List of Minisymposia
Plenary and Semi Plenary lectures will be complemented by Minisymposia organized by recognized experts in targeted research areas and related to all the important topics of the conference.
Each Minisymposium is expected to consist of at least one 2-hour session (6 presentations of 20 minutes each). The number of sessions of each MS will be determined by the multiples of six papers submitted.
In each MS a Keynote lecture is allowed every two full sessions of the MS, where a keynote presentation normally comprises two presentation slots. This means you should have at least 10 confirmed presentations to schedule a KL.
The above scheduling includes time for questions and discussion.
Participants interested in organizing a Minisymposium as part of DTE 2027 Conference are invited to send an email to dte_sec@cimne.upc.edu
The list of confirmed Minisymposia follows:
Data-driven modelling of computation of materials, solids and fluids
Recent years have seen a rapid growth in digital twin technologies that integrate simulation, data, and remote sensing to enable predictive, adaptive, and real-time decision-making for complex engineered systems. At the same time, there has been an explosion in the development of data-driven models that can augment or replace expensive high-fidelity simulations, uncover hidden physics, and enable rapid updates as new data become available. These trends motivate the development of hybrid digital twins that couple conventional physics-based models with data-driven components. Such approaches can provide a mechanism to rigorously integrate data-driven models and methods into modeling and simulation toolchains, as well as improve the trustworthiness of data-driven modeling in science and engineering.
Achieving hybrid couplings in a robust and reliable manner raises significant challenges, including consistency across heterogeneous model fidelities, stability of coupled workflows, treatment of nonconforming discretizations, and preservation of key physical structures such as conservation laws, symmetries, and stability properties. This minisymposium will highlight recent advances in methodologies for creating hybrid couplings involving data-driven and conventional models within digital twin frameworks. We welcome submissions featuring a wide range of data-driven approaches, including but not limited to projection-based, operator inference, dynamic mode decomposition and neural network-based reduced order model techniques. Of particular interest are both overlapping and non-overlapping domain decomposition strategies that assign different model types to different regions or physics. Topics of interest include: multi-scale and multi-physics hybrid models, structure-preserving coupling methods, adaptive model switching, optimization-based coupling, iterative schemes such as Schwarz-based methods for information exchange between heterogeneous components, and software frameworks that support heterogeneous model integration.
Keywords:
computational mechanics, data-driven modeling, digital twins, physics-informed AI, Uncertainty Quantification
Novel materials can serve as sensors or actuators within the closed feedback loop of digital-twin technologies. Hence, desired passive material properties, such as elastic modulus, Poisson’s ratio, mass density, electric permittivity, magnetic permeability, thermal conductivity, heat capacity, and others, are critical to the successful deployment of a digital-twin service. Furthermore, active materials may provide actuation as the digital-twin control demands. Thermodynamic bounds must be satisfied in the passive materials. However, when materials are allowed to have internal microstructures and processes, they may exhibit negative characteristics, such as negative stiffness due to postbuckling processes or in Landau phase transitions. Several physical considerations, such as violation of conservation laws in the non-Hermitian, non-reciprocal systems, or odd elasticity, are adopted to examine their stability or exceptional points, where eigenvalues change from real to complex numbers as internal variables are suitably tuned. Possible applications of materials with negative characteristics are abundant, e.g. vibration mitigation, noise reduction, cloaking in electromagnetic/acoustic/elastic waves, or unbounded effective material properties in composites. In this minisymposium, all aspects, numerical, theoretical, and experimental viewpoints in understanding such materials are welcome, including but not limited to machine learning techniques to generate internal microstructures or mechanisms, and novel analytical or computational methods in calculating physical properties for materials in solid, liquid, or other states. Experimental or theoretical studies to correlate numerical results are also welcome.
Keywords:
non-Hermitian, non-reciprocal, passive materials, sensors, active materials
This minisymposium provides a platform to discuss recent advances in machine learning and data-driven methodologies for material modeling, design, and manufacturing. With the rapid growth of artificial intelligence and high-performance computing, these approaches are transforming traditional paradigms in computational mechanics, materials science, and manufacturing engineering. The minisymposium brings together researchers developing and applying data-driven techniques to address complex multiscale and multiphysics problems. Key directions include the integration of physics-informed machine learning with continuum and discrete models, the development of surrogate and reduced-order models for efficient simulation, and data-driven discovery of constitutive laws and governing equations. Contributions leveraging experimental and industrial data for predictive modeling, digital twins, and real-time decision-making are particularly encouraged. Topics include, but are not limited to:
• Machine learning-based surrogate models and foundation models for materials and manufacturing
• Physics-informed and physics-guided machine learning for solid and structural mechanics
• LLM and AI-agentic application for material modeling, design, and manufacturing
• Data-driven computational mechanics without explicit constitutive laws
• Discovery of constitutive relations and governing equations from data
• Data-driven modeling of heterogeneous, anisotropic, and multiscale materials
• Machine learning for inverse problems and parameter identification
• Classical numerical method hybrid with AI/ML
• Reduced-order modeling and real-time simulation
• Integration of experimental data and AI for material characterization
• Digital twins and smart manufacturing systems
• Additive manufacturing and process optimization using data-driven approaches
• Semiconductor manufacturing and micro/nano-scale material modeling
• Uncertainty quantification, probabilistic modeling, and robustness analysis
• Interpretable and explainable AI for engineering applications
Applications may span solid/fluid mechanics, additive manufacturing, semiconductor processes, biomechanics, and other advanced material systems. Overall, this minisymposium aims to foster interdisciplinary dialogue and collaboration, highlighting the potential of machine learning and data-driven approaches to complement
Keywords:
composite manufacturing, computational mechanics, data-driven modeling, digital twins, high-fidelity simulations, hybrid modeling, inverse design, materials and structures, multiscale modeling, physics-informed AI, scientific machine learning
Computational fluid dynamics has become an essential tool for understanding, predicting, and controlling complex flow phenomena in science and engineering. This minisymposium covers state-of-the-art computational and data-based approaches for flow prediction and flow control, including direct numerical simulation, large-eddy simulation, reduced-order modeling, data analysis, data assimilation, system identification, and control-oriented modeling. Topics of interest include, but are not limited to, drag reduction, enhancement of mixing, heat and mass transfer control, multiphysics flow problems, turbulent flow prediction, coherent-structure analysis, optimal and feedback control, and applications to canonical, environmental, biomedical, and industrial flows.
The aim of this minisymposium is to provide a broad forum for researchers working on experimental and computational fluid dynamics, flow control, and data-based analysis to exchange ideas on current progress and future developments in this field. Particular emphasis is placed on discussions that connect physical understanding, predictive modeling, and control-oriented applications. Contributions from students and young researchers are especially welcome, and the minisymposium is intended to encourage active discussion across different methodologies, flow configurations, and application areas.
Keywords:
Data-driven modeling
Digital twins are emerging as a key paradigm in materials and structural engineering, enabling real-time prediction, monitoring, and optimization of complex physical systems. A central challenge is the development of surrogate models that are both computationally efficient and physically consistent while capturing high-dimensional and nonlinear behavior. Neural operators have recently gained attention as a powerful framework for learning mappings between function spaces and approximating solution operators of governing equations, making them well-suited for digital twin applications.
Despite their promise, several challenges remain, including handling multi-input parametric variability, extending to multiphysics settings, and ensuring temporal consistency through causality-aware formulations. Additional issues include spectral bias, limited data efficiency, and the need to systematically incorporate physical constraints.
