betaFeedback

WorksComputer Science

Scalar Entropic Damage Transport: Well-posed softening with a single damage variable

Abstract

Entropic Damage Transport (EDT) represents material softening via a local continuum damage model while ensuring the well-posedness of the mechanical boundary value problem. It is based on a convexification of the time-incremental potential associated with a rate-limited scalar damage model. A one-dimensional distribution of damage states serves as the internal variable, and the number of these states grows with the number of time steps. We show that the emergence of the damage distribution is an artifact of the time discretization. Surprisingly, the distribution concentrates at a single damage state in the time-continuous limit, and EDT converges to a rate-limited damage model with a scalar damage variable, whose free energy is jointly convex in the strain and the reciprocal degradation. We exploit this hidden convexity by a frozen-metric discretization in terms of the reciprocal degradation, which renders every increment jointly convex. The resulting scalar update preserves damage irreversibility, the prescribed damage-rate limitation and thermomechanical consistency, reduces to a scalar equation and provides a symmetric and positive definite algorithmic tangent in closed form. Computational experiments assess the accuracy and the efficiency of both formulations from the material point to the microstructure level, and demonstrate that the scalar update permits a finite-element computation of industrial size with more than 14 million degrees of freedom.

Introduction

State of the art

Local continuum damage mechanics describes the progressive loss of load-carrying capacity of a material by a degradation variable affecting the free energy. Its first concepts emerged from creep-rupture studies [1, 2] and were subsequently embedded into thermodynamically consistent constitutive frameworks [3, 4, 5], see the review by Krajcinovic [6] and the monograph by Murakami [7]. Various variables are in use to carry the damage, among them scalar degradations and damage tensors [8] as well as stiffness- or compliance-based variables [9, 10].

Softening renders such models useful for failure prediction [11], for instance of fiber-reinforced composites [12, 13, 14]. Computations which resolve the microstructure, be it of a lab-size coupon [15] or of the unit cells of a multiscale chain [16, 17, 18], typically evaluate a local constitutive law at every material point. In such a finite-element computation, however, damage tends to localize into a single band of elements whose width decreases with the element size. Thus, the material response depends on the mesh, and, without a regularization, the energy dissipated in the band typically tends to zero under refinement, as the dissipation per unit volume remains bounded whereas the volume of the band vanishes [19]. This deficiency of rate-independent softening models was recognized early [20] and motivated nonlocal remedies [21]. For a comparison of these remedies, see Jirásek [22], and for a review, see Bažant and Jirásek [23]. Mathematically, softening typically deprives the incremental potential of a local damage model of its quasiconvexity, so that minimizers of the associated boundary value problem need not exist [24, 25]. At the level of rates, localization may be accompanied by a loss of ellipticity of the boundary value problem for the displacement rate, which is detected when the acoustic tensor formed with the tangent stiffness ceases to be positive definite for some direction of propagation [26].

The remedies proposed in the literature fall into four classes.

  • Mesh adjustment. Engineering practice commonly adapts the constitutive law to the mesh. The crack-band model [27] ties the band width to a characteristic element length, which fixes the dissipated energy by a deliberate adjustment of the softening law to this length. Phenomenological approaches such as GISSMO [28], widespread in crashworthiness and sheet metal forming, reduce the mesh sensitivity by tabulating failure parameters over the element size. These parameters are tied to the discretization for which they were identified and have to be identified anew for every element size. The adjustment regularizes the dissipated energy, not the boundary value problem, whose solution still localizes into a single row of elements aligned with the mesh.

  • Rate limitation. This class starts from the observation that inertia bounds the speed at which a crack propagates [29] and thus how fast the stored energy may be dissipated. A quasi-static computation neglects inertia, so that nothing prevents the damage from jumping within a single time step, and the constitutive law has to supply this limit instead. Viscous models penalize a rapid damage evolution [30, 31, 32], and the delay-damage models bound the damage rate by a characteristic time [33, 34, 35]. In dynamics, this time and the elastic wave speed combine into an internal length [19], whose regularizing effect fades for slow loading and may be absent for dynamic failure out of rest [36]. Quasi-statically, the rate limitation bounds the damage increment per time step without introducing a length and is thus no spatial regularization by itself. Moreover, it generally alters the elastic domain, i.e., the range of driving forces under which the damage remains inactive, unless the threshold is adjusted accordingly. Without a term linear in the damage rate, the elastic domain is lost altogether, as the damage evolves under any positive driving force unless the free energy provides a threshold [37, 38]. Viscous regularization was also combined with relaxation [39, 40], where the required viscosity may be unphysically high and its effect depends on the loading rate [41, 42].

  • Nonlocal and gradient-extended models. If the damage at a material point depends on its neighborhood, a localization band cannot become narrower than an internal length, and the dissipated energy

does not vanish under mesh refinement. This class comprises nonlocal averaging [43, 44, 45], gradient-extended [46, 47, 48] and micromorphic formulations [49, 50] as well as phase-field fracture [51, 52, 53]. Mathematically, the direct method requires coercivity and weak lower semicontinuity [25], and without quasiconvexity the latter fails, so that minimizing sequences oscillate on ever finer scales. Nonlocal and gradient terms upgrade weak to strong convergence of the damage field, by a compact Sobolev embedding or by the convolution with a fixed kernel. For a coercive functional, minimizers thus exist [54], which, however, need not be unique or depend continuously on the load. In practice, the incremental energy of gradient-damage and phase-field models remains nonconvex jointly in the displacement and the damage, so that the solution of a time step may be nonunique and jump under a steady increase of the prescribed loading [55, 56]. Moreover, the internal length is typically not accessible to measurement [53], and if it approaches the size of the microstructure, first-order homogenization lacks the separation of scales it requires [57, 58, 59]. Finally, the additional global field enlarges the system solved in every increment [41, 60].

