{"id":"71d845de-cf0e-4fa2-9006-aefdfa781e57","arxiv_id":"1908.03540","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A validated LES is shown to complement MRI measurements of an inclined jet in crossflow, supplying near-wall and turbulence data at three velocity ratios.","lead":"This paper pairs 3D magnetic resonance measurements of an inclined jet in crossflow with large eddy simulations, validating the simulation against the experiment and then using the simulation to fill in data the experiment cannot see. If the approach works, future MRI studies can obtain near-wall and turbulence statistics in complex flows without new measurement hardware.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The enrichment claim rests on unvalidated extrapolation: mean-field agreement is used to certify near-wall and turbulent statistics that are never directly checked.","rationale":"The reader correctly identifies the weakest assumption: mean-flow agreement over the validated domain is used to infer reliability of near-wall and turbulent quantities that are never directly checked. This is genuinely load-bearing because those quantities are the entire point of the enrichment. The paper deserves credit for long averaging times, a coarse/fine mesh comparison, extensive 3D validation of mean fields, and explicit admission of many limitations. Those strengths make the simulation plausible but do not close the epistemic gap. A simulation can match mean momentum and scalar fields while misrepresenting Reynolds stresses or scalar fluxes, especially when the SGS model, inlet fluctuation energy, and the scalar Schmidt number are not independently constrained. The circulation overprediction in Fig. 17 is an additional reminder that the validation is not perfect. I would therefore not reject the paper; it is a useful method demonstration. The verdict should be conditional on either adding a direct check of at least one enriched quantity or explicitly softening the claim from 'validated' to 'mean-flow-validated simulation data.' This is a modest adjustment, not a rejection.","tokens_in":16770,"tokens_out":6018,"duration_ms":73912,"concrete_test":"Perform independent measurements of at least one turbulence statistic in the same geometry at matched Reynolds number, e.g., stereo-PIV for u'v' (and PLIF for v'c') in the x/D = 2 and x/D = 5 planes, and compare the profiles directly with the LES. If the LES deviates by more than the combined experimental and statistical uncertainties, the mean-field validation is insufficient to support the enrichment claim; if it agrees within uncertainty, the concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is that because the LES matches MRV/MRC mean fields in 3D, it can be trusted for near-wall values and turbulent statistics. The load-bearing premise is an extrapolation: agreement in resolved mean velocity and concentration is treated as evidence for quantities the experiments cannot measure. The paper states this explicitly for the in-hole flow: \"Given that the LES matches the experiments elsewhere, this is a region where the LES results are more trustworthy than the MRI data\" (Sec. 4.2). The same logic underlies the wall concentration results (Fig. 11) and the turbulence statistics presented in Sec. 4.3. No independent measurement constrains any of these enrichment products. The circulation comparison (Fig. 17b) shows a systematic LES overprediction of the experimental values, so the validation is imperfect even for mean-derived metrics. A second specific weak point is the scalar Schmidt number: the LES sets Sc = 1 and the paper notes that molecular diffusion \"might compete with turbulent mixing\" extremely close to the bottom wall, adding \"there are no scalar data to be matched there\" (Sec. 3.1). This is exactly the near-wall enrichment region, so the scalar wall value is sensitive to a modeling choice that is not validated. The enriched quantities are plausible, but the Conclusion's phrase \"validated set of simulations\" overstates what the validation actually supports.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a methodology to use highly resolved Large Eddy Simulations (LES) to complement and enrich three-dimensional mean flow measurements obtained by Magnetic Resonance Velocimetry (MRV) and Magnetic Resonance Concentration (MRC). The approach is demonstrated on a circular, 30-degree inclined jet in crossflow at velocity ratios r = 1, 1.5, and 2. The LES models the same geometry and flow conditions as the experiments, including the plenum and hole, and uses an iterative procedure to set inlet conditions that reproduce hot-wire profiles of the incoming boundary layer. The numerical meshes are highly refined (40-48 million cells), the subgrid-scale viscosity is small, and a coarse/fine mesh convergence study is performed. A validation campaign against the MRV/MRC mean fields is carried out using qualitative contour comparisons, one-dimensional profiles, and an integral circulation metric with misalignment-based error bars. After validation, the paper presents enrichment products that MRI cannot provide directly: wall concentration (adiabatic effectiveness), in-hole velocity, and turbulence statistics such as Reynolds stresses and turbulent scalar fluxes.","tokens_in":17079,"tokens_out":4395,"duration_ms":45662,"significance":"The proposed enrichment strategy is timely and potentially very useful: it exploits the full 3D volumetric validation that MRI uniquely offers, and then uses the validated LES to supply near-wall and second-moment data that MRI cannot measure. The paper demonstrates