This minisymposium invites contributions addressing these challenges through methodological advances and engineering applications. While the primary focus is on neural operators, contributions based on other deep learning approaches are also strongly encouraged to foster a broader discussion on the capabilities and limitations of different methodologies for digital twin modeling. Of particular interest are physics-informed and hybrid approaches that combine learning-based models with classical numerical methods.
Overall, the goal is to advance data-driven modeling strategies as enabling technologies for robust, efficient, and physics-consistent digital twins in materials and structural engineering.
Keywords:
digital twins, materials and structures, neural operators, Physics-informed machine learning
This minisymposium (MS) focuses on the integration of physics-based modeling and data-driven methodologies to develop digital twins for granular materials across scales—from particle-level interactions to industrial and geotechnical systems. Granular media pose unique challenges due to their discrete nature, complex flow behavior, and sensitivity to initial and boundary conditions. Traditional simulations (e.g., DEM, MPM) offer high fidelity but remain computationally intensive, while purely data-driven approaches risk violating physical constraints.
The MS objectives are: (i) to present recent advances in hybrid frameworks that embed physical information (e.g., conservation laws, contact mechanics, pore collapse models) into machine learning architectures (e.g., temporal graph neural networks, operator learning, encoder-decoder models); (ii) to explore digital twin implementations for real-time prediction, material tracking, and process optimization in mining, rotary kilns, blast furnaces, and warehouse logistics; and (iii) to discuss methods for uncertainty quantification, model reduction, and validation using X-ray microtomography (CT), 3D particle reconstruction, and in-situ experiments.
Key topics of this MS include: CT-based digital twins, physics-constrained deep learning simulators, surrogate modeling for multiscale granular flows, digital twins for asteroid regolith and space manufacturing, carbon-aware warehouse management, and the role of additive manufacturing and large vision models in generating high-fidelity grain morphologies. The MS aims to bridge computational mechanics, materials science, and industrial engineering, fostering discussion on scalable, trustworthy digital twins that respect physical laws while leveraging real-time data.
Keywords:
Data-driven modeling, Digital image correlation, digital twins, granular flow, granular materials; , Physics-informed machine learning, sensors, X-ray computed tomography
This mini-symposium focuses on recent advances in AI-enhanced computational mechanics for fracture analysis and material modeling for material strength. Topics of interest include fracture and crack propagation prediction, material constitutive modeling for damage and related phenomena, and the advancement of these approaches using AI technologies. Both fundamental and applied studies are invited to foster knowledge exchange and to promote the development of novel AI-enhanced computational frameworks.
Keywords:
fracture analysis, material modeling, material strength
In recent years, machine learning approaches have been widely used in many engineering fields, thus leading to complex behavior. Vehicle development and estimation require structural stiffness, material behavior, design optimization, occupant injury, and more. This mini symposium is dedicated to a broad view of vehicle performance using machine learning techniques.
Numerical simulations are used to model various mechanical behaviors, including vehicle structural analysis, material behavior, crash deformation, injury estimation, and design optimization. Simulations are generally based on the finite element method, and the target object is discretized into a mesh. Moreover, the simulations deal with different scales, including the full vehicle, components, and test pieces. Hence, the challenges of numerical simulations for predicting such phenomena are complex and involve mechanics and numerical techniques.
Data-driven modeling is powerful for reducing computational costs in various situations using surrogate models. Preparing numerical results and selecting appropriate machine learning techniques are necessary for constructing models. In addition, generative AI is a novel method of broader engineering assistance and automation.
Therefore, the main topic is machine learning techniques that efficiently and accurately handle conventional numerical results. Subjects include the structural deformation, occupant injury, network model construction, data argumentation, and data generation. Finally, we hope for submissions that provide solutions to improve computational accuracy and efficiency, reduce training time, and handle complex structural behavior.
Keywords:
Data-driven modeling, Machine learning, structures
Digital twins are emerging as powerful tools for real-time simulation, monitoring, and predictive control of oceanic and atmospheric systems. Creating effective digital twins for such complex environments requires computational efficiency without sacrificing interpretability and accuracy. This mini-symposium focuses on advanced Reduced Order Models (ROMs) and hybrid, physics-informed machine learning approaches, offering real- time performance while preserving the critical physical characteristics of environmental processes.
Oceanic and atmospheric phenomena, characterized by multiscale and multiphysics interactions, pose substantial computational challenges, particularly for high-resolution and real-time applications. Hybrid approaches leveraging physics-informed neural networks (PINNs), dynamic mode decomposition (DMD), proper orthogonal decomposition (POD), and generative AI techniques bridge the gap between computational feasibility and physical fidelity. These methods embed domain-specific knowledge into data-driven frameworks, enhancing reliability, interpretability, and predictive capabilities of digital twins.
Target applications of interest include digital twins for offshore energy systems (e.g., wind farms, wave energy converters, oil/gas platforms), aerosol-cloud interactions in climate modeling, ocean-atmosphere coupling dynamics (e.g., hurricane forecasting, air-sea interactions), and atmospheric pollution transport. Contributions highlighting advancements in multi-fidelity modeling, data assimilation techniques, uncertainty quantification, and computational efficiency improvements are particularly encouraged.
Keywords:
digital twins, model order reduction, Physics-informed machine learning
Bio … Medical, …
Digital twin technology is emerging as a unifying paradigm across engineering and biomedical domains, enabling virtual representations of complex systems. In healthcare, digital twins extend core engineering principles—such as multiphysics modeling, system identification, and control—toward patient-specific applications, including disease progression modelling, therapy optimization, treatment planning and optimization of clinical decision-making.
Minisymposia will highlight how methodologies traditionally developed in computational engineering—such as high performance computing, model order reduction, physics informed AI and hybrid AI–physics modeling—are being translated and adapted to biomedical and healthcare applications. Digital twins integrate multimodal data sources such as imaging, multi-omics, wearable sensors, and electronic health records. Multiphysics modeling plays a central role in advancing digital twins for biomedical applications by enabling the integration of interacting physical, chemical, and biological processes within a unified computational framework. Advances in high performance computing allow treatment planning, including real time simulations.
In silico modeling is rapidly reshaping drug discovery by enabling the construction of high-fidelity digital representations of biological systems and disease processes. Within a digital twins framework, patient-specific or population-level virtual models can integrate multi-omics data, pharmacokinetics/pharmacodynamics (PK/PD), and mechanistic pathway simulations to predict therapeutic response and optimize candidate selection.
The session will feature contributions spanning multiple scales, from cellular and organ-level simulations (e.g., brain, liver, cardiovascular, oncology, and infectious disease models), applications of high performance computing and AI for treatment plannning. Application of in silico models for drug discovery will be also discussed.
Keywords:
high-performance computing, in silico modeling, Physics-informed machine learning, treatment planning
Research has shown that compression and mechanical strain lead to tissue damage and can be used to evaluate injury risk. Finite element modeling has made significant progress in mechanobiological modeling. Consequently, strain mechanisms are better understood.
However, more progress is needed in multiscale modeling, coupled models between
organs or bones, and in digital twinning for patient specific modeling. The computational complexity of finite element models hinders progress and the incorporation of more observational data into numerical models.
This minisymposium aims to discuss recent results in model order reduction in this field, especially when strain prediction in organs belongs to a nonlinear manifold with a nonlinear latent representation.
All strategies incorporating recent algorithms related to dimensionality reduction, self-
supervised learning, autoencoders, or reduced-basis dictionaries are welcome at this
minisymposium
Keywords:
AI driven modeling, mechanobiological modeling, model order reduction, soft tissues
The convergence of advanced computational modelling and digital twin technology is
revolutionizing healthcare, enabling unprecedented precision, personalization, and
predictive capabilities. Digital twins—virtual replicas of physical entities or systems—
are increasingly leveraged in medicine to simulate physiological processes, optimize
treatments, and enhance decision-making.
