ISRERM2026 10^th International Symposium on Reliability Engineering and Risk Management

The long-term reliability of structural systems pose critical research challenges on a global scale. Over their life cycles, structural systems face deterioration and exposure to hazardous loads, leading to significant uncertainties that are both time-dependent and space-dependent. These uncertainties give rise to a variety of complex failure modes that require robust analytical approaches for comprehensive evaluation.

This Mini-Symposium aims to unite leading experts in structural reliability to share and discuss recent advancements in the field. The forum will provide a platform for addressing a broad spectrum of topics while emphasizing the distinct characteristics that define and advance our understanding of structural reliability.

Topics for potential contributions include, but are not limited to:

• Structural reliability and risk assessment
• Time-dependent reliability analysis methods
• Dynamic reliability analysis methods
• Advanced stochastic models for structural reliability
• Reliability-based design and optimization methodologies

The safety and reliability of engineering structures critically depend on the accurate identification of system parameters, boundary conditions, and external loads. However, real-world engineering systems are inherently affected by various sources of uncertainty, including material variability, measurement noise, model inadequacy, and environmental randomness. Conventional deterministic inverse problem methods often fall short in such settings. This mini-symposium focuses on the integration of uncertainty inverse problems and stochastic system identification, aiming to develop robust and efficient computational and experimental approaches to quantify the propagation of uncertainty throughout the inversion process. Such efforts are essential to enhance the credibility of structural condition assessment, damage diagnosis, and life prediction, and to provide a solid scientific basis for risk-informed decision-making in reliability engineering. contributions on the following topics (but not limited to) are invited:

• Problem setting and theoretical frameworks for uncertainty inverse problems;
• Bayesian inference and Markov chain Monte Carlo (MCMC) methods;
• Stochastic optimization and robust inversion techniques;
• Methods and techniques for stochastic system identification;
• Hybrid identification methods integrating multi-source data (e.g., sensor networks, images, physics-based models);
• Machine learning and data-driven approaches for uncertainty quantification;
• Deep learning in inverse problems, including Bayesian neural networks;
• Model reduction and surrogate modeling for accelerated stochastic computation;
• Stochastic parameter identification from vibration responses;
• Identification of time-varying and nonlinear systems under uncertainty;
• Random field modeling and spatial variability quantification;
• Stochastic loading identification;
• Engineering applications and case studies, including: Stochastic damage identification in civil infrastructure (bridges, buildings, dams); Inverse reliability analysis of aerospace and mechanical structures; Stochastic response reconstruction under extreme events (earthquakes, wind, impact).

We strongly encourage and warmly welcome contributions that focus on pioneering theories and real-world applications in engineering mechanics, civil engineering, mechanical engineering, and related fields.

Reliability engineering and risk management increasingly rely on exploring diverse models (e.g., data-driven, physics-based, and hybrid frameworks) to enable robust analysis and decision-making under limited information and uncertainty. The predictive credibility of these models—critical for trustworthy design and risk assessment—hinges on rigorous inference frameworks. Bayesian model inference offers a principled paradigm for model validation, calibration, selection, fusion, and credibility-enhancing experimental design, bridging theoretical foundations with engineering practice. This Mini-Symposia aims to convene researchers and practitioners to explore cutting-edge advances in Bayesian model inference, addressing the following interconnected themes (but not limited to):

• (1) Theoretical Foundations, including novel Bayesian nonparametric methods, hierarchical modeling for multi-scale systems, Bayesian model calibration/selection/averaging for uncertainty quantification, credibility metrics integrating physical constraints, structural health monitoring, high-precision learning from small informative datasets or large, heterogeneous and (spatially and temporally) correlated dataset, handling the statistical nature of the measurement uncertainty, etc.
• (2) Numerical Innovations, focusing on scalable MCMC algorithms, deep learning-driven variational inference, approximate Bayesian computation for complex physics models, and computational efficiency strategies for high-dimensional parameter spaces.
• (3) Engineering Applications, spanning reliability assessment in aerospace propulsion systems, risk-informed design for civil infrastructure under climate uncertainty, fault diagnosis in energy grids, and safety-critical systems in healthcare.
• (4) Emerging Frontiers, welcoming interdisciplinary ideas at the intersection of Bayesian inference with AI-driven automated model construction, quantum-informed inference, and real-time data assimilation for dynamic systems.

By fostering cross-disciplinary dialogue, the symposium seeks to advance methodological rigor, promote practical implementations, and identify new frontiers in leveraging Bayesian inference to enhance model credibility and their applications in risk & reliability. This platform will facilitate knowledge exchange, collaborative problem-solving, and the translation of theoretical insights into actionable engineering solutions.

Amidst the development of sensor technologies and structural health monitoring (SHM) systems, various model updating methods have been developed to update the in-situ properties of structures and their future performance based on measured data. The response of structures under working conditions is complex and influenced by operational/environmental variations. The need to explicitly consider the uncertainty due to these factors is well recognized. A Bayesian perspective, i.e., viewing probability as a plausibility of propositions conditional on given information instead of relative frequency of occurrence, provides a powerful approach for mathematical formulation, delivering the ‘posterior’ probability distribution of system parameters given the information from data and consistent with modeling assumptions. Recent years have seen significant progress in methodologies that combine traditional domain-specific models with machine learning techniques. Challenges still remain, however, such as extracting interpretable information and learning governing physical principles from monitoring data, improving computational efficiency when dealing with high-dimensional systems and large SHM datasets, etc.

This min-symposium arouses contributions to Bayesian approaches of model updating and risk management with applications such as structural health monitoring. Possible topics include but are not limited to: (1) Bayesian system identification of engineering structures, (2) AI-based methods for data processing, (3) performance assessment and risk managements in SHM, (4) physics-data fusion model combing machine learning and physics, (5) new technologies for operational modal analysis.

