{"id":"0e898c11-2938-40b7-8fc5-02c1dd82a0b8","arxiv_id":"2607.26274","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Gravity-driven crevasse propagation in glaciers is a dynamic instability; only a dynamic phase-field model with degraded mass density produces sharp, localized cracks.","lead":"This paper argues that the slow, quasi-static equations normally used to simulate crevasse growth in glaciers break down, and that a dynamic model with inertia gives the physically realistic sharp cracks. If right, it changes how calving—a key driver of sea-level rise—should be simulated.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Dynamic sharp-crack result may hinge on nonstandard degradation of gravitational body force; no glacier run with undegraded body force is shown.","rationale":"The reader identified the same weakest assumption, and I agree. This is load-bearing because the paper's central claim is instantiated only through a formulation that degrades the sole driving force (gravity) inside the crack band. The self-acknowledged numerical motivation in Remark 1 and Appendix A makes the choice an assumption, not a derived consequence. The O(ε) argument is plausible, but for the glacier problem ε=10 mm gives a 4ε=40 mm band; the local effect on crack-tip energy release rate and on phase-field localization is not quantified. The lack of a glacier simulation with an undegraded body force means the sharp-crack result could be an artifact of this choice. The proposed concrete test would settle whether the conclusion is robust. I also note the internal timescale inconsistency (Fig. 8 caption says t=1 ms while the text says t=5 µs, both implying crack speeds beyond the elastic wave speed) as a secondary concern, but the body-force degradation is the more fundamental issue. Since the reader's conditional verdict already requires additional checks, my recommendation is unchanged: CONDITIONAL acceptance pending the sensitivity test and the other reproducibility items.","tokens_in":20867,"tokens_out":10667,"duration_ms":115006,"concrete_test":"Repeat the dynamic glacier simulation of Section 4.2 with the gravitational body force left undegraded (i.e., replace ⟨(v^2+η)b, w⟩ by ⟨b, w⟩ in Eq. 13/18), keeping b=2 for the inertial term, using a sufficiently small time step (e.g., Δt=0.0125 µs, the value that removes spurious branching in Sec. 4.1.2) and η=10^-4 to suppress mesh distortion. Compare the phase-field profile width vs. depth (as in Fig. 9) and crack speed with the published b=2 results. If the profile remains localized near 4ε and the speed is comparable, the body-force degradation is not load-bearing. If diffuse widening or severe distortion appears, the central claim rests on the nonstandard degradation and the paper needs a physical derivation or a demonstrated ε→0 convergence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that post-nucleation crevasse propagation is a dynamic instability, and that inertia is what converts quasi-static diffuse damage into a sharp 4ε crack (Figs. 8–9). Every simulation supporting this claim—both the quasi-static pathology (Eq. 8) and the dynamic glacier runs (Eq. 13, Section 4.2)—uses a nonstandard phase-field degradation of the gravitational body force: b is replaced by v^2 b in the momentum balance (Remark 1, Eq. 13, Appendix A). This means the material inside the regularized crack band has no weight. Physically, fractured ice still has mass and gravity still acts on it; a crevasse is an open traction-free surface, not a massless band. The paper justifies the choice only by numerical convenience (avoiding ill-conditioning/mesh distortion, citing Nguyen et al. [51]) and by an O(ε) regularization-error argument, not by a derivation from mass conservation or from a sharp-interface limit. Because gravity is the sole driving force in this problem, the removed body force in the 4ε band directly changes the energy budget near the tip. The b=0 vs b=2 comparisons in Section 4.1 are all in nongravitational benchmark problems; the glacier problem is never run with an undegraded body force. Thus the sharp localized crack may be a consequence of removing the driving force in the crack band rather than of inertia. The claim would survive robustly only if the glacier result is insensitive to this modeling choice or if the degradation is shown to vanish in the ε→0 limit in a gravity-loaded geometry.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that quasi-static phase-field simulations of gravity-driven crevasse propagation in grounded glaciers are pathologically overdriven after nucleation, producing diffuse, thickening damage bands, and that post-nucleation propagation is intrinsically a dynamic instability. It introduces a dynamic phase-field formulation (following Liu et al.) in which elastic stiffness, inertial mass, and gravitational body force are all degraded by v^2 in the fracture band. Benchmarks (1D wave, 2D plate, Kalthoff-Winkler, branching) show that degrading density removes spurious force transmission, mesh distortion, and artificial crack widening. Dynamic glacier simulations produce sharp 4-epsilon localized cracks, whereas quasi-static runs do not. The paper claims this establishes that accounting for