Fractocohesive Length Scale

• Fractocohesive length scale is a family of related metrics that mark the transition from local cohesive-damage processes to large-scale elastic crack mechanics.
• It is quantified via experimental techniques like fracture-surface analysis, size-effect tests, and cohesive modeling across materials such as metals, ceramics, elastomers, and composites.
• This scale governs critical transitions in fracture behavior, including shifts from multi-fractal to mono-affine roughness and from flaw-insensitive to flaw-sensitive failure modes.

The fractocohesive length scale is an intrinsic or effective length that marks the crossover between fracture governed by localized cohesive, damage, or dissipation mechanisms and fracture governed by larger-scale elastic crack mechanics. In the literature, the same underlying idea appears under several notations and operational definitions: as a crossover scale extracted from fracture-surface morphology, as an effective fracture process zone length obtained from size-effect tests, as a cohesive or nonlocal regularization length in continuum models, and, in elastomers, as the ratio between fracture toughness and work to fracture. Across these usages, it separates regimes such as multi-fractal versus mono-affine roughness, coalescence-dominated versus elastic-line crack propagation, flaw-insensitive versus flaw-sensitive failure, or diffuse damage versus sharp-interface fracture (Vernede et al., 2014, Santucci et al., 2010, Najmeddine et al., 3 Dec 2025).

1. Concept and terminological scope

The term does not denote a single universally fixed quantity across all fracture subfields. Instead, it denotes a family of closely related length scales that quantify the spatial extent of the fracture process zone, the width of the dissipative region, or the scale at which continuum elastic descriptions become valid.

Context	Representative definition	Physical role
Fracture-surface morphology	$\xi$ extracted from spatial correlations of local slopes	Crossover from multi-fractal damage patterns to mono-affine roughness
Planar crack fronts	$\delta^*$ where $\sigma(\delta^*)/\delta^* = \alpha$	Transition from coalescence-dominated to fluctuating-line behavior
Quasibrittle size effect	Effective FPZ length $c_f$	Controls deviation from LEFM and pseudo-ductile-to-brittle crossover
Elastomers	$\xi = G_c/W_c$	Critical flaw size and transition from flaw-insensitive to flaw-sensitive response
Hydraulic fracture	$L_{coh} = E' w_c/\sigma_c$	Characteristic size of the cohesive/process zone
Cohesive phase-field models	$l_{ch} = E G_c/\sigma_0^2$	Physical process-zone width decoupled from numerical regularization

In fracture-surface studies, $\xi$ is obtained from the spatial correlations of “steep cliffs” on the post-mortem fracture surface (Vernede et al., 2014). In planar crack-front studies, the analogous quantity is the crossover length $\delta^*$, defined operationally by the local slope criterion $\sigma(\delta^*)/\delta^* = \alpha$ (Santucci et al., 2010). In quasibrittle structural fracture, the corresponding quantity is usually the effective FPZ length $\delta^*$0, extracted through Bažant-type size-effect laws (Li et al., 2017). In elastomers, the fractocohesive length is defined directly as $\delta^*$1, a material-specific ratio between fracture toughness and work of fracture (Najmeddine et al., 3 Dec 2025). In cohesive and gradient models, comparable lengths appear as $\delta^*$2, $\delta^*$3, $\delta^*$4, or $\delta^*$5, each parameterizing the finite spatial width of dissipation or nonlocality (Liu et al., 2020, Roch et al., 2022, Rezaei et al., 2021).

Taken together, these works suggest that the fractocohesive length scale is best understood as a crossover metric rather than a single formula: it identifies the scale below which failure is controlled by local damage, cohesive tractions, bond scission, or disorder, and above which scale-free elastic fracture descriptions become appropriate.

2. Morphological and crack-front crossover scales

A particularly explicit definition arises from the statistics of fracture surfaces. In turbulent-like fracture morphologies, the local slope field is defined by

$\delta^*$6

with spatial correlations

$\delta^*$7

At short distances,

$\delta^*$8

while for $\delta^*$9 the correlations are negligible. The length $\sigma(\delta^*)/\delta^* = \alpha$0 is therefore the distance at which the logarithmic correlation extrapolates to zero (Vernede et al., 2014).

