Plenary Lectures
Cláudio Ruggieri
University of São Paulo
Computational Fracture Mechanics (CFM) has undergone significant development, from early analytical approximations to advanced numerical techniques capable of addressing complex three-dimensional (3D) crack problems. The presentation provides an overview of the progress and challenges in CFM, drawing insights from historical developments to recent achievements. The lecture will highlight key numerical methods, with a strong focus on the widely employed Finite Element Method (FEM), but also providing some key aspects of the Boundary Element Method (BEM) and the extended FEM (XFEM) approaches in addressing crack problems in the framework of CFM. Special attention will be given to numerical simulations of cracked components, the role of fracture criteria and the integration of high-performance computing in modern CFM. Despite powerful computational resources now available, there are still restrictions imposed by our current understanding of the local fracture mechanism, including fracture under brittle and ductile regimes, fatigue crack growth, environment assisted cracking, such as hydrogen embrittlement, and multiscale effects in heterogeneous materials to obtain meaningful and effective structural integrity predictions which are reasonably invariant to model details and component geometry. Clearly, the need to incorporate all the relevant mechanical and metallurgical microfeatures into the model to obtain a more accurate description of the physical mechanism (thereby placing emphasis on developing a more fundamental, computational model) should be tempered by the need to develop a relatively simpler, but yet highly effective, engineering model capable of dealing with complex situations with only modest computational effort.
Elena Atroshchenko
University of New South Wales
Using energy harvested from the environment as an additional or the only power source for a sensing system can prolong its service life by reducing battery replacement efforts, enabling its use in remote locations, and facilitating the ubiquitous employment of the Internet of Things (IoT). Piezoelectric energy harvesting (PEH) is a widely used technique that can convert mechanical vibrations into electricity; however, the power produced by conventional PEHs is insufficient to compete with batteries. In order to increase energy generation, we explore metamaterials—periodic assemblies of unit cells engineered to achieve novel functionalities. We show how two types of metamaterials—locally resonant structures and auxetic metamaterials—can be designed using isogeometric shape optimisation. Moreover, the voltage signal produced by a PEH carries information about the vibration source and can be used for sensing, replacing a dedicated sensor. We demonstrate that a voltage-based unsupervised sensing framework, based on a variational autoencoder, can be highly efficient in bridge damage identification, demonstrating a promising path toward self-powered, intelligent monitoring systems.
Hadi Hajibeygi
TU Delft
Underground hydrogen storage (UHS) for energy supply-demand management is a relatively new topic compared to the natural gas storage. It is gaining increasing interests due to the ‘hope’ that hydrogen may indeed be the missing link of scalable low-carbon energy systems. Successful deployment of UHS depends on reliable performance analyses, which depend on rigorous multiscale modeling and simulation of the relevant cyclic processes at various scales. To address this, we present a multiscale experimental-numerical framework, addressing the thermo-chemical properties at molecular scale, trapping mechanisms at micro-meter scale, and its performance and recoverability in continuum reservoir scale. The nonlinear, time-dependent mechanical response of the host rocks is also addressed, with the focus on model construction and parameter calibration, including field validation. Emphasizing the importance of reliable performance assessments, some key gaps in this evolving technology will be also addressed.
Irina Tezaur
Sandia National Laboratories
Multi-scale methods are crucial for understanding and predicting the behavior of engineering systems where small-scale events influence overall performance. This talk presents a novel approach for concurrent multi-scale coupling in finite deformation solid mechanics using the Schwarz alternating method (SAM). This methodology leverages known solutions to partial differential equations (PDEs) in simpler domains to iteratively construct solutions for more complex domains.
The first part of this talk focuses on overlapping and non-overlapping variants of SAM for high-fidelity finite element models in solid mechanics. I will demonstrate on some large-scale numerical examples simulated using Sandia’s Sierra/Solid Mechanics code that this method can accelerate mod/sim workflows and simplify the meshing of complex geometries.
In the second part of the talk, I will discuss some recent extensions of overlapping SAM to couple high-fidelity models with non-intrusive Operator Inference (OpInf) reduced order models (ROMs), facilitating the integration of data-driven models into existing multi-scale frameworks. Numerical examples will illustrate runtime reductions, as well as the potential for improving the predictive viability of projection-based ROMs through spatial localization and online integration of high-fidelity information.
Finally, time permitting, I will introduce a novel contact enforcement approach rooted in non-overlapping SAM. By treating each body as a separate, non-overlapping domain and preventing interpenetration using an alternating Dirichlet-Neumann iterative process, our approach eliminates the need for contact constraints and offers great flexibility with respect to meshing. I will show through numerical examples that the approach yields enhancements in accuracy and energy conservation compared to traditional methods.
John Dolbow
Duke University
We provide an overview of several recent advances in models for the fracture mechanics of elastic brittle materials. These advances are changing the focus in the community from models and simulations of crack propagation to those for crack nucleation. At this point it is clear that at least three intrinsic macroscopic material properties govern fracture nucleation in elastic brittle materials: i) the elasticity; ii) the strength; and iii) the fracture toughness or critical energy release rate. Of these, material strength has been the most misunderstood and overlooked, despite its central role in governing crack nucleation under a broad set of conditions. Precisely how cracks nucleate and transition to propagation, and how descriptions of strength should be interlaced with Griffith-like energetics in models for fracture has been a subject of study for decades. We will focus attention on an emerging class of regularized models of the phase-field type that are complete in the sense that they allow for the representation of an arbitrary strength surface. This completeness is what has permitted these models to move beyond mere calibration against particular experimental observations. Select examples from quasi-static and dynamic fracture that illustrate the interplay between strength and energetics will be shown. The challenges and opportunities this new perspective provides for research in numerical methods will be discussed, as well as an outlook into modeling fracture nucleation beyond the basic setting of elastic brittle materials.
Maria Jose Pontes
Federal University of Espírito Santo
Victor Yepes
Universitat Politècnica de Valéncia
The increasing frequency and intensity of natural and anthropogenic hazards have underscored the need to develop resilient infrastructure systems. This plenary lecture presents recent advances in resilient optimization and decision-making frameworks for structural engineering, integrating robustness, sustainability, and lifecycle management. The discussion begins by redefining resilience as a time-dependent property, encompassing both structural robustness and recovery capacity, and its interdependence with environmental and social sustainability. The lecture then introduces computational frameworks that combine robust and distributionally robust optimization (DRO) with reduced-order modeling, hybrid metaheuristics, and surrogate modeling to achieve efficient, uncertainty-aware structural designs. Practical applications are illustrated through case studies on progressive collapse prevention, durability optimization in coastal environments, and cost-efficient design of hybrid steel I-beams. Finally, the presentation explores multi-criteria decision-making (MCDM) approaches—using fuzzy, neutrosophic, and Bayesian methods—to incorporate subjective, ethical, and participatory dimensions into engineering decisions. By bridging advanced optimization, digital twins, and human-centered decision processes, the lecture outlines a comprehensive vision for designing resilient, sustainable, and socially responsive structures in the 21st century.