{"id":"275aa94b-9fbb-4f6a-8cad-6661a2f0f132","arxiv_id":"2501.04154","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":1.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A review of particle-resolved DNS results that concludes the method has matured to produce high-fidelity data for turbulent particulate flows.","lead":"This paper reviews how particle-resolved direct numerical simulation (PR-DNS) has become a reliable tool for studying turbulent flows laden with solid particles. It surveys results on fluidized beds, settling particles, and sediment transport, arguing that PR-DNS now complements laboratory experiments.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The only quantitative PR-DNS-vs-experiment validation quoted (Section 6.2, Fig. 6.1) has non-overlapping error bars (29±1 vs 33±2), so the Section 6.5 'equal footing' conclusion is not established by the evidence shown.","rationale":"Read in good faith, this is a review chapter (published 2022, posted to arXiv 2025) whose central claim is that PR-DNS data are as trustworthy as laboratory measurements. For that claim to hold, the numerical method must reproduce known quantitative results across the covered regimes. The paper's own evidence includes several qualitative correspondences that are real independent support: the columnar-clustering transition at Ga larger than about 170 was experimentally confirmed by Huisman et al. [55], and the sediment transport simulations reproduce the Shields threshold and a power-law flux (Section 6.4.2). However, the only concrete quantitative benchmark presented with numbers is the wave-speed comparison, and it misses by roughly 14% with non-overlapping error bars. That is exactly the kind of mismatch that determines whether 'equal footing' is justified. I therefore agree with the reader's weakest_assumption. The additional textual tension between Section 6.1's 'without further modelling assumptions' and Section 6.2's explicit lubrication and collision models reinforces the concern: in dense suspensions the modelled lubrication force is a dominant dissipation channel, so describing the method as assumption-free is too strong. A targeted convergence and model-sensitivity study on the one quantitative validation case would settle the point. Because the paper is a book-chapter review rather than a new research preprint, the existing UNVERDICTED verdict stands; the concern affects the strength of the conclusions, not the preprint status.","tokens_in":27287,"tokens_out":4655,"duration_ms":47401,"concrete_test":"Re-run the Fig. 6.1 wave-instability case [24] as a convergence and model-sensitivity study: keep angle brackets phi = 0.51, density ratio 4.4, and Ga = 133 fixed, and vary grid resolution dp/dx from about 8 to 24 and lateral domain size from 4dp to 16dp with the same lattice-Boltzmann/IBM method. If the dimensionless wave speed c~ remains at 33±2 across converged settings, the mismatch with Duru et al.'s 29±1 is genuine and the 'equal footing' claim must be weakened; if c~ moves toward 29, the original validation was grid- or domain-limited, which would directly undermine the claim. As a complementary check, repeat with the lubrication force switched off to bound the model dependence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The maturation claim ('equal footing with laboratory experimental measurements', Section 6.5) requires that PR-DNS predictions be quantitatively accurate across the configurations reviewed. The chapter's only explicit quantitative validation is the liquid-fluidized-bed wave speed: experiment c~=29±1 [30] versus simulation c~=33±2 [24] (Section 6.2, Fig. 6.1). The error bars do not overlap, so this is not a successful quantitative validation by the paper's own criterion. Qualitative correspondences, such as columnar clustering confirmed experimentally by Huisman et al. [55] and the sediment flux scaling and Shields threshold in Section 6.4.2, are genuine independent support but do not establish quantitative fidelity. Moreover, Section 6.1's assertion that PR-DNS solves the Navier-Stokes and Newton-Euler equations 'without further modelling assumptions' is internally contradicted by Section 6.2, which states that radial lubrication and spring-type soft-sphere collisions are modelled, with most dissipation occurring 'as a result of lubrication modelling'. These are modelling assumptions that directly affect dense flows such as the wave case. The central conclusion therefore overstates the evidence; the review should present PR-DNS as promising but not yet on demonstrably equal footing.