Claudia Zelaschi

Seismic risk assessment is more challenging for infrastructure systems than for single, non-interacting components, especially when network-level interaction is included in the analysis. Accounting for the interdependencies of the spatially distributed demands and of the capacity does significantly affect reliability and risk assessment. The spatial distribution of the ground motion intensity is one of the main sources of dependences among the components' response, and models for predicting the correlation of the most commonly used intensity measures (peak ground acceleration, spectral acceleration at the fundamental period of vibration, spectral acceleration at multiple periods, etc.) at different locations are available in literature. However, when these models are applied to system-level analysis, it is still to be determined if the capacity of different components (modelled by the fragility function) should be considered as dependent or independent. Dependence can derive from two sources: on one hand, epistemic uncertainty can be common across many components, on the other, the seismic intensity may be just an incomplete summary of the seismic event, not able to capture relevant features of the demand. Because of this, an assessment of the response that only accounts for the interdependence in the seismic intensity may underestimate the actual dependence. To investigate such a loss of information about the correlation between structural responses, we perform non-linear analyses of a RC bridge, using sets of ground motions recorded during the same event at two different locations, and compare actual and predicted interdependence. From the simulations' results, we investigate the accuracy of system-level analysis.

When carrying out seismic loss assessment of an infrastructure network within a certain region, the characterization of the fragility of bridges, which represent nodes of the network, is one of the most important aspects. The fragility assessment largely depends on the characterization of the seismic demand that such structures face and, for this reason, a proper intensity measure (IM) is necessary. Many different IMs have been proposed over the years and recent studies have proposed new approaches based on different parameters or on the coupling of previously tested IMs on an individual fashion (vector-based IMs). These studies have however been focused only on building structures whereas the available literature on analysis of IMs for bridges has considered a limited number of structural configurations. Indeed, both the seismic response of bridges and the corresponding fragility are strongly dictated by the structural configuration, material properties and seismic excitation intensity, which in turn will significantly influence the performance of the IM under analysis. In order to take into account such variability and to overcome the aforementioned limitations of the current state-of-the-art, the study presented herein considered optimized numerical simulation techniques, through the Latin Hypercube sampling method, for the generation of a population of reinforced concrete bridges. Such case study was used to assess the correlation between a considerable amount of traditional and innovative IMs and the nonlinear structural response of bridges, estimated through nonlinear dynamic analysis using a selected set of suitable ground motion records as input excitation. The capability of predicting the seismic response of the structure using distinct IMs has been extensively scrutinized and compared for the population of generated bridges, in terms of practicality, efficiency and proficiency. An optimal IM has been sought calculating the Pearson Moment Correlation Coefficient (PMCC). The comparative study outlined the Fajfar Index, peak ground velocity and root mean square velocity as the most promising intensity measures. Typically employed IMs (peak ground and spectral accelerations) performed reasonably well. On the other hand, using vector-based intensity measures did not significantly improve the results in terms of efficiency, practicality and proficiency.

The most recent developments in earthquake engineering design are based on the concept of prescribed performance levels, rather than the traditional prescriptive approaches. A relatively well-known software tool for conducting PBEE studies for bridge-ground systems (BridgePBEE), which is based on three-dimensional (3D) finite element model analyses carried out in OpenSees, considering a simplified soil-structure interaction macro element, has been recently applied to a limited number of individual bridge structures. This study aims to apply a Performance-Based Earthquake Engineering (PBEE) methodology to the seismic assessment of an existing road network, in which the nodes correspond to bridges, rather than to a single structural configuration, using BridgePBEE to study the response of different real bridges. Such versatile Finite-Element (FE) environment provided the possibility of processing detailed information on effective damage, safety and economic assessment. In addition, by definition of PBEE, informed decisions of the evolution of seismic losses in terms of cost and time repair quantities were derived at different Peak Ground Acceleration (PGA) levels. Nonlinear dynamic analyses were then run by using properly selected seismic records as input, by means of the conditional spectrum method applied to the seismic hazard outputs for the considered target site, using peak ground acceleration as reference intensity measure. The conclusions provide performance considerations of the investigated population of bridges in terms of repair costs and time, deriving relevant considerations for design procedures .

The nodes of a road transportation network of a certain region can represent a population of bridges. In view of a sound characterization of corresponding fragility or vulnerability functions, the knowledge of real data, such as geometrical and material properties available from a national bridge inventory database, is a crucial element. In fact, the outcome of damage seismic assessment can be strongly affected by the variability of such information. When only limited information for each node is known and the relevant structural behaviour needs to be assessed for e.g. loss estimation studies at macro-scale level, the use of statistical tools is of extreme importance. Accordingly, the present paper furnishes a double-fold contribution: (1) provide the systematic statistical characterization of geometrical and material properties of typical Italian reinforced concrete (RC) bridges and (2) simplified formulae for the quick evaluation of the fundamental elastic period of vibration of RC bridges. This allows the development of typological fragility curves for RC bridge classes, which might be chosen according to their fundamental period of vibration. As such, the bridges can be identified with a specific bridge class, characterized by a representative fragility curve, used for expedite large-scale seismic assessment.

The seismic assessment of a road network depends largely on the characterization of the fragility of its bridge components. The accuracy of bridge seismic demand estimates and the use of proper intensity measures (IM) will significantly influence such task. The available literature has mainly focused on buildings or a limited number of bridge configurations and IMs, which may not be representative for bridge portfolio assessment studies. In this paper, the correlation quality between a larger pool of traditional and innovative IMs and the nonlinear dynamic response of typical Italian RC bridges is investigated to identify the best performing IMs.

When carrying out loss assessment of a road network within a certain region, the characterization of the vulnerability of the existing population of bridges is one of the most relevant aspects. To such extent, in the last decades, the engineering community has recognised fragility curves as one of the fundamental and most effective tools in seismic risk assessment, correlating the probability of exceeding of specific limit states, which can instead be correlated with damage and loss, for different levels of intensity measures (IMs). The available literature mainly provides such curves as function of peak ground acceleration (PGA) or spectral acceleration (Sa), often preferred for the availability of national hazard information in terms of such parameters. However, when dealing with a bridge population of reinforced concrete bridges, Fajfar Index, peak ground velocity and root mean square velocity have been recently identified as the most promising intensity measures, in terms of eff...

In macro-scale roadway networks, bridges are the main point-site components and tend to be among the most vulnerable elements in case an earthquake occurs. Given this weakness, the proper assessment of the seismic vulnerability is of high importance to obtain reliable estimates of losses in the immediate aftermath of seismic events or to manage emergency situations. In this sense, the use of real data (e.g. geometrical and material parameters drawn from national bridge inventory databases), accounting for its variability, is preferable and increases the accuracy of estimated fragility or vulnerability functions of the bridge portfolio of a certain region. However, complete information for each single bridge is typically not available. One way to address this limitation is to use statistical tools to handle large-scale loss estimation studies and to gauge the parametric characterization of relevant structural behaviour properties. In view of these considerations, the present paper addresses the first step towards the development of supporting tools for improved large-scale seismic assessment of bridges by characterizing the geometrical and material properties of typical Italian reinforced concrete (RC) bridges. With this characterization, we generate a large bridge stock for a specific bridge typology using the Latin Hypercube sampling scheme. Such bridge stock is then assessed with a view to establish simplified relationships between the fundamental period of vibration and the geometrical configuration of the bridge. The developed formulae, defined as a function of two geometrical variables, can be used to quickly estimate the period of vibration and the seismic demand (and fragility) of the structure in expedite large-scale seismic assessment studies.