condition assessment of bridge structures using structural health monitoring. ... Engineering, Design and Managing Urban Infrastructure, 26 March 2009, Queensland .... consider the damage detection methods that don't require a comparison of the intact ..... tools (Wong & Chen, 2001) because of its many distinct merits.
QUT Digital Repository: http://eprints.qut.edu.au/
Wang, Liang and Chan, Tommy H.T. (2009) Review of vibration-based damage detection and condition assessment of bridge structures using structural health monitoring. In: The Second Infrastructure Theme Postgraduate Conference : Rethinking Sustainable Development: Planning, Engineering, Design and Managing Urban Infrastructure, 26 March 2009, Queensland University
© Copyright 2009 Queensland University of Technology
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring Liang WANG1, T.H.T. CHAN2
Abstract As a part of vital infrastructure and transportation networks, bridge structures must function safely at all times. However, due to heavier and faster moving vehicular loads and function adjustment, such as Busway accommodation, many bridges are now operating at an overload beyond their design capacity. Additionally, the huge renovation and replacement costs always make the infrastructure owners difficult to undertake. Structural health monitoring (SHM) is set to assess condition and foresee probable failures of designated bridge(s), so as to monitor the structural health of the bridges. The SHM systems proposed recently are incorporated with Vibration-Based Damage Detection (VBDD) techniques, Statistical Methods and Signal processing techniques and have been regarded as efficient and economical ways to solve the problem. The recent development in damage detection and condition assessment techniques based on VBDD and statistical methods are reviewed. The VBDD methods based on changes in natural frequencies, curvature/strain modes, modal strain energy (MSE) dynamic flexibility, artificial neural networks (ANN) before and after damage and other signal processing methods like Wavelet techniques and empirical mode decomposition (EMD) / Hilbert spectrum methods are discussed here.
Keywords: structural health monitoring (SHM); vibration-based damage detection (VBDD); dynamic analysis; condition assessment, bridge structures
Introduction Owing to the global roaring economy and fast urban sprawling over last decades, bridge structures are regarded as very important portions of transportation networks. They are taking more obligations of life care than ever before. In other words any damages or even collapse of bridges resulting from the poor performance might disrupt the transportation system and directly result in tragedies of life and property loss. (Rantucci & Seismology, 1994; Sohn, 2003) Many bridges in Australia built in several decades ago are now subjected to heavier and faster moving loads than when it was designed and still suffering deterioration with age. Some bridges being refunctioned in a transportation manner might conduct load pattern changes, e.g. Victoria Bridge in Brisbane, Queensland has been redesigned as accommodation of bus lanes on one single side (Fig.1) (Todd & 1
PhD Candidate, School of Urban Development, Queensland University of Technology
2
Associate Professor, School of Urban Development, Queensland University of Technology
1
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Rodney, 2005) and from general knowledge this will lead to asymmetrical load to the bridge structure. Therefore, these factors will probably result in localised distress and, if not paid attention to, bridge failure with adverse consequences – disruption to normal life and costly renovation. The retrofit and reconstruction of bridges is always costly to the infrastructure owners. On the other hand, every civil structure, no matter what materials used, has limited life and cannot be well-determined under the uncurtain variation of environment or bridge functions. Health Monitoring is considerably a cost-efficient way to maintain the infrastructures and diagnose damage at early stage. Consequently it is imperative to guarantee safety and efficiency of bridge structures throughout their life spans or beyond by monitoring the health condition and accordingly adopt appropriate maintenance. To solve the problem, a comprehensive research program is currently underway to develop an innovative structural health monitoring (SHM) system to monitor the structural health of the bridge(s) and attend to any distress under all operating conditions. Besides, the new SHM technique employing innovative sensors, damage detection models, health status assessment and dynamic computer simulation techniques is based on the principle that the structural health of a bridge can be assessed through observing the changes of its vibration properties. This will be the most cost effective means of preventing bridge failure and ensuring their safe and efficient performance.
