If you still want to submit a Special Session proposal, in a theme not covered below, please send ASAP an e-mail to André T. Beck with the name of the proposed Special Session, the name, affiliation and e-mail address of the organizers and an abstract for the SS. SS organizers are expected to lead the review of abstracts and full papers submitted to their SS. Please use the TEMPLATE available HERE to submit your proposal.

Advanced simulation methods for uncertainty quantification and risk analysis

Edoardo Patelli, Alejandro Diaz de la O, Siu-Kui Au, Michael Beer


Engineering problems have uncertainties in various forms and of various nature arising from, for example, limited information, human factors, subjectivity and experience, imprecise measurements and unknown physics. Despite this, decision makers still need to make definite choices based on the available information. To deliver products with commissioned reliable performance, complex technological installations, engineering systems and components have to be designed to cope with risk and uncertainty. Probabilistic and non-probabilistic as well as mixed concepts of imprecise probabilities have been developed, applied and achieved a new level of acceptance.

The solutions of such problems are increasingly leveraging on the availability of efficient simulation methods. While new algorithms are still arising in different fields, Monte Carlo methods are now established and ready for applications in the real practical engineering setting, thanks to the advent of modern computer technology and advancements in algorithm efficiency. Examples of new methods include Subset Simulation, Line Sampling, sparse-grid stochastic collocation methods, Bayesian nested sampling and new adaptive Monte Carlo and quasi-Monte Carlo methods. Nevertheless many issues are still encountered in real applications and new expectations or endeavors are forming upon recognition of the availability of these powerful tools.

This Special Session aims at bringing together researchers, academics and practising engineers, providing a forum for discussion on theoretical and practical issues in the development, implementation and scalability of Monte Carlo methods. Contributions to theory development, applications, and implementation in engineering practice, are welcome. The issues of numerical efficiency and applicability to industry-size problems are of particular interest.

Approaches for Uncertainty Quantification in Structural Dynamics

Michael Beer, Hector A. Jensen, Edoardo Patelli, Ioannis A. Kougioumtzoglou, Marcos A. Valdebenito


Analysis of structural systems subject to dynamic loadings is extremely common in several disciplines such as aerospace, civil and mechanical engineering, among others. Usually, this type of analysis is carried out by means of highly detailed and numerically involved computational models. In order to ensure such models provide meaningful results, it is necessary to explicitly take into account the effects of the uncertainty on loadings and structural parameters that affect the performance. This may become a daunting task, as it adds an additional level of complexity over the (already challenging) structural dynamic deterministic analysis. Hence, there is an evident need for numerical methods that allow quantifying uncertainty in structural dynamics. Then, the aim of this special session is bringing together the latest developments on this field, with emphasis on approaches which exhibit a high numerical efficiency and that allow to model realistic systems of engineering interest.

The scope of this special session is wide, at it comprises: approaches for uncertainty quantification involving both probabilistic models and non-traditional models (such as intervals, fuzzy analysis and imprecise probabilities); analytical approaches and simulation methods; application of approximations and meta-models; forward problems (such as reliability analysis, sensitivity analysis, reliability-based optimization, robust risk analysis, etc.) and inverse problems (such as model identification and model updating); etc. Both theoretical developments and applications involving systems of engineering interest are particularly welcomed in this session.

This mini-symposium 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.

Complex Engineered Networks and Infrastructure Systems

Konstantin M. Zuev, Edoardo Patelli, Michael Beer, Matteo Broggi, Frank Coolen


Complex engineered networks are a pervasive feature of modern society. Examples include transportation systems (road, rail, and airlines), electric power grids, networks of natural gas pipelines, cellular grids, and the internet. These distributed infrastructure systems with many interconnected components provide critical services for everyday life, such as water, food, energy, transport, communication, banking, and finance. Moreover, most of these critical infrastructures are interconnected and interact with and depend on social networks. As a result of technological progress and worldwide urbanization, the dependence of our society on these complex systems spanning cities, countries, and even continents, constantly grows. Given the critical role that engineered networks play in the functioning of our societies, there is an increasing demand for these systems to be highly reliable. A deep understanding of their actual capabilities to withstand natural hazard, such as earthquakes, tsunamis and hurricanes, and man-made threats, e.g. accidents and terrorism, is crucial. The related issues of resilient network design and operation are also closely related to sustainability problems which are of increasing importance today. In particular, the degree to which a technological network subjected to internal or external stresses (e.g. cascading failures or seismic hazards) is capable of keeping (or recovering) the service demanded needs to be quantitatively estimated. In this respect, cascading failures, where external perturbations trigger some initial local failures that lead to eventual global network failure, are especially hazardous. Quantitative assessment of network reliability and associated risks and uncertainties is therefore a key aspect of system design, optimization, and operation.

The main objective of this Special Session is to bring together experts working in the interdisciplinary area of engineered networks and infrastructure systems to present and discuss the latest developments in the field. Some relevant topics include reliability, risk, vulnerability and resilience analyses of critical infrastructures, multi-sector interdependencies of infrastructure networks, survival signature, common cause failure, and cascading failures.

This mini-symposium 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.

Imprecise probability

Scott Ferson, Frank Coolen, Edoardo Patelli, Michael Beer


Reliability and performance analyses of complex systems become increasingly complicated due to limited, vague and imprecise information. This problem has generated significant developments on generalized approaches for uncertainty quantification with the key question of how to model epistemic uncertainty. In many practical cases only ranges or bounds are available for some parameters so that set-theoretical descriptors provide an appropriate model. In combination with probabilistic information this leads to imprecise probabilities as the theme of this Mini Symposium. Whilst previous developments were heavily focused on modeling with often simple applications to demonstrate basic features, the approaches have more recently reached a stage of sophistication that enables the solution of real-size problems. This step has been achieved by combining imprecise probabilities with established and emerging concepts and techniques from the traditional probabilistic field, specifically, with advanced stochastic modeling and Monte Carlo simulation. In view of model capabilities and numerical efficiency, concepts and technologies from mathematics and computer science have been adopted to meet the engineering requirements.

