Residual imbalance is not release evidence.

1. The imbalance number is not the whole decision

A residual-imbalance value answers a narrow question: did the rotor clear a numeric threshold under the balancing conditions used? It does not automatically answer why that threshold was the right one for this rotor, this speed range, this bearing or support condition, and this failure consequence.

For rigid-behaviour rotors, ISO 21940-11 defines procedures around permissible residual unbalance, correction planes, tolerance planes, and process-error accounting. The same standard explicitly separates rigid-behaviour rotors from flexible-behaviour rotors, which are handled under ISO 21940-12.

2. Balance grade needs a rationale

A balance quality grade can be useful, but it is not self-justifying. The release question is not only “what grade was achieved?” It is “why is this grade appropriate for this rotor and what evidence supports that choice?”

ISO 21940-11 allows unbalance tolerance to be derived through several routes: balance quality grades, experimental evaluation, bearing-force limits, vibration limits, or established experience. That matters because the grade is one path to a tolerance, not a universal proof of release authority.

3. Correction path matters

Two rotors can reach the same residual-imbalance value through different correction paths and still carry different release risk. The correction method, correction planes, trial-mass sequence, speed condition, and mode sensitivity affect what the final number means.

For flexible rotors, ISO 21940-12 describes modal balancing and influence-coefficient approaches. It also shows that response near resonance depends on modal unbalance, proximity to resonance speeds, and damping in the rotor-support system. A correction mass may strongly affect one mode and weakly affect another depending on its axial location relative to the mode shape.

The Foiles / Allaire / Gunter review reinforces the same point at the method-family level: rotor balancing includes influence-coefficient, modal, unified, no-phase/no-amplitude, least-squares, and constrained approaches. There is no single universal “balance number” detached from method and context.

4. Component balance is not assembly release evidence

Balancing components individually is not the same claim as releasing the assembled rotor. Assembly errors, runout, play, reassembly angle, coupling, and service-bearing conditions can change the unbalance state and the vibration response.

The evidence trail must therefore distinguish component-level balance, assembly-level balance, and service-equivalent balance. A result from one level should not silently authorise another level.

5. Flexible / supercritical rotors need dynamic evidence

Standard balancing intuition is often built around subcritical rigid-rotor behaviour. High-speed precision rotors can operate close to, through, or above bending critical speeds. In that regime, the relevant evidence is not only residual unbalance; it is vibration response over the operating envelope, including start-up, shut-down, overspeed margin, damping, mode participation, and support conditions.

Recent work on supercritical rotor balancing shows why this is difficult in practice: high-speed machines often pass critical speeds, but full-speed measurement can be blocked by temperature, inaccessibility, and hazardous conditions. Low-speed balancing that assumes rigid behaviour may fail to identify the imbalance projection on high-speed flexible modes.

6. Balance-error sources are part of the evidence

A measured residual unbalance is not a perfect fact. It is a measurement produced under a specific setup. ISO 21940-14 exists because balance measurements contain errors, and significant errors must be identified, assessed, and accounted for.

For AURA, that becomes a balance-error ledger: tooling, machine support, sensing system, readout, radial and axial runout, component inclination, reassembly differences, mandrel/service-shaft differences, thermal distortion, looseness, interface clearance, and repeatability.

Measurement apparatus is therefore an evidence object. The balancing machine type, soft-bearing or hard-bearing suspension, mounting and drive arrangement, software filtering, reading repeatability, and tolerance consumption affect what the residual imbalance number is allowed to prove. A rotor that is merely inside G0.4 does not automatically have the same release authority as a rotor that is inside G0.4 with repeatable readings and unused practical margin.

7. The balancing machine is part of the evidence

The balancing result is also conditioned by the balancing machine itself. A soft-bearing setup, a hard-bearing setup, drive arrangement, suspension behaviour, sensor chain, mounting state, and software filtering can change how the residual-imbalance reading is produced and repeated.

For release evidence, the question is therefore not only whether the reading is inside a tolerance such as G0.4. The question is how much of the tolerance was consumed, whether repeated readings remain stable, what filtering was used, and why a particular reading inside the tolerance is accepted as release-worthy rather than merely numerically compliant.

8. The balancing evidence trail

9. What AURA should preserve

The same discipline AURA applies to coefficient evidence applies here: preserve not only the final number, but what the number depends on, how it was produced, what it was checked against, and what it is allowed to prove.

The balancing evidence layer now has a published specification, expanded Field Guide, and locked Gate 4 preview. It is still not a shipped production module. The public surface shows how the same AURA release-evidence architecture applies to a second high-speed rotor decision: balance.