{"id":"53725e8b-ef06-4d13-ae02-183ef2a2b24e","arxiv_id":"2507.03187","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Across six Reynolds numbers and eight angles of attack, the leeward flow over a 6:1 prolate spheroid takes one of three topologies, proto-vortex, coherent vortex, or recirculating wake, whose axial evolution sets the body's loads.","lead":"The paper maps 48 large-eddy simulations of flow past a 6:1 prolate spheroid and organizes the leeward recirculation into three states: a weak proto-vortex, a coherent rolling vortex, and an incoherent recirculating wake. A generalist gets a clean story connecting these states, their axial growth, and the forces and moments on the body.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The stagnation-pressure threshold in section 3.6 is derived from an algebraic error; the resulting 0.49 P_inf_s criterion defines A_t, Gamma_t, and the load model, so the central scalings and load-topology link may be artifacts of that threshold.","rationale":"Both the reader and I identify serious weaknesses, but the most load-bearing is the threshold derivation. The reader's weakest_assumption is the LES transition prediction (Re approximately 1.5M versus experimental 2-3M). That is a valid concern about external validity, but it affects the mapping of Re to regime boundaries; the qualitative three-state taxonomy might survive a shift in transition location. The threshold error, by contrast, is an internal inconsistency in the definition of the measured quantities themselves. Every quantitative result in the paper - A_t(x), Gamma_t(x), vortex radius/circulation, and the load integral - is computed from the 0.49P_inf_s iso-surface. If that threshold is arbitrary, the log-log slopes in Figs. 40 and 42, and the force decomposition in section 5.9, are not robust measures of the flow but artifacts of a parameter choice. The algebraic error in section 3.6 means the derivation provides no physical justification for the threshold; the 'consistency with delta_99' claim is false. This is exactly the kind of condition that must be checked before accepting the central quantitative claim. The fix is straightforward: either re-derive a correct threshold (e.g., from the actual velocity profile at the delta_99 edge, which will be a function of local pressure, not a universal constant) or perform a sensitivity sweep. If the scalings persist, the paper's claims stand; if not, the central contribution is weakened. Since the reader already assigned CONDITIONAL and flagged the threshold, my independent assessment does not move the verdict; it strengthens the justification for the conditional requirements.","tokens_in":23124,"tokens_out":8069,"duration_ms":86165,"concrete_test":"Recompute A_t(x), Gamma_t(x), and the linear force f_y(x) for three representative cases (alpha = 30, 40, 60 degrees at Re = 4M) using thresholds c = 0.3, 0.4, 0.6, 0.7 P_inf_s. If the log-log slopes of A_t vs x (Fig. 40) and Gamma_t vs x (Fig. 42) remain within +/-10% of their reported values and the axial location of peak f_y is unchanged, the threshold error is not load-bearing. If the slopes or peak location vary systematically with c, the central scalings and load-topology conclusion are artifacts of the arbitrary threshold.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In section 3.6 the authors derive the recirculation threshold by combining the delta_99 definition with a stagnation-pressure relation. They write P_s = P(1.0 - 0.992) + 0.992 P_inf_s approximately 0.49 P_inf_s, implying that the threshold is a constant fraction of freestream stagnation pressure. This step is algebraically inconsistent: with P_s = 0.008P + 0.992P_inf_s, the approximation P_s approximately 0.49P_inf_s would require P approximately -63P_inf_s, an unphysical pressure. No normalization or expansion justifies collapsing the local pressure dependence. The value 0.49 is therefore arbitrary, and the claimed consistency with the boundary-layer thickness definition is not demonstrated. This is not cosmetic. The 0.49P_inf_s criterion defines the recirculation area A_t (P_s < 0.49P_inf_s and r > delta_99), the primary vortex boundary (largest closed stagnation-pressure iso-surface, presumably the same value), and enters directly into the load integral in section 5.9.2, f_y^recirculation(x) = integral(0.49P_inf_s - p_wall) d_phi. The headline results - quadratic growth of A_t, linear growth of Gamma_t, and the conclusion that suction peaks where swirl is high rather than where recirculation is largest - are all extracted from this threshold-dependent object. The control-volume argument in section 5.7 also assumes the boundary is a constant-stagnation-pressure material surface; if the threshold