  • Relaxation. A nonconvex incremental potential favors the formation of microstructures on ever finer scales [61], which a finite-element computation resolves only down to the element size. Relaxation replaces the incremental potential by a generalized convex envelope – the quasiconvex, the rank-one convex or the convex one [62]. In contrast to nonlocal and gradient terms, it does not upgrade weak to strong convergence but renders the energy compatible with weak limits, as the quasiconvex envelope is weakly lower semicontinuous [25]. Relaxation does not restore coercivity, however. As the degradation of a softening model tends to zero, the incremental energy loses its coercivity in the strain, and minimizing sequences need not be bounded. The envelope stands for a microstructure on a finer scale [61, 63, 64], which relaxed damage models carry along in the evolution [39, 65, 66]. No efficient algorithm is known for the quasiconvex envelope [67, 68], one reason being that quasiconvexity lacks a local characterization [69]. Hence, the rank-one convex envelope [67, 70] or the convex one [38, 68] is taken instead, although rank-one convexity does not imply weak lower semicontinuity. Convexity alone ensures neither existence nor uniqueness. A lower semicontinuous and strongly convex incremental energy, however, admits a unique minimizer, which depends Lipschitz-continuously on the prescribed load [71]. Relaxation was also combined with a viscous regularization [40].

The work at hand is concerned with Damage Transport, which belongs to the fourth class but builds on the second. It originates from the work of Schwarz et al. [38], who took up the bounded damage rate of the delay-damage models [33, 35] to supply the coercivity which relaxation alone lacks, limited the damage increment of a time step accordingly and approximated the convex envelope of the resulting condensed energy numerically. Lauff et al. [68] computed this envelope in closed form, revealing that the internal variable becomes a distribution of damage states and the dissipation a transport cost between such distributions [72]. Subsequently, they introduced an entropic regularization, as is common in computational optimal transport [73, 74], which renders the condensed incremental potential uniformly strongly convex [75]. As a consequence, the incremental boundary value problem of small-strain elasticity is well-posed in the sense of Hadamard, i.e., a solution exists, is unique and depends continuously on the prescribed load.

For the resulting model, a fast material-point solver [76] and a redistribution strategy onto a fixed set of damage states at exact stiffness [77] are available. The redistribution permits adaptive time steps and reduces the memory footprint of the implementation, at the price of an additional approximation of the distribution. Entropic Damage Transport thus combines a softening response with a local model, requires neither an additional field nor an internal length and leads to well-posed incremental boundary value problems, which renders it a promising approach. Its internal variable, however, remains a discrete distribution of damage states, no matter how efficiently it is updated.

Contributions

Entropic Damage Transport was formulated for a fixed time discretization, and its damage distribution emerges from the time steps. We therefore asked ourselves whether more suitable discretizations exist, for instance ones which approximate the damage distribution by a fixed number of states.

To this end, we identified the continuous-time model which Entropic Damage Transport discretizes, and the answer is surprising: The effective degradation of Entropic Damage Transport converges to that of a scalar rate-limited damage model, and the damage distribution concentrates at the damage state of this model, with a mean square distance bounded by a multiple of the time step size, see Appendix § 10.§ 1 for the argument at material-point level. Hence, there is no need to approximate a distribution, and a scalar internal variable suffices, provided its time discretization retains the convexity of the increment.

To transfer the favorable properties of Damage Transport to a scalar damage variable, we reveal a hidden convex structure of the underlying scalar model. Used as the internal variable, the reciprocal degradation enters the damaged compliance linearly, and the free energy inherits the joint convexity of the Hooke energy in the strain and the compliance [78]. The relaxed free energy of Damage Transport carries the linear average of this very quantity over the damage distribution [68, 75]. Thus, both free energies share the same convex structure. This joint convexity permits eliminating the damage distribution altogether, instead of updating it more efficiently [76, 77].

However, in the scalar setting, the dissipation potential becomes state-dependent, as a metric converts the rate of the reciprocal degradation into the damage rate. As a result, the fully implicit increment loses its joint convexity beyond a critical time step size wherever the second derivative of the reciprocal degradation is positive.

We propose the frozen-metric discretization, which ensures strong convexity of the condensed incremental potential. In more detail, it evaluates the metric at the previous state and adapts the semi-implicit schemes used for gradient systems with a state-dependent kinetic structure [79, 80] to the derivative of the reciprocal degradation with respect to the damage. The frozen-metric discretization is a time-discrete model in its own right, which shares the continuous-time limit of Entropic Damage Transport at material-point level but differs from Entropic Damage Transport at finite time step sizes. The scalar model underlying Entropic Damage Transport may be considered as a delay-damage model [33, 35] whose damage-rate limitation is enforced by a logarithmic barrier. The key novelty lies in its internal variable, which reveals the hidden convex structure, and in its time discretization, which preserves this structure. In particular, our findings reveal that Entropic Damage Transport belongs, in fact, to the second class of mitigation strategies. Its well-posedness stems from the damage-rate limitation, combined with a suitable internal variable and time discretization, and not from the relaxation. In retrospect, the relaxation of Schwarz et al. [38] was dispensable, as a more suitable time discretization of the rate-limited model would have sufficed.

The designed model offers the following advantages.

  • Each material point carries a single scalar internal variable, the reciprocal degradation.

  • For an initial damage distribution concentrated at a single state, the effective degradation of Entropic Damage Transport and the frozen-metric states converge to one and the same scalar evolution at material-point level as the time step size vanishes, see Appendix § 10.

  • The increment is jointly convex in the strain and the reciprocal degradation for every positive time step size. The consistent algorithmic material tangent is symmetric and uniformly positive definite. Moreover, it is available in closed form.

  • The admissible states of one increment form an interval known in closed form. The update preserves damage irreversibility and the prescribed damage-rate limitation, and it satisfies the reduced discrete Clausius–Duhem inequality. Computing the update requires solving a single scalar equation, for which the undamped Newton method converges monotonically.

In § 2.1, we describe the scalar model underlying Damage Transport, § 2.2 introduces the reciprocal degradation as the internal variable and § 2.3 the frozen-metric increment. § 3 compares both formulations at material-point level, on a fiber-reinforced unit cell and on a notched specimen, and applies the scalar update to a jet-engine bracket with more than 14 million degrees of freedom. In Appendix § 10, we provide the convergence arguments at material-point level, and Appendix § 11 collects the arguments for the properties of the frozen-metric update.

