1. Introduction
After opening up the country and lifting more than two centuries of rigorous trade restrictions under the Tokugawa Shogunate, modern technologies for the construction of infrastructure were introduced in Japan in 1868 during the Meiji Restoration. During this period, people began to use metal as a structural material. New bridges built around the 1870s began to be made using cast and/or wrought iron with riveting techniques; Kurogane Bridge in Nagasaki (with a span of 27 m) is known as the first modern cast iron bridge in Japan. Starting in 1895, steel quickly replaced wrought iron as the favored material for metal bridges. However, the Great Kanto Earthquake disaster in 1923 revealed the many drawbacks of wrought iron, and high-tensile steels were later adopted for bridges. Kiyosu Bridge—an eyebar-chain self-anchored suspension bridgespanning 183 m over the Sumida River in Tokyo—is a representative example of riveted bridges from this era [1].
In the 1900s, the use of reinforced concrete for infrastructure became more common, and pre-stressing technology was widely used for numerous types of simple structural elements in the 1950s. The availability of advanced steel and reinforced concrete technologies provided an impetus for the construction of longer-span bridges and taller buildings [2].
In an archipelago country such as Japan, long-span bridges are an essential element of transportation links to connect islands and circumvent bays. Long-span bridge construction commenced after the end of World War II, when Saikai Bridge, a 243.7 m steel arch bridge in Nagasaki, was completed in 1955. The construction of long-span bridges intensified rapidly afterward, and peaked during the Honshu–Shikoku Bridge Project (HSBP) [3], [4]. The HSBP is a national project that was carried out between 1975 and 1999 to link the Honshu and Shikoku Islands. The links comprise major long-span bridges, with the Akashi Kaikyo Bridge as the longest suspension bridge and the Tatara Bridge as the longest cable-stayed bridge [5]. At present, there are 15 suspension bridges, three cable-stayed bridges, and one truss bridge in Japan with spans greater than 500 m. A part of these bridges are listed in Table 1 [6].
Table 1. Bridges in Japan with a longest span over 500 m and special control or maintenance features.
| Bridge name | Year of completion | Longest span (m) | Bridge type | Remarks |
|---|---|---|---|---|
| Akashi Kaikyo | 1998 | 1991 | Suspension | Active control during construction/wake galloping of hangers, dry air injection for cable, permanent monitoring system |
| Minami Bisan–Seto | 1988 | 1100 | Suspension | Dry air injection for cable |
| Third Kurushima Kaikyo | 1999 | 1020 | Suspension | Active control during construction/dry air injection for cable |
| Kita Bisan–Seto | 1988 | 990 | Suspension | Dry air injection for cable |
| Shimotsui–Seto | 1988 | 940 | Suspension | Dry air injection for cable |
| Tatara | 1999 | 890 | Cable-stayed | — |
| Onaruto | 1985 | 876 | Suspension | Dry air injection for cable |
| Innoshima | 1983 | 770 | Suspension | Dry air injection for cable |
| Akinada | 2000 | 750 | Suspension | Dry air injection for cable |
| Hakucho | 1998 | 720 | Suspension | Active control during construction, dry air injection for cable, permanent monitoring system |
| Kanmon | 1973 | 712 | Suspension | During rehabilitation |
| First Kurushima Kaikyo | 1999 | 600 | Suspension | Active control during construction, dry air injection for cable |
| Meiko Central | 1998 | 590 | Cable-stayed | Active control during construction |
| Rainbow (Tokyo Port) | 1993 | 570 | Suspension | Active control during construction, dry air injection for cable, seismic monitoring system |
| Oshima | 1988 | 560 | Suspension | Dry air injection for cable |
| Toyoshima | 2008 | 540 | Suspension | Dry air injection for cable |
In 1964, Tokyo was selected to host the Summer Olympics, which marked the start of a construction boom in Japan. This event transformed major cities with new infrastructure development after their successful emergence from the postwar period of basic reconstruction. Technological developments in structural engineering were remarkable in the 1970s–1990s, especially in the area of seismic/wind design and construction. Until the 1960s, regulations regarding earthquake resistance were prohibitive for the construction of high-rise buildings. It was not until 1968 that the first high-rise office tower was constructed. The 147 m, 36-story Kasumigaseki Building was completed in 1968 as a modern office high-rise building in Tokyo, remarkably ending the perception that tall buildings were unfitting for Japan and other earthquake-prone regions. Since then, many more tall buildings have been constructed in urban areas, and the number of buildings with heights exceeding 200 m has now reached more than 40.
