{"id":"16f38e51-5615-4b87-a368-4b905edf8014","arxiv_id":"1908.06577","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A three-region empirical model, closed with one fitted boundary-layer scaling, predicts Taylor-Couette torque grows as Ta^{1/2} divided by the square of a Lambert W function.","lead":"This paper studies the swirling flow between a rotating inner cylinder and a stationary outer cylinder, and proposes a simple model of the turbulent flow made of two thin wall layers around a core region with nearly constant spin. The model predicts that the torque needed to keep the cylinders spinning grows almost like the square root of the Taylor number, but with a logarithmic correction that no simple power law can capture.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Lambert-W torque law and the asymptotic free-vortex state both rest on the unvalidated closure δ_i = K u_τ_i/Ω_i; one DNS point does not establish that K is universal in η and Ta.","rationale":"The reader's weakest_assumption identifies exactly the load-bearing point: Eq. (4.7) is a one-parameter closure calibrated on a single DNS point, and the asymptotic Nu(Ta,η) scaling and the vanishing-wall-layer state are derived from it. I considered whether the assumed L=1/2 constant-angular-momentum core is more vulnerable, but that assumption has direct support from Fig. 6 and from the matching structure of the model; the thickness closure is the least constrained element. The paper is transparent about the closure, and the comparisons with experiments at several η provide meaningful but indirect support. A direct test of the constancy of K against existing DNS at other radius ratios would settle whether the central claim survives. Since the concern is not a demonstrated contradiction but an under-supported extrapolation, the existing CONDITIONAL verdict is the right one: no change is needed.","tokens_in":18338,"tokens_out":9707,"duration_ms":103486,"concrete_test":"Use existing DNS mean angular-momentum profiles at η=0.5 and η=0.714 (e.g., Ostilla-Monico et al.) over Rei=10^5–3×10^5; define δ_i as the radius where L = rU_θ/(Ω_i R_i^2) first reaches 0.5; compute the ratio δ_i Ω_i/u_τ_i and test whether it is constant at 0.25 across η and Rei. If it varies by more than about 20%, the closure (4.7) fails and the Lambert-W asymptotic scaling is not established; if it is constant, the model's central claim gains independent support.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (4.7), δ_i = K u_τ_i/Ω_i, is the only closure in the model. With α=2K=0.5 it is calibrated against a single DNS value, Re_τi=1410 at Rei=10^5, η=0.909. Substituting (4.8) into the inner matching equation (4.5) produces Eq. (4.10), whose Lambert-W solution (4.12) directly yields the central Nu(Ta,η) law (4.15) and the asymptotic state of §4.6. If K actually depends on η or on Rei/Ta, then (4.12), (4.15), and the 'not a power law' conclusion do not follow. The paper's own comparisons cannot fully certify (4.7): Table 2's δ_i/d values are model outputs, and Fig. 10 compares a different, criterion-dependent δ_99 defined by (4.19), not the δ_i appearing in (4.7). Agreement with bulk Nu at η=0.5, 0.72, 0.909 is encouraging but is an integrated measure that may be insensitive to the precise wall-layer thickness scaling. Thus the most load-bearing assumption is that K is a universal constant of order 0.25.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents wall-resolved large-eddy simulations (LES) of Taylor–Couette flow with a rotating inner cylinder and a stationary outer cylinder at radius ratio η = 0.909 and inner Reynolds numbers up to Rei = 3×10^6, using the stretched-vortex subgrid model. The LES are verified against DNS at Rei = 10^5 and 3×10^5 and are then used to characterize the mean azimuthal velocity, turbulence intensities, and angular-momentum profiles. The authors propose a one-dimensional, three-region empirical model: log-law wall layers on each cylinder separated by a central region of constant angular momentum L = Ωi Ri^2/2. The model is closed by the scaling hypothesis δi = K uτ_i/Ωi, with the constant fixed to K = 0.25 (α = 2K = 0.5) using one DNS point. The model yields an analytic Lambert-W formula for Nu(Ta,η), predicts that Nu grows as Ta^{1/2} divided by the square of a Lambert-W function, and describes an asymptotic state of nearly constant angular momentum with vanishingly thin