is mis-derived, that assumption fails and the mass-balance rationale for A_t ~ x^2 collapses. Even a perfectly resolved LES with exact transition prediction would produce threshold-dependent scalings and load decomposition unless the criterion is independently justified. A sensitivity check is therefore essential before the quantitative central claims can be accepted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a parametric large-eddy simulation study of flow over a 6:1 prolate spheroid at six Reynolds numbers (Re = 0.15M, 1M, 1.5M, 2M, 3M, 4M) and eight angles of attack (alpha = 10 to 90 deg). The authors classify the leeward recirculation into three states (proto-vortex, coherent vortex, recirculating wake) and describe how the axial evolution of vortex radius, circulation, stagnation pressure, and swirl changes with Reynolds number and incidence. They connect these flow-topology features to the normal force and pitching moment using a control-volume mass balance and a Riabouchinsky-type cavity model, and report power-law scalings A_t ~ x^2 for recirculation area and Gamma_t ~ x^alpha for total circulation. The central claims are that separation is always symmetric in the investigated parameter range, that the coherent-vortex state converts azimuthal momentum into axial momentum, and that maximum suction occurs where swirl is high rather than where recirculation is largest.","tokens_in":23437,"tokens_out":6509,"duration_ms":67335,"significance":"If the quantitative results hold, the paper provides a valuable taxonomy and a large, openly described LES dataset for a canonical separated flow, building on a novel stagnation-pressure-based vortex definition (Plasseraud & Mahesh 2024a). The grid-convergence check (Table 1, 218M vs 470M: 0.11% difference in normal force) and the use of multiple consistent observables (vorticity, stagnation pressure, skin friction, helicity) lend qualitative support to the three-state classification. However, the quantitative layer--quadratic area growth, circulation scaling, and the load-topology link--is built on a threshold derived in Section 3.6 that appears to contain an algebraic error, and the Reynolds-number dependence rests on an assumed transition location that the LES does not resolve. These concerns are central to the paper's main claims, not peripheral. The qualitative taxonomy and the dataset itself are likely to be useful to the community even if the specific scalings require revision.","major_comments":[{"comment":"The derivation of the recirculation threshold P_s ≈ 0.49 P_inf_s is algebraically inconsistent. From P_s = P + 1/2 ρ u^2 and u^2/u_p^2 = 0.992, the exact expression is P_s = 0.008 P + 0.992 P_inf_s. The stated approximation P_s ≈ 0.49 P_inf_s would require P ≈ −63 P_inf_s, which is unphysical. Moreover, the factor 0.992 does not follow from the 0.01 tolerance in Section 3.5; the consistent value would be u^2/u_p^2 = (0.99)^2 = 0.9801. This threshold is load-bearing: it defines A_t (used for the x^2 scaling in Section 5.7 and Figures 39–40), the total circulation Γ_t (Figures 41–42), the primary-vortex boundary in Section 3.7, and the load integral f_y^recirculation(x) = ∫(0.49 P_inf_s − p_wall) dφ in Section 5.9.2. The authors should either correct the derivation or demonstrate that the reported scalings and load conclusions are insensitive to the threshold choice, for example by recomputing A_t, Γ_t, and the load integral for thresholds of 0.4 and 0.6 P_inf_s.","section":"Section 3.6"},{"comment":"The attribution of the jump in δ99 and Re_θ between Re = 1M and 1.5M to a laminar-to-turbulent transition at Re ≈ 1.5M conflicts with the cited experimental critical regime of Re = 2–3M (Ahn 1992) and with the authors' own earlier finding that trips do not guarantee a turbulent boundary layer at high incidence (Plasseraud et al. 2023). The LES uses a dynamic-Smagorinsky SGS model with a trip inherited from prior work, so the transition location is not resolved or independently modeled. The separation lines (Figures 6 and 9), the boundary-layer state (Figures 10–11), and the load trends (Figures 48–49) all depend on this assumed transition. Please provide a sensitivity check, such as a comparison against experimental separation locations in the critical regime, or discuss how the main conclusions would change if transition occurs at Re = 2–3M rather than at 1.5M.","section":"Section 5.2.2"},{"comment":"The control-volume mass balance that leads to the quadratic growth of A_t assumes that the surface S_Ps, an iso-surface of constant stagnation pressure, is a material surface with zero mass flux. In the turbulent, time-averaged flow considered here, the mean stagnation pressure is not constant along mean