{"id":"6239990f-a230-4c16-8a9d-019c044cab66","arxiv_id":"1908.04640","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"An ANSYS-based iterative simulation method reproduces COMSOL results for trapped currents in superconducting disks under zero-field and field cooling, for both critical-state and flux-creep models.","lead":"This paper demonstrates a new way to simulate the trapped electric currents in high-temperature superconducting disks using the commercial engineering software ANSYS, matching results from a different reference solver. It matters because engineers designing superconducting magnets and trapped-field devices need faster, restartable simulation tools that can include magnetic materials and stress effects.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Critical-state agreement with COMSOL is partly built in: the IAM forces |J|=Jc and the final inner -Jc layer by hand, so only the penetration front is genuinely tested; Section 5 shows the result also depends on a hand-picked ρ0/T1 pair.","rationale":"The reader's weakest assumption and my concern coincide: the critical-state IAM pins J to ±Jc, so the agreement with COMSOL mainly tests the penetration front, not the constitutive model. I sharpen this with two specifics: the final inner -Jc layer is inserted by an extra BFE step, and Section 5 shows sensitivity to ρ0/T1 that is not explained by any convergence theory. These are correctness risks, not accusations: with the published evidence I cannot rule out that the algorithm is a legitimate time-discrete critical-state scheme for this geometry. The proposed slab test would settle it. Since the reader already conditioned acceptance on additional validation and code release, my finding does not change the verdict.","tokens_in":16405,"tokens_out":6278,"duration_ms":67209,"concrete_test":"Run the same IAM on an infinite-slab or long-bar Bean problem with an exact analytic solution (e.g., a 1D slab in a parallel applied field) over a full ZFC field cycle, using the same APDL workflow but without any final BFE forcing; report load-step-converged |J| profiles and the magnetization loop against the analytic Bean prediction for several (ρ0, T1, N1) combinations. If the penetration front is off by more than one mesh element or the loop area changes by more than, say, 5% across the ρ0/T1 scans, then the method is not a general critical-state solver and the central claim should be narrowed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the IAM 'can simulate the critical state model' is only weakly supported because the algorithm manufactures the Bean profile rather than solving for it. After each load step, every over-trapped element is switched to ET-1 and its current is pinned to Jc by the BFE command (Section 2.1c); at the end of the ZFC descent, the inner -Jc layer is explicitly set by an additional BFE forcing step (Section 2.2). Equation (5) is an invented nodal-voltage condition whose only stated role is to keep that inner layer from decaying, with no derivation from Maxwell's equations or the Bean model. Thus the only genuinely predictive quantity in the comparison with COMSOL is the location of the penetration front in one axisymmetric disk geometry. The agreement in Figure 2 is therefore partly circular: both plots look like Bean states because one was forced to look like one. Section 5 reinforces the concern: for the same nominal critical-state case, ρ0=1e-15 Ω·m or T1=5000 s produces a bulk that is 'not well penetrated at the ends' (Figure 16f,i), so the correct-looking solution requires a hand-picked ρ0/T1 pair. Without a derivation that the forcing rule plus Eq. (5) converges to the Bean solution in general, the method's agreement is an interpolation of one benchmark, not a validated critical-state solver.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes an iterative algorithm method (IAM) embedded in ANSYS for simulating magnetization currents in disk-shaped ReBCO bulk superconductors. The method uses the A-V formulation in superconducting regions and the A-formulation elsewhere. For the critical-state model, after each load step the trapped current density in penetrated elements is forced to Jc; for the flux-creep model, element resistivities are updated according to the E-J power law with smoothing coefficients. The authors compare ZFC and FC results with COMSOL H-formulation simulations, report extensions to ferromagnetic inserts, Jc(B), and Jc(ε), and study the influence of load-step number, initial resistivity, ramping time, and updating coefficient.","tokens_in":16666,"tokens_out":5954,"duration_ms":58595,"significance":"If correct, the method offers a practical ANSYS-based alternative for bulk superconductor magnetization modeling, with adjustable computation time, restart capability, and straightforward inclusion of material nonlinearities. The benchmark is performed against a COMSOL H-formulation model with matched parameters rather than fitted to the benchmark, which is a strength. However, the critical-state validation is weakened by the fact that the algorithm enforces |J|=Jc and the final inner -Jc layer by construction, so the comparison tests mainly the penetration-front location. The reported agreement is