{"id":"19774e28-da32-42c6-a1ce-e98ed0bf80ce","arxiv_id":"1908.05080","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"DEM simulations of slow granular flow in a bin and on an incline yield power-law rheological scaling relations that form a continuum model for dense granular flows.","lead":"Using simulations of grains flowing out of a bin, this paper measures how the local viscosity and pressure of a dense granular material depend on shear rate, and fits simple power-law formulas to the results. It then proposes a continuum model that could let engineers simulate hoppers and silos faster than particle-by-particle methods.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Power-law model's only continuum validation is a tuned fit to the same inclined-surface data, and the no-yield-stress conclusion extrapolates beyond I>0.003 where low-I deviations are visible.","rationale":"The paper's empirical core—DEM data collapse for 1/η̂, u^2, and φ with I across outlet sizes—is a credible and useful finding, and the paper is honest about the difference between bin and inclined-surface flows and about low-I complexity. However, the strongest claim goes beyond data collapse: it asserts a constitutive model that can replace yield-stress descriptions and validates it with the inclined-surface velocity profile. This assertion is load-bearing because the no-yield-stress conclusion depends on it. The validation is weakened by the explicit tuning of n from 1.922 to 1.905 in Fig. 12 and by the fact that the parameters were obtained from the same flow geometry. A leave-one-out cross-validation using existing data would settle whether the model truly reproduces the profile. The reader's chosen weakest assumption—the 2D isochoric transformation—is not the main risk: the measured dilation is below 10% of the shear rate and the paper states L-independence checks, so residual compressibility is a second-order bias rather than a threat to the central no-yield-stress claim. For these reasons I keep the CONDITIONAL verdict rather than raising or lowering it.","tokens_in":17040,"tokens_out":10413,"duration_ms":112898,"concrete_test":"Perform a leave-one-out cross-validation on the four inclined-surface velocity profiles in Fig. 12: for each angle in turn, fit m and n of Eq. (33) using only the other three angles (same I>0.003 cutoff and standard-error filter), then compute Eq. (34) for the held-out angle with no adjustment of n and compare with the DEM vx(y). If the untuned predictions fall outside the simulation error bars, the tuned comparison in Fig. 12 does not validate the continuum model, and the no-yield-stress conclusion would require independent low-I data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—Eq. (33) as a no-yield-stress viscosity law that can replace a yield-stress description and, with Eqs. (12)–(13), (37), (38), reproduce the inclined-surface profile—rests on two poorly secured steps. First, the validation in Fig. 12 is not independent: Eq. (34) is derived from the same constitutive model, uses m=2.024 and n fitted to the inclined-surface flow whose velocity it is then compared with, and agreement is obtained only after changing n from 1.922 to 1.905 'to obtain a good match' (Sec. III.C). A fit that is manually adjusted to a target it was also used to construct cannot demonstrate predictive power. Second, the fit underlying Eq. (33) is restricted to I>0.003 (Sec. III.C, Fig. 11), while the paper's own Fig. 7 reports 'complex yielding behaviour' for I<0.02 and Fig. 9(b) shows deviations from the power law at the low-I end. The conclusion that a power law 'may be used over the entire range of I instead of considering a yield stress' is therefore an extrapolation into a regime where the present data do not exclude a finite yield stress. The 2D isochoric transformation flagged by the reader is a lesser concern here, because dilation is measured below 10% of the shear rate and L-independence is checked.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"Using soft-particle DEM, the paper studies steady gravity-driven discharge from a rectangular bin (Do = 6d, 7d, 8d) and compares it with bumpy inclined-surface flows. After establishing that the flow is nearly isochoric, that the stress tensor is symmetric, and that the principal axes of stress and rate of strain are nearly aligned, the authors introduce a pointwise coordinate transformation that reduces any two-dimensional isochoric velocity gradient to the canonical form of Eq. (16), enabling direct extraction of an effective viscosity and pressure components. They report power-law fits for the inverse scaled viscosity (1/eta = a I^b), squared scaled fluctuation velocity (u^2 = g I^h), shear-rate-scaled viscosity (1/eta-hat = m I^n), and solid fraction (phi = phi_m - r I^s), with good collapse across outlet sizes for I above about 0.003-0.01. These correlations are assembled into a continuum model (Eqs. 12, 13, 33, 37, 38). The model is tested against inclined-surface velocity profiles via Eq. (34), and agreement is obtained after tuning the exponent n from 1.922 to 1.905. The paper concludes that a power-law viscosity relation such as Eq. (33) can be used over the entire range of I in place of a yield-stress description.","tokens_in":17520,"tokens_out":5405,"duration_ms":54277,"significance":"The paper contains substantial careful work: coarse-grained averaging with explicit standard-error filtering, checks of independence of the bin thickness L, a clean argument that every two-dimensional isochoric flow can be locally transformed to the canonical form, detailed data collapse for three outlet widths, and direct comparison with the Jop et al. mu(I) model. If the proposed constitutive model were independently validated, it would be a useful empirical closure for slow dense granular flows, especially because the fluctuation-velocity scaling obviates the need for a separate energy equation. The main limitation is that the single continuum validation is in-sample and hand-tuned, and the no-yield-stress conclusion extrapolates beyond the fitted inertial-number range. The central claim is therefore defensible only after re-scoping or after an out-of-sample test.","major_comments":[{"comment":"The velocity-profile comparison is not an independent validation. The parameters m = 2.024 and n = 1.922 are obtained by fitting 1/eta-hat versus I for the inclined-surface flow (Fig. 11), and the profile in Eq. (34) is then evaluated using that same power law and compared with the same flow class. When the initial prediction differs substantially, the exponent is changed to n = 1.905 'to obtain a good match' (Sec. III.C). With a parameter adjusted to the target curve, Fig. 12 demonstrates consistency rather than predictive skill. Please provide an out-of-sample test (for example, a different geometry, a different flow rate, or a profile computed with the directly fitted m and n before any tuning), or explicitly and prominently frame Fig. 12 as a consistency check rather than as validation.","section":"Section III.C, Fig. 12 and Eq. (34)"},{"comment":"The conclusion that Eq. (33) 'may be used over the entire range of I instead of considering a yield stress' is an extrapolation. The fit of Eq. (33) is restricted to I > 0.003 (Sec. III.C), while Fig. 7 shows a region of steep rise and 'complex yielding behaviour' for I < 0.02, and Fig. 9(b) shows low-I deviations from the power-law form. The data as presented do not exclude a finite yield stress as I approaches zero. Please restrict the claim to the fitted range I > 0.003, or add low-I data and a quantitative comparison between the power-law and yield-stress forms (for example, residual analysis or an information criterion applied over the full I range).","section":"Section III.C, Fig. 11, and Conclusions"},{"comment":"The discussion around Eqs. (35)-(36) implies that the dash-dotted curve in Fig. 7 provides independent evidence for a no-yield-stress mu(I). However, Eq. (36) is an algebraic restatement of Eq. (33) combined with the definition 1/eta-hat = m I^n and eta = mu P / gamma-dot; it is not a new prediction. The text should make clear that the mu(I) power law is derived from, not tested against, the 1/eta-hat data, and should not present the agreement of the dash-dotted line with part of the mu-I data as independent support.","section":"Section III.C, Eqs. (35)-(36)"}],"minor_comments":[{"comment":"The last factor in Eq. (A4) is written as (H - y)^{1/2} but should evidently be (H - y)^{3/2}; the main-text Eq. (34) has the correct form.","section":"Appendix A, Eq. (A4)"},{"comment":"The sentence 'with the bin flow data fitted in two inertial number ranges, i.e., I <= 0.03 and I > 0.03; phi_m.' is incomplete and should be rewritten.","section":"Section III.C, after Eq. (37)"},{"comment":"In the Conclusions, 'the inertial member' should read 'the inertial number'.","section":"Section IV, Conclusions"},{"comment":"The caption and text should state explicitly that n = 1.905 is the tuned value rather than the value obtained from the direct fit in