{"id":"bd608192-5840-4e8a-892c-6d730f519a2a","arxiv_id":"2506.09517","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A two-step, point-cloud-based coarse-graining mapping for CFD-DEM reduces grid-size dependence and improves pore-pressure feedback in dense particle-fluid flows.","lead":"This paper presents a new mapping strategy for CFD-DEM simulations that first converts particle data into smooth fields on a point cloud, then couples those fields to the fluid grid. The motivation is to reduce grid-size dependence and to capture pore-pressure feedback in dense, particle-fluid systems such as underwater landslides and fluidized beds.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Dense-collapse evidence is confounded: the proposed method adds the Appendix B semi-implicit momentum-exchange scheme, and the KBM/DPVM comparisons do not state whether they used it, so the point-cloud mapping may not be the cause of the captured pore-pressure feedback.","rationale":"The reader's weakest assumption concerns the fixed point-cloud parameters (N_o = 96, spacing 0.25d) and the absence of a sensitivity study. That is a valid secondary concern about the quantitative grid-independence claim, especially for fine grids where the point cloud may under-sample grid cells. However, the more load-bearing issue is the uncontrolled comparison behind the paper's strongest claim. The proposed method is not just a new mapping; it also introduces an extended semi-implicit momentum-exchange scheme in Appendix B, and the paper does not state whether the comparison methods KBM and DPVM employed the same treatment. Since the semi-implicit scheme is known to improve stability in dense particle-fluid systems and Appendix B derives a restrictive stability criterion for explicit coupling, the observed difference in pore-pressure feedback and collapse initiation time could be due to that algorithmic change rather than to the point-cloud mapping. This directly threatens the causal claim that the mapping 'enables' the capture of pore-pressure feedback. The paper otherwise has useful strengths: the 1D and 2D weight-allocation tests are instructive, the Ergun test is a reasonable benchmark, and the implementation is based on the open-source CFDEM framework. Those strengths do not resolve the confound. A controlled re-run of the collapse cases with KBM/DPVM using the same semi-implicit scheme would settle the issue, so a conditional verdict with this additional required check is appropriate.","tokens_in":21237,"tokens_out":6444,"duration_ms":75694,"concrete_test":"Re-run the dense granular column collapse cases of Section 5.3 (phi_i = 0.6233 and the loose cases phi_i = 0.55, 0.571, 0.60) with standard KBM (w/d = 2) and DPVM, but using the same Appendix B semi-implicit momentum-exchange scheme, with <v>_j computed by each method's own projection (kernel-weighted for KBM, local DPVM average for DPVM). Compare t_init versus phi_i and the t = 0-0.06 s pore-pressure profile along L_A-B (Figs. 14c and 15b). If KBM/DPVM with semi-implicit coupling reproduce the delayed initiation and negative pore pressure, the claim that the point-cloud mapping is the enabling ingredient is unsupported. The paper should also explicitly report whether the original KBM/DPVM runs used explicit or semi-implicit coupling.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that the point-cloud mapping enables accurate detection of subtle particle displacements and the associated pore-pressure feedback (Sections 5.3 and 7) is supported mainly by the immersed granular column collapse comparison in Figs. 14 and 15. However, the comparison is not controlled. The proposed solver is augmented with an extended semi-implicit momentum-exchange term, Eq. (B.6), which uses the coarse-grained particle velocity <v>_j from Eq. (26). Appendix B further shows, via Eq. (B.5), that explicit momentum exchange becomes severely time-step limited in dense packings. The paper never states whether the KBM and DPVM runs in the collapse test used the same semi-implicit treatment or the default explicit scheme. If KBM/DPVM used explicit coupling, their failure to capture dilation-induced negative pore pressure and delayed initiation could be a numerical artifact of the explicit scheme rather than a consequence of the point-cloud mapping. Thus the causal attribution of the improvement to the mapping strategy is not established by the presented evidence.