Figure 1 Asymmetrical Vehicle Loads on Victoria Bridge – Busway Accommodation (Courtesy of Suart Hill)
2
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Traditional Damage Detection The traditional damage detection strategies consist of visual inspection and localised non-destructive evaluation such as radio X-ray, radiographic, eddy current and ultrasonic techniques. All these methods, based on the foreseen damage locations and requiring that the vicinity of the damage is known a priori and the portion of the structure being inspected is readily accessible (S. W. Doebling et al., 1998), can hardly detect the in-depth damages or other inaccessible components. Besides, the previous approaches are only validated through a simple beam or plate type structures and rarely real structures which are more complicated. On the other hand, the damages are normally modelled by structural parameter reduction (like Young’s modulus) other than establishing the damage (crack) scheme. To extend SHM to larger and more complex structures such as truss-girder bridges, the global damage detection methods which examine changes in the global characteristics of structures are needed (S. W. Doebling et al., 1998). Structural Health Monitoring Techniques The basic thought as to what SHM should attempt to achieve goes with Balageas et al. (2006)’s definition: to give, at every moment during the life of the structure, a diagnosis of the ‘state’ of the constituent materials, of the different parts, and of the full assembly of these parts constituting the structure as a whole. The core of global damage detection methods assumes that the structural modal parameters are functions of the physical parameters, such as mass, stiffness and damping matrices, which means changes in these physical parameters such as global stiffness resulting from invisible cracks, consequently cause changes in modal parameters. Thus, tracing these variations is much helpful to inspect the occurrence, location and severity of structural damages. Although had a slow start, SHM of bridges has been under way for decades and the developing computation performance of computers has made the overall data collecting process quicker and easier (Olund & DeWolf, 2007). Worden et al., (2007) proposed that researchers have mutually agreed upon five general realisations through the years with respect to SHM and the process leading up to it: (1) Damage assessment needs to be made by comparison of two structural health states; (2) A trade-off exists between sensor precision (sensitivity) and its disturbance of a changing environment and general noise. (3) The size or amount of damage detected is inversely proportional to the frequency being measured. The other two require more explanation and can be found in the noted reference (Worden et al., 2007). According to the amount of information provided regarding the damage state, the damage identification can be classified into four levels (Rytter, 1993) to identify: Level 1 - Damage Existence; Level 2 – Level 1+Damage Location; Level 3 – Level 2+Damage Severity and Level 4 – Level 3+Structural Health Condition/Remaining 3
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Life Prediction. However, the existing investigation regarding the long-term SHM approaches are limited and most falling into the level 1 to level 3 (Park et al., 1997; N. Stubbs et al., 2000; Xia, Brownjohn, 2003; Yao & Natke, 1994). Some of them, Maeck, De Roeck, (1999) applied a direct stiffness approach to damage detection, localization, and quantification for the Z24 prestressed concrete bridge in Switzerland and noted that curvatures were rather small for the side spans and cause numerical difficulties in calculating bending stiffness in these spans. Peeters & De Roeck, (2000) compared a classical sensitivity-based updating technique with a direct stiffness calculation approach using data from a prestressed, 60m long concrete bridge with three-span. Macdonald & Daniell, (2005) monitored a cable-stayed bridge during its construction period and the modal parameters of many modes were identified from ambient vibration measurements. VanZwol et al., (2008) investigated the long-term structural health monitoring of steel-free deck bridges and made the recommendations on bridge health maintenance. Yang et al., (2008) investigated an existing PC box girder bridge with distributed HCFRP sensors in a destructive test. A. E. Aktan et al., (1996); A. E. Aktan et al., (2001) had mentioned monitoring and managing the health of infrastructure systems which could be regarded as part of level 4 in a, however, management or integrity manner other than information collected through level 1 to level 3. Procedure for damage detection and condition assessment A typical procedure for damage detection of civil structures based on vibration measurement is depicted as below:
Figure 2 Typical Procedures for Damage Detection
First, a structure model is set up in the FE environment according to the as-built drawings. Then, the baseline FE model can be established where we can get the required dynamic parameters at the same time. Then, required modal parameters can be calculated from the analytical model. The experimental detection will be followed after this step to update the numeric model by comparing the extracted modal properties with experimental results. However, the accomplishment of this step needs the integration of experimental technologies and analytical arts which has been 4
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
widely addressed in literature (A. E. Aktan et al., 1997). Next, from the updated FE model the damage detection results and the safe condition of the structure will be summarised. Finally, the reliability or the remaining load-carrying capacity will be analysed to help authorities to choose maintenance strategies. However, for some structures which might hardly support the baseline data, we should specifically consider the damage detection methods that don’t require a comparison of the intact and damaged state of the bridge (Cruz & Salgado, 2008). For instance, try the methods using curvature, wavelet techniques. Damage Detection Methods •
Methods using Natural Frequencies
There are two types of frequency analysis, the forward identification and the inverse identification that can be used for damage identification (Arvid, 2004). Both methods use the hypothesis that natural frequencies of a structure shifts when the physical properties change. However, Doebling et al., (1996) pointed out that the frequency shifts had significant practical limits for civil structures due to its insensitivity to damage unless when there was a severe damage or accurate measurement applied. Mares et al., (1999) proposed a damage identification procedure based on rigid body constraints. They simulated a crack in a two-dimensional finite element model of a cantilevered beam and found that when the external load was applied at the damaged element, there was no difference in the frequency response functions (FRF) between the damaged and undamaged state of the cantilever beam, but if the load was applied to the other undamaged elements, there was a change in the FRF. Crema & Mastroddi, (1998) correlated measured FRF with mass, stiffness, and viscous and hysteretic damping matrices. They claimed that it was possible to identify structural properties directly from the measured FRF data and to detect possible local stiffness or mass variations. The structural parameters were estimated by constructing inverse problems at different frequencies and input-output pairs. The authors performed numerical simulations on truss and beam structures such as airplane wings to validate the approach. As always, they found that noise could corrupt the resolution. Agneni, (2000) used the measured FRF for model updating and damage detection. The mass and stiffness matrices were estimated from the FRF and they investigated the truncation effect on FRF when the time signal was truncated at the end of the time and transferred into the frequency domain via Fast Fourier Transform (FFT). They concluded that the truncation could cause a significant change in FRFs even when no damage there and concealed the effects caused by damage. Besides, the truncation effects depended not only on the time window size used in the FFT calculation but also the decaying rate of various modes. Therefore, consideration must be given to these facts when using FRFs in damage detection applications.
