{"id":"363fa754-32c0-4df3-a132-03be0ffd03be","arxiv_id":"2607.10164","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"A first 3D integral surface-tension discretisation for sharp front tracking conserves momentum, handles variable σ, and outperforms CSF and smoothed FT on several two-phase benchmarks.","lead":"The authors present the first three-dimensional integral surface-tension scheme for sharp front-tracking, computing force from interface–cell intersections with an explicit Laplace pressure correction. It matches CSF on spurious currents while improving accuracy for low-Oh oscillations, Marangoni migration, and strongly deforming rising bubbles.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified beyond the reader's already-flagged geometric-reconstruction dependence.","rationale":"The reader's weakest_assumption correctly identifies the single point on which the sharp force balance and low-Oh stability rest. All other potential concerns (abstract wording of the thermocapillary gain, lack of formal accuracy theory, generic remeshing) are secondary and already reflected in the CONDITIONAL verdict. Because the paper supplies concrete external benchmarks that succeed under precisely those conditions, no stronger objection is warranted and the verdict needs no adjustment.","tokens_in":28162,"tokens_out":454,"duration_ms":5267,"concrete_test":"Re-run the Oh=0.005 oscillating-droplet case (Fig. 14b) and Case 3 rising bubble (Mo=1.31) while deliberately varying only the marker count in Algorithm 1 (e.g., 12 vs 48 markers) and the Gaussian weight scale; if both amplitude/frequency and skirt length remain within a few percent of the published curves, the geometric-reconstruction assumption is robust for the claimed regimes.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim holds under the paper's own evidence: the 3D integral scheme (Eqs. 6–13, 38; Algorithm 1) is fully specified, conserves momentum by construction, and demonstrably outperforms CSF and classic FT on the low-Oh oscillating droplet (Fig. 14b), thermocapillary terminal velocity (Table 1, Fig. 16), and low-Mo rising-bubble shapes/Re (Figs. 19–20, Table 3). The only load-bearing soft spot is exactly the one the reader already isolates—accuracy of the local quadratic fits that supply ϕ, nΣ and κ for both intersections and the pressure-correction β (Eqs. 12, 18). That dependence is real, but the multi-benchmark results (including second-order Ca_rms convergence and experimental shape agreement without specialised roughness smoothing) already supply empirical support that the fits remain adequate under the remeshing and deformations tested. No deeper internal inconsistency or missing control is required to accept the claim as stated.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript presents the first three-dimensional integral surface-tension scheme for front-tracking frameworks, extending the 2D formulation of Popinet & Zaleski. Surface tension is evaluated as a line integral of σ m_Σ at the intersections of the Lagrangian front with Eulerian cell faces; a pressure-jump correction β (derived from face fractions γ and local curvature) is added so that the discrete pressure gradient and force remain consistent. Intersections, normals and curvatures are obtained from local weighted quadratic surface fits (Algorithm 1). The scheme is implemented in a sharp FT/PPIC solver and benchmarked against CSF and classic (kernel-smoothed) FT on static/translating Laplace spheres, Lamb oscillations (Oh = 0.05 and 0.005), Young thermocapillary migration, and three Bhaga–Weber rising-bubble cases. Spurious currents are comparable to CSF; accuracy and stability are superior on low-Oh oscillations, variable-σ flows and strongly deforming bubbles, with terminal-velocity errors reduced by factors of ~3–6 and bubble shapes closer to experiment.","tokens_in":28466,"tokens_out":1134,"duration_ms":32688,"significance":"A genuine 3-D integral scheme that conserves momentum by construction and treats variable surface tension without extra terms fills a gap repeatedly noted in the 2-D literature. The multi-benchmark campaign (analytic Young–Laplace, Lamb and Young solutions plus independent Bhaga–Weber experiments), second-order Ca_rms convergence, and the ability to run low-Oh oscillations and low-Mo skirted bubbles without specialised roughness smoothing constitute strong empirical support. Open Zenodo data further raise the standard of reproducibility. If the geometric reconstruction remains robust under more extreme topologies, the method should become a standard option in sharp FT codes.","major_comments":[{"comment":"Section 2.2.1 / Algorithm 1 and Eqs. (12), (16)–(18): every geometric quantity required by the scheme (φ, n_Σ, κ, l_Σ, m_Σ and therefore β) is obtained from a single local weighted least-squares quadratic fit through a fixed number of markers (24–48). The multi-test results already supply empirical evidence that these fits remain adequate under