Analysis and Testing of Bridges

The Institution of Structural Engineers

The Institution of Structural Engineers 11 Upper Belgrave Street, London SWlX 8BH

01712354535

The Institution of Structural Engineers

Analysis & testing of bridges

Papers presented at a 1-day seminar organised jointly by the

Institution of Structural Engineers and the

Society for Earthquake and Civil Engineering Dynamics

Wednesday, 26 April 1995 London

The Institution of Structural Engineers 11 Upper Belgrave Street, London SWlX 8BH

01712354535

Organising committee

David Doran, BSc, DIC, FCGI, CEng, FIStructE, FICE (Consultant) Chairman George Davison, BSc, CEng, MIStructE, MICE (Consultant) Frank Jones, MEng, CEng (Consultant) John Maguire, BSc, PhD, CEng, FIStmctE, MICE (Lloyd’s Register) Andy Lorans, BSc(Eng), ACGI (The Institution of Structural Engineers) Secretary I

Published by SET0 Ltd, 11 Upper Belgrave Street, London, SWlX 8BH.

ISBN 1 874266 33 6

0 1997 The Institution of Structural Engineers

The Institution of Structural Engineers and the members who served on the Committee which produced this report have endeavoured to ensure the accuracy of its contents. However, the guidance and recommendations given in the report should always be reviewed by those using the report in the light of the facts of their particular case and specialist advice should be obtained as necessary. No liability for negligence or otherwise in relation to this report and its contents is accepted by the Institution, the members of its Committee, its servants or agents.

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means without prior permission of the Institution of Structural Engineers who may be contacted at 1 1 Upper Belgrave Street, London SWlX 8BH.

2 IStructWSECED seminar ‘Analysis & testing of bridges’

Contents

Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . , . . . . . , . . . . . . . , . . . . 5

Testing to enhance structural analysis and appraisal: an overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Dr J. G. M. Wood Structural Studies & Design Ltd

Long-span bridges under dynamic loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Professor R. T. Severn Dr C. A. Taylor A. Vann University of Bristol

Seismic response of RC bridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Professor A. Elnashai Imperial College, London

Recent experiences on concrete and cast iron bridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Dr D. MacKay Lloyds Register

Potential applications of NDT methods for bridge assessment and monitoring . . . . . . . . . . . . . . . . . . . . . . . . 21 Dr P. C. Das The Highways Agency N. C. Davidson C. Colla University of Edinburgh

Arch bridge testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Dr W. J. Harvey University of Dundee

Investigation of post-tensioned structures, particularly considering intrusive inspections of tendons and their ducts . . . 33 S. W. Kemp Technotrade Ltd

Assessment of the dynamic behaviour of the Aberfeldy GRP cable-stayed footbridge . . . . . . . . . . . . . . . . . . . . 38 R. Pimentel Professor P. Waldron Universiq of Shefield Dr W. J. Harvey University of Dundee

Recent experience in dynamic monitoring of a multispan bridge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 G. P. Roberts W. S. Atkins, Bristol

Attendance list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

IStructBSECED seminar ‘Analysis & testing of bridges’ 3

Foreword

The Institution’s Chief Executive, Dr John Dougill, in his briefing to me emphasised the need to achieve a seminar programme with a fair balance between traditional and more advanced methods of testing. I believe the Organising Committee and the speakers responded well to this challenge.

This topic is of great interest to employing authorities and the consultants and contractors who seek to serve them. This is particularly so for those involved in bridge assessment which flows from the need to consider heavier loading, deterioration and other phenomena. Most bridge engineers have experienced the gap which may exist between the theoretical consideration of a design and observed performance in the field. Testing can play a major part in helping to reconcile these divergences.

The discussion at this seminar was wide ranging with anxiety expressed that assessment contracts were being let at low prices which precluded a sufficiently high professional approach. With this went the implied request for employing authorities to review their specifications for assessment and in particular their criteria for obtaining adequate quality.

Some speakers felt that bridges fell into two distinct categories - ‘very satisfactory’ and ‘those causing real concern’ with few cases between these extremes. In a general obser- vation one speaker suggested that 20% of bridges exhibited significant problems.

Enthusiasm was expressed about the potential of dynamic testing if the techniques could be refined and results presented in ways that non-specialist civiVstructura1 engineers could understand.

By general consent the seminar was a success (75 attendees). It is hoped that these published proceedings will provide engineers with a useful reference. In some cases authors have taken the opportunity to enhance the papers circulated at the time of this seminar.

I would like to thank all those who took part in the event, in particular, the organising committee, the two chairmen Brian Simpson and John Maguire, the speakers and Andy Lorans and his staff.

David Doran Chairman of Organising Committee

IStructUSECED seminar ‘Analysis & testing of bridges’ Previous page

is blank 5

appraiial: Testing to enhance structural analysis and

an overview

J. G. M. Wood, BSc, PhD, CEng, FIStructE, MICE Structural Studies & Design Ltd

Introduction The simplified methods and idealisations used in normal design and structural appraisal are necessarily conservative. They can give a misleading estimate of strength for certain types of structure. The testing of full-sized or model structures in the laboratory or in the field enables us to improve our estimates of strength. Tests incorporate many features which are simplified out in analysis. Tests have application in both the development of improved design and appraisal rules and in the appraisal of specific structures which do not comply with current Standards andlor are deteriorating.

Often it is only necessary to test parts or core samples of structures to provide improved data for incorporation in analysis, but the statistical analysis of the variability of samples must be utilised in interpreting the data. In the assessment of structures with potentially serious problems there should be a cycle of analysis, testing, then more refined analysis, etc. This provides a cost effective route to more reliable assessment, often with substantial savings in remedial works cost and a reduction in traffic disruption.

The development of sophisticated, but highly idealised computer analysis of structural behaviour and the introduction of design standards based on gross simplification of real structural behaviour, have left many engineers with little awareness of the real behaviour of structures. Although the primary duty of structural engineers is to design so that failures do not occur, many engineers have little experience of the detailed analysis of real structural failures produced by deliberate testing or accidental failures. This makes it important that testing to failure is a central element in structural engineering training and that the analyses of service failures of structures are fully published’. 2 , 3.

Why test? The testing of structures provides opportunities for Consulting Engineers, Contractors and Owners of structures to work with specialists in academic, research and specialist testing organisations to cost effectively resolve practical problems. When particular difficulties arise with the assessment of a structure, the first step in refining the normal assessment procedures is to refer to the research literature. Sometimes published research will be directly applicable to the problem and can be used directly. More often the configuration of the structure is significantly different to that on which data has been published. The next stage is to identify from the literature the best research worker or testing organisation in the particular field and, with the Client’s agreement, engage them as a specialist sub-consultant.

Their knowledge and data will usually be more comprehensive than the published literature, which is invariably summarised. They should also be aware of research in progress in the UK and internationally, which may resolve the difficulties in appraisal. If the potential costs involved in remedial work to the structure are substantial, the cost of obtaining further information from testing the structure in the field, or replicas of the structure in the laboratory, should be estimated. The costs and possible benefits of a testing programme should be discussed with the Client and a clear brief and budget agreed.

The benefits to the research community of direct involvement in the engineering and operational management of bridge

IStructE./SECED seminar ‘Analysis & testing of bridges’

structures are substantial. Structure specific testing can often lead on to separately funded more general research programmes. The benefits to the Client and Highway Manager of more reliable and rapid assessment of the condition of the bridge stock from testing, will often be 10 or 100 times the cost of a well-targeted and managed testing programme. Testing programmes, unlike traditional University research, can produce reliable information in months.

Over the last 20 years I have developed and applied the above approach, often with former colleagues at Flint & Neill or Mott MacDonald, to a diverse range of bridge structures, some of which are listed and referenced below.

Examples The collapse of steel box girder bridges at Milford Haven, Yarra and Koblenz initiated a major testing programme4, combined with the development of more refined analysis. This lead to the development of SBG6A Memson Appraisal Rules and BS 5400, Part 3.

Testing during the early studies of deterioration of the hangers on the Severn Bridge included measuring the hanger loads during removal, the testing for the load elongation characteristics of the hangers to failure, and the very detailed analysis of fracture and corrosion characteristics of individual strands by RAE Farnborough. In situ stresses in the hangers were determined from data on strains in a steel bar fitted in place of a short hanger, which were logged and analysed by Bristol University.

Corrosion problems suspected in half-joints on motorway bridges were not detectable using conventional surface half-cell corrosion cells. New equipment was developed6 with Aston University, to measure half-cell potentials at different depths up holes drilled to obtain chloride analysis samples from the soffit. This technique of sampling and measuring from below has substantially reduced the need for traffic closures and has precisely detected the location of severe corrosion.

Hundreds of motorway bridges constructed in the 1960s incorporated ‘Freyssinet’ type hinges in the decks. Concern about the ultimate load and fatigue behaviour of these joints lead to the construction of three full-size replicas which were tested at Wimpey laboratories. Computer analysis suggested that they were badly overstressed, but testing showed them to be robust under fatigue and ultimate loads. Subsequently the specimens were used for trials of high pressure water jetting and epoxy injection.

Some early motorway bridges included flat slabs with reinforcement detailing above columns which does not meet current requirements and with low ‘assessment’ strengths7. The testing of a series of one-third scale replicas at Birmingham University permitted both the ultimate punching shear load behaviour and the load deformation characteristics of these slabs to be determined. These results justified the continuation of the use of a number of structures for full lorry traffic, despite non-compliance with simplified assessment rules. These tests lead on to more detailed researchK.

The decision to specify a high quality epoxy-coated reinforcing bar cage for the tunnel segments for Storebaelt tunnel9 raised questions on the bond, cracking and ultimate behaviour of the joints between tunnel segments. A series of load tests on model joints at BCA rapidly produced data for comparison with similar tests on model joints at British Cement

Previous page is blank

7

Association (BCA) rapidly produced data for comparison with similar tests on conventionally reinforced units which confirmed the suitability of the design.

The development of procedures for the assessment of structures with Alkali-Silica Reaction (ASR)’O,’ I required a range of new tests. These included free and restrained expansion testing12 to provide quantitative data for the stress analysis of structures and compressive13 and tensile14 stiffness and strength testing. Load testing in situ and of elements removed from structures has been applied to assess strength reserves parts of several bridges with ASR15.

Simple procedures have been introduced for monitoring movements across joints in structures, either continuously with transducers, at regular intervals using Demec or vernier gauges, or using slide gauges to record maxima and minima. These have been compared with analytically estimated values to estimate structural behaviour and the condition of bearings. Demec gauge crack monitoring has enabled long term trends of crack movement to be reliably recorded in structure with ASR.

The premature deterioration of concrete structures, especially when exposed to chlorides, has lead to the development of improved testing for concrete chloride resistance for new construction using the bulk diffusion test16. However, the achievement of durability reliabilityi7, comparable to that for structural reliability, will need a major programme of testing in the field and laboratory to develop quantitative durability design.

Conclusion Testing is a valuable, but under-utilised tool for appraisal and design. The over-simplification of our codes and analysis methods and the difficulties in applying them to unusual details incorporated in old and deteriorating structures can lead to gross errors in estimates of strength. We also need more reliable test data on the overall failure behaviour of innovative structures, so that new materials and design configurations can be reliable and cost-effectively introduced with matching improvements to our methods of structural analysis. However we should not lose sight of the fact that in many areas, like post-tensioned concrete bridgesis, satisfactory testing procedures have yet to be developed.

References 1.

2.

3.

4.

5.

6.

7.

8.

9.

8

Sibley, P. G. & Walker, A. C.: ‘Structural accidents and their causes,’ Proc Inst. Civ. Engrs Pt 1, 62, pp 191-208, 1977 Report of the Royal Commission Failure of West Gate Bridge, Melbourne, 1971 Woodward, R. J. & Williams, F. W.: ‘Collapse of Ynys-Y-Gwas Bridge,’ West Glamorgan. Proc. Inst. Civ. Engrs, Pt 1,4. August, pp635-669, 1988 Institution of Civil Engineers ‘Steel box girder bridges,’ Proc. Int. Con$ ICE. London, 1972 Rockey, K. C. & Evans, H. R.: The design of steel bridges. London. Granada 198 1 Wood, J. G. M., Ecob, C. R., Page, C. L. & Lambert, P.: ‘The development and application of new inspection techniques for corroding reinforced concrete structures,’ Structural Faults and Repair Conference, London, June 1987 Wood, J. G. M., Johnson, R. A. & Ellinas, C.: ‘Reliability analysis applied to deteriorating bridge structures,’ International Conference on Bridge Management, University of Surrey, March 1990 Ng, K. E.: ‘Effect of Alkali-Silica Reaction on the punching shear capacity of reinforced concrete slabs,’ PhD Thesis, University of Birmingham, 1991 Ecob, C. R., King, E. S. & Rostam, S.: ‘Epoxy coated reinforcement cages in precast concrete segmental tunnel linings - Durability,’ 3rd Int. Symp. Corrosion of

10.

11.

12.

13.

14.

15.

16.

17.

18.

Reinforcement in Concrete Construction, The Society of Chemical Industry, May 1990 The Institution of Structural Engineers, Structural effects qf Alkali-Silica Reaction: Technical guidance on the appraisal of existing structures, London, 1992 Wood, J. G. & Johnson, R. A.: The appraisal and maintenace of structures with alkali silica reaction. The Structural Engineer, Vol. 7 1, No 2, January 1993 Wood, J. G. M., Norris, P. & Leek, D. S.: ‘Physical behaviour of AAR damaged concrete in structures and in test conditions,’ 8th International Conference on Alkali-Aggregate Reaction, Kyoto, Japan, July, 1989 Wood, J. G. M., Chrisp, T. M. & Crouch, R. S.: ‘The stiffness damage test - a quantative method of assessing damaged concrete,’ IStructEBRE Conference, The Life of Structures, Brighton, April, 1989 Wood, J. G. M., Norris, P. & Barr. B. I. G.: ‘Concrete shear strength determination using cores subject to torsional loading,’ Magazine of Concrete Research, December 1990 Wood, J. G. M., Johnson, R. A. & Abbott, R. J.: ‘Monitoring and proof load testing to determine the rate of deterioration and the stiffness and strength of structures with AAR’, IStructEBRE Seminar - Structural assessment based on full and large scale testing, Watford, April 1987 Wood, J. G. M., Wilson, J. R. & Leek, D. S.: ‘Improved testing for chloride ingress resistance of concretes and relation of results to calculated behaviour,’ 3rd International Conference on Deterioration and Repair of Reinforced Concrete in the Arabian Gulf, Bahrain S E and CIRIA, October 1989 Wood, J. G. M.: ‘Towards quantified durability design for concrete.’ Paper to Structures in Distress. QMWC Symposium, 31st January 1995, pp139-159 in Proc. French, W. J., Ed. Improving Civil Engineering Structures - Old and New. Geotechnical Publishing Ltd. Basildon 1994 Wood, J. G. M.: ‘The significance of site investigations, post-tensioned concrete bridges, assessing the risks,’ Lindsell Seminar, New College. Oxford 10 January 1994

IStructWSECED seminar ‘Analysis & testing of bridges’

Long-span bridges under dynamic loads

R. T. Severn, PhD, FEng, FICE C. A. Taylor, BSc, PhD, CEng, MICE A. Vann Earthquake Engineering Research Centre, University of Bristol

Introduction For more than ten years earthquake engineering research at Bristol has been concerned with long-span bridges, using the three-pronged approach of theoretical analysis, laboratory studies and full-scale measurements. The three approaches are complementary, and even taken together it is seldom possible to answer all questions with confidence. For example, in the crucial matter of damping, the only satisfactory approach is by full-scale measurement, but since these are necessarily made at low amplitude levels, their relevance to actual earthquake conditions must always be questioned. And in both the laboratory and the analysis, modelling contains a substantial element of judgment.

