{"id":"3957b3d9-6a72-4eac-b5cd-41a219d7c95a","arxiv_id":"2411.16005","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Vortex shedding onset behind a cylinder in a Bingham plastic fluid is disturbance-dependent (subcritical), producing hysteresis in drag and heat transfer, and the average Nusselt number follows a Carreau-Yasuda-like curve in Bingham number.","lead":"This paper uses computer simulations to show that a heated cylinder in a Bingham plastic fluid can switch between steady flow and vortex shedding at different Bingham numbers depending on how the flow is started, a sign of subcritical bifurcation. The result matters because yield-stress fluids are common in food, cosmetics, and drilling, where predicting when unsteady heat transfer begins changes equipment design.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Hysteresis gap may be a finite-time or regularization artifact: without a linear stability or disturbance-amplitude study, the BncI–BncD gap does not prove subcritical bifurcation.","rationale":"The paper's strongest claim is the subcritical bifurcation and hysteresis. The evidence is the different Bnc values from two continuation protocols. The logic is: if the only difference between the protocols is disturbance intensity, then different outcomes imply a finite-amplitude threshold. However, the protocols may differ in other ways: the IB branch carries a fully developed vortex-shedding field into the next case, while the DB branch starts from a converged steady state whose unyielded regions contain extremely high regularized viscosity. The finite simulation window is a confound because the DB steady state may simply not have been integrated long enough for a small linear instability to emerge; the regularized viscosity may suppress that instability artificially. Tables 1 and 2 verify time-step and M convergence only for a single unsteady shedding case, not for the transition threshold itself, and no mesh-independence study is reported anywhere. The reader's weakest assumption already identified this class of issue; my stress test sharpens it into a specific, testable failure mode. A linear stability analysis or a controlled disturbance-amplitude study would settle the matter, but the paper does neither. Therefore the central claim is conditional: it is plausible and qualitatively consistent with Mossaz et al.'s larger-disturbance BncI trend, but it is not yet established. The verdict should remain CONDITIONAL, with the concrete checks above as required before the hysteresis claim can be accepted.","tokens_in":22367,"tokens_out":6511,"duration_ms":60826,"concrete_test":"Pick a representative hysteresis point, e.g., Re=100 with Bn midway between BncI and BncD. Run the DB protocol but (i) extend the simulation time to at least 50–100 times the apparent shedding period used to declare statistical stationarity, and (ii) repeat with M U∞/D = 10^6 and with a 2× refined mesh. If Clrms grows from ~0 to a finite value within the extended time in any of these runs, the 'steady' DB state is a finite-time or regularization artifact and the subcritical-bifurcation claim fails. Independently, perform a disturbance-amplitude test at the same Bn: start from the steady DB solution, add a finite-amplitude perturbation of controlled size (e.g., brief cylinder rotation or an impulsive vertical velocity patch), and determine whether a finite amplitude threshold exists below which the flow returns to steady and above which it sheds.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that Bingham-fluid shedding onset is subcritical rests entirely on the observation that BncI (increasing Bn from a Newtonian shedding state) exceeds BncD (decreasing Bn from a steady Bingham state) for Re ≥ 60. This inference presumes that each initialization protocol converges to the true asymptotic state of the governing equations within the simulation window. That presumption is not tested. Near a bifurcation the growth or decay rate of perturbations is small, so a finite integration window (the paper declares stationarity without quantifying transient duration) can misclassify a slowly growing instability as steady in the DB branch, or a slowly decaying perturbation as sustained shedding in the IB branch, producing a spurious hysteresis gap. The risk is compounded by the Papanastasiou regularization: with M U∞/D = 10^5, the effective viscosity in nominally unyielded regions reaches O(μB(1 + M·Bn)), an enormous artificial viscosity that can artificially stabilize the steady branch during the DB continuation. The convergence checks in Tables 1 and 2 are performed at (Re,Bn)=(100,5), a shedding case, not at the critical transition, so they do not constrain either artifact at BncD or BncI. No linear stability analysis, controlled finite-amplitude disturbance study, M-sensitivity study at the