convenient finite-element mesh. The two rows of .0938 0. It can be seen that this breadth is taken to be from midway between adjacent members on either side. 6. R7 R4. These values are presented for all of the grillage members in Table 6.66): 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.5 m past the centre of the bearing. Figure 6. 6. R9 R2.0470 Torsion constant (m4) 0.2 (b) shows a cross-section of the slab with the grillage members superimposed.0371 0.7.87− 0. when determining the value of the torsion constant of the longitudinal members in rows R1 and R9. Similarly. a reduced breadth of (0. Example 6. R6 R5 Transverse Members End members All intermediate members 0.5/2+0.1. However.0938 0.0537 0. R8 R3.1.24)=0.24 m for the calculation of the torsion constant. This is used to determine the breadth of slab attributable to each longitudinal grillage member.2.2 (b). For the transverse end members.0862 0.0534 0.0640 0. The longitudinal members have been grouped by row as R1 to R9 and the transverse members have been grouped as end members and all intermediate members as illustrated in Fig.0964 0.1 Grillage member properties for Example 6.Page 202 Table 6.

in the case of concrete bridges. 6. may be precast to ensure a good quality of finish. the appropriate stiffness is determined by first finding the neutral axis location for the complete deck. All of the elements are assigned a depth of 0. either to stiffen the edge. In slab bridges.5 shows the cross-section of a deck with a long slender edge cantilever with an upstand at its edge. causing the bridge neutral axis to rise. 6. This is a somewhat arbitrary choice. such as those illustrated in Figs. This type of construction is chosen partly for its reduced self weight and partly for its slender appearance (see Section 1. but the extra number of elements in the model chosen is not considered to be excessive.1. E=35×10 6 kN/m2. equal to the average breadth of the elements.3 Edge cantilevers and edge stiffening Slab bridge decks often include a portion of reduced depth at their edges known as an edge cantilever.1 elements at each edge of the model could be replaced with one row of 1. finding the location of the neutral axis may not be straightforward. the only geometric property which has to be assigned to the elements is their depths. Figure 6. are often included at the edges of the slab.8 m which is equal to the actual depth of the bridge slab. the neutral axis will not remain straight as the upstand tries to bend about its own axis. Bridge decks of this type are discussed further in Chapter 7. Only decks where the neutral axis remains substantially straight are considered here. The effect of an edge cantilever or an integral upstand/downstand is to change the stiffness of the bridge deck. These are frequently important aesthetically and. The properties of each part are then calculated about this axis. 6. and had the length been taken as.3 Finite-element mesh for bridge of Fig. As for Example 6. to carry a protective railing. where the edge cantilever is relatively short or stocky . Cross-sections of typical slab decks with edge cantilevers are illustrated in Fig.4. The length of the elements along the span of the bridge was chosen as 1.Page 203 Fig.4(c) and (d). 6. These will be similar to those illustrated in Fig.4. In such cases. It is not necessarily conservative to ignore the additional stiffness provided by them. the elastic modulus is taken to be that of the slab. 6. or simply for aesthetic reasons. In such a case. In some bridge decks.2 m which is equal to the breadth of the widest element. If they are made integral with the deck. then the increased stiffness which they provide generally needs to be considered. a similar degree of accuracy could be expected. 6.5 m breadth. As this is an isotropic bridge slab. Upstands or downstands. say.8). the upstand may not be integral with the bridge deck and can simply be considered as an additional load on it.

Example 6.3: Grillage analysis of slab with edge cantilever The cross-section of a prestressed concrete bridge slab with edge cantilevers is illustrated in Fig. The neutral axis is then taken to be straight across the complete deck and to pass through its centroid.6. 6. the neutral axis is found to be 563 mm below the top of the bridge deck. . 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. Details of a general approach to this calculation are given in Appendix C. which has a constant cross-section through its length.4 Typical cross-sections of slab decks showing cantilevers and upstands Fig. This can be determined by hand or by using one of many computer programs available for such purposes. In this case. It is required to design a suitable mesh of grillage members to model the structure. 6.Page 204 Fig. The bridge deck. 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 second row of grillage members from the edge. R7. • Two rows of grillage members. R4. but that chosen here seems reasonable.6 Bridge deck of Example 6. of grillage members are located to coincide with the supports to the bridge deck. Figure 6. Note that row R4 is not exactly at the centre of the portion it represents. 6.3 times the depth of the deck (0.Page 205 Fig. This is in keeping with recommendation number 6 of Section 5. is placed at a distance of 0.7(a) shows the divisions chosen and the corresponding grillage members. these members represent a portion of bridge slab of breadth 1000 mm and they are located at the centre of that portion. (b) plan The cross-section is divided into a number of segments.7(b).3 (dimensions in mm): (a) section.3. is located at the centre of the portion of cantilever which it represents. • The fourth row. The spacings of longitudinal grillage members is given in Fig. This distance corresponds to 0. In each case. are chosen between the supports. and middle row.3×1200=360 mm) from the midpoint of the sloping edge of the main deck. • The third row of members from the edge.7. . is placed at a distance of 90 mm from the edge of the cantilever. The location from which this distance is taken is somewhat arbitrary. 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 outermost row of grillage members. R2.3 times the average depth of cantilever. each of which is represented by a row of grillage members. Row R1. R3. R5 and R6 (and R8 and R9).

Due to the variation in depth between rows R2 and R3. 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 . the transverse members between these rows have been modelled as two separate members with a row of nodes where they join.27. it gives a good longitudinal to transverse spacing ratio. (b) schematic of cross-section showing spacing between members. This is a very dense mesh having a spacing less than the slab depth. (c) plan of mesh Figure 6.7 Grillage model (dimensions in mm): (a) cross-section showing grillage members and corresponding segments of deck. each row of longitudinal grillage members is considered separately. between 1:1 and 1:1.7 (c) illustrates a plan of the grillage mesh with dimensions in mm. Twenty one rows of transverse members with a spacing of 1000 mm were chosen. 6. However. For this example.Page 206 Fig. The second moment of area about the centroid (of the bridge) of each portion of deck is determined.

