Wind Turbine Drivetrain Vibration Analysis
Evaluates drivetrain vibration health across three subsystems: main bearing, gearbox, and generator.
When to Use
Load this skill when the user wants to:
- Assess drivetrain vibration health from CMS or SCADA data
- Interpret RMS, peak-to-peak, or spectral findings for main bearing, gearbox, or generator
- Correlate vibration alarms with operational events
- Decide whether to continue operating, increase monitoring, or shut down
Drivetrain Components
| Component |
Sensor Location |
Key Frequencies |
| Main Bearing |
Non-drive end, drive end |
BPFO, BPFI, BSF, FTF |
| Gearbox LSS |
Low speed shaft |
Gear mesh (LSS x teeth), bearing defect freqs |
| Gearbox IMS |
Intermediate shaft |
IMS gear mesh harmonics |
| Gearbox HSS |
High speed shaft |
HSS gear mesh, bearing defect freqs |
| Generator NDE |
Non-drive end bearing |
Electrical harmonics, bearing defect freqs |
| Generator DE |
Drive end bearing |
Bearing defect freqs, rotor unbalance |
Vibration Thresholds (ISO 10816 / CMS Reference)
| Location |
Normal |
Warning |
Critical |
| Main Bearing RMS (g) |
< 0.3 |
0.3 - 0.8 |
> 0.8 |
| Gearbox HSS RMS (g) |
< 0.5 |
0.5 - 1.5 |
> 1.5 |
| Gearbox LSS/IMS RMS (g) |
< 0.3 |
0.3 - 1.0 |
> 1.0 |
| Generator RMS (g) |
< 0.5 |
0.5 - 1.2 |
> 1.2 |
| Peak-to-peak step change |
< 10% |
10-30% |
> 30% |
Note: Always evaluate against site-specific baseline. A 20% rise from stable baseline is more significant than an absolute value alone.
Frequency Fault Signatures
| Fault |
Frequency Signature |
| Bearing outer race (BPFO) |
(N/2) x (1 - d/D x cos a) x RPM |
| Bearing inner race (BPFI) |
(N/2) x (1 + d/D x cos a) x RPM |
| Gear mesh |
number of teeth x shaft RPM |
| Gear mesh sidebands |
GMF +/- shaft frequency |
| Rotor unbalance |
1x RPM dominant |
| Misalignment |
2x RPM dominant, axial component |
| Looseness |
Sub-harmonics (0.5x, 1.5x) or high harmonic content |
Severity Scale
| Severity |
Label |
Description |
Action |
| 1 |
Healthy |
All values normal, stable trend |
Continue normal operation |
| 2 |
Early warning |
1-2 parameters in warning zone, stable |
Increase CMS polling frequency |
| 3 |
Moderate |
Multiple warning flags or single critical |
Inspect within 2 weeks |
| 4 |
Significant |
Critical zone or rapid trend growth |
Plan shutdown within 48-72 hours |
| 5 |
Critical |
Multiple critical flags, step-change |
Immediate shutdown required |
Procedure
- Collect inputs: CMS trend (last 30-90 days), current RMS and peak-to-peak per component, frequency spectrum findings, SCADA alarms, operational context.
- Evaluate RMS values against thresholds. Flag Warning or Critical zones.
- Analyze trend:
- Stable: value in warning zone but flat for >30 days = lower urgency
- Gradual rise: value increasing steadily = schedule inspection
- Step change: sudden jump >30% = treat as Critical regardless of absolute value
- Interpret frequency spectrum if available:
- Match dominant peaks to fault signatures table
- Note sidebands around gear mesh frequencies
- Note sub-harmonics or 1x/2x dominance
- Correlate with SCADA alarms and operational events.
- Assign severity per component, then determine drivetrain-level severity as highest.
- Generate output report using the format below.