For many years, the highest building in Japan was the Yokohama Landmark Tower, an office and shopping complex with a height of 296 m that was built in 1993. In 2014, a new building complex of offices and a shopping center named Abeno Harukas was completed with an overall height of 300 m [7]. This is currently the tallest occupied building in Japan. Meanwhile, the tallest structure in Japan is now the Tokyo Sky Tree (a broadcasting, restaurant, and observation tower), which was built in 2012 with a height of 634 m.
Civil structures have an important characteristic that differentiates them from other industrial products. Each structure is unique from others. They are designed according to different local geographical and geological conditions and are built from different construction materials using different construction technologies. None of the civil structures are exactly the same; they are very different from mass-produced goods. Another important characteristic is that civil structures are built as assets of society, and are expected to function for a long time. The lifetime of structures is generally long, and can be on the order of hundreds of years for some. Therefore, it is essential to ensure that the built infrastructure satisfies the design requirements and design assumptions. It is also important to examine the condition of the actual structure in order to ensure high durability and a long service life. From another perspective, developments in the design and construction of long-span bridges and tall buildings have always involved sophisticated models, analysis, and advanced technologies. The uncertainties associated with these new models, analysis, and technologies must be quantified and monitored to ensure the correctness and efficacy of their application. For these purposes, long-term or short-term campaign-type structural monitoring has been conducted in large and important structures, and very valuable information has been obtained [8].
Structural monitoring in general covers many aspects and methodologies. In this paper, however, we emphasize the application of structural monitoring to bridges, buildings, and roadway pavement using vibration techniques. Vibration is used as a measure in structural monitoring because the vibration response represents both the global and local performance of a structure. Vibration-based monitoring is an important work in the advancement of research on earthquake- and wind-resistant structures. In relation to seismic monitoring, the Japan Meteorological Agency (JMA) has conducted seismic motion observation, mainly with displacement seismographs. However, acceleration records are more convenient than displacement records for research on earthquake-resistant structures, because accelerations become direct input motions to a structure in its equation of motion. In addition, conventional seismographs might record vibrations with saturated amplitudes during large earthquakes.
The development of strong-motion instruments in Japan began after the 1948 Fukui Earthquake (see Table 2 for detailed information on the earthquakes discussed in this paper), which was a large and damaging inland earthquake. In 1953, the Strong-Motion Accelerometer Committee (SMAC) developed the SMAC accelerometer (named after the committee) [9]. The SMAC accelerometer is an analog device that can record triaxial accelerations up to 1 g (g = 9.8 m·s−2). A total of 25 SMAC accelerometers were installed in buildings in 1956, marking the start of earthquake-induced response monitoring of structures in Japan. By 1975, about a thousand SMAC accelerometers had been installed in Japan [10]. The installations acquired valuable earthquake measurements during events such as the 1964 Niigata Earthquake, the 1968 Tokachi-oki Earthquake, and the 1978 Miyagi-oki Earthquake. Some lessons learned from these measurements have been utilized for the design of tall buildings in Japan.
Table 2. List of the earthquakes mentioned in this paper.
| Earthquake names | Date of occurrence | Epicenter | Magnitude | Maximum intensity (JMA intensity) and maximum recorded peak ground acceleration (PGA) |
|---|---|---|---|---|
| 1948 Fukui Earthquake | 28 Jun 1948 | 36.16° N, 136.22° E | Mw = 7.1 | JMA intensity 6 |
| 1964 Niigata Earthquake | 16 Jun 1964 | 38.37° N, 139.22° E | Mw = 7.6 | JMA intensity 6, PGA > 450 cm·s−2 |
| 1968 Tokachi-oki Earthquake | 16 May 1968 | 40.90° N, 143.35° E | Mw = 8.2 | JMA intensity 5 in Aomori and Hakodate |
| 1978 Miyagi-oki Earthquake | 12 Jun 1978 | 38.19° N, 142.03° E | Mw = 7.4 | JMA intensity 5 in Sendai |
| 1995 Great Hanshin (Kobe area) Earthquake | 17 Jan 1995 | 34.59° N, 135.07° E | Mw = 6.9 | JMA intensity 7, PGA 833 cm·s−2 |
| 2001 Geiyo (near Hiroshima) Earthquake | 24 Mar 2001 | 34.083° N, 132.526° E | Mw = 6.7 | JMA intensity 5−, PGA 852 cm·s−2 |
| 2003 Tokachi‐Oki (Hokkaido) Earthquake | 26 Sep 2003 | 41.78° N, 143.86° E | Mw = 8.3 | JMA intensity 6−, PGA 988 cm·s−2 |
| 2004 Chuetsu-Niigata (Niigata Prefecture) Earthquake | 23 Oct 2004 | 37.3° N, 138.8° E | Mw = 6.6 | JMA intensity 7, PGA 800–1700 cm·s−2 |
| 2004 Kushiro-Oki (Hokkaido) Earthquake | 29 Nov 2004 | 42.946° N, 145.274° E | Mw = 7.0 | JMA intensity 5+, PGA 879 cm·s−2 |
| 2004 Kii Peninsula (Mie Prefecture) Earthquake | 5 Sep 2004 | 33.15° N, 137.14° E | Mw = 7.3 | JMA intensity 5−, PGA 400 cm·s−2 |
| 2007 Chuetsu-Niigata-Oki Earthquake | 16 Jul 2007 | 37.5° N, 136.6° E | Mw = 6.6 | JMA intensity 6 upper, PGA 670 cm·s−2 in Kashiwazaki City |
| 2011 Great East Japan (Tohoku) Earthquake | 11 Mar 2011 | 38.322° N, 142.369° E | Mw = 9.0 | JMA intensity 7 (max), PGA 2000–2900 cm·s−2 |
Mw: moment magnitude.