The proper quantitative treatment of uncertainties is a fundamental prerequisite for accurately assessing the response and reliability of various engineering systems. Over the past few decades, the stochastic dynamics community has focused on determining the system responses and evaluating the associated risks in complex dynamic systems subjected to stochastic excitations. In this context, many researchers have made significant contributions to determining the response and reliability of diverse linear and nonlinear systems, including those exhibiting hysteretic behaviour, as well as systems incorporating fractional derivative elements. However, challenges remain regarding the precision and computational efficiency of the associated analytical and numerical methodologies, particularly in high-dimensional systems and systems exposed to non-Gaussian, non-white, and non-stationary excitations.

The primary goal of this MS is to highlight recent breakthroughs and emerging interdisciplinary approaches within the broad field of computational techniques aimed at addressing current challenges in stochastic engineering dynamics, with a specific focus on uncertainty modelling and propagation. Additionally, this MS seeks to establish a platform for the exchange of ideas and collaboration among diverse technical and scientific disciplines. We invite contributions that encompass both fundamental research and practical applications in the domain of stochastic dynamics, including order reduction techniques, risk and reliability assessment applications, joint time-frequency analysis methods and tools, sparse representation methodologies, compressive sampling, and data driven approaches. Furthermore, we encourage submissions related to stochastic and fractional calculus modelling and applications, Monte Carlo simulation methods, stochastic structural design applications, performance-based engineering methodologies, and related subjects.

Ensuring the reliability of dynamic structures under environmental loads requires a strong understanding of uncertainties and their effects on structural performance. Civil, offshore, and energy infrastructure, such as high-rise buildings, long-span bridges, offshore wind farms, and industrial facilities, must endure variable and often extreme environmental forces, including wind, waves, earthquakes, and temperature changes. These uncertainties, present in both external loads and structural characteristics, create major challenges in predicting structural responses and managing risk. This mini-symposium will highlight recent advances in the quantification and management of uncertainties that affect the behavior of dynamically sensitive structures. Topics will include, but are not limited to:

• Probabilistic and stochastic modeling techniques for representing environmental loads and their effects on structures, including approaches such as surrogate modeling, advanced Monte Carlo methods, and data-driven techniques
• Risk analysis and reliability assessment, covering methods for quantifying failure probabilities, evaluating resilience under extreme events, and integrating risk-informed decision-making into engineering practice
• Uncertainty quantification and propagation, exploring how epistemic and aleatory uncertainties affect structural response predictions and how computational methods can enhance their characterization
• Strategies for robust and adaptive structural design, focused on developing systems that maintain reliable performance under variable and unforeseen conditions
• Real-world applications and case studies demonstrating how uncertainty quantification and risk assessment methodologies are applied in practice

This mini-symposium aims to facilitate interdisciplinary exchange and highlight new methods for improving the performance of structures exposed to uncertain environmental conditions.

Ensuring reliable infrastructure performance under uncertainty is one of the central challenges in modern infrastructure asset management. Infrastructure asset management is fundamentally defined as a set of coordinated activities aimed at realizing value from assets by systematically guiding investment, operation, and maintenance decisions throughout the life cycle of complex, interdependent infrastructure systems. These systems, comprising networks of roads, bridges, water systems, and other essential assets, undergo deterioration processes that are uncertain, spatially and temporally heterogeneous, and often coupled through shared risks and external stressors. This mini-symposium focuses on uncertainty-informed approaches to infrastructure asset management, where both statistical deterioration modeling and (stochastic) control theory play a critical role. We invite contributions that address uncertainty in data, models, and system dynamics, arising, for example, from inspection errors, limited monitoring, environmental variability, or model misspecification. Topics of interest include, but are not limited to, probabilistic performance prediction, reliability-based decision frameworks, robust and adaptive maintenance strategies, risk-informed life-cycle optimization, and deep reinforcement learning methods. A particular emphasis is placed on methodological advances that enhance decision-making under deep uncertainty, and on integrative approaches that treat infrastructure asset portfolios as complex systems requiring coordinated, system-wide management. By bringing together researchers and practitioners from reliability engineering, infrastructure systems analysis, and decision science, this session aims to advance the theory and practice of infrastructure asset management toward more resilient and value-driven infrastructure development under uncertainty.

The rational assessment of risk and reliability in marine and offshore structures is of critical importance for various reasons—for example, to prevent catastrophic events, to protect the ocean environment, and to attain efficient maneuvering and operation. Taking the safety of a floating structure (e.g., a marine vessel or a floating offshore wind turbine) as an example, the floating structure in ocean are subjected to stochastic excitations from waves, currents, and wind, which are often complex and mutually interacting. Consequently, the response characteristics exhibit significant complexity on top of the structural nonlinearities, motion nonlinearities, material nonlinearities, and other contributing factors. Furthermore, uncertainties due to limited data or knowledge hinder us from perfectly characterizing the complex nonlinearities involved. Proper consideration of these aspects is also essential for the advancement of autonomous operation technologies.

This mini symposium aims to provide a platform for researchers working on the assessment of risk and reliability in maritime and ocean engineering. A wide range of applications can fall within the scope of this session, including but not limited to marine vessels, offshore platforms, autonomous surface vehicles, offshore wind turbines, and wave energy converters.

This session welcomes contributions related to, but not limited to, the following topics:

• Probabilistic estimation of extreme responses
• Uncertainty quantification
• Surrogate modeling of complex phenomena
• Advanced numerical techniques
• Reliability-based design approaches
• Data-driven and robust operations
• Autonomous systems
• Machine learning (ML) and artificial intelligence (AI) techniques
• Risk and reliability of AI/ML in safety-critical applications

Both theoretical developments and practical applications are highly encouraged.