This $\sigma(\delta^*)/\delta^* = \alpha$1 separates two scaling regimes. Below $\sigma(\delta^*)/\delta^* = \alpha$2, fracture surfaces are multi-fractal: local slope amplitudes are strongly correlated, and the moments of height differences

$\sigma(\delta^*)/\delta^* = \alpha$3

display $\sigma(\delta^*)/\delta^* = \alpha$4-dependent scaling. Above $\sigma(\delta^*)/\delta^* = \alpha$5, the surface becomes mono-affine, with $\sigma(\delta^*)/\delta^* = \alpha$6 approximately constant at $\sigma(\delta^*)/\delta^* = \alpha$7 (Vernede et al., 2014). The same work reports $\sigma(\delta^*)/\delta^* = \alpha$8 for an aluminum alloy, $\sigma(\delta^*)/\delta^* = \alpha$9 for mortar, and $c_f$0 for ceramics, with correlated “steep cliff” clusters having fractal dimension $c_f$1 and a power-law cluster-size distribution $c_f$2 with $c_f$3 (Vernede et al., 2014).

A parallel picture appears in planar crack fronts. Large in-plane crack fronts propagating along weak interfaces exhibit a crossover at $c_f$4. Below $c_f$5, the fronts are multi-affine with $c_f$6, consistent with coalescence models. Above $c_f$7, they are mono-affine with $c_f$8, consistent with fluctuating-line or depinning descriptions. Experimentally, $c_f$9 is observed around $\xi = G_c/W_c$0, although it varies with disorder (Santucci et al., 2010). The operational definition,

$\xi = G_c/W_c$1

with $\xi = G_c/W_c$2, makes the crossover a geometric slope criterion rather than a purely statistical fitting parameter (Santucci et al., 2010).

Numerical constrained-crack-growth studies reinforce this duality. In an extended fiber-bundle model, the large-scale regime follows the pinned elastic line universality class with roughness exponent $\xi = G_c/W_c$3, whereas the small-scale regime is governed by hole coalescence and gradient percolation with roughness exponent $\xi = G_c/W_c$4 and crack-front fractal dimension $\xi = G_c/W_c$5, consistent with $\xi = G_c/W_c$6 (Gjerden et al., 2013). The crossover length is interpreted there as a fractocohesive length separating local, damage-controlled rupture from nonlocal, elasticity-controlled crack-front motion.

These results support a recurring interpretation: below the fractocohesive scale, crack advance is governed by nucleation, growth, and merging of cavities or holes; above it, the crack can be idealized as an elastic interface in a random medium.

3. Effective process-zone length in quasibrittle size effect

In quasibrittle solids and composites, the fractocohesive scale is commonly identified with the effective fracture process zone length $\xi = G_c/W_c$7. The central equivalent-fracture-mechanics ansatz is

$\xi = G_c/W_c$8

so that the actual crack behaves as if it were longer than the initial notch by the effective FPZ length. For mode-I size-effect analyses, the energy release rate is written as

$\xi = G_c/W_c$9

and the nominal strength at failure follows the generalized form

$L_{coh} = E' w_c/\sigma_c$0

or, in the classical Bažant form,

$L_{coh} = E' w_c/\sigma_c$1

In this framework, $L_{coh} = E' w_c/\sigma_c$2 governs the crossover from pseudo-ductile, FPZ-dominated behavior to brittle LEFM scaling (Ko et al., 2018, Li et al., 2017).

For discontinuous fiber composites, geometrically scaled SENT tests combined with stochastic finite element modeling yielded $L_{coh} = E' w_c/\sigma_c$3 and $L_{coh} = E' w_c/\sigma_c$4 for $L_{coh} = E' w_c/\sigma_c$5 platelets, $L_{coh} = E' w_c/\sigma_c$6 and $L_{coh} = E' w_c/\sigma_c$7 for $L_{coh} = E' w_c/\sigma_c$8 platelets, and $L_{coh} = E' w_c/\sigma_c$9 and $l_{ch} = E G_c/\sigma_0^2$0 for $l_{ch} = E G_c/\sigma_0^2$1 platelets. A quasi-isotropic laminate made from the same constituents had $l_{ch} = E G_c/\sigma_0^2$2 and $l_{ch} = E G_c/\sigma_0^2$3 (Ko et al., 2018). The reported interpretation is that DFCs possess a much larger FPZ than laminated composites, and that the FPZ length is a key scale governing the pseudo-ductile-to-brittle transition.

