Tài liệu Fundamental study on structural damage detedtion in vibration response of long span suspension bridge

DEPARTMENT OF CIVIL ENGINEERING YOKOHAMA NATIONAL UNIVERSITY FUNDAMENTAL STUDY ON STRUCTURAL DAMAGE DETECTION IN VIBRATION RESPONSE OF LONG-SPAN SUSPENSION BRIDGE （長大吊橋の振動応答 （長大吊橋の振動応答からの 振動応答からの構造 からの構造損傷検知に関する 構造損傷検知に関する 基礎的研究） By NGUYEN DANH THANG A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering Academic Advisors: Prof. Hitoshi Yamada Prof. Hiroshi Katsuchi Assoc. Prof. Eiichi Sasaki September, 2010 ABSTRACT Nowadays, a great number of long-span bridges were constructed all over the world. Most of long-span bridges which are common over the sea are particularly difficult to maintain because of their specific conditions: a severe natural environment including strong winds, strong tidal currents and salt air, a large degree of continuing deformation of structures, an extremely large variety of structural members and materials, and the need to cope with the fatigue of structural steel especially in the case of any bridges which carry trains as well as road traffic. However, their service time have to be more than 100 years because these bridges are very expensive to design, construct and maintain. As a result, the health, durability, and safety of these bridges in a long-term service period are now attracting a lot of scientists and engineers. An issue arising will be a methodology how the structural damages can be detected from the monitoring data. Besides, with increasing of span and slenderizing of structure, long-span bridges become more and more sensitive to wind. For a long-span bridge, with limited torsional stiffness, wind-induced forces, such as self-excited force and buffeting force, can cause destructive phenomena. Self-excited forces causing flutter are in general dependent on the geometric profile of the bridge deck section, angle of wind attack and wind velocity expressed as reduced frequency. Meanwhile, buffeting is defined as the unsteady loading of a structure by velocity fluctuations in the oncoming flow. In addition, earthquake is also extreme excitation for long-span bridge and can cause a lot of structural damage. Therefore, both wind and earthquake are required special attention for long-span bridge. i Although many advances in design, construction as well as maintenance have been developed day to day, many problems of structure still remain unknown or unsolved. The ability to detect structural damages in a bridge before it endangers the structure has been of interest to engineers for many years. This study was carried out to investigate how the structural damage affects the windinduced and earthquake responses of a long-span suspension bridge. Besides, this study was focused on how to detect damage of long-span bridge in vibration response. To illustrate this purpose, a detailed finite element model of a long-span bridge was developed and verified using field data, making this model as accurate as possible in representing the actual structural behavior. Using this finite element model, the reliability analysis of the bridge is performed considering dead load, wind load and earthquake loading. After that, based on the realistic deteriorations, various types of structural damages of a long-span bridge are simulated to facilitate the discussion. All of the dynamic data for comparing damaged with undamaged cases were generated numerically from the finite element model. The obtained results showed that monitoring data can be used for detecting some damages cases, but existing monitoring systems are not sufficient for damage detection. Lessons archived from this study are expected not only to maintain this bridge but also to improve our understanding of the real bridge performance as well as to provide useful feedbacks for future design. ii ACKNOWLEDGEMENTS After going through almost three years of hard work, it is time to thank all those who have pulled me through this period and made my stay at Yokohama National University a pleasant one. First of all, I would like to express my sincere gratitude and thanks to my advisors: Professor Hitoshi Yamada, Professor Hiroshi Katsuchi and Associate Professor Eiichi Sasaki for their kind advices, valuable suggestions, invaluable guidance, moral support and effective encouragement throughout the course of this study. I would like to extend my gratitude to Professor Tatsuya Tsubaki and Associate Professor Kimitoshi Hayano for their helpful comments, suggestions and serving as members of the examination committee. I take this opportunity to thank Ms. Matsuda, secretary of Wind and Structures Laboratory, for her kindness, support and spiritual encouragement. My special gratitude is due to all my dear friends for making my time spent at