The objectives of the mini-symposium will address the state-of-art of biomechanical modelling and simulation studies using finite element method (FEM) and their combination with artificial intelligence (AI) and mixed reality (MR) for evidence-based diagnosis, clinical decision in Healthcare.
Advanced modelling techniques, such as physics-based simulations, machine learning (ML), and deep learning, are integrated to capture the complexity of biological systems. Hybrid models, which combine mechanistic and data-driven approaches, offer a powerful solution to address the limitations of traditional methods. For instance, physics-informed neural networks (PINNs) merge differential equations with neural networks to improve the accuracy of simulations, while ensemble models aggregate multiple algorithms to enhance robustness. These advancements enable the creation of patient-specific digital twins, which can simulate organ-level interactions, predict disease progression, and evaluate treatment outcomes in silico.
In healthcare, digital twins are applied across a spectrum of domains: from personalized medicine, where patient-specific models predict disease progression, to surgical planning, where virtual replicas guide interventions.
However, challenges persist, including the need for high-quality, interoperable data, computational efficiency, and ethical considerations and translational research (integration of digital twins into routine clinical practice).
The objectives of the mini-symposium will address the state-of-art of biomechanical modelling and simulation studies using finite element method (FEM) and their combination with artificial intelligence (AI) and mixed reality (MR) for evidence-based diagnosis, clinical decision in Healthcare.
Advanced modelling techniques, such as physics-based simulations, machine learning (ML), and deep learning, are integrated to capture the complexity of biological systems. Hybrid models, which combine mechanistic and data-driven approaches, offer a powerful solution to address the limitations of traditional methods. For instance, physics-informed neural networks (PINNs) merge differential equations with neural networks to improve the accuracy of simulations, while ensemble models aggregate multiple algorithms to enhance robustness. These advancements enable the creation of patient-specific digital twins, which can simulate organ-level interactions, predict disease progression, and evaluate treatment outcomes in silico.
In healthcare, digital twins are applied across a spectrum of domains: from personalized medicine, where patient-specific models predict disease progression, to surgical planning, where virtual replicas guide interventions.
However, challenges persist, including the need for high-quality, interoperable data, computational efficiency, and ethical considerations and translational research (integration of digital twins into routine clinical practice).
Technologies enabling Digital Twins: MOR, Informed-AI, Hybrid models, …
Digital twins are emerging as a transformative paradigm for predictive modeling, real-time monitoring, and decision-making in engineering systems. Their development requires the integration of physics-based modeling, data-driven approaches, and robust uncertainty quantification.
This minisymposium aims to bring together researchers working at the intersection of computational mechanics, multiscale modeling, artificial intelligence, and uncertainty quantification to advance the next generation of digital twin technologies. Topics of interest include, but are not limited to, physics-informed machine learning, hybrid modeling approaches, reduced-order and multiscale models, data assimilation, uncertainty quantification and propagation, and applications to materials, structures, and complex systems.
By bridging fundamental modeling approaches with emerging AI techniques, this minisymposium seeks to foster cross-disciplinary discussions and promote the development of predictive, reliable, and scalable digital twins for engineering applications.
Keywords:
data-driven modeling, digital twins, materials and structures, multiscale modeling, physics-informed AI, Uncertainty Quantification
Digital twins often depend on high fidelity data upon which a digital representation of a physical state is built. However, for many applications, capturing data – or capturing sufficient high-quality data for building digital representations – is prohibitively costly or physically unrealistic. This minisymposum highlights current efforts to build digital twins when data is scarce, unreliable or unattainable. Examples of such efforts include problems include multimodal or multifidelity algorithms; incorporating first-principles or constitutive equation modeling; transfer learning; data augmentation; and Bayesian or statistical methods. Towards this end, this minisymposium will convene world-class researchers in a forum to present advances in data science, digital modeling, scientific computing, and machine learning, drawing upon expertise in these fields, specific domain applications, and digital workflows, focusing on digital twins for systems in mechanics and materials sciences.
Keywords:
digital twins, materials and structures, physics-informed AI
Deploying digital twins in practice requires models that can run fast enough to be useful, update as new data arrives, and produce predictions that engineers can trust. Physics-based simulations rarely meet all three criteria on their own, and black-box machine learning models often fail the moment conditions shift outside the training set. Scientific machine learning addresses this gap by building physical knowledge directly into the learning process — not as a post-hoc correction, but as a structural constraint.
This minisymposium focuses on SciML methods that work in the context of live, operational digital twins — not just offline surrogate training. We are particularly interested in contributions that deal with what happens after the model is built: keeping it synchronized with a real system, handling sensor noise and missing data, and knowing when its predictions can be trusted. Contributions are invited on:
• Physics-informed neural networks, neural operators, and equation-informed learning architectures
• Hybrid and multi-fidelity model architectures
• Reduced order models combined with or enhanced by machine learning
• Data assimilation and model updating for real-time synchronization
• State estimation under noisy or incomplete observations
• Uncertainty quantification and prediction reliability
• Graph-based and topology-aware learning for spatially distributed systems
• Applications across structural, mechanical, aerospace, energy, and manufacturing domains
The session welcomes both methodological contributions and case studies from real engineering problems. By bringing together researchers from nonlinear dynamics, computational science, and machine learning, this minisymposium aims to accelerate the development of SciML methods that are robust, efficient, and impactful for twinning practice.
Keywords:
digital twins, physics-informed AI, scientific machine learning, structural health monitoring
Digital twins have emerged as a transformative paradigm in engineering, driven by the rapid advancement of computational power and data-driven methodologies. A digital twin represents a dynamic, high-fidelity virtual counterpart of a physical system, enabling real-time monitoring, predictive maintenance, and accelerated design cycles. In this context, the high-fidelity numerical simulations and artificial intelligence (AI) frameworks that underpin these twins play a pivotal role in establishing the foundation for robust digital twin technology.
This mini-symposium aims to explore the synergistic integration of AI and physics-based simulations for the development of effective, efficient, and trustworthy digital twins in multiscale design. We specifically seek to address the computational bottlenecks inherent in high-fidelity models through the incorporation of machine learning surrogate modeling, physics-informed neural networks (PINNs), verification and validation (V&V), uncertainty quantification (UQ), and advanced optimization frameworks.
We invite contributions focusing on the development and application of digital twin technologies, as well as their foundational methodologies. We are particularly interested in research that bridges multiple scales, ranging from microscale material behavior to macroscale structural performance and manufacturing processes. By fostering a dialogue between domain experts in computational mechanics and data science, this session aims to define the future of predictive engineering.