Output Format
=== DRIVETRAIN VIBRATION REPORT ===
ASSET : [Turbine ID]
SITE : [Site name]
DATA PERIOD : [Date range of CMS/SCADA data]
MISSING DATA : [List any unavailable inputs]
MAIN BEARING:
RMS : [value] g - [Normal / Warning / Critical]
Trend : [Stable / Gradual rise / Step change]
Spectrum : [Key findings or not available]
SCADA Alarms : [Count and type]
Severity : [1-5] - [Label]
GEARBOX (LSS / IMS / HSS):
RMS : LSS [value] g / IMS [value] g / HSS [value] g
Trend : [per shaft]
Spectrum : [Key findings]
SCADA Alarms : [Count and type]
Severity : [1-5] - [Label]
GENERATOR (DE / NDE):
RMS : DE [value] g / NDE [value] g
Trend : [per bearing]
Spectrum : [Key findings]
SCADA Alarms : [Count and type]
Severity : [1-5] - [Label]
DRIVETRAIN SEVERITY : [1-5] - [Label]
SHUTDOWN : [Yes / No / Conditional]
FAULT HYPOTHESIS:
- [e.g., HSS bearing outer race defect - BPFO peak confirmed at X Hz]
- [e.g., Gear mesh sideband modulation - possible gear wear or load variation]
RECOMMENDED ACTIONS:
- [e.g., Increase CMS polling to daily for HSS channel]
- [e.g., Oil sample with ferrography within 72 hours]
- [e.g., Plan HSS bearing replacement at next scheduled outage]
ESCALATION TRIGGERS:
- [e.g., RMS exceeds 1.5 g on HSS - immediate shutdown]
- [e.g., Step change >30% on any channel - treat as critical]
- [e.g., New BPFO or BPFI peak confirmed in spectrum - escalate to Severity 4]
Cross-Skill Correlation
If gearbox visual data is available, load wind-turbine-gearbox skill and cross-correlate:
- High Fe ppm + rising HSS vibration = active wear confirmation
- Spalling in borescope + BPFO peak in spectrum = bearing failure progression
- Normal oil + rising vibration = early fault not yet generating debris (higher urgency)
If blade inspection data is available, check for rotor imbalance:
- 1x RPM dominant in main bearing spectrum + blade damage = aerodynamic imbalance
- Asymmetric blade damage across A/B/C = mass or aerodynamic imbalance source
Pitfalls
- Do not evaluate vibration in isolation. Cross-reference with oil analysis and visual inspection.
- A single high RMS reading during a storm or grid fault is not a fault indicator. Check operational context.
- Spectrum analysis requires RPM-normalized data. Raw frequency peaks are meaningless without shaft RPM.
- Generator electrical faults can appear as vibration. Check electrical data before attributing to mechanical cause.
- Stable high RMS is less urgent than rapidly rising moderate RMS. Trend rate matters more than absolute value.
Verification
After generating the report, confirm with the user:
- Does the severity match CMS system alerts or OEM recommendations?
- Is shaft RPM data available to normalize spectrum frequencies?
- Are there recent maintenance events that could explain vibration changes?
- Is SCADA power curve deviation consistent with vibration findings?
1---2name: wind-turbine-vibration-analysis3description: Analyzes wind turbine drivetrain vibration data (main bearing, gearbox, generator) from CMS trends, RMS/peak values, frequency spectrum, and SCADA alarms. Classifies severity (1-5) and recommends shutdown or monitoring actions.4license: MIT5---6
7# Wind Turbine Drivetrain Vibration Analysis
8
9Evaluates drivetrain vibration health across three subsystems: main bearing, gearbox, and generator.
10
11## When to Use
12
13Load this skill when the user wants to:
14- Assess drivetrain vibration health from CMS or SCADA data
15- Interpret RMS, peak-to-peak, or spectral findings for main bearing, gearbox, or generator
16- Correlate vibration alarms with operational events
17- Decide whether to continue operating, increase monitoring, or shut down
18
19## Drivetrain Components
20
21| Component | Sensor Location | Key Frequencies |
22|-----------|----------------|-----------------|
23| Main Bearing | Non-drive end, drive end | BPFO, BPFI, BSF, FTF |
24| Gearbox LSS | Low speed shaft | Gear mesh (LSS x teeth), bearing defect freqs |
25| Gearbox IMS | Intermediate shaft | IMS gear mesh harmonics |
26| Gearbox HSS | High speed shaft | HSS gear mesh, bearing defect freqs |
27| Generator NDE | Non-drive end bearing | Electrical harmonics, bearing defect freqs |
28| Generator DE | Drive end bearing | Bearing defect freqs, rotor unbalance |
29
30## Vibration Thresholds (ISO 10816 / CMS Reference)
31
32| Location | Normal | Warning | Critical |
33|----------|--------|---------|----------|
34| Main Bearing RMS (g) | < 0.3 | 0.3 - 0.8 | > 0.8 |
35| Gearbox HSS RMS (g) | < 0.5 | 0.5 - 1.5 | > 1.5 |
36| Gearbox LSS/IMS RMS (g) | < 0.3 | 0.3 - 1.0 | > 1.0 |
37| Generator RMS (g) | < 0.5 | 0.5 - 1.2 | > 1.2 |
38| Peak-to-peak step change | < 10% | 10-30% | > 30% |
39
40Note: Always evaluate against site-specific baseline. A 20% rise from stable baseline is more significant than an absolute value alone.
