A unified variational framework for phase-field fracture and third-medium contact in finite deformation hyperelasticity

This paper introduces a unified variational framework that integrates phase-field fracture and third-medium contact within finite deformation hyperelasticity by regularizing both crack topology and contact interfaces, thereby eliminating the need for explicit tracking algorithms while successfully simulating complex coupled phenomena like secondary crushing in Brazilian disk tests.

The Gist

Imagine you are trying to simulate how a piece of brittle material (like a ceramic tile or a rock) breaks when you press down on it with a heavy object.

In the real world, two things happen simultaneously:

1. Contact: The heavy object touches the tile.
2. Fracture: The tile cracks, and those new cracks might touch each other or the heavy object.

For decades, computer scientists have struggled to simulate this because the math for "touching" and the math for "breaking" are like two different languages that don't speak well to each other. Usually, you have to tell the computer exactly where the crack is and where the touch happens. If the crack moves or the touch area changes shape, the computer gets confused and crashes.

This paper introduces a clever new way to solve this problem by treating both touching and breaking as the same kind of "fuzzy" problem.

The Core Idea: The "Fuzzy" Approach

Think of the old way as trying to draw a perfect, razor-sharp line for a crack and a perfect, hard edge for contact. It's rigid and hard to manage when things move.

The authors propose a "fuzzy" approach using two main tricks:

1. The "Softening" Trick for Cracks (Phase-Field Fracture)

Instead of a sharp, jagged crack, imagine the crack is a foggy zone.

• Intact material is clear air.
• Broken material is thick fog.
• The crack is the transition zone where it's getting misty.

As the material breaks, the "fog" spreads out. The computer doesn't need to track the exact edge of the crack; it just calculates how "foggy" the area is. This makes it incredibly easy to simulate cracks branching, merging, or changing direction without the computer getting lost.

2. The "Ghost Jelly" Trick for Contact (Third-Medium Contact)

Usually, when two objects touch, the computer has to constantly check: "Are they touching? Are they overlapping?" This is computationally expensive.

The authors introduce a fictitious "Ghost Jelly" (the Third Medium) that fills the tiny gap between the objects.

• Imagine the gap between a hammer and a nail is filled with a super-soft, squishy jelly.
• When you push the hammer down, the jelly squishes.
• As the jelly gets thinner and thinner, it pushes back harder and harder, mimicking the force of contact.
• The computer doesn't need to check if the hammer "touched" the nail; it just calculates how much the jelly is squished.

The Magic Synergy:
The genius of this paper is that they realized both the crack (the fog) and the contact (the squishy jelly) are just "fuzzy fields." They combined them into one single mathematical recipe.

• When the hammer presses down, the "jelly" squishes, creating stress.
• That stress turns the "fog" (the crack) thicker.
• If the crack opens up and the two sides touch, the "jelly" automatically fills that new gap and pushes back, just like it did with the hammer. No extra instructions needed!

Why This Matters: The "Brazilian Nut" Example

The authors tested this on a classic experiment called the Brazilian Disk Test (imagine crushing a round cookie between two flat plates).

• Old Models: They usually pretend the plates are perfectly flat and the force is applied at a single, fixed point. They miss what happens right under the plates.
• This New Model: Because the "Ghost Jelly" naturally fills the gap as the cookie deforms, the simulation shows that the cookie doesn't just split down the middle. It also gets crushed and crumbled right under the plates where the pressure is highest.

This matches what happens in real life (and in real experiments) but was very hard for computers to predict before. It's like the simulation finally learned to see the "crumbs" forming at the edges, not just the big crack in the middle.

The Bottom Line

This paper builds a unified "operating system" for simulating how things break and touch. By replacing sharp, hard edges with smooth, fuzzy fields (fog for cracks, jelly for contact), they eliminated the need for complex, error-prone tracking algorithms.

In simple terms: They stopped trying to draw perfect lines and started simulating how materials "smudge" and "squish." This allows engineers to predict how things will fail in complex scenarios—like car crashes, battery electrodes cracking, or biological tissues tearing—with much higher accuracy and less headache.

Technical Summary

1. Problem Statement

The paper addresses the computational challenge of simulating coupled contact and fracture phenomena within a finite deformation hyperelastic setting. Traditional computational mechanics approaches face significant hurdles when these two phenomena interact:

• Contact Complexity: Classical contact algorithms require explicit detection of contact pairs, master-slave surface designation, and enforcement of non-penetration constraints (via penalty, Lagrange multipliers, or augmented Lagrangian methods). These become computationally expensive and algorithmically complex when surfaces evolve dynamically (e.g., crack faces closing or debris interaction).
• Fracture Complexity: Phase-field fracture (PFF) eliminates the need for explicit crack tracking by regularizing sharp cracks into diffuse damage fields. However, standard PFF formulations do not inherently handle the contact of newly formed crack faces or the interaction between a rigid indenter and a fracturing substrate without additional algorithmic layers.
• The Gap: There is a lack of a unified variational framework that naturally couples the regularization of cracks (PFF) with the regularization of contact interfaces (Third-Medium Contact, TMC) to simulate contact-induced fracture without explicit interface tracking.

