Safety of Bridges 9780727725912, 0727725912

Citation preview

Safety of bridges

Edited by Parag C. Das

THE

INSTITUTION O F

CIVIL E N G I N E E R S

ft^^I ^F^L

H I G H W A Y S A G E N C Y

Thomas Telford

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Published by ICE Publishing, 40 Marsh Wall, London E14 9TP. Distributors for ICE Publishing books are USA: Publishers Storage and Shipping Corp., 46 Development Road, Fitchburg, MA 01420

www.icevirtuallibrary.com A catalogue record for this book is available from the British Library ISBN: 978-0-7277-2591-2 © Thomas Telford Limited 2011 ICE Publishing is a division of Thomas Telford Ltd, a whollyowned subsidiary of the Institution of Civil Engineers (ICE). All rights, including translation, reserved. Except as permitted by the Copyright, Designs and Patents Act 1988, no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying or otherwise, without the prior written permission of the Publisher, ICE Publishing, 40 Marsh Wall, London E14 9TP. This book is published on the understanding that the author is solely responsible for the statements made and opinions expressed in it and that its publication does not necessarily imply that such statements and/or opinions are or reflect the views or opinions of the publishers. Whilst every effort has been made to ensure that the statements made and the opinions expressed in this publication provide a safe and accurate guide, no liability or responsibility can be accepted in this respect by the author or publishers.

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Preface

This book consists of a number of papers, most of which are based on presentations made at the symposium The Safety of Bridges, held in London on 4 - 5 July 1996. The symposium was sponsored by the Institution of Civil Engineers, London, and the Highways Agency, an executive agency of the Department of Transport, United Kingdom. The book is not intended to be the proceedings of the symposium, as many of the papers have now been enhanced to include more detailed information and new ones have been added to fill in perceived gaps. The main purpose of this book, as it was for the symposium, is to discuss in some detail the safety concepts which form the basis of modern bridge design and assessment codes, a subject which does not often feature in published materials. The secondary purpose is to describe the background work carried out in the development of the new UK bridge and route-specific traffic loading requirements, and the proposed whole life performance-based assessment rules. The work mentioned above represents a major departure from the current bridge design and assessment philosophy in that, instead of considering only the safety of a new structure in its as-built condition, or of an existing structure in its at-present condition, future whole life safety requirements will now need to be taken into account in an explicit manner. This will make it possible to compare alternative design and maintenance proposals on a more rational basis than at present. The book is aimed, first and foremost, at practising bridge engineers. How­ ever, it also contains sufficiently detailed information for it to be of use to research workers and specialist bridge engineers—particularly to those involved in research in the areas of whole life reliability analysis and prob­ abilistic loading. Although the book is concerned primarily with bridges and other highway structures, the principle of whole life performance related assessment and the concepts of safety are relevant to buildings and other structures as well. Hence, it is hoped that the book will also prove to be of interest to structural engineers in general. Apart from engineers, those involved with safety regulations for public struc­ tures, and also those interested in the economics of costing and allocation of maintenance funds for structures, may also find some of the papers interesting. Part 1 of the book starts with an introduction by Mr Lawrie Haynes, Chief Executive of the Highways Agency, which is based largely on the welcome address he delivered at the symposium. This is followed by papers giving an overview of the background developments, for example the current national bridge assessment and strengthening programme and the increasing use of

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PREFACE

design, build, finance and operate (DBFO) contracts, and how the bridge design and assessment rules now need to be updated to allow greater flexibility. In Part 2, an introduction is given to the safety concepts implicit in the current bridge codes. It also contains a paper on socially acceptable mini­ mum safety levels and another on the calibration of the new North American bridge design codes. Part 3 is devoted to the development of the new probabilistic bridge-specific short span assessment loading requirements. These requirements are now being incorporated into the current UK bridge assessment code BD 21. Recent developments in bridge engineering are taking place as a result of operational needs for the highway infrastructure where large numbers of structures are involved, and options and risks have to be considered care­ fully before action is taken. Available alternatives need to be evaluated in terms of capital costs as well as the hidden costs of traffic disruptions result­ ing from bridgeworks. Part 4 contains papers which give a flavour of various operational issues, maintenance options and whole life costing methods. It also contains a paper on a whole life cost optimisation procedure being developed in the United States. Part 5 follows on with a number of papers describing the basic principles behind the proposed whole life performance related assessment rules, which are intended to form an integral part of future bridge management procedures, and the first part of the current development work. The final part, Part 6, contains three papers which deal with overall risk analysis of bridges, thereby widening the discussions to safety issues other than those mainly involving traffic loading. The new developments which feature in this book, perhaps the first of their kind anywhere as far as practical application is concerned, have been possible only through contributions from top bridge engineers and academics from the UK and overseas, many of whom are authors of the papers in this book. I should like to take this opportunity to thank person­ ally, and on behalf of the publisher, the authors of the papers, all of whom have contributed with enthusiasm. Thanks are also due to Dr A.R. Flint, Mr B. Pritchard and Dr J.B. Menzies, who, as my co-members of the steering committee for the above mentioned symposium, took part in selecting the topics presented there, which now form the basis of the papers in this book. Finally, I should like to register my, and the publisher's, gratitude to Mr Lawrie Haynes, Chief Executive of the Highways Agency, for giving permission to publish the papers relating to the agency's projects and also for kindly allowing me to act as editor for this book.

Parag C. Das Group Manager for Structures

Highways

Management,

Agency

iv

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Contents

Part 1. Overview

1

Introduction. L. HAYNES Trunk road bridges—current needs for design and maintenance. T. A. ROCHESTER Bridge assessment and strengthening—local authority perspective. A. LEADBEATER DBFO contracts and bridge performance considerations.

3

J. BENT-MARSHALL

7 12 20

Part 2: Safety concepts and codes

27

Safety concepts and practical implications. P. C. DAS Bridge failures, hazards and societal risk. J . B . MENZIES Application of bridge reliability analysis to design and assessment codes. A. S. NOWAK

29 36 42

Part 3. Bridge-specific loading

51

Development of bridge-specific assessment and strengthening criteria. P.C. DAS Current UK bridge assessment rules and traffic loading criteria. A . R . F L I N T Development of short span bridge-specific assessment live loading. D.I.COOPER Traffic data for highway bridge loading rules. J. PAGE Collection of statistical data. N.J. RICKETTS Reliability evaluation of short span bridges. M.K. CHRYSSANTHOPOULOS, T.V. MICIC and

G.M.E. MANZOCCHI

53 58 64 90 99 110

Part 4. Options and economics

129

Bridge management problems and options. T.K. WILLIAMS Principles of whole life costing. G.P. TILLY A whole life cost model for the economic evaluation of durability options for concrete bridges. P.R. VASSIE Application of life cycle reliability-based criteria to bridge assessment and design. D.M. FRANGOPOL

131 138

Part 5. Whole life assessment

159

Whole life performance-based assessment of highway structures: proposed procedure. P.C. DAS

161

145 151

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

CONTENTS

Whole life performance-based assessment rules: development programme and issues. A.R. FLINT Revised rules for concrete bridges. P. THOFT-CHRISTENSEN, F.M. JENSEN, C R . MIDDLETON and

166

A. BLACKMORE

Lifetime reliability model for steel girder bridges.

175 C.-H. PARK

and

A. S. NOWAK

Data collection report.

189 A. BLACKMORE

and

C R . MIDDLETON

Part 6. General risk assessment An overall risk-based assessment procedure for sub-standard bridges. N.K. SHETTY, M.S. CHUBB and D. HALDEN Probabilistic risk assessment of corrugated steel buried structures—comparison of reliability between standard span and long span structures. H.A.D. KIRSTEN Probabilistic assessment of bridge scour and other hydraulic risks. A. McCRACKEN

203

223 225

236 243

vi

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Part 1. Overview

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Introduction L. Haynes,

Chief

Department

of

Executive

of the Highways

Agency,

Transport

The Highways Agency and bridges The Highways Agency is responsible for making the most of the existing trunk road network in England. That means maintenance and improvement of the existing network of which bridges form an important part. Their importance comes not only from their visual impact on the surrounding environment but also from the substantial resources required for their maintenance. Trunk road bridges assumed a rather prominent place in my mind recently while the subject was being investigated by the National Audit Office. Last year, Sir Patrick Brown, Permanent Secretary at the Department of Trans­ port, and I appeared before the Parliamentary Public Accounts Committee. This had the effect of focusing my thoughts even more. The single, most important message that Sir Patrick and I sought to get across to the committee was that safety is our number one priority. Natu­ rally, the debate was very wide-ranging, but whatever the issue—provision for heavier lorries, rating of parapets, allocation of funds, or management information systems—our retort was the same: safety is paramount! The following figures are evidence of the scale of our commitment to safety. Since 1988, capital expenditure on the trunk road bridge maintenance programme has been in excess of £800 million. The main driver for more than 80% of this expenditure has been the need to improve safety, whether it be large-scale works, such as the recent highly successful Marsh Mills project, or the simple replacement of a sub-standard parapet. About £145 million has already been spent on the assessment and strengthening of trunk road structures, and this element of our overall programme will predominate as 1999, and the increase in maximum lorry weights from 38 to 40 tonnes, approaches. Our objective here is to ensure that all structures carrying trunk roads, and other important routes over trunk roads, can be used safely by these heavier lorries from 1 January 1999.

Bridge assessment Clearly, the 15 year bridge rehabilitation programme for trunk road bridges represents a major endeavour by the Agency aimed at improving the bridge

3

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

stock so that it continues to perform adequately under the very onerous traffic conditions of today. Local authorities and other bridge owners are also carrying out similar exercises involving their own bridges. From the results coming out from these exercises so far, it is becoming clear that the current assessment requirements, which are based on the worst possible traffic and other conditions that may occur in any part of the country, are not flexible enough to deal with bridges in less demanding situations. This is a very important limitation, considering the enormous costs of dealing with the so-called substandard bridges, some of which may, in reality, be at extremely low risk. Many bridges without any sign of distress, for instance, have failed these assessments, so it is important to know whether the work is necessary to ensure public safety, or is being justified purely in terms of whole life cost considerations. This is where clear engineering judgement comes to the fore. The Highways Agency has undertaken a major revision of the bridge assessment rules through a number of projects involving experts from both here and abroad. The teams are examining the fundamental principles of the current assessment and strengthening procedures. The aim is to produce flexible whole life related rules. These proposed rules will also have considerable impact on future bridge design principles, and the way bridge maintenance activities are managed and funded.

Objectives The projects involved touch on a number of difficult issues, the most problematic one being the definition of an acceptable safety level for bridges. Any conclusions reached on these matters will need to he based on a wide consensus. The themes of last year's symposium—flexibility, risks and options—are particularly relevant as we now need to concentrate more on the manage­ ment of the network and on getting the most out of the existing roads, using all the resources and innovation that we can muster. In the early stages of building any infrastructure, engineering serves by providing the technology for building. The engineering objectives at this stage are quite straightforward: in the case of the highway network, how to build safer and better roads and bridges in the most (whole life) cost effective way. However, the benefits derived from any money spent on management activities on an established network are not so clear. The benefits of any remedial or preventative work are very complex. If there appears to be any immediate risk of collapse or failure, the work would obviously be essential. But, if the benefits lie only in terms of reducing future costs, these are likely to be apparent only in the long term.

4

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

This transition from the building phase to the maintenance phase for the network requires a change of emphasis in highway engineering, both for roads and bridges. Instead of pursuing engineering excellence purely as an end in itself, engineers should look at everything they do in the light of the overall requirements that society demands of the network. What are these requirements? First, as I explained earlier, maintaining and improving public safety is paramount. Second, the road is there to serve the road user. Keeping roads open for use must also therefore be a primary objective. Any maintenance work has to be designed to minimise any traffic disruption caused by it. Disruptions in traffic flow are not only inconvenient to the road user. At slow speeds, pollu­ tion from exhaust fumes is at its worst. Minimising the harmful effects on the environment in whatever we do is also therefore an important consideration in how the Agency manages its activities. Finally, the public and the taxpayers expect us to ensure value for money in whatever we spend. In determining value for money, we must look beyond the immediate or short term returns. Whole life performance in terms of future maintenance costs must be taken into account. More durable materi­ als and methods are therefore desirable as a matter of principle. Whatever engineers and others plan or implement, all these objectives have to be carefully considered, and the results must reflect a satisfactory balance between them.

Options I realise that technically it is a very difficult task to determine the extent to which any engineering work meets these core objectives. It is particularly difficult because long-term estimates are not only less reliable, but they are also affected by issues such as the extent to which real capital expendi­ ture can now be compared with, for example, future traffic delay costs. However, this is a challenge that the bridge engineering community now faces. Engineers have to be aware of the wider implications of their work, not only in the immediate term, but also for the future generations. They must consider, as far as practicable, all the options available and the risks associated with each of them, and recommend the best possible course of action. I have been told that, as a result of research that is being carried out around the world, it is now possible to develop much more sophisticated methods for considering risks and options in bridge engineering. The impetus for this work has come from the issues faced by highway authorities everywhere, the main problem being how to prioritise scarce funds among the multitude of urgently needed tasks.

5

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

It is very encouraging that the work presented here is the result of extensive co-operation by international experts, and is aimed at pulling together state-of the-art knowledge from many countries. I sincerely hope that such international co-operation will continue to flourish. I am also very pleased to note that the projects involved are not just academic exercises, but are producing techniques and criteria that benefit the practising engineer, and which will have real impact on the strategy and economics of managing road networks. I believe the new loading criteria, which in some cases represent a reduction of as much as 4 0 % from current requirements, are already being considered for use for the structures on the Midlands Links and the M4 elevated sections in London, not the most straightforward elements of the trunk road network from a structural maintenance perspective!

The future The new developments will mean a change of culture for the engineering profession. Codes and standards in future may allow engineers greater freedom to base their recommendations on comparative risks. It is essential to start training new generations of engineers so that they are fully equipped to take such decisions. The engineering institutions, the industry and, of course, government agencies, all have a responsibility for this. In this context, I am very pleased to be able to report that three major con­ sulting firms have joined the Highways Agency recently to provide funding for university students to carry out post-graduate studies on bridge risk and reliability. Partnership and co-operation are the key to a better future. The Agency is putting partnership alongside innovation to ensure that we really do get the most out of our existing and future roads and bridges.

6

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Trunk road bridges—current needs for design and maintenance T. A.

Rochester,

Chief

Highway

Engineer,

Highways

Agency,

London

Introduction Structural design and assessment have two essential aspects: analysis of the behaviour of the structure under load, and the intended levels of safety. For bridges, considerable research has been carried out through the years to improve the methods of analysis. By now, using the sophisticated computer based methods available today, and the results of the numerous laboratory and full scale tests carried out by the Transport Research Laboratory (TRL) and others, many structure types can now be analysed with a great degree of precision. Consideration of safety targets, however, has largely been left to the drafters of the design codes. For the modern bridge design codes, including BS 5400, the safety targets were duly considered, but in practice were deter­ mined mainly to be consistent with the earlier codes. This procedure is quite reasonable in itself since design requirements should incorporate the fruits of past experience, both of successes as well as of failures, and should avoid initiating drastic changes of practice. However, safety levels implicit in the design codes tend to be conservative because they have to adequately cover a large number of application situa­ tions as well as uncertainties regarding the materials and workmanship which will be used in the actual construction. Furthermore, they also have to ensure, not only the immediate safety of the structure, but also that safety can be maintained for the whole of the functional life without the need for major rehabilitation work. In recent years a number of difficult issues relating to safety levels have emerged in the context of procuring new highway structures, and in the course of maintaining the trunk road bridge stock. These difficulties have made it increasingly necessary to iden­ tify, as far as target safety is concerned, what is essentially required from the ethical (risk to life) point of view, and how much can be left to be determined by considerations of whole life economics. The Agency has put in place a number of research and development pro­ jects in the area of bridge safety, risk and reliability with the purpose of clarifying the basic objectives of design and management and to provide rational working procedures. This presentation recaps some of the issues that have led to these projects, which will be described in much of this book.

7

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDCES

Past developments The number of bridges in UK administered directly by central government form only a small proportion (some 10 % ) of the national stock. Nevertheless, these are on the strategic trunk road network and have to be maintained to the highest standard. Furthermore, central government also contributes through the Transport Supplementary Grant towards the bridge manage­ ment activities of the local highway authorities. As such, through the years, various government departments have taken the lead in commission­ ing R&D work in the area of bridge design and assessment. In 1933, owners of road bridges in UK were given legal powers to restrict the weights of vehicles crossing their bridges, and they did so using the methods of assessment available at the time. Experience showed however that vehicles very much in excess of the permitted weights could use the bridges without apparent ill effects and it was suspected that the rules were perhaps over-conservative. As a result, from that period onward, there has been continuous effort in this country to improve the methods of bridge assessment and make them more precise. From 1936, the Building Research Station commenced an extensive investi­ gation, sponsored by the Ministry of Transport, into the behaviour under load of various types of bridges. During the war years the problem became an important and urgent matter and, in 1942, the Ministry of War Transport initiated investigations into the load carrying capacity of bridges, which among other things, resulted in the MEXE method of arch bridge assessment. From the early 1960s, the increases in the volume and weights of heavy goods vehicles were causing serious concern to the bridge authorities. The ageing stock of older bridges, predominantly built during the days of canal and railway expansion, were particularly considered to be at risk, and even the more modern bridges were showing signs of increasing dete­ rioration. The Ministry of Transport then produced the earlier versions of the Bridge Assessment Code culminating in BE 4 in 1968, following which the first formal national bridge rehabilitation programme, Bridgeguard, was launched. Soon afterwards, the other bridge owners also started their programmes. By the late 1970s re-examination of traffic loading on the roads demon­ strated that the earlier load requirements, both for design and assessment, were becoming grossly inadequate as a result of successive increases in permitted vehicle weights in the Construction and Use Regulations and the general growth of road traffic. In the early 1980s, the Department of Transport (DoT/DTp) led a committee comprising all the main bridge autho­ rities in the country to revise the assessment code, the result of which was the present assessment code BD 2 1 . In 1988, the current 15 year bridge rehabilitation programme for trunk roads was launched. It was preceded by a bridge census and sample 1

1

8

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW 2

survey of the older bridges. The survey showed that some 5 0 - 6 0 % of the older concrete and steel bridges, of various types such as jack arches and bent plate decks, and about 1 0 % of the country's 33 000 masonry arch road bridges were likely to be found to be sub-standard. This meant a major impending financial and logistical burden for the country. Hence, the Department of Transport commissioned a multimillion pound research programme for improving the assessment methods for such bridges. In parallel, considerable resources have been employed by the DTp and, subsequently, by the Agency to develop assessment rules for more modern concrete and steel bridges.

New issues Assessment The purpose of all this work has been to eliminate as far as possible any unnecessary strengthening or repairs of bridges. As the rehabilitation pro­ gramme progressed, however, it became increasingly apparent that there were some fundamental limitations in the current assessment rules which made it difficult to achieve this aim. Firstly, the current rules are based on the worst possible scenarios of load­ ing and clearly were too onerous for some bridge situations. For instance, the same assessment rules are applicable equally to bridges carrying a motor­ way as well as a minor road, to a 20 year old bridge as well as to a 90 year old bridge, and to bridges of different types irrespective of their deterioration rates. Secondly, as mentioned earlier, if a bridge is assessed to be inadequate, the nature of the inadequacy is not clear—whether it needed to be strength­ ened for immediate safety or for whole life economic benefits. Thirdly, the options for a bridge assessed to be sub-standard are limited. It has to be replaced or strengthened as soon as possible, and in the interim period it has to be either weight restricted, width restricted, or propped, and monitored. For most bridges carrying busy trunk roads, such options are costly in terms of traffic disruption, and cannot be continued for very long. Hence, if a bridge is considered to be sub-standard, and the reason for this is merely economical, ideally one should be able to adopt any of a number of alternative strategies to immediate strengthening, for example to 'do-minimum' and accept a reduced remaining life for the bridge. In order to prioritise bridge maintenance funds effectively and to be able to forecast future needs of the bridge stock rationally, it is necessary to have assessment rules which answer the above questions. The Agency is at present putting in place a number of research projects to develop such a set of rules. This work is beginning to produce results and many of the papers in this book are related to these projects.

9

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Design 3

At present when a bridge is designed in accordance with BS 5400, the code merely says that it may be 'expected' to remain safe for 120 years without requiring any major rehabilitation. This is largely a guess, since no rigorous prediction of long term behaviour was possible at the time of developing the code. In the course of the present work it has been found that analytical pre­ diction of long term behaviour is now becoming possible and such methods are being used to produce whole life target profiles for different types of bridges. This may be of considerable significance for bridge design. Although the design life is assumed to be 120 years, many bridge compo­ nents such as bearings and waterproofing membranes can last different lengths of time. In recent years, it has become increasingly necessary to consider options at the design stage of different materials and components which sometimes have very different characteristics from each other. Leaving aside the need to choose between the main bridge materials such as steel and concrete, various durability options are also available nowadays, for example in the form of epoxy coated reinforcement, galvanised steel corru­ gated structures and soil reinforcement for retaining walls. Furthermore, new materials such as plastics, and some not so new materials such as masonry, are also being proposed for bridge construction. A l l these options represent different whole life maintenance needs and performance characteristics. The question of having flexibility in the expected life of bridges as a long term management tool also cannot be ignored. Experience has shown that different bridge types can be expected to have a different engineering life in any case. Such options, whether they occur at the design stage or when determining maintenance strategy, can only be compared rationally if whole life target pro­ files are available for each design to act as a yard stick. The current design rules will therefore need to be revisited and further developed with this fea­ ture. This means that in addition to the present tasks relating to assessment, the design code BS 5400 and the current versions of the bridges parts of the Structural Eurocodes will need to be modified to incorporate the whole life target profiles for various bridge and component types and materials. Whole life targets w i l l also provide a straightforward client specification for bridges built and maintained under the private finance initiatives such as the design, build, finance and operate (DBFO) schemes. The multitude of prescriptive standards used now could then be replaced by simpler core objectives based on whole life targets.

Conclusions To summarise, from the early part of this century, central government has taken a leading interest in improving bridge assessment methods and

10

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

rules. This work is still at the forefront of the Agency's activities in the area of bridge engineering. Past work has benefited not only the trunk road network but also the rest of the nation's road infrastructure. In today's context of procuring and managing bridges, new issues have appeared on the horizon and these need to be addressed. The Agency, on behalf of the Department of Transport, is again taking the lead in finding solutions using the best available expertise from home and abroad.

References 1.

B D 2 1 (1977). T h e assessment roads a n d bridges, H M S O ,

2.

Department

and strengthening

of h i g h w a y structures. Design m a n u a l for

London.

of T r a n s p o r t (1987). H M S O ,

London.

Sample

survey

and census

of

highway

structures. 3.

B S 5 4 0 0 ( 1 9 7 8 ) . T h e d e s i g n of concrete, tution,

steel

a n d composite

bridges.

British S t a n d a r d s Insti­

London.

11

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Bridge assessment and strengthening—local authority perspective A. Lead beater, Oxfordshire

County

formerly

Deputy

County

Engineer,

Council

Introduction This paper outlines the local authority view of the current UK bridge assess­ ment and strengthening programme and extensively draws on discussions that have taken place at the Bridges Group of the County Surveyors' Society. This group, although sponsored by the society, has representation from local authorities in England, Scotland and Wales as well as the Highways Agency and the Northern Ireland office. Railtrack and the Transport Research Laboratory (two very important players in the local authority bridges scene) are also permanent group members. This paper looks briefly at the background to the programme, at the issues which continue to surface, at solutions and arrives at some conclusions. It can only be seen as a review of the programme on what has become an ever-changing scene.

Background Comparisons

between

local authority

and government

bridge

stock

Mallett, in his paper on Bridge maintenance initiatives, presented to the Institution of Highways and Transportation Conference in 1 9 8 6 , gave the following information with respect to UK bridge stock (Table 1). 1

Table 1. UK bridge stock as at 1986

Type

No.

Motorway

5 000

Trunk road

8 000

Local authority Railway British waterways

Government owned

/ /

129 000 12 000 1000

12

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW 2

A later survey, covering England only, is shown in Table 2. 2

Table 2. Bridge stock in England

Government owned and maintained public highway bridges

9 515

Government owned 'other structures', culverts, tunnels

1 472 10 987

It is likely that owing to the construction of new trunk roads and motor­ ways, and consequent detrunking, local authority owned bridges have increased substantially since the 1986 report. The later survey gives the split of bridge type for government owned bridges in England (Table 3). 2

Table 3. Type of bridge stock owned by the

Government

Type

%

No.

Predominantly concrete

80

8 790

Predominantly steel

15

1648

5

549

Masonry arches

Information on local authority bridge stock shows a much greater varia­ tion, typically as shown in Table 4 (although this will vary from area to area). Table 4. Type of bridge stock owned by local

authorities

Type*

%

Masonry/brick arches conforming t o MEXE

22

28 380

Masonry/brick arches non-conforming

26

33 540

Steel/iron including wrought iron

10

12 900

Concrete

34

43 860

8

Stone slab

No.