In the first part of this paper the results of measurements on three large suspension bridges, the Humber and the two Bosporus bridges, are presented, comparing them with predictions made using both 2-, and 3-dimensional finite element calculations of natural frequencies and modal shapes, the dynamic forces here being a combination of traffic and wind. The measurements on all three bridges was of acceleration, but for the Humber a new computer-vision system was developed, which allowed dynamic displacement of the mid-span of the bridge to be measured directly, giving data from which natural frequencies and mode shapes could also be obtained for corroboration of those obtained from acceleration measurements. An array of 13 anemometers across the span allowed displacement to be correlated with wind speed and direction, from which lift and drag coefficients were deduced.

In matters of analysis and computation, this paper deals with asynchronous input to bridges, that is, where the size of the structure allows a finite time to elapse between the ground input at different piers. Also, in view of the non-deterministic nature of earthquake ground input and the consequent need to use a statistical approach for assessing response, the stochastic method

of analysis is explored, finding that the very low natural frequency of long- span bridges makes the use of such methods problematic.

The Humber Bridge The ambient vibration measurements were made to validate an extensive theoretical investigation which involved 2- dimensional finite element idealisations, one in the horizontal plane and one in the vertical plane, as well as a full 3-dimensional solution involving 3500 degrees of freedom. Fig 1 shows the measurement principle used, always involving a stationary reference accelerometer, with others deployed for 1 hour periods at various stations within the deck or towers as indicated in Figs 1 & 2, the whole measurement programme lasting 10 days. Fig 3 gives a typical result, showing clear peaks of response for vertical and torsional modes.

For the calculation of vertical deck modes the finite element model allowed three degrees of freedom, vertical and longitudinal displacements with rotation about the lateral (upstream-downstream) axis. The second of these was found to be a key component in agreement between calculation and measurement. Fig 4 shows this agreement for the first 8 vertical deck modes; 25 such modes were actually measured and computed. The agreement in these first 8 modes is very good, but special interest centres on the first two of these modes because this good agreement only became possible by alteration of the design deck support conditions at the towers. This specifies free longitudinal movement, and for such conditions the calculations give an antisymmetrical first mode at 0.108 Hz and a symmetrical second mode at 0.1 15 Hz. Actual observations at the bridge showed no such movement, and if one end of the deck

rave ing acceleromeler

11 12 13

(4

vclling accclcromcler

I1 12 13 (b)

Fig 1. Measurements using: (a ) a reference accelerometer and two independent travelling accelerometers to determine vertical and lateral response; (h) a reference accelerometer and two travelling accelerometers to determine by two simultaneous

measurements at extremes of width or height, the vertical, lateral and torsional response.

IStmctE/SECED seminar ‘Analysis & testing of bridges’ 9

I

Barlon sidc-span Hessle side-span Main-span (280 m) (1410m) (530 m)

111

Fig 2. 2-dimensional finite element mesh and measurement stations

is assumed to be hinged in the calculations, the first mode becomes symmetrical at 0.11 5 Hz and the second antisymmetrical at 0.155 Hz, see Table 1. Of course, with larger excitation, such as an earthquake would produce, the dynamic behaviour might revert to the design condition, and the measured damping values would probably increase.

The full sequence of measurements in these first (1 985) studies on the Humber included also the lateral and torsional modes of the main and two side spans, and the lateral, longitudinal and torsional modes of the two towers, 136 modes in all. The full

2o O O i I I

"1500

9 5-00

I . . . . I I I I I , I I I I

000 2Qo 4 0 0 603 8.00 )o Frequenq HZ

(a)

I - 1

i

0.00 J O-OO 0.20 0-40 0-M) 0.80 1.W . 1.20 1.40

Freauencr: Hz

L 120-00

0-00 4 I L' O-OO 020 0.40, OM) 080 1-00 1-20 T-40

Creouencf: HI

Fig 3. Vertical and torsional main span response: (a) main span vertical response at R1; (b) acceleration spectrum at

centre span - vertical component (12 i- 13)/2; ( c ) acceleration spectrum at centre span - torsional component ( I 2 - 13/12

results are given in Ref 1 and corresponding studies on the Bosporus bridges are given in Refs 2 and 3.

Video displacement tracking system The studies reported previously included some measurements of wind speed and direction and their correlation with modal frequency, amplitude and damping, but these were appreciably extended in the 1988 work now to be described.

Mode: V I Measured Frequency: 0116Hz Period: 8.58s Damping: 3.92% Predicted Frequency: 0.1 15 Hz Period: 6.67 s

I

Mode: v2 MeasuredFrequency: 0154 Hz Period: 6-51 s Damping: 3.58% PredinedFrequency: 0155Hz Period: 6.46s

. . I

Mode:V3 MeasuredFrequency: 0.177Hr. Period:. 564s Damping: 3.62% Predicted Frequency: 0.175 Hz, Perlcd: 5-60s .

I

. ,

Mode:V4 MeasuredFrequency: 0218Hz . Period: 4-595 Damping: 3-11% Predicted Frequency: 0220 Hz Period: 464s

Mode:V5 MeasuredFrequency: .0.240Hz Period: 4-17s Damping: 2.05% Piedicied Frequency: 0235Hz Period: 4.18s

Mode:V6 MeasuredFrequency: 0310Hz Period: 3-23s Damping: 181% PredicledFrequency: 0307Hz Pen'&: 325s

I

Fig 4. Comparison of measured and predicted vertical deck modes

10 IStructE/SECED seminar 'Analysis & testing of bridges'

I

Solution A (design condition) ' Solution B (hinged condition, basis for comparison in this report) I ,

integrate acceleration twice to obtain displacement with which to compare the direct measurement of the latter, and this is shown in Fig 6. Overlay of the two traces shows that the peak-to-peak amplitudes are in close agreement, but the use of a digital high-pass filter for the integration process has removed the lower frequency trend in the lower trace. This correspondence between direct displacement and twice-integrated acceleration goes some way, at least, towards showing that no serious errors are inherent in the vision system due to atmospheric refraction or other perturbations along the sight line. Continuous measurement of the rmd-span deflection made it possible to plot an experimental influence line as a large vehicle crossed the bridge; thls is shown in Fig 7.

Fig 8 shows the correlation of lateral and vertical displacement with the lateral wind velocity Vcos 8; the former shows a clear parabolic variation, but the latter is rather less certain, probably due to temperature variation during the course of the measurement. In this respect, a solid-state temperature probe installed In one of the main cables showed that the centre span displaces downwards by 64mm for each OC increase in main cable temperature. The measurements given in Fig 8 can be used Fig 5. Target on Humber Bridge

Experimental mode

A transputer- based visual tracking system had been developed for dynamic displacement measurements of models on the laboratory shaking table, and by changing the lens system it was used to track the mid-span deflection of the Humber Bridge from a station sited at the base of the Hessle tower; over this distance of 750m an accuracy of Imm was achieved. The target was a 500mm diameter illuminated roundel, shown in Fig 5, which allowed simultaneous vertical and horizontal measurement; details are given in Ref 4. Behind the roundel in Fig 5 can be seen a much larger rectangular board containing a black horizontal stripe. This was the target for vertical (only) displacement measurement in a system used by the Polytechnic of Milan, with whom we were collaborating in a programme carried out on behalf of Stretto di Messina Spa in anticipation of the construction of the 3300m main span suspension bridge between Sicily and Calabria. The Milan displacement measurement system involved two such targets, another with a vertical illuminated black stripe for horizontal measurements. The principal purpose of these studies was to provide a large body of data to validate and calibrate the mathematical modelling of the aerodynamic response of the proposed Messina bridge, for which physical models had been tested in Milan. Necessarily therefore, much more attention was paid to the measurement of wind velocity and its correlation with response parameters, particularly displacement; for this purpose 1 3 anemometers were distributed across the Humber span and the measurement of wind velocity and corresponding bridge displacements were continuous over a 8-month period.

Considerable interest is attached to the comparison, where possible, of results obtained from the earlier acceleration measurements and the new displacement measurements. For example, the response time-histories obtained in the two cases both contain modal information which can be abstracted using data processing techniques. Table 2 gives a summary of some natural frequencies obtained by the two approaches, and it will be seen that good agreement is obtained. It is possible also, to

~ _ _ _ ~ Table 2 Humber Bridge deck vibration modes ' Frequency displacment spectrum '

vision system) modal suvey (Hz)

0.054 1 L1 0.056 1 0.247 L6 0.260 0.309 T1_ 0.3 11

Vertical 0.1 15 0.116 0.154 0.177

0.158

0.309 i V6 0.310 ! T1 0.311

I ( Lateral

0.176 ~ v 3

0 50 ,

I 0 ! , i , , , ' , / / , , , , , , , ' I / , , . / / I , , , I , : . I ; , I . . . , , . . . 1 , , I . , # . ! !

a 900 1000 1100 1200 1300 1400 1500 Time ( 5 1

0.25 I

! I

900 1000 1100 1200 1300 1400 1500 b Time ( S I

Fig 6. Comparison of data from video system and processed acceleration signals. ( a ) displacement from video system; (b)

displacement from integration of accelerometer signal.

IStructE/SECED seminar 'Analysis & testing of bridges 11

-0 .2

- 0 . 3

-0.4

i i

1 '

4 80 54 0 600 T ime (sl

Influence of HGVs on centre span vertical deflection

-0 5 420

a

Fig 7. Comparison of influence lines obtained by measurement and analysis. (a ) time history at centre span during crossing of

vehicle of unknown weight; (b) influence line of centre span displacement for vertical load on deck.

in conjunction with calculations using the finite element models to compute lift and drag coefficients; such calculations obtain the vertical and horizontal forces distributed uniformly across the span required for unit centre-span deflections. Here, a value of 0.35 was calculated for CL, and 0.067 for CD. These compare favourable with values 0.26 and 0.078 which were obtained from model studies in the NPL laminar flow wind-tunnel studies at the design stage for Humber.

1

As is well-known, the flexibility of a suspension bridge produces coupling in the modes of vibration, and to obtain a full picture of how this coupling operates and to evaluate the accuracy of theoretical model predictions for full-scale application, it is necessary to have displacement and rotation measurement. The Bristol system has recently been expanded to measure stereoscopically f rom t w o cameras, so that 3-dimensional displacements can be obtained and this system is currently being used on a research study of the Second Severn Crossing cable-stayed bridge.

Asynchronous input Earthquakes pass through the ground at different speeds in the range approximately 300-800m/s. The piers of a long-span bridge would therefore receive the input with a time lag, possibly of several seconds. The ground also modifies the earthquake through which it passes. Analysis, andor model studies, must therefore consider the extra computational dimension caused by different inputs to the two piers and the two anchor blocks.

The analytical procedure is described in Ref 5, the major new feature being introduced beyond the conventional (i. e. synchronous) analysis is that so-called 'r-vectors' have to be calculated for each ground degree-of-freedom at the support points. Each r-vector is the displacement shape which the bridge takes up when the degree of freedom is given unit value. Fig 9 shows such vectors for unit vertical displacements of the Hessle anchor block and tower of the Humber Bridge. The actual displacements (referred to here as pseudo-static) from the vertical component of the ground motion can be obtained from these r-vectors, followed by calculation of forces in the members of the bridge. Such forces are additional to those caused by synchronous input, and the purpose of the study on the Humber and both Bosporous bridges was to assess the magnitude of these displacements and forces in comparison with those produced by standard seismic analysis, with the speed of travel of the earthquake as a prime parameter.

The results presented use the SI6E component of the Pacoima Dam record made during the 1971 San Fernando earthquake, assumed to travel across the bridge at 250,500, 1000,2000m/s; the infinite speed provides the conventional analysis. The use of only one record means that the results must be assessed with caution. Fig 10 refers to vertical input across the First Bosporus bridge (main span 1074m) and (a) gives the maximum values of the vertical pseudo-static displacement, the largest values occumng for 500m/s. The conventional dynamic displacements are given in (b), and the total in (c), which is not obtained simply

E v

3

,.. .+ I . . .

. .

1 1 1 1 1 1 1 1 I I I I 1 1 l I I . I \ . . 0 . 5 . ..I0 15 . . . . 20.. 25

b . . I(cose,(ms-!) .

Fig 8. Correlation of 64s average displacements with wind speed. (a ) p lateral displacement px = 0.0018 (V COS e)', (6 ) p vertical displacement, ~ J I = -0.00081 (V cos

12 IStructEYSECED seminar 'Analysis & testing of bridges'

Fig 9. Humber Bridge: r-vectors for the Hessle anchor block and tower; vertical input

by adding (a) and (b) together because maximum values occur at different times. Fig 1 1 gives the total bending moments in the Hessle tower of the Humber Bridge for different input speeds of the Pacoima earthquake applied in the vertical plane; the 5 O O d s input gives the largest values throughout the tower, which are several times greater than the infinite speed conventional case.

Stochastic analysis The two principal methods of calculating the response of long-span bridges to earthquakes are time-history analysis and

Veloctites' rn ls

- *

cc0:TI -

/I \:, 02251 (C )

. " Fig 10. Bosporus Bridge deck: (a ) vertical pseudo-static displacements; (b ) vertical dynamic displacements; (c)

vertical total displacements (all for vertical input)

the response spectrum technique; both have shortcomings. The former is expensive, produces an embarrassing amount of information which it is difficult to use, and has to be carried out for each earthquake record which is thought to be relevant. The deficiencies entailed in the use of response spectra are that they only give maximum values occumng and there are difficulties in combining the contributions in different modes.

It would be possible, in principle at least, and at great expense of time and money, to carry out a statistical analysis of the results obtained from the many required time-history analyses, but the stochastic approach attempts to provide equally useful information at the expense of greater mathematical complexity (Ref 6) in treating the time-history data. Fig 12 shows the vertical displacement time-history at mid-span of the Dolerw, 50m main span suspension bridge subjected to a 13.4s earthquake applied in the vertical plane. The absolute maximum value is about 165 mm, but many other maxima occur and it would be useful to have some measure of these, as well as a measure of the average value during a given time-span. For the last parameter, the RMS value is easily calculated, but this requires a timespan to be specified. The dilemma here is that the response continues to be non-zero for some time after the end of the input, and the RMS value clearly tends to zero with time; judgment is therefore required in specifying the time interval over which the RMS is to be calculated. Returning to maximum values, these too can be averaged to produce a 'mean-of-maxima' displacement value, but this also tends to zero as the time interval increases. To go further in abstracting useful parameters from the time-history of Fig 12, the frequency of occurrence of maxima can be obtained, together with a cumulative distribution function of displacement values, which is useful in considering fatigue behaviour.