2. Methodology

The authors propose a unified variational framework based on the principle that both crack topology and contact interfaces can be regularized using continuous field descriptions. The methodology integrates three distinct components into a single total potential energy functional ($\Pi$):

A. Domain Decomposition

The computational domain ($\Omega_0$) is split into three non-overlapping subdomains:

1. Substrate ($\Omega_1$): The deformable body where fracture occurs. Modeled using Phase-Field Fracture (PFF) with a damage variable $d \in [0, 1]$.
2. Indenter ($\Omega_2$): A rigid or stiff elastic body that does not fracture.
3. Third-Medium ($\Omega_3$): A compliant fictitious material filling the gap between $\Omega_1$ and $\Omega_2$. It acts as the contact interface.

B. Unified Energy Functional

The total potential energy is minimized with respect to displacement ($u$), damage ($d$), and auxiliary fields ($p, q$):
$$\Pi(u, d, p, q) = \Pi_{int} + \Pi_{frac} - \Pi_{ext}$$

• Substrate Energy ($\Omega_1$): Uses a spectral strain energy decomposition to separate tensile ($\Psi^+$) and compressive ($\Psi^-$) energies. Only the tensile part is degraded by the damage function $g(d)$, preventing crack healing under compression.
• Fracture Energy ($\Omega_1$): Uses the AT2 crack density functional (smooth energy landscape) to regularize the sharp crack surface into a diffuse band of width $2\ell$.
• Third-Medium Energy ($\Omega_3$): Modeled as a Neo-Hookean hyperelastic material with extremely low stiffness (scaling parameter $\gamma \ll 1$). As the bodies approach, the medium compresses ($J \to 0$), naturally transmitting contact forces without explicit detection.
• Auxiliary-Field Regularization: To prevent mesh distortion and element inversion during extreme compression of the third medium, auxiliary fields ($p$ for local rotation, $q$ for volumetric deformation) are introduced. These fields regularize the kinematic quantities, ensuring numerical stability.

C. Numerical Implementation

• Discretization: Continuous Lagrange finite elements are used for all fields.
• Solution Strategy: A staggered (alternate minimization) scheme is employed due to the non-convexity of the total energy.
  1. Mechanical/Contact Step: Solve for displacement ($u$) and auxiliary fields ($p, q$) monolithically with damage ($d$) fixed.
  2. Damage Step: Solve for damage ($d$) with kinematics fixed, subject to irreversibility constraints ($\dot{d} \geq 0$).
• Software: Implemented using FEniCS with the multiphenics library for block-structured variational forms.

3. Key Contributions

1. Unified Variational Formulation: The paper establishes the first framework that seamlessly couples PFF and TMC within a single energy functional, eliminating the need for explicit contact detection algorithms or crack tracking.
2. Auxiliary-Field Regularization for Contact: It extends the TMC method with auxiliary fields to handle severe mesh distortion in the third medium, enabling robust simulation of finite deformation contact where $J \to 0$.
3. Natural Handling of Evolving Topology: The framework automatically resolves the contact of newly formed crack faces and the evolving contact area between an indenter and a substrate, as the third medium adapts to the changing geometry without algorithmic intervention.
4. Spectral Decomposition in Finite Deformation: It successfully integrates spectral strain energy decomposition (to prevent crack healing) with finite deformation hyperelasticity in the contact region.

4. Results and Validation

The framework was validated through four numerical examples:

• 2D C-Box Benchmark (TMC Validation): Demonstrated that the third medium successfully transmits contact forces and maintains mesh quality under severe compression. The auxiliary fields were shown to be essential; without them, the medium either acted as a rigid barrier or deformed excessively.
• 2D Single-Edge Notched (SEN) Benchmark (PFF Validation): Successfully reproduced Mode I (straight crack) and Mode II (kinked/curved crack) fracture paths, validating the spectral decomposition and damage evolution.
• 2D Three-Point Bending Test: Simulated a rigid indenter pressing on a pre-cracked beam. The model captured the transition from point contact to a finite contact area as the indenter conformed to the substrate, driving the pre-crack to open and propagate upward.
• 3D Brazilian Disk Test:
  • Reproduced the characteristic vertical splitting crack along the loading diameter.
  • Key Finding: The simulation naturally captured secondary crushing-type fracture zones near the contact regions (loading platens). This phenomenon, caused by stress concentration from the progressive growth of the contact area, is consistent with experimental observations but is impossible to reproduce with simplified models that prescribe fixed contact geometries (e.g., line loads).

5. Significance and Future Outlook

• Predictive Capability: The framework allows for the predictive simulation of complex contact-fracture interactions (e.g., indentation-induced failure, fragmentation) without prior assumptions about contact geometry or crack paths.
• Experimental Fidelity: By capturing secondary crushing zones in the Brazilian disk test, the model bridges the gap between simplified theoretical loading models and complex experimental reality.
• Limitations:
  • Exact zero-gap contact is not achieved; a small residual gap exists due to the finite thickness of the compressed third medium (controlled by $\gamma$).
  • Computational cost is high due to the additional degrees of freedom (third medium + auxiliary fields) and the staggered solution scheme.
• Future Work: The authors plan to extend the framework to include friction, adhesion, inertia (impact), anisotropic materials, and thermal/chemical coupling. A specific application target is nuclear fuel rod analysis (pellet-cladding mechanical interaction).

In conclusion, this work provides a robust, thermodynamically consistent, and algorithmically elegant solution for simulating coupled contact and fracture problems in large deformations, offering a significant advancement over traditional methods that rely on explicit interface tracking.