Meta-models/surrogate models for uncertainty propagation, sensitivity and reliability analysis

Abstract

Structural reliability methods and more generally, methods that aim at taking into account model- and parameters uncertainty in computational mechanics and civil engineering have received much attention in the past two decades. Some well-known methods such as FORM/SORM for reliability analysis, spectral methods for stochastic finite element analysis, global sensitivity analysis (Sobol’ indices), etc. are nowadays applied in the industrial context, e.g. in nuclear, aerospace or automotive engineering.
However, accurate computational models for industrial structures or systems are usually based on finite element or finite difference methods. Thus a single run of the model may last several minutes to several hours, even on recent computers. In order to use these models for uncertainty and reliability analysis, which requires repeated calls to the computational code, it is necessary to develop a substitute that may be evaluated thousands to millions of times at low cost: these substitute are called meta-models or surrogate models.
The aim of this mini-symposium is to confront various kinds of meta-modelling techniques in the context of uncertainty propagation including classical response surfaces, polynomial chaos expansions, kriging, support vector regression, neural networks, sparse grid interpolation, etc.
Papers that present new developments in the methodology as well as large industrial applications that make use of meta-models are welcome.

Organizers

Bruno Sudret, Phimeca Engineering, Paris, France, (sudret@phimeca.com), Jean-Marc Bourinet & Nicolas Gayton, IFMA/LaMI, Clermont-Ferrand, France Marc Berveiller, EDF R&D, Moret-sur-Loing, France