{"id":"c44151d2-5f3f-44a2-86a6-fc80f9878fa7","arxiv_id":"2501.01753","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A streak-angle visualization technique with a calibrated standard-deviation threshold and KL-divergence analysis distinguishes laminar, transitional, and turbulent states in particulate pipe flow.","lead":"This paper introduces a low-cost laser and camera method that detects whether flow in a particle-carrying pipe is laminar, transitional, or turbulent by measuring how much particle streaks angle away from the pipe axis. It could give labs and industry a cheap way to check flow regimes where expensive PIV systems are impractical.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Critical-Reynolds estimate rests on asymmetric KL divergences and unspecified histogram estimators; the claimed crossover may be an artifact of the analysis rather than a flow transition.","rationale":"I read the paper as proposing an accessible streak-angle classifier with two related claims: a standard-deviation threshold for regime discrimination and a KL-divergence crossover for the critical Reynolds number. The first claim is supported by agreement with PIV and pressure-drop data at six Reynolds numbers, though the 0.04 threshold is calibrated from only two reference states. The second claim is less supported. Section 5.3 neither specifies histogram construction nor acknowledges KL asymmetry, and it references a Re=4500 case that is outside the range stated in Section 2. The crossover is exactly the kind of quantity that finite-sample plug-in estimators and asymmetric divergences can distort, so the strongest antecedent of consistency with previous studies is not established. This does not invalidate the entire paper; it means the novel critical-Re result should be conditional on an estimator/reference-robustness check, while the classification claim can stand with its existing validation. I partially agree with the reader's weakest-assumption: the threshold generalization worry is real, but the most load-bearing gap is the unvalidated KL computation behind the critical-Reynolds claim.","tokens_in":16445,"tokens_out":4166,"duration_ms":44347,"concrete_test":"Recompute Fig. 8 from the raw streak-angle data with a fixed histogram rule (e.g., Scott's rule) and with kernel density estimates; compute both DKL(P||Q) and DKL(Q||P), the Jensen-Shannon divergence, and bootstrap 95% confidence intervals for the crossover Reynolds number. Also verify that a Re=4500 reference run exists. If the crossover shifts by more than about 200 in Re or depends on divergence direction or binning, the critical-Reynolds claim should be withdrawn or substantially qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5.3's estimate of the critical Reynolds number is the least secure part of the central claim. The KL divergence in Eq. (6) is computed from binned streak-angle histograms, but the paper never specifies the bin width, the number of frames or particles per distribution, or how zero-count bins are handled. Plug-in KL estimates are biased for finite samples and vary with binning, so the curves in Fig. 8 could shift. More fundamentally, DKL(P||Q) is asymmetric: the crossover between DKL(·||laminar) and DKL(·||turbulent) is not a reference-independent measure of distance to either regime. If the references are swapped, or if a symmetric divergence such as Jensen-Shannon is used, the intersection can move. The turbulent reference at Re=4500 also lies outside the stated measurement range [1120,2980] and no supporting run is described. Without uncertainty quantification or a check of estimator/reference sensitivity, the claim that the crossover gives a critical Reynolds number consistent with previous studies is not yet supported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a low-cost streak-visualization method for detecting laminar, transitional, and turbulent regimes in dilute particulate pipe flow. Particle streaks are recorded with a laser-and-camera setup, processed by background subtraction, adaptive thresholding, Canny edge detection, and a probabilistic Hough transform, and characterized by the distribution of streak angles relative to the pipe axis. The standard deviation of streak angles over a five-frame moving window is used to classify flow regimes against a reference threshold, and the Kullback-Leibler divergence between observed angle distributions and reference laminar/turbulent distributions is used to estimate the critical Reynolds number. The method is validated against Particle Image Velocimetry (PIV) and pressure-drop friction-factor measurements for Reynolds numbers between 1120 and 2980 at a fixed volume fraction and particle size