From component behavior to network performance, this research thrust leverages and develops machine-learning-based surrogates, network science tools, and UQ frameworks to accelerate from structural reliability to system lifecycle assessment.
The question
Infrastructure performance emerges across scales: how a component behaves shapes how a network performs, yet the two are usually modeled in separate silos, or integrated together using high-fidelity models that make the analysis too expensive to run at scale.
What we do
We develop algorithmic methods that simulate heterogeneous infrastructure from the structure to the network scale. For example, we use machine-learning-based surrogate models to make expensive structural models tractable and couple them to network-level performance. To reduce the computational burden of integrated models, we investigate computational methods that trade off efficiency and accuracy by identifying which interactions most influence system outcomes, and we propose objective methods for such assessments. We also study smart-modeling techniques that enable ML/AI-based models to autonomously improve their predictive capabilities, accelerating model training in risk and resilience contexts.
Publications
2025
ICOSSAR’25
Collective behaviors in regional seismic responses: insights from phase transitions in statistical physics
Sebin Oh, Raul Rincon, Jamie Ellen Padgett, and Ziqi Wang
In 14th International Conference on Structural Safety and Reliability (ICOSSAR’25), 2025
Parameterized Fragility Assessment of Coastal Structures: Capturing the Influence of Neighboring Structures
J. M. Patel, Raul Rincon, and Jamie Ellen Padgett
2025
Journal of Structural Engineering, in review (July 2025)
2024
Earthquake Spectra
Fragility modeling practices and their implications on risk and resilience analysis: From the structure to the network scale
Although fragility function development for structures is a mature field, it has recently thrived on new algorithms propelled by machine learning (ML) methods along with heightened emphasis on functions tailored for community- to regional-scale application. This article seeks to critically assess the implications of adopting alternative traditional and emerging fragility modeling practices within seismic risk and resilience quantification to guide future analyses that span from the structure to infrastructure network scale. For example, this article probes the similarities and differences in traditional and ML techniques for demand modeling, discusses the shift from one-parameter to multiparameter fragility models, and assesses the variations in fragility outcomes via statistical distance concepts. Moreover, the previously unexplored influence of these practices on a range of performance measures (e.g. conditional probability of damage, risk of losses to individual structures, portfolio risks, and network recovery trajectories) is systematically evaluated via the posed statistical distance metrics. To this end, case studies using bridges and transportation networks are leveraged to systematically test the implications of alternative seismic fragility modeling practices. The results show that, contrary to the classically adopted archetype fragilities, parameterized ML-based models achieve similar results on individual risk metrics compared to structure-specific fragilities, promising to improve portfolio fragility definitions, deliver satisfactory risk and resilience outcomes at different scales, and pinpoint structures whose poor performance extends to the global network resilience estimates. Using flexible fragility models to depict heterogeneous portfolios is expected to support dynamic decisions that may take place at different scales, space, and time, throughout infrastructure systems.
npj Nat. Hazards
Future cities demand smart and equitable infrastructure resilience modeling perspectives
Risk-informed decisions that promote infrastructure resilience (or the ability to withstand, recover from, and adapt to stressors like natural hazards) require confident predictions of system performance now and into the future. We propose a perspective shift–one capable of handling uncertain and dynamic conditions, leveraging emerging observations from smart systems, and guided by demands for social equity. This shift requires collective efforts, but our future cities demand and deserve it.