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a two-step mapping strategy for unresolved and semi-resolved CFD-DEM. In the first step, each DEM particle is represented by a multi-layer Fibonacci point cloud and spread with a truncated Gaussian kernel, producing grid-independent coarse-grained fields (volume fraction, momentum density, velocity). In the second step, these point-based fields are projected onto fluid grids through a topology-based indicator, while fluid variables are interpolated back to particle centers for the drag closure. The method is compared with conventional kernel-based mapping (KBM) and the divided particle volume method (DPVM) in 1D/2D weight-allocation tests, an Ergun pressure-drop test, single-sphere sedimentation, a bi-disperse fluidized bed, and an immersed granular column collapse. The central claims are improved accuracy, stability, and grid independence over conventional kernel-based mapping, and, in very dense granular collapse, the ability to detect sub-grid particle displacements and capture dilation-induced negative pore pressure and delayed initiation.","tokens_in":21441,"tokens_out":5471,"duration_ms":61983,"significance":"If the central claims survive scrutiny, the method would be a useful addition to semi-resolved CFD-DEM: it offers a smoother, grid-independent coupling over a wide range of grid-to-particle size ratios and extends unresolved methods into dense regimes where pore-pressure feedback matters. The manuscript has clear strengths: the point-cloud quadrature and two-way coupling equations are derived explicitly; the application suite spans dilute to dense systems; the efficiency analysis with a look-up table variant is included; and the no-flux boundary treatment in Appendix A is a practical contribution. The main evidence for the pore-pressure claim, however, rests on a comparison that is not controlled with respect to the semi-implicit momentum-exchange scheme, and the point-cloud resolution parameters are fixed without a sensitivity study. These gaps currently limit the strength of the central claims.","major_comments":[{"comment":"The immersed granular column collapse test (Figs. 14 and 15) is the principal evidence for the claim that the point-cloud mapping captures dilation-induced negative pore pressure and delayed initiation. However, the proposed solver is augmented with the extended semi-implicit momentum-exchange term, Eq. (B.6), which uses the coarse-grained particle velocity <v>_j from Eq. (26). The paper never states whether the KBM and DPVM runs in this test used the same semi-implicit treatment or the default explicit scheme. If KBM and DPVM used explicit coupling, their failure to capture the negative pore pressure and delayed initiation could be a numerical artifact of the time-step limitation described by Eq. (B.5) rather than a consequence of the mapping strategy. This is load-bearing for the Section 7 conclusion. Please provide a controlled comparison: run KBM and DPVM with the same semi-implicit scheme, or run the proposed method with explicit momentum exchange, and report pore pressure and initiation times for both configurations.","section":"5.3 and Appendix B"},{"comment":"The point-cloud parameters N_o=96 points on the outermost layer, layer spacing Delta_r=0.25d, and outer radius r_o=2d are fixed throughout the paper, and no sensitivity study is reported for N_o or Delta_r. The core claims of grid independence and detection of subtle sub-grid particle displacements depend on the point clouds providing a sufficiently fine and uniform quadrature of the truncated Gaussian kernel. If these parameters are too coarse, the method would degrade toward the behavior of grid-centered kernel methods. Please report the convergence of the weight-allocation error, Eq. (32), and of representative collapse outputs (e.g., initiation time or pore pressure) with respect to N_o and Delta_r.","section":"3.2 and 4.1"},{"comment":"The Ergun pressure-drop test is largely a self-consistency check: the Gidaspow drag model in Eq. (9) uses the Ergun correlation for epsilon_f <= 0.8, and the comparison in Eq. (34) is against the same Ergun correlation. Agreement therefore supports the numerical consistency of the mapping and the drag implementation, but it does not independently validate the physical accuracy of the point-cloud mapping. The paper should either present this test explicitly as a consistency check or validate against a drag closure that is not derived from the Ergun equation.","section":"4.3, Eqs. (9) and (34)"}],"minor_comments":[{"comment":"The definition of W(x_j) is ambiguous for the proposed method: the text says the weight is evaluated at a grid center, which is the KBM operation, while the proposed method sums point-cloud weights inside each grid cell. Please clarify how W(x_j) is computed for each method in this error metric.","section":"4.1, Eq. (32)"},{"comment":"Eq. (26) is undefined when a grid cell contains no point-cloud points, because the denominator vanishes. This can occur near the free surface in the collapse simulations; please specify the treatment used in that case (e.g., setting <v>_j to zero or using a small regularization).","section":"3.3, Eq. (26)"},{"comment":"The wording that only the proposed approach captures the pore-pressure feedback is too strong given the uncontrolled comparison noted in the major comments; please qualify the claim until the comparison with KBM and DPVM is made on equal footing.","section":"5.3 and 7"},{"comment":"The notation in Eq. (B.6) is