Advancements in artificial intelligence (AI) are rapidly transforming the field of rock mechanics, offering unprecedented capabilities for characterizing the mechanical behavior, heterogeneity, and structural evolution of rock masses, especially in complex underground environments such as deep and water-rich tunnels. This Mini-Symposium aims to bring together leading researchers and practitioners to explore cutting-edge methodologies that integrate AI techniques—including deep learning, physics-informed neural networks, and Bayesian optimization—with traditional geo-mechanical modeling and field monitoring. Topics of interest include, but are not limited to: AI-based recognition of rock mass structures from multi-source data (e.g., 3D point clouds, acoustic signals, borehole imagery), intelligent prediction of mechanical parameters and deformation responses, real-time tunnel face characterization, and hybrid modeling frameworks that bridge data-driven inference and constitutive rock mechanics principles. Particular attention will be given to case studies and engineering applications that demonstrate the effectiveness and robustness of AI-augmented methods in enhancing design, risk evaluation, and decision-making in tunnel construction. By fostering interdisciplinary dialogue between AI scientists, geotechnical engineers, and underground space researchers, this Mini-Symposium seeks to catalyze a new wave of intelligent, adaptive approaches for addressing the geo-mechanical challenges of modern tunneling projects. Participants are encouraged to contribute both theoretical advances and field-validated innovations, as we collectively reimagine the future of intelligent geomechanics in subterranean environments.

Recent years have seen an emerging shift towards data-driven approaches for modeling complex and nonlinear systems, alongside an increased recognition of the associated uncertainties. This trend has been further accelerated by innovations in e.g., scientific machine learning and deep generative models. Besides, these data-driven techniques have the potential to foster new frontiers in forward and inverse uncertainty quantification, particularly with regard to high-dimensional and/or highly nonlinear problems.

This mini-symposium aims to bring together experts, researchers and practitioners from academia and industry to explore and discuss the latest developments, methodologies and applications related to data-driven modeling and uncertainty quantification for understanding complex and nonlinear systems. Topics for potential contributions include but are not limited to:

• Identification of (reduced order) models and governing equations from data.
• Fast and accurate inferences under uncertainties.
• Machine/deep learning-aided forward and inverse uncertainty analysis.
• Quantification of mixed aleatory and epistemic uncertainties.
• Efficient solutions for high-dimensional and/or highly nonlinear problems.
• Applications in the fields of structural engineering, geotechnical engineering, engineering seismology and others.

Geotechnical engineering practice is always involved with a considerable level of uncertainty. It is primarily attributed to (1) inherent randomness in geologic materials (aleatory uncertainty) and (2) imperfect knowledge of the subsurface conditions (insufficient data to build a three-dimensional ground model) and the soil-structure interaction (idealization of complicated soil behavior and its interaction with the structure) (epistemic uncertainty). In principle, aleatory uncertainty cannot be reduced, as naturally occurring geomaterials manifest itself on both spatial and temporal scales; however, it can be characterized more accurately by integrating the data and information from geologic survey, geophysical investigation and geotechnical (laboratory and in-situ) testing. Epistemic uncertainty can be considered to include (1) measurement error, (2) statistical uncertainty in the determination of soil and rock properties, and (3) model uncertainty in the calculation of soil behavior and its interaction with the structure. Better understanding and characterization of the uncertainties is critical for developing reliability-based geotechnical design procedure and risk management. National or benchmarking geo-test sites worldwide are an efficient tool to do this work. These sites represent a wide range of geologic materials and were investigated by a variety of methods. They provide a lot of subsurface data to characterize the spatial variability and statistical uncertainty in the estimation of soil properties. Furthermore, field instrumented tests have also been performed on full-scale geo-structures (e.g., embankment, shallow and deep foundations), which can be effectively used for the calibration of analysis and design methods, ranging from empirical/semi-empirical, theoretical to advanced numerical models. As such, this session aims to present benchmarking examples of using these sites to quantify aleatory and epistemic uncertainties in site characterization and geotechnical analysis, and ultimately, their effect on geotechnical decision.

Proving a quantitative reliability target is a challenging issue in both research and industrial practice. Different reliability requirements must be taken into account at various product levels, such as the system, assembly, or component level. Uncertainties must be considered in both test planning and test evaluation, which in turn affect planning parameters such as test duration and sample size. The consequences for the confidence level must also be examined. Failure-oriented tests in particular are often performed as accelerated tests, which represents another aspect of reliability testing.

Topics for potential contributions include but are not limited to:

• Reliability Demonstration Testing (zero failure testing)
• Failure Testing (e.g. bases on Weibull)
• Quantitative Accelerated Life Testing
• Testing on different system levels
• Consideration of prior knowledge (Bayes)
• Merging testing and simulation
• Uncertainties in Reliability Testing
• ...practical and theoretical investigations

High-fidelity simulations are central to tackling the complex, nonlinear, and uncertain behavior of geomaterials and geotechnical structures. However, conventional simulation-based models, while rigorous, are often computationally expensive-posing bottlenecks for uncertainty quantification, time-dependent reliability analysis, and optimization tasks. Recent advances in data-driven surrogate modeling are revolutionizing how geotechnical problems are approached. By learning the input-output mapping of a costly simulator or directly exploiting monitored data, surrogate modeling enables rapid uncertainty propagation, optimization, and adaptive control. Despite their promise, several critical challenges hinder the widespread adoption and robust application of surrogate models in practice. These include: the curation of training data, ensuring generalizability across diverse geological contexts, handling data scarcity, maintaining physical consistency and interpretability, and rigorous validation. This Mini-Symposium aims to bridge this gap by showcasing advancements in surrogate modeling techniques. The sessions will delve into innovative methodologies, practical implementations, and emerging trends, fostering interdisciplinary collaboration to accelerate their adoption in geotechnical engineering. Contributions are welcomed in areas including, but not limited to:

1. Development of advanced surrogate modeling techniques for geotechnical applications, including data-driven, statistical, and machine learning approaches
2. Physics-informed and hybrid surrogate models that integrate domain knowledge and physical constraints for enhanced prediction accuracy and interpretability
3. Strategies for surrogate model training and validation under practical constraints such as data scarcity, high dimensionality, and uncertainty
4. Efficient surrogate-based frameworks for computationally intensive geotechnical simulations and uncertainty quantification
5. Applications of surrogate models in geotechnical analysis, design, and decision-making processes
6. Integration of surrogate models with emerging technologies for real-time monitoring, digital twins, and intelligent geotechnical systems
7. Case studies, benchmarking, and practices for the implementation and validation of surrogate models in geotechnical engineering

Engineering structures are commonly vulnerable to catastrophic natural disasters, e.g., seismic motion and strong wind. Structural failure due to catastrophic disasters can cause numerous economic losses and casualties. With the increasing frequency and compounding effects of natural hazards, engineering structures are often exposed to multiple types of disastrous effects either simultaneously or sequentially. The multi-hazard disastrous effect modelling and dynamic reliability analysis are therefore of essential importance for the purpose of hazard mitigation. Nevertheless, high-fidelity and efficient techniques regarding this topic are still challenging. On the one hand, randomness inevitably involves both disastrous effects and structures. On the other hand, the nonlinear behavior of engineering structures (especially under the dynamic disastrous effect) makes the problem even more difficult. To this regard, this mini-symposium (MS) is devoted to reporting the recent advances and emerging approaches related to stochastic dynamics and reliability evaluations of complex engineering structures under single and multiple disastrous effects. Special attention is paid to the modelling, analysis, and reliability assessment of structures subjected to multi-hazard scenarios, including but not limited to earthquake-wind, earthquake-tsunami, and wind-rain interactions. This MS aims to foster interdisciplinary discussions and collaborations in the development of advanced computational methods, probabilistic models, and reliability-based design strategies against extreme events.

Topics for potential contributions include but are not limited to:

• Modelling of disastrous excitations, e.g., the modelling of seismic ground motions and strong wind;
• Pragmatic techniques for stochastic dynamics of complex nonlinear structures;
• Distinct failure mechanisms of complex engineering structures under disastrous effects;
• Dynamic reliability analysis under multi-hazard scenarios involving cascading or compounding effects;
• Dynamic reliability analysis of complex engineering structures;
• Dynamic reliability analysis involving multiple kinds of failure mechanisms;
• Data- and physics-informed approaches for multi-hazard reliability evaluation.

Maintenance of infrastructure is an important and urgent issue because there are many structures to be maintained. However, it is not easy to maintain all the structures in good condition due to many problems. Moreover, natural disasters, such as earthquakes, tsunamis, typhoons, heavy rain, and volcanic eruptions, have been becoming more frequent. Many infrastructures are at these risks. However, information collection, such as manual inspection and survey after a disaster, is extremely difficult, and the convenient Health Monitoring Systems are expensive. Therefore, innovative technologies for these problems are expected. New technologies have been developing using AI (Artificial Intelligence) and IoT, such as deep learning and UAVs. Deep learning can be promising because it can find important feature characteristics automatically. IoT sensing, such as low-cost MEMS, UAV, satellite images, cloud servers, and so on, can be applied widely for real-time monitoring and collecting big data for decision making in natural disasters. This special session aims to bring together academics and practitioners to discuss the future of AI/IoT technologies for maintenance and natural disaster prevention of infrastructures.

Modern engineering systems often exhibit complex dynamic behaviors characterized by strong nonlinearity, high dimensionality, and multi-scale interactions. These systems operate under multi-source uncertainties, including external excitations (e.g. wind, seismic, wave, operational loads, etc.), internal randomness in material properties, geometry, and boundary conditions, as well as epistemic uncertainties arising from incomplete knowledge. Under such conditions, their responses typically display pronounced stochastic dynamic characteristics, with intricate uncertainty propagation across both temporal and spatial scales. Accurately characterizing propagation mechanisms and quantifying these uncertainties in nonlinear, high-dimensional, and even fractional-order dynamic systems remains an open challenge, which is critical for advancing the safety, reliability, and resilience of next-generation engineering systems.

This Mini-Symposium will serve as an interdisciplinary forum for researchers and practitioners to discuss theoretical and methodological advances in stochastic dynamics. Topics for potential contributions include but are not limited to:

• Full probabilistic modeling of random fields and stochastic processes.
• Uncertainty propagation in high-dimensional nonlinear dynamical systems, including emerging challenges at the computational mechanics-stochastic dynamics interface under exascale simulation conditions.
• Model-order reduction techniques for efficient stochastic analysis in large-scale systems.
• Complex excitation and structural mechanisms, including fractional-order dynamics, Poisson and Lévy-type excitations, and multi-source stochastic inputs.
• Representation, quantification, and propagation of epistemic uncertainty in dynamical problems.
• Advanced theoretical and computational methods in stochastic dynamics, ranging from physics-driven and physics-enhanced/encoded approaches to data-driven and AI-aided techniques.
• Applications and innovative solutions for real-world engineering systems (e.g., critical infrastructure, energy systems, and advanced equipment).

Infrastructure systems today face unprecedented challenges over their life cycles, including aging, environmental degradation, and increasingly severe, climate-induced hazards. Climate change has become a dominant driver of risk, intensifying the frequency and magnitude of extreme events and introducing new patterns of uncertainty that traditional engineering approaches are ill-equipped to manage. Ensuring long-term performance under compounded climatic, environmental, and operational stressors is now a critical imperative. This Mini-Symposium (MS) offers a focused platform for advancing methods and applications in reliability analysis, probabilistic risk assessment, resilience quantification, and life-cycle performance management, with a particular emphasis on climate adaptation and sustainable infrastructure development.