the remeshing and deformations examined, yet a short sensitivity study to N_marker or the Gaussian weight scale would make the robustness claim fully quantitative and would address the only load-bearing modelling assumption that is not itself derived from the integral principle.","section":null},{"comment":"Section 3.1, paragraph after Eq. (38): the face-fraction switch γ ≽ 0.5 is replaced by the cell-centred volume-fraction switch α ≽ 0.5 “to ensure consistency of β across the three components”. While this is a pragmatic and volume-based choice, it is no longer identical to the geometric face fraction that appears in the derivation of Eqs. (9)–(12). A one-sentence quantification of the discrepancy (or a demonstration that it vanishes under grid refinement) would close the only remaining formal gap between the continuous integral statement and the discrete well-balanced source term.","section":null}],"minor_comments":[{"comment":"Abstract and §5: the phrase “up to an order of five” matches the factors ~3–6 reported in Table 1, but could be misread as five orders of magnitude; a parenthetical “(factor of ~5)” would remove any ambiguity.","section":null},{"comment":"Figure 7: colour-bar ranges for S*_x differ between the classic/CSF panels and the integral panel, which slightly obscures the visual comparison of sharpness; a common scale (or an inset) would help.","section":null},{"comment":"Eq. (54): the inertia-tensor proxy for the polar radius is elegant and removes remeshing noise; a brief remark on its accuracy for the chosen a0 = 0.025 would be useful for readers who wish to reuse it.","section":null},{"comment":"Section 4.5: AMR criteria (interface + vorticity) are mentioned but not quantified; a single sentence on the refinement threshold or the resulting cell count would aid reproducibility.","section":null},{"comment":"Occasional typographic inconsistencies (en-dashes in “two–phase”, missing spaces around some equation references, and the dual citation style for Saini et al.) should be cleaned in production.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The geometric-fit dependence flagged by the reader is real but already well-supported by the evidence; I do not regard it as grounds for major revision. The paper is a clean, well-executed contribution that fits JCP / IJNMF / JFM Computational Fluids scope. No citation or novelty concerns."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is the first working 3D integral surface-tension discretisation for front tracking. That is the real news. Popinet–Zaleski and the later 2D level-set/VOF extensions never crossed into three dimensions because of the intersection geometry; Gennari and van Wachem solve it with signed-distance quadratic fits on the face corners, a pressure-jump correction β, and a well-balanced face-centred source that slots into their existing sharp FT/PPIC solver.\n\nWhat they do well is the validation suite. Static and translating Laplace spheres give Ca_rms comparable to CSF and better than classic smoothed FT, with near-second-order convergence for the integral scheme. The low-Oh=0.005 oscillating droplet is the cleanest win: only the integral method stays stable and tracks Lamb’s frequency and damping without specialised roughness smoothing. Thermocapillary migration (Young et al.) and the Bhaga–Weber rising bubbles (especially the low-Mo skirted case) show clear quantitative and shape improvements over both CSF and classic FT. Momentum conservation is by construction, variable-σ needs no extra machinery, and the algorithm (including the quadratic fit) is fully specified. Data are released.\n\nSoft spots are real but limited. The abstract’s “five orders of magnitude” thermocapillary claim does not match Table 1 (roughly one order at best); the conclusions are more careful. Everything still rests on the local quadratic fits (24–48 markers, Gaussian weights) for ϕ, n and κ that feed both the intersections and β. That is the load-bearing geometric assumption. The paper does not supply a formal accuracy theory for those fits under remeshing and large deformation, and they deliberately avoid case-tuned roughness smoothing. Empirically the multi-benchmark results already push that assumption hard and it holds for the regimes they test; it is not a hidden flaw, just the place where the method can still break.