Probabilistic methods and seismic risk

Probabilistic methods and seismic risk

Background

In the recent years, nonlinear transient analyses have become the preferred approach for the evaluation of seismic performance of buildings.  Such analyses rely on “best-estimate” structural models and thus have to be accompanied by sensitivity analysis, confidence intervals and other pertinent statistical quantities. When a great number of deterministic analyses is needed for the statistical estimations then it can be useful to construct a metamodel (also called response surface). The latter is a reduced statistical model that describes part of the physics by providing simplified input-output relations.

Objectives

Moreover, Performance-Based Earthquake Engineering (PBEE) is becoming the state of the art for risk-informed decision making.  It provides robust tools and methods for design and verification of engineered structures whose seismic performance has to meet diverse economic and safety needs of owners and society.  Both the uncertainty related to the seismic hazard and the uncertainty of data and structural models have are accounted for in the probabilistic performance based approaches. In the nuclear domain, the seismic Probabilistic Risk Assessment (PRA) methodology has become the most commonly used approach for the evaluation of seismic risk and is now applied worldwide.  The key elements of a seismic PRA studies are seismic hazard analysis, seismic fragility evaluation (for each component and substructure) and system analysis including the construction of logical fault tree model. These three elements allow for the evaluation of the failure probability of the industrial plant.

The structural reliability is generally characterized by fragility curves, expressing the failure probability of a structure or component as a function of seismic ground motion intensity (for example peak ground acceleration).  In this context, the following research topics are addressed

  • Model reduction (both metamodels and physics-based reduced models) and uncertainty propagation
  • Choice of hazard consistent ground motion time histories for transient structural analysis
  • Development of robust and efficient methods for the evaluation of fragility curves in the industrial  context