streamlines because of Reynolds stresses, and the mean mass flux across the iso-surface need not vanish. The authors state that this holds in the inviscid limit, but the LES flow is viscous and turbulent. Please quantify the neglected turbulent flux across S_Ps using the LES data (for example, by computing the actual mean normal velocity on the iso-surface), or provide a justification that the error is small enough to support the derived scaling.","section":"Section 5.7"}],"minor_comments":[{"comment":"The Introduction states that 'the primary vortex is attached and coherent at low angles of attack (10°, 20°)', but Section 5.4 classifies the 20° case as a proto-vortex without a distinct center of rotation; please reconcile this terminology.","section":"Introduction / Section 5.4"},{"comment":"The Burgers vortex profile is written as (u_x, u_theta, u_r) = (−a r, ..., 2 a z); the radial and axial components appear to be interchanged, since a standard Burgers vortex has u_r = −a r and u_z = 2 a z.","section":"Section 5.5.2"},{"comment":"The notation for the freestream stagnation pressure appears both as P∞s and P_inf_s in different places; please unify the notation.","section":"Section 3.6"},{"comment":"The y-axis label of Figure 10 is ambiguous; please state the units and whether the value is normalized by L.","section":"Figure 10"}],"recommendation":"major_revision","confidential_remarks":"The qualitative taxonomy and the parametric dataset are valuable and likely of interest to the JFM readership. However, the algebraic error in Section 3.6 undermines the quantitative scalings and the load model, and the transition assumption in Section 5.2.2 needs support. These issues are fixable within the scope of the manuscript, so I recommend major revision rather than rejection. If the authors can either correct the threshold derivation or show threshold-independence of the results, and provide a sensitivity analysis for transition, the paper could become a strong contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Marc – here's my take on Plasseraud & Mahesh. The paper is worth your time, but the headline scalings rest on a threshold whose derivation in §3.6 is wrong. They define the recirculation by P_s < 0.49 P∞_s, claiming this follows from the δ99 edge condition. It doesn't: at the edge, P_s = 0.008P + 0.992P∞_s, which only equals 0.49P∞_s if P ≈ −63 P∞_s. No normalization makes that legitimate. Since A_t, Γ_t, the vortex boundary, and the load integral in §5.9.2 all use this 0.49 value, the A_t ~ x^2 and Γ_t ~ x scalings and the suction-peaks-with-swirl conclusion are threshold-dependent until shown otherwise. The stress-test note lands.\n\nWhat is genuinely good: the 48-case (Re, α) map is a real step beyond the isolated points in the literature, and the three-state taxonomy (proto-vortex, coherent vortex, recirculating wake) is well illustrated with consistent vorticity, stagnation-pressure, skin-friction, and helicity fields. The separation-line results and the observation that boundary-layer state at separation controls topology are plausible. Grid convergence is decent: 0.11% normal-force change from 218M to 470M, though it's a single point. The paper is honest about using the 2024a boundary definition and about validation inherited from prior work.\n\nThe other soft spots are real but milder: the Re ≈ 1.5M transition inference in §5.2.2 conflicts with the cited experimental critical range (Re = 2–3M, Ahn 1992), and the trips don't guarantee turbulent layers at high incidence; no statistical error bars on any integral quantity. Those are fixable. The central qualitative taxonomy, however, is probably robust: it's based on visual and field-consistent state descriptions, not the suspect threshold alone.\n\nMy recommendation: this deserves a serious referee. The dataset is valuable and the taxonomy is a useful organizing result for a canonical geometry. But the quantitative claims need major revision: correct or replace the 0.49 criterion, show sensitivity of A_t, Γ_t, and the load balance to threshold choice, add convergence/error-bar checks, and validate one matched case against measurements before the scalings are accepted. If the threshold sensitivity is benign, this becomes a reference-grade paper; if not, the qualitative map still stands but the load-topology story weakens. I would not cite the scalings until then; I might cite the classification.","headline":"A genuinely useful 48-case LES taxonomy of leeward flow on the 6:1 spheroid, undermined by an algebraic error in the threshold that defines its headline scalings.","tokens_in":24099,"tokens_out":2920,"would_cite":false,"duration_ms":31918,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76D10","76D17","76F65","76M12"],"pacs":["47.32.C-","47.27.Ep","47.27.nb"],"model":"deepseek-v4-flash","headline":"Across 