also qualitative. These limitations do not invalidate the engineering contribution but require substantial additional verification before the central claims can be accepted.","major_comments":[{"comment":"The critical-state ZFC result is in part manufactured by the algorithm rather than solved. After each load step, over-trapped elements are switched to ET-1 and their current is pinned to Jc (Section 2.1c), and at the end of the descending ramp the inner -Jc layer is explicitly forced by an additional BFE step (Section 2.2). Equation (5) is an ad hoc nodal-voltage condition introduced solely to keep that inner layer from decaying, with no derivation from Maxwell's equations or the Bean model. Please provide a derivation or a convergence argument showing that the forcing rule plus Eq. (5) reproduces the Bean critical state for general geometries and field histories; otherwise the only genuinely predictive quantity in the comparison with COMSOL is the location of the penetration front in a single axisymmetric disk.","section":"Sec. 2.1(c), 2.2; Eq. (5)"},{"comment":"The critical-state solutions are not independent of the numerical parameters ρ0 and T1. Figure 16(f) and (i) show that ρ0=10^-15 Ω·m or T1=5000 s produce a bulk that is 'not well penetrated at the ends,' and the text recommends re-assigning a smaller ρ0. For the Bean critical state, the final current distribution should be independent of these parameters, so the good agreement in Figure 2 is conditional on a hand-picked ρ0/T1 pair. Provide a selection criterion derived from the relevant time constants and demonstrate convergence to a parameter-independent solution (e.g., as ρ0 is reduced at fixed ramp time, or as the ramp time is shortened at fixed applied-field history).","section":"Sec. 5.1.2 and 5.1.3; Fig. 16"},{"comment":"The flux-creep results depend strongly on the smoothing coefficient k1, which is not part of the E-J power law. k1=0.9 already produces 'abnormal penetrating elements,' and k1=0.8 or 0.7 yield inhomogeneous, nonconverged profiles. Since the paper claims easy convergence as an advantage, the method needs either a principled rule for choosing k1 (and by extension k2,k3) or a demonstration that the converged solution is independent of these coefficients over a suitable range.","section":"Sec. 5.2.4; Fig. 17(j)-(l)"},{"comment":"All agreement claims with the COMSOL H-formulation are made visually or by counting penetrated elements in the mid-plane; no quantitative error measure is reported. Please add a quantitative comparison (for example the L2 or L∞ norm of the current-density difference over the bulk, or the trapped-field profile) and a convergence study with respect to load-step count and mesh size. Without such measures, 'agree well' remains unsubstantiated.","section":"Sec. 3; Figs. 6, 9, 10"}],"minor_comments":[{"comment":"The caption of Figure 8(b) says 'flux creep model based ZFC magnetization,' but the section and figure describe FC magnetization; correct the label.","section":"Fig. 8 caption"},{"comment":"The Figure 9 caption lists panels (a), (c), (d), (e) while the text refers to (a)-(d); renumber the subfigures consistently.","section":"Fig. 9 caption"},{"comment":"Equation (5) should define V explicitly as the electric scalar potential and state the units of each quantity; the reader should not have to infer the meaning from context.","section":"Eq. (5)"},{"comment":"The initial resistivity is given as ρ0=10^-16 Ω·m in Section 2.1 and ρ0=10^-17 Ω·m in Section 3.1; clarify whether this difference is intentional and how the value should be chosen in practice.","section":"Secs. 2.1 and 3.1"},{"comment":"The current-density color maps have no visible color bars or labeled scales; add color bars and indicate the Jc value on each map.","section":"Figs. 2, 6, 9, 10"},{"comment":"The bracket and brace formatting in Eq. (13) appears garbled; typeset the strain-dependent Jc expression carefully.","section":"Eq. (13)"},{"comment":"Reference [53] is incomplete: it lacks the year, volume, and page numbers; complete the citation.","section":"Reference [53]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is an engineering-oriented numerical implementation paper rather than a new-physics contribution. The main barrier to acceptance is the lack of a derivation or convergence proof for the critical-state forcing procedure and Eq. (5), together with the demonstrated sensitivity to ρ0, T1, and k1. If the authors can supply quantitative convergence tests and parameter-selection criteria, the paper would be a useful contribution to the applied superconductivity modeling literature."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a practical engineering paper, not a physics advance, and the most useful part is the flux-creep implementation. The critical-state validation is weaker than it looks because the algorithm manufactures the Bean profile rather than solving for it. But the method is credible enough to warrant a serious referee, especially if the authors release the APDL scripts and tighten the comparison.