ABSTRACT. Superconducting accelerator magnets are being developed for enhanced magnetic fields in the Future Circular Collider (FCC) at CERN, which means that, during operation, the superconductive coils will be exposed to even higher electromagnetic (Lorentz) forces. To ensure the mechanical stability and quench protection during normal operation, in this study, the epoxy-impregnated 10-stack made from Nb3Sn Rutherford cables, which can be regarded as a unit cell of superconductive coils, is investigated under compression for its anisotropic mechanical characterization at room and cryogenic temperatures.

In addition, detailed constitutive relations for each Rutherford-cable component (epoxy resin, annealed copper, Nb3Sn superconductor and insulation mica/glass layer) will be determined, in order to build a 2D representative finite element (FE) model at the 10-stack level. This will enable us to predict thermo-mechanical behavior of 10-stack Rutherford cables under cyclic loadings. The homogenized constitutive response of these 10-stack cubes is of significance for the feasibility of coil design and construction.

An in situ full-field deformation measurement is performed at the level of cable stacks via image-based analyses. Rather than the stress-strain curve fit, the image analysis will provide strain measurements at the local strand level, which will be used to validate the FE simulations and, ultimately, explain the failure from knowledge of the detailed mechanical loading.

ABSTRACT. The current trend in transportation industries such as aerospace, automotive, or railway towards the use of recyclable materials has increased the manufacturing of composite parts made of thermoplastic materials through thermoforming processes. Considering the limitations in shaping these types of materials, the development of a simulation-based methodology is necessary to study the manufacturing process and make informed decisions to ensure the quality of the parts. In this scenario, numerical analysis emerges as a fundamental resource in reducing time and costs compared to a purely experimental procedure. Simulation provides a new approach, allowing for faster iterations and the prediction of defects such as wrinkles, delaminations, or distortions, enabling corrective actions to enhance process efficiency. This article focuses on constructing a virtual model to study and optimize the manufacturing of a structural component through a thermoforming process using PAEK thermoplastic material reinforced with unidirectional carbon fiber, employing the PAM-COMPOSITES software from ESI Group. The development of Forming Limit Curve based on shearing and tensile deformation is also explored to assess anomalies in the manufacturing process of composite components through thermoforming. The obtained results are compared with the experimentally manufactured part, analyzing the presence of defects and distortion phenomena, thereby validating the model and establishing a methodology to optimize various phases of the process.

ABSTRACT. Background: Natural fiber reinforced polymers composites (NFRP) are currently very popular in the field of engineering, due to their excellent mechanical and physical properties and low manufacturing costs [1]. The study of the mechanical and failure behavior of thin flax/epoxy composites has been the subject of several research papers [1-2]. The objectives of this paper are to address the mechanical behavior and failure mechanisms that occur of thick flax/epoxy composite laminates, so far lacking in the scientific literature. Procedure: To achieve the objectives mentioned previously, experimental inter-laminar shear tests are proposed. These tests have been conducted on various thickness of the laminates (2mm, 4 mm and 8.5 mm). For the analysis of the mechanisms of failure, the tests have been recorded by CCD camera and the images have been used to measure the strain fields thanks to the DIC technique. The experimental results have been correlated to those obtained by Finite Elements Approach. The numerical model has been developed with cohesive elements in order to predict the damage between layers. Key findings: From the experimental results and the finite element analysis, the following conclusions can be drawn: • Failure of the laminates occurred due to delamination at the interface between the plies and the decohesion of the middle layer for all the specimens tested. • The Finite Element model show a good agreement with experimental results indicate the validity of the FE model at the macro-scale. However, at the meso-scale, some mechanisms of failure are not captured by the model. References: [1] I. El Sawi et al. An investigation of the damage mechanisms and fatigue life diagrams of flax fiber-reinforced polymer laminates, Journal of materials science, 2014. https://doi.org/10.1007/s10853-013-7934-0 [2] K. Mayssa et al. Evaluation of interlaminar shear of laminate by 3D digital holography, Optics and Lasers in Engineering, 2017. https://doi.org/10.1016/j.optlaseng.2016.12.014

ABSTRACT. Measuring the appearance and propagation of damage during impact testing remains a difficult task. A recent study (by the authors) using high-speed infrared thermography has shown that this measurement technique gives access to in-situ damage measurement during low velocity impact tests. In particular, matrix cracking in the surface ply and delamination of the first interface below the non-impacted surface can be monitored. The aim of this study is to evaluate the possible extension of this methodology to high velocity impacts. For that purpose, ballistic impact tests were carried out using a gas gun and a rigid 16 mm diameter ball. Various impact velocities were tested between 40 m.s-1 and 75 m.s-1, preferably below the ballistic limit. An infrared mirror was used to monitor the opposite surface using an infrared camera. A metrological investigation with an extended blackbody was performed to ensure that temperature measurement is not disturbed by this specific configuration. The stacking sequence used for this experimental investigation is a [0/+60/-60]ns made from T700/M21 UD ply leading to a total thickness of approximately 3.25 mm. This may allow direct comparison with the results available in the literature on a triaxially braided composite material with the same fibre orientations. Although the number of images captured for these high-speed impact configurations is limited, they nevertheless provide some information about the damage scenario during these tests.

ABSTRACT. Fibre Reinforced Polymers (FRPs) are extensively used in structural applications, due to their outstanding specific mechanical properties. To guarantee safety, knowledge of their damage mechanisms is essential.

Interlaminar fracture, or delamination, is a critical damage mechanism for laminated FRPs. Characterization of interlaminar fracture toughness (IFT) follows international standards, which recommend using unidirectional (UD) specimens, where delamination is propagated along the fibre direction. In real applications, however, multidirectional (MD) laminates are used, and delamination may initiate at any interface and grow in any direction, with a different IFT.

Due to several problems (three-dimensional effects, thermal residual stresses, undesired energy dissipation mechanisms, delamination migration) there is no agreement on how to characterize IFT using MD specimens. Researchers have been trying to design optimal MD specimens for decades. A major recent development was the introduction of Fully-Uncoupled Multi-Directional (FUMD) specimens, featuring unprecedented thermoelastic uncoupling properties and enabling testing of any desired interface. Preliminary studies demonstrated the potential of FUMD specimens, making them interesting candidates for standardisation purposes.

The number of feasible FUMD specimens increases with the specimen ply number, and it can become large. Wide adoption and standardisation require knowledge of all feasible designs, their evaluation and selection. In this study, we derive the full set of layups for FUMD specimens design, up to current computational limitations. Specifically, since FUMD layups are obtained from quasi-trivial (QT) quasi-homogeneous layups, we formalise the process to select, from a complete database, those QT layups usable for FUMD specimens, and implement a code to perform the selection at scale. We classify and analyse the sequences obtained to glean insight on aspects of practical interest to specimen design, such as number of usable orientations, possibility to include 0º layers and their number, feasible delamination interfaces as a function of the ply number.

ABSTRACT. Carbon composite materials are ideal candidates for High Energy Physics (HEP) applications due to their low density, high stiffness-to-weight ratio and excellent thermal properties. They are widely used in the support structures of tracking detectors, where they play a key role in the thermal management of the silicon sensors and readout electronics. In state-of-the-art trackers such as those installed in the experiments at CERN’s Large Hadron Collider, lightweight composite structures provide the main heat path between the silicon modules and a network of metallic or plastic pipes containing a cooling fluid. However, despite the good results obtained with this approach, the performance targets of future HEP experiments call for even lighter and more efficient technologies. In this respect, the use of sacrificial materials to create micro-vascular networks in the composite laminates represents a very appealing solution to integrate the cooling circuit in the support structure. In this work, both experimental and numerical methods have been used to investigate the pressure resistance of channels embedded in carbon composite laminates. Modified poly(lactic) acid (PLA) preforms have been combined with a post-cure vaporization technique [1] to manufacture plates with longitudinal channels. Destructive tests have been carried out to determine the burst pressure of the plates as a function of the layup and the cross-section geometry of the channels. The deformation of the composite plates during the tests has been monitored using Digital Image Correlation (DIC). In parallel, a finite element model has been developed to predict the resistance of the plates, relying on cohesive elements to simulate the failure of the channels subject to internal pressure. Experimental delamination results obtained with mode I double cantilever beam (DCB) test specimens have been used to determine the input parameters for the numerical model.

ABSTRACT. Crystal plasticity (CP) is extensively used to model microstructure-sensitive mechanical response of polycrystalline metals. Fast Fourier Transform (FFT)-based methods are attractive due their higher efficiency compared with CP-Finite Elements, and their direct use of voxelized microstructural images. In this presentation, we will report recent advances in the integration of FFT-based formulations with advanced characterization techniques, both for micromechanical modelling and for improved data reduction. Specifically, we will show: a) applications of the large-strain elasto-viscoplastic FFT-based (LS-EVPFFT) model [1] to interpret micromechanical characterization in nano-metallic laminates, including the formation of localization bands observed in nano-pillar experiments, and b) a novel FFT-based methodology to impose micromechanical constraints to arbitrary voxelized stress fields obtained by x-ray diffraction [2]. The proposed stress filtering method consists in finding the equilibrated stress field closest to a non-equilibrated field, posed as an optimization problem. References: [1] M. Zecevic., R.A, Lebensohn R.A., L. Capolungo, Non-local large-strain FFT-based formulation and its application to interface-dominated plasticity of nano-metallic laminates. JMPS 173, 105187 (2023). [2] H. Zhou, R.A. Lebensohn, P. Reischig, W. Ludwig, K. Bhattacharya, Imposing equilibrium on measured 3-D stress fields using Hodge decomposition and FFT-based optimization. Mechanics of Materials 164, 104109 (2022).

ABSTRACT. In the present work, a coupled chemo-mechanical model for hydrogen embrittlement is developed in a FFT-based environment. The problem is solved using an implicit staggered method. Thus, the three problems (mechanical, damage, and hydrogen diffusion) are solved sequentially until convergence is reached. The mechanical problem is solved for finite strains using the Fourier Galerkin method. In order to account for the fracture behaviour of the material, a phase-field method is implemented. To model the diffusion of hydrogen, the hydrogen population is subdivided into two classes: trapped hydrogen and (mobile) lattice hydrogen. According to Oriani postulate, these two populations must be in equilibrium. The resulting non-linear differential equation is solved for the chemical potential by means of a linearization.

ABSTRACT. The mechanical response of nanostructured metals is influenced by microstructure, characteristic timescales, and lengthscales. Notable examples include single crystalline pillars and nanoporous foams. Computational homogenization tools are useful for linking macroscopic response with microstructural behavior by means of representative volume elements of the system. FFT spectral methods offer improved numerical performance over standard Finite Element based homogenization, thus allowing for detailed RVE models. FFT homogenization for metals relies on the crystal plasticity model, which assumes the homogenization of dislocation ensembles at the nano-scale to define a flow-rule for each slip system. However, at sub-micron scales, the plastic behavior of metals is dominated by few discrete slip events of stochastic nature producing discontinuities along the slip planes involved. Therefore, standard crystal plasticity is not adequate for these sizes. This study introduces a stochastic approach, considering the slip of individual planes, implemented in an FFT-based homogenization software by incorporating each slip event as a shear eigenstrain field. The implementation is initially applied to predict the behavior of tungsten micropillars and then extended to simulate nanoporous tungsten with random bicontinuous open-cell representative volume elements obtained using levelled-wave methods. The model results are compared against experimental observations. It is shown that both macroscopic stress-strain curves and deformation patterns agree with experimental results.

ABSTRACT. The predictive capabilities of crystal plasticity models depend critically on the availability of constitutive parameters that reflect the behavior of the material at hand. The large amount of parameters used in typical crystal plasticity models and the strong non-linear interactions between them renders the identification of these parameters a tedious task. This holds especially for the case of multiphase materials where the contribution of the individual phases to the overall mechanical behavior is hard to untangle. Here, we present results from parameter identification of a phenomenological crystal plasticity model for the hexagonal α-phase of Ti-6Al-4V α+β titanium alloy produced by laser powder bed fusion (LPBF). The lattice strain obtained from in-situ synchrotron diffraction experiments is used as input. An optimization procedure is used to adjust the constitutive parameters of a phenomenological crystal plasticity model until the experimental reference results are reproduced. The performance and robustness of the approach is investigated and the obtained parameters are discussed in comparison to data from the literature.

ABSTRACT. Interpenetrating phase composites (IPCs) based on architected media allow for mechanical properties well beyond the bounds of their constituent phases. The arising mechanical response depends on a series of underlying influential design features, which include the material properties of the phases involved, their architectural design, their volume fraction, as well as loading-related parameters, such as the strain-rate of the loading. In the current contribution, extensive numerical and experimental insights on the dependence of the effective composite material performance on the aforementioned design parameters are provided, for different IPC materials that include polymer and soft matrix phase composites. The data are used as a reference for the development of dedicated tree and deep learning modelling architectures that can, not only accurately capture the effective composite performance, but also be used as surrogate models for subsequent explainability analysis tasks. In particular, dedicated high-accuracy and low computational cost machine learning models are elaborated and employed to assess the significance of the underlying influential design parameters, as well as their interaction, classifying their importance for different base material combinations and loading scenarios.

