Fast validation of stochastic structural models using a PGD reduction scheme

In this work we propose to use a reduced model based on the Proper Generalized Decomposition (PGD) method to obtain a quick estimate of the response of a structure whose material parameters are considered to be uncertain. First, we give a detailed description of the construction of this reduced model in a finite element framework. Then, the performance of the proposed method is illustrated through two examples. Finally, we show how this reduced model can be used in an inverse approach in order to estimate the dispersion of the material parameters based on measurements taken from a family of structures.

• We build a reduced mechanical model function of structural parameters.
• We use the reduced model to obtain an accurate estimation of a probability of failure.
• We use the reduced model in the context of parameters identification.
• Our approach efficiency is demonstrated using a CPU-Time criterion.