\n\nWhat is new: the iterative algorithm extends the old ANSYS Resistivity-Adaption-Algorithm from Gu et al. to transient magnetization, flux creep relaxation, Jc(B), Jc(ε), and ferromagnetic inserts. The external benchmark against the COMSOL H-formulation, run by Mark Ainslie with matched Jc, n, geometry, and field ramps, is real. The flux-creep results (n=20, n=40) agree well in peak currents and penetration depths at several times. The restart capability and easy convergence are practical advantages that matter for complex models. Section 5 is honest about sensitivity to load steps, ρ0, T1, and k.\n\nThe soft spots: in the critical-state runs, elements with |J| > Jc are forced back to Jc via BFE commands and the inner -Jc layer is hand-set; Eq. (5) is an invented nodal-voltage condition with no derivation. So the COMSOL agreement for the critical state mostly tests the position of the penetration front, not whether the Bean profile emerges from the field dynamics. Section 5 reinforces this: for ρ0=1e-15 Ω·m or T1=5000 s the bulk is 'not well penetrated at the ends,' meaning the correct-looking solution depends on a hand-picked ρ0/T1 pair. That is not fatal for a workflow paper, but it does mean 'can simulate the critical state model' is overstated as written. The flux-creep model is not subject to the same circularity charge—the current is not pinned to Jc—so that part of the validation stands. What is missing is quantitative error norms, mesh/time-step convergence, and the actual APDL code. The stress-test note's circularity point is fair but too broad if applied to the whole paper.\n\nWho this is for: groups doing trapped-field simulations of bulks or tape stacks in ANSYS, especially accelerator-magnet contexts. It is a useful benchmark-case workflow, not a new physical result.\n\nRecommendation: send it to peer review. Require the authors to (1) release scripts, (2) add error norms and convergence checks, and (3) either justify Eq. (5) or replace the critical-state forcing with a field-solving step and validate on a non-axisymmetric case or tape stack. With those, the paper would be a solid contribution.","headline":"A genuine, useful ANSYS workflow for trapped-field simulations, with solid flux-creep benchmarking but a critical-state validation that is partly hand-imposed rather than independently solved.","tokens_in":17288,"tokens_out":4006,"would_cite":true,"duration_ms":34946,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"An iterative load-step algorithm lets ANSYS simulate bulk superconductor magnetization in both critical-state and flux-creep models.","keywords":["A-V-A formulation","iterative algorithm method","critical state model","flux creep model","magnetization current","trapped field","ReBCO bulk superconductor","finite element analysis"],"falsifier":"Run the same iterative loop on a pulsed-field magnetization (ramp to 1 T in about 50 ms) or on a rectangular tape stack, and compare the trapped-field profile and penetration depth against an H-formulation simulation or a magnetization measurement; any systematic front mismatch would show the clamping step distorts time-dependent penetration.","tokens_in":16061,"feed_emoji":"🧲","tokens_out":9721,"duration_ms":85462,"temperature":0.7,"pith_summary":"This paper introduces an iterative algorithm method (IAM) that lets a standard finite-element eddy-current solver simulate magnetization currents in high-temperature superconductors. The method alternates transient electromagnetic solves with a post-processing step: for the critical state model it forces every penetrated superconductor element to carry exactly the critical current density $J_c$ after each load step, and for the flux creep model it updates each element's resistivity according to the E-J power law with a smoothing coefficient. The authors show that for a disk-shaped ReBCO bulk, the trapped current profiles during zero-field-cooled and field-cooled magnetization match those produced by the H-formulation in COMSOL. If the match holds, the method gives engineers an ANSYS-based route to trapped-field magnet design with adjustable computation time and restart capability.","feed_headline":"ANSYS now simulates bulk superconductor magnetization","feed_subtitle":"An iterative A-V-A loop reproduces COMSOL H-formulation results for trapped currents in ReBCO disks.","key_machinery":"The carrying mechanism is the A-V-A formulation combined with an outer iterative loop. In superconductor regions the solver uses the magnetic vector potential $\\mathbf{A}$ together with the electric scalar potential $V$, while non-superconductor regions use $\\mathbf{A}$ alone, which shortens computation time. After each applied-field load step the method either clamps the trapped current density to $J_c$ using force commands (critical state model) or updates the resistivity with the E-J power law $\\rho = \\max\\{\\rho_0, (E_c/J_c)(|J|/J_c)^{n-1}\\}$ and a mixing coefficient $k$ (flux creep model). For the field-descending branch of the critical state model, the method imposes the nodal voltage $V = -2\\pi r \\rho_0 J_c$ on penetrated elements to keep the inner-layer current from decaying while the outer layer reverses.","core_discovery":"The central claim is that magnetization currents in bulk superconductors can be computed by a load-step loop in which each step solves the A-V-A formulation (A-V in the superconductor, A-only