10 320 129 000

* These statistics relate to England, Scotland and Wales.

These statistics (although approximate) demonstrate the very consider­ able difference between government and local authority bridge stock. The former is predominantly modern, predominately post-second world war and concrete. The latter has over half likely to be more than 100 years old and is a very mixed bag of structure type. It is interesting to note that both the Highways Agency and the local authorities estimate the cost of 2

13

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

3

assessment and strengthening to be in excess of £2 billion. It is important to understand, also, that such sums of money are not likely to be available for completion of the programme by 1 January 1999. Thus the total cost of assessment/strengthening of the country's bridge stock may be £4 billion.

Two tier highway network versus cordons It is generally assumed that it is unnecessary to strengthen bridges on minor roads which are physically unsuitable for larger lorries or which simply do not carry much traffic. Such assumption may not be valid in view of the requirements given in the Road Traffic Regulation Act 1985 S122 and the Highways Act 1984 S41 and S130 (both backed by case law). The key phrases in all of these sections is in S122 (RTRA 1984), which states that it is the duty of (every) local authority to ensure 4

5

• •

the desirability of securing and maintaining reasonable access to premises that the amenities of the area through which the roads run is preserved or improved.

It is almost certain that commercial pressures from road hauliers and also pressures from those concerned with protecting their existing amenity with respect to noise, pollution and road safety, would fiercely resist any whole­ sale change in the road system (such as the two tier system suggested above), unless confirmed by the government in legislation. Nevertheless, cordons have been created by local authorities which suc­ cessfully reduce the movement of HGVs through an area 'except for access'. Three key points are essential here. (a) The chosen through route must be clearly the best route from the point of view of the haulier (i.e. the carrot). (b) Enforcement by the constabulary must be regular (i.e. the stick). (c) There must be a reduction in annual and maximum loading to be taken on bridges within the cordon which is recognised by the assessment code. This, in my view, must be a key element of any strategy for the future and should be based on risk assessments and engineering judgement—a key element of this conference. However, the national standards should recognise this.

Private bridges A further extremely important issue is the status and responsibility for all those private bridges that have not been constructed by an Act of Parliament. These include bridges carrying public highways but which are owned by third parties. These third parties can include: Environment

14

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

Protection Agency (NRA arm), privatised utility companies (particularly water authorities and large chambers for telecommunication companies), Land Drainage Board, private quarries and British Coal, Sustrans, district councils and individuals who own one or two bridges carrying the highway. It is likely that there may be as many as 3/4000 such bridges in the UK. The local authority clearly has a duty of care under the Highways Act 1980 to ensure the safe use of the public highway. On the question of funding such work, legislation is, it is believed, silent. This is an issue which needs to be resolved quickly. It will be reasonable that whoever gains the benefit of the bridge should pay the costs of updating the bridge to modern standards of capacity and safety.

Issues Progress of the assessment and strengthening programme The base documents for the 15 year rehabilitation programme are TRMM 2/90 Appendix 2 for government owned structures and Circular Roads 2/91 issued by the Department of Transport in 1990 and 1991 respectively. TRMM 2/90 anticipates strengthening to be completed by 1998 and upgrad­ ing by 2002. Roads 2/91 states 6

7

Local highway

authorities

are also required

he ready to receive the 40 tonne lorries

to ensure that their own roads

by 1999 (and that includes

all

will

privately

owned bridges).

With the present level of funding it appears likely that the trunk road/ motorway target will not be met, and virtually certain that the local road target will not be met. Thus, many alternatives will need to be considered in order properly to manage the use of the highway network by all vehicles, including those measures to be brought in on 1 January 1999. It is hoped (now that agreement has been reached with Railtrack) that all assessments will be completed on local authority bridges in England by the end of 1998. The funding position is not so clear in Scotland and Wales and the break up of all existing bridge units in Welsh counties and Scottish regions, by the creation of unitary authorities, is likely to cause further delay. 2

Comparisons of the earlier assessment standard BE 4 and the B D 2 1 current national assessment codes 8

BE 4: 196 7 had 36 pages of text and some distribution graphs. The assessment work was based on well-known British Standards. The new assessment standards incorporate 6 BDs, 5 BEs, BS 5400 and the ability and necessity to carry out many calculations by means of a computer.

15

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Bridgeguard (which used BE 4) was an exercise where all local authority bridges were assessed between 1968 and 1975. The state of the bridge stock today demonstrates quite clearly that BE 4 worked (although by today's standards i t is a blunt instrument). It is a great pity that we cannot produce a simplified, updated BE 4 which would enable quick assess­ ment and confidence by the assessor.

What is the risk factor when using the national codes? The conference is greatly associated with risk and, in many senses, BE 4 covered that area; some examples are detailed below • •

carriageway widths in excess of 18 ft and not more than two lanes of vehicles shall be considered except at the Engineer's discretion many may comments, i.e. dependent on Engineer judgement

The present national codes lead to the assessment of a local authority bridge on an unclassified road to be the same in the following conditions • • • •

the route to and from a large waste disposal/incinerator site, 500 HGVs/day the route to and from an urban bus station, 20 HGVs/day plus 100 PSVs/day an urban housing estate carrying 10 HGVs/week a rural road accessing a farm only 30 HGVs/year.

Common sense (or engineering judgement) indicates that each bridge is likely to deteriorate at a different rate, that the bridge at most risk is the one carrying most HGVs per year, and yet all those bridges would be assessed the same. Many other variables exist where national codes do not recognise local conditions, e.g. impact on a flat, straight, well-surfaced joint free bridge against impact on a curved, humped, poorly surfaced bridge.

Credibility Many bridges w i l l fail under the national codes in a theoretical rather than in a significant sense. Often they w i l l look in pristine condition and will consis­ tently carry public service vehicles and HGVs. They will be known to the local council member, the bridge owner and, more particularly, his adviser the local authority client manager, who w i l l have to make a decision on what to do. Often the wrong decision (i.e. to weight restrict) will devalue the reputation of the adviser. This issue must be very carefully worked through with both politicians and local people.

16

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

Solutions The following is a possible way ahead regarding the financial crisis, the time problem, the credibility problem and the inflexibility of national codes. It is outlined in Fig. 1. Solutions (h) and (i) (weight restriction and demolish/ rebuild) are seen as the final answer only to be entertained when all other solutions have demonstrably failed. A few comments on each solution is given below. (a) Reduce loaded length: by propping structure when sufficient room exists or by external prestressing.

(b) Review national code for local conditions:

in effect, the subject of this

conference. Allows engineering judgement on local effects and gives a sound method of actual assessed strength. (c) Traffic management: depending on traffic density can reduce width, con­ trol with traffic signals, even provide electronic control using axle load detection pads and remote control. (d) Monitoring: as simple as inspection at regular intervals of condition cracks etc. or as complicated as sophisticated electronic methods of strains etc. under actual loading. (e) Research and development: much research has been carried out over the past three or four years managed and funded (often jointly) by County Surveyors' Society, other local authorities, industry and government 9

Existing bridges

-Assessment pass

Assessment fail

Overcome by

More rigorous assessment fail

> reduce loaded length —• pass

Is the failure theoretical or significant?

• Traffic review management national code for local condition (width restriction, traffic signals, weight controlled system)

> monitoring —•

• Load R & Dtesting Strengthen Weight restriction—unenforceable Demolish/rebuild—soft option

AIM Monitoring, testing, managing, strengthening rather than rebuilding (i) Economic (bridge work) reduces cost (ii) Economic (traffic) reduces traffic delay (iii) Environmental (popular with local) (saves material)

Fig. 1. Possible

solution

17

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

agency. All is of extreme value (e.g. CSS work on masonry/brick parapets). (/) Load testing: the National Steering Committee on Bridge Testing is working towards creating a guideline for all types of highway bridge with support from all bridge owners in the UK. (g) Strengthening: in many cases the sub-structure is adequate, and to repair or strengthen the super-structure is a viable alternative to rebuilding it. Anything which extends the life of a bridge should be deemed strengthening, e.g. water proofing concrete decks. (h) Weight restriction: the enforcement of weight restriction has low priority for the constabulary. It is dangerous to rely on weight restric­ tions to protect weak bridges. (i) Demolish/rebuild: essential in some cases but to many it may appear to be a soft option which does not require any particular engineering judgement. 10

Effect on urban centres of two tier system Any two tier system inherently implies transhipment depots, and, if they are to be sited in environmentally friendly areas, then that implies the building of such depots on countryside in green belt areas. The basis of two tier systems is to reduce large vehicles travelling in urban areas (where com­ merce, business, shops and offices exist). However, that will, in effect, increase the number of HGVs allowed to use the urban roads, increase costs, increase congestion and severely increase pollution. Thus, on economic and environmental grounds such a proposal is negative. It would be much better to restrict journey and delivery times.

Development of whole life costing for design of new and repair of old The true consideration of whole life costing is something that is often discussed or mentioned but rarely acted upon. Lip service is paid to the concept. The Concrete Bridge Development Group seminar on whole life c o s t i n g demonstrated this very clearly. Dr Rigden's paper at that con­ ference demonstrated very clearly the Treasury position with respect to whole life costing and many of the other papers demonstrated clearly what should be done. Until we get an accepted methodology for assessing whole life performance then repair costs will be very high, traffic delay costs and congestion will occur and credibility will diminish. This process has begun in the U S A and the Federal Highway Administration have pro­ duced a g u i d e which shows their approach. Similar systems should be developed in this country; if not, a national strengthening programme will be necessary every 2 0 - 3 0 years or so. But such a programme can, it is believed, be avoided. 10

11

18

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

Conclusions At present, it is clear that most elected members of local authorities and nearly all of the public see bridges as part of the road system which is totally safe, and, furthermore, which is managed by well trained professionals. Those very few collapses which do occur in the UK are seen as a result of natural (or man-made) disaster such as scour, mining subsidence, fire (Mancunian W a y ) or even explosion (Staples Corner). Therefore, the design and maintenance service has a very low profile. Things with a low profile do not attract high profile funding; thus it is essential that all bridge managers seek to go out and tell the public about their work. Accordingly, the way forward must be to encourage the following key factors • • • • • •

greater use of risk analysis and engineering judgement where possible, use of simplified design methods use of whole life costing analysis on all bridgework economic solutions for all proposed work informing people (stakeholders and the general public) of the issues raising the profile of bridge engineering in all its facets.

References 1 . M a l l e t G . P . ( 1 9 8 6 ) B r i d g e m a i n t e n a n c e value

for m o n e y . I n s t i t u t i o n o f H i g h w a y s a n d T r a n s ­

portation w o r k s h o p , 8 April. 2. N a t i o n a l A u d i t Office ( 1 9 9 6 ) . H i g h w a y s A g e n c y — t h e b r i d g e p r o g r a m m e . H M S O , L o n d o n . 3.

L e a d b e a t e r A . D . ( 1 9 9 6 ) . T h e financial

s e t t l e m e n t 1996/97.

Fourth Annual Surveyor Bridges

Conf., 1 5 F e b r u a r y . 4. T h e R o a d Traffic R e g u l a t i o n A c t : 1 9 8 4 . H M S O , L o n d o n . 5. T h e H i g h w a y s A c t : 1 9 8 4 . H M S O , L o n d o n . 6 . T R M M 2 / 9 0 A p p . 2 ( 1 9 9 0 ) . P r o g r a m m e of s t r u c t u r e s m a i n t e n a n c e — 1 5 Year litation

Programme,

Bridge

Rehabi­

bridges and

struc­

D e p a r t m e n t of T r a n s p o r t .

7. C i r c u l a r R o a d s 2 / 9 1 ( 1 9 9 1 ) . A s s e s s m e n t a n d s t r e n g t h e n i n g of highway tures, D e p a r t m e n t of T r a n s p o r t . 8.

BE4

(1967). T h e assessment

o f highway

bridges

for c o n s t r u c t i o n a n d

use

of

vehicles,

Ministry of T r a n s p o r t . 9.

Buro

Happold

Consulting

Engineers, Load

a s s e s s m e n t o f e x i s t i n g b r i d g e stock.

management/risk

assessment—approach

to

Unpublished.

10.

C o n c r e t e B r i d g e s 1 9 9 5 . W h o l e life c o s t i n g s e m i n a r . C o n c r e t e B r i d g e D e v e l o p m e n t

11.

F H A (1992). V a l u e engineering for h i g h w a y s . U S D e p a r t m e n t of T r a n s p o r t a t i o n , W o r k i n g t o n .

Group.

19

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

DBFO contracts and bridge performance considerations J. B e n t - M a r s h a l l ,

Sir William

Halcrow

& Partners

Ltd

Introduction Design, build, finance and operate (DBFO) is a new form of contract, being used increasingly to procure road schemes in the UK. Such contracts have two principal aims •

•

To develop a private sector road operating industry which provides at least an equivalent level of public safety, service and value for money to that which exists throughout the rest of the trunk road network. As part of the Private Finance Initiative, to transfer to the private sector most of the risks of constructing new roads and operating and maintain­ ing new and existing roads for 30 years.

In DBFO contracts, the company is permitted to use innovative standards and methods for building and maintaining bridges and other structures on the project road, subject to compliance with the core client requirements. This form of contract brings into focus the requirements which are usually implicit in the existing codes. It is becoming increasingly important to define these requirements more explicitly in order to deliver the primary objectives in terms of safety, customer service and long term performance. The purpose of this paper is to discuss the current position regarding DBFO contracts in England, and to point out the new issues that such contracts are bringing to the fore. The first four DBFO contracts, which made up tranche 1, have now been awarded and are under way. These are • • • •

A69 Newcastle to Carlisle A419/A417 Swindon to Gloucester A1(M) Alconbury to Peterborough M l - A l Motorway Link, Leeds.

In total these represent £400 million of new construction and a 30 year operation and maintenance (O&M) value of approximately £300 million. The schemes involve construction of 125 new bridges (including nine Category 3 bridges) and their maintenance plus another 130 existing structures on sections of road which w i l l not be improved. The second tranche of projects (1A) comprises the •

A50/A564 Stoke to Derby Link

20

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

• • •

M40 Junctions 1-15 A30/A35 Exeter to Bere Regis A19 Dishforth to Tyne Tunnel.

These projects represent £165 million of new build, £1-1 billion of O&M cost and include 50 new bridges plus more than 600 existing structures including the Tees Viaduct. Tranche 2 DBFO projects, which are expected to be awarded in 1997-98, comprise the • • •

A65/A629/A650 Cumbria to Bradford A21/A26/A27/A259 Weald and Downland A6/A43/A421/A428 South Midlands Network.

The new build element of these projects is approximately £350 million and the 30 year O&M value is around £700 million. A total of 175 new structures will be built and 500 existing structures will become the day-to-day respon­ sibility of the DBFO company (DBFO Co.). It is important to remember that the Secretary of State for Transport remains the Highway Authority for DBFO roads and will be answerable to Parliament to the same degree as for any other trunk road. These first three tranches will eventually contain nearly 1600 bridges and other significant highway structures, 350 of which will have been built by the DBFO Co. This represents a considerable investment, to which the DBFO contract has paid particular attention.

The DBFO contract The DBFO contract is a negotiated contract complying with EC Works Directive (93/37/EEC) for Public Works Contracts. It contains the Agreement (Conditions of Contract) plus a number of technical, procedural and com­ mercial schedules, which have been amended and developed during the negotiations. The principal technical schedules are: • • • • •

Schedule 3: Land Schedule 4: Design and Construction Schedule 5: Quality Assurance Schedule 6: Operation and Maintenance Schedule 17: Communications.

Schedules 4, 5 and 6 contain specific requirements relevant to bridges. The technical requirements in the tender invitation documents are defined as either Core or Illustrative Requirements. Core Requirements are a set of high level principles which define the standards for safety and the level of service to be delivered. These are supplemented by project specific require­ ments, e.g. line orders, CPOs, environmental statements, and undertakings

21

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

given at public inquiry. Core requirements are kept to the minimum and are not negotiable. Examples relating to bridges include loading standards and minimum headroom plus a ban on half-joints and grouted post-tensioned construction (which has been relaxed for tranche 2). Illustrative Requirements define a set of general technical standards which are one way of delivering the core requirements and include design standards (DMRB), specification (SHW), road width and alignment, bridge type and layout etc. Tenderers are encouraged to innovate and propose alter­ natives which provide an equivalent or improved level of service. Where these are accepted during negotiations, they become the Requirements of the contract. Any contract w i l l involve changes, especially one which lasts for 30 years and the DBFO contract allows both parties ample provision for change. The Department can change the Requirements, including the specification and design standards, while the DBFO Co. can propose changes under the review procedure. This allows the Department to accept the change, with or without comments, or reject it on certain specified grounds. These include failure to satisfy Core Requirements, proposing a solution of lower standard than the Requirements or failing to match good industry practice. The contract also contains a number of possible remedies ranging from penalty points for a variety of non-compliances, through warning notices to termination for major events of default or persistent non-compliance.

Design and construction of new bridges Illustrative Requirements for the design and construction of new bridges and other highway structures are contained in Schedule 4 and include • •

• •

Design manual for roads and bridges and the Specification for highway works, both of which have been adopted by all DBFO companies to date. Technical appraisal forms (TAFs) which are very similar to AIPs and are used to define the form of the structure, its design parameters and appearance. Departures from standards already accepted. General arrangement drawings (as part of the TAFs).

Where design work has already been undertaken, the existing designs are made available to tenderers. Tenderers may adopt these designs, with or without modification, or propose alternatives providing they satisfy the Core Requirements. Before construction starts the DBFO Co. must commission the preparation of the detailed design for the new works. This will be prepared by the designer, who must be independent of the DBFO Co., and, in the case of a category 3 structure, checked by an independent checker. The output from

22

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

this design, called the design data, must be submitted to the Department's Agent (DA) together with design and check certificates. The design data will include detailed drawings but not calculations or bar schedules unless specifically requested. The DA can only reject the design data on grounds similar to those stated earlier for Core and Illustrative Requirements. The construction will be supervised by the designer while the Department's Agent has a monitoring role similar to that of a design and build contract. All activities throughout the design, construction and O&M phases are to be carried out under the umbrella of an agreed quality system and the con­ tract lists a number of quality plans which must be accepted before work starts. A robust and workable quality system is a key element of the DBFO contract. Probably the most effective incentive for achieving a high level of perfor­ mance quality is the contract itself. The DBFO Co. must maintain the road for 30 years and hand it back with a specified residual life. It is expected that banks and other credit providers will seek to protect their investment by organising their own monitoring of the DBFO Co.'s design and construc­ tion performance, in addition to that of the Department. It is expected that the DBFO Co. will design the new works to minimise maintenance costs by selecting durable design methods and long life materials. An example of this is the enthusiastic adoption of BD 57 Design for durability. In certain cases existing designs have been modified to incorporate integral abutments, thereby dispensing with some of the less durable elements of traditional bridge design. The search for reduced maintenance and construction costs, however, must not be achieved at the expense of appearance. The standards defined in the Construction Requirements will be particularly important in this respect. The principle of 'equivalence' must be rigorously applied, when considering proposed changes, if DBFO designs are to be aesthetically satisfactory.

Operation and maintenance The contract contains Core O&M Requirements similar to those specified for construction. The Illustrative O&M Requirements are largely based on the department's Trunk Road Maintenance Manual (TRMM), which although not a specification as such, does define current good practice. Schedule 6 of the contract amends and supplements the TRMM. It defines the perfor­ mance standards required for each aspect of O&M while allowing scope for flexibility in delivering a safe, adequately maintained highway. Schedule 6 contains requirements for • •

records of highway structures principal, general and special bridge inspections

23

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

• • •

routine maintenance and repairs assessment strengthening and refurbishment.

The Construction Requirements contained in Schedule 4 are invoked as necessary to set the standards for these activities. The DBFO Co. is responsible for all aspects of O&M and is likely to appoint an operator, usually a firm of consulting engineers with the neces­ sary experience, to manage the process. The Department will appoint the Department's Representative (DR) to monitor performance and compliance with O&M requirements. He will issue penalty points for non-compliance with the O&M requirements. The DBFO Co. is required to adopt and fulfil the 15 year bridge rehabilita­ tion programme, as described in TRMM, and should have made the neces­ sary financial provision in its tender. In some cases, assessments were not available at tender invitation but sufficient experience has been gained in applying the current standards for bidders to judge the likely need for strengthening. In extreme cases there is provision in the contract for compensation in the event of major unforeseeable defects (latent defects). Where a sub-standard bridge is contained within the DBFO project road, any reassessment, risk analysis and reprogrammed strengthening (in accor­ dance with revised standards) would be carried out by the DBFO Co. follow­ ing a Department's Change. If no such change is made and the current requirements of the contract are adhered to, bridges on DBFO roads may be strengthened more quickly than those elsewhere on the network. The contract seeks to encourage the DBFO Co. to improve the efficiency and life span of maintenance and renewal works. The company is likely to seek methods and techniques which reduce the timescale of maintenance operations, or permit more operations to be carried out at night as this would incur lower lane closure charges, which are a feature of the contract. Additionally such measures would help to maximise traffic throughput and hence the shadow toll payments. The DBFO Co. will be seeking maintenance solutions which give the best value for money on a whole life cost basis. New materials and methods will probably be tested but the DBFO Co. is likely to demand a performance warranty from suppliers.

Handback In the last five years of the contract the DR will carry out a series of handback inspections to identify any works necessary to achieve the required residual life of each element. There are provisions in the final years of the contract to retain payments in order to build up an adequate fund in case of default by the DBFO Co. The principal structural elements of bridges must have a residual life of 30 years beyond the end of the contract and

24

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 1. OVERVIEW

it is expected that the industry will have the techniques necessary to accurately assess residual life by the time they are needed.

Risk transfer and management A fundamental principal of the Private Finance Initiative is that risk should be carried by the party best able to manage it. Individual PFI projects are judged on their overall value for money when compared with traditional pro­ curement methods and the degree to which risk is transferred from the public to the private sector. For DBFO projects the risks transferred to the DBFO Co. include • • • • • • •

construction risk, e.g. unforeseen conditions, delays and overruns, price rises, adverse weather conditions O&M risk, e.g. construction and material defects, defects in the existing road delays caused by utility companies, statutory authorities etc. third party risk from adjacent landowners, road users, etc. traffic risk (i.e. if the DBFO Co.'s predicted volumes are not achieved then shadow toll payments will be lower than expected) accidental damage environmental and archaeological issues.

Certain risks will be retained by the Department and these include •

• • •

planning risk (although tranche 2 will seek to transfer some risk to the DBFO Co. for schemes which have not completed their statutory proce­ dures) Department's changes change of law significantly affecting the DBFO Co.'s income or operation certain events of force majeure

Some risks will be shared, with the Department bearing the costs above an agreed threshold, e.g. • • •

protester action latent defects (early contracts only) the introduction of user paid tolls.

Where risks are transferred, the DBFO Co. should have made a judgement about the likely cost of each risk and allowed for it in the bid. The company should be seeking to manage the risks so as to minimise their cost and this will require a high level of foresight and quality management during all design, construction and maintenance operations. Contractors are experi­ enced at managing hard risks, e.g. poor ground conditions, but will also need to fully understand soft risks such as driver behaviour.

25

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

There is an inevitable wish to transfer as much risk as possible from the public to the private sector, but this may be false economy. A private sector organisation must price for uncertainty and if the risk is one which it cannot effectively predict or manage, the price may be unrealistically high. The hidden costs of risk management, e.g. litigation to resolve problems, can also be much greater than the apparent costs. Alternatively, the public sector w i l l be less able to manage or mitigate the effects of transferred risks and will, therefore, have less control over situations that may arise. Only time w i l l tell whether the degree and balance of risk transfer has been correct.

The future There is no doubt that the DBFO contract is here to stay for the foreseeable future. On completion of tranche 2 more than 10% of the trunk road network in England w i l l be managed by DBFO companies and most major schemes are, in future, likely to be procured by this method. The introduction of networks in tranche 2, rather than single links, is another expansion of the philosophy but these projects still contain a sizeable new build element. The performance of DBFO companies w i l l be compared with traditional maintenance agents (County Councils or consulting engineers) and the public w i l l be critical of any obvious differences. One potential difference is that maintenance levels have traditionally been controlled by available funding while the DBFO contract seeks to encourage a 'needs led' whole life maintenance policy. New design and maintenance standards and methods of working will need to be introduced under the Department's change procedure. Some of the initiatives discussed at the symposium are particularly important to DBFO contracts, e.g. bridge specific assessment loading, risk analysis and whole life performance evaluation and management. Whatever the future brings, I am convinced that the responsibility of a 30 year maintenance liability will produce more durable and maintenance-free bridges than either traditional procurement or design and build contracts.

Acknowledgement Halcrow acts as technical adviser to the Highways Agency, who are respon­ sible for developing and managing the DBFO procurement process. The Highways Agency has given consent for this paper to be published but any opinions expressed are those of the author and not necessarily the Agency.

