{"id":"d52041d1-77fa-479c-b31b-9ec64bce4745","arxiv_id":"2607.18608","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An isochoric (constant-volume) flash preconditioner resolves transient vaporization caused by abrupt fracture opening in porous-media flow simulations, a transient that conventional adaptive time stepping skips.","lead":"This paper introduces a numerical preconditioner that snaps the fluid state to the correct thermodynamic condition immediately after a fracture suddenly widens, so that transient steam formation is not skipped by large simulation time steps. The method lets both pressure-based and volume-based reservoir simulators capture short-lived vaporization events that ordinary adaptive time stepping misses.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unvalidated uv-flash free-expansion assumption (Eq. 3.9) sets every reported transient quantity; the isenthalpic alternative [44] is acknowledged but never tested, leaving the physical claim conditional.","rationale":"The reader's weakest_assumption identifies exactly the concern that I judge most load-bearing: the physical accuracy of the instantaneous isochoric free-expansion flash. This assumption is not merely a numerical detail; it directly sets the amplitude of every reported transient (gas saturation, pressure drop, temperature drop) and the duration of the transients, because the preconditioner's flash is the only mechanism that injects the phase change. The paper is transparent that this is an idealization and even contrasts it with the isenthalpic treatment of Sanchez-Alfaro et al., but it does not validate that the uv-flash is the correct singular limit of a rapid aperture change. The central methodological contribution—that a physics-based preconditioner can detect transients that adaptive time stepping otherwise skips—is supported by the simulations and does not depend on the exact flash model, but the paper's physical conclusions (monotonic trends with aperture, thermal effects, substantial pressure reductions) are conditional on the flash being faithful. The proposed smooth-ramp convergence test would settle this directly by checking whether the uv-flash is recovered as the opening becomes arbitrarily fast. Because this is the same concern the reader already raised and the CONDITIONAL verdict is appropriate, no adjustment to the verdict is needed. Credit is due for the reproducible code (Zenodo) and the internally consistent formulation; the weakness is a validation gap, not an internal inconsistency.","tokens_in":26039,"tokens_out":8075,"duration_ms":94255,"concrete_test":"Replace the discontinuous aperture jump (Eq. 4.1) with a smooth ramp a(t) = a_r [1 + (γ−1) H_τ(t−t⋆)] with transition time τ, and run unpreconditioned simulations with time steps Δt ≪ τ for τ = 10^4, 10^3, 10^2, 10, 1, 0.1 s (using the thermal case vT-3.0 and isothermal i-vT-3.0). Record peak gas saturation, minimum pressure, and minimum temperature for each τ; extrapolate to τ→0 and compare with the uv-flash preconditioned results (vT(uv)-3.0, i-vT(vT)-3.0). If the extrapolated quantities match the flash results, the free-expansion assumption is the correct singular limit; if they instead approach an isenthalpic flash or a different state, the preconditioner's physical predictions are not validated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing concern is the unvalidated isochoric free-expansion flash that defines the post-opening state. In Section 3.2, the preconditioner sets v* = γ v(t^n) (Eq. 3.9) and, in thermal cases, keeps u fixed, then performs a uv-flash (Algorithm 1, lines 14–15). This flash is the only physics that distinguishes the preconditioned from the unpreconditioned simulations; it therefore fixes the reported peak gas saturation, temperature drop, pressure drop, and transient durations (Figures 5–12). The manuscript explicitly acknowledges this is 'a modeling idealization' and that a different isenthalpic choice was made by Sanchez-Alfaro et al. [44], but provides no resolved reference solution, no convergence-to-the-singular-limit study, and no experimental comparison. If the true post-opening thermodynamic path is not the constant-u free expansion (e.g., because the fracture walls do work on the fluid, or finite-rate opening allows partial re-equilibration), all quantitative claims—including the monotonic trends in Section 4.3—would shift. The numerical claim that the preconditioner detects transients otherwise missed by adaptive stepping is more robust, but the paper's headline physical results are conditional on this assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a persistent-variable formulation for thermal compositional flow in fractured porous media in which specific volume is promoted to an independent transport variable. From a fully coupled flow–transport–equilibrium system, the authors derive volume-based models and show that classical pressure-based formulations are recovered by eliminating local thermodynamic variables. To handle abrupt fracture opening, they introduce a nonlinear preconditioner that assumes instantaneous free expansion of a closed local fluid amount: specific volume is set to v* = γ v(t^n) (Eq. 3.9), internal energy is held fixed in thermal cases, and a uv- or vT-flash is performed before the coupled transport solve (Algorithm 1, lines 14–15). Numerical examples show that unpreconditioned simulations step over the transient vaporization event, while preconditioned vT- and ph-based runs resolve it, and that gas content, cooling, and transient durations increase monotonically with aperture ratio. The paper also reports that pressure- and volume-based formulations