inertia is necessary and sufficient for sharp through-thickness crevasse propagation.","tokens_in":21270,"tokens_out":5854,"duration_ms":53148,"significance":"If the central claim holds, the paper makes a substantive contribution: it identifies a failure mode of quasi-static phase-field fracture under fixed self-weight and proposes a dynamic regularized-fracture treatment with consistent degradation. The benchmark evidence is compelling and the finite-element formulation is described in enough detail to be reproducible with FEniCSx. The sharp-crack result is in principle falsifiable against glacier observations. However, the load-bearing modeling choice of degrading the gravitational body force with the phase field is physically nonstandard and is only benchmarked in nongravitational settings; until the glacier result is shown to be insensitive to this choice, or the choice is derived from a sharp-interface limit, the central claim remains conditional.","major_comments":[{"comment":"The dynamic formulation degrades the gravitational body force as (v^2+eta)b, removing the weight of material inside the regularized crack band. The paper justifies this by numerical convenience and an O(epsilon) mass-loss argument, not by a physical derivation; a sharp crack is an open traction-free surface, not a massless band. Because gravity is the sole driving force, removing body force in a 4-epsilon band (epsilon = 10 mm) changes the energy budget ahead of the tip and could itself produce sharp, fast propagation. All b=0 vs b=2 comparisons are in nongravitational benchmarks; no glacier simulation with undegraded body force is shown. Please add a glacier sensitivity study varying b (e.g., b=0, 1, 2) and/or a sharp-interface asymptotic derivation showing that body-force degradation does not alter the propagating crack solution at leading order.","section":"Eq. (13), Remark 1, Appendix A"},{"comment":"The quasi-static pathology is described as 'overdriven' because the strength surface is exceeded in a widening zone ahead of the tip. However, the same v^2 body-force degradation is already active in those quasi-static runs. It is therefore unclear whether the diffuse damage is a property of quasi-static evolution under self-weight or an artifact of the load removal in the damaged band. A quasi-static run with undegraded body force, if numerically manageable with stabilization, would disentangle the two mechanisms; without it, the claim that inertia is the specific remedy is not fully isolated.","section":"Section 3.2, Fig. 3"},{"comment":"The final comments acknowledge idealized 2D geometry, omitted ocean pressure, tidal forcing, and evolving basal traction, but the central claim is also not compared quantitatively with field observations. The cited calving timescale of 10-100 ms is not used to validate the simulated propagation time; Fig. 8 only shows t = 1 ms and no through-thickness time or crack speed is reported. Adding such a comparison would strengthen the assertion that the dynamic instability is the physically relevant regime.","section":"Section 4.2, Fig. 8, Section 5"}],"minor_comments":[{"comment":"The glacier dimensions are inconsistent: Fig. 1 says L=300 m, H=75 m; Section 2 says H=100 m; Fig. 2 caption says '100 m long and 75 m tall.' Please harmonize.","section":"Fig. 1, Section 2, Fig. 2"},{"comment":"The text and Fig. 2 caption disagree on whether the reduced-body-force case is (c) or (d). Please correct the cross-reference.","section":"Section 3.2, Fig. 2 caption"},{"comment":"References [18] and [19] are the same paper (Kumar et al., JMPS 2020). Please merge or renumber.","section":"References"},{"comment":"Fig. 8(b) reports sigma_ts = 0.2 MPa, while Section 4.2 says the weak boundary layer uses one-third of 0.7 MPa, i.e., approximately 0.233 MPa. Please reconcile.","section":"Fig. 8 caption vs Section 4.2"},{"comment":"The illustrative gravity-loaded specimen in Fig. 11 is not described with dimensions, material parameters, or loading details. Please either add these or explicitly state that the figure is schematic only.","section":"Appendix A, Fig. 11"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern identified by the reader is on point: the body-force degradation is the crux of the glacier result and is not investigated in a gravitational setting. I would ask for a b-sensitivity study or a sharp-interface derivation before publication; without it, the central claim is conditional on a nonstandard and physically unmotivated modeling choice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper is worth your time. The new bit is the diagnosis: quasi-static phase-field fracture of a gravity-loaded glacier, with realistic strength, becomes overdriven after crack initiation, so the damage band grows instead of forming a sharp crack. They show this in both the Kumar et al. model and a classical variational model, and they trace it to a growing over-stressed region ahead of the tip. Then they propose a dynamic formulation with consistently degraded density (b=2) and show that inertia gives a localized 4ε crack. That is a coherent story and a plausibly useful advance for calving modeling.