R8. 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.e.021 0. running from the edge as far as the row of nodes indicated in Fig.71) as this is an orthotropic deck : Table 6.013 0.2.178 0.110 0. R13 R2. These are labelled Tc in Fig.7.Page 207 of deck.029 0. Tc. 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).7(c).002 0. The first group are those in the cantilever portion.2. R10 R5.261 0.034 0. These are labelled Tm in the figure. are also calculated about their own centroids as it is about these that they will bend. R9 Transverse members Tc—End members Tc—Intermediate members Tm—End members Tm—Intermediate members 0. The depth of these members is taken as the average depth of the cantilever. 6.144 0.278 .019 0. The second group are those in the main portion of the deck and account for all of the other transverse members. Tm. 300 mm.290 0. The torsion constants for the members are determined in accordance with equation (5. R11 R4.144 0. The second moment of area of the transverse members in the cantilever.010 0. For example the second moment of area of row R7 is given by: All of the longitudinal grillage member second moments of area are presented in Table 6. The transverse members are divided into two groups. 6. R6.002 0. R7.2 Grillage member properties for Example 6.146 Torsion constant (m 4) 0.131 0. are taken about their own centroids as they will bend (transversely) about their own centroids. i. The results are presented in Table 6. R12 R3.143 0.3 Second moment of area (m4) Longitudinal members R1.

the second moment of area per unit breadth of the longitudinal members (with reference to Table 6. In doing this.2) is given by: Therefore the torsion constant per unit breadth of the longitudinal members.3 is not satisfied which required that the torques per unit breadth in the grillage members in the longitudinal and transverse directions be of the same magnitude. Tc. an approximation is made by taking an average value for the torsion constant per unit breadth of the transverse members. as the two distinct values are very close. Tc. Hence. the average value is considered acceptable.2) is given by: The second moment of area per unit breadth of the transverse members is 0. Tc.2) is given by: . is therefore: Considering the longitudinal members in row R3 and the transverse members Tm. To apply this equation. However. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6. R1.3. is given by: Considering next the longitudinal members in row R2 and the transverse members Tc. the X direction is arbitrarily chosen as the longitudinal direction. the condition of Section 5.002 m3. R2. the second moment of area per unit breadth of the longitudinal members (with reference to Table 6. At this stage. 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.Page 208 where and are the second moments of area per unit breadth in the X and Y directions respectively. the torsion constant per unit breadth of the longitudinal members. and the transverse members. The torsion constant per unit breadth of the transverse grillage members. and the transverse members. Considering the longitudinal members in row R1 and the transverse members Tc.

3×0.9=0.36 m. is given by: This value is adopted for longitudinal members R4 to R10.8(c) shows a plan of the finite element model with rows of elements labelled r1 to r14. the breadth is reduced by 0. 6.6.2.6(a) is divided into a number of segments in a similar manner to the grillage model. the torsion constant per unit breadth of the longitudinal members. the need to average two dissimilar values of torsion constant was avoided. R3.2=0. R4 to R10.3 and Fig. and the transverse members. These values are given in Table 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. the division of the deck for the finite-element model varies somewhat from that of the grillage. The cross-section of Fig. and the transverse members. The other longitudinal members.3×0. 6.Page 209 The second moment of area per unit breadth of the transverse members is 0. Example 6. Tm. The length of the elements (in the longitudinal direction) is taken as 1000 mm. Tc. and the longitudinal members in rows R1 and R13.8(a) shows the division of the deck and Fig.2) which is: Hence. Figure 6. The depths of the elements have not been drawn to scale in this figure. For the end transverse members. 6. For the end transverse members.09 m. R4 to R10. 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. Tm. the breadth is reduced by 0.3×1.4: Finite-element analysis of slab with edge cantilever It is required to prepare a finite-element model for the bridge deck of Example 6.8(b) shows a cross-section through the finite-element model. . Tm. have the same second moment of area per unit breadth (with reference to Table 6. For the longitudinal members in rows R3 and R11.3=0. is given by: This value is adopted for the longitudinal members in row R3. This results in 20 elements in each of the 14 longitudinal rows.144 m3 and therefore the torsion constant per unit breadth of the longitudinal members. As the nodes form the boundaries of the elements and the location of the supports must coincide with nodes. the breadth is reduced by 0.27 m. Figure 6.

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

1440 Ec delem (m) 0.0561 0. The shear modulus. r8. terms of the elastic modulus of the concrete.1138 0.76)): Equation (5.876 1. r12 r4.1456 0. 068 Ec 0. Ec .1456 (m3) 0. within the formwork before casting the concrete.204 0 . 414 Ec The bridge deck is geometrically orthotropic.2 for concrete. These are created by placing void formers. in The elastic moduli in the two directions and the equivalent depths of each row of elements are given in Table 6. usually made from polystyrene.3. . r7.3.77) then gives an expression for the elastic modulus in the Y direction. The variation of second moment of area in the two directions is allowed for by specifying two different elastic moduli. r13 r3. Ec : . as the second moments of area vary in two orthogonal directions. 414 Ec 0.4 Finite element row number r1. r10 (m 3) 0. is calculated using equation (5. In the finite-element program.989 Ec 0.4 Voided slab bridge decks Longitudinal voids are often incorporated into concrete slab bridge decks to reduce their self weight while maintaining a relatively large second moment of area.3 Finite-element properties for Example 6. then the equivalent depth. the elastic modulus and the second moments of area per unit breadth. 6. 106 Ec 0.989 Ec 0.109 1. it is common practice to model the voided slab using the same methods as are used for . delem . it is modelled as materially orthotropic with a single value for element depth.0490 0.204 1.0833 Ec 0. When the void diameter is less than about 60% of the slab depth.027 Ec 0 .838 0. r11 r5 r6. Arbitrarily choosing the elastic modulus in the X direction. It is common practice to discontinue the voids over the supports which has the effect of creating solid diaphragm beams there.1440 Ec 0.Page 211 Table 6.064 Ec 0. Assuming a Poisson’s ratio of 0. r9. r14 r2.0036 Ec 0. These values are also given in Table 6.732 Ec 0 .0013 Ec 0. 356 Ec 0. Figure 6. to be equal to the elastic modulus of the concrete.80) by substituting values for the Poisson’s ratio. values of were arrived at for each row of elements. .9 shows a cross-section through a typical voided slab bridge deck with tapered edges. to be used for the finite elements is found by equating the second moments of area of the element and the slab (equation (5.