ABSTRACT. Structural hierarchy in honeycombs is a design scheme that introduces geometrical complexities at different length scales, which in turn influence the mechanical behaviour by altering the unit cell mechanics. One way to realize a hierarchical honeycomb is to replace the monolithic cell walls with sandwich-structured walls with enhanced mass-specific properties. The effective elastic properties of such type of honeycombs, also known as sandwich-structured honeycombs (SSHC), were recently investigated [1], revealing great enhancements in the in-plane stiffness as compared to monolithic honeycombs of equal weight. Building on that knowledge, we present an analytical model of the in-plane compressive strength of SSHCs considering four distinct collapse modes: face yielding, core shear, elastic buckling and face wrinkling. Closed-form expressions of the collapse stress were derived in uniaxial, biaxial and shear loading cases, and collapse mechanism maps were constructed for a wide range of architectural parameters. The latter maps showed dominance of the core shear mode followed by face yielding over a practical design space encompassing a broad range of core and face sheet thickness ratios. Numerical simulations and experimental findings were compared to the results predicted by the analytical model, showing good agreement in terms of collapse load and mechanism. The results also showed that the incorporation of sandwich-structured cell walls is an effective strategy for enhancing the compressive strength of honeycomb structures, reporting enhancements in the collapse stress by factors of 4-20 when compared to conventional honeycombs with monolithic cell walls of equal weight. REFERENCES [1] El Khatib, O., Kumar, S., Cantwell, W.J. and Schiffer, A., 2023. Effective Elastic Properties of Sandwich-Structured Hierarchical Honeycombs: An Analytical Solution. International Journal of Mechanical Sciences, p.108883.

ABSTRACT. Hierarchical materials are complex, multi-scale systems where structural patterns are repeated across scales in a self-similar fashion. Hierarchical microstructural patterns are often credited as a determinant factor in the high fault tolerance of biological and bio-inspired architected materials. For the effective modelling of such meso-scale systems, discrete hierarchical lattice/network models have been used in recent years, where the network constituents are elastic load-carrying elements, subject to a failure criterion. They capture how hierarchical structures redistribute stress over multiple microstructural levels, leading to arrest of crack propagation, the emergence of diffuse damage, and the increase of fault tolerance. In this work, we specifically address the case of interface adhesion and failure/detachment of hierarchically structured thin films in contact with heterogeneous substrates under quasi-static tension. To this end, we introduce a three-dimensional hierarchical network model, where discrete links/elements fail based on a maximum distortion energy (von Mises) criterion with Weibull distributed thresholds, modeling inhomogeneities in local cohesive and adhesive strengths. Element elasticity is modeled in terms of the scalar Random Fuse Model. Our statistical analysis of fracture surfaces indicates that the hierarchical organization is responsible for a substantial enhancement in the localization of damage near the interface. We discuss the origin of this localization phenomenon using concepts of spectral graph theory and multiscaling analysis techniques as well as the influence of hierarchical structuring on performance in terms of peak stress and work of fracture. Further insights into stress redistribution are gained by means of Green functions, an approach that extends beyond these scalar models. While our study of hierarchical fracture and failure is motivated by examples of fibrous materials of biological interest, our results indicate that hierarchical patterns can be useful in engineering scenarios where the focus is on tuning and optimizing adhesion properties or on predetermining locations of failure.

ABSTRACT. Although nanocrystalline metal nitride coatings are currently used for their outstanding properties such as hardness and wear resistance, improving these properties remains a challenge in metalworking industries, to lengthen tool life and increase cutting performance. Metal nitrides like TiN, (Ti,Al)N and metal carbides such as TiC, WC are commonly used. To move towards better performing coatings, one solution consists in depositing TiAl/TiAlN metal/nitride multilayered coatings using a recent sputtering technique (Reactive Gas Pulsing Process RGPP). This process allows to easily modulate the stacking of the coating at the nanometer scale to take advantage of the properties of the nitride, the metal and the interfaces to improve the fracture toughness of the coating while maintaining very high hardness. Ti0.67Al0.33 and Ti0.54Al0.46N (labelled TiAl and TiAlN) monolithic coatings and (TiAl/TiAl)n multilayered coatings were deposited by radio-frequency magnetron sputtering from a single sintered titanium/aluminium target using conventional and RGPP processes with two different periods (10 and 50 nm) respectively. A finite element model reproducing the mechanical behaviour during nanoindentation tests of those as-deposited coatings was developed. Identification of the material’s elasto-platic behaviour of the metal and nitride compounds in the nano-stacking was performed using experimental indentation of TiAl and TiAlN monolithic coatings for the elastic part. The plastic part was determined by finite element model updating of dual Berkovich and cube corner indentations with an experiment design guided by an identifiability index. The stacking of the multilayered coatings was specified by N-K-edge Electron Energy-Loss Spectroscopy to introduce in the numerical model an interface between each metal and nitride layers. The elasto-plastic properties of these interface layers were considered as a rule of mixtures of the metal and nitride properties using two hypotheses: parallel or serial. We proved that the properties of the interface layers are intermediate between the two hypotheses.

ABSTRACT. In recent decades, the engineering field has been focused on finding novel materials and structural configurations that can enhance mechanical performance while reducing costs. The goal is to develop structures that are stronger, lighter, and more resilient to meet specific load conditions and boundary requirements. Additive manufacturing has opened up new possibilities, enabling the investigation of novel geometric configurations and leading to the emerging field of metamaterials. One kind of those metamaterials are the auxetic structures, which are characterized by a negative Poisson's ratio. They exhibit exceptional shear strength, indentation resistance, high fracture toughness, and significant energy dissipation capacity. These properties make them suitable for various purposes, including shock absorbers, filters, biomedical implants, and protective elements. Understanding the behaviour of auxetic structures under different types of loading is crucial for their development and the design of different applications. This understanding relies on the fundamental 'unit cell' of auxetic structures, which determines their mechanical characteristics. The objective of this study is to investigate the effect of varying unit cell parameters on the mechanical behaviour of polymer-based auxetic structures, manufactured using a stereolithography (SLA) 3D printer. To that end, quasi-static compression tests were performed on auxetic structures with different beam lengths, cross-sectional thicknesses and re-entrant angles of the unit cell. Furthermore, the effect of the number of cells within the structure was examined. The data and images obtained during the tests allowed to analyse the maximum crushing force, the energy absorption capacity, the auxetic behaviour or the failure mode of the structures and hence to compare the effectiveness of studied structures. The conclusions reached contribute to the comprehension of the behavior of auxetic structures and their potential for energy absorption across diverse structural designs.

ABSTRACT. High Pressure Die Casting (HPDC) using Aluminum (Al) has emerged as an option for producing parts with high mechanical strength, high geometric precision and low weight for the automotive industry. In this type of process, injection molds are subjected to pressure and temperature cycles, suffering high thermal fatigue in their structure, especially in inserts exposed to the Al flow. In this sense, many scenarios require the definition of a minimum number of cycles to shape cavity components. Some traditional materials are used in injection molds for this process and the use of processes such as Additive Manufacturing (AM) to manufacture components for this type of mold, although it is an interesting option, presents some issues that must be considered. This work presents a study of the application of the additive manufacturing process in the manufacture of mold inserts for Aluminum HPDC in the automotive industry. To analyze the performance of the insert produced by AM, a set of inserts was also produced using the conventional manufacturing process, that is, machining and using traditional steel. These inserts were tested in an HPDC production mold and were periodically replaced after each number of cycles defined by the company. The inserts manufactured by the AM process were produced in DIN 1.2709 steel (Maraging 300) and underwent subsequent machining, heat treatment and surface coating processes. The costs involved in AM and conventional manufacturing were compared and field tests are being carried out at the partner company with the aim of validating the option of applying AM in aluminum injection molds. AM inserts have so far been used in up to ten thousand cycles, this number is the average use of inserts produced with traditional materials, which validates the use of AM in this condition without identifying apparent thermal cracks.

ABSTRACT. Current state of the art in L-PBF printing has demonstrated a dependance of the parts density to the layer thickness being employed. This fact limits the achievable productivity of the technology due to very thin layers being needed for consolidation. In addition, the variability sources of the printing equipment, including sub-systems proximity, part region or thermal history can lead to a heterogeneous defect distribution within the parts, which are more prominent in the case of thicker layers. The objective of the present work is to develop novel adaptative printing strategies that are able to increase the quality of AlSi10Mg parts printed with 90 µm layer thickness. For this purpose, a quantification of the defects encountered through conventional printing strategies is made through microstructure and porosity characterization. In addition, process monitoring tools are employed to understand the energy being deposited. As a result, a correlation between printing parameters and part’s defects is made. Finally, and based on the observed results, an alternative 3D printing strategy consisting of variable printing parameters across the part is proposed and assessed.

ABSTRACT. Aluminum alloys are gaining significant attention in additive manufacturing. However, most of available alloys primarily target Laser Powder Bed Fusion (L-PBF) applications i.e., they are optimized to perform exceptionally well under the high solidification and cooling rates characteristic of this additive manufacturing technique. When these alloys are employed in Direct Energy Deposition (DED) applications, such as LMD, WAAM and WLAM, they often fall short in terms of mechanical performance due to the relatively lower cooling rates compared to L-PBF. In this study, we present the development of novel Al-Mg-Zn alloys specifically designed for DED applications, both for additive printing and repair purposes. The compatibility of these candidate alloys with DED was rapid remelting on cast ingots without the need to atomize all of them. The selected alloy, which exhibits resistance to solidification cracking and a refined microstructure, underwent testing within the LMD process. Initially, a parameterization study was carried out, revealing that the development of porosity was largely due to the hydrogen content in the initial powder. The use of a dry powder yielded densities exceeding 99.9%. This new material was subsequently characterized for both bulk printing and repair applications. Despite the loss of Zn, the printed coupons exhibited low anisotropy, a yield strength of 400 MPa, and a total elongation exceeding 4%. Even in the as-built state, the achieved yield strength surpassed that of LMDed Scalmalloy after the corresponding heat treatment. The research on repairing pre-existing 7075 alloy structures focused on (i) the deposition parameters, (ii) parameters to avoid liquation cracking at the interface, and (iii) the development of an in-situ laser heat treatment to increase the strength in the refurbished zone without the need to fully re-heat treat the entire component. Tensile testing across the interface was used to assess the mechanical properties of the various repair approaches.

ABSTRACT. The cheese of Swiss origin “Tete de moine” is generally cut by scraping its surface with a sharp tool attached to a shaft. Scraping the cheese produces thin sheets of cheese that are strongly wrinkled at the edge. In this work we experimentally demonstrate that these wrinkles are produced by the change of mechanical properties in the radial direction of the cheese and the scraping process itself, establishing an analog between our cheese scraping process and mechanical machining widely used in the industry. To do this, we built an experimental system that allows us to carry out the scraping process systematically, for controlled load levels on the cutting tool. We found that the wrinkle formation process is strictly plastic and that due to the change in mechanical properties in the radial direction, it is reflected in a change in the compression ratio in the radial direction of the cut cheese sheet. Additionally, we independently measure the surface friction and the elastic limit of the material, to explain from mechanical parameters the change in compression ratio found when producing the cheese sheets and show that the process is strictly plastic.

ABSTRACT. Indented circular thin sheets floating on water can exhibit wrinkles induced by radial compression at their edge. At low indentation depth, wrinkles cover an annulus but can cover the whole surface for more pronounced indentation. On a highly viscous fluid, the indentation force and the growth of wrinkles are affected by the dynamics. For instance, new wrinkles, much smaller than the static ones, are observed. This wavelength is determined by the viscous force and more classical parameters such as bending modulus. At sufficiently short time, wrinkles are radially uniform and the problem can be solved by considering a linear compression of a rectangular sheet. We discuss the impact of viscous force on the initial wavelength and the time relaxation of those unstable wrinkles by solving the Reynold equation combined with the beam equation.

ABSTRACT. We study the mechanical and geometrical behavior of two ribbons that are joined together at one extremity in the form of a dowsing Y-rod. The ribbons are pulled apart at their free ends in opposite directions. We consider the angle $\theta$ made by the connected end with the normal to the direction of the pulling forces.

If both ribbons are identical, the connected end is oriented normal to the pulling direction, $theta=0$. Breaking the symmetry of the system with different bending stiffnesses modifies that angle, with the joined end pointing towards the pulling force on the weaker ribbon. Surprisingly, over a wide range of forces, this angle is independent of the load and is a function of the stiffness asymmetry alone.

We rationalize this observation with a boundary layer analysis in the framework of two coupled Kirchhoff beams. The analytical solution found for the shape of the boundary layer allows us to use this simple test as a quite accurate measurement of relative bending stiffness. We present this model, as well as its limits for forces large enough to violate hypotheses, introducing either plasticity or three-dimensional effects in the boundary layer.

Finally, by allowing an inhomogeneous cross-section of one of the ribbons along its length, and thus a curvilinear variation of its bending stiffness, we prevent the angle from being load-independent. The dependence of the angle $\theta$ with the applied force can be chosen by figuring out the corresponding spatial evolution of the beam’s cross-section. We formulate this inverse problem in the framework of the tapered Elastica, and present a direct application to a visual linear force sensor made with two thin pieces of Mylar and one piece of double-sided tape.

ABSTRACT. Fracture propagation in brittle thin sheet often follows very reproducible fracture paths. Since our schoolyear we all intuitively know how to cut of a sheet of paper laid on a table with a ruler A description of this mundane situation is however difficult because of the incompatibility in fundamental assumptions for thin plate mechanics (averaging stress field along the thickness) and Fracture mechanics (requiring a detailed description of the divergent stress field close the crack tip). I will show how a variational approach can help explain the robustness and the geometry of torn fracture trajectories, and rationalise the controlled cutting of a paper sheet so that the cut follows the clamping ruler.

ABSTRACT. Magnetorheological elastomers (MREs) are two-phase composite materials obtained by dispersing metallic magnetic particles in a soft elastomer matrix. The elastomer matrix can be further modified to be a foam, thereby yielding a three-phase porous material exhibiting millimeter-sized voids. Such magnetorheological foams, akin to MREs, show variable stiffness and can be deformed upon application of a magnetic field, but they have the additional advantage to be even more lightweight than MREs. This study investigates the fabrication of magnetorheological foams and addresses their characterization. In particular, their characterization involves the analysis of pore structure and distribution, alongside with the calculation of porosity and particle filling factor, as a function of fabrication parameters. Finally, the magneto-mechanical behavior of the obtained magnetorheological foams is assessed via a dedicated experimental setup.

