Abstract :

The main goal of the performance-based earthquake design (PBED) is the probabilistic evaluation of the system-level performance of structures due to earthquakes. To this aim, a framework was proposed by the Pacific Earthquake Engineering Research Center based on the convolution of all levels of ground motion, structural damage, and loss. One of the key components in this estimation of structural performance is related to the possible complexity of the wave-field motion to be defined as an input to the SSI analysis. Traditionally, the estimation of seismic structural performance is decoupled in two separate steps: i) hazard analysis for ground motion (GM) estimation  generally defined in a single point at the site of interest), and ii) the SSI analysis based on the point-wise definition of the input. However, a limitation of this approach is related to the absence of the “local’’ geology in the consideration of the seismic input motion, possibly highly influencing the spatial variability of the seismic GM and thus the input signal to be defined.

With this regard, the aim of this thesis is to highlight the impact of each component on the seismic performance of critical structures, relying on state-of-the-art numerical tools for the prediction of earthquake ground motion as well as the assessment of damage for structural and nonstructural components (NSCs). The proposed approach makes use, on one hand of 3D physics-based simulations (PBS) for source-to-site wave propagation to provide an accurate description of spatial variability of ground shaking and, on the other hand, of systemic approaches for the evaluation of advanced SSI analysis, accounting for a complex 3D input motion excitation, for the damage analysis of structural and NSCs. The spectral element method (SEM) and finite element method (FEM) in time domain are used as numerical tools for the propagation on a regional and site/structural scales, respectively.

As a first step, and so as to be able to introduce a realistic 3D input motion to SSI analysis, a SEM-FEM weak coupling based on the Domain Reduction Method (DRM) is implemented and verified. Numerical verification is provided for a canonical case study existing in the literature. The implemented version allows to account for both complex regional and local geology in SEM and FEM models, via a not-honoring approach. An optimization procedure, based on the decimation of SEM output signal, is proposed to accelerate the computational time of the coupled approach while maintaining an accurate prediction of numerical results.

The evaluation of seismic hazard in the first step of the PBED methodology is defined here as the numerical prediction of the ground motion for a site-specific application. The case of the Kashiwazaki-Kariwa nuclear power plant (KKNPP) in the Niigata region of Japan is considered as the case study. Two rupture scenarios inspired by recorded aftershocks of magnitude 𝑀” close to 4 are numerically modeled via the RIK kinematic source model in order to investigate the effect of the dynamic excitation source. In addition, three geological models are examined for the region, allowing to evaluate the effect of regional and surface geology on the obtained surface ground motions. The comparisons are performed in terms of several intensity measures for the “free-field” motion at the KKNPP site.

Following the hazard analysis, the structural and damage analysis are performed assuming the reactor building of the Unit 7, a well-studied structure subject of the international benchmark of OECD/NEA Karisma. In addition to the identified geology effects, different SSI approaches are examined at first to evaluate the influence of a complex 3D input excitation imposed via the DRM instead of standard approaches (i.e., plane wave with vertical incidence or BEMFEM coupling) on both the structural and NSCs responses.

The relative average spectral acceleration (𝐴𝑆𝐴%&) is later used as an efficient and sufficient IM for the evaluation of the dynamic response of a hypothetical electric cabinet situated in the structure of interest. Given the time-consuming character of the 3D PBS, a reduced order model based on Synthetic Green’s Functions (SGF) is deployed in the last part in order to accelerate the computational chain and evaluate a series of plausible earthquake scenarios for the structure of interest. The cumulative complementary distribution function is computed for the case of this hypothetical electric cabinet.

Finally, this work highlights the importance of the: i) complexity of a realistic source, ii) wave propagation path and iii) local site effects, on the uncertainty of both the obtained dynamic excitation and the structural response.

————————————————————————————————————————————————————————————–

Un lien de connexion Teams est disponible pour ceux souhaitant suivre la soutenance à distance :

Click here to join the meeting