Applying the stochastic method (Ref 6) to the Dolenv bridge with the same input, results shown in Fig 13 are obtained using the 13.5s of the earthquake as the time interval. Comparison of Figs 12 and 13 and allows the following comments.

Fig 11. Humber Bridge: bending moments in Hessle towerfor different speeds of vertical input

lStructE/SECED seminar 'Analysis & testing of bridges' 13

0.20 1 I I

w : 0.10

z 5 li $- -0.10 b'

E T -0.00

I

-0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 1&00 18.00 20.00

Time /seconds

Fig 12. Dolerw Bridge: vertical displacement time-history at mid-span

The CQC method is the modal combination approach in the response spectrum method which allows for modal coupling; its mid-span value of 177mm is only slightly larger than the 165mm given by the time-history analysis, and this small difference could well be due to the different computer programs used for the two solutions. Otherwise, the sequence of values given in Fig 13, i.e. 41.7, 107, 177mm are in the expected order.

A simple visual inspection of Fig 12 shows that the mean-of-maxima value given in Fig 13 at mid-span, of 107mm is close to the truth.

Similarly, the 1.25 value given in Fig 13 for the average frequency of occurrence of maxima at mid-span is supported by the evidence of Fig 12.

V 17.7 _ - -

/aJ

301 HI. ibl

RMS o f dispiacements

Mean-of-maxima displocemenls

Abso!ute mox,mum displocemenls by COC method

Fig 13. Dolerw Bridge: (a ) vertical displacements of the deck; (b) average jrequencies of occurrence of vertical displacements

maxima qf the deck.

The stochastic approach applied to the Humber Bridge gives a less reassuring picture as Fig 14 indicates. Here, the expected sequence of results does not always occur; over the mid-span region the mean-of-maxima is greater than both the time-history and CQC values. No doubt some of these anomalies are due to different software for the different methods; time-history for example, using about 25 times as many degrees of freedom as the stochastic approach. But there is no doubt that the principal reason is to be found in Fig 15 which shows the vertical displacement time-history at mid-main span of the Humber. Within the 13.5s of the earthquake there simply is not enough information on which to base reliable statistics. The situation can be improved by continuing the stochastic analysis beyond 13.5s (see Ref 6); how far beyond is a matter for more research, but our own work suggests that one-quarter of the fundamental period should always be added to the timespan of the earthquake, which would be 2.15s for the Humber. Certainly a duration of 20s produces the anticipated sequence of results. It should also be said that both Bosporus bridges give the expected order of results (Refs 6 and 3) from the stochastic analysis even when only the 13.5s time-span is used for the calculation. But both these bridges have a shorter main span, with no suspended side spans to complicate the response.

From these few results, the only general conclusion which can be drawn is that the results of applying the stochastic method to

0 : ; J

Fig 14. Humber Bridge: (a ) vertical displacements of the deck; ( b ) average frequencies of occurrence.

14 IStructWSECED seminar 'Analysis & testing of bridges'

- _- 0.00 5.00 10.00 15.00

7irne/seconds

Fig 15. Humber Bridge: vertical displacement time-history at mid main span.

long span bridges must be applied cautiously, and with background knowledge of the principles involved.

References 1 .

2.

3.

4.

5 .

6.

Brownjohn. J. M. W . , Dumanoglu, A. A., Severn, R. T. & Taylor C. A.: ‘Ambient vibration measurements of the Humber Suspension Bridge and comparison with calculated characteristics.’ Proc. Instn. Civ. Engrs, Part 2, 1987,83, Sept, 561 -6002 Brownjohn, J. M. W., Dumanoglu, A. A. & Severn, R. T.: ‘Ambient vibration survey of the Fatih Sultan Mehmet (Second Bosporus) suspension bridge. ’ E’quake Eng. and Struct. Dgn. Vol21,907-924 (1992) Dumanoglu, A. A., Brownjohn, J. M. W. & Severn, R. T.: ‘Seismic analysis of the Fatih Sultan Mehmet (Second Bosporus) suspension bridge.’ E ’quake Eng. and Struct Dgn. Vo12 1,88 1-906 (1992) Stephen G. A., Brownjohn, J. M. W. & Taylor, C. A.: ‘Measurement of static and dynamic displacement from visual monitoring of the Humber Bridge.’ Eng’ng. Strucr.

Dumanoglu, A. A. & Severn, R. T.: ‘Seismic response of modern suspension bridges to asynchronous vertical ground motion.’ Proc. Instn. Civ. Engrs. Pt 2, 1987, 83, Dec, 701-7306 Dumanoglu, A. A. & Severn, R. T.: ‘Stochastic response of suspension bridges to earthquake forces.‘ E ’quake Eng. andStruct. Dyn. Vol 19, 133-1 152 (1990)

1993, Vol 15, NO 3, 197-208

IStructWSECED seminar ‘Analysis & testing of bridges’ 15

Seismic response of RC bridges

A. S. Elnashai, BSc, MSc, PhD, CEng, MIStructE. MASCE Imperial College, London

Summary Seismic performance of reinforced concrete bridges has been attracting considerable attention following the collapse of several structures in the San Fernando (1969), the Loma Prieta (1989) and the Northridge (1994) earthquakes in California as well as the Hanshin (1 995) earthquake in Japan. Concerns have been expressed regarding the safety of such structures in other parts of the world, considering that the California Transportation Department (CALTRANS), which controls bridge seismic design, is a leading authority in bridge design under earthquake loading and that Japanese seismic design practice is one of the most advanced in the world. Notwithstanding the direct human and economic losses from the heavy damage or collapse of a major bridge structure, the loss of an important transportation artery causes great damage to business, especially in affluent societies. This, reinforced by the field observations which indicate that there is a lot still to be learnt regarding seismic design of RC bridges, lends weight to the concerted effort dedicated to studying the inelastic seismic response of such structures, not only in California, but worldwide. Moreover, the recent introduction of the Eurocode 8 chapter on Bridges has stimulated interest in this subject in Europe.

To evaluate the supply of RC bridges, required to meet the evaluated demand imposed by strong ground motion, experimental testing of components is required. Since small- scale testing, particularly under dynamic loading, encounters problems of similitude, full, or near-full scale testing is preferable, rendering investigations of seismic supply of bridge components prohibitively costly. Therefore, analytical procedures, using material and member models which incorporate the salient behaviourial representations, are of great significance. Analytical results may be verified by experimentation in a targeted and an optimally-steered fashion. Issues of demand are not discussed herein, but it suffices to state that the case for utilisation of analytical methods in demand assessment is even stronger than in supply estimation.

In this presentation, after highlighting the vulnerability of RC bridges to damage from earthquake ground motion effects, attention is focused on issues of analytical evaluation of the ductility supply and demand of RC bridge structures. First, the limit states of yield and ultimate are discussed and alternatives are presented for their definition, as below.

In order to determine seismic response parameters in both the static and dynamic analyses, it is necessary to define a set of limit states for yield and ultimate. The limit state definition remains controversial and results obtained from one study are not, in general, comparable to another because of the wide variation in this definition. These are discussed hereafter, in the light of the wider discussion presented in other publications by the writer.

Yield limit state The criteria initially considered for yield were two established definitions: firstly, yield of the longitudinal reinforcement; and secondly, the secant of the load deflection curve.

First yield First yield is defined as the point at which the main longitudinal reinforcement reaches the material yield strain. Problems exist with this definition in terms of numerical estimates when compared to experimentally obtained values. Generally, values observed experimentally tend to be greater than those obtained

numerically. This arises from bond slip between the longitudinal reinforcement and the concrete, combined with early stiffness degradation due to cracking of the section.

It was decided that the point of first yield would not in fact be taken when the outermost pair of reinforcement bars yielded, as this was seen to have minor overall effect. Instead, the point of first yield was taken to coincide with yielding of five reinforcement bar pairs. Assuming the yield strain for steel

= 2.15 x 10-3, this definition corresponded to E~ = 2.40 x 10.) in the strain-monitored outer reinforcement. In terms of structural response this definition showed good agreement with the first substantial departure of the load-deflection curve. By using this definition an attempt is made to compensate for the disparity between earlier yield in numerical results compared to experiments.

Secant yield To define the point of secant yield the load-deflection curve must firstly be replaced by an equivalent elasto-plastic system. The initial stiffness of this system corresponds to the secant passing through the point of 75% peak lateral load. Yield is subsequently defined as the point where transition occurs from elastic to purely plastic. In general secant yield is considered more realistic than first yield but values still tend to be lower than experimentally observed results.

Ultimate limit state Even more than the yield limit state, the definition of ultimate structural response is a particularly subjective decision. Consequently a number of ultimate criteria were considered, as follows.

Concrete failure This is defined as the maximum available curvature. This curvature is assumed to correspond to the post-peak value of curvature at 0.85 times the peak moment, Mpek. Multiplying this value for curvature by a fictitious, but conservative, value of plastic hinge length U4 one arrives at the corresponding maximum available rotation. In the case of the pier under investigation the axial load remains unaltered. This implies that the curvature at failure would be constant and thus the value of strain at 0.85 times M F ~ would also correspond to failure. Finally, the stress in the confined concrete is related to the moment, therefore the ultimate strain could alternatively be calculated from a 15% drop in the peak stress of the confined concrete.

Significant drop in resistance A commonly cited criterion based on prescribing the ultimate limit state as the deformation corresponding to a drop in the load carrying capacity of the load-deflection curve, typically in the range of 10-1576.

Fracture of tensile reinforcement 'This criterion defines the ultimate limit state using a limiting value of tensile strain in the main longitudinal reinforcement. The limiting value of strain is taken as that permitted in Eurocode 8, which is 12%

Limit on excessive rotation An empirical procedure essentially limiting element rotation may be utilised. Using such a definition of ultimate limit state constrains the rotational ductility to a predetermined value.


The response parameters monitored in this study are the curvature and displacement ductilities, the energy dissipated, plastic hinge length and the load multipliers corresponding to yield and ultimate limit states. Under dynamic loading, the behaviour factors, calculated as the ratio between the yield peak ground acceleration and that corresponding to the ultimate limit state, as calculated and compared for various choices of ground motion.

Two studies were undertaken, one involving the variation of steel yield, reinforcement arrangement, applied axial load and pier slenderness of the bridge pier shown in Fig 1 below, whilst the other involved investigating various modelling assumptions. These included the modelling of the pile groups, soil stiffness and deck-to-pier connection stiffness.

The first study yielded general trends concerning the effect of the investigated parameters on the ductility of piers, which are the main energy-dissipation component of RC bridges. It was shown that increasing the yield strength of the reinforcing bars causes a drastic reduction in ductility, hence QA of steel properties is essential. This effect is less significant in the presence of high axial forces. However, the detrimental effect of high axial load, which increases yield displacement and decreases ultimate displacements, were emphasised by the results.

The study on the effect of inclusion, or otherwise, of the soil-pile system and the connection between deck and pier indicated that the modelling of the whole system is most important. However, the results were almost insensitive to the assumption of soil stiffness or pier-deck connection stiffness and strength. Therefore, detailed studies aimed at assessing these stiffnesses are not warranted.

The study showed also that the selection and scaling of earthquake records for dynamic analysis is an essential ingredient in the seismic assessment process. Table 1 shows the behaviour factors calculated from a set of records selected on the basis of their peak ground acceleration-to-velocity ratio.

The values of behaviour factor evaluated for the six earthquakes, given in Table 1, show the extent to which ground motion input may cause variation due to predominant frequency content, duration and number of cycles corresponding to the

Record label

,4 O r n ’- I

Yield PGA Ultimate I Behaviour i Ratio to

w- S ~ c l i a n A A

-~____ Std. dev. (3 52.3 2.14

Section B-E

24-024

cover=somm D’om 65m FOUNDATION CYLINDER REINFDRCEUENT

I 46.7 ~ 2..20

Fig I . Detuils of RC bridge pier studied


cov = olp 0.3 6

Table 1 Behaviour factors for RC bridge pier

0.54 , 0.25 0.36

Friuli 2.500 j ::: 1 4.0 \ 0.909 Gazli ~ 0.305 3.3 i 0.75 -

Loma Prieta ~ 0.160 0.570 3.6 0.819 EW

El Centro 0.130 0.930 7.2 1.636 -

_ _ _ . _ _ _ ~ , LomaPrieta Spitak . ~ 0.100 0.130 ~ I 0.665 0.320 1 i *ET-

NS I

Mean

Table 2 Structural response for PGA and SI scaling

Structural response (2) procedures --

__

(mm) ~ x 10-3

~

Table 3 Response parameters for PGA and SI scaling

Response parameters Record 1 p A ( d e m d pga) i &(demand pga) F A (dcmand SI) I & (demand SI)

procedures- __

0.50

Loma 2.20 4.20 1 3.20 8.00 Prieta i

! 2.00

EW i 3.80

w- 6.50 2.70 I 6.00

1 El 1.60 2.40 2.20 Centro ,

2.50 ~ 2.60 I 5.50

Prieta 1 i Mean,p 1 1.78 3.28 i 2.45 5.06

Std. 0.67 ~ 1.94 ~ 0.55 2.03 Dev.,o ~

cov = ’ 0.38 0.59 0.22 I 0.40 _ _ _ ~

I -.aL- particular frequencies of structural significance. The results, given in Table 2, highlight the danger of using natural earthquakes in seismic assessment studies when the suite of records is not carefully selected. For instance, if Friuli, El Centro and the Spitak records were chosen, a mean q factor would have been 5.4. On the other hand, the three remaining records, if used on their own, would have indicated a mean behaviour factor of 3.3, more than 35% lower than the former value. Moreover, the results show that El Centro, which is by far the most widely used

17

earthquake record, is by no means a conservative loading scenario.

Scaling of the records, required for behaviour factor calculation, was also investigated; direct acceleration scaling as well as scaling to acommon velocity spectral intensity. It became evident that direct acceleration scaling takes no account of levels of spectral response at frequencies other that that associated with the peak acceleration pulse. Consequently two records which possess equal peak ground acceleration will be scaled with the same factor irrespective of their frequency content.

Alternatively, scaling to spectral intensity ensures that all records possess the same intensity across the response spectrum band for periods from 0.1 to 2.5s, and for structures of moderate to long period, constitute a more appropriate approach. The demand imposed by the selected set of records on the piers, expressed as displacement and rotation ductility, is given in Table 3.

Finally, the presentation includes early results from the global analysis of a large curved RC bridge using advanced adaptive dynamic analysis techniques. The structure was subjected to three components of earthquake motion, with records generated from the EC8 recommended spectrum. Comparison of supply and demand imposed on the ten piers comprising the bridge are presented and discussed. It is shown that having a single ductility factor for the entire structure leads to very uneven distribution of demand. The shear demand imposed, as a function of the instantaneous axial load is also discussed.

