برآورد کاهش چرخه عمر لرزه نگاری و تجزیه و تحلیل حساسیت جهانی بر اساس مدل سازی حرکت زمین تصادفی

The assessment of seismic losses for structural systems through adoption of stochastic ground motion models for characterization of the seismic hazard is the focus of this study. An assembly-based vulnerability methodology is adopted for earthquake loss estimation that uses the nonlinear time–history response of the structure under a given excitation to estimate damages in a detailed, component level. Description of the earthquake acceleration time–history through stochastic ground motion models is considered in this context. The parameters of these models are connected to the regional seismicity characteristics (such as moment magnitude and rupture distance) through predictive relationships. Description of the uncertainty for these characteristics and for the predictive relationships, by appropriate probability distributions, leads then to quantification of the life-cycle seismic losses by its expected value. Because of the complexity of the adopted models, estimation of this expected value through stochastic simulation is suggested and techniques for improvement of computational efficiency are discussed. An innovative global sensitivity analysis is also reviewed, based on advanced stochastic sampling concepts. This analysis aims to identify the importance of each of the uncertain parameters, within the seismic hazard description, towards the overall seismic risk (life-cycle cost). The benefits in terms of detailed, versatile description of seismic risk and the computational challenges of the overall simulation-based, probabilistic framework are extensively discussed. The methodology is illustrated through application to a four-storey moment-frame concrete building for estimation of life-cycle repair cost. Emphasis is placed on the results from the sensitivity analysis for investigating the impact on the estimated repair cost of the ground motion model characteristics and of the fragility features of the different assemblies.

Seismic life-cycle cost assessment requires proper integration of (i) loss-estimation methodologies for evaluating the structural performance using socioeconomic criteria, (ii) probabilistic approaches for treating the uncertainties related to the seismic hazard and to the structural behavior over the entire life-cycle of the building, as well as (iii) algorithms for efficient evaluation of the resultant multidimensional integrals ultimately quantifying seismic cost. The modeling of earthquake losses for a specific seismic event and the characterization of the earthquake hazard, describing the likelihood of occurrence of each event as well as the resultant seismic forces (in the specific format required for structural analysis, for example as ground motion time-history), constitute undoubtedly the most important components of this process. Earlier methodologies for seismic loss evaluation expressed these losses in terms of the global reliability of the structural system. Recent advances in performance-based engineering quantify more appropriately repair cost, casualties, and downtime in relation to the structural response, using fragility curves to develop such a relationship [1]. In this context, approaches have been proposed that approximately describe the nonlinear structural behavior by the static pushover response [2] and [3] and/or estimate earthquake losses in terms of global response characteristics [4]. Other researchers [5], [6] and [7] have developed analytical tools that evaluate seismic vulnerability on a detailed, component level (such as partitions, beams and columns), using the nonlinear time–history response of the structure under a given earthquake excitation to ultimately calculate seismic damages. This latter approach requires description of the entire ground motion time history for seismic events. The most popular methodology [8], [9], [10] and [11], to facilitate such a description relies on adoption of Intensity Measures (IMs) that represents the dominant features of the seismic excitation, and subsequent scaling/selection of ground motion records to different hazard levels (different IM values), as prescribed by a probabilistic seismic hazard analysis. These ground motions are taken to represent samples of possible future ground motions for each hazard level (a limited number of such levels is typically considered). To reduce computational burden a small number of ground motions is used (for each level) and some probability distributions is fitted over the samples of the structural response to obtain their statistical description for the specific hazard level [6]. An alternative approach, gaining increased interest within the structural engineering community [12], [13], [14] and [15], especially in light of recent concerns related to ground motion scaling [16], is to use stochastic ground motion models [17] and [18]. Such models modulate a high-dimensional stochastic sequence through functions that address spectral and temporal characteristics of the excitation, to ultimately provide samples of the earthquake acceleration time–history. The parameters of these functions, for example, duration of strong motion, can be related to earthquake (type of fault, moment magnitude and rupture distance) and site characteristics (shear wave velocity, local site conditions) by appropriate predictive relationships [12] and [19]. Description of the uncertainty for the earthquake characteristics (moment and rupture distance) and for these predictive relationships, through appropriate probability models, leads then to a complete and detailed probabilistic description of potential future ground-motion time–histories. The focus of these studies has been, though, primarily on development of stochastic ground motion models. Limited attention has been given to the impact of such a seismic hazard characterization within the context of performance-based engineering and life-cycle cost estimation. This paper discusses a simulation-based, comprehensive computational approach that aims to bridge this gap. It focuses on seismic loss estimation through adoption of stochastic ground motion models for the seismic hazard description and investigates the potential benefits in terms of detailed, versatile description of seismic risk as well as the challenges in terms of computational efficiency. A probabilistic framework for assessment of life-cycle repair cost is initially presented based on the concepts discussed above for loss estimation and probabilistic earthquake hazard description. In this context, life-cycle seismic cost is quantified by its expected value over the established probability models and stochastic simulation is suggested for its evaluation. Techniques for improvement of computational efficiency are examined and suggestions are provided for their implementation in practical applications. An innovative probabilistic global sensitivity analysis is also reviewed, based on advanced stochastic sampling concepts. This analysis, demonstrated for seismic risk applications first in [20] and recently extended in [21] to groups of parameters (not necessarily constrained to individual scalar parameters), aims to identify the importance of the various risk-factors (i.e., uncertain model parameters) towards the overall performance of the structural system. This analysis is implemented here in the context of seismic loss estimation, i.e., not constrained to simplified performance evaluation in terms of system reliability as in the aforementioned two studies.

