Localized Fracture Modeling of Steel-Concrete Interfaces: A Novel Approach to Size-Independent Shear Strength and Energy

Reinforced concrete plays an indispensable role in shaping modern infrastructure, largely due to its combined strength, long-term durability, and relatively low construction cost. But behind its apparent simplicity lies a more intricate internal interaction—specifically, the bond between the embedded steel reinforcement bars and the surrounding concrete matrix. This interface is responsible for transferring load between the two materials and ensuring their deformation remains compatible under stress. The safety, stiffness, and resilience of entire structures often hinge on the integrity of this bond. And yet, even after decades of research and widespread application, the finer mechanics of this interaction—particularly under localized failure conditions—are still not fully understood.

A major source of this uncertainty stems from how interfacial shear strength (τ_f) and shear fracture energy (G_IIf) are defined and measured. These two parameters, which govern how cracks initiate and grow during pull-out or splitting failures, are usually estimated using models that assume the interface is smooth and uniform. That assumption, though mathematically convenient, misses the mark. The real interface is anything but regular—it’s rough, filled with aggregates of varying size, and disrupted by the ribs of the steel bar. This inconsistency gives rise to size effects: predicted values of τ_f and G_IIf fluctuate depending on bond length, specimen dimensions, and bar geometry. In engineering terms, this means design values lack consistency and may not reflect the actual performance of a structure under load. The issue becomes more pronounced as materials evolve. High-strength concrete mixes, recycled aggregates, and new rebar designs introduce additional variability that traditional models are ill-equipped to handle. More critically, widely used pull-out tests and empirical formulations do little to illuminate the fracture process at the microscale. They offer bulk averages, but fail to capture where and how energy is dissipated as the bond begins to fail. That’s not just a theoretical problem—it affects how we design for failure and durability in real-world applications.

To break from these limitations, new research paper published in Journal Structures and led by Professor Shutong Yang, Ruiyang Pang, Zhongke Sun, Enhui Fang, Zhenhua Ren from the Department of Civil Engineering, College of Engineering at Ocean University of China developed a new analytical fracture model. Their framework moves away from assumptions of continuity and instead incorporates interfacial heterogeneity directly into the equations. By using meso-scale descriptors—such as an effective aggregate size (d_eff) and discrete stepwise coefficients (β and C)—the model replicates how shear cracks propagate around ribs and through coarse concrete. The novel model links the maximum pull-out force directly to τ_f and G_IIf in a way that neutralizes size dependency and reflects the actual, physical processes occurring at the interface.

In order to capture the complexities of interfacial fracture between steel bars and concrete, the research team carried out a well-structured experimental program involving 85 pull-out specimens. These included three types of HRB400 deformed rebars—R12, R08, and DR08—each differing in rib spacing and surface configuration. The bars were embedded in concrete blocks of two distinct sizes: 150 mm and 100 mm. After accounting for a few early failures, 82 specimens were retained for analysis. The aim wasn’t just to replicate standard tests, but rather to probe how bond length, bar geometry, and cover thickness influence the mechanisms of bond failure under controlled conditions. The authors used PVC sleeves to define the bonded length, ensuring that unbonded zones remained consistent across specimens. To track slip behavior with high fidelity, displacement sensors were mounted at both the loaded and free ends of the rebar. Tests were conducted at a constant loading rate, and force-slip data were collected in real time with sufficient resolution to identify transitions in mechanical behavior. Two failure patterns emerged across the dataset. In most of the larger concrete blocks, failure occurred by bar pull-out—this was the more gradual mode and allowed for observation of post-peak softening and residual frictional resistance. In contrast, many of the smaller blocks—particularly those with longer bonded lengths—exhibited concrete splitting. These specimens failed abruptly, with a sudden loss in capacity and no residual phase in the load-slip curve. The contrast between the two modes was not only structural but had a measurable effect on derived parameters like shear strength and fracture energy.

In pull-out failures, the bond-slip response matched the proposed model remarkably well. The τ_f values derived from the analytical framework were consistently higher than the maximum average bond stresses obtained using traditional methods. This discrepancy grew with bond length, underscoring how conventional averaging obscures true local strength. Notably, τ_f stayed relatively stable across different lengths for the same bar type—evidence that the model successfully neutralized the size effect. G_IIf, however, proved more sensitive. It declined when the crack-tip approached specimen boundaries, revealing a boundary effect. But once the crack-tip was situated at least 16 times the effective meso-scale parameter (d_eff) from the edge, G_IIf leveled off. In that range, it could be interpreted as an intrinsic material property. Interestingly, bar type also played a role—τ_f was modestly affected, but G_IIf varied more noticeably, especially with changes in rib geometry and axial stiffness.

The significance of the research work of Professor Shutong Yang and colleagues  successfully build a new analytical model that accounts for heterogeneity and discontinuity at the interface, the authors have provided a transformative shift in fracture characterization and enabled the prediction of τ_f and G_IIf as intrinsic, size-independent properties. This marks a decisive departure from conventional methods, which often mask the underlying physics with averaged metrics that vary with specimen dimensions or geometry. We believe one of the clearest implications is in structural design and with this new model, practitioners gain access to more precise, context-specific data on bond performance—data that reflects the true interfacial response, not a diluted average. This has direct relevance to design codes and safety factors, particularly in performance-based design frameworks where accurate fracture energy estimation becomes crucial. We believe another key outcome is the resolution of boundary effect artifacts in shear fracture energy assessment. By identifying the distance threshold (16×d_eff) beyond which edge effects no longer distort G_IIf, the study provides a clear guideline for specimen preparation and data interpretation, ensuring that experimental results reflect true material behavior rather than geometric constraints. This is valuable for researchers developing or validating numerical models, as it defines the spatial limits within which local fracture energies can be considered reliable.