Since the late 1980s, the SMAC seismograph system has been considered outdated. New types of accelerometers were developed with the advancement of information technology and digital recording. Modern accelerometers are now often of the small micro-electromechanical systems (MEMSs) type. More recently, wireless sensor networks have been under very active research and development. Applications for the practical use of wireless sensor networks for structural monitoring in Japan have been demonstrated in several studies [11], [12]. There are two issues in the application of wireless sensor networks: the robustness of wireless communication and the reduction of power consumption. More development in these aspects is expected in future.
During their development, the data accumulated from vibration-based structural monitoring systems has been used in many ways, including monitoring structure behavior during extreme events and providing feedback for the redevelopment of design and new construction. As longer bridges and taller buildings were increasingly built in the 1990s, structural monitoring systems were supplemented into the construction process and vibration controlbecame more common for large structures. As structures become more advanced and more monitoring systems are implemented, data accumulates on structures’ behavior under various environmental and loading conditions. This data can be used to evaluate structural condition, indicate possible damage, and facilitate decision-making on repair and/or retrofit. Monitoring data can also be used to support the maintenance and management of structures.
In this paper, we provide a review on the structural monitoring strategies and practices of civil structures with an emphasis on research works and implementations in Japan and on the authors’ experiences. Some new and unexpected findings from the vibration monitoring data of real bridges and buildings are reported in order to emphasize the importance of monitoring. This paper consists of two main sections that address the monitoring of bridge structures and building structures, respectively; the monitoring cases in each section are divided based on the type, strategy, and purpose.
2. Monitoring of bridge structures
2.1. Monitoring for the design verification of long-span bridges
Dynamic performance is an important consideration in the design of long-span bridges. Because of their flexibility and low damping, various types of vibration may occur during the lifetime of a long-span bridge. Aerodynamic stability and seismic response are the main concerns in design; hence, dynamic tests during the design phase and after the completion of construction are common in the early stages of the development of long-span bridges in Japan. In some cases, monitoring systems are installed during construction and remained in place for several years after construction completion. The data from this type of monitoring has been utilized to confirm design assumptions regarding seismic and wind loads. In the following section, we describe several research works related to monitoring for the design verification of loading and structural response against wind and earthquakes.
2.1.1. Monitoring for wind-induced design verification
During the initial development of long-span bridges in Japan, issues associated with the quantification of forces—especially wind loading—were significant in the design process. Limited past experiences and large uncertainties within design assumptions made verification in scaled experimental testing an important design step. From 1973 to 1975, a one-tenth sectional bridge girdermodel was constructed for the HSBP, which includes the Akashi Kaikyo Bridge, in order to verify the wind-resistant design method. The truss-stiffening girder with a length of approximately 8 m was tested in natural wind (Fig. 1). The measured drag coefficients were in agreement with the estimation based upon wind tunnel testing [13].
Fig. 1. (a) Large-scale bridge model (courtesy of the Honshu–Shikoku Bridge Authority); (b) comparison between in situ model response (observation) and estimation based on wind tunnel experiment (average wind speed 12.6 m·s−1, elevation 0°).This attempt during an earlier stage of bridge design eventually developed into the use of permanent measurement installation for monitoring during the service life of a bridge, with advances in sensor technology and information systems. For example, Fig. 2 [14], [15] shows the instrumentation at the Akashi Kaikyo Bridge at the time of completion [16]. This figure shows an example of the relationship between average wind speed and lateral displacement, as measured by global positioning system (GPS). Displacement of the bridge can be measured with reasonable accuracy by GPS due to the prominent length of the bridge span. The observed values are close to the design average values and the maximum values are conservative, with a reasonable margin [17]. In addition, the measured data of the power spectrum, turbulence intensity, and spatial correlation of natural wind at various long-span bridges in the HSBP has been studied and verified against the design assumptions and has been found to be within reasonable conservative margins [14], [15].