Fig. 11, so that readers do not mistake the comparison for an independent prediction.","section":"Fig. 12 caption and surrounding text"},{"comment":"The statement that 'The values obtained for I0 are lower as compared to those reported earlier' lacks a specific citation or numerical comparison; please name the earlier values or the reference to which the comparison is made.","section":"Section III.C, Fig. 7 discussion"}],"recommendation":"major_revision","confidential_remarks":"The paper is technically careful and likely publishable after revision. The main issue is not the DEM analysis or the data collapse but the overstated validation and the extrapolated no-yield-stress conclusion. Please ask the authors either to add an out-of-sample validation or to re-scope the central claim to the fitted inertial-number range; I would not insist on new simulations if the abstract and conclusions are appropriately weakened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things. First, this is a careful DEM study of a quasi-2D bin flow with two velocity components, and the empirical scalings it reports are plausible. Second, the continuum model is not actually validated: the one \"prediction\" shown is an amplitude fit with a tuned exponent, so the no-yield-stress conclusion goes beyond the data.\n\nThe genuinely useful parts: the coordinate transformation of Sec. III.B lets them extract viscosity and normal pressure components in a flow that is not unidirectional, and the collapse of 1/η̂ against inertial number I is clean for the three outlet sizes. The fluctuation-velocity scaling u^2 = g I^h and the correlation between normal stress differences and fluctuation anisotropy are in line with earlier work and give useful calibration targets. The data handling is careful — standard-error cuts, checks on bin thickness, and clear reporting of excluded points.\n\nWhere it goes soft: the validation in Sec. III.C is not independent. Eq. (34) is derived from the same constitutive law, uses m from the bin fit, and n is changed from 1.922 to 1.905 to make the inclined-surface profile match. Worse, the profile shape in Eq. (A4) is independent of n: the exponent on (H-y) is always 1/2, so tuning n only changes the prefactor. That makes Fig. 12 an amplitude fit, not a test of the rheology. The authors should either present it as a fit or produce a genuine prediction, parameters from one geometry tested on another without adjustment.\n\nSecond, the claim that a power law may replace a yield stress \"over the entire range of I\" is an extrapolation. The fit is restricted to I > 0.003, and the paper's own figures show deviations at low I, including \"complex yielding behaviour\" below I ≈ 0.02. The data do not exclude a finite yield stress. A more honest statement would say the power law works above I ≈ 0.003 and that low-I behavior is unresolved.\n\nThe φ-I correlation is piecewise, which is ad hoc but acceptable for an empirical model.\n\nOverall, the empirical content is worth publishing after revision. A serious referee should ask for toned-down conclusions and a genuine validation, but the DEM data and collapse are solid enough to merit review.","headline":"Careful DEM data and a useful coordinate transform, but the continuum model is validated only by a tuned amplitude fit, so the no-yield-stress claim outruns the evidence.","tokens_in":17985,"tokens_out":5188,"would_cite":true,"duration_ms":46151,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["45.70.Mg","83.80.Fg"],"model":"deepseek-v4-flash","headline":"This paper establishes that slow dense granular flow with two velocity components behaves as a slightly compressible power-law fluid, with viscosity, volume fraction, and fluctuation velocity collapsing onto single curves of the inertial…","keywords":["granular rheology","discrete element method","inertial number","power-law fluid","quasi-two-dimensional flow","bin flow","inclined surface flow","normal stress differences"],"falsifier":"Measure the effective friction coefficient $\\mu=|\\tau_{xy}|/P$ in a quasi-two-dimensional bin flow at inertial numbers below 0.003: the power-law claim predicts $\\mu\\to 0$ as $I\\to 0$, while a yield-stress material would level off at a finite $\\mu_s$; a companion check is to repeat the bin simulation at thickness $L=16d$ and see whether the extracted viscosity and pressure relations shift outside the stated error, which would expose the isochoric two-dimensional assumption.","tokens_in":16874,"feed_emoji":"⏳","tokens_out":7134,"duration_ms":62493,"temperature":0.7,"pith_summary":"This