Conclusions regarding the ductility supply (capacity) and demand (global design requirement) of RC bridges are given, and the areas requiring development for code application are highlighted.

References 1.

2.

Calvi, G. M., Elnashai, A. S. & Pavese, A., ibid.: ‘Influence of regularity on the response of RC bridges.’ Elnashai, A. S. & Beith, J. G.: ‘Seismic design limit states and ductility supply of reinforced concrete bridge piers,’ Proc. 2nd International Workshop on the Seismic Design of Bridges, New Zealand, 9- 12 August 1994

18 IStructE/SECED seminar ‘Analysis & testing of bridges’

Recent experiences on concrete and cast iron A

bridges

D. C. Mackay, BSc, MSc, PhD, CEng, MBCS Lloyd's Register

Introduction This paper describes selected specialist bridge assessments undertaken recently by Lloyd's Register. The first part of the paper outlines the overall approach adopted which is illustrated in the second part by means of three case histories.

Bridge assessments undertaken by Lloyd's Register have involved a combination of structural analysis and supplementary load testing. They have followed basically an analytical approach, within the context of the UK Highways Agency assessment standards, and have used data from static or dynamic load tests to validate an analytical model. This type of approach has been found to be beneficial when assessment by 'traditional' methods leads to an unduly pessimistic evaluation of structural integrity.

Overall approach The overall approach to analysis and test is:

an assessment which involves both

(a)

(b)

(c) (d)

Static or dynamic load testing to measure the structural response of the bridge Preparation of a theoretical analysis model, e.g. grillage or finite element Validation of the model using the test data Strength assessment using the validated model to determine the load capacity of the bridge.

The overall approach using dynamic (or modal) testing is illustrated in Fig 1.

Types of load test There are two types of load tests for bridges':

Proving load tests involve the application of test loads to demonstrate the integrity of a bridge under assessment loads. The level of loading should provide an acceptable increase over assessment levels. At the present time, acceptable levels have not been clearly defined. Such tests have been undertaken in the UK and Canada. However, this form of testing is not at present favoured by the Highways Agency due to the possibility of causing damage to a bridge during testing.

Supplementary load testing involves the application of test loads no greater than the traffic loads already known to have been taken by the structure, in order to reduce the possibility of damaging the structure. Test measurements are used to adjust the analytical model used in the assessment t o BD2 1/93. This process involves extrapolation of the test results to higher load levels, where non-linear response may be exhibited, and consequently introduces other uncertainties into the assessment process. Nevertheless, this approach tends to be favoured by the Highways Agency. Lloyd's Register have undertaken both static and dynamic supplementary load tests.

1 7"EORETICAL ANALYSIS E L

MODAL TESTING

HItiHWAYS AGENCY ASSESSMENT STANVARDS

I I

INTEGRITY ASSESSMENT

FINITE ELEMENT MODEL

.".*-"_Id

Fig 1. Flow chart of a bridge structural integrity assessment using dynamic load testing

IStructWSECED seminar 'Analysis & testing of bridges' 19

Dynamic test procedure The following sources of dynamic excitation are used in bridge testing:

(a)

Lloyd’s Register recommended that minor strengthening works be implemented by the Client. The scope of the strengthening was considerably less than would have been required if only the traditional assessment had been made.

Woolslope Bridge, Dorset, UK Woolslope Bridge is situated on the B3072 between West Moors and Ferndown and is owned by Dorset County Council. It was constructed in 1957. It has a single span of 9.4m and carries a two lane, single carriageway road. The deck consists of prestressed concrete inverted tee beams with an in situ concrete

Routine traffic - the response to the passage of normal traffic loading is measured Single lony excitation - temporary profiledramps are fixed to the carriageway and a l o w driven Over the ramp at speeds between 10 and 40km/h.

(b)

The dynamic response of the bridge, and abutments if required, is measured using accelerometers attached to the structure. Signals from the accelerometers are continuously monitored on an oscilloscope and a spectrum analyser, and recorded on a digital tape recorder.

The test data is post-processed off-site to produce the shapes and frequencies of the key modes of vibration of the bridge.

Static test procedure Kentledge weights are used to apply HB loading to a bridge deck. The weights are delivered to site on heavy goods vehicles and positioned using heavy lifting equipment - an operation which takes about 6 hours. The full test load is maintained overnight.

The deflection of the deck is measured at various positions using deflection transducers, such as dial gauges. The measurements are recorded during the loading operation and on an hourly basis at full load until the deflection of the deck becomes insignificant.

Compared to dynamic testing, a static load test involves greater effort (and hence cost) to organise and, as it requires a larger duration, leads to greater disruption of traffic.

Analysis model A finite element or grillage model, as appropriate, is prepared to simulate the load test. In the case of a dynamic load test, a natural frequency (free vibration) analysis is required to determine mode shapes and their frequencies. The model is validated by modifying stiffness and mass properties in a systematic fashion to obtain reasonable agreement with the test results. To date, this has been undertaken manually but Lloyd’s Register is currently investigating the use of computer software to be able to increase the amount of test data that can be considered.

Strength assessment The strength assessment is undertaken using the validated model with the applied loads and the strength of the bridge deck determined in accordance with Highways Agency Standards BD21/932 and BD44/903.

Case histories

St Leonard’s Bridge, Dorset, UK St Leonard’s Bridge is situated on the A31 trunk road and is owned by the Highways Agency. The bridge is believed to have been built in 1936. It has a central span of 10.7m and two ground bearing side spans and carries one two-lane carriageway. Its abutments and deck are constructed from in situ reinforced concrete. The deck consists of seven beams and a slab cast on the beams.

Dynamic and static load tests were conducted by Lloyd’s Register and provided data on the behaviour of the bridge. A finite element model was prepared and validated against the test data. The validated model was then used in a strength assessment to determine the load capacity of the bridge.

infill. The deck is supported by two capping beams, each supported on four piles.

The bridge had previously been assessed by a firm of consulting engineers who had found that the bridge could not carry the full assessment live loading. They had recommended that a static load test was undertaken to determine the behaviour of the bridge under a known loading regime.

In a static load test, Lloyd’s Register loaded the bridge using calibrated weights and measured deflections using dial gauges. A grillage model was prepared and validated against the test data before determining the ultimate load capacity of the deck. The bridge was shown to be capable of carrying the full assessment live loading.

Little Bridge, Tonbridge, Kent, UK Little Bridge is located in the centre of Tonbridge and is owned by Kent County Council. It is a single span bridge, constructed in the late 19th century and consists of eleven cast iron arch girders built into brick abutments.

Another consultant had found that calculation of the load rating of the bridge was sensitive to the amount of horizontal movement assumed at the springings. Lloyd’s Register undertook a dynamic load test, instrumenting the arch girders with accelerometers to measure the deflection of the bridge due to the passage of a heavy goods vehicle over a ramp installed at the mid-span of the bridge. The relationship between the deflection at the crown and translation at the springings was established by comparing time histories recorded at these points. This relationship was used to develop an analytical model of the bridge. The model will be used shortly to analyse the bridge under assessment loads.

Acknowledgements The author wishes to thank the Committee of Lloyd’s Register for permission to publish this paper and is grateful to the following organisations for allowing the case histories to be included:

Highways Agency Dorset County Council Dorset Engineering Consultancy Kent County Council.

References 1 . Chalkley, C.: ‘Modelling problems in short-span bridges

and the basics of load testing.’ LoBEG Load Testing Seminar, 25 February 1994. Highways Agency Standard BD21/93, The assessment of highway bridges and structures. Highways Agency Standard BD44/90, The assessment of concrete highway bridges and structures.

2.

3.

20 IStrucWSECED seminar ‘Analysis & testing of bridges’

Potential applications of NDT methods for bridge assessment and monitoring

P. C. Das The Highways Agency

N. C. Davidson Department ojCivil and Environmental Engineering, University of Edinburgh

C. Colla Department of Civil and Environmental Engineering, University of Edinburgh

Synopsis This paper describes the basic principles of radar, sonar and dynamic testing methods, and how such methods could potentially be used for the assessment and long-term condition monitoring of bridges.

Non-destructive testing (NDT) methods can essentially provide a qualitative picture of the interior of a structure or the surrounding materials. Such information could be used for identifying hidden or unrecorded structural features, which may indicate hidden sources of reserve strength.

NDT methods can also be used for monitoring material deterioration or development of faults, either by directly identifying the faults, or by recording changes in the dynamic characteristics of the bridge.

The appendix of this paper contains the details of a number of applications of these methods carried out by Edinburgh University.

Introduction In recent years, a number of NDT methods have been used, mainly on trial basis, for a variety of structures-related applications, including some involving bridges. Although such applications have been sometimes intended for the precise location of relatively small features, such as reinforcement bars and structural cracks, in general the results from these methods are expected to be qualitative. This is because most of these methods rely on reflected electromagnetic or acoustic signals or dynamic responses, which give a very complex but imprecise picture of the reality.

When assessing the load-carrying capacity of an existing bridge, the calculations used are of a precise nature. The results also have to be fairly decisive, in that the bridge authority has to take specific actions based on the results of the assessments. In this context, at first sight, there would seem to be no place for ‘imprecise’ qualitative testing methods in bridge assessment. Indeed, there have been attempts in the past to test the comparative ‘accuracy’ of such methods for locating small structural features for assessment purposes, the results of which have not been encouraging.

Nevertheless, the current national bridge assessment and strengthening programme has highlighted a number of situations where the precision of the conventional assessment methods cannot be relied upon. In such cases, a few examples of which are given later, instead of the ‘pass or fail’ type of results from the assessments, it may be more logical to allow a ‘pass, fail or monitor’ type of result. The monitor option could be used when a bridge fails assessment by a small margin but may have features which are likely to lead to a reserve for strength, which cannot at present be rationally taken into account in the calculations.

When deciding whether to adopt a ‘monitor’ option for an assessment-failed bridge, the bridge authority’s confidence would be reinforced if some form of qualitative examination could confirm the possibility of there being a hidden reserve of strength. NDT methods such as the sonar, radar or the dynamic methods could be of great value in arriving at such decisions. Furthermore, once the monitor option is decided upon, some of these methods could also be used for examining the bridge periodically to see if any significant structural deterioration is taking place.

The Highways Agency has for some time been actively examining the various NDT methods for use in different aspects of the management of the network. There is already advice available on the use of ground penetrating radar (GPR) for road pavement assessment (HA 72/94, Design manual for roads and bridges, the Highways Agency). The present intention is to eventually provide similar guidance for bridge-related use. The purpose of this paper is to highlight the areas of bridge assessment and monitoring where non-destructive testing techniques could be fruitfully used. It also describes what is needed at present in terms of the specifications for equipment and test procedures, so that bridge authorities can make use of these methods with confidence.

Methods A variety of methods have been used in recent times in a number of different ways. Those intended to be covered in this paper fall into the general categories of radar, sonar and dynamic testing methods.

The applications so far have been mainly exploratory or in the context of research and therefore competent specifications for commissioning such testing widely are not at present available. The common characteristics of all these methods are:

Most of the methods were originally developed for purposes other than structural testing, for example, for quick testing of aircraft bodies or machine parts. Highly sophisticated equipment is available commercially. The instruments come with wide ranges of capability, a large part of which could be inappropriate for any particular application. Most equipment is portable, and the tests can be carried out quickly and conveniently without much disruption to traffic.

The general principles of such methods are that:

(1) Radar will propagate through most materials, including air, but not metals; however it is rapidly attenuated in saline conditions and in clays.


Sonic will propagate through most materials including metals, but not air. The conductivity technique is essentially a high powered metal detector and will identify changes in ground conductivity. Shallow metal would shield deeper penetration due to its high conductivity. Dynamic testing methods generally involve the vibrational characteristics of the whole bridge.

Basic radar and sonic theory The same basic principle of transmitting a wave into the structure lies behind the radar and sonic techniques though different forms of energy are used - electromagnetic in the case of impulse radar, acoustic for sonic testing. The idea of such NDT methods is presented schematically in Fig 1. A transmitter of the source signal and a receiver, to record reflected energy, are required. Initially a pulse is transmitted into the medium of interest. If the region were perfectly homogeneous no signals would be returned to the receiver. However, any feature which has contrasting physical properties, in relation to the type of energy used, causes a proportion of the signal to be reflected while some energy is also transmitted through the boundary. If the reflected signal returns in the direction of the receiver and is of sufficient amplitude it will be recorded at the receiver. By measuring the time taken for the energy to arrive at the receiver the location of structural defects and boundaries between different media may be determined. Using the equation shown below, the depth,D, of the defect or material boundary may be found.

T 2

D=-V

where T is the time taken for the arrival of the reflected energy and V the velocity. If the transmitted energy encounters a further interface the process of reflection and transmission will be repeated and thus deeper features may be detected.

The data may also be analysed in the frequency domain to give a different perspective on the interpretation. To achieve this waveforms are transformed in the frequency domain using a Fast Fourier Transform (FFT) algorithm and subsequently the response from the structure is divided by the force input in a frequency domain function, yielding the Frequency Response Function (FRF).

Distances to boundaries and dimensions of discontinuities can be computed from the expression:

L = VI ( 2 x A F)

where:

V = velocity through material AF = interval between resonant frequencies as indicated in

Transmitter Receiver A i r

A i r I v

Fig 1. The principle underlying radar and acoustic non-desfructive methods

the frequency domain plot.

This method of analysis can give a clearer interpretation of some signals.

Theory of electrical conductivity measurement Electromagnetic (EM) survey methods make use of the response of the ground to the propagation of electromagnetic fields. These response fields reveal in both phase and amplitude the presence of the conductor and provide information on its geometry and electrical properties (Fig 2). The induction of current flows results from the magnetic components of the EM field, consequently there is no need for physical contact with the ground. The EM field measured is generally a complex function of the coil spacing s of the conductivity meter, the operating frequency f and the conductivity distribution of the subsurface 0. The depth of penetration depends upon s and is independent from the conductivity distribution of the subsurface. The depth also depends on the frequency used and the lower the frequency, the deeper the penetration but the poorer the resolution (amplitude decreases exponentially with depth).

Vibrational analysis In this method the vibrational response of the bridge is determined using an accelerometer mounted on the face of the structure which is excited by an instrumented hammer. Vibrational characteristics can indicate, for example, the degree of end-fixity of a bridge.

Dynamic testing techniques In one form of this technique the whole bridge is excited by dropping a relatively heavy weight (say 50kg) through a height of l m on the bridge. Its natural frequency is then determined using an accelerometer placed somewhere on the bridge. By taking such measurements periodically, a time related picture of the behaviour of the bridge can be built up.

Applications

General The electromagnetic investigative techniques of impulse radar and conductivity may be applied to the assessment of bridges

receiving c o i l

1

Fig 2. Principles of electromagnetic survey


constructed of concrete, brick or stone masonry and which include soil fill. The sonic method may be used for these materials and in addition metal structures may be evaluated, but bearing in mind that sonics will not cross an air boundary such as a full width crack or discontinuity.

