{"id":"99dd9f8f-64ee-4097-971c-59ab802c8923","arxiv_id":"2607.03776","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":5,"one_line_summary":"A double-porosity phase-field model with independent crack and matrix pressures, fixed-stress splitting, and a variational inequality for non-negative pressure reproduces K-, M-, and O-vertex KGD hydraulic-fracture solutions including fluid lag.","lead":"A dual-pressure phase-field model separates crack fluid pressure from matrix pore pressure so hydraulic fractures can be simulated in the viscosity-dominated regime, including fluid lag. That regime dominates real subsurface injections and had been out of reach for prior phase-field fracture codes.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"The hard phase-field threshold that confines fracture flow is an unproven free parameter whose effect on mass balance and lag size is not quantified.","rationale":"The Reader correctly isolates the hard threshold χ(v) and the forced pc=pp outside the fractured zone as the weakest link. That construction is indispensable for the dual-continuum claim: without it the fracture-flow equation would be solved everywhere and the pressure discontinuity that stabilizes the viscosity-dominated regime would be lost. Because the paper supplies neither the numerical value of vcr nor any sensitivity study, and because the lag under-prediction is already visible in the published figures, the concern is concrete and load-bearing. The recommended concrete test is a minimal, one-parameter re-run that directly quantifies the dependence; if the front location proves insensitive, the Reader’s residual reservation can be retired and the verdict can move toward unconditional acceptance. Until that check (or an equivalent mass-conserving reformulation) is performed, CONDITIONAL remains the appropriate verdict. No stronger objection (e.g., inconsistency of the microporomechanical derivation or failure of the fixed-stress split) is visible in the manuscript.","tokens_in":26237,"tokens_out":726,"duration_ms":6164,"concrete_test":"Re-run Case 1 of Table 2 (Gc=60 Pa·m, Km=0.593) at fixed mesh h=0.5 m for three values of the threshold, e.g. vcr∈{0.1,0.3,0.5}, extract the dimensionless fluid-front location ξf at t=18 s (the post-processing time used for Fig. 8), and compare against the analytical family ξf∈{0.3–0.6}. If ξf changes by more than ~0.05 across the three thresholds, the lag-zone size is numerically controlled by the free cutoff and the O-vertex verification is not robust.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that the dual-continuum model is the first phase-field formulation to stably reproduce M-vertex (Km<0.7) and O-vertex fluid-lag KGD solutions rests on the numerical construction in §3.2–3.3. Fracture flow (Eq. 63) is multiplied by the discontinuous indicator χ(v)=1 only when v≤vcr (Eq. 65); outside that zone pc is forcibly set equal to pp (Eq. 67). The variational inequality then enforces pc≥0 inside the active set. The paper never reports the value of vcr used in any of the four lag cases of §4.5, nor does it supply a mesh- or threshold-convergence study showing that the fluid-front location ξf (and therefore the lag length) is independent of vcr and of the element size h=0.5 m. The authors themselves attribute the systematic under-prediction of lag size (Figs. 8b,d,f,h) in part to “non-strict enforcement of the mass conservation condition \times zero flux from the pressurized zone to the lag zone.” Because the M- and O-vertex verifications are precisely the regimes in which a sharp fluid front must be resolved, an uncalibrated free cutoff that artificially advances that front is load-bearing: if the reported agreement with Garagash asymptotics is an artifact of a particular vcr, the claim that the formulation “accurately captures \times transient fluid lag” is overstated.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper develops a dual-continuum phase-field model for hydraulic fracturing in double-porosity media, with independent mesoscale crack pressure pc and micropore pressure pp derived from microporomechanics (effective stiffness, Biot tensors, and Biot moduli degraded by the phase field). A modified fixed-stress split handles the two-pressure hydromechanical system; a phase-field indicator χ(v) gates fracture (Reynolds) flow to a damaged subdomain; and a variational inequality enforces non-negative pressures to capture fluid lag without front tracking. Plane-strain KGD verifications are reported against Garagash–Detournay asymptotics in the toughness-dominated (K), viscosity-dominated (M, Km≈0.59), and early-time lag (O) regimes, with spatial aperture and pressure profiles compared for several Km values.","tokens_in":26782,"tokens_out":1153,"duration_ms":14971,"significance":"If the formulation is robust, this is a genuine advance: existing phase-field hydraulic-fracture models have largely been limited to toughness-dominated or near-M regimes and cannot represent a sharp pressure drop or fluid lag under a smeared continuous pressure. The microporomechanical derivation of phase-field-dependent poroelastic coefficients, the two-pressure fixed-stress scheme, and the variational-inequality lag treatment are concrete technical contributions. Direct, external comparison to closed-form KGD solutions (injection pressure, aperture, length, and spatial profiles) is the right standard of evidence and is largely met