S17: AAdvanced modeling techniques: probabilistic, reduced order and data driven approaches
Ludovic Noels (U Liege, Belgium) firstname.lastname@example.org
Francois Willot (Mines ParisTech, France) email@example.com
Elias Cueto (U. Zaragoza) firstname.lastname@example.org
Felix Fritzen (U. Stuttgart) email@example.com
In the recent years non-determinism has been acknowledged as a major issue which affects structural and material performance and reliability. One source of uncertainties is at the material level itself and its mechanical properties, which should thus be quantified. For example, in safety critical applications it is important to be able to estimate the probability of rare realizations, such as a significant decrease of strength due to the deviation of the material properties from the design. However experimental characterization alone would require an excessive number of tests to capture the tails of the distributions. Therefore, there has recently been a growing interest in stochastic virtual testing.
Because of the different involved length scales, this however requires the development of combined experimental/numerical frameworks to up-scale the uncertainties at the material scale resulting from the manufacturing process to the structural scale.
Numerous challenges are identified in this context with, as a non-exhaustive list
Quantifying the material uncertainties at the different scales;
Predicting material uncertainties from simulation of the manufacturing process;
Determining how random microstructures and heterogeneities may be characterized statistically, and identifying parameters relevant to process or properties;
Developing virtual micro-structure generators using random sets theory, multi-point methods, machine learning techniques and others;
Identifying model parameters with their uncertainties from a limited number of measurements;
Developing efficient micro-structure numerical analysis tools etc;
Conducting efficient stochastic multiscale analyses using order reduction, surrogate model, machine learning theories…;
Developing models linking structural responses to manufacturing process for process optimization;
Assessing the homogenized properties of random media by change of scale using e.g. meshless spectral solvers.
The purpose of this mini-symposium is to discuss the recent advances in the related topics for different type of materials. Participants will deliver scientific talks on their material of interest, including, but not limited to:
Metallic and polymeric materials obtained by additive manufacturing, including metamaterials;
Dense to porous ceramics;
Polycrystals with defects including pores, intra-and inter-granular precipitates, grain boundary phases, additional solid phases;
Materials with compositional and gradients of properties.