|Home | About | Journals | Submit | Contact Us | Français|
Aimed at the health monitoring and evaluation of bridges based on sensing technology, the monitoring contents of different structural types of long-span bridges were defined. Then, the definition, classification, selection principle, and installation requirements of the sensors were summarized. The concept was proposed that new adaptable long-life sensors could be developed by new theories and new effects. The principle and methods to select controlled sections and optimize the layout design of measuring points were illustrated. The functional requirements were elaborated on about the acquisition, transmission, processing, and management of sensing information. Some advanced concepts about the method of bridge safety evaluation were demonstrated and technology bottlenecks in the current safety evaluation were also put forward. Ultimately, combined with engineering practices, an application was carried out. The results showed that new, intelligent, and reliable sensor technology would be one of the main future development directions in the long-span bridge health monitoring and evaluation field. Also, it was imperative to optimize the design of the health monitoring system and realize its standardization. Moreover, it is a heavy responsibility to explore new thoughts and new concepts regarding practical bridge safety and evaluation technology.
As lifeline engineering, bridges are in a stage of rapid development at present, and the community has attached great importance to the operational safety of bridges. Therefore, how to ensure the healthy operation of long-span bridges has become a hot topic that is difficult to address in this field. Through the service period that ranges from decades to even a hundred years, bridges will be affected by external loads like cars, climate, and rivers. Bridges will also be affected by internal effects like material aging, fatigue, and other factors . These factors will result in inevitable damage and serious accidents. For example, on 1 August 2007, the I-35W Mississippi River Bridge in the United States collapsed . On 14 July 2011, the Wuyishan Mansion Bridge in the Fujian province of China collapsed . The lessons taught by a series of disasters have made health monitoring and evaluation for bridges a major issue.
Sensing technology belongs to modern science and its purpose is to obtain and identify the information from natural sources by sensors. With the innovation of technology, sensors that are original, intelligent, and reliable have been widely used for health monitoring of long-span bridges. For instance, sensors were installed on the Foyle Bridge , which is a continuous steel bridge with variable heights and a total length of 522 m. Their purpose was to monitor the response of vibration, deflection, and strain under the vehicle and wind load in the operation stage. Numerous acceleration sensors and strain gauges were attached to the Tsing Ma Bridge with a set of GPS (Global Position System) devices, which were used for the long-term monitoring of the service safety . Monitoring systems have also been installed in numerous bridges. For example, in Japan, the Akashi Kaikyo Bridge  has a main span of 1991 m. In Denmark, there exists the Great Belt East Suspension Bridge and the Faroe Sea-Crossing Cable-Stayed Bridge, of which the main spans are 1624 m and 1726 m, respectively. In Norway, the Skarnsundet Cable-Stayed Bridge has a main span of 530 m. Additionally in China, there exists the Su Tong Yangtze River Highway Bridge and the Run Yang Yangtze River Bridge, of which the main spans are 1088 m and 1490 m, respectively. In this paper, the Caijia Jialing River Bridge for the subway in Chongqing city, China, will be taken as an example. Its total length is 1250 m. Also, constant, on-line, and dynamic health monitoring is conducted so as to achieve the objective of real and effective evaluation of its structural state [7,8,9].
With the progress and development of technology, the current health monitoring program focuses on smart sensors and methods of reliability and safety evaluation. The implementation relies on the monitoring system . A set of complete monitoring systems has functions such as acquisition, transmission, processing, management, evaluation, early warning, etc. The whole process is completed by six systems .
With respect to bridge safety evaluation technology, the applied theories mainly include the reliability theory, analytic hierarchy process, fuzzy theory, etc.
In summary, the health monitoring of long-span bridges has achieved great progress. To obtain standardized management, many countries have promulgated relevant guidelines and procedures. For example, ISIS (Intelligent Sensing for Innovative Structures) in Canada released “The Structural Health Monitoring Guide” . Drexel University in America completed “A Case Study on the Health Monitoring Guide for Significant Bridges”  with the FHWA (Federal Highway Administration). SAMCO (Structural Assessment, Monitoring and Control Organization) released “The Structural Health Monitoring Guide”  with the European Union. In China, the department of transportation released “Technical specification for structural safety monitoring systems of highway bridges” . However, when they were applied to engineering practices, there were still some obvious shortcomings, such as the mechanism and technology of monitoring and sensing, monitoring theory and technology of the service lifetime, the processing of monitoring information , structural status evaluation, etc. Also, the combined effects of the complex service environment and disaster loads can accelerate the damage evolution of long-span bridges, which will have negative results on the health and safety of the overall structure. This should be considered with the planning and construction of super bridge projects across long-span seas or rivers, such as the Shanghai Railway Yangtze River Bridge, the Egongyan Special Track Bridge in China, the Pulau Muara Besar Bridge, etc. The monitoring and evaluation tasks will become more burdensome and difficult. As a result of the health monitoring and evaluation of long-span bridges based on sensing technology, the research status, positive innovation, and the scientific grasp of damage evolution laws need to be recognized to ensure the safety of the structures .