The MS will explore advanced methodologies for managing climate-driven multi-hazard risks across infrastructure life-cycle phases, from design through construction, operation, maintenance, and decommissioning under evolving climatic conditions. Key themes encompass: (1) probabilistic life-cycle performance models integrating material deterioration, structural aging, and climate change impacts; (2) climate-resilient design frameworks addressing non-stationary hazards and future climate scenarios; (3) multi-hazard risk assessment for infrastructure networks under climate uncertainty; (4) data-driven approaches enabling real-time reliability monitoring and climate-adaptive predictive maintenance; (5) decision frameworks optimizing resource allocation in climate-conscious asset management; and (6) sustainable infrastructure solutions reconciling circular economy principles with reliability requirements amid environmental change.

This symposium convenes researchers, practitioners, and policymakers to disseminate cutting-edge findings, address critical climate-related infrastructure challenges, and define future research trajectories. We especially encourage submissions on methodological innovations, climate modeling integration, adaptation techniques, case studies, and interdisciplinary approaches—particularly for structural, transportation, and coastal infrastructure exposed to climate impacts.

The increasing frequency and intensity of natural hazards under climate change, including earthquakes, high winds, and floods, pose unprecedented threats to main engineering structures and critical infrastructure. This reality calls for a paradigm shift in structural engineering, moving beyond conventional design objectives such as life-cycle safety and collapse prevention, toward the emphasis on resilience that encompasses the ability to anticipate, absorb, adapt to, and rapidly recover from disruptive events.

Advanced structural control employs innovative technologies, such as base isolation systems, active and semi-active control systems, to enhance the resilience and safety of engineering structures and infrastructure. These systems effectively mitigate dynamic responses caused by earthquakes, winds, and other natural hazards, reducing damage and risk as well as ensuring functionality post-event. Integration with smart materials and real-time monitoring represents the next frontier in adaptive structural control.

This mini-symposium is dedicated to exploring the transformative role of next-generation structural control systems in achieving these goals. It will showcase cutting-edge research and innovative applications that are shaping the future of adaptive and intelligent structures and infrastructure. Discussions will focus on control technologies that not only mitigate dynamic structural responses but also enhance structural robustness, support multi-hazard risk assessment and control strategies, and ensure rapid functional recovery while minimizing economic losses in the aftermath of extreme events.

The scope of the mini-symposium is broad, encompassing the following topics, though not limited to:

• Innovative control devices and materials
• AI-driven and adaptive control algorithms
• Reliability-based design optimization methods
• Resilience-informed decision-making frameworks
• Hybrid simulation and real-time testing
• Cyber-physical systems and digital twins
• Field applications and case studies

Probability concepts have long been applied in wind-resistant design of structures. Current design codes and standards take its basis in the formulation of wind load modeling with such probability concepts. Performance-based wind-resistant design has recently attracted more attention, and extensive research has been devoted to establishing and sophisticating the design procedure. Probability, reliability, and risk concepts have also been applied to decision making in the context of wind utilization as well as wind disaster mitigation. Moreover, the probability concept has also been applied to wind environmental assessments; however, it is recently used to better understand the statistical features of the wind environment in urban areas. Whereas all of these research work should fall in the category of wind engineering, communications within the wind engineering communities over different applications seem not to be enough; typically research work is divided according to the target wind speed ranges; lower wind speed for wind environment, moderate wind speed for wind energy, higher wind speed for structural design and disaster mitigation.

Motivated by this, this mini symposium offers the opportunity for researchers in different “segments” of wind engineering, working on the applications of probability, reliability and risk concepts, to gather, present ongoing studies, and share their recent advancements. Thereby, the mini symposium facilitates homogenizing the use of probability, reliability and risk concepts for unified understanding of wind and its responses in society.

Engineering systems in civil, mechanical, aerospace, energy, and infrastructure domains operate under uncertain, evolving, and often harsh environments. Their safety and performance are not static but evolve over time, influenced by stochastic loads, progressive material degradation, operational variability, and unforeseen hazards. Traditional reliability methods, while powerful, are often limited by their off-line and static nature, making them insufficient for capturing the continuously changing reliability state of complex systems.

Recent advances in sensing technologies, digital twins, and intelligent monitoring platforms are enabling real-time data collection and system state awareness. This creates new opportunities for real-time reliability updating, where prior probabilistic models are dynamically updated with in-situ data to provide accurate, timely assessments of system safety and performance. Approaches such as filter–based methods, stochastic simulation methods, dynamic Bayesian networks, surrogate-assisted inference, and machine learning–driven updating strategies are becoming essential tools to integrate monitoring data with physics-based models.

This Mini-Symposium will provide a forum to discuss state-of-the-art methods, computational strategies, and applications of real-time reliability updating. Contributions are welcome in areas including, but not limited to:

1. Methodological advances in real-time reliability updating, including filter–based methods, stochastic simulation methods, and dynamic Bayesian networks.
2. Surrogate modeling and machine learning approaches for efficient online reliability updating.
3. Integration of monitoring data, sensor networks, and digital twin platforms into reliability assessment frameworks.
4. Addressing key challenges such as data scarcity, sensor noise, incomplete observations, and model bias in real-time updating.
5. Reliability-informed decision-making for structural health monitoring, predictive maintenance, and risk management.
6. Applications in civil, mechanical, aerospace, offshore, and energy systems under time-varying and uncertain environments.
7. Case studies, benchmarking, and best practices for implementing real-time reliability updating in engineering decision-making.

Life-cycle performance of civil structures is affected by multiple hazards including sudden disasters such as earthquakes and floods, but also by aging deterioration over time. There are uncertainties in estimating the extent of damage and deterioration based on inspection and monitoring, as well as in the structural performance assessment of damaged, deteriorated or repaired components. Therefore, probabilistic approaches which consider various uncertainties in the assessment are essential.