5
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring •
Methods using Mode Shapes
West, (1984) and Wolff & Richardson, (1989) suggested the use of the modal assurance criterion (MAC) to detect the existence and the location of structural faults. The MAC was defined as
MAC (i, j ) =
(
{φ A} {φ B } T i
j
)
2
({φ } {φ } ) ({φ } {φ } ) A
T i
A
B
j
T i
1
B
j
{ }
Where φ A is the modal vector of the intact structure’s ith mode whilst {φ B } is the jth i
j
modal vector of the structure after damage. MAC is a scale quantity ranging from 0 to 1.0, representing that the degree of correlation between two sets of modal vectors is uncorrelated at all or perfectly correlated respectively. The method is based on the assumption that changes in modal vectors at the degrees of freedom near the damage are relatively larger than others located far from the damage. Fox & C. H. J., (1992) showed that MAC was insensitive to mode shape changes due to damage, however a MAC based on measurement point close to node points for a particular mode was sensitive to mode shape changes due to damage. As the MAC only uses one pair of modes for damage localisation, the problem of how to choose appropriate modes for MAC calculation induces the similar COMAC methods for damage localisation, which stands for Coordinate Modal Assurance Criterion (Lieven & Ewins, 1988) and is defined as: 2
⎛ m i i ⎞ ⎜ ∑ φ ArφBr ⎟ ⎠ 2 COMAC (i ) = m ⎝ r =1 m ⎛ i i ⎞⎛ i i ⎞ ⎜ ∑ φ Arφ Ar ⎟ ⎜ ∑ φBrφBr ⎟ ⎝ r =1 ⎠ ⎝ r =1 ⎠ i i Where φ Ar and φ Br are modal component of the rth mode shape at location i for the two paired mode shapes respectively and m is the number of the measured modes. The location where a COMAC value is close to zero is the possible damage location. Besides, Ko et al., (1994) presented a method using a combination of MAC, COMAC and sensitivity to detect damage in steel-framed structures. Salawu & Williams, (1995) conducted modal tests of a full-scale bridge before and after rehabilitation and concluded that the natural frequencies of the bridge did not change much as a result of structural repairs whilst both MAC and COMAC performed good to indicate the location of the repairs.