Keywords:
ai, high-fidelity simulations, inverse design, multiscale modeling, physics-informed AI, Uncertainty Quantification, verification and validation
The Minisymposium “Image-based computational modelling of multi-phase and multi-physics systems” aims to showcase the latest progress in the integration of full-field imaging techniques with computational simulations for the study of complex multiscale and multiphysics problems. This class of challenges is characterized by the occurrence of simultaneous processes that span multiple scales, in which the characteristic sizes and times can differ by orders of magnitude [1]. Traditionally, simplifications are commonly considered as means to obtain a closed solvable form of the equations that describe the physical process of interest, namely the averaging theories of homogenization [2]. Meanwhile, in the latest years, full-field imaging techniques, such as neutron and x-ray tomography, have emerged as powerful tools for capturing the spatio-temporal evolution in various scientific and engineering domains [3, 4, 5, 6]. By providing unique insights into local processes, these techniques facilitate improving the understanding of the mechanisms that interact at different levels leading to the observed behaviour at the macroscopic scale [3]. Simultaneously, the advent of new methods and algorithms have enabled the possibility of obtaining conformal meshes with accurate morphology representing specific features of the microstructure being studied, including even several open-source and freely available options. Thus, the combination of full-field techniques with these computational methods allows for the explicit representation of the real heterogeneities and the capability of simulating their impact on underlying processes. The present minisymposium will focus on the sharing of the application of these integrated innovative approaches in distinct multidisciplinary areas where spatial-temporal evolution of key parameters plays a crucial role, such as moisture transport in porous media, mechanical behaviour of biological material and others. By bridging the gap between experimental observations and numerical simulations, these innovative approaches enable enhanced model validation and the ability to perform “virtual experiments”. The presentations within this session will showcase cutting-edge research that highlights the potential of these new innovative methods.
Keywords:
Data-driven modeling, digital twins, experimental mechanics, physics-informed AI, sensors
Recent advances in data-driven modeling and artificial intelligence have enabled the discovery of governing equations directly from data, opening new possibilities for understanding and predicting complex engineering systems. Equation discovery techniques, such as sparse regression (e.g., SINDy), symbolic regression (e.g., AI Feynman), and physics-informed neural networks (PINNs), provide powerful tools to uncover hidden physical laws and construct interpretable models from limited or noisy data.
This minisymposium aims to bring together researchers working on equation discovery and physics-informed learning in the context of engineering applications. Particular emphasis will be placed on the integration of data-driven methods with physics-based modeling for digital twin technologies, where interpretability, generalization, and robustness are essential.
Topics of interest include, but are not limited to:
・Equation discovery and symbolic regression methods for engineering systems
・Physics-informed machine learning and hybrid modeling approaches
・Discovery of governing equations for nonlinear, multiscale, and coupled phenomena
・Applications to solid mechanics, fracture mechanics, fluid dynamics, and multiphysics problems
・Integration of equation discovery with numerical simulation and high-fidelity models
・Uncertainty quantification and robustness of discovered models
By bridging the gap between data and governing equations, this minisymposium aims to advance the development of interpretable, reliable, and efficient digital twins for next-generation engineering systems.
Keywords:
Physics-informed machine learning, Equation discovery, Interpretable AI for Digital Twins
Digital models (DMs) are designed to be replicas of systems and processes. At the core of a digital model (DM) is a physical/mathematical model that captures the behavior of the real system across temporal and spatial scales. One of the key roles of DMs is enabling “what if” scenario testing of hypothetical simulations to understand the implications at any point throughout the life cycle of the process, to monitor the process, to calibrate parameters to match the actual process and to quantify the uncertainties. This mini-symposium presents the latest developments in real-time forecast and calibration approaches for digital twins. Approaches that are equipped with uncertainty quantification are especially highlighted in the mini-symposium.
Keywords:
digital models, digital twins, forecast, scientific machine learning, calbration
Reduced-order models (ROMs) have become a key technology for the efficient simulation of complex systems governed by partial differential equations. By constructing low-dimensional representations of high-fidelity models, ROMs enable rapid evaluations in many-query settings such as optimization, control, and uncertainty quantification, and are central to the development of digital twins. However, ensuring accuracy and robustness in strongly nonlinear and multi-parameter regimes remains a major challenge for classical approaches based on linear subspaces.
Recent advances in scientific machine learning, particularly generative methods but also broader learning-based approaches, open new perspectives for reduced-order modeling. These approaches provide flexible tools to learn complex solution distributions and nonlinear manifolds, going beyond the limitations of traditional projection-based techniques. Machine Learning techniques, such as generative models, operator learning, data-driven closure modeling, can enhance ROMs by improving the representation of nonlinear dynamics, enabling data augmentation, and supporting the development of hybrid physics–data-driven surrogates.
This mini-symposium aims to bring together researchers working on reduced-order modeling, Machine Learning for physical systems, and digital twin technologies. The focus is on emerging methodologies that leverage a wide range of AI techniques, with a focus on generative models, to build accurate, robust, and scalable ROMs. Topics of interest include nonlinear representation learning, generative modeling of solution manifolds, hybrid physics–machine learning methods, uncertainty quantification, learning-based closure strategies, operator learning, and applications to fluid dynamics, structural mechanics, and multi-physics systems.
By promoting contributions ranging from theoretical advances to industrial applications, this mini-symposium seeks to foster cross-disciplinary exchange and to advance the development of next-generation reduced-order modeling techniques driven by artificial intelligence and scientific machine learning.
Keywords:
scientific machine learning
Digital twins are rapidly evolving from descriptive and predictive tools into intelligent, adaptive systems that integrate data, models, and decision-making across the lifecycle of engineering systems. Recent advances point to a paradigm shift from reactive digital representations toward autonomous, agentic twins capable of learning, reasoning, and acting within a closed loop between the physical and virtual domains. In this emerging landscape, digital twins are no longer passive mirrors of reality, but active computational entities that enable discovery, optimization, and control under uncertainty.
Positioned at the interface of computational mechanics, uncertainty quantification, and scientific machine learning, this mini-symposium aims to explore the foundations and enabling technologies required to equip engineering systems with predictive and adaptive capabilities through digital twins. This session seeks contributions advancing next-generation digital twins, with emphasis on: model order reduction and surrogate modelling for real-time applications; uncertainty quantification, propagation, and management; physics-enhanced machine learning; and autonomous, agent-based digital twin frameworks. Particular attention is given to approaches that enable adaptive model updating, automated knowledge extraction, and self-learning digital twins that actively seek information and optimize their interaction with the physical environment.
Keywords:
digital twins
The realization of a functional Digital Twin (DT) hinges on resolving the fundamental conflict between physical rigor and real-time responsiveness. While high-fidelity simulations provide necessary precision, their prohibitive computational costs often preclude real-time synchronization with physical assets. Conversely, simplified models or purely data-driven approaches often lack physical consistency and predictive reliability in unobserved regimes. To bridge this gap, Multi-Fidelity (MF) modeling [1] has emerged as a critical enabler. By strategically integrating diverse information sources—ranging from multi-physics simulations and experimental data to real-time sensors and expert knowledge—MF approaches offer a robust framework to balance accuracy with computational efficiency, particularly as industries transition toward large-scale and safety-critical digital twin ecosystems.
This session aims to move beyond simple error correction, focusing on how MF frameworks can enhance and deepen information to transform disparate data into actionable insights for high-stakes decision-making. We invite contributions that address the core engineering requirements of modern DTs, including scalable mathematical structures that facilitate hierarchical fusion.
Furthermore, we seek innovative research on rigorous uncertainty quantification across fidelity layers, the hybrid integration of deductive physical laws with inductive observation data, and emerging AI-driven approaches, including agent-assisted modeling and autonomous workflow coordination, within unified probabilistic environments. We also welcome studies on the strategic application of these models to data-driven optimization, robust design, and efficient decision-making under uncertainty. By exploring these theoretical foundations and practical implementations across fields such as aerospace, infrastructure, and energy, this session seeks to define the next generation of reliable, scalable, and design-ready digital twins.