for K and M. The work would open phase-field modeling to the viscosity-dominated conditions that dominate field injections.","major_comments":[{"comment":"§3.2, Eqs. (65)–(67): Fracture flow is gated by a hard threshold χ(v)=1 only for v≤vcr, with pc forced equal to pp outside that zone. The value of vcr used in any verification (especially the four lag cases of §4.5) is never stated, and no mesh- or threshold-sensitivity study is given for the fluid-front location ξf or lag length. Because the M- and O-vertex claims rest on resolving a sharp fluid front, and the authors themselves attribute lag underestimation partly to non-strict mass conservation across the pressurized/lag interface created by this construction, independence of the reported fronts from vcr and h must be demonstrated (or vcr fixed by a clear physical/numerical criterion and shown to be non-influential).","section":null},{"comment":"§4.5, Figs. 8b,d,f,h: The model systematically underestimates lag size (simulated ξf more advanced than the Garagash family). The abstract and conclusions state that the model “accurately captures … transient fluid lag.” That wording overstates the evidence; either the claim should be qualified to match the documented bias, or the mass-conservation / cavitation treatment should be tightened so that lag length converges to the asymptotics before the accuracy claim is retained.","section":null},{"comment":"Table 1 and §4.2–4.5: All KGD verifications set αm=0, ϕm=0, Kp=10^{-19} m² so leak-off and inter-scale exchange vanish. The dual-continuum machinery is then used mainly to allow pc≠pp and to gate fracture flow. That is legitimate for impermeable KGD benchmarks, but the paper’s framing as a double-porosity model for realistic viscosity-dominated fracturing is not yet supported by any case with nonzero matrix storage/exchange. At least one permeable double-porosity demonstration (or an explicit statement that verification is impermeable-only and double-porosity exchange remains untested) is needed so the central claim is not read as broader than the evidence.","section":null}],"minor_comments":[{"comment":"Introduction: “viscousity-dominated” is a typo (should be “viscosity-dominated”).","section":null},{"comment":"§2.2, Eq. (26): The argument list of Eℓ is written “Eℓ(ε, , ζc, ζp, v)” with a missing field; clean the notation.","section":null},{"comment":"§3.1: Unconditional stability of fixed-stress is cited for single-porosity isotropic cases (Kim et al.); a short remark that the two-pressure anisotropic extension (53)–(56) is used without a new stability proof would set expectations correctly.","section":null},{"comment":"Fig. 7 caption: “(c) and (d)” are labeled inconsistently with the body text (“phase-field (b) and pressure (c)”); align labels.","section":null},{"comment":"Throughout: report the numerical value of vcr, the shape factor S (if used), and whether gravity is active in the KGD runs so results are reproducible.","section":null}],"recommendation":"major_revision","confidential_remarks":"The “first phase-field model for M-vertex with Km<0.7 and fluid lag” claim is fair relative to the cited Wilson & Landis and Santillán works, which the authors correctly note sit above the Garagash M-vertex threshold. The main risk is not novelty but that the lag results may be sensitive to an unreported free cutoff; requiring a short sensitivity appendix would protect the journal and the claim without demanding a full reformulation."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is the first phase-field hydraulic-fracture paper that actually reaches the viscosity-dominated regime (Km < 0.7) and resolves fluid lag against Garagash asymptotics, rather than stopping at toughness or claiming near-M with Km ~1. The dual continuum (independent pc and pp from double-porosity microporomechanics), the two-pressure fixed-stress split, and the variational inequality for non-negative pressure are the concrete pieces that make it work. They remove the smeared-pressure instability that has kept earlier phase-field codes out of the field-relevant M-vertex.\n\nWhat they do well is clear. The K-, M-, and O-vertex comparisons (Figs. 4–8) are against independent closed-form solutions, not self-generated curves. Injection pressure, aperture, length, and spatial profiles line up for toughness and viscosity cases; the lag cases show the correct qualitative trend with Km. The microporomechanical state equations are taken from the Dormieux/Pichler line and specialized cleanly; the energy and Biot-moduli derivations are variationally consistent. Citation pattern is honest about prior phase-field HF limits.\n\nThe soft spot the stress-test flags is real and load-bearing for the lag claim: fracture flow is gated by a hard indicator χ(v) = 1 only for v ≤ vcr, with pc forced to pp outside. They never report the vcr value used in the four lag cases, nor any mesh/threshold study showing ξ f is independent of that cutoff or of h = 0.5 m. They themselves note the lag zone is systematically under-predicted and attribute part of it to non-strict mass conservation across the pressurized/lag interface. That weakens the “accurately captures transient fluid lag” language. Double-porosity exchange is also not exercised in the impermeable benchmarks, so the dual-continuum claim is stronger than the evidence for permeable media. No code is shipped.