paper uses discrete element simulations of steady gravity-driven flow in a rectangular bin, a flow with two nonzero velocity components, to test whether a simple local rheology describes slow dense granular flow. The central claim is that the material behaves as a slightly compressible power-law fluid: the shear-rate-scaled viscosity follows $1/\\hat{\\eta}=m I^n$, the solid fraction follows $\\varphi=\\varphi_m-r I^s$, and data from three outlet sizes collapse onto single curves of the inertial number $I$. A coordinate transformation casts any two-dimensional isochoric flow into a canonical form that makes the local viscosity and pressure components directly measurable, and the resulting continuum model reproduces the full velocity profile of an inclined-surface flow. The authors conclude that a power-law relation can be used over the whole range of $I$ studied, without invoking a yield stress.","feed_headline":"Slow granular flow is a power-law fluid, not a yield-stress one","feed_subtitle":"DEM runs of a quasi-2D bin collapse viscosity, density and fluctuation speed onto one inertial-number curve.","key_machinery":"The load-bearing object is the pointwise coordinate transformation of Sec. III.B, which maps any two-dimensional isochoric velocity gradient tensor into the canonical form of Eq. (16) by solving the eigenvalue problem of the rotated isochoric gradient. For a generalized Newtonian fluid with stress $\\sigma=P I-2\\eta D$, this transformation makes the shear stress and the individual pressure components directly computable: $\\eta=k_1\\cdot\\sigma\\cdot k_2/(k_1\\cdot D\\cdot k_2)$, $P_{x'}=k_1\\cdot\\sigma\\cdot k_1$, and similarly for the other components. The resulting scalings with the inertial number $I=\\dot{\\gamma}d/\\sqrt{P/\\rho}$—power laws for $1/\\hat{\\eta}$, $\\bar{u}^2$, and $\\varphi$—are the empirical content that closes the continuum model.","core_discovery":"The local rheology of slow dense granular flow in the bin is a slightly compressible power-law fluid, not a yield-stress fluid. The paper derives this from DEM data by transforming the local velocity gradient to the canonical form $v_{x'}=\\dot{\\gamma} y'$, $v_{y'}=\\psi \\dot{\\gamma} x'$, which is possible for every two-dimensional isochoric flow, and then reading off viscosity and pressure from the stress tensor. The viscosity scaled by shear rate obeys $1/\\hat{\\eta}=m I^n$ with $m=1.713$, $n=1.872$ for bin flow, and the solid fraction obeys $\\varphi=\\varphi_m-r I^s$ with separate fits below and above $I=0.03$. The same power-law form describes flow on a rough inclined surface with a slightly different prefactor and a lower viscosity caused by particle layering. Because the power law extends to $I\\approx 0.003$, the paper argues that a yield stress is not needed for the range of slow flows studied.","pith_inferences":["If the power law holds all the way to $I\\to 0$, the apparent yield stress seen in many granular measurements may be a finite-shear-rate or finite-size artifact; this could be tested with extremely slow creep experiments or DEM at $I<10^{-3}$.","The layering-induced offset between inclined-surface and bin-flow rheology suggests that rheological parameters calibrated in unidirectional, layered shear flows may mis-predict non-layered complex flows; a bin-flow-calibrated model should be tested against hopper or silo discharge.","The transformation's assumption of isochoric two-dimensional flow limits the scheme's accuracy wherever out-of-plane motion or dilation contributes; extending the extraction to weakly compressible three-dimensional flows would require a third invariant or a volumetric correction term.","The equality of in-plane pressure components and the anisotropy of pressure with fluctuation components suggests a possible closure modeling the full pressure tensor from the granular temperature tensor rather than a single scalar pressure."],"forward_implications":["Continuum simulations of slow dense granular flows in complex geometries can use Eq. (33) and Eq. (37) directly, eliminating the need for a yield-stress term in the constitutive relation.","The collapse of bin-flow data for three outlet widths onto a single curve means the local rheology is fully characterized by the inertial number and a few material parameters, at least in the slow regime studied.","The explicit correlation for fluctuation velocity ($\\bar{u}^2=g I^h$) closes the momentum balance without solving a separate fluctuation-energy