Keywords:
AI-driven design optimization, Autonomous digital twins, Hybrid models, Multi-fidelity modeling, Scalable surrogate modeling
Solving differential equations using high-fidelity solvers incurs significant computational costs, especially for real-time and many-query applications. Motivated by this, machine learning models are increasingly used for the approximation, simulation, and analysis of systems governed by ordinary and partial differential equations. The scope ranges from purely data-driven approaches, e.g., neural operators, neural ODEs, and deep learning-based surrogate models and solvers, to methods that embed prior physical knowledge, such as structure-preserving architectures, symmetry- and conservation-aware models, and generative AI with geometric or variational priors. Such structural properties may enter at different levels: in the design of the model architecture, in the construction of the training data, or in the learning process itself. Of particular interest are contributions that address the role of partial observability and measurement design, including sparse and sensor-informed models where data limitations fundamentally shape model architecture and training strategy. This includes operator learning and surrogate modeling approaches that explicitly account for sensing limitations, and contributions that characterize how data quality, observability, and measurement design constrain the performance of learned models. Contributions of both theoretical and computational nature are welcome, with emphasis on approaches that go beyond black-box learning by leveraging the mathematical structure of the underlying problem.
Keywords:
scientific machine learning, sparse sensing, structure preservation, surrogate modeling
Digital twins are continuously updated virtual representations of physical objects, processes or systems that support simulation-based decision-making by establishing a two-way link between physics-driven models and real-world observational data. Although they first gained prominence in the manufacturing industry under the Industry 4.0 paradigm, digital twins have since found application in a wide range of domains among others in critical civil infrastructure such as road networks, bridges, wastewater treatment facilities, and energy grids: systems characterized by high cost, long service lives, and stringent safety requirements. Thereby, the predictive capability and operational utility of digital twins depend fundamentally on sophisticated computational mechanics, the ability to run simulations in real time, and seamless integration between data and models.
Building dependable digital twins for civil infrastructure systems requires deep interdisciplinary collaboration, integrating domain-specific modeling, numerical simulation, data assimilation, and machine learning to construct, validate, and interconnect the necessary sub-models, datasets, and interfaces. In contrast to conventional simulation workflows, digital twins face the added challenges of satisfying real-time performance requirements, incorporating streaming data, and continuously updating their underlying models – tasks that become especially demanding in high-dimensional, multi-physics, or nonlinear problem settings. In safety-critical applications, achieving the levels of robustness and interpretability demanded by stakeholders places further constraints on numerical stability, uncertainty quantification methods, and the design of model hierarchies.
Keywords:
Digital Twins for Infrastructure, Geometric Semantic Models, Numerical Modeling Strategies, Sensor Data Collection and Management, Synchronization and Twinning, System Architectures, Uncertainties
MS044 – Shape Intelligence: Digital Twins for Explicit Geometry and Multiphysics Design Optimization
Digital twin technologies are undergoing a significant transition from passive system representation and predictive monitoring toward active engineering frameworks capable of supporting design generation, geometry adaptation, and intelligent decision making. Simultaneously, recent advances in explicit geometry representation, non-parametric optimization, artificial intelligence, and multiphysics simulation have created new opportunities for accelerating engineering design while improving manufacturability and real-world applicability.
This minisymposium aims to establish an interdisciplinary forum for researchers and engineers working at the convergence of digital twins, geometry driven optimization, and advanced computational engineering. The session seeks to explore how digital twins may evolve into active design environments capable of continuously interacting with simulation models, updating geometry, and guiding engineering performance throughout the product lifecycle.
Particular emphasis will be placed on explicit geometry-based methodologies and optimization approaches that directly manipulate engineering representations while integrating physical modelling with data driven techniques. The minisymposium welcomes contributions spanning fundamental developments, computational methodologies, and industrial applications.
Topics of interest include, but are not limited to: explicit geometry and shape optimization; thermofluid, thermal, and structural design optimization; topology and non-parametric optimization; reduced order and surrogate modelling; digital twin enabled design updates; artificial intelligence assisted engineering design; physics informed machine learning; uncertainty quantification and verification; manufacturability aware optimization; computational infrastructure; and industrial deployment.
The objective of this minisymposium is to stimulate discussion on the next generation of digital twins as intelligent design systems capable of linking simulation, optimization, artificial intelligence, and engineering implementation toward faster, more adaptive, and more reliable multiphysics design workflows.
Keywords:
AI Driven Design, Data Driven Engineering, Data-driven modeling, Digital Manufacturing, digital models, digital twins, Explicit Geometry Optimization, Industrial AI, Smart Engineering Systems, Thermofluid Optimization
The development of reliable and real-time digital twins for engineering structures remains a major challenge due to the high computational cost of conventional numerical simulations, limited sensor data availability, model uncertainties, and difficulties in integrating physics-based models with data-driven approaches. Recent advances in scientific machine learning, particularly Physics-Informed Neural Networks (PINNs) and related physics-guided learning frameworks, have opened new opportunities for constructing robust and efficient digital twins in computational solid mechanics.
This minisymposium aims to bring together researchers working on physics-informed and hybrid machine learning methodologies for digital twin applications in solid mechanics. The session will focus on emerging computational frameworks that integrate governing physical laws, experimental observations, sensor data, and numerical simulations within unified predictive environments. Particular emphasis will be placed on PINNs, neural operators, reduced-order modeling, operator-learning techniques, and hybrid finite-element–machine-learning approaches for forward and inverse problems.
Topics of interest include, but are not limited to, digital twins for structural systems, constitutive parameter identification, structural health monitoring, uncertainty quantification, multiscale and multiphysics modeling, real-time simulation, surrogate modeling, data assimilation, and AI-enhanced computational mechanics. Contributions addressing scalability, robustness, computational efficiency, industrial deployment, and real-world engineering applications of PINN-based digital twin frameworks are especially encouraged.
The minisymposium intends to foster interdisciplinary discussions among researchers in computational mechanics, scientific machine learning, applied mathematics, and engineering sciences, while highlighting current challenges and future research directions for next-generation physics-informed digital twins in solid mechanics.
Keywords:
computational mechanics, digital twins, Physics-informed machine learning
Digital Twins: Engineering applications
ABSTRACT
Non‑destructive testing (NDT) and structural health monitoring (SHM) are essential for ensuring the safety, reliability, and service life of engineering structures. Recent advances in sensing, multi‑physics modelling, and data analytics have accelerated the development of digital twins (DTs) for real‑time monitoring and predictive maintenance. Digital‑twin‑enabled NDT/SHM frameworks integrate physics‑based simulations with continuous sensing to support damage detection and structural diagnostics, and state‑of‑the‑art reviews show that digital twins significantly enhance monitoring accuracy and support real‑time model updating in civil infrastructure [1]. Parallel advancements in AI‑assisted NDT—such as automated ultrasonic and radiographic data interpretation—further strengthen DT pipelines by improving defect detection accuracy and reducing operator dependence [2].
Contributions are invited for following topics: (i) multi‑physics and data‑driven DT modelling; (ii) sensing and data‑fusion frameworks; (iii) feature extraction and damage indicators; and (iv) AI‑enabled NDT and DT‑driven diagnostics. Other relevant topics are also welcome.
REFERENCES
[1] Qiuting Wang et al., Current Status and Prospects of Digital Twin Approaches in Structural Health Monitoring, Buildings, 15(7), 1021, 2025
2 Amine el Mahdi Safhi et al., AI‑Driven Non‑Destructive Testing Insights, Encyclopedia MDPI, 4(4), 1760-1769, 2024
Keywords:
computational mechanics, digital twins, non-destructive testing, structural health monitoring
Digital twins (DTs) - high-fidelity, continuously updated virtual representations of physical systems - are transforming the field of aerospace engineering. This mini-symposium brings together researchers and practitioners to explore ways in which digital twin technologies can improve the design, operation and maintenance of such systems.
We invite contributions that leverage digital twins across a broad spectrum of aerospace applications. Topics may include, but are not limited to:
- Digital twins for design, testing, and certification use virtual replicas to speed up the design process, cut down on testing, and facilitate certification via physics-based simulations or hybrid approaches combining physical and efficient AI models.
- Structural health monitoring and predictive maintenance leverage digital twins to detect damage, estimate remaining life, and optimize maintenance.
- Geometric modeling develops high-fidelity geometric representations as the backbone of aerospace digital twins.
- Human-in-the-loop systems and decision support integrate human expertise with digital twins to enhance decision-making and situational awareness.
- Digital continuity, interoperability, and lifecycle management establish standards that ensure consistent data flow, compatibility, and maintainability.
- Fusion of system models, sensor data, and geometric models to combine physics-based models and real-time sensor data with geometric representations.
- Cloud-Edge-IoT solutions for aerospace applications and missions.
- Case studies from industry and large-scale deployments showcase real-world implementations, benefits, and lessons learned (e.g. mission operation control, mission design, system of systems).
While these examples highlight prominent use cases, the minisymposium encourages submissions on all aspects of digital twins for aerospace applications. This includes emerging techniques, cross-domain transferability, and the theoretical approaches of DTs tailored to aerospace systems.
Through this session, we aim to foster interdisciplinary dialogue between the computational mechanics, aerospace engineering, and machine learning communities.
Keywords:
Aerospace, data-driven modeling, digital twins, physics-informed AI
The main characteristic of a city or a territory, in terms of predictability, is its extreme complexity. The existence of numerous interconnected and interacting systems, of large spatial and temporal dimensions, presenting different granularities (systems of systems) and subjected to uncertainties and imprecisions that propagate within, makes modeling particularly complex. If we add to this the main protagonists, humans, whose behavior eludes deterministic models, the objective of achieving the desired level of diagnosis, prognosis, and support for decision-making is considerably limited.
This mini-symposium revisits the different technological bricks enabling the construction of digital twins of cities and territories, needing for an efficient alliance between physics-based and data-driven models, the former involving advanced model order reduction and regression technologies based on physics-based model solutions, and the latter based on data-assimilation and physics-informed machine-learning to fill the predictions to observations gap.
The large time and space scales that those systems involve, need important computational resources (high performance and quantum computing), the collection of smart (useful) data properly assimilated into the physics-based and data-driven model, as well as an appropriate verification and validation technologies for supporting control and decision making.
The resulting digital twins (prototypes and instances) should serve for optimal planning and operation, the former asking for advanced generative AI techniques, the latter more concerned by using predictive AI for static and dynamic systems enabling monitoring, diagnosis, prognosis and decision-making, covering both every-day operation and crises management and mitigation. The role of humans will be carefully considered.
Keywords:
digital twins, physics-informed AI, scientific machine learning
Modeling and simulation of composite materials' behavior and their advanced manufacturing processes (3D printing, injection molding, resin transfer molding, etc.) often involve complex and coupled physics with little to no available data on the actual material properties under processing conditions. The ignorance associated with the modeling of those processes propagates through the simulation and deviates the digital twin solutions from the reference measurement data [1,2]. Recent developments in the field of composite materials aim to complement the physics-based digital twins’ solutions with data-driven approaches, while others intend to refine the models through parameter identification or model augmentations. Real-world uncertainties and variability are also being addressed through different technologies.
The aim of this mini-symposium is to collect and discuss recent developments and advances addressing the build-up of digital twins for composite materials and their transformation processes, as well as models’ augmentations using data measurements and in-situ knowledge embedding within closed loops. The main topics of the mini-symposium are, but not limited to, uncertainty propagation, parameter identification for in-situ processes, closed-loop real-time modeling, physical models’ augmentation with data, and ignorance characterization in the simulation of composite manufacturing processes (e.g. resin transfer molding, automated fiber placement, etc.).
REFERENCES
[1] Ghnatios, C., Gérard, P. & Barasinski, A. An advanced resin reaction modeling using data driven and digital twin techniques. Int J Mater Form 16, 5 (2023). https://doi.org/10.1007/s12289-022-01725-0.
[2] Rodriguez, S., Monteiro, E., Mechbal, N. et al. Hybrid twin of RTM process at the scarce data limit. Int J Mater Form 16, 40 (2023). https://doi.org/10.1007/s12289-023-01747-2.
Keywords:
closed-loop, composite manufacturing, composite materials, computational mechanics, hybrid modeling, physics-informed AI, uncertainty propagation
In many engineering applications involving digital twins (DTs), using high-fidelity forward models can render the solution of inverse and control problems computationally intractable. This is typically due to the high-dimensionality and complexity of high-fidelity models, often accompanied by high-dimensional inference parameters and decision spaces. Many-query applications, such as optimization and control, however, require a fast, yet accurate model output response, as well as information about the response uncertainty. Surrogate models and reduced-order models (ROMs) can help make these problems tractable, provided they achieve sufficient accuracy and can be built from a limited number of forward model evaluations. This MS aims at addressing several key challenges that arise in the usage of surrogate models in DT settings:
1. Goal-oriented modeling: Surrogates and ROMs do not necessarily need to reproduce the full spatio-temporal dynamics of the system. Instead, they may only need to capture the control objectives and data assimilation observables accurately. Determining what appropriate methodologies are for the design of such models remains an open challenge.
2. Structure preservation: DTs often simulate system dynamics over extended periods; therefore, the design of corresponding ROMs could exploit physics-based knowledge to preserve important structural properties and invariants, such as energy or mass.
3. Trustworthy neural network surrogates: Neural network representations have shown strong potential as high-dimensional surrogate models, but further work is needed to establish their reliability for dynamical systems, especially when training data is limited.
4. Nonlinear low-dimensional representations: Many surrogate and ROM techniques rely on the existence of an intrinsically low-dimensional parameter-to-output map or solution manifold. However, linear subspace methods may fail to represent this structure efficiently for important problem classes, e.g., advection-dominated flow and transport.
5. Accurate gradient approximation: When surrogates are trained only on samples of high-fidelity input–output maps, and not on their Jacobians, they may approximate gradients poorly. This can, in turn, lead to inaccurate solutions of the optimization problems that underpin data assimilation and optimal control.
Keywords:
Data-driven modeling, digital twins, scientific machine learning
The increasing complexity of infrastructure systems and their exposure to uncertain, evolving environments call for integrated, lifecycle-aware approaches to enhance resilience and sustainability. Recent advances in data-centric engineering and artificial intelligence have enabled new possibilities for developing digital twins that connect physical systems with their virtual counterparts across different stages of the lifecycle. This minisymposium aims to bring together contributions focused on methodologies for developing digital twins for infrastructure systems. Such developments may involve model formulation, data integration, simulation, validation, deployment, and continuous updating. The scope includes approaches applicable to various lifecycle stages, such as design, construction, operation, monitoring, and post-disaster response, without requiring full lifecycle coverage. Emphasis is placed on methods that enhance system understanding, predictive capability, and decision-making under uncertainty.
Topics of interest include, but are not limited to: data-driven and hybrid modeling approaches, sensing and monitoring techniques, model updating and data integration, predictive analysis, and decision-support methodologies for resilient infrastructure systems. Contributions addressing the interaction between physical models and data, as well as approaches enabling real-time or near-real-time system representation, are particularly encouraged. This minisymposium seeks to foster exchange across disciplines, including computational mechanics, data-centric engineering, and infrastructure applications. It aims to advance lifecycle-aware digital twin frameworks for resilient and intelligent infrastructure systems.
