🔬 materials science

Identification of the length scale parameter of simplified strain gradient elasticity from standard Mode I fracture tests

This paper presents a method to identify the single length scale parameter of simplified strain gradient elasticity from standard Mode I fracture tests, deriving explicit formulas for brittle materials and regression relations for quasi-brittle materials to capture non-classical size effects using experimental data.

Original authors: Yury Solyaev, Kirill Shelkov, Pavel Polyakov

Published 2026-08-24

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yury Solyaev, Kirill Shelkov, Pavel Polyakov

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

Materials like glass, ceramics, and certain composites are often brittle, meaning they can shatter suddenly when a crack forms. For engineers designing bridges, aircraft, or medical implants, predicting exactly when and how these materials will fail is a matter of safety. For decades, the standard tool for this prediction has been a theory called linear elastic fracture mechanics. This approach works remarkably well when a crack is long, treating the crack tip as a mathematical point where stress becomes infinitely high. However, this theory breaks down when cracks are very small or when the material itself has a complex internal structure, like a porous ceramic or a composite filled with fibers. In these cases, the material does not behave as if the crack is a sharp, singular point; instead, the material's internal "grain" or microstructure smooths out the stress, preventing the infinite spike that classical theory predicts. To fix this gap, scientists have developed a more advanced framework called strain gradient elasticity, which adds a single, crucial piece of information to the equations: a length scale. This length scale represents the size of the material's internal features, acting as a ruler that tells the math how small the cracks are relative to the material's own structure. Until now, however, figuring out exactly what this length scale is for a specific material has been a difficult, open-ended problem.

In a recent study, researchers set out to solve this puzzle by creating a practical method to identify this hidden length scale using standard laboratory tests. They focused on three common ways engineers test how materials break: pulling a plate with a crack in the middle, pulling a plate with a crack on the edge, and bending a beam with a crack on the edge. The team performed highly detailed computer simulations of these tests, using a sophisticated numerical technique that could capture the smooth, non-infinite stress fields predicted by the advanced theory. By comparing their simulation results with the known behavior of materials, they discovered a direct, simple relationship for brittle materials. They found that for long cracks, the mysterious length scale is essentially a fixed multiple of a value already known to engineers, which is calculated from the material's toughness and its ultimate strength. This means that for many brittle materials, scientists can now estimate this new parameter without needing complex new experiments, simply by plugging in existing data.

The researchers went further to address materials that are not perfectly brittle, such as porous ceramics or fiber-reinforced composites, where the transition from small cracks to large cracks is more gradual. For these "quasi-brittle" materials, the standard rules do not apply as neatly. To handle this, the team developed a set of regression formulas—essentially mathematical curves derived from their massive library of computer simulations. These curves act as a bridge, allowing engineers to take experimental data from standard fracture tests and work backward to find the correct length scale parameter for that specific material. The team tested these formulas against real-world data from chopped fiber composites and various types of porous and dense ceramics. In every case, the method worked. For the fiber composites, the identified length scale turned out to be very close to the physical width of the chopped fibers used in the material. For the dense ceramics, the calculated length scale matched the size of the material's grains, confirming that the parameter truly reflects the material's internal microstructure.

The study does not claim to have replaced the old theories, but rather to have extended them. The researchers explicitly showed that for long cracks, their new method converges perfectly with the classical approach, ensuring that the new tool does not contradict established engineering wisdom. Instead, it fills the void for short cracks and complex microstructures where the old theory fails. By providing a clear, step-by-step way to identify this length scale from standard test data, the work offers a practical path forward. Engineers can now use these refined models to predict the strength of components with cracks of any size, from microscopic flaws to large fractures, using a single, consistent set of material properties. This brings a unified understanding to how materials fail, bridging the gap between the smooth, continuous world of classical mechanics and the granular, complex reality of the materials we build with.

Technical Summary: Identification of the Length Scale Parameter of Simplified Strain Gradient Elasticity from Standard Mode I Fracture Tests

Problem Statement
Strain Gradient Elasticity (SGE) theories offer a framework to regularize stress singularities at crack tips and capture non-classical size effects in brittle and quasi-brittle materials. However, the reliable identification of the additional material constants required by SGE, particularly the length scale parameter ( $l$ ), remains a significant challenge. While previous studies have suggested relations between $l$ and microstructural features or used indirect methods (e.g., indentation, pull-out), a robust methodology for identifying $l$ using standard Mode I fracture mechanics test data (CCT, SENT, SENB) has been lacking. Specifically, there is a need to bridge the gap between classical Linear Elastic Fracture Mechanics (LEFM) predictions for long cracks and the non-classical size effects observed in short cracks within a unified SGE framework.