transition, or mesh-independence study is provided. Hence the BncI–BncD gap may reflect finite-time statistics or regularization/basin effects rather than a true subcritical bifurcation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a numerical study of flow and heat transfer past a heated circular cylinder in a Bingham plastic fluid over Re = 10–180, Pr = 1–100, and Bn = 0–10^4, using the Papanastasiou regularization in a finite-volume solver. The authors validate their method against Newtonian cylinder data (Rec = 46.1, Stc = 0.1168, drag and Nusselt curves) and steady Bingham results from Nirmalkar & Chhabra and Mossaz et al. The central new claim is that, for Re ≥ 60, the onset of vortex shedding is subcritical: the critical Bingham number depends on the initialization protocol, with BncI (increasing Bn from a Newtonian shedding state) larger than BncD (decreasing Bn from a steady Bingham state), leading to hysteresis and abrupt jumps in the time-averaged drag coefficient and Nusselt number. The paper also reports a Carreau-Yasuda-like correlation for Nu as a function of Bn, and various local and global heat transfer characteristics.","tokens_in":22759,"tokens_out":4760,"duration_ms":44618,"significance":"If the subcritical-bifurcation claim is correct, it is a significant and non-obvious result for viscoplastic fluid mechanics, with practical consequences for heat-transfer control and for interpreting simulations of Bingham flows past bluff bodies. The manuscript's Newtonian validation is convincing, and the steady-state Bingham drag and Nusselt comparisons with published data are within 3–10%, giving reasonable confidence in the baseline solver. The paper also provides a broad parametric map of yielded/unyielded regions, shear-rate distributions, and local Nusselt numbers that could be useful for benchmarking. However, the central hysteresis claim is not yet backed by direct stability or amplitude-disturbance evidence; the present evidence is consistent with a subcritical bifurcation but also with finite-time and regularization artifacts. Because the central claim is load-bearing, the current evidence is not sufficient for acceptance without revision.","major_comments":[{"comment":"The subcritical-bifurcation conclusion rests entirely on the observation that BncI and BncD differ for Re ≥ 60. The paper does not establish that the DB branch has reached the true asymptotic state. Near a bifurcation, perturbation growth rates are small, so a finite integration window can misclassify a slowly growing instability as steady in the DB branch and a slowly decaying disturbance as sustained shedding in the IB branch. The manuscript states only that 'the flow reaches statistical stationary state' (Section 2.3) without giving a quantitative stationarity criterion or reporting the transient duration at the transition points (e.g., Re = 180, Bn = 1.8 vs. 1.77). Without a linear stability analysis, a controlled finite-amplitude disturbance study, or at minimum a time-history check showing that the observed state is invariant over an extended window, the BncI–BncD gap does not prove subcritical bifurcation and may reflect finite-time statistics.","section":"Section 3.2.1, Fig. 6"},{"comment":"The Papanastasiou regularization parameter M U∞/D = 10^5 gives an effective viscosity in nominally unyielded regions of order 1 + Bn·M U∞/D, which for Bn ≈ 2–3 is O(10^5) times the plastic viscosity. This large artificial viscosity can artificially stabilize the steady branch and thereby shift BncD downward. The M-sensitivity check in Table 2 is performed at (Re, Bn) = (100, 5), a strongly unsteady shedding case, and does not constrain the behavior near BncI or BncD. A dedicated M-independence study at the transition points, or the use of an augmented-Lagrangian / unregularized method for at least a subset of cases, is needed to rule out a regularization-induced hysteresis interval.","section":"Section 2.3, Table 2"},{"comment":"The mesh-independence of the reported hysteresis and jumps is not demonstrated. The mesh description (209,600 cells, first-cell spacing 0.0025D) is given, but no comparison with a finer or coarser mesh is presented. The Newtonian and steady Bingham validations at moderate parameters do not guarantee that the sharp transitions near Bnc are mesh-independent, especially because the yielded/unyielded boundary location is sensitive to the resolved velocity gradients in the regularization framework. A mesh-refinement study for a representative case such as Re = 180 near BncD and BncI is required to support the quantitative claims about abrupt jumps in Cd and Nu.","section":"Section 2.1"},{"comment":"The claim that the Nu–Bn data 'fits well with the Carreau-Yasuda-like non-Newtonian viscosity model' is a fitting statement, not a predictive validation. The parameters Nu0, Nu∞, n, and