This mini-symposium aims to bring together experts, researchers and practitioners to explore and discuss the latest developments, methodologies and applications related to the life-cycle performance of civil structures under multiple hazards. Topics will include, but are not limited to:

• Hazard assessment
• Time-dependent reliability analysis
• Decision-making based on inspection and monitoring
• Machine learning and artificial intelligence techniques
• Performance assessment of repaired members
• Resilient structures and infrastructures

In the face of increasing frequency and severity of natural hazards driven by climate change and rapid urban development, advanced modeling and simulation techniques are essential to support hazard assessment, reliability analysis, and risk-informed decision making and disaster mitigation. In recent years, machine learning (ML) has emerged as a powerful tool in the reliability engineering and natural hazards community, offering data-driven insights, enhanced predictive capabilities, and improved computational efficiency. This mini-symposium aims to bring together researchers at the forefront of ML and natural hazard studies to exchange ideas, showcase state-of-the-art applications, and discuss emerging challenges.

We invite submissions on a wide range of topics related to ML applications in natural hazard modeling, simulation, and assessment, including but not limited to:

• ML-enhanced and Physics-Informed hazard modeling: Integration of ML with physics-based models for improved simulation of earthquakes, tropical cyclones, floods, wildfires, and other natural hazards.
• Big data integration in hazard science: Application of ML to extract insights from large-scale datasets such as satellite imagery, climate reanalysis products, sensor networks, post-hazard damages, and hazard databases.
• Uncertainty quantification and surrogate modeling: Application of ML techniques (e.g., GPR, ANN, Transformer, VAE, GAN, and Diffusionx) for probabilistic hazard analysis, model surrogates, and high-dimensional sensitivity analysis.
• Multi-hazard risk assessment and decision support: Use of ML for joint hazard modeling, damage and loss estimation, and supporting resilient infrastructure planning and risk mitigation under climate change.

The assessment and optimization of engineering structures and infrastructure systems under uncertainty are critical to ensuring safety, functionality, and cost-effectiveness over their entire life cycle. This minisymposium will bring together researchers and practitioners to present and discuss recent advances in the theory, methodology, and applications of reliability assessment and reliability-based optimization. Topics of interest include advanced reliability analysis methods, reliability-based design and maintenance optimization, and the integration of surrogate modelling and machine learning techniques to improve computational efficiency and decision-making under uncertainty.

Applications span a wide range of civil engineering structures, transportation systems, and energy infrastructure, including but not limited to bridges, onshore/offshore wind turbines, and other critical assets. Contributions may address all phases of the life cycle—from conceptual design and construction, through operation and inspection, to maintenance planning and end-of-life decision-making. The minisymposium aims to foster cross-disciplinary exchange between experts in structural engineering, reliability theory, computational mechanics, optimization, and data science.

Stochastic finite element method and surrogate model, which establish the relationship between random output and random input by using explicit functions, have been widely used in many aspects such as uncertainty quantification, reliability analysis, global sensitivity analysis and model updating et al. At present, various stochastic finite element methods and surrogate models have been developed. As typical representatives among these methods, the perturbation method, intrusive/non-intrusive polynomial chaos expansions, Kriging method and Gaussian processes have been successfully applied to solve the stochastic problems in extensive industrial fields. However, highly accurate and efficient stochastic finite element methods and sample-efficient surrogate models are still developing to implement the stochastic analysis of large-scale structural models. This research direction has been attracting many researchers’ attention.

Model updating is a key content in structural health monitoring. When considering the randomness of structural modelling and measurement data, the stochastic model updating becomes unavoidable. To solve stochastic model updating problems, the stochastic finite element methods and surrogate models can play a significant role instead of the usually used Bayesian methods. This direction is a hot topic in current structural model updating, which has a very good application prospect in practical structural health monitoring.

The symposium aspires to create a conducive environment for scholars, researchers, and practitioners to exchange insights and findings, fostering collaborative endeavors and innovations in the domain of stochastic finite element methods and surrogate models. The inclusion of diverse theoretical explorations and practical applications is anticipated to provide a comprehensive perspective on the evolving role of stochastic finite element methods and surrogate models to address the inverse problem in health monitoring.

Topics for potential contributions include but are not limited to:

• (1) Recent advances in stochastic finite element methods and surrogate models.
• (2) Advanced computational techniques in stochastic static and dynamics.
• (3) Recent mathematical and numerical developments in stochastic structural engineering static and dynamics.
• (4) New surrogate modeling techniques tailored to some computationally-demanding problems.
• (5) Stochastic finite element method and surrogate models in reliability analysis, global sensitivity analysis.
• (6) Stochastic model updating.
• (7) Model update with structural health monitoring.
• (8) Engineering practice of structural health monitoring accommodating uncertainties.
• (9) Novel uncertainty quantification techniques in model updating.
• (10) Data-driven approaches for model updating.

Various sources of risks, such as the deterioration effects, weather- or geo-related hazards, and manmade shocks, challenge the safety and serviceability of critical structures and infrastructure in uncertain environments. It is essential to accurately assess and properly manage these risks throughout the design, construction, and operation stages. High-fidelity numerical simulations provide a solid foundation for evaluating and understanding the performance of engineering systems. Meanwhile, machine learning and surrogate modeling have gained significant attention as powerful tools for addressing the ever-increasing computational complexity. In recent years, a variety of advanced methods have emerged, including those grounded in physics-informed learning, operator learning, transfer learning, and multi-fidelity learning. These approaches provide promising avenues for more efficient and effective reliability and risk assessments, therefore informing and facilitating the subsequent decision making.

The proposed session will encompass various areas associated with reliability and risk analysis. Areas of interest include but are not limited to high-fidelity simulation methods for evaluating and predicting structural performance; physics-informed and data-driven machine learning for reliability and risk assessments; intelligent and digital methods in infrastructure management; statistical inference integrating physics data and scientific computing; surrogate modeling for structural health monitoring and digital twin.