Hidden convexity and frozen-metric time discretization

The scalar model underlying Damage Transport

We consider the rate-limited scalar damage model underlying the relaxation approach by Schwarz et al. [38] and the subsequent Damage-Transport framework [68, 75]. The model is formulated as a generalized standard material [81] in an isothermal small-strain setting. Let

w0(ε)=12ε:C:ε,ε∈Sym⁡(3),(1)w_0(\varepsilon) = \frac{1}{2}\varepsilon: \mathbf{C} : \varepsilon,\quad\varepsilon\in\operatorname{Sym}(3), \tag*{(1)}

denote the undamaged elastic energy, where Sym⁡(3)\operatorname{Sym}(3) stands for the space of symmetric second-order tensors in three dimensions and the elasticity tensor C:Sym⁡(3)→Sym⁡(3)\mathbf{C} : \operatorname{Sym}(3) \to\operatorname{Sym}(3) is symmetric and positive definite. Let the scalar damage variable take values in the interval

d∈D:=[0,dmax⁡),0<dmax⁡≤∞,(2)d \in\mathcal{D} := [0,d_{\max}),\quad0 < d_{\max} \le\infty, \tag*{(2)}

where the maximum damage dmax⁡d_{\max} may be finite or infinite. We consider a given degradation function

f:D→(0,1](3)f : \mathcal{D} \to(0,1] \tag*{(3)}

which is continuous and strictly monotonically decreasing and satisfies the conditions

f(0)=1andlim⁡d→dmax⁡f(d)=0.(4)f(0) = 1 \quad\text{and} \quad\lim_{d \to d_{\max}} f(d) = 0. \tag*{(4)}

Following the structural assumptions of Damage Transport [68, 75], we introduce the reciprocal degradation

g:D→[1,∞),g(d)=1f(d),(5)g : \mathcal{D} \to[1,\infty),\quad g(d) = \frac{1}{f(d)}, \tag*{(5)}

and assume that the function gg is strictly convex. The conditions (2.4) directly imply

g(0)=1andlim⁡d→dmax⁡g(d)=∞.(6)g(0) = 1 \quad\text{and} \quad\lim_{d \to d_{\max}} g(d) = \infty. \tag*{(6)}

As the reciprocal of the degradation function ff, which is positive for all damage values, the function gg is continuous, strictly increasing and, by the boundary conditions (2.6), a bijection from the interval D\mathcal{D} onto the interval [1,∞)[1,\infty). For the differential arguments below, we additionally suppose that the conditions

g∈C2(D),g′(d)>0andg′′(d)≥0,d∈D,(7)g \in C^2(\mathcal{D}),\quad g'(d) > 0 \quad\text{and} \quad g''(d) \ge0,\quad d \in\mathcal{D}, \tag*{(7)}

hold. A prominent example satisfying the stated assumptions on an unbounded damage interval is the exponential degradation [82]

f(d)=exp⁡(−d)andg(d)=exp⁡(d),d∈[0,∞).(8)f(d) = \exp(-d) \quad\text{and} \quad g(d) = \exp(d),\quad d \in[0,\infty). \tag*{(8)}

With the undamaged energy (2.1) and the degradation function (2.3), the free energy is defined by

w(ε,d):=f(d)w0(ε)=w0(ε)g(d),ε∈Sym⁡(3)andd∈D.(9)w(\varepsilon,d) := f(d)w_0(\varepsilon) = \frac{w_0(\varepsilon)}{g(d)},\quad\varepsilon\in\operatorname{Sym}(3) \quad\text{and} \quad d \in\mathcal{D}. \tag*{(9)}

The corresponding stress computes as

σ:=∂w∂ε(ε,d)=f(d)C:ε.(10)\boldsymbol{\sigma} := \frac{\partial w}{\partial\varepsilon}(\varepsilon,d) = f(d)\mathbf{C} : \varepsilon. \tag*{(10)}

The damage rate is restricted by a constitutive damage-rate limitation d˙max⁡\dot{d}_{\max}. With the unregularized damage-initiation threshold rr and the entropic regularization parameter λ\lambda intrinsic to Entropic Damage Transport [75], subject to the restrictions

d˙max⁡>0,r≥0andλ>0,(11)\dot{d}_{\max} > 0,\qquad r \ge0 \qquad\text{and}\qquad\lambda> 0, \tag*{(11)}

we consider the dissipation potential of the damage rate d˙\dot{d}

ψ(d˙):={rd˙−λd˙max⁡log⁡(1−d˙/d˙max⁡),0≤d˙<d˙max⁡,+∞,otherwise.(12)\psi(\dot{d}) := \begin{cases} r\dot{d} - \lambda\dot{d}_{\max}\log\left(1-\dot{d}/\dot{d}_{\max}\right), & 0 \le\dot{d} < \dot{d}_{\max},\\ +\infty, & \text{otherwise}. \end{cases} \tag*{(12)}

The dissipation potential (12) is a barrier version of the bounded-rate concept [35, 38]. Here, the damage-rate limitation d˙max⁡\dot{d}_{\max} carries the unit of an inverse time, whereas the threshold rr and the parameter λ\lambda are energy densities, i.e., they carry the unit of a stress. For a damage rate below the damage-rate limitation, differentiating the dissipation potential (12) yields the identities

ψ′(d˙)=r+λ1−d˙/d˙max⁡andψ′′(d˙)=λ/d˙max⁡(1−d˙/d˙max⁡)2>0.(13)\psi'(\dot{d}) = r + \frac{\lambda}{1-\dot{d}/\dot{d}_{\max}} \qquad\text{and}\qquad \psi''(\dot{d}) = \frac{\lambda/\dot{d}_{\max}}{\left(1-\dot{d}/\dot{d}_{\max}\right)^2} > 0. \tag*{(13)}

At a vanishing damage rate, both identities hold for the right-sided derivatives. Thus, the dissipation potential is strictly convex on its effective domain. Moreover, the logarithmic term diverges as the damage rate d˙\dot{d} approaches the damage-rate limitation d˙max⁡\dot{d}_{\max} and therefore prevents the upper damage-rate bound from being attained. The thermodynamic force associated with the damage variable reads