41
42## Frequency Fault Signatures
43
44| Fault | Frequency Signature |
45|-------|-------------------|
46| Bearing outer race (BPFO) | (N/2) x (1 - d/D x cos a) x RPM |
47| Bearing inner race (BPFI) | (N/2) x (1 + d/D x cos a) x RPM |
48| Gear mesh | number of teeth x shaft RPM |
49| Gear mesh sidebands | GMF +/- shaft frequency |
50| Rotor unbalance | 1x RPM dominant |
51| Misalignment | 2x RPM dominant, axial component |
52| Looseness | Sub-harmonics (0.5x, 1.5x) or high harmonic content |
53
54## Severity Scale
55
56| Severity | Label | Description | Action |
57|----------|-------|-------------|--------|
58| 1 | Healthy | All values normal, stable trend | Continue normal operation |
59| 2 | Early warning | 1-2 parameters in warning zone, stable | Increase CMS polling frequency |
60| 3 | Moderate | Multiple warning flags or single critical | Inspect within 2 weeks |
61| 4 | Significant | Critical zone or rapid trend growth | Plan shutdown within 48-72 hours |
62| 5 | Critical | Multiple critical flags, step-change | Immediate shutdown required |
63
64## Procedure
65
661. Collect inputs: CMS trend (last 30-90 days), current RMS and peak-to-peak per component, frequency spectrum findings, SCADA alarms, operational context.
672. Evaluate RMS values against thresholds. Flag Warning or Critical zones.
683. Analyze trend:
69 - Stable: value in warning zone but flat for >30 days = lower urgency
70 - Gradual rise: value increasing steadily = schedule inspection
71 - Step change: sudden jump >30% = treat as Critical regardless of absolute value
724. Interpret frequency spectrum if available:
73 - Match dominant peaks to fault signatures table
74 - Note sidebands around gear mesh frequencies
75 - Note sub-harmonics or 1x/2x dominance
765. Correlate with SCADA alarms and operational events.
776. Assign severity per component, then determine drivetrain-level severity as highest.
787. Generate output report using the format below.
79
80## Output Format
81
82=== DRIVETRAIN VIBRATION REPORT ===
83
84ASSET : [Turbine ID]
85SITE : [Site name]
86DATA PERIOD : [Date range of CMS/SCADA data]
87MISSING DATA : [List any unavailable inputs]
88
89MAIN BEARING:
90 RMS : [value] g - [Normal / Warning / Critical]
91 Trend : [Stable / Gradual rise / Step change]
92 Spectrum : [Key findings or not available]
93 SCADA Alarms : [Count and type]
94 Severity : [1-5] - [Label]
95
96GEARBOX (LSS / IMS / HSS):
97 RMS : LSS [value] g / IMS [value] g / HSS [value] g
98 Trend : [per shaft]
99 Spectrum : [Key findings]
100 SCADA Alarms : [Count and type]
101 Severity : [1-5] - [Label]
102
103GENERATOR (DE / NDE):
104 RMS : DE [value] g / NDE [value] g
105 Trend : [per bearing]
106 Spectrum : [Key findings]
107 SCADA Alarms : [Count and type]
108 Severity : [1-5] - [Label]
109
110DRIVETRAIN SEVERITY : [1-5] - [Label]
111SHUTDOWN : [Yes / No / Conditional]
112
113FAULT HYPOTHESIS:
114 - [e.g., HSS bearing outer race defect - BPFO peak confirmed at X Hz]
115 - [e.g., Gear mesh sideband modulation - possible gear wear or load variation]
116
117RECOMMENDED ACTIONS:
118 - [e.g., Increase CMS polling to daily for HSS channel]
119 - [e.g., Oil sample with ferrography within 72 hours]
120 - [e.g., Plan HSS bearing replacement at next scheduled outage]
121
122ESCALATION TRIGGERS:
123 - [e.g., RMS exceeds 1.5 g on HSS - immediate shutdown]
124 - [e.g., Step change >30% on any channel - treat as critical]
125 - [e.g., New BPFO or BPFI peak confirmed in spectrum - escalate to Severity 4]
126
127## Cross-Skill Correlation
128
129If gearbox visual data is available, load wind-turbine-gearbox skill and cross-correlate:
130- High Fe ppm + rising HSS vibration = active wear confirmation
131- Spalling in borescope + BPFO peak in spectrum = bearing failure progression
132- Normal oil + rising vibration = early fault not yet generating debris (higher urgency)
133
134If blade inspection data is available, check for rotor imbalance:
135- 1x RPM dominant in main bearing spectrum + blade damage = aerodynamic imbalance
136- Asymmetric blade damage across A/B/C = mass or aerodynamic imbalance source
137
138## Pitfalls
139
140- Do not evaluate vibration in isolation. Cross-reference with oil analysis and visual inspection.
141- A single high RMS reading during a storm or grid fault is not a fault indicator. Check operational context.
142- Spectrum analysis requires RPM-normalized data. Raw frequency peaks are meaningless without shaft RPM.
143- Generator electrical faults can appear as vibration. Check electrical data before attributing to mechanical cause.
144- Stable high RMS is less urgent than rapidly rising moderate RMS. Trend rate matters more than absolute value.
145
146## Verification
147
148After generating the report, confirm with the user:
149- Does the severity match CMS system alerts or OEM recommendations?
150- Is shaft RPM data available to normalize spectrum frequencies?
151- Are there recent maintenance events that could explain vibration changes?
152- Is SCADA power curve deviation consistent with vibration findings?
modbender