range.","tokens_in":16640,"tokens_out":3884,"duration_ms":41896,"significance":"If fully supported, the paper would offer a simple, inexpensive tool for regime detection in particulate pipe flows at low particle concentration, where established techniques such as PIV, LDV, and UIV are costly, intrusive, or ineffective. The use of two independent external benchmarks, PIV and pressure-drop friction factor, is a genuine strength, as is the explicit description of the image-processing pipeline and its parameters. However, the central classification claim currently rests on a calibration step whose reference runs are only partially documented, and the critical-Reynolds-number claim depends on an underspecified, reference-dependent KL-divergence estimator. The significance of the paper would be materially increased by resolving these issues.","major_comments":[{"comment":"The KL-divergence calculation is not reproducible as reported. Equation (6) defines D_KL for continuous distributions, but the actual computation is performed on binned histograms of streak angles, and the bin width, the number of frames or particles used to form each distribution, and the treatment of zero-count bins are never stated. Plug-in KL estimates are known to be biased and binning-dependent, so without this information the curves in Fig. 8 could shift substantially. In addition, D_KL(P||Q) is asymmetric, so the crossover between D_KL(·||laminar) and D_KL(·||turbulent) is not a reference-independent measure of distance to either regime; swapping the references or using a symmetric divergence such as Jensen-Shannon could move the intersection. The claim that the crossover gives a critical Reynolds number consistent with previous studies is therefore not yet supported and needs either a precise estimator definition plus sensitivity checks or a reformulation of the criterion.","section":"Section 5.3, Eq. (6), Fig. 8"},{"comment":"The classification threshold σ = 0.04 is calibrated from reference runs at Re = 1120 and Re = 7500, but the stated Reynolds-number range of the experiments is [1120, 2980], and no details of the Re = 7500 run are given. Moreover, Section 5.3 and Fig. 8 refer to a turbulent reference at Re = 4500, which is inconsistent with the Re = 7500 reference mentioned in Section 5.2 and is also outside the stated measurement range. The threshold is presented as a red line with no uncertainty quantification or sensitivity analysis, even though the classification of intermediate cases such as Re = 1980 depends directly on it. The authors should document both reference runs and provide at least a simple sensitivity check of the threshold.","section":"Section 5.2, Fig. 7"},{"comment":"The experimental setup section states that two particle diameter ranges are utilized, 425–500 µm and 212–250 µm, but every reported measurement in Table 3 and Figs. 4–8 is for the 212–250 µm range only. No data are presented for the larger particles, so the abstract's and conclusion's implicit generalization of the method across particle sizes is unsupported by the current results. Either include experiments with the second particle size or explicitly restrict the claims to the tested range.","section":"Section 2 and all experimental results"}],"minor_comments":[{"comment":"The notation in Eq. (5) uses \\bar{\\theta}_i, but the denominator and the surrounding text suggest a single global mean \\bar{\\theta}; please make the indexing consistent. Also, the y-axis label in all panels of Fig. 7 reads \"< (radian)\", which appears to be a typo for the standard deviation σ (in radians).","section":"Section 5.2, Eq. (4)-(5), Fig. 7"},{"comment":"The caption of Fig. 3(f) says \"showing the two lines counted as one per actual streak,\" but the main text states that Canny edge detection followed by Hough transform is used precisely to avoid multiple lines per streak; please clarify what the caption means.","section":"Section 3, Fig. 3 caption"},{"comment":"The PIV and streak-visualization flow states in Table 3 are identical by construction of the table, but the two systems are located 4.5 m apart and puffs can grow or decay between them. Please state how the comparison was synchronized and whether the pressure-drop data were used to resolve any ambiguity in evolving puffs.","section":"Section 4.1, Table 3"},{"comment":"The claim that the method is especially efficient precisely where other methods are less effective at low particle concentration is not demonstrated, since only a single volume fraction Φ = 1.2×10^-3 is tested and no comparison with another method at that concentration is shown.","section":"Abstract and Conclusion"},{"comment":"There are several typographical errors that should