26

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Part 2. Safety concepts and codes

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Safety concepts and practical implications P. C . D a s , Highways Agency,

London

Introduction A bridge is considered to be unsafe if pieces of concrete from it start falling on to vehicles and road users below, or when serious damage or deteriora­ tion is found in the main structural members. Moreover, a bridge is also con­ sidered to be unsafe if a mainly calculation based assessment indicates that it may be structurally inadequate. In the first case, the safety of the bridge is as perceived and the risks are real; in the second case, safety is essentially conceptual. When dealing with conceptual safety, justification for the safety targets used in design and assessment, and the practical implications for an existing bridge not meeting these requirements, are not very clear. The purpose of this paper is to consider these implications and to discuss the principles which form the basis of much of the design and assessment rules for structures. In many of the papers in this book extensive references are made to relia­ bility analysis and reliability index. Although practising bridge engineers are broadly aware of the significance of such terms, it may be helpful to consider their definitions, albeit in a schematic form, so that the contents of these papers can be appreciated more fully. Opportunity therefore has been taken in this paper to define these terms which nowadays are commonly used in the formal definition of structural safety.

Evaluation of structural safety Safety levels in the earlier structural codes (pre-1970) were defined in terms of overall factors of safety. In permissible stress designs the factors of safety were those between the permissible stresses and the ultimate stress capacity of the materials concerned. In the code for earth retaining structures CP2, however, overall factors of safety were recommended in explicit forms. Most modern structural codes (e.g. BS 5400) nowadays ensure safety levels through recommended partial safety factors. Whenever practicable, these factors have been derived by establishing the safety levels implicit in the earlier codes using reliability analysis, and then by calibrating the partial safety factors in the new codes, again through reliability analysis. The safety levels used for such comparisons are described in terms of probability of failure (p ) or reliability index ((5). f

29

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Fig. 1. Vehicle load data

Reliability

analysis

Reliability analysis techniques are still in many ways a developing science and can take various forms. However, the following principles are generally followed in such analyses. Firstly, a failure criterion is selected. The commonly used form is that R, the structural resistance at any point (for example the mid-span moment of resistance), must be greater than S, the corresponding load effect (for example the applied bending moment at mid-span). This is expressed as g = H- S

(1)

where g < 0 represents failure. Since R and S are subject to various uncertainty sources, they are random variables and can be described in terms of probability density functions (PDFs) based on available information and data (e.g. statistical data such as those in Fig.l). A PDF is a probability distribution such that the total area under it is unity. The PDFs for R and S are then formed, which are shown schematically in Fig. 2. Furthermore, since R and S are random variables, g is also a random variable and the probability of failure corresponds to the event g < 0. The objective of reliability analysis is to calculate this probability. For certain cases, of which the most common is to assume that R and S are normally distributed independent variables, it is easy to obtain the PDF of g. In this case g is also normally distributed and the probability of failure, p , is obtained from f

Pf

=

* ( - / ? )

(2)

30

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

Load effect and resistance

Fig. 2. PDFs of R and S

where $ is the standard normal distribution function and (3 is the reliability index, which in some cases may be defined as (3

= m/cr

(3)

where m and a are the mean and standard deviation of g. This is schemati­ cally shown in Fig. 3. The same procedure can be shown in the normalised R — S space. Normalised in this case means expressing the two dimensions in terms of the standard deviations of R and S. Since variable loads such as traffic loading and wind loading are measured in terms of return periods, both j3 and p are expressed in terms of time intervals, e.g. a pf of 0.6 x 1 0 ~ in one year. The basic aim of design rules is to recommend the design values of R and S so that the relative positions of the PDFs of R and S will provide the required target reliability relating to a given failure condition. Since, for any structure, there may be a number of different limit states, it is convenient to recommend characteristic or nominal loads and strengths with various partial safety f

6

Unsafe

Safe

9=0

K

9 I3c

(

Fig. 3. PDF of g

31

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

factors to arrive at the required design values for the corresponding limit states. In very general terms, the implications of the above procedure are (a) the farther the PDFs of R and S in Fig. 2 are from each other, the less is the likelihood of failure (b) if load effects increase generally, i.e. S moves to the right, or, if structural resistance deteriorates, i.e. R moves to the left, the structure becomes less safe, resulting in a reduction of (3. The above, in short, is the general procedure used for the reliability analysis of any structure, including bridges, whether for design or for assessment.

Safety levels as operational tools Structural codes invariably have to recommend that certain level of safety is acceptable, below which a structure may be deemed to be unsafe. It may be implicit in a bridge code, for example, that such a level may be a p of 10~ per annum. There is no special significance in such a figure, because two similar bridges with p of 10~ and 1 0 per annum may not be dramatically different, although one could pass the assessment, while the other would not. The real significance of the acceptable p of 1 0 per annum is that it may be implicit in the codes and rules which have been in use for many years, and which have been found to be satisfactory. Hence, such a target may be considered as an operational tool proven through long term use in the past. In order to illustrate the significance of an operational tool, let us assume that the weather forecast for a particular day says that there is a 30% prob­ ability of rain that day. This figure is meaningless in a practical sense; it does not mean that it will rain for 30% of the day, nor that the rainfall will be 30% of a particular level. An individual going out for the day, if he considers taking an umbrella with him, has only one of two options to choose from—to take it or not to take it. However, he can use such weather forecasts by choosing an operational target, say a 2 0 % probability of rain, above which he will take an umbrella. Only time will tell whether this target gives him a satisfactory level of protection against rain or not. If not, he will need to try out another target figure. The nature of the safety levels implicit in structural design and assessment rules is no different. These levels have been arrived at through past experience of both successes and failures, and that is essentially their justification. This is also the reason why, when new rules are drafted, drastic changes from past practice are not considered to be advisable.

6

f

5

- 7

f

- 6

f

32

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

Significance of assessment failure All assessment failed bridges are not expected to actually fail under a critical ultimate load condition. Assessments are essentially assessment of the likelihood of a bridge being safe, or unsafe, if and when a critical load condition occurs. Referring to Fig. 3, the real value of the safety margin g = R — S cannot be known without testing the structure to its ultimate load condition. A bridge which barely passes assessment may in reality prove to be inadequate for the ultimate condition, although the likelihood is very small. Alternatively, bridges which just fail assessment, may often be found to be adequate if tested to the ultimate limit state load, because their probability of failure is also very small. Thus, it can be said that if all assessment 'failed' bridges are left without any remedial action, only a few of these are in reality likely to fail during a given period of time, although more will fail with time due to deterioration. The reason why, however precisely the assessments are carried out, a significantly large number of bridges fail assessment is that unfortunately, the rare situations where extreme loads may coincide with weaknesses in bridges to result in actual failure cannot be identified or predicted precisely. Otherwise only those few bridges where this is going to occur would need to be strengthened, and the rest could be considered as having passed assessment. In short, if the present operating level of safety is lowered and large num­ bers of assessment failed bridges are left without any code recommended remedial measures, there will be a real possibility that a few of the bridges, and not necessarily the weakest ones, will fail in service.

Live load return periods Randomly occurring loads such as traffic and wind load are expressed in terms of return periods, such as a 100 year or 1000 year maximum. This is essentially a convenient way of denoting relative levels of extreme loads. A 100 year return wind load does not mean that it will only occur at the end of 100 years, in fact it can occur at any time. It however means that it has a high probability of occurring at least once in any 100 years. Moreover, it is less likely to occur more than once in such a period, but it will not be safe to assume that it is completely unlikely to occur during a shorter period. Such time related implications are meaningless as far as traffic load of return periods greater than the foreseeable future are concerned, since vehicular traffic is unlikely to remain in the present form beyond a relatively short time horizon. Thus, a traffic load level of 100 year return period is no more than a convenient way of denoting extreme levels.

33

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

To summarise, just because live loads of very long return periods are used in the design and assessment calculations, it does not mean that an assess­ ment failed bridge will be completely safe for a limited period and can be left without effective interim action.

Ductile versus brittle failure It is commonly believed that when considering interim actions for sub­ standard bridges, one ought to be more careful regarding a bridge that is weak in shear (potentially brittle failure) than a bridge which is weak in bend­ ing (potentially ductile failure). This notion originates from controlled tests of bridges or beams to failure, where the loads are applied in increments, and bending failure is usually preceded by extensive deformation and, in concrete bridges, by extensive cracking. Nevertheless, if the full critical load is applied instantaneously, as will be the case if a critical heavy vehicle appears on the bridge, both types of failure may take place fairly rapidly. However, the limit of bending failure is set in the codes at such a level that at the ultimate limit state (ULS) the bridge may not collapse, but will still have extensive damage. Thus, loss of life from a bending failure at the ULS load is less likely than from a shear failure, but the bridge may still have to be closed until repairs are carried out. Although, in general, monitoring may not provide adequate safeguard against either shear or bending failure if the bridge is assessed to be inade­ quate in respect of extreme vehicle loading which may occur without warn­ ing, it is believed by many bridge engineers that monitoring may provide valuable information regarding the response of the structure to real traffic loading, and also a degree of confidence in its adequacy under day-to-day loading.

Conclusions The paper discusses the evaluation and practical implications of conceptual safety as far as bridges are concerned. It notes that if a bridge is assessed to be inadequate, there will be some risk that a critical load may occur even within a short period, and that monitoring alone may not necessarily prevent failure. The paper also concludes that the safety targets implicit in the design and assessment codes are essentially operating tools, which have been proven to be satisfactory through long term use in practice.

Acknowledgement The author is grateful for comments and suggestions received from the members of a Highways Agency steering committee at the draft stage,

34

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

most notably from Professor M. J. Baker of Aberdeen University, Mr M. Chubb of W . S. Atkins, Dr M. K. Chryssanthopoulos of Imperial College, Dr A. R. Flint of Flint & Neill Partnership, Professor D. Frangopol of Colorado University, Dr J. Menzies of J. Menzies Consultants, Dr C. R. Middleton of Cambridge University, Professor A. Nowak of Michigan University, and Mr J. Wallbank of G. Maunsell and Partners.

35

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Bridge failures, hazards and societal risk J. B. Menzies, Engineering

Consultant,

John

B.

Menzies

Introduction This paper is based on a study, undertaken for the Highways Agency, of the safety of highway bridges in relation to the risk to life associated with bridge collapse among the existing population of approximately 100 000 bridges in the United Kingdom and with particular reference to normal bridges of short span, say 20 m or less. The objective of the study is to determine the mini­ mum socially acceptable safety level for bridges. The study is part of a wider project to determine whole life performance targets for bridges. A bridge collapse is a rare event in the United Kingdom and loss of life associated with a collapse is an even rarer event. For normal highway bridges of short span, such rare events do not generally trigger widespread or prolonged public concern. It might be concluded, therefore, that highway bridges have an acceptable level of safety. The first concern of the engineer when considering the safety of an exist­ ing bridge, is to ensure that it will not collapse or at least that collapse is very unlikely. This requirement is generally met by providing maintenance and repair to keep and, where necessary, restore structural adequacy to a level implied by good practice as manifest through standards and codes of prac­ tice for bridge design and assessment. It is well understood among engineers that absolute safety is unattainable and inevitably there are risks of collapse and loss of life associated with any bridge. The risks are, however, not quantified in practice. The relative mag­ nitude of these risks compared to other risks in everyday life may provide indications of the answers to important questions facing engineers How unlikely should bridge the maximum Are acceptable in

socially

collapse

acceptable

levels of these risks

and associated

risk

to life relating

substantially

loss of life be, i.e. what is to bridge

different

collapse?

from those

existing

practice?

Current resources for the maintenance, repair and strengthening/replace­ ment of normal highway bridges in the United Kingdom provide a high level of structural safety. However, various options are available in all these activities, including the 'do minimum' option. In order to form a main­ tenance strategy comprising a rationally determined mix of such options for the whole bridge population, it is essential to know which items of work are necessary for maintaining minimum acceptable safety levels, and which are justified by economic reasoning alone. Examination of information on

36

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

acceptable levels of risk to life and the incidence of bridge collapses and loss of life is discussed therefore with the aim of identifying acceptable bridge safety targets.

Risks to life in perspective Throughout life diseases cause more deaths than accidents. Road travel presents a significant risk at all stages of life and it is only after middle age that diseases present a more significant risk. Comparisons between the levels of risk to individuals for different activities are usually made on the basis of the fatal accident rate (FAR). The FAR is the risk of death per 100 million hours of exposure to the activity or, alternatively, per hour of activity per 100 million people exposed. In the United Kingdom there are about 4000 deaths each year on the roads in a population of 56 million which equates to a FAR of about 20. Risks in terms of the FAR involved in different forms of highway travel are approximately • • • • •

travel by motorcycle: 300 travel by pedal cycle: 60 walking beside a road: 20 travel by car: 15 travel by bus: 1

In comparison the background FAR at home is about 1 and the FAR relating to building collapse is about 0-002. The FAR at home of 1 is suggested here as a suitable benchmark for considering individual risks relating to bridge collapse. The collapse of a normal bridge (as opposed to a major large span one) is likely to be considered by the public from a safety point of view similarly to a road traffic accident. A collapse might result in a multi-fatality traffic acci­ dent. In terms of the FAR the risk of being killed in a multi-fatality accident generally (all causes) is minute compared to daily risks on the roads.

Acceptable levels of risk Perception and acceptability of risk are affected by many factors concerning the nature of the hazards, the voluntary or involuntary nature of exposure to risks, the possible consequences of undesirable outcomes and the benefits associated with the risks. Acceptable risk levels cannot be found solely by analysis because they depend fundamentally on value judgements. For this reason decisions on acceptability must take account of societal values. There is generally a greater expectation that public places should be safer than private ones and that places seen as refuges, e.g. home, hospitals, should be very safe. For most users of normal highways bridges, perhaps

37

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

particularly motorists, the safety and integrity of the bridge they are passing over is not in question and the fact that they are passing over a bridge may not even be noticed. Although collapses of normal bridges do occur which have potential to cause loss of life, they are rare and are not seen as major disasters. Public aversion to risk is related to the number of people involved and the circumstances of death; the greater the loss of life in any one event, the greater generally is the public alarm. The acceptable level of risk of a bridge collapse causing multiple loss of life may be expected to be very much smaller (probably two or three orders of magnitude) than for a bridge collapse where little or no actual loss occurs. These general factors affecting acceptability of risks to life suggest that acceptable risk to life and social reaction to bridge collapse depends on • • •

the size of the bridge and its public profile the class of road carried the cause of collapse. 1

Since CIRIA Report 63 was published in 1977 there has been substantial progress in defining the principles for determining acceptable risk. Three approaches are discussed below •

Comparisons of FAR. The safety of structures supporting roads is largely taken for granted by the public. The collapse of a bridge as vehicles cross it and resultant fatalities of vehicle occupants would be perceived more as an involuntary activity than a voluntary one. Com­ parisons of the FAR for a range of voluntary and involuntary activities suggest that an upper limit for an acceptable FAR for 'bridge collapse while travelling over or under it' would be less than the FAR of 15 for travel by car and more than the background FAR at home which is about 1. The FAR statistics for all types of accident suggest that a FAR of about 2 would be an acceptable value relating to bridge collapse. These comparisons point overall to a FAR of 2, which corresponds to an annual risk of about 20 x 10~ , as acceptable. Subjective attitudes. Attitudes to reliability and their relationship to various degrees of voluntary or involuntary exposure to risk have been proposed elsewhere. These criteria (which have been found to be generally confirmed in the literature) suggest, on the basis that the risk of loss of life caused by bridge collapse is an involuntary one, that the acceptable probability for such an event is in the region of 1 in 10 to 1 in 10 annually. It seems unlikely that a probability of 1 in 10 would be acceptable. From review of the criteria, a probability of 1 in 10 annually is deduced to be an acceptable value. Tolerability of risk. The primary principle laid down in the HSW Act is that the employer must do whatever is reasonably practicable to reduce 6

•

7

6

5

6

•

38

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

risk (the ALARP principle). This principle, first applied to hazards such as radioactive materials and asbestos, has recently been applied in deriving the strategic objectives for safety in the railway industry. To determine criteria for acceptance, comparisons of risk from a wide range of natural and man-made activities have been made. The approach used is first to fix a level of risk which is regarded as tolerable and must not be exceeded. This risk is regarded as being so small that no further precaution is necessary. At the other extreme an intolerable level is identified where the given risk is so great or the out­ come so unacceptable that it must be refused altogether. If the risk falls between these two states, then the law requires that it must be reduced to a level which is 'as low as reasonably practicable' provided the cost/ effort is not grossly disproportionate to the benefits. For the railway industry this approach has led to an acceptable level of 1 in 1 0 per annum fatality risk being set for all employees, passengers or public. The industry now has milestone targets to reduce existing risks down to the acceptable level. In addition effort is being made to identify actions which might reasonably be taken to reduce potential for multifatality accidents. By analogy with the targets set by the railway industry, an acceptable level of risk of death associated with highway bridge collapse is deduced to be 1 in 1 0 per annum. 6

6

Considering the indications from the above approaches overall, it is proposed that a maximum socially acceptable annual risk of loss of life associated with bridge collapse would be 1 in 1 0 (single life) and 1 in 1 0 (many lives). 6

7

The incidence of bridge collapse in the United Kingdom The stock of highway bridges in the United Kingdom is approximately 100 000 and the stock of bridges owned by Railtrack PLC comprises some 25 000 railway under/bridges, 6500 public highway bridges and about 11 500 private road bridges and footbridges. Collapses of bridges in this stock over recent decades have been rare and available information is incom­ plete and insufficient to enable failure rates and risk to life to be estimated statistically. In the absence of statistical estimates, available reports on all types of bridge failure suggest the incidence of bridge collapses in the United King­ dom and associated loss of life may be assumed for the purposes of consid­ ering acceptability of risk to life and safety targets as follows •

The failure rate of UK bridges from all causes is in the region of one collapse every 1-2 years. Experience worldwide including the United Kingdom indicates that more than half of all collapses are due to

39

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

•

accidental impact or scour/flooding. The collapse rate due to structural deficiency or overloading for the total stock of highway bridges is about one collapse every five years. Loss of life associated with bridge collapse in the United Kingdom is rare. No reports of cases relating to structural deficiency or overloading have been found but it is believed that several lives were lost in recent decades in two collapses, one a railway bridge collapse caused by scour and another a highway bridge collapse caused by vehicle impact. From the limited data it is reasonable to assume loss of life associated with highway bridge collapse may be in the region of one life lost every ten years, i.e. one life lost on average every two collapses.

Safety targets The assumed rate of one collapse every five years for a population of 100 000 highway bridges implies that 1 in 5000 will collapse in a lifetime of 100 years, i.e. an annual risk of 1 in 5 x 1 0 . If, on average, one in two of the collapses causes a fatality, then the FAR for the travelling population in the United Kingdom (taken as 25 x 1 0 travelling on average one hour a day) is 0-1. This is less than the background FAR at home (about 1). It equates to a per annum fatality risk for the total UK population of about 0*2 x 10~ , i.e. 1 in 5 x 1 0 . This is a smaller risk than the proposed maximum socially acceptable value of 1 in 1 0 per annum. This comparison suggests the current risk to life associated with highway bridge collapse is inside the acceptable region and, therefore, less than that which would give rise to public concern. It may be concluded, therefore, that normal UK highway bridges (i.e. those with short spans comprising the major proportion of the stock of about 100 000) are currently safe in relation to the acceptable risk to life of 1 in 1 0 per annum. However, it^ would not be advisable to deliberately lower the safety levels implicit in current bridge assessment rules, or to reduce maintenance effort, because the difference between the two figures, i.e. 0-2 x 1 0 a n d 1 x 10~ is very small and they relate to the average probability for the bridge population as a whole. The actual probability for individual bridges will vary widely depending on bridge type, location, lane arrangements and condition. For the record, safety targets derived in this study can be expressed as follows 5

6

8

8

6

6

_8

•

6

Annual maximum socially acceptable risk of accidental death to members of the public associated with normal highway bridge collapse: 1 in 1 0 Annual risk of collapse of bridges of span less than 20 m due to struc­ tural deficiency or overloading: 1 in 5 x 1 0 6

•

5

40

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

It should be noted that these targets relate to the normal highway bridge population as a whole. In addition they do not relate to long span bridges where more stringent requirements may be appropriate. Although the available information is limited and an element of judgement has been introduced to establish the safety targets, it is suggested that they provide a suitable basis for identifying items of bridge maintenance, repair and strengthening/replacement work needed to preserve acceptable safety. Application of the safety targets is envisaged to include the use of priorities based on the importance, from a safety point of view, of individual bridges and structural elements. Greatest priority for use of resources should be given to those bridges in the highway network carrying much traffic and whose collapse would cause the greatest danger and disruption to the traffic network, and to those bridges and elements which are likely, on failure, to collapse suddenly without warning. These general factors to be taken into account in setting priorities are already recognised to some extent in the priorities used for committing resources to bridge assessment and maintenance. A methodology for linking the priorities to the proposed safety targets is being developed,

which

together with the criteria for work justified only by economic reasoning, will enable a more cost effective deployment of resources. The resulting whole life performance based assessment rules are dis­ cussed in Part 5. The safety target will form an important criteria within the rules and the proposed target life-time reliability profiles

(developed

using time-dependent reliability analysis taking account of deterioration processes and management activities) will provide a basis for assessing whole life bridge management options.

Conclusion The present risk of loss of life associated with bridge collapses in the United Kingdom is in the socially acceptable region. This implies that current design, assessment and maintenance procedures are generally adequate and should be kept at their present levels.

Reference 1.

CIRIA (1977).

Rationalisation of safety and serviceability factors in structural codes.

Report

63. Construction Industry Research a n d Information Association, London.

41

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Application of bridge reliability analysis to design and assessment codes A. S. N o w a k ,

Professor

of Civil

Engineering,

University

of Michigan,

USA

Introduction This paper presents the calibration of new load and resistance factor design (LRFD) bridge codes in the United S t a t e s and C a n a d a , based on a probabil­ istic approach. The major steps include selection of representative struc­ tures, calculation of reliability for the selected bridges, selection of the target reliability index and calculation of load and resistance factors. Load and resistance factors are derived so that the reliability of bridges designed using the proposed provisions will be at the predefined target level. 1

2

The load models are developed using the available statistical data, surveys and other observations. Load components are treated as random variables. Live load covers a range of forces produced by vehicles moving on the bridge. For multi-lane bridges, the maximum load effect is determined by simulations. The dynamic load is a function of three major parameters: road surface roughness, bridge dynamics (frequency of vibration) and vehicle dynamics (suspension system). The derivations are based on the numerical simulations. The capacity of a bridge depends on the resistance of its components and connections. Structural performance is measured in terms of the reliability index. The reliability indices are calculated for girder bridges, including non-composite steel, composite steel, reinforced concrete and prestressed concrete girders. The results show a considerable degree of variation. The calculated reliability indices served as a basis for the selection of the target reliability index.

Code calibration procedure 3

4

The calibration procedure was based on Nowak et al. ' The work on the new bridge design code was formulated including the following steps (a)

Selection of representative bridges. Representative structures were selected from various geographical regions of the United States (AASHTO) and the Province of Ontario (OHBDC). These structures cover materials, types and spans which are characteristic for the region. Emphasis is placed on current and future trends, rather than

42

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

very old bridges. For each selected bridge, load effects (moments, shears, tensions and compressions) are calculated for various components. Load carrying capacities are also evaluated. (b) Establishing the statistical database for load and resistance parameters. The available data on load components, including results of surveys and other measurements, is gathered. Truck survey and weigh-in-motion (WIM) data are used for modelling live load. There is little field data available for dynamic load, therefore a numerical procedure is developed for simulation of the dynamic bridge behaviour. Statistical data for resis­ tance include material tests, component tests and field measurements. Numerical procedures are developed for simulation of behaviour of large structural components and systems. (c)

Development of load and resistance models. Loads and resistance are treated as random variables. Their variation is described by cumulative distribution functions (CDF) and correlations. For loads, the CDFs are derived using the available statistical data base (step (b)). Live load models include multiple presence of trucks in one lane and in adjacent lanes. Multi-lane reduction factors are calculated for wider bridges. Dynamic load is modelled for single trucks and two trucks side-byside. Resistance models are developed for girder bridges. The variation of the ultimate strength is determined by simulations. System reliability methods are used to quantify the degree of redundancy.

(d)

Development of the reliability analysis procedure. Structural perfor­ mance is measured in terms of the reliability, or probability of failure. Limit states are defined as mathematical formulas describing the state (safe or failure). Reliability is measured in terms of the reliability index, (3. Reliability index is calculated using an iterative procedure. The developed load and resistance models (step (c)) are part of the reliability analysis procedure. Selection of the target reliability index. Reliability indices are calculated for a wide spectrum of bridges designed according to the previous editions of A A S H T O and OHBDC. The performance of existing bridges is evaluated to determine whether their reliability level is ade­ quate. The target reliability index, /3 , is selected to provide a consistent and uniform safety margin for all structures. Calculation of load and resistance factors. Load factors, 7 , are calculated so that the factored load has a predetermined probability of being exceeded. Resistance factors, 0, are calculated so that the structural reliability is close to the target value, /3 .