unusual care in matching experimental and numerical conditions: the iterative inlet-condition generation, long averaging times with a convergence check, mesh convergence study, and explicit propagation of misalignment uncertainty into the circulation comparison are all commendable and set a high standard for this type of hybrid experimental-numerical dataset. If the extrapolation to unmeasured quantities is accepted, the resulting datasets would be valuable for turbulence model development in film-cooling flows. The main risk is that this extrapolation rests entirely on agreement of mean fields; the paper should either strengthen that link or qualify the claims accordingly.","major_comments":[{"comment":"The abstract and conclusion describe the LES as a 'validated set of simulations,' but the validation evidence is confined to mean velocity and mean scalar fields. The turbulence statistics in Sec. 4.3 (u'v' and v'c') and the near-wall concentration in Fig. 11 are enrichment products that are not checked against any independent measurement. The argument that agreement in resolvable mean 3D fields justifies trust in unmeasured fluctuation and near-wall quantities is plausible but not automatic; a simulation can match first moments while misrepresenting second moments. I recommend either adding a quantitative check of at least one turbulent correlation (e.g., using hot-wire or PIV in a representative plane) or explicitly labeling the turbulence and near-wall results as 'LES predictions not directly validated' and softening the wording in the abstract and conclusion.","section":"Abstract and Sec. 5 (Conclusion); Sec. 4.3"},{"comment":"The text states that setting the molecular Schmidt number to unity instead of the true value (of order 1000 for copper sulfate in water) has no practical effect because molecular diffusion is negligible compared to turbulent mixing, 'except extremely close to the bottom wall, where molecular diffusion might compete with turbulent mixing.' This caveat coincides exactly with the wall concentration (adiabatic effectiveness) presented as an enrichment result in Fig. 11, and the paper concedes 'there are no scalar data to be matched there.' Because the near-wall scalar value is sensitive to an unvalidated modeling choice, the authors should quantify the sensitivity (for example, by running the r = 1 case with Sc = 1000, or by a boundary-layer scaling argument) or else remove the wall concentration from the set of validated enrichment outputs.","section":"Sec. 3.1, Eq. (5), and Fig. 11"},{"comment":"The circulation comparison shows a consistent LES overprediction for all three velocity ratios, in several streamwise locations exceeding the 2-sigma misalignment error bars. This is the only quantitative integral metric used to validate the mean velocity field, so the systematic bias is a substantive discrepancy rather than minor scatter. The manuscript attributes it to 'other experimental uncertainties... or possible differences in the in-hole flow,' but no supporting evidence is provided. I ask the authors to investigate this bias (e.g., through a sensitivity study of the inlet-condition scaling factors or of the integration contour location) and to revise the phrase 'excellent agreement' in the conclusion so that it reflects this boundary of the validation.","section":"Fig. 17(b)"}],"minor_comments":[{"comment":"The word 'Velocimety' in the first sentence is a typo and should be 'Velocimetry.'","section":"Abstract"},{"comment":"The quantity SNR in Eq. (2) is not defined there; please define it explicitly as the signal-to-noise ratio measured as described in the text, ideally with the same symbol and definition used in the uncertainty estimate.","section":"Sec. 2.1, Eq. (2)"},{"comment":"The phrase 'perceived time averages' is ambiguous; please replace it with 'running time averages' to clarify that the statistics accumulated up to each time step are compared with the final average.","section":"Fig. 8 caption"},{"comment":"The text and figure legend use 'MRI data' while elsewhere the specific acronym 'MRV' is used; please be consistent throughout the manuscript.","section":"Sec. 4.2, Fig. 17 caption"}],"recommendation":"major_revision","confidential_remarks":"This is a strong paper that deserves publication, but the central claim of a 'validated set of simulations' goes beyond what the evidence supports. The authors should either add a direct check of a turbulent quantity or carefully recast the enrichment products as predictions that are plausible but not directly validated. The Sc = 1 sensitivity is a concrete, fixable issue that should be addressed. I would not reject the manuscript; it makes a genuine methodological contribution and the experimental/numerical effort is impressive."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this if you work with MRV/MRC or film cooling datasets. The genuinely new part is the workflow, not the physics: for r=1.5 and r=2 (plus re-presented r=1) they run well-resolved LES of the same geometry, document an iterative inlet-condition procedure using hot-wire profiles and a developing channel LES, and validate against 3D MRV/MRC data using qualitative contours, 1D profiles, and a circulation integral with misalignment error bars. The mesh convergence study and 440 D/Uc averaging are serious; subgrid viscosity below molecular viscosity in the interaction region supports the claim that the simulations are effectively very well resolved. This is a useful recipe for the MRI fluid mechanics community.