Seismic fragility analysis with the generalized lambda model

In earthquake engineering, the performance of a structural is typically assessed by a fragility model. The latter depicts the failure probability of the structure as a function of the intensity measure (or a set of intensity measures) of seismic loads. More precisely, the failure probability domain is defined by a given engineering demand parameter (EDP) (e.g., the maximal inter-story drift for a multi-story building) exceeding a certain threshold for a given value of the intensity measure. Due to the high cost of experiments and numerical simulations, performing fragility analysis is usually intractable.

In this project, we apply the generalized lambda model (external pageZhu and Sudret, 2021call_made) for emulating the distribution of the EDP conditioned on the selected intensity measure. This surrogate relies on the use of the flexible generalized lambda distribution to represent the conditional distribution. The distribution parameters are functions of the intensity measure and are modeled by polynomial chaos expansions. Once constructed, the surrogate can be post-processed to obtain the fragility function. Moreover, this model can be directly integrated into the framework evaluating the decision variable (e.g., monetary losses) of structures.