Via physical and mechanical properties, the health monitoring of long-span bridges is aimed at monitoring the overall structures non-destructively and constantly. Then, the location and degree of structural damages can be diagnosed. Also, the service situation, reliability, durability, and bearing capacity can also be evaluated. If the structure has an emergency or severely abnormal usage, warning signals will be triggered to provide a basis and guidance for maintenance, management, and decision-making . Due to different types of long-span bridges combined with differences in the operating environment as well as the need for monitoring functions, the monitoring contents should be chosen appropriately, which are shown in Table 1.
The sensor is a device that can detect certain specified parameters and convert them into available signals. According to the principle of operation, it can be classified by resistance, raster, piezoelectric, laser , etc. As the leading edge of health monitoring systems, the selection of the sensor should follow the principle of “stable and reliable performance which is cost-effective” and should be convenient for the integration of the health monitoring system. Also, some evaluation indexes, such as sensitivity, linearity, signal-to-noise ratio, and resolution should also meet the monitoring needs. Before the installation of the sensor, it is necessary to carry out its calibration. During the installation process, damage to the structure should be reduced.
With the rapid development of science and technology, there has been some breakthroughs in the long-term stability, reliability, durability, and other aspects of sensors. Meanwhile, some new types of practical sensors have emerged, such as the multi-directional dynamic stress monitoring sensor for concrete . However, the information acquisition of bridge health monitoring and evaluation are greatly restricted because of the complexity of the working environment, the mismatch between the long life of the bridge and the short life of the sensors, and the monitoring signals which result from multi-effects. Therefore, by using new theories and new effects, in the future it will be a hot topic to study and develop new types of long-life sensors with stronger adaptability.
Due to the limited number of measuring points in health monitoring systems, the full freedom data of long-span bridges cannot be obtained. Therefore, it is necessary to optimize the sensor configuration so as to achieve effective damage identification.
Considering the comprehensiveness, accuracy, and economic factors regarding constant response information, when sensors are arranged, the controlled sections should be selected by following multiple principles of analyzing internal forces and deformation, vulnerability, and geometric dimensions. Methods include technology of static control analysis, dynamic control analysis, and monitoring overall bridge state acquisition via finite points.
According to the criterion of minimum errors of identification (transmission), model reduction , interpolation fitting, and MAC (Modal Assurance Criterion) , some methods can be used to optimize the layout of sensors. The methods include genetic algorithms, data fusion, inversion theory, and optimization which is combined with construction monitoring.
According to the structure type of long-span bridges, the number of health monitoring points, the type of sensor, the acquisition mode of sensing information, etc., should be rationally designed:
The acquisition device can be divided into hardware and software. The hardware device should follow the standard protocol and interface with the function of constant acquisition, automatic storage, and instant display. The software device can achieve manual intervention acquisition and adjust parameters, preliminarily dealing with the output contents of sensing information and the identification of abnormal information.
Sensor devices with special software to measure and control should be provided with the software running platform, communication protocol, and response interface. As for the acquisition software of the development platform, the language should meet the requirements of health monitoring visualization where LabVIEW  and LabWindows  can be adopted.
Combined with the transmission distance, network coverage, communication facilities, and other factors, reasonable transmission modes of sensing information are respectively selected, which are then followed.
Based on the sensing information, distinguishing the abnormity caused by structural and nonstructural damage is a hot topic in this field of research. Due to the sensor deviation, noise, system failures, and other coupling factors, abnormal sensing information caused by nonstructural damage must be obtained so as to provide effective information for structural damage identification and safety evaluation.
In engineering applications, as for the processing of sensor information, the methods are aimed at sensing signals. Specifically, the time domain, frequency domain, and time-frequency domain method can be adopted for noise reduction and filtering the gross error and accidental error. Also, the distorted information can be reconstructed by the trend curve method or the neural network method. Then, the missing information can be repaired by way of interpolation, substitution, and weight adjustment .
The management of sensing information is carried out on a server which requires a fast display, efficient storage, report generation, and other functions. With an effective connection between information sensing devices and the Internet, system technologies regarding the Internet of Things can be adopted for management.