very hard to parse: the absolute-value bars, the fraction inside the angle brackets, and the implied split into implicit and explicit parts are not defined clearly. Please rewrite the equation with explicit definitions of all norms and bracket operations.","section":"Appendix B, Eq. (B.6)"},{"comment":"The text states that the near-wall volume-fraction profile is in 'good agreement with theoretical and experimental results,' but the reference for this theoretical/experimental profile is not identified and no quantitative error is reported. Please add the reference and a quantitative comparison.","section":"Appendix A, Fig. A.18"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and presents a workable algorithmic improvement to semi-resolved CFD-DEM. The main blockers are not novelty but experimental control and parameter sensitivity: the collapse-test comparison is confounded by the simultaneous introduction of the Appendix B semi-implicit scheme, and the point-cloud resolution is fixed without sensitivity analysis. I would be willing to review a revised version. The authors may also consider making the CFDEM/OpenFOAM/LIGGGHTS implementation or the test-case setups available, since reproducibility would substantially increase the value of the paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the point-cloud coarse-graining mapping is a real and well-tested idea, but the paper's flagship claim—that it captures dilation-induced negative pore pressure in dense collapse—rests on a comparison that is not controlled. The confound flagged in the stress-test is real. Appendix B adds a semi-implicit momentum-exchange scheme that relies on the coarse-grained particle velocity; the paper never states whether the KBM and DPVM collapse runs used the same scheme or the default explicit one. Given Eq. (B.5), explicit exchange is severely time-step-limited in dense packings, so KBM/DPVM failure to see pore-pressure feedback could be a numerical artifact of the exchange scheme, not a deficiency of their mapping. As written, the causal attribution to the point-cloud mapping is not established.\n\nWhat is genuinely new: using a multi-layer Fibonacci point cloud as an intermediate, grid-independent layer between DEM particles and the fluid grid. The 1D/2D weight-allocation tests are clean and show real advantages over KBM in grid-size dependence and oscillation. The no-flux boundary treatment for physical and processor boundaries is a nice addition, and the efficiency analysis with the table-based look-up is useful. The method is plausible and likely a step forward for semi-resolved CFD-DEM.\n\nSoft spots beyond the confound: no sensitivity study for the point-cloud parameters (N_o=96, layer spacing 0.25d, outer radius w=2d), on which the claimed grid-independence depends. The dense-collapse reference is from the authors' own group, and the initial packing density differs slightly (0.6233 vs 0.6277); no error bars or multiple realizations are shown. The Ergun test is mostly a self-consistency check, though it does demonstrate stability. No code or data are released.\n\nAudience: CFD-DEM practitioners who care about grid robustness and dense suspensions. The paper deserves peer review, but the revision needs a controlled dense-collapse comparison with the same momentum-exchange scheme for all methods, plus a parameter sensitivity study. I would not take the pore-pressure claim as evidence in its current form.","headline":"Genuinely novel point-cloud coarse-graining mapping for CFD-DEM, but the flagship pore-pressure claim is undercut by an uncontrolled comparison that never states whether KBM/DPVM used the same semi-implicit momentum-exchange scheme.","tokens_in":21978,"tokens_out":2786,"would_cite":true,"duration_ms":32425,"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 paper claims that a two-step, point-cloud coarse-graining mapping removes grid-size dependence in unresolved CFD-DEM and is the only tested approach that captures dilation-induced negative pore pressure and delayed collapse in dense…","keywords":["Eulerian-Lagrangian","coarse graining","two-way coupling","two-step mapping","granular column collapse","CFD-DEM","pore pressure feedback","grid independence"],"falsifier":"Repeat the dense immersed granular column collapse ($\\phi_i=0.6233$) and the 1D weight-allocation error test with much coarser point clouds, e.g., $N_o=24$ or layer spacing $0.5d$; if the negative pore pressure, delayed initiation, and the two-orders-of-magnitude error reduction all survive, the point-cloud parameters are not load-bearing, and if they degrade, the central claim fails.","tokens_in":20980,"feed_emoji":"🌊","tokens_out":10185,"duration_ms":102233,"temperature":0.7,"pith_summary":"The paper proposes a two-step mapping strategy for volume-averaged CFD-DEM simulations of particle-fluid systems. Instead of sending particle data straight to fluid grid centers, it first spreads each particle onto a multi-layer Fibonacci point cloud