Peridynamics based model of anticrack-type fracture in brittle foams

Shucheta Shegufta, Michael Zaiser · 2025

Read Peridynamics based model of anticrack-type fracture in brittle foams by Shucheta Shegufta, Michael Zaiser (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-09784.txt.

Introduction

Shucheta Shegufta and Michael Zaiser model an odd failure mode of porous brittle materials: cracks that advance under compression instead of tension. Called anticracks, these have been seen in snow, porous sandstone, and brittle foams. Ordinary fracture intuition does not cover them, since squeezing usually closes cracks rather than growing them.

The authors build their model inside bond-based peridynamics, a framework that replaces local stresses with bonds between points. Porosity enters as randomly deleted bonds, representing disorder well above the pore scale. They then apply the framework to experimental measurements of anticrack propagation in silicate foams. The work aims to give engineers and snow scientists a predictive handle on compressive collapse.

Catalog record

CREATORShucheta Shegufta, Michael Zaiser
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE31 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-09784

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Peridynamics based model of anticrack-type fracture in brittle foams Shucheta Shegufta, Michael Zaiser Abstract A particular failure mode of highly porous brittle materials consists in the propagation of cracks under uniaxial compressive loads. Such ’anticracks’ have been observed in a range of materials, from snow and porous sandstone to brittle foams. Here we present a computational model for the formation and propagation of anticrack-type failure in porous materials within the general computational framework of bond-based peridynamics. Random porosity is represented, on a scale well above the characteristic pore size, by random bond deletion (dilution disorder). We apply our framework to experimental data on anticrack propagation in silicate foams. ∎ 1 Introduction In the regime of brittle or quasi-brittle behavior, highly porous materials allow for an at first glance counter-intuitive failure mode: failure may occur under loading conditions when the principal stresses are purely negative, a situation precluded by conventional wisdom. In case of uniaxial mode-I loading, such failure implies on the macroscopic scale a negative crack face displacement (’anticrack’), which is possible because of the formation of a damage zone around the crack faces in which the material porosity strongly decreases while microscale contact of the opposing crack faces is not yet established. Such behavior has been observed in a range of porous materials, for example sandstone ( Mollema and Antonellini 1996 ; Sternlof et al. 2005 ) , snow ( Heierli et al. 2008 ; Heierli and Zaiser 2008 ; Gaume et al. 2018 ; Adam et al. 2024 ) , and glass foams ( Heierli et al. 2012 ) . Simulation studies have reported similar behavior in cemented granular materials ( Yamaguchi et al. 2020 ) . A similar phenomenon in geosciences are pressure solution surfaces ( Fletcher and Pollard 1981 ) where negative crack face displacement is made possible by pressure induced dissolution and removal of material. We note that a second use of the term ’anticrack’, namely to denote rigid lamellar inclusions in a soft matrix ( Dundurs and Markenscoff 1989 ) , is not directly pertinent to the present work. Computational and theoretical models of anticrack-type failure have used a range of methods, including analytical fracture mechanics arguments ( Sternlof et al. 2005 ; Heierli et al. 2008 ) , discrete element method (DEM) ( Ritter et al. 2020 ; Yamaguchi et al. 2020 ) , and continuum modelling. Of particular interest for the present work are elastoplastic models that allow for volume reduction as developed for snow by Barraclough et al. 2017 , Gaume et al. 2018 , Löwe et al. 2020 and implemented for dynamic anticrack propagation in snow using the material point method by Gaume et al. 2018 . This implementation is of particular interest since it naturally takes into account dynamic fragmentation phenomena subsequent to the release of an avalanche by anticrack propagation. More generally speaking, the handling of complex fracture geometries and the contact between fracture surfaces is an indispensable feature of anticrack modelling (where ultimate contact of fracture surfaces is intrinsically part of the problem), and particle based methods like DEM or MPM are natural candidates for such problems. Here we use another mesh-free, particle-based simulation approach that allows for dealing with spontaneous emergence and self-organized development of fracture surfaces, namely bond-based peridynamics enhanced by contact forces. We parameterize and validate this approach by reference to the excellent experimental study of Heierli et al. 2012 who investigated anticrack-type failure in brittle foam glass. A description of the model is provided in Section 2 alongside a parameterization of the model for the foam glass material of Heierli et al. 2012 . Results for different loading modes are presented in Section 3 and compared with experimental findings, along with post-failure behaviour of the material. The results are summarized in Section 4 2 Model formulation and parameterization 2.1 Bond based peridynamics: basic model We use a bond-based peridynamic model as originally defined by Silling and Lehoucq 2010 . Since the experiments of Heierli et al. 2012 can be envisaged as plane strain deformation, we consider quasi-two-dimensional systems where the displacement field is a two-dimensional vector {\boldsymbol{\mathnormal{u}}}({\boldsymbol{\mathnormal{x}}})={\boldsymbol{\mathnormal{y}}}({\boldsymbol{\mathnormal{x}}})-{\boldsymbol{\mathnormal{x}}} where {\boldsymbol{\mathnormal{y}}}({\boldsymbol{\mathnormal{x}}}) is the current location of the point with material coordinates {\boldsymbol{\mathnormal{x}}} . The force balance equation for the point {\boldsymbol{\mathnormal{x}}} is given by \rho({\boldsymbol{\mathnormal{x}}})\ddot{{\boldsymbol{\mathnormal{u}}}}({\boldsymbol{\mathnormal{x}}})=\int_{{\cal H}_{{\boldsymbol{\mathnormal{x}}}}}{\boldsymbol{\mathnormal{f}}}\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right)\text{\rm d}{\boldsymbol{\mathnormal{x}}}^{\prime}+{\boldsymbol{\mathnormal{b}}}({\boldsymbol{\mathnormal{x}}}), (1) where {\boldsymbol{\mathnormal{f}}}\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right) is the pair force between {\boldsymbol{\mathnormal{x}}}^{\prime} and {\boldsymbol{\mathnormal{x}}} , {\boldsymbol{\mathnormal{b}}} is a body force field, and interactions are restricted to a family {\cal H}_{{\boldsymbol{\mathnormal{x}}}} which we take to be a circle of radius \delta , the so-called horizon, around {\boldsymbol{\mathnormal{x}}} , \left|{\boldsymbol{\mathnormal{x}}}-{\boldsymbol{\mathnormal{x}}}^{*}\right|\leq\delta\;\forall\;{\boldsymbol{\mathnormal{x}}}^{*}\in{\cal H}_{{\boldsymbol{\mathnormal{x}}}} . The pair force is specified constitutively. We introduce the notations \boldsymbol{\xi}\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right)={\boldsymbol{\mathnormal{x}}}^{\prime}-{\boldsymbol{\mathnormal{x}}} and \boldsymbol{\eta}\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right)={\boldsymbol{\mathnormal{u}}}\left({\boldsymbol{\mathnormal{x}}}^{\prime}\right)-{\boldsymbol{\mathnormal{u}}}\left({\boldsymbol{\mathnormal{x}}}\right) . The pair force is then taken linearly proportional to the bond stretch s , and pointing in the direction of the vector {\boldsymbol{\mathnormal{e}}}_{u} connecting both points in the current configuration: {\boldsymbol{\mathnormal{f}}}\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right)=c\left(\boldsymbol{\xi}\right)s{\boldsymbol{\mathnormal{e}}}_{u}\left(\boldsymbol{\xi},\boldsymbol{\eta}\right)\quad,\quad s=\frac{\left|\boldsymbol{\eta}+\boldsymbol{\xi}\right|-\left|\boldsymbol{\xi}\right|}{|\boldsymbol{\xi}|}\quad,\quad{\boldsymbol{\mathnormal{e}}}_{u}\left(\boldsymbol{\xi},\boldsymbol{\eta}\right)=\frac{\boldsymbol{\eta}+\boldsymbol{\xi}}{\left|\boldsymbol{\eta}+\boldsymbol{\xi}\right|} (2) Here c(\boldsymbol{\xi}) is the so-called bond micro-modulus which for an isotropic bulk material depends on the bond length \xi=|\boldsymbol{\xi}| only. In the following we use for simplicity a constant micro-modulus, c(\xi)=c_{0} , which we choose such that the behavior of the material under homogeneous small deformations matches an isotropic linear-elastic material of bulk modulus k , hence c_{0}=12k/\pi\delta^{3} . The bond energy can then be written in terms of the bond length \xi and bond stretch s as e\left({\boldsymbol{\mathnormal{x}}},{\boldsymbol{\mathnormal{x}}}^{\prime}\right)=e\left({\boldsymbol{\mathnormal{x}}}^{\prime},{\boldsymbol{\mathnormal{x}}}\right)=\frac{c_{0}}{2}s^{2}\xi. (3) Elastic-brittle behavior is normally introduced by defining a critical bond stretch s_{\rm c} in such a manner that, once s>s_{\rm c} , the bond fails: in this case, its micro-modulus is irreversibly set to zero. The elastic energy \frac{c_{0}}{2}s_{\rm c}^{2}\xi of the failed bond is then converted into defect energy E_{\rm d} ; for our later considerations it is irrelevant whether the defect energy is considered as some kind of microscopically stored, non recoverable internal energy, or simply as heat. In the present work we consider the mechanical behavior of a microscopically brittle material, namely closed-cell glass foams, for which a modified failure rule is needed as two failure thresholds are required: Bond failure can occur either for s<-s_{\rm cb} , representing compressive buckling of the cell walls with immediate brittle failure, or for s>s_{\rm tf} , representing tensile failure. For numerical implementation, we consider discretization of the continuous medium in terms of a regular grid of collocation points with spacing \Delta . This numerical implementation converges to the limit of Eq. ( 1 ) as m=\delta/\Delta\to\infty (so-called m convergence). 2.2 Porosity and elastic properties To mimic randomly distributed voids, we build upon the model of Chen et al. 2019 . These authors proposed a mesoscale ’intermediately-homogenized’ (IH) peridynamic model and applied it to wave propagation in glass foams. Their basic idea is to formulate a model which does not resolve single voids or other microstructural details, but nevertheless accounts for the porosity dependency of elastic properties as well as for microstructural disorder typically associated with large degrees of porosity. For a material with porosity \phi and critical porosity \phi_{\rm c} (the porosity where the material loses cohesion), this is done by the simple expedient of randomly deleting a fraction d_{\phi}=1-(1-\phi/\phi_{\rm c})^{2} of all bonds. For the limit of sufficiently large m , Chen et al. 2019 demonstrated that elastic moduli E_{\phi} and wave

Read the complete text