A comparison of stochastic and data-driven FEM approaches to problems with insufficient material data

In many applications the experimental results are not sufficient for material characterization. Nonetheless, predictive numerical simulations require to harness the available stress-strain data and to use them, e.g., as input parameters for finite element computations. In this contribution two strategies, a stochastic and a data-driven finite element method are compared and their advantages and disadvantages with respect to material uncertainties are studied. Examples in one and three dimensions are computed to illustrate the methods' properties. In particular, the data-driven method has shown to be a new and promising approach to the problem of imprecise material data. (C) 2019 Elsevier B.V. All rights reserved.