First-Out and Root-Cause Presentation
In a cascade, showing the first alarm to fire and the inferred root cause turns a confusing flurry into an actionable diagnosis.
Finding the cause in the cascade
When a fault cascades, the alarm that matters is usually the first one — the initiating event — but it is instantly buried under its consequences. First-out logic captures and preserves that initiating alarm, and root-cause presentation infers and foregrounds the underlying cause, so the operator diagnoses instead of firefighting symptoms. This is the diagnostic complement to flood suppression: suppression makes the list readable, root-cause makes it meaningful.
First-out capture
At the onset of a cascade, the system latches the first alarm in the causal sequence and marks it distinctly. Even as dozens of downstream alarms follow, the operator can see what started it. For the breeder, the first-out might be a specific stability-margin breach that preceded the disruption cascade; for the burner, a plug-density drop that preceded the confinement and DEC alarms. Knowing the initiator is what makes the response correct rather than merely reactive.
Root-cause inference
Beyond the first alarm, the twin's causal structure and the anomaly ensembles infer the most likely root cause and present it as a header over the collapsed cascade, with the evidence that supports it and a confidence. Because inference can be wrong, the root-cause header always shows its confidence and lets the operator expand the full evidence and override the diagnosis. It is a hypothesis offered with support, not a verdict imposed.
- First-out: the latched initiating alarm, always preserved and highlighted
- Root cause: inferred underlying cause with evidence and confidence
- Cascade: collapsed under the cause, fully expandable, nothing discarded
- Override: the operator can reject the inferred cause and see the raw sequence
First-out and root-cause presentation build on flood suppression and the precursor timeline, and every cascade is preserved intact for incident replay so the inferred cause can be checked against what actually happened.