ABSTRACT. This research proposes a novel multidisciplinary framework for designing and controlling robotically actuated elastic shape-changing material systems, defined as Elastic Robotic Structures (ERS). ERS encompass a wide range of hybrid structures combining bending, tensile and inflatable elements that are mechanically and pneumatically actuated. ERS are designed to operate at human scale for various design applications, including adaptive building systems, creative robotics and interactive objects. The goal is to develop everyday intelligent systems capable of interacting with humans and responding to different parameters.

Elastic materials’ capacity to undertake large deformations under different load conditions makes them inherently adaptive. However, their non-linear behaviour makes these systems challenging to predict. Given the complexity of designing continuously operating elastic systems at human scale, ERS research sits at the intersection of architecture, engineering and robotics. Recent advancements in computational and numerical methods have enhanced the design process, facilitating the creation of complex, structurally efficient elastic structures with significant design potential. However, the lack of methods for controlling continuously operating systems means that most of these structures remain static or display limited changes.

Shape-changing elastic systems, explored in various engineering fields like soft robotics, human-computer interaction and structural mechanics, often face limitations in terms of scale and shape diversity, driven by the specialised approach of engineering applications. Soft robotics offers solutions that can be implemented for the control of elastic shape changing systems with complex shapes and continuously operating systems. The ERS framework integrates methods used to design, characterise and control soft robots with simulation and design approaches used in architectural design and structural engineering. The work offers an overview of these approaches, illustrating how they were used to design different ERS. It also aims to be a guide for the design of similar systems.

ABSTRACT. This work presents a comprehensive exploration of interface damage in engineering materials characterized by an anisotropic/isotropic phase combination. Examples include twins impinging on a grain boundary, crystalline-amorphous interface or martensite-ferrite interface in advanced multi-phase steels, where martensite islands typically deform by sliding on the retained austenite films. In these cases the anisotropic deformation of one phase, when favourably oriented, induces interface damage through a jagged deformation mechanism, complementing the classical cohesive interface deformation mechanism. To capture these two microscopic mechanisms, i.e. jagged damage and cohesive opening, at the mesoscale, a novel computational homogenization framework is developed. The mesostructure, comprising multiple anisotropic particles in an isotropic matrix, is modelled with interfaces represented by enriched cohesive zones. The microscopic interfacial zone unit cell resolves the laminated structure of the anisotropic phase, defining effective interface separation and internal kinematic quantities associated with jagged and cohesive deformation mechanisms. The generalized Hill-Mandel condition yields tractions work-conjugated to these internal kinematic quantities, leading to a mesoscale enriched cohesive law identified through representative microscopic unit cell simulations. Comparison to a fully resolved model of mesoscopic interfacial zones demonstrates the efficacy of the developed enriched cohesive interface model, emphasizing the importance of considering both microscopic mechanisms in predicting anisotropic/isotropic interface separation and overall material failure. The model is applied to a dual-phase steel microstructure, where the model performance is validated against microscale experimental results. The proposed microphysics-based effective interface model provides a valuable tool for understanding and predicting interface damage in complex materials.

ABSTRACT. The talk that delves into the origins of fracture toughness in amorphous silica, focusing on the influence of a rounded crack tip and the limitations of linear elastic fracture mechanics. Griffith's theory states that in the absence of a sharp crack, the energy release rate becomes zero, rendering linear elastic fracture mechanics inadequate for assessing resistance in geometries featuring a rounded crack. To overcome this limitation, the talk employs coupled criterion and phase-field simulations to assess fracture initiation.

Through extensive large-scale atomic-scale simulations, the research identifies damage within the atomic structure. A finite element model update scheme is utilized to pinpoint the critical energy release rate and the regularization length scale during crack propagation.

In conclusion, the talk offers a comprehensive comparison of the identified properties, examining their consistency with the predictions of the homogeneous phase-field solution and the material's tensile strength. Furthermore, the research endeavors to contrast the outcomes of all three methodologies, thereby shedding light on the foundational assumptions that underlie continuum models, including phase-field and finite fracture mechanics. By exploring the interplay between crack geometry and fracture resistance, this study advances our understanding of fracture toughness in amorphous silica and contributes to the ongoing development of more accurate models for predicting fracture initiation and propagation in complex materials.

ABSTRACT. Linear elastic fracture mechanics (LEFM) theory has long postulated that the speed of crack growth is constrained by the Rayleigh wave speed. While numerous experimental and numerical studies have generally supported this prediction, some exceptions have raised questions about its validity and the underlying factors influencing dynamic crack behavior. In this work, we present new numerical results showing that tensile (mode I) cracks can surpass the Rayleigh wave speed and exhibit propagation at supershear velocities. The key to this finding lies in incorporating geometric non-linearities into the material model. While such non-linearities are inherent in most materials, their effects on dynamic fracture growth have been largely overlooked in previous work. Our results reveal that accounting for geometric non-linearities is sufficient to enable supershear crack propagation. In addition, we show that these non-linearities induce modifications in the crack-tip singularity, leading to unconventional crack-tip opening displacements, cohesive zone behavior, and altered energy flow dynamics toward the crack tip. These observations suggest that the elastic fields and energy budgeting in the vicinity of the crack tip of geometrically non-linear materials have a completely different behavior than that of linear elastic materials, which is commonly assumed in LEFM theory. Consequently, this provides a novel perspective on dynamic crack growth that challenges existing theoretical frameworks.

ABSTRACT. In this work, localizing gradient damage model is combined with smoothed finite element method (SFEM) to investigate the fracture behavior of quasi-brittle materials. The cell-based strain smoothing approach is considered over the domain to convert the classical domain integration to line integration along the each boundary of the smoothing cell, which eliminates the requirement of derivatives of shape functions in the computation of field gradients. The gradient damage framework uses the stress-based evolving anisotropic nonlocal interaction domain, which helps to maintain the localized damage bandwidth during later stages of loading and overcomes the limitations associated with conventional gradient damage models. The anisotropy in nonlocal interactions is modelled using an anisotropic gradient tensor, which governs the orientation of the nonlocal interaction domain based on the principal stress state at a given material point. The stress tensor obtained through SFEM is utilized for determining the principal stress state, to enforce a properly oriented interaction across the bandwidth of the damage process zone throughout the loading process. The proposed framework is extended to simulate the standard fracture mechanics problems of mode-I, mode-II and mixed mode. The obtained results are compared with the traditional FEM counterpart and literature. The comparison of results clearly shows improvement in the computational efficacy over the traditional FEM, and also shows a great potential of it for simulating large deformation problems where element distortion is a critical issue.

ABSTRACT. The development of architected metallic metamaterials produced through additive manufacturing has introduced a vast array of properties beyond those observed in traditional bulk alloys called to revolutionise multiple critical sectors like the aerospace or biomedical. However, the intricate geometries and high surface-to-volume ratio inherent in these architected metamaterials, coupled with the surface characteristics derived from the additive manufacturing (AM) process, give rise to a more complex fatigue behavior when compared to conventional bulk alloys. This complexity poses a significant challenge in the technological application of architected AM materials.

This research addresses this issue through a systematic multiscale study that investigates the interconnection between metamaterial design, defects, and fatigue behavior through the combination of experimentation and modelling. The base material used is a commercial aluminum alloy, AlSi10Mg, processed through selective laser melting. The study employs a combination of fatigue experimentation, computational modeling, and defect identification to analyze the impact of processing conditions and design geometry on microstructural defects and surface quality. By systematically connecting these factors with the fatigue life of the metamaterials, the research aims to provide insights that can help mitigate the challenges associated with the fatigue behavior of metamaterials. The results show the importance of neglecting at the design stage the strut geometrical deviations, broken struts, and porosity in the overestimation of the fatigue performance and how these defects are produced by large unsupported elements and small feature sizes. Ultimately, this work seeks to set the path to enhance the reliability and performance of architected metallic metamaterials, addressing concerns and advancing their potential for use in various technological applications.

Grant PID2020-116440RA-I00 funded by MICIU/AEI/10.13039/501100011033.

ABSTRACT. The investigated study proposes a mechanical design tool that it can be used for fatigue life prediction. The fatigue life estimation of structures under multiaxial cycle’s loadings on FGS cast iron was simulated using the 3D finite element method. A CAD Modeler, which predicts the admissible residual stress using fatigue criteria, was used to assure the mechanism security in the integrated design phase. A new approach to simultaneous engineering was applied using the components mechanical design with considerate the residual stress. The admissible residual stress calculate after one both a loading and a cycles number is conducted on FGS cast iron specimens, and the fatigue cracks initiation zones as well as the proposed models shall be compared with those obtained from experimental fatigue tests, where satisfactory prediction capabilities on both the fatigue crack initiation locations, which it needs to be introduced by different treatments. The fatigue life of the model is demonstrated

ABSTRACT. While the interaction between 2D materials and cells is of key importance to the development of nanomedicines and safe applications of nanotechnology, still little is known about the biological interactions of many emerging 2D materials. Here, an investigation of how hexagonal boron nitride (hBN) interacts with the cell membrane is carried out by combining molecular dynamics (MD), liquid-phase exfoliation, and in vitro imaging methods. MD simulations reveal that a sharp hBN wedge can penetrate a lipid bilayer and form a cross-membrane water channel along its exposed polar edges, while a round hBN sheet does not exhibit this behavior. It is hypothesized that such water channels can facilitate cross-membrane transport, with important consequences including lysosomal membrane permeabilization, an emerging mechanism of cellular toxicity that involves the release of cathepsin B and generation of radical oxygen species leading to cell apoptosis. To test this hypothesis, two types of hBN nanosheets, one with a rhomboidal, cornered morphology and one with a round morphology, are prepared, and human lung epithelial cells are exposed to both materials. The cornered hBN with lateral polar edges results in a dose-dependent cytotoxic effect, whereas round hBN does not cause significant toxicity, thus confirming our premise.

ABSTRACT. Falls among the elderly cause a huge number of hip fractures worldwide. Energy absorbing floors (EAFs) represent a promising strategy to decrease impact force and hip fracture risk during falls among the elderly. However, little is known about how the design features of floors influence biomechanical effectiveness. A high biofidelity while-body finite element model (THUMS) with updated soft tissue near the hip region was implemented in this study, to measure the attenuation in femoral neck force provided by four commercially available EAFs. The simulations were performed at the most serious sideways falling posture for a small-size elderly woman with a -10◦ trunk angle and +10◦ anterior pelvis rotation. At a pelvis impact velocity of 3 m/s, the peak force attenuation provided by four EAFs ranged between 4% and 20%. Similarly, the risk of hip fractures also demonstrates a comparable attenuation range. The results also exhibited that floors with higher internal energy during impact demonstrated greater force attenuation in sideways falls. By comparing the protective performance of existing EAFs, these results can improve the floor design that offers better performance and clinical effectiveness in high-fall-risk environments for the elderly.

ABSTRACT. Biological systems often create materials with intricate structures to achieve specialized functions. In comparison, precise control of structures in man-made materials has been challenging. Here, we report a serendipitous discovery of insect cuticle peptides (ICPs) spontaneously forming nanocapsules through a single-step solvent exchange process, where the concentration gradient resulting from mixing of water and acetone drives the localization and self-assembly of the peptides into hollow nanocapsules. The underlying driving force is the intrinsic affinity of the peptides for a particular solvent concentration, while the diffusion of water and acetone creates a gradient interface that triggers peptide localization and self-assembly. This gradient-mediated self-assembly offers a transformative pathway towards next-generation drug delivery systems based on peptide nanocapsules.

ABSTRACT. We investigated the stab resistance of Bombyx mori silk cocoons to authenticate their purpose as natural material structures with an inherent protective functionality. The cocoon wall is a nonwoven fibroin-sericin composite material architected by strands of continuous silk fibres, which are known to be tough and resilient. Both full ovaloid cocoons, and rectangular flattened specimens of the cocoon wall, were subjected to both static and dynamic stab tests. The blades used for stab testing met HOSDB standards, which using a "Home Office Engineered Knife (01-07164)". Three scenarios were considered: (1) stab tests through the entire ovaloid cocoon, (2) stab tests through only the first wall of the ovaloid cocoon, and (3) stab tests through the flattened rectangular cocoon wall cut from the cocoon. The final test scenario was special as it allowed for direct measurement of the knife penetration depth. Our study examined the structural components of the cocoon in providing stab resistance. We discuss stab resistance from tow perspectives: (a) the micromechanism of stab resistance as a function of silk fibre arrangement and (b) the macromechanisms of stab resistance as a function of cocoon geometry. Our work provides new insights into the inherent hierarchical protective characteristics of silk fibre architectures.

ABSTRACT. Plasma membranes form the physical barrier that separates the intracellular contents from the extracellular environments and are essential to the homeostasis of the cell. It has been identified that lipid peroxidation is associated with multiple pathological conditions, including neurodegenerative diseases, atherosclerosis, diabetes, preeclampsia, aging, cancer, etc. However, how exactly lipid peroxidation contributes to those pathological conditions remains largely elusive. Especially, contradictory results regarding its regulation of membrane properties have been reported in experiments. In this study, we first performed molecular dynamics simulations to systematically investigate how lipid peroxidation modulates the structural and mechanical properties of planar lipid membranes. We then extended our simulation systems to nanosized lipid vesicles to interrogate how the change in membrane mechanics caused by lipid peroxidation might alter the vesicle shape. Our results reveal that lipid peroxidation modulates lipid bilayer mechanics in a peroxidation site-specific manner: peroxidation at sites in the bilayer interior disturbs and softens the membrane, whereas peroxidation at sites near the membrane-water interface results in a more ordered and stiffer membrane. Interestingly, we find that the increase in the area per lipid as a result of lipid peroxidation at sites in the bilayer interior leads to a dramatic shape transformation of nanosized vesicles from spherical to prolate. Our finding is essential to a more accurate understanding of the initiation of lipid peroxidation-induced downstream biochemical events in various pathological conditions. In addition, the lipid peroxidation-induced vesicle shape transformation suggests a novel way to fabricate non-spherical nanoliposomes that may serve as a more efficient drug delivery system.