elsewhere) and then directly enforces the constitutive law: set $|J| = J_c$ in penetrated elements for the critical state model, or update the element resistivity with $\\rho = \\max\\{\\rho_0, (E_c/J_c)(|J|/J_c)^{n-1}\\}$ using a relaxation coefficient for the flux creep model. The paper reports good agreement with H-formulation results in COMSOL for a 25 mm diameter, 10 mm thick ReBCO disk in zero-field cooling, field cooling, and field reversal, and shows that the same loop can absorb ferromagnetic inserts, field-dependent $J_c(B)$, and strain-dependent $J_c(\\varepsilon)$. It further claims practical advantages: adjustable computation time (about 2 to 3 seconds per load step), multi-frame restart analysis, and easy convergence even at large $n$-values.","pith_inferences":["If the equivalence to H-formulation persists for multi-cycle or pulsed-field magnetization, the same clamping loop could turn any commercial eddy-current solver into a critical-state solver, removing the need for custom partial-differential-equation interfaces.","The ad hoc node-voltage condition used during field descent is only justified by the single disk benchmark; a natural stress test is to compare the method against measured AC loss in a coil or stack, where repeated flux reversal exercises the inner-layer decay handling.","The smoothing coefficients $k_1$-$k_3$ resemble numerical filters; a systematic study of their optimal values versus mesh size and time step would turn the current practical recipe into a calibrated convergence criterion."],"forward_implications":["A trapped-field magnet designer can run magnetization simulations in ANSYS with a freely chosen trade-off between accuracy and speed by setting the number of load steps.","The same iterative loop handles ferromagnetic flux guides around the bulk, so realistic magnet assemblies can be modeled without changing the core algorithm.","Because $J_c(B)$ and $J_c(\\varepsilon)$ enter as per-element updates after each load step, the method can couple electromagnetic solving to mechanical stress analysis during the magnetization process.","The restart capability means a multi-day simulation can be stopped and resumed after a workstation interruption without losing intermediate results.","For the flux creep model, selecting a relaxation coefficient $k$ near 1 (for example 0.99) is necessary to keep the resistivity update stable; too-small $k$ (0.8 or 0.7) yields discontinuous current profiles."],"supporting_citations":[{"why":"Defines the critical state model that the method reproduces by clamping the current density to the critical value.","marker":"[44]"},{"why":"Provides the E-J power law on which the flux creep resistivity update is based.","marker":"[46]"},{"why":"Supplies the benchmark disk geometry and magnetization case used to build the axisymmetric finite-element model.","marker":"[54]"},{"why":"An H-formulation implementation used as the reference simulation for comparison.","marker":"[33]"},{"why":"Earlier ANSYS Resistivity-Adaption-Algorithm that this iterative method extends.","marker":"[50]"},{"why":"Previous iterative resistivity-adaption algorithm for critical state modeling in ANSYS.","marker":"[53]"},{"why":"Provides the field-dependent critical current density relation and benchmark references used in the $J_c(B)$ extension.","marker":"[47]"}],"fun_headline_variants":["A-V-A iterative loop reproduces COMSOL superconductor magnetization","ANSYS A-V-A simulates critical state and flux creep in bulk superconductors","Iterative A-V-A method matches H-formulation for ReBCO disks","A-V-A formulation in ANSYS: fast superconductor magnetization simulation","Loop over A-V-A steps captures trapped currents in superconductors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method assumes that clamping the current density to $J_c$ after each load step, together with the node-voltage condition of Eq. (5), gives the same penetration dynamics as the true critical state model; the paper supports this only by matching one COMSOL disk profile.","fun_headline_variants_meta":{"raw":{"variants":["A-V-A iterative loop reproduces COMSOL superconductor magnetization","ANSYS A-V-A simulates critical state and flux creep in bulk superconductors","Iterative A-V-A method matches H-formulation for ReBCO disks","A-V-A formulation in ANSYS: fast superconductor magnetization simulation","Loop over A-V-A steps captures trapped currents in superconductors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000633,"raw_usage":{"total_tokens":2954,"prompt_tokens":1010,"completion_tokens":1944,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":1847}},"tokens_in":626,"tokens_out":1944,"duration_ms":14801,"temperature":1.0,"reasoning_tokens":1847,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:36:54.716675+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same iterative loop on a pulsed-field magnetization (ramp to 1 T in about 50 ms) or on a rectangular tape stack, and compare the trapped-field profile and penetration depth against an H-formulation simulation or a magnetization measurement; any systematic front mismatch would show the clamping step distorts time-dependent penetration.","supporting_citations":[],"review_version":1}