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

Fig. This is quite often a reasonable assumption when considering transverse bending.61 0.80 0.66 0.1981) 0.78 0.68 0.64 0. 6.70 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 (6.72 0.9 0. Clearly this equation is only applicable to slabs with a sensible void spacing.85 0.55 0.76 0. the transverse stiffness can be approximated as being equal to the longitudinal stiffness. (1981) recommend using the method of Elliott which gives this quantity in terms of the depth of the slab. When the void diameter to slab depth ratio is 0.69 0.1) Equation (6. Examination of equation (6. d.7 0.79 0.75 0.51 0. 6.11): (6.8 0.88 0.80 0.11 Cross-section through segment of voided slab bridge Table 6.48 0.Page 213 transverse second moment of area.86 0.87 0. to that of solid slab. Bakht et al.56 0.77 0.1) shows that the presence of the voids reduces the transverse stiffness by only 12% for a ratio of 0.5 0.6 or less.82 0.6 0.70 0. dv (Fig.75 0.71 0.58 0. islab (from Bakht et al.65 0.64 0.6.90 0.84 0.62 0.90 .60 0.1).81 0.86 0. This equation assumes that the centre of the voids and the deck centroid (for transverse bending) are located at mid-depth.85 0. and the diameter of the voids. iv-slab.4 Ratio of torsional stiffness of voided slab.1) does not take into account the spacing of the voids as the authors maintained that this was not a significant factor.82 0.74 0.89 0.84 0.45 0.

This gives the values presented here in Table 6.5: Grillage model of voided slab bridge Figure 6. For a grillage model. The neutral axis passes through the centroid of the deck which is located at mid-depth as the voids are located there.5 .13 Grillage mesh for bridge of Example 6.6 Fig. However.12 shows the cross-section of a prestressed concrete bridge deck which incorporates circular voids along its len gth. 6. conclude that. jv-slab. Bakht et al.71) and Table 6. 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. to that of a solid slab of the same depth.Page 214 For the torsional stiffness of voided slabs per unit depth. Thus the total bridge is 25 m long consisting of 23 m of voided section and two 1 m diaphragms . It was suggested that the values given in Table 6. Bakht et al. Fig.4 are only applicable to internal voids in an infinitely wide slab because those at the edges possess much lower torsional rigidities.4 for the ratio of torsional stiffness of the voided slab jv-slab. The deck spans 24 m between the centres of supports and is supported on four bearings at either end as illustrated in the figure. (1981) recommend using the method of Ward and Cassell. j slab. in most practical cases.65) or (5.4 can then be used to determine jv-slab. The layout and member properties are required for a grillage model.5 and 6. 6. jslab can be determined from equations (5. The voids stop short at each end forming solid diaphragm beams 1 m wide over the supports.12 Cross-section through bridge of Examples 6. Example 6.

i.14 Segment of voided slab Figure 6.: The edge longitudinal grillage member represents a portion of deck equal to exactly half that of the internal members with the result that its second moment of area is given by: The second moments of area of the internal transverse members are determined using equation (6. The second moment of area of this member is found by subtracting the second moment of area of the circle from that of the rectangle. It is not considered appropriate to locate these grillage members at 0.e.1): Hence. As the void diameters are in excess of 60% of the slab depth. The internal longitudinal grillage members represent the portion of deck illustrated in Fig. 6. 6.Page 215 Fig. The longitudinal direction is taken to be the X direction.13 shows a suitable grillage mesh. the second moment of area is: . The transverse grillage members are located in 17 rows.3 times the depth of the slab from the edge as this location is within the void. for the internal transverse members.14. The longitudinal members are located midway between voids.5 m apart. the supports coincide with the locations of nodes in the grillage mesh. By using this arrangement. 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. 1. the slab is treated as an orthotropic plate and the properties of the longitudinal and transverse members are determined separately.

2 m square elements is chosen. will be 1.5 and Fig.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. Each longitudinal row of elements represents a strip of the deck from midway between one void to midway between the next. 6. The torsion constant per unit breadth for the diaphragm is given by equation (5. a mesh consisting largely of 1. the next row of transverse members.71): Example 6. as illustrated in Fig.5 m.Page 216 For the 1m wide end diaphragms.6: Finite-element model of voided slab bridge A finite-element model is required for the 25 m long voided slab deck of Example 6.5 m wide. Both the ratio dv/sv and dv/d are 0. are used to represent the diaphragm. The transverse rows of elements adjacent to the diaphragms at each end are 1. The second moment of area per unit .12. Interpolating in the table gives a ratio for the torsion constants per unit breadth of: Taking equation (5. 6.3 m wide in order to make up the correct total length.4.75 m wide and will have a second moment of area of: The torsion constant for the grillage members is found from Table 6. At the ends.67. each 0.15. the second moment of area is simply: As the diaphragm is only 1 m wide and the transverse members are spaced at 1. For convenience. adjacent to the diaphragm. two transverse rows of elements.

de . E x.1) gives: The slab is geometrically orthotropic. 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.2 m wide strip of the deck. To model this as a materially orthotropic plate. equation (6. equal to the modulus for the concrete.77) gives an expression for the elastic modulus in the Y direction: . as the second moments of area (rather than the moduli of elasticity) are different for the longitudinal and transverse directions. Selecting the modulus of elasticity in the X direction. then equation (5. the second moment of area per unit breadth is: For the transverse direction. it is necessary to calculate a single equivalent value for slab depth.15 Finite element mesh for bridge of Example 6.Page 217 Fig.6 breadth in the longitudinal direction can be found by considering a 1. 6.76) implies a depth of element of: Equation (5.

16 (a) and (b). During construction. 6. The extent of this load sharing is largely dependent on the stiffness of the slab. Many other methods exist. Transverse diaphragm beams can be used to provide additional load sharing between longitudinal beams. Wide diaphragms also serve to improve the shear . Therefore they can be analysed by similar methods to those proposed for slab decks in the preceding sections. In addition to this. 6. Consequently. This phenomenon is indicated in Fig. 6. The main load-carrying component of a beam and slab deck is the longitudinal spanning beams. On completion.80): Taking a Poisson’s ratio of 0.2 m thick and have moduli of elasticity in both directions equal to that of the concrete. a precast concrete slab or even a completely in-situ beam and slab as illustrated in Fig. the beams generally act alone and must be capable of carrying their self weight.67). 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. The shear modulus is calculated from equation (5.Page 218 where E v-slab is the modulus of elasticity of the concrete in the voided slab. an overly stiff slab may lead to a prediction of load sharing between adjacent beams which does not occur in reality. The slab acts to transmit applied loads to the beams by spanning transversely between them.16 (c). such as steel beams with a composite steel and concrete slab.5 Beam and slab bridges Beam and slab decks are used for a wide variety of modern bridges. the structural action of these decks is considered to be two-dimensional. it is important that the slab be idealised correctly in the model as.17. the most obvious being the casting of an in-situ concrete slab on steel or precast concrete beams as shown in Fig. the weight of the slab and any construction loads present. 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 slab provides a means for load sharing between longitudinal beams. The shear modulus for the diaphragms is given by equation (5. Beam and slab decks may be formed in a number of ways. 6.2. 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. this gives: The diaphragm beams are solid so the corresponding elements are 1. for example.

