Endowing variational phase-field fracture models with custom strength criteria

This paper proposes a novel variational phase-field fracture framework that incorporates arbitrary elastic domains and custom strength criteria by introducing a state-dependent dissipation potential, thereby preserving the variational structure while allowing independent control over elastic degradation and crack nucleation under multiaxial stress states.

The Gist

Imagine you are trying to predict how a piece of glass or concrete will crack when you squeeze, stretch, or twist it. For a long time, scientists have used a clever mathematical tool called the "phase-field" method to simulate this. Think of this method as a high-tech weather map for cracks: instead of drawing a sharp, jagged line where a crack will appear, it paints a blurry, soft zone that gradually turns from "solid" to "broken."

However, there was a problem with the old maps. They were like a one-size-fits-all suit. They assumed that a material breaks the same way whether you are pulling it apart (tension) or squeezing it (compression). In reality, materials are picky. Concrete, for example, hates being pulled apart but is quite tough when squeezed. The old models couldn't easily tell the difference between these different "personalities" of stress without breaking the mathematical rules that made them work in the first place.

The New Idea: A Customizable Suit

The authors of this paper propose a new way to build these models. They call it "endowing variational phase-field fracture models with custom strength criteria." In plain English, they figured out how to give these crack-predicting models a custom-tailored suit that fits any specific material's rules.

Here is how they did it, using a simple analogy:

The Two-Part System: The Jacket and the Shield

Imagine a material is wearing two layers:

1. The Jacket (Free Energy): This layer represents the material's stiffness. As the material gets damaged (like a jacket getting holes in it), it gets weaker and less stiff. In the old models, the jacket and the rules for when it rips were glued together. If you changed the jacket, you accidentally changed the rules for when it ripped.
2. The Shield (Dissipation Potential): This layer represents the material's strength or its "breaking point." It decides exactly how much force is needed to start a crack.

The Innovation:
The authors realized they could let the Shield change its shape based on how you are pushing or pulling the material, without messing up the Jacket.

• Old Way: If you wanted the material to be stronger in compression than tension, you had to rewrite the whole math of the jacket. It was messy and often broke the mathematical "variational structure" (the internal logic that keeps the simulation stable).
• New Way: They made the Shield "state-dependent." This means the Shield can look like a circle, an oval, or a weird blob depending on the direction of the force.
  • If you pull the material, the Shield might be small (easy to break).
  • If you squeeze it, the Shield might be huge (hard to break).
  • Crucially, the Jacket (stiffness) stays exactly the same. The two are now independent.

The "Elastic Domain" Map

The paper talks a lot about the "elastic domain." Imagine a map of a safe zone. As long as the forces on the material stay inside this zone, the material is safe and won't crack.

• In the old models, this safe zone was always a perfect, symmetrical circle (or half-circle).
• In the new models, the authors can draw this safe zone in any shape they want.
  • They can make it a Double-Ellipse (like a peanut shape) to handle different limits for stretching vs. squeezing.
  • They can make it a Drucker-Prager cone (like an ice cream cone) to model rocks and soil that behave differently under pressure.
  • They can make it a Huber shape that allows the material to be squeezed without cracking (non-interpenetration) but still breaks easily if pulled.

Why This Matters (According to the Paper)

The authors tested their new method with several different "recipes" (models M1 through M5). They simulated a disk of material being pulled and pushed from different angles.

1. Flexibility: They showed that they could create a model where the material breaks easily when pulled but is very strong when squeezed, and vice versa, all while keeping the math clean and stable.
2. Independence: They proved that you can tune the "stiffness" (how much it bends) and the "strength" (when it breaks) separately. Before, changing one often forced you to change the other.
3. Accuracy: The simulations showed that the cracks started exactly where the custom "safe zone" map said they should, matching different complex loading conditions (like twisting and squeezing at the same time).

The Bottom Line

This paper doesn't claim to cure diseases or build new bridges immediately. Instead, it provides a new, more flexible mathematical toolkit. It allows scientists to build computer simulations that respect the specific, quirky rules of different materials (like concrete, rock, or biological tissue) without breaking the fundamental laws of physics that make the simulations reliable. It's like upgrading from a generic, pre-made map to a GPS that can draw custom routes for any terrain you throw at it.

Technical Summary

Technical Summary: Endowing Variational Phase-Field Fracture Models with Custom Strength Criteria

Problem Statement
While phase-field fracture modeling has become a consolidated theory for capturing complex crack patterns, a significant limitation remains in the ability to arbitrarily prescribe strength criteria and consistently describe the evolution of the corresponding elastic domain under multiaxial stress states. Existing approaches generally fall into three categories: (a) splitting the free energy, which retains variational structure but lacks flexibility in defining arbitrary elastic domains; (b) directly modifying the phase-field driving or resisting forces, which offers versatility but sacrifices the variational structure; and (c) coupling with plastic or nonlinear strains, which preserves variational consistency but introduces additional internal variables and formulation complexity. Crucially, in most existing methods, elastic degradation behavior and the strength criterion are strongly coupled, preventing independent tuning of stiffness degradation and material strength.

Methodology
The authors propose a fundamentally different strategy formulated within the energetic framework of Generalized Standard Materials (GSM). The core innovation is the introduction of a state-dependent dissipation potential. Unlike standard models where the dissipation potential is constant or depends only on the phase-field variable, this approach allows the dissipation potential to depend on the current state of the material (specifically, the strain or stress invariants).

Key methodological components include:

• Separation of Mechanisms: The free energy density ($\psi$) is responsible solely for encoding elastic degradation (stiffness loss), while the state-dependent dissipation potential ($\phi$) encodes the strength criterion and the shape of the elastic domain. This decoupling allows these two distinct material properties to be controlled independently.
• Fracture Energy Dependence: The fracture energy $G_c$ is treated as a function of the loading direction (represented by an angle $\vartheta$ in the volumetric-deviatoric strain or stress plane). This dependence is incorporated via a fracture function $G_f(\varepsilon, \alpha)$ within the dissipation potential.
• Variational Consistency: By deriving the evolution equations from a global energy balance and stability condition (Energetic Formulation), the proposed model retains the variational structure of the problem, preserving associated theoretical and computational advantages.
• Numerical Implementation: A staggered solution scheme is employed, solving the equilibrium and damage problems alternately. The state-dependent fracture function is frozen at the previous converged load step during iterations to approximate the state-dependent dissipated energy.

Key Contributions and Material Models
The paper demonstrates the flexibility of this framework by constructing and analyzing five specific material models (M1–M5) that combine different degradation laws and strength criteria:

1. M1 (Standard AT1): Full elastic degradation (FED) with a standard isotropic strength criterion.
2. M2 (Double-Ellipse): FED with a tunable double-ellipse strength criterion, allowing arbitrary volumetric and deviatoric limits.
3. M3 (Drucker–Prager): FED with a Drucker–Prager criterion, suitable for pressure-sensitive materials.
4. M4 (Drucker–Prager with Non-Interpenetration): Partial elastic degradation (PED) where the compressive bulk modulus is not degraded, yet the Drucker–Prager strength criterion is recovered.
5. M5 (Huber with Non-Interpenetration): PED combined with a Huber-type criterion, allowing independent tuning of tensile-volumetric and deviatoric limits while ensuring non-interpenetration under compression.

Results
The paper presents analytical and numerical results for a biaxial disk problem under various loading directions ($\vartheta$):

• Homogeneous Responses: Analytical solutions confirm that the models successfully reproduce distinct elastic domain evolutions. For instance, M3 and M4 share the same elastic domain shape but exhibit different mechanical responses due to the presence or absence of bulk modulus degradation. M5 demonstrates non-symmetric degradation behavior, distinguishing between tensile and compressive volumetric strains.
• Crack Nucleation: Numerical simulations of crack nucleation under multiaxial loading show that the models accurately predict nucleation points corresponding to the prescribed strength surfaces. The results illustrate that the proposed framework can capture mixed-mode nucleation behaviors (e.g., isotropic expansion, uniaxial tension, pure shear, uniaxial compression) consistent with the specific strength criteria defined for each model.
• Independence of Features: The results validate that elastic degradation (governed by the free energy) and the strength criterion (governed by the dissipation potential) can be tuned independently, a feature not easily achievable in previous approaches.

Significance and Claims
The authors claim that this work paves the way for future developments by providing a flexible, variationally consistent framework for incorporating arbitrary strength criteria into phase-field fracture models. The primary significance lies in the ability to:

1. Preserve the variational structure of the problem while introducing complex, state-dependent strength criteria.
2. Decouple stiffness degradation from strength, allowing for more realistic and independently controllable material modeling.
3. Systematically construct models for complex materials (e.g., quasi-brittle, frictional, or pressure-sensitive materials) without resorting to non-variational source terms or excessive internal variables.

The paper concludes modestly, noting that while the framework handles nucleation and stable evolution well, capturing unstable crack propagation events will require future integration of time-regularizing strategies, such as inertial or viscous terms. The current work establishes the theoretical foundation for these extensions.