The appendix contains details of a series of case studies carried out by the University of Edinburgh using these investigative techniques. In general terms the techniques might be applied as follows.

Radar applications in concrete include identifying:

rebar locations in concrete concrete member thickness voiding or honeycombing in concrete voids in non-metallic grouted duct post-tensioned bridges.

Radar applications in masonry include identifying:

wall thickness voids springer shapes arch barrel thickness retaining wall thickness bridge structure behind spandrel walls - structure capillary rise in masonry.

Dynamic testing includes determination of structural integrity and deterioration

Bridge assessment and monitoring There are many types ofbridges where assessments are not likely to be conclusive and are therefore amenable to the use of NDT methods in some form for added confidence. The following are some typical examples of such bridges and other situations where such methods could be useful.

Masonry arch bridges Masonry arch bridges are extremely difficult to analyse. Not only that most are very old and without proper records, the strength of such bridges is also greatly influenced by the quality and strength of the backfill and the presence of additional structural features such as internal spandrel walls. Recently the results from a number of assessment methods, including the finite element method, were compared with the results from a series of ten full- scale bridge collapse tests. The outcome of the exercise is that all methods are considered to be very approximate. As such there is a strong case for using the ‘monitor’ option for masonry bridges which fail assessment by narrow margins.

Jack arch and trough-deck bridges These types of bridges are also difficult to analyse because the structural details are not amenable to straight-forward idealisation. The assessment calculations therefore can be suspect.

Older composite bridges

fill or cellular

General radar applications include identifying bridge scour Applications of impulse sonic testing include:

wall thickness in concrete and masonry void identification and mapping wall dimensions, spandrel wall and arch thicknesses void detection in metallic grouted post-tensioned ducts (with further research).

Applications of non-contact conductivity testing include identifying:

moisture movement detection over time

layering within masonry non-homogeneity.

In some of these bridges concrete or other types of fill were placed directly on steel plates of various configurations, with or without shear connections. Assessment calculations tend to underestimate the composite action in such cases.

Arching action In many relatively modem short-span bridges, the decks were cast directly on to the abutment. This can provide a considerable amount of lateral support, thereby inducing very strong arching action as well as enhancing the shear capacity of the deck. Such effects are usually not taken into account in assessments. Some forms of dynamic or vibrational tests can indicate the presence of endrestraint and how that changes with time.

Bridge deterioration Physical deterioration of bridges in general, and that of problem bridges such as the post-tensioned bridges in particular, could potentially be monitored using vibrational or dynamic methods

t Voided region + t Voided region + Length of section 5 m

3 200 mm P _x -

Fig 3. Example of I GHz radar data recorded over the slab

IStrucWSECED seminar ‘Analysis &testing of bridges’ 23

I

Fig 4. Estimated regions of the slab in need of repair

in order to diagnose any changes to the overall dynamic characteristics of the bridges. Furthermore, the progress of deterioration can be more directly monitored by using the other NDT methods.

Monitoring following load tests In some forms of load-testing of bridges, substantial magnitudes of loads may be applied to a bridge. In such cases there is a risk of hidden damage being caused to the bridge during the tests.

Fig 5. Upstream sight ($Middleton Bridge

24

Such damage, although being small to start with, may increase with time, and seriously affect the bridge at a later date. Therefore, a rigorous regime of subsequent monitoring is essential if such tests are carried out. NDT test methods can play an important part in this.

Present needs In order for the bridge authorities to use the methods described in this paper with confidence, the following aspects need to be addressed.

Trials The methods need to be applied, on a trial basis, to a variety of bridge types and situations in order to identify the scope for useful application.

Specification of equipment and procedures Based on the trial applications, specification for use to particular bridges and problems, including the specification of equipment, the procedures to be followed for the testing, and the expertise required of the personnel to be involved should be established.

Presentation of results and verification Criteria should be developed so that the testing bodies can follow certain common standards for reporting the results. Guidance should also be provided so that non-specialist assessing engineers and clients can verify their accuracy.

IStructE/SECED seminar 'Analysis & testing of bridges'

c: P

9 E. p. 8 rn approx.

Fig 6. Radar cross section obtained with lOOMHz antennae

0--

e-

Fig 7. Formation of scour hole at bridge pier


Conclusions It can be seen from the discussions in the paper and the examples of application given in the appendix that NDT methods can potentially be employed for bridge assessment and monitoring purposes in a number of ways.

The main difficulties envisaged are that the results of the tests are generally of a qualitative nature and require expert knowledge for their interpretation. Furthermore, the available ranges of equipment and testing methods are such that the scope and specification for each area of application need to be drawn up carefully.

Acknowledgements This paper is being presented with the kind permission of the Chief Executive of the Highways Agency, Department of Transport. The authors are also indebted to Professor Michael C. Forde, Head of the Department of Civil Engineering, University of Edinburgh, and Ms Caroline Ellick of the Department of Civil Engineering, University of Reading, for their valuable advice during the preparation of this paper.

Appendix

Case studies - impulse radar investigations

Bridge deck assessment An assignment was undertaken to assess a reinforced concrete bridge deck for suspected voiding. As a strengthening measure a new slab had been lain on top of the existing bridge. From a visual examination at the slab edge incomplete filling between the rebars was evident. Edinburgh University was therefore commissioned to conduct a radar survey to estimate the extent of voiding throughout the slab.

A high frequency transceiver (antenna) was used to investigate the 200mm thick slab. The reinforcing cage had a 50mm cover of concrete with rebars centred at 150mm. Steel reinforcement is a very strong reflector of electromagnetic energy and if the bars are placed too closely it may not be possible to investigate a structure. The design of the slab, quoted above, is close to the limit within which radar is effective.

An example of the data recorded at the site is shown in Fig 3. This data was taken along the edge of the slab where the voiding was visible. The displayed regular undulations are reflections from the reinforcement. Areas where the concrete has not filled the structure is indicated by a more disrupted, smaller magnitude sequence of reflections (darker colours) than is observed for the rest of the slab. Taking the darker, more irregular signals to be indicative of voids the complete slab was surveyed and the suspected problem areas identified. The plan of Fig 4 shows the estimated regions of the slab in need of repair. Two cores were removed and the radar assessment of the slab was proved to be correct at these locations.

Study of stone masonry bridge In the evaluation of masonry arch bridges, impulse radar testing is starting to be used and has advantages over other techniques in certain circumstances, but very real problems can exist with regard to the interpretation of radar data. Where multiple changes of interface between dielectric layers take place, the change of dielectric constant will be identified but the resolution of the final target layer may prove particularly elusive. A good example of this would be when the radar antenna is drawn along the surface of a masonry arch bridge with the objective of identifying hidden geometry and character of the soil fill with precision. Such investigation would have to proceed through the various layers of material and if the soil fill is heterogeneous the problem is compounded to the extent that the data is usually uninterpretable.

The system used on this survey was a GSSI Sir System 10 digital radar set. Antennas of 100 and 300MHz centre frequency

25

U 111 -

I 0 111 -

5 5 0

2 c) 111 -

3 0 I l l -

5 111 c-----+ NB. The third pier was dry

Fig 8. Section of radar data recorded over gravel river bed

were used, operated with different orientation and in various transmitting and receiving modes.

Each antenna was moved along vertical and horizontal survey lines along the surface of the abutment walls of the bridge (Fig 5). The survey had to overcome a few practical difficulties due to the configuration of the bridge and the rough surface of the stone masonry work. The antennas were lowered vertically with ropes from the top of the parapet, or manually moved horizontally across both the upstream and downstream sides of the abutment.

The use of radar to investigate the bridge fill material has produced interesting data, as shown in Fig 6. Significantly higher dielectric constants were calculated than might have been expected from published literature. This may be due to moisture/salinity factors and is currently under investigation.

Evaluation of bridge scour Bridge scour is the removal of river bed material at the foot of piers due to the vortex system set up by the current, see Fig 7. The phenomenon may seriously impair the stability of the bridge. Recent bridge failures due to scour have demonstrated the importance of evaluation and monitoring of the problem. For structures spanning rivers and other water courses, impulse (or ground-penetrating) radar may be deployed to assess the problem.

Researchers at Edinburgh University have conducted impulse radar surveys for scour at two sites. In addition to the radar

equipment a sonar device was used to detect the river bed profile. A dinghy carrying the transceivers was towed behind a small boat which contained the instrumentation. Various traverses, across and up and down the river, were then made in the vicinity of the bridge. The whole procedure was recorded on video to provide information on the position of the boat during traverses.

From these investigations it was found that the nature of the river bed affected the strength of signals returned from within the bed. At the site of a bridge crossing a river with a gravel bed only the surface of the bed was shown on the radar section, demonstrated in Fig 8. However, the texture of the bed was resolved indicating the position of rock armour around the piers. In the case of a river with a sandy bed, see Fig 9, signals from the bed itself and subsurface interfaces were received. This allowed aproposal for the history of scour formed during past floods to be put forward (Fig 10). This information allows the extent to which the hole may be opened up during subsequent flood events to be predicted and the risk of failure assessed.

The research showed that impulse radar produces useful, accurate information when applied to the problem of bridge scour. Future objectives include the classification of sediment types using multifrequency imaging, determination of limitations of the technique in conductive water and improvements in processing techniques.

20

40

0 a - 4

26

Fig 9. Radar data taken over sandy river bed

IStructHSECED seminar ‘Analysis & testing of bridges’

111

> 1.5

P a !s

Vy -7.20000e-3

z a

- I

-

111

I

T/D 5.00e-4 TL -1.960e-4

-

Fig 10. Interpretation of radar section over sandy river bed

8 I .- - - 4

Sonic study

Concrete bridge abutment In the upgrading of a highway bridge it was necessary to determine the thickness of the supporting abutments. To provide the required thickness information a sonic survey was undertaken.

To determine the distance a reflected wave has travelled, it is necessary to know the velocity with which it has moved through the medium. Therefore the velocity of acoustic waves passing through the concrete abutment was required before a sonic survey could be used to estimate its thickness. Average values of velocity through concrete may be used, such as 400Ods for good quality concrete. However, where possible, it is advisable to measure the velocity of acoustic waves through the structure

/ h

I J , i

I

-

under examination. At one end the bridge abutments protruded out from the embankment. This allowed a transmitter and receiver of ultrasonic emissions to be placed either side and as the distance between the two transducers was known the velocity through the concrete was determined. A value of 1350ds was found which is a very low value and indicates poor quality concrete.

The abutment was then tested using an instrumented hammer, force transducer and digital oscilloscope. At various locations along the abutment the hammer was used to impart energy into the structure while returning signals were picked up by the transducer and recorded within the oscilloscope storage facility. By measuring the time for the arrival of the signal reflected from the back of the abutment wall an estimate of its thickness was calculated using the determined value of velocity. Fig 11 shows an example of the data. The time for the reflection was measured and an estimate of 0.9m obtained for the thickness of the abutment wall at this test location.

Assessment by conductivity measurement

Stone masonry bridge A novel application of this, until now, geophysical, archaeological and agricultural method has been carried out in the civil engineering field and a conductivity survey has been undertaken on one abutment wall of an historical stone masonry structure.

The digital conductivity meter used - Geonics EM38 - has intercoil spacing of Im and provides a depth of exploration of 1.5m in Vertical Dipole Mode operating with a frequency of I4.6kHz.

The meter has been used on both the upstream and downstream sides of this 2-span bridge and on the wall beneath the main vault. The measurement stations followed a grid marked on the walls, in an area far from any evident metallic objects (drains, reinforcing beams). For maximum accuracy and good spatial resolution, measurements have been overlapped to have readings every 0.5m. Contacting and non-contacting (at 0.5m from the wall) conductivity measurements were taken, to obtain data at different depths. Data were collected using adigital data recorder

Fig 11. Sonic test of concrete bridge abutment


Fig 12. In situ conductivi5 data collection

(Fig 12) and later transferred to a PC. The results have been plotted to produce contour maps of the conductivity distribution.

The values obtained are in a high and very wide range (Fig 13), indicating heterogeneity in soil filling in the abutment, made of rock such as argillites, wet clays or alluvium and sand or variations in moisture contenthalinity. Conductivity is usually determined by clay content, moisture and salinity. Of these the most complex is usually the moisture profile, the conductivity increasing approximately as the square of moisture content. Both dielectric constant and conductivity increase with water content and the presence of salts in pore water will increase even more the conductivity, without affecting the dielectric constant too much. (c.f. problems described aboutradarapplication on masonry).

The survey was repeated after 6 months and differences have been noticed in particular behind the wall under the vault (Figs 14 - 15). Comparison of results from data taken at lm depth lead to the hypothesis that a significant moisture/water movement is taking place at the rear face of that wall, with concern about the possible loss of the finest part of the filling. Coring is to be carried out in the next few months to verify the internal situation.

The future and developments at Edinburgh University Research into NDT techniques at Edinburgh University is on-going. Objectives include a better understanding of the response of heterogeneous civil engineering bridge materials, through conducting large scale laboratory experiments and calibrating findings with field surveys. Specific problems with regard to concrete and masonry arch bridges are being investigated, including masonry bridge construction, component thickness and bridge scour.

Custom equipment for determining conductivity at variable depth is to be developed and the enhancement of interpretation software and 3-D modelling is being carried out.

\

Fig 13. Conductivity map of downstream side

2x IStructE/SECED seminar 'Analysis & testing of bridges'

Fig 14. Under the vault: conductivity survey in August 1994

Fig 15. Under the vault: conductivity survey in February 1995

IStructWSECED seminar 'Analysis &testing of bridges' 29

Arch bridge testing

W. J. Harvey, BSc, PhD, CEng, MICE Department of Civil Engineering, University of Dundee

Introduction To be invited to present a paper on arches in a day programme dedicated to the testing and analysis of bridges is a somewhat daunting prospect. The abstract originally offered does not do justice to the title and the structure of the day. I will therefore attempt to offer a brief but broad overview of the work which has been done in the arch field over the past decade or so, but present it from a very personal point of view in the hope of kindling some enthusiastic discussion.

The motive for much of the work that has been carried out was the difficulty experienced in assessing arch structures for the new higher loads imposed which resulted from joining the EU. Over the period a considerable number of people have been involved in the programme of work. The long term aim is presumed to be improving our ability to analyse the behaviour of arch bridges. For some that means predicting with the highest possible accuracy the anticipated failure load of the bridge. For others it means being able to predict confidently the failure mode. It is my view that we have not actually progressed enormously towards either of these goals, but we have succeeded delivering techniques which, whilst not delivering the accuracy that was hoped for, can with some certainty offer a lower bound, and therefore safe solution.

Assessment of arches is necessarily a matter of both analysis and judgment. The traditional approach has been quick and crude, with the aspect of judgment apparently concentrated in the final factor which covers the condition of the bridge and which has an enormous effect on the output of the assessment.