Methodology
The authors employ a simplified SGE theory (Aifantis form) characterized by a single length scale parameter $l$ alongside classical elastic constants. The study proceeds through the following steps:

Numerical Implementation: The authors utilize an enriched $C^1$ -continuous Finite Element Method (FEM). This formulation incorporates additional shape functions based on the asymptotic solutions of SGE crack problems. The method allows for the calculation of regular (non-singular) stress and strain fields at the crack tip, as well as the amplitude factors ( $K_1, K_2$ ) and the $J$ -integral.
Convergence and Validation: A rigorous convergence analysis is performed regarding integration points (requiring 42 points for enriched elements to converge amplitude factors) and mesh refinement near the crack tip. The numerical solutions for the $J$ -integral are validated against semi-analytical solutions and shown to converge to classical LEFM results for long cracks ( $a/l \gg 1$ ).
Two-Pronged Identification Approach:
- Long Cracks (Brittle Materials): For cracks significantly larger than the length scale parameter, the authors derive an explicit relation between the SGE length scale $l$ and the LEFM characteristic length $L_{LEFM} = K_{Ic}^2 / (\pi \sigma_{ult}^2)$ . By comparing SGE predictions with the classical LEFM solution for an infinite plate, they refine the proportionality coefficient.
- Short Cracks (Quasi-Brittle Materials): For finite specimens and short cracks where size effects are pronounced, the authors develop regression relations. These relations approximate the nominal strength ( $\sigma_n$ ) as a function of normalized crack length ( $a/l$ ) and specimen geometry ( $a/b$ ). The regression functions are constructed to satisfy two asymptotic limits: the LEFM power law for long cracks and the "strength plateau" (approaching ultimate strength $\sigma_{ult}$ ) for vanishingly small cracks.

Key Contributions

Refined Proportionality Relation: The paper establishes a precise closed-form expression for identifying the length scale parameter in brittle materials: $l \approx 0.362 (K_{Ic}/\sigma_{ult})^2$ . This refines previous estimates (e.g., $l = L_{LEFM}/8$ ) by utilizing rigorous numerical simulations that strictly satisfy SGE boundary conditions.
Generalized Regression Relations: The authors propose specific regression formulas for three standard test configurations (CCT, SENT, SENB). These relations incorporate classical geometric correction factors and an additional SGE-specific correction factor to capture the transition from strength-dominated to toughness-dominated failure regimes.
Validation via Experimental Data: The proposed identification methods are applied to experimental data from literature and new experiments conducted by the authors. This includes:
- Random chopped fiber composites (SENT tests).
- Highly porous SiO $_2$ ceramics (SENB tests).
- Dense polycrystalline ceramics (Al $_2$ O $_3$ , SiC, Si $_3$ N $_4$ , Sialon) from existing literature.

Results

Brittle Materials: For materials like glass, steel, and ceramics, the identified length scale parameters ( $l$ ) generally range from tens to hundreds of microns. The study finds that $l$ is often of the same order of magnitude as the grain size or Representative Volume Element (RVE) size, though not always identical (e.g., in nanocomposites).
Quasi-Brittle Materials: The regression relations successfully describe the non-classical size effect observed in chopped fiber composites and porous ceramics. For the chopped fiber composites, the identified $l$ was found to be close to the minimum dimension of the chopped tape. For porous ceramics, $l$ values were relatively large (0.76 mm to 1.5 mm), correlating with pore size and geometry.
Dense Ceramics: In dense ceramics, the identified $l$ values (5.8 to 39 $\mu$ m) correlate well with grain sizes, with a ratio $l/d$ ranging from approximately 1.95 to 3.00. This supports the interpretation of $l$ as a microstructure-related characteristic length governing the interaction between small flaws and surrounding grains.
Accuracy: The regression relations demonstrate high accuracy when compared to numerical SGE solutions, with Mean Absolute Percentage Errors (MAPE) generally below 2.6% for the fitted curves. When applied to experimental data, the models capture the transition from LEFM behavior to the strength plateau effectively.

Significance and Claims
The paper claims to provide a practical and robust methodology for identifying the SGE length scale parameter using standard fracture test data, thereby addressing a long-standing open problem in the field. The significance lies in:

Unification: The ability to describe the strength of bodies with cracks of arbitrary length (from short to long) within a single linear elastic formulation using regular stresses.
Simplicity: The proposed identification methods do not require complex inverse numerical modeling for every new material; instead, they rely on explicit formulas or simple regression fits to experimental data.
Predictive Capability: Once $l$ is identified, the simplified SGE framework can be applied to analyze structural elements with complex geometries and stress concentrators (notches, holes) where classical LEFM fails to predict strength accurately due to size effects.

The authors modestly note that while the simplified SGE with a single parameter works well for Mode I loading, more general constitutive relations or failure criteria may be required for mixed-mode loading or complex stress states. Furthermore, the dependence of SGE amplitude factors on crack length (linear for long cracks, exponential decay for short cracks) is highlighted as an interesting result requiring further theoretical justification.

More like this