\n\nThe benchmarks are mostly solid. The 1D bar and 2D plate tests show that undegraded density transmits a spurious reaction force, and the Kalthoff–Winkler and branching tests show that b=2 removes spurious branching and mesh distortion. These are real, reproducible-looking numerical observations, and they support the choice of b=2 as a numerical fix.\n\nNow the soft spots. The biggest one is exactly what the stress-test flagged: the gravitational body force is degraded with v^2 in every glacier run, and the paper never shows a glacier simulation with undegraded body force. The authors argue this is an O(ε) regularization error, and that is plausible, but it is not a derivation. Because gravity is the only driving force, a skeptic can reasonably ask whether the sharp crack is partly an artifact of removing weight from the crack band. The quasi-static vs dynamic comparison in Section 4.2 holds the degradation fixed, so the comparison itself is not confounded. But the physical claim that the sharp crack is 'the' dynamic behavior would be much stronger if they ran the dynamic glacier with b=0 (or a smaller ε) and showed the sharp crack persists. That is an experiment the paper should have done. I'd call this a moderate concern, not a fatal one.\n\nThere are also some embarrassing inconsistencies: the glacier height is 75 m in Figure 1, 100 m in Section 3.2, and 75 m again in the Figure 2 caption; the text says t=5 μs where the caption says t=1 ms; figure references are occasionally off. These are fixable but they make the paper harder to trust. And no code or data is released despite the FEniCSx implementation being mentioned.\n\nBottom line: the paper deserves serious refereeing. The central idea is interesting and the benchmarks are useful, but the referee should ask for a sensitivity study on the body-force degradation, a mesh-convergence check for the glacier run, and a cleanup of the text. I would send it to review, and I'd expect it to come back stronger.","headline":"Plausible dynamic phase-field fix for glacier fracture, but the sharp-crack result hinges on a body-force degradation that needs a sensitivity check, and the manuscript is sloppier than it should be.","tokens_in":21689,"tokens_out":4075,"would_cite":true,"duration_ms":39070,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["74R10","86A40","74S05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Gravity-driven crevasse propagation in grounded glaciers is a dynamic instability: once a crevasse nucleates, inertia carries a sharp, localized crack through the ice thickness, while quasistatic phase-field models produce unphysical diffus","keywords":["glacier fracture","crevasse propagation","phase-field fracture","dynamic fracture","gravity loading","strength surface","calving","brittle ice"],"falsifier":"Re-run the glacier crevasse simulation with the dynamic formulation but leave the gravitational body force undegraded in the fractured band, using a contact condition or small residual stiffness to keep the problem well posed; if the crack no longer propagates as a sharp, localized band through the full thickness, then the dynamic-instability conclusion is an artifact of the body-force degradation. Alternatively, a field measurement of a calving event showing surface crevasse deepening over minutes to hours rather than elastic milliseconds would contradict the claim.","tokens_in":20791,"feed_emoji":"🧊","tokens_out":6505,"duration_ms":61721,"temperature":0.7,"pith_summary":"This paper tries to establish that fracture propagation in grounded glaciers under self-weight cannot be treated as a quasistatic process. Using a phase-field fracture model—a description in which a scalar field smears a crack over a small width—the authors show that quasistatic simulations after crevasse initiation become \"overdriven\": gravity keeps the stress state ahead of the crack tip beyond the material's strength surface, so the diffuse crack band progressively thickens and produces diffuse damage patterns. They argue that the post-nucleation phase is fundamentally a dynamic instability, and that including inertia yields sharp, localized cracks that run through the ice thickness on millisecond timescales. A reader should care because calving events contribute to sea-level rise, and the finding changes how crevasse propagation should be modeled: dynamic fracture, not quasistatic linear elastic fracture mechanics, is the right framework for the fast final stage.","feed_headline":"Crevasse growth is a dynamic instability, not a quasistatic crack","feed_subtitle":"Accounting for inertia turns diffuse damage into sharp, through-thickness cracks that can calve glaciers.","key_machinery":"The load-bearing object is a phase-field fracture model with a strength surface: a scalar field v in [0,1] smears the crack over a regularized width proportional to a length parameter ε, while a pressure-sensitive strength criterion such as F(σ) = √J2 + γ1I1 + γ0 = 0 governs nucleation and fracture-energy competition governs propagation. The paper's dynamic extension adds an inertial term to the balance of linear momentum and degrades elastic, gravitational, and inertial contributions consistently: Div((v²+η)σ) + (v²+η)b = (v^b + ηρ)ρ ü, with b = 2. The argument is carried by comparing b = 0 (undegraded density) with b = 2: undegraded density lets the regularized crack carry momentum and tra","core_discovery":"The paper's central claim is that, for a grounded glacier loaded by fixed self-weight, once a crevasse nucleates or is seeded, the gravitational pre-stress overdrives the stress field ahead of the crack tip far beyond the strength surface, so the resulting growth is an unstable dynamic event rather than a slow quasistatic one. In the phase-field model used here, the quasistatic formulation responds by widening the regularized crack band instead of advancing a sharp crack. The dynamic formulation—in which elastic stiffness, gravitational body force, and mass density are all degraded by the same phase-field factor—produces a crack of optimal regularized width that propagates through the full t","pith_inferences":["Editorial inference: glacier models that treat crevasse deepening with quasistatic or viscous damage mechanics may systematically overestimate damage-zone width and underestimate propagation speed near the terminus; a practical extension would be a dynamic switching criterion triggered when the strength surface is exceeded over a critical zone.","Editorial inference: the paper's choice to degrade the gravitational body force with the phase field is not physically derived from a cracked body's mass balance; if real crack faces still carry self-weight, removing body force inside the band lowers the energy needed to open the crack, and I would test whether the sharp, fast propagation survives with an undegraded self-weight plus a contact cond","Editorial inference: the dynamic-instability picture suggests calving events may radiate detectable stress waves, and measured crevasse-deepening rates from seismic or fiber-optic instruments could discriminate dynamic versus quasistatic propagation.","Editorial inference: because the predicted critical crevasse depth is insensitive to tensile strength over the realistic range, the model makes a testable prediction that surface crevasse depths in grounded glaciers should be bounded near this depth unless meltwater pressure or other processes intervene."],"forward_implications":["Quasistatic phase-field simulations of grounded glaciers under self-weight should not be used for post-nucleation crevasse propagation: they produce diffuse, thickening damage zones rather than sharp cracks.","A dynamic formulation with consistently degraded elastic, inertial, and gravitational terms yields sharp, localized cracks that propagate through the ice thickness, matching the rapid (10–100 ms) timescales reported for calving events.","Leaving mass density undegraded in dynamic phase-field fracture creates spurious force transmission across the regularized crack, artificial crack branching, mesh distortion, and crack widening; degrading density removes these artifacts.","The critical crevasse depth for arrest in grounded ice is a strength-governed quantity (about 72 m for the adopted parameters), and it matches linear elastic fracture mechanics predictions without invoking fracture toughness.","Multiple surface crevasses under gravitational pre-stress grow competitively, with some arresting and others shielding, so the dynamic model can capture crevasse–crevasse interaction."],"fun_headline_variants":["Glacier crevasses are dynamic cracks, not quasistatic ones","Crevasses grow as dynamic instabilities, not slow cracks","Dynamic phase-field model yields sharp glacier cracks","Gravity overdrives crevasses into dynamic fracture"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the gravitational body force, and hence mass density, can be degraded with the phase field inside the crack band; if a real crevasse's faces still carry self-weight, removing that body force artificially lowers the energy cost of opening the crack and could be what makes the sharp, fast dynamic propagation appear.","fun_headline_variants_meta":{"raw":{"variants":["Glacier crevasses are dynamic cracks, not quasistatic ones","Crevasses grow as dynamic instabilities, not slow cracks","Dynamic phase-field model yields sharp glacier cracks","Gravity overdrives crevasses into dynamic fracture"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000138,"raw_usage":{"total_tokens":977,"prompt_tokens":720,"completion_tokens":257,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":464,"completion_tokens_details":{"reasoning_tokens":202}},"tokens_in":464,"tokens_out":257,"duration_ms":3154,"temperature":1.0,"reasoning_tokens":202,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T00:17:10.292235+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the glacier crevasse simulation with the dynamic formulation but leave the gravitational body force undegraded in the fractured band, using a contact condition or small residual stiffness to keep the problem well posed; if the crack no longer propagates as a sharp, localized band through the full thickness, then the dynamic-instability conclusion is an artifact of the body-force degradation. Alternatively, a field measurement of a calving event showing surface crevasse deepening over minutes to hours rather than elastic milliseconds would contradict the claim.","supporting_citations":[],"review_version":1}