15 years ago, the only approach to assessment in common use was the Military Engineering Experimental Establishment’s (MEXE) modified method, based on Pippard’s empirical boundaries to allow an analysis based on Castigliano’s work. Despite being analytically indefensible, MEXE remains in regular use because it has proved to be safe and easy in application without setting too onerous conditions. Whilst the analysis on which MEXE is based is mathematically sound, it is known to bear no relation to the actual behaviour of the structure. The three primary assumptions are not satisfied; that is, the arch ring itself is not elastic, it does not stand on pin supports, and failure is not, except in relatively rare cases, controlled by the strength of the material. Thus MEXE combines a series of numbers which demonstrably describe the geometry of the structure in a way which leads to results which are intuitively sensible and which through empiricism take into account all the complex interactions present in an arch bridge.

For these reasons no modem approach to analysis can hope to provide a similar degree of correlation between analytical and practical results. Nonetheless it is worthwhile exploring what approaches are on offer and how close to the desired result they manage to come. That begs the question ‘what are the desired results?’ and since some millions of pounds have been spent on arch bridge testing overrecent years, a brief discussion of testing is in order here.

Arch bridge testing - background Three possible reasons for carrying out tests are to prove that structural capacity is above a desired minimum, to develop understanding of behaviour in a way which may well not be quantitative, or to test hypotheses about behaviour. Sometimes tests which are intended to fall into one of these categories actually fit into another. I believe that the majority of testing of

arch bridges that has gone on over the decades, and this includes the programmes of tests carried out in the 1930s, ‘40s and O OS, were essentially proof tests, not in the sense that they were all non-destructive but that their value lay chiefly in demonstrating the enormous capacity of arches without showing how that capacity was delivered. Careful inspection of the tests and results which came from them actually leads to a considerable degree of understanding of arch behaviour and particularly of the load paths which provide the unexpected capacity.

A modest number of tests were carefully designed to test hypotheses. This essentially cannot be done on real bridges because the parameters cannot be checked before the test and sometimes the information is lost during the test. Of the control tests, only the results of those we carried out ourselves are fully available to us and the notes below will therefore lean most heavily on our own work.

In 1986 under a contract from the Department of Transport and the Science & Engineering Research Council, we built five models. For political and economic reasons the first of these was the most complex. It was a model of a small bridge rather than a scale model of a bridge. The span was 4m, the arch was semi- circular and 6m wide. It was built from precast concrete voussoirs, the spandrel walls from concrete bricks, the fill contained at the ends with very heavily stiffened shutters and then surfaced. 50 soil pressure gauges were installed in the spandrels and the upper surface of the arch. The bridge was first tested using 50kN patch loads representing a single wheel placed at various positions so that the distribution of wheel loads through the fill and its effect on the arch could be explored.

The results were a surprise, not qualitatively but quantitatively. The cone of distribution from the surface patch to the extrados of the arch was very narrow and in an area immediately surrounding the patch load, pressure on the back of the arch went down as the arch deflected slightly away from the fill. The pressure distributions on other parts of the arch indicated a much greater transverse stiffness than had originally been anticipated. Figs 1 and 2 show the results from two of these patch load tests.

In Fig 1 the load is off-centre across the span but on the centreline of the bridge longitudinally. The diagram shows the developed extrados of the arch with the position of the soil pressure gauges and a contour map of the pressures produced. It is very clear from this diagram that the effect of the load fans through the arch structure from the point of application out towards the abutments. What is not clear but becomes so under analysis is that the soil pressures are rather less than would be anticipated and that the deformations were extremely small. It is clear from a closer study of the results that even with these patch loads, the spandrel walls were providing substantial stiffening for the arch. This is perhaps clearer in Fig 2 where the patch was applied close to the spandrels and the pressure changes show distribution across the full width of the arch with any only local active pressure changes where the arch is able to deflect downwards away from the spandrels.

The ultimate test of this bridge was carried out using a line load at one-third span. Fig 3 shows the pressures that resulted; the heavy curved lines show the pressure changes which seem necessary to support the structure while the more angular broken line indicates the pressures actually recorded. Here it becomes very clear that a secondary load path is in operation and that whilst the fill perhaps could provide the stability the arch needs, in this case it certainly did not. The transverse stiffness of the arch is clearly so great that the majority of the load takes the stiff path into the spandrels rather than the more flexible path into the fill.

30 IStructWSECED seminar ‘Analysis & testing of bridges’

Four further tests were carried out of structures designed to behave in a 2-dimensional fashion. The first was a l m wide slice of the structure above. Once again, fill was applied above the arch, this time contained laterally between a shutter-type structure but one which was heavily lubricated to prevent undesired structural interaction between the shutters and the soil. The results were most satisfactory. The soil pressure distributions were now much closer to those anticipated in relation to 2-dimensional analysis with a substantial development of passive pressure on the unloaded part of the bridge and rather more distribution of the live load through the fill than in the full-scale bridge models.

Three further tests explored the same effects in progressively shallower spans, and in these tests the major changes were that the shallower the span, the greater the distribution of load between the patch on the surface and the contact between the soil and the arch extrados. It seems clear that this reflects the increasing flexibility of the progressively flatter arches.

The conclusions we may draw from these tests are that there is a support mechanism available to the structures involving substantial distribution of load through the fill and then substantial interaction between the arch and the fill to provide arch stability. These effects may easily be modelled in a 2-dimensional analysis. In an actual bridge the spandrels will provide substantial edge stiffening which modifies the soi1:structure interaction to the point where its effect is almost negligible. This also dramatically increases the actual strength of the structure.

It is my contention that this alternative load path is not quantifiable in our present state of knowledge but that it will always provide an increased capacity over that predicted using the simpler 2-dimensional model. Provided, therefore, a 2-dimensional model can show a modest factor of safety for all anticipated loads, the alternative load path will provide that extra capacity which allows an engineer to rest easy in his bed.

Analysis There are three basic levels of analysis available: the empirical which relate capacity to leading dimensions of the bridge via carefully judged factors, equilibrium approaches which simply seek to find an adequate load path through the structure, and stiffness analyses which attempt to relate deformation of the arch to its mechanical properties.

The MEXE method and the computerised Pippard-MEXE method fall into the first category. The analysis treats the arch as an elastic ring on pinned supports. Only vertical loads are applied to the ring to represent the fill and the live load. The limit to capacity is set by limiting the compressive stress allowed in the arch whilst ignoring the tensile stress which the arch is known to be incapable of carrying. In some cases compressive stresses above the limit would be present in the arch without any live load, but that is also ignored. The limiting compressive stress is chosen so that the results of analysing tested bridges correspond as closely as possible with the test results.

Elastic and elasto-plastic analyses offer a considerable range of complexity. For example, the arch may be treated as an elastic material which has no tensile strength in a method developed by Castigliano. The load applied to the arch may be purely vertical gravity load or there may be some model for the interaction with the fill, typically springs which have different ratios in compression and tension. The same general effect can be produced using finite elements to model the arch in one form or another. In these relatively simple models, one degree of additional complexity might be to allow for changing geometry in the structure as the arch is loaded. The finite element model may be constructed of both the arch and the fill; a number of standard packages would offer this capability and the program developed by Mike Chrisfield at Transport Research Laboratory is perhaps the most refined, offering a Mohr-Coulomb failure model for the soil and a full elasto-plastic response for the arch masonry which allows the user to study the development of cracks as the lozd is incremented. A number of major finite element packages allow 3-dimensional models to be constructed

I

I 1

0 ! .__ I ---0 -0 I

I 0 0 ; I

I I

I

+4 +3 +2 +I

10 -1 -2 -3 -4 -5

Figs 1 & 2. Soil pressure distribution result from patch loads

Fig 3. Soil pressure round an arch ring

IStructE/SECED seminar 'Analysis & testing of bridges' 31

with the same sort of level of complexity; that is to say a 3-D Mohr-Coulomb soil and 3-dimensional elasto-plastic elements for the arch and spandrel walls, even including the possibility of taking account of change in geometry as the structure is loaded.

Equilibrium methods The equilibrium method is relatively straightforward. The approach is simply to try to find a path through the structure which can transmit the forces without causing material failure. The geometry of the structure is presumed not to change significantly, and the various complex interactions are treated by working to limits rather than attempting to model the interaction explicitly. Failure is deemed to occur when the load path can no longer be contained within the arch by the chosen limits to the applied soil pressure. Naturally material strength must be taken into account and this is done by ensuring that the load path stays far enough from the boundaries of the structure to avoid crushing the material.

Discussion The following remarks are essentially a personal review. The reader should be aware that they are written by the vendor of a program designed for arch assessment and they are made with the deliberate intention of provoking comment.

The first point to make is that increasing complexity necessarily increases cost. The software itself becomes more expensive and also more expensive to use. The labour involved in data acquisition and data preparation increases, as does the scale of computer required and the time taken to do a run. Against that background we should consider the value to be extracted from the results of the various methods.

The first comment is that except when taking account of geometric non-linearity, all the analytical rather than empirical methods should produce very similar answers given similar input data. So if the specification for the soils is set to give similar soil pressures, and the live load is incremented to the point of collapse, the bridges will collapse in a way very similar to a mechanism and the collapse load will be similar to that derived from an equilibrium mechanism based analysis. Satisfying the conditions for this similarity is actually quite difficult since each approach models the soil in a slightly different way. Some versions of the equilibrium analysis assume that the soil can attain full passive pressure before failure, neglecting the fact that to develop such a pressure the arch must deform grossly and the gross deformation will alter the structure which has to contain the load path. Spring models of the soil can only be used on the basis of calibration so that once it has been decided that the arch model must carry a certain load at failure, the spring can be calibrated to produce that result. The more tests against which the calibration is done, the better the correlation is likely to be, but once again there is a certain empiricism in the process. The more refined models involving Mohr-Coulomb elements for the soil present even more difficulty because there are more soil parameters to be input and calibration is therefore that much more difficult.

Bridge assessment is very different from bridge analysis. The data that is available is extremely limited or very expensive to obtain. Even the true shape of the arch extrados is often unknown and the cheaper methods of exploration which usually involve drilling from the underside of the bridge are notoriously imprecise. Similarly the content of the fill may well be unknown so ascribing properties to it is quite difficult.

My own belief is that there is some danger in using an analytical method which is more complex than the data which are available justify. The engineer is led to a false assumption of accuracy, even if this assumption is backed up by the knowledge that his model predicts loads similar to those from actual failures. Underlying the whole procedure one must remember that if the correct load is predicted in the wrong way, it is not possible to extrapolate from one set of results to another single result in any direction.

Arch assessment remains a difficult process. Engineers are confident the arches are sound but cannot prove that by analysis. In the limit, we depend economically on the engineer using his judgment backed by relatively unrefined analyses. Any codified rules which discourage him from using that judgment will necessarily lead to lower assessments. By the same token, any attempt to put pressure on the fees available for assessment will lead to the same conclusion. Saving even 80% of the assessment fee is of tiny value if the outcome is to condemn a bridge.


Investigation of post tensioned structures, DarticuIarlv coniidering intrusive inspection of

I

iendons and their duct;

S. W. Kemp, BSc, CEng, MICE Technotrade Ltd.

Introduction Before discussing methods of investigation it is necessary to establish what one is trying to achieve by undertaking an investigation:

surely the first item has to be a suitable degree of confidence against imminent failure there may also be aspects of estimation of future requirements for maintenance and liabilities from this in some instances the adaptability of the structure needs consideration and further information is being sought. For instance, the road carried may need to be upgraded.

Unfortunately, as with other bridge testing, there are signs that the owners or agents for the bridge stock are more interested in a transfer of liability than in determining the condition of the structure to any degree of accuracy.

With the current financial climate the final criteria for an investigation is cost. It must be cheap -does this inevitably mean it will be nasty as well? Value for money is more important in the long run than absolute cost.

Scope of survey Having established the overall brief and considering primarily the first two points - immediate safety and maintenance liabilities, the scope and extent of the investigation has to be planned.

What should be examined during an investigation The most obvious elements to be examined are the tendons. If the tendons are corroded and broken we do not have a prestressed bridge but a poorly reinforced structure. If the tendons are in good condition is the sheathing, if any, also in good condition?

What conditions are the anchorages in? Does it matter? If the overall structural quality is good, with high strength, well- compacted concrete, the tendons are in fully grouted ducts and still in good condition, is there any benefit in examining the anchorages as they are redundant for a fully grouted system?

In order to examine the anchorages, it is usually necessary to remove road surfacing and deck waterproofing, a movement joint, ballast walls and protective encasement. In general this means sustained traffic management even if the anchorages are in the verges. It is normal that an anchorage inspection will take the best part of a week or even longer.

Having repaired the breakouts at an anchorage is there left a permanent weakness? Failures in movement joints are not exclusive to post-tensioned bridges. Indeed there is doubt whether any joints have a life expectancy adequate for the design life of a bridge. It is more probable than not that ,over a modest period from the investigation, contaminated water will reach the face of the beam ends. It is questionable whether any repair will be as good as original concrete properly placed, although some

Fig 1. Typical anchorage repair in verge requiring long term traffic management

Fig 2. Typical anchorage -- note missing strand top right

IStructE/SECED seminar ‘Analysis & testing of bridges’ 33

Fig 3. Well grouted tendon in percussive breakout

repairs may be better than some original concrete protection. I feel very strongly that too much is being made of anchorage inspection at unnecessary cost in the immediate and long terms. Only inspect if there are specific reasons to do so.

It may be convenient to take the opportunity of a fairly extensive investigation to catch up on some other items, particularly joints. There are two groups of joints, which I consider for the purposes of post-tensioned investigations as either primary structural or secondary. The primary joints are between segmental precast units or precast to in situ joints such as on transverse diaphragms or stiffeners. The secondary class, for the purposes of inspection, are the overall movement joints. Obviously these latter can have a significant influence on a bridge, but their importance is common to bridges of all types of construction.

The quality or condition of the primary joints directly affects, and is affected by, the integrity of the post-tensioning. The visual appearance of these joints may be sufficient alone to indicate problems exist. It may be worthwhile inspecting some of the apparently sound areas as well.

The secondary joints may be visually inspected, with varying degrees of success, by videoprobe or CCTV.

Having examined the tendons, joints and anchorages there may be benefit in checking the existing residual stress in tendons or concrete, or both. Of the two methods the author’s preference is for measurement of concrete. After all the tendons may all be present but never have been stressed or they may have been stressed but 50% have since corroded. An appraisal of the structure can be greatly assisted by measuring the concrete stresses. Although the relative accuracy of a steel stress measurement is better than for concrete the author considers the latter to be more meaningful. One of the problems with steel measurement is the uneven initial stresses caused by stage or sequential stressing. There are also losses along a cable. The concrete measurement can tell, within its limits of accuracy, the current reserves of the structure in serviceability terms. Creep will have tended to even out any inequalities in the initial transferred stresses.

Although the ultimate capacity of the structure is independent of current stress levels it is obvious that there can be considerable durability problems for structures where the prestress in the concrete has been lost.

Consideration should also be given at this stage to further testing of the concrete generally. The level of access for a phase I11 investigation is generally much better than for a principal inspection. One item rarely specified, but vital for understanding the serviceability response of the structure, is the modulus of elasticity.