This mini-symposium aims at bundling the most recent developments in the area of imprecise probabilities, including strategies for bounding probabilities, in the context of challenging geotechnical and structural engineering problems. Contributions may have a mathematical, a computer science or engineering nature. The models may include all variants of imprecise probabilities such as interval probabilities, p-box approach, evidence theory, fuzzy probabilities and so forth. The issues of numerical efficiency and applicability to industry-size problems are of particular interest.

Operations Research Applications in Resilience Modeling

Hiba Baroud, Camilo Gomez


Modeling and analysis of the resilience of systems garnered an increased interest among practitioners and researchers in recent years. These systems include infrastructure networks, communities, and processes, among others. As resilience definitions and metrics are developing and becoming more extensive, there is a need to frameworks and tools that provide the ability to quantify, measure, and analyze resilience as a function of deterioration and uncertainty due to natural and man-made hazards, as well as constraints such as the availability of resources.

This session is focused on advances in operations research, more specifically mathematical modeling, data analytics, and optimization approaches that are applied to model the resilience of systems. Methodological topics include but are not limited to risk analysis, reliability, graph theory, stochastic optimization, network optimization, statistical learning, interdependency modeling, and decision analysis. Application areas include critical infrastructure systems such as energy, water, transportation, communications, and manufacturing, among others, as well as community resilience and social vulnerability.

Software for Uncertainty Quantification

Stefano Marelli, Bruno Sudret, Edoardo Patelli


The awareness of the role of uncertainty in virtually all fields of applied sciences has grown steadily over the past decades. The inclusion of uncertainty quantification (UQ) in predictive models is a technical challenge that fostered the development of techniques such as probabilistic/non-probabilistic modelling of the sources of uncertainty, surrogate modelling, sensitivity analysis, model calibration, robust optimization, etc. The deployment and further diffusion of such techniques, however, is closely related to the availability of proper software that can be incorporated by researchers and practitioners into their own workflows.

This mini-symposium aims at bringing together leading and innovative players in the international uncertainty quantification software scene so as to foster discussions and exchange of ideas between developers and perspective users. Contributions are welcome on the following topics: non-intrusive UQ techniques, surrogate modelling, multi-fidelity modelling, high-performance computing in UQ, general-purpose UQ software and case studies and applications to real-scale industrial problems.

Stochastic inverse problems in linear and nonlinear dynamics

Sondipon Adhikari, Anas Batou, Thiago Ritto


Inverse problems are fundamental to the development and validation of numerical models of complex dynamic systems. Such problems are, for instance, encountered in engineering design and model updating of complex dynamic systems. In presence of uncertainties or variabilities in the dynamic systems, these problems should be redefined and suitable methods need to be developed for solving them. In this context, this Mini-symposium will be focused on recent advances in theoretical, numerical and experimental methods related to stochastic inverse problems in linear and nonlinear dynamics. The topics to be covered will include, but are not limited to:

– Parameter identification.

– Structural modification.

– Model updating.

– Stochastic model updating.

– Dynamic reanalysis.

– Uncertainty identification.

– Bayesian approach.

– Sensitivity analysis.

– Model reduction and metamodeling.

– Design of experiments.

– Experimental uncertainties.

– Non probabilistic approaches.

– Robust optimization and design problems (RBO, RBDO).

Surrogate models for uncertainty quantification, reliability analysis and robust design

Stefano Marelli, Bruno Sudret, Sankaran Mahadevan, Alex Taflanidis


In the last decade, uncertainty quantification has emerged as a research topic at the boundary of statistics and probability, applied mathematics, computational science and engineering. In many fields of applied sciences, it is now customary to study the impact of model- and parameter uncertainties on the predictions of the computational models used for designing, operating or monitoring engineering or environmental systems.

However, accurate computational models (e.g., finite element analysis) of complex structures and systems are often costly. A single run of the model may require minutes to hours of CPU time, even on powerful computers. The direct use of these models for reliability analysis and reliability-based design optimization, which require repeated calls to the computational code, is therefore infeasible. As a means to overcome this limitation, it is necessary to develop substitutes that may be evaluated thousands to millions of times at low cost: these substitutes are referred to as meta- or surrogate models.

The aim of this mini-symposium is to compare various families of meta-modelling techniques in the context of uncertainty propagation. These include polynomial chaos expansions, Kriging (Gaussian process modelling), support vector regression, neural networks, low-rank tensor approximations, etc. New methodologies and algorithms related to structural reliability, sensitivity analysis, reliability-based design as well as large scale industrial applications that make use of surrogate models are equally welcome in this mini-symposium.

Uncertainty Quantification in Resilience Modeling

Hiba Baroud, Henry Burton


The importance of having robust and resilient systems has gained the attention of decision makers and government officials in the past decade. Resilience is often thought of as the ability exhibited by a system to “bounce back” following a disturbance. Modeling the resilience of a particular system is generally focused on the behavior of the system during the response and recovery phase which has a stochastic nature. As such, accounting for the uncertainty is a significant component of studying systems resilience.

The goal of this session is to bring together researchers and practitioners that are working on advances in computational and mathematical methods of uncertainty applied to resilience modeling. Topics include but are not limited to uncertainty quantification in risk analysis, vulnerability analysis, recovery simulation and interdependency modeling with applications in buildings, transportation, energy, water, freight and other infrastructure systems.