be corrected in revision, including \"intantaneous\" (Section 5.1), \"deteting\" (Introduction), \"rapdidly\" (Introduction), and \"the the flow\" (Section 4.2).","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The core experimental method is promising and the PIV/pressure-drop validation is a genuine strength, but the KL-divergence section is the least secure part of the paper and currently reads as an underspecified post-hoc analysis rather than a demonstrated quantitative tool. I would encourage the editor to request a revision that either supplies the missing estimator and reference-run documentation with sensitivity checks or scales back the critical-Reynolds-number claim. The manuscript is within the scope of the journal and does not raise concerns about attribution or improper citation practice; the main issue is technical completeness."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful core here is genuinely simple: in a dilute particulate pipe flow, the standard deviation of particle streak angles separates laminar, transitional, and turbulent states, and it is checked against two independent diagnostics (PIV and pressure drop). That validation is real. The agreement in Table 3 and the friction-factor comparison in Fig. 4 give me confidence that the basic classification claim is not a curve-fit artifact. The method is cheap and easy to implement, which is a legitimate practical niche given how cumbersome PIV and LDV are for this kind of flow. The paper also deserves credit for showing that a puff at Re=1980 produces clear intermittent peaks in sigma, and for reporting the comparison at two perturbation strengths.\n\nWhere the paper is soft, in increasing order: First, the sigma=0.04 threshold is calibrated from two runs on the same rig, and the authors give no uncertainty on sigma or on the threshold. That means the method is a reasonable laboratory classifier, but the abstract's language about reliability for flowmeter calibration overstates what is demonstrated. Second, the claim about low-concentration advantage is asserted but not actually shown against any competing method at low concentration; it is plausible, not proven. Third, and most important, the KL-divergence critical Reynolds number in Section 5.3 is the weakest part. The histogram binning is unspecified, the estimator for discrete KL is not given, the turbulent reference is at Re=4500 even though the measurement range is stated as [1120,2980], and the asymmetry of DKL means the crossover between DKL(·||laminar) and DKL(·||turbulent) is not a reference-independent distance. The stress-test note is right: the crossover could shift with binning or with a symmetric divergence like Jensen-Shannon. So I would treat the critical Reynolds number as suggestive, not established.\n\nI also agree with the reader that the paper is not circular: the PIV and pressure-drop checks provide external grounding, and the calibration issue is a limitation, not a fatal flaw. The writing needs polishing (several typos, some section numbering errors), but nothing in the argument is incoherent.\n\nWho should read this: experimentalists working on transition in particle-laden pipe flow, especially people who need a cheap regime indicator in transparent fluids. It deserves a serious referee, but the referees should push hard on the KL analysis and on qualifying the abstract's claims.","headline":"A simple, low-cost streak-angle classifier that mostly works as advertised, but the KL-based critical Reynolds estimate is not yet solid enough to be the paper's headline.","tokens_in":17146,"tokens_out":988,"would_cite":true,"duration_ms":12952,"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":"Particle streak angles, measured with a low-cost laser and camera, classify laminar, transitional, and turbulent particulate pipe flow.","keywords":["laser-camera setup","streak visualization","turbulence detection","particulate pipe flow","laminar-turbulent transition","particle streak angles","Kullback-Leibler divergence","critical Reynolds number"],"falsifier":"Set up the same rig at Re = 1530 with the larger 425–500 µm particles and compute the five-frame $\\sigma$ from the streak images; if $\\sigma$ routinely exceeds 0.04 while simultaneous PIV and pressure-drop measurements indicate laminar flow, the calibrated threshold does not generalise. A broader version scans the whole Re range for any condition where $\\sigma>0.04$ coincides with a laminar PIV profile and laminar friction factor.","tokens_in":16206,"feed_emoji":"📷","tokens_out":9418,"duration_ms":82970,"temperature":0.7,"pith_summary":"This paper contends that the statistical spread of particle streak angles is enough to tell