Page 219 Fig.1 Grillage modelling Grillage modelling of beam and slab decks generally follows the same procedures as for slab decks. 6.17 Load sharing in beam and slab decks: (a) thin slab—little load sharing. but quite often.16 Forms of beam and slab construction: (a) in-situ slab on steel beams. In precast concrete beam construction. a diaphragm beam is constructed over intermediate supports to provide additional continuity. continuity between adjacent spans may be provided by the slab alone. The obvious exception is that grillage beams should normally be . 6. 6. (b) thick slab— increased load sharing capacity by extending the portion of the bridge near a support which is solid. (b) in-situ slab on precast concrete beams. (c) in-situ beam and slab Fig.5.

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

20.Page 221 Fig. for a rectangular section according to Ghali and Neville (1997) is: (6.347 m The torsion constant is generally not given and must be determined by the analyst. Figure 6. 6. For the purposes of determining the torsion constant. the beam cross-section is approximated as two rectangles as illustrated in Fig. (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).2) . 6. The torsion constant. in this case.19 Beam and slab bridge deck: (a) cross-section. 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. the properties are: Area = 0. The section properties of the precast beam are generally given by the manufacturer.374 m 2 Second moment of area = 0.19 (b) shows the exact dimensions of the precast beam. J.0265 m2 Height of centroid above soffit = 0. but to factor the stiffness of the slab by this modular ratio.

Supports are located at the ends of each longitudinal beam (other than the dummy beams). If this is not the case. Figure 6. This gives a transverse to longitudinal member spacing ratio of 2:1 which is acceptable. Some grillage programs allow the definition of ‘dummy’ beams. 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. 0.20 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. 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. Additional transverse beams are located at 2 m centres between these to represent the transverse stiffness of the slab. then these beams should be assigned very small section properties relative to those used elsewhere in the grillage (say. both . 6.20 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). As the grillage model is planar. Transverse members are positioned at each end to model the diaphragms.Page 222 Fig. Applying this equation to the rectangles of Fig. This is a convenient method for applying loads such as those due to parapet railings. For the interior longitudinal members.21 shows a suitable grillage layout for this bridge deck.5%). 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. A longitudinal grillage member is positioned at the location of each Y-beam.

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

6. 6.22.3 times the beam spacing: Hence the centroid is: above the soffit. For the second moment of area: The torsion constant is: The slab acts as a flange to the diaphragm beams.Page 224 Fig. . 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.22 Section through end diaphragm beam Hence the second moment of area of the edge section is: For the transverse members. the properties are determined in the usual manner. For the slab bending about its own axis. The recommended flange breadth is the sum of the web breadth plus 0.

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

For the longitudinal grillage members.16 m which is equal to the depth of the slab. The finite elements continue to the edge of the deck resulting in a row of elements 0. Supports are provided at the ends of each longitudinal grillage member. 6. Example 6. The beams are then modelled by grillage members with the properties of the actual beams excluding the contribution of the slab. the second moment of area of the combined section is: Fig.8: Finite-element model of beam and slab bridge A finite-element model is required for the beam and slab bridge of Example 6. In the second approach. 6.24 shows a suitable finite-element mesh incorporating grillage members longitudinally. From Example 6. The finite elements are assigned a thickness of 0. One of two approaches can be taken. Grillage members are used for each of the Y-beams and for each of the end diaphragms. the slab is modelled using orthotropic finite elements with the true transverse and longitudinal properties applied in both directions.19.5 m wide at each side. Care should be taken when determining the properties of the finite elements representing the slab. This is generally straightforward to implement and follows the recommendations made for slab bridge decks. the slab is modelled using isotropic elements which are assigned a thickness equal to the depth of the actual slab. The modulus of elasticity and Poisson’s ratio for the beams are used for these members.7.Page 226 members. 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. In the first approach. 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 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. They are also assigned the elastic properties of the slab.24 Combined finite-element and grillage mesh . Figure 6. An element length of 1 m in the longitudinal direction results in a maximum element aspect ratio of 1:2 which is considered to be acceptable. The stiffness of the slab which has already been applied through the finite element is subtracted.

such an approach results in a great quantity of reinforcement and has been shown to be .Page 227 The second moment of area of the 0. From equations (6. 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.7 by adding the individual torsion constants of the Y-beam and slab. 6. However. Hence. the second moment of area for the end diaphragms in the grillage model was calculated as (refer to Fig. the torsion constant to be assigned to the grillage members is simply that of the Ybeam: In Example 6.7. the elements are present up to the centre of the diaphragm to represent the transverse stiffness of the slab about its own axis.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 6.3 Transverse behaviour of beam and slab bridges The top slab in a beam and slab bridge is often designed transversely as a one-way spanning slab supported by the longitudinal beams (Fig. 6.65): 6. As the slab is represented by the elements.25).22): For the finite-element model.5.2) and (5.

and the extent (if any) of transverse bracing to the cells. The result is that load is transferred from the slab to the beams by arching action rather than bending action alone.Page 228 Fig. Canadian bridges have been built without any transverse slab reinforcement but using steel straps to guarantee confinement.4. 6. 6. 6.5. alternative methods are available for their analysis which are generally more convenient. The most common type are box girder decks. Figure 6. the slab depth to beam spacing ratios were 1:12 and 1:13.6 Cellular bridges Cellular bridge decks are formed by incorporating large voids within the depth of the slab. The fourth form. In these cases. as was discussed in Section 6. The first two of these are longitudinal and transverse bending. as indicated in Fig. as illustrated in Figs. It is the transverse distortional behaviour that makes the analysis of cellular decks different from other forms. with large diameter circular voids. 1997).27(d). The behaviour is similar to that observed in Vierendeel girders. the stiffness of the individual webs and flanges. However. Clearly the provision of transverse diaphragms along the span of a cellular deck will significantly reduce the degree of transverse distortion. 1992) allows for the provision of much less reinforcement than would be found by an assumption of bending. is transverse cell distortion. . with single or multiple rectangular cells. The third form of behaviour is twisting. In two reported cases (Bakht and Jaeger. Voided slab bridges. To account for observed arching action. There are four principal forms of structural behaviour associated with cellular bridges.25 Detail of section in beam and slab deck quite conservative. 6. as indicated in Fig. 6. can also be considered to be of a cellular form. the Ontario Highway Bridge Design Code (OHBDC.26 shows a number of commonly used cellular deck forms. This distortion is caused by the localised bending of the webs and flanges of the individual cells. The principal factors affecting the distortion are the dimensions of the cells relative to the deck depth.27(a) and (b).27(c). The beams have a considerable lateral stiffness and have the effect of confining the slab. which characterises cellular structures.