\n\nThe validation is multi-faceted and the paper is transparent about its imperfections. The circulation comparison shows the LES consistently overpredicts, sometimes beyond the misalignment error bars; the abstract's \"excellent agreement\" is a bit generous, though the trends—velocity-ratio scaling, streamwise decay, asymmetry direction—are captured. In-hole flow agreement is qualitative at best because MRV is unreliable there. The citation pattern is fine; they credit Coletti et al. and Bodart et al. for the r=1 work.\n\nThe real soft spot is exactly what the stress-test note says. The enrichment products—wall concentration, in-hole velocity, Reynolds stresses, scalar fluxes—rest on an extrapolation from validated mean fields to unvalidated near-wall and second-moment quantities. The paper states this directly in Sec. 4.2 (\"Given that the LES matches the experiments elsewhere...\"), and the Sc=1 caveat for near-wall scalar is a fair example: the wall concentration is precisely the region where the modeling choice is least constrained by data. I do not think this is a load-bearing flaw for the paper's actual claim, which is a method demonstration, not a claim that the turbulence data are experimentally certified. But \"validated set of simulations\" in the Conclusion overstates it; \"validated mean flow, plausible enrichment\" is closer.\n\nMinor: the solver is proprietary and data are on request, which limits reproducibility. That lowers the reusability score but does not undercut the method demonstration.\n\nWould I send this to referees? Yes. It is carefully documented, the comparisons are quantitative, and the limitations are declared. A referee should push on the extrapolation claim and ask for any direct check of turbulence quantities if available, but the paper deserves serious review. If you are in the niche, cite it; if you are outside, it is still a clean example of LES-experiment validation.","headline":"A solid, honest method demonstration: validated LES enriches MRI mean fields for an inclined jet in crossflow at three velocity ratios; the main caveat is that near-wall and turbulence quantities are inferred, not directly validated, and the paper knows this.","tokens_in":17622,"tokens_out":3243,"would_cite":true,"duration_ms":32368,"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":"LES enriches MRI flow data with near-wall turbulence","keywords":["Magnetic Resonance Velocimetry","Magnetic Resonance Concentration","Large Eddy Simulation","jet in crossflow","film cooling","turbulent statistics","near-wall flow","inlet conditions"],"falsifier":"A direct test would measure near-wall velocity and concentration and turbulent fluctuations in the same geometry using an independent technique such as particle image velocimetry or a hot-wire probe, and check whether the LES agrees there as closely as it does for the mean fields away from the wall; systematic disagreement in those quantities while mean fields agree would invalidate the enrichment claim.","tokens_in":16618,"feed_emoji":"🌀","tokens_out":5200,"duration_ms":47170,"temperature":0.7,"pith_summary":"MRI-based velocimetry and concentration measurements map three-dimensional time-averaged flows but cannot see turbulent fluctuations or values at solid walls. This paper argues that highly resolved large-eddy simulations, matched to the same geometry and inlet conditions, can supply those missing data, provided the mean fields agree with MRI throughout the three-dimensional domain. To establish this, the authors simulate an inclined jet in crossflow at three velocity ratios and validate against MRV/MRC measurements. The central claim is that the near-wall and turbulence data from the validated LES can be treated as experimental data, enriching the MRI datasets.","feed_headline":"LES supplies the turbulence MRI cannot see","feed_subtitle":"Validated simulation adds near-wall and fluctuation data to experimental mean fields.","key_machinery":"The central mechanism is the enrichment loop: a fine-mesh LES whose subgrid-scale viscosity is negligible almost everywhere (so it approaches DNS resolution) but which, unlike DNS, is feasible for this geometry; an iterative inlet-condition procedure that tunes a synthetic turbulence generator against hot-wire data until a channel-flow LES matches the measured boundary-layer profile and momentum thickness; and validation against the three-dimensional MRI fields as a whole. The load-bearing identity is the matching of mean velocity and mean concentration between LES and MRV/MRC in the region where both are reliable; agreement there is taken as evidence that the LES is a physically realistic representation of the flow. Quantitative comparisons use integrated quantities, such as the circulation $\\Gamma$ of the counter-rotating vortex pair, which are robust to MRI noise and misalignment.","core_discovery":"The paper demonstrates that, when the simulation domain and inlet conditions are carefully matched to the experiment, large-eddy simulation reproduces the three-dimensional mean velocity and scalar fields of Magnetic Resonance Velocimetry and Concentration closely enough that the simulation can be trusted where MRI is blind. The validated LES then provides wall concentration (adiabatic effectiveness), in-hole velocity, and turbulent correlations such as Reynolds stresses and scalar fluxes that the experiments cannot measure. The validation is grounded in the full 3D fields rather than single-point comparisons; quantitative metrics like the circulation of the counter-rotating vortex pair are computed independently from both datasets. The simulations reveal that in this geometry the $r=1$ jet reattaches to the wall after