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This review chapter surveys particle-resolved direct numerical simulation (PR-DNS) results for turbulent particulate flows, covering dense fluidized systems, dilute settling and turbulence interaction, vertical and horizontal wall-bounded channel flows, sediment transport, and pattern formation. The manuscript argues that PR-DNS has matured so that it can be applied successfully to a wide range of configurations and concludes that PR-DNS is \"on an equal footing with laboratory experimental measurements,\" enabling complementary use of simulation and experimental datasets.","tokens_in":27523,"tokens_out":6482,"duration_ms":59502,"significance":"The chapter is a useful, comprehensive overview written by leading practitioners, and it provides a valuable entry point to a large body of recent PR-DNS literature. Its strengths include a broad coverage of configurations, clear separation of regimes (dilute, dense, unbounded, wall-bounded), and an honest discussion of computational cost and sub-models such as collision and lubrication treatments. The main weakness is that the central claim of quantitative fidelity—the abstract's \"matured\" statement and the \"equal footing\" conclusion in Section 6.5—is not supported by the validation evidence presented, most notably the wave-speed comparison with non-overlapping error bars.","major_comments":[{"comment":"The paper cites the dimensionless wave speed in a liquid-fluidized bed as a validation of PR-DNS against experiment, but the reported values are c̃ = 29 ± 1 (experiment [30]) and c̃ = 33 ± 2 (simulation [24]). These two intervals do not overlap: the simulation's lower bound is 31, while the experiment's upper bound is 30. The text presents this comparison without comment and Section 6.5 uses it as part of the basis for the \"equal footing\" conclusion. The authors should either supply additional successful quantitative comparisons (for example, drag or settling-velocity data) or explicitly qualify the wave-speed result as a partial validation and state that the method is not yet demonstrated to match experiment within uncertainty in all configurations.","section":"Section 6.2 / Figure 6.1"},{"comment":"Section 6.1 states that PR-DNS solves the Navier-Stokes and Newton-Euler equations \"without further modelling assumptions – except for those which relate to solid-solid contacts.\" This is contradicted by Section 6.2, which describes radial lubrication and spring-type soft-sphere collisions as modelled forces and notes that most dissipation between approaching particles \"takes place in the liquid prior to collision, either in the resolved flow or as a result of lubrication modelling.\" Lubrication and collision sub-models are not resolved from first principles and can influence dense-flow statistics such as those used for the wave-speed validation. Please revise the characterization in Section 6.1 and add a discussion of the sensitivity of the reviewed results to these sub-models.","section":"Section 6.1 vs. Section 6.2"},{"comment":"The review supports its maturity claim with several qualitative correspondences—for instance, columnar clustering confirmed experimentally by Huisman et al. [55] and the recovery of the critical Shields number and sediment flux scaling in Section 6.4.2. These are genuine independent checks, but they do not establish quantitative fidelity for averaged statistics such as mean velocity and concentration profiles in the absence of direct experimental comparison. To make the \"equal footing\" conclusion defensible, Section 6.5 should include an explicit statement of which classes of configuration have direct quantitative validation and which rely on qualitative or indirect evidence.","section":"Sections 6.4.2 and 6.5"}],"minor_comments":[{"comment":"The phrase \"the turbulence structure is signifcantly modified\" contains a typo: \"signifcantly\" should be \"significantly.\"","section":"Section 6.4.1"},{"comment":"The cross-reference \"as already discussed in subsection 6.6\" appears to be incorrect; the nonlinear drag mechanism is discussed in Section 6.3.2.","section":"Section 6.4.1"},{"comment":"The notation for the density ratio is inconsistent: the text uses ρp/ρf while the caption of Figure 6.2 uses ρP/ρf; please standardize the notation.","section":"Section 6.2 / Figure 6.2"},{"comment":"The sentence \"Wrel is then the value obtained once ensemble averaging (6.4) over all particles\" is grammatically incomplete; please rephrase to \"once ensemble averaging of (6.4) over all particles is performed.\"","section":"Section 6.3.1"},{"comment":"The modified Stokes number in Equation (6.2) is introduced with a tilde that is not visible in the rendered text; please define it explicitly (for example, as \\(\\widetilde{St}\\)) or describe it in words.","section":"Section 6.2 / Equation (6.2)"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern lands: the wave-speed comparison with non-overlapping error bars is a genuine problem for the central claim. This is a review chapter, and the fix is within scope—temper the \"equal footing\" conclusion, add a discussion of validation status, and correct the modelling-assumption characterization. I recommend major revision rather than rejection because the chapter's comprehensive overview and open discussion of limitations are valuable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this as a book chapter, not a research preprint: it reviews prior PR-DNS results and presents no new simulations, derivations, or data. Its value is as a status report. The organization is good—drag parameterization, dilute settling, wall-bounded and sediment transport—and the compiled comparisons in Figures 6.4, 6.7, and 6.8 are genuinely useful. It also cites independent experimental support where it exists: Huisman et al. confirming columnar clustering, and the critical Shields threshold and bedload flux scaling from PR-DNS matching experiments. That is real credit and not self-promotion.