{"id":"749a4168-f043-409a-9123-01e12df150b0","arxiv_id":"2505.10899","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The paper proposes a new LB discretization of a profile-corrected, level-set-enhanced Cahn-Hilliard equation and reports improved local volume conservation in four two-phase test cases.","lead":"The paper presents a multiple-relaxation-time lattice Boltzmann model for a modified Cahn-Hilliard equation that combines profile correction with a level-set distance function. Four two-phase benchmarks suggest the model keeps droplet volumes and interface shapes better than the standard Cahn-Hilliard LB model.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"LB recovery of the modified C-H equation is asserted, not derived: Eqs. (30)-(32) lack the multiscale analysis needed to exclude extraneous terms, so the volume-conservation claim is not yet established.","rationale":"The reader's conditional verdict is grounded in the missing recovery derivation, and my stress-test identifies the same load-bearing concern. The paper's novel contribution is the LB discretization of Eq. (17); if the LB evolution does not actually reproduce Eq. (17), the favorable volume-conservation results in Section 4 may reflect properties of the discretization rather than the modified C-H model. The absence of a Chapman-Enskog analysis is particularly significant because the intro itself documents prior LB-C-H models that failed to recover their target equations due to exactly such extraneous terms. A second, related fragility is the unquantified truncation in Eq. (10), but it is subordinate to the recovery question because the LB scheme is the paper's claimed contribution. The paper has no machine-checked proofs or released code, and the numerical evidence is qualitative, so the missing multiscale analysis is not compensated by independent verification. My proposed Chapman-Enskog test would settle the matter directly; if the expansion yields exactly Eq. (17), the conditional acceptance becomes well-founded, whereas if extraneous terms appear, the central claim weakens substantially. Therefore I do not change the reader's verdict.","tokens_in":14269,"tokens_out":12192,"duration_ms":127562,"concrete_test":"Perform a second-order Chapman-Enskog expansion of Eqs. (30)-(33), keeping every Δt and O(Kn²) term, and compare the resulting PDE with Eq. (17). Specifically, verify that the λ-contribution from equilibrium (31) plus forcing (32) amounts exactly to λM[∇²φ−∇·((1/W)[1−tanh²(2ψ/W)]∇ψ/|∇ψ|)] with no residual terms such as Δt(τ_g−1/2)λ∇·(φu) or λM∇·(∂_t(φu)/c_s²). If any extra term survives, the LB model must be amended with a correction force before the volume-conservation comparison in §4 can support the paper's central claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central assertion is that the MRT phase-field LB system (30)-(33) recovers Eq. (17) macroscopically. The paper states this is 'inspired by' [31,37] but provides no Chapman-Enskog or multiscale expansion. This is not a formality: the RHS of Eq. (17) contains a divergence-form source, λM[∇²φ−∇·((1/W)[1−tanh²(2ψ/W)]∇ψ/|∇ψ|)], whereas the new forcing term in Eq. (32) is a vector field inserted through ω_i c_i·[...]. The equivalence between a divergence-form source and such a forcing is exactly what must be demonstrated: the chemical potential in Eq. (31) is shifted by λφ, so the recovered diffusion contains ∇·(M∇(ℏ+λφ)); the forcing must cancel the unwanted part and reproduce the flux term to all required orders. Any residual term of order Δt, e.g. from the discrete ∂_t(φu) treatment or from the ψ-based flux, changes the equation actually simulated and can alter the local-volume behavior. The paper's own introduction recounts how He et al. and Lee et al. failed to recover the intended C-H equation for exactly this reason, so the omission is material. Secondary, Eq. (10) drops the first term of δ εp/δφ in Eq. (9) without estimating its size; when the interface profile deviates from the equilibrium tanh, this term is not zero and modifies the volume-preserving mechanism.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes a modified Cahn-Hilliard (C-H) equation for incompressible two-phase flows that combines a profile-correction penalty term with the signed-distance function from level-set methods, and then constructs a multiple-relaxation-time lattice Boltzmann (MRT-LB) model to solve it together with the incompressible Navier-Stokes equations. The central claim is that this LB model preserves the local volume of each phase and the interface shape better than the classical-C-H LB model of Liang et al. [31], especially for small droplets. The paper reports four numerical tests (stationary droplets, single vortex, Rayleigh-Plateau instability, and droplet deformation in a shear flow) and concludes that the proposed model exhibits superior volume conservation and interface fidelity.","tokens_in":14605,"tokens_out":9632,"duration_ms":96519,"significance":"The paper