Mechanical and Civil Engineering Seminar

Uncertainty in simulations is due to the stochastic nature of geometric and physical parameters, indeterminate nature of initial/boundary conditions, and inadequacy of physical models coupled with discretization errors. These uncertainties can be classified into two types -- parametric (aleatory) uncertainty and model form (epistemic) uncertainty. Whereas the probability theory can deal with the parametric uncertainty, some generalization of probability theory is required to deal with the model form uncertainty. Among the generalizations of probability theory, evidence theory is relatively well developed. The presentation will briefly discuss some techniques based on these theories to quantify uncertainty in the context of some representative problems.