48 large-eddy simulations of a 6:1 prolate spheroid, the lee-side boundary layer always separates symmetrically, and the recirculation is always one of three states: a proto-vortex, a coherent vortex, or a recirculating wake.","keywords":["prolate spheroid","large-eddy simulation","boundary layer separation","vortex topology","recirculating wake","Reynolds number effects","angle of attack","normal force"],"falsifier":"A wind-tunnel test on a 6:1 prolate spheroid at Re=1.5M and 2.5M, alpha=20 and 40 degrees, measuring the azimuthal separation line and the presence or absence of a coherent lee vortex: the paper predicts a sharp boundary-layer-state jump between those Reynolds numbers, so if the separation-line shift or vortex state is identical at both, the central Re-dependence claim is wrong.","tokens_in":140,"feed_emoji":"🌀","tokens_out":5753,"duration_ms":84098,"temperature":0.7,"pith_summary":"This paper uses 48 large-eddy simulations to map the flow behind a 6:1 prolate spheroid over Reynolds numbers from 150,000 to 4 million and angles of attack from 10 to 90 degrees. It claims that in every case the boundary layer separates symmetrically and the leeward recirculation appears in just one of three states: a proto-vortex with no distinct rotation center, a coherent vortex aligned with the body axis, or an incoherent recirculating wake. In the coherent state the separated shear layer rolls into a vortex with a pressure-minimum core, and along the body the recirculation area grows quadratically while the total circulation grows roughly linearly. These axial evolutions, not the total size of the recirculation, set the normal force and pitching moment. The result matters because it reduces a complex, three-dimensional separated flow to a small set of topology types whose growth laws connect directly to vehicle loads.","feed_headline":"48 simulations collapse spheroid lee flow into three states","feed_subtitle":"At every Reynolds number and angle of attack, the lee recirculation is one of three vortex states that set the forces.","key_machinery":"The central object is the coherent vortex, identified by the largest closed iso-surface of stagnation pressure; this definition supplies a stable boundary for measuring radius, circulation, and area. The argument's engine is an axial control-volume balance in which the constant-stagnation-pressure vortex boundary acts as a material surface, so the mass added across the separated shear layer per unit length drives the axial mass flow to grow linearly while the separation-length geometry ($L_s \\sim x$) makes the area grow quadratically. The load connection comes from a Riabouchinsky-style balance between shear-layer stress and wall suction, and from Crocco's equation expressing the in-plane vortex force as the Lamb-vector integral, which isolates the left/right turbulent-stress asymmetry as the only contributor.","core_discovery":"For all 48 combinations of Reynolds number and angle of attack considered, the boundary layer of the inclined 6:1 prolate spheroid separates symmetrically and its leeward recirculation belongs to one of three categories: proto-vortex, coherent vortex, or recirculating wake. In the coherent-vortex state, the separated shear layer rolls into a three-dimensional vortex aligned with the spheroid axis, whose center is a pressure minimum and which converts azimuthal momentum into axial momentum. Along the axis, the total recirculation area grows as $A_t \\sim x^2$ while the total circulation grows roughly as $\\Gamma_t \\sim x$, with the exponent depending on angle of attack. The paper further shows that the normal force and pitching moment follow from this axial evolution: suction is highest where swirl and vortex stretching are greatest, not where the recirculation is largest, and the overturning moment comes from the first half of the body.","pith_inferences":["Editorial inference: the same three-state classification and area/circulation scaling may apply to other slender axisymmetric bodies at incidence; a quick test would be to compute $A_t(x)$ and $\\Gamma_t(x)$ for a 3:1 ellipsoid or a submarine-like hull.","Editorial inference: the finding that suction tracks swirl rather than recirculation size suggests a control-surface design rule: to maximize side force, promote early vortex inception and high swirl, not a large recirculation bubble.","Editorial inference: because the paper's boundary-layer state is produced by an inherited trip, a transition-resolving simulation at Re=1.5M-2.5M would either confirm or shift the state boundaries and the inferred critical Reynolds number."],"forward_implications":["At