Y(ε,d):=−∂w∂d(ε,d)=−f′(d)w0(ε)=g′(d)g(d)2w0(ε).(14)Y(\varepsilon,d) := -\frac{\partial w}{\partial d}(\varepsilon,d) = -f'(d)w_0(\varepsilon) = \frac{g'(d)}{g(d)^2}w_0(\varepsilon). \tag*{(14)}

The evolution of the damage variable is governed by Biot’s equation [83]

Y(ε,d)∈∂ψ(d˙),(15)Y(\varepsilon,d) \in\partial\psi(\dot{d}), \tag*{(15)}

where ∂ψ\partial\psi denotes the subdifferential of the convex dissipation potential (12). On the active branch, inserting the first relation in Eq. (13) into Biot’s equation (15) yields the result

Y(ε,d)=r+λ1−d˙/d˙max⁡.(16)Y(\varepsilon,d) = r + \frac{\lambda}{1-\dot{d}/\dot{d}_{\max}}. \tag*{(16)}

At a vanishing damage rate, the subdifferential of the dissipation potential (12) consists of its right derivative together with the normal cone of the effective domain at its lower endpoint, which adds every smaller value,

∂ψ(0)=(−∞,r+λ].(17)\partial\psi(0) = (-\infty,r+\lambda]. \tag*{(17)}

Thus, the solution of Biot’s equation (15), obtained by solving Eq. (16) for the damage rate and accounting for the subdifferential (17), is the closed-form evolution law

d˙=d˙max⁡⟨Y(ε,d)−r−λ⟩+Y(ε,d)−r,(18)\dot{d} = \dot{d}_{\max}\frac{\left\langle Y(\varepsilon,d)-r-\lambda\right\rangle_+}{Y(\varepsilon,d)-r}, \tag*{(18)}

where ⟨⋅⟩+\langle\cdot\rangle_+ denotes the positive part and the fraction is defined to be zero whenever its numerator vanishes. Together with the definition (14), the right-hand side of Eq. (18) depends only on the current strain and damage state. In particular, damage remains inactive as long as the driving force does not exceed the total threshold,

Y(ε,d)≤r+λ,(19)Y(\varepsilon,d) \le r+\lambda, \tag*{(19)}

whereas the inequalities

0<d˙<d˙max⁡(20)0 < \dot{d} < \dot{d}_{\max} \tag*{(20)}

hold on the active branch. The sum r+λr+\lambda is thus the driving force at which damage initiates, and the parameter λ\lambda sets the scale of the overstress.

The scalar evolution (18) constitutes the continuous rate-limited damage model underlying the time-discrete Entropic Damage-Transport model by Lauff et al. [75]. At material-point level and for an initial damage distribution concentrated at a single damage state, the effective degradation of Entropic Damage Transport converges to the degradation along the scalar evolution (18) as the time step size tends to zero, see Appendix § 10.§ 1.

Reciprocal degradation and hidden convexity

The free energy (9) is not jointly convex in the strain and the damage variable, as its Hessian is indefinite wherever the strain does not vanish and the second derivative of the reciprocal degradation is positive. On the other hand, the Hooke energy

wHooke ⁣:Sym⁡(3)×Sym⁡+(Sym⁡(3))→R≥0,(ε,S)↦12ε ⁣:S−1 ⁣:ε,(21)w^{\mathrm{Hooke}} \colon\operatorname{Sym}(3) \times\operatorname{Sym}_{+}(\operatorname{Sym}(3)) \to\mathbb{R}_{\geq0}, \qquad(\varepsilon,\mathsf{S}) \mapsto\frac{1}{2}\varepsilon\colon\mathsf{S}^{-1} \colon\varepsilon, \tag*{(21)}

where Sym⁡+(Sym⁡(3))\operatorname{Sym}_{+}(\operatorname{Sym}(3)) denotes the symmetric and positive definite linear maps on $\operatorname{Sym}(3), is jointly convex in the strain and the compliance tensor [78], as it is a matrix-fractional function [84]. In the scalar damage model at hand, the damaged compliance tensor takes the form

S(d):=1f(d)C−1=g(d)C−1,d∈D.(22)\mathsf{S}(d) := \frac{1}{f(d)}\mathsf{C}^{-1} = g(d)\mathsf{C}^{-1}, \qquad d \in\mathcal{D}. \tag*{(22)}

Thus, the reciprocal degradation value g(d)g(d) acts as a scalar compliance variable. The same structure underlies the relaxed free energy of Damage Transport [68, 75], where the effective reciprocal degradation is obtained as a linear average over the damage distribution, so that a damage distribution stores the free energy of the single reciprocal degradation value at this average, see Figure 15 in Appendix § 10.§ 1. These observations suggest using the reciprocal degradation value (5) directly as the scalar internal variable, i.e., we define

Damage distribution and scalar state of equal free energy at one instant

Figure 15. Damage distribution and scalar state of equal free energy at one instant, for the exponential degradation (2.8).

z:=g(d),d∈D.(23)z := g(d), \qquad d \in\mathcal{D}. \tag*{(23)}

As the reciprocal degradation is a bijection onto the interval [1,∞)[1,\infty), the inverse function

g−1 ⁣:[1,∞)→D(24)g^{-1} \colon[1,\infty) \to\mathcal{D} \tag*{(24)}

is well-defined, and the original damage variable may be recovered from the reciprocal degradation variable

