How to Check Wind Turbine Gearbox Condition Monitoring: the Design Frame (IEC 61400-4:2012 for Gearboxes Above 500 kW, Rolling-Element Bearings, Lubrication and Prototype Testing — Withdrawn 15 April 2025 in Favour of IEC 61400-4:2025), the Vibration Standards (ISO 10816-21:2015 for Geared Turbines Above 200 kW — Now Withdrawn in Favour of ISO 20816-21:2025; China's NB/T 31004-2011, Current Since November 2011), Why the Gearbox Is the Component to Monitor (Longest Downtime per Failure in ~15,400 Turbine-Years; 0.154 Major Replacements per Turbine-Year Offshore, 95% of Major Replacements With the Generator; Not Always Meeting 20-Year Design Life), What Fails (36 of 37 Incidents in One Small Database Involved Bearings — Hardening Cracks, Abrasion, Scuffing) and How Good Vibration Diagnosis Was in One Blind Test (NREL's Round Robin: Best of Sixteen Partners Found Five of Seven Damages; Most Had More Misses Than False Alarms; One Partner Found Six of Seven in Hindsight)

Short answer. A gearbox condition-monitoring (CM) check has three layers. Standards: the gearbox design standard is IEC 61400-4 — the 2012 edition (above 500 kW, rolling-element bearings, lubrication and prototype/production testing in scope) was withdrawn on 15 April 2025 and a 2025 edition exists; its vibration is measured and evaluated on non-rotating parts under the ISO 10816-21 family — the 2015 edition is now withdrawn and ISO 20816-21:2025 is the current version — and, in China, under NB/T 31004-2011, current since 1 November 2011; its downtime is reported through the IEC 61400-26-1 availability model. Why the gearbox: in ~15,400 turbine-years of German data, electrical systems failed most often but gearboxes caused the longest downtime per failure; offshore, gearbox major replacements run at 0.154 per turbine-year and, with the generator, make up 95% of major replacements; NREL's collaborative was founded because gearboxes 'are not always meeting 20-year design life'. What a CMS can see: in NREL's blind Round Robin on one damaged gearbox, the best of sixteen diagnostic teams found five of seven confirmed damages, most teams missed more than they falsely alarmed, and one team found six of seven once told where to look. A CMS report should be read with that one result in mind.

1. The design and measurement frame

CNWattHub reading: three questions follow from the table. Which editions the certificate and the CMS specification cite — IEC 61400-4:2012 was withdrawn in April 2025 and a 2025 edition exists, and if the CMS specification still names ISO 10816-21:2015, the reference is withdrawn and ISO 20816-21:2025 should be asked for; whether a Chinese project's CMS documents cite NB/T 31004-2011, which has been current for fifteen years and whose text this guide has not read; and whether gearbox stoppages are categorised in the IEC 61400-26-1 model, because that is what turns a CMS-triggered stop into an availability number a contract can act on. These are readings of the abstracts and the record, not of any clause.

2. Why the gearbox is the component to monitor

CNWattHub arithmetic (on the tagged figures): at 0.154 major replacements per turbine-year, a 100-turbine offshore fleet would expect about 15 gearbox major replacements a year, against about 10 for generators; and the survey pattern — a quarter of faults producing 95% of downtime — is the statistical reason a CMS is pointed at the gearbox rather than at the electrical systems that fail more often. The populations differ (German onshore survey data; about 350 turbines of one European offshore geared type), so the numbers are indicative of ranking, not of any specific fleet.

3. What fails inside the gearbox

NREL's 2011 brief reports a small failure database of 37 incidents, 36 involving bearings and 22 gears — so 'may need to pay attention to both'; the top bearing modes were hardening cracks, abrasion and adhesion (scuffing), the top gear modes fretting corrosion and high-cycle bending fatigue. The GRC's own field gearbox is the case study: it suffered two oil-loss events in the field that damaged internal bearings and gears; because the damage was not catastrophic it was retested in the 2.5 MW dynamometer with vibration and oil-debris monitoring before disassembly. The brief's lesson: simple causes may lead to complex damage; O&M improvements are needed to minimise the possibility of oil loss; assembly needs additional attention.

CNWattHub reading: 36 of 37 incidents in that database involving a bearing is a reason to ask how a vibration CMS covers bearing fault frequencies and whether an oil-debris channel sits beside it — the GRC retest ran both. The oil-loss case also says that an early CMS check is the lubrication record: both GRC damage events began as an oil event.

4. How good vibration diagnosis is: the blind Round Robin

NREL shared the retest vibration data with sixteen partners (seven universities, nine companies) who had two months and no knowledge of the damage; the gearbox had no seeded faults. After disassembly all partners agreed seven damage instances were detectable by vibration; the best partner identified five of the seven, and most partners had more missed detections than false alarms — 'there is room for the industry to improve vibration-based diagnostic algorithms'. When the inspection results were disclosed, one partner (Impact Technologies) revisited its data and found six of the seven, its blind-stage misses having included the intermediate-speed-shaft upwind bearings and the high-speed-shaft downwind bearing.

CNWattHub reading: this is the most useful number here for reading a CMS report. On one gearbox with real (not seeded) damage, the best blind detection rate was 5/7 ≈ 71%, and the typical error was a miss, not a false alarm. The practical consequences for a check, drawn from that one test: (1) a clean CMS report should not be read as strong evidence of a clean gearbox — in this test, silence is the error mode the study found most often; (2) one partner's hindsight detection of 6/7 shows the signatures were in the data, so a review of historical CMS data after a borescope finding is worth commissioning; (3) that partner's blind misses were bearings on the intermediate and high-speed shafts — one indication of where to ask about sensor coverage or a borescope. The study is one 2012 research gearbox with 2012-era algorithms and no negative control; it bounds expectations rather than rating any current product.

5. A checking sequence built from the sources

1. Design basis. Which IEC 61400-4 edition the gearbox certificate cites (2012, withdrawn April 2025, or 2025); bearing type and lubrication system in scope.
2. Vibration standard. ISO 20816-21:2025 (not the withdrawn 2015 text); NB/T 31004-2011 on Chinese sites.
3. Lubrication record first. Oil-loss, top-up and filter events; oil-debris channel alongside vibration.
4. Sensor placement. Ask how intermediate- and high-speed-shaft bearings are covered — one partner's blind misses.
5. Alarm reading. Expect misses as well as alarms; corroborate silence with borescope or oil analysis on a schedule; re-run history after any finding.
6. Downtime accounting. Gearbox stops categorised in the IEC 61400-26-1 model; compare against the 0.154/yr offshore and WMEP downtime benchmarks for ranking only.

Scope limits

This guide reads the IEC 61400-4:2012 (withdrawn) and IEC 61400-26-1:2019 abstracts, the ISO 10816-21:2015 catalogue page, the NB/T 31004-2011 record (text not read), NREL's 2012 Round Robin report, NREL's 2011 failure-modes brief and 2013 database survey, and Carroll et al.'s offshore paper. It quotes no CMS vendor specification or alarm-threshold table (the ISO 20816-21:2025 text was not read), no oil-analysis standard, no Chinese fleet gearbox statistic, and no OEM warranty or CMS contract clause. Not engineering or investment advice.