laminar, transitional, and turbulent regimes apart in a particulate pipe flow. Using a 50 mW laser sheet and an ordinary camera, it records short-exposure streaks of neutrally buoyant particles (Stokes number 0.2–0.6) and measures each streak's angle relative to the pipe axis. Laminar flow gives angles clustered near zero; turbulent flow gives a wide spread. The paper shows that the standard deviation of these angles, computed over a moving five-frame window, separates the regimes with a threshold of 0.04 and that transient turbulent puffs appear as clear peaks; it then uses the Kullback-Leibler divergence between measured and reference angle distributions to locate the critical Reynolds number. The claim matters because the method works at low particle concentrations, where PIV, LDV, and similar techniques lose reliability, and it can be built from inexpensive parts.","feed_headline":"Particle streaks tell laminar from turbulent pipe flow","feed_subtitle":"A single laser and camera measure streak-angle spread to classify the flow and locate the transition point.","key_machinery":"The load-bearing object is the distribution of streak angles $\\theta_i$ measured by the Hough transform on background-subtracted, thresholded camera frames. From each five-frame window the paper forms the sample standard deviation $\\sigma$ and uses $\\sigma=0.04$, calibrated from reference laminar and turbulent runs, as the regime threshold. The second piece of machinery is the Kullback-Leibler divergence $D_{KL}(P\\parallel Q)=\\int p(x)\\log[p(x)/q(x)]\\,dx$ between the current angle distribution and fixed reference distributions; evaluated against both a laminar and a turbulent reference, the point where the two divergence curves cross defines the critical Reynolds number. This two-stage pipeline carries the entire argument: the threshold does the classification, the divergence does the transition-point estimate, and both are validated against independent PIV and pressure-drop measurements.","core_discovery":"The central discovery is that particle trajectories, captured as light streaks, carry a usable one-dimensional signature of the fluid phase's state: the angular dispersion of streaks. For a laminar fluid the streaks are nearly parallel to the pipe axis; for a turbulent fluid they are visibly misaligned, and this difference persists across the Reynolds-number range studied (Re = 1120–2980). Quantifying the dispersion by the standard deviation $\\sigma$ of streak angles over five consecutive frames, the paper establishes a calibrated threshold $\\sigma = 0.04$ that classifies each frame as laminar, transitional (a puff), or turbulent, matching classifications from simultaneous PIV measurements and from friction-factor versus Reynolds-number curves. Going further, the paper computes the Kullback-Leibler divergence $D_{KL}(P\\parallel Q)$ between the measured angle distribution and two reference distributions, one laminar (Re = 1120) and one turbulent (Re = 4500); the crossing of the two divergence curves gives a critical Reynolds number that depends on the perturbation amplitude, with stronger perturbations triggering an earlier transition. This critical value is consistent with the $Re_c \\sim \\epsilon^{-1}$ scaling for particulate pipe flow reported in earlier work.","pith_inferences":["Because the KL-divergence crossover already replaces the hand-set threshold with a data-driven criterion, a natural extension is a fully unsupervised classifier that needs no reference runs; the manual calibration step in the paper is the main obstacle to that.","The method's stated advantage at low concentration suggests a testable boundary: at higher volume fractions, particle-particle collisions may widen streak-angle distributions even in laminar flow, so the 0.04 threshold would need re-calibration or would fail.","Applying the same streak-angle pipeline to the 10 µm tracer particles, rather than the larger inertial particles, would connect the angle statistics directly to fluid-phase velocity fluctuations and could test whether the threshold reflects fluid turbulence or particle inertia."],"forward_implications":["A laboratory can classify particulate pipe flow regimes with a 50 mW laser and a standard camera, without PIV-grade lasers, high-speed cameras, or synchronisation hardware.","At volume fractions around $10^{-3}$, where PIV and ultrasound methods lose sensitivity, the streak-angle classifier still separates laminar, puff, and turbulent states.","The KL-divergence crossover gives a quantitative critical Reynolds number from the same streak images, so no separate velocity-field measurement is needed to locate transition.","Peaks in the five-frame standard