wall layers. An extension to sand-grain roughness predicts a fully rough regime with Nu ∼ Ta^{1/2} and Cf independent of Rei.","tokens_in":18586,"tokens_out":12567,"duration_ms":125863,"significance":"If the central closure is accepted, the paper provides a simple, closed-form description of the torque scaling in turbulent Taylor–Couette flow with a stationary outer cylinder, including a concrete non-power-law asymptote that could organize existing experimental and numerical data. The new wall-resolved LES data at Rei up to 3×10^6 for η = 0.909 are a useful resource, and the rough-wall extension gives a testable Moody-diagram analogue for this geometry. The algebraic derivations are transparent and the final formulas are easy to use. The main caveat is that the quantitative predictions and the asymptotic state are contingent on the empirically calibrated closure (4.7), and the paper would be strengthened by a clear statement of that conditionality and by a sensitivity analysis of the fitted constant.","major_comments":[{"comment":"The closure δi = K uτ_i/Ωi is the sole load-bearing assumption of the model, yet it is introduced without derivation, and the constant is fixed by matching a single DNS point (Rei = 10^5, Reτi = 1410). That same DNS point reappears in the validation (Table 2 and Fig. 7), so the agreement there is a consistency check rather than an independent confirmation. The comparisons at other η (experiments) and at higher Rei (present LES) do provide indirect support for K being independent of η and Ta over the tested range, but the asymptotic claims in §4.6 and the non-power-law form in Eq. (4.15) rely on (4.7) holding at arbitrarily large Ta, far beyond the validated range. Please add a sensitivity study with respect to α (e.g., how Nu(Ta,η) and the inferred asymptotic state vary for α = 0.25, 0.5, 0.75, 1.0), and either present direct evidence for the scaling δi ∝ uτ_i/Ωi from DNS/LES at several η and Rei or explicitly frame the asymptotic state and the Lambert-W law as model-based conjectures in the abstract and conclusions.","section":"4.1"},{"comment":"The approximate analytical solution leading to Eq. (4.15) is obtained after substituting α = 1/2 into Eq. (4.9) and then dropping subdominant terms, and the final formula contains no dependence on α. This may mislead readers into thinking the Lambert-W expression is parameter-free. In reality, for α ≠ 1/2 the argument of the Lambert function acquires a factor √α (through the 2αη term in the logarithmic argument), and the quantitative prediction changes. Please state explicitly that the functional form—the logarithmic correction to Ta^{1/2}—is insensitive to the calibrated value, but that the specific formula (4.15) and the numerical comparisons are tied to α = 0.5.","section":"4.2"},{"comment":"The LES validation of the model at high Rei rests on a computational domain of Δθ = π/10 and Ly = 2πd/3. As the authors themselves note in §3.4, Ostilla-Monico et al. (2015b) found that finite-domain effects on the near-wall log region and on turbulence statistics are non-negligible at moderate Reynolds numbers. Because the high-Rei LES (up to Rei = 3×10^6) are used to support the model’s Ta dependence, the paper should quantify or at least discuss the possible influence of this narrow domain on the mean azimuthal velocity and torque, or clearly state the resulting uncertainty in the high-Rei validation.","section":"3.1"}],"minor_comments":[{"comment":"The heading 'Numerical method' appears twice; the second occurrence should be renumbered or renamed to distinguish the curvilinear formulation.","section":"§2.1–2.2"},{"comment":"The citation 'Cheng et al. 2017, 2018, ?' contains a bare question mark; the reference list is incomplete and should be fixed.","section":"§2.2"},{"comment":"The definition of δ99 uses the inequality U(r) - U_target < 0.01, which is not sign-definite; it should likely be |U(r) - U_target|/U_target < 0.01.","section":"§4.5, Eq. (4.19)"},{"comment":"The spelling of the author name 'Ostilla-Monico' is inconsistent (e.g., Ostilla-M´onico vs. Ostilla-Mnico) in the text and in the reference list.","section":"throughout"},{"comment":"The sentence 'Once the parameters Reτi and ηi have been determined' should refer to Reτi and δi rather than ηi.","section":"§4.5"},{"comment":"The abstract contains the sentence 'With Ri,Ro the inner and outer radii respectively, the radius ratio is η = 0.909', which reads as if the paper only treats one radius ratio, whereas the model is applied for η = 0.5, 0.72, and 0.909. Please rephrase to avoid confusion.