{"id":"eed2d4f9-3a3a-4973-8dd3-41272bd242d7","arxiv_id":"2507.11657","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Simulations of a non-Newtonian slurry in a semicircular open flume show that particle size distribution, flume tilt, and bubble size and amount all change how fine and coarse sand settles.","lead":"This paper uses 3D computer simulations of a thick, clay-based slurry flowing down an open trough to see how sand grains of different sizes settle to the bottom. The results suggest that the mixture of particle sizes, the trough tilt, and tiny gas bubbles all change how much sand settles, which matters for recycling water from oil-sands tailings.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 21 s simulated time is shorter than the ~35 s advection time to the analysis plane, and no initial-condition or steady-state check is reported; if the flow has not reached the plane, validation and parametric trends are transient artifacts.","rationale":"The reader's weakest assumption, the flat no-shear free-surface lid, is a genuine transferability concern, but the 21 s simulation time is more fundamental: it threatens the validity of the numerical model even inside the simplified domain. The manuscript repeatedly presents results at z = 14.5 m; for Case 1 at 0.41 m/s the advective timescale is about 35 s, longer than the total simulated time. Since no initial condition is described and no steady-state convergence is shown, the agreement in Figs. 4–7 could be a transient crossing rather than a converged solution, and the parametric differences in Figs. 8–17 could be transient artifacts. This is directly checkable by running longer. I agree with the reader's overall CONDITIONAL verdict and with the accompanying list of minor issues, but I weight the time horizon as the single most load-bearing condition; hence partial agreement on the weakest assumption. The verdict remains CONDITIONAL pending the longer-time check.","tokens_in":16089,"tokens_out":5929,"duration_ms":76874,"concrete_test":"Rerun Case 1 and one parametric case (e.g. PSD 3 or inclination 6°) with identical numerical settings, extending the simulation to at least 60 s of physical time while recording the plane-averaged solid volume fraction and centre-line carrier velocity at z = 14.5 m every 1 s. Compare the time-invariant profiles with the t = 21 s profiles used in Figs. 4–7 and the relevant parametric figures. If any point shifts by more than the validation tolerances (±15% volume fraction, ±5% velocity), or if the sign of a bubble/inclination trend reverses, the central claim is not established; if the profiles are unchanged, the short run time is not a material issue.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The most load-bearing gap is not the free-surface lid itself but the simulation horizon. Table II fixes 21,000 time steps at Δt = 0.001 s, i.e. 21 s of physical time. The analysis and validation plane is at z = 14.5 m, and Case 1 bulk velocity is 0.41 m/s, giving an advective time of roughly 35 s to that plane and about 45 s to the outlet. The paper reports no initial condition, no steady-state criterion, and no time-history check. Thus the profiles shown in Figs. 4–7 and all parametric comparisons in Figs. 8–17 may represent a partially developed transient state rather than the quasi-steady open-channel flow used to draw dewatering conclusions. Because every validation claim and every bubble/inclination trend is evaluated at this same fixed t = 21 s, an unacknowledged transient would not just perturb one quantity; it would undermine the central claim that the model reproduces the experiments and that the parametric trends transfer to the physical flume.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a 3D unsteady Eulerian–Eulerian (E–E) simulation of multisize solid particles settling in a non-Newtonian (Bingham, with Herschel–Bulkley n=1) slurry inside a semicircular open flume. The model is validated against Spelay's experimental data for four operating cases, with claims of agreement within ±15% for solid volume fraction and ±5% for carrier velocity. The authors then perform parametric studies varying particle size distribution, flume inclination, bubble size, and bubble volume fraction, concluding that inclination and bubble characteristics can control settling and be used for dewatering design.","tokens_in":16228,"tokens_out":4152,"duration_ms":45438,"significance":"If the model and its parametric predictions are correct, the paper provides a useful engineering guideline for slurry dewatering in open-channel tailings transport, addressing a real industrial problem. The work combines a poly-dispersed granular E–E framework with Herschel–Bulkley rheology, performs a drag-model sensitivity analysis, and compares against a published experimental dataset. The parametric trends (PSD, inclination, bubbles) are potentially actionable. However, the significance is tempered by two load-bearing issues: the simulation horizon is shorter than the advective time to the analysis plane, and the drag model is selected and validated on the same dataset. These issues affect the credibility of both the validation and the parametric conclusions.","major_comments":[{"comment":"The simulation time is limited to 21,000 steps at Δt = 0.001 s, i.e. 21 s of physical time. The analysis plane is at z = 14.5 m and the Case 1 bulk velocity is 0.41 m/s, giving an advective time of roughly 35 s to this plane and about 45 s to the outlet. The paper reports no initial condition, no steady-state criterion, and no time-history check for any monitored quantity. Consequently, the profiles in Figures 4–7 and all parametric comparisons in Figures 8–17 may be evaluated before the flow has reached a quasi-steady state. This is not a minor matter: every validation claim and every bubble/inclination trend is evaluated at this same fixed time, so an unacknowledged transient would undermine the central claim that the model reproduces the experiments and that the parametric trends transfer to the physical flume. The authors should provide time histories of solid volume fraction and velocity at the analysis plane, demonstrate that the flow is converged to a quasi-steady state (or at least invariant over a meaningful interval), and if necessary extend the simulation horizon.","section":"§II.D.2, Table II; §III.A"},{"comment":"The symmetric drag model is selected because it gives the lowest mean absolute error against the Spelay 2007 dataset (Fig. 3(f)), and the same dataset is then used as the validation benchmark in §II.H. This is a selection-on-validation circularity: the reported ±15% (±5%) agreement for solid volume fraction (carrier velocity) is in-sample, not out-of-sample. The drag-model sensitivity study would be much stronger if performed on a hold-out subset of the data or against an independent experimental dataset. As written, the validation claim overstates the predictive power of the model.","section":"§II.F, Fig. 3(f); §II.H, Figs. 4–7"},{"comment":"The physical flume is replaced by a truncated domain of constant flow depth with a flat, zero-shear free-surface lid (Fig. 1(b)), while the real flume has five open sections and a deformable free surface. The paper acknowledges this simplification but does not quantify its impact on the settling conclusions. In particular, the bubble phase accumulates at this artificial lid in the simulations (as indicated in §III.D, where the authors state that 'accumulation of bubbles near the top free surface' occurs), which may alter the predicted bubble–solid interaction and therefore the bubble size and volume fraction trends in Figures 15 and 17. The authors should either assess the sensitivity of the bubble and particle distributions to the free-surface treatment or temper the conclusions about bubble effects on settling.","section":"§II.C, Fig. 1(b); §III.C–D"}],"minor_comments":[{"comment":"Equation (15) is used for two different relations (granular temperature dissipation and kinetic energy transfer), leading to duplicate equation numbers; the second expression should be renumbered.","section":"Equations (15)"},{"comment":"In the text 'Figure 4(e) presents the variation of the solid particle velocity' should be 'Figure 3(e)', since Figure 4 is reserved for the validation comparisons.","section":"§II.F, Fig. 3(e)"},{"comment":"The spelling 'Hershel-Bulkley' should be 'Herschel–Bulkley' throughout.","section":"Abstract; §II.B"},{"comment":"The statement that the model predicts the solid volume fraction within the ±15% uncertainty band for approximately 93% of data is vague; please specify the number of measurement points and the cases included in this statistic.","section":"§II.H, Fig. 5"},{"comment":"The abstract phrase 'the increase in flume inclination progresses the settling and dissipation of fine and coarse particles, respectively' is ambiguous and should be rephrased to match the more precise description in §III.B (fine particles settle more while coarse particles dissipate more with increasing inclination).","section":"§III.B; Abstract"},{"comment":"The wall shear stress range '9.87e-08 to 0.44 Pa' for particles from 75 to 296 µm appears unusually wide and should be checked for unit consistency; if correct, the authors should explain the physical origin of this twelve-order-of-magnitude spread.","section":"§IV, Conclusion"},{"comment":"There are several typographical errors, including 'Futhermore' (§III.B), 'non-newtonian' (Abstract, Introduction), and 'twice of bulk flow velocity' (§II.H) which should read 'twice the bulk flow velocity'.