•
Methods using Curvature/Strain Modes
An alternative to using mode shapes to obtain the spatial location of damage is to utilize the mode shape derivatives, such as curvatures. There is a direct relationship between curvature and bending strain. Pandey et al., (1991) demonstrated that mode shape curvature could be a useful indicator to damage detection of beam structures. 6
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Chance et al., (1994) investigated the measured strain mode shape and found it was much feasible for damage localisation. Wang et al., (2000) presented a numerical study of damage detection of Tsing Ma Bridge in Hong Kong SAR by utilising the changes of the mode shape curvatures. •
Methods using Modal Strain Energy
N. Stubbs et al., (1992) presented the pioneer work on using Modal Strain Energy (MSE) for damage localisation. The general definition of MSE of a structure on the rth mode could be demonstrated as
1 MSEr = ΦTr K Φ r 2
3
where K is the stiffness matrix of a structure. N Stubbs & Kim, (1996) and Zhang et al., (1998) improved the method by using modal strain energy to localise the damage and estimated the damage size without baseline modal properties. They defined the contribution of element j to the rth MSE as Crj =
Tr k Φ r Φ j Φ Tr K Φ r
=
Tr k Φ r Φ j
ωr2
4
where Φ r is the rth mass normalized mode shape. Carrasco et al., (1997) discussed using changes in modal strain energy to locate and quantify damage within a space truss model and claimed that the magnitude of the changes could be used as an indicator of the overall magnitude of the damage. Results of the test showed that this method did very well at localising damaged elements within the truss structure. •
Methods using Dynamic Flexibility
Based on that higher modes contribute more to the system stiffness matrix than lower modes (Berman & Flannelly, 1971), to obtain good stiffness matrix estimation or its changes need a large number of dynamic modes, especially the higher modes to be measured. However, measuring the higher frequency response is very difficult due to practical limitations. To avoid the problem, a method using dynamically measured flexibility matrix is proposed to estimate the changes in structural stiffness. By extracting the first m mode shapes and modal frequencies, they proposed the index written as m
F = ΦΛ −1ΦT = ∑ i =1
1
ω
2 i
Φ i ΦTi
5
where ωi is the ith frequency, Φ i the ith mode shape. Reich & Park, (2000) focused on the use of localized flexibility properties for structural damage detection. The strain-based substructural flexibility matrices measured from before/after a damage 7
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
event were compared to identify the location and relative degree of damage. Bernal, (2000) set up a numerical example of a 39-DOF truss and obtained the accurate results of identifying the modes. He concluded that changes in the flexibility matrix were desirable to monitor than the changes in stiffness matrix. By tracing the changes in modal flexibility matrices before/after the structural damage occurred, Pandey & Biswas, (1994, 1995) presented a damage localisation algorithm and showed that only first several modes correlated the damage localisation. Wang et al., (2000) defined the normalised changes in modal flexibility for achieving more reliable damage detection. Gao & Spencer, (2006) discussed the issues relating to the synthesis of modal flexibility matrix from ambient and forced vibration data and implemented damage locating vector (DLV) method for online damage localisation. The experimental validation of various damage index methods had also been conducted by researchers through field testing. Farrar & Jauregui, (1998) conducted experimental and numerical comparison of five index methods using the simulated and experimental data from I-40 Bridge. Park et al., (2001) compared the result of damage index methods with visual inspection results by using the periodical measurement data from a concrete box-girder bridge and found the environmental factors might affect the accuracy of damage index methods. Huth et al., (2005) mentioned that it was difficult to localise the damage at the earlier stage even using the modal flexibility matrix. •
Artificial Neural Network (ANN)
Many damage detection schemes utilize neural networks to detect, localise, and quantify damage in structures. Ritter A & Kirkegaard, (1997) evaluated two neural networks for damage assessment, namely the multilayer perceptron (MLP) network with back propagation and the radial basis function (RBF) network and concluded that the MLP network demonstrates might be used in connection with vibration-based inspection whereas the RBF network completely failed. The authors also cautioned that the performance of the RBF network was highly dependent on an appropriate selection of damage cases used in the training. Huang & Loh, (2001) observed that a nonlinear neural network could be regarded as a general type of nonlinear auto-regressive moving-average (NARMA) model and discovered that the nonlinear model was able to predict the structure’s behaviour during earthquakes if the network was trained using past earthquakes with equivalent peak amplitudes. Choi & Kwon, (2000) developed a damage detection method for a steel-truss bridge based on neural network analysis. Chang et al., (2000) used an iterative neural network for structural health monitoring. The network was first updated using initial training data sets consisting of assumed structural parameters as target outputs and their corresponding dynamic characteristics as inputs.
8
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Feature Extraction Feature extraction is the process of identifying damage-sensitive properties, derived from the measured dynamic response, which allows one to distinguish between the undamaged and damaged structures and normally it performs some form of data compression and data fusion (Sohn, 2003). Table 1 Summary of Main Damage Detection Methods Approaches
Limitations
Virtues
The forward methods are mature
Only sensitive in the case that the measurement point is close to node points for a particular mode
The combination of MAC and COMAC work sensitively in damage detection
New sensor technology must be applied to register the strain field over the whole structure due to the strain is a local entity
Feasible for damage localisation
Higher derivatives of mode shapes are more sensitive to damage
Cannot detect damage in a structure when the damage is located in an element not sensitive to modal parameter changes
Feasible by localising damaged elements within the truss structure
More precise than the flexibility method
Require a sufficient number well distributed sensors
Only first few modes are needed
If damages are too small, it could masked by numerical errors
Better to detect damage severity
Large training sample is needed for accurate detection
Do not need knowing the physical relationships between the structural properties and damage occurrence
Limited physical meanings
Fast computation algorithms are needed
Can be applied to nonlinear and non-stationary time series
Insensitive unless severe damage happened or accurate measurement applied.