recover the same physical solution but differ in nonlinear robustness.","tokens_in":26343,"tokens_out":6444,"duration_ms":78277,"significance":"If the central modeling premise is accepted, the paper makes a useful methodological contribution: it shows a thermodynamically motivated initialization strategy that prevents adaptive time stepping from skipping short-lived phase-change events, demonstrates that the global equilibrium specification can be chosen independently of the thermodynamic preconditioner, and provides a clean derivation connecting volume-based and pressure-based formulations. The work is reproducible (Docker/Zenodo archive), the derivations are parameter-free, and no model parameters are fitted to the reported results. The main caveat is that every reported transient quantity is a consequence of the assumed free-expansion thermodynamic path; the paper acknowledges this idealization but does not validate it or quantify its influence. The numerical claim about detecting transients is robust in the sense that the preconditioner changes the solver behavior, but the physical claim about mechanically induced vaporization is conditional on the unvalidated uv-flash assumption.","major_comments":[{"comment":"The post-opening state is fixed by v* = γ v(t^n) and, for thermal runs, u(t^n), followed by a uv-flash. This free-expansion assumption is the only mechanism that distinguishes preconditioned from unpreconditioned simulations, so all reported transient quantities — peak gas saturation, temperature drop, pressure drop, and transient durations in Figures 5–12 and Section 4.3 — inherit it. The paper explicitly states that the transition is 'not uniquely prescribed by the governing equations' and acknowledges the isenthalpic choice of Sanchez-Alfaro et al. [44], but it provides no resolved reference solution, no comparison with the isenthalpic alternative, and no sensitivity study. If the actual path involves work done on the fluid by the moving fracture walls or finite-rate opening, the quantitative claims (e.g., 6.51–66.36% gas, 9.25–9.36 MPa pressure drop) would shift. Please add a compari","section":"§3.2, Eq. (3.9), Algorithm 1"},{"comment":"The statement that unpreconditioned simulations 'miss transient vaporization' is partly a statement about adaptive time stepping: as the paper explains, the time step spanning t_-1 and t* exceeds the transient duration τ_g, so the solver advances directly from the pre-expansion state to the post-transient state. The preconditioner does not detect an independently known physical event; it imposes a thermodynamic path and then resolves the consequences. The abstract's phrasing 'preconditioned models resolve it' should be qualified as 'resolve the vaporization predicted by the free-expansion model.' This distinction is central because the paper's headline physical result is not a validation of mechanically induced vaporization but a demonstration that a prescribed uv-flash can be embedded in the nonlinear solver.","section":"§4.2, Abstract"},{"comment":"The principal conceptual claim that 'equilibrium specification controls the numerical properties of the nonlinear problem rather than the recovered physical response' is supported only by comparing the vT and ph global formulations on a single scenario. The uv global formulation is explicitly not run because 'we do not expect' it to change the results. That expectation is plausible from the governing-equation structure, but it remains a conjecture. Given that this coordinate-invariance claim is highlighted in the abstract and conclusion, either run the uv formulation or restrict the claim to the formulations actually tested, and report quantitative differences between the vT and ph solutions rather than relying on visual 'indistinguishability.'","section":"§4.3, conclusion"}],"minor_comments":[{"comment":"The caption labels the first panel 'v [mol m^{-3}]', but v is defined in Eq. (2.29) as specific volume, so the units should be m^3/mol (or the label should be changed to density).","section":"Figure 5 caption"},{"comment":"The text refers to 'the enthalpy mobility λ_u'; the symbol is defined as λ_h. Please correct the typo.","section":"Eq. (2.8) and text after Eq. (2.9)"},{"comment":"The pseudocode calls NPIPM with σ_pre but does not explicitly show the computation of v* and, in the thermal case, u(t^n), which the text in §3.2 describes. Adding these steps would make the algorithm self-contained.","section":"Algorithm 1, line 15"},{"comment":"The abstract says pressure–enthalpy and volume–temperature formulations 'recover identical physical solutions,' while Section 4.3 says 'virtually identical' or 'indistinguishable.' Please choose one wording and, ideally, give a quantitative measure (e.g., maximum relative difference in gas saturation or pressure) to support the claim.","section":"Abstract and §4.3"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the numerical machinery is, as far as I can tell, carefully constructed and reproducible. My main reservation is not the method but the interpretation: the preconditioner imposes the free-expansion thermodynamic path, and every quantitative result follows from that choice. I would not reject the paper, but I would ask the authors to either add a sensitivity/validation study against at least one alternative path or a resolved reference, or to reframe the central claims as conditional on the free-expansion idealization. The omitted uv global run also weakens the coordinate-invariance claim. With these revisions, the paper would be a solid contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the central methodological claim checks out: in their fracture-opening scenarios, the unpreconditioned runs step over the vaporization transient, and the preconditioned runs catch it. That is shown clearly for both pressure- and volume-based formulations. Second, everything quantitative about that transient — peak gas saturation, temperature drop, transient durations — is fixed by the assumed post-opening thermodynamic path (v* = γ v(t^n), u constant, uv-flash), and that assumption is acknowledged as an idealization but never tested against the isenthalpic alternative or any resolved reference. So treat the numbers as conditional, the mechanism as solid.