ABSTRACT. Hydrogen is today mainly obtained by hydrocarbon steam reforming, which produces large CO2 quantities. Because of the rising concern about greenhouse gas emissions, the widespread use of hydrogen as an energy carrier requires the development of a carbon-free production chain. In case it makes use of renewable electricity sources, hydrogen production by electrolysis may be the key to trigger the expansion of this promising sector. However, only a few percent of the total hydrogen production comes today from water electrolysis, mainly because of its cost, which is about four times higher than the cost of hydrogen obtained by steam reforming. Electrolysis requires an electrocatalyst, typically platinum, which is rare and expensive. Electrolytic production of H2 is thus handicapped by its dependence on platinum and by the adverse role played by the hydrogen bubbles produced at the electrode surface in the hydrogen production itself. It is therefore crucial, in order to minimize the cost and energy losses, to avoid materials like platinum as much as possible, and to limit the adverse effects of bubbles production. It has already been demonstrated that elastic strains can modulate the electrocatalytic activity of metals [1-3], so that more abundant materials could be strained in order to compare with platinum in terms of electrocatalytic activity. The question of the optimal position in the 6-dimensional strain space is however open, and we propose an experimental approach based on an original imaging technique to address this issue. [1] Mavrikakis et al. Phys. Rev. Lett. 81 (13), 1998. [2] Kibler et al. Angew. Chem. Int. Ed. 44, 2005. [3] Martinez-Alonso et al. PCCP 24, 2022.

ABSTRACT. Composed of an aggregate of grains with different sizes and local orientations, the deformation of polycrystalline metals exhibits heterogeneity at the microstructure scale under loading. The investigation of this heterogeneous localization has been greatly facilitated by the development of full-field measurement techniques such as Digital Image Correlation (DIC). However, the measurement of stress fields still remains an open problem. To access the local stress, some methods have been reported, such as inverse methods or data-driven identification based on DIC measurements, which require extensive simulation work or mathematical calculations or thermomechanical coupling measurements. In this study, a more straightforward method is proposed to estimate the local stress distribution in 316L austenitic polycrystalline stainless steel using the Ludwik hardening model. This approach requires the identification of local elastic limits and hardening strength coefficients. For this purpose, we first develop a strategy to identify the moments of plasticity onset based on strain evolution measured by DIC at the grain-scale, where all quantities are averaged within each grain. Assuming isotropic material properties, the grain-scale elastic limits are then identified using Hooke's law with the strain information at plasticity onset. Moreover, based on the qualitative relationship between the hardening strength coefficient and the plastic strain rate, another strategy is proposed to identify the grain-scale strength coefficients. A correlation between grain-scale von Mises strain and strength coefficients has been established. With the identified elastic limits and strength coefficients, the grain-scale stress fields are estimated, and their averaged values align well with the engineering ones, validating the reliability of the proposed approach. The strategies for elastic limits and hardening strength coefficients can be further extended to each measurement point, allowing the estimation of microscale stresses that highlight intragranular heterogeneities in the stress distribution.

ABSTRACT. In experimental solid mechanics, measurement of stress is rather associated with diffraction techniques that inherently measure lattice (elastic) strain. Recently, with the microscopic applications of techniques like digital image correlation (DIC), however, extensive information is gathered on a polycrystal with inherently higher spatial resolution and continuous material coverage with strain mapping. As real space techniques, these inherently measure total strain, which is quickly dominated by plastic strain after loading. Nevertheless further, DIC rarely has the sensitivity to explore elastic strains in common metals. With DIC over a polycrystal with intragranular resolution, the sensitivity can be increased by local averaging, with the averaging domains functionally chosen as grains. Thus, grain-averaged strain sensitivity can, in principal, reach the elastic strain orders. DIC, however, can only elaborate elastic strains if they are equal to total strains. Once a polycrystal is plastically strained, this state is uniquely realized over unloads. Hence, here, we will employ a functional unload over such a polycrystal, measuring the in-plane components of elastic strain release with an area-scanning in situ variant of microscopic DIC [1]. For the stress interpretation of these values (after grain-averaging) for each grain, their orientations are predetermined by electron backscatter diffraction. Pure FCC Nickel (with anisotropy ratio 2.5) is selected for this fundamental study. Crucially, the in situ measurement capability is utilized to conduct DIC imaging at multiple points over the unload to evaluate the linearity of the response. A formalism is presented that elaborates the nature of local stress information that is available through a macroscopic unload. [1] N.A. Özdür, I.B. Üçel, J. Yang, C.C. Aydıner, Exp Mech 61 (2021) 499–514.

ABSTRACT. The usage of engineering polymers, including rubbers, spans a wide range of applications in various industries, from aerospace to medicine. The comprehension of rubber elasticity, evolving from Hooke's law in the 1660s to Flory's ground-breaking work on polymer chains in the 1930s and 1940s (earning him the 1974 Nobel Prize), culminated in the formalization of understanding through the Neo-Hookean model in the 1940s. This model established the notion that, under isothermal conditions, stress is (non)linearly correlated with strain, with no consideration for other state variables. However, our research, utilizing an innovative X-ray tomography method, challenges this foundational concept. We perform in-situ experimentation in X-ray tomography to present evidence that prompts a re-evaluation of the established understanding. Our findings reveal the presence of a mobile un-crosslinked phase within rubbers and various engineering polymers, such as Nylon and HDPE, introducing an additional state variable into the equation. Consequently, spatial variations of stress (or strain) emerge as influential factors in determining their mechanical behaviour. This challenges the prevailing belief that the material properties of polymers can be solely defined on a local scale.

ABSTRACT. The versatile residual-stress-actuated on-chip tensile testing method has proven over the last years its potential for the characterization of the link between the mechanical behaviour and microstructure at the micro- and nanoscale. Recently, the on-chip method has appeared promising to tackle challenges related to strain engineering studies. Strain engineering studies refer essentially to the exploration of the evolution of functional properties under elastic distortions. Nano- and micro-scale specimens are particularly suited as they usually are stronger than macroscopic specimens and thus allow larger elastic strain without fracture. The difficulty is to impose well-defined mechanical conditions at small scale. Two practical examples are presented in this work.

First, experimental validation of theoretical band gap calculations (first-principles calculations based on density functional theory) is provided through photoluminescence spectroscopy measurements performed on on-chip microfabricated monocrystalline Si specimens deformed up to ∼1% under uniaxial tensile stress.

Second, while uniaxial tension is highly relevant to determine several mechanical properties, the first-principles calculations have highlighted the interest of biaxial and shear loading configurations to improve more efficiently the band structure of silicon. On-chip shear and biaxial configurations have thus been designed. Experimental proof of concept of these new structures is provided on monocrystalline silicon.

ABSTRACT. Small-scale metals have been extensively studied under quasistatic conditions. In recent years, there has been an effort to increase strain rates in micro/nanomechanical tests, but they are mostly limited to impact experiments. The acquisition of quantitative stress-strain signatures of micro/nanoscale metals under extreme mechanical loading conditions, e.g., strain rates beyond 10^3/s, remains unexplored. Moreover, simulation tools such as molecular dynamics (MD), which allow the visualization of internal features in materials and their evolution in silico can only be performed at very high strain rates: 10^6/s-10^9/s. Advanced experiments closer to these extreme strain rates could validate computational findings and further enhance the understanding of how size, defect density, temperature, and strain rate influence the deformation mechanisms. We report for the first time in situ mechanical testing performed on nickel microparticles at strain rates up to 10^4/s, coupled with MD simulations on particles with corresponding shapes, closing the strain rate gap to two orders of magnitude. Well-defined, defect-free microparticles with the expected equilibrated Wulff shape were obtained by annealing nickel thin films via solid-state dewetting. The microparticles were compressed at strain rates ranging from 10^-3 to 10^4/s at room temperature and between 10^-3 and 10/s at cryogenic temperatures. Based on the stress-strain signatures, thermal activation analysis, and pre-and post-test microstructural characterization, together with near-direct comparison to MD simulations, an extensive deformation map of pristine microscale nickel as a function of temperature and strain rate has been obtained. The properties of the microparticles reveal a convolution between size effect, strain rate effect and temperature. We have a transition in strain rate sensitivity that increases 10-fold beyond 10/s, and this transition is even stronger in smaller particles. Finally, based on the low activation volumes ascertained ~10b^3 from experiments and the MD results, surface nucleation of defects is suggested as the dominating main deformation mechanism.

ABSTRACT. Elastic strains can be used to modify adsorption energy barriers (and, thus, the catalytic activity) of surfaces. This strategy is used here to modify the catalytic activity of Au thin films for the Hydrogen Evolution Reaction (HER) and the Oxygen Reduction Reaction (ORR), that are key processes to produce hydrogen by water splitting and the generation of energy from hydrogen in fuel cells.

Au thin films of 100 nm thick and a strong <111> texture were manufactured by DC magnetron sputtering on Si, polyamide, and NiTi substrates. Prior to deposition, the shape memory NiTi substrates with martensitic microstructure were loaded in four-point bending at ambient temperature. The Au thin films were deposited on the surfaces deformed in tension or compression, and the substrate/thin film was then heated above the phase transition temperature. The NiTi substrate was transformed to the austenitic phase, recovering the initial shape (before four-point bending loading), and transferring the elastic strains to the Au film.

The microstructure of the Au thin films was analyzed by means of atomic force and transmission electron microscopy and the residual strains induced in the film were assessed using X-ray diffraction. The electrocatalytic activity of the thin films with different residual strains was determined in acidic media towards the HER and ORR using linear sweep voltammetry and cyclic voltammetry. Accordingly, improved electrocatalytic performance (i.e., smaller reaction overpotential and smaller Tafel slopes) was observed in the presence of tensile strains, while compressive strains led to opposite behavior (particularly for HER). This is in full agreement with theoretical predictions based on density functional theory. Furthermore, the impact of the surface morphology on the catalytic activity is studied.

ABSTRACT. A technique for the highly localised measurement of residual stresses at the surface of metallic samples is presented in this paper. Residual stresses are relaxed by removing a small volume of material via focused ion beam (FIB) milling. The new free surfaces of the milled geometry produce displacements in the neighbouring material, which generates micro-strains. These strains are detected by applying digital image correlation (DIC) algorithms to images captured using a scanning electron microscope and related to residual stresses via finite element simulations. Owing to the material removal and strain detection methods it uses, this technique is named FIB-DIC. The spatial resolution of the technique is of the order of tens of microns, and the sensitivity of the displacement measurement of the order of tens of nanometres. Previous measurements using this technique have been focused on validating the technique, by comparing its results satisfactorily to X-ray diffraction measurements. In this work, the residual stress distribution generated at the surface of an aluminium sample during face turning is characterised. Process parameters are related to microscale residual stress patterns, and local measurements using FIB-DIC are compared with the standard X-ray diffraction residual stress measurements.

ABSTRACT. This article deals with moist rigid porous media and studies the propagation of long acoustic waves in presence of evaporation/condensation. At equilibrium, the solid walls are covered by a thin water film and water vapour in the air is at its equilibrium-temperature-dependent saturation pressure. Under acoustic excitation, water vaporises or condenses. The reversibility of this phase change and the assumed small acoustic harmonic perturbation leads to a linear problem where the usual local poro-acoustics physics (Fourier-Stokes system) is enriched with the (i) Clapeyron relation (CR) which links the liquid-wall temperature, the vapour pressure, and the latent heat of vaporisation, (ii) heat transfer of the latent heat in the solid, (iii)diffusion equation for water vapour in air, (iv) water vapour’s equation of state.

Using the two-scale homogenisation method, it is first shown that the gas pressure is locally constant. Then, the method yields the usual visco-inertial Darcy’s law and the mass and heat fluxes associated with the phase change. The latter are obtained from a coupled thermo-diffusion problem with specific boundary conditions at the gas/liquid interface : (i) liquid-wall temperature imposed by CR and (ii) discontinuity in heat flux due to water phase change induced heat flux. Finally, the gas effective compressibility is derived from the overall mass conservation of dry air and water vapour in the pores. The role of dimensionless parameters and characteristic frequencies on the speed of sound and attenuation is also discussed with respect to the equilibrium temperature.

ABSTRACT. The issue of void stability in multiply twinned particles is reconsidered in terms of thermodynamics of irreversible processes. The evolution equation of the particles is derived according to the principle of maximum dissipation of non-equilibrium thermodynamics [1] to provide the kinetic model of the strain-induced void evolution. It is demonstrated that the relaxation of the surface and strain energies aroused by either void growth or shrinkage strongly determines the stability of hollow multiply twinned particles. Particularly, the void tends either to reach equilibrium state or shrink with subsequent collapse in dependence to initial conditions and material moduli. Analysis of void evolution modes are performed to reveal the critical and optimal parameters of this process. The latter ones are in good agreement with available data on the experimental investigations of the hollow multiply twinned particles synthesized by different methods [2]. Acknowledgements. This work was supported by the Russian Science Foundation (grant No. 22-11-00087, https://rscf.ru/en/project/22-11-00087/). Reference [1] Fischer F.D., Svoboda J. High temperature instability of hollow nanoparticles. J Nanopart Res, 2008, 10, 255-261. [2] Yasnikov I.S., Vikarchuk A.A. Voids in icosahedral small particles of an electrolytic metal. JETP letters, 2006, 83, 42-45.