Fig. 2. (a) Akashi Kaikyo Bridge monitoring system; (b) relationship between ten-min average wind velocity and mid-girder lateral displacement at the Akashi Kaikyo Bridge; (c) comparison between measurement and design code (courtesy of the Honshu–Shikoku Bridge Authority). 1A and 4A refer to anchorages; 2P and 3P refer to main pylons; β is wind skew angle (i.e. angle between oncoming wind with respect to the normal of the bridge axis. Reproduced from Ref. [14] with permission of J-STAGE, © 2010 and from Ref. [15] with permission of J-STAGE, © 2006.Wind tunnel experiments are commonly performed using a sectional model in order to determine the aerodynamic stiffness and damping. The variations of factors with respect to wind velocity are known from the wind tunnel test, but are rarely confirmed from full-scale testing of completed long-span bridges. In an effort to understand and confirm the aerodynamic stiffness and damping, structural monitoring using wind-induced responses was carried out at the Hakucho Bridge—a three-span suspension bridge with a total length of 1380 m (330 m + 720 m + 330 m) (Fig. 3). After the construction was completed, and before being opened to traffic in 1998, the bridge was densely instrumented, with accelerometers placed every 30–55 m. The wind-induced structural responses were recorded for several weeks for a wide range of wind velocities.
Fig. 3. (a) Hakucho Bridge and (b) campaign-type sensors layout for ambient vibration measurement (Z1–Z19 denote the locations of sensors). Identified change in: (c) aerodynamic damping and stiffness with respect to wind velocity; (d) damping and stiffness due to friction force with respect to wind velocity. Reproduced from Ref. [18] with permission of American Society of Civil Engineers, © 2005 and from Ref. [19] with permission of Elsevier Ltd., © 2007.Structural performance during ambient vibration and strong wind was evaluated using inverse analysis [18], [19]. The results showed that in general, the natural frequencies decrease as the wind velocities increase, while the damping ratios increase as the wind velocities increase. The contributions of aerodynamic and friction forces with respect to wind velocity were quantified. The results suggested that the contribution of aerodynamic force was much smaller than the effect of friction force at the bearing. The aerodynamic force’s contribution was about one percent of the friction force, and its behavior agreed with that of the aerodynamic forces obtained from wind tunnel results (Fig. 3(c)). These findings, to authors’ knowledge is the first attempt to clarify and compare the aerodynamic forces obtained from a wind tunnel with the results of full-scale monitoring of a long-span bridge in the world.
In addition, the locality effect of the phase difference concentrated mainly at the edge of the girders was observed. This finding can be used to determine the contribution of additional damping and stiffness caused by friction force at the bearings (Fig. 3(d)). The damping and stiffness due to friction force at the bearings display clear trends; namely, low damping and high stiffness during small vibration. When the wind speed increases, the damping also increases, which is when the bearings become unstuck, whereas the stiffness decreases due to the increasing flexibility of the structure. Detailed explanation on the effect of wind speed on stiffness and damping are given in Ref. [18]. The influence of additional stiffness and damping due to the friction force at the bearings was also observed from long-term seismic monitoring of the bridge, as reported in Ref. [20].
2.1.2. Monitoring for seismic design verification
Seismic responses obtained from structural monitoring have been used to verify seismic design. One example of such a case is the Tatara Bridge (Fig. 4), the longest cable-stayed bridge in Japan, which was strongly excited by the 2001 Geiyo (near Hiroshima) Earthquake (moment magnitude Mw = 6.7). The maximum ground acceleration at the bridge site was 144 cm·s−2. Observation of the seismic responses revealed that the actual seismic load in terms of the response spectra calculated from the recorded ground motion was below the design specification. The seismic behavior of the bridge was studied by means of simulation analysis in order to verify the structural model and assumptions. The simulation results were found to be in good agreement with the observed responses [21].