the inverse method is still only being investigated theoretically
Mode Shapes
Curvature/Strain
Frequencies Response Function
Modes
Modal
Strain Energy
Dynamic Flexibility
Artificial Neural Network
of be
(ANN) EMD & Hilbert spectrum
In cases of either abrupt or accumulative damages, occurrence of damage and the moment when the damage occurs can be clearly determined by data decomposition. Due to the length of the paper, we hereby just introduce the feature extraction 9
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
methods based on Wavelet technique and empirical mode decomposition (EMD) incorporated Hilbert spectrum method •
Wavelet technique
Wavelets can decompose any signal and some when transferred by Wavelets method can be shown to be more sensitive to local changes in structural properties. It also can be viewed as an extension of the traditional Fourier transform with adjustable window location and size. Over the past 10 years, wavelet theory has become one of the emerging and fast-evolving mathematical and signal processing tools (Wong & Chen, 2001) because of its many distinct merits. Using a selected analysing or mother wavelet function Ψ (t ) , the continuous wavelet transform of a signal f (t ) is defined as (Chui & Jian Zhong, 1992; Daubechies, 1992).
(Wf )(a, b) =
1 a
∫
+∞
−∞
⎛ t −b ⎞ f (t )Ψ ⎜ ⎟ dt ⎝ a ⎠
6
where a and b = dilation and translation parameters, respectively. Both are real numbers and a must be positive. The bar over Ψ (t ) indicates its complex conjugate. The mother wavelet should satisfy an admissibility condition to ensure existence of the inverse wavelet transform such as +∞
FΨ (ω )
−∞
ω
CΨ = ∫
2
dω < ω
7
where FΨ (ω ) denotes the Fourier transform of Ψ (t ) . The signal f (t ) may be recovered or reconstructed by an inverse wavelet transform of (Wf )(a, b) as defined by
1 f (t ) = CΨ
+∞ +∞
⎛ t −b ⎞ 1 ⎟ da ⋅ db a ⎠ a2
∫ ∫ (Wf )(a, b)Ψ ⎜⎝
−∞ −∞
8
A wavelet family associated with the mother wavelet Ψ (t ) is generated by two operations: dilation and translation. The translation parameter, b, indicates the location of the moving wavelet window in the wavelet transform. Shifting the wavelet window along the time axis implies examining the signal in the neighbourhood of the current window location. Therefore, information in the time domain will still remain, in contrast to the Fourier transform, where the time domain information becomes almost invisible after the integration over the entire time domain. The dilation parameter, a, indicates the width of the wavelet window. A smaller value of a implies a higher-resolution filter, e.g., the signal is examined through a narrower wavelet window in a smaller scale. By selecting different dyadic scales, a signal can be broken down into many low-resolution components, referred to as the wavelet decomposition tree. The wavelet tree structure with details and approximations at various levels may reveal 10
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
valuable information of the signal characteristics that may not be clearly seen in the original data or the results from other approaches. The analysis may be modified to the wavelet packet analysis for the purpose of multi-resolution analysis, which provides an alternative tree structure for the original data. Higher dimensional versions can also be easily extended (Chui & Jian Zhong, 1992). Different from the Gabor transform, the wavelet transform can be used for multi-scale analysis of a signal through translation, such that the time-frequency features of a signal can be extracted effectively. It is then used in a sensitivity-based inverse problem for structural damage detection with sinusoidal or impulsive excitation and acceleration and strain measurements. The wavelet coefficient is shown more sensitive than the other responses and however it is further found that not sensitive to different errors in the initial modelling including the support stiffness, mass density, flexural rigidity, damping ratio and excitation force. •
EMD and Hilbert spectrum method