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of JFM and makes a useful contribution: new wall-resolved LES data at Rei up to 3×10^6 and a compact analytical model with a clearly stated closure. The main risk is that the central closure is calibrated on one DNS point, and the asymptotic claims are extrapolations far beyond the validated parameter range. I believe the paper can be made publishable by adding the requested sensitivity analysis and by explicitly marking the asymptotic state and the Lambert-W law as model predictions. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The Lambert-W torque law is the real product here, and it is new. The model reduces TC torque to a closed form involving the ProductLog function, and it does a decent job matching DNS, experiments, and the authors' own wall-resolved LES across a useful range of eta and Ta. The rough-wall extension to a Moody diagram for TC flow is also a nice piece of work, and the LES itself, while not groundbreaking, extends the resolved range to Re_i = 3e6 and seems carefully done. The paper is honest about the modeling assumptions, which goes a long way.\n\nThe soft spots are real but not fatal. The stress-test note lands: Eq. (4.7), delta_i = K u_tau_i / Omega_i, is the entire closure, and it is introduced with no derivation. Alpha = 2K = 0.5 is calibrated against a single DNS point at eta = 0.909, and the same point later appears in the validation. That makes the quantitative asymptotic claims conditional on K being universal in eta and Ta, which is not established. The comparisons to bulk Nu at other radius ratios are encouraging, but as the note says, Nu is an integrated quantity and could be insensitive to the precise wall-layer scaling. The authors also import kappa = 0.4 and A = 4.5 from flat-wall turbulence without justification, and the LES domain is a narrow sector with a domain-size issue that they themselves acknowledge. No code or data are shipped, which limits reproducibility of the LES.\n\nThe central derivation is algebraically self-consistent, and the paper is clear that the model is empirical. The 'not a power law' conclusion should be read as a prediction of the model, not an established fact. But the model has enough explanatory power to deserve a serious referee, and the limitations can be addressed with sensitivity tests and comparison to independent data at other radius ratios. The paper is for the TC and wall-turbulence community, and a receptive reader will find it useful. A skeptical reader will find the closure ad hoc, but that is not a reason to desk reject.\n\nRecommendation: send it to peer review. The referee should press on the universality of K, the domain-size effects, and the absence of validation data outside eta = 0.909, but the Lambert-W result is a legitimate contribution that should be on record.","headline":"A genuinely new analytic torque law for TC flow, but the central asymptotic claim hangs on a single fitted closure constant; worth refereeing.","tokens_in":19124,"tokens_out":1549,"would_cite":true,"duration_ms":18447,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["47.27.Cn"],"model":"deepseek-v4-flash","headline":"This paper claims that in turbulent Taylor-Couette flow with a stationary outer cylinder, the torque ratio grows as the square root of the Taylor number divided by the square of a Lambert-W function, a logarithmic-corrected non-power law.","keywords":["Taylor-Couette flow","turbulent torque scaling","large-eddy simulation","constant angular momentum","Lambert W function","logarithmic wall layers","rough-wall turbulence","Nusselt number"],"falsifier":"Measure $\\delta_i$ and $u_{\\tau i}$ separately in DNS or experiment at a second radius ratio, say $\\eta=0.5$, over a range of $Re_i$: if $\\delta_i\\Omega_i/u_{\\tau i}$ deviates from approximately 0.5 by more than the uncertainty, the closure—and with it the Lambert-W torque law—is falsified. Alternatively, bin existing $Nu(\\mathrm{Ta})$ data at fixed $\\eta$ by decade and check that the local slope drifts downward toward $1/2$ as $1/(\\ln\\mathrm{Ta})^2$ rather than settling on a constant exponent.","tokens_in":18126,"feed_emoji":"🌀","tokens_out":17490,"duration_ms":151403,"temperature":0.7,"pith_summary":"This