6. (d) transverse distortion . (c) twisting.26 Sections through alternative cellular bridge decks Fig. (b) transverse bending.27 Behaviour of cellular decks: (a) longitudinal bending.Page 229 Fig. 6.

28(b).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. 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. from equation (6.6.Page 230 6.28 Distortion of single cell with stiff webs: (a) applied loading. 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.3) where I is the second moment of area and E is the modulus of elasticity. The vertical deflection due to the bending of a beam of length l. 6. where d is their thickness. 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. fixed against rotation at both ends and subjected to a vertical force P/2. then the distorted shape of the cell is as shown in Fig. then the load acting on each can be taken as P/2. Hence. is: (6. Figure 6. In this method. the deflection due to flange distortion is: (6. The method is illustrated below by means of an example. If it is assumed for now that the webs are stiff and that transverse distortion is caused by bending of the flanges only. the deck is idealised as a grillage of beam members in the usual manner. If the flanges are of equal thickness.3).4) Fig. (b) distorted shape . 6.

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

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

9) where a is the area enclosed by the centre line of the wall.29 would give: However. A formula suggested by Hambly (1991) halves the constant and removes the web term: (6.10) Example 6. The torsion constant for a thin-walled box section is given by: (6. It is assumed that the deck is continuously supported transversely at each support. the contribution of the webs is accounted for through the shear forces in the longitudinal beams and should not be accounted for again here. is an increment of length and di is the thickness of that increment. Transverse grillage members are located at the ends and at the central support to represent the transverse diaphragms. three-cell bridge deck with edge cantilevers. The two edge members represent the portion of deck from the edge to halfway between the first and second webs (Fig. one at the centre of each web.65 m . Applying equation (6. By summing moments of area about any point in the section.Page 233 case as illustrated in Fig. 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. There are 2 m thick solid diaphragms at the end and central supports.9: Shear flexible grillage model of a cellular bridge deck Figure 6.32). the centroid can be shown to be located at 0.9) to the single cell of Fig. 6. 6.11.32 illustrates a two-span.31(b) except in the edge cantilevers. Figure 6.33 shows a convenient grillage mesh. 6. Four longitudinal members are chosen. l i. Additional transverse members are placed at 2 m centres giving a longitudinal to transverse member spacing ratio of 1:1. A grillage model is required. The two internal members represent the portion of deck from halfway between the first and second webs to the centre.

Page 234 Fig.32 Cellular bridge of Example 6.8): . 6. 6. (b) longitudinal section Fig.33 Plan view of grillage mesh above the soffit. For the edge longitudinal members: For the internal longitudinal members: For the transverse members.9 (dimensions in m): (a) cross-section. the second moment of area per unit breadth. The second moments of area for the longitudinal members about this axis are then determined. itrans is given by equation (6.

36 m4. a common approximation for I-sections.67) gives: which results in a shear area of: The breadth of the transverse members is 2 m. giving: . Then equation (5.10): This gives a torsion constant for the interior longitudinal members of 0.7): For concrete. a Poisson’s ratio of 0.2 is assumed. 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. giving: For the longitudinal members. giving: The torsion constant per cell is given by equation (6. the shear area is taken as the area of the webs.Page 235 The breadth of the transverse members is 2 m.

The transverse grillage members should generally be oriented perpendicular to the longitudinal members. In highly skewed decks. uplift can occur at acute corners which is generally to be avoided. Figure 6. the span length and the width of the deck.1 Grillage modelling A suitable grillage model of a skew deck will depend largely on the angle of skew.3% of the actual area. An exception to this is in concrete decks where the transverse reinforcement is not .Page 236 The end and central diaphragm beams are 1. The greatest reactions will tend to occur at obtuse corners in skew decks and the smallest reactions at acute corners. Hence: 6. Large reactions at obtuse corners lead to high shear forces which can also be difficult to design for.2 m deep by 2 m wide. This deck will tend to span in the skew direction so the longitudinal grillage members are aligned in that direction.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. For a rectangular section. 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. 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. in reinforced concrete. Care is needed in modelling the support system in such cases as any flexibility will cause a redistribution of reactions. 6. Significant skew in bridge decks leads to a non-uniform distribution of reactions between supports. the Wood and Armer equations can dictate a requirement for top reinforcement near supports where hogging would not normally be expected. As a result. A high degree of twisting accompanied by large torsional moments (mxy ) are also associated with skew decks. 6.34(b) shows a suitable grillage layout.7 Skew and curved bridge decks Many bridge decks incorporate some degree of skew and others are curved in plan. 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. the shear area can be shown to equal 83.

6.34(c). This deck will tend to span perpendicular to the supports rather than along the skew direction. wide bridge deck with a small angle of skew and Fig. If significant edge beams or stiffening is provided to the bridge deck. Once again.36 shows a suitable grillage mesh for a curved bridge deck. follow the curved layout closely due to the fineness of the mesh. Bridge decks which fall between the extremes of Figs. Curved decks pose no particular problem for grillage modelling. although straight. (c) alternative grillage layout Fig. the transverse grillage members are orientated perpendicular to the longitudinal members. Care should be taken with the edge grillage members which generally will have to be orientated in the skew direction. highly skewed bridge deck: (a) plan view. (b) grillage layout. In such cases. 6.35(b) shows a suitable grillage layout. Some analysis programs will allow the use of curved beams. narrow. 6.35 will require a greater amount of judgement by the analyst in choosing a suitable grillage layout. Consequently. but straight beams will be sufficiently accurate if the grillage mesh is fine enough. then this should be allowed for when assigning the properties of the edge beams in the grillage. 6. Figure 6. 6. the longitudinal grillage members are orientated in this direction.35 Short. (b) grillage layout .Page 237 perpendicular to the longitudinal reinforcement. it is generally more appropriate to orientate the transverse members parallel to the transverse reinforcement as illustrated in the alternative grillage layout of Fig. The longitudinal members.34 and 6.34 Long.35(a) shows a short. wide bridge deck with small skew: (a) plan view. Figure 6. The Fig.