λ in Eqs. (38a)–(38d) are calibrated on the same simulation data that are then compared with Eq. (37) in Fig. 19. An error of less than 5% for the steady branch only confirms that the four-parameter function is sufficiently flexible to interpolate the data. The paper should state explicitly that Eq. (37) is a correlating equation and should give its intended range of applicability; it should not be presented as evidence for a physical analogy with the Carreau-Yasuda model without independent data.","section":"Section 3.2.2, Eqs. (37)–(38d)"}],"minor_comments":[{"comment":"The yield criterion 'the flow yields when μ/μB < 10^5.4' is introduced without justification. The paper should explain how this threshold was chosen and whether the yielded-region morphology is sensitive to the choice within, say, one order of magnitude.","section":"Section 3.2.1"},{"comment":"The criterion for 'statistical stationary state' is qualitative. Please provide a quantitative definition, such as convergence of time-averaged Cd and Nu over a sliding window, and report the total simulation time in convective units for the reported cases.","section":"Section 2.3"},{"comment":"The piecewise Nu–Re correlation for Newtonian flow is discontinuous at Rec; the paper attributes this to the steady-to-unsteady transition. It would be helpful to state that the correlation is not intended to be continuous at Rec and to clarify whether the two branches are valid in the immediate vicinity of Rec.","section":"Section 3.1, Eq. (28)"},{"comment":"There is a typographical error: 'Bin gham' should be 'Bingham' in the first paragraph of the Conclusion.","section":"Section 4"},{"comment":"The paper contains several grammatical and punctuation errors (e.g., 'It should be note that', 'the flow is checked to be steady'). A careful language edit is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a relevant problem and the validation effort is commendable, but the central subcritical-bifurcation claim currently rests on a numerical-hysteresis observation that could be an artifact of the regularization and finite-time statistics. The requested additional studies (M-sensitivity at the transition, longer-time checks, mesh refinement, and ideally a linear stability or forced-disturbance analysis) are within the scope of the manuscript and would either confirm the claim or significantly change the interpretation. The Carreau-Yasuda-like correlation is a useful engineering fit but should be framed as such. I recommend major revision rather than rejection because the issue is fixable with additional numerical evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a careful numerical parameter sweep of vortex shedding and heat transfer for a heated cylinder in a Bingham fluid, and the genuinely new result is the hysteresis reported between two continuation protocols (IB vs DB) for Re ≥ 60, with BncI ≠ BncD and abrupt jumps in drag and Nusselt number. That claim is plausible but not yet nailed down.\n\nThe paper does a lot right. Newtonian validation is solid: Rec = 46.1, St = 0.1168, and the drag and Nusselt curves match published data. The steady Bingham results agree with Nirmalkar & Chhabra and Mossaz et al. within 3–10%, and the BncI line is close to Mossaz et al.'s linear fit. The Carreau–Yasuda-like Nu–Bn correlation (Eq. 37) is a useful engineering fit with tabulated coefficients, and the authors are honest that it is a fit rather than a derived law.\n\nThe soft spot is the load-bearing inference. The subcritical-bifurcation conclusion rests entirely on the observation that two initialization protocols give different critical Bn values. That is suggestive, but it is not a stability analysis. Near a bifurcation, transient growth or decay is slow, and the paper does not quantify how well the time-marching has converged to the asymptotic state near BncI and BncD. The Papanastasiou parameter M was checked at (Re, Bn) = (100, 5), a strongly shedding case, not at the transition, where artificial viscosity in nominally unyielded regions could artificially stabilize the steady DB branch. There is also no mesh-independence study, only time-step and M checks. The authors do not claim to have done a linear stability or disturbance-amplitude analysis, but without it the hysteresis gap is a numerical observation in search of a mechanism. The close match with Mossaz et al. for BncI helps, but it does not validate the gap itself.