\n\nThe soft spots are real but not fatal. The only explicit quantitative validation against experiment—the liquid-fluidized-bed wave speed, Section 6.2—has non-overlapping error bars: experiment 29±1, simulation 33±2. The text presents this as validation without comment, which it is not. That alone weakens the Section 6.5 claim that PR-DNS is 'on an equal footing' with laboratory measurements. There are other independent supports, as above, but they are mostly qualitative or indirect; they do not establish quantitative fidelity across the reviewed configurations.\n\nSecond, Section 6.1 says PR-DNS solves Navier-Stokes plus Newton-Euler 'without further modelling assumptions—except solid-solid contacts.' Section 6.2 then says particles interact via radial lubrication and spring-type soft-sphere collisions, with most dissipation arising from lubrication modelling. Those are modelling assumptions for dense flows, and the wave case is exactly such a flow. The sentence overstates the method's first-principles character.\n\nI would not call the paper's central argument broken. It is a review, so there is no circularity and no fitting to its own conclusions; Equation (6.1) is attributed properly to Rubinstein et al. The issue is calibration of the rhetoric: 'promising and increasingly reliable' is supported; 'equal footing' is not.\n\nWho gets value: modelers and newcomers who want a map of the PR-DNS literature—especially drag correlations, finite-size particle settling, and sediment transport—and a sense of computational cost. Experts in the area will find little they do not already know.\n\nRecommendation: as a review chapter it deserves peer review, not desk rejection, but the referees should push for a revised conclusion and an honest discussion of the wave-speed mismatch and the modelling assumptions in the collision and lubrication treatment.","headline":"A competent review chapter with no new content; its 'equal footing' conclusion overreaches the validation evidence it actually shows.","tokens_in":28070,"tokens_out":2690,"would_cite":true,"duration_ms":26751,"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":"This review argues that particle-resolved direct numerical simulation has matured to the point where its datasets can be used on equal footing with laboratory experiments for turbulent particulate flows.","keywords":["particle-resolved direct numerical simulation","turbulent particulate flows","fluidized beds","settling suspensions","sediment transport","bedform formation","drag correlations","finite-size particles"],"falsifier":"A controlled comparison of PR-DNS predictions against laboratory measurements for the same dimensionless parameters, such as the wave speed in a liquid-fluidized bed at the cited conditions and then at several other volume fractions and particle-fluid density ratios, would settle the question: if the simulation results systematically fall outside experimental error bars in more than one configuration, the equal-footing claim is false.","tokens_in":27095,"feed_emoji":"🌊","tokens_out":10159,"duration_ms":95624,"temperature":0.7,"pith_summary":"This review argues that particle-resolved direct numerical simulation (PR-DNS) has matured into a source of high-fidelity data for turbulent particle-laden flows. PR-DNS solves the fluid motion and each particle's rigid-body motion together, resolving the flow around every particle without sub-grid models except for solid-solid contact. The paper surveys results across dense fluidized beds, dilute settling suspensions, wall-bounded channels, and sediment beds, and concludes that simulation datasets can be used side by side with laboratory measurements to clarify mechanisms and to build simplified engineering models. The practical payoff, if the claim is right, is a numerical laboratory that can supply space-and-time-resolved data in configurations where experiments are difficult or impossible.","feed_headline":"Particle-resolved simulation now rivals lab experiments","feed_subtitle":"Two decades of PR-DNS results now stand beside laboratory data for fluidized beds, settling, and sediment transport.","key_machinery":"The central object is the PR-DNS method itself: the Navier-Stokes equations for the fluid are solved together with Newton-Euler equations for each rigid particle, with the no-slip condition enforced at each particle's surface, so no modelling assumptions are introduced except for solid-solid contacts. This resolves the micro-scale flow around every particle and gives direct access to forces, torques, and local flow structure. Supporting techniques discussed by the paper include collision algorithms, Voronoi tessellation analysis (a space-partitioning statistic that measures clustering), and the dimensionless parameters governing settling and sediment entrainment; the validation strategy relies on quantitative comparisons with specific experiments such as the fluidized-bed wave speed.","core_discovery":"The central claim is that PR-DNS is “on an equal footing with laboratory experimental measurements” and that both data sources can be used in a complementary way. The review supports this by showing that PR-DNS has reproduced a wide range of phenomena—including wave instabilities in liquid-fluidized beds, columnar clustering of settling spheres, and the formation of sediment ripples—while also providing detailed local quantities such as drag forces, settling statistics, and momentum budgets that experiments cannot easily measure. According to the paper, the method