✈️ Turbofan Forensic Intelligence
Overview
NASA N-CMAPSS turbofan engine run-to-failure dataset: 9 engines, 2 fault modes, flight-level sensor aggregates. The logic engine builds a provable 5-layer inference chain, tracing failures in complex turbofan systems directly to their root cause.
Data Structure
| Metric | Value |
|---|---|
| Engines | 9 turbofan units |
| Fault Modes | 2: HPT-only (3 engines), HPT+LPT cascade (6 engines) |
| Sensors | 14 per timestep (temperatures, pressures, speeds, flows) |
| Flights | 1,000 flight-level aggregates |
| Source | NASA DASHlink: Public Domain |
Causal Reasoning Model
Five-layer inference pyramid traces degradation to root causes:
- Sensor Predicates: raw data from 1,000 flights
- Intermediate Facts: real-time anomaly identification
- Pathway Analysis: mapping stressors to physical wear
- Failure Synthesis: tracing HPT-to-LPT degradation cascades
- Root Cause Conclusion: final diagnostic proof
Four primary failure tracks:
- Thermal Overload: throttle cycles driving critical wear
- Operational Intensity: identifying the harsh flight profile tax
- Cascade Mechanics: HPT degradation forcing LPT failure
- Thermo-Mechanical Fatigue: compound heat and pressure stress
Sample Questions
- "Why did Engine 2 fail 15% earlier than expected?"
- "Show the causal chain from HPT degradation to LPT failure."
- "How did severe flight conditions accelerate thermal wear?"
Prescriptive Recommendations
Ask what to do: get maintenance recommendations with cost-benefit analysis. The engine recommends HPT inspection scheduling for cascade-risk engines, throttle reduction for thermal stress, and engine derating to extend remaining useful life.
- "What maintenance actions should we take for the highest-risk engines?"
- "How can we extend the remaining useful life of engines showing thermal stress?"