(e)

5

6

T

if]

T

Load and resistance models Load and resistance parameters are random variables. For steel girder bridges (non-composite and composite), reinforced concrete T-beams, and

43

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

prestressed concrete girder bridges (pretensioned), the statistical models of resistance were developed in refs 7-12. It was determined, that the bias factor (ratio of mean to nominal) for dead load is A = 1-03-1 05, and coefficient of variation is V = 0 08-0*10. For live load, depending on span length, for AASHTO A = 1 - 6 - 2 1 , for OHBDC A = 1-0-1-25, and V = 0-12. The nominal live load is represented by the HS20 truck (AASHTO 1992) and OHBDC truck (1983). HS20 loading consists of either three axles: 35 kN, 142 kN and 142 kN, spaced at 4-3 m, or a uni­ formly distributed lane load of 9*3 kN/m with a moving concentrated force of 80 kN. In the new LRFD AASHTO code, live load is a combination of HS20 truck and a uniformly distributed load of 9-3 kN/m. Therefore, the bias factor for live load is A = 1-25-1-35. OHBDC truck is a five axle vehicle: 6 0 - 1 4 0 - 1 4 0 - 2 0 0 - 1 6 0 kN, or a combination of 70% truck plus a uniformly distributed load of 10 kN/m. To make the bias factor more uniform, in OHBDC the design truck has tandem axles increased to 160 kN each (instead of 140 kN). The corresponding bias factor is A = 0-95-1-10. Dynamic load associated with an extreme value of truck load is about 0 1 0 - 0 1 5 of the static portion of live load, with V = 0-80. For a combined static and dynamic live load V = 0.18. Design dynamic load in AASHTO LRFD is specified as 3 3 % of the truck load effect (with zero assigned to the uniform load). In OHBDC, dynamic load is assumed equal to 0*25 of static live load, except for very short spans governed by a single axle or a tandem. The basic random variables considered in development of resistance models are dimensions, concrete compressive strength, and properties of structural steel, prestressing and non-prestressing strands. The parameters for moment carrying capacity are A = 1-12 and V = 0-10, for non-composite and composite steel girders; A = 1*14 and V = 0-13, for reinforced concrete T-beams; and A = 1-05 and V = 0-075, for prestressed concrete AASHTOtype girders. For shear capacity the parameters are A = 1-14 and V = 0-105 for steel girders; A = 1*12 and V = 0*155 for reinforced concrete T-beams; and A = 1*15 and V = 0*14 for prestressed concrete AASHTOtype girders. 1

2

2

1

2

Reliability analysis procedure Reliability indices, /?, are calculated using a specially developed computer procedure based on the first order reliability method. The available reliabil­ ity methods are reviewed in several textbooks. ' The methods vary with regard to accuracy, required input data, computational effort and special features (time-variance). In some cases, a considerable advantage can be gained by use of the system reliability methods. The structure is considered as a system of components. In the traditional reliability analysis, the analysis is performed for individual components. Systems approach allows to 13

14

44

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

quantify the redundancy and complexity of the structure. The new AASHTO LRFD and OHBDC codes are based on element reliability. However, system reliability methods are used to verify the selection of redundancy factors. Structural performance is measured in terms of the reliability index f3. 1

2

13

Reliability

analysis for AASHTO

(1992)

To develop a reference spectrum of the reliability indices, (3, they were calcu­ lated for* girders designed using the AASHTO (1992) and OHBDC (1983) codes. In AASHTO (1992), the basic design requirement is expressed in terms of moments or shears (load factor design) l-3D + 2-17(L + I) < cbR

(1)

where D, L and I are moments (or shears) due to dead load, live load and impact, R is the moment (or shear) carrying capacity, and cb is the resistance factor. Values of the resistance factor are cb — 1*00 for moment and shear in steel girders; cb — 0*90 and 0-85 for moment and shear in reinforced concrete T-beams, respectively; cb — 1*00 and 0-90 for moment and shear in prestressed concrete AASHTO-type girders, respectively. In OHBDC (1983), the basic design requirement is l - l D + 1-2D + 1-5D + 1-4(L + I) < cbR a

2

(2)

3

where D is the dead load moment (or shear) due to factory-made compon­ ents; D is the dead load moment due to cast-in-place components; D is the dead load moment due to asphalt; L and I are moments (or shears) due to live load and impact; R is the moment (or shear) carrying capacity, and cb is the resistance factor. Values of the resistance factor are specified for material rather than components, and cb = 0-90 for moment and shear in steel; cb = 0-70 in concrete in composite steel girders; cb = 0-85 for steel rebars and prestressing steel; cb = 0-75 for shear capacity of rebars. For AASHTO (1992), the results of calculations show a considerable variation in reliability indices depending on limit state and span length, from about 2 for short span (10 m) and short girder spacing (1.2 m) to over 4 for larger spans and girder spacing. The target reliability index was selected = 3-5. For OHBDC (1983), reliability indices vary from 3 for short span (20 m) to 4 for spans of 4 0 - 6 0 m, for steel girders and reinforced concrete T-beams. For prestressed concrete girders, (3 is about 5. The same target reliability index was selected (3 — 3*5. t

2

3

T

New load and resistance

factors 1

The results of the reliability analysis for the current AASHTO served as a basis for the development of more rational design criteria for the considered

45

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

1

girders. The load factors developed for the LRFD AASHTO are 1-25D + 1-50D + 1-75(L + I) < (j)R A

(3)

n

where D is the deal load; D is the dead load due to asphalt wearing surface; L is the live load (static); I is the dynamic load; R is the resistance (load carrying capacity), and (f> is the resistance factor. In the selection of resistance factors, the acceptance criterion is closeness to the target value of the reliability index, /? . The recommended resistance factors are (j) = 1 0 0 for moment and shear in steel girders; (p = 0*90 for moment and shear in reinforced concrete T-beams; (j) = 1 0 0 and 0-90 for moment and shear in prestressed concrete AASHTO-type girders, respec­ tively. Reliability indices calculated for bridges designed using the new LRFD AASHTO are close to the target value of 3-5 for all materials and spans. The calculated load and resistance factors produce a uniform spectrum of reliability indices. For comparison, the ratio of the required load carrying capacity by the new LRFD AASHTO and the AASHTO varies from 0-9 to 1-2. For OHBDC, the load factors were not changed from 1983 edition, but recommended resistance factor for prestressing steel is (j) = 0-95. The result­ ing reliability indices are about 3-5. A

n

T

1

1

8

2

6

5

o o •

0

1

o o

o

o

o

•

2

•

Steel

O

Reinforced Concrete

O

Prestressed Concrete

3

4

Reliability Index for AASHTO (HS20)

Fig. 1. Reliability indices for AASHTO (1992) and LRFD AASHTO

(1994)—Moment.

46

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

0 0

••! '•o\*>Skfe& ° !

•

Steel

0

Reinforced Concrete

O

Prestressed Concrete

3

i

i 2

3

Reliability Index for AASHTO (HS20)

Fig. 2. Reliability indices for AASHTO (1992) and LRFD AASHTO

2

(1994)—Shear.

•

Steel

O

Reinforced Concrete

O

Prestressed Concrete

3

Reliability Index for OHBDC (1983)

Fig. 3. Reliability indices for OHBDC (1983) and OHBDC

(1991)—Moment.

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

5

,

4

^—

•

:

f

%!

_

•

i 0

1

isft.

Steel

O

Reinforced Concrete

O

Prestressed Concrete

i 2 3 Reliability Index for OHBDC (1983)

4

5

Fig. 4. Reliability indices for OHBDC (1983) and OHBDC (1991)—Shear.

Conclusions The calculated load and resistance factors in the calibration part of the paper provide a rational basis for the design of bridges. They also provide a basis for comparison of different materials and structural types. The study has several important implications. The calculated load and resistance factors provide a uniform safety level for various bridges. The statistical analysis of load and resistance models served as a basis for the development of more rational design criteria. Bridge components designed using the proposed AASHTO LRFD and OHBDC have reliability index from 3*5 to 4-0, as shown in Figs 1-4. 1

2

Acknowledgments AASHTO LRFD was carried out in conjunction with the National Coopera­ tive Highway Research Program (NCHRP) Project 12-33. Research related to OHBDC was sponsored by the Ontario Ministry of Transportation. The opinions and conclusions expressed or implied in the paper are those of the writers and are nqt necessarily those of the sponsoring organizations.

48

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 2. SAFETY CONCEPTS AND CODES

References 1.

AASHTO

(1994). L R F D bridge design specifications.

w a y a n d T r a n s p o r t a t i o n Officials, W a s h i n g t o n , 2.

OHBDC

(1991).

American Association

Ontario highway bridge design code.

view, Ontario, C a n a d a , 3rd

of State

High­

DC. Ministry

of Transportation,

Downs-

edn.

3.

N o w a k A . S . ( 1 9 9 5 ) . C a l i b r a t i o n o f L R F D b r i d g e c o d e . A S C E J . S t r u c t . E n g n g , 121(8), 1 2 4 5 -

4.

N o w a k A . S. a n d

1251. G r o u n i H. H. (1994). C a l i b r a t i o n o f the O H B D C - 1 9 9 1 .

Canadian

J. Civ.

E n g n g , 21, 2 5 - 3 5 . 5.

AASHTO

(1992).

Standard specifications for highway bridges.

S t a t e H i g h w a y a n d T r a n s p o r t a t i o n Officials, W a s h i n g t o n , 6.

OHBDC

(1983).

Ontario highway bridge

view, Ontario, C a n a d a , 2 n d

design

code.

American

Association

of

DC.

Ministry

of Transportation,

Downs-

edn.

Safety, 13(1,2), 5 3 - 6 6 . Canadian ]. Civ. E n g n g , 21, 3 6 - 4 9 .

7. N o w a k A . S . ( 1 9 9 3 ) . L i v e l o a d m o d e l f o r h i g h w a y b r i d g e s . / . S t r u c t . 8.

N o w a k A . S. ( 1 9 9 4 ) . L o a d m o d e l f o r b r i d g e d e s i g n c o d e .

9.

N o w a k A . S . a n d H o n g Y - K . ( 1 9 9 1 ) . B r i d g e l i v e l o a d m o d e l s . A S C E J . S t r u c t . E n g n g , 117(9), 2757-2767.

10.

H w a n g E - S . a n d N o w a k A . S. ( 1 9 9 1 ) . S i m u l a t i o n o f d y n a m i c l o a d for b r i d g e s . A S C E J. S t r u c t .

11.

N o w a k A . S . e t al ( 1 9 9 4 ) . P r o b a b i l i s t i c m o d e l s f o r r e s i s t a n c e o f c o n c r e t e b r i d g e g i r d e r s . ACI

12.

T a b s h S. W . a n d N o w a k A . S. ( 1 9 9 1 ) . R e l i a b i l i t y a n a l y s i s o f h i g h w a y g i r d e r b r i d g e s . A S C E J.

13.

Thoft-Christensen

E n g n g , 117(5), 1 4 1 3 - 1 4 3 4 .

Struct. ]., 91(3),

269-276.

Struct. Engng, 117(8), Springer-Verlag, 14.

Thoft-Christensen

theory.

2373-2388.

P. a n d B a k e r M . } . (1982). S t r u c t u r a l

reliability theory and its application.

267. P. a n d M u r o t s u Y . ( 1 9 8 6 ) .

Application of structural

systems

reliability

Springer-Verlag.

49

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Part 3. Bridge-specific loading

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Development of bridge-specific assessment and strengthening criteria P. C .

Das,

Highways

Agency,

London

Introduction The current 15 year bridge rehabilitation programme for trunk road bridges in the UK was started in 1988, and was soon accompanied by similar programmes for bridges on local authority roads. Prior to launching these programmes, the current bridge assessment code, BD 21, and the associated advice note BA16, were developed, based on an earlier standard, primarily to take into account the latest construction and use (C&U) vehicle configura­ tions and the resulting increased traffic loading. These documents were further amended in 1989 to include the predicted effects of the vehicles to be permitted into the UK from 1999 under the new EU directives for interna­ tional transport road vehicles. The rehabilitation programme has focused the attention of bridge engineers and highway authorities in UK on the practical implications of the assessment methods and criteria, in terms of the vast resources required for dealing with the sub-standard bridges and the large scale inconvenience caused to the public by the remedial works. It has become increasingly necessary to justify why such work is necessary, what options are available to the bridge manager, and what risks are involved in each of these options including in not doing the work. An extensive R&D programme has been put in place by the Highways Agency in recent years with the specific purpose of reviewing the assessment rules and methods to find answers to the above questions and also to ensure that the rules were not unduly conservative. Past projects in this area relating to individual problem bridges have resulted in a systematic improvement of the rules. However, the develop­ ments have now reached a stage where any further work on improving the rules in the current format is likely to produce diminishing results. The experiences gained from the rehabilitation programme, as well as from commissioning and managing the road network, now point to the need for a radically new bridge assessment methodology, incorporating much greater flexibility than at present, based on consideration of risks and options. Procedures based on such a methodology are now being developed by the Agency through a number of projects, and the following is an account of these developments. 1

53

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Levels of bridge assessment Bridge assessments, or indeed the assessment of any existing structure, can be carried out, provided the means are available, in stages of increasing sophistication, aiming at greater precision at each higher level. The R&D work in this subject area, commissioned by the Department of Transport and the Highways Agency in recent years, has been aimed at providing the tools for carrying out bridge assessments at these successively higher levels. Hence, for discussing these developments, it will be convenient to follow these levels. Broadly speaking, the following are the possible such levels of assess­ ment. • • •

• •

Level 1. Assessment using simple analysis and codified requirements and methods. Level 2. Assessment using more refined analysis. Level 3. Assessment using better estimates of bridge specific load and resistance values, using probabilistic estimates where possible. Level 4. Assessment using bridge specific target reliability. Level 5. Assessment using full-scale reliability analysis.

Level 1 This is the simplest level of assessment, giving a conservative estimate of load capacity. At this stage, only the simplest analysis is necessary, and the full factors of safety given in the codes are used. To facilitate such assessment, the assessment code BD 21 and the accompanying advice note BA16 have been developed. These documents themselves represent considerable relaxation compared to the design codes, which would be the only alternative in their absence, and the use of which would have resulted in more bridges failing assessment unnecessarily. In addition to the main assessment code, assessment versions of the design codes for concrete and steel bridges, BD 44 and BD 56 respectively, have also been developed for the purpose of providing relaxation for those types of bridges. The implementation documents for the 15 year bridge rehabilitation programme, BD 34, BD46 and BD 5 0 also contain valuable guidance. 2

3

4 - 6

Level 2 If a bridge is found to be inadequate in the more straightforward Level 1 assessment, it then needs to be assessed using more refined analysis and better structural idealisation. Load and material tests and non-destructive tests (NDTs) can also be very useful at this stage. To assist the engineer in

54

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

carrying out such assessments, the Department of Transport and more recently the Agency have commissioned an extensive research programme into the behaviour of various bridge types, involving both model and full scale testing and analysis methods. The type of bridges which this work has covered are masonry arch bridges and other older bridge types such as jack arch and trough deck bridges. More modern bridge types with special problems, such as the shear capacity of voided slab concrete bridges, have also been investigated. This research, which is still continuing, has resulted in better understand­ ing of bridge behaviour. For instance, the arch bridge assessment methods were thoroughly examined and better guidance is now available in BD 21 and BA16. Most other results are available in the numerous TRL reports of recent years. Level 3 The Level 1 and Level 2 assessments are carried out using the same loading requirements for all bridges. The strength parameters used are also as given in the codes. In Level 3, further refinement can be made by adopting bridge specific loading requirements as well as bridge specific worst credible values of the strength parameters. The bridge specific loading may depend upon traffic conditions and com­ positions, road surface irregularities and the dynamic response of specific bridges. For long span bridges (i.e. over 50 metres span), methods for deter­ mining bridge specific assessment live loading have already been available for some years, and such loading is permitted to be used in accordance with BD 50. The current short span loading is a single set of worst credible loading, based on estimated maximum dynamic impact effects, worst overloading of vehicles and maximum lateral bunching of vehicles on a bridge. It does not take into account different traffic conditions on different types of roads, different impact characteristics, or the lower probability of maximum impact effects occurring at the same time as lateral bunching. In order to eliminate some of these limitations, the Agency has recently developed probabilistic bridge specific traffic loading criteria for short span bridges through the following three projects 5

• • •

TRL collected statistical vehicle impact data for a large number of road and bridge situations and vehicle configurations Flint & Neill developed the probabilistic load models using the TRL supplied data, and earlier traffic data Imperial College carried out the calibration of the load models using reliability analysis of a number of bridge spans and configurations.

The papers in Part 3 contain further background to these projects and

55

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

details of the work carried out in them. The outline new rules are also given in the papers. Level 4 Level 3 assessments, although involving better estimates (probabilistic, when possible) of load and strength parameters, are still carried out on the basis of code implicit target reliability. For example, the worst traffic load level in the above mentioned load development, is determined by equat­ ing the corresponding reliability to that related to the current assessment live loading level. Such code implied target reliability, being design related, may be over-conservative for particular bridges. Furthermore, the current code requirements do not take into account whether the bridge is a new bridge or an old one with limited life requirement, or whether the type of bridge is likely to deteriorate significantly in the future or not. Level 4 assessments, should ideally address some of these limitations, and should therefore be a more refined form of bridge specific assessments which consider the target reliability for the type of bridge concerned and the age of the specific bridge. To allow such assessments to be carried out, whole life reliability based target performance profiles are being developed under four research projects commissioned by the Highways Agency. The work carried out so far in these projects is described in Part 5.

Level 5 The outputs from the work being carried out in relation to Level 3 and Level 4 assessments are intended to be in a form readily usable by the assessing engineers. For this purpose, the new rules will be in terms of the live load capacity factor K defined in BD21. Both the new bridge specific loading requirements and the whole life target profiles will be given in terms of K factors. For developing the K profiles, extensive use is being made of full scale reliability analysis of many bridge types and configurations. For important bridges, or for some special cases, it may be worthwhile using such analysis directly instead of using the specified K based new requirements. This more refined and fundamental level of assessment using reliability analysis is the Level 5 assessment, the basic techniques for which are also becoming available as an outcome of the present work.

Conclusions In conclusion, it can be said that the problems faced in the context of the current bridge rehabilitation programme, as well as in procuring and

56

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

managing the bridge stock in recent years, have necessitated a critical examination of the current bridge assessment rules and procedures. As a result, considerable advances have been made in the way highway struc­ tures are assessed and it is now becoming possible to carry out assessments at levels of increasing sophistication.

References 1. B D 2 1 a n d B A 1 6 ( 1 9 9 7 ) . T h e a s s e s s m e n t m a n u a l f o r roods

and

bridges.

2. B D 4 4 ( 1 9 9 5 ) . T h e a s s e s s m e n t HMSO, 3.

HMSO,

and strengthening

of h i g h w a y

structures. Design

London.

of c o n c r e t e s t r u c t u r e s . D e s i g n m a n u a l for r o a d s a n d

bridges.

London.

BD 56 (1996) The assessment

o f s t e e l b r i d g e s . D e s i g n m a n u a l f o r roads

and

bridges.

HMSO,

London. 4.

BD 34 (1990). Implementation

of the 1 5 y e a r bridge rehabilitation p r o g r a m m e — S t a g e 1 older

short s p a n bridges. Design m a n u a i for r o a d s a n d bridges. H M S O , 5.

BD46

(1992).

Implementation

of

the

15 year

bridge

London.

rehabilitation

programme—Stage

m o d e r n short s p a n bridges. Design m a n u a l for r o a d s a n d bridges. H M S O , 6.

BD 50 (1992). Implementation

2

London.

of the 1 5 y e a r bridge rehabilitation p r o g r a m m e — S t a g e 3 long

s p a n b r i d g e s . D e s i g n m a n u a l f o r roads

and

bridges.

HMSO,

London.

57

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Current UK bridge assessment rules and traffic loading criteria A. R. Flint,

Flint

& Neill

Partnership

Current design and assessment directives Design of highway bridges in the United Kingdom is commonly based on the provisions of BS 5 4 0 0 which, in addition to defining design loadings, encompass associated requirements for materials and workmanship. Direc­ tions on the application of that standard to bridges for the Department of Transport are given in a number of departmental standards and additional guidance is provided in advice notes within the Design manual for roads and bridges. Although these do not modify the basic principles of the British Standard they do vary some important design parameters, such as vehicular loading. The committee B/525/10 of the British Standards Institution will continue to maintain BS 5400 at least until the corresponding Euro Norms are developed and adopted. In consequence the developments described in this book are at present intended for application in conjunction with the above mentioned UK directives. Assessment of the structural adequacy of highway bridge stock in service is generally based on the design codes but with a number of permissible departures described in various departmental standards and advice notes which have been issued from time to t i m e . 1-5

6-9

10-14

The philosophy and principles of the design rules Until the advent of BS 5400 there had been no comprehensive United Kingdom design code for bridges. B S 1 5 3 was a 'permissible stress' specification strictly only applicable to common plate- and truss girder bridges of spans up to 300 ft., first published in 1923. However, it contained nominal highway loading applicable to other forms of construction which was updated in 1954. In 1967 it was decided to develop the new standard adopting limit state, partial safety factor principles and to revise vehicular loading. As stated in Part 1 of BS5400 it is implied by the use of constant load and resistance factors that the total consequences of failure of bridges of different types are the same. The safety factors do not vary with the location or importance. The design resistance equations and the associated partial factors in the current codes are in general appropriate only when material properties 15-17

15

1

58

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

BRIDGE SPECIFIC LOADING

and workmanship comply with the standard. With the exception of a design allowance for corrosion loss of inaccessible steel parts, no future deteriora­ tion of mechanical properties nor loss of steel material is catered for. How­ ever, for concrete the material factor is deemed to allow for some deterioration due to chemical attack, etc. Although a design life of 120 years was defined, its only applications relate to return period of the design winds and temperatures and to minimum fatigue life. The design HA loading is intended to produce load effects from construc­ tion and use vehicles having a return period of about a year on a heavily trafficked trunk route (representing approximately the characteristic loading divided by 1-2) with an allowance of 1 0 % for future growth. 10

Departures from design rules in assessment Possibly one of the most radical departures from design rules sometimes adopted for assessment are waivers on compliance with serviceability criteria, particularly for concrete bridges. BD 21/93 suggests that bridges built before 1965 do not in general need to be checked for the serviceability limit state since they are deemed to be likely to have previously exhibited distress if they were deficient but younger structures should be so checked. Since significant increases in potential loading have occurred in the past 25 years and since the nominal serviceability loading has a probability of being exceeded of about 1 in 120 per annum that policy deserves review. There is at present no consistency in the treatment of serviceability. In application of the design rules in assessment, provision is made to take into account information not available at the design stage, including known material properties obtained from construction records or tests for assess­ ment. In such cases there are recommended procedures for estimating in situ strengths and deriving worst credible values. Alternatively, condition factors based on judgement may be applied to the nominal resistance. Measured component shapes and sizes, locations of reinforcement and any losses of section may also be used. In such cases some beneficial adjustments to the factors on resistance are made to reflect decreased uncertainty, although no rigorous procedures exist to relate these changes to the uncertainties in mechanical properties. In the context of the topics to be discussed in this seminar it should be noted that BA44/90 suggests that 7 could be reduced from 1-2 for new concrete to 1-0 for concrete which is not expected to deteriorate further. 12

11

m 2

Vehicular traffic loading A prime cause of the current assessment and strengthening programme is the perceived need for bridges to safely carry highway traffic containing vehicles

59

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

180

0

T

| i i i i | • i i i | i i i i | i i i 0

10

20

30

i

|

i

i

i

40

i

|

i

i

i

i

50

[

i

i

i

i

60

|

i

70

i

i

i

|

80

i

i

i

i

|

90

i

i

i

i

|

100

Loaded length, m

Fig. 7. Equivalent UDL plotted simply supported

against loaded length for bending

moment

at mid-span of a

bridge: lane 7

providing more onerous load effects than those covered by the current Con­ struction and Use (C&U) Regulations. BD37/88 provides design HA loading suited to this but the majority of the bridge stock on trunk roads were designed to BS153 or BS5400 loading. The maximum laden weights of vehi­ cles permitted by the regulations have increased over the years. Although per­ mitted axle configurations and loads have been to some extent tailored to restrict the growth in potential load effects on short spans there has neverthe­ less been such growth and for long-loaded lengths substantial increase in effects due to congested traffic. Fig. 1 illustrates the variation with loaded length of the equivalent uniformly distributed HA loading for lane 1 of 3*65 m width from the current and earlier British design and assessment rules derived for mid-span moment in a simply supported member. The HA loading models in BS 153, BS 5400 and BD 37/88 each consist of uniformly distributed lane loads in conjunction with a knife edge load. The main changes from BS 153 incorporated in BS 5400 Part 2 were to truncate the intensity of the distributed loading for spans below 30 m and for those greater than 380 m and to require that all designs be capable of carrying at least 25 units of HB loading, the HB loading model being changed to have alternative spacings between inner axles. The current loading in 9

2

60

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

BRIDGE SPECIFIC LOADING

B S 37/88 does not truncate the distributed loads, increases the minimum HB loading to 30 units and greatly increased the load intensity above the B S 5400 levels for long spans.

Current design loading The current design loading is contained in BD 37/88 and was derived in dif­ ferent ways for short and long loaded lengths, the transition between the two occurring in the region of 40 m below which the flowing traffic with dynamic effects govern and beyond which congested traffic is the more onerous.

Short span loading The static loading model was derived by deterministic analysis of load effects on representative influence lines due to the range of goods vehicles permitted by the C&U Regulations from which envelopes were p r o d u c e d . The static loads on the vehicles and axles were taken as the legal weight limits multiplied by upper bound overload factors obtained for 2 axle rigid and 4 axle articulated vehicles from weighbridge measurements at selected sites. The extreme effects for spans greater than 20 m were found to be due to convoys of 4 axle rigid vehicles. Diversity of overload w a s allowed for by reducing the overload factor from 1-4 to 1-0 over the span range 10—60 m. For spans of up to 20 m allowance is made for potential lateral bunching of three lanes of traffic into two notional lanes, the lane loading being varied with lane and carriageway width. The density of the bunching w a s assumed to diminish for spans greater than 20 m, until for spans greater than 4 0 m there was no bunching allowance. An impact factor of 1-8 was applied to the worst axle in the single vehicle cases, derived from loads measured on the rear wheels of a 2 axle rigid vehicle. T h e loadings derived were assumed to represent the maximum credible values and were divided by 1-5 to give nominal design values and increased by 1 0 % to allow for unspecified future changes in traffic. 18

Medium-long span design loading For spans greater than about 4 0 m extreme load effects are governed by conditions with more than one vehicle in a lane, either in flowing traffic with dynamic effects u n c o r r e c t e d between vehicles or in congested traffic without impact. T h e design loading for these w a s derived by Monte Carlo analysis using statistical models of traffic mixes, axle and vehicle weights, axle and vehicle spacings, and jam frequencies and l e n g t h s . The characteristic (1 in 2400 probability per year) loading w a s derived from analysis of effects on various influence lines for different spans. T h i s was divided by 1*2 to give nominal design loadings. 19

61

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

The data on traffic mixes and vehicle characteristics were based on census observations on trunk routes and weighbridge measurements in the United Kingdom. An allowance of 1 0 % was added for possible future growth in loading. Following the Armitage report further analysis was undertaken with traffic containing vehicles outside the C&U Regulations considered in the report. It was concluded that such traffic would not provide more onerous load effects than current C&U traffic. Special Types General Order (STGO) vehicles, which have dimensions or weights not in compliance with C&U Regulations, were not included in the traffic mixes. Subsequent studies showed that STGO category 1 vehicles could be accommodated by the HA loading. 21

Assessment loading The assessment of UK highway bridges in accordance with BD 21/93 is currently undertaken to modified HA loading without any allowance for HB vehicles. Provision is made to assess for capacity to carry 40 tonne EU vehicles, normal C&U traffic or traffic excluding heavy categories of vehicles. No provision is made for future growth in loading. Load reduction factors, K, are applied to HA loading from BD 37/88. For 40 tonne capacity the factor is 0*91. For other categories the factor for spans less than 20 m varies with span being derived from load effect envelopes, calcu­ lated for the relevant range of C&U vehicles covered by the regulations in 1986 without the smoothing adopted for the design load model, but including the same assumptions regarding dynamic effect, overloading and bunching, constant K factors applying to longer spans. No allowance is made for STGO category 2 or category 3 vehicles, which have a weight of up to 80 tonnes and 150 tonnes respectively, which implies effective prohibition of such vehicles on assessed bridges. The partial factors on the modified HA loadings and dead loads are main­ tained at the design values. The value of the factor 7 is maintained at 1-1 for other than cast iron bridges. Values for assessment for serviceability are not given. Provision is made for alternative loading for long span bridges applicable to specific bridges based on analysis of traffic data obtained for the relevant route. f 3

Features of assessment rules under review The following aspects of current assessment rules for the United Kingdom are under review and discussed in this book •

short span bridge loading, taking account of anticipated traffic and dynamic effects

62

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

BRIDGE SPECIFIC LOADING

• • •

allowances for bunching of multiple lane traffic load reduction factors allowances for deterioration.

References 1.

B S 5 4 0 0 (1978).

Steel, concrete and composite bridges.

S t a n d a r d s Institution, 2.

B S 5 4 0 0 (1978). Steel, S t a n d a r d s Institution,

3.

4.

5.

6.

7.

8.

statement.

British

concrete and composite bridges.

P a r t 2:

Specification for loads.

British

London.

concrete and composite bridges. P a r t 3 : C o d e of practice for the design of steel bridges. B r i t i s h S t a n d a r d s I n s t i t u t i o n , L o n d o n . B S 5 4 0 0 ( 1 9 9 0 ) . Steel, concrete and composite bridges. P a r t 4 : Code of practice for the design of concrete bridges. B r i t i s h S t a n d a r d s I n s t i t u t i o n , L o n d o n . B S 5 4 0 0 ( 1 9 7 9 ) . Steel, concrete and composite bridges. P a r t 5 : Code of practice for the design of composite bridges. B r i t i s h S t a n d a r d s I n s t i t u t i o n , L o n d o n . B D 2 4 ( 1 9 9 2 ) . D e s i g n o f c o n c r e t e b r i d g e s . U s e o f B S 5 4 0 0 : P a r t 4 : 1 9 9 0 . D e s i g n manual for roads and bridges. H M S O , L o n d o n . B D 1 3 ( 1 9 9 0 ) . D e s i g n o f s t e e l b r i d g e s . U s e o f B S 5 4 0 0 : P a r t 3 : 1 9 8 2 . D e s i g n m a n u a l for roads and bridges. H M S O , L o n d o n . B S 5 4 0 0 (1982). SteeJ,

B D 1 6(1982). Design of composite

roads and bridges. 9.

P a r t 1: G e n e r a l

London.

BD37

HMSO,

bridges. U s e of B S 5 4 0 0 : P a r t 5: 1 9 7 9 . D e s i g n m a n u a l

for

London.

(1988). L o a d s for h i g h w a y bridges.

Design manual

for roads and bridges.

HMSO,

London. 10.

B D 4 4 (1995). T h e assessment

for roads and bridges. H M S O ,

of concrete h i g h w a y bridges a n d structures. Design

11.

B A 4 4 (1995). The u s e of BD 44/95. Design m a n u a l

12.

B D 2 1 (1993). The assessment

13.

BD56

and bridges. (1996).

bridges.

HMSO,

for roads and bridges. H M S O ,

of h i g h w a y bridges a n d structures. Design m a n u a l

of steel

highway

bridges.

Design

manual

for roads and

London.

14.

B A 5 6 (1996). T h e u s e of B D 56/96. Design m a n u a l for r o a d s a n d bridges. H M S O ,

15.

B S 1 5 3 (1954). Girder bridges. Part 3 A : L o a d s . British S t a n d a r d s Institution,

Steel girder bridges. Steel girder bridges.

16.

B S 1 5 3 (1954).

17.

B S 153 (1954).

18.

D e p a r t m e n t of T r a n s p o r t (1992). Interim revised loading specification

tution,

London.

London.

Part 3B: Stresses. British S t a n d a r d s Institution,

London.

P a r t 4: D e s i g n a n d c o n s t r u c t i o n . B r i t i s h S t a n d a r d s Insti­

London.

A: Revision 19.

London.

for roads

London.

The assessment

HMSO,

manual

London.

of short span loading. Department of Transport,

for bridges:

Appendix

November.

Flint & Neill P a r t n e r s h i p (1986). Interim d e s i g n s t a n d a r d : long s p a n bridge

loading.

Trans­

port Research Laboratory Contractor Report 1 6 , M a y . 20.

21.

B D 5 0 ( 1 9 9 2 ) . Technical r e q u i r e m e n t s for the a s s e s s m e n t a n d s t r e n g t h e n i n g p r o g r a m m e for highway structures—Stage 3 long span bridges. H M S O , London. D e p a r t m e n t o f T r a n s p o r t ( 1 9 8 1 ) . L o r r i e s , people and the environment. D e p a r t m e n t o f T r a n s ­ port House of C o m m o n s Paper 8439, H M S O ,

London.

63

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Development of short span bridge-specific assessment live loading D. I. Cooper,

Flint

& Neill

Partnership

Introduction Background The UK bridge stock is being assessed for adequacy to carry current and future heavy vehicle types, i.e. the EU 40 tonne vehicles that will be permitted from 1 January 1999. The current live load criteria used in the assessments are very similar to the design criteria, and for short span bridges both criteria are derived from a worst credible loading scenario. Probabilities of occurrence of such loads are not considered. The Highways Agency has recently commissioned a number of projects to develop site-specific reliability-based assessment methods for short span bridges, including revised live loading criteria. This paper describes the development of these loading criteria. 1

Methodology for production of load model Probabilistic

models

It was envisaged that a set of probabilistic load models would be derived to represent different site-specific traffic, road surface characteristics (which affect axle weight dynamic variations), and bridge dynamic response. It was stipulated by the Highways Agency that the final deterministic load levels to be used in assessments would be defined in terms of load reduction factors, K, as provided by BD 21/93. It was also stipulated that the worst load level would be the current assessment live load level; i.e. HA KEL and UDL multiplied by a K factor of 0-91. In order to derive the required assessment load levels, probabilistic models of vehicle static and dynamic load effects were first developed. It was found that six such models were sufficient to describe all situations for a given loaded length and traffic lane. Characteristic load effects were determined from the probability distribu­ tions. The term characteristic was defined as an effect with a 5% probability of being exceeded in a 120 year period (or 1 in 2400 probability per year), assuming no change in the nature of the traffic. It is convenient to specify design loads by such infrequent values. This is because reliability analysis 2

64

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3 . BRIDGE SPECIFIC LOADING

has shown that the variability in partial factors on loads (and resistances) tends to be lower when the factors are defined relative to rare values than when they are defined relative to frequent values. In this study, the precise definition of characteristic was immaterial, since eventually the assessment load reduction factors were to be determined from the relative values of the products of these characteristic loads and the partial factors which would be required in order to provide uniform reliability across the six live loading scenarios. The probability distributions describing the live load models were provided to Imperial College for calibration using reliability analysis.

General

principles

Bridge traffic load effects result from vehicle axle loads, interacting via road roughness and structural discontinuities, with bridge static and dynamic response. All these parameters needed to be considered. It is convenient to separate static from dynamic responses. The static response implies a load effect which is caused by the interaction between the weights of moving or stationary vehicles and bridge elastic response independent of bridge or vehicle inertial effects. A static load effect model could be derived from probabilistic models of site-specific and time-period related vehicle dimensions and wheel forces. The dynamic response implies a load effect that results from deformations due to the interaction of a time-period related set of loads with an elastic structure, inclusive of inertial effects. It may cause variations above and below the static load effect, and is probabilistic, in that its magnitude in any particular case cannot be predicted with absolute certainty, but statistical measures of its variations may (conceptually) be determined. The Transport Research Laboratory is able to obtain real time measure­ ments of wheel loads from moving vehicles. This is done by using strain gauges to measure axle bending, with accelerometers to correct for inertia of the unsprung masses. The TRL method appeared to be capable of provid­ ing the basis for realistic dynamic load models, and the Highways Agency commissioned TRL to obtain data from a variety of vehicles on a selection of bridges and on different road surfaces. 3

4

Derivation of load models Static load effect Vehicle

data

Modern weigh-in-motion (WIM) systems allow large data samples to be obtained. Measurements are instantaneous point-in-space samples from

65

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

the continuously varying wheel forces on the road, so they do not perfectly represent static weights. The differences between WIM records and station­ ary vehicle forces (which are often called errors) increase the apparent dispersion of the data. This results in conservative overestimates of extreme load effects. We favour using the data as measured, rather than making attempts to correct for errors in dispersion. When and if improved data become available, uncertainty might be reduced and further rationalisation of load models might then be considered. The static model was based on a two week long sample of data from the M6 motorway near Warrington in 1990, representative of industrial inter­ city traffic. Studies of rural motorway traffic (Bridgwater, M5) showed very similar distributions of weights for each type of heavy goods vehicle (although numbers were smaller). Weight distribution histograms demon­ strate smooth distributions out into their tails. Goods vehicle weight and length distributions Fig. 1 shows the distribution of weights of the first working day's goods vehicle traffic from the Warrington WIM apparatus. The figure is a conventional dot plot, on which each dot indicates the weight and length of an individual goods vehicle. Overlaying the dot plot are two histograms. The vertical rectangular columns represent the numbers of vehicles within each cell along the horizontal weight axis, and the horizontal rectangles represent numbers of vehicles within each cell up the vertical length axis. Numbers of vehicles counted within each cell range are also shown. The heaviest, shortest vehicles were viewed on a simultaneous video record, and most proved to be mobile cranes. The regulations classify these as special types, and they were filtered out of the data and deleted from the simulations. These abnormal vehicles are frequent users of major roads, and it is most important to note that these assessment load models do not cater for them.

Extrapolations

It is convenient to assume that the probability distribution of static extreme traffic effects can be modelled by an Extremal Type I distribution, even though this assumption is only theoretically justifiable if the underlying random process is characterised by a single probability distribution which falls off in an exponential manner in its upper tail. Development of a model of extreme loading involves extrapolations, in order to obtain the probability distribution of such extreme events. When traffic data samples are relatively small, mathematical models for each vehicle type must be used. Two extrapolation stages are required: one for vehicles, and a second stage for load effects. There is, thus, uncertainty in both the vehicle model and the extrapolation process. 5

66

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

45

r

50

55

weight, t

I O J O T - O - I - O O T - O I - O C

Numbers of vehicles within ranges of 1.25 tonnes (excluding cars)

Fig. 7. Typical goods vehicle weight and length

distributions

Much larger data samples are available now than in the past, and there is less need to use mathematical models in order to simulate vehicles. Furthermore, use of continuously measured data ensures that complex effects due to load variations with time of day, and road lane chosen by each driver, are automatically available. Extrapolation may be confined to dealing with effects, unless future changes to the vehicle population are to be modelled.

Traffic

modelling

The W I M data provided the flowing traffic model: including weights, dimen­ sions, speeds, close following, and side-by-side behaviour. The vehicles in the data were run over bridge influence lines in a simulation process, and histograms of individual and joint lane load effects were recorded. The histograms of static load effects were used as the basis from which the para­ meters of the Extremal distribution for any desired return period could be derived. The magnitude of each single event was defined as being the peak load effect reached between the times that each individual vehicle left the bridge.

67

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Flow

rate

effects

The characteristic load effect values were recalculated for reduced traffic flows, in order to provide the reductions in assessment loading for lightly used routes. The reductions in flows were related to assumed annual average hourly HGV flows (AAHHGVF) as follows • • •

high flows: AAHHGVF > 70 medium flows: 7 < AAHHGVF < 70 low flows: AAHHGVF < 7.

High flow is equivalent to over 600 000 heavy goods vehicles (two-way total) per year. The characteristic loadings were found to change by approximately 10% for a ten-fold change in flow rates, so it is not necessary to be very precise in the estimates of goods vehicle flows. The intention of using these values was to represent motorway traffic, main A-road or busy IBroad traffic, and minor local roads respectively. Bridge

influence

lines

Various compromises are inevitably introduced by codification: notably because of the inability of simplified models to perform equally well on all influence line shapes. Load effects on three influence lines were studied. The first was that for mid-span bending in a simply supported beam. This was used as a proxy for all approximately symmetrical lines, with the bulk of their area close to their centres. The second line was that for end shear in a simply supported beam. This is a proxy for all asymmetrically skewed influence lines. The third was that for bending over the central pier of a two-span bridge beam. Flowing traffic simulations on the three chosen influence lines were performed at loaded lengths of 2-5 m, 5 m, 10 m, 16 m, 20 m, 30 m, 40 m and 50 m. In addition, traffic jam simulations were run at 40 m and 50 m. Axle and axle group statistics were also used to derive load models for 2-5 m and 5 m. Results

of static

analyses

For each of the selected loaded lengths and influence line shapes, histograms were obtained for 1-lane, 2-lane, 3-lane and 4-lane load effects, as well as for each lane individually. The 2-lane effects were obtained by combining the heaviest loaded lanes in each direction, using the second week's traffic as a proxy for opposite direction vehicles. Each histogram contained load effect counts obtained in a 2-week period of continuous flowing traffic. Fig. 2 presents an example plot of one such histogram. Each histogram was then converted into a cumulative frequency distribu­ tion (CDF). The values of the CDF were each raised to the power of a number

68

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3 . BRIDGE SPECIFIC LOADING

2000

-n

1000

0 0

303

150 Shear effect, kN

Fig. 2. Tail of distribution

ot shear effects for Lane 1

equal to the daily traffic flow, in order to obtain the CDF of daily maxima as shown in Fig. 3. The CDF of daily maxima was then plotted in Gumbel form, as in Fig. 4. The straight line shows the shape of the selected model superimposed of the data points. The parameters of the Extremal distribution of static load effects for this loaded length, influence line shape and traffic lane selection, were obtained from the straight line fit shown. These Extremal Type 1 parameters repre­ sented the distribution of maximum load events for two week periods, and the distributions of annual maxima were calculated from these. Dynamic load effects Bending moment stresses are due to overall bridge curvature, in forms that approximate to the simple modes of beam vibration. Dynamic amplification of moments occurs due to bridge transient response and oscillation, and can be predicted by superimposing results from analysis of the responses in each vibration mode. 1.0 - T O.B - CDF of Simulations

0.6 - -

0.4

m m

-i

^

C D F (Daily maxima)

0.2

0

100

400

200

fig. 3. CDF of Lane 1 shears plus CDF

69

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

350-rU = 298, a = 0.092

1

I -6

.

1 4

-

1

**H

2

0

2

-Log [-Log (CDFN)] e

Fig. 4. Gumbel plot for 20 m daily maximum

e

shears—flowing

traffic: Lane 7

Vehicle dynamic behaviour can be separated for practical purposes into two distinct parts. There are the oscillations which involve the mass of the whole vehicle on its suspension: the so-called heave or bounce response; and there are the oscillations of individual axles, responding to road rough­ ness and discontinuities: the wheel hop response. Typically, the heave mode has a frequency between about 2 Hz and 3 Hz, whereas wheel hop frequency is between about 12 Hz and 16 Hz. Modal responses were obtained by time history analysis, using the 'Wilson-f? algorithm to stabilise the theoretical responses without requiring the use of inconveniently small time increments in the process. The T R L d a t a were used in dynamic response analysis of simple bridge models. It has been established by o t h e r s that the first mode natural bending frequency of highway bridges may be assumed to be a function of span, rather than of construction type. Thus, bridges have natural frequen­ cies in their first bending mode of approximately 8 2 L , where L is in metres. For a 10 metre bridge, this gives 10 Hz, and for a 4 0 metre bridge it gives 3 Hz. 4

6

- 0 9

Dynamic effects were found to be somewhat larger at higher speeds, but the difference w a s not large enough to justify additional rules on this basis. T h i s conclusion is similar to that arrived at by J. Page of T R L where, at a bridge with particularly poor surface, individual wheel load impact factors as high as 1-4 were measured at speeds of only 2 0 k p h . 8

Bridge

frequency

responses

Modal response analyses were performed for the instrumented vehicles on theoretical bridge models. W h i l e there were differences between static analysis results (using the constantly varying wheel loads but excluding modal responses) and the dynamic results (including modal responses), the overall set of static and dynamic results were very similar. Bridge dynamic response w a s thus very much less important than vehicle dynamic behaviour for most short span bridges.

70

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3 . BRIDGE SPECIFIC LOADING

5-Axl. Steel susp. artic. 4-Axl. Steel susp. artic. Normal

I 5-Axl. Air susp. artic. I 4-Axl. Steel + rubb. susp. artic. I 5-Axl. Steel, susp. tippr.

Normal curve: mean = 1.08, SD = 0.15

0.8

Fig. 5. Distribution

0.9

of impact values from modal analysis: all vehicles

Results of dynamic

analysis

Figure 5 shows the combined histogram of all dynamic amplification factors calculated from the recorded vehicle wheel forces on the bridge transits, using theoretical simple beam models for each bridge. The dynamic amplifi­ cation factor (DAF) is defined as the ratio between the largest load effect inclusive of dynamic response to occur in a bridge transit to the largest static effect in that transit. Each different tone of shading represents a different vehicle type (as described by Ricketts ). Apart from the values for the air suspension articu­ lated vehicle (for which amplification factors were significantly smaller than the others), there were no significant differences between the distributions of factors generated by the different vehicles. A conceptual problem arises when considering how to use the results of the dynamic analysis. Failure probability is the likelihood of occurrence of some event which is so large that it exceeds the capacity of the structure. Such an event will be rare, and will almost certainly be the largest event occurring for a considerable period of time. That being so, the statistics of extreme events may be used in the predictive model. However, all static events are associated with a dynamic component, and the question arises as to whether some form of extreme value analysis is needed for the dynamic model, or for the joint static plus dynamic effect. It is reasonable to suppose that bridges exhibit different dynamic responses due to physical differences (e.g. road surface profiles, response frequencies, etc.). Thus, particular bridges will tend to exhibit particular levels of response for a particular vehicle type, and the magnitude of that response is independent of the number of transits of that vehicle type. There­ fore, the DAF is independent of time: a characteristic that it shares with the 4

71

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

structural variables, rather than with the variables that describe other load variables such as vehicle static weights. Such a proposition is conservative in the sense that it implies that the bridges with highest DAF values always exhibit these highest responses, even on the rare occasions when a particu­ larly heavily overloaded vehicle passes. In the probabilistic model, therefore, the D A F is modelled by a single distribution of its variability, irrespective of exposure time or number of vehicle transits.

Full scale bridge tests The objectives of bridge instrumentation were • •

to confirm that bridge behaviour matches theoretical predictions, derived from applying measured loads to modal analysis bridge models to determine the amount of dynamic magnification in the bearing and end shear effects, neither of which is directly calculable in such a theoretically straightforward manner as the bending effects.

Selected

bridge

sites

Two bridges were instrumented. T h e s e were Holton Mill Bridge on the M407 A40, and Cuttle Bridge on the T h a m e Bypass. •

Holton Mill Bridge: T R L Bridge 69: O.S. Map: 165: Map Ref: 6112056.

This is a zero skew angle, three span bridge carrying the dual 3-lane carriageway A 4 0 over the River Thame. Construction is of parallel pre­ stressed concrete beams with in situ slab and black top road surface. The side spans of this bridge were over dry land (except in times of flood), and provided an opportunity to instrument a road carrying a major motor­ way, without the inconvenience of motorway regulations. •

Cuttle Bridge (Thame Bypass): T R L Bridge 72: O.S. Map: 165: Map Ref: 698067.

This is a three span bridge, carrying the single carriageway 2-lane Thame B y p a s s over the River Thame. T h e river is here smaller than at Bridge 69, but the centre span is also 15 metres, and the two side spans are 10 metres each. A c c e s s to side spans is easy, and soffit levels are very conveni­ ent. The deck slab is at about 1-8 metres, and no staging is needed for access. The bridge is composite, with steel beams (approximately 500 mm deep and 2-5 m centres) carrying in situ slab with black top surface. This bridge is ideal for instrumentation. It was possible to install simple L V D T s to measure deflections under any position of either side span. A description of the measurements at the Cuttle Bridge site follows.

72

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

Modal analysis moments 5AA/Steel: 10 mph

Moments from average wheel loads: 5AA/Steel

Max. modal moment: 42.52 Max. average moment: 40.92

Instant static moments 5AA/Steel: 10 mph

DAF: 1.0391

Fig. 6. Modal and static analyses for 10 mph transit

Measurements: Cuttle Bridge Measurements were obtained during east-bound transits, from the first span encountered. T h e wheel force records were used as input to the simple modal analysis model of a 10 metre span simply supported bridge having typical natural frequency, and the bending moments during 10 mph and 40 mph transits were assessed. The 5-axled articulated steel suspension vehicle results are described below. Figure 6 illustrates results of three bending moment analyses for the 5-axle articulated vehicle transits. The first plot illustrates the dynamic analysis, showing the effect of superimposed responses from the first three modes of vibration caused by the variable wheel force record. The second plot shows the moments resulting from transit of steady wheel forces with magnitudes equal to the average recorded values. T h e third plot charts the moments resulting from static analysis of a transit of the variable wheel forces actually recorded. The plots in Fig. 6 show that the results for the 10 mph transit the modal response analysis and the two static analyses were similar. The apparent dynamic amplification factor (calculated as the of the effects of the average forces and the modal analysis) w a s less 1-04.

from very ratio than

Figure 7 shows that for the faster transits, the modal analysis moment w a s over 4 0 % higher than the static moment effect of the average constant wheel forces: a D A F of 1-42. The peak moment calculated from a sequence of static analyses using the measured wheel loads was very similar to the modal ana­ lysis result, which was to be expected since the modal frequencies of this bridge were high relative to the vehicle bounce frequency. Bending strains were also measured for these vehicle transits. Fig. 8 compares these results at the beam mid-span. T h e dynamic amplification factor w a s recorded as being 1-37, which is very close to the 1-42 value calculated in the modal analysis.

