{"id":"07808cc0-d776-436b-a090-0b8585dbb5b0","arxiv_id":"1908.08383","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Using volume-of-fluid simulations with and without a collision-suppression force, the authors show that droplet coalescence is responsible for the negative curvature of the emulsion viscosity curve and that suppressing coalescence makes emulsions behave like suspensions of deformable particles.","lead":"This numerical study shows that when droplets in an emulsion can merge, the emulsion becomes less viscous at high droplet concentrations, while preventing merging makes it more viscous. The difference is traced to the amount of droplet surface area, which shrinks during coalescence and reduces the surface tension contribution to the shear stress.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The repulsive collision force used to define the non-coalescing branch contributes about 10% of the total stress and changes wall-normal droplet distribution, so the curvature reversal may be partly a model artifact rather than a pure coalescence effect.","rationale":"The reader identified the innocuousness of the Eulerian collision force as the weakest assumption, and the full text supports that assessment. The force is introduced as a numerical device, calibrated by trial and error, and its direct contribution to the stress budget is not negligible: the paper reports about 10% of the total stress at phi=0.3. Moreover, the force changes the spatial distribution of the dispersed phase, with a substantial increase of the near-wall volume fraction in the non-coalescing cases. Since the reported effective viscosity is obtained from the wall shear stress, this redistribution enters the measured quantity directly. Thus the non-coalescing branch is not a clean counterfactual that isolates coalescence alone. The paper has genuine strengths: the comparison with Caserta and Guido (2012) is qualitative but relevant, the surface-area time histories in Fig. 9 directly support the proposed interfacial-area mechanism, the r/2 test checks size effects for the coalescing branch, and the Eilers-formula collapse in Appendix B supports the interpretation of the non-coalescing branch as suspension-like. Yet none of these checks addresses the force-amplitude sensitivity of the rheology curves or the force-induced wall accumulation. Because the concern is load-bearing but not fatal, the conditional verdict is appropriate; no change to the reader's verdict is needed.","tokens_in":18633,"tokens_out":6458,"duration_ms":78826,"concrete_test":"Re-run the non-coalescing rheology cases at Ca=0.1, lambda=1 and phi=0.2 and 0.3 with two additional collision-force amplitudes, e.g., halved (a=27.5, b=1.75) and doubled (a=110, b=7), while verifying that no coalescence occurs, and compare the effective viscosity, stress decomposition, and wall-normal volume fraction profiles. If the effective viscosity changes by more than roughly the 10% direct force contribution, or if the positive curvature of mu_e(phi) weakens or disappears, then the force model is not an innocuous control and the central comparison is compromised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the non-coalescing branch be a clean control that differs from the coalescing branch only by the absence of merging. In this paper that control is produced by the Eulerian repulsive force of Eq. (3.1), with coefficients a=55 and b=3.5 fixed by trial and error. The force is not passive: Eq. (4.8) adds its wall-normal integral C to the stress balance, and Fig. 8 shows that this term is about 10% of the total shear stress at phi=0.3, i.e., of the same order as the physical stress contributions being compared. In addition, Fig. 9 (right) shows a significant increase of the near-wall volume fraction in the cases with the collision force. Because the effective viscosity is evaluated from the wall shear stress via Eq. (4.1), this force-induced redistribution directly contributes to the measured viscosity increase. The two-drop calibration test in Fig. 4 shows only that two different force amplitudes give similar post-collision trajectories; it does not test the sensitivity of the rheology curves to the force amplitude in dense systems, nor does it quantify how much the force changes drop deformation and near-field flow in the many-drop simulations. The paper itself reports the 10% stress contribution and acknowledges that only the two limiting coalescence efficiencies are considered, but no convergence study in force strength is provided for the suspension rheology. Therefore the sign change of the viscosity curvature may be partly caused by the specific force model rather than by inhibition of coalescence alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"De Vita et al. present volume-of-fluid simulations of shear-driven emulsions in a Couette cell at vanishing Reynolds number, scanning volume fraction φ=0.00164–0.3, viscosity ratio λ=0.01–1, and capillary number Ca=0.05–0.2. Two branches are compared: one in which droplets coalesce naturally, and one in which an