using a truncated Gaussian kernel, producing smooth coarse-grained fields that do not know about the fluid grid. Only then are those fields projected onto the grid for coupling. The authors argue that this removes the grid-size dependence and particle-crossing oscillations of conventional kernel-based methods, and that it is the only tested approach that captures the dilation-induced negative pore pressure and delayed initiation of very dense immersed granular column collapse. If right, the method extends unresolved CFD-DEM reliably from dilute suspensions to nearly jammed granular packs.","feed_headline":"Point clouds capture pore-pressure feedback in dense collapse","feed_subtitle":"A two-step coarse-graining strategy cuts grid-size dependence from dilute to dense particle-fluid systems.","key_machinery":"The load-bearing object is the multi-layer Fibonacci point cloud attached to each particle. Points are positioned on concentric spherical layers by the Fibonacci lattice, with layer spacing $0.25d$ and the outermost layer at radius $w=2d$; the number of points per layer scales with the squared layer radius, so each point owns an equal-area Voronoi cell on its sphere. A truncated Gaussian kernel is evaluated at every point and renormalized over the point cloud, so the weights sum correctly without any grid involvement. A topology-based indicator tells which fluid grid contains each point, enabling grid-averaged void fraction and particle velocity fields, while the same point weights interpolate fluid quantities back to the particle center for drag. This two-way, grid-independent quadrature of the kernel is what lets the method detect tiny volume changes of the granular phase.","core_discovery":"The central discovery is that a particle can be represented by roughly 300 evaluation points arranged on concentric Fibonacci-lattice spheres (outermost layer radius equal to the kernel width $w=2d$, layer spacing $0.25d$, 96 points on the outer layer), and that evaluating the coarse-graining kernel at those points rather than at fluid grid centers decouples the mapping from the grid. The paper shows that this point-based coarse graining lowers the weight-allocation error relative to the analytical truncated Gaussian by orders of magnitude, suppresses temporal oscillations as a particle crosses grid boundaries, yields convergence above first order in a sphere-sedimentation test, and reproduces both a bi-disperse fluidized bed and Ergun pressure-drop data. Its strongest validation is the immersed granular column collapse: for initial solid fraction $\\phi_i=0.6233$, only the proposed method gives the delayed initiation and negative pore pressure along the column diagonal that the resolved LBM-DEM reference shows, while DPVM and KBM collapse too early. The claim is that this sensitivity comes from resolving sub-grid particle displacement in the coarse-grained volume fraction and momentum fields before they are projected to the fluid grid.","pith_inferences":["One consequence the authors leave implicit is that any Eulerian-Lagrangian transfer quantity—heat, species mass, charge—could be routed through the same grid-free point-cloud layer with the same expected insensitivity to grid size.","The paper's premise that the coarse-grained fields are smooth and grid-free suggests a route beyond drag-only coupling: the same point-cloud weights could carry contact stress or granular temperature into the fluid-phase equations, which the authors flag as future work.","A testable extension is to make point-cloud resolution adaptive—fewer points in dilute regions, more in dense packs—since the fixed roughly 300 points per particle is the main overhead on coarse grids."],"forward_implications":["Weight allocation to grids matches the analytical kernel far more closely than standard KBM, with errors reduced by at least two orders of magnitude in the 1D test.","Particle motion across grids produces much smaller periodic oscillations in the mapped weight field, for both fine and coarse grids.","Settling velocity of a single sphere converges with grid refinement at better than first order and agrees with experiment from $\\Delta x/d=3$ down to $0.25$.","In bi-disperse fluidized beds, the method reproduces measured expansion heights on both fine and coarse grids, where DPVM and KBM shift with grid resolution.","For very dense immersed granular collapse, the method tracks the resolved LBM-DEM initiation times over initial packing densities $0.55$ to $0.6233$, capturing the regime where collapse is delayed by negative pore pressure."],"supporting_citations":[{"why":"Identifies the coarse-grid limitations of conventional KBM and DBM and supplies the dynamic-KBM method that the point-cloud mapping is contrasted against.","marker":"[46]"},{"why":"Establishes the semi-resolved kernel-based approach and the fine-grid DPVM oscillation problem that motivates the two-step mapping.","marker":"[31]"},{"why":"Provides the Euler-Lagrange mollification and