73

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Finally, the DAF for deflections was assessed by comparing the LVDT measurements, and the results are shown in Fig. 9. The dynamic ampli­ fication factor for deflections was thus measured to be 1-3, which is some­ what less than the 1-4 value found in the moment effect analysis and Modal analysis moments 5AA/Steel: 4 0 mph

Moments from average wheel loads: 5AA/Steel

Max. modal moment: 58 Max. average moment: 40.75

Instant static moments 5AA/Steel: 4 0 mph

DAF: 1.42331

Fig. 7. Modal and static analyses for a 40 mph transit Mid-span strains: 10 mph

Mid-span strains: 4 0 mph

Comparison of mid-span bending strains 5-axled steel suspension artic. DAF: 1.37

Fig. 8. Comparison

of results at mid-span

10 mph 5-axle artic steel + steel

4 0 mph 5-axle artic steel + steel

4 0 mph deflection: -1.6632 10 mph deflection: -1.276

Fig. 9. Comparison

of the dynamic amplification

factor for deflections

74

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

measurement. The difference may be because the peak moment is caused by response in several modes, giving higher local curvature than that implied by simple interpretation of deflection measurements.

Shear

results

Owing to the configuration of the run-on slab on Cuttle Bridge, local wheel loads measured on the vehicle on the start of the bridge (above the strain gauges) were lower in the higher speed transits than in the lower speed transits. However, the strain gauge results during the high speed transits were also lower, which demonstrated that the beam stiffener above the end bearing did not respond like a simple mass-spring system under a step function load. The shear vibration frequency must be so high that there is little magnifi­ cation due to bridge dynamic response. Shear forces reflect the vehicle wheel force variations, so the shear (and bearing force) dynamic response model can be taken to be the same as that for bending.

Conclusions

•

• •

•

•

•

•

from

tests on instrumented

bridges

The bridge tests demonstrated that the predictions of dynamic amplifi­ cation factors that were produced by time-history modal dynamic analysis using force input from measured wheel loads were realistic. Shear responses were similar to those for bending. The calculated dynamic strain responses for the two bridges were between 5% and 7% higher than the measured dynamic deflection responses. Dynamic response determined by strain measurements agreed more closely with the moment calculations than with the deflection measure­ ments. Bridge flexural natural frequencies, for spans below about 40 metres, are higher than the predominant vehicle frequencies, so bridge elements do respond directly to short duration changes in vehicle wheel loads. Thus, load effects obtained from instantaneous static analysis using measured wheel loads provide realistic estimates of dynamic behaviour. A bridge with an uneven approach road (such as that at the Thame Bypass site) may exhibit quite different dynamic amplifications for different vehicles. However, for any particular vehicle, the dynamic response w i l l be similar for all transits. Therefore, the dynamic allowance must be based on a worst vehicle type, but a probabilistic model will then be independent of exposure time. Theoretical analysis of longer span bridges demonstrated sufficiently more dynamic amplification for lower frequency structures to justify some additional allowance in the probabilistic load model. However,

75

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

at longer spans traffic jam effects start to dominate flowing traffic effects. Details of probabilistic load model Basic parameters

considered

by

model

Load effects were defined by •

A model to give a set of static load effects. The geometrical configuration of load model 1 in Eurocode 1, Part 3, was used for this purpose, since it includes a tandem axle configuration that applies geometrically realistic force distributions on to structural models. A statistically defined factor, to provide the required probability distri­ bution of static load effects relative to the basic static load model. A statistically defined factor, to provide the required dynamic enhance­ ment of static load effects. 8

• •

The parameters of the statistical factors were derived for different bridge spans, road roughnesses and traffic flows. Separate statistical models were derived for • • •

extreme lane 1 effects, due to passage of an individual single vehicle of extreme weight joint extreme multi-lane effects, due to coincident passage of heavy vehicles heavily compressed jammed traffic, stationary and closely spaced side by side.

These models were derived from simply supported bridge studies, for which the terms span and loaded length may be assumed to be equivalent. We consider the final assessment loading to be suitable for multi-span bridges, since shorter spans for the same overall length will tend to reduce dynamic effects slightly. roughness Real time records of dynamic wheel forces were obtained by TRL from poor, medium and good surface roughness sections of the TRL test track. It became evident that (for the range of roughnesses that were considered), significant reductions in bridge response from normal conditions were only obtained on the good test track, and the rules reflect this conclusion. Road

5

Calibration The most onerous current assessment scenario is for a bridge with relatively poor road surface and heavy traffic flow rates (exclusive of abnormal indivisible loads) similar to those met on major motorways. Characteristic load effects were derived for a range of spans and numbers of lanes, and the preliminary values of the lane load adjustment factor K

76

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

were defined as the ratios between the characteristic effects and the HA effect for 2*5 m lane width, multiplied by a constant normalising factor to ensure no change in assessment loading in the most onerous scenario. The 2*5 m lane width was chosen as the load on this width represented the effects of a single line of vehicles without any lateral bunching effect. Calculations of reliabilities resulting from use of the statistical load models were carried out for a limited number of bridge spans and construc­ tion types. It was found that the optimum live load partial factor values (i.e. the ratios between design point values and the characteristic load effect values) were not constant, which indicated that modifications to the preliminary K values would be desirable in some cases. These modifications were carried out, and the final K values were prepared. More details of this exercise are given in the following paper. 9

Basis of assessment load Target

reliability

The Highways Agency decision was that it was to be assumed that current assessment rules which cater for normal road traffic (as opposed to those which cater for abnormal indivisible loads) provide satisfactory structures, even when such structures are subject to the most onerous conditions of use. Abnormal vehicle effects are subject to separate assessment. For each bridge configuration, the new assessment load model for the worst case scenario was to be identical to that already given in BD 21/93. Other targets might have been chosen. For example: a uniform target reliability index would have allowed further reductions in assessment load levels in some scenarios. The assessment load model comprises a definition of both loads and partial factors. Furthermore, it was stipulated by the Highways Agency that the present values of partial factors should be kept. Since the loading (i.e. the product of load and partial factor) on the most onerous case structures was to remain unchanged, the remaining question was to con­ sider how the loading should vary for less onerous cases, in order to retain consistent notional reliability for each span and bridge type. 2

Combined

static

and dynamic

effect

The overall distribution of load effects is a mixed distribution comprising an Extremal Type 1 distribution (for which the maximum w i l l depend on number of repetitions of the load events) and a normal distribution (which is independent of the number of events). For the purposes of reliability analysis, it was not necessary to provide a combined static plus dynamic load effect distribution. However, it is

77

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

convenient to calculate lane load factors from the characteristic values. These can be calculated by simulation, but in each case they were found to be very close to the product of the 2400 year mode of the Extremal, and the 5% probability of exceedence value of the normally distributed dynamic multiplier.

New assessment live loading levels Results of reliability analysis Introduction 9

The reliability analyses are described and reported elsewhere. The following notes are intended to illustrate the principal conclusions of the study.

Reinforced

concrete

slab

bridges

Spans of 5 m, 10 m and 16 m were selected for study. As spans increased, and live to dead load ratio decreased, there was found to be a marked reduction in the sensitivity of the reliability index to uncertainty in live load. Optimum sets of partial factors for permanent properties and for permanent and transient loads were obtained for six scenarios of road and traffic conditions, with the live load factor being adjusted so as to produce uniform reliability. The sets of factors were not identical to those currently presented in BS 5400, Part 2 and required for use in BD 21/93 (because the BS factors were calculated on a slightly different basis), but there was a substantial amount of uniformity in all except the live load factors across the range of road conditions. The most obvious difference was that the live load partial factors required for the poor road surface scenarios were, on average, 7-5% greater than those for the good surfaces. Therefore, for the good road surface cases, the assessment live load could be reduced by a further 7% below that reduced level already determined from changes in characteristic loading. 1 0

Reinforced

concrete

beam

plus

slab

2

bridges

In the reinforced concrete beam + slab cases, the live load partial factors for the good surface cases were again generally lower than those for the poor surface cases, although here there were more significant flow-related changes. There was approximately a 6% increase in the partial factor required for the low flow cases, which cancelled out part of the benefit of the 14% reduction in characteristic load effects in these cases.

78

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

Assessment load reduction factors Derivation

of lane load

factors

The statistical load model that was used in the reliability analysis contained a realistic pattern of wheel loads, with patch loads separately defined at the ends of notional axles. This is quite unlike the knife edge load provided by BD 21/93. The result was that the critical condition was caused by adjacent vehicles from lanes 1 and 2, since adjacent vehicles' wheels can approach closer side by side than wheels on the same vehicle. The Eurocode tandem axles have wheels 2 0 metres apart, whereas wheels from adjacent vehicles can approach to 0*5 metres (assuming a 3-0 metre centre-to-centre vehicle spacing). Slab bridges have relatively good lateral dispersion and are governed by the effects of 2-lane loading, but beam bridges which have relatively poor lateral dispersion are also governed by 2-lane loading: this time on the beams below and between adjacent vehicles. It follows that the assess­ ment lane load factors that cater for the joint 2-lane condition cannot be reduced without degrading reliability. The BD 21/93 lane load factors for lanes 1 and 2 (which provide for equal design load on both) must, therefore, remain unchanged, if the principle of not allowing reductions in loading for current worst case scenarios is to be followed. Since the lane load factors for lanes subsequent to the first two were only found to be significantly reduced from the BD 21/94 values at the shortest loaded lengths, there is no strong justification for altering the existing values.

Lateral

bunching

Short span bridge loads in individual lanes are governed by individual, or small groups of, vehicles travelling at normal road speeds. These govern because there are generally very many more flowing traffic than stationary traffic events, even on the busiest routes, and it is the moving vehicles which provide the highest load effects on short spans due to their increased dynamic contribution. The BD 21/93 load model for the shortest spans caters for the possibility that, for short distances at least, vehicles might come side by side much more closely than on long stretches of road. The current design rules incorporate full impact on laterally bunched traffic because (as noted previously) large dynamic wheel load amplifications were measured at speeds as low as 20kph (approximately 12*5 mph). However, recent TRL research has established that vehicles are much more closely spaced when stationary than when moving, however slowly. Thus, the model needs to cater either for moving vehicles w i t h dynamic effects but no lateral bunching, or stationary vehicles with bunching but no dynamics. 11

79

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

T h e s e effects were compared. It w a s found that the simply supported span moment effects of a single lane of moving vehicles were greater than the moment effects of the single lane share of two lanes of stationary vehicles. For example: for a 5 metre span, the minimum value (i.e. with the smooth road impact factor) of characteristic moving vehicle moment effect in the heaviest lane loaded w a s found to be 4 5 7 kNm. Assuming a very high ( 2 5 % ) fraction of traffic to be not only in jammed conditions, but also closely laterally bunched, and assuming a contraflow case where the worst lanes in each direction were adjacent, the charac­ teristic static worst lane effect w a s 3 5 1 kNm, and the second lane effect w a s 2 2 0 kNm. The load from bunched traffic within a 3*65 m lane would then be 3 5 1 - f 1-15/2-5 X 2 2 0 = 4 5 2 kNm which is less than the 4 5 7 kNm found in the flowing traffic case with mini­ mum impact; and this even assumes onerously (and quite unrealistically) that all vehicles which run side by side do so only 2-5 metres centre to centre. So, even with a very high assumed proportion of traffic in jams, and very onerous lane disposition, the flowing traffic case without lateral bunching exceeded the jammed traffic case with lateral bunching. Therefore, no lateral bunching allowance is needed. Since the current assessment rules allow loading to be reduced in proportion to lane width down to the value for 2-5 metre wide lanes, the value currently used for 2-5 metre wide lanes is satisfactory for all lane widths.

Vehicle

widths

Actual H G V s are typically just under 3-0 metres measured across their wing mirrors, although they are 2-5 metres across their bodywork. Since flowing traffic conditions are critical at short spans, there seems to be no justification for allowing notional lane widths to fall below 3 metres, except where only one 2-5 metre wide lane is on a structure, or two lanes which can just fit into 5-4 metres with 0-4 metres vehicle separation, as required by Eurocode 1, part 3. T h e problem with factoring down the load to remove the bunching allow­ ance is that it is not the intensity of load per unit area on bridge decks which ought to be reduced in the ratio of 2-5 to the notional lane width: but, merely, the width of the area loaded within each notional lane should be reduced to 2-5 metres. T h e vehicles need not be placed closer than 3 metres centre to centre. For slab bridges, with good lateral dispersion, it would be acceptable to load the whole lane width as required by BD 21/93, and factor down the live load by the ratio of the BD 21/93 assessment load calculated for 2-5 metres to the original notional lane width. W h e r e a s s e s s m e n t s have already

80

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

been carried out, this allows the original assessment live load effects to be reduced in the same ratio. For beam plus slab bridges without good lateral dispersion, it is only possible to justify reducing the widths of applied load to strips of 2-5 metre width, while retaining the original BD 21/93 intensity. E a c h load strip ought to be placed at 3-0 metres centre to centre.

Load reduction

factors

The final values of reduction factors were obtained from the relationship between characteristic values, modified by the changes in partial factors found in the reliability analysis. T h e factors are intended for all conventional construction. Load reduction factors for different scenarios were all calcu­ lated with reference to the most onerous current a s s e s s m e n t case, for which no load reduction factor would be allowed, apart from the 0-91 value given in BD 21/93. The analysis using the M6 set of data was taken to be appropriate to the 40 tonnes assessment live loading case, and all other assessment load magnitudes were calculated by reference to this case. Values for the 38 tonnes assessment loading case were obtained deterministically by factoring the 4 0 tonnes values by the ratios between the 38 and 40 tonnes values given in BD 21/93, Fig 5/2. Analysis of the 25 tonnes a s s e s s ­ ment live loading condition w a s performed by repeating the static load effect simulations with the original data, but filtering out all vehicles having more than 3 axles. T h e 17 tonnes case w a s similarly obtained by filtering out all vehicles with more than 2 axles, and the 7-5 tonnes case w a s obtained deterministically from the 17 tonnes case results using the BD 21/93 ratios as before. 2

Figure 10 compares the assessment load reduction factors given in BD 21/93 with those derived by the probabilistic study for low traffic and good road K Factors from BD 21/93

K Factors for low traffic, good surface, Lg 1.00

0.50

~jr—~~r-\-

0.30 -4-t

• •

;

i •

0.20 - ( • : • • • 0.10

:

!

—l

0.00 -H 0

—t—

1

1

20 30 40 Loaded length Assessment loading _ o - 4 0 t - o - 3 8 t - ^ - 2 5 t —M— 17 t

Fig.

10.

Load

reduction

factors

0.00 4 — 0

h-

10

for

50

—I I I 20 30 Loaded length

Assessment loading -o-40t -o-38t -*-25t

7.5 t

BD21/

1 10

and

low

traffic,

good

' 40

1 50

17 t *

7.5 t

surface

81

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

surfaces. T h e shapes of the lines differ somewhat. T h i s may be due to the progression of changes of bridge sensitivity to the effects of axles, axle groups, vehicles (with axle groups at each end) and vehicle groups. Table 1 shows the assessment load reduction factors derived for other scenarios. Some modifications to these reduction factors were made for the following purposes •

No probabilistic model for 3 tonnes assessment loading was produced, and it w a s decided that the existing rules should be unchanged. This limited the lowest allowable level for the 7-5 tonnes load model.

Table 1. Load reduction

factors for road

conditions

K Factors for heavy traffic, poor surface Length 401 381 251 17t 7-51 2-5 5 7-5 10 15 20 30 40 50

0-91 0-91 0-91 0-91 0-91 0-91 0-91 0-91 0-91

0-84 0-84 0-84 0-85 0-88 0-91 0-91 0-91 0-91

0-84 0-84 0-84 0-85 0-88 0-81 0-91 0-91 0-91

0-64 0-60 0-69 0-75 0-70 0-62 0-75 0-75 0-75

K Factors for heavy traffic, good surface Length 401 381 251 17t 7-5t

0-37 0-35 039 0-44 0-37 0-31 0-37 0-37 0-37

2-5 5 7-5 10 15 20 30 40 50

K Factors for medium traffic, poor surface Length 401 381 25t 17t 7-51 2-5 5 7-5 10 15 20 30 40 50

0-90 0-89 0-89 0-90 0-91 0-89 0-90 0-89 0-89

0-83 0-82 0-82 0-84 0-88 0-89 0-90 0-89 0-89

0-62 0-59 0-67 0-72 0-68 0-60 0-71 0-71 0-71

0-74 0-74 0-74 0-75 0-78 0-81 0-83 0-81 0-81

0-74 0-74 0-74 0-75 0-78 0-72 083 0-81 0-81

0-56 0-53 0-61 0-66 0-62 0-55 0-73 0-73 0-73

0-33 0-31 0-35 0-39 0-33 0-28 0-37 0-37 0-37

K Factors for medium traffic, good surface Length 401 381 251 17t 7-5t

0-36 0-34 038 0-43 0-36 0-30 0-36 0-36 0-36

2-5 5 7-5 10 15 20 30 40 50

K Factors for low traffic, poor surface Length 401 381 25t 17t

7-51

K Factors for low traffic, good surface Length 401 381 251 17t 7-5t

2-5 5 7-5 10 15 20 30 40 50

0-35 0-33 0-37 0-41 0-34 0-29 0-33 0-33 0-33

2-5 5 7-5 10 15 20 30 40 50

0-88 0-87 0-87 0-88 0-90 0-87 0-88 0-87 0-86

0-81 0-80 0-80 0-82 0-87 0-87 0-88 0-87 0-86

0-83 0-82 0-82 0-83 0-88 0-79 0-90 0-89 0-89

0-81 0-81 0-81 0-81 0-81 0-81 083 0-81 0-81

0-81 0-80 0-80 0-81 0-85 0-77 0-87 0-87 0-87

0-60 0-57 0-64 0-69 0-65 0-58 0-67 0-67 0-67

0-79 0-79 0-79 0-79 0-80 0-79 0-81 0-79 0-78

0-77 0-77 0-77 0-77 0-79 0-77 0-79 0-77 0-76

0-73 0-73 0-73 0-74 0-77 0-79 0-81 0-79 0-78

0-71 0-71 0-71 0-72 0-76 0-77 0-79 0-77 0-76

0-73 0-73 0-73 0-73 0-77 0-70 0-81 0-79 0-78

0-71 0-71 0-71 0-71 0-75 0-68 0-79 0-77 0-76

0-55 0-52 0-59 0-64 0-60 0-53 0-69 0-69 0-69

0-53 0-50 0-56 0-61 0-57 0-51 0-65 0-65 0-65

0-32 0-30 0-33 0-37 0-32 0-26 0-35 0-35 0-35

0-31 0-29 0-32 0-36 0-30 0-25 0-33 0-33 0-33

82

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

•

The 7*5 tonnes assessment live load model was adjusted upwards so as to merge with the long span requirements at 50 metres.

Nominal single axle and single wheel loads The BD 21/93 values for nominal single axle and single wheel loads for the different classes of assessment loading are in the same proportion to one another as are the reduction factors for UDL and KEL which are given i n Fig. 5/2 for the shortest span listed. The same principle was followed when preparing the values for the new loading rules. As in BD 21/93, the final values were rounded off slightly. Accidental loads The BD 21/93 accidental load criteria were not considered in this study.

Probabilistic load model Introduction This section describes the probabilistic load model that was used in the reliability analysis which resulted i n the new assessment live loading. This live load model can also be applied directly in reliability analysisbased assessments, e.g. in Level 5 assessments. Probabilistic models are presented for flexural and shear effects due to highway traffic load effects on simply supported bridges of spans from 5 metres to 40 metres. The models provide modification factors on a general traffic highway load model which is similar to that presented in Eurocode ENV 1991-3: 1994. The models do not cater for the effects of abnormal vehicle traffic. Each comprises a basic deterministic load model, and statistically defined factors that are applied in order to provide probabilistic effects. 12

2

Load combinations The following load combinations are provided for • •

Lane 1 loaded with flowing traffic (plus dynamic allowance). A l l lanes loaded with flowing traffic (plus dynamic allowance).

Lane widths The carriageway width is to be divided into notional lanes according to the rules given in BD 21/93. The bridge shall first be assessed for the effects of the lane 1 load model, placed at the most adverse location along the

83

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

centrelines of each of the notional traffic lanes in turn. T h e bridge shall then be a s s e s s e d for the effects of load on all traffic lanes, with each lane load model placed at the most adverse location along the centrelines of each of the marked traffic lanes in turn. Lane loadings in this model are interchangeable.

Statistical load

model

Statistical load modification factors are defined. These are applied to the load effects produced by the static load models. T h u s •

•

T h e lane 1 loaded model comprises the lane 1 basic load model, applying the factors shown for lane 1 in T a b l e 2 and applying the appropriate dynamic amplification factor. T h i s model is applied to a single 3-0 m width, with no e x c e s s area loading. T h e all lanes loaded model comprises the basic load model for joint effects on lanes 1 plus 2, plus the additional lanes load model; applying the factors shown for all lanes loaded in T a b l e 2 and applying the appro­ priate dynamic amplification factor. This model is applied to 3-0 m wide lanes, with no e x c e s s area loading.

Statistical model parameters are provided for heavily trafficked UK motor­ way sites, and also for sites where flow rates are substantially less than this.

Basic static load

model

Lane 1 basic load model

27 kN/m uniformly distributed load along a single

lane, placed centrally within the lane, and distributed on a width equal to the smaller of the notional lane width and 3-0 metres. Plus: 2 axle loads, each of 300 kN, spaced at 1-2 metres along carriageway; each axle having 2 wheels at 2-0 m track width placed centrally within the lane. Basic load model for joint effects on lanes 1 plus 2: apply model used for lane 1 to both lanes 1 plus 2. Additional

lanes load model

7-5 kN/m uniformly distributed load along each

of the remaining lanes, distributed on widths equal to the smaller of the notional lane widths and 3-0 metres. Plus: 2 axle loads, each of 100 kN, spaced at 1-2 metres along carriageway; each axle having 2 wheels at 2-0 m track width. There is no requirement in these load models to apply a distrib­ uted load to the excess lane width between 3 0 m and the full lane width.

Statistical

modifications

to static

model

The static load models are factored by statistically defined variables as in T a b l e s 2 a n d 3. T h e E x t r e m a l Type 1 distribution is used to model the probability distribution of extreme traffic load effects.

84

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

Table 2. Lane 1 load effect factor*

Span

Load effect

Traffic volume

1 year parameters

flow per direction

for Extremal Type 1

per day

distribution U

5

Moment

Shears and end reactions

10

Moment

Shears and end reactions

16

Moment

Shears and end reactions

20

Moment

Shears and end reactions

40

Moment

Shears and end reactions

alpha ( a )

50 000/motorway

0-423

56

10 000/main road

0-383

56

2000/minor road

0-354

56

50 000/motorway

0-443

53

10 000/main road

0-400

53

2000/minor road

0-370

53

50 000/motorway

0-387

62

10 000/main road

0-350

62

2000/minor road

0-324

62

50 000/motorway

0-407

58

10 000/main road

0-368

58

2000/minor road

0-340

58

50 000/motorway

0-444

52

10 000/main road

0-399

52

2000/minor road

0-368

52

50 000/motorway

0-490

42

10 000/main road

0-435

42

2000/minor road

0-396

42

50 000/motorway

0-411

77

10 000/main road

0-381

77

2000/minor road

0-360

77

50 000/motorway

0-421

58

10 000/main road

0-381

58

2000/minor road

0-354

58

50 000/motorway

0-373

102

10 000/main road

0-351

102

2000/minor road

0-335

102

50 000/motorway

0-382

67

10 000/main road

0-347

67

2000/minor road

0-323

67

* These models are subject to dynamic amplification effects.

85

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Table 3. All lanes load effect factor*

Span

Load effect

Traffic volume

1 year parameters

flow per direction

for Extremal Type 1

per day

distribution U

5

Moment

Shears and end reactions

10

Moment

Shears and end reactions

16

Moment

Shears and end reactions

20

Moment

Shears and end reactions

40

Moment

Shears and end reactions

alpha

50000/motorway

0-311

64

10 000/main road

0-275

64

2000/minor road

0-250

64

50 000/motorway

0-307

64

10 000/main road

0-271

64

2000/minor road

0-246

64

50 000/motorway

0-341

58

10000/main road

0-302

58

2000/minor road

0-275

58 56

50 000/motorway

0-334

10 000/main road

0-294

56

2000/minor road

0-265

56

50000/motorway

0-388

52

10000/main road

0-344

52

2000/minor road

0-313

52

50 000/motorway

0-389

58

10 000/main road

0-349

58

2000/minor road

0-322

58 52

50000/motorway

0-386

10 000/main road

0-341

52

2000/minor road

0-310

52

50 000/motorway

0-386

55

10 000/main road

0-344

55

2000/minor road

0-315

55

50 000/motorway

0-376

52

10 000/main road

0-332

52

2000/minor road

0-301

52

50000/motorway

0-353

61

10000/main road

0-315

61

2000/minor road

0-289

61

(a)

* These models are subject to dynamic amplification effects.