Eulerian repulsive force (Eq. 3.1) suppresses coalescence, intended to mimic surfactant-stabilized emulsions. The main finding is that the effective viscosity versus φ curve has negative curvature when coalescence is allowed, with a maximum near φ≈0.2 and values below unity for λ<1, and positive curvature when coalescence is suppressed, resembling suspensions of deformable particles. The authors attribute the difference to the reduction of interfacial area upon merging, which lowers the interfacial-tension contribution to the shear stress. Normal stress differences, flow-topology statistics, and droplet size distributions are also reported.","tokens_in":18890,"tokens_out":6665,"duration_ms":63593,"significance":"If the central comparison were clean, the paper would be a valuable contribution: it gives a mechanistic explanation of the experimentally observed negative curvature (Caserta & Guido 2012) and demonstrates that neglecting coalescence in emulsion simulations changes the qualitative rheology. Strengths include the careful stress-budget derivation (Eqs. 4.2–4.8), grid-resolution checks, and benchmarks against Einstein, Taylor, and the Pal model in the dilute regime. The coalescing branch is compared to experiments with reasonable qualitative agreement. The principal caveat is that the non-coalescing control is not a passive variation: the collision force contributes directly to the stress balance and alters the wall-normal droplet distribution, so the strength of the causal claim (that coalescence causes the curvature reversal) depends on the inertness of that force model.","major_comments":[{"comment":"The effective viscosity is computed from the wall shear stress via Eq. (4.1), while the bulk stress balance in Eq. (4.8) includes the collision-force integral C. Because the collision force is a numerical body force whose streamwise component is about 10% of the total shear stress at φ=0.3, the reported µ_e for the non-coalescing branch is not strictly a material property of the emulsion. The authors should provide a convergence study in the force strength (e.g., increasing a and b beyond 55 and 3.5) to demonstrate that the suspension viscosity is independent of the force amplitude as long as coalescence is prevented. Without this, the positive curvature of the non-coalescing curve could be partly an artifact of the model force.","section":"§4.1, Eqs. (4.1), (4.8), Fig. 8"},{"comment":"The wall-normal profile of the average volume fraction is substantially modified by the collision force, with a marked increase of <φ> near the walls. Since µ_e is evaluated from wall quantities, this force-induced segregation directly contributes to the measured viscosity increase, independently of the interfacial-area mechanism emphasized in the paper. The statement in §4 that the dispersed phase is 'approximately homogeneous' is difficult to reconcile with the changes in near-wall <φ> shown in Fig. 9(right). The authors should quantify the wall-region contribution to µ_e (e.g., by computing µ_e from the bulk stress instead and comparing) or demonstrate that the results are insensitive to the redistribution.","section":"§4.1, Fig. 9(right)"},{"comment":"The calibration of the collision force coefficients a and b (Eq. 3.1) is done by trial and error on a single two-drop collision, and the only reported sensitivity test compares two force amplitudes on the post-collision trajectory. This does not establish that the force is dynamically inert in the dense many-drop suspensions: the force acts in a shell of thickness 3Δ around every interface, can alter the local flow in the gaps, and changes the deformation of droplets. A sensitivity test of the suspension rheology and of the stress budget to a and b in a dense case (e.g., φ=0.3) is needed to support the claim that the difference between the two branches is due solely to the presence or absence of coalescence.","section":"§3, Fig. 4"}],"minor_comments":[{"comment":"The advection equation for the VoF function appears to contain a typo: ∂(u_j H)/∂x_j should likely be ∂(u_j T)/∂x_j, or the right-hand side should involve H ∂u_j/∂x_j, since H is the color function and T is the cell average. Please check the notation.","section":"Eq. (2.3)"},{"comment":"The rows in the caption are mislabeled: the third row is described twice as '(d)-(e)-(f)', and the fourth row should be '(g)-(h)-(i)' or '(j)-(k)-(l)' accordingly.","section":"Fig. 3 caption"},{"comment":"The affiliation contains a typo: 'Stockhom' should be 'Stockholm'.","section":"Author affiliations"},{"comment":"The black diamonds representing the experiments of Caserta & Guido are not identified in the caption; please state explicitly what they denote.","section":"Fig. 14 caption"},{"comment":"The statement that the ratio of N1 and N2 is 'almost constant' with volume fraction is supported by the figure, but for λ=1 the scatter is non-negligible; consider providing a quantitative measure such as a best-fit slope with confidence intervals.","section":"§4.2, Fig. 13"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the coalescing branch is well benchmarked. The main decision-relevant issue is the non-coalescing control: the collision force is not demonstrably passive, and the wall-normal concentration change could affect the viscosity measured at the wall. I would not reject if the authors can supply a force-amplitude convergence study and quantify the wall-segregation effect, but the current manuscript leaves that load-bearing point open."