Uncertainty and Probabilistic Analysis

Backgrounds

Every finite element analysis is subject to numerous uncertainties. These may be related to the numerical tool itself (discretization, element formulation, solver,...) or to the physics of the problem.

Part of the uncertainty in a FE model can be eliminated or at least reduced by means of prototype testing and model validation and updating, but part of the uncertainty remains. For example the variability of the input parameters due manufacturing processes or environmental conditions may or may not have been taken into account when doing prototype testing.

Uncertainty analysis is finding the relation between given variability and probability on input parameters and the variability and probability of output parameters. With this information, the engineer assesses if the structure is a robust design. Inversely, it is possible to optimize a design to satisfy given constraints on the variability of the responses and thus derive acceptable tolerances on the input parameters. This directly leads to conclusions with respect to manufacturing processes or operating conditions.

Related Tools and Applications

Monte Carlo sampling.
Statistical correlation.
Variability propagation
Probabilistic model updating.
Uncertainty classification.
Robust design.
Design space exploration.
Design of experiments.
Reliability analysis.

Contact DDS for custom solutions in this area.