Research — Illustrative (draft, unvalidated)
Failover race
Each row is a failover sequence you supply: steps that run one after another, and the capacity it restores. Drag the time budget and the required capacity, or retime any step, and see which sequences still qualify. The function scores your candidates; it never writes one.
Caller-supplied sequences against the time budget (illustrative durations)
Time budget
Required capacity
Unit-free: kW, GPM or whatever you reason in.
Transfer switch ride-through · optional second function
Failover scoring
Fastest passing
Gates (yours)
Transfer-switch margin
How the check read it
Client-side port:the logic and table in this page are a TypeScript port of the harness function, checked against values exported from the Python tests. Not a live API call. Domain maturity, quoted from the research page: “Draft; no reliability or controls engineer has reviewed it yet. The availability math is cross-checked against closed-form series, parallel and 2-of-3 cases, and an AI-assisted pre-review found and fixed three issues. IEEE 493 and ISA-75.19.01 were not read directly.” The step names follow the function’s own docstring example; every duration, capacity, budget and transfer or ride-through time on this page is a value you set, not equipment data. The fastest-passing pick is the harness’s disclosed, narrow tie-break (minimum total time among candidates that meet both gates).Real functions this maps to:
evaluate_failover_sequence_candidates, calculate_transfer_switch_ride_through_margin.Not modeled: sequence generation (the function never proposes one), interlock logic, sensor debounce, valve or pump dynamics, overlapping or parallel steps (steps are summed end to end), equipment stress or fail-safe ordering, and any standard-set time budget. The gates are yours; the page does not say whether they are the right ones.See also:Failover race, auto sweep · K-of-N availability