37 Alternative finite-element meshes: (a) skewed quadrilateral finite elements. especially for Fig.7. 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.2 Finite-element modelling Finite-element modelling of skew or curved decks should be carried out according to the recommendations for right decks. Generally.Page 238 Fig. This is an advantage that the finite-element method has over the grillage method. 6.36 Grillage layout for curved bridge deck transverse members radiate from the centre of the curve. 6. 6. (b) alternative triangular elements . In this way. they are approximately perpendicular to the longitudinal members.

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

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. the problems associated with bridge decks such as those with wide edge cantilevers are discussed. Figure 7. the . when the bending moment in a flanged beam varies from one point to another. Bridge decks with edge cantilevers are considered but it is stipulated that only those with short cantilevers should be analysed by the methods proposed. 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. the analysis of bridge decks using planar models is discussed.2 Shear lag and neutral axis location When a bridge deck flexes. the edges do not receive the same amount of axial stress as those near the centre of the bridge. a common centroid can be found and the entire bridge is often assumed to bend about a neutral axis passing through this point. In this chapter. When flanges or cantilevers are wide and slender. These are distributed transversely from one part of the deck to adjacent parts by interface shear stresses. 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. independently of the rest.2(b). This common neutral axis can be seen in Fig. If the bridge deck is now rejoined. 7.1 Introduction In Chapter 6. 7. Thus. each part would bend about its own centroid. The theoretical background is reviewed and a number of solutions are suggested including three-dimensional methods of analysis.Page 240 Chapter 7 Three-dimensional modelling of bridge decks 7.1. If a load were applied to the deck in this condition. the bridge deck has a noncontinuous neutral axis as indicated in the figure. longitudinal bending stresses are set up. In this condition. 7. As the rejoined bridge bends.

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

A three-dimensional analysis can automatically account for shear lag as it allows for variations in neutral axis location directly. a two-dimensional analysis.4. Hambly (1991) presents a chart for the determination of effective flange widths for beams subjected to distributed and concentrated loads. as it is from these points that longitudinal stresses begin to spread out into the cantilevers. The correct effective flange width to be used for the cantilever is largely dependent on the ratio of the actual cantilever width to the length between points of zero moment (points of contraflexure).3. 7. as described in Chapter 6. Hence. as illustrated in Fig. analysed with no allowance for shear lag. is often used which does not take account of shear lag. can be used to determine the maximum stress in the cantilever. but this tendency causes the overall bridge deck neutral axis to move towards the centroid of the cantilevers at the edges.2(d). Obviously they are not free to do this. The chart. 7. be . reproduced here as Fig. b. Such a non-uniform neutral axis is illustrated in Fig. There is a strong link between shear lag and neutral axis location.3 Actual and calculated distributions of longitudinal bending stress at top of flanged deck . relates the ratio of effective flange width. a two-dimensional model with an effective flange width.Page 242 centroidal axes. 7. 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. It is possible to overcome this problem by assuming an ‘effective flange width’ for the edge cantilevers. and actual flange width. 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. 7.3 Effective flange width In the design of bridge decks. The effective flange width is also dependent on the form of the applied loading. 7. to the ratio of actual flange Fig.

Example 7.1: Effective flange width Fig. 7. and common approximations (dashed line) width. The cantilevers are 2. 7.5 Cross-section of bridge deck of Example 7. b.Page 243 Fig.5(a) shows the cross-section of a bridge deck with edge cantilevers. Also shown in the figure are the popular approximations for this relationship: and: which can be seen to be reasonably accurate for relatively wide flanges. and length between points of contraflexure.1: (a) showing actual cantilever widths. Figure 7. (b) showing effective flange width .4 m wide and the deck has a single simply supported span of 20 m.4 Effective flange width for different loadings (solid line). L.

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. One such technique is three-dimensional finite-element analysis using solid ‘brick’ type elements. is equal to the span length in this case. Inclusion of voids.12. is 2.93 and 0. A constant stress is assumed in the modelled portion of the cantilever and that part of it outside the effective flange width is ignored.6 shows such a model of a portion of bridge deck with edge cantilevers.23 m and 1. Hence.5(b) shows this effective flange width for one of these load cases. From Fig. Figure 7. Unfortunately. the length between points of contraflexure. Figure 7. the model automatically allows for any variations in the location of the neutral axis and hence allows for shear lag in edge cantilevers. L. the ratio of the cantilever width to this length.Page 244 As the span is simply supported. a cellular structure or transverse diaphragms pose no particular problems. 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. In addition to this. the use of such models is currently limited mainly to research and highly specialised Fig.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. 7. When the effects of shear lag are significant.67 for the uniformly and point loaded cases respectively. This results in effective flange widths of 2.4/20=0.61 m respectively. 7. some form of threedimensional model is necessary to achieve an accurate representation of the behaviour of the structure.6 Portion of bridge deck modelled with solid brick elements .4 the ratios of be/b are 0.

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

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

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

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

11(b) shows the cross-section with a suitable upstand grillage model superimposed. The grillage members representing the main part of the deck are located at the centroid of that part which is at 0. Figure 7.2 m below the top.4 m. The properties of the members in the upstand grillage model are easily determined. (c) plan view of grillage mesh Figure 7. The X direction is arbitrarily chosen to be parallel to the span of the bridge. the second moment of area per unit breadth is calculated .3 times the depth of the side (at that location) from the edge in accordance with the recommendations of Section 5.7. This gives vertical members with a length of 0. Seventeen rows of transverse members are provided at a constant spacing of 1.6 m from the top. The members representing the edge cantilevers are located at the centroid of the cantilevers which is 0. (b) crosssection with grillage members superimposed. Assuming the main deck slab to be isotropic.5 m.2 (dimensions in mm): (a) cross-section.3.11(c) shows a plan of the upstand grillage mesh. the portion of bridge deck associated with each grillage member is indicated by the broken lines.11 Upstand grillage model of Example 7.Page 249 Fig. 7. The grillage members representing the cantilevers and the outermost members in the main part of the deck have been positioned at 0.