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this is a careful numerical study that does something new—it suppresses droplet coalescence with an Eulerian repulsive force and shows the viscosity-vs-volume-fraction curve flips from negative to positive curvature. The mechanism, that coalescence reduces total interfacial area and thereby lowers the interfacial stress contribution, is physically sensible. The coalescing branch is independently benchmarked against the Pal model and Caserta & Guido's experiments, with reasonable agreement in the dilute regime.\n\nThe paper earns credit for the stress-budget derivation that explicitly includes both the interfacial and collision-force contributions, for the grid-convergence check, and for the two-drop calibration showing that doubling the force changes the post-collision trajectory by only about 3%. The collapse of the non-coalescing data onto the Eilers formula via an effective volume fraction is a nice supporting result.\n\nThe soft spot is exactly what the stress-test note says: the collision force is not passive. At phi=0.3 it contributes about 10% of the wall shear stress, and Fig. 9(right) shows it pushes droplets toward the wall, which directly affects the wall-based viscosity measurement. The authors report both facts honestly, but they do not provide a force-amplitude sensitivity study in dense suspensions. The two-drop test does not tell us whether the suspension rheology itself is sensitive to the force coefficients. So the clean-control assumption is only partially validated. I would want at least one dense run at half and double force amplitudes to see whether the curvature sign is robust.\n\nThat said, the concern is not fatal. The coalescing branch is benchmarked independently, and the non-coalescing branch behaves like deformable particles in multiple ways: the Eilers collapse, the emergence of extensional-flow regions, and the droplet size distribution staying monodisperse. The mechanism is consistent with the stress budget. The main caveat is quantitative: the exact viscosity shift is probably contaminated by the force model, but the sign change is likely real.\n\nOne minor issue: the abstract says the first normal stress difference is positive without noting the exception for the lowest viscosity ratio, which the main text reports. That should be fixed.\n\nThis paper deserves a serious referee. It is a useful, honest contribution for people working on emulsion rheology or drop-laden flows, and I would cite it if I were studying coalescence effects. It would also make for a good reading-group discussion about what constitutes a clean numerical control. My recommendation: send it to peer review, ask for a force-amplitude sensitivity test and an abstract correction.","headline":"A solid VoF study showing coalescence flips the curvature of the emulsion viscosity curve, but the artificial repulsive force that defines the 'no-coalescence' branch is not a fully clean control.","tokens_in":19431,"tokens_out":2921,"would_cite":true,"duration_ms":30675,"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":"Coalescence, not just droplet deformation, is what makes emulsion viscosity rise to a peak near 20% volume fraction and then fall.","keywords":["emulsion rheology","coalescence","effective viscosity","interfacial tension","volume of fluid method","collision force","normal stress differences","shear flow"],"falsifier":"A decisive check is to repeat the volume-fraction series with a second, force-free way of suppressing coalescence—for example an interface model with insoluble surfactant transport and strong Marangoni resistance—and compare the effective-viscosity curve and stress budget. If the positive curvature and the roughly 10% collision-force stress contribution do not reproduce, the reported curvature flip is an artifact of the repulsive model. A more local test: in the two-droplet collision of figure 4, the forced case differs by about 6% in final vertical displacement from the double-resolution unforced case; refining the grid should reduce this difference if the force is inert, or expose that the force itself alters the collision outcome.","tokens_in":18375,"feed_emoji":"💧","tokens_out":7256,"duration_ms":63795,"temperature":0.7,"pith_summary":"The paper tries to establish that coalescence of droplets is the mechanism behind an observed anomaly in emulsion rheology: the effective viscosity increases with disperse-phase concentration up to about 20% and then decreases, producing a negatively curved viscosity curve. Using interface-resolving simulations of shear flow, the authors compare emulsions in which droplets merge freely