For Marcellus shale, size-effect tests on three-point-bending specimens showed pronounced anisotropy. Using SEL, the effective FPZ lengths were $l_{ch} = E G_c/\sigma_0^2$4 in arrester orientation, $l_{ch} = E G_c/\sigma_0^2$5 in divider orientation, and $l_{ch} = E G_c/\sigma_0^2$6 in short-transverse orientation, with corresponding fracture energies $l_{ch} = E G_c/\sigma_0^2$7, $l_{ch} = E G_c/\sigma_0^2$8, and $l_{ch} = E G_c/\sigma_0^2$9 (Li et al., 2017). The divider orientation therefore exhibited the largest FPZ and the least brittle response, while the arrester orientation showed the smallest FPZ and the most brittle response.

A related program for mode-II interlaminar fracture in geometrically scaled ENF composite specimens combined microscopic and macroscopic DIC with through-thickness in-plane displacement analysis. There, Bažant’s type-II size-effect law was used to extract both a unique mode-II fracture energy and an effective process-zone length $\xi$0, while full-field DIC supplied local separations and FPZ size estimates for cohesive-law calibration (Kim et al., 2023). The main methodological conclusion is that global load–displacement data alone are insufficient; local full-field measurements are needed to determine a physically meaningful traction–separation law and characteristic length.

4. Cohesive, nonlocal, and phase-field formulations

Cohesive and regularized fracture models elevate the fractocohesive scale from an inferred experimental quantity to an explicit constitutive or modeling parameter. In plane-strain hydraulic fracture with a rough cohesive zone, the traction–separation law is

$\xi$1

with

$\xi$2

The same model modifies Poiseuille flow through a roughness correction

$\xi$3

Here $\xi$4 is the characteristic size of the cohesive zone, and the coupled hydro-mechanical problem is organized by a nucleation time scale $\xi$5, which separates nucleation, intermediate, and late-time stages. At early and intermediate times, the fracture is shorter, wider, and more pressurized than predicted by LHFM, and these deviations increase with dimensionless toughness, in-situ-to-cohesive stress ratio, and roughness exponent (Liu et al., 2020).

Dynamic cohesive fracture theory introduces a related dissipation length as the static process-zone size

$\xi$6

for a linear slip-weakening law. In heterogeneous cohesive materials, this finite process-zone size acts as a low-pass filter on heterogeneities: features smaller than $\xi$7 are averaged out, and the front deformation amplitude depends on both the toughness contrast and whether heterogeneity enters through peak strength or process-zone size (Roch et al., 2022). Crack dynamics further contract the process zone according to

$\xi$8

so scale effects are mitigated at higher propagation speeds (Roch et al., 2022).

In nonlocal cohesive peridynamics, the governing intrinsic scale is the horizon $\xi$9, the ratio of nonlocal interaction length to the characteristic sample dimension. The process zone is defined as the set of points for which a sufficient fraction of bonds inside the horizon exceed a critical strain threshold, and its measure obeys the explicit bound

$\delta^*$0

As $\delta^*$1, the process zone concentrates onto a set of zero volume, and the nonlocal energy $\delta^*$2-converges to Griffith fracture energy (Lipton, 2014). In that sense, the fractocohesive length is the parameter controlling the transition from smeared cohesive fracture to sharp-interface LEFM.

Cohesive phase-field models adopt yet another form,

$\delta^*$3

with the goal of making structural predictions almost insensitive to the numerical regularization length $\delta^*$4. The central idea is to encode both fracture energy and peak strength in the degradation law, so the physical process-zone width is set by the cohesive properties rather than directly by $\delta^*$5 (Rezaei et al., 2021). This formulation is extended to anisotropic materials by making both $\delta^*$6 and $\delta^*$7 direction dependent.

5. Elastomers, soft solids, and intrinsic dissipation widths

In elastomers, the fractocohesive length is defined directly as

$\delta^*$8

where $\delta^*$9 is fracture toughness and $\sigma(\delta^*)/\delta^* = \alpha$0 is work of fracture (Najmeddine et al., 3 Dec 2025). As discussed there, the concept originates with Thomas (1955) and Chen et al. (2017), and the ratio corresponds to a critical flaw size below which fracture becomes flaw-insensitive. The same study reports that, for thermally aged Styrene Butadiene Rubber and Silicone Rubber, both $\sigma(\delta^*)/\delta^* = \alpha$1 and $\sigma(\delta^*)/\delta^* = \alpha$2 decrease significantly with aging, yet their ratio remains statistically invariant across time, temperature, and material system. The measured values were $\sigma(\delta^*)/\delta^* = \alpha$3 for SBR aged at $\sigma(\delta^*)/\delta^* = \alpha$4, $\sigma(\delta^*)/\delta^* = \alpha$5 for SBR aged at $\sigma(\delta^*)/\delta^* = \alpha$6, and $\sigma(\delta^*)/\delta^* = \alpha$7 for SR aged at $\sigma(\delta^*)/\delta^* = \alpha$8 (Najmeddine et al., 3 Dec 2025). For the AT1 phase-field model, the same work derives

$\sigma(\delta^*)/\delta^* = \alpha$9

linking the experimentally measured fractocohesive length to a computational regularization parameter (Najmeddine et al., 3 Dec 2025).