\n\nWhat is new and well done: the persistent-variable volume-based framework is coherent. The derivation from the coupled system to the volume-based models, and the recovery of pressure-based formulations, is clean. The isochoric preconditioner embedded in the time loop is new relative to the cited work — Sanchez-Alfaro et al. prescribe the post-opening state separately, and prior volume-based formulations don't use this free-expansion initialization. The PorePy implementation, code, and Docker data are there, and no parameters are fitted; γ is a prescribed scenario input.\n\nSoft spots: the biggest is exactly what the reader's report flags. Eq. (3.9) sets v* = γ v(t^n) and the uv-flash sets every reported transient quantity. The paper explicitly says this is a modeling idealization and contrasts it with the isenthalpic choice of [44], but no isenthalpic run is shown. There's also no convergence study to the singular limit of instantaneous opening, and no experimental comparison. If the walls do work on the fluid during opening, or opening is finite-rate, the quantitative trends in Section 4.3 shift. The qualitative detection result is robust to that, but the physical numbers are not. Minor typo: Figure 5 lists v in mol/m^3, which should be m^3/mol (or the reciprocal). The citation pattern is fine; self-citations to [45,46] are appropriate given the flash machinery is their own.\n\nWho this is for: anyone working on compositional reservoir simulation, coupled flow-geomechanics, or dilation-induced phase change. The paper deserves a serious referee. My recommendation: send it out, but ask for a sensitivity study around the free-expansion assumption — at minimum run the isenthalpic preconditioner and show how the transient changes. That would turn a conditional result into a robust one.","headline":"The isochoric preconditioner is a genuinely new idea and the detection claim holds; the quantitative transient results ride on an acknowledged but untested free-expansion assumption.","tokens_in":26788,"tokens_out":2926,"would_cite":true,"duration_ms":95431,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76S05","80A22","65H10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Sudden fracture opening can vaporize water faster than adaptive time stepping can see; this paper's isochoric free-expansion preconditioner resolves the transient in both volume- and pressure-based flow models.","keywords":["fractured porous media","non-isothermal compositional flow","phase transition","persistent-variable formulation","nonlinear preconditioning","isochoric equilibrium","volume balance","mixed-dimensional model"],"falsifier":"Run the same fracture-opening scenario with a fully coupled solver that takes time steps far smaller than the transient (for example, milliseconds around the jump) without the preconditioner, or with a finely resolved reference simulation; if the small-step solution produces no vapor or a materially different pressure drop or gas fraction, the preconditioner's predicted transient is an artifact of the free-expansion assumption. A laboratory check would be a rapid-dilation experiment that measures on-fault pressure and vapor content during the opening and compares them to the isochoric-flash pr","tokens_in":25960,"feed_emoji":"♨️","tokens_out":4674,"duration_ms":58857,"temperature":0.7,"pith_summary":"Rapid fracture opening can change pore volume faster than fluids can flow, and conventional implicit time stepping can step straight over the resulting vaporization transient without ever seeing it. This paper argues that the missing piece is a properly chosen thermodynamic path for the instantaneous volume change: it treats the event as a free expansion of a closed fluid parcel, jumps the specific volume to match the new pore volume, keeps internal energy (or temperature) fixed, and runs an isochoric equilibrium flash to set the post-opening state before the transport solve. On that basis it introduces volume-temperature and volume-internal-energy formulations alongside the classical pressure-enthalpy one, all sharing a persistent-variable equilibrium solver that handles disappearing and appearing phases without switching primary variables. In the studied fracture-opening cases, preconditioned models resolve transient vaporization that unpreconditioned ones miss, and larger aperture increases produce monotonically stronger gas content, cooling, and transient durations. If correct, the message is that choosing how you initialize thermodynamics after a mechanical event, not just which equations you solve, determines whether fast phase change is captured, and that pressure- and volume-based formulations recover the same physics with different solver behavior.","feed_headline":"A thermodynamic flash catches vaporization that fast solvers skip","feed_subtitle":"Treating fracture opening as free expansion lets simulations resolve steam that adaptive time stepping jumps over.","key_machinery":"The central object is the isochoric equilibrium calculation: an equilibrium flash at fixed