References

  1. [1]L. M. Kachanov, “Rupture Time Under Creep Conditions,” International Journal of Fracture, vol. 97, no. 1–4, pp. 11–18, 1999.DOI
  2. [2]Y. N. Rabotnov, “On the Equation of State for Creep,” Proceedings of the Institution of Mechanical Engineers, Conference Proceedings, vol. 178, no. 1, pp. 2–117–2–122, 1963.
  3. [3]J. Lemaitre, “A Continuous Damage Mechanics Model for Ductile Fracture,” Journal of Engineering Materials and Technology, vol. 107, no. 1, pp. 83–89, 1985.DOI
  4. [4]J. C. Simo and J. W. Ju, “Strain- and stress-based continuum damage models—I. Formulation,” International Journal of Solids and Structures, vol. 23, no. 7, pp. 821–840, 1987.
  5. [5]J. L. Chaboche, “Continuum Damage Mechanics: Part I—General Concepts,” Journal of Applied Mechanics, vol. 55, no. 1, pp. 59–64, 1988.DOI
  6. [6]D. Krajcinovic, “Damage mechanics,” Mechanics of Materials, vol. 8, no. 2–3, pp. 117–197, 1989.
  7. [7]S. Murakami, Continuum Damage Mechanics: A Continuum Mechanics Approach to the Analysis of Damage and Fracture, vol. 185 of Solid Mechanics and Its Applications. Dordrecht: Springer, 2012.
  8. [8]G. Z. Voyiadjis and P. I. Kattan, “A Comparative Study of Damage Variables in Continuum Damage Mechanics,” International Journal of Damage Mechanics, vol. 18, no. 4, pp. 315–340, 2009.DOI
  9. [9]M. Ortiz, “A constitutive theory for the inelastic behavior of concrete,” Mechanics of Materials, vol. 4, no. 1, pp. 67–93, 1985.DOI
  10. [10]D. Krajcinovic, “Selection of damage parameter – Art or science?,” Mechanics of Materials, vol. 28, no. 1–4, pp. 165–179, 1998.DOI
  11. [11]J. Lemaitre, “Local approach of fracture,” Engineering Fracture Mechanics, vol. 25, no. 5–6, pp. 523–537, 1986.DOI
  12. [12]A. Puck and H. Schürmann, “Failure analysis of FRP laminates by means of physically based phenomenological models,” Composites Science and Technology, vol. 62, no. 12–13, pp. 1633–1662, 2002.
  13. [13]H. Rolland, N. Saintier, and G. Robert, “Damage mechanisms in short glass fibre reinforced thermoplastic during in situ microtomography tensile tests,” Composites Part B: Engineering, vol. 90, pp. 365–377, 2016.DOI
  14. [14]I. Hanhan and M. D. Sangid, “Damage propagation in short fiber thermoplastic composites analyzed through coupled 3D experiments and simulations,” Composites Part B: Engineering, vol. 218, p. 108931, 2021.DOI
  15. [15]V. Laheri, P. Hao, and F. A. Gilabert, “Direct embedment of RVE-based microscale into lab size coupons to research fracture process in unidirectional and bidirectional composites,” Composites Science and Technology, vol. 235, p. 109949, 2023.DOI
  16. [16]J. Spahn, H. Andrä, M. Kabel, and R. Müller, “A multiscale approach for modeling progressive damage of composite materials using fast Fourier transforms,” Computer Methods in Applied Mechanics and Engineering, vol. 268, pp. 871–883, 2014.DOI
  17. [17]J. Köbler, N. Magino, H. Andrä, F. Welsching­er, R. Müller, and M. Schneider, “A computational multi-scale model for the stiffness degradation of short-fiber reinforced plastics subjected to fatigue loading,” Computer Methods in Applied Mechanics and Engineering, vol. 373, p. 113522, 2021.DOI
  18. [18]N. Magino, J. Köbler, H. Andrä, F. Welschinger, R. Müller, and M. Schneider, “A multiscale high-cycle fatigue-damage model for the stiffness degradation of fiber-reinforced materials based on a mixed variational framework,” Computer Methods in Applied Mechanics and Engineering, vol. 388, p. 114198, 2022.DOI
  19. [19]R. de Borst, L. J. Sluys, H.-B. Mühlhaus, and J. Pamin, “Fundamental issues in finite element analyses of localization of deformation,” Engineering Computations, vol. 10, no. 2, pp. 99–121, 1993.DOI
  20. [20]Z. P. Bažant, “Instability, ductility, and size effect in strain-softening concrete,” Journal of the Engineering Mechanics Division, vol. 102, no. 2, pp. 331–344, 1976.DOI
  21. [21]Z. P. Bažant, T. B. Belytschko, and T.-P. Chang, “Continuum theory for strain-softening,” Journal of Engineering Mechanics, vol. 110, no. 12, pp. 1666–1692, 1984.DOI
  22. [22]M. Jirásek, “Nonlocal models for damage and fracture: Comparison of approaches,” International Journal of Solids and Structures, vol. 35, no. 31–32, pp. 4133–4145, 1998.DOI
  23. [23]Z. P. Bažant and M. Jirásek, “Nonlocal integral formulations of plasticity and damage: Survey of progress,” Journal of Engineering Mechanics, vol. 128, no. 11, pp. 1119–1149, 2002.
  24. [24]C. B. Morrey, “Quasi-convexity and the lower semicontinuity of multiple integrals,” Pacific Journal of Mathematics, vol. 2, no. 1, pp. 25–53, 1952.DOI
  25. [25]B. Dacorogna, “Quasiconvexity and relaxation of nonconvex problems in the calculus of variations,” Journal of Functional Analysis, vol. 46, no. 1, pp. 102–118, 1982.DOI
  26. [26]A. Benallal, R. Billardon, and G. Geymonat, “Bifurcation and Localization in Rate-Independent Materials. Some General Considerations,” in Bifurcation and Stability of Dissipative Systems, pp. 1–44, Vienna: Springer, 1993.
  27. [27]Z. P. Bažant and B. H. Oh, “Crack band theory for fracture of concrete,” Matériaux et Constructions, vol. 16, no. 3, pp. 155–177, 1983.DOI
  28. [28]F. X. C. Andrade, M. Feucht, A. Haufe, and F. Neukamm, “An incremental stress state dependent damage model for ductile failure prediction,” International Journal of Fracture, vol. 200, no. 1–2, pp. 127–150, 2016.DOI
  29. [29]L. B. Freund, Dynamic Fracture Mechanics. Cambridge: Cambridge University Press, 1990.