with identical emulsions in which an Eulerian repulsive force prevents merging. With merging suppressed, the viscosity curve becomes positively curved and always lies above the matrix viscosity, matching the behaviour of suspensions of deformable particles. The paper attributes the difference to a geometric-stress mechanism: coalescence reduces the total interfacial area, and the interfacial-tension contribution to the bulk shear stress shrinks with that area. If the claim is right, any prediction of emulsion rheology that neglects merging will be wrong outside the dilute regime.","feed_headline":"Coalescence alone bends an emulsion's viscosity curve downward","feed_subtitle":"Shear simulations show viscosity peaking near 20% concentration and dropping when droplets merge; blocking merging flips the curve.","key_machinery":"The load-bearing device is an Eulerian repulsive force of lubrication form, Fc = μ0 r U (a/ψ + b/ψ²) n, written in terms of the signed distance ψ from the interface; it is applied only where two different droplet indices are detected within a stencil, and it models the effect of surfactants by preventing film drainage and merging. The force is hand-tuned (a = 55, b = 3.5) so that coalescence is fully suppressed with minimal side effects. The second piece is the stress-budget identity that equates the wall shear stress with the bulk average of viscous, interfacial-tension, and collision-force stresses; combined with the time history of the total interface area, this identity is what lets the authors attribute the viscosity change to a reduction of interfacial area. A third element is the flow-topology parameter Q = (D² − Ω²)/(D² + Ω²), used to show that the matrix flow is nearly pure shear when droplets merge and gains extensional regions when merging is suppressed.","core_discovery":"The central claim is that droplet coalescence causes the negative curvature of the effective-viscosity versus volume-fraction curve in emulsions. In the simulations, free merging produces a maximum of the effective viscosity near φ ≈ 0.2; for viscosity ratio λ = 0.01 the normalized viscosity even falls below 1 at high volume fraction. Applying the repulsive collision force (3.1) with coefficients a = 55 and b = 3.5 prevents merging altogether, and the same system then shows a monotonic, positively curved viscosity curve that is well described by the Eilers formula once droplet deformation is folded into an effective volume fraction. The paper traces this to the stress budget: with coalescence the total interface area drops by up to 80% of its initial value, lowering the interfacial-tension contribution to the shear stress; without coalescence the interface area stays nearly constant, and the interface-tension term accounts for about half of the effective viscosity, with the collision force itself contributing about 10%.","pith_inferences":["If the mechanism is generic, coalescence efficiency should be treated as a rheological state variable as important as volume fraction and capillary number; any processing change that alters drop-size distribution—surfactant dose, pre-shear history, compatibilizer—will slide the system along the family of curves between the two limits computed here.","The reported 10% stress contribution of the artificial repulsive force is a directly testable artifact check: a second, force-free way to inhibit coalescence (for instance an interface with strong Marangoni resistance) should reproduce the positive curvature and the same stress-budget decomposition.","The area-reduction explanation predicts a quantitative scaling: the drop in effective viscosity should track the loss of total interfacial area, so simultaneous measurement of drop size and shear stress during coalescence would give a direct test of the mechanism.","For viscosity ratios below one, coalescence can make an emulsion less viscous than its continuous phase; engineering controlled coalescence could therefore be a purely microstructural route to viscosity reduction without changing chemistry."],"forward_implications":["Emulsion constitutive curves must be reported with coalescence efficiency controlled; two emulsions with identical composition, drop size, and shear rate can have qualitatively different viscosity curves depending on whether merging is allowed.","Analytical viscosity formulas that assume positive curvature, such as Pal's equation and the Eilers formula, should only be applied to non-coalescing emulsions; the paper shows they fail when coalescence is active.","Suppressing coalescence makes an emulsion behave like a suspension of deformable particles: the viscosity data collapse onto the Eilers formula using an effective volume fraction based on droplet deformation, with about 6% error at the highest concentration.","The ratio of first to second normal stress difference stays roughly constant with volume fraction, and the first normal stress difference changes sign at the lowest viscosity ratio—an experimentally checkable rheological fingerprint.","Flow topology changes with coalescence: merging droplets