ABSTRACT. The presentation compares two approaches for modeling the inelastic behavior of foam structures. The first approach is classical in nature, where analytical representations for the yield surface and the yield potential are used in a thermodynamically consistent modeling concept. The difficulty here is to formulate the analytical approaches. By adapting the formulation for a yield surface by Ehlers, which was originally developed for geomaterials, a very good agreement with numerically determined yield surfaces could be achieved. The parameters of the analytical yield surface can be represented as functions depending on a hardening variable. The second approach is purely data-driven and uses approximations such as bivariate splines or neural networks for the representation of yield surface and yield potential. The neural networks used are so-called feed forward networks and are used as universal approximators or regressors. The data required for the training of the neural network or the calibration of the analytical model are obtained from finite element simulations of representative volume elements of generic foam structures, whereby the stress-controlled load paths are systematically varied. The approaches used are compared in terms of their suitability, flexibility and accuracy for describing the inelastic behavior of foam structures.

ABSTRACT. Granular materials is a particular example of a porous material seen as a complex system. While individual grains can be modeled as solid bodies in interaction through contact forces, the large number of grains involved at the engineering scale result in a collective complex behavior that needs to be modeled within the framework of continuum mechanics. In the change of scale, some properties emerge fundamentally different from those at lower scales. With respect to constitutive modeling of granular plasticity, the concept of critical state emerge from the detailed balance of conformational transitions. Reorganizations in the microstructure makes it possible for granular materials at critical state to be sheared with no volume change under a constant stress state. As such, critical state stands as a cornerstone of constitutive theory for granular materials. To better understand the microscale origin of critical state, we analyze the rates of mesostructural transformations in simulations based on the discrete element method (DEM). Then, a deactivation/reactivation procedure acting on the local mesoscale is proposed to enrich a specific multiscale constitutive model for granular materials (the H-model). This procedure mimics the microstructure changes resulting from contact grain and loss that modifies the contact network with no volume change. We show how this procedure enable to make stationary regimes emerge naturally in multi-scale constitutive modeling without drawing from any empirical law at the macroscopic scale.

ABSTRACT. We analyze the band structure of a single-phase metamaterial involving low-frequency flexural resonances by combining asymptotic homogenization and Bloch–Floquet analysis. We provide the closed-form expression of the dispersion relation in the whole Brillouin zone. The dispersion relation involves two effective, frequency-dependent, mass densities associated with symmetric and antisymmetric flexural resonances of the beams at the microscopic scale. We demonstrate that our simple locally-resonant structure produces at low-frequency band-gaps and, in the hyperbolic regions of the dispersion diagram, negative refraction. Our findings are validated by direct numerical calculations.

ABSTRACT. Multi-scale simulation of lattices, cellular materials and meta-materials faces the difficulty of handling the local instabilities which correspond to a change of the micro-structure morphology. On the one hand, first order computational homogenisation, which considers a classical continuum at the macro-scale, cannot capture localisation bands. On the other hand, second-order computational homogenisation, which considers a higher order continuum at the macro-scale, introduces a size effect with respect to the Representative Volume Element (RVE) size.

By reformulating second-order computational homogenisation as an equivalent homogenised volume, non-uniform body forces arise at the micro-scale and act as a supplementary volume term over the RVE. Contrarily to the original uniform body forces resulting from an asymptotic homogenization [1], the devised non-uniform body forces arise from the Hill-Mandel condition and are expressed in terms of the micro-scale strain localization tensor, i.e. the relation between the micro-scale and macro-scale deformation gradients [1].

The consistency and accuracy of the approach are illustrated by simulating non-linear elastic meta-materials and elasto-plastic cellular materials under compressive loading.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 862015.

REFERENCES [1] V. Monchiet, N. Auffray, J. Yvonnet, Strain-gradient homogenization: A bridge between the asymptotic expansion and quadratic boundary condition methods, Mechanics of Materials 143 (2020) 103309. [2] L. Wu, S. M. Mustafa, J. Segurado and L. Noels. Second-order computational homogenisation enhanced with non-uniform body forces for non-linear cellular materials and metamaterials. Computer Methods in Applied Mechanics Engineering, 407: 115931, 2023.

ABSTRACT. Mechanical metamaterials are gaining attention for their unique responses, offering novel possibilities in elastic wave control. Researchers focus on developing new unit cells that produce unconventional macroscopic responses, such as band-gaps, cloaking, and negative refraction, etc [2, 3, 4]. To model large samples, homogenisation techniques establish equivalent continuum models for macroscopic metamaterials. A common approach assumes a classical linear Cauchy continuum, considering frequency-dependent parameters to capture complex responses in the frequency domain. However, these models may exhibit negative effective masses or effective elastic coefficients near resonance frequencies, making the Cauchy continuum non-positive-definite. We propose a procedure to convert frequency-dependent models into enriched continuum models of the micromorphic type [1]. The enriched model’s parameters are constant, and the well posedness of the model is ensured across all frequencies. The work bridges upscaling techniques with the idea that micromorphic continua are suitable for modelling metamaterial responses at the macroscopic scale.

References [1] G. Rizzi, M.V. d’Agostino, J. Voss, D. Bernardini, P. Neff, & A. Madeo. From frequency-dependent models to frequency-independent enriched continua for mechanical metamaterials. To appear in European Journal of Mechanics - A/Solids (2024) [2] C. Bellis, & B. Lombard. Simulating transient wave phenomena in acoustic metamaterials using auxiliary fields. Wave Motion, 86, 175-194. (2019) [3] L. Liu, A. Sridhar, M. G. D. Geers, & V. G. Kouznetsova. Computational homogenization of locally resonant acoustic metamaterial panels towards enriched continuum beam/shell structures. Computer Methods in Applied Mechanics and Engineering, 387, 114161. (2021) [4] A. Sridhar, V. G. Kouznetsova, & M. G. Geers. Homogenization of locally resonant acoustic metamaterials towards an emergent enriched continuum. Computational mechanics, 57, 423-435 (2016)

ABSTRACT. This paper describes the long wavelength dynamic behaviour of a periodic structure whose irreducible period consists of an Euler beam supported by two co-located springs (one rotational and one compressive).

The homogenisation method for discrete periodic media is used to obtain the macroscopic behaviour of the equivalent continuous multi-supported beam, provided that the wavelength is much larger than the period size. Under this assumption, each elementary beam remains in the quasi-static regime. Depending on the orders of magnitude of the stiffnesses of the springs relative to the bending stiffness of the elementary beam, the equivalent continuous structure exhibits different dominant order behaviours, governed either by shear, by bending, or by both. In an infinite medium, the long-wavelength dispersion relations obtained for each model agree with those obtained by Bloch waves.

In a finite domain, it remains to be specified how the boundary conditions actually applied to the structure can be transferred to the homogenised models. This is necessary in order to perform the modal analysis of the real system using the homogenised model. To overcome this issue, which is still open for beam systems, we propose the following method: local kinematic conditions are directly transferred to the macroscopic kinematic variables. However, local force-type conditions cannot be directly related to macroscopic forces. It is therefore necessary to express the exact local equilibrium conditions in function of the macroscopic kinematic variables. This yields the conditions on the macroscopic kinematic variables that effectively express the real force conditions. This procedure makes it possible to recover the modal analysis results obtained from numerical calculation, both in terms of frequency and deformation. It is also observed that the boundary conditions only alter the modal kinematics on a boundary layer of the order of the period and located at both ends.

ABSTRACT. The present contribution provides an analysis of surface effects in the mechanical response of architected materials modeled in the framework of strain gradient mechanics. The classical and strain gradient properties are evaluated by relying on a dedicated discrete homogenization method to upscale the microstructural information toward an effective strain gradient continuum at the macroscopic level. The formulation of the strain gradient model formulated via Hill extended macrohomogeneity condition allows a proper surface expression of the effective strain gradient kinematic and static variables. The scaling law of the strain gradient moduli with the edge contribution is obtained from their closed-form expressions versus the lattice microstructural parameters, and recoursing to the notion of shape derivative. The sensitivity of the strain gradient moduli to the relative amount of bulk versus surface effects is evaluated, showing that absolute size effects are well captured by strain gradient moduli. The energetic formulation of a second strain gradient continuum allows revisiting the notion of anisotropic surface energy, thereby providing a generalization of Mindlin’s model of surface energy.

References

R. D. Mindlin. Second gradient of strain and surface-tension in linear elasticity. International Journal of Solids and Structures, vol. 1, no. 4, pp. 417–438, Nov. 1965, doi: 10.1016/0020- 7683(65)90006-5.

N. Mawassy, J.F. Ganghoffer, H. Reda, S.E. Alavi, H. Lakiss. Analysis of surface effects based on first and second strain gradient mechanics. Mechanics of Materials 175, 104462.

ABSTRACT. Often, data-driven models require sampling complex, high-dimensional probability distributions. For that, Markov chain Monte Carlo (MCMC) methods are the standard tools. Naïve implementations of these algorithms have a huge cost and thus, many optimized methods have been proposed. One of the most popular ones, the Hamiltonian Monte Carlo (HMC) method, transforms the sampling process into a surrogate dynamical problem that needs to be integrated in time. The discrete solution of these trajectories is then interpreted as proposal samples that are later accepted or discarded depending on detailed rules that achieve the desired distribution.

In this talk, we will present the energy-stepping Monte Carlo (ESMC) method. It is an HMC method where the time-integration phase of the algorithm is performed using the energy-stepping scheme. As a result, the new method possesses remarkable properties: in particular, no proposed sample is ever discarded irrespective of the dimensional and complexity of the sampled distribution, hence being more efficient than other HMCs.

The talk will describe the theory behind the method, numerical examples showcasing its favorable properties as compared with other MCMC methods, and how the latter can impact data-driven simulations.

ABSTRACT. In the last decade, clinical research has made important progress on the use of machine learning for Traumatic Brain Injury (TBI) outcome predictions in large cohorts of patients. While these efforts mainly focus on the identification of biomarkers for injury identification and evolution, the question asked by forensic investigators in the context of law enforcement is different. Instead, police forces tend to focus on the likelihood that a given impact scenario leads or not to an observed injury. Traditional methods in forensic analysis, combining biomechanics and neuro-clinical knowledge, often do not provide an objective probabilistic assessment. Here, we propose to bridge this gap by introducing a comprehensive numerical framework coupling biomechanical simulations of various injurious impacts with machine learning algorithms. The model was trained against 200 finite element simulations representing various impact scenarios, alongside 53 detailed criminal assault reports provided by UK’s Thames Valley Police and the National Crime Agency's National Injury Database. Once trained, the proposed framework takes, as input, police reports data and predicts the risk of TBI for three specific symptoms: skull fracture, loss of consciousness and intracranial haemorrhage. The model demonstrates exceptional predictive accuracy, with rates exceeding 92% for skull fractures, 74% for loss of consciousness, and 87% for intracranial hemorrhages, with very high sensitivity and specificity. A notable feature of this research is its ability to identify which inputs, including specific mechanical properties and regions of the human head, are most influential in predicting targeted injuries. This insight underscores the critical role of the mechanics perspective in enhancing the model's accuracy. Despite its current limitation due to the available data on head injury cases, the framework shows remarkable predictive power and potential for future expansion and refinement.

ABSTRACT. FE2 complexity renders multiscale cellular meta material simulations impractical on account of excessive time and (computational) resource requirements. Especially the rate dependent, dissipative material nature of the base material alongside the fine discretization of the underlying repeated lattices necessitates acceleration of the numerical scheme. Resolution of the micro scale boundary value problem by a surrogate is investigated and its applicability is demonstrated using lattice based meta materials.

An effective surrogate model sensitive to (strain) rate and (microscale) geometrical parameters using a recurrent neural network (RNN) is trained (offline) on a dataset populated by performing full-field simulations. Populating the dataset, including identification of generation parameters, establishing bounds for spanning a functional space, designing of the surrogate model and tuning of the training parameters is presented.

The quality of the trained surrogate is evaluated by means of testing data and FE2 counterparts by substitution in equivalent multiscale simulations. Comparisons are made on the predictions demonstrating the sensitivity on (strain) rate, local constitutive behaviour, local (lattice) geometrical parameters using various loading scenarios. Performance gains in (simulation) time are reported alongside the modalities for population of a useful dataset and training of a reliable surrogate.

One potential area of application for surrogated multiscale modelling is microscale level optimization to maximize / minimize an objective function defined on macroscale level. This is achieved by the reduction in computational resources enabling fast and cheap evaluation of the objective function (multiscale FE simulation).

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 862015

ABSTRACT. Machining titanium alloys presents formidable challenges due to their high thermal strength, low thermal conductivity, low modulus of elasticity, and heightened chemical reactivity [1]. Micro-pillar textured tools have emerged as a proven method for enhancing the machinability of titanium alloys. These tools demonstrate the ability to reduce tool-chip contact length through improved chip curling and enhanced heat dissipation from the interface to the surroundings [2,3]. Despite their effectiveness, the exploration of texture shapes and dimensions on performance remains limited due to constraints imposed by fabrication techniques.

To address this gap, finite element analysis serves as an efficient alternative to experimental studies, overcoming fabrication challenges and minimizing practical costs and resource requirements. The accuracy of the numerical model relies on the material behaviour model, leading to the experimental validation of cutting forces with ten different Johnson-Cook parameter sets. Parameters with minimal deviation from experimental results are selected for further investigation.