addresses a relevant and well-documented limitation of phase-field LB methods: local phase-volume loss for small droplets. The construction in Sec. 2, based on an existing penalty-energy idea [27,50] and the level-set reformulation [17], is clearly presented, and the chosen benchmarks are standard and appropriate. If the MRT-LB system in Sec. 3 is rigorously shown to reproduce Eq. (17), the model could be a useful contribution to interface-capturing LB methods. However, the manuscript currently leaves the load-bearing multiscale derivation as an assertion, and the numerical evidence is largely qualitative. The interface-shape preservation is partly by construction because the penalty term in Eq. (17) explicitly targets the hyperbolic-tangent profile (4); the decisive open question is whether the discrete LB equations actually realize Eq. (17) accurately. These gaps are addressable, but they are central rather than cosmetic.","major_comments":[{"comment":"The central assertion that the LB system recovers the modified C-H equation (17) is not derived. The text states only that the evolution equation is 'inspired by' Refs. [31,37]. This is load-bearing because Eq. (31) shifts the chemical potential by lambda*phi and Eq. (32) inserts the psi-dependent flux as a forcing term; one must show, via a Chapman-Enskog or multiscale expansion, that the discrete moments produce exactly div(M*grad(hbar)) + lambda*M*[Laplacian(phi) - div((1/W)(1 - tanh^2(2*psi/W))*grad(psi)/|grad(psi)|)] and produce no extraneous terms of order Delta t. In particular, the recovered diffusion of the shifted potential includes div(M*grad(lambda*phi)), and the force in Eq. (32) must cancel the unwanted part and reproduce the intended flux to all required orders. The time-derivative term dt(phi*u) in Eq. (32), approximated by the backward difference in Eq. (38), is a specific source of such discretization errors. The introduction itself (Sec. 1, Refs. [28-30]) documents earlier LB-C-H models that failed to recover the intended equation for exactly this class of reason, so the derivation is not a formality. Without it, the claim that the simulations solve Eq. (17) is unverified.","section":"Section 3.2, Eqs. (30)-(32)"},{"comment":"The derivation of Eq. (17) drops the first term on the right-hand side of Eq. (9), with the justification that 4*phi*(1-phi)/W - |grad(phi)| tends to zero as the interface approaches the hyperbolic-tangent profile. No error bound, asymptotic estimate, or numerical measurement of this term is given. In the deformed-interface tests of Secs. 4.2-4.4 the interface is not at the equilibrium profile, so the dropped term is not obviously negligible; it contributes to the effective penalty and could modify the local-volume behavior. The authors should either bound this term in the regimes simulated or explicitly state that Eq. (17) is the model to be solved, independent of the gradient-flow derivation from Eq. (8).","section":"Section 2, Eqs. (9)-(10)"},{"comment":"The quantitative evidence for the central claim is insufficient. Only Fig. 3(b) in Sec. 4.1 provides volume-evolution curves; the single-vortex (Sec. 4.2), Rayleigh-Plateau (Sec. 4.3), and shear-flow (Sec. 4.4) comparisons consist of qualitative interface snapshots. The paper reports no relative volume-change errors, no grid-convergence study, and no comparison with a well-resolved finite-difference solution of the continuum Eq. (17). A convergence test (e.g., the single-vortex case at time T) and time series of the relative volume error for each case are needed to substantiate the statement that the model 'achieves more precise volume conservation for each phase' (Abstract). As written, the superiority claim is only visually supported for three of the four benchmarks.","section":"Section 4"}],"minor_comments":[{"comment":"The radii r1 and r2 in Eq. (41) are not defined; please specify their values.","section":"Section 4.1, Eq. (41)"},{"comment":"The quantity 'local volume' is never defined precisely; please state whether it is the integral of phi over the domain, the area enclosed by the phi = 0.5 contour, or another measure, since the main quantitative claim is expressed through this quantity.","section":"General"},{"comment":"The density rho used in Eqs. (20), (21), (28), and (29) is never expressed as a function of the order parameter phi; although all tests use density ratio 1, the method is presented for general incompressible multiphase flows, so the interpolation formula (e.g., rho = rho_l*phi + rho_g*(1-phi)) and the way rho is updated should be stated.","section":"Section 3.1"},{"comment":"The regularization in Eq. (37) and the clipping of phi to [0,1] before computing psi change the signed-distance function; the effect of these choices on the forcing term (32), especially near the bulk regions, should be discussed or tested.","section":"Section 3.2, Eqs. (36)-(37)"},{"comment":"The boundary conditions (periodic, solid walls, moving walls) are not stated for any of the four simulations, which hampers reproducibility.","section":"General"},{"comment":"The text contains numerous typos, including 'surface tansion', 'casued', 'presssion', 'conservaed', 'latice spacing', and the repeated phrase 'more consistent representation ... more consistently' in the abstract; a careful proofread is needed.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of Physical Review E. The key risk is the missing multiscale derivation for the LB recovery of Eq. (17); if the authors supply a rigorous derivation and add quantitative volume-error and convergence studies, the manuscript would be considerably strengthened. I did not find evidence of deliberate misrepresentation; the gaps are omissions rather than demonstrable errors."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a new multiple-relaxation-time lattice Boltzmann discretization of the profile-corrected Cahn-Hilliard equation that Li and Yang introduced in 2024. The paper's selling point is that this LB model preserves the local volume of small droplets much better than the classical-C-H LB model of Liang et al. The numerical pictures support that claim qualitatively, but the paper never shows that the discrete scheme actually converges to the target equation (17) in the hydrodynamic limit. That missing Chapman-Enskog expansion is not a formality; the forcing term (32) is a vector field while the intended source is divergence-form, and as the introduction recounts, He et al. and Lee et al. failed for exactly that reason. I would not trust the claimed improvement until this is worked out.\n\nWhat is genuinely new: the MRT-LB scheme with the λ-shifted equilibrium and the ψ-based forcing. That is a legitimate new numerical scheme for an existing PDE, and the choice of test cases (stationary droplets, single vortex, Rayleigh-Plateau, shear) covers the relevant failure modes. The comparison with Ref. [31] is appropriate, and the observed interface behavior matches what the modified C-H equation is designed to do.\n\nSoft spots, in order of severity. First, the lack of any multiscale derivation for the phase-field LB equation. The sentence 'Inspired by previous work...' is not a derivation. The reader should ask the authors to supply the standard Chapman-Enskog or asymptotic analysis, or at least show numerically that a known exact solution is recovered to a specified order. Second, Eq. (10) drops the first term of the penalty variation without estimating its size; that term is not zero when the interface deviates from the equilibrium tanh profile, and the whole volume-preservation argument depends partly on that approximation. Third, the numerical evidence is qualitative for three of the four cases; only the stationary-droplet case has volume-loss curves. No grid-convergence study, no error norms, no quantitative long-time volume error for the vortex, Rayleigh-Plateau, or shear cases. The figures are suggestive, but the claim is stated much stronger than the data.\n\nThe citation pattern is fine: the relevant LB-C-H history is cited, and the PDE source [50] is clearly identified. The paper is not a derivative duplicate; it is an application-oriented extension.\n\nWho is this for? Researchers in lattice Boltzmann multiphase flow who care about the droplet-shrinkage problem. The paper deserves serious referee time, but the referee should push hard on the missing derivation and request quantitative checks. If those hold, it becomes a solid contribution.","headline":"A plausible new LB scheme for a profile-corrected Cahn-Hilliard model, but the key recovery claim is asserted rather than derived and the numerical support is only qualitative.","tokens_in":15136,"tokens_out":3547,"would_cite":false,"duration_ms":35317,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A modified Cahn-Hilliard equation with a signed-distance profile correction keeps each phase's volume nearly constant in two-phase flow simulations, where the classical equation lets small droplets shrink.","keywords":["lattice Boltzmann method","Cahn-Hilliard equation","local volume conservation","phase-field method","level-set signed distance function","interface capturing","multiphase flow","multiple-relaxation-time"],"falsifier":"A decisive check is a formal multiscale (Chapman-Enskog-style) expansion of the phase-field distribution update, Eqs. (30)-(32): if the recovered macroscopic equation contains terms beyond Eq. (17), or misses the level-set flux, the volume-conservation advantage would be an artifact of the discrete method. A purely numerical falsifier is a grid-refinement study of the two-droplet test in Section 4.1: if the scheme truly solves Eq. (17), the per-droplet volume error should shrink toward zero as the lattice spacing and time step are reduced at fixed physical parameters; if the error instead saturates at a level set by the dropped term in Eq. (10), the correction is not converging to the stated continuum target.","tokens_in":14052,"feed_emoji":"💧","tokens_out":10508,"duration_ms":90874,"temperature":0.7,"pith_summary":"The paper addresses a known failure of the Cahn-Hilliard phase-field equation: although it conserves the total mass of the binary mixture, each individual phase can lose volume because the interface drifts from its equilibrium profile and small droplets shrink. To fix this, it adopts a modified Cahn-Hilliard equation that adds a penalty term pulling the interface back to a hyperbolic-tangent profile, and rewrites the sharpening flux using the signed distance function of a level-set formulation so the flux stays smooth. The paper then constructs a multiple-relaxation-time lattice Boltzmann model whose collision, equilibrium, and forcing terms are chosen to recover this modified equation. In numerical tests, including stationary droplets, a single vortex, Rayleigh-Plateau breakup, and droplet deformation in shear, the new model keeps each droplet's volume nearly constant and preserves interface shape more faithfully than the classical-Cahn-Hilliard lattice Boltzmann model, with the largest gains for small droplets.","feed_headline":"Modified interface equation keeps phase volumes nearly constant","feed_subtitle":"A lattice Boltzmann solver built on the new equation beats the classical model, especially for small droplets.","key_machinery":"The central object is the modified Cahn-Hilliard equation (Eq. (17)) and the penalty flux carried by the signed distance function $\\psi$. The penalty energy of Eq. (6) pushes the interface toward the hyperbolic-tangent profile $|\\nabla\\phi| = 4\\phi(1-\\phi)/W$; truncating its variation (Eq. (10)) and substituting $\\phi(1-\\phi) = \\frac{1}{4}[1-\\tanh^2(2\\psi/W)]$ along with $\\nabla\\phi/|\\nabla\\phi|=\\nabla\\psi/|\\nabla\\psi|$ yields the smooth level-set-based flux that performs the volume-conservation work. The numerical engine is the multiple-relaxation-time lattice Boltzmann update (30)-(32), with equilibrium (31) and forcing (32) chosen so that the scheme is intended to recover Eq. (17) in the continuum limit.","core_discovery":"The central claim is that local volume conservation follows from correcting the interface profile, not from adding a global mass-correction term. The paper's target equation, Eq. (17), is the classical Cahn-Hilliard equation plus a penalty term $\\lambda M[\\nabla^2\\phi - \\nabla\\cdot(W^{-1}(1-\\tanh^2(2\\psi/W))\\nabla\\psi/|\\nabla\\psi|)]$, where $\\psi$ is the signed distance function tied to $\\phi$ by $\\psi = (W/4)\\ln(\\phi/(1-\\phi))$. With this term, the equilibrium profile is enforced as a soft constraint, and the level-set substitution removes the nonlinear sharpening singularity that affects earlier profile-correction formulations. The lattice Boltzmann model of Section 3 is designed so that its macroscopic limit is this modified equation; the simulation section compares it against the classical-Cahn-Hilliard lattice Boltzmann model in four benchmarks and reports consistently smaller volume drift and better interface morphology, most visibly for droplets below the critical radius where the classical model exhibits Ostwald-ripening-like shrinkage.","pith_inferences":["The penalty term acts as a soft reinitializer: it pushes the phase field toward the tanh profile that the level-set signed-distance function encodes. Viewed this way, the method is a phase-field cousin of conservative level-set reinitialization, and one could test whether explicit reinitialization every few steps at the same cost recovers the same volume-conservation curves.","Because the penalty strength $\\lambda$ is fixed at 0.5 in all tests, the reported gains have not been shown at the optimal penalty strength; a sweep over $\\lambda$ at fixed interface thickness, mobility, and grid spacing would reveal how much of the improvement comes from the correction term itself and how much from level-set smoothing.","The formulation is limited to two phases; for three or more phases, one signed-distance field per phase and a triple-junction rule for the penalty fluxes would be needed, so the mechanism does not transfer automatically."],"forward_implications":["Long-time simulations of droplet breakup and coalescence can track small satellite droplets without the artificial shrinkage that the classical Cahn-Hilliard model imposes below a critical radius.","Because the signed-distance flux remains smooth where the original profile-correction flux has $\\phi(1-\\phi)$ gradients, the interface-capturing scheme can run at thinner interfaces without the precision loss caused by sharpening-flux jumps.","The two-distribution-function architecture is unchanged, so incompressible multiphase lattice Boltzmann solvers can adopt the modified interface equation by replacing only the phase-field distribution and its forcing term.","In surface-tension-dominated regimes, per-phase volume conservation is maintained alongside global mass conservation, which is what allows the stationary-droplet, vortex, and shear benchmarks to report near-constant