leave the matrix in nearly pure shear, whereas non-coalescing emulsions develop extensional flow in the gaps between droplets."],"supporting_citations":[{"why":"Supplies the experimental observation of negative curvature in the viscosity curve and the vorticity-banding context that motivates the study.","marker":"Caserta & Guido (2012)"},{"why":"Supplies the Volume-of-Fluid solver, validation, and the emulsion simulation framework this paper extends to viscosity ratios and coalescence control.","marker":"Rosti et al. (2019)"},{"why":"Supplies the lubrication-force form on which the Eulerian repulsive collision model (3.1) is based.","marker":"Bolotnov et al. (2011)"},{"why":"Supplies the deformable-particle suspension data and the effective-volume-fraction scaling that the non-coalescing results collapse onto.","marker":"Rosti & Brandt (2018)"},{"why":"Supplies the concentrated-emulsion viscosity equation used as the analytical baseline that fails when coalescence is active.","marker":"Pal (2003)"},{"why":"Supplies the bulk-stress decomposition used to compute normal stress differences from the simulation fields.","marker":"Batchelor (1970)"},{"why":"Shows that the wall-normal velocity gradient equals the bulk shear stress, justifying the effective-viscosity measurement.","marker":"Yang et al. (2016)"},{"why":"Supplies the conceptual collision framework (external flow versus film drainage) that defines when coalescence occurs.","marker":"Chesters (1991)"}],"fun_headline_variants":["Droplet merging bends emulsion viscosity curve downward","Coalescence drags emulsion viscosity down at high volume fractions","Merging droplets cut emulsion viscosity by shrinking interface area","Blocking coalescence flips the viscosity curve upward, simulations show","Surfactant effect: stopping droplet merging makes emulsions thicker"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the hand-tuned Eulerian repulsive force with coefficients a = 55 and b = 3.5 prevents coalescence completely without otherwise changing the droplet dynamics, interface deformation, or the stress budget; the paper itself reports that this force contributes about 10% of the total stress, so a perturbing effect large enough to alter the comparison cannot be excluded from the reported evidence.","fun_headline_variants_meta":{"raw":{"variants":["Droplet merging bends emulsion viscosity curve downward","Coalescence drags emulsion viscosity down at high volume fractions","Merging droplets cut emulsion viscosity by shrinking interface area","Blocking coalescence flips the viscosity curve upward, simulations show","Surfactant effect: stopping droplet merging makes emulsions thicker"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000579,"raw_usage":{"total_tokens":2781,"prompt_tokens":1053,"completion_tokens":1728,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":669,"completion_tokens_details":{"reasoning_tokens":1645}},"tokens_in":669,"tokens_out":1728,"duration_ms":11253,"temperature":1.0,"reasoning_tokens":1645,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:39:47.292534+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check is to repeat the volume-fraction series with a second, force-free way of suppressing coalescence—for example an interface model with insoluble surfactant transport and strong Marangoni resistance—and compare the effective-viscosity curve and stress budget. If the positive curvature and the roughly 10% collision-force stress contribution do not reproduce, the reported curvature flip is an artifact of the repulsive model. A more local test: in the two-droplet collision of figure 4, the forced case differs by about 6% in final vertical displacement from the double-resolution unforced case; refining the grid should reduce this difference if the force is inert, or expose that the force itself alters the collision outcome.","supporting_citations":[{"cited_title":"Langmuir 28 (47), 16254–16262","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental observation of negative curvature in the viscosity curve and the vorticity-banding context that motivates the study."},{"cited_title":"Acta Mechanica 230 (2), 667–682","cited_arxiv_id":null,"evidence_quote":"Supplies the Volume-of-Fluid solver, validation, and the emulsion simulation framework this paper extends to viscosity ratios and coalescence control."},{"cited_title":"& Podowski, Michael Z","cited_arxiv_id":null,"evidence_quote":"Supplies the lubrication-force form on which the Eulerian repulsive collision model (3.1) is based."},{"cited_title":"Journal of Fluid Mechanics 41, 545—-&","cited_arxiv_id":null,"evidence_quote":"Supplies the bulk-stress decomposition used to compute normal stress differences from the simulation fields."},{"cited_title":"2016 Numerical simulations of the rheology of suspensions of rigid spheres at low volume fr action in a viscoelastic fluid under shear","cited_arxiv_id":null,"evidence_quote":"Shows that the wall-normal velocity gradient equals the bulk shear stress, justifying the effective-viscosity measurement."}],"review_version":1}