Journal of Engineering Mechanics

The long-range monitoring of civil infrastructure systems monitored with dense sensor arrays that are capable of generating voluminous amounts of data from continuous online monitoring requires the implementation of a proper data processing and archiving scheme to maximize the benefits of structural health monitoring operations. This paper focuses on the areas of data management, data quality control, and feature extraction of meaningful parameters to describe the response of large-scale infrastructure systems to ambient excitation in the context of structural health monitoring (SHM). Recordings from the monitoring system installed on the Vincent Thomas Bridge (VTB) in San Pedro, California form the database of the proposed data-management and archiving methodology. The data processing methodology for the VTB is based on the calculation of the sensor array acceleration covariance matrices for every hour of available data and the subsequent orthogonal decomposition of the covariance matrices. The dominant proper orthogonal modes of the bridge are determined, and their statistical variations over an extended observation period covering several months of continuous data are quantified and analyzed. The empirical probability density functions for the mean daily bridge accelerations are computed and used to compare the statistical variations in different periods of operation of the bridge (working days, weekends, holidays). It is shown that the computed statistical distributions of the bridge response can provide a quantitative baseline through which to facilitate the early detection of any anomalies indicative of a possible structural deterioration resulting from fatigue (service loads) or extreme loading events, i.e., earthquakes, artificial hazards, or other natural hazards.

Verification of vibration-based damage detection through a full-scale actual structural testing is an important learning opportunity. From such a test, the evolution of dynamic characteristics can be observed, damage detection methods can be validated, and baseline criteria for typical structural damage can be formulated. This paper describes a case study on a full-scale destructive testing of an overpass reinforced concrete bridge. Damage is introduced by cutting one of the bridge piers at the footing level allowing vertical settlement. This type of damage is expected to simulate the condition in which a bridge suffers from nonuniform pier settlement or hidden damage inside piles of buried foundations. By applying time and frequency domain vibration analysis, as well as a system identification technique, changes in dynamic characteristics caused by the damage are evaluated. The results clearly indicate the changes in frequencies as an indicator of damage presence, while the change in mode shapes can be used to locate the damage. The paper also discusses the application of the damage detection method based on outlier analysis of the autospectra function using the bridge ambient acceleration responses. The results indicate that the presence of damage at an early stage can be detected by observing the outliers in multivariate data, and the detection accuracy improved when damage has significantly changed the dynamic characteristics of the structure.

Infrastructure is subject to continuous aging. This has given life-cycle management of the infrastructure an increasing role. Therefore, reliable inspection and monitoring tools are in increasing demand. The paper provides a short overview of the major activities in this area in Europe, followed by a presentation of current projects in the field of civil engineering in which topics related to combinations of nondestructive testing (NDT) methods and modular control and data acquisition approaches are discussed. In this paper, the two major construction materials in civil engineering, steel and concrete, are discussed. Regarding reinforced as well as prestressed concrete, a combination of various NDT methods is often required, which allow reliable results for material characterization, flaw detection, and the determination of component-specific geometry parameters. Therefore, a multisensor measurement approach with a high degree of automation is required. Two recently developed measurement devices (OSSCAR and BetoScan) will be described. Regarding steel constructions, an example from the field of pipeline inspection is shown where the combination of NDT with fracture mechanical assessment is directly possible. Both fields deliver the required interfaces in which the fracture risk is evaluated in probabilistic terms. Therefore, this procedure can be used as a tool for inspection-based lifetime conceptions. The paper concludes with an outlook on future trends for degradation monitoring.

Complex analysis and design of structures, especially landmark structures such as long-span bridges, have been conducted by many engineers and researchers. Currently, it is possible to collect more and precise monitoring data as well as to develop complex three-dimensional (3D) FEM models. These models, which can be calibrated using structural health-monitoring (SHM) data, can be used for the estimation of component and system reliability of bridges. However, the uncertainties related to the data, analysis, and nonstationary nature of the structural behavior need to be better incorporated by using a set of models that are continuously updated with monitoring data. This set of models constitutes a family as a result of the approach by which the models are obtained and the relationships among them. The objective of this paper is to explore the impact of uncertainty in predicting the system reliability obtained by a one-time, initially calibrated FEM model as well as by a family of FEM models continuously calibrated with monitoring data. To explore the uncertainty effects, a laboratory structure that has a combined system configuration with main and secondary elements is monitored. The monitoring data are employed for the FEM model calibration by using artificial neural networks (ANNs) to obtain parent (calibrated) FEM models from which a set of offspring FEM models is generated to incorporate the uncertainties. It is shown that the use of parent-offspring FEM models becomes important especially when critical parameters that have an impact on the model responses cannot be precisely defined. Finally, it is shown in a comparative fashion that the prediction of reliability using a family of FEM models and a single model can be quite different because the family of models provides a more realistic estimate of the structural responses and probability of failure.