\n\nStill, the central advance holds: they have a stable formulation that reproduces M-vertex and lag solutions where previous phase-field work did not. This is for people who write or use phase-field HF codes and need the viscosity-dominated regime. It deserves a serious referee; the vcr and mass-conservation issues are fixable in revision. I would engage with it and expect to cite the method once the cutoff sensitivity is quantified.","headline":"Solid dual-pressure phase-field HF that finally hits verified M- and O-vertex KGD solutions; the hard vcr cutoff is a real soft spot but does not erase the advance.","tokens_in":27319,"tokens_out":588,"would_cite":true,"duration_ms":5930,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A dual-pressure phase-field model stably captures viscosity-dominated hydraulic fractures and fluid lag for the first time.","keywords":["phase-field","double porosity","hydraulic fracture","micromechanics","viscosity-dominated regime","fluid lag","fixed-stress split","variational inequality"],"falsifier":"A mesh-refinement and threshold-sensitivity study of the lag-zone length in the early-time KGD problem that either shows the computed fluid-front location converging to the analytical ξ f values of Garagash or demonstrates a persistent, threshold-dependent under-estimate of the lag.","tokens_in":27113,"feed_emoji":"💧","tokens_out":623,"duration_ms":5379,"temperature":0.7,"pith_summary":"Existing phase-field models of hydraulic fracturing smear pressure across a diffuse crack zone. That works when toughness dominates and pressure is nearly uniform inside the fracture, but fails under the high injection rates of real subsurface operations: viscous pressure drops become large, the fluid front can lag behind the crack tip, and the smeared continuous pressure drives numerical instability. This paper replaces the single pressure with two independent fields—mesoscale crack pressure and micropore pressure—derived from double-porosity microporomechanics, together with a modified fixed-stress split and a variational inequality that enforces non-negative pressure. The resulting dual-continuum model is shown to reproduce closed-form KGD solutions across the toughness-dominated, viscosity-dominated, and early-time fluid-lag regimes, thereby making phase-field methods usable for the viscosity-dominated conditions that actually govern field hydraulic fracturing.","feed_headline":"Dual-pressure phase-field model captures viscous hydraulic fractures","feed_subtitle":"First phase-field formulation to match analytical solutions when fluid lag and large pressure drops appear","key_machinery":"The dual-continuum pressure pair (pc, pp) together with the dynamic indicator χ(v) that confines fracture flow to the damaged zone and the variational inequality that forces both pressures to remain non-negative; these three ingredients together remove artificial pressure continuity and allow a lag zone to form without explicit front tracking.","core_discovery":"A dual-continuum phase-field formulation that evolves mesoscale crack pressure and micropore pressure independently, with phase-field-dependent poroelastic coefficients taken from microporomechanics, a fixed-stress split adapted to the two-pressure system, and a variational inequality that enforces non-negative pressures, is the first phase-field model to stably recover analytical KGD solutions in the viscosity-dominated (M-vertex) and early-time fluid-lag (O-vertex) regimes as well as the toughness-dominated regime.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Dual-continuum phase-field resolves viscous fractures and fluid lag","Phase-field separates crack and pore pressures in viscosity regime","First phase-field model recovers fluid lag and M-vertex solutions","Dual-pressure phase-field captures viscosity-dominated hydraulic fracs","Variational dual-continuum model stably handles fluid lag in fracking"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"Fracture flow is switched on only when the phase-field drops below a free numerical threshold, and outside that zone the two pressures are forced equal; the paper does not prove that mass balance is independent of that threshold or of mesh size.","fun_headline_variants_meta":{"raw":{"variants":["Dual-continuum phase-field resolves viscous fractures and fluid lag","Phase-field separates crack and pore pressures in viscosity regime","First phase-field model recovers fluid lag and M-vertex solutions","Dual-pressure phase-field captures viscosity-dominated hydraulic fracs","Variational dual-continuum model stably handles fluid lag in fracking"]},"model":"grok-4.5","effort":"low","cost_usd":0.004086,"raw_usage":{"total_tokens":1333,"prompt_tokens":882,"num_sources_used":0,"completion_tokens":74,"cost_in_usd_ticks":40860000,"prompt_tokens_details":{"text_tokens":882,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":377,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":882,"tokens_out":74,"duration_ms":3128,"temperature":1.0,"reasoning_tokens":377,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-12T00:01:25.248345+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"A mesh-refinement and threshold-sensitivity study of the lag-zone length in the early-time KGD problem that either shows the computed fluid-front location converging to the analytical ξ f values of Garagash or demonstrates a persistent, threshold-dependent under-estimate of the lag.","supporting_citations":[],"review_version":1}