This practice has proven that a modular structure in the form of the database can achieve the hierarchical and classified management of sensing information. This database mainly includes static and dynamic databases.
Based on the response information, theoretical calculation analysis, mathematics, and mechanics, the safety evaluation of structures can be achieved via the changing characteristics indexes of the structure itself, its response, and trends. Moreover, the individual control index can be used for a comprehensive evaluation. Some advanced methods of safety evaluation are briefly introduced as follows.
(1) Safety Evaluation Based on Reliability Theory
The structural function Z is constructed and the sensing information is used to modify the structural load effect model S and resistance model R. Via the reasonable statistical analysis method, the probability of exceeding the limit state during the period of usage can be obtained, which can implement the structural safety evaluation.
When Z = R – S > 0, it indicates that the structure is in a state of reliability.
When Z = R – S < 0, it indicates that the structure has been deactivated or destroyed.
When Z = R – S = 0, it indicates that the structure is in a limited state.
(2) Safety Evaluation Based on the Deterioration Effect
According to numerous historical monitoring data, the random process Zi is constructed, searching for and extracting the characteristic information which reflects the structural state. Then, the deterioration law of the structural performance is determined. Therefore, the following Equation (1) is used to implement the structural safety evaluation.
where: μ indicates the process mean value and ξi indicates randomly varied parameters.
In Figure 1, when the structure is in a normal state, μ will be unchanged and ξi will randomly vary with no obvious trend through the whole process. When damage or safety problems occur, the changing amplitude of Zi will continue to increase and μ will sustain an irreversible monotonic trend.
(3) Safety Evaluation Based on Envelope Theory
According to the limit state principle of bearing capacity based on the plastic theory, the design value of the unfavorable combination of the loading effect must be less than or equal to that of the structural resistance. It can be expressed as Equation (2).
where: indicates the importance factor of the structure, S indicates the loading effect function, R indicates the resistance effect function, fd indicates the design value of material strength, adc indicates the geometric parameter of concrete, ads indicates the geometric parameter of rebar, Z1 indicates the calculating coefficient of the bearing capacity, ξe indicates the deterioration coefficient of the bearing capacity, ξc indicates the cross section reduction coefficient of the reinforced concrete structure, and ξs indicates the cross section reduction coefficient of rebar .
Therefore, the loading effect diagram of each section reflected by the sensing information can be compared and analyzed with the modified limit resistance effect diagram of the bridge. The purpose is to obtain the structural safety evaluation.
(4) Safety Evaluation Based on Dynamic Response
According to the sensing information, the dynamic characteristics parameters  of the structure can be obtained as well as the judgment, localization, and quantification [30,31] of the structural damage. Then the finite element model can be corrected via the reanalysis of the ultimate bearing capacity, implementing the structural safety evaluation.
Regarding the fatigue damage evaluation for a steel box girder based on sensing information, the S-N curve (fatigue curve) of the construction details can be adopted to evaluate the fatigue status according to the Miner principle. Under special meteorological conditions (such as the wind, rainfall, etc.) regarding natural factors which will affect the normal service of the bridge, it is necessary to carry out this special evaluation.
Based on the sensing information, although bridge safety evaluation has obtained some great achievements which have been successfully applied to some long-span bridges, most of them focus on the overall evaluation of historical data accumulation and accidents. So we are confronted with a technical bottleneck  for early damage warning and online safety evaluation.
The Caijia Jialing River Bridge is located in the Liangjiang New Area of the Chongqing Municipality in China. It is designed in the Rail Transit Line 6 (second-phase project) between the Jinshan Temple Station and Caojiawan Station. The total length of the bridge is 1250 m, and the main structure is a cable-stayed concrete bridge with twin towers and double cable planes where tower-beam consolidation is adopted. The layout of the spans is 60 m + 135 m + 250 m + 135 m + 60 m = 640 m, and the transverse layout is 1.5 m (cable area) + 1.4 m (maintaining road) + 4.6 m (carriageway) + 4.6 m (carriageway) + 1.4 m (maintaining road) + 1.5 m (cable area). The shape of the tower is a diamond, and the auxiliary piers are rectangular cross-sectional hollow piers. Cables adopt the steel strand and the standard strength fpk = 1860 MPa, which are protected by an HDPE (High Density Polyethylene) tube. The single cell box girder with a constant height is adopted as the main girder, and the concrete grade is C55. The layout of the main bridge is shown in Figure 2.