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a relevant industrial problem and the parametric study is potentially useful, but the validation is partly in-sample and the simulation horizon is shorter than the advective time to the analysis plane. These are technical issues, not evidence of misconduct. The fit to the journal's scope is appropriate. I would encourage the authors to address the simulation-time and drag-model circularity concerns thoroughly before the manuscript can be considered for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a competent, clearly written CFD parametric study that extends a known Eulerian-Eulerian modeling program to a semicircular flume with polydisperse solids and bubbles. It earns credit for grid independence, drag-model sensitivity, and an external validation anchor (Spelay's experiments) with about 93% of solid fraction points within ±15% and carrier velocity within ±5%. What's new: the flume geometry, PSD screening, inclination sweep, and bubble size/fraction parametric trends. The paper is honest about the constant-depth lid simplification, though not about all of its consequences.\n\nThe soft spots, in order. First and most important: the simulation time. Table II fixes 21,000 steps at Δt = 0.001 s, i.e. 21 s. The analysis plane is 14.5 m downstream; with a bulk velocity of 0.41 m/s the advective time is about 35 s, and longer for the slower cases. The paper never states the initial condition, shows no time traces, and gives no steady-state criterion. So the profiles in Figs. 4–7 and all parametric trends are at best a snapshot of a possibly developing flow. The agreement with Spelay could be explained by a fortuitous initial condition, but without documentation it is not reproducible. This gap does not destroy the paper but makes the central claim conditional: until the authors show that the plane of interest is in quasi-steady state, the validation and trends are not fully trustworthy.\n\nSecond, the drag closure is chosen by lowest MAE against the same Spelay dataset later used for validation. That's common practice but weakens the independence of the test. A separate holdout case would have been stronger. Minor.\n\nThird, the parametric runs are single simulations with no uncertainty quantification. Given the many modeling constants (restitution, drag model, rheology), the trends in Figs. 8–17 are indicative, not quantitative.\n\nFourth, the bubble phase accumulates at the artificial flat free surface. That is a known limitation of the lid, but the bubble-induced settling trends in Sections III.C–D could be sensitive to this boundary. I'd want a sensitivity test on the free-shear condition before using those trends for dewatering design.\n\nWho this is for: researchers doing CFD of non-Newtonian slurry transport, especially in oil-sands tailings. It's a useful benchmark study, and the drag-model comparison is worth seeing. The paper deserves peer review; the issues are fixable with additional simulations and documentation. If the authors show time-converged results and one holdout validation, it becomes a solid engineering guideline. As is, treat the trends as provisional.","headline":"Useful engineering parametric study, but the missing steady-state check on a 21 s simulation of a 35 s advection problem makes the trends provisional until time convergence is shown.","tokens_in":16882,"tokens_out":2559,"would_cite":false,"duration_ms":30980,"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":"A 3D Eulerian-Eulerian model captures solids settling in a non-Newtonian flume within ±15%, and predicts that particle size distribution, flume tilt, and bubble size and loading are control levers.","keywords":["Eulerian-Eulerian model","Herschel-Bulkley rheology","non-Newtonian slurry","solids settling","open-channel flume","particle size distribution","bubble effects","slurry dewatering"],"falsifier":"Measure the vertical solid-volume-fraction and carrier-velocity profiles at the 14.5 m station of the real 18.5 m flume with its five open sections and free surface exposed, under the four Spelay operating conditions; if more than roughly 7% of points fall outside the model's ±15% solid-fraction band, or velocities deviate beyond ±5% in a way that grows with bubble volume fraction or open-section location, the flat-lid simplification is load-bearing and the parametric conclusions fail.","tokens_in":15800,"feed_emoji":"🫧","tokens_out":8163,"duration_ms":86315,"temperature":0.7,"pith_summary":"Slurry moving through open channels carries solids of many sizes, and knowing where they settle is what makes dewatering of thickened tailings practical. This paper argues that a 3D unsteady Eulerian-Eulerian model, with a Herschel-Bulkley (here Bingham) carrier fluid and the symmetric drag closure, reproduces measured vertical solid-concentration and velocity profiles in a semicircular flume within ±15% and ±5%, respectively. With that validation