In recent decades increased urbanization and mass migrations towards cities have contributed to population shifts and infrastructure growth –leading in extreme cases to the formation of megacities– in some of the world’s most hazard-prone areas. The inevitable result is particularly large life and economic loss potential, something that has unfortunately been confirmed far too often by the thousands of lives lost and communities devastated in recent events like hurricanes Katrina and Ike and earthquakes in Haiti, Chile, New Zealand and Japan. Unfortunately, accurate assessment and mitigation of risk in such complex environments are non-trivial to achieve using traditional approaches. Even in the absence of extreme events, maintenance of our aging infrastructure network is emerging as another critical engineering challenge. Although Civil Infrastructure is second only to the health care industry in annual expenditures in the United States, it does not efficiently implement similar advanced diagnostic and decision support tools. While embedded sensors for continuous monitoring of bridges could help to deliver the data necessary for diagnostic and prognostic efforts, a rational and consistent framework that can use the vast assimilated data to guide the decisions about optimal infrastructure maintenance has yet to be delivered. These threats posed by natural hazards and aging civil infrastructure are acknowledged as grand challenges of the 21st Century by both the National Academy of Engineering and American Society of Civil Engineers, as they have the potential to undermine the most fundamental pillar of our society.
Motivated by this, the research, teaching and outreach of the HIgh Performance system Analysis and Design (HIPAD) Laboratory is an integrated effort to counter these threats, related to natural hazard risk assessment and mitigation and optimal infrastructure maintenance, through the implementation of novel probabilistic methodologies (focusing on stochastic simulation and sampling), advanced simulation-based engineering science tools, cyber-collaborations, and a variety of knowledge diffusion mechanisms and venues, including outreach in Haiti. In parallel our research extends to the probabilistic analysis and design of any engineering system warranted to exhibit higher performance, i.e. optimal life-cycle cost/benefit, maximum reliability, or minimal downtime, under regular operation and/or extreme loading conditions (for example, optimization of offshore wind turbines and offshore energy conversion devices). Uncertainties related to the characteristics of these systems and their operational environment significantly impact their performance and ultimately their optimal design.
Our work focuses on the robust (i) analysis, (ii) design and (iii) Bayesian model updating in presence of probabilistically characterized model uncertainties. This research also addresses the probabilistic quantification of model uncertainties but primarily focuses on efficient computational methodologies for the propagation of these uncertainties to calculate the system probabilistic performance (for example, risk assessment), for optimization of that performance (for example, risk mitigation) or for updating it when additional knowledge becomes available through monitoring data (for example, infrastructure condition assessment through health monitoring implementation).
The laboratory is equipped with Persephone, a high-performance computer cluster (forty-two nodes, each equipped with eight 2.57 Nehalem computational cores and 12 GB of RAM), Prometheus, a 1,792 CUDA core GPU (Graphical Processing Unit) personal supercomputer, and variety of tools for structural and stochastic simulation/optimization, including dedicated licenses for the TOMLAB optimization environment. This provides great opportunities to investigate applications of simulation-based engineering science to address modeling uncertainties in natural hazard risk mitigation, infrastructure condition assessment, and in the analysis and design of complex engineering systems, while adopting high-fidelity models to describe these systems and their environment (no constraints on model complexity). We additionally address the development of automated assessment tools for knowledge dissemination and for allowing non-technical end-users to leverage the full potential of the established research advancements.
For more details please see Research, Outreach and Projects tabs
Department of Civil Engineering and Geological Sciences
University of Notre Dame