Recently, the empirical mode decomposition (EMD) incorporated the Hilbert spectrum method has been proposed to identify the dynamic characteristics of linear structures. EMD is a time series analysis method that extracts a custom set of basis sets to describe the vibratory response of a system. In conjunction with the Hilbert Transform, EMD method provides some unique information about the nature of the vibratory response. Some simulation results also indicated that the Hilbert phase is proportional to mass and stiffness properties between two successive degrees of freedom (Salvino et al., 2003). Hence, the Hilbert phase can be used to infer, locate and possibly quantify the amount of damage in a structure. Model Updating Methods With the use of system identification concepts, the measured modal properties can be used to set up or modify the analytical structural model as precisely as possible to diagnose damages. There are two categories of structural system identification, one is to establish an analytical model, so-called direct system identification; The other one is to modify an existing analytical model or so-called indirect system identification (Natke, 1988a). The purpose of model improvement is to achieve an analytical model which is dynamically equivalent to the tested structure and damage detection aims to identify the changes in structural physical parameters due to damage rather than modelling errors. Research falls on model updating has been carrying on for 30 years. Among them, (Gorl & Link, 2001; Hart & Yao, 1977; Liu & Yao, 1978; Natke, 1988b; Zimmerman et al., 1992) are worthwhile of attention. The model updating methods for damage detection shall be classified into four categories: 1) Optimal Matrix Updating Methods; 2) Eigenstructure Assignment Methods; 3) Sensitivity-Based Updating Methods; 4) Stochastic Model Updating Methods. Condition Assessment of Bridge Structures Condition assessment is defined as measuring and evaluating the state properties of a constructed facility and relating these to the performance parameters (A. E. Aktan et al., 1996). It includes identification of any defects, deterioration and damage. 11
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Condition assessment should first focus on the evaluation of the global state of health, and then on regional or local damage. Some of the critical steps in condition assessment related to damage diagnostics may be summarized as (1) setting up the parameters limits which categorises the condition status; (2) diagnosing the damage existence; (3) localising the damage; (4) defining the damage severity; (5) assessing the effects of damage on structural reliability. These condition assessment steps are, in other words, exactly consistent with structural health monitoring Levels mentioned previously. Therefore SHM should be considered within the realm and the most important portions of condition assessment. Conclusion This paper briefly discusses the vibration-based damage detection methodologies used in structural health monitoring of bridge structures. For the bridge structures in the real-world, no single method or technique can be utilised for either global or localised health monitoring due to their different limitations. Therefore, the investigation of comprehensive methods is needed in future to keep SHM system feasible under more damage patterns and to be more practical. Besides, new signal processing methods proposed recently allow the VBDD methods distinguish the damage localisation/ patterns in a clear manner and more efficiently. From the review, it can be concluded that although damage detection, from Level 1 to Level 3, is a vital part of the condition assessment of bridge structures and has been investigated by a number of peers, few papers mention the SHM systems that have fallen into the structural condition assessment (Level 4) based on the interface supported by Level 1 to Level 3. Therefore, these all prompt an urgent need to build comprehensive damage detection approaches and a technical condition assessment system for a practical and efficient SHM system. References Agneni, A. (2000). On the use of the Gauss filter in modal parameter estimation. Mechanical Systems and Signal Processing, 14(2), 193-204. Aktan, A. E., Daniel, N. F., Arthur, J. H., David, L. B., Victor, J. H., Kuo-Liang, L., et al. (1997). Structural Identification for Condition Assessment: Experimental Arts. Journal of Structural Engineering, 123(12), 1674-1684. Aktan, A. E., Farhey, D. N., Brown, D. L., Dalal, V., Helmicki, A. J., Hunt, V. J., et al. (1996). Condition assessment for bridge management. Journal of Infrastructure Systems, 2(3), 108-117. Aktan, A. E., Chase, S., Inman, D., & Pines, D. (2001). Monitoring and managing the health of infrastructure systems. Arvid, H. (2004). Structural Health of Bridges, Monitor, Assess and Retrofit. Unpublished Licentiate Thesis, Department of Civil and Environmental Engineering, Division of Structural Engineering. Balageas, D. L., Fritzen, C.-P., & Güemes, A. (2006). Structural health monitoring. London ; Newport Beach, CA: ISTE.