  30. [30]A. Needleman, “Material rate dependence and mesh sensitivity in localization problems,” Computer Methods in Applied Mechanics and Engineering, vol. 67, no. 1, pp. 69–85, 1988.DOI
  31. [31]L. J. Sluys and R. de Borst, “Wave propagation and localization in a rate-dependent cracked medium—model formulation and one-dimensional examples,” International Journal of Solids and Structures, vol. 29, no. 23, pp. 2945–2958, 1992.DOI
  32. [32]J.-F. Dubé, G. Pijaudier-Cabot, and C. L. Borderie, “Rate Dependent Damage Model for Concrete in Dynamics,” Journal of Engineering Mechanics, vol. 122, no. 10, pp. 939–947, 1996.DOI
  33. [33]O. Allix, P. Feissel, and P. Thévenet, “A delay damage mesomodel of laminates under dynamic loading: basic aspects and identification issues,” Computers & Structures, vol. 81, no. 12, pp. 1177–1191, 2003.DOI
  34. [34]A. Suffis, T. A. A. Lubrecht, and A. Combescure, “Damage model with delay effect: Analytical and numerical studies of the evolution of the characteristic damage length,” International Journal of Solids and Structures, vol. 40, no. 13–14, pp. 3463–3476, 2003.
  35. [35]O. Allix, “The bounded rate concept: A framework to deal with objective failure predictions in dynamic within a local constitutive model,” International Journal of Damage Mechanics, vol. 22, no. 6, pp. 808–828, 2013.DOI
  36. [36]J. Zghal and N. Moës, “Analysis of the delayed damage model for three one-dimensional loading scenarii,” Comptes Rendus. Physique, vol. 21, no. 6, pp. 527–537, 2020.DOI
  37. [37]P. Junker and D. Balzani, “An extended Hamilton principle as unifying theory for coupled problems and dissipative microstructure evolution,” Continuum Mechanics and Thermodynamics, vol. 33, no. 4, pp. 1931–1956, 2021.
  38. [38]S. Schwarz, P. Junker, and K. Hackl, “Variational regularization of damage models based on the emulated RVE,” Continuum Mechanics and Thermodynamics, vol. 33, no. 1, pp. 69–95, 2021.
  39. [39]P. Junker, S. Schwarz, J. Makowski, and K. Hackl, “A relaxation-based approach to damage modeling,” Continuum Mechanics and Thermodynamics, vol. 29, no. 1, pp. 291–310, 2017.
  40. [40]K. Langenfeld, P. Junker, and J. Mosler, “Quasi-brittle damage modeling based on incremental energy relaxation combined with a viscous-type regularization,” Continuum Mechanics and Thermodynamics, vol. 30, no. 5, pp. 1125–1144, 2018.DOI
  41. [41]P. Junker, J. Riesselmann, and D. Balzani, “Efficient and robust numerical treatment of a gradient-enhanced damage model at large deformations,” International Journal for Numerical Methods in Engineering, vol. 123, no. 3, pp. 774–793, 2022.arxiv.org/abs/2102.08819
  42. [42]K. Langenfeld, K. Purzeja, and J. Mosler, “How regularization concepts interfere with (quasi-)brittle damage: a comparison based on a unified variational framework,” Continuum Mechanics and Thermodynamics, vol. 34, no. 6, pp. 1517–1544, 2022.DOI
  43. [43]G. Pijaudier-Cabot and Z. P. Bažant, “Nonlocal damage theory,” Journal of Engineering Mechanics, vol. 113, no. 10, pp. 1512–1533, 1987.
  44. [44]Z. P. Bažant and G. Pijaudier-Cabot, “Nonlocal Continuum Damage, Localization Instability and Convergence,” Journal of Applied Mechanics, vol. 55, no. 2, pp. 287–293, 1988.
  45. [45]E. Lorentz and S. Andrieux, “Analysis of non-local models through energetic formulations,” International Journal of Solids and Structures, vol. 40, no. 12, pp. 2905–2936, 2003.DOI
  46. [46]R. H. J. Peerlings, R. de Borst, W. A. M. Brekelmans, and J. H. P. de Vree, “Gradient enhanced damage for quasi-brittle materials,” International Journal for Numerical Methods in Engineering, vol. 39, no. 19, pp. 3391–3403, 1996.doi.org/10.1002/(sici)1097-0207(19961015)39:19<3391::aid-nme7>3.0.co;2-d
  47. [47]R. H. J. Peerlings, M. G. D. Geers, R. de Borst, and W. A. M. Brekelmans, “A critical comparison of nonlocal and gradient-enhanced softening continua,” International Journal of Solids and Structures, vol. 38, no. 44–45, pp. 7723–7746, 2001.DOI
  48. [48]B. J. Dimitrijevic and K. Hackl, “A regularization framework for damage-plasticity models via gradient enhancement of the free energy,” International Journal for Numerical Methods in Biomedical Engineering, vol. 27, no. 8, pp. 1199–1210, 2011.
  49. [49]S. Forest, “Micromorphic Approach for Gradient Elasticity, Viscoplasticity, and Damage,” Journal of Engineering Mechanics, vol. 135, no. 3, pp. 117–131, 2009.DOI
  50. [50]T. Brepols, S. Wulfinghoff, and S. Reese, “Gradient-extended two-surface damage-plasticity: Micromorphic formulation and numerical aspects,” International Journal of Plasticity, vol. 97, pp. 64–106, 2017.DOI
  51. [51]G. A. Francfort and J.-J. Marigo, “Revisiting brittle fracture as an energy minimization problem,” Journal of the Mechanics and Physics of Solids, vol. 46, no. 8, pp. 1319–1342, 1998.
  52. [52]C. Miehe, M. Hofacker, and F. Welschinger, “A phase field model for rate-independent crack propagation: Robust algorithmic implementation based on operator splits,” Computer Methods in Applied Mechanics and Engineering, vol. 199, no. 45–48, pp. 2765–2778, 2010.DOI
  53. [53]J.-Y. Wu and V. P. Nguyen, “A length scale insensitive phase-field damage model for brittle fracture,” Journal of the Mechanics and Physics of Solids, vol. 119, pp. 20–42, 2018.DOI
  54. [54]A. Mielke and T. Roubíček, Rate-Independent Systems: Theory and Application, vol. 193 of Applied Mathematical Sciences. New York: Springer, 2015.DOI