Contributions that apply new methods and strategies to realistic engineering problems or present a clear pathway for their application are highly encouraged.

Structural reliability analysis offers a critical tool for evaluating the safety, serviceability and durability of engineering structures under various uncertainties arising from, such as geometrical dimensions, material properties and external loads. Despite great progress, traditional reliability methods often suffer from limited accuracy or computational inefficiency when applied to practical engineering problems. With the rapid development of artificial intelligence (AI) techniques, new avenues are emerging for tackling these challenges. This mini-symposium aims to bring together recent advances in AI-empowered methods for addressing computational challenges in structural reliability analysis. Possible topics of interest include (but are not limited to):

• AI-assisted uncertainty modelling;
• (Bayesian) active learning reliability methods;
• Neural networks-based methods for reliability analysis;
• Physics-informed methods for reliability analysis;
• Gradient-enhanced methods for reliability analysis;
• Dimension-reudction methods for high-dimensional reliability analysis;
• Novel applications to challenging engineering problems.

Mathematical models of complex physical systems are highly computationally demanding and are usually affected by various sources of uncertainty. Quantifying the uncertainty of such systems further increases the computational cost of the analysis, and therefore it is often necessary to develop suitable and computationally efficient surrogate models. Several surrogate modeling methods are available, with popular ones including Gaussian process regression, support vector machines, polynomial chaos expansions, and artificial neural networks.

Unfortunately, the accuracy of standard data-driven surrogates is highly dependent on the quality and size of the training data. Recently, there has been significant interest in methods that incorporate additional constraints, typically known physical principles, into the training process to ensure realistic and physically meaningful surrogate model behavior despite limited training data sets. Surrogate models capable of satisfying physical constraints, broadly referred to as physics-informed machine learning methods, offer superior accuracy and numerical efficiency, and represent powerful tools for uncertainty quantification and reliability analysis of physical systems.

This mini-symposium aims to highlight recent advances in physics-informed machine learning, including but not limited to:

1. Novel algorithms for physics-informed machine learning (PiML)
2. Applications of PiML for uncertainty quantification and reliability analysis
3. Bayesian inference and PiML for parameter identification
4. Statistical sampling for optimal PiML training, including adaptive algorithms
5. Algorithms for rare-event estimation combined with PiML

Modern engineering systems are increasingly complex and are operating in unpredictable environments. This complexity introduces multiple types of uncertainties, often referred to as polymorphic uncertainties, which are characterized by the co-existence of both aleatory (inherent, probabilistic) and epistemic (knowledge-deficient, non-probabilistic) uncertainties. Traditional uncertainty quantification (UQ) and design optimization methods, which often treat all uncertainties as purely aleatory, can become inadequate or even misleading under such conditions.

This Mini-Symposia aims to explore the latest theoretical advances and computational methodologies for effectively quantifying and propagating these mixed uncertainties. We will focus on robust and efficient frameworks that go beyond pure probabilistic analysis, incorporating techniques from evidence theory, possibility theory, interval analysis, and imprecise probabilities to create hybrid models. A key challenge is the computational cost associated with these advanced UQ methods, particularly when embedded within an optimization loop. Therefore, this Mini-Symposia will also highlight novel strategies for design optimization under polymorphic uncertainty. Topics of interest include, but are not limited to, reliability analysis, sensitivity analysis, multiphysics model validation and updating, reliability-based design optimization (RBDO) and uncertainty-based multidisciplinary design optimization (UMDO) under polymorphic uncertainty, especially in high-dimensional spaces.

We invite contributions that present innovative algorithms, demonstrate applications in real-world engineering problems (e.g., aerospace, automotive, energy systems), or address the theoretical foundations for reasoning with and optimizing under imperfect knowledge. The goal of this Mini-Symposia is to foster discussion and collaboration on moving the field beyond traditional paradigms towards more comprehensive and credible design under uncertainty.

Modern engineering systems are increasingly complex and are operating in unpredictable environments. This complexity introduces multiple types of uncertainties, often referred to as polymorphic uncertainties, which are characterized by the co-existence of both aleatory (inherent, probabilistic) and epistemic (knowledge-deficient, non-probabilistic) uncertainties. Traditional uncertainty quantification (UQ) and design optimization methods, which often treat all uncertainties as purely aleatory, can become inadequate or even misleading under such conditions.

This Mini-Symposia aims to explore the latest theoretical advances and computational methodologies for effectively quantifying and propagating these mixed uncertainties. We will focus on robust and efficient frameworks that go beyond pure probabilistic analysis, incorporating techniques from evidence theory, possibility theory, interval analysis, and imprecise probabilities to create hybrid models. A key challenge is the computational cost associated with these advanced UQ methods, particularly when embedded within an optimization loop. Therefore, this Mini-Symposia will also highlight novel strategies for design optimization under polymorphic uncertainty. Topics of interest include, but are not limited to, reliability analysis, sensitivity analysis, multiphysics model validation and updating, reliability-based design optimization (RBDO) and uncertainty-based multidisciplinary design optimization (UMDO) under polymorphic uncertainty, especially in high-dimensional spaces.

We invite contributions that present innovative algorithms, demonstrate applications in real-world engineering problems (e.g., aerospace, automotive, energy systems), or address the theoretical foundations for reasoning with and optimizing under imperfect knowledge. The goal of this Mini-Symposia is to foster discussion and collaboration on moving the field beyond traditional paradigms towards more comprehensive and credible design under uncertainty.

Structural collapse induced by extreme events, such as earthquakes, explosions, vehicular impacts, fires, and terrorist attacks, can lead to catastrophic loss of life and severe property damage. Consequently, to prevent structures from collapse is the bottom-line engineering demanding, and ensuring structural safety through adequate design is paramount to mitigate collapse risk, but still of great challenge. Within the framework of performance-based design and optimal decision-making, a quantitative assessment of structural safety against collapse is essential. This entails the integrated evaluation of structural fragility, reliability, robustness, risk, and resilience enhancement.