Yokohama National University an unforgettable memory as well as for what they helped for me to overcome lot of difficulties in foreign environment. iii I would like to thank to the Ministry of Education, Science and Culture of Japan (Monbukagakusho) for the full financial support and the research facilities they provided during my study. Lastly, but not the least, I want to express all my gratitude to my wife and my son, my father and mother, and other members of my great families for all the trust, support that they gave me. Without their love, encouragement, inspiration and sacrifice, this work could hardly be completed. Nguyen Danh Thang September, 2010 iv TABLE OF CONTENTS ABSTRACT i ACKNOWLEDGMENTS iii TABLE OF CONTENTS v LIST OF FIGURES viii LIST OF TABLES xiv 1 2 INTRODUCTION 1 1.1 General 1 1.2 Necessary of study 4 1.3 Objectives and outline of study 6 MONITORING SYSTEM AND DAMAGE DETECTION OF LONGSPAN BRIDGE 7 2.1 Monitoring system of long-span bridge 7 2.1.1 Structural health monitoring 7 2.1.2 Application of structural health monitoring for long-span bridges 2.2 3 11 Damage detection for long-span bridge 21 2.2.1 Deterioration of long-span bridge 21 2.2.2 Damage detection for long-span bridge 25 NUMERICAL SIMULATION OF LONG-SPAN BRIDGE v 31 3.1 Finite element model of long-span bridge 31 3.2 Wind speed and wind loading simulation 35 3.2.1 Introduction 35 3.2.2 Aeroelastic forces on long-span bridge 38 3.2.3 Wind speed simulation 41 3.2.4 Wind loading simulation 43 3.2.5 Wind-induced response of long-span bridge 43 3.3 4 49 DAMAGE DETECTION BY WIND-INDUCED RESPONSE 52 4.1 Damage detection by global vibration response 52 4.1.1 Damage assumption 53 4.1.2 Changes in structural behaviors 54 4.1.3 Other results of damage analysis 65 4.2 5 Earthquake simulation Damage detection by local vibration response 77 4.2.1 Introduction 77 4.2.2 Power Spectral Density functions 78 4.2.3 Changes in local frequencies 79 DAMAGE DETECTION BY EARTHQUAKE RESPONSE 101 5.1 Degree of nonlinearity method 102 5.1.1 Frequency response function 102 5.1.2 Hilbert transform of FRF 102 5.1.3 Degree of nonlinearity 103 5.2 Damage detection by earthquake response 104 5.2.1 Case of damage study 104 5.2.2 Change in DON value 105 vi 5.2.3 6 Change in local frequency 106 CONCLUSION REMARK 110 6.1 Conclusions 110 6.2 Recommendations 111 REFERENCES 113 vii LIST OF FIGURES Page Figure 1.1 The collapse of Tacoma Bridge 2 Figure 1.2 The collapse of I-35W Bridge 3 Figure 2.1 Akashi Kaikyo Bridge 12 Figure 2.2 Monitoring system of Akashi Kaikyo Bridge 14 Figure 2.3 Recorded wind parameters and responses of Akashi Kaikyo Bridge during typhoon 16 Figure 2.4 Tatara Bridge 17 Figure 2.5 Installed accelerometer in Tatara Bridge 18 Figure 2.6 Sensor system of Tatara Bridge 18 Figure 2.7 Tsing Ma Bridge 19 Figure 2.8 Sensor system of Tsing Ma Bridge 19 Figure 2.9 Location of sensor in cross frame of Tsing Ma Bridge 20 Figure 2.10 Mean speed and direction of onset wind recorded in Tsing Ma Bridge 20 Figure 2.11 Damage on cable by corrosion 21 Figure 2.12 Bridge deterioration prediction in Japan 21 viii U α Figure 2.13 Seto Bridge 22 Figure 2.14 Deteriorating grade of maintenance ways of Seto Bridge 23 Figure 2.15 Health of long-span bridge in the USA 23 Figure 2.16 Broken wires and reduction of cross section of Alvsborg Bridge, Sweden 24 Figure 2.17 Fatigued strand of Severn Bridge, England 25 Figure 3.1 3D finite element model of Akashi Kaikyo Bridge 32 Figure 3.2 Cross section of Akashi Kaikyo Bridge deck 33 Figure 3.3 Static aerodynamic coefficients of Akashi Kaikyo Bridge as a function of wind angle of attack 33 Figure 3.4 First symmetric lateral deflection mode of model 34 Figure 3.5 First symmetric vertical deflection mode of model 34 Figure 3.6 First symmetric torsion mode of model 34 Figure 3.7 Three-degree-of-freedom model of bridge deck 39 Figure 3.8 Maximum wind speed at 10m of altitude at Kobe City and Akashi Town 42 Figure 3.9 Location of Akashi Kaikyo Bridge and two wind recorded places 42 Figure 3.10 Two components of time history wind speed fluctuation at 20 m/s 44 Figure 3.11 Applied three components of wind forces on the bridge deck 44 Figure 3.12 Time history wind forces at 20 m/s applied at middle span of the bridge 45 Figure 3.13 Time history wind-induced response at middle main span of the bridge at 20 m/s Figure 3.14 Time history wind-induced velocity at middle main span of the bridge at ix 46 47 20 m/s Figure 3.15 Time history wind-induced acceleration at middle main span of the bridge at 20 m/s 48 Figure 3.16 Time history ground motion of two applied earthquakes on the bridge 49 Figure 3.17 Earthquake applied on the bridge 50 Figure 3.18 Response at middle span of the bridge during EQ1 50 Figure 3.19 Response at middle span of