A stochastic extension of Navier's equations of elasticity is considered that accounts for uncertainties in the domain and material parameters. We consider a setting where a set of measurement data is combined with a simulated surrogate model formed using sparse grid stochastic collocation. By using this approach, we obtain a solution that recovers both the simulated model and the set of measurements. The accuracy of this approach is discussed both theoretically and in numerical experiments.

This presentation is part of Contributed Presentation “CP11 - Contributed session 11”

Authors:

Vesa Kaarnioja (University of Helsinki)

Harri Hakula (Aalto University)

Keywords:

elasticity problem, numerical linear algebra, partial differential equation models, sparse grids, stochastic collocation, uncertainty quantification, unstructured grids