\n\nThese are not fatal flaws in an engineering paper, but they mean the central claim should be read as conditional. The correlation and the Bnc maps are worth having regardless. I would send this to peer review, with a request that the authors either add a linear stability or controlled disturbance-amplitude test, or soften the subcritical-bifurcation wording. People working on viscoplastic bluff-body flows and heat-transfer correlations will get real value from the data.","headline":"Solid, well-validated numerics with a plausible but unproven subcritical-bifurcation claim that needs a stability analysis before it is taken as settled.","tokens_in":23345,"tokens_out":1896,"would_cite":true,"duration_ms":19349,"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":"The paper claims that vortex shedding past a heated cylinder in a Bingham plastic fluid is a subcritical bifurcation: for $Re \\geq 60$, steady and shedding wakes coexist over a range of Bingham numbers, so drag and heat transfer jump…","keywords":["Bingham plastic fluid","vortex shedding","subcritical bifurcation","hysteresis","yield stress","circular cylinder","heat transfer","Nusselt number"],"falsifier":"Linearize the steady Bingham base flow around the cylinder at $Re \\geq 60$ and compute its most unstable eigenmode. If the steady state is linearly unstable for $Bn$ below $Bn_{cI}$, then a fully converged steady-state initialization should spontaneously develop shedding, contradicting the claimed stable steady branch up to $Bn_{cI}$. Conversely, if the steady state is linearly stable down to $Bn_{cD}$ and shedding appears only after finite-amplitude disturbances, the subcritical claim is confirmed. An independent check would initialize the flow at a $Bn$ inside $(Bn_{cD}, Bn_{cI})$ with a controlled disturbance sweep and demonstrate two coexisting stable states separated by an unstable threshold.","tokens_in":22116,"feed_emoji":"🌊","tokens_out":13473,"duration_ms":116696,"temperature":0.7,"pith_summary":"This paper tries to establish that vortex shedding behind a heated circular cylinder in a Bingham plastic fluid—a fluid that only deforms once the local shear stress exceeds a yield stress—is history-dependent. For $Re \\geq 60$, the authors find two different critical Bingham numbers depending on whether the Bingham number is increased from a Newtonian shedding state or decreased from a steady yield-stress state, so steady and unsteady wakes coexist on an interval of $Bn$. If the claim is right, the transition is a subcritical bifurcation, and the mean drag coefficient and mean Nusselt number jump discontinuously at the transition, not smoothly. That matters because it means mixing and heat transfer in industrial yield-stress fluids can be controlled by operating history rather than only by the current flow conditions.","feed_headline":"Vortex shedding in Bingham flow is history-dependent","feed_subtitle":"At Re ≥ 60, steady and vortex-shedding wakes coexist over a band of Bingham numbers, so drag and heat transfer jump abruptly.","key_machinery":"The argument is carried by two coupled tools. The first is a regularized Bingham constitutive model (Papanastasiou regularization) that replaces the discontinuous yield-stress law with a smooth high-viscosity approximation, allowing a single finite-volume solver to represent both yielded and unyielded regions. The second is the pair of initialization protocols, IB and DB, which act as controlled disturbance levels: IB starts from the fully developed Newtonian vortex-shedding field and raises $Bn$, so the shedding perturbation is carried into the yield-stress regime; DB starts from a converged steady Bingham state and lowers $Bn$, so only intrinsic instability can seed new motion. The difference between the two critical values, $Bn_{cI}$ and $Bn_{cD}$, is the signature of the subcritical bifurcation, and the Carreau-Yasuda-like fit for $\\overline{Nu}(Bn)$ is the engineering formula built on the same data.","core_discovery":"The central discovery is that the onset of vortex shedding in a Bingham plastic flow around a cylinder is not a unique critical event but a subcritical bifurcation with a hysteresis loop. For $Re \\geq 60$, increasing $Bn$ along the IB path suppresses shedding at $Bn_{cI}$, while decreasing $Bn$ along the DB path restores shedding only at the smaller value $Bn_{cD}$; between the two, the same parameters admit both a steady wake and a vortex-shedding wake. The root-mean-square lift coefficient jumps from a finite value to zero at $Bn_{cI}$ and from zero to a finite value at $Bn_{cD}$, and the mean drag coefficient and mean Nusselt number follow with abrupt jumps: for example, at $Re=180$ on the IB branch the mean Nusselt number drops from $7.4749$ to $7.3652$ as $Bn$ changes from $2.68$ to $2.7$. Away from the transition, drag and heat transfer vary smoothly with $Bn$, and the Nusselt-number data over the whole range are represented by a Carreau-Yasuda-like correlation (Eq. 37) that reduces to Newtonian and fully plastic limits.","pith_inferences":["Editorial inference: the width of the hysteresis interval, $Bn_{cI}-Bn_{cD}$, should be proportional to the minimum finite-amplitude disturbance needed to trigger shedding; a controlled amplitude-continuation study