the bridge during EQ2 51 Figure 4.1 54 Figure 4.2 Figure 4.3 Figure 4.4 Figure 4.5 Figure 4.6 Figure 4.7 Figure 4.8 Figure 4.9 Assumed severe damaged cases of Akashi Kaikyo Bridge Changes in mean relative value of structural response for type 1 of damage Changes in mean relative value of structural response for type 2 of damage Changes in mean relative value of structural response for type 3 and type 4 of damage Changes in mean relative value of structural velocity for type 1 of damage Changes in mean relative value of structural velocity for type 2 of damage Changes in mean relative value of structural velocity for type 3 and type 4 of damage Changes in mean relative value of structural acceleration for type 1 of damage Changes in mean relative value of structural acceleration for type 2 of damage x 56 57 58 59 60 61 62 63 Figure 4.10 Changes in mean relative value of structural acceleration for type 3 and type 4 of damage 64 Figure 4.11 Changes in whole structure basic natural frequencies of structure 65 Figure 4.12 Changes in higher order of whole structure natural frequencies 66 Figure 4.13 Changes in STD of structural response for type 1 damage 68 Figure 4.14 Changes in STD of structural response for type 2 damage 69 Figure 4.15 Changes in STD of structural response for type 3 and type 4 damages 70 Figure 4.16 Changes in STD of structural velocity for type 1 damage 71 Figure 4.17 Changes in STD of structural velocity for type 2 damage 72 Figure 4.18 Changes in STD of structural velocity for type 3 and type 4 damages 73 Figure 4.19 Changes in STD of structural acceleration for type 1 damage 74 Figure 4.20 Changes in STD of structural acceleration for type 2 damage 75 Figure 4.21 Changes in STD of structural acceleration for type 3 and type 4 damages 76 Figure 4.22 Three selected typical hanger for damage detection 80 Figure 4.23 Damage occur in the medium hanger 81 Figure 4.24 Response at damage point in case of 5% of hanger area was lost 82 Figure 4.25 Local frequencies at damage point in case of 5% of hanger area was lost 83 Figure 4.26 Damage point response in case of 30% of hanger area was lost 84 Figure 4.27 Local frequencies at damage point in case of 30% of hanger area was lost 85 Figure 4.28 Local frequencies at upper neighbor point in case of 30% of hanger area was lost Figure 4.29 Local frequencies at lower neighbor point in case of 30% of hanger area xi 86 87 was lost Figure 4.30 Damage occur in the longest hanger 88 Figure 4.31 Damaged point response in case of 50% of hanger area was lost 89 Figure 4.32 Local frequencies at damaged point in case of 5% of hanger area was lost 90 Figure 4.33 Local frequencies at neighbor point in case of 50% of hanger area was lost 91 Figure 4.34 Damage occur in the longest hanger 92 Figure 4.35 Damaged point response in case of 50% of hanger area was lost 93 Figure 4.36 Local frequencies at damaged point in case of 5% of hanger area was lost 94 Figure 4.37 Local frequencies at neighbor point in case of 5% of hanger area was lost 95 Figure 4.38 Change in frequency between healthy and damage condition 96 Figure 4.39 PSD error for three analyzed cases 97 Figure 4.40 Changes in local frequencies at middle span of bridge for all severe cases 98 Figure 4.41 Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Changes in local frequencies at middle side span of bridge for all severe cases Analysed earthquake cases of long-span bridge Changes in DON value in case of the medium hanger was damaged under EQ1 Changes in DON value in case of the medium hanger was damaged under EQ2 Changes in DON value in case of the shortest hanger was damaged under EQ1 xii 99 105 105 106 106 Figure 5.5 Figure 5.6 Changes in DON value in case of the shortest hanger was damaged under EQ2 Changes in DON value along the bridge in case of the medium hanger was damaged under EQ2 106 107 Changes in local frequency in case of the medium hanger lost 30% of Figure 5.5 cross section area for upper, damage and lower neighbor point, 108 respectively Figure 5.6 PSD error in case of the medium hanger damaged xiii 109 LIST OF TABLES Page Table 2.1 Table 2.2 Measurement items at Akashi Kaikyo Bridge Criteria for deterioration evaluation of galvanized structure of Seto Bridge 13 22 Table 3.1 Comparison of natural frequencies between FEM and measured 33 Table 4.1 Change in natural frequencies according to damages cases 67 Table 4.2 Parameter of selected hangers 80 xiv CHAPTER 1 INTRODUCTION 1.1. General As an indispensability of development, a great number of long-span bridges were constructed all over the world today. Most of long-span bridges which are common over the sea are particularly difficult to maintain because of their specific conditions: a severe natural environment including strong winds, strong