Fig. 4. (a) Tatara Bridge; (b) comparison between design spectra and observed response spectra in the 2001 Geiyo (near Hiroshima) Earthquake; (c) Tatara Bridge monitoring system (unit: m) (courtesy of the Honshu–Shikoku Bridge Authority). EW: east–west; NS: north–south; V: vertical. P1, P2, P3, and P4 are the piers.An important engineering problem in the seismic analysis of long-span bridges is the spatial variation of seismic ground motions. This variation is a result of the time lag of seismic wave propagation, since the foundations of long-span bridges are separated by the long span. One such analysis was performed on Onaruto Bridge using observed responses during the 1995 Great Hanshin (Kobe area) Earthquake (Mw = 6.9). The study revealed that the spatial variation of the ground motions increased the vertical response of the girder [22]. A similar tendency of the increase of the vertical girder response was observed at other long-span bridges, including the Akinada Bridge during the 2001 Geiyo (near Hiroshima) Earthquake [23].
Also observed during the 2001 Geiyo (near Hiroshima) Earthquake was the failure of the First Kurushima Kaikyo Bridge [24]. The observed seismic ground motion was applied to a dynamic three-dimensional finite element analysis, and it was verified that the failed center stay rods performed as they had been designed. Reanalysis of observed data from extreme events can provide valuable information for verifying and updating the design.
Another important case of design verification is the estimation of the appropriate damping values and mechanisms. Estimation of damping values and mechanisms is quite difficult due to the complexity of the mechanisms involved and the sensitivity of the estimated values to excitation conditions. Nevertheless, some studies have used the seismic records of instrumented long-span bridges to clarify damping mechanisms and estimate the values. For example, Kawashima et al. [25], [26] utilized seismic records from over 33 earthquakes on the Suigo Bridge—a 290.45 m long steel box-girder two-span continuous cable-stayed bridge—to clarify the damping characteristics of the tower and deck. It was found that the damping ratio correlated with the measured accelerations, depending on the structural components and the direction of the excitations.
At the Tsurumi Tsubasa Bridge, strong motion observation was performed starting from the opening of the bridge, and several records of significant earthquakes have been obtained. During the 23 October 2004 Chuetsu-Niigata (Niigata Prefecture) Earthquake, the records show that vibration continued for a long time, indicating that the damping in the response displacement amplitudewas small [27]. Based on seismic records from ten earthquakes on the Yokohama Bay Bridge, it was found that the damping ratios for lower modes in both the vertical and lateral direction have an increasing trend with an increase in earthquake magnitude [28]. For a small earthquake magnitude, the average damping ratios are found to be 2%; these increase significantly, up to 4%–5%, as the earthquake magnitude increases, resulting in a larger value than the previously suggested 2%.
2.2. Bridge monitoring for the verification of seismic isolation system performance
Seismic protective technology using isolation has been implemented on bridges in Japan for over 30 years. The first seismically isolated bridge in Japan was the Miyagawa Bridge. The bridge girder is a three-span continuous non-composite steel girder with a length of 105.8 m. Located in Haruno-cho, Shizuoka Prefecture, the bridge was opened in March 1991 and was one of eight bridges selected from across the country for the pilot construction project of a base-isolation system. Lead rubber bearings (LRBs) were adopted as a seismic isolation device. To examine the earthquake response characteristics of the seismic isolation bridge, strong-motion accelerometers were installed on the Miyagawa Bridge at the pier cap, girder, and free-field. On April 25, 1992, an earthquake with a magnitude of 4.9 on the JMA scale was recorded, with the earthquake epicenter in Shizuoka. These were the first seismic records from the monitoring system of a seismically isolated bridge in Japan. Analysis of the records helped to confirm some important aspects that had been adopted in the design of the base-isolated bridge [29].
The following section describes several case studies of the monitoring of seismically isolated bridges with short-medium spans and long spans. Just before the large 1995 Great Hanshin (Kobe area) Earthquake, base-isolation systems had been implemented on a few bridges in Japan, some of which were instrumented with seismic monitoring systems. The 1995 Great Hanshin (Kobe area) Earthquake was the first time that such base-isolation technologies experienced strong shaking. Structural monitoring systems for seismically isolated bridges were installed with the original objective of confirming the performance of the isolation system under seismic excitation. Since the isolation technique was considered to be new and advanced, it was necessary to verify the accuracy of the design procedure and models for such bridges with recorded responses obtained from actual events.