would map the unstable threshold separating the two attractors.","Editorial inference: if the hysteresis is physical, a transient disturbance such as a short cylinder rotation or acoustic pulse could switch an otherwise steady Bingham wake into persistent vortex shedding, enhancing convective heat transfer without sustained energy input.","Editorial inference: applying the same IB/DB protocol to other yield-stress bluff-body geometries might show that some previously reported steady regimes are metastable, because earlier studies generally used a single start-up condition.","Editorial inference: the paper's discrete $Bn$ steps bracket, rather than resolve, the exact jumps; arc-length continuation or refined sweeps would reveal whether the jumps are true discontinuities or steep but continuous folds."],"forward_implications":["For $Re \\geq 60$, identical flow parameters can produce either a steady or a shedding wake, so numerical and experimental studies of Bingham flow around cylinders should report the initialization or operating history.","The abrupt jumps in $\\overline{C_d}$ and $\\overline{Nu}$ at $Bn_{cI}$ and $Bn_{cD}$ mean that correlations fitted to one branch will fail near the critical Bingham number, and a small change in yield stress or flow rate can switch heat-transfer performance sharply.","The approximately linear critical curve $Bn_{cI} \\approx 0.0201 Re - 0.9993$ provides a quick rule of thumb for when yield stress suppresses vortex shedding.","The Carreau-Yasuda-like Nusselt correlation extends useful engineering prediction from $Bn=0$ to the fully plastic limit for steady flow, with the stated fitting error below 5%."],"supporting_citations":[{"why":"Supplies the steady-state drag and Nusselt number data against which the present solver is validated, and the modified-Reynolds-number scaling that the low-Re* drag law extends.","marker":"Nirmalkar & Chhabra4"},{"why":"Provides the earlier criteria for recirculation and non-stationary regimes behind a cylinder, the linear BncI-Re fit, and the idea of triggering instability with a large disturbance.","marker":"Mossaz et al.15"},{"why":"Introduces the exponential regularization of the yield-stress constitutive law that removes the viscosity discontinuity in every simulation.","marker":"Papanastasiou18"},{"why":"Supplies the parameter-continuation initialization strategy on which the IB and DB protocols are based.","marker":"Peng et al.17"},{"why":"Provides the benchmark Strouhal-number correlation for Newtonian vortex shedding used to validate the unsteady solver.","marker":"Williamson27"},{"why":"Supplies the Newtonian root-mean-square lift coefficient data used to validate the fluctuation statistics.","marker":"Qu et al.34"}],"fun_headline_variants":["Hysteresis in vortex shedding of Bingham fluids","Abrupt jumps in drag and heat for Bingham cylinders","Same flow, two wakes: Bingham hysteresis","Subcritical transition in Bingham cylinder flow","Bingham shedding: history decides the wake"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion rests on the assumption that the IB and DB initialization protocols differ only in the strength of the disturbance they carry, so the gap between $Bn_{cI}$ and $Bn_{cD}$ is a genuine physical hysteresis rather than an artifact of the regularization, time step, or finite-time statistics.","fun_headline_variants_meta":{"raw":{"variants":["Hysteresis in vortex shedding of Bingham fluids","Abrupt jumps in drag and heat for Bingham cylinders","Same flow, two wakes: Bingham hysteresis","Subcritical transition in Bingham cylinder flow","Bingham shedding: history decides the wake"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000663,"raw_usage":{"total_tokens":3127,"prompt_tokens":1143,"completion_tokens":1984,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":759,"completion_tokens_details":{"reasoning_tokens":1916}},"tokens_in":759,"tokens_out":1984,"duration_ms":13722,"temperature":1.0,"reasoning_tokens":1916,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:38:46.668036+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Linearize the steady Bingham base flow around the cylinder at $Re \\geq 60$ and compute its most unstable eigenmode. If the steady state is linearly unstable for $Bn$ below $Bn_{cI}$, then a fully converged steady-state initialization should spontaneously develop shedding, contradicting the claimed stable steady branch up to $Bn_{cI}$. Conversely, if the steady state is linearly stable down to $Bn_{cD}$ and shedding appears only after finite-amplitude disturbances, the subcritical claim is confirmed. An independent check would initialize the flow at a $Bn$ inside $(Bn_{cD}, Bn_{cI})$ with a controlled disturbance sweep and demonstrate two coexisting stable states separated by an unstable threshold.","supporting_citations":[],"review_version":1}