According to the principle and method of selecting controlled sections and the layout optimization of the measured points, the overall layout of the sensors in the Caijia Jialing River Bridge is shown in Figure 3.
Combined with the construction progress of the bridge, the health monitoring sensors were appropriately installed at the corresponding positions, as follows.
The health monitoring for the Caijia Jialing River Bridge considered loading factors including live train load, transverse rocking force, braking force, temperature force, wind load, water pressure [35,36], etc. Via the finite element simulation analysis by the professional program Midas/Civil, the sub loads were combined to establish the upper and lower bound of the envelope interval. For example, for the main girder of the bridge, the vertical deformation and the longitudinal stress of point 2 (middle position of the top slab) and point 5 (middle position of the bottom slab) were measured.
These results are shown in Figure 5, Figure 6 and Figure 7, where Section A–Section G successively represent: L/2 Secondary Sidespan next to Jinshan Temple, L/2 Sidespan next to Jinshan Temple, L/4 Mid-span, L/2 Mid-span, 3L/4 Mid-span, L/2 Sidespan next to Caojiawan, and L/2 Secondary Sidespan next to Caojiawan.
Based on the upper and lower bounds of the envelope interval, the safety coefficient r = 0.75, was determined. Then the envelope intervals  of a normal state, critical state, and degradation evaluation were established. For instance, for the vertical stiffness of the mid-span, the calculation results for the deformation in the Z-direction (DZ) are shown in Table 2.
So in the mid-span of the main girder, the envelope of the vertical stiffness was [−166 mm, 77 mm] in the normal state and was [−221 mm, −166 mm) (77 mm, 103 mm] in the critical state.
Similarly, the envelope intervals of other physical parameters such as the stress and cable forces could be established. The time-varying effect among the material characteristics was considered. With the continuous operation time, the results of the daily inspection were also organically integrated. Then, the coefficient of checking and calculation, the deterioration coefficient of the bearing capacity, the reduction coefficient of the sections, and the influence coefficient of the train live load  were introduced and the envelope interval could be reasonably revised.
Meanwhile, based on the finite element model, the static and dynamic characteristics were analyzed. The natural vibration characteristics were taken as an example . The first-order vibration mode is shown in Figure 8, and the first five orders of the frequency and vibration mode are shown in Table 3.
When installed completely, the health monitoring system should be debugged as a whole. Before the passage of the train, the time 02:00 in the early morning would be regarded as the reference time, which would commence the acquisition of monitoring information constantly, online, and dynamically. For example, the GNSS automatically collected data at 10 s intervals. Meanwhile, the acceleration sensors automatically collected data continuously. Other monitoring contents including the stress and the cable force were collected at 10 min intervals. Then, the monitoring information could be transmitted to the server by way of wired transmission. Ultimately, the processing and evaluation of data was implemented.
One day is taken as an example. The monitoring and evaluation of some measured parameters are explained.
In order to clearly describe the results of the static characteristics, the evaluation coefficient is defined in Equation (3).
where: indicates the measured value of the loading effects, indicates the theoretical value of the loading effects. When is less than or equal to 1, it indicates that the working performance of the structure is good. When is greater than 1, it indicates that the working performance is not ideal and some defects may exist.
In a word, when combined with the results above, the evaluation coefficient was less than 1. So the static characteristics of the bridge were normal.
Meanwhile, based on Figure 15, the measured first mode frequency (0.835 Hz) was greater than the calculated frequency (0.686 Hz). It indicated that the dynamic characteristics were normal.
In summary, the static and dynamic characteristics of the Caijia Jialing River Bridge were all normal. The bridge structure was also in a normal state.
This work was supported by the National Science Fund for Distinguished Young Scholars (51425801), the National Key Research and Development Program of China (2016YFC0802202), the Chinese Academy of Engineering Consulting Project (2015-XZ-28, 2016-XY-22), the National Natural Science Foundation of China (51278512, 51508058, 11404045), the Social Livelihood Science and Technology Innovation Special of Chongqing (cstc2015shmszx30012), the Science and Technology Project of Guizhou Provincial Transportation Department (2016-123-006, 2016-123-039, 2016-123-040), the Science and Technology Project of Yunnan Provincial Transportation Department (2014 (A) 27), and the Communications Science and Technology Project of Guangxi Province of China (20144805).
J.Z. arranged all the work in the project and provided keen insight for this manuscript; X.L. and H.Z. investigated the current situation in this research field; H.Z. established the finite element model and obtained the calculated results; J.Y. analyzed the data; X.L. and R.X. wrote the manuscript.
The authors declare no conflict of interest.