Original title: Small-Sample MTBF Reliability Modelling of Wind Turbine Main Bearings Based on Three-Way Expansion Bootstrapping
Chenyu Wu, Yuanyuan Wu, Jianxiong Gao, Yiping Yuan · September 2026
Wind turbines rely on main bearings to spin smoothly. These parts last a long time and rarely break, meaning engineers have very little breakdown data to study. This makes it tricky to predict when a bearing will fail or need maintenance. To fix this, researchers created a new mathematical expansion tool paired with a failure curve called a three-parameter Weibull distributionA curve that models how likely a part is to fail at different ages.. Their method expands the small dataset in three directions at once: by filling in gaps between existing data points, stretching past the earliest points, and stretching past the latest points. This lets them squeeze more clues out of both the core data and the outer edges. Testing showed the expanded data matched the original data without significant differences. The new strategy achieved the lowest deviation for a key shape measurement of the curve, while also delivering far more stable failure-time predictions than standard expansion techniques.
You're trying to guess how long your bus ride takes, but you've only done it twice. Two trips aren't enough to tell if the typical ride is 20 or 30 minutes, so any estimate feels like a coin flip. Here, those two trips are the few failure records from a wind turbine bearing. The method stretches each record into a fuller set—adding similar times inside the gap, plus extra possibilities at both ends—so the average becomes something you can trust.
Wind turbine main bearings are expensive to replace, so operators need accurate reliability estimates to schedule maintenance. With so few failure samples available, traditional methods produce uncertain answers. This new approach stretches those rare samples further, giving engineers more confidence in when bearings are likely to fail, which directly affects maintenance costs and downtime.
The study relies entirely on simulated failure intervals from Romax fatigue-life simulations, not real operational data. While the three-way expansion minimizes deviation in the scale parameter, its MTBF relative deviation is actually the largest of the three methods, and two-way expansion produced the smallest MTBF standard deviation. These mixed results suggest the method's advantage is specific to certain metrics and conditions, and it remains unclear whether it generalizes to field data or other turbine designs.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Small-Sample MTBF Reliability Modelling of Wind Turbine Main Bearings Based on Three-Way Expansion Bootstrapping — Chenyu Wu, Yuanyuan Wu, Jianxiong Gao et al. — Machines — doi:10.3390/machines14091084
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.