Strength-degradation phase-field regularization of cohesive fracture: the antiplane case

Impact Assessment

Core Contribution. This paper provides the antiplane-shear analysis and first systematic numerical implementation of a recently proposed variational fracture model (Bourdin, Marigo, Maurini, Zolesi, arXiv:2506.22558) in which the phase field degrades the *strength* rather than the *stiffness*. The central payoff is conceptual and practical: strength, stiffness, and toughness become three *independent* material inputs, dissolving a well-known structural limitation of standard phase-field fracture where the strength surface is dictated by the elastic energy and its magnitude is tied to the regularization length ℓ. In this formulation ℓ becomes a purely numerical parameter (when small relative to the elasto-cohesive length), and limit analysis, perfect plasticity, cohesive fracture, and brittle fracture emerge as regimes of one variational model with data (μ, τc, Gc). The paper derives closed-form homogeneous and localized (cohesive-law) solutions for a simple-shear problem, proposes a numerical scheme, and validates the equivalent cohesive law, its ℓ-independence, and the L/ℓch size effect.

Methodological Rigor. This is the paper's strongest dimension. The authors derive exact reference solutions (equivalent cohesive law, no-snap-back conditions, elasto-cohesive length ℓch = μGc/τc²) and then verify numerics against them quantitatively—matching peak strength within 1%, reproducing the analytical mesh-toughening factor (1+h/4cwℓ) to better than 1% for h≤ℓ/5, and recovering the Griffith–Barenblatt size-effect transition. The numerical scheme itself is a genuine and elegant contribution: exploiting separate convexity, each alternate-minimization subproblem is recast as a second-order cone program and solved to global optimality with MOSEK, without smoothing or penalization of the non-smooth strength term. The V-notch ("Pac-Man") study convincingly demonstrates the SSY→cohesive→brittle transition and compares against an independent sharp-interface model, matching to ~6% after discounting mesh toughening. The surfing experiment cleanly separates reversible (Geff = Gc) from irreversible (Geff ≈ 1.3Gc, plastic wake) behavior. This is careful, internally consistent work.

Potential Impact. The independence of strength from the regularization length addresses a genuine, frequently-cited bottleneck: standard phase-field requires ℓ to be tuned as a material property, becoming impractical for stiff/low-strength materials and for nominally brittle materials at large scales, and cannot reproduce measured multiaxial strength surfaces. If the broader program succeeds, this could meaningfully influence how the computational-fracture community regularizes cohesive fracture and unifies plasticity with fracture. However, the *concept* originates in the companion paper, and closely related models were independently introduced by Vicentini et al. and Feng & Li around the same time—indicating a converging research front rather than a singular breakthrough. The present paper's incremental contribution is the analytically transparent antiplane case plus the conic-programming machinery.

Timeliness & Relevance. Highly timely. The 2024–2026 literature shows intense, parallel activity on flexibly-tunable strength surfaces in phase-field fracture (multiple independent groups, Γ-convergence proofs by Maggiorelli et al. 2025, Alessi–Colasanto–Focardi three-part series). The work sits squarely in an emerging, competitive subfield.

Strengths & Limitations.

Strengths: (i) exact analytical benchmarks paired with matching numerics; (ii) a robust, smoothing-free conic-optimization solver that extends to multiaxial/plastic-irreversibility variants; (iii) clear physical interpretation unifying four classical fracture regimes; (iv) reproducibility aided by open tools (FEniCSx, Gmsh, MOSEK) and fully detailed conic reformulations in appendices.

Limitations: (i) the antiplane/scalar setting is deliberately simplified—crucially, it sidesteps the jump-compatibility condition that the authors admit is "one of the most delicate features" of the full tensorial model; generalizability to 3D is asserted but deferred to a companion paper; (ii) the core model is not original to this paper; (iii) results confirm theoretical expectations rather than surprising the field; (iv) no released code repository link, though method detail is high; (v) the effective-toughness inflation in the irreversible case is observed but not analytically characterized (acknowledged as open).

Other observations. The framing that ℓ can be treated as a pure numerical parameter is practically valuable and partially challenges the now-standard "ℓ as internal length" interpretation—more a reframing/qualification than a refutation. The scalar setting makes every derivation explicit, giving strong pedagogical and reference value. Resource requirements are modest (single lab, standard FE + convex-optimization tooling), lowering the barrier for others to build on the scheme. The work functions as a load-bearing verification step and numerical toolkit for a larger multi-paper program whose ultimate impact hinges on the forthcoming tensorial treatment.

Overall: a rigorous, timely, well-executed contribution that is somewhat incremental relative to its parent model and confined to a simplified setting, but which provides valuable analytical benchmarks and a reusable numerical methodology for a fast-moving subfield.