Keywords:
ai, civil engineering, digital twins
Digital twin technologies require the seamless integration of physics-based simulation, data-driven modeling, uncertainty quantification, and real-time computational workflows. However, complex engineering systems often involve multiscale and multiphysics phenomena, including fluid-structure interaction, fracture, impact, manufacturing processes, infrastructure systems, and coupled thermo-mechanical behavior. No single numerical method is sufficient to address all aspects of such problems with the robustness, accuracy, and computational efficiency required for practical digital twins.
This minisymposium focuses on multimethod computational mechanics as a methodological foundation for next-generation physics-informed digital twins. Topics of interest include, but are not limited to, finite element, finite volume, finite difference, particle, meshfree, lattice Boltzmann, immersed boundary, Eulerian, Lagrangian, and hybrid numerical methods; coupling strategies for multiphysics and multiscale problems; model order reduction and surrogate modeling; data assimilation and sensor-informed simulation; physics-informed and hybrid AI models; uncertainty quantification; generative AI and generative design for engineering systems; and high-performance computing platforms for real-time or near-real-time digital twin applications.
The objective of this minisymposium is to bring together researchers working on computational mechanics, numerical methods, AI for science, and digital twin applications, and to discuss how diverse numerical and data-driven approaches can be integrated into reliable, interpretable, and deployable digital twin frameworks. Particular emphasis will be placed on cross-method comparison, method coupling, practical limitations, validation, and the development of new computational paradigms beyond conventional disciplinary boundaries. MS Topics include technologies enabling digital twins, hybrid models, informed AI, data assimilation, model order reduction, uncertainty quantification, generative AI and generative design, engineering applications, and AI-HPC integration.
Computational electromagnetics has long been a foundational technology for the design and analysis of electrical systems, including transformers, electric machines, and printed circuit boards. As modern engineering demands increasingly complex geometries, multi-physics coupling, and higher accuracy, the limitations of conventional solvers such as the finite-difference time-domain (FDTD) and finite element methods (FEM) have become more pronounced due to their rapidly growing computational cost. Moreover, the intrinsic complexity of solving partial differential equations derived from Maxwell’s equations poses fundamental challenges that are difficult to overcome by algorithmic improvements alone.
This mini-symposium envisions a paradigm shift in computational electromagnetics driven by machine learning. By integrating data-driven models with physical laws, emerging approaches such as physics-informed neural networks (PINNs), neural operators, and AI-assisted inverse modelling offer new possibilities for fast, scalable, and adaptive electromagnetic analysis. The symposium aims to bring together researchers exploring how artificial intelligence can complement or redefine traditional numerical methods, enabling real-time simulation, uncertainty-aware modelling, and automated discovery in electromagnetic field analysis. Through this dialogue, we seek to illuminate future directions toward next-generation electromagnetic solvers that seamlessly combine physics, data, and computation.
Keywords:
Computational electromagnetics, Inverse design, Physics-informed machine learning
Digital twin technology is emerging as a transformative framework for the monitoring, prediction, and management of engineering infrastructure. While digital twins have seen rapid development in manufacturing, aerospace, and structural engineering, their applications in geomechanics and geoengineering remain comparatively underdeveloped due to the complexity of geomaterials, multiphysics coupling, and substantial uncertainties. Recent advances in computational geomechanics, monitoring technologies, artificial intelligence, and physics-informed learning provide new opportunities to overcome these challenges. By integrating physics-based numerical modelling, field monitoring, data assimilation, and machine learning, digital twins offer the potential to continuously update predictions of infrastructure performance and support adaptive, risk-informed decision-making throughout the infrastructure lifecycle.
This mini-symposium aims to bring together researchers and practitioners working on digital twin technologies for geomechanics and geoengineering applications. Contributions are invited on the development of physics- and data-driven frameworks, coupled multi-physics modelling, uncertainty quantification, monitoring integration, and intelligent predictive systems for geotechnical infrastructure. Both methodological developments and practical engineering applications are welcome.
This minisymposium addresses methods and tools for the Engineering contains deployment and operation of industrial digital twins, a cornerstone technology for smart manufacturing and transformation of industrial processes. The conceptions of industrial digital twins are complex and multidisciplinary systems that integrate physical assets, data, models, and services, raising significant challenges in interoperability, cybersecurity, and scalability. The session aims to present and discuss applied research contributions that tackle these challenges through structured engineering approaches and open, interoperable platforms.
The minisymposium will highlight results from the JNI applied research program led by IRT SystemX, which brings together academic and industrial partners from sectors including aerospace, energy, mobility, telecommunications, and defense. Contributions will showcase methodological framework for designing and implementing digital twins of complex industrial systems, emphasizing the coordination of domain expertise with systems and software engineering practices and a cyber-by-design approach [1].
Specific topics covered in the session include reference architecture and standards for industrial digital twins, data management and traceability based on the industry 4.0 Asset Administration Shell. Several workflow models for orchestration, simulation, and the integration of analytics and artificial intelligence services will be presented. Particular attention will be given to open and interoperable platforms, zero trust architectures, and standards such as the industry 4.0 Asset Administration Shell [2].
Through representative industrial processes such as predictive maintenance, smart energy management, and secure autonomous systems, the minisymposium aims to illustrate how these methods and tools enable robust, secure, and scalable digital twin implementations. Overall, the session seeks to foster knowledge exchange between researchers and practitioners, identify best practices, and contribute to the maturation of engineering approaches for industrial digital twins. It also encourages discussions on validation, lifecycle management, governance, and sustainability of digital twins, while aligning industrial expectations with emerging research results and supporting the creation of trustworthy, reusable, and evolvable digital twin ecosystems across academic and industrial communities.
Keywords:
Asset Adinistration Shell, Interoperability, Methodological framework, Open source Digital Twin Plateforme
Mathematical foundations
Digital twins are rapidly emerging as a central paradigm for real-time prediction, monitoring, and control of complex engineering systems. A key challenge in constructing reliable digital twins is the development of data-driven models that remain faithful to the mathematical and physical structures governing the underlying systems. These may include positivity, monotonicity, convexity, and other functional constraints, as well as physical invariants such as conservation of mass, momentum, or energy. Digital twins further demand that models respect symmetries, stability characteristics, and geometric structure to promote reliability. Without such structure preservation, learned models may exhibit nonphysical behavior, violate conservation laws, or become unstable, especially when deployed in predictive settings.
This minisymposium focuses on methodologies for building structure-preserving data-driven models that enable robust, interpretable, and trustworthy digital twins. We will bring together researchers developing and applying novel methods involving structure-preserving model reduction, operator inference, physics-constrained scientific machine learning, and related areas. Topics of interest include: physics-constrained dynamical systems learning, data-driven reduced-order models for digital twins, data-driven variational integration, equivariant machine learning, structure-preserving neural operators, constrained system identification, and other techniques for ensuring invariant preservation, stability and passivity in learned models. We welcome contributions from a variety of science and engineering applications, including but not limited to fluid dynamics, solid mechanics, quantum mechanics, fluid-structure interactions, energy systems, advanced manufacturing, and Earth system modeling.
Keywords:
data-driven modeling, digital twins, scientific machine learning, structure preservation
This minisymposium explores recent developments in scientific machine learning methods for solving partial differential equation (PDE)-based problems on complex geometries and topologies. Classical numerical methods address these challenges using advanced meshing, spatial adaptivity, and embedded techniques; while highly effective, they often entail significant computational cost and numerical complexity.