A versatile, simulation-based, framework was discussed in this work for assessment of life-cycle repair cost for structural systems. The framework addresses the variability in future seismic events through stochastic ground motion models, formulated by modulating a high-dimensional white noise sequence through functions that address spectral and temporal characteristics of the excitation. The parameters of these models are connected to the regional seismicity characteristics through predictive relationships. Description of the uncertainty for the seismicity characteristics and for the predictive relationships, by appropriate probability distributions, leads then to a complete probabilistic description of the seismic hazard, expressed in terms of the acceleration time–history. An assembly-based vulnerability approach was then adopted to quantify earthquake losses based on the nonlinear time–history structural response. The life-cycle repair cost was quantified by its expected value over the space of the uncertain parameters for the structural and excitation models and estimation of this expected value through stochastic simulation was considered. For reduction of the computational burden importance sampling and high-performance computing concepts were used. An efficient probabilistic sensitivity analysis was also presented. This analysis aims to identify the importance of the various uncertain parameters (risk factors) within the seismic hazard description and is performed computationally efficiently (without requiring additional structural simulations) through advanced stochastic simulation concepts. The framework was illustrated through application to a four-storey moment-frame concrete building. Results were presented with respect to the total cost as well as with respect to the cost of different damageable assemblies. Repairing partitions and repainting damaged surfaces were shown to have the biggest contributions to the overall cost with repairs to structural systems also having a significant one. The moment magnitude was shown to be the most influential parameter for the repair cost, especially for the structural components, with the rupture distance also having a significant impact. The risk factors related to the predictive relationships of the ground motion model had negligible importance apart from the uncertainty for the higher frequency of the source spectrum and, for acceleration sensitive components, the uncertainty in the duration of the strong ground motion. This is an important result, demonstrating that the uncertainty in the rest of the predictive relationships can be neglected with minor only impact on the calculated repair cost. When considered as uncertain, the characteristics for the fragility curves were also shown to have an influence on the estimated repair cost, though reduced compared to the excitation characteristics (even compared to the high frequency of the source spectrum). The study also stresses the importance of a stochastic ground motion model that can adequately describe seismic events for smaller magnitude ranges, up to 5–5.5, since they were shown to contribute significantly towards the overall cost, especially with respect to the nonstructural components. Finally the impact of the seismic hazard variability, related to the probability distributions assumed for the moment magnitude and rupture distance, on the estimated cost was investigated. A computationally efficient scheme was discussed for this task, within the stochastic-simulation based framework, that requires no additional structural simulations. It was shown in this case that a decrease in the regional seismicity increases the importance of the seismic hazard characteristics towards seismic risk (repair cost). The entire study demonstrated the great benefits of the proposed modeling framework for the life-cycle losses, based on stochastic ground motion models; the framework offers great versatility in terms of calculating different aspects of the overall repair cost (including exploring the sensitivity with respect to the assumed fragility characteristics) while also allowing for direct control of the accuracy of the calculations and for an integrated global sensitivity analysis.