ILLUSTRATIVE · SYNTHETIC DATA · NOT VALIDATED BY AN AVIATION SME · SHOWS OUTPUT SHAPE ONLY
Evidenx does not score. This card presents bounded facts; significance and causation are for the investigation team to determine.
Step 1 of 8Scope & signal reduction
What was examined
Six evidence sources were examined for this serious incident and reduced to one subsystem and a small set of candidate causes to test. As in the DFIR layer, nothing is discarded. Every parameter and record remains an observation; this layer selects and correlates, it does not conclude.
~1,300
parameters & records across 6 sources
→
1
subsystem (right-engine fuel delivery)
→
3
candidate causes tested
FDR
1,120 parameters
flight data recorder
QAR / EHM
prior 40 FH
engine trend monitoring
Maintenance
prior 60 FH
logbook & task cards
ADS-B track
flight leg
position / altitude
CVR / aural
event window
warnings & callouts
Component / fluids
pending
exam & samples
Step 2 of 8The finding
The anchor: a pattern, not a cause
Right-engine in-flight power loss during climb, with a pattern
consistent with an interruption of fuel delivery (HP fuel pump / fuel metering). Corroborated across FDR, engine trend monitoring, and the maintenance record. Causation is not established. This is for the investigation team to determine.
reference matchSB-ENG-4471 (illustrative) · HP fuel pumpfailure signature: FF drop → N2 rollback
evidence inFDR QAR/EHM maintenance
Reference lists here (applicable Service Bulletins / Airworthiness Directives, known failure signatures) are the aviation analogue of the DFIR reference lists, authored and curated by a domain SME, not by this tool.
Step 3 of 8Corroborating evidence
Co-located / correlated facts for the team to interpret
FDR
temporal
Fuel flow on the right engine dropped ~2.1 s before N2 began to roll back; throttle-lever angle unchanged across the event (no commanded reduction).
QAR / EHM
trend
HP fuel-pump vibration trend rising over the prior ~40 flight hours, above the fleet baseline band.
Maintenance
provenance
HP fuel pump replaced ~60 flight hours prior; a fuel-filter bypass indication was logged on the preceding turnaround; SB-ENG-4471 open.
Each item is a neutral fact plus why it was surfaced. The system does not assert that the pump caused the power loss. It places the correlated evidence in front of the investigator.
Step 4 of 8Timeline
Maintenance history & flight events
T-60 FH HP fuel pump replaced (right engine)maintenance
T-40 FH → T HP fuel-pump vibration trend rising above baselinetrend
T-1 turn Fuel-filter bypass indication loggedmaintenance
T+00:00 Climb; right-engine fuel flow dropflight
T+00:02 Right-engine N2 rollback begins; throttle unchangedflight
T+00:05 Master caution: 'ENG 2 FAIL'aural
T+00:40 Altitude/airspeed loss consistent with asymmetric thrustflight
T+14:00 Single-engine approach and landing; no injuriesflight
Step 5 of 8Alternative hypotheses considered
Competing causes, tested against the evidence
Fuel contamination / quantity
Recorded fuel quantity was normal and consistent with flight plan; a boost-pump or crossfeed issue would present differently on the FDR. Not distinguished yet. Fuel sample laboratory analysis is pending (see next page).
FADEC / sensor fault
Throttle-lever angle was unchanged and dual-channel FADEC data show no commanded reduction or sensor divergence at the event. Weakened relative to a fuel-delivery interruption.
Mechanical / bearing failure
No oil-debris or chip-detector activation was recorded, and oil pressure/temperature stayed in band through the event. Weakened, but borescope not yet performed.
This is where the model earns its keep for aviation: competing causes are held side by side and tested against what the evidence does and does not distinguish, the same alternative-hypotheses / expected-but-absent machinery used in the DFIR layer.
Step 6 of 8Expected but not yet collected
Absence, recorded honestly
Fuel sample laboratory analysis
not_collected
Would distinguish contamination from a delivery interruption. Sample retained; result pending. Its absence carries no weight for or against any hypothesis.
Engine borescope inspection
not_collected
Would address the mechanical/bearing alternative. Not yet performed.
Oil-debris / chip-detector readout
absent
Searched for and not present, mildly reducing the mechanical-failure alternative.
absent (searched, not found) is evidentiary; not_collected (never gathered) is a collection-scope fact and is never weighed against a hypothesis, the same distinction we just added to the DFIR schema.
Step 7 of 8Understanding
Bounded & factual, no score
Indicators consistent with a right-engine fuel-delivery interruption were observed across three independent sources (FDR, engine trend monitoring, maintenance record), with a fuel-flow drop preceding N2 rollback and the throttle unchanged. Two alternative causes (FADEC/sensor, mechanical) are weakened by the available data; a third (fuel contamination) is not yet distinguished pending laboratory analysis. Key evidence (fuel sample, borescope) has not been collected. Causation is not established and is for the investigation team to determine.
3 corroborating sourcesalternatives: 2 weakened, 1 pendingevidence gaps: 2 not collected
Step 8 of 8Conclusion
The call is the investigation board's. Evidenx stops at understanding
Evidenx has taken six streams of accident evidence to
data, to
information, to
understanding, the structured, corroborated, bounded picture on the preceding pages. It stops here.
The conclusion is the investigation board's to make.data → information → understanding → their conclusion
Evidenx has not decided, and does not decide: probable cause, fault, or blame. There is no score, rank, or confidence value in this card.
For the investigation board to determine:
- Whether the corroborated sequence is causal or incidental in this accident.
- Which alternative causes survive scrutiny.
- Whether outstanding data (recorders, teardown, weather) must be in hand first.
- The probable cause, contributing factors, and any safety recommendations.
The investigation board determination
Left intentionally blank. This is the investigation board's conclusion to record. Evidenx provides the evidence, not the verdict.