WCEE 2024
Intelligent learning paradigms to enable adaptable seismic fragility and restoration models
Raul Rincon and Jamie Ellen Padgett
In 18th World Conference on Earthquake Engineering (WCEE 2024), Nov 2024
2022
Empirical fragility assessment of adobe and rammed earth walls subjected to seismic actions
Raul Rincon, Juan C. Reyes, Julian Carrillo, and Alejandra Clavijo-Tocasuchyl
Earthquake Engineering & Structural Dynamics, Nov 2022
In many Latin American countries, most constructions from the colonial period were built in earth masonry. During past earthquakes, some of these buildings collapsed even with moderate magnitude events resulting in significant loss of human lives. Nevertheless, empirical fragility functions at the component level (FFs) for these structures have not been comprehensively investigated in the literature. This paper presents the probabilistic characterization of the uncertain damage sustained by earthen vertical wall segments subjected to seismic actions. Drift-based FFs are constructed for in-plane (IP) actions using 48 test results from previous studies. Three damage states, corresponding to cracking, extensive cracking, and collapse, are proposed for this loading direction. Only the collapse state is analyzed for out-of-plane (OOP) forces and a total of 26 experimental results comprise the database for this case; peak ground acceleration is selected to represent the intensity measure for OOP demands. Analysis of variance (ANOVA) tests aided to confirm that, for in-plane demands, the compressive strength is one of the most influential parameters for the cracking limit state. ANOVA tests also helped to identify that only the slenderness and boundary conditions seem to have a slight influence in the FFs for the OOP demands. Finally, the derived FFs were used to recommend drift and acceleration-based limit states for earthquake-resistant assessment of earthen components. The developed fragility functions may be used to predict the probability of occurrence of a certain damage state at a component level which can be used in risk and resilience assessments of vernacular buildings.
2018
Practical seismic microzonation in complex geological environments
Luis E. Yamin, Juan C. Reyes, Rodrigo Rueda, Esteban Prada, Raul Rincon, and 3 more authors
Soil Dynamics and Earthquake Engineering, Nov 2018
The seismic design of buildings and infrastructure components requires the estimation of the hazard considering the dynamic response of the soil deposits, which substantially modifies the characteristics of the input motion at the rock basement. Seismic microzonation studies attempt to identify geologic zones of an area of interest with similar seismic hazard at a local scale. This paper presents a methodology to obtain seismic spectral amplification factors within each soil zone characterization considering the main sources of uncertainty. Results are presented in terms of spectral amplification factors for various seismic intensities and soil profile vibration periods. Design soil amplification factors can then be mapped using the measured vibration period of the soil profile at each location and the seismic intensity at bedrock for a given design return period. Response and design spectra may then be estimated at surface level for every location. Results can be easily integrated into probabilistic risk assessment platforms such as CAPRA (www.ecapra.org) for hazard and risk evaluations.
2017
Probabilistic seismic vulnerability assessment of buildings in terms of economic losses
Luis E. Yamin, Alvaro Hurtado, Raul Rincon, Juan F. Dorado, and Juan C. Reyes
The probabilistic seismic risk assessment in terms of economic losses for building portfolios aims to the estimation of the probability distribution functions (PDF) of economic losses for a set of stochastic events representing the seismic hazard at a particular geographic zone. This paper proposes a methodological approach to evaluate and integrate, in a consistent and rigorous way, the economic losses as a function of the seismic hazard intensity for prototype building constructions. Prototype building models are designed and characterized with a set of reference parameters. By means of 3D structural models, detailed nonlinear response history analyses are performed for a set of seismic records at several increasing intensities. Seismic records are selected to represent particular seismological and geotechnical conditions at the site of analysis. Then, a component-based model is conformed considering structural, non-structural, and content components potentially susceptible to damage. Each component type is assigned a fragility specification for various damage states in terms of costs and times of repair. Using Monte Carlo simulations, the different sources of uncertainty are included in the assessment of the costs and times of repair at different seismic intensities. Uncertainties in the hazard, model response, damage states, and costs and times of repair are considered. Aspects such as geographical variations in the hazard, scale economy, special commercial conditions, minimum or total intervention costs, and business interruption costs are included in the assessment. Finally, the results are represented by means of vulnerability functions for specific building typologies. To illustrate the methodology, a case study is presented in detail for a typical 5-story reinforced concrete moment resisting frame building designed for special seismic code level and located in a typical soft soil deposit of Bogotá, Colombia. Additional results are presented for six (6) different building typologies illustrating variations in results due to different story heights and seismic code levels. The resulting vulnerability functions are compared with equivalent results from other similar methodologies. Conclusions and possible potential applications related to probabilistic risk assessment are summarized.
