🔬 materials science

Strength-degradation phase-field regularization of cohesive fracture: the antiplane case

This paper introduces a strength-degradation phase-field model for antiplane fracture that unifies limit analysis, plasticity, and various fracture regimes by decoupling strength, stiffness, and toughness, thereby allowing crack nucleation and propagation to be governed by an arbitrary convex strength surface and an elasto-cohesive length scale rather than being constrained by elastic energy.

Original authors: Blaise Bourdin, Corrado Maurini

Published 2026-08-03

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Blaise Bourdin, Corrado Maurini

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to predict how a piece of glass, a block of clay, or even a metal beam will break. For a long time, scientists have had two very different ways of looking at this problem. On one hand, there's the "brittle" view: materials snap suddenly, like a dry twig, and the only thing that matters is how much energy it takes to create a new crack surface. On the other hand, there's the "ductile" view: materials squish and stretch before breaking, like taffy, and the key is how much the material yields or deforms under pressure.

The tricky part is that real-world materials often do both. A piece of metal might stretch a little (plasticity) before it finally snaps (fracture). For decades, computer models struggled to handle this mix. The popular "phase-field" models, which are like digital simulations that let cracks grow smoothly instead of jumping from one pixel to another, had a major flaw: they tied the material's strength directly to how "fuzzy" the crack looked on the computer screen. If you wanted to simulate a strong material, you had to make the computer's "fuzziness" tiny, which required supercomputers. If you made the fuzziness bigger to save time, the material would magically become weaker. It was a frustrating game of "pick two": you could have strength, you could have speed, or you could have a realistic crack shape, but you couldn't have all three at once.

This paper introduces a new way to model breaking materials that finally untangles this knot. The authors, Blaise Bourdin and Corrado Maurini, propose a model where the material's strength is a separate, independent ingredient, just like its stiffness or its toughness. They tested this idea in a simplified setting called "antiplane shear" (think of sliding a deck of cards sideways) and found that their new model works beautifully. It can simulate everything from a metal stretching and squishing (plasticity) to a clean, sudden snap (brittle fracture) without needing to change the rules of the game. Most importantly, they showed that the "fuzziness" of the computer simulation is just a numerical tool to help the math work, not a physical property that changes how strong the material is. This means scientists can finally simulate complex breaking behaviors on standard computers without the results changing just because they tweaked the grid size.

The New "Strength-First" Approach

The core of this paper is a new mathematical recipe for simulating how things break. The authors call it "strength-degradation phase-field regularization." That's a mouthful, so let's break it down with a simple analogy.

Imagine a material is like a crowd of people holding hands. In old models, when the crowd started to get stressed, the people's arms (the stiffness) would get weaker and weaker until they let go. The problem was that the point at which they let go depended entirely on how tightly you were watching them (the "regularization length"). If you watched closely, they held on longer; if you watched from far away, they let go sooner.

In this new model, the people don't weaken their arms. Instead, they have a maximum grip strength. As long as the pull is below that strength, they hold firm. Once the pull hits that limit, the grip starts to degrade, and they eventually let go. Crucially, this "maximum grip strength" is a fixed property of the people, not a side effect of how you are watching them. This allows the model to handle materials that are strong but not very stretchy, or materials that stretch a lot before breaking, all within the same framework.

The Three Stages of Breaking

The authors ran simulations to see how their model behaves under different conditions. They discovered that their single model can naturally transition through three distinct stages of failure, depending on the size of the object and the material properties:

The "Squishy" Phase (Small-Scale Yielding): When a crack starts to form in a small area, the material around the tip doesn't just snap. It stretches and deforms, creating a "process zone" where the material is yielding. The model captures this perfectly, showing a zone of deformation that matches classical theories of plasticity.
The "Cohesive" Phase: As the crack opens up, the material doesn't break instantly. Instead, it holds on with a "cohesive" force, like a sticky tape that is slowly peeling apart. The force required to pull it apart decreases as the gap gets wider. This is the "cohesive crack" behavior, which is essential for understanding how real materials fail.
The "Brittle" Phase: Finally, if the gap gets wide enough, the material gives up completely and snaps. The crack becomes a clean, sharp break, just like in the simplest models of brittle fracture.