in hand, it predicts that the particle size distribution changes where solids accumulate, that tilting the flume more steeply settles fine particles while dispersing coarse ones, and that bubble size and bubble volume fraction switch between suspending and settling regimes. If these predictions hold, flume inclination, bubble size, and bubble loading are practical levers for designing slurry disposal and water recovery systems.","feed_headline":"Fine bubbles suspend slurry solids, coarse ones settle them","feed_subtitle":"A 3D flume model validated to within ±15 percent shows which settings favour dewatering.","key_machinery":"The load-bearing machinery is a 3D unsteady Eulerian-Eulerian multiphase model with seven phases: one non-Newtonian carrier fluid, five solid size classes (75, 105, 149, 210, and 296 µm), and a bubble phase. Solid-phase stresses are closed by kinetic theory of granular flow, the carrier rheology is the Herschel-Bulkley model with $n=1$ (a Bingham fluid with yield stress 40 Pa), and every interphase exchange—carrier-solid, solid-solid, and solid-bubble—uses the symmetric drag model, which the paper selects by mean absolute error against the experiments. This combination is what makes the settling profiles, velocity fields, and wall-shear-stress distributions reproducible in the simulations.","core_discovery":"The paper's central claim is that the full multisize settling behaviour of a non-Newtonian thickened slurry in a horizontal semicircular flume can be captured by treating carrier fluid, five solid size classes, and bubbles as interpenetrating continua, with the carrier obeying the Herschel-Bulkley model at $n=1$ (a Bingham fluid) and all phase interactions closed with the symmetric drag model. Against the four experimental cases, the model places about 93% of solid-volume-fraction predictions inside a ±15% uncertainty band and carrier velocity within ±5%. The paper then uses the validated model to claim that flume inclination is a differential lever: fine particles (75–149 µm) settle more as inclination rises, while coarse particles (210–296 µm) disperse; and that bubbles act by size, with 5–50 µm bubbles suspending solids of all sizes and 500–1000 µm bubbles behaving nearly like the no-bubble case, while raising bubble volume fraction from 0.0025 to 0.03 increases overall settling.","pith_inferences":["A natural extension the paper does not spell out is to treat the bubble phase as a deliberate actuator: pulsed or staged bubble injection at selected sizes could create alternating suspension and settling zones along one flume, potentially improving water recovery over a fixed geometry.","Because the free surface is a flat zero-shear lid in the model, the physical flume's open sections and deformable surface may entrain air and let bubbles escape; if so, the simulated bubble-volume-fraction effects are likely an upper bound on suspension and a lower bound on settling enhancement in the field.","The same symmetric-drag, Bingham closure could be tested at Herschel-Bulkley flow indices $n<1$; the paper's validation only covers the $n=1$ Bingham limit, so extending to shear-thinning slurries would require fresh validation against data.","The inclination result implies a two-way trade-off: a steeper flume settles fines but disperses coarse solids, so an optimal dewatering design may need a segmented slope rather than a single angle."],"forward_implications":["If the model is right, the particle size distribution is a primary design variable: PSDs weighted toward coarse sizes produce higher settled beds and distinctly higher wall shear stress (up to about 0.44 Pa for 296 µm particles), which must be accounted for in slope and liner design.","Raising flume inclination from 3° to 6° at constant depth increases average wall shear stress by about 1.6 times and shifts behaviour from coarse-particle segregation toward fine-particle settling, giving operators a tilt-based way to target fine capture.","Introducing fine bubbles (5–50 µm) at low volume fractions keeps solids suspended, while coarse bubbles (500–1000 µm) leave settling close to the no-bubble case, making bubble size a possible control knob rather than just a disturbance.","Increasing bubble volume fraction to 0.03 enhances solids settling because the added bubbles lower the mixture density and raise terminal settling velocity, so bubble loading can be tuned to favour dewatering.","The validated model supplies an engineering route to evaluate intermediate water exclusion from thickened slurry in open-channel tailings systems without building and testing each case experimentally."],"supporting_citations":[{"why":"Supplies the experimental flume geometry, the four validation operating conditions, the particle size distribution, and the measured solid-volume-fraction and velocity profiles the model