This paper investigates the model order determination problem in the identification of dynamic characteristics of long-span bridges subjected to ambient excitation. Based on a stochastic state-space model framework, a new approach for state variable estimation is proposed, which is developed for the purpose of properly determining the order of a mathematical model of the structure under consideration. Comparing the newly developed approach with existing ones, their performances for system identification are evaluated with respect to their ability to highlight structural properties against noise ones in terms of the solution of a singular value problem, from a theoretical point of view and in applications to numerical and field measurements of a suspension bridge. From these applications, it is demonstrated that the newly developed approach is the most effective among the existing ones in discriminating structural modes, including weakly excited and closely spaced modes, from noise ones in terms of singular values, even when dealing with low signal-to-noise ratio signals and nonwhite wind excitation.

A damage detection approach is developed using nonlinear autoregressive with exogenous inputs (NARX) neural networks and a statistical inference technique. Within a large spatially extended dynamic system, an instrumented local substructure may be represented by a neural network, to predict the dynamic response of a given sensor from that of its neighbors. Without change in the system properties, the network prediction error will follow a stable statistical distribution. To infer damage, change in the prediction error variance as evaluated by the statistical inference standard $F$ test is utilized as a sensitive indicator. Validation of the described procedure is undertaken using two experimental data sets (from the Los Alamos National Laboratory in Los Alamos, NM). Reduced stiffness and nonlinear response of a mass-spring system is documented in the first set, while joint damage in a frame structure is explored in the second. Favorable results are obtained in both cases with linear/nonlinear and single/multidamage patterns. Overall, the proposed framework may be particularly efficient for large spatially extended sensor network situations, where local condition assessment may be conducted based on the response of a few neighboring sensors.

Structural identification (St-Id) of long-span bridges by ambient vibration testing provides a starting point for quantitatively characterizing the actual in-service mechanical characteristics and behaviors of these complex constructed systems. However, various uncertainties involved in the experimental and identification processes impact the reliability of St-Id, especially if vibration testing is the sole experiment. Such uncertainties represent perhaps the most fundamental barrier to more widespread applications of measurements in civil engineering practice. The goal of this paper is to leverage a vibration test of a long-span suspension bridge to illustrate a number of possible strategies for coping with the uncertainties confronted when identifying the dynamic characteristics of large-scale constructed systems. The design and implementation of a field test in the context of St-Id are first presented to illustrate how uncertainties can be mitigated by following a disciplined approach to designing the experiment. Next, data preprocessing strategies including data inspection, time window selection, band-pass filtering, averaging, and windowing are proposed to reduce data errors. Three separate postprocessing methods, including Peak Picking, PolyMAX, and Complex Mode Indicator Function, are executed independently to verify the reliability of the data processing results. Experimental results for both the towers and suspended spans are correlated with simulations from three-dimensional finite-element analysis of the long-span bridge for St-Id.

This paper addresses the issue of connectivity- and cost-based optimal scheduling for maintenance of bridges at the transportation network level. Previous studies in the same field have considered the connectivity just between two points or other network performance indicators, such as the total travel time. In this paper, the maximization of the total network connectivity is chosen as the objective of the optimization, together with the minimization of the total maintenance cost. From a computational point of view, several numerical tools are combined to achieve efficiency and applicability to real cases. Random field theory and numerical models for the time-dependent structural reliability are used to handle the uncertainties involved in the problem. Latin hypercube sampling is used to keep the computational effort feasible for practical applications. Genetic algorithms are used to solve the optimization problem. Numerical applications to bridge networks illustrate the characteristics of the procedure and its applicability to realistic scenarios.