paper studies the torque needed to sustain turbulent flow between a rotating inner cylinder and a stationary outer cylinder. It builds a minimalist three-zone picture: thin logarithmic wall layers at each cylinder, separated by a bulk region of constant angular momentum. The central result is a closed-form formula for the Nusselt number $Nu$ (the torque ratio): at fixed radius ratio $\\eta<1$, $Nu$ grows like $\\mathrm{Ta}^{1/2}$ divided by the square of the Lambert-W function of a quantity proportional to $\\mathrm{Ta}^{1/4}$. That is a logarithmic correction to a half-power law, not a power law, so reported exponents below $0.5$ are only finite-range fits. If the model is right, the ultimate high-$\\mathrm{Ta}$ state is nearly uniform angular momentum with vanishingly thin wall layers.","feed_headline":"Taylor-Couette torque: log-corrected half-power, not power law","feed_subtitle":"A three-region model predicts the torque law and an ultimate flow of uniform angular momentum with thin wall layers.","key_machinery":"The load-bearing construction is a one-dimensional, three-region radial model of the mean azimuthal velocity: region I is an inner-cylinder log layer with profile $U_\\theta=\\Omega_i R_i-u_{\\tau i}(\\kappa^{-1}\\ln((r-R_i)u_{\\tau i}/\\nu)+A)$, region II has constant angular momentum $U_\\theta=\\frac{1}{2}\\Omega_i R_i^2/r$, and region III is an outer log layer. The model is closed by the scaling relation $\\delta_i=K u_{\\tau i}/\\Omega_i$, written as $\\delta_i/d=\\alpha Re_{\\tau i}\\eta/(Re_i(1-\\eta))$ with $\\alpha=2K=0.5$; this turns velocity matching at the inner-layer edge into a single algebraic equation for $Re_{\\tau i}$. Solving that equation in the large-$Re_i$ limit introduces the Lambert-W function, whose sub-logarithmic growth produces the log-corrected torque law.","core_discovery":"The paper's claim is that the mean state of high-Reynolds-number Taylor-Couette flow with the outer cylinder at rest is captured quantitatively by matching two log-law layers against a known constant-angular-momentum core, $r u_\\theta = \\frac{1}{2}\\Omega_i R_i^2$. The matching produces an algebraic equation for the inner friction Reynolds number $Re_{\\tau i}$, and in the large-$Re_i$ limit the solution is $Re_{\\tau i}\\propto Re_i/W(Z_1)$ with $Z_1\\propto \\eta^{1/2}Re_i^{1/2}/(1-\\eta)^{1/2}$. Converting to torque gives $Nu(\\mathrm{Ta},\\eta)=\\kappa^2\\eta^3\\mathrm{Ta}^{1/2}/(4(1+\\eta)^2 W(Z_2)^2)$ with $Z_2\\propto \\eta^{3/2}\\mathrm{Ta}^{1/4}/((1-\\eta)^{1/2}(1+\\eta)^{3/2})$. At enormous $\\mathrm{Ta}$ this becomes $Nu\\sim 4\\kappa^2\\eta^3\\mathrm{Ta}^{1/2}/((1+\\eta)^2(\\ln\\mathrm{Ta})^2)$, a genuine non-power-law asymptote. The same model, extended with a Colebrook roughness function, predicts that a rough inner wall produces a fully rough plateau with $C_f$ independent of $Re_i$ and $Nu\\sim\\mathrm{Ta}^{1/2}$.","pith_inferences":["If the Lambert-W form is generic, then single-exponent fits to $Nu(\\mathrm{Ta})$ are inherently temporary; the local slope should decrease logarithmically with $\\mathrm{Ta}$, a drift that can be checked by decade-binning existing torque data.","The same three-region construction could be carried over to other rotating wall-bounded flows with roll-driven angular-momentum mixing, but the closure constant $K$ would need independent measurement rather than being assumed universal.","The rough-wall branch gives a concrete engineering target: torque measurements on roughened cylinders should plateau as $Re_i$ grows, a Moody-diagram-style saturation that existing Taylor-Couette facilities could test directly.","The deeper physical bet is that the constant-angular-momentum bulk persists to arbitrarily large $\\mathrm{Ta}$; if Taylor rolls or the bulk itself change character at extreme driving, the asymptotic state would differ even if the wall-layer closure held."],"forward_implications":["At fixed $\\eta<1$, $Nu(\\mathrm{Ta})$ has no power-law asymptote; the true large-$\\mathrm{Ta}$ behavior is $Nu\\sim \\mathrm{Ta}^{1/2}/(\\ln\\mathrm{Ta})^2$, so local power-law fits with exponent below $0.5$ must drift downward as $\\mathrm{Ta}$ increases.","The mean azimuthal flow approaches $u_\\theta=\\frac{1}{2}\\Omega_i R_i^2/r$ across almost the whole gap, while the wall-layer thicknesses $\\delta_i/d$ and $\\delta_o/d$ shrink like the inverse of a Lambert-W function.","The model reproduces