must match.","marker":"[9]"},{"why":"Establishes the symmetric drag closure and the bubble-size/volume-fraction treatment for poly-dispersed non-Newtonian slurry that this study extends to an open flume.","marker":"[18]"},{"why":"Provides the prior poly-dispersed slurry pipeline model and the inclined-pipe result that the flume-inclination trends are compared with.","marker":"[24]"},{"why":"Justifies the Herschel-Bulkley (Bingham) rheological model for fine-rich clay slurries and supports the yield-stress and consistency parameters used here.","marker":"[44]"},{"why":"Explains why particles concentrate at the flume invert through shear-rate zones, the mechanism the paper invokes for its settling profiles.","marker":"[56]"},{"why":"Supports the claim that higher bubble volume fractions lower mixture density and raise terminal settling velocity, the mechanism for the bubble-fraction results.","marker":"[33]"},{"why":"Supplies the bubble-particle attachment efficiency argument used to explain why fine particles rise with bubbles at higher bubble volume fractions.","marker":"[32]"}],"fun_headline_variants":["Tiny bubbles suspend slurry solids, large ones let them fall","Slope boosts fine slurry settling, scatters coarse particles","Higher bubble fraction speeds up slurry solids settling","Flume model backs slurry settling predictions within 15%","Multisize slurry settling tuned by bubble size and slope"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The physical flume has five open sections and a deformable free surface, while the simulation uses a truncated domain of constant depth with a flat, zero-shear lid; if free-surface deformation, open-section air entrainment, or bubbles collecting at that lid materially change particle settling, the predicted PSD, inclination, and bubble trends will not transfer to the real flume.","fun_headline_variants_meta":{"raw":{"variants":["Tiny bubbles suspend slurry solids, large ones let them fall","Slope boosts fine slurry settling, scatters coarse particles","Higher bubble fraction speeds up slurry solids settling","Flume model backs slurry settling predictions within 15%","Multisize slurry settling tuned by bubble size and slope"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000425,"raw_usage":{"total_tokens":2230,"prompt_tokens":1047,"completion_tokens":1183,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":663,"completion_tokens_details":{"reasoning_tokens":1104}},"tokens_in":663,"tokens_out":1183,"duration_ms":13479,"temperature":1.0,"reasoning_tokens":1104,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:05:47.634257+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the vertical solid-volume-fraction and carrier-velocity profiles at the 14.5 m station of the real 18.5 m flume with its five open sections and free surface exposed, under the four Spelay operating conditions; if more than roughly 7% of points fall outside the model's ±15% solid-fraction band, or velocities deviate beyond ±5% in a way that grows with bubble volume fraction or open-section location, the flat-lid simplification is load-bearing and the parametric conclusions fail.","supporting_citations":[{"cited_title":"Zheng , author D","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental flume geometry, the four validation operating conditions, the particle size distribution, and the measured solid-volume-fraction and velocity profiles the model must match."},{"cited_title":"Matsuhisa \\ and\\ author R","cited_arxiv_id":null,"evidence_quote":"Establishes the symmetric drag closure and the bubble-size/volume-fraction treatment for poly-dispersed non-Newtonian slurry that this study extends to an open flume."},{"cited_title":"Feng , author Y","cited_arxiv_id":null,"evidence_quote":"Provides the prior poly-dispersed slurry pipeline model and the inclined-pipe result that the flume-inclination trends are compared with."},{"cited_title":"Parvathaneni \\ and\\ author V","cited_arxiv_id":null,"evidence_quote":"Justifies the Herschel-Bulkley (Bingham) rheological model for fine-rich clay slurries and supports the yield-stress and consistency parameters used here."},{"cited_title":"Huilin \\ and\\ author D","cited_arxiv_id":null,"evidence_quote":"Explains why particles concentrate at the flume invert through shear-rate zones, the mechanism the paper invokes for its settling profiles."},{"cited_title":"Zhou , author Z","cited_arxiv_id":null,"evidence_quote":"Supports the claim that higher bubble volume fractions lower mixture density and raise terminal settling velocity, the mechanism for the bubble-fraction results."},{"cited_title":"Zhang , author A","cited_arxiv_id":null,"evidence_quote":"Supplies the bubble-particle attachment efficiency argument used to explain why fine particles rise with bubbles at higher bubble volume fractions."}],"review_version":1}