• Matthias Faes
• Pengfei Wei
• Xiukai Yuan
• Jingwen Song
• Marcos Valdebenito, Adolfo Ibanez University
• Michael Beer

Computational mechanics offers the tools for constructing numerical models that predict the behavior of complex engineering systems. These models can be quite involved and elaborate, as they must be able to represent appropriately issues such as nonlinearity and time-variant behavior, among others. However, in practical applications, one is usually confronted with varying degrees of uncertainty for defining and analyzing a model. Hence, in order to characterize the performance of an engineering system, it is necessary to perform uncertainty quantification. Such task can be quite challenging from a numerical viewpoint, as it implies conducting repeated analyses of the numerical model under different scenarios. In addition to this challenge, uncertainty associated with a problem may stem out of different sources. In some cases, uncertainty may be due to inherent randomness (that is, aleatory uncertainty) and is best described by means of probability. In some other cases, uncertainty is caused by imprecision due to lack of data, measurement errors, conflicting sources of information, etc. The latter class corresponds to epistemic uncertainty and can be described resorting to intervals, fuzzy variables, etc. Naturally, in the more general case, one may be faced with the task of coping with both randomness and imprecision. Such situation imposes an additional challenge, as it is necessary to propagate both types of uncertainty (but without mixing them) during the analysis.

Explicit consideration of randomness and imprecision offers a powerful framework for coping with uncertainties. In essence, it provides a collection of probabilistic analyses (performed under aleatory uncertainty) which are indexed by the model describing epistemic uncertainty. Nonetheless, its practical implementation is far from trivial, as it demands increased numerical efforts when compared with purely aleatoric analysis. Therefore, the aim of this mini-symposium is addressing the very latest development on approaches for uncertainty propagation under randomness and imprecision. The scope of the mini-symposium is broad, as it covers: different models for representing uncertainty such as classical probabilities, intervals, fuzzy analysis, imprecise probabilities, evidence theory, etc.; novel formulations for coping with aleatoric and epistemic uncertainty; advanced simulation methods; development and application of surrogate models, etc. Both theoretical developments and applications involving systems of engineering interest are particularly welcomed in this session.

This activity is organized under auspices of the Committee on Probability and Statistics in Physical Sciences (C(PS)^2) of the Bernoulli Society for Mathematical Statistics and Probability and the Risk and Resilience Measurements Committee (RRMC) of the Infrastructure Resilience Division (IRD) at the American Society of Civil Engineers (ASCE).