To ensure the study's simplicity and comprehensiveness, this research presents a numerical analysis of 3D orthogonal machining of Ti6Al4V using plain and micro-pillar textured tools of various shapes (circular, rectangular, elliptical, etc.). The reverse micro electrical discharge machining (RµEDM) is employed for fabricating micro-pillar-shaped textures on tungsten carbide (WC) cutting tools. The challenges and critical aspects of micro-texture fabrication are extensively covered. All these shapes are explored under varying texture dimensions. The micro-pillar textures are specifically designed to address challenges associated with other patterns like micro-dimples and micro-grooves [3]. Consequently, a comparative study is conducted among these three texture patterns to elucidate the pros and cons of each.

References

[1] https://doi.org/10.1016/j.jmatprotec.2008.06.020 [2] https://doi.org/10.1016/j.jmatprotec.2018.05.032 [3] https://doi.org/10.1016/j.mfglet.2023.04.001

ABSTRACT. Bending performances for automotive body sheet is a key feature for validating the use of an aluminum alloy in an industrial application. Understanding the impact of the texture on the bendability of the material is then of interest to help industrials tuning composition and milling process parameters in order to match the bending performances with the application expectations. The first part of the paper presents the methodology adopted to evaluate the bending performances of a virtual texture before doing any experimental test. It lies on the use of a two-scale model approach: a macro-scale with a FEM modeling of a VDA bending test and a micro scale, representative of the volume element, with a crystal plasticity simulation thanks to the open-source DAMASK software. By coupling these two scales, it is possible to study the response of the virtual texture at the grain size. By using different type of textures, it is then possible to rank them according to their bending performances. The main assumptions, constitutive laws and model’s preparation will be detailed to understand the whole frame of the methodology. The first results on ideal textures (i.e. with a high proportion of a unique component such as Goss or Copper) give bending performances and ranking in accordance with the existing literature. The application of this methodology on three real industrial and significantly different textures for which experimental campaigns have been done shows as well a good correlation between virtual and experimental ranking. The paper will be concluded with a proposition of criterion to evaluate the bending performance of a texture. Finally, the role of iron particles will be discussed as an opportunity for further studies.

ABSTRACT. Profilometry-based indentation plastometry (PIP) can be used to determine the stress-strain characteristics of metallic materials from indentation testing of a small, localised area. It is currently very well suited for testing on isotropic, fully dense and homogeneous metals. The procedure uses the residual indent profile and an (accelerated) iterative finite element simulation of the indentation process. The plasticity parameters in a constitutive law (within an indentation finite element model) are iteratively changed until optimum agreement between measured and predicted residual profile shapes is obtained. The technique characterises the full uniaxial stress-strain relationship, including the yield stress and up to the ultimate tensile strength. Comparisons are drawn with results from tensile coupons and hardness testing to demonstrate the equivalency with tensile and improvements over hardness testing. One of the attractions of PIP is that it allows stress-strain curves to be obtained for relatively small volumes of material, such that local properties can be mapped in regions where they are changing over short distances. This makes it uniquely well-suited for measuring local variations in properties in, for example, the vicinity of welds or in functionally graded materials. This can be further expanded to include testing different regions of a single part that have undergone different processing conditions. The additional information provided by PIP can then enhance design optimization by offering insights beyond those obtained solely from witness coupons, highlighting the importance of considering alternative approaches for comprehensive part characterization.

ABSTRACT. Prismatic slip of the screw <a> dislocation in magnesium at temperatures ≳ 150 K is understood to be governed by double-cross-slip of the stable basal screw through the unstable prism screw and back to the basal screw, with the activation energy controlled by the formation energy of two basal-basal kinks. However, atomistic studies of the double-kink process predict activation energies roughly twice those derived experimentally. Here, a new mechanism of prism glide is proposed, analysed theoretically, and demonstrated qualitatively and quantitatively via direct molecular dynamics (MD) simulations. The new mechanism is intrinsically 3d, and involves the nucleation of a single kink at the junction where a 3d prismatic dislocation loop transitions from the basal screw segment to non-screw prismatic character. The relevant kink energies are calculated using recently-developed Neural-Network Potentials (NNPs) for Mg that show good agreement versus DFT for basal and prism <a> dislocations, enabling a parameter-free analytic model for the activation barrier. Direct MD simulations show both operation of the precise proposed mechanism and a stress-dependent activation barrier that agrees reasonably with the analytic model. Predictions of dislocation velocity compare very well with in-situ TEM data, and macroscopic strength versus temperature can also be understood. Overall, the new intrinsically 3d mechanism for dislocation glide due to single kink nucleation rather than double-kink nucleation explains key features of the prismatic slip in Mg and may have broader applicability in other metals where kink nucleation processes control thermally-activated flow.

ABSTRACT. Most biological materials and many soft engineering materials are made from fibers stochastically distributed and oriented in two or three dimensions. These materials are referred to as ‘network materials’, as their mechanical behavior is controlled by the underlying fiber network. The parameters controlling the mechanical response are the structure of the network, fiber properties and fiber-fiber interactions. In this work we are concerned with the effect of inter-fiber adhesion on the overall mechanical behavior. We show that, in crosslinked networks, a yield point appears due to fiber-fiber adhesion and the yield stress scales linearly with the adhesion strength. The large deformation response is only weakly affected by the adhesive forces. Non-crosslinked networks evolve under the action of adhesive forces to produce cellular networks of fiber bundles. In the absence of friction, such networks are stabilized exclusively by adhesion. A phase diagram capturing the range of network stability in the space of system parameters will be discussed.

ABSTRACT. The extraordinary large traction force needed to separate two stacks of interleaved paper sheets (popularly known as the phonebook enigma) is caused by a geometric amplification of friction. Although its origin is well described in the literature (Alarcon et al. 2016), this remarkable interfacial effect has not been further explored in terms of possible applications. We consequently create higher level assemblies of paper sheets’ stacks connected together solely by interleaving and evaluate their tensile behavior. In particular, we compare manually assembled macroscopic membranes with different types of woven structure. Through mechanical tests, FE simulations and analytical modelling we demonstrate superior traction force for a double-weave assembly due to an increase in self-induced compression at the stack level which is orders of magnitude larger than what is reported in the literature. Moreover, via nanoindentation measurements we are able to characterize the effect of increased compression force at the paper sheet surface, thus highlighting how macroscopic loading conditions (membrane tension) translate at the microscopic level (inter-sheet friction) and contribute to the overall friction amplification effect of the entire assembly. Finally, we demonstrate a possible usage of such structures by building a structurally stable 5 meters span hanging bridge. Our work thus shows how to program the overall mechanical behavior of a structure via geometric design at multiple hierarchical levels, which is held together by interfacial friction forces alone.

ABSTRACT. Many devices, including sports shoes and robotic hands, involve frictional soft contacts. Optimizing those devices requires fine control of the interface’s friction law. We lack systematic methods to create dry contact interfaces whose frictional behaviour satisfies preset specifications. In this talk, a generic surface design strategy to prepare dry rough interfaces that have predefined relationships between normal and friction forces will be presented. Such metainterfaces circumvent the usual multiscale challenge of tribology, by considering simplified surface topographies as assemblies of spherical asperities. Optimizing the individual asperities’ heights enables specific friction laws to be targeted. Through various centimeter-scaled elastomer-glass metainterfaces, different types of achievable friction laws, including linear laws with a specified friction coefficient and unusual non-linear laws will be illustrated. This design strategy represents a scale- and material-independent, chemical-free pathway toward energy-saving and smart soft interfaces.

ABSTRACT. Surface adhesion is a key material property in several engineering applications. In the past decades, many efforts have been made in mimicking nature to achieve outstanding adhesion performances. A well-known example is given by geckos, which are able to climb walls thanks to the hierarchical structure of their feet, that allows them to quickly adjust surface adhesion. Drawing inspiration by geckos, a controlled tuning of interface adhesion could potentially be crucial to a broad range of applications, such as soft robotics, space technology, micro-fabrication, and flexible electronics. When considering soft interfaces, an intricate correlation between viscoelasticity and adhesion exists: parameters like loading history, unloading speed and layer thickness can affect the detachment (i.e., pull-off force) of a rigid indenter from a soft viscoelastic substrate. Moreover, recently it has been demonstrated that surface adhesion of PDMS to a glass sphere can be successfully regulated via mechanical micro-vibrations: it is possible to enhance the apparent adhesion strength up to 77 times or reduce it to 0 with respect to the case with no vibrations, within a timescale comparable to that of geckos. Following up on these results, we have built a customised experimental set-up, that performs indentation tests on soft polymers films, mounted on a vibrating substrate (excited by shakers). We aim at studying the role of materials properties (i.e., crosslinking ratio), film thickness, loading/unloading velocities, and indenter geometry on the adhesive performances of the system. Subsequently we will explore the effects of micro-patterning of these soft interfaces. Micro-structured surfaces were manufactured via two‐photon polymerization lithography, using a Photonic Professional GT2 3D printer (NanoScribe), with sub-micron precision. We believe that our findings will contribute to the development of a new class of soft smart interfaces, with tuneable adhesion properties.

ABSTRACT. Dynamic crack propagation is the key mechanism leading to catastrophic material failure. Linear Elastic Fracture Mechanics (LEFM) provides a predictive theoretical framework [1] and describes fracture experiments well as long as its application hypothesis are fulfilled and single cracks are considered. However, at high enough speeds, crack propagation involves small-scale, high-frequency, microcracking events that are difficult to account for via conventional continuum fracture mechanics. To shed light on these mechanisms, we re-examined an experimental dataset obtained on acrylic glass (PMMA) [2,3,4], which includes both measurements of crack speed and fracture energy at the macroscale, and catalogs with the micrometer/nanosecond resolved dynamics of the associated near-tip microcracking events. In the presentation, we will see how a simple geometrical model succeeds in reproducing quantitatively all the statistical features observed in the microcracking dynamics, and how macroscale crack speed and fracture energy emerge from the proper upscaling of this model.

References: [1] L.B. Freund, Dynamic Fracture Mechanics, Cambridge University Press, New York (1990). [2] J. Scheibert, C. Guerra, F. Célarié, D. Dalmas, and D. Bonamy. Brittle-quasibrittle transition in dynamic fracture: An energetic signature. Physical Review Letters, 104(4):045501 (2010) [3] C. Guerra, J. Scheibert, D. Bonamy, D. Dalmas, “Understanding fast macroscale fracture from microcrack post-mortem patterns” PNAS 109, 390-394 (2012). [4] D. Dalmas, C. Guerra, J. Scheibert, and D. Bonamy. Damage mechanisms in the dynamic fracture of nominally brittle polymers. International Journal of Fracture, 184(1-2):93–111 (2013).

ABSTRACT. When brittle hydrogels fail, several mechanisms conspire to alter the state of stress near the tip of a crack, and it is challenging to identify which mechanism is dominant. In the fracture of brittle solids, a sufficient far-field stress results in the complete loss of structural strength as the material ‘unzips’ at the tip of a crack, where stresses are concentrated. Direct studies of the so-called small-scale yielding zone, where deformation is large, are sparing. Using hydrogels as a model brittle solid, we probe the small-scale yielding region with a combination of microscopy methods that resolve the kinematics of the deformation. A zone over which most of the energy is dissipated through the loss of cohesion is identified in the immediate surroundings of the crack tip. With direct measurements, we determine the scale and structure of the process zone, and identify how the specific loss mechanisms in this hydrogel material might generalize for brittle material failure.

ABSTRACT. This work introduces NeoMag, a system designed to enhance cell mechanics assays in substrate deformation studies. NeoMag uses multidomain magneto-active materials and external magnetic fields to mechanically actuate the substrate, transmitting reversible mechanical cues to cells. The system boasts full flexibility in alternating loading substrate deformation modes, seamlessly adapting to both upright and inverted microscopes. The multidomain substrates facilitate mechanobiology assays on 2D and 3D cultures. In addition, the integration of the system with nanoindenters allows for precise evaluation of cellular mechanical properties under varying substrate deformation modes. The system's efficacy is demonstrated by studying the impact of substrate deformation on astrocytes, simulating mechanical conditions akin to traumatic brain injury and ischaemic stroke. The results reveal local heterogeneous changes in astrocyte stiffness, strongly influenced by the orientation of subcellular regions relative to substrate strain. These stiffness variations, exceeding 50% in both stiffening and softening, and local deformations significantly alter calcium dynamics. Furthermore, sustained deformations induce actin network reorganization and activate Piezo1 channels, leading to sustained calcium influx that inhibits calcium events. Conversely, fast and dynamic deformations transiently activate Piezo1 channels and disrupt the actin network, causing cell softening over 24 hours. These findings unveil mechanical and functional alterations in astrocytes during substrate deformation, illustrating the multiple opportunities this technology offers.

ABSTRACT. Over 33 million people in the EU suffer from diabetes. According to data from the International Diabetes Federation (IDF), the absolute number of diabetics in the EU is projected to increase from approximately 33 million in 2010 to 38 million in 2030. Many patients require multiple-dose insulin therapy to achieve acceptable glycemic control. Evidence suggests that a noninvasive or minimally invasive treatment option will help improve adherence for individuals whose reasons for nonadherence range from the inconvenience of therapy to needle aversion, especially in pediatric patients. Furthermore, insulin-induced lipohypertrophy is the most common local complication observed in diabetic patients undergoing insulin therapy, which reduces insulin absorption. This study couples a computational model capable of simulating insulin fluid dynamics through the port and subcutaneous tissue with a model of insulin absorption in the tissue. The computational model, based on finite element method, was utilised to analyse the operating conditions of the injection port. Specifically, the impact of an obstruction within the cannula delivering insulin to the subcutaneous tissue was examined. Moreover, the model can integrate both the physiological heterogeneity of the tissue and account for pathological tissue conditions such as lipodystrophies. The results highlight the influence of both obstruction and tissue heterogeneities on tissue absorption. Furthermore, the model can be employed to optimise the design of the injection port, thereby enhancing the device's effectiveness in diabetes therapy.