This mini-symposium aims to bring together researchers, academics, and practicing engineers to present and discuss advances in theoretical frameworks, experimental investigations, numerical modeling techniques, and practical design methodologies for collapse prevention in both new and existing structures under uncertainties. The scope is broad and inclusive, welcoming contributions on (but not limited to) the following topics:

1. Uncertainty quantification and propagation in complex engineering structures
2. Probabilistic hazard analysis of various accidental and extreme loads
3. Component- and system-level reliability analysis in terms of collapse prevention
4. Fragility, vulnerability and risk assessment of structures against collapse
5. Alternate load path analysis against progressive collapse
6. Structural robustness quantification
7. Aging and deterioration on structural safety against collapse under uncertainty
8. Resilience analysis of building structures against collapse
9. Reliability-, risk-, or resilience-based design optimization in terms of collapse prevention

Both theoretical developments and applications involving different structural systems are particularly welcomed in this session.

The rational design, performance evaluation, and long-term reliability of engineering systems, which are key to supporting national core strategic needs such as high-end equipment manufacturing, major infrastructure construction, and aerospace development, are fundamentally underpinned by the mechanical behavior of their constituent materials and structures. In real-world operations, however, these behaviors are invariably influenced by diverse uncertainties: randomness in material microstructures and constitutive properties, stochastic dynamic loads (e.g., wind gusts, seismic waves, cyclic fatigue), geometric deviations from design specs, and time-varying environmental factors (e.g., temperature fluctuations, corrosion). Conventional deterministic mechanics approaches, relying on fixed parameters and idealized conditions, often fail to account for the propagation of such uncertainties in mechanical systems, leading to either overly conservative, resource-wasting designs or underestimated risks that compromise structural safety.

This mini-symposium highlights recent advances in stochastic mechanics, with the goal of bridging theoretical innovation and engineering practice. It emphasizes the development of novel modeling frameworks, efficient computational methods, and experimentally validated approaches for quantifying, propagating, and mitigating uncertainty in mechanical systems. Such advancements are essential for enhancing the accuracy of structural response prediction, reliability-based design, and risk assessment, as well as for establishing a rigorous scientific foundation for decision-making in complex engineering applications. Contributions are invited on the following and related topics:

1. Theoretical frameworks in stochastic mechanics (e.g., probabilistic continuum mechanics, stochastic damage mechanics, and random vibration theory);
2. Stochastic constitutive modeling of materials (e.g., random elasticity, viscoelasticity, and plasticity in metals, composites, and geomaterials);
3. Computational methods in stochastic mechanics (e.g., stochastic finite element methods, Monte Carlo simulation and variance reduction techniques, polynomial chaos expansion);
4. Multi-scale stochastic mechanics (e.g., linking microstructural randomness to macroscopic mechanical behavior);
5. Stochastic dynamics and vibration control (e.g., structural response under stochastic excitation, robust vibration suppression);
6. Multi-physics coupling under uncertainty (e.g., thermo-mechanical, hydro-mechanical, and electro-mechanical systems with stochastic parameters);
7. Data-driven and machine learning approaches in stochastic mechanics (e.g., surrogate modeling for stochastic response prediction, AI-enhanced uncertainty quantification);
8. Extreme event analysis (e.g., rare-event simulation for structural failure under impacts or earthquakes);
9. Model validation, calibration, and uncertainty quantification for stochastic mechanical models;
10. Stochastic optimization for reliability-based design of mechanical systems and structures;
11. Engineering applications and case studies, including: reliability assessment of mechanical components (aircraft structures, gearboxes, turbines); stochastic mechanical behavior of energy systems (wind turbine blades, nuclear reactor components);
12. Uncertainty-aware damage prognosis and remaining useful life prediction of mechanical structures.

We strongly encourage and warmly welcome contributions that present pioneering theoretical advances, innovative computational methodologies, and practical engineering applications in stochastic mechanics. Submissions from fields such as engineering mechanics, aerospace engineering, civil engineering, materials science, and related disciplines are highly welcomed.

Large language models (LLMs) offer a unique capability to integrate diverse data types—ranging from text-based reports and guidelines to visual imagery and sensor outputs—thereby providing a comprehensive perspective on engineering risk. For example, retrieval-augmented generation can improve the interpretation of risk-related standards, while back-analysis of engineering failures benefits from LLMs’ ability to process ambiguous, multimodal data. Through conversational interfaces, LLMs also make reliability insights more accessible, reducing barriers to probabilistic risk assessment and complementing established model-based approaches.

Agentic AI frameworks extend these capabilities by enabling LLMs to operate as autonomous agents. Such agents can proactively analyze real-time monitoring data, update reliability assessments, and support adaptive risk management strategies. They can interact with heterogeneous information sources to enable autonomous reliability design, assign dynamic risk grades to engineering structures, and provide decision support in safety-critical applications.

This mini-symposium invites contributions on theoretical developments, methodological advances, and applied case studies that demonstrate the role of LLMs and agentic AI in engineering reliability and risk management. By fostering technical exchange across disciplines, the symposium aims to define pathways toward more resilient, adaptive, and data-driven engineering systems.

This session explores the growing role of machine learning (ML) and artificial intelligence (AI) in geotechnical engineering.
Topics include data-driven site characterization, predictive modeling, real-time monitoring, observational method and decision support for design and construction.
Contributions demonstrating practical applications, methodological advances, and case studies that highlight the value of data in decision-making are especially welcome.
Discussions on AI governance in geotechnical practice, including transparency and explainability of ML/AI methods, benchmarking examples and guidance frameworks are also welcome.