deviation signal reveal localized turbulent puffs and their passage, enabling time-resolved monitoring of transitional features."],"supporting_citations":[{"why":"Provides the original rig and the combined PIV/PTV system that the streak setup extends and uses as an independent validation.","marker":"[53]"},{"why":"Supplies the perturbation-intensity scaling $Re_c\\sim\\epsilon^{-1}$ and friction-factor behaviour used to benchmark the measured critical Reynolds number.","marker":"[30]"},{"why":"Establishes the particle-laden transition context and the comparison point for the observed transition around Re ≈ 2260.","marker":"[20]"},{"why":"Defines the 'closer to turbulence than to laminar' criterion whose spirit the KL-divergence crossover implements.","marker":"[15]"},{"why":"Gives the critical-point methodology for pipe-flow transition that the KL-divergence estimate is modelled on.","marker":"[62]"},{"why":"Justifies that the perturbation and measurement stations lie in fully developed flow, a prerequisite for the regime classification.","marker":"[54]"},{"why":"Supplies the probabilistic Hough transform that detects the streak lines from which all angles are measured.","marker":"[59]"}],"fun_headline_variants":["Streak angles expose laminar-turbulent transition in particle pipes","Particle streak spread reveals flow regime in pipe","Simple laser-camera method reads flow state from particle streaks","Streak-angle standard deviation flags turbulent pipe flow","Low-cost streak visualization identifies laminar, turbulent, and transition"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the 0.04 standard-deviation threshold, calibrated from just two reference runs, correctly separates laminar from turbulent behaviour for every Reynolds number, perturbation strength, particle size, and concentration studied here, even though the particles do not follow the fluid exactly.","fun_headline_variants_meta":{"raw":{"variants":["Streak angles expose laminar-turbulent transition in particle pipes","Particle streak spread reveals flow regime in pipe","Simple laser-camera method reads flow state from particle streaks","Streak-angle standard deviation flags turbulent pipe flow","Low-cost streak visualization identifies laminar, turbulent, and transition"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000875,"raw_usage":{"total_tokens":3827,"prompt_tokens":1031,"completion_tokens":2796,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":2716}},"tokens_in":647,"tokens_out":2796,"duration_ms":22554,"temperature":1.0,"reasoning_tokens":2716,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:20:30.951804+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Set up the same rig at Re = 1530 with the larger 425–500 µm particles and compute the five-frame $\\sigma$ from the streak images; if $\\sigma$ routinely exceeds 0.04 while simultaneous PIV and pressure-drop measurements indicate laminar flow, the calibrated threshold does not generalise. A broader version scans the whole Re range for any condition where $\\sigma>0.04$ coincides with a laminar PIV profile and laminar friction factor.","supporting_citations":[{"cited_title":"Review of Scientific Instruments 91(9) (2020)","cited_arxiv_id":null,"evidence_quote":"Provides the original rig and the combined PIV/PTV system that the streak setup extends and uses as an independent validation."},{"cited_title":"Physical Review Fluids 7(4), 042301 (2022)","cited_arxiv_id":null,"evidence_quote":"Supplies the perturbation-intensity scaling $Re_c\\sim\\epsilon^{-1}$ and friction-factor behaviour used to benchmark the measured critical Reynolds number."},{"cited_title":"Physical Review Letters 122(11), 114502 (2019)","cited_arxiv_id":null,"evidence_quote":"Establishes the particle-laden transition context and the comparison point for the observed transition around Re ≈ 2260."},{"cited_title":"Science 333(6039), 192–196 (2011)","cited_arxiv_id":null,"evidence_quote":"Defines the 'closer to turbulence than to laminar' criterion whose spirit the KL-divergence crossover implements."},{"cited_title":"Journal of Fluid Mechanics 839, 76–94 (2018)","cited_arxiv_id":null,"evidence_quote":"Gives the critical-point methodology for pipe-flow transition that the KL-divergence estimate is modelled on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Justifies that the perturbation and measurement stations lie in fully developed flow, a prerequisite for the regime classification."},{"cited_title":"Computer vision and image understanding 78(1), 119–137 (2000)","cited_arxiv_id":null,"evidence_quote":"Supplies the probabilistic Hough transform that detects the streak lines from which all angles are measured."}],"review_version":1}