Skip to content

Use

Uses

Debriefing, EBT/CBTA training, flight safety with FDM and ASR, research; one case, different questions.

A finished case serves several purposes. They share the same rules and differ in the question asked. None of them turns PHASE into a parallel investigation or an assessment of individuals.

Debriefing

The point is not to comment on the outcome. It is to understand the moments when the situation changed. The chart supports a discussion of what was known at each moment, what was observable, which options were still open, which margins were shrinking, which recoveries were attempted, and what remains undetermined.

The debriefing question becomes:

At this precise moment, with this information, these margins and these constraints, which competencies could be brought to bear, and which options were still available?

The debrief is about an object, not a person: a window, a bar, a diamond, a segment. Instructor and trainee look at the same surface, point at it, come back to it. “You should have…” becomes “look at what was active there”. The discovery reading is the default.

A crew can also review its own flight with the same representation. This self-debrief calls for more care: the object-not-person rule applies to yourself first, and the case is reviewed by a trained person before it is shown to anyone else.

EBT/CBTA training

PHASE uses the EBT/CBTA competencies as its framework for the crew. It does not replace them: it gives them a time structure. A competency is linked to a threat, to a moment in the sequence, to an observable behaviour, to an aircraft state, to a margin.

Training can then work on the competency in its real conditions. Not just “decision-making”, but decision-making as time runs short and an option closes. Not just “communication”, but escalation as the flight path degrades and the first interventions get no response.

Three formats recur:

  • the constructed scenario, designed from the operator’s EBT objectives, labelled as constructed and never mistaken for a real case;
  • the contrasted pair: two charts on the same axis, with the same threats, differing only in the crew;
  • the shared reading: a reading of one observable behaviour is proposed, each trainee answers “agree”, “reservation” or “disagree”, and explains why.

The shared reading is not looking for one right answer. It brings out differences in reading between instructors, which makes it a TRI/TRE standardisation tool.

PHASE works without a simulator, a replay system or an EBT tracking tool. The simulator makes crews act and decide in real time; PHASE is for building and analysing the dynamics of targeted situations. Example: the diversion in degraded configuration demo.

Flight safety: FDM and ASR

A flight safety office receives flight data monitoring (FDM) output on one side and pilot reports (ASR, Air Safety Report) on the other. The data tells you what crossed a threshold, to the second. The report tells you what the pilot perceived and why they acted, with approximate timing.

PHASE synchronises both on a single timeline. Every fact keeps its source. Measured times and times known only by their order are kept distinct. The dashed outline shows what the crew did not know yet; the cursor shows what was active as a gate was crossed.

The case then feeds into training:

  • an EBT reading of the event, with the behaviours at stake and their level of evidence;
  • a scenario to replay in the simulator;
  • the questions to ask the crew;
  • where useful, a comparison with a similar public case.

Beyond a single event, several cases read the same way bring out recurring patterns: defences that come into play late, windows where margins tighten, tipping points that look alike.

Example: the FDM/ASR demo, an unstabilised approach read from both sources.

PHASE is not an investigation tool. It re-reads source material; it does not generate new evidence.

Research

For researchers, PHASE offers an explicit grammar, a declared status for every object and an open data format. A case is a readable JSON file described by a schema (proposed version 1.0), shared with research partners and designed to be loaded and exchanged without loss. The real Darwin case is one example: a coding proposed by the analysis chain, not yet reviewed by an analyst.

Several questions remain open and lend themselves to study:

  • agreement between analysts coding the same report;
  • agreement between instructors on Control Index placement;
  • the effect of the chart, the discovery reading or the contrasted pair on how a case is understood.

The details are in Status and limits.

A shared language

The same event is read differently by a safety analyst, an instructor, a pilot or an operations manager. PHASE does not erase those viewpoints. It organises them around a common reference. The discussion is no longer only about an overall conclusion: it can focus on one link, one window, one defence or one uncertainty.

Who carries out analyses

Analyses are carried out on request. An AI-assisted analysis chain produces a first structured draft of the case; an analyst checks, corrects and validates it. AI proposes and structures; the expert checks, corrects and decides. The public demo runs no AI at all: it shows precomputed cases.