\n\nThis is for people who actually run 3D DNS of surface-tension-dominated or Marangoni flows with front tracking. It deserves a serious referee. I would cite it when I next need a sharp, conservative 3D surface-tension option, and I would send it out for review without hesitation.","headline":"First solid 3D integral surface-tension scheme for front tracking; real gains on low-Oh and Marangoni cases, with one abstract overclaim and geometry dependence that the benchmarks already stress-test.","tokens_in":29049,"tokens_out":566,"would_cite":true,"duration_ms":7307,"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 first 3D integral surface-tension scheme for sharp front tracking matches CSF spurious currents and beats it on low-Oh oscillations, Marangoni migration, and deforming bubbles.","keywords":["two-phase flows","surface tension","integral formulation","front tracking","Marangoni flows","spurious currents","rising bubbles"],"falsifier":"Re-run the Oh = 0.005 oscillating-droplet case and the lowest-Morton rising-bubble case with deliberately coarser or noisier front meshes (or with the quadratic fit replaced by a lower-order one); if the scheme then loses stability or reverts to the large shape and velocity errors of the classical methods, the geometric premise fails.","tokens_in":29064,"feed_emoji":"💧","tokens_out":667,"duration_ms":8481,"temperature":0.7,"pith_summary":"Surface tension is usually discretised as a volumetric force that needs a smoothed delta function and special handling for variable coefficients. This paper shows that the older integral line-integral form can be made to work robustly in three dimensions inside a sharp front-tracking method. The force is applied only on the true intersections of the Lagrangian interface with the faces of each Eulerian cell, and a simple pressure correction enforces the Laplace jump across those faces. Because the scheme never spreads the force, it conserves momentum cell by cell and treats Marangoni stresses with no extra machinery. Across standard tests the method produces parasitic currents no larger than the continuous-surface-force approach, yet it is markedly more accurate for oscillating drops at low Ohnesorge number, for thermocapillary migration (terminal-velocity errors drop by orders of magnitude), and for rising bubbles that form thin skirts at low Morton number. A reader who needs reliable free-surface dynamics without mesh-dependent smoothing therefore has a new practical option.","feed_headline":"First 3D integral surface tension for sharp front tracking","feed_subtitle":"Matches CSF parasitic currents, cuts Marangoni and low-Oh errors by orders of magnitude","key_machinery":"The face-wise integral force (Eq. 6) together with the pressure-correction factor β (Eqs. 11–12) obtained from a quadratic surface fit (Algorithm 1) that supplies signed distance, normal and curvature on every cut face; the resulting staggered source is then made well-balanced for a collocated finite-volume solver.","core_discovery":"The first three-dimensional integral surface-tension scheme, realised inside a sharp front-tracking framework that reconstructs interface–cell intersections from local quadratic fits, conserves momentum locally and globally, needs no delta-function discretisation, and automatically handles variable surface tension. Relative to continuous-surface-force and classical smoothing methods it yields comparable spurious velocities while delivering substantially lower errors on low-Ohnesorge oscillations, thermocapillary terminal velocities, and the steady shapes of strongly deforming rising bubbles.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["First 3D integral surface tension for sharp front tracking","3D integral scheme brings momentum-conserving surface tension to front tracking","Integral surface tension extended to 3D complex interfaces","Front-tracking gains first 3D Popinet-Zaleski surface tension","3D integral method cuts Marangoni and low-Oh errors sharply"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"That ordinary quadratic fits through a few dozen nearby front markers always give signed distances, normals and curvatures accurate enough for both the geometric intersections and the Laplace pressure correction, even after remeshing and large three-dimensional deformation.","fun_headline_variants_meta":{"raw":{"variants":["First 3D integral surface tension for sharp front tracking","3D integral scheme brings momentum-conserving surface tension to front tracking","Integral surface tension extended to 3D complex interfaces","Front-tracking gains first 3D Popinet-Zaleski surface tension","3D integral method cuts Marangoni and low-Oh errors sharply"]},"model":"grok-4.5","effort":"low","cost_usd":0.00627,"raw_usage":{"total_tokens":1619,"prompt_tokens":813,"num_sources_used":0,"completion_tokens":74,"cost_in_usd_ticks":62700000,"prompt_tokens_details":{"text_tokens":813,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":732,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":813,"tokens_out":74,"duration_ms":5293,"temperature":1.0,"reasoning_tokens":732,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T13:48:08.660563+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Re-run the Oh = 0.005 oscillating-droplet case and the lowest-Morton rising-bubble case with deliberately coarser or noisier front meshes (or with the quadratic fit replaced by a lower-order one); if the scheme then loses stability or reverts to the large shape and velocity errors of the classical methods, the geometric premise fails.","supporting_citations":[],"review_version":1}