12
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Berman, A., & Flannelly, W. G. (1971). Theory of incomplete models of dynamic structures. AIAA Journal, 9(8), 1481-1487. Bernal, D. (2000, 2000). Damage localization using load vectors. Paper presented at the COST F3 Conference, Madrid. Carrasco, C. J., Osegueda, R. A., Ferregut, C. M., & Grygier, M. (1997). Damage localization in a space truss model using modal strain energy, Orlando, FL, USA. Chance, J. E., Worden, K., & Tomlinson, G. R. (1994). Processing signals for damage detection in structures using neural networks, Orlando, FL, USA. Chang, C. C., Chang, T. Y. P., Xu, Y. G., & Wang, M. L. (2000). Structural Damage Detection Using an Iterative Neural Network. Journal of Intelligent Material Systems and Structures, 11(1), 32-42. Choi, M. Y., & Kwon, I. B. (2000). Damage detection system of a real steel truss bridge by neural networks. Proceedings of SPIE - The International Society for Optical Engineering, 3988, 295-306. Chui, C. K., & Jian-Zhong, W. (1992). A general framework of compactly supported splines and wavelets. Journal of Approximation Theory, 71(3), 263-304. Crema, L. B., & Mastroddi, F. (1998). A direct approach for updating and damage detection by using FRF data, Leuven, Belgium. Cruz, P. J. S., & Salgado, R. (2008). Performance of vibration-based damage detection methods in bridges. Computer-Aided Civil and Infrastructure Engineering, 24(1), 62-79. Daubechies, I. (1992). Wavelets in signal analysis, Brussels, Belgium. Doebling, Scott W. Farrar, & Charles R. (1996). Computation of structural flexibility for bridge health monitoring using ambient modal data. Doebling, S. W., Farrar, C. R., & Prime, M. B. (1998). Summary review of vibration-based damage identification methods. Shock and Vibration Digest, 30(2), 91-105. Farrar, C. R., & Jauregui, D. A. (1998). Comparative study of damage identification algorithms applied to a bridge: I. Experiment. Smart Materials and Structures, 7(5), 704-719. Fox, & C. H. J. (1992). The location of defects in structures: a comparison of the use of natural frequency and mode shape data. Paper presented at the Proceedings of the 10th International Modal Analysis Conference. Gao, Y., & Spencer, B. F., Jr. (2006). Online damage diagnosis for civil infrastructure employing a flexibility-based approach. Smart Materials and Structures, 15(1), 9-19. Gorl, E., & Link, M. (2001). Identification of damage parameters of a full-scale steel structure damaged by seismic loading, Madrid, Spain. Hart, G. C., & Yao, J. T. P. (1977). SYSTEM IDENTIFICATION IN STRUCTURAL DYNAMICS. 103(6), 1089-1104. Huang, C.-C., & Loh, C.-H. (2001). Nonlinear Identification of Dynamic Systems Using Neural Networks. Computer-Aided Civil and Infrastructure Engineering, 16(1), 28-41. Huth, O., Feltrin, G., Maeck, J., Kilic, N., & Motavalli, M. (2005). Damage identification using modal data: Experiences on a prestressed concrete bridge. Journal of Structural Engineering, 131(12), 1898-1910. Ko, J. M., Wong, C. W., & Lam, H. F. (1994). Damage detection in steel framed structures by vibration measurement approach. Paper presented at the Proceedings of the 12th International Modal Analysis Conference, Bethel.
13
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Lieven, N. A. J., & Ewins, D. J. (1988). Spatial correlation of mode shapes: the coordinate modal assurance criterion (COMAC). Paper presented at the Proceedings of the 6th International Modal Analysis Conference, Bethel. Liu, S. C., & Yao, J. T. P. (1978). STRUCTURAL IDENTIFICATION CONCEPT. 104(12), 1845-1858. Macdonald, J. H. G., & Daniell, W. E. (2005). Variation of modal parameters of a cable-stayed bridge identified from ambient vibration measurements and FE modelling. Engineering Structures, 27(13), 1916-1930. Maeck, J., & De Roeck, G. (1999). Dynamic bending and torsion stiffness derivation from modal curvatures and torsion rates. Journal of Sound and Vibration, 225(1), 153-170. Mares, C., Mottershead, J. E., & Friswell, M. I. (1999). Damage location in beams by using rigid-body constraints. Key Engineering Materials, 167-168, 381-390. Natke, H. G. (1988a). Application of System Identification in Engineering. Vienna: Springer. Natke, H. G. (1988b). Updating computational models in frequency domain based on measured data: a survey. Probabilistic Engineering Mechanics, 3, pp. 28-35. Olund, J., & DeWolf, P. E. (2007). Passive structural health monitoring of Connecticut's bridge infrastructure. Journal of Infrastructure Systems, 13(4), 330-339. Pandey, A. K., & Biswas, M. (1994). Damage detection in structures using changes in flexibility. Journal of Sound and Vibration, 169(1), 3-17. Pandey, A. K., & Biswas, M. (1995). Experimental verification of flexibility difference method for locating damage in structures. Journal of Sound and Vibration, 