A detailed investigation on the performance of base-isolated bridges during a large earthquake was conducted on the Matsunohama Viaduct (Fig. 5) [30], an elevated bridge located on the Hanshin Expressway Bayshore Route in the Kansai area in the west of Japan. The bridge was opened in 1994 and was the first application of base-isolated bridges within the Hanshin Expressway. The viaduct is a four-span continuous steel box-girder bridge with a bridge length of 211.5 m and a curvilinear radius of 560 m. It is located about 35 km east–southeast from the epicenter of the 1995 Great Hanshin (Kobe area) Earthquake. The Matsunohama Viaduct has two base-isolated bridges: Bridge A and Bridge B. In a study by Chaudhary et al. [30], the performance of the isolation system during the 1995 Great Hanshin (Kobe area) Earthquake was investigated by means of a system identification method. The study showed that it was possible to capture the overall behavior of the base-isolated Matsunohama Viaduct bridges with simple equivalent linear, two degrees-of-freedom (2-DOF) lumped mass models. The study demonstrated that the base-isolation system performed satisfactorily because it effectively decoupled the superstructurefrom the substructure, such that the spectra of the girder contained only the dominant superstructure frequency and filtered out other frequencies.
Fig. 5. Seismically isolated Matsunohama Viaduct. (a) General layout and strong-motion instrumentation; (b) photo of the bridge pier caps and seismic bearing; (c) location of sensors of the pier cap and girder. BH: bore hole. P20–P32 denote the location of piers. Reproduced from Ref. [30] with permission of American Society of Civil Engineers, © 2000.Fig. 6 shows the results of observation from 1995 Great Hanshin (Kobe area) Earthquake main shock and aftershocks. It was observed that the natural frequencies decreased with increasing earthquake intensity for both bridges. The decrease in the first modal frequency was related to the reduction of the bearing stiffness as the seismic isolation functioned. The reduction in the second modal frequency was due to the reduction in stiffness of the substructure. The damping ratio of the first mode, which is associated with the isolator, was larger in Bridge B than in Bridge A. This difference was due to the properties of the isolation system that was employed in both bridges. A similar seismic monitoring system and response analysis was established and carried out for Yama-age Bridge, which was seismically isolated using high-damping rubber (HDR) bearings, and which was affected by the 1995 Great Hanshin (Kobe area) Earthquake. The performance of the isolation bearings identified from the actual earthquake reasonably agreed with the predicted performance based on loading testing conducted prior to installation within the possible range of modeling uncertainties—that is, the effect of friction [31].
Fig. 6. Variation of modal parameters with earthquake intensity. (a) natural frequency (ω0); (b) damping ratio (ξ). Subscripts A and B denote the Bridges A and B, respectively; while subscripts 1 and 2 denote the 1st and 2nd modes, respectively.In both of the cases mentioned above, the performance of the isolation bearings was evaluated further by comparing the identified stiffness and damping coefficient with the equivalent linearized experimental values. Friction force caused by minor structural elements can affect the dynamic behavior of the superstructure and increase the modeling uncertainty considerably, which may degrade the effects of the base isolation. The influence of minor structural elements, as observed from this study, was utilized as feedback to improve the design of the seismic isolation system and its implementation on highway viaducts [32].
Based on the observations of the seismic responses of isolated bridges during the 1995 Great Hanshin (Kobe area) Earthquake, the seismic isolation system was considered to be more advantageous than the lateral force distribution structure using rubber bearings, because the damping performance reduced the response displacement to a greater extent. As a result, the use of seismic isolation systems increased significantly after the 1995 Great Hanshin (Kobe area) Earthquake. On the national highways administrated by the Japan’s Ministry of Land, Infrastructure and Transport and Tourism (MLIT), the seismic isolation design was adopted for about 120 bridges and 200 sections of newly constructed bridges. In addition, seismic isolation was applied to retrofitexisting bridges [32].
Long-span bridges are more flexible than short- and medium-span bridges. The seismic load of a long-span bridge is usually considered to be lower than the wind load. However, the inertia load caused by the superstructure of a long-span bridge can be significant due to the heavy weight of a long-span girder. Therefore, reducing the seismic load by further lengthening the natural periods is commonly explored by employing the isolation technique. This can be achieved by isolating a girder from the tower using a specially designed tower–girder connection system. However, while isolation minimizes the seismic load, unexpected excessive displacement due to flexibility may result. Therefore, treatment of excessive motion should be carefully considered for long-span bridge isolation.
Strategies for lengthening the natural period of a long-span bridge using a tower–girder connection have been used in some of the long-span cable-stayed bridges in Japan. For example, the Meiko Triton cable-stayed bridge in Nagoya utilizes an elastic cable connecting the main tower with the girders in the longitudinal direction in order to elongate the natural period to about 2–3 s. The Tsurumi Tsubasa Bridge in Yokohama adopts an elastic restriction cable system between the tower and girders, and uses a vane-type oil damper to control the motion. Another example is the Higashi-Kobe Bridge, in which a concept involving all-free movable supports in the longitudinal direction at all bearing supports at the towers and pier caps was adopted to lengthen the natural period. To improve the safety and increase damping, a vane-type oil damper was installed at the girder end [32].