has matured enough to be applied successfully to diverse fluid/particle configurations, at the price of large but manageable computational cost.","pith_inferences":["A systematic benchmark suite comparing PR-DNS against laboratory measurements across many configurations would be a natural extension of the equal-footing claim; the review's validation rests on a single wave-speed comparison, so the suite would test how broadly the claim holds.","PR-DNS could become a causal-inference tool: by switching off rotation, collisions, or turbulence forcing one at a time, researchers can identify mechanisms that are entangled in physical experiments.","The volume of resolved data produced by PR-DNS is well suited to data-driven closure discovery, a direction the review mentions as promising for future model development.","If PR-DNS datasets are pooled as community reference data, they could serve as synthetic experiments for parameter regimes that are hard to realize in the laboratory, such as very high particle volume fractions or extreme density ratios."],"forward_implications":["Drag correlations used in engineering models can be made more faithful by including the Stokes-number dependence that PR-DNS has quantified, interpolating between fixed-particle and empirical fluidization limits.","PR-DNS results can be used to test and calibrate simplified Euler-Lagrange and Euler-Euler models in regimes where experiments are difficult.","Sediment transport models can be improved with PR-DNS data, which show that algebraic flux laws fail locally because the particle flux lags the shear stress.","The method can deliberately simulate “wrong physics”—for example, suppressing particle rotation—to isolate the causal role of individual mechanisms.","As computational capacity grows, PR-DNS is expected to extend to non-spherical particles, larger domains, and a broader sweep of the parameter space, including long-time bedform evolution."],"supporting_citations":[{"why":"PR-DNS of dense liquid-fluidized beds that produces the wave-instability results used as a validation case.","marker":"[24]"},{"why":"Experimentally measured wave speed in a liquid-fluidized bed used as the quantitative benchmark for validation.","marker":"[30]"},{"why":"PR-DNS-based drag study that quantifies how Stokes number shifts drag between fixed-particle and empirical correlations.","marker":"[110]"},{"why":"Supplies the high-Stokes-number drag correlation for fixed particle assemblies used as one limit in the drag interpolation.","marker":"[133]"},{"why":"Supplies the empirical fluidization drag correlation used as the other limit in the interpolation.","marker":"[143]"},{"why":"PR-DNS that first reproduced sediment pattern formation, demonstrating a capability experiments cannot easily provide.","marker":"[71]"},{"why":"PR-DNS of ripple formation and evolution, used to assess bedform predictions and scaling.","marker":"[73]"},{"why":"Laboratory experiments confirming columnar clustering of settling spheres, supporting the claim that PR-DNS clustering is physical.","marker":"[55]"},{"why":"PR-DNS of sediment bed erosion that reproduces the critical Shields-number threshold and the cubic particle-flux scaling.","marker":"[72]"},{"why":"Experimental determination of the critical Shields number used to validate PR-DNS erosion thresholds.","marker":"[97]"}],"fun_headline_variants":["PR-DNS now stands on equal footing with labs","Particle-resolved simulations match lab experiments","High-fidelity simulations complement lab data","PR-DNS matures into a lab-standard tool"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that particle-resolved simulation can stand beside laboratory measurements assumes that the numerical codes in the reviewed studies really do reproduce the physics of fluid and particles, and the one head-to-head validation shown—a wave speed in a fluidized bed—does not by itself prove that.","fun_headline_variants_meta":{"raw":{"variants":["PR-DNS now stands on equal footing with labs","Particle-resolved simulations match lab experiments","High-fidelity simulations complement lab data","PR-DNS matures into a lab-standard tool"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000196,"raw_usage":{"total_tokens":1247,"prompt_tokens":721,"completion_tokens":526,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":337,"completion_tokens_details":{"reasoning_tokens":468}},"tokens_in":337,"tokens_out":526,"duration_ms":5498,"temperature":1.0,"reasoning_tokens":468,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:39:33.036483+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A controlled comparison of PR-DNS predictions against laboratory measurements for the same dimensionless parameters, such as the wave speed in a liquid-fluidized bed at the cited conditions and then at several other volume fractions and particle-fluid density ratios, would settle the question: if the simulation results systematically fall outside experimental error bars in more than one configuration, the equal-footing claim is false.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the high-Stokes-number drag correlation for fixed particle assemblies used as one limit in the drag interpolation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the empirical fluidization drag correlation used as the other limit in the interpolation."}],"review_version":1}