equation.","The same constitutive relations, with parameters refit for the layered surface flow, reproduce the inclined-surface velocity profile through Eq. (34).","In-plane normal stresses are nearly equal and the out-of-plane component is smaller, with each pressure component proportional to the corresponding fluctuation-velocity component, so pressure anisotropy carries no independent information beyond the fluctuation anisotropy."],"supporting_citations":[{"why":"Supplies the coordinate-transformation method that makes local viscosity and pressure components directly measurable.","marker":"45"},{"why":"Gives the mu(I) constitutive law compared against and replaced by the power-law form in the slow regime.","marker":"31"},{"why":"Kinetic-theory scaling of viscosity and pressure with fluctuation velocity underpins the nondimensionalization and the fluctuation-velocity correlation.","marker":"53"},{"why":"Establishes that normal stress differences originate in anisotropic velocity fluctuations, which the paper confirms and uses to correlate pressure components.","marker":"57"},{"why":"Coarse-graining method used to compute averaged continuum fields from particle data.","marker":"77"},{"why":"Provides the collisional-plus-streaming stress tensor computation used for all stress data.","marker":"79"},{"why":"Basis for the power-law solid-fraction vs inertial-number correlation fitted in Eq. (37).","marker":"88"}],"fun_headline_variants":["Slow granular flow: power-law fluid, not yield-stress","DEM shows slow granular flow is a power-law fluid","Granular flow follows power-law rheology, no yield stress","Power-law, not yield-stress: slow granular flow rheology"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire extraction of viscosity and pressure assumes the flow is exactly two-dimensional and volume-preserving at the point of measurement, even though the bin has finite thickness (8 particle diameters) and measured dilation reaches about 10% of the shear rate.","fun_headline_variants_meta":{"raw":{"variants":["Slow granular flow: power-law fluid, not yield-stress","DEM shows slow granular flow is a power-law fluid","Granular flow follows power-law rheology, no yield stress","Power-law, not yield-stress: slow granular flow rheology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000305,"raw_usage":{"total_tokens":1787,"prompt_tokens":1021,"completion_tokens":766,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":696}},"tokens_in":637,"tokens_out":766,"duration_ms":6781,"temperature":1.0,"reasoning_tokens":696,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:24:14.085790+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the effective friction coefficient $\\mu=|\\tau_{xy}|/P$ in a quasi-two-dimensional bin flow at inertial numbers below 0.003: the power-law claim predicts $\\mu\\to 0$ as $I\\to 0$, while a yield-stress material would level off at a finite $\\mu_s$; a companion check is to repeat the bin simulation at thickness $L=16d$ and see whether the extracted viscosity and pressure relations shift outside the stated error, which would expose the isochoric two-dimensional assumption.","supporting_citations":[{"cited_title":"Bhateja \\ and\\ author D","cited_arxiv_id":null,"evidence_quote":"Supplies the coordinate-transformation method that makes local viscosity and pressure components directly measurable."},{"cited_title":"Kumaran ,\\ title title Kinetic theory for sheared granular flows , \\ @noop journal journal C","cited_arxiv_id":null,"evidence_quote":"Kinetic-theory scaling of viscosity and pressure with fluctuation velocity underpins the nondimensionalization and the fluctuation-velocity correlation."},{"cited_title":"Saha \\ and\\ author M","cited_arxiv_id":null,"evidence_quote":"Establishes that normal stress differences originate in anisotropic velocity fluctuations, which the paper confirms and uses to correlate pressure components."},{"cited_title":"Weinhart , author R","cited_arxiv_id":null,"evidence_quote":"Coarse-graining method used to compute averaged continuum fields from particle data."},{"cited_title":"Tripathi \\ and\\ author D","cited_arxiv_id":null,"evidence_quote":"Provides the collisional-plus-streaming stress tensor computation used for all stress data."},{"cited_title":"Hatano ,\\ title title Power-law friction in closely packed granular materials , \\ @noop journal journal Phys","cited_arxiv_id":null,"evidence_quote":"Basis for the power-law solid-fraction vs inertial-number correlation fitted in Eq. (37)."}],"review_version":1}