The beauty of this work is that it doesn't need to switch between different mathematical formulas to describe these three stages. It's all one continuous process. The model simply evolves from one state to the next based on the physics, not on a human programmer's choice.

The "Surfing" Experiment and the Plastic Wake

One of the most fascinating parts of the paper is a simulation the authors call "surfing." Imagine a crack that is already there, and you are dragging it through the material at a constant speed, like a surfer riding a wave. The authors ran this simulation in two ways: once where the material's deformation was reversible (like a rubber band) and once where it was irreversible (like clay that stays squished).

In the reversible case, the energy needed to break the material matched exactly what the theory predicted. But in the irreversible case, something interesting happened. As the crack moved, it left behind a "plastic wake"—a trail of permanently deformed material, like the wake of a boat. Because the material had to be squished to let the crack pass, it took more energy to break the material than the theory for a simple brittle crack would suggest. The authors measured this "effective toughness" and found it was about 30% higher than the base toughness of the material. This explains why some materials seem tougher in real life than their basic formulas predict: the plastic deformation ahead of the crack tip is doing extra work.

Bridging the Gap with a V-Notch

To prove their model works for complex shapes, the authors simulated a "V-notch"—a sharp cut in a piece of material, like the corner of a Pac-Man shape. They applied force to this notch and watched what happened.

They found that as the load increased, the material went through the same three stages: first, a small plastic zone formed at the tip (the "squishy" phase); then, a cohesive crack started to grow (the "sticky tape" phase); and finally, a brittle crack suddenly nucleated and raced through the material.

The most impressive result here is that the force required to start the crack followed a "universal law." Whether the notch was very sharp (like a crack) or very blunt (like a straight edge), the model's predictions collapsed onto a single curve. This curve smoothly connects the strength of the material (how hard you have to pull to break it) with its toughness (how much energy it takes to break it). This suggests that the model can predict failure for any shape without needing to guess which formula to use.

Why This Matters

The authors are careful to note that these results come from simulations and mathematical proofs, not physical experiments on real materials. However, the simulations are robust and match known theoretical limits perfectly. They have shown that by changing how the model degrades strength instead of stiffness, they have created a tool that is both flexible and accurate.

Previously, if you wanted to simulate a material that was strong but brittle, you were stuck. You either had to use a model that was computationally impossible (requiring a mesh so fine it would crash your computer) or a model that gave the wrong answer. This new approach allows the "regularization length" (the computer's fuzziness) to be just a numerical tool. You can make it small to get a sharp crack, or larger to save computing power, and the material's strength and toughness will remain exactly the same.

In short, this paper offers a unified language for fracture. It brings together the worlds of plasticity (squishing), cohesive fracture (peeling), and brittle fracture (snapping) into a single, consistent framework. It suggests that we don't need to choose between these different types of failure; they are just different chapters in the same story of how materials break. For engineers and scientists, this means we might soon be able to simulate complex failures in bridges, aircraft, or biological tissues with a level of detail and accuracy that was previously out of reach.

Technical Summary: Strength-degradation phase-field regularization of cohesive fracture: the antiplane case

Problem Statement
Traditional phase-field fracture models, while successful in regularizing brittle fracture, suffer from a fundamental limitation: the material strength surface is not an independent constitutive parameter. Instead, it emerges from the interplay of elastic stiffness, fracture toughness, and the regularization length ( $\ell$ ). This coupling forces the regularization length to act as a material property to fit nucleation loads, rendering the approach impractical for stiff materials or large-scale structures where $\ell$ must be resolved by the mesh. Furthermore, the shape of the resulting strength surface is dictated by the elastic energy form, failing to reproduce arbitrary multiaxial strength data. Conversely, while plasticity theory allows for arbitrary convex strength domains, standard softening plasticity models suffer from localization pathologies and mesh dependence, and sharp-interface cohesive models struggle with numerical implementation and the lack of a unified framework for crack nucleation along unknown paths.

Methodology
The authors analyze a variational phase-field model introduced in a companion work (Bourdin, Marigo, et al., 2025), specifically tailored to the antiplane shear setting. The core innovation of this model is the degradation of strength rather than stiffness.