86

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 3. BRIDGE SPECIFIC LOADING

The probability density function of this distribution (PDF) is given by f (y) = a e x p ( - a ( y - U) - exp(-c*(y - U)))

(l)

Y

and the cumulative distribution function (CDF) by F (y) = e x p ( - e x p ( - a ( y - U)))

(2)

Y

where U is the mode of the distribution, and a is an inverse measure of dispersion. If U is the mode of the distribution of largest annual events, and the mode of the distribution of the largest events from sets of N years is U then 5

N

U

N

= U-(log (-log (l-l/N))/a e

(3)

e

which for large N becomes U

= U + l/alog (N)

N

Probabilistic

dynamic

(4)

e

modification

to static

model

Note: bridge natural frequencies are closely correlated with spans. In the absence of structural dynamic analysis, the mode 1 natural frequency, F , may be taken as: F = 8 2 L ' , where L is in metres. The dynamic amplifica­ tion factor is taken to be a normally distributed random variable. Values for different scenarios are presented in Table 4. Road surface roughness categories Road surface roughnesses are categor­ ised by means of the methodology used by TRL. In order for a road to be classified as having good surface quality for bridge assessment purposes, the variances of the moving average deviations of the profile at the centres of gauge lengths of 3 m, 10 m and 30 m must respectively be less than 4-5 mm , 30 mm and 180 mm , measured over the length of the bridge plus a 50 metre length extending 50 metres prior to the vehicle arrival point. In a

-0

9

a

13

2

2

2

Table 4. Dynamic amplification

Lane(s)

Frequency, Span

factors

Road

Mean

Standard

roughness

DAF

deviation D A F

1

>5Hz,

5H

5Hz,

5Hz,

/////////////

Section A-A

Fig. A.2.1. River Ken net Bridge (Not to scale)

209

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A3. Bridge No. 3 — W i n d m i l l Bridge Windmill Bridge is in Cambridgeshire (A14/122.2, DTP 10035). It carries the A14 trunk road over a river. It was constructed in 1975, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A.3.1 and shown schematically in Fig. A.3.1. The cover to the main longitudinal reinforcement is 40 mm.

Table A.3.1. Details hr Windmill

Span

Bridge

9-60 m

Width

18-65 m

Skew

34-5°

Slab thickness Concrete strength, /

(2 slabs of 18-65 m with joint at centre)

550 m m 30 N / m m

c u

2

Reinforcement strength, f

250 N / m m

Carriageway w i d t h

13-0m

Surfacing depth

100 m m

y

(nominal) 2

(nominal)

Reinforcement: Bottom longitudinal 32 mm at 100 mm c/c Bottom transverse

ft

12 mm at 200 mm c/c

0.55oJ Section A-A

Fig. A.3.1.

Windmill

Bridge (Not to scale)

210

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

AA. Bridge No. 4—South

Hinksey

Bridge

South Hinksey Bridge is in Oxfordshire (A34/80.1, DTP 7372). It carries a principal road over the A34 trunk road. It was constructed in 1973, and designed to carry 37-5 units of HB type loading. The bridge has a twospan deck slab continuous over central pier and simply-supported at the ends, as described by Table A.4.1 and shown schematically in Fig. A.4.1. The cover to the main longitudinal reinforcement is 40 mm.

Table A.4.1. Details for South Hinksey

Bridge

Span

2 x 20-20 m

Width

9-80 m

Skew

0° 1100 m m

Slab thickness Concrete strength, /

26-9 N / m m

c u

Reinforcement strength, f

410 N / m m

Carriageway w i d t h

7-3 m

Surfacing depth

100 m m

y

/////////

2

2

(nominal)

Reinforcement: Bottom longitudinal 4 0 mm at 125 mm c/c (central) 25 mm at 125 (near supports) Bottom transverse

2 0 mm at 2 0 0 mm c/c

Top longitudinal

16 mm at 125 mm c/c

(plus some 2 5 & 4 0 mm at 125 mm c/c over central pier)

Central pier

Top transverse

2 0 mm at 2 0 0 mm c/c

LU ///////// Plan

Fig. A.4.1.

South Hinksey Bridge (Not to scale)

211

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A.5. Bridge No. 5—River

Wang

Bridge

River Wang Bridge is in Suffolk (A12/159.7, DTP 9940). It carries the A12 trunk road over the River Wang. It was constructed in 1954, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A.5.1 and shown schematically in Fig. A.5.1. The cover to the main longitudinal reinforcement is 30 mm.

Table A.5.1. Details for River Wang Bridge

Span

6-70 m

Width

18-14m

Skew

46-6°

Slab thickness Concrete strength, /

460 m m 30 N / m m

c u

2

Reinforcement strength, f

230 N / m m

Carriageway width

7-42 m

Surfacing depth

250 m m

y

Plan

(nominal) 2

(nominal)

Section A-A

Fig. A.5.1. River Wang Bridge (Not to scale)

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

A.6. Bridge No. 6 — W e s t e r n l/C East Bridge Western l/C East Bridge is in Suffolk (A14/142.6, DTP 10014). It carries the A14 trunk road over a minor road. It was constructed in 1975, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A.6.1 and shown schematically in Fig. A.6.1. The cover to the main longitudinal reinforcement is 40 mm.

Table A.6.1. Details for Western l/C East Bridge

Span

16-2 m

Width

26-1 m

Skew

12°

Slab thickness

700 m m

Concrete strength, /

37-5 N / m m

c u

Reinforcement strength, /

y

410 N / m m

Carriageway width

2 x 9-3 m

Surfacing depth

100 m m

2

2

(nominal) (nominal)

Reinforcement: Bottom longitudinal 32 mm at 100 mm c/c Bottom transverse

2 0 mm at 325 mm c/c

26.1 A

A 0.700

Section A-A

Fig. A.6.1. Western l/C Fast Bridge (Not to scale)

213

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A.7. Bridge No. 7 — H a r d s t o f t - H a r d w i c k Road Bridge Hardstoft-Hardwick Road Bridge is in Derbyshire (Ml/224.8, DTP 512). It carries a minor road over the Ml motorway. It was constructed in 1967, and designed to carry 45 units of HB type loading. The bridge has a simply-supported beam and slab deck as described by Table A.7.1 and shown schematically in Fig. A.7.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.7.1. Details hr Hardstoft-Hardwick Span

18-63 m

Width

9-13m

Skew

0°

Road Bridge

Beam depth

1090 m m

Slab thickness

200 m m

Concrete strength, f

40 N / m m

Reinforcement strength, f

410 N / m m

Carriageway width

5-5 m

Surfacing depth

100 m m

cu

y

2

2

(nominal)

Main reinforcement: Bottom beams longitudinal 3 layers of eighteen 35 mm (internal beam) 3 layers of fifteen 35 mm (external beams)

18.63

Top beam and slab reinforcements: varies

9.13

ft I 0.200

777777777 1.450 Section A-A

Fig. A.7.1.

Hardstoft-Hardwick

Road Bridge (Not to scale)

214

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

A.8. Bridge

No. 8—Chesterfield-Bolsover

Bridge

Chesterfield-Bolsover Bridge is in Derbyshire (Ml/232.9, DTP 5024). It carries the A632 road over the M l motorway. It was constructed in 1967, and designed to carry 45 units of HB type loading. The bridge has a simply-supported beam and slab deck as described by Table A.8.1 and shown schematically in Fig. A.8.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.8.1. Details for Chesterfield-Bolsover Span

Bridge

18-6 m

Width

13-96 m

Skew

0°

Beam depth

1090 m m

Slab thickness

200 m m

Concrete strength, f

40 N / m m

cu

2

Reinforcement strength, f

410 N / m m

Carriageway w i d t h Surfacing depth

10-06 m 100 m m

y

(nominal) 2

(nominal)

Main reinforcement: Bottom beam longitudinal 3 layers of sixteen 35 mm (internal beams) 3 layers of fifteen 35 mm (external beams) Top beam and slab reinforcement: varies

V A

A

//////////// Plan

Fig. A.8.1. Chesterfield-Bolsover

Bridge (Not to scale)

215

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A.9. Bridge No. 9—Belper Road Bridge Belper Road Bridge is in Derbyshire (A38/456.9, DTP 7845). It carries the A38 trunk road over a minor road. It was constructed in 1977, and designed to carry 45 units of HB type loading. The bridge has a simply-supported beam and slab deck as described by Table A.9.1 and shown schematically in Fig. A.9.1. The cover to the main longitudinal reinforcement is 40 mm.

Table A.9.1. Details hr Belper Road Bridge

Span

14-30 m

Width

12-05 m

Skew

11°

Beam depth

900 m m

Slab thickness

(2 slabs of 13-95 m with joint at centre)

225 m m

Concrete strength, /

37-5 N / m m

c u

Reinforcement strength, f

410 N / m m

Carriageway width

9-3 m

Surfacing depth

100 m m

y

2

(nominal)

2

(nominal)

Plan

Section A-A

Fig. A.9.1. Belper Road Bridge (Not to scale)

216

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

A.10.

Bridge No. 10—Park

Lane Bridge

Park Lane Bridge is in Derbyshire (A38/460.0, DTP 7848). It carries the A38 trunk road over a minor road. It was constructed in 1977, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A.10.1 and shown schematically in Fig. A.10.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.10.1. Details for Park Lane Bridge

Span

6-12 m

Width

12-55 m

Skew

3°

Slab thickness Concrete strength, /

(2 slabs of 12-55 m with joint at centre)

400 m m 37-5 N / m m

c u

Reinforcement strength, f

410 N / m m

Carriageway width

9-3 m

Surfacing depth

120 m m

y

2

(nominal)

2

(nominal)

Reinforcement:

n A

Bottom longitudinal 25 mm at 100 mm c/c Bottom transverse

2 0 mm at 150 mm c/c

A

//V/V/V/V/Z/VZ

0.400 J

Fig. A.10.1. Park Lane Bridge (Not to scale)

217

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A.11. Bridge No. 11—Asher

Lane Bridge

Asher Lane Bridge is in Derbyshire (A38/463.0, DTP 7855). It carries the A38 trunk road over a minor road. It was constructed in 1977, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A . l l . l and shown schematically in Fig. A.11.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.11.1. Details for Asher Lane Bridge Span

22-9 m

Width

24-6 m

Skew

54°

Slab thickness

900 m m

Concrete strength, /

37-5 N / m m

c u

Reinforcement strength, f

410 N / m m

Carriageway w i d t h

2 x 9-3 m

Surfacing depth

100 m m

y

2

2

(nominal) (nominal)

Reinforcement: Bottom longitudinal 32 mm at 100 mm c/c

r A

22.9

Bottom transverse

A

2 0 mm at 100 mm c/c

24.6

Plan

Section A-A

Fig. A.11.1. River Asher Lane (Not to scale)

218

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

Bridge No. 12—Flood

A.12.

Relief Bridge

Flood Relief Bridge is in Derbyshire (A50/111.5, DTP 10477). It carries the A50 trunk road over a flood channel. It was constructed in 1975, and designed to carry 45 units of HB type loading. The bridge has a simplysupported slab deck as described by Table A.12.1 and shown schematically in Fig. A.12.1. The cover to the main longitudinal reinforcement is 30 mm.

Table A.12.1. Details for Flood Relief Bridge

8-30 m

Span Width

12-20 m

Skew

0° 700 m m

Slab thickness

30 N / m m

Concrete strength, f

cu

Reinforcement strength, f

410 N / m m

Carriageway width

7-7 m

Surfacing depth

100 m m

y

//////////////

n A

2

(nominal) 2

(nominal)

Reinforcement:

*

Bottom longitudinal 2 5 mm at 100 m m c/c Bottom transverse

16 mm at 2 0 0 mm c/c

A

•

111

mn

Fig. A.12.1. Flood Relief Bridge (Not to scale)

219

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A.73. Bridge No. 73—River Trent South Floodplain

No. 3 Bridge

River Trent South Floodplain No. 3 Bridge is in Derbyshire (Ml/187.9, DTP 432). It carries the Ml motorway over a flood plain. It was constructed in 1965-66, and designed to carry 45 units of HB type loading. The bridge has a simply-supported slab deck as described by Table A.13.1 and shown schematically in Fig. A.13.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.13.1. Details for River Trent South Floodplain No. 3 Bridge Span

9-86 m

Width

37-0 m

Skew

9°

Slab thickness

530 m m

Concrete strength, f

28 N / m m

cu

2

22

Reinforcement strength, f

250 N/mm

Carriageway width

2 x 14-5 m

Surfacing depth

100 m m

y

(nominal)

Reinforcement: Bottom longitudinal 32 mm at 76 mm c/c

37.0

Bottom transverse

22 mm at 300 mm c/c

0.530

Plan

Fig. A.13.1. River Trent South Floodplain

Section A-A

No. 3 Bridge (Not to scale)

220

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 5. WHOLE LIFE ASSESSMENT

A.74. Bridge No. 7 4 — K i r k Hallam-Sandiacre

Road

Bridge

Kirk Hallam-Sandiacre Road Bridge is in Derbyshire (Ml/196.4, DTP 455). It carries the M l motorway over the C6 minor road. It was constructed in 1966, and designed to carry 45 units of HB type loading. The bridge has a simplysupported slab deck as described by Table A.14.1 and shown schematically in Fig. A.14.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.14.1. Details hr Kirk Hallam-Sandiacre

Span Width

13-80 m 18-13 m

Skew

15°

(2 slabs of 18-13 m with joint at centre)

680 m m

Slab thickness Concrete strength, f

Road Bridge

40 N / m m

cu

Reinforcement strength, f

410 N / m m

Carriageway width

14-3 m

Surfacing depth

100 m m

y

(nominal)

2

(nominal)

2

Reinforcement: Bottom longitudinal 2 layers of 25 mm at 76 mm c/c

18.13

Bottom transverse

A

22 mm at 152 mm c/c

0.680 15° Section A-A

Plan

Fig. A.14.1. Kirk Hallam-Sandiacre

Road Bridge (Not to scale)

221

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

A . 7 5 . Bridge No. 15—Morton-Hilcote

Road Bridge

Morton-Hilcote Road Bridge is in Derbyshire (Ml/219.7, DTP 503). It carries the Ml motorway over a minor road. It was constructed in 1966, and designed to carry 45 units of HB type loading. The bridge has a simplysupported slab deck as described by Table A.15.1 and shown schematically in Fig. A.15.1. The cover to the main longitudinal reinforcement is 50 mm.

Table A.15.1. Details for Morton-Hilcote

Span W

i

d

t

Road Bridge

13-10 m h

18-00 m 6°

Skew Slab thickness Concrete strength, /

(2 slabs of 18-00 m with joint at centre)

610 m m 40 N / m m

c u

Reinforcement strength, f

2

410 N / m m

y

Carriageway w i d t h

14-3 m

Surfacing depth

100 m m

(nominal) 2

(nominal) (3 lanes)

Reinforcement: Bottom longitudinal 2 layers 2 5 mm at 76 mm c/c

18.00 A

4

A

Bottom transverse

2 2 mm at 152 mm c/c

Top longitudinal

19 mm at 6 1 0 mm c/c

Top transverse

19 mm at 305 mm c/c

6° Plan

Fig. A.15.1.

Morton-Hilcote

Section A-A

Road Bridge (Not to scale)

222

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Part 6. General risk assessment

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

An overall risk-based assessment procedure for sub-standard bridges N. K. S h e t t y a n d M. S. C h u b b , W. 5. D. Halden, Transport Research Laboratory,

Atkins

Consultants

Ltd;

Crowthorne

Introduction Bridges need to be assessed to establish limits for their safe and economic operation. In the UK, a major bridge assessment programme has identified a significant number of bridges which do not satisfy the current load carrying requirements, and the implication is that some sort of remedial action is needed. The enormity of the task is such that work has to be spread out; hence, priorities have to be set. In so doing, the costs and risks to the public should be assessed alongside other priorities on the road network. This paper proposes a framework for a risk-based approach for assess­ ment of bridges, and simplified methods for risk evaluation. The procedure will be most appropriate for bridges which do not satisfy the requirements of a standard assessment of load carrying capacity using current assess­ ment codes. The development described here is part of an ongoing research project being carried out by W. S. Atkins Consultants on behalf of The Scottish Office. It is intended to be dovetailed into the strategic research programme being conducted by the Highways Agency in the area of bridge assessment. The purpose of the proposed methodology will be to identify the low risk bridges for which traffic disruptive interim measures (i.e. measures applied prior to strengthening) may not be necessary, and to select appropriate remedial actions.

Overview of the risk-based framework A risk-based framework for assessment and prioritisation of bridges is being developed by Shetty et al. and this is summarised here. The overall pro­ cedure involves two main parts 1

• •

assessment and ranking of bridges prioritisation of bridges for remedial work.

The main steps involved are summarised below.

225

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Assessment

and ranking of bridges

Screening

In the UK screening is not necessary at present since all national bridges are being assessed in stages. However, screening can be expected to play an increasing role as bridge managers seek to optimise the efficiency of their inspection and assessment regimes.

Standard

assessment

of load carrying

capacity

This is performed for each bridge according to standard documents such as BD 21, BD 44 etc. If a bridge is found to possess the required load carrying capacity, no further action would be required.

Risk

assessment

For a bridge which does not satisfy the requirements of a standard assess­ ment (sub-standard bridge), the risk of failure is evaluated considering the probability and consequences of failure. The level of sophistication adopted for the risk analysis will vary, but simplified methods should be acceptable in the majority of situations.

Ranking

All sub-standard bridges are then ranked in terms of their risk values.

Prioritisation Remedial

of bridges for remedial

work

options

For each bridge, the effects on load carrying capacity and risk are evaluated for a number of interim and remedial options, ranging from simple monitor­ ing to extensive repair or replacement.

Whole-life

costs

For each selected option, whole life costs are evaluated including: inspec­ tion and maintenance costs; repair and rebuilding costs, and traffic delay costs.

Rank

options

The remedial options are then ranked using a multi-criteria decision analysis a p p r o a c h considering all relevant costs and risks under each option. 2

226

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

Prioritisation Using consistent information about whole life costs and risks for each bridge, remedial works can then be prioritised on the bridge stock such that the benefit (expected utility) from the money spent is maximised. The methodology for risk assessment, which forms an essential part of the above framework, is described in the following.

Risk evaluation The demarcation of the risk levels needs to be based on wider consensus within the profession. Of necessity, standards-based assessments are usually conservative and the level of conservatism can vary between different bridges. Therefore, when a bridge does not satisfy a standard assessment, it is useful to quantify the risk of failure by taking into account bridge-specific information such as actual traffic density and composition, etc. Methods for doing this are described in Part 3 of this book. The accept­ ance levels for assessments to current standards do not take account of the importance of the structure, degree of redundancy and the consequences of failure. These factors can be taken into account in a risk-based procedure. The most widely accepted engineering definition of the term risk, which is used here, is Risk = Probability

of failure

x Consequence of

failure

Probability The probability of bridge failure is influenced by uncertainties in the deter­ mination of the extreme traffic loads, the inherent randomness in material properties and uncertainties in the analysis methods used for determining load effects and capacities. Modern methods of structural reliability analysis, see Thoft-Christensen and Baker, are now widely accepted as a rational way of accounting for uncertainties in the parameters. A reliability analysis provides a number of sensitivity and importance factors which are very useful when assessing probabilities. For example, Fig. 1 shows the relative importance of several variables when related to the probability of failure of a composite steel girder/RC slab section under bending and shear failure modes. From these results it can be seen that the yield strength of steel is the most important variable for bending, while the thickness of the web is very important for shear failure. Although, such advanced reliability/risk analyses methods such as those described in Part 3 are available for evaluating the risk of a bridge failure, they are not generally suitable for use on a routine basis due to their complexity, and the general lack of accepted standards and base data. 3

227

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Bending

Fig. 1. Relative importance composite

Shear

oi variables to the probability

of failure oi a steel girder/RC

slab

deck

A simple and consistent risk scoring system is therefore being developed which could be used by engineers. Consequences 4

The various consequences of failure are grouped into four categories and evaluated as summarised below •

•

•

Human Factors affecting human consequences (principally loss of life or injury) include: the maximum number of vehicles or pedestrians which could be passing over the bridge at the time of the incident; the number of vehicles or pedestrians passing under the bridge; the type of failure (e.g. collapse of a span, or spalling concrete), and the likely effects on road accidents due to the failure (e.g. the potential for a pile up of vehicles). The warning of failure strongly influences the extent of human consequences. Environmental The environmental consequences depend on similar factors as above but, in addition, the proportion of vehicles carrying hazardous substances, the nature of the crossing (road, rail or river) and the nature of the adjacent environment (rural, urban residential, industrial, etc.) can influence the amount of pollution damage and result­ ing clean-up costs. Traffic Traffic consequences arise mainly from delays and detours caused by the failure (partial or complete) of a bridge. Traffic delay costs can be evaluated using available computer programs such as QUADRO, considering the traffic flow-rate, proportion of vehicles 5

228

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

•

diverted, average length and quality of detour, and the duration over which the traffic will be disrupted before the bridge is restored to normal service. Economic Economic consequences are evaluated by estimating the direct and indirect costs arising from failure including: repair, reconstruction, vehicle damage, environmental clean-up, legal and compensation costs.

Risk scoring system The aim of the scoring system is that risk analysis can be carried out simply and quickly for a wide range of bridges so that information about risk can be more widely used in bridge assessment, repair and strengthening decisions. By ensuring that the system is calibrated against bridge performance data, the level of sophistication adopted in the analysis can be the choice of the assessing engineer, with simple scores being adopted where possible and advanced techniques being used for more complex or unusual situations. Development of the scoring system is not yet complete but, in line with the risk evaluation concepts discussed above, it is envisaged that the risk score will be calculated by combining a probability score and a consequence score, both of which are expressed on a scale of 1 to 10.

Consequence

scores

The consequence of failure is evaluated using parametric formulae developed using advanced risk assessment techniques. The calibration of this is underway and it is planned that the consequence score will be a logarithmic function of the total cost of failure, calculated as a sum of the human, environmental, traffic and economic consequences, evaluated in terms of cost. Such consequence analysis principles are common to asset management in general, including roads and bridges, and are already familiar to most engineers. Therefore the remainder of this paper con­ centrates upon the development of a scoring system for evaluating the probability score. 1

Probability

scores

The probability score is calculated by defining a base score, and applying modifying factors to take account of the expected performance of the bridge or component under consideration. The development of the probability scores and modifying factors, is based upon the results of quantitative risk analysis for a number of representative bridge types, using results from advanced methods such as those illustrated

229

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Table 1. Bridge and component

types used for reliability

Structural form

Component

R C solid slab deck

R C slab

analyses

Failure mode Bending—midspan

(continuous)

Bending—support Shear

Steel beam deck with

Steel beam with slab

Bending

composite R C slab

Longitudinal shear

(simply-supported)

Vertical shear

Pretensioned beam deck

Pretensioned beam with

Bending

with composite R C slab

composite R C slab

Cracked section shear

(simply-supported)

in Fig. 1. In order to provide inputs for the development of the risk scoring system, comprehensive reliability analyses are being carried out on many of the most common types of bridge, although in some instances it is necessary to supplement this with appropriate engineering judgement and experience where data are not available. To date, results are available for the three types of short span bridges shown in Table 1. For each failure mode, the components were dimensioned such that the capacity ratio was exactly equal to unity, when assessed using BD 21 for the 40 tonne HA assessment loading. The results are limited to bridges of less than 20 m span. Preliminary analysis of the sensitive factors has identified that: the base score should depend upon the capacity ratio, the component type and the failure mode; and the modifying factors should depend upon several factors including the type of road, the span, pavement condition, redundancy, method of analysis, etc. The final choice of modifying factors and scores w i l l be based upon comprehensive calibration using a wide range of bridge types. However, to illustrate the scoring system, a preliminary version of the system has been developed, which is described below. The scores shown are subject to change as results of additional studies become available, and as such these should not be used at present for any practical purposes.

Probability scoring system The probability score P is numerically expressed as (10—reliability index). This provides a means of relating the score to the probability of failure as shown in Fig. 2. As the score increases, the probability of failure increases exponentially. Based on the probability score alone, it is possible to categorise the various bridges into high, medium, low and negligible risk s

230

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

Fig. 2. Proposed relation between probability

score and probability

of failure

categories. The results so far have shown that different bridge components, all with a capacity ratio of unity, can lead to risk scores in the range of 2 to 5, implying a wide variation in the probabilities of failure. A probabilistic model for traffic loading on short span bridges was developed for use within reliability analysis based on 1991 traffic data from surveys carried out on the M5 and M6 motorways in the UK. In addition, a number of parameters such as impact factor, dead load, super­ imposed dead load, analysis uncertainty, strength of steel reinforcement or tendons, strength of concrete, area of steel, cross-section dimensions, etc. have been treated as random variables and described using appropriate probability distributions. The format of the scoring system for evaluating the probability score is shown in Table 2. A base probability score is initially obtained from Fig. 3 as a function of the capacity ratio for the component type and the failure 4

Table 2. Probability scoring sheet Bridge name:

Span:

Structural form:

Road type:

Element number:

Element type:

Failure mode:

Capacity ratio:

Parameter

Score

Base probability score Modification for road type Modification for impact effect Modification for span and LLF Modification for reserve strength and redundancy Modification for inspection regime Total probability score

231

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Capacity ratio 1 0

0.9

0.8

0.7

0.6

0.5

Fig. 3. Base probability score plotted against capacity ratio (provisional)

mode considered. The base score is then modified to take into account a number of bridge-specific features, which were identified from sensitivity analysis to have a major influence on the probability of bridge failure. The modification factors given below are in the form of additions or subtractions to the base score.

