Main bearings (MB) in wind turbines are prematurely failing, sometimes before 6 years
in service, yet the exact damage mechanisms are still debated. This research hypothesises premature main bearing failures may be linked to repetitive force changes driven
by the passage of energy-dominant atmospheric turbulence eddies through the wind
turbine rotor in the atmospheric boundary layer. A high-resolution large-eddy simulation (LES) of a daytime atmospheric boundary layer was developed using the AMRWind finite volume code, and validated against the existing literature. The [Brasseur
and Wei, 2010] framework was employed to improve the accuracy in the surface layer,
where LES typically struggles. The overshoot in the prediction of the normalised mean
velocity gradient was minimised, though not be fully removed, likely due to the contributions from numerical dissipation in the AMR-Wind code. Moderately convective
boundary layers (MCBL) were simulated for several eddy-turnover times until quasistationarity was achieved. A novel methodology was developed to quantify the eddy
passage time of the MCBL, showing good agreement with the literature. A sensitivity
analysis parametrised a state-of-the art actuator line model (