fixed Reynolds number and incidence, the lee-side suction and lift are larger on the forward half of the spheroid, which produces the overturning pitching moment despite the body's nose-tail symmetry.","Increasing angle of attack raises normal force up to about 70 degrees; beyond that the vortex loses coherence and suction drops.","Increasing Reynolds number delays separation and shrinks the recirculation, lowering lift at a given incidence.","The quadratic area growth with near-linear circulation growth implies a decreasing mean axial velocity and swirl along the vortex, consistent with the observed decay of vortex strength toward the tail.","The three-state classification (proto-vortex, coherent vortex, recirculating wake) is exhaustive for the studied range, so a simulation or experiment at any new point in this Re-alpha range should land in one of these states."],"supporting_citations":[{"why":"Supplies the stagnation-pressure iso-surface method used to define the vortex boundary and compute its circulation and area.","marker":"Plasseraud & Mahesh (2024a)"},{"why":"Provides detailed experimental measurements of the primary vortex and boundary layer that the coherent-vortex description extends.","marker":"Chesnakas & Simpson (1997)"},{"why":"Establishes the experimental trend of separation location and vortex circulation versus Re/alpha that the simulations reproduce.","marker":"Fu et al. (1994)"},{"why":"Defines the experimental critical Reynolds-number regime (2M-3M) against which the paper's Re-dependence is compared.","marker":"Ahn (1992)"},{"why":"Provides DNS of crossflow past the spheroid and the linear growth of separation length used for the wake comparison.","marker":"El Khoury et al. (2012)"},{"why":"Supplies the Riabouchinsky cavity model connecting shear-layer stress to wall suction used to explain loads.","marker":"Roshko (1993)"},{"why":"Is the prior trip-resolved LES baseline whose trips and boundary-layer state are inherited in the present simulations.","marker":"Plasseraud et al. (2023)"},{"why":"Provides the reconstruction-velocity definition of boundary-layer thickness (delta_99) used for separation characterization.","marker":"Griffin et al. (2021)"}],"fun_headline_variants":["Three lee-side vortex states govern all 48 spheroid flows","48 simulations reveal three universal lee vortex states","Three states explain all lee vortices on 6:1 spheroid","Spheroid lee flow always one of three vortex states"],"cache_read_input_tokens":25984,"weakest_assumption_plain":"The boundary-layer state at separation is the paper's explanatory variable for topology and Re-dependence, but that state is produced by dynamic-Smagorinsky LES with a trip inherited from prior work, not by resolving or modeling laminar-to-turbulent transition; if the LES misplaces transition, the separation lines, state boundaries, and Re-dependence of loads are all affected.","fun_headline_variants_meta":{"raw":{"variants":["Three lee-side vortex states govern all 48 spheroid flows","48 simulations reveal three universal lee vortex states","Three states explain all lee vortices on 6:1 spheroid","Spheroid lee flow always one of three vortex states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000873,"raw_usage":{"total_tokens":3834,"prompt_tokens":1058,"completion_tokens":2776,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":674,"completion_tokens_details":{"reasoning_tokens":2702}},"tokens_in":674,"tokens_out":2776,"duration_ms":21003,"temperature":1.0,"reasoning_tokens":2702,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:20:14.163474+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A wind-tunnel test on a 6:1 prolate spheroid at Re=1.5M and 2.5M, alpha=20 and 40 degrees, measuring the azimuthal separation line and the presence or absence of a coherent lee vortex: the paper predicts a sharp boundary-layer-state jump between those Reynolds numbers, so if the separation-line shift or vortex state is identical at both, the central Re-dependence claim is wrong.","supporting_citations":[{"cited_title":"1992 An experimental study of flow over a 6 to 1 prolate spheroid at incidence","cited_arxiv_id":null,"evidence_quote":"Defines the experimental critical Reynolds-number regime (2M-3M) against which the paper's Re-dependence is compared."},{"cited_title":"Journal of Fluid Mechanics 960, A3","cited_arxiv_id":null,"evidence_quote":"Is the prior trip-resolved LES baseline whose trips and boundary-layer state are inherited in the present simulations."},{"cited_title":"P., Fu, L","cited_arxiv_id":null,"evidence_quote":"Provides the reconstruction-velocity definition of boundary-layer thickness (delta_99) used for separation characterization."}],"review_version":1}