specific volume with either fixed temperature (vT specification) or fixed specific internal energy (uv specification). In the persistent-variable formulation the flash is cast as a constrained minimization solved through a semi-smooth KKT system in which phase fractions can be zero, so no phase stability checks or primary-variable switching are needed. The preconditioning step uses the free-expansion path v* = gamma v(t^n) (with u or T held) to initialize the secondary thermodynamic variables before the coupled Newton solve; a Schur complement reduction then feeds thermodynamically consistent derivativ","core_discovery":"The paper's central claim is that abrupt pore-volume change should be handed to the flow solver as a thermodynamic event, resolved by isochoric equilibrium, before transport is advanced. Concretely, when the fracture aperture jumps by a factor gamma, the fluid specific volume is set to v* = gamma v(t^n), the pre-event specific internal energy (or temperature, in the isothermal case) is held fixed, and an equilibrium calculation at fixed (u,v) or (v,T) produces the new pressure, temperature, saturations, and phase fractions. This free-expansion flash is a nonlinear preconditioner that applies equally to volume- and pressure-based global formulations. The paper reports that without it, both pr","pith_inferences":["The same preconditioner should detect rapid pore-volume compaction as well as dilation: compacting a closed parcel with fixed internal energy would raise pressure and temperature, and the flash would similarly initialize the compressed state, a testable extension the paper does not run.","Comparing the paper's isochoric (constant-internal-energy) free expansion against an isenthalpic (constant-enthalpy) initialization on identical fracture-opening cases would show how much the choice of thermodynamic path matters and which idealization is closer to a fully resolved solution.","By making the thermodynamic event explicit, the preconditioner doubles as an automatic event detector: it could flag time steps where the physically implied state change is large, guiding adaptive stepping instead of relying on trial-and-error step cuts.","If the claimed coordinate invariance holds generally, it suggests that reservoir simulators can keep their established pressure-based transport codes and still capture mechanically induced phase change simply by adding an isochoric flash as a pre-step, no rewrite of the global solver required."],"forward_implications":["Unpreconditioned pressure- and volume-based simulations step over the vaporization transient; the isochoric preconditioner makes both formulations resolve it, so the fast event can be captured without prior knowledge of its timing.","Within the tested aperture range (gamma = 1.1 to 3.0), peak gas content, expansion-induced cooling, and the durations of the gas and pressure transients all increase monotonically with gamma.","Thermal and isothermal runs give nearly identical transient durations; the thermal runs differ in the phase-evolution path and leave the fracture slightly warmer after recovery.","Pressure-enthalpy and volume-temperature formulations recover the same physical solution while differing in nonlinear convergence, so the equilibrium specification can be chosen for numerical convenience without changing the physics.","The modeled transient pressure reduction (roughly 10 MPa to 0.8 MPa) is large enough that ignoring geomechanical feedback in fully coupled simulations could misrepresent the subsequent mechanical response."],"fun_headline_variants":["Free-expansion flash resolves vaporization fast solvers skip","Isochoric preconditioner catches phase change missed by adaptive steps","Thermodynamic flash in fracture opening reveals hidden vaporization","Preconditioner treats fracture opening as free expansion, catching steam"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The post-opening state is fixed by assuming an instantaneous free expansion of a closed local fluid parcel: specific volume jumps to the pore-volume ratio while specific internal energy (or temperature) stays fixed, with no mass, energy, or heat exchange during the event; every reported transient quantity inherits this idealization.","fun_headline_variants_meta":{"raw":{"variants":["Free-expansion flash resolves vaporization fast solvers skip","Isochoric preconditioner catches phase change missed by adaptive steps","Thermodynamic flash in fracture opening reveals hidden vaporization","Preconditioner treats fracture opening as free expansion, catching steam"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000593,"raw_usage":{"total_tokens":2641,"prompt_tokens":795,"completion_tokens":1846,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":539,"completion_tokens_details":{"reasoning_tokens":1776}},"tokens_in":539,"tokens_out":1846,"duration_ms":14775,"temperature":1.0,"reasoning_tokens":1776,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T14:50:47.226652+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same fracture-opening scenario with a fully coupled solver that takes time steps far smaller than the transient (for example, milliseconds around the jump) without the preconditioner, or with a finely resolved reference simulation; if the small-step solution produces no vapor or a materially different pressure drop or gas fraction, the preconditioner's predicted transient is an artifact of the free-expansion assumption. A laboratory check would be a rapid-dilation experiment that measures on-fault pressure and vapor content during the opening and compares them to the isochoric-flash pr","supporting_citations":[],"review_version":1}