coarse-graining formulation from which the kernel-smoothing step descends.","marker":"[44]"},{"why":"Supplies the diffusion-based coarse-graining alternative and the no-flux boundary treatment adapted for physical walls.","marker":"[43]"},{"why":"Supplies the convolution-based coarse-graining formalism that the first mapping step turns into a computational quadrature.","marker":"[56]"},{"why":"Demonstrates that initial volume fraction controls immersed granular collapse via pore pressure feedback, the phenomenon the method must capture.","marker":"[10]"},{"why":"Shows that dilation-induced negative pore pressure governs the initiation of underwater granular avalanches, the mechanism tested in the dense collapse validation.","marker":"[11]"},{"why":"Provides the fully resolved LBM-DEM reference data for the immersed granular column collapse used to validate pore pressure and initiation times.","marker":"[71]"},{"why":"Supplies the experimental single-sphere settling data used to validate the method's grid convergence.","marker":"[66]"},{"why":"Provides the bi-disperse fluidized-bed experimental data used for validation on fine and coarse grids.","marker":"[69]"}],"fun_headline_variants":["Fibonacci point clouds decouple CFD-DEM from grid size","Two-step coarse graining captures pore-pressure delay","Point-cloud mapping ends grid-size dependence in CFD-DEM","Sub-grid particle displacement key to dense collapse delay"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method assumes that the fixed point-cloud resolution (96 outermost points, layers every $0.25d$) samples the truncated Gaussian kernel finely and uniformly enough for every particle; the paper reports no sensitivity study for these parameters, so if they are too coarse the claimed grid-independence and pore-pressure sensitivity would weaken.","fun_headline_variants_meta":{"raw":{"variants":["Fibonacci point clouds decouple CFD-DEM from grid size","Two-step coarse graining captures pore-pressure delay","Point-cloud mapping ends grid-size dependence in CFD-DEM","Sub-grid particle displacement key to dense collapse delay"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000906,"raw_usage":{"total_tokens":3936,"prompt_tokens":1022,"completion_tokens":2914,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":638,"completion_tokens_details":{"reasoning_tokens":2851}},"tokens_in":638,"tokens_out":2914,"duration_ms":22896,"temperature":1.0,"reasoning_tokens":2851,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:46:48.433998+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the dense immersed granular column collapse ($\\phi_i=0.6233$) and the 1D weight-allocation error test with much coarser point clouds, e.g., $N_o=24$ or layer spacing $0.5d$; if the negative pore pressure, delayed initiation, and the two-orders-of-magnitude error reduction all survive, the point-cloud parameters are not load-bearing, and if they degrade, the central claim fails.","supporting_citations":[{"cited_title":"Eshraghi, E","cited_arxiv_id":null,"evidence_quote":"Identifies the coarse-grid limitations of conventional KBM and DBM and supplies the dynamic-KBM method that the point-cloud mapping is contrasted against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the semi-resolved kernel-based approach and the fine-grid DPVM oscillation problem that motivates the two-step mapping."},{"cited_title":"Capecelatro, O","cited_arxiv_id":null,"evidence_quote":"Provides the Euler-Lagrange mollification and coarse-graining formulation from which the kernel-smoothing step descends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the diffusion-based coarse-graining alternative and the no-flux boundary treatment adapted for physical walls."},{"cited_title":"Goldhirsch, Stress, stress asymmetry and couple stress: from discrete particles to continuous fields, Granul","cited_arxiv_id":null,"evidence_quote":"Supplies the convolution-based coarse-graining formalism that the first mapping step turns into a computational quadrature."},{"cited_title":"Rondon, O","cited_arxiv_id":null,"evidence_quote":"Demonstrates that initial volume fraction controls immersed granular collapse via pore pressure feedback, the phenomenon the method must capture."},{"cited_title":"Pailha, M","cited_arxiv_id":null,"evidence_quote":"Shows that dilation-induced negative pore pressure governs the initiation of underwater granular avalanches, the mechanism tested in the dense collapse validation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the fully resolved LBM-DEM reference data for the immersed granular column collapse used to validate pore pressure and initiation times."},{"cited_title":"Ten Cate, C","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental single-sphere settling data used to validate the method's grid convergence."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the bi-disperse fluidized-bed experimental data used for validation on fine and coarse grids."}],"review_version":1}