The decision or acceptance boundary is not explicit.
If the team cannot state what must be selected, rejected, frozen or proved next, the screen has no defensible authority target.
Decision asset · evidence boundary
Where an Architecture Screen must stop.
A screening result becomes unsafe when it is asked to support a decision outside the evidence it controls. Use this boundary map to separate what can be screened now, what must be proved next and where release authority must remain downstream.
The ruleStop when the next conclusion depends on evidence the screen does not control.
Eight stop conditions
Do not let the model outrun the decision basis.
Each condition marks a boundary where the answer may still be useful for direction, but more evidence is required before authority can increase.
The operating envelope is not bounded.
Speed, duty, radial/axial load, gas state, temperature or packaging remain too open to identify the governing state.
The candidate identity is no longer controlled.
Geometry revision, clearance, supply state, shaft-line context or coordinate basis changed without being carried into the result.
The support model is assumption-dominant.
Stiffness, damping, state dependence or another decision-critical support behaviour lacks sufficient provenance for the requested conclusion.
Rotor consequence remains unresolved.
Mode-family separation, response, stability or operating-state consequence is uncertain enough to change the architecture decision.
Detailed CFD / FE cannot be reconciled to the same state.
High-fidelity evidence uses a different geometry, boundary condition, state identity or extraction basis and cannot be re-coupled cleanly.
Manufacturing and tolerance basis can change the answer.
Clearance, thermal state, tolerances, material or interface definition is not controlled tightly enough for the next irreversible step.
Test / acceptance / release authority is undefined.
The evidence does not define what test or acceptance proof is required before prototype release, freeze or production transfer.
Evidence authority
A result is useful only inside its authority.
The objective is not to maximise the status of every result. It is to state what the present evidence may support and keep the next proof explicit.
Do not promote the claim
Six evidence jumps that should trigger a stop.
These are not wording preferences. Each jump changes the engineering authority of the statement.
Screening result≠Production release
One CFD field≠Validated architecture
Component PASS≠System freeze
Assumed damping≠Stability evidence
One operating state≠Full duty envelope
Historical case deviation≠Universal error bound
Use this before the next irreversible step
Bring one decision. State what the current evidence can support - and what must be proved next.
An Architecture Screen should reduce uncertainty without pretending to own release authority it has not earned.