184(2), 311-328. Pandey, A. K., Biswas, M., & Samman, M. M. (1991). Damage detection from changes in curvature mode shapes. Journal of Sound and Vibration, 145(2), 321-332. Park, S., Stubbs, N., Bolton, R., Choi, S., & Sikorsky, C. (2001). Field verification of the damage index method in a concrete box-girder bridge via visual inspection. Computer-Aided Civil and Infrastructure Engineering, 16(1), 58-70. Park, S., Stubbs, N., & Sikorsky, C. (1997). Linkage of nondestructive damage evaluation to structural system reliabilitySmart Systems for Bridges, Structures, and Highways. Paper presented at the The International Society for Optical Engineering, Bellingham. Peeters, B., & De Roeck, G. (2000). One year monitoring of the Z24-bridge: Environmental influences versus damage events, San Antonio, TX, USA. Rantucci, G., & Seismology. (1994). Geological disasters in the Philippines : the July 1990 earthquake and the June 1991 eruption of Mount Pinatubo : description, effects and lessons learned. Roma: Presidenza del Consiglio dei Ministri, Dipartimento per l'informazione e l'editoria. Reich, G. W., & Park, K. C. (2000). Experimental application of a structural health monitoring methodology. Proceedings of SPIE - The International Society for Optical Engineering, 3988, 143-153. Ritter A, & Kirkegaard, P. (1997). Vibration based inspection using neural networks. Paper presented at the Proceeding of 2nd international conference on structural damage assessment using advanced signal processing procedures. Rytter. (1993). Vibration based inspection of civil engineering structures. University of Aalborg, Denmark. Salawu, O. S., & Williams, C. (1995). Bridge assessment using forced-vibration testing. Journal of Structural Engineering, 121(2), 161-173. Salvino, L. W., Pines, D. J., Todd, M., & Nichols, J. (2003). Signal Processing and Damage Detection in a Frame Structure Excited by Chaotic Input Force, San Diego, CA, United States.
14
Review of Vibration-Based Damage Detection and Condition Assessment of Bridge Structures using Structural Health Monitoring
Sohn, H. (2003). A Review of Structural Health Monitoring Literature: 1996-2001. Stubbs, N., & Kim, J. T. (1996). Damage localization in structures without baseline modal parameters. AIAA Journal, 34, pp. 1644-1649. Stubbs, N., Kim JT, & K., T. (1992). An efficient and robust algorithm for damage location in offshore platforms. Paper presented at the Proceedings of the ASCE 10th Structures Congress. Stubbs, N., Park, S., Sikorsky, C., & Choi, S. (2000). A global non-destructive damage assessment methodology for civil engineering structures. International Journal of Systems Science, 31(11), 1361-1373. Todd, B., & Rodney, S. (2005). Words to Walk by: Exploring Literary Brisbane: Univ. of Queensland Press. VanZwol, T. R., Cheng, J. J. R., & Tadros, G. (2008). Long-term structural health monitoring of the Crowchild Trail Bridge. Canadian Journal of Civil Engineering, 35(2), 179-189. Wang, J. Y., Ko, J. M., & Ni, Y. Q. (2000). Modal sensitivity analysis of Tsing Ma Bridge for structural damage detection, Newport Beach, CA, USA. West, W. M. (1984). Illustration of the use of modal assurance criterion to detect structural changes in an orbiter test specimen. Paper presented at the Proceedings of the Air Force Conference on Aircraft Structural Integrity, Los Angles. Wolff, T., & Richardson, M. (1989). Fault Detection in Structures from Changes in their Modal Parameters. Paper presented at the Proceedings of the 7th International Modal Analysis Conference. Wong, L. A., & Chen, J. C. (2001). Nonlinear and chaotic behavior of structural system investigated by wavelet transform techniques. International Journal of Non-Linear Mechanics, 36(2), 221-235. Worden, K., Farrar, C. R., Manson, G., & Gyuhae, P. (2007). The fundamental axioms of structural health monitoring. Paper presented at the Proceedings of the Royal Society of London, Series A (Mathematical, Physical and Engineering Sciences). Xia, P.-Q., & Brownjohn, J. M. W. (2003). Residual stiffness assessment of structurally failed reinforced concrete structure by dynamic testing and finite element model updating. Experimental Mechanics, 43(4), 372-378. Yang, C., Wu, Z., & Zhang, Y. (2008). Structural health monitoring of an existing PC box girder bridge with distributed HCFRP sensors in a destructive test. Smart Materials and Structures, 17(3), 035032. Yao, J. T. P., & Natke, H. G. (1994). Damage detection and reliability evaluation of existing structures. Structural Safety, 15(1-2), 3-16. Zhang, L.-M., Wu, Q., & Link, M. (1998). A structural damage identification approach based on element modal stain energy, Leuven, Belgium. Zimmerman, T. J. E., Colquhoun, I. R., Price, P. S. J., & Smith, R. J. (1992). Development of a limit states guideline for the pipeline industry, Calgary, Alberta, Can.
15