injection while $r=1.5$ and $r=2$ remain detached, and that the short $4.1D$ feed hole produces a strongly non-uniform in-hole velocity with a separation bubble, rather than developed pipe flow.","pith_inferences":["The same enrichment approach could be extended to quantities MRI cannot provide at all, such as wall shear stress or pressure, but each would need its own validation target since mean-field agreement does not guarantee derivative quantities.","A natural testable extension is to run the same matched LES/MRI protocol with an added planar PIV or LDV check of turbulent statistics in one region; if fluctuations also match, the claim that simulations can be treated as data becomes much stronger.","The systematic overprediction of circulation suggests the in-hole flow or its coupling to the crossflow differs subtly between simulation and experiment; quantifying the sensitivity to plenum feed details would bound the enrichment's uncertainty.","The weakly positive $u'v'$ and $v'c'$ under the jet where mean gradients are positive is a concrete signature of gradient-diffusion model failure, giving turbulence modelers a targeted test for new closures."],"forward_implications":["For this geometry, the $r=1$ jet detaches briefly and reattaches by $x/D\\approx3$, while $r=1.5$ and $r=2$ remain detached, placing the reattachment threshold between $r=1$ and $r=1.5$.","The LES supplies wall concentration (adiabatic effectiveness) that MRC cannot measure reliably, including the sharp drop after injection and the partial recovery for $r=1$.","In-hole flow is far from fully developed: a separation bubble on the plenum-side corner causes high axial velocity on the opposite side and secondary flows up to about 20% of the bulk velocity, with little change across $r$ in this range.","Turbulent statistics from the LES, such as the Reynolds stress $u'v'$ and scalar flux $v'c'$, are now part of the dataset for data-driven turbulence modeling in film cooling flows.","The counter-rotating vortex pair circulation in the LES follows the experimental trends but is consistently overpredicted, occasionally beyond misalignment-induced uncertainty."],"supporting_citations":[{"why":"Establishes MRV as the experimental method that supplies the 3D mean velocity fields being enriched.","marker":"[1]"},{"why":"Establishes MRC as the experimental method for the mean concentration fields being enriched.","marker":"[2]"},{"why":"Provides the r=1 experimental dataset and analysis that this work extends to r=1.5 and r=2.","marker":"[14]"},{"why":"Supplies the subgrid-scale eddy-viscosity model used in the LES momentum equations.","marker":"[22]"},{"why":"Prior similar simulation whose averaging time is shown to be too short for converged scalar flux statistics, justifying the longer averaging used here.","marker":"[26]"},{"why":"Provides the synthetic inflow turbulence generation method used to set time-varying inlet conditions for the LES.","marker":"[28]"}],"fun_headline_variants":["LES fills in turbulence details MRI cannot capture","Validated LES enriches MRI flows with near-wall stats","Pairing LES with MRI reveals flow blind spots in jets","LES adds fluctuation data to MRI mean flow fields"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on the premise that close agreement in time-averaged velocity and concentration throughout the measured three-dimensional domain implies the simulation is also accurate for near-wall values and turbulent correlations, which are never directly checked against experimental data.","fun_headline_variants_meta":{"raw":{"variants":["LES fills in turbulence details MRI cannot capture","Validated LES enriches MRI flows with near-wall stats","Pairing LES with MRI reveals flow blind spots in jets","LES adds fluctuation data to MRI mean flow fields"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000537,"raw_usage":{"total_tokens":2589,"prompt_tokens":968,"completion_tokens":1621,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":584,"completion_tokens_details":{"reasoning_tokens":1559}},"tokens_in":584,"tokens_out":1621,"duration_ms":12467,"temperature":1.0,"reasoning_tokens":1559,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:08:51.991599+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would measure near-wall velocity and concentration and turbulent fluctuations in the same geometry using an independent technique such as particle image velocimetry or a hot-wire probe, and check whether the LES agrees there as closely as it does for the mean fields away from the wall; systematic disagreement in those quantities while mean fields agree would invalidate the enrichment claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes MRV as the experimental method that supplies the 3D mean velocity fields being enriched."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes MRC as the experimental method for the mean concentration fields being enriched."},{"cited_title":"Coletti, M","cited_arxiv_id":null,"evidence_quote":"Provides the r=1 experimental dataset and analysis that this work extends to r=1.5 and r=2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the subgrid-scale eddy-viscosity model used in the LES momentum equations."},{"cited_title":"Bodart, F","cited_arxiv_id":null,"evidence_quote":"Prior similar simulation whose averaging time is shown to be too short for converged scalar flux statistics, justifying the longer averaging used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the synthetic inflow turbulence generation method used to set time-varying inlet conditions for the LES."}],"review_version":1}