Dense-array permanent seismic monitoring systems have been installed in several long-span bridges in Japan, including those that use a seismic isolation system. An example of such a monitoring system was deployed at the Yokohama Bay Bridge (Fig. 7). The bridge was constructed on soft soil, which is one example of a situation in which a special seismic isolation system may be needed. The bridge is located near an active fault and is close to the epicenter of the 1923 Great Kanto Earthquake. These conditions have made seismic performance a major concern. Therefore, to confirm the seismic design and to monitor the bridge performance during earthquakes, a comprehensive dense-array monitoring system was installed. The objectives of the monitoring system are to evaluate seismic performance, perform verification and comparison with the seismic design, and observe for possible damage. The analysis of the seismic records focused on the local component link–bearing connection (LBC), which is a seismic isolation device.
Fig. 7. (a) Yokohama Bay Bridge and the layout of the permanent seismic monitoring system; (b) location of the link–bearing connections (LBCs) of Yokohama Bay Bridge; (c) typical LBC at the tower (left) and at the end pier caps (right). Two of the three typical first modes of the Yokohama Bay Bridge identified from earthquakes: (d) hinged–hinged mode; (e) fixed–fixed mode. S: sensor of girder; T: sensor on tower; K: sensor on foundation; B: sensor on end piers; G: sensor on ground; φ: relative modal displacement ratio (non-dimensional); R: right; L: left. Reproduced from Ref. [28] with permission of John Wiley & Sons, Inc., © 2005.As part of its dynamic monitoring system, the bridge is equipped with 85 channels of accelerometers at 36 locations (Fig. 7). Seismic records with varying amplitudes obtained from six major earthquakes between 1990 and 1997 were analyzed to evaluate the global and local performance of the bridge [28], [33]. System identification of a long-span bridge under seismic excitation requires the consideration of non-unique ground-excitation records measured along the bridge and excitation in multiple directions. An investigation of the performance of the bridge’s LBCs was primarily carried out by observing the first longitudinal mode, analyzing the response between the pier caps and girder, and conducting an analysis using a finite element model. Based on these analyses, the following findings were obtained (Fig. 7): ① System identification resulted in three typical first longitudinal modes that were different on the relative modal displacement between the end pier caps and girder. These modes are the hinged–hinged mode, the mixed hinged–fixed mode, and the fixed–fixed mode. The latter two modes are variations on what was the strongly expected mechanism (i.e., the hinged–hinged mode). The response analysis of the relative displacement between the end pier caps and girder confirmed these findings. ② The LBC has yet to function as a full-hinged connection during small earthquakes. Therefore, higher natural frequencies due to the stiffer connection were observed. The mixed hinged–fixed mode was observed during a moderate earthquake. Full-hinged connections at both of the end pier caps were mostly observed during large earthquakes.
2.3. Bridge monitoring for the verification of a structural retrofit
After the 1995 Great Hanshin (Kobe area) Earthquake, a recommendation to reconstruct and repair the highway bridges that suffered damage from the earthquake was issued by the Ministry of Construction on 27 February 1995. The three-year retrofit program was completed in 1997. Other important bridges that were designed prior to the 1995 Great Hanshin (Kobe area) Earthquake were also retrofitted in subsequent years to meet the standards and specifications determined after the 1995 Great Hanshin (Kobe area) Earthquake. These included the three large cable-supported bridges of the Metropolitan Expressways: the Yokohama Bay Bridge, Rainbow Bridge, and Tsurumi Tsubasa Bridge [34].
Structural monitoring can provide insights into the retrofit process and can verify the efficiency of retrofit action. This has been the case for the Yokohama Bay Bridge, a cable-stayed bridge with a central span of 460 m that has been continuously monitored by a densely distributed sensor system since 1990. In 2005, a seismic retrofit program was implemented on the bridge for Level 2 earthquake safety assurance according to Japan’s bridge seismic code. The retrofit program considered two types of maximum credible earthquakes: magnitude 8 far-field or moderately far-field large earthquakes taking place in the subduction zone of the Pacific Plate and near-field inland earthquakes occurring beneath the site or close to the site.