5 m past the centres of the bearings. giving member breadths of 1.11. the bridge extends 0.65): The second moments of area and torsion constants for the grillage members are then determined by multiplying these values by the relevant breadths of the members shown in Fig. resulting in a second moment of area of: a torsion constant of: and an area of: .Page 250 as for a beam: The torsion constant per unit breadth for longitudinal and transverse members is calculated according to equation (5. 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. For the longitudinal members in the main deck. 7.

216 0.432 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 0.173 J (m4) 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 (5.0066 0.0073 0.65): 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.50 .0121 0.346 0.0161 0.173 0.276 0.60 1.1 Upstand grillage member properties for Example 7.0042 0. other than those at the ends of the deck.180 0.Page 251 For the edge cantilevers.32 1. the second moment of area is: the torsion constant is: Table 7.2 I (m4) Longitudinal members Cantilever Main deck (interior) Main deck (edge) Transverse members Cantilever (interior) Main deck (interior) Cantilever (ends) Main deck (ends) 0.3 times the depth (at that location) which gives: The area of the longitudinal cantilever members is given by: For the transverse cantilever members.0080 0.50 1.44 1.80 0.256 A (m2) 0.

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

7.12 Upstand finite-element model Fig.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. An upstand FE model is required.10. .3: Upstand finite-element model Figure 7.13 Calculated longitudinal bending stresses at span on top surface of deck The bridge deck of Fig. Similar results were found at midspan and for all other cases considered. 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).Page 253 Fig. 1998).14(a) shows the cross-section of a bridge deck with wide edge cantilevers. 7. Figure 7. Example 7. The deck is continuous over two spans of 24. 7.9 was analysed by the authors using an upstand FE model (O’Brien and Keogh. It can be seen that the upstand FE model predicts an almost identical stress distribution to the elaborate three-dimensional brick FE model.8 m and is supported along the entire width of the main part of the deck at each support location. The distributions predicted by the elaborate three-dimensional brick FE model and the plane and upstand grillage models described previously are also shown. 7.

6 m in from the edge of the cantilever.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. Figure 7. This is caused by the inability of the planar model to allow for the rising neutral .2 m below the top of the bridge deck. 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. Those representing the main part of the deck are located at the centroid of that part which is 0. All of the elements are 1. As the model is symmetrical about the central support.Page 254 Fig. 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. 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.15(b) illustrates the corresponding distribution along a line 0. zero stress close to span and maximum tensile stress above the central support.4 m.2 m wide and 1. This results in vertical members with a length of 0. 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. (b) finite-element mesh Figure 7.3: (a) cross-section (dimensions in mm).14 Upstand finite-element model of Example 7. only one span is shown in the figure. The elements in the main part of the deck are given a depth of 1. This stress distribution follows the expected pattern with zero stress at the ends. This model was analysed by the authors under the action of self weight. 7.14(b) shows a three-dimensional view of a suitable upstand finite-element mesh. maximum compressive stress close to span.24 m long (in the span direction). Figure 7. The plane FE model predicts a significantly greater stress at both the mid-span and central support locations. The elements representing the edge cantilevers are located at the centroid of the cantilevers which is 0.4 m.6 m from the top of the deck.2 m and those in the edge cantilevers a depth of 0.

Page 255 Fig. This example shows the benefits of three-dimensional modelling over planar modelling for bridge decks of this type. Alternatively this can be viewed as the inability of the planar model to allow for shear lag.6 m in from edge of cantilever axis in the edge cantilever. (b) 0. Most FE programs only allow the specification of a depth for the finite elements which does not .3: (a) at centre.15 Longitudinal bending stress at top fibre for bridge of Example 7. as the stiffness of each part of the deck is made up of a combination of both of these.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. 7. 7.

This is sufficient when dealing with solid slabs. Figure 7. The second moment of area of this is: and the area is: .Page 256 allow the independent specification of area and second moment of area. the equivalent depth of the elements will generally be quite close to (but smaller than) the actual depth of the voided slab.2 m long elements in the longitudinal direction would be appropriate for this model. a finite element with a depth chosen by considering the second moment of area of the voided slab will have an excessive area. 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.4: Upstand FE model of voided slab Figure 7.16(b) shows the cross-section of a suitable upstand FE model for this bridge deck. When considering the longitudinal direction. A solution to this problem is to reduce the area of the elements. 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. but causes problems when dealing with voided slabs. Clearly a member with negative area has no physical meaning and. A choice of 20. This is not the case when considering the cross-sectional area which is greatly reduced by the presence of the voids. most computer programs will not allow this. In other words. The deck is simply supported with a 24 m span and is supported continuously across its breadth at each end. The X direction is chosen as the longitudinal direction. 1. As the voids are generally located close to mid-depth of the slab. In this case the vertical members are 0. For the elements in the main part of the deck. this will result in an overly stiff model. this could be done by incorporating additional grillage members into the model with a negative area and zero second moment of area. the presence of the voids does not greatly affect the longitudinal second moment of area of the deck. In theory. They should also have zero in-plane second moment of area as the in-plane behaviour is still modelled by the finite elements.4. quite sensibly. Example 7.2 m wide with one void. The additional grillage members should have zero area. An upstand FE model is required. Therefore. Modelling of voided slabs by the plane FE method is discussed in Section 6. 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.16(a) shows the cross-section of a voided slab bridge deck with wide edge cantilevers.35 m long. 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. each element represents a portion of deck 1.

is: This gives a shortfall in second moment of area which has to be made up by additional grillage members.1): . The second moment of area of these additional members. 7. These elements have the same equivalent depth of 0. of: The second moment of area of this equivalent solid element.16 Upstand finite-element model of Example 7. is: To incorporate the additional members in the model.17. 7.879 m and the longitudinal grillage members have second moments of area of 0. deq . Ieq .093 m 4. each finite element in the main part of the deck is replaced by four elements and four grillage members as illustrated in Fig. (b) section through finite-element model Equating this to an equivalent solid element with the same area gives an equivalent element depth.4 (dimensions in mm): (a) cross-section. The required transverse second moment of area per unit breadth is given by equation (6.Page 257 Fig. .

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.18 Upstand finite-element model with additional grillage members (half) .

