← Latest papers
🔬 materials science

Nonlocal regularization of sharp-crack energetics: a variational bridge from Griffith geometry to emergent constitutive degradation

This paper presents a variational nonlocal regularization framework that derives emergent constitutive degradation directly from tensile strain-energy averaging without independent fracture fields, successfully recovering Griffith surface energy and enabling structure-preserving dynamic simulations of crack branching and impact.

Original author: Qing Cheng, Yuqi Sun, Xuejun Xu

Published 2026-11-18
📖 7 min read🧠 Deep dive

Original author: Qing Cheng, Yuqi Sun, Xuejun Xu

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

Cracks in solid materials are not merely lines of separation; they are the result of a fundamental trade-off between the energy stored in a deformed object and the energy required to create a new surface. For nearly a century, scientists have understood that when a material breaks, it releases the elastic energy built up by stretching or squeezing it, paying for this release with the cost of forming a fresh crack face. This balance, known as Griffith's theory, explains why a sharp crack propagates, but it struggles to describe the messy, complex reality of how a crack actually moves through a solid, especially when that solid is being hit, shaken, or twisted. Traditional computer models often treat the crack as a sharp, mathematical line that must be tracked explicitly, a method that becomes computationally difficult when the crack branches or changes direction suddenly. Other approaches smooth the crack out into a fuzzy zone, but these often lose the precise connection to the original physics of sharp cracks or require artificial rules to decide when and how the material weakens.

A team of researchers at Tongji University has developed a new way to simulate these fractures that bridges the gap between the sharp, idealized theory and the smooth, computer-friendly models. Their work focuses on a specific behavior of materials: they usually break when pulled apart, but they often remain stiff and strong when squeezed together. By separating the energy of stretching from the energy of squeezing, and by averaging the stretching energy over a small neighborhood rather than looking at a single point, the researchers created a model that naturally predicts how a material degrades without needing to add extra, artificial rules. They tested this new approach against two classic, high-speed fracture scenarios: a crack that splits into two branches under tension, and a crack that kinks sharply under a sudden impact. The results show that their method captures the complex paths of these cracks with high accuracy, while also preserving the correct physical laws of energy conservation, offering a more reliable tool for understanding how solids fail under extreme conditions.

The core of this new approach lies in how it handles the energy within a material. In the real world, if you pull a piece of glass, it stores energy until it snaps; if you push on it, it usually just gets stiffer. Many existing computer models treat all energy the same way, or they use a separate "damage" variable that acts like a switch to turn off the material's strength. The researchers instead built a system where the material's strength fades naturally based on the energy it has stored. They focused on the energy from stretching, which is the dangerous kind that leads to breaking, and applied a special mathematical filter to it. This filter looks at the stretching energy not just at one point, but as an average over a small surrounding area. This averaging smooths out the sharp edges that cause numerical problems in computers, but it does so in a way that still respects the underlying physics of a sharp crack. Crucially, they left the energy from squeezing completely untouched, ensuring the material stays strong under compression, just as real solids do.

This method produces a "constitutive degradation," which is a fancy way of saying the material's ability to carry a load decreases automatically as it stretches. In older models, scientists often had to invent a separate rule to say, "When the crack gets this wide, the material is half as strong." Here, the weakening emerges directly from the math of the energy itself. As the material stretches, the averaged energy rises, and a built-in function limits how much energy can be stored, effectively turning the material's stiffness down. The researchers proved that if you look at a smooth, unbroken piece of material, this new model behaves exactly like standard elasticity. But if you introduce a crack, the model naturally recovers the correct amount of energy needed to create a new surface, matching the classic theory of Griffith. The shape of the mathematical function they used to limit the energy determines exactly how strong the material is before it starts to weaken, allowing the model to be tuned to specific materials.

To test if this theory works in the chaotic world of real-time breaking, the team ran two difficult simulations. The first involved a plate of material with a small cut in it, which was then pulled apart until the crack raced through the material and split into two branches. This is a notoriously hard problem for computers to solve because the crack path is unpredictable. The researchers compared their new method against a standard, well-known time-stepping algorithm. Both methods predicted that the crack would travel straight for a while and then split into two symmetric branches, with the angles of those branches matching closely. The new method, which uses a specific technique to keep energy calculations precise, showed that the crack tip speeds and the timing of the split were nearly identical to the standard method, but with a mathematical guarantee that the total energy of the system was being conserved correctly at every step.

The second test was even more demanding, involving a block of material with two notches that was struck by a high-speed impact. In this scenario, the material is subjected to a mix of pulling and shearing forces, causing the cracks to kink and turn at sharp angles. The researchers found that their model, which separates stretching from squeezing, correctly predicted that the cracks would turn outward, away from the center of the impact. When they tried the same simulation without separating the stretching and squeezing energies, the model failed to capture this behavior; instead, it predicted the cracks would turn inward and the material would lose its ability to resist compression, which is physically incorrect for this type of brittle fracture. The split model preserved the material's stiffness under compression, leading to a realistic outward kinking pattern that matched experimental observations. The timing of the crack initiation and the angle of the kink were consistent across different levels of detail in the simulation, showing that the method is robust.

The researchers also paid close attention to the mathematics of how the simulation moves forward in time. They developed a new way to calculate the next step in the motion that guarantees the energy balance remains exact, even on a computer where numbers are rounded. This "structure-preserving" approach ensures that the energy lost to creating a new crack surface is perfectly accounted for by the work done by the external forces, without any artificial drift or error building up over time. In their tests, the difference between the calculated energy and the actual work done was vanishingly small, confirming that the simulation is not just visually correct but physically faithful. This level of precision is vital for dynamic problems where small errors can quickly lead to completely wrong predictions about how a structure will fail.

Ultimately, this work offers a cleaner, more physically grounded way to simulate how things break. By letting the material's weakening emerge naturally from the energy it stores, rather than imposing it through artificial rules, the model captures the complex dance of crack propagation with fewer assumptions. The separation of stretching and squeezing energies proves essential for mixed-mode failures, where the material is both pulled and pushed, ensuring that the simulation does not falsely soften the material under compression. While the model has been tested on specific scenarios like branching and impact, the underlying principles suggest it could be extended to more complex materials and shapes in the future. For now, it stands as a significant step toward a unified understanding of fracture, where the sharp, sudden nature of a break is reconciled with the smooth, continuous mathematics needed to predict it.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →