{"id":"f44f2131-2af5-44de-b9e6-ccecb19d8bc3","arxiv_id":"2607.13996","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The RFQC real-fluid solver's liquid-upwind anomaly is a start-up failure caused by a jump in the affine slope xi and by thermodynamic re-projection canceling the pressure rise; smoothing the initial data fixes it.","lead":"Real-fluid CFD can fail when a high-pressure liquid is advected into a low-pressure vapor: the RFQC method leaves the initial break stuck and the downstream pressure too low. The authors identify the cause in the method's pressure-recovery step and show that smoothing the initial discontinuity restores the correct expansion wave.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central mechanism is derived only under fully upwind condition (u_L>c_L), yet the paper never states or tests the sonic threshold; LUA scope is therefore undefined, and Eq. 37 may not apply to subsonic liquid-upwind cases.","rationale":"Reader's weakest assumption already flags non-supersonic velocities; this pass sharpens it into a falsifiable scope condition. The paper's own numbers show every test is just past Mach 1, which strongly suggests the true threshold is u_L = c_L, yet the paper never states it. The central new contribution—identifying Y_b and Y_p as the cause—is conditioned on Eq. 30; establishing when Eq. 30 applies is therefore load-bearing. A single sweep across the liquid Mach number settles it. No other concern outweighs this: the algebra in Appendix A is coherent, the diagnostics in Fig. 8 match the decomposition for tested cases, and the empirical gamma limitation is acknowledged by the authors. The verdict should remain conditional pending this scope test; the paper should either define the LUA threshold as the liquid sonic point and restrict claims, or extend the mechanism to subsonic cases.","tokens_in":20140,"tokens_out":11189,"duration_ms":108345,"concrete_test":"Run the Fig. 4 flash-evaporation Riemann problem (n-dodecane, p_L=2 MPa, ρ_L=300 kg/m^3, p_R=0.1 MPa, ρ_R=2 kg/m^3, first-order HLLC, N=500, CFL=0.5, t=0.6 ms) for u_L = 0, 50, 100, 120, 130, 140, 150, 200 m/s, bracketing c_L≈130 m/s. For each, check for the residual discontinuity and non-monotonic density, and compute L1 error against the exact solution. Clean subsonic results would confirm the LUA threshold is the liquid sonic point and confine the Eq. 37 mechanism to u_L>c_L; subsonic anomalies would invalidate the fully-upwind derivation as the sole mechanism.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The pivotal step is Eq. 30, which assumes both cell interfaces reduce to one-sided upwind states. For this to hold in the liquid cell L, all acoustic characteristics must be right-going, i.e. u_L > c_L. The paper's only justification is 'c_L = 130 m/s; thus the cell L enters a fully upwind state under the condition of u_L = 150 m/s' (Sec. 5.1). All presented tests are supersonic: 1D cases u=150/300 vs c_L=130; 2D jets u=150 vs c=147 and u=300 vs c=295 (Mach 1.02–1.15). The abstract's 'exceeds a certain threshold' is never defined. Therefore the Y_b/Y_p feedback mechanism (Eqs. 46, 53) is established only for supersonic liquid-upwind starts. If LUA does not occur for u_L < c_L, the paper's scope is narrower than claimed and the threshold is essentially the sonic point; if LUA does occur subsonically, Eq. 30 fails and Eq. 37 cannot be the mechanism. Either way the central claim lacks a verified domain. The algebra of Eq. 37 itself is internally consistent and Fig. 8 diagnostics support the decomposition for the tested supersonic cases, but that does not resolve the scope gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper revisits pressure oscillations in finite-volume methods for real fluids and interprets the Real Fluid Quasi-Conservative (RFQC) method as recovering a mechanical-equilibrium pressure by evolving the affine coefficients (ξ,E0) of the internal-energy–pressure relation along pathlines, with thermodynamic re-projection absorbing the deviation from the isentropic trajectory. The authors then identify a Liquid-Upwind Anomaly (LUA) in phase-change Riemann problems when a liquid-upwind velocity is superimposed on a high-pressure-liquid/low-pressure-vapor initial discontinuity. They derive a single-step pressure-update equation under a fully upwind assumption (Eq. 37), decompose it into four terms, and argue that the orders-of-magnitude jump in ξ during phase change makes the flux-induced term Y_b deficient, while thermodynamic re-projection removes the positive increment Y_p, producing a feedback loop that delays pressure rise in the downstream cell. A tanh-based initial-condition regularization is proposed and validated on 1D Riemann problems and 2D sonic phase-change jets.","tokens_in":20537,"tokens_out":12951,"duration_ms":129670,"significance":"If the claims hold, the paper offers a concrete physical/numerical mechanism for a previously unreported failure mode of RFQC and provides a practical remedy. The algebraic derivation in Appendix A is detailed and self-consistent, and the numerical diagnostics in Figs. 8–10 and 13 substantiate the decomposition for the tested cases. The paper's explicit comparison with the double-flux method and its use of exact Riemann solutions strengthen the numerical evidence. The authors are also candid about the empirical nature of the regularization. However, the central mechanism is derived under a restrictive fully upwind assumption whose domain is not established, and some of the conceptual claims in Section 3 are partly restatements of the method's construction rather than independent physical derivations. These issues do not invalidate the reported numerical results, but they limit the generality of the paper's central claim.","major_comments":[{"comment":"The derivation of the pressure-increment equation assumes that both interfaces of cell R satisfy the fully upwind one-sided flux condition Eq. (30). For the liquid cell L this requires right-going acoustic characteristics, u_L > c_L. The only justification given is 'c_L = 130 m/s; thus the cell L enters a fully upwind state under the condition of u_L = 150 m/s' (§5.1), and all presented test cases are sonic/supersonic (u_L/c_L ≈ 1.02–1.15 in §4 and §7). The abstract's 'exceeds a certain threshold' is never defined. If LUA also occurs for u_L < c_L, then Eq. (30) is violated and Eq. (37) cannot be the operative mechanism; if it does not, the paper's scope is narrower than claimed. Please define the threshold and provide tests for subsonic liquid-upwind starts, including a case with u_L < c_L.","section":"§5.1, Eq. (30)"},{"comment":"The Φ update is written as \\tilde ξ^+ = (1-b)ξ + bξ_L and \\tilde E0^+ = (1-b)E0 + bE0,L, with b=θu_L. In contrast, the mass and energy updates use a=θu for the outflow weight (Eqs. 33, 62, 65). For the advection equation (25), a standard first-order upwind discretization with right-going flow would give (1-a)ξ + aξ_L. The choice of b in Eqs. (35)–(36) is not justified and directly enters Eq. (37), the decomposition (38)–(42), and the subsequent Y_b/Y_p feedback argument. Please state the exact discrete form of Eq. (25) used by RFQC; if the scheme genuinely uses u_L for the outflow coefficient, explain the derivation; if not, correct the analysis.","section":"§5.1, Eqs. (35)–(36)"},{"comment":"The 'mechanical-equilibrium recovery' claim is partly a restatement of the construction: p is defined by (ρe−E0)/ξ in Eq. (3), and Eq. (12) simply extracts a common pressure from the affine relation under an assumed pressure-equilibrium average. The paper's physical interpretation of ξ and E0 along pathlines is useful, but it should clearly distinguish (i) the definition of the algorithm, (ii) the assumption that pressure equilibrium holds inside a cell, and (iii) the empirical/predictive content that justifies the method for real fluids. As written, Section 3 may overstate the explanatory power of the derivation.","section":"§3, Eqs. (3), (12)–(13)"}],"minor_comments":[{"comment":"The notation is ambiguous: F_{L+1/2} and F_{R-1/2} denote the same interface in Fig. 7, yet Eq. (30) writes them as F_L and F_R respectively. Clarify whether the intended statement is that the left interface of cell R has flux F_L and the right interface of cell R has flux F_R.","section":"Eq. (30)"},{"comment":"The phrase 'exceeds a certain threshold' is never quantified in the body. Please either define the threshold or rephrase to match the actual condition used in §5.1.","section":"Abstract"},{"comment":"The regularization introduces γ and N_s as free parameters, but no sensitivity study is reported. The paper later states that the remedy is technical rather than fundamental; please add a brief sensitivity discussion or explicitly state that no predictive criterion is established.","section":"§6"},{"comment":"The theoretical analysis assumes first-order spatio-temporal discretization, while Section 7 uses third-order WENO, second-order MUSCL, and SSP-RK3. The text acknowledges that high-order analysis is difficult; this limitation should be stated more prominently so the reader does not infer that the Eq. (37) mechanism directly proves the 2D high-order results.","section":"§5, §7"},{"comment":"Typo: 'Eo' should be 'E0' in the discussion of reconstructed variables.","section":"§6, last paragraph"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the unverified domain of the fully upwind assumption in §5.1. I would ask for at least one subsonic liquid-upwind test and a clarification/correction of the Φ advection update in Eqs. (35)–(36), since the central pressure-increment equation depends on that discrete choice. If those points are resolved, the paper could be a solid contribution; the empirical nature of the regularization should be acknowledged in the final version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does one genuinely useful thing: it identifies a reproducible failure mode of its own RFQC scheme, works out a plausible mechanism for it, and gives a practical fix that restores correct behavior in the tested cases. That is worth a serious referee.\n\nThe new material is the Liquid-upwind Anomaly and its explanation. The derivation of the single-step pressure-increment equation is careful, and the decomposition into the flux term, velocity-work term, kinetic mixing, and pressure impulse is checked against the numerics. The feedback loop involving the thermodynamic re-projection error is clearly argued and supported by the diagnostics in Figures 8–10 and 13. The regularization remedy is simple and appears to work for the Riemann problems and the two sonic jet cases.\n\nThe soft spots are real but not fatal. The most important one is the scope gap: the mechanism is derived under the fully upwind assumption, which for the liquid cell requires u_L > c_L. All the test cases are just barely supersonic (Mach 1.02–1.15), and the abstract's mention of a \"certain threshold\" is never pinned down. Either LUA also occurs subsonically, in which case Eq. 30 is not the right basis for the mechanism, or it does not, in which case the paper overstates its domain. That needs to be resolved before the claim is fully solid. The remedy's gamma is empirical, no code or data are shipped, and the exact-solution benchmark is the authors' own previous work; none of these are disqualifying, but they are the standard requests for a revision.\n\nI disagree with the harshest reading that the mechanical-equilibrium interpretation is empty. It is largely a reformulation, but it is a coherent one and it clarifies why the advection of the affine coefficients works. The analysis is honest; the paper explicitly admits the regularization is technical rather than fundamental. The math in the appendix is self-consistent, and the citations to the broader literature look appropriate. The in-text limitation statements are consistent with what I found in the derivations.\n\nFor peer review: yes, send it. Ask the authors to define the upwind threshold, test a subsonic case or argue why the mechanism extends, and ideally release the code and exact-solution data. This is publishable after revision, and it will be useful to people building real-fluid schemes.","headline":"The LUA finding is real and the mechanism is plausible, but the claimed scope rests on a fully upwind condition that the paper never defines or tests outside a narrow supersonic window.","tokens_in":20951,"tokens_out":1566,"would_cite":true,"duration_ms":19155,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76M12","76N15","80A22"],"pacs":[],"model":"deepseek-v4-flash","headline":"A real-fluid finite-volume scheme fails on liquid-upwind phase change because its thermodynamic re-projection removes the pressure rise a downstream cell needs, stalling the rarefaction wave.","keywords":["pressure oscillation","real fluid","phase change","quasi-conservative method","liquid-upwind anomaly","Riemann problem","thermodynamic re-projection","finite volume method"],"falsifier":"Add the removed re-projection increment Yp back into the pressure update, or disable the re-projection step, for the n-dodecane flash-evaporation case with u = 150 m/s and check whether the downstream cell's pressure rises to about 1.9 MPa and the rarefaction fan develops; if it does not, the Yb/Yp feedback mechanism is not the cause. Alternatively, run the same case with an equation of state for which xi is nearly constant across phase change and see whether the anomaly disappears.","tokens_in":20074,"feed_emoji":"💧","tokens_out":4445,"duration_ms":41394,"temperature":0.7,"pith_summary":"The paper argues that pressure oscillations in real-fluid finite-volume methods reflect a physical choice of equilibrium assumption: recovering pressure from averaged conservative variables assumes thermodynamic equilibrium, which real fluids violate, whereas the Real Fluid Quasi-Conservative (RFQC) method recovers pressure from a mechanical-equilibrium affine relation between internal energy and pressure. The paper then identifies and explains a failure mode: in liquid-upwind phase-change Riemann problems, once the downstream cell enters the two-phase region, the affine slope jumps by orders of magnitude, so the flux-driven pressure increment becomes tiny, and the method's thermodynamic re-projection removes an additional positive increment, trapping the cell in a delayed low-pressure cycle. The paper proposes regularizing the initial discontinuity over a few cells, which restores the correct rarefaction fan and makes the method work for sonic phase-change jets. A sympathetic reader would care because it clarifies what the method is physically doing and gives a practical, if empirical, remedy for a reproducible failure.","feed_headline":"Re-projection starves pressure rise in liquid-upwind phase change","feed_subtitle":"A quasi-conservative scheme's own thermodynamic re-projection removes the increment its downstream cell needs; a few-cell smoothing fixes it","key_machinery":"The central object is the isentropic affine relation rho e = xi p + E0, with xi = h/c^2 the isentropic slope and E0 the zero-pressure internal-energy intercept. The RFQC method advects (xi, E0) along pathlines instead of solving their isentropic source-term equations, reconstructs pressure as p = (rho e - E0)/xi, and re-projects the thermodynamic state each step; the re-projection converts the off-isentropic drift into an internal-energy error. The analysis of the anomaly is carried by a single-step pressure-increment equation (Eq. 37), whose terms Yb, Ys, YK, YH, plus the re-projection loss Yp, show that Yb is small and Yp removes the needed positive increment.","core_discovery":"The central claim is that the Liquid-upwind Anomaly is not a random instability but a start-up singularity caused by the affine slope jump xi_R >> xi_L at a fully upwind phase-change interface. Under fully upwind conditions, the dominant positive pressure-increment term Yb = b xi_L (p_L - p) / xi_tilde^+ shrinks because the denominator is dominated by the large two-phase xi, and the thermodynamic re-projection, which is supposed to keep the method thermodynamically consistent, removes a positive increment Yp = -(1-a) epsilon_p / xi_tilde^+. With both contributions suppressed, the downstream cell's pressure rises far more slowly than the exact rarefaction solution requires, and the persistent","pith_inferences":["The paper leaves implicit that any quasi-conservative scheme that advects affine thermodynamic parameters and re-projects may show a similar failure; the Yb/Yp decomposition could generalize to other real-fluid methods.","The tanh regularization is empirical (gamma = 0.2 to 1); a predictive threshold criterion based on the xi ratio and upwind Mach number could be derived and tested, turning the anomaly into a design rule.","The paper's physical interpretation suggests re-projection error should be a controlled energy residual; a more fundamental fix might limit or redistribute the re-projection correction where isentrope curvature is extreme rather than smoothing initial data.","Because the speed-of-sound drop in the two-phase region is physical, the same xi jump could affect any homogeneous-equilibrium-model scheme using a similar affine closure, so the LUA may appear in cavitation or flashing flows beyond n-dodecane."],"forward_implications":["If correct, RFQC's pressure recovery is mechanical-equilibrium recovery, and pressure oscillations in real fluids are a conflict between thermodynamic-equilibrium averaging and mechanical equilibrium, not an energy-equation closure issue.","The LUA is a start-up anomaly tied to initial discontinuities, so practical real-fluid simulations with liquid-upwind phase-change starts should regularize initial data rather than modify the flux or time integrator.","The regularized RFQC method reproduces the exact flash-evaporation Riemann solution and restores first-order convergence rates consistent with a Godunov method.","The method remains stable and accurate for two-dimensional sonic phase-change jets, whereas the double-flux method shows severe oscillations even at first order.","Because the mechanism depends on xi_R >> xi_L and full upwind, phase-change flows with lower advection speeds or supersonic liquid states may not exhibit LUA, explaining the observed threshold behavior."],"fun_headline_variants":["Re-projection starves pressure rise in liquid-vapor phase change","Affine slope jump triggers start-up anomaly in quasi-conservative flow","Scheme's own fix removes pressure bump upwind cells need","Liquid-upwind trap: how re-projection delays pressure recovery","Pressure starvation at phase-change fronts: a start-up effect"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The mechanism analysis assumes that both interfaces of the anomalous cell are fully upwind, so the upstream liquid cell stays constant and the downstream cell is updated only from its own state and the liquid state; if this degeneracy, or the closure q = xi p + E0 after re-projection, fails for other equations of state, higher-order reconstruction, or non-supersonic liquid velocities, the derived pressure-increment equation need not describe the failure.","fun_headline_variants_meta":{"raw":{"variants":["Re-projection starves pressure rise in liquid-vapor phase change","Affine slope jump triggers start-up anomaly in quasi-conservative flow","Scheme's own fix removes pressure bump upwind cells need","Liquid-upwind trap: how re-projection delays pressure recovery","Pressure starvation at phase-change fronts: a start-up effect"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000287,"raw_usage":{"total_tokens":1566,"prompt_tokens":833,"completion_tokens":733,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":577,"completion_tokens_details":{"reasoning_tokens":645}},"tokens_in":577,"tokens_out":733,"duration_ms":7217,"temperature":1.0,"reasoning_tokens":645,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T03:04:31.517844+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Add the removed re-projection increment Yp back into the pressure update, or disable the re-projection step, for the n-dodecane flash-evaporation case with u = 150 m/s and check whether the downstream cell's pressure rises to about 1.9 MPa and the rarefaction fan develops; if it does not, the Yb/Yp feedback mechanism is not the cause. Alternatively, run the same case with an equation of state for which xi is nearly constant across phase change and see whether the anomaly disappears.","supporting_citations":[],"review_version":1}