The retrofit program utilized the previous monitoring results and simulations of identified potential damage for both types of ground motion, and concluded that significant damage would occur on the towers and bearings under such excitations. Furthermore, the far-field ground motion would create more damage and induce 1.5 m longitudinal displacement of the girder. Accordingly, five retrofit strategies and a fail-safe design concept were introduced [34].
As explained in the previous section, it was realized from the seismic monitoring of the Yokohama Bay Bridge that there is a possibility that the LBCs may not function properly during a large earthquake. In such a case, an excessive moment at the bottom of the end pier cap may result, and the LBCs may fail and create uplift deformation at the girder. To prevent such conditions, a seismic retrofit of the bridge was conducted and a fail-safe scenario was provided. The seismic retrofit of the bridge, which was performed in 2005, employed a fail-safe design in which the girder ends are connected to the footing using pre-stressed cables to prevent uplift of the girder end, as shown schematically in Fig. 8 [35].
Fig. 8. (a) Photos and (b) schematic figure of a fail-safe design system using pre-stressed cables that connect the girder end and the ground to prevent uplift at the Yokohama Bay Bridge [35].2.4. Bridge monitoring for the verification of structural control systems
Vibration control is commonly applied to long-span bridges in order to suppress wind-induced vibration. Conventional methods are passive vibration control (e.g., oil dampers for girder motion) and tuned mass dampers for tower oscillation. As structures become larger and more flexible, greater capacity is required for control devices, and active control—which introduces an artificial external force to suppress vibration—becomes an attractive option. Because active control naturally requires measurement in order to modulate the control force, monitoring forms the basis of this new technology.
From a practical standpoint, active control is considered to be superior to passive devices when: ① multiple vibration modes are present; ② natural frequencies change, as is typically observed during construction; and ③ installation space is limited, so compact devices are preferred. These three conditions apply to flexible long-span bridges, especially during the construction stage. Hakucho Bridge is one of the bridges that was actively controlled during construction. A pendulum-type control device, shown in Fig. 9(a), was installed near the top of the tower, as illustrated in Fig. 9(b) [36]. This system is a so-called “hybrid” system, which incorporates both passive control effects provided by the pendulum motion and an active control force provided by a rack and pinion with electric motors.
Fig. 9. Vibration control system of Hakucho Bridge tower. (a) Pendulum-type control device; (b) location of installation at Hakucho Bridge tower (unit: mm). T.P: height.Monitoring is also important to ensure that the control systems deployed on long-span bridges function as intended, and to provide feedback on the efficacy of the control system performance. Table 1 lists a few long-span bridges for which active control was applied during construction.
2.5. Bridge monitoring during extreme events
Compared with other developed regions in the world, such as Europe and North America, Japan is widely known as a country that is prone to natural disasters. The high intensity of seismic activity and the frequent occurrence of seasonal strong winds and typhoons have made evaluation against such extreme events the focus of structural monitoring in Japan. After verifying assumptions related to the initial design, retrofit, and structural control performance, a bridge monitoring system is implemented to monitor the structural performance during extreme events and to determine their effects on the structures. Monitoring of structural performance during extreme events has two main purposes: ① to test the appropriateness or limitations of the design assumptions during extreme loading conditions, and ② to observe the possibility of new structural behaviors that were not considered in the design. Both purposes are very useful as feedback for the improvement of future structural designs. The following sections describe cases involving two main extreme events: strong winds or typhoons, and large earthquakes.
2.5.1. Bridge monitoring during strong winds and typhoons
The number of typhoons that hit Japan varies per year; however, on average, about 11 typhoons make landfall in Japan every year. The typhoon season typically starts in summer—in August—and can bring heavy rainfall and strong devastating winds. Several strong typhoons have caused heavy casualties and damaged civil infrastructures. The monitoring of infrastructure such as long-span bridges during strong wind and typhoon events is important, not only to provide assurance of bridge performance, but also to give insight into actual structural behavior and unexpected phenomena that may require special treatment, as feedback for future design. Some important case studies are explained below.
At the Akashi Kaikyo Bridge, two strong typhoons passed through about two and a half years after its opening. The bridge’s structural monitoring system recorded the wind and bridge response during Typhoon Vicky (7th typhoon of 1998) and Typhoon Bart (18th typhoon of 1999). (See Table 3 for detailed information on the typhoons mentioned in this paper.) Miyata et al. [15]conducted an analysis of the full-scale recorded data with a focus on the power spectral density (PSD), spatial correlation of the wind-speed fluctuation, and response of the deck. This was the first opportunity to evaluate the response of such a long-span bridge under actual strong wind conditions, and the study notes that actual static lateral deflections of the deck in the middle of the center span agreed well with the analytical value (Fig. 10).