Learning-based approaches, including neural operators and neural network surrogates, offer a promising alternative. Although they often avoid explicit meshing, geometric information must still be encoded and incorporated into the model and training process, posing challenges in representing geometry, enforcing boundary and coupling conditions, and capturing small-scale features. Moreover, surrogate models mapping from geometry to solution require suitable parameterizations of the domain, ranging from pixel or voxel representations to (signed) distance functions and latent embeddings. There is also growing interest in methods that operate directly on mesh-based representations, such as graph neural networks.
This minisymposium brings together researchers from machine learning, numerical analysis, and computational science to discuss architectures, training strategies, theoretical foundations, and applications. We are particularly interested in neural network-based discretization approaches (e.g., physics-informed neural networks), methods building on classical numerical bases, as well as neural operator and surrogate modeling architectures, with an emphasis on robust and scalable methods for PDEs in complex geometries.
Keywords:
complex geometries, neural operators, scientific machine learning
The evaluation of critical buckling loads is one of the fundamental problems in the stability analysis of slender structural elements, particularly in applications related to earthquake engineering, structural design, and computational mechanics. This work presents the application of the Transfer and Transport Matrix Method (TMM) as an analytical and numerical approach for determining critical buckling loads in bars subjected to axial forces under different boundary conditions, including restrains, free and sliding supports. The method has recently been successfully applied in other fields of engineering [1].
This article aims to present a methodology based on the classical differential formulation of elastic stability, incorporating matrix techniques to describe the buckling modes of the structural element. Transfer matrices establish relationships between displacements, rotations, and internal force at different stations of the bar. The generation and plotting of buckling modes are performed using the kinematic variables of the station vectors.
Although methods such as the Finite Element Method (FEM) and the Direct Strength Method (DSM) have become standard tools in structural analysis due to their versatility and computational implementation. Transfer matrix formulations have experienced renewed interest due to mathematical elegance, reduced computational cost, and suitability for parametric and segmented structural systems [2]. Particularly, the TMM provides a natural and efficient framework for the stability analysis of slender members.
The research also considers the implementation of the methodology within scientific programming environments to automate calculation procedures and support parametric simulations. From an interdisciplinary perspective, the method combines concepts from differential equations, matrix algebra, and structural stability theory with applications relevant to modern engineering and seismic-resistant design.
The expected results aim to contribute to the development of advanced computational methodologies for structural stability assessment and to promote the integration of matrix-based approaches into research, teaching, and technological applications in civil and computational mechanical engineering.
Keywords:
Buckling modes, civil engineering, computational mechanics, structures, Transfer Matrix Method
Computation, platforms and infrastructures, …
With the rapid development of science and technology, applications of mechanics have expanded across various engineering fields. This minisymposium focuses on the applications of mechanics in civil engineering, with particular emphasis on theoretical developments, computational methods, numerical simulations, and experimental approaches. The minisymposium provides a platform for researchers from academia and industry to present recent advances, innovations, and practical applications of mechanics. We sincerely invite contributions related to the aforementioned topics to exchange ideas and findings through presentations and discussions.
Keywords:
civil engineering, experimental mechanics, structures, theoretical analysis, computational mechanics
ΜL methods for uncertainty quantification
Digital twin models enable continuous, near real-time assessment of the risk of damage and failure in complex engineering systems by tightly coupling high-fidelity simulations with live sensor data. However, high-fidelity digital twin models are often computationally expensive, making quick-turnaround risk evaluation, control, and many-query reliability analysis prohibitively costly. Machine learning (ML) has emerged as a key enabling technology to bridge this gap by accelerating uncertainty propagation, learning surrogate dynamics from heterogeneous data, and quantifying both aleatoric and epistemic uncertainties under realistic operating conditions.
This Minisymposium aims to bring together recent advances in ML-driven uncertainty quantification, risk assessment, and reliability analysis for digital twin applications. We invite contributions on, but not limited to: (i) physics-informed and data-driven surrogates, reduced-order models, polynomial chaos, Gaussian process, and operator-learning approaches for stochastic dynamical systems; (ii) multifidelity methods for estimating tail-risk measures such as conditional value-at-risk, probability of failure, and other rare-event statistics for nonlinear systems with dependent and high-dimensional inputs; (iii) deep quantile, Bayesian, ensemble, and generative models for uncertainty-aware prediction and decision-making; (iv) active learning, adaptive sampling, and optimal experimental design for rare events and reliability-based design optimization; (v) sensor fusion, data assimilation, and online model updating for time-evolving twins; (vi) uncertainty-aware control, prognostics and health management, and predictive maintenance; and (vii) cross-disciplinary applications in structural and mechanical systems, energy and combustion plants, additive manufacturing, aerospace, and civil infrastructure.
By bridging methodological developments with engineering practice, this session aims to foster a community discussion on trustworthy, computationally efficient digital twins capable of robust, risk-informed decision-making across a broad spectrum of disciplines.
Keywords:
digital twins, Multi-fidelity Learning, scientific machine learning, uncertainty propagation, Uncertainty Quantification
AI and High-Performance Computing
The application of artificial intelligence (AI) technologies in computational mechanics has a long and rich history. However, the integration of recent AI advances—particularly deep learning, physics-informed neural networks (PINNs), generative models, and multi-fidelity learning—into computational mechanics is still in its early stages and rapidly evolving. The objective of this mini-symposium is to explore how AI techniques, including deep learning and other machine learning approaches, can be effectively applied to address fundamental and applied problems in computational mechanics. We warmly welcome contributions that advance the synergy between these two fields, aiming to develop impactful and innovative methodologies. Of particular interest are studies where AI enables the simulation of previously intractable physical phenomena, accelerates large-scale or multi-physics simulations, supports data-driven scientific discovery, or significantly improves the accuracy and efficiency of existing computational models.
Keywords:
Data-driven modeling, Multi-fidelity Learning, AI driven modeling, Machine learning
Environmental flows, including atmospheric boundary layers, urban flows, and pollutant dispersion, are characterized by strong nonlinearity, turbulence, and multiscale interactions. These challenges are encountered across a wide range of spatial scales and physical settings, from street-level urban environments to regional and global atmospheric flows. Numerical simulations, such as large-eddy simulation (LES) and mesoscale models, provide detailed insights; however, they remain computationally expensive and are often unsuitable for real-time applications and operational forecasting.
Recent advances in artificial intelligence (AI), machine learning, and data-driven modeling have enabled new approaches to accelerate simulations and construct reduced-order models (ROMs) that retain essential physical characteristics. In particular, neural operators, autoencoder-based ROMs, operator inference, and hybrid physics–AI models offer promising frameworks for bridging the gap between high-fidelity simulations and real-time prediction.
This minisymposium aims to bring together researchers working on AI-driven model reduction and data-driven approaches for environmental flows across diverse physical settings and scales, ranging from urban microenvironments to mesoscale and global climate flows. The focus is on developing scalable, physically consistent, and interpretable models that integrate numerical simulations, observational data, and machine learning techniques.
Topics of Interest
Data-driven reduced-order modeling (POD, autoencoders, neural operators)
AI-based prediction of wind, temperature, and humidity fields
Bias correction of numerical weather prediction models
Physics-informed machine learning
Data assimilation and sensor integration
Multi-fidelity and surrogate modeling across scales
Urban climate and air quality modeling
Urban heat island analysis and mitigation using data-driven methods
Pedestrian-level wind and thermal comfort prediction
Environmental digital twins and real-time forecasting
Keywords:
ai, civil engineering, Data-driven modeling, Machine learning