ABSTRACT. In topology optimization it is essential to account for the real-world material behavior in order to result in application-oriented and safe engineering structures. Therefore, we develop the inclusion of hardening material behavior into the thermodynamic topology optimization. Simulation results show the influence of different material behaviors. The optimized structures are experimentally validated to ensure the usage of our method for real-world applications.

We present the method of thermodynamic topology optimization including hardening material behavior. Parameters for the material model are determined by tensile tests. An experimentally testable boundary value problem is optimized due to the hardening material behavior. In addition, a reference structure with linear elastic material behavior is optimized. Both structure specimens are additively manufactured and experimentally tested. The test results verify the influence of the material behavior on optimized structures. Therefore, we demonstrate that it is beneficial for designing real-world engineering structures to include a precise material behavior in the topology optimization.

ABSTRACT. The Laser Powder Bed Fusion process is of particular interest to the aerospace industry because of its ability to produce metal parts with complex geometries. Structurally hardened aluminum alloys produced by L-PBF exhibit specific metallurgical characteristics due to non-equilibrium solidification. These include the formation of metastable phases and of thermally induced dislocations pinned to nanometric precipitates. As a result, the as-built material exhibits remarkable properties that make it possible to avoid a post-fabrication heat treatment. This opens the route for the development of specific alloy grades that fully take advantage the unique features of the L-PBF process. This work focuses on a model binary Al-Fe alloy produced by L-PBF. The objective is to better understand the metallurgical and thermal mechanisms associated with the rapid solidification and cooling rates that characterize L-PBF manufacturing route and lead to the structural hardening of the alloy in the as-built condition. To this end, detailed characterization by scanning and transmission electron microscopy and X-ray diffraction has been carried out on samples obtained using different processing parameters. Nano- and micro-indentation measurements and macroscopic tensile tests have also been performed. Our results reveal the heterogeneity of the mechanical behavior and show that the macroscopic behavior is largely controlled by the size of the solidification cells. We have also performed a transmission electron microscopy study of the hardening mechanisms such as the pinning of dislocations to nanometric precipitates. Finally, the contributions of all the hardening mechanisms have been considered to develop a macroscopic hardening model for Al-Fe alloys produced by L-PBF. The parameters of the phenomenological model have been carefully selected using dedicated measurements and literature data. The aim is now to use this model to link the processing parameter to the mechanical behavior and to use this link to optimize these alloys in the as-built state.

ABSTRACT. Unsaturated wet granular materials exhibit intricate microstructures composed of solid particles, liquid phases, and void spaces. Understanding the morphological and rheological characteristics of these materials is essential for various applications, from geotechnical engineering to environmental science. By employing X-ray tomography, we aim to understand the relationship between the microstructural features and rheological properties of these materials, shedding light on their behavior and interactions in various conditions. Our focus is on slightly polydisperse polystyrene beads mixed with a minimal amount of liquid, specifically in the pendular regime (the ratio between the liquid volume and the volume of polystyrene beads is less than 0.075), where capillary bridges are the primary liquid morphology, although other morphologies also exist in smaller proportions. A custom shear device, compatible with an X-ray microtomography imager, has been designed to observe microstructural evolution under imposed confining stress and shear rate. Through these X-ray microtomography experiments, we capture detailed 3D images at different deformation stages. Employing advanced image segmentation techniques, which integrate machine learning and deep learning, we can analyze these complex microstructures accurately and comprehensively. Our segmented images offer a deeper understanding of grains and liquid distribution, as well as the different liquid morphologies. In particular, we have developed an automatic tool for classifying the different liquid morphologies within the sample. This method enables us to analyze the 3D spatial distribution of the grains and liquid fractions, in addition to the changes in the liquid morphologies, providing insights into their responses under shear conditions.

ABSTRACT. Under conditions favoring grain boundary sliding (GBS), metallic polycrystalline aggregates tend to preserve equiaxed grain shapes. The motion of the grain centroids then differs significantly from the affine displacement field that could be deduced from the macroscopic strain. Instead, if GBS operates, grains tend to switch neighbours and this involves cooperative displacements of the surrounding grains. The present work assumes that the overall viscosity of a polycrystal subjected to GBS is dictated by such redundant motion of the grains, i.e. by accommodating displacements which do not contribute to the macroscopic strain. Based on a simplified numerical modeling of diffusion-aided GBS (Rachinger creep) inside a 2D polycrystal, we estimate the energy that is dissipated when prescribing the motion of a single grain and allowing accommodation only in the close neighbourhood. The focus of the analysis is set on the influence of the presence porosity at triple junctions, which is characteristic of the final densification stage during the sintering of metallic alloys. The ratio of the macroscopic bulk viscosity and the macroscopic shear viscosity is shown to evolve as a function of porosity.

ABSTRACT. We present a new model, based on the discrete element method, designed to simulate a dynamic rod system. Originally designed for the study of plant cell tissues, our approach shows remarkable adaptability to a variety of applications. The model is based on a set of equations governing the motion of bars in two-dimensional space. These equations are solved using a robust numerical integration scheme, with the resulting motion serving as the basis for calculating the forces distributed to the nodes by the bars. The results demonstrate a commendable fidelity in reproducing the anticipated behavior, attesting to the model's accuracy and reliability. Beyond its application to the elucidation of plant cell tissue dynamics, our model holds promise for the study of a wide range of systems. In conclusion, our proposed model, anchored in the discrete element method, not only successfully captures the subtleties of plant cell tissue dynamics, but also lends itself to broader scientific investigations.

ABSTRACT. Among different types of avalanches, the "slab avalanches" initiate by a long crack perpendicular to the slope, and rapidly propagate downhill during the flow. Modeling these avalanches presents challenges, including predicting the threshold for crack initiation in such a cohesive granular material, and understanding the impact of snow properties on fracture propagation.

To address these questions, we conducted experiments using a cohesion-controlled granular material encompassing a wide range of cohesion. We established an experimental setup specifically designed to explore the formation of fractures in a quasi static regime. In our setup, a layer of the cohesion-controlled material undergoes flexural deformation. Upon reaching a threshold in tensile stress, cracks emerge with a distinct wavelength that increases with cohesion. We compare these measurements to phase-field model used in damage mechanics.

ABSTRACT. The Compressed Earth Block (CEB) represents a low-carbon construction material for masonry buildings, permitting the reuse of local soil, removed after leveling and other earthworks, and then reducing the energy consumption related to its collection, transport, recovery, and disposal. The CEB masonry buildings can be constructed if fine-grained soil is available at the construction site. The soil is mixed with water, cement-stabilized, and then compressed in-situ by a machine to achieve a target compressive strength. A pilot project of a CEB masonry building is analyzed. It is built in Southern France in 2023, in a medium-high seismic hazard zone. The mechanical parameters of CEB and mortar are obtained by experimental tests. Whereas, the mechanical parameters of CEB masonry are obtained using homogenization formulas according to the Eurocode, considering the regularity of masonry. Even if the CEB is a promising construction material, contributing to a more sustainable building industry, the assessment of structural performance in seismic zones can limit their use. The unreinforced masonry building is modeled according to the equivalent frame (EF) approach, accepted by the Eurocode for earthquake design of buildings. Deformable macro-elements representing the wall panels are connected by rigid nodes (modeling the stiffer zones between them). The wall element response to shear, bending, and axial force is coupled according to strength domains. The intersection of strength domains gives information about the expected in-plane failure mode of the masonry panel (bending-rocking or shear mechanisms). The modal characteristics, ductility, and expected seismic performance of the CEB masonry building are investigated. An ambient vibration recording campaign is conducted in the building and the elastic building behavior, simulated using the EF model, is validated by comparing the dynamic properties obtained by numerical and operational modal analysis. Consequently, the elastic mechanical parameters of CEB masonry used in the model are validated.

ABSTRACT. Understanding the transport of hydrogen within metallic nanomaterials is crucial for the advancement of energy storage and the mitigation of hydrogen embrittlement. Using nanosized magnesium particles as a model, recent studies have revealed several highly nonlinear phenomena that occur over long time periods. The time scale of these phenomena is beyond the capability of established atomistic models such as molecular dynamics. In this work, we present a new approach, referred to as diffusive molecular dynamics (DMD) [1,2], to the simulation of long-term diffusive mass transport at the atomic scale. The basic assumption underlying DMD is that the time scale of diffusion is much larger than that of microscopic state transitions. In terms of numerical implementation, our approach involves the numerical integration of the master equation, and the numerical solution of a highly nonlinear optimization problem at every time-step. By working with atomic fractions, the characteristic time-step size of our DMD simulations can be much larger than those based on either AMD or KMC methods. In the present work, we focus on the characterization of thermodynamic and kinetic properties of magnesium hydride nanoparticles. Modeling phase transitions of Mg/MgH2 systems introduce an additional difficulty due to the change of lattice structure. We also note that the scope of DMD is not limited to metal hydrides and a broad range of multi-species systems of practical interest suggest themselves as worthwhile foci for future studies. REFERENCES [1] X. Sun, M.P. Ariza, M. Ortiz, K. Wang, Journal of the Mechanics and Physics of Solids, 125: 360-383, 2019. [2] G. Venturini, K. Wang, I. Romero, M.P. Ariza, M. Ortiz, Journal of the Mechanics and Physics of Solids, 73: 242-268, 2014.

ABSTRACT. Polymer matrix composites are widely used in various industrial fields such as aeronautics and transport in general. Modelling their time-dependant macroscopic mechanical behaviour as a function of their microstructural organization is a common objective in the scientific community. Whatever the application, and before considering other dissipative processes, the very first stage remains a correct description of the viscoelasticity. Mean-field homogenization is here employed with new developments of the incremental variational approach originally proposed by Lahellec and Suquet (2007), also called EIV approach for “Effective Internal Variable” approach. The present paper has two objectives. The first one is to illustrate the possible association of the EIV approach to different linear schemes and, as a result, its possible application to various microstructures (fiber or particle-reinforced composites, strand-based wood composites). The second objective is to enlarge the application domain of the original EIV approach in terms of constituents’ viscoelastic laws that may be considered. Initially applied to phases described by simple viscoelastic laws involving a single, traceless, internal variable, (ex. Maxwell model and constant bulk modulus), the original formulation of the EIV model is here generalized to phases described by more representative viscoelastic laws involving several internal variables with both spherical and deviatoric parts. The generalized EIV approach is evaluated by comparison to reference solutions obtained by full-field finite element (FE) simulations for a fiber-reinforced microstructure. Linear viscoelastic laws with increasing complexity are considered for the matrix. At first, a classical Maxwell model is used as in Lahellec and Suquet’s original work, but now the internal variable (viscous strain) has a spherical part. Then, the matrix is described by a generalized Maxwell model with several branches. Different distributions of relaxation times are studied for bulk and shear moduli. In every case, the accuracy is satisfactory.

ABSTRACT. In order to describe the mechanical behaviour of solid, microstructurally complex, materials multi-scale modelling techniques can be used. There exist a large number of conceptually different multi-scale strategies, ranging from phenomenological models, to continualisation or homogenisation approaches, etc. In this talk two classes of different approaches will be presented. First, the conventional homogenisation methodology will be discussed, through either analytical homogenisation, e.g. so called gradient enriched models, or numerical homogenisation. The main difference between these methodologies is the existence (or lack of) an explicitly given constitutive equation on the macro-level. In both methodologies, by means of hierarchical multi-scale procedures, homogenised information of the detailed (heterogeneous, random) micro-structural description is brought to the macro-level in the form of effective properties. Thus, the homogeneous macro-structural behaviour is driven by the heterogeneous micro-structure.

A second, alternative methodology that will be discussed, is more data driven approach. It is based on an analysis of existing (experimental) data, such as, for example, digital images of material micro-sections, it uses fuzzy sets theory to estimate effective macroscopic properties of a material.

We will discuss advantages and potential limitations of aforementioned approaches, new findings and open-ended questions.

ABSTRACT. This study presents an innovative approach for developing a reduced-order model (ROM) tailored specifically for nearly incompressible materials at large deformations. The formulation relies on a three-field variational approach to capture the behavior of these materials. To construct the ROM, the full-scale model is initially solved using the finite element method (FEM), with snapshots of the displacement field being recorded and organized into a snapshot matrix. Subsequently, Proper Orthogonal Decomposition (POD) is employed to extract dominant modes, forming a reduced basis for the ROM. Furthermore, we efficiently address the pressure and volumetric deformation fields by employing the k-means algorithm for clustering. A well-known three-field variational principle allows us to incorporate the clustered field variables into the ROM.

To assess the performance of our proposed ROM, we conduct a comprehensive comparison of the ROM with and without clustering with the FEM solution. The results highlight the superiority of the ROM with pressure clustering, particularly when considering a limited number of modes, typically fewer than 10 displacement modes. Our findings are validated through two standard examples: one involving a block under compression and another featuring Cook's membrane. In both cases, we achieve substantial improvements based on the three-field mixed approach. These compelling results underscore the effectiveness of our ROM approach, which accurately captures nearly incompressible material behavior while significantly reducing computational expenses.

ABSTRACT. The traditional approach to modeling the non-linear constitutive behavior of soft materials at finite strains relies on hyperelasticity, a paradigm that inherently involves uncertainties in phenomenological constitutive modeling. This method often results in information loss since experimental data is not directly integrated into calculations, and the engineer's selection of a suitable strain energy function heavily depends on experience.

In contrast, data-driven approaches present a promising alternative. We recently introduced an innovative data-adaptive approach for modeling hyperelastic materials at finite strains allowing for more efficient inclusion of experimental data. The proposed modeling procedure ensures adherence to crucial constraints such as thermodynamic consistency, material objectivity, frame indifference, and material symmetry a priori. The fundamental concept involves formulating an approximation of the strain energy function using a sum of basis functions multiplied by parameters. The basis functions are expanded over the space of invariants, and support points, where the parameters are defined, are distributed in the invariant space. These parameters are determined through a nonlinear optimization problem, minimizing the 2-norm of the residual vector, representing the difference between measured and computed displacements and reaction forces. Importantly, our approach does not depend on measured stresses, setting it apart from many existing data-driven and model-free methods.

