MS17.12 - Structural Health Monitoring

The Bundesanstalt für Materialforschung und -prüfung (BAM) collaborates with TNO to develop a software framework for automated calibration of structural models based on monitoring data. The ultimate goal is to include such models in the asset management process of engineering structures. As a basis for developing the framework, a multi-span road bridge consisting of ten simply supported steel box girders was selected as a test case. Our group measured output-only vibration data from one box girder under ambient conditions. From the data, we determined eigenfrequencies and mode shapes. In parallel, we developed a preliminary structural model of the box girder for the purpose of predicting its modal properties. In this contribution, we provide an overview of the measurement campaign, the operational modal analysis, the structural modeling and qualitatively compare the identified with the predicted modes. As an outlook, we discuss the further steps in the calibration process and future applications of the calibrated model.

Over the past fifty years, external post tensioning tendons have been extensively used for the design and retrofitting of concrete bridges. Despite its multiple advantages (such as economical construction, ease of installation, re stressing and substitution in case of failure), the behavior of these post tensioning systems is sensitive to both the deterioration and damage caused by three main factors: (i) the increasing traffic loads; (ii) the exposure to adverse environmental conditions (the tendon corrosion is a current significant problem); and (iii) the structural aging. In order to guarantee the adequate performance of these post tensioning systems during the overall life cycle of the bridge, their fatigue assessment must be checked. One key aspect to be considered for an accurate prediction of these external post tensioning systems' fatigue behaviour is quantifying the uncertainties associated with the variation of their modal properties and the increasing traffic loads. In order to compute numerically these uncertainties, information extracted from structural performance monitoring (SPM) can be used. In particular, modal data can be obtained from continuous long term monitoring. Thus, these data (time series) can be considered to perform a Bayesian finite element model (FEM) updating of the structure. As a result of this updating process, the main physical parameters of the post tensioning system can be estimated as random variables. Thus, data obtained from SPM allow for predicting the structural capacity of the tendons accurately. Additionally, information extracted via SPM can be used to analyze the uncertainty associated with the increasing traffic load (the demand). Finally, a reliability fatigue analysis, based on the experimentally estimated capacity and demand, can be employed to compute the expected life cycle of the bridge. In this study, a fatigue reliability analysis of the post tension system of a high speed train bridge has been performed in detail. For this purpose, long term monitoring of the system has been implemented. As a result of this dynamic monitoring, both modal data and information about the real traffic have been extracted. On the one hand, these modal data have been used to perform a Bayesian FEM updating. A Markov Chain Monte Carlo method has been considered to sample the probabilistic functions (normal distribution) which characterize the random behavior of the main physical parameters of the post tensioning system. On the other hand, the effect of traffic loads has been modelled as a random process. Data extracted via SPM has been considered to characterize this process. Additionally, the S N curves (assuming a Weibull distribution to describe the uncertainty of the fatigue strength) and the Palmgren Miner hypothesis were chosen for the definition of the fatigue strength of the considered post tensioning system. Thus, a limit state function for the fatigue assessment can be defined. Finally, the variability of the random variable damage can be computed from this limit state. As a conclusion, this manuscript illustrates how the information extracted via SPM allows quantifying the uncertainty associated with the post tensioning system, and consequently how these data allow computing more accurately

In the field of civil infrastructures, Structural Health Monitoring generally suffers from a scarcity of labelled data describing the damage classes and rare environmental conditions. To solve this issue, the proposed work adopts a Transfer Learning approach for leveraging information from a source structure, characterised by a rich class of damage labels, to im-prove inferences on a target structure with limited knowledge and sparse datasets. The goal is to train a machine learning algorithm on a bridge undergoing damage and to afterwards transfer these labelled damage-state data across the members of the investigated population. Given possible differences in the underlying distributions of each structure, Domain Adapta-tion techniques are applied to match the domains in a shared feature space. The methodolo-gy is validated on a heterogeneous population composed of two numerical bridges of differ-ent geometry and materials. Finite Element Models are built to simulate artificial data stemming from operational conditions and multiple damage scenarios. Results prove the ef-fectiveness of the presented approach to successfully exchange health-state information within a population of bridges, thus reducing the computational burden and enabling popu-lation-based SHM.

Operational Modal Analysis for identifying dynamic parameters, together with Finite Element Model Updating algorithms, is a promising and powerful tool for detailed analysis of complex civil engineering structures. It is also an integral part of vibration-based methods in Structural Health Monitoring. In this work, the identification of the dynamic parameters was performed based on output-only vibration data recorded in a testing campaign on a 36m high steel structure used as an observation tower for tourists in Latvia. This slender structure is constructed on a shallow foundation consisting of a concrete slab freely supported on the ground. Hence, another task of the work is that of investigating possible foundation rocking effects on the dynamic behavior of the tower. To interpret correctly the experimental results, a Finite Element Model of the tower has been developed within the Ansys environment. After a sensitivity analysis is carried out to understand which parameters mostly affect the dynamics of the tower (elastic modulus of materials, stiffness of restraints, masses, etc,), the Model Updating is performed by adopting an Artificial Intelligence algorithm called Particle Swarm Optimization. The calibration is performed with the aim of obtaining a numerical model that simulates the real dynamic behavior of the case study with high accuracy and can be used for the development and design of the Structural Health Monitoring system of the tower.