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. 7.18.6 m intervals to join the meshes on the different planes. This approach has the advantage of simplicity as there is a direct correspondence between each member and Fig.6 elements to give nodes at 0. this is clearly an approximation as the exact location of the neutral axis will depend on the flange widths and the relative stiffnesses of the members. The properties of the remaining parts of the deck are then calculated. 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.19(b).6. Only one-half of the model is shown as it is symmetrical. 7.Page 259 The edge cantilevers are modelled as finite elements with a depth of 0. an upstand FE analysis can be used to represent the behaviour more accurately than the alternative planar models. (b) section through upstand finite-element model .2 elements in the cantilever were replaced with four 0.19(a) shows a beam and slab bridge.6 m intervals. the originally proposed 1. Figure 7. Rigid.2×1. 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. each about its own centroid. 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. In such cases where the location of the neutral axis is unclear.6×0. or very stiff vertical grillage members are specified at 0. The final upstand FE model with grillage members shown as dark lines is illustrated in Fig. However.5 m which is equal to the actual depth of the cantilever. Therefore. and are represented by grillage members at the levels of those centroids as illustrated in Fig. provided care is taken to ensure that good similitude exists between the model and the actual structure.19 Upstand finite-element model of beam and slab bridge: (a) cross-section. 7. 7.

20 Plate finite-element model of cellular bridge: (a) original bridge. Transverse diaphragms could also be incorporated into this model with ease. Figure 7. The calculated moment for each beam member is only applicable to bending about its own centroid. (b) finite-element model a part of the structure. the number of elements required to achieve this is very large and this. combined with the tedium of interpreting the results. If reinforcement is to resist the stresses in a beam and the adjacent elements. 7. 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 interpretation of the output can be tedious. This model. has the advantage of automatically allowing for transverse cell distortion as discussed in Section 6. 7. The out-of-plane behaviour is affected by the vertical components of tendon force and by the moments induced by tendon eccentricity. Unfortunately. 7.20(a) shows a cellular bridge deck and Fig.6.20(b) shows a suitable model based on a variation of the upstand FE analogy. The bridge deck is then analysed to determine the effects of the equivalent loading. The stresses determined . 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.7 Prestress loads in three-dimensional models When analysing for the effects of prestress in bridge decks. it is usual to uncouple the in-plane and out-of-plane behaviours. However. as well as dealing with a varying neutral axis. 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. often rules out its use. to correctly model longitudinal bending there.Page 260 Fig.

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

upstand FE analyses with equivalent loading calculated in the traditional way (as described in Chapter 2) did not always give accurate results. (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.21 Portion of prestressed concrete deck: (a) original deck. In particular.Page 262 Fig. . 7.

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

Page 264 .

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

Page 266 .

72×106 6.1) 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 − 23.1) are given in Table C.40×106 0 0 0. .1 where Top and Bottom refer to the numerator and denominator respectively of the fraction specified in the equation.76×10 9 Bottom − 39. 6.1) where xi and y i are the co-ordinates of point i and n is the number of co-ordinate points. Table C. of any section can be found from the co-ordinates of the perimeter points using the formula: (C. 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).1 Evaluation of equation (C.60×106 0 0 0. For the purposes of this calculation.Page 267 Appendix C Location of centroid of section The centroid. point n+1 is defined as equal to point 1.19×10 9 2. The terms of equation (C.93×10 9 0 0 − 20. For the section of Fig.6.48×106 0 0 − 32.64×10 9 .

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

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

This force is illustrated in Fig.1 for a segment of cell between points of contraflexure.Page 270 where V is the total shear force and i t and ib are the second moments of area per unit breadth of the top and bottom flanges respectively. 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. 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.

this becomes equation (6.7): .Page 271 If the second moments of area per unit breadth are expressed in terms of the flange and web depths .

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

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

42.Page 274 Index Page numbers appearing in bold refer to major entries AASHTO 40 Aesthetics 34–9 Aluminium deck 42 Analysis. 147–50 Cross-section 2–8 box 5–7 older concepts 7− 8 solid rectangular 2–4 T. post-tensioning 14 Creep 12. 228 Box section 5–7 Bridge bashing see Loading. 169. box Centroid. location of 267–8 Collision loads see Loading. 231. 180. 147–50 Computer implementation of grillages 179–80 Concrete. 185 elastomeric 31 pot 30–1 sliding 17. 265–6 Blister 16 Box culvert 21–3. 72–4 Box girder 212. 29–30 spherical 30 see also Supports Bending moments due to applied loading 263–4. 228–36 grillage modelling 230–6 three-dimensional finite element model 260 transverse cell distortion 228. 13–18 Contraction of integral bridges 128–33 Coupler. 17. impact Composite 25. 260. 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. 18. 28. 78. impact Buried joint 32 Cable-stayed 25–6 Cantilever 3 balanced 14–17 Cellular bridge 212. 229. 82.4–5 voided rectangular 4 . 182. lightweight 42 Continuous beam/slab 10. 269–71 Cellular section see Cross-section. 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. introduction to 67–120 Anisotropic 151.

54 Imposed traffic loading 40. 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. 211.Page 275 Culvert. 230 similitude with bridge slab 171–3 sources of inaccuracy 180–2 U-beams. 184. 107 Eurocode 40 Expansion of integral bridge see Integral bridge. 49. 228 Differential settlement 9. 180–2. 260–1 Integral bridge 21–3. 237. 180. 75–7 Differential temperature 47–51. modelling of 225 see also Upstand grillage modelling Halving joint 24 Impact loading 41. 41. 220. 177–9. 218–19. 178. 245–7. 240. 88–9. 262 mesh 189–91 properties of elements 186–9 recommendations for modelling 189–91 similitude with bridge slab 186 slab bridges 185–91 see also Upstand finite element modelling Foundation. dynamic Geometrically orthotropic 152. 189. sources of inaccuracy Incremental launch 17 In-plane effects 162. 237 Effective flange width 242–4. 184 recommendations for modelling 182–5 shear flexible 212. 187–9. 92. 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. 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 . 246. 252 Edge stiffening 203–11. 51–2. 185. box 21–3. 244–6. 90. 40. 252. expansion FEA see Finite element analysis Finite element analysis (FEA) application of planar 200–39 beam and slab bridge 225–7 brick elements 244. 211. 72–4 Curved bridge 236–9 finite element modelling 238–9 grillage modelling 236–8 Dead loading 40. 245 Elastomeric bearing 31 Elevations 8–26 Equivalent loading due to prestress 54–66. sources of see Grillage. 253. dry 127 Diaphragm 10. 260–1 Equivalent loading due to temperature/ thermal effects 47. 43–5 Inaccuracy. 95. shallow strip 130 Forced vibration 53 Frame bridge 21–3 Frequency see Loading. 42 Density.

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

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

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