A gradient-enhanced damage formulation for elastomers arrives at the same ratio,

$\delta^*$00

but interprets it as the full width of the damage zone where irreversible dissipation through bond scission occurs (Mousavi et al., 30 Aug 2025). In that model, the imposed nonlocal length $\delta^*$01 controls the core highly damaged region, whereas the emergent $\delta^*$02 matches the width of the overall damage zone and the transition from flaw-insensitive to flaw-sensitive behavior. Numerically, flaws with sizes $\delta^*$03 produce notch-size-insensitive response, while $\delta^*$04 recover classical flaw sensitivity (Mousavi et al., 30 Aug 2025).

Soft brittle solids also exhibit a distinct nonlinear elastic scale,

$\delta^*$05

derived as the distance from the crack tip where weakly nonlinear elastic corrections become comparable to LEFM fields (Bouchbinder et al., 2013). This scale sets the wavelength of a high-speed oscillatory crack instability and provides an intrinsic near-tip length absent from LEFM (Bouchbinder et al., 2013). In three-dimensional brittle hydrogels with multiple interacting fracture planes, a stabilized crack-front step height $\delta^*$06 was found to vary linearly with both $\delta^*$07 and a dissipation length $\delta^*$08, according to

$\delta^*$09

with $\delta^*$10 (Wang et al., 2024). A plausible implication is that, in soft solids, the fractocohesive scale often coexists with a nonlinear elastic scale, and observable defect sizes may be selected by both.

6. Interpretation, misconceptions, and extensions

A common misconception is that the fractocohesive length is a single universally defined material constant. The literature shows instead that the name refers to several operationally distinct but physically related quantities: $\delta^*$11, $\delta^*$12, $\delta^*$13, $\delta^*$14, $\delta^*$15, $\delta^*$16, or $\delta^*$17, depending on whether the emphasis is on morphology, size effect, cohesive modeling, nonlocal regularization, or soft-matter dissipation (Vernede et al., 2014, Li et al., 2017, Rezaei et al., 2021).

A second misconception is that classical LEFM alone is sufficient once fracture energy is known. Multiple studies explicitly show that this is not generally the case. Fracture surfaces exhibit a scale-dependent transition from correlated multi-fractality to Gaussian mono-affinity (Vernede et al., 2014); crack fronts display separate coalescence and elastic-line regimes (Santucci et al., 2010, Gjerden et al., 2013); quasibrittle structures require $\delta^*$18 to explain strength scaling (Ko et al., 2018, Li et al., 2017); and cohesive hydraulic fractures can remain far from LHFM predictions over long nucleation and intermediate stages (Liu et al., 2020).

A third misconception is that internal length parameters in regularized models are purely numerical artifacts. Cohesive phase-field, peridynamic, and gradient-enhanced damage formulations all attempt to relate these lengths to measurable dissipation widths, material strengths, or process-zone sizes (Lipton, 2014, Rezaei et al., 2021, Mousavi et al., 30 Aug 2025). The extent to which such parameters are fully material-intrinsic remains model dependent, but the cited works do not treat them as arbitrary smoothing constants.

The broader relevance of the concept extends beyond tensile cracks. In disorder-controlled slip at frictional interfaces, the largest avalanche size that can still be arrested by disorder is

$\delta^*$19

and the cutoff is explicitly interpreted as a fractocohesive length scale separating disorder-arrested avalanches from system-spanning fracture events (Geus et al., 2022). This suggests that the concept generalizes to driven dissipative interfaces whenever one needs to identify the scale beyond which local disorder can no longer arrest collective failure.

Across these formulations, the central meaning remains stable: the fractocohesive length scale is the length below which fracture cannot be idealized as a sharp, scale-free elastic crack. It is the scale of correlated damage, cohesive tractions, bond scission, process-zone filtering, or disorder arrest, and it governs the transition from local dissipation physics to macroscopic crack propagation.