Tendons Having decided to inspect the tendons one needs to determine the scale of the investigation.

Fig 4. Radar survey of bridge deck

A single span structure will typically have three critical locations per tendon; at mid-span and near each of the anchorages. A multispan structure may have three critical points per span, per tendon, and is more likely to have complex transverse tendons with similar multiplicity of critical points. A twelve span continuous structure carrying a dual carriageway could easily have over three thousand critical sites.

It is important to consider carefully how many locations are inspected. Fortunately, experience suggests that there are good and bad bridges, rather than a small number of defects in each bridge. It is possible that insufficient work has been done on the ‘good’ bridges to find defects, but it is considered that experience is reasonably valid. This distribution of defects seems to reflect rather on the quality of site supervision than inherent problems in stressing techniques. On this basis we have found that one in two or three of the tendons inspected on a bad bridge will show defects but none on a good bridge. Therefore if no defects are found during inspection of twenty or thirty tendon locations it is unlikely that significant problems exist.

Methods The most important problem to be overcome is the location of the tendons.

In many instances there are some drawings of cable profile available. Unfortunately these are not always converted to as- built drawings. Even when accurate profiles of the duct fixings are available there can be significant variations in the profile between the duct restraints due to flotation. For this reason, some form of non-destructive location is required to minimise damage.

The current methods for duct location are dead reckoning, which as mentioned may be unreliable, radar, covermeter, ultrasonics or stitch drilling. The latter of course is hardly non- destructive.

Where the unstressed reinforcement layout is known to be simple and the expected cover is low, say 75mm or less, it may be sufficient to use a covermeter to confirm duct location. Where reinforcement is heavy, such as near anchorages, or where cover is greater, other methods are required.


There is considerable work being undertaken at present on the use of ultrasonics for imaging of concrete. Unfortunately concrete is not the kindest medium for sound transmission at the higher frequencies required for good resolution. It is likely that within two years there will be working systems but in the immediate future these should be disregarded.

Stitch drilling can locate ducts with luck. However, before relying on drilling for location consideration must be given to how much damage is acceptable before giving up. Great care is also required in inspection of steel contacts. There have been instances where ‘tendons’ which have been visually inspected through 25mm diameter holes, have subsequently been broken out, only to find that the link was in excellent condition. Nonetheless there are occasions when the restraints upon cable location are such that there is little room for errors in placement. In these cases stitch drilling may suffice.

Radar has the advantage of being non-destructive and having the capacity for much greater penetration than covermeters. The disadvantages are cost, difficulty of interpretation and reliability. High steel densities can block any detail behind, rendering the technique not always successful.

In the absence of any alternative method it is advised that it is good practice to start by requiring the use of a radar survey. If satisfactory reasons can be found to reverse this decision without reducing the chances of finding the ducts then by all means do so. But it is considered that it is better and more reliable to assume that radar is required and then prove otherwise rather than try to justify radar, with its pros and cons, as an afterthought.

You will note that radiography has not been considered as a duct location method. This is primarily due to the small area of coverage and time per exposure. I will discuss radiography more later.

Penetration and initial inspection The normal procedure for obtaining the first contact with the duct is by a 20 or 25mm percussion drilled hole. This has the advantages of speed and low cost with little chance of damage to the tendon. I must take exception to the guidance in BA50 regarding completion of the drilling by cold chisel. If the hole depth exceeds 75mm there is not enough room to manoeuvre a cold chisel successfully. Entering the duct can only be carried out using a drill at greater depths. If any water is found it should be collected.

After making contact with steel the hole should be blown clean with compressed air and a careful study made using a forward looking endoscope. A videoprobe can be used but is generally too cumbersome at this stage. Photography is possible but the results are not always as good as hoped. It is very important to ensure that the endoscope is well braced before exposure.

If the contact is confirmed as a tendon a pressure test should be carried out. The use of pressure rather than vacuum techniques has the advantage of not pulling any contamination into voids. The most popular method seems to be the simple Boyle’s Law apparatus where a known volume of pressurised air is discharged into the duct and the change of pressure recorded. Leaks from the duct can be calculated by recording the variation of pressure with time. Simple apparatus which cannot record the pressure decay is not really adequate.

Detailed inspection Before enlarging the pilot hole, care should be taken to check that there are no other cables closer to the surface. For shallow cables, the hole can be enlarged by stitch drilling or saw cutting and percussive breaking. Greater depths are generally best reached by overcoring to within l0mm of the duct and finishing them by percussive methods.

The quality of the sheath exposed should be recorded before removal and sampling of the grout, noting its condition. Sufficient grout should be removed to allow inspection of the tendons. Photographs can be taken, subject to limitations of depth.

IStrucWSECED seminar ‘Analysis B testing of bridges’

Fig 5. Corroded tendon through cored access

If voids are found they may be inspected using a videoprobe. These differ from the older fibrescopes in that optical fibres are only used to transmit illumination. The image is recorded by a video camera at the probe tip. This allows a much greater resolution, typically 300000 pixels rather than 10 to 30000 fibres in an optical bundle. The output is viewed on a monitor during the survey and recorded for future use. Beware that many of the systems offered for sale are not suitable for use on 110 volt site supplies.

There are many suppliers and combinations of diameter and length. The smaller the diameter obviously the smaller the hole that can be inspected, however, also the more vulnerable the probe becomes. At a cost in the tens of thousands of pounds, we do not wish to lose a probe because it is too delicate. We have found the most practical diameter to be 10 to 12mm.

The probe length should be as long as possible. The furthest we have had our probe along a duct is some 3m. However, remember that the control unit is not likely to be hard against the concrete and the minimum realistic length for a probe becomes

Fig 6. Pneumatic drill for stress measurement in steel

35

4m from the steering controls, not from the monitor. Again practicalities suggest that 6m is the upper limit of length.

An allowance should be included in tender documents for post- editing and annotating video recordings. It is possible to title on site but elaborate commenting is not practical. Indeed the space available, often in a hydraulic hoist, is limited and the minimum quantity of equipment is preferable. Some labelling of the video is of course necessary at the time of inspection.

At this stage it is possible to carry out stress measurements on the steel. This is not for the faint hearted. A hole of 150mm diameter is required to provide access to the tendons for gauge fixing and the equipment. Although the hole drilled is only about 1.5mm diameter by Imm deep this can be quite significant on a 4 or 5mm diameter wire.

Having completed the testing, the hole can then be made good. Consideration should be given to protection of the cables if there were voids. Should inspection tubes be installed to allow further inspection of voids? It is not generally possible to replace the duct, indeed it is highly probable that there is no sheath left to replace. In more vulnerable areas, i.e. holes drilled from the top of the deck it may be beneficial to provide a bituminous seal against the remaining duct and tendon.

Wherever the duct is located the repair should be made with a good quality repair concrete. The size of the hole will determine the maximum aggregate size that can realistically be placed. Proprietary repair materials are advisable and an aggregate size greater than 8mm is unlikely to be suitable.

Where waterproofing and surfacing have been removed an adequate overlap of the repair is required. We note that many specifications require reinstatement to match but neglect to identify the waterproofing in advance. Unless the original membrane was very new it is questionable how well any patch repair will bond. Our preference is generally for a layer of hot bitumen as the most practical and reasonably effective. Cold applied materials such as Stirling Lloyd Eliminator, Hand applied Grade, can be used when appropriate.

Residual stress measurement Residual stress measurements can be made in steel or concrete. The principles are the same although the size and interpretation differ greatly. The principle is that drilling a hole allows the strains in surrounding material to relax. The strains can readily be measured and stresses calculated from strains using standard formulae. The procedures for stress relief in steel are covered by ASTM E837. However, the practice is rather more difficult than the standard suggests.

On the practical side the installation of the gauges on steel is difficult to achieve. The foil strain gauges were developed as laboratory tools. Their use on site requires considerable experience. Drilling of the hole for strain relief is also a delicate operation. It is usual to use a microscope to position the drill accurately. Hole diameters for steel vary from 0.8 to 3.5mm with hole depths of 60 to 100% of the depth.

Interpretation of steel stresses is significantly affected by surface treatment of the steel, such as rolling, drawing or grit blasting. Again, the experienced companies can take sufficient measurements to allow compensation for this. The construction of the cable and the make-up of the strands also influences stress distribution.

Measurement of concrete stresses on site are somewhat easier due to the increase in scale from 1.5 to 75mm diameter. The interpretation is correspondingly more complex. Unfortunately concrete is an inherently inhomogeneous material with further inhomogeneities due to any reinforcement in the vicinity of the test. The stress fields are likely to be 3-dimensional rather than the predominantly uniaxial stresses in a bar or wire. Considerable research has been carried out by a few people on stress measurement in concrete and a very few companies have the expertise required to obtain meaningful results. Track record is vital here,

In addition to the hole drilling, a slot cutting technique can be used on suitable concrete members. A circular segment is sawn from the concrete and the uniaxial strains relieved in the concrete

Fig 7. Strain gauge array for measurement of residual stress in concrete

measured. This is a somewhat simpler site technique. Problems are the long-term stress raising effects and the lack of space to physically cut the slot in the depths of cover available.

With all stress relief results the elastic properties of the materials are required. Steel values may be assumed but it is usual to carry out measurements on the concrete in situ, by jacking, or to remove the cores and carry out measurements in the laboratory.

As a guide, the realistic accuracy of measurements in steel are likely to be of the order of 25N/mm2 and in concrete 0.5 to 1.5 N/mm2.

Bearing in mind the residual stresses in the concrete are likely to be less than 10N/m2 and in steel 100 to 600N/mm2 the steel test can be seen to be relatively more accurate. Nonetheless the concrete value is probably more relevant to an assessment of the serviceability of the structure.

Radiography There are various forms of radiography used for concrete but essentially they all provide an image similar to the medical X-rays with which we are all familiar. Some of the equipment, e.g. the ‘scorpion’, use electronic detectors rather than photographic film but the principles are the same.

A powerful stream of radiation, X-rays, beta or gamma particles is focused through the concrete and a shadowgram produced.

The technique clearly shows voids in ducts and can distinguish between more than one duct within the thickness of the concrete. The image becomes less sharply focused as the thickness through the concrete increases with the sharpest objects being those nearest the film. This is currently the only technique which can locate voids within the tendon sheath successfully.

Unfortunately radiography has some very serious drawbacks. The primary of these has to be the inherent danger of ionising radiation. It kills! Health and Safety requirements demand, rightly, great care. Public must be excluded from the immediate vicinity and the site cleared of non-essential personnel. The use of electrically driven sources rather than naturally radioactive materials is of benefit here as the source can be immediately shut down with the flick of a switch rather than withdrawing a source back into a shield and capping.


The required safety zone depends upon circumstances. but is likely to be 70 to 200m in the direction of the main beam of radiation

The area to be photographed is limited by the plate or detector size. Electronic detectors should be more versatile. Exposure times can exceed 2 hours, which in limited processions greatly restrict the number of survey points per shift. Fairly obviously radiography can only be used when one can access both sides of the member.

There is a significant mobilisation cost including extensive liaison with the Health & Safety Executive (HSE). Additional personnel are required to police the exclusion zone. Preliminary liaison and mobilisation costs can be of the order of E1500 and the cost per day/night may be E1200.

Interpretation of the plates produced appears easy. However, because of the long exposures sometimes required it is easy to overlook some of the smaller sizes of ordinary reinforcement. This is particularly so near to anchorages, which are in my opinion, the most appropriate place for radiography, where the steel may be so dense that bar edge details are lost by superimposition. Before drilling to prove the presence of voids it is important to consider the engineering implications of the location.

Summary There are some well-established and some less reliable techniques available for inspection of post-tensioned bridges.

It is important in preparation of tender documents, to consider that if, as the Client or the Engineer, you do not know exactly what is in your structure and what condition it is in, is it reasonable to expect clairvoyance from the test house?

It will almost certainly be necessary to have some provisional items in the documents to cater for contingencies. If these are extra over to items which will be required this is perfectly acceptable. However, the practice of asking for a unit rate such as lm2 of radar survey is to nobody’s benefit. Many of the items will have a significant mobilisation element or be priced on a day rate. These should be allowed for in documents.


Assessment of the dvnamic behaviour of

Natural frequency Mode (&-- 1 .oo H1 1.59 v1 1.92 v 2 2.59 v 3 2.81 H1 3.14 v 4 3.44 T1 3.63 v 5

V6 T2

4.00 4.31

Aberfeldy GRP plashc cable-stayed footbridge

Direction

lateral vertical vertical vertical lateral

vertical torsional vertical vertical

torsional

~ _ _ _ _ ~

R. L. Pimentel, MSc University of Shefield (formerly Lecturer, Federal University of Paraiba, Brazil)

P. Waldron, MSc, DIC, PhD, CEng, MICE Department of Civil & Structural Engineering, University of ShefSield

W. J Harvey, BSc, PhD, CEng, MICE Department of Civil Engineering, University of Dundee

Introduction The application of Advanced Composite Materials (ACM) in bridges is relatively new. One of the advantages of ACM is its high strength to weight ratio, which reduces the self-weight of the structure and makes handling and installation much easier (Mufti et al, 1991). However, this much reduced self-weight requires attention to be paid to the dynamic performance since such composite structures may be more susceptible to vibrations.

The aim of this presentation is to describe the (GRF') dynamic tests carried out on the Aberfeldy glass reinforced plastic cable-stayed footbridge. Specifically, the tests comprise an ambient vibration survey for the determination of the natural

(bi Mode V i Natural Freqmcy 1 % H z

(:)Plan

\

.& , /F

Fig 1. Footbridge main mode shapes

frequencies and mode shapes whereas damping values are determined by additional tests (heel drop and free-vibration decay). Pedestrian tests are performed, enabling an assessment of the footbridge with respect to the vibration serviceability requirements of BS 5400.

Ambient vibration survey

Measurement of the natural frequencies and mode shapes of the deck An ambient vibration survey was used for the determination of the natural frequencies and mode shapes of the structure using two accelerometers positioned at several points along the deck. The main natural frequencies are given in Table 1. The mode shapes associated with the first natural frequencies in both the vertical and horizontal directions are given in Fig. 1.

Fig. 2 shows the relationship between first natural frequency and span length for a variety of 67 footbridges (CEB, 1991). The test result of the first natural frequency for the Aberfeldy bridge (1.59 Hz, 63m span) compares very favourably with the trend line in Fig. 2, indicating that the constituent material of the bridge is much less important than span length in determining the fundamental frequency.

Measurement of damping values of the deck in the vertical direction Equivalent viscous damping values were determined from heel drop tests and from the free-vibration decay part ofjumping tests. The results for the first three vertical modes are presented in Table 2.


-0.73 f o = 33.6 L

0 Steel

A Composite 3 I Concrete

m

6- '$ Aberfeldy Footbnd

Walking Range U

0 , , I I ~ / ' I ' I ' I

0 10 20 30 40 50 60 Span L Iml

Fig 2. Footbridgefundamental frequency as a function of span

Table 2 Equivalent viscous damping values ( in % of critical damping)

1st mode ( V l ) 2nd mode (V2) 0.94 3rd mode (V3) 1.20 1.09 ~

* not possible to identify from the filtered response

Additional results from jumping tests Additional findings from the jumping tests were:

In the expression above, f, is the pacing rate of the pedestrian, which is to be equal to the fundamental natural frequency of the footbridge whereas 1 SON is the amplitude of the pedestrian load related to its first harmonic. It can be concluded that these serviceability calculations would be completely applicable to footbridges made of any construction material. This comment is important since the Code is originally intended to be applicable to bridges made of steel andor concrete.

The filtered acceleration time-response of the footbridge on the antinode of the first mode shape (middle of the main span) was recorded. Comparing with the code limits, we have:

slim = 0.5 5= 0.5

ameS = 2.1 4 d s 2 (measured value)

= 0.63m/s2 (Code limit)

which is more than three times higher. Some possible solutions (based on the principles of dynamics) for the improvement of the vibrational performance of the tested footbridge will be checked out against new tests and numerical calculations.

Frequency tuning of the structure This is aimed at making the natural frequencies of the structure outside the range of the normal range of pacing rates of pedestrians (1.4 - 2.6Hz). The fundamental natural frequency of the structure can be either increased (high tuning) or lowered (low tuning) in the attempt to avoid the critical range.

Low tuning This is achieved by adding mass to the structure. From a dynamic point of view, adding mass implies an increase of the time in which the resonance condition is reached when the structure is excited at its natural frequency. On the other hand, when applying the Code requirements to evaluate the level of vibrations, two drawbacks occur when adopting this solution:

the Code acceleration limit decreases, as it is a function of the natural frequency of the structure.

the time a pedestrian would have to cross the footbridge increases as he is to walk with a slower pacing rate equal to the reduced natural frequency.

(a) There is no significant change in damping values along the free-vibration decay part of the jumping tests, which is an indication of linear behaviour of the structure inside the range of excitation under serviceability conditions. Jumping tests done with an increasing number of subjects revealed a significant alteration of the natural frequencies of the structure, which is a consequence of the relative lightness of the construction materials.

This compensates for the increase in the time taken for the resonance condition to be reached.

The conclusion is that low tuning does not appear to be a feasible solution for the problem.

(b)

Serviceability tests The main aim of this group of tests was to check the serviceability of the structure against pedestrian-induced vibrations, according to the requirements of BS 5400, the Code of Practice for bridge structures. To do this, a pedestrian was required to walk with a pacing rate which coincides with the fundamental natural frequency of the footbridge.

The Code specifies a limit for the maximum acceleration the footbridge can experience when its fundamental natural frequency in unloaded conditions is below 5Hz, with a correction factor for frequencies in the range 4 to 5Hz:

'[im = 0.5 6 (m/s2), where fo is the first vertical natural frequency of the footbridge

Calculation of the acceleration is performed (analytically) assuming that the dynamic pulsating load applied by a pedestrian moves with a speed vt = 0.9 fo and is determined by the expression below:

F = 180 sin(27tfot) (N) ( 1 )

High tuning This is achieved by increasing the stiffness of the footbridge. The static deflections decrease and so the dynamic behaviour is improved. Two possibilities can be considered:

increasing the diameter of the cables: considering the design of the footbridge, in which the back stays are fixed to the ground by aluminium ties, an increase in their stiffness implies a direct reduction on the flexibility of the tower top and consequently a reduction on the amplitude of the vertical dislocations of the deck. increasing the height of the edge beams: this can be done without compromising the aesthetics of the bridge.

Use of Tuned Mass Dampers (TMDs) This established solution has been used successfully to damp vibrations of lively footbridges. A relation of 1/100 between the mass of the structure and the mass of the damper has been proved to give satisfactory results (Jones et al, 1979). In this case, due to the relatively low self-weight of ACM bridges, a TMD could be effectively designed having a relatively small mass.

IStructWSECED seminar 'Analysis & testing of bridges' 39

Concluding remarks The footbridge has shown a dynamic performance which is apparently similar to the one expected if conventional materials were adopted in its construction. The association of low damping values and (expected) low natural frequencies is responsible for the high level of accelerations measured. Damping values of near constant value were found during measurements at different vibration levels. This is an indication of linear behaviour inside the range of in-service excitations. The vibration serviceability requirements of BS 5400 seem appropriate to be applied to ACM structures. However, as a consequence of the relative lightness of the construction materials, the influence of the mass of the pedestrians should therefore be considered in the dynamic design due to the relatively high live-to-dead load ratio.

Acknowledgments The authors would like to thank the Engineering and Physical Sciences Research Council for financial support of this research. Mr. Pimentel is sponsored by CNPq and by the Federal University of Paraiba, Brazil, to undertake his studies at Sheffield University. The assistance of the Aberfeldy Golf Club for providing access to the footbridge and permission for the tests, and A. Pavic and M. Petkovski of the University of Sheffield and G. Ripley and T. Sullivan of the University of Dundee for technical assistance during the tests is gratefully acknowledged.


Recent experience in dynamic monitoring of a multispan bridge

G. P. Roberts, BSc, CEng, MIMechE W. S. Atkins, Bristol

Introduction This paper gives a summary of some recent work which uses the traffic-induced vibration of a bridge as a means of integrity monitoring.

Dynamic monitoring Dynamic monitoring is a form of non-destructive testing which uses the vibration characteristics of a structure to indicate its integrity. Any change in the structure will result in a change its natural vibration modes. So, at least in principle, monitoring the vibration behaviour allows deterioration to be detected.

______ Monitonng 1 Change in the bndge

I structure charactenstics ~ Change in vibration

I

Past experience Attempts to use dynamic monitoring for diagnosing structural deterioration in bridges have up to now only met with limited success. The general requirements for a dynamic monitoring system are:

a representative analytical model of the bridge, to predict its natural modes of vibration - for example, using finite element methods a monitoring system, which provides measurements of the natural modes of the bridge - for example, using accelerometers criteria by which significant changes in the vibration behaviour can be judged.

Often in the past, dynamic monitoring of bridges has been limited by:

.

. 0

the use of analytical models which only predict the simplest natural modes the use of large vibrators to excite the modal response of the bridge - this often involved partial or total closure of the bridge to traffic the difficulty involved in gathering sufficient vibration data to adequately identify anything more than the simplest modes changes in the vibration behaviour due to ‘normal’ environmental effects, which tend to mask other ‘abnormal’ structural changes changes in strength are not always associated with changes in stiffness, and are therefore not reflected in the vibration behaviour.

In fact it is the higher modes of the structure which are more sensitive to structural changes and until recent years the computing power to analyse such modes cost-effectively simply was not available.

As with most computer-based technology, major advances have been made in recent years in finite element modelling, and in vibration measurement and analysis, bringing them into the domain of day-to-day engineering tools. These advances provide the engineer with powerful analytical tools to understand the higher order natural modes of the bridge.

Recent experience The following section summarises the results of work carried out recently on Friarton bridge in Perth. This is a 9-span box girder motorway bridge on the M90.

The results of the work have been encouraging. The main achievements are summarised in the following paragraphs:

Finite element techniques have been used to model the bridge’s dynamic behaviour sufficiently accurately to gain a clear understanding of how the bridge works. A typical vertical bending mode of the bridge is shown in Fig 2. It has been shown that normal traffic loading on the bridge can be relied upon to excite a useful number of modes of the structure. A typical vibration spectrum is shown in Fig 3. Peaks in the spectrum have been numbered and correspond

Fig I . General view of bridge

7 - Mode5, 143Hz \

Fig 2. Analytical predictions of bridge mode shapes

IStructEYSECED seminar ‘Analysis & testing of bridges’ 41

to natural modes of the bridge. This result clearly showed that shakers were not necessary to excite the natural modes of the bridge. An efficient and effective method has been demonstrated for measuring the mode shapes of a long span bridge. Using rucksack mounted equipment the first fifteen mode shapes of the 800m multiple-span bridge were measured in 2 days. The global natural modes of the bridge have been measured in cold weather in March and compare well with the predictions from the analytical model. The comparison of natural frequencies is shown in Fig 4 below. A second set of measurements was carried out on the bridge in warm weather in July. These measurements showed that the first 12 mode shapes were repeatable. An overall decrease in the natural frequencies of 2.5% occurred, which is associated with the increased temperature of the bridge. A comparison between the mode shapes is shown in Fig 5 below. At this stage it was clear that ‘normal’ environmental changes, which could account for up to 3-4% changes in natural frequency, would effectively mask any ‘abnormal’ structural changes, which might result in natural frequency changes of as little as 1 %. Monitoring was therefore carried out continuously over a 12 month period, to understand, and hence isolate, the environmental effects on the bridge natural modes. Temperature was found to be the primary cause of natural mode change. The effect on frequency was found to be reasonably linear, and affected all modes by the same proportion. Mode shapes were found to be reasonably stable, independent of the temperature effect. The temperature effect on natural frequency is shown in Fig 6.

The well defined nature of the temperature effect on the natural modes means that this effect can be removed from the measurements, giving the potential for ‘abnormal’ structural changes to be detected. Analytical predictions indicate that the natural frequencies and mode shapes of the higher order modes are sensitive to structural change. Methods have been developed which have the potential to detect ‘abnormal’ structural change using measured natural frequencies and mode shapes, independent of temperature effects.

Conclusions Bridge natural modes have been successfully measured under normal traffic loading conditions. Shakers have not been necessary to excite the natural modes of the bridge. Quick and efficient methods have been developed for measuring full mode shapes of a long span structure. The natural modes have been measured with a high degree of repeatability on different days.

The results to date show that there is potential, using the higher order modes of the bridge, to detect structural change. ‘Normal’ variations in the bridge vibration modes, due to environmental changes, show a linear trend in natural frequency with temperature, but do not affect the mode shapes.

Methods have therefore been developed which have the potential to detect ‘abnormal’ structural change. These methods use the measured natural frequencies and mode shapes of the bridge and are unaffected by changes in temperature.

Work is currently in hand to demonstrate that predictable changes to the natural frequencies and mode shapes do occur, when structural changes to the bridge are made. This will be demonstrated during a major strengthening programme on the bridge which is now in progress.

B a In 2

r 2 10 * -

H 0 ,

2 0-

(0

a

n z I I I Frequency ( H r )

Fig 3. Spectrum of measured vibration with modes indicated

i lini of / perfect match

,111 2 1

Predicted Frequency (Hz)

Fig 4. Measured and predicted natural frequency comparison

42

Fig 5. Repeatability of mode shapes

Fig 6. Overall effect of temperature on natural frequency


It is believed that the work described here significantly advances the possibilities of dynamic monitoring as an aid to detection of deterioration. Perhaps one day all long-span bridges will be monitored this way.

Acknowledgement W. S. Atkins gratefully acknowledge the funding and support of the Roads Directorate of The Scottish Office Industry Department. The project is being managed on their behalf by TFU Scotland. The views expressed in this paper are not necessarily the views of T U nor of The Scottish Office.

IStructE/SECED seminar ‘Analysis & testing of bridges’ 43

Attendance list

Mr M. Ahluwalia, Sir William Halcrow & Partners Mr I. Ahmed, Acer Consultants Ltd Mr C. S. Atkins, Surrey County Council

Mr C. P. Barker, Flint & Neill Partnership Mr A. P. Barwell, Union Railways Ltd Mr R. D. Bellamy, W S Atkins Consultants Ltd Dr A. Bensalem, Napier University Mr G. T. Bessant, London Underground Ltd Mr R. A. Blackwell, Rodney Blackwell consulting Engineers

Mr P. K. Brooke, Ove Arup & Partners Mr J. S. Carlton, Lloyd’s Register Mr K. C. Chan, Highways Agency Ms C. Colla, University of Edinburgh Mr J. S. Coutts, Fugro Ltd

Dr N. Davidson, University of Edinburgh Mr G. Davison, IStructE Informal Study Group on Management & Maintenace of Bridges Dr P. Das, Highways Agency Mr T. M. Da Silva, London Underground Ltd Mr D. K. Doran, Consultant

Professor A. Elnashai, Imperial College

Mr P. D. Fox, Acer Consultants Ltd Ms J. B. F. Frandsen, Dar Al-Handasah

Mr J. Hambly, Defence Research Agency Mr R. Harben, British Waterways Mr C. Harding London Underground Ltd Mr K. M. Hanison, Transport Research Laboratory Dr W. Harvey, Dundee University Mr P. Healy, Parkman Consulting Engineers Mr D. Hemmingfield, Hertfordshire County Council Mr A. Hough, British Rail Research

Dr T. J. Ibell, University of Cambridge

Mr P. Jenkins, London Underground Ltd Mr C. C. Jones, Acer Consultants Ltd Mr F. Jones, Consultant

Mr N. Kain, Railtrack Mr S . Kemp, Technotrade Mr M. Kennard, W S Atkins - London Mr G. King, Lothian Regional Council Ms E. Kirkham, Gloucestershire County Council Mr S . A. Knight, Mott MacDonald Mr R. H. Y. KO, Highways Agency

Dr J. D. Littler, Building Research Establishment Mr A. K. A. Lorans, Institution of Structural Engineers Dr L. A. Louca, Queen Mary & Westfield College Dr M. Lowe, Imperial College

Dr J. Maguire, Lloyd’s Register Dr D. Mackay, Lloyd’s Register Mr J. Maheswaran, London Underground Ltd Dr I. F. Markey, Norwegian Public Roads Mr M. Marques, W S Atkins - London Dr J. B. Menzies, SCOSS Dr C. R. Middleton, University of Cambridge Mr J. M. Moriarty, London Underground Ltd

Dr J. S. Owen, University of Nottingham

Mr A. Packham, Railtrack

Mr B. Pavlakovic, Imperial College Mr S. Penny, DHV (UK) Ltd Mr G. Perks, Northumberland County Council Mr A. Pickett, Highways Agency Mr R. Pimentel, University of Sheffield Dr C. D. Posner, Hertfordshire County Council

Mr M. J. Ray, Rendel Palmer & Tritton Mr S. Revess, S R Design Associates Mr G. Roberts. W S Atkins

Mr B. Sadka, Highways Agency Mr S. Sanmugalingam, Sir William Halcrow & Partners Professor R. Severn, University of Bristol Mr B. Simpson, OBE, Vice-President, The Institution of Structural Engineers Mr S. Stahl, Corporatiton of London Mr A. Stowe, Defence Research Agency Mr P. J. Swanton, Mott MacDonald

Mr B. Temple, British Rail Research

Dr C. Williams, University of Plymouth Dr M. S. Williams, University of Oxford Dr J. Wood, Structural Studies & Design Ltd

44 lStructE/SECED seminar ‘Analysis & testing of bridges’

Analysis and Testing of Bridges

Documents

fatih sultan mehmet

imperial college mr

london swlx 8bh01712354535

mott macdonald mr

highways agency mr

hessle anchor block

alternative load path

finite element model