DNS, LES, and experimental $Nu(\\mathrm{Ta},\\eta)$ data for $\\eta=0.5$, $0.72$, and $0.909$ across $\\mathrm{Ta}\\simeq10^{10}$ to $10^{13}$, including the decline of inner-wall boundary-layer measures with $Re_i$.","For a sand-grain-rough inner wall, the model predicts a fully rough plateau: $C_f$ independent of $Re_i$, $\\delta_i/d$ independent of $Re_i$, and $Nu\\sim\\mathrm{Ta}^{1/2}$.","The model is intended for practical radius ratios roughly $0.6\\le\\eta<1$; it does not cover the plane-Couette limit $\\eta\\to1$, where the predicted $\\delta_i/d$ would diverge."],"supporting_citations":[{"why":"It supplies the DNS benchmark used to verify the LES and to calibrate the closure constant alpha = 0.5.","marker":"Ostilla-Mónico et al. (2016)"},{"why":"It provides the definitions of Ta and Nu and the review context for ultimate-regime torque scaling.","marker":"Grossmann et al. (2016)"},{"why":"It provides high-Reynolds-number experimental torque data at eta = 0.909 used to validate Nu(Ta).","marker":"Van Gils et al. (2011)"},{"why":"It provides experimental torque-scaling data at eta = 0.72 and 0.909 used as model comparison.","marker":"Van Gils et al. (2012)"},{"why":"It provides wide-gap torque measurements at eta = 0.5 used as model comparison.","marker":"Merbold et al. (2013)"},{"why":"It supplies the Lambert-W asymptotic expansion used to derive the log-corrected large-Ta behavior.","marker":"Corless et al. (1996)"},{"why":"It provides the rough-wall velocity formulation behind the Colebrook roughness function used in the rough-wall extension.","marker":"Jiménez (2004)"}],"fun_headline_variants":["Taylor-Couette torque law: not a power law, a Lambert W twist","Torque in TC flow: half-power law masked by log corrections","TC torque: Ta^1/2/W^2, not a simple power law","TC torque: Lambert W scaling beats power-law fits","Taylor-Couette torque: non-power-law scaling via Lambert W"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quantitative torque law rests on the assumption that the inner wall-layer thickness is proportional to the local friction velocity divided by the cylinder rotation speed, with the constant fixed at 0.5 from one DNS point; if that ratio depends on radius ratio or Reynolds number, the claimed asymptote fails.","fun_headline_variants_meta":{"raw":{"variants":["Taylor-Couette torque law: not a power law, a Lambert W twist","Torque in TC flow: half-power law masked by log corrections","TC torque: Ta^1/2/W^2, not a simple power law","TC torque: Lambert W scaling beats power-law fits","Taylor-Couette torque: non-power-law scaling via Lambert W"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001233,"raw_usage":{"total_tokens":5208,"prompt_tokens":1229,"completion_tokens":3979,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":845,"completion_tokens_details":{"reasoning_tokens":3884}},"tokens_in":845,"tokens_out":3979,"duration_ms":28237,"temperature":1.0,"reasoning_tokens":3884,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:40:12.203829+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure $\\delta_i$ and $u_{\\tau i}$ separately in DNS or experiment at a second radius ratio, say $\\eta=0.5$, over a range of $Re_i$: if $\\delta_i\\Omega_i/u_{\\tau i}$ deviates from approximately 0.5 by more than the uncertainty, the closure—and with it the Lambert-W torque law—is falsified. Alternatively, bin existing $Nu(\\mathrm{Ta})$ data at fixed $\\eta$ by decade and check that the local slope drifts downward toward $1/2$ as $1/(\\ln\\mathrm{Ta})^2$ rather than settling on a constant exponent.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the DNS benchmark used to verify the LES and to calibrate the closure constant alpha = 0.5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides high-Reynolds-number experimental torque data at eta = 0.909 used to validate Nu(Ta)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides experimental torque-scaling data at eta = 0.72 and 0.909 used as model comparison."},{"cited_title":"Physical Review E\\/ 87 (2), 023014","cited_arxiv_id":null,"evidence_quote":"It provides wide-gap torque measurements at eta = 0.5 used as model comparison."},{"cited_title":"Advances in Computational mathematics\\/ 5 (1), 329--359","cited_arxiv_id":null,"evidence_quote":"It supplies the Lambert-W asymptotic expansion used to derive the log-corrected large-Ta behavior."}],"review_version":1}