# Disposition of the bounded computational continuation

**No verified lighter installed propulsion unit has been established. The continuation materially changes the engineering argument: the mass/heat balance can defeat thinning before a higher knee field helps, and the assumed doubled-current line is not a predicted fault.** These are conditional public-reference findings, not conclusions about an identified production motor.

## What has been resolved computationally

- Direct equal-torque roots replace consequential current interpolation. Thin-SUS is heavier under the stated equal-copper-heat accounting at all three chosen loads. Other constructions retain budgets, not installed-mass savings.
- Coherent specimen Br values and a direct inverse root connect the hardware comparison to a material property: the thin generic-ring zero-budget boundary is about 1.221 T. The selected HRE-free reading leaves 0.556 g nominal budget and only 6.04 mT Br headroom, comparable with source-reading uncertainty. The older interpolation boundary is superseded.
- Thirty-two successful field–circuit scenarios cover all five constructions and bounded speed/current/resistance/inductance/angle/duration/discretization variations. A failed no-load attempt remains retained alongside its diagnosed numerical correction. The source snapshots distinguish solver versions; physics was not tuned to repair the numerical normalization failure.
- The baseline's final angular/time refinement changes the material-history field screen by 0.032%. The thin-steel candidate's doubled radial resolution changes it by 0.090%. The three tested segmented-rotor comparisons fail the 3% field criterion; their exact requirements are unaccepted despite much smaller current changes.
- Material-frame exposure tracking, winding coupling, circuit time-integration behavior and outer-rotor torque sign have independent numerical checks. Two duration extensions from 2 to 7 electrical periods leave the checked peak quantities unchanged. The static ring-free outer-domain check is small; it is not a whole enlarged-domain trajectory.
- A separate four-case conducting-ring calculation passes its stated numerical controls and whole-solution grid checks. Its approximately 1 kA/m paired field change is smaller than absolute grid shifts, so its frozen effect-resolution result is inconclusive. It is an assumed-conductivity/infinite-axial model, not a measured processed ring.
- A simplified three-coordinate axial problem is numerically corroborated. It demonstrates sensitivity to source/receiver geometry; it supplies no native-motor correction.
- A prospective known-current physical fixture comparison meets central point tolerances but fails interval informativeness. It does not validate local magnet H or the native motor. The original failed two-prototype evaluation remains failed and exposed.

## What prevents a component claim

The comparison loads are not an established mission or continuous rating. The voltage calculation is a steady-dq estimate, not a harmonic voltage solution; the inertia calculation is a consistency estimate from imposed speed, not a coupled speed trajectory. Actual connection, protection, shaft dynamics, steel properties after processing, axial end geometry, losses, cooling, retention, manufacturing variation and installed mass remain unavailable.

For the segmented constructions, real inter-segment gaps, adhesive, corner radii and magnetization transition/orientation are specifically missing. A finer calculation of ideal adjoining blocks would not establish the finished-part field. No rescue campaign was launched to obtain a favorable result.

For the thin generic-ring branch, the nominal positive budget is smaller than consequential property/measurement uncertainty before any integration burden. This supports neither a robust opportunity nor proof of physical impossibility. A short duty with existing thermal headroom could support a different comparison.

## Next evidence ranked by decision consequence

1. **Duty, voltage and thermal/installed boundary.** Determines whether the magnet saving can pay for current, copper, cooling and retention. This can eliminate the candidate before material work.
2. **Known-current or fault record with a drawing and winding definition.** Determines which field scenario is relevant and whether the model predicts observed current/flux transfer. Actual processed-ring properties and axial geometry belong to this evidence packet.
3. **Hot same-batch parent versus finished arc.** Determines whether the parent misses the joint Br/knee requirement or processing degrades an otherwise suitable material. Include a stated knee criterion, recoil/irreversible response, shape correction and measurement uncertainty. No such measurement has been made in this continuation.
4. **Repeated finished parts and structural integration.** Determines whether a manufacturing route can supply the material/geometry capability reliably and whether retention/variation consumes the remaining mass budget.

A potential process investigation should follow those outcomes. Parent material already adequate: qualify or improve finishing, orientation and integration. Parent inadequate: investigate structure/composition/processing with explicit Br–knee tradeoffs, without assigning commercial curves to research specimens. Finished material adequate but motor inadequate: challenge field, control, loss and thermal explanations. HRE-free chemistry, domestic origin, finished-part capability and procurement evidence stay separate.

## Reusable work and evidence boundaries

Retain the nonlinear field core, circuit coupling, moving material exposure map, conducting-ring transport, inverse/equal-torque root methods, source snapshots and independent review protocol. Adapt their physical inputs and re-establish application validity for a new unit. No generic trained motor twin, physical process qualification or demonstrated customer benefit is claimed.

The original public export is preserved. The v2 notebook includes new figures/data and retains the old failed evaluations and assumptions. No private hardware input, customer-specific stand-in, physical experiment, external message, publication or new paid workload was introduced.

Primary review records: `validation_owner/INDEPENDENT_REVIEW.md`, `FAULT_REFINEMENT_REVIEW.md`, `FINAL_FAULT_CASE_AUDIT.json`. Numerical source of truth: `fault/NUMERICAL_CHECKS.json`, `DURATION_CHECKS.json`, `all_case_summary.json`, the mass ledgers and all per-case inputs/traces/fields.