Energy Functional: The model utilizes an energy functional where the elastic energy density is defined via a minimization over a nonlinear deformation field $p$ . This formulation incorporates a convex strength domain $K_0$ (defined by shear strength $\tau_c$ ) and a damage variable $\alpha$ that degrades the strength domain ( $K(\alpha) = k(\alpha)K_0$ ) while leaving the elastic stiffness ( $\mu$ ) unchanged.
Variational Structure: The model is interpreted as a phase-field regularization of softening plasticity (à la Hencky) without an irreversibility condition on the plastic strain $p$ in the primary formulation. The damage variable $\alpha$ drives the evolution, governed by a gradient-type regularization term.
Numerical Scheme: To solve the non-smooth, separately convex energy minimization problem, the authors propose an alternate minimization scheme. The subproblems are recast as Second-Order Cone Programs (SOCP) and solved using conic optimization (MOSEK), avoiding smoothing or penalization of the strength term.
Analytical Framework: The study derives closed-form solutions for a simple shear problem, establishing the equivalent cohesive law and analyzing the conditions for snap-back in both homogeneous and localized regimes.

Key Contributions and Results

Independent Material Parameters: The analysis demonstrates that in the limit $\ell \to 0$ , the model recovers a cohesive fracture behavior where the strength ( $\tau_c$ ) and toughness ( $G_c$ ) are independent material data, decoupled from the regularization length $\ell$ and the elastic stiffness $\mu$ . The regularization length acts purely as a numerical parameter.
Unified Variational Framework: The model unifies limit analysis, perfect plasticity, cohesive fracture, and brittle fracture into a single framework. The transition between these regimes is governed by the ratio of structural size to the elasto-cohesive length ( $\ell_{ch} = \mu G_c / \tau_c^2$ ) and the regularization ratio $\ell/\ell_{ch}$ .
Closed-Form Solutions: For the antiplane shear problem, the authors derive explicit expressions for:
- The equivalent cohesive law, which is independent of $\ell$ and $\mu$ .
- The conditions for the absence of snap-back, which depend on the brittleness ratio $L/\ell_{ch}$ for structural response and $\ell/\ell_{ch}$ for material response.
- The surface energy density, showing it is a Barenblatt-type cohesive energy with initial slope $\tau_c$ and plateau $G_c$ .
Numerical Verification:
- Simple Shear: Simulations verify the recovery of the analytical cohesive law and the independence of the peak strength from $\ell$ . The model correctly captures the size effect, transitioning from progressive failure (short bars) to brittle snap-back (long bars) as $L/\ell_{ch}$ increases.
- Surfing Simulation: By simulating a crack propagating under a prescribed stress intensity factor, the authors distinguish between reversible and irreversible nonlinear deformation. In the irreversible case, a residual plastic wake forms, increasing the effective toughness ( $G_{eff}^c \approx 1.3 G_c$ ) compared to the prescribed $G_c$ .
- V-Notch: The model bridges small-scale yielding (SSY) plasticity, cohesive fracture, and brittle fracture. It reproduces the transition from a confined plastic zone (Hult-McClintock) to a cohesive crack (Dugdale/Barenblatt) and finally to a brittle crack. The nucleation loads for various notch angles collapse onto a "universal law" interpolating between the strength limit and the Griffith limit.

Significance and Claims
The paper claims that this formulation resolves the long-standing issue of strength regularization in phase-field models. By degrading strength rather than stiffness, the model allows for:

Arbitrary Strength Surfaces: The ability to prescribe arbitrary convex strength domains, a feature previously exclusive to plasticity theory but difficult to implement in fracture.
Mesh Independence: The equivalent cohesive law and material strength are independent of the regularization length, allowing $\ell$ to be chosen solely for numerical resolution.
Unified Phenomenology: The model provides a single variational description that naturally encompasses the spectrum of failure modes—from ductile-like plastic localization to brittle fracture—without requiring a priori hypotheses or switching between different constitutive models.

The authors note that while the antiplane setting simplifies the analysis by ensuring jump compatibility, the extension to 3D vectorial elasticity involves tensorial jump compatibility conditions which are the subject of ongoing work. Additionally, while the model handles ductile-like behavior via plastic irreversibility, a complete theory of ductile fracture would require further development regarding hardening and permanent plastic deformations.