volumes."],"supporting_citations":[{"why":"Introduces the penalty energy whose gradient-flow variation becomes the profile-correction term in the model's target equation.","marker":"[27]"},{"why":"Provides the signed-distance reformulation that replaces the nonlinear sharpening flux with the smooth tanh expression.","marker":"[17]"},{"why":"Derives the local-volume-conservation-improved interface equation that Eq. (17) is built on, and supplies two of the benchmark configurations.","marker":"[50]"},{"why":"Is the classical-Cahn-Hilliard multiple-relaxation-time lattice Boltzmann model used as the comparison baseline in every numerical test.","marker":"[31]"},{"why":"Provides the interface-capturing lattice Boltzmann equilibrium, forcing, and time-difference discretization that Eqs. (30)-(32) follow.","marker":"[37]"},{"why":"Establishes spontaneous shrinkage of drops below a critical radius in phase-field simulations, the small-droplet failure the tests target.","marker":"[49]"},{"why":"Supplies the shear-flow droplet-deformation benchmark and the equilibrium-solution analysis the test relies on.","marker":"[60]"},{"why":"Documents the interface-thickening and volume-drift failure of classical Cahn-Hilliard capture that motivates the correction.","marker":"[12]"}],"fun_headline_variants":["Modified Cahn-Hilliard locks in phase volumes for LB flow","Lattice Boltzmann with corrected interface keeps droplets from shrinking","Profile-corrected Cahn-Hilliard halts volume loss in multiphase LB","New LB model conserves phase volumes locally via level-set fix"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim rests on the assumption that the lattice Boltzmann update of Section 3, with equilibrium (31) and forcing (32), truly reproduces the modified Cahn-Hilliard equation (17) in the continuum limit, including the new level-set-based source term, and that dropping the first piece of the penalty variation in Eq. (10) is harmless; the paper states both but does not prove the recovery.","fun_headline_variants_meta":{"raw":{"variants":["Modified Cahn-Hilliard locks in phase volumes for LB flow","Lattice Boltzmann with corrected interface keeps droplets from shrinking","Profile-corrected Cahn-Hilliard halts volume loss in multiphase LB","New LB model conserves phase volumes locally via level-set fix"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000245,"raw_usage":{"total_tokens":1559,"prompt_tokens":992,"completion_tokens":567,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":491}},"tokens_in":608,"tokens_out":567,"duration_ms":5807,"temperature":1.0,"reasoning_tokens":491,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:01:08.904352+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check is a formal multiscale (Chapman-Enskog-style) expansion of the phase-field distribution update, Eqs. (30)-(32): if the recovered macroscopic equation contains terms beyond Eq. (17), or misses the level-set flux, the volume-conservation advantage would be an artifact of the discrete method. A purely numerical falsifier is a grid-refinement study of the two-droplet test in Section 4.1: if the scheme truly solves Eq. (17), the per-droplet volume error should shrink toward zero as the lattice spacing and time step are reduced at fixed physical parameters; if the error instead saturates at a level set by the dropped term in Eq. (10), the correction is not converging to the stated continuum target.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the penalty energy whose gradient-flow variation becomes the profile-correction term in the model's target equation."},{"cited_title":"Jain, Accurate conservative phase-field method for simulation of two-phase flows, J","cited_arxiv_id":null,"evidence_quote":"Provides the signed-distance reformulation that replaces the nonlinear sharpening flux with the smooth tanh expression."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Derives the local-volume-conservation-improved interface equation that Eq. (17) is built on, and supplies two of the benchmark configurations."},{"cited_title":"Liang, B.C","cited_arxiv_id":null,"evidence_quote":"Is the classical-Cahn-Hilliard multiple-relaxation-time lattice Boltzmann model used as the comparison baseline in every numerical test."},{"cited_title":"Liang, R","cited_arxiv_id":null,"evidence_quote":"Provides the interface-capturing lattice Boltzmann equilibrium, forcing, and time-difference discretization that Eqs. (30)-(32) follow."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes spontaneous shrinkage of drops below a critical radius in phase-field simulations, the small-droplet failure the tests target."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the shear-flow droplet-deformation benchmark and the equilibrium-solution analysis the test relies on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the interface-thickening and volume-drift failure of classical Cahn-Hilliard capture that motivates the correction."}],"review_version":1}