•

Modification for road type The modifications are given in Table 3 for

•

three road types. These have been derived by modifying the probability distribution of the annual extreme traffic loading for the three values of traffic flow-rates assumed, as given in the table. Modification for impact effect The impact effect depends on the dynamic characteristics of the bridge and the vehicle and the roughness of the pavement. For simplicity, the modification has been expressed in terms of span and pavement condition as in Table 4.

•

Modification

for method of analysis, redistribution

and redundancy A

bridge normally possesses considerable reserves of strength beyond the theoretically computed capacity of the element considered. To account for this, values of reserve strength factor (RSF), expressed as Tab/e 3. Provisional modification score for road type

Road type

Flow-rate (vehicles/day)

AScore

Motorway

50 000

00

Major road

10 000

-025

Minor road

2000

-050

232

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

Table 4. Provisional modification

score for impact effect

Span

Pavement condition Good

•

•

Average

Poor

10 m

- 1 00

000

15 m

-0-75

-0-15

+ 1 00 +075

20 m

-0-50

-0-25

+0-50

a ratio of the ultimate load carrying capacity of the bridge to the capacity of the component, are suggested in Table 5 along with the modifications to the score for different combinations of component and system failure behaviour. If the assessment engineer estimates a different value of RSF for the particular bridge considered, the corresponding modification to the score could be used. Modification for inspection regime These are given in Table 6 for different levels of inspection and the expected warning of failure. These are based on judgement. Other /actors In addition to the above, modifications to the score are being developed for the live load factor (ratio of the live load effect to the total load effect) and lane width.

The most appropriate method for taking account of uncertainty in material properties is still being considered. Assessment codes already seek to manage this in various ways, for instance through the collection of core sample data on material properties, and the use of approaches such as worst credible strength and modified partial factors. If these approaches are carried out correctly and thoroughly then further change to the prob­ ability of failure as a result of lower material variability is likely to be small. It may therefore be that a separate modification factor for material properties is not necessary. Table 5. Provisional modification

score for method of analysis, redistribution

and

redundancy

System behaviour

Element behaviour

RSF

AScore

Element failure

Brittle

100

-0-00

causes collapse

Ductile

1-05

-0-50

Gradual

1-15

-100

Element failure causes

Brittle

1-15

-100

significant damage

Ductile

1-20

-1-25

Gradual

1-25

-1-50

Element failure causes

Brittle

1-30

-1-75

local damage only

Ductile

1-40

-2-25

Gradual

1-50

-2-50

233

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Table 6. Provisional modification score for inspection Inspection level Uninspectable element Routine principal inspection Special inspection before assessment

Continuous monitoring

regime

Warning of failure

AScore

Little or none

+0-50

Significant

+0-25

Little or none

-000

Significant

-0-25

Little or none

-025

Significant

-0-50

Little or none

-0-25

Sufficient

-050

Significant

-100

Summary and conclusions Recent bridge assessments in the UK have identified a significant number of bridges for which the rated capacity is unsatisfactory for service requirements. In view of the large numbers involved, bridges have to be priori­ tised for strengthening works. In addition, risks need to be assessed for those bridges for which strengthening may not be possible in the near future. The paper has proposed a risk-based framework for prioritising bridges for remedial works. Both simplified and advanced methods of risk evalua­ tion have been proposed for use within this framework. A simplified risk scoring system is being developed so that risk analysis can become practical in a much wider range of situations. An approach for evaluating the probability scores has been proposed based on the results of comprehensive reliability analyses (Level 5 assess­ ments, see Parts 2 and 3). At present, results for three of the most commonly used bridge types are available and an illustrative probability scoring system is presented based upon these results. The system involves obtaining a base probability score as a function of the capacity ratio from a series of graphs for the different component types and failure modes. Modifications to the base score have been given to take into account a number of bridgespecific features such as: the type of road, span, pavement condition, live load factor, load redistribution, redundancy and inspection regime. The proposed system allows bridges which have failed the standard assessment requirements to be classified into high, medium, low and negligible risk categories.

Acknowledgements The permission of The Scottish Office, the Transport Research Laboratory and W . S. Atkins to publish this paper is gratefully acknowledged. The

234

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

contents of this paper are the responsibility of the authors and do not necessarily represent the views or policies of The Scottish Office or the Transport Research Laboratory. The authors would also like to acknowledge the contribution of colleagues at W. S. Atkins—Dr N. C. Knowles, Mr R. D. Bellamy and Miss K. Flaig. The copyright of the paper remains with W. S. Atkins Consultants and the Transport Research Laboratory.

References 1. Shetty N. K. et al (1996). A risk-based framework for assessment and prioritisation of bridges. 3rd Intnl Conf. on Bridge Management, Surrey. 2. Shetty N. K. et al (1996). Risk considerations in decision-making for re-assessment of bridges and planning of remedial works. 15th IABSE Congress, Copenhagen, Denmark. 3. Thoft-Christensen P. and Baker M. J. (1982). Structural reliability theory and its applications. Springer-Verlag. 4 . Shetty N. K. et al (1995). Performance based capacity of bridges and other structures: Final Report. W. S. Atkins Report No. AST/M2832/FINAL. 5. Department of Transport (UK) (1982). QUADRO—QUeues And Delays at ROad works, Version 2, Users Manual.

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Probabilistic risk assessment of corrugated steel buried structures—comparison of reliability between standard span and long span structures H. A. D. Kirsten, Steffen

Robertson and Kirsten (UK) Ltd

Introduction Corrugated steel buried structures (CSBS) have been in use in the United Kingdom in various forms for many decades. The smaller structures are largely circular or near circular in form and are usually enclosed by the corrugated steel sheeting. Arch forms known as super-span structures have been introduced in recent years to span wider crossings. The Highways Agency has an existing design standard, BD 12/88, for CSBS up to 8 m span. A new design standard, BD 66/95, has been drafted to cover long span structures up to a maximum span of 14 m. In view of the lack of long term and widespread use of this size of CSBS to date, doubts were raised about the reliability of the super-span structures. Doubts were alternatively expressed that the new standard may be unduly conservative. These concerns in addition raised a general interest in the relative reliability of CSBS compared to other bridge types. The aims of the project were to 1

•

• •

evaluate the comparative reliability of long span structures designed in terms of the draft standard and short span structures designed in terms of the existing standard evaluate the nature and degree of possible undue conservatism in the proposed draft standard evaluate the relative reliability of CSBS compared to other bridge types.

The findings were to include recommendations on possible improvements in the design criteria and methods, and on further research that may be appropriate.

Reliability measures considered Design reliability may be evaluated in terms of two measures, i.e. probability of failure and reliability index. The probability of failure may be defined as the likelihood that the probability density function of the difference between

236

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

the capacity and demand with regard to a particular design aspect is negative. The reliability index is defined as the ratio between the mean and the standard deviation of the probability density function of the difference between capacity and demand. The standard deviation of the probability density function of the difference between capacity and demand may be expressed in terms of the standard deviations of capacity and demand and the coefficient of correlation between capacity and demand.

Design aspects considered The design aspects in BD 12/88 comprise buckling of the shell, compression of the shell, shearing of the bolted seam, increase in span and soil pressures at the top, corner and bottom of the structure. In BD 66/95, the design aspects are buckling of the plates above the thrust beam, compression of the plates below the thrust beam, shearing of the seams above and below the thrust beam, shearing of the thrust beam and the thrust beam connection, live load crown deflection and competence of the foundation soils. Seam strength and buckling of the crown plates are the aspects in CSBS construction that are subject to the highest likelihood of failure. The investigation was accordingly undertaken in terms of these aspects.

Description of work The conservatism in the draft standard was investigated for three different shapes, two spans and three depths of cover of CSBS construction. The different shapes comprised ellipses and low and high profile arches. The spans correspond to 8m and 13-5 m and the depths of cover to 2 0 % , 3 5 % and 5 0 % of span. Fig. 1 shows typical design cases. The degree of conservatism in the draft standard for 8 m span was evalu­ ated in terms of reductions in the safety margins and increases in plate thick­ ness that would result in probabilities of failure for seam strength and plate buckling that would correspond to those determined for the cases analysed in terms of the existing standard. The effects of span and depth of cover on the degree of conservatism in the draft standard were evaluated in a similar way.

Comparative evaluation of standards The essential differences between the two standards with regard to the purposes of this study may be presented briefly as follows •

The demand effects are completely different and are determined by very different methods in the two standards. The terms in the draft

237

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

7760 Scale 1:150 (a)

238

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

(e)

Fig. 1. (d) case 4—super-span

•

high profile arch (ref: 44-A-9-16); (e) case 5—super-span

high

standard are far greater in number and complexity. The calculational model in BD 12/88 is based on a hydrostatically loaded circular section and in BD 66/95 on a set of empirically observed characteristics for arched sections assisted across the crown by thrust beams. Allowable deflections are 5% and 0*4% of the span in the existing and draft standards respectively. The specifications for construction of the trench and fill in the existing standard are not compatible with allowable deflections of 5% of the span. The calculational model in the draft standard in addition does not accurately represent the interaction between the structure and the surrounding soil. It is questionable in view of these uncertainties whether the provisions in BD 12/88 constitute a reliable basis against which BD 66/95 can be evaluated.

239

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

Description of workbooks Two spreadsheet-based workbooks were developed to enable the probabil­ ities of failure and reliability indexes for the various design aspects in the two standards to be determined. The workbooks each comprised three worksheets respectively for capturing the input data, performing the various calculations and for reporting the results. The distribution functions for the various input parameters were specified in the first worksheet in terms of custom-built single command statements. The means, standard deviations and histograms of the output values for demand and capacity were determined for a specified number of iterations in the second worksheet. The correlation coefficient for the output distribu­ tions of capacity and demand was also calculated at this stage. The results presented in the third worksheet comprised averages for capacity and demand and probabilities of failure and reliability indexes for the various design aspects.

Software employed The workbooks were designed in terms of EXCEL, version 5, and the statistics underlying the probabilities of failure and reliability indexes, determined by means of @RISK, version 3.1a. BESTFIT was employed to fit curves to the resulting histograms. Three separate goodness-of-fit tests were used in BESTFIT to evaluate the accuracy with which the probability of failure was calculated.

input values for parameters The values for the parameters were all assumed to be normally distributed. The mean values for the parameters that represent the shapes and dimen­ sions of the structures and of the surrounding fills, and the strengths and stiffnesses of the steel components, were obtained from manufacturers' design guides. The mean values for the properties of the fills were taken to represent typical conditions. The standard deviations were in most instances estimated to represent the expected variations in the parameters. Sufficient information was available in the case of some parameters to enable the standard deviations to be properly calculated. The mean of the HA, HB and construction loads were determined from simplified criteria relating the pressure at the top of the structure to the depth of overburden.

Findings and recommendations The existing standard was found to be adequate with regard to seam strength in that the corresponding probabilities of failure for the various

240

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

shapes and depths of cover were very small. The probabilities of failure for seam strength for structures designed in terms of the draft standard, were of considerable magnitude for different spans, shapes and depths of cover, which confirmed the conservatism of the draft standard with regard to seam strength. The degree of conservatism in the draft standard for seam strength was on average 29% for 8 m span and reduced on average to 18% for 13-5 m span for the horizontal ellipse and high profile arch. The draft standard was not found to be conservative for seam strength at either 8 m or 13*5 m span for the low profile arch. The probabilities of failure for seam strength for 8 m span for the high profile arch were small and did not vary with depth of cover up to 50% of span in terms of the existing standard. These probabilities and the corre­ sponding degrees of conservatism for 8 m and 13*5 m spans did, however, increase slightly with depth of cover in terms of the draft standard. The existing standard was also found to be adequate with regard to plate buckling in that the corresponding probabilities of failure for the three shapes considered were very small. Regarding plate buckling design, the draft standard was less conservative for both 8 m and 13*5 m spans than the existing standard. The degree of under conservatism in the draft standard for plate buckling was on average 25% for 8 m span and 1 1 % for 13-5 m span for the horizontal ellipse. The corresponding figures for the high profile arch were 4 % and 6*5%. For the low profile arch, the draft standard gave a degree of under conservatism for plate buckling of 1 1 % for 8 m span and a degree of conservatism of 4*5% for 13*5 m span. The relative under conservatism in the draft standard compared to the existing standard for plate buckling is very much affected by cross-sectional shape. This is due more to differences in curvature of the top plates than to overall shape. The top plates in the structures considered in terms of the draft standard are less curved than those considered in terms of the existing standard. This finding is peculiar to the standard shapes manufactured by Asset International for super-span structures. The relative under conservatism for plate buckling in the draft standard is, however, not a cause for concern, because the relevant probabilities of failure are small. Therefore, although the comparative degree of conserva­ tism for seam strength in the draft standard is almost as much as the degree of under conservatism for plate buckling in some instances, long span structures designed in terms of the draft standard are reliable. The draft standard is only conservative with regard to seam strength and then to a comparatively limited extent. It would not be advisable to reduce this degree of conservatism without further specific investigation and observation. It would be important in further studies to extend this evaluation to the products of a broader

241

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

range of manufacturers. It w o u l d also be advisable in d e s i g n s in terms of the draft s t a n d a r d to determine the probabilities

of failure for the various

aspects considered a n d to evaluate these in terms of internationally accepted n o r m s . T h e m e a n s a n d s t a n d a r d deviations for the input parameters will require to be determined reliably for this p u r p o s e . It is not p o s s i b l e in terms of the s t u d y u n d e r t a k e n to determine the relative reliability of C S B S construction compared to other types of bridges. M o r e studies of the k i n d considered are required for these p u r p o s e s . Accurate numerical simulations of the l o a d s a n d deflections for the v a r i o u s cases will, furthermore, be required i n this regard. C o m p a r i s o n s of the results of s u c h numerical models w i t h the observed behaviour of properly mented structures will provide invaluable i n s i g h t s into the

instru­

fundamental

p r o v i s i o n s of the two d e s i g n s t a n d a r d s .

Acknowledgements T h e assistance a n d s u p p o r t given to the project b y D r P a r a g D a s a n d D r D a v i d B u s h from the H i g h w a y s A g e n c y are sincerely

appreciated.

M r D e n i s S m i t h of A s s e t International provided considerable information, a r r a n g e d site visits a n d attended a n u m b e r of meetings, all of w h i c h are gratefully a c k n o w l e d g e d . T h a n k s

are due to members of staff of

Transport

in

Research

Laboratories

Crowthorne;

Partners a n d Flint & Neill Partnership for

Sterling

the

Maynard &

extensive d i s c u s s i o n s

and

information provided.

Reference 1.

B D 1 2 (1988). T h e design of c o r r u g a t e d steel buried s t r u c t u r e s . Design m a n u a l

bridges.

HMSO,

for roads and

London.

242

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

Probabilistic assessment of bridge scour and other hydraulic risks A. McCracken, Steffen

Robertson

and Kirsten

(UK)

Ltd

Introduction A significant proportion of trunk road and rail bridges in the UK are over water. In many cases this means that the opportunity for inspection is restricted when compared with bridges over land. Consequently, quantita­ tive assessment of any risk posed to the bridge is often difficult and the prioritisation of maintenance works is not straightforward. In the past decade a few bridges have failed with a consequent loss of life highlighting the requirement for a means of having foreknowledge of any threat posed to a bridge. The problems of inspection and evaluation have long been recognised and the Highways Agency have let a number of contracts investigating the situation. Increasingly, the assessment methodologies have turned to probabilistic methods to determine and rank the risk of failure of a bridge system. Steffen Robertson and Kirsten are currently finalising a Highways Agency contract entitled Probabilistic assessment of hydraulic action on bridges. This paper presents the approach adopted to quantitatively determine the hydraulic risk, and in particular the scour risk, the data requirements and the spreadsheet models developed and the application of the methodology to a variety of test case bridges. The work has integrated the impacts of the hydraulic action with (a) the capacity of the engineering geological environment of the river bed and bank to withstand scour (b) the interaction of scour with the various bridge foundation types.

Objectives of study The main objective was the development of a quantitative procedure to evaluate the risk of scour and other hydraulic related failure of a bridge system. Ultimately, this would be translated into a practical manual for use by bridge inspection engineers to identify and prioritise risks. This objective was accordingly met by •

development of a classification method for hydraulic risks

243

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

•

• •

development of quantitative risk assessment for various failure modes affected by hydraulic conditions, e.g. collapse of piers, abutments, deck or approach caused by foundation failure, structural failure, instability or washaway production of a user manual for a hydraulic risk-based evaluation procedure proving of the procedures by applying them to a variety of test case bridges.

Methodology The direct aspects of hydraulic action have been well studied by others and formulae for depths of general and local scour, and forces on piers and superstructure have been developed. The components of the risk methodology adopted were • • •

identification of the hydraulic risks and bridge failure modes evaluation of the hazards through capacity-demand models interpretation of the results through interactive spreadsheet programs.

The main event is the failure of a bridge system. This would generally result from failure of one or more of the integrated sub-systems of the bridge, i.e. failure of the piers, abutments, deck or approaches. Within each of the bridge sub-systems there are numerous components. Failure of these components would lead to failure of the sub-system. The focus of this study is confined to the risk of failure of foundations and embankments under the threat of scour. The method used is based on deterministic capacity-demand models and probabilistic interpretation of the variations of the parameters pertaining to the hydraulic system, the bridge structure and the soil sub-systems. For each of these events capacity and demand functions can be deter­ mined. Failure occurs where demand exceeds capacity or alternatively where the safety margin (capacity less demand) is negative. The safety margin is itself a probability density function. The probability of failure is the likelihood that the safety margin is negative.

Description of work The work undertaken is based on other studies on flood hydrology, river engineering, bridge site hydraulics, bridge design and bridge scour. In parti­ cular, reference was made to Binnie and Partners work for the Highways Agency reported in Advice Note Assessment of scour at highway bridges and the H. R. Wallingford report to the Transport Research Laboratory entitled Hydraulics of highway structures. In particular, the work of Neill 1

244

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

which relates the average river bed particle size to depth of scour and velocity for a particulate medium, is integral to the methodology. The major problem confronting the bridge inspection engineer is that it is often impractical to see or measure the occurrence and extent of scour. Furthermore, scour may have taken place and subsequently been infilled and obscured by low strength material. T h e work now undertaken, however, shows that the existence of previous scour is not a critical factor. The risk of failure due to scour can be measured relative to the undisturbed state prior to the occurrence of any scour damage. The fact that subsequent scour may have occurred at a bridge site does not increase the initial risk of failure. It is simply significant of an inherently high risk of failure to that particular structure. In other words, the extent of the scour

does not form part of the risk analysis measured accurately.

and therefore does not have to he

The probability distribution of depth of scour however is directly related to the river hydraulics through the probability of occurrence of flood events. The interaction between the probability of failure of a foundation sub-system at a particular depth and the probability of occur­ rence of scour, defines the probability of failure of the bridge at that depth of scour. The probabilities of failure were assessed for foundation failure, hydraulic conditions and forces caused by loading.

Loading

criteria

The structural loading applied to a bridge structure consists of the combina­ tion of thrusts and forces on the components as follows • • •

piers: hydrodynamic, debris, impact loading abutments: hydrodynamics, retained earth loading bridge deck: traffic loading, hydrodynamic loading.

The combination of loading applied to the foundations of the structure due to these loads represent demand on the structure.

Foundation

failure

criteria

The various capacities, demands and parameters for spread foundations and piled foundations require to be identified. T h e likelihood of failure through any of the mechanisms were then identified and modelled in terms of closed form solutions (in simple cases) with random variation of parameters. T h e probability of bridge foundation collapse, P , due to hydraulic action (Fig. 1) is the product of the probability of scour, P , to a particular depth and the probability of structural failure, P f , at that depth. c

s

245

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

1.2 -

« o =

0.8 0

6

0.4 0.2

1

2

3

Depth of scour, ds, m (a)

1

2

3

4

Depth of scour, ds, m

(b) Fig. 1. (a) probability

Hydraulic

of foundation

failure; (b) probability

of depth of scour

criteria

Estimates of flood flows both for river flow and tidal flow are required together with calculations of depth and flow velocities for the river channel both upstream and through the bridge. A classification of overbank flow is also required based on an overall flood plain rating system and flood plain factors. The demand effects of each of these w a s assessed. The probability of a flood flow Q, velocity V and depth of flow d, can be calculated from river hydraulics theory subject to the constriction to flow of the bridge structure. The estimation of scour depth d related to particular flood flow events can then be determined from the relationship derived by Neill (Fig. 2). The figure derives the relationship between particle size p (defined by the D of the bed material), depth of scour d and competent velocity (the velocity at which a scour depth will be attained). The probability, P , of a scour depth being attained (Fig. 1(b)) is related directly to the probability of the occurrence of the flood event. s

1

5 0

s

s

246

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

0.3

0.5 0.7 T

1—I—I

.001

1.0

2

I I I

.002

1

1

.005

3 1

5

1

1

1

.01

7 1

I

10 I

20

I I

1

.02

30

I

I

I

.05

50 I

I

.10

70 I

I

.20

100 I

I

200 I

.50

I

300 I

1

1.0

Bed material grain size, ft

Fig. 2. Relationship between partical size, scour depth and flow velocity at which scour attained (Neill

depth

diagram)

In estimating general and local scour depth, the effects of river geometry, pier shape factors and direction of flow were taken into account.

Probability

of bridge

failure

The product of the probability of depth of scour, P , and the probability of structural foundation failure, P , defines the probability of bridge support collapse, P (Fig. 3). The probability of bridge support collapse varies with the depth, of scour and it can be shown that it has a maximum turning s

c

f

0.003

t

\\

0.0025

6

0.002

"5

0.0015

\

| | o

-seriesl

0.001

\

/

0.0005

0.5

1

1.5

2

2.5

\

3

3.5

Depth of scour, ds, m

Fig. 3. Probability of bridge

collapse

247

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

SAFETY OF BRIDGES

point that reflects the distribution of peak river flows at the bridge site based on the relationship between depth of scour and flow rate.

Data input Data capture worksheets have been defined to suit all bridge types and river conditions. The main input data required for the risk assessment comprises • configuration and depth of the foundation • geotechnical characteristics of the underlying materials • river hydraulics of the bridge site • flood hydrology. The data input sheets require the following to be gathered • • • • •

Geotechnical data: soil/rock profile for founding materials and abut­ ment fill Foundation data: piers and abutments Substructure: piers and abutments Superstructure: design and loadings Hydraulics: hydrology and hydraulics

•

River borne loading: silt, armour stone, ships, ice, debris, tides, etc.

Prototype applications Seven bridges in total were examined. These comprised a variety of conditions, types of construction and foundations including river and tidal situations; single and multi-span; spread footings and piles; iron trestle, concrete piers and brick arch within a variety of geotechnical conditions. Bridge design and river hydraulic data is generally available. Geotechnical data fundamental to foundation analysis and scour resistance is however less available, especially for structures over 25 years old. To overcome data deficiencies probabilistic methods again come into use to determine likely parameter values and distributions. Knowing geology and probable founding soil types together with information from published sources and various records, estimates of strength and other input parameters can be obtained. By conducting sensitivity analyses on the less well-known parameters an understanding of the criticality of the parameters, and hence the risk, is obtained. Conclusions and findings The probability of failure measurement has been shown to lend itself to the assessment of comparative risk for bridges. The relative probabilities of

248

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.

PART 6. GENERAL RISK ASSESSMENT

failure for a number of bridges can be utilised both when prioritising maintenance and remedial work. The complexity of most bridge systems in terms of the variables of design, failure mechanisms and input parameters was clear from an early stage in the study. The complexity is such that standardised assessment within a simple manual is probably not appropriate. Each bridge requires to be evaluated individually and then the more likely failure mechanisms selected from the worksheets and examined for their risk.

Acknowledgements The authors sincerely appreciate the assistance provided by Dr Parag Das, Mr Yogesh Patel and Mr Alan Kirkdale of the Highways Agency, Dr Steven Magenis of Posford Duvivier who assisted with the hydraulic aspects of the project and the County Council and Environment Agency Officers of Nottingham, Shropshire and Suffolk for providing the test case data.

Reference 1.

Neill C. R. (1973).

Guide to bridge

C a n a d a , U n i v e r s i t y of T o r o n t o

hydraulics. Roads

and Transportation

Association

of

Press.

249

Downloaded by [ UNIVERSITY OF WISCONSIN - MADISON] on [10/03/24]. Copyright © ICE Publishing, all rights reserved.