In our initial investigation, we demonstrated the efficiency and flexibility of our method based on numerical experiments and linear finite-element-like basis functions. The focus of this contribution is on the applicability to real experimental data obtained from Digital-Image-Correlation and basis functions with higher continuity. Specifically, we concentrate on a substantial experimental database encompassing various soft materials, particularly ELASTOSIL, VHB, and DOWSIL, with the aim of identifying a set of strain energy functions characterizing these materials in a fully automated and reliable manner.

ABSTRACT. Multi-scale modelling is the method of choice for modelling materials with complex internal architectures. One of the main assumptions often used in such models is material periodicity. However, in some categories of materials, such as 3D woven composites, the material loses periodicity during manufacturing. The interaction between the weave architecture and structural geometry leads to a non-periodic internal architecture which is unique to each structure. A novel approach based on geometric informed material clustering is introduced to address the challenges arising from material non-periodicity. The proposed approaches identifies repeatable patterns that appear in the material architecture on the sub-meso-scale. These patterns called “Material Clusters” are grouped using data clustering and 3D image registration algorithm. Here, an objective metric for the cluster similarity is used to categorise the cluster based on their morphology and local material properties. The material response is sampled using meso-scale models of a number of structural features and under different loading conditions. The output from these models is used to populate database of material clusters, consisting of the clusters geometric information, their elastic response, and their history-dependent damage response. On the macro-scale, structural scale simulations are carried out by querying the cluster database for previously stored responses, instead of solving the complete structure in real-time. The proposed approach benefits is twofold. First, is a significant reduction in the computational cost of structural scale models while still representing the material architecture, with no periodicity assumption. Second, the proposed framework continually accumulates knowledge about the material response as the number of simulations carried out increases. The framework can identify previously unseen material clusters / loading conditions which are then simulated and added to the database for future use. In this work, the details of the framework development, application to 2D and 3D woven composites and benchmarks will be presented.

ABSTRACT. Numerical homogenization methods are widely used in science and industrial applications to predict the effective behavior of engineering materials in structural components based on their microstructure. For nonlinear material behavior, this requires solving coupled boundary value problems for the microscale and macroscopic scale in a concurrent way. The most flexible approach for this purpose is the utilization of the finite element method on both scales, known as the FE² method. The high flexibility and generality comes along with high computational costs, which motivated numerous techniques to reduce these costs. In particular, neural network (NN)-based surrogate models or reduced-order models (ROM) have attracted a lot a research effort. Both types of models are driven by training data from expensive fully-resolved microscale simulations as their input. But while NN-based models need to be augmented to capture physical constraints, ROM-FE models form variational approximations to the microscale problem and thus inherit their fundamental physical behavior, though still at higher computational costs than surrogate models. The costs do not only comprise the online computational time during the actual structural simulation, but also the so-called offline costs for generating the training data. The present contribution shows how a hyperintegrated ROM method can be combined with a monolithic solution strategy to reduce the online costs, in conjunction with a clustered training strategy to lower offline costs.

ABSTRACT. To extend the use of single crystals or additively manufactured materials to sensitive components, for instance in aerospace industry where catastrophic failure must be avoided at all costs, it is crucial to have models capable of reliably determining how and when a crack propagates for anisotropic materials. While Linear Elastic Fracture Mechanics (LEFM) has been certified experimentally for isotropic materials, its extension to the anisotropic case lacks close comparison with experiments.

To fill this gap, we have developed experiments designed to precisely test and identify the material parameters of the generalized energy release rate criterion, based on Griffith's energy balance, and an associated phase field variational approach (Li and Maurini, 2019).

The reference material used is Polycarbonate printed by Fused Deposit Modeling to fit within the framework of LEFM. As an initial study, the printing strategy has been tuned to get some 2D isotropic elasticity, that permits the use of standard tools (Stress Intensity Factors and Digital Image Correlation) while exhibiting some strong fracture anisotropy (Corre and Lazarus, 2021).

References:

Corre, T., Lazarus, V., 2021. Kinked crack paths in polycarbonate samples printed by fused deposition modelling using criss-cross patterns. Int. J. Fract. 230, 19–31. https://doi.org/10.1007/s10704-021-00518-x

Li, B., Maurini, C., 2019. Crack kinking in a variational phase-field model of brittle fracture with strongly anisotropic surface energy. J. Mech. Phys. Solids 125, 502–522. https://doi.org/10.1016/j.jmps.2019.01.010

ABSTRACT. We study the ability of phase field (PF) and the coupled criterion (CC) to predict crack initiation and propagation under opening or shear modes. A confrontation of both approaches reveals that the internal length used in PF is intrinsically correlated to the material tensile strength used in the CC. This correlation also involves the material Poisson’s ratio and the local principal stress state [1,2]. Based on this correlation, a length-free (LF) implementation of the PF approach for fracture is proposed [3]. The inputs of the LF-PF model are similar to the CC inputs, namely the critical energy release rate and tensile strength (or more generally, a strength surface). The proposed approach is tested and compared to the CC on several benchmark examples. The proposed LF-PF approach can be considered as a PF implementation of the CC, both models may be used in a complementary manner since they share the same input parameters and provide similar results regarding crack initiation.

[1] Molnár G, Doitrand A, Estevez R, Gravouil A. 2020. Toughness or strength? Regularization in phase-field fracture explained by the coupled criterion. Theoretical and Applied Fracture Mechanics, Volume 109: 102736. [2] Molnár G, Doitrand A, Jaccon A, Prabel B, Gravouil A. 2022. Toughness or strength? Thermodynamically consistent linear-gradient damage model in Abaqus. Engineering Fracture Mechanics, Volume 266: 108390. [3] Doitrand A, Molnár G, Estevez R, Gravouil A. 2023. Strength-based regularization length in phase field fracture. Theoretical and Applied Fracture Mechanics, Volume 124, 103728

ABSTRACT. New types of concrete have emerged in the last decades, which require research about their fracture behaviour. As an example, a recen study carried out in collaboration with the Spanish Association of Railway Sleeper Element Manufacturers manifested that the size-effect curves of concrete should be extended in order to be applied to new mixtures. In that work a new implementation of the method proposed by Planas, Elices and Guinea, which combines diagonal splitting tensile tests and three-point bending tests of notched specimens to determine the main fracture properties of quasi-brittle materials, was offered with the appropriate modifications in order to perform the tests by using the factory facilities, and a round-robin test was carried out with the collaboration of ten laboratories, obtaining satisfactory results. However, it was found that some mixtures present a diameter to brittleness length ratio with values out of the ranges considered by Rocco et al in their size-effect study, manifesting the need of extending the available curves. The current works presents a numerical study for the extension of the size-effect curves of cylindrical specimens for low diameter to brittleness length ratios, by using elements with an embedded cohesive crack within the finite element framework COFE (Continuum Oriented Finite Element) and bi-dimensional models of the specimens in dimensionless numerical simulations. In the presentation, the main aspects of the simulations will be explained together with the analysis of the main cases for verification of the method, the extended curves of cylindrical specimens will be presented for the ranges of width of bearing bands and diameter of cylinders considered by Rocco el al, and it will be discussed how to proceed in the interpretation of the experimental results to determine the tensile strength and brittleness length of concrete.

ABSTRACT. Etching is used to improve the tensile strength of glass by blunting or even eliminating surface defects. It involves immersing test specimens in a hydrofluoric acid bath at varying concentrations and for varying durations. So, an initial surface crack is gradually transformed into a U-notch, the radius of which increases with immersion time. Under these conditions, Griffith’s criterion traditionally used for crack-like defects becomes ineffective for predicting the critical failure stress. The Coupled Criterion (CC) based on a twofold condition in stress and energy can take over. It requires the knowledge of the tensile strength of the material together with its toughness and can predict crack initiation at stress concentration points, e.g. V- or U-notch tips, holes, inclusions. Experiments have been carried out on specimens made of soda lime glass for construction industry. In a first step, a scratch is produced with a conical 120° diamond indenter, leading to a crack of controlled depth. The specimens are then immersed in fluoric acid baths of varying concentrations for varying durations, cracks are therefore transformed into U-notches. Some specimens are reserved for measuring the notches depth and radius, while others are subjected to double ring bending tests until failure. The CC is then used in a reverse manner to identify the intrinsic strength of the soda lime glass, i.e. its resistance when all surface defects (extrinsic defects) have been eliminated. It relies on the inner structure of the material and, even though glass is classified as an amorphous material, it is a rather hasty assumption to suppose that it has no micro-structure at all, crystallites may be present for instance, not forgetting impurities, bubbles, among others. A value around 1000 MPa is derived, far below the molecular strength which is at least one order of magnitude higher.

ABSTRACT. We compare two different formulations for the construction of inductive biases in the framework of scientific machine learning. The objective is to ensure that a neural network's prediction of a mechanical system's behavior inherently satisfies thermodynamic principles (energy conservation, non-negative entropy production). At least two different approaches can be considered either by using a single generator (the generalized free energy of the system, F), giving rise to the so-called single bracket formalism. This generalized free energy potential is a combination of the internal energy and the generalized entropy of the system, i.e., F=E+S. This opens the possibility to separate into two generators, giving rise to the so-called GENERIC or metriplectic formalism, where an energy potential associated with conservative energy (E) and the non-conservative effects (S) are defined (1). Through developed neural networks, the dynamics of different non-conservative systems has been learned by reconstructing the two described formalisms: Single Bracket (SB) and double bracket or GENERIC (G) (2). The impacts of different hyperparameters and their computational costs were analyzed in data reconstruction, revealing the advantages, and limitations of each formulation. Both methods effectively reconstruct the dynamics of various studied problems, although the SB formalism does not impose energy conservation explicitly (3). Despite formalism similarities, results exhibit significant variations under diverse network conditions. The SB formalism improves precision with increasing data but relies heavily on network capacity. Conversely, the G formalism takes advantage of the separation of energy terms to directly impose the first and second laws of thermodynamics. This effect enhances system robustness, enabling valid reconstruction with less data, improving generalization, and limiting overfitting. 1. C. Eldred, F. Gay-Balmaz, J. Phys. A Math. Theor. 53, 395701 (2020). 2. Q. Hernández, et al., J. Comput. Phys. 426, 109950 (2021). 3. H. Yu, et al., Phys. Rev. Fluids. 6 (2021).

ABSTRACT. Deep Material Network (DMN) has recently emerged as a data-driven surrogate model for heterogeneous materials. Compared to other neural networks, DMN distinguishes itself by the capability to encode directly the morphology of a particular microstructure through its fitting parameters. After an offline training solely based on linear elastic data generated by computational homogenization, the trained model is able to accurately extrapolate to history-dependent complex inelastic behaviors such as plasticity.

In this work, a novel micromechanics-informed parametric DMN (MIpDMN) architecture is proposed for multiscale materials with a varying microstructure described by several parameters. A single-layer feedforward neural network is used to account for the dependence of DMN parameters on the microstructural ones. Micromechanical constraints are prescribed both on the architecture and the outputs of this new neural network. Offline training is performed by minimizing a loss function aggregating the data generated at various morphologies. The proposed MIpDMN has been tested numerically with success for different parameterized microstructures. The trained MIpDMN is able to predict accurately the structure-property relationships for linear and nonlinear behaviors and demonstrates satisfying generalization capabilities when morphology varies. In addition, significant speed-up can be obtained in terms of computational time compared to full scale finite element simulations.

The proposed MIpDMN is also recast in a multiple physics setting, through an adequate redefinition of the DMN laminate homogenization function (building block of the neural network). Numerical simulations indicate that physical properties other than the mechanical ones such as thermal conductivity and coefficient of thermal expansion can also be predicted using the same model trained previously on isothermal data. This shows that MIpDMN learns the parameterized microstructure per se, and not a physical property in particular.

ABSTRACT. Multi-scale simulations can be accelerated by substituting the micro-scale problem resolution by a surrogate trained from off-line simulations. In the context of history-dependent materials, recurrent neural networks have widely been considered to act as such a surrogate, e.g. [1], since their hidden variables allow for a memory effect.

However, defining a training dataset which virtually covers all the possible strain-stress state evolution encountered during the online phase remains a daunting task. This is particularly true in the case in which the strain increment size is expected to vary by several orders of magnitude. Self-Consistent recurrent networks were thus introduced in [2] to reinforce the self-consistency of neural network predictions when small strain increments are expected. This new cell was applied to substitute an elasto-plastic material model.

However when considering a representative volume element response in the context of multi-scale simulations, it was found that the Self-Consistent recurrent networks requires a long training process. In this work, we revisit the Self-Consistent recurrent unit to improve the training performance and reduce the number of trainable variables for the neural network to act as a composite surrogate model in multi-scale simulations.

This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement No. 101056682.

[1] L. Wu, V. D. Nguyen, N. G. Kilingar, L. Noels, A recurrent neural network-accelerated multi-scale model for elasto-plastic heterogeneous materials subjected to random cyclic and non-proportional loading paths, Computer Methods in Applied Mechanics and Engineering (2020). [2] C. Bonatti, D. Mohr, On the importance of self-consistency in recurrent neural network models representing elasto-plastic solids, Journal of the Mechanics and Physics of Solids (2022). [3] L. Wu, L. Noels, Self-consistent Reinforced minimal Gated Recurrent Unit for surrogate modeling of history-dependent nonlinear problem: application to history-dependent homogenized response of heterogeneous materials. In Preparation.