  55. [55]K. Pham, J.-J. Marigo, and C. Maurini, “The issues of the uniqueness and the stability of the homogeneous response in uniaxial tests with gradient damage models,” Journal of the Mechanics and Physics of Solids, vol. 59, no. 6, pp. 1163–1190, 2011.DOI
  56. [56]P. Farrell and C. Maurini, “Linear and nonlinear solvers for variational phase-field models of brittle fracture,” International Journal for Numerical Methods in Engineering, vol. 109, no. 5, pp. 648–667, 2017.arxiv.org/abs/1511.08463
  57. [57]W. J. Drugan and J. R. Willis, “A micromechanics-based nonlocal constitutive equation and estimates of representative volume element size for elastic composites,” Journal of the Mechanics and Physics of Solids, vol. 44, no. 4, pp. 497–524, 1996.DOI
  58. [58]I. M. Gitman, H. Askes, and L. J. Sluys, “Representative volume: Existence and size determination,” Engineering Fracture Mechanics, vol. 74, no. 16, pp. 2518–2534, 2007.DOI
  59. [59]E. W. C. Coenen, V. G. Kouznetsova, E. Bosco, and M. G. D. Geers, “A multi-scale approach to bridge microscale damage and macroscale failure: a nested computational homogenization-localization framework,” International Journal of Fracture, vol. 178, no. 1–2, pp. 157–178, 2012.DOI
  60. [60]Y. Navidtehrani, C. Betegón, and E. A. Martínez-Pañeda, “A simple and robust Abaqus implementation of the phase field fracture method,” Applications in Engineering Science, vol. 6, p. 100050, 2021.arxiv.org/abs/2104.08132
  61. [61]C. Carstensen, K. Hackl, and A. Mielke, “Non-convex potentials and microstructures in finite-strain plasticity,” Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences, vol. 458, no. 2018, pp. 299–317, 2002.DOI
  62. [62]D. Balzani and M. Ortiz, “Relaxed incremental variational formulation for damage at large strains with application to fiber-reinforced materials and materials with truss-like microstructures,” International Journal for Numerical Methods in Engineering, vol. 92, no. 6, pp. 551–570, 2012.DOI
  63. [63]R. V. Kohn and S. Müller, “Relaxation and regularization of nonconvex variational problems,” Rendiconti del Seminario Matematico e Fisico di Milano, vol. 62, no. 1, pp. 89–113, 1992.DOI
  64. [64]C. Miehe, “Analysis of material instabilities in inelastic solids by incremental energy minimization and relaxation methods: evolving deformation microstructures in finite plasticity,” Journal of the Mechanics and Physics of Solids, vol. 52, no. 12, pp. 2725–2769, 2004.DOI
  65. [65]M. Köhler and D. Balzani, “Evolving microstructures in relaxed continuum damage mechanics for the modeling of strain softening,” Journal of the Mechanics and Physics of Solids, vol. 173, p. 105199, 2023.DOI
  66. [66]J. Melchior, M. Köhler, and D. Balzani, “Variationally consistent microstructure evolution and microsphere-based reconvexification for damage with application to arterial tissues,” Mechanics of Materials, vol. 212, p. 105495, 2026.DOI
  67. [67]D. Balzani, M. Köhler, T. Neumeier, M. A. Peter, and D. Peterseim, “Multidimensional rank-one convexification of incremental damage models at finite strains,” Computational Mechanics, vol. 73, no. 1, pp. 27–47, 2024.arxiv.org/abs/2211.14318
  68. [68]C. Lauff, M. Schneider, and T. Böhlke, “Numerically robust local continuum damage models with softening response via convex relaxation,” Computer Methods in Applied Mechanics and Engineering, vol. 452, p. 118759, 2026.DOI
  69. [69]J. Kristensen, “On the non-locality of quasiconvexity,” Annales de l’Institut Henri Poincaré C, Analyse non linéaire, vol. 16, no. 1, pp. 1–13, 1999.DOI
  70. [70]M. Köhler, T. Neumeier, M. A. Peter, D. Peterseim, and D. Balzani, “Hierarchical rank-one sequence convexification for the relaxation of variational problems with microstructures,” Computer Methods in Applied Mechanics and Engineering, vol. 432, p. 117321, 2024.arxiv.org/abs/2405.16866
  71. [71]R. T. Rockafellar and R. J.-B. Wets, Variational Analysis, vol. 317 of Grundlehren der mathematischen Wissenschaften. Berlin, Heidelberg: Springer, 1998.DOI
  72. [72]C. Villani, Optimal Transport: Old and New, vol. 338 of Grundlehren der mathematischen Wissenschaften. Berlin, Heidelberg: Springer, 2009.
  73. [73]M. Cuturi, “Sinkhorn Distances: Lightspeed Computation of Optimal Transportation Distances,” in Advances in Neural Information Processing Systems, vol. 26, pp. 2292–2300, 2013.arxiv.org/abs/1306.0895
  74. [74]G. Peyré, “Entropic Approximation of Wasserstein Gradient Flows,” SIAM Journal on Imaging Sciences, vol. 8, no. 4, pp. 2323–2351, 2015.arxiv.org/abs/1502.06216
  75. [75]C. Lauff, A. Schlichting, T. Böhlke, and M. Schneider, “A well-defined local continuum damage model with softening via convexification and entropic regularization,” Continuum Mechanics and Thermodynamics, 2026. in press.DOI
  76. [76]M. Schneider, “A fast solver for entropy-regularized damage transport,” Computational Mechanics, vol. 78, no. 4, pp. 1663–1672, 2026.DOI
  77. [77]M. Schneider, “Damage Transport with Redistribution,” Computer Methods in Applied Mechanics and Engineering, vol. 462, p. 119281, 2026.DOI
  78. [78]J. Görtthofer, M. Schneider, A. Hrymak, and T. Böhlke, “A convex anisotropic damage model based on the compliance tensor,” International Journal of Damage Mechanics, vol. 31, no. 1, pp. 43–86, 2022.DOI
  79. [79]A. Zaitzeff, S. Esedoglu, and K. Garikipati, “High order, semi-implicit, energy stable schemes for gradient flows,” Journal of Computational Physics, vol. 447, p. 110688, 2021.DOI
  80. [80]A. Mielke, “An introduction to the analysis of gradient systems,” Tech. Rep. 3022, Weierstrass Institute for Applied Analysis and Stochastics, 2023. arXiv:2306.05026.arxiv.org/abs/2306.05026
  81. [81]B. Halphen and Q. S. Nguyen, “Sur les matériaux standard généralisés,” Journal de Mécanique, vol. 14, no. 1, pp. 39–63, 1975.
  82. [82]B. J. Dimitrijevic and K. Hackl, “A method for gradient enhancement of continuum damage models,” Technische Mechanik, vol. 28, no. 1, pp. 43–52, 2008.
  83. [83]M. A. Biot, Mechanics of Incremental Deformations. New York: Wiley, 1965.DOI
  84. [84]S. Boyd and L. Vandenberghe, Convex Optimization. Cambridge: Cambridge University Press, 2004.
  85. [85]A. Ben-Tal and M. Teboulle, “Hidden convexity in some nonconvex quadratically constrained quadratic programming,” Mathematical Programming, vol. 72, no. 1, pp. 51–63, 1996.
  86. [86]Y. Xia, “A survey of hidden convex optimization,” Journal of the Operations Research Society of China, vol. 8, no. 1, pp. 1–28, 2020.arxiv.org/abs/1902.10921
  87. [87]E. Whalen, A. Beyene, and C. Mueller, “SimJEB: Simulated Jet Engine Bracket Dataset,” Computer Graphics Forum, vol. 40, no. 5, pp. 9–17, 2021. Data set under the Open Data Commons Attribution License, doi:10.7910/DVN/XFUWJG, https://simjeb.github.io/. Bracket number 270 is the design butterfly-bracket-ii by the designer Lumas, submitted to the GE jet engine bracket challenge hosted by GrabCAD, https://grabcad.com/library/butterfly-bracket-ii-1.DOI
  88. [88]H. Moulinec and P. Suquet, “A fast numerical method for computing the linear and nonlinear mechanical properties of composites,” Comptes Rendus de l’Académie des Sciences. Série II, vol. 318, no. 11, pp. 1417–1423, 1994.
  89. [89]H. Moulinec and P. Suquet, “A numerical method for computing the overall response of nonlinear composites with complex microstructure,” Computer Methods in Applied Mechanics and Engineering, vol. 157, no. 1–2, pp. 69–94, 1998.DOI
  90. [90]M. Schneider, F. Ospald, and M. Kabel, “Computational homogenization of elasticity on a staggered grid,” International Journal for Numerical Methods in Engineering, vol. 105, no. 9, pp. 693–720, 2016.DOI
  91. [91]L. Gélébart and R. Mondon-Cancel, “Non-linear extension of FFT-based methods accelerated by conjugate gradients to evaluate the mechanical behavior of composite materials,” Computational Materials Science, vol. 77, pp. 430–439, 2013.DOI
  92. [92]M. Kabel, T. Böhlke, and M. Schneider, “Efficient fixed point and Newton–Krylov solvers for FFT-based homogenization of elasticity at large deformations,” Computational Mechanics, vol. 54, no. 6, pp. 1497–1514, 2014.DOI
  93. [93]D. Wicht, M. Schneider, and T. Böhlke, “On quasi-Newton methods in fast Fourier transform-based micromechanics,” International Journal for Numerical Methods in Engineering, vol. 121, no. 8, pp. 1665–1694, 2020.DOI
  94. [94]M. Schneider, “A review of nonlinear FFT-based computational homogenization methods,” Acta Mechanica, vol. 232, no. 6, pp. 2051–2100, 2021.DOI
  95. [95]S. R. Williams and A. P. Philipse, “Random packings of spheres and spherocylinders simulated by mechanical contraction,” Physical Review E, vol. 67, no. 5, p. 051301, 2003.pubmed.ncbi.nlm.nih.gov/12786140
  96. [96]C. F. Shih, “Relationships between the J-integral and the crack opening displacement for stationary and extending cracks,” Journal of the Mechanics and Physics of Solids, vol. 29, no. 4, pp. 305–326, 1981.DOI
  97. [97]W. T. Carter, D. J. Erno, D. H. Abbott, C. E. Bruck, G. H. Wilson, J. B. Wolfe, D. M. Finkhousen, A. Tepper, and R. G. Stevens, “The GE Aircraft Engine Bracket Challenge: An Experiment in Crowdsourcing for Mechanical Design Concepts,” in Proceedings of the 25th Annual International Solid Freeform Fabrication Symposium, (Austin, TX, USA), pp. 1402–1411, 2014.
  98. [98]I. A. Baratta, J. P. Dean, J. S. Dokken, M. Habera, J. S. Hale, C. N. Richardson, M. E. Rognes, M. W. Scroggs, N. Sime, and G. N. Wells, “DOLFINx: the next generation FEniCS problem solving environment.” Zenodo preprint, 2023.
  99. [99]C. Geuzaine and J.-F. Remacle, “Gmsh: a 3-D finite element mesh generator with built-in pre- and post-processing facilities,” International Journal for Numerical Methods in Engineering, vol. 79, no. 11, pp. 1309–1331, 2009.DOI
  100. [100]J. Nocedal and S. J. Wright, Numerical Optimization. New York: Springer, 2 ed., 2006.
  101. [101]P. Vaněk, J. Mandel, and M. Brezina, “Algebraic multigrid by smoothed aggregation for second and fourth order elliptic problems,” Computing, vol. 56, no. 3, pp. 179–196, 1996.DOI
  102. [102]L. Lecarme, E. Maire, A. Kumar K.C., C. D. Vleeschouwer, L. Jacques, A. Simar, and T. Pardoen, “Heterogeneous void growth revealed by in situ 3-D X-ray microtomography using automatic cavity tracking,” Acta Materialia, vol. 63, pp. 130–139, 2014.DOI
  103. [103]M. Ruiz de Sotto, V. Doquet, P. Longère, and J. Papasidero, “Anisotropic, rate-dependent ductile fracture of Ti-6Al-4V alloy,” International Journal of Damage Mechanics, vol. 31, no. 3, pp. 374–402, 2022.
  104. [104]W. Weibull, “A statistical distribution function of wide applicability,” Journal of Applied Mechanics, vol. 18, no. 3, pp. 293–297, 1951.DOI
  105. [105]J. M. Ortega and W. C. Rheinboldt, Iterative Solution of Nonlinear Equations in Several Variables. New York: Academic Press, 1970.

Paper details

vlnc:W_7uet87zh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Claude (Anthropic)

View statement
Looking for related papers…
Contents