- name
- wind-turbine-gearbox
- description
- Assesses wind turbine gearbox health from multi-sensor and inspection data. Classifies damage severity (1-5), identifies root cause, and recommends shutdown or monitoring actions.
- version
- 1.0.0
- author
- Sertug17
- license
- MIT
- metadata
- hermes
- tags
- [Energy, Maintenance, Wind-Turbine, Gearbox, Condition-Monitoring, Predictive-Maintenance, Vibration, Acoustics]
- related_skills
- []
Wind Turbine Gearbox Intelligence
Evaluates gearbox condition using five input parameters and produces a structured maintenance report.
When to Use
Load this skill when the user wants to:
- Assess gearbox health from on-site inspection or sensor data
- Classify damage severity on a 1-5 scale
- Determine whether a turbine should be shut down or kept running under monitoring
- Generate a structured maintenance or escalation plan
Quick Reference
| Input Parameter | What to Collect |
|---|---|
| Visual inspection | Surface cracks, pitting, spalling, discoloration, debris |
| Oil iron (Fe ppm) | Iron particle concentration in gear oil (ppm) |
| Temperature (C) | Bearing/gear temperature, normalized to baseline |
| Vibration | RMS or peak-to-peak acceleration (g), frequency anomalies |
| Acoustic / Sound | Noise type: grinding, knocking, whining, clicking |
Fault Thresholds (Reference)
| Parameter | Normal | Warning | Critical |
|---|---|---|---|
| Oil Fe (ppm) | < 100 | 100 - 300 | > 300 |
| Temp above baseline | < 5 C | 5 - 15 C | > 15 C |
| Vibration RMS (g) | < 0.5 | 0.5 - 1.5 | > 1.5 |
| Acoustic | No anomaly | Intermittent | Continuous |
| Visual | Clean surface | Minor pitting | Spalling/crack |
Common Failure Modes
| Failure Mode | Typical Indicators |
|---|---|
| Micropitting | High Fe ppm, slight vibration increase, no visible cracks |
| Spalling | High Fe ppm, elevated vibration, visible surface damage |
| Fatigue crack | Knocking sound, vibration spike at gear mesh frequency |
| Bearing wear | Whining noise, high temperature, broadband vibration increase |
| Oil contamination | Very high Fe ppm, discolored oil, possible foaming |
Procedure
- Collect inputs across all five parameters. If any are unavailable, note as "not measured" and proceed.
- Evaluate each parameter against the thresholds table. Flag Warning or Critical zones.
- Cross-correlate symptoms:
- Fe ppm + vibration increase → wear / spalling progression - Knocking sound + vibration spike → fatigue crack - Temperature + whining sound → bearing failure - Multiple Critical flags → Severity 5
- Assign severity:
- 1 Healthy: All parameters normal. No action required. - 2 Early wear: 1-2 parameters in warning zone. Increase monitoring frequency. - 3 Moderate damage: 2-3 parameters in warning/critical. Inspect within 2 weeks. - 4 Significant damage: Multiple critical flags. Plan shutdown within 48-72 hours. - 5 Critical: Imminent failure risk. Immediate shutdown required.
- Determine root cause from the failure modes table.
- Generate the output report using the format below.
Output Format
=== GEARBOX HEALTH REPORT ===
ROOT CAUSE : [e.g., Progressive spalling on intermediate shaft gear] SEVERITY : [1-5] - [Healthy / Early Wear / Moderate / Significant / Critical] SHUTDOWN : [Yes / No / Conditional]
MONITORING STRATEGY: - [e.g., Repeat oil sample in 72 hours] - [e.g., Daily vibration trend monitoring for 1 week]
ESCALATION TRIGGERS: - [e.g., Fe ppm exceeds 400 - immediate shutdown] - [e.g., Vibration RMS exceeds 2.0 g - immediate shutdown]
Pitfalls
- Never assign Severity 5 based on one parameter alone. Cross-validate with at least two sources.
- Temperature readings can be misleading in extreme ambient conditions. Ask for baseline-normalized values.
- Acoustic descriptions are subjective. Ask for noise type and whether continuous or intermittent.
- If sensor data is unavailable, rely on visual and oil condition as primary indicators.
- Do not conflate oil change interval with oil health. New oil can still show high Fe ppm.
Verification
After generating the report, confirm with the user:
- Does the severity match their on-site observations?
- Are escalation thresholds feasible for their monitoring setup?
- Are there additional data points (CMS trending, historical Fe ppm) that could refine the assessment?