Skip to content
fr

Institutions and research

A representation of flight that states what it knows, and what it doesn’t.

PHASE lays out a crew’s information processing along the time axis, with a coding in which every object carries its source and its status. The method is formalised and has been applied to real cases. Its scientific validation is still to be done, and it is open.

Explore the real casePropose a partnership

What researchers and institutions run into

Representations that look surer than they are

A timeline suggests links that no source supports. Alignment passes for causation. Uncertainty vanishes in the drawing.

Codings that cannot be compared

Each analysis brings its own categories and vocabulary. Without a fixed grammar, two analysts cannot set their readings of the same report side by side.

Data locked inside a document

The coding of an event stays in a report or a slide deck. It cannot be reused, for another reading or for a study.

What PHASE brings

1

A formal grammar

Ten fixed strata, three shapes, three fills, four marks. Each visual variable carries one piece of information only. The rules are written down; a summary is available in the Method section.

2

Traceability for every fact

Every object carries its source, its analytical status and its level of evidence. Relations between objects live in the record, with a precise predicate and a status, and are never drawn.

3

Reservations with two readings

Where the material is inconclusive, the case says so: at what moment, what the source can support, at least two possible readings and what each would change downstream.

4

A readable data format

A case is a readable JSON file described by a schema (proposed version 1.0), shared with research partners. It is designed to be loaded, exchanged and recoded without loss.

Demo

A real case: an unstabilised approach into Darwin.

Embraer E190, February 2025, based on the ATSB’s public report. Storm avoidance, a mode reversion, a lateral deviation below the minimum safe altitude, then the 1,000 ft and 500 ft gates. This is a coding proposed by the PHASE analysis chain and not yet reviewed by an analyst: some forty coded objects over about twenty-five minutes, with reservations left open, each with its two readings.

Explore the real case
PHASE chart : Unstabilised approach, Embraer E190 VH-UYO, Darwin
Descent and weather avoidanceDes…The mode reversionThe mode rev…Lateral deviation below the MSALateral deviation below the MSA“Established” and the final approach fix“Established” and the final ap…The 1,000 ft gateThe 1,000 ft…The 500 ft gate and the configurationThe 500 ft gate and the con…Cues of perceived time pressurePM: deviation reported to ATCPM: partial readbackPF: “established” call outside tolerancePF: continuation decided at the gatePM: “stable” assessment at the gatePM: flap 4 detectedPF: corrections via selectors, AP engagedPM: flap 2 selected 800 ft below profilePM: flap 4 selectedPM: before landing checklist reported donePM: flap 5 restoredPF: AP disconnected, manual landingPF: autopilot not disconnectedPF: approach mode not re-armedPM: incomplete readbackPF: non-compliant position reportPM: flap 4 selected instead of flap 5PF: go-around not initiatedPM: no “unstable” callIncorrect aircraft mode: basic modesIncorrect flight path: lateral deviationVertical position: above the glideslopeIncorrect configuration: flap 4Stabilisation not achievedIncorrect speed: above VAPP + 10Guidance reverts to basic modesIMC down to about 1,200 ftStationary storm cell over the aerodromeMaintain 2,000 ft until glideslope interceptStabilisation heights differ by conditions16:35:3016:36:0016:36:3016:37:0016:37:3016:38:0016:38:30LAPAR, initial approach fixLAPAR, initial approach fixOutside lateral tolerance, below the MSAOutside toleranceTOROT, final approach fixTOROT, final approach fixIMC stabilisation height, 1,000 ft1,000 ftVMC stabilisation height, 500 ft500 ftTouchdownTouchdown16:34:51

What we offer

Validating PHASE is an open programme. We are looking for teams to carry it out with us.

  • Reliability studies

    Double coding of the same report, inter-rater agreement on the coding and then on Control Index placement.

  • Training effect studies

    Do the chart, the discovery reading or the contrasted pair change how a case is understood? No study measures this yet.

  • Access to the corpus and the format

    The method’s documents, the case schema and the demo cases, for academic or institutional work.

Frequently asked questions

A research project, a methodological question?

Laboratory, authority, investigation body or training organisation: tell us what interests you. The founder will reply to discuss it.

Propose a partnership