tidal currents and salt air, a large degree of continuing deformation of structures, an extremely large variety of structural members and materials, and the need to cope with the fatigue of structural steel especially in the case of any bridges which carry trains as well as road traffic. However, their service time have to be more than 100 years because these bridges are very expensive to design, construct and maintain. As a result, the health, durability, and safety of these bridges in a long-term service period are now attracting a lot of scientists and engineers. Some longspan suspension bridges are monitored for the purpose of their health monitoring. An issue arising will be a methodology how the structural damages can be detected from the monitoring data. Besides, with increasing of span and slenderizing of structure, long-span bridges become more and more sensitive to wind. For a long-span bridge, with limited torsional stiffness, wind-induced forces, such as self-excited force and buffeting force, can cause destructive phenomena and need special attention, especially after the collapse of suspension Tacoma Bridge on November 7th, 1940 by normal wind (Figure 1.1). Self-excited forces causing flutter are in general dependent on the geometric profile of the bridge deck section, angle of wind attack and wind velocity expressed as reduced frequency. Meanwhile, buffeting is 1 defined as the unsteady loading of a structure by velocity fluctuations in the oncoming flow. Figure 1.1. The collapse of Tacoma Bridge Because of important role and expensive cost, long-span bridges are require inspection from time to time to ascertain that they are still safe and capable of withstanding various environmental effects. Such inspections and associated non-destructive testing procedures can reveal progressive damage, and allow appropriate repair measures to be taken before the damage deteriorates to the extent of making the structure unserviceable. Even for new infrastructure, particularly large structures with high initial construction costs, it is now recognized that monitoring programs are desirable right from the outset in order to detect any signs of damage as early as possible, and allow appropriate interventions to be taken. Programs of this nature, if properly implemented, can extend the useful life of the structure quite considerably, with the utility value gained more than justifying the costs of the monitoring itself. This philosophy has gained considerable momentum in areas such as Japan, China, and Korea, where long-span bridges are abundant. However, this thinking is more widespread, and much research on the issues of monitoring, damage detection and long-term performance of structures is going on not only in Asia, but also in America (USA and Canada) and Europe (Germany, Belgium, UK, etc.). On the other hand, structural health monitoring of bridges is a very complicated issue. The principal developments concentrate on issues that the ordinary bridge owner is not interested in. A common language between technology achievements and bridge owners has not been found and the method statement that appeal to bridge owners are lacking. 2 Besides, the development community has not been able to explain the new methods do not eliminate the problem of aging or damaged bridges but are only better at being able to identify problems. However, the performed monitoring campaigns are often so expensive that they are only scientific interest. In addition, the cost of Structural Health Monitoring of bridges is also expensive. Cost will depend on the depth of investigation and vary from simple quick investigation until permanent online Structural Health Monitoring. In general, an in-depth inspection currently costs approximately 10,000 Euro per 100 m of bridge. In order to be able to equip Structural Health Monitoring system, bridge owners are required sufficient capital to invest in the expensive monitoring equipment necessary. The cost for a 32-channel Structural Health Monitoring system is in the region of 100,000 Euro with a life expectation of 3 years [1]. Figure 1.2. The collapse of I-35W Bridge (photo by BBC) Although many advances in design, construction as well as maintenance have been developed day to day, many problems of structure still remain unknown or unsolved. The ability to detect structural damages in a bridge before it endangers the structure has been of interest to engineers for many years. Currently, bridge condition assessment is largely carried out by visual inspection at intervals of one to five years, followed by more detailed 3 examination and analysis if necessary. However, it is possible for significant damage to have developed in the intervening period, putting structures at risk. There have been some disastrous failures of bridges due to undetected progressive damage in the past, e.g. the collapse of I-35W Bridge in Minneapolis, USA without any warning on August 1st, 2007 (Figure 1.2). Therefore there is considerable interest in continuous monitoring of bridges. 1.2. Necessary of study The design of civil structures is characterized by two main features: load-carrying capacity and serviceability. However, each structural system undergoes various environmental and loading influences during its service life, which can cause a significant damage accumulation. Consequently, the structural carrying capacity and serviceability are enormously affected. Therefore, the need for reliable nondestructive evaluation technique and detection of damage at the earliest possible stage has been pervasive throughout the civil engineering community in the last decade. The process of implementing damage detection strategies can be referred to as “structural health monitoring”. The so-called vibration-based health monitoring techniques rely on the fact that damage causes changes in the local structural damping (energy dissipation) and stiffness. As a consequence, the global dynamic properties of the structure, e.g. eigen frequencies, mode shapes, modal damping, etc., should be influenced. Structural Monitoring is basically an activity where actual data related to civil structures is measured and registered. This has been performed through all times by responsible designers, contractors and owners with almost identical objectives - to check that the structures behave as intended. Historically the activity has required specialists, has been time consuming and hence costly and as a result hereof only a limited number of performance indicators - typically geometry - have been measured a periodically and supplemented by regular visual observations. At the core of any structural health monitoring framework system are the diagnostic and prognostic algorithms used to detect the presence, magnitude and extent of structural faults. The emergence of this field has led to a variety of diagnostic methods for detecting, locating and quantifying varying degrees of damage. 4 Several methods of structure lifetime estimation are known today. However, their performance is heavily influenced by the quality of the recorded data: length of the time series, presence of measurement and system noise, system excitation, etc. In addition, different types of energy dissipation could be present at any given time instant. Some of them can be associated with material properties, others with the system boundary conditions. Effect of contact friction can be observed in some cases as well. Thus, the estimation procedure requires a very careful use of numerical procedures. Moreover, an engineering understanding and critical considerations are important for a reliable identification of the presented damping properties. To have a better design for long-span bridge, the study of the wind load and earthquake load on bridge is of vital importance. Many works have been conducted on the study of damage of steel bridges [2]. However, there has been very little research on the windinduced damage especially for long-span suspension bridges. With the increase of span length of modern suspension bridges, the investigation for evaluating wind induced damage becomes more and more significant for long-span suspension bridges, which were common located at a typhoon prone region. The dynamic response against strong wind and earthquake are subjected to unknown factors those are uneasy to predict. Therefore, it is necessary to establish a monitoring system that can collect data on dynamic response of the bridge in order to verify the assumptions and constant used for the design due to strong wind and earthquake. The wind load for long-span bridges has great importance in their structural design. It usually consists of time averaged wind force and some contribution of the dynamic response due to the wind fluctuation, but there still remain uncertainties in expression of wind characteristics to define the accurate and reliable wind load. To overcome this it will be important to compile information of the wind at many bridge site. Here, as the example of monitoring results, the deformation characteristics of the bridge response due to strong wind are elucidated. By comparing the analyzed simulation results through wind tunnel test and field measured results, the reliability of the current monitoring system is confirmed. Besides, with the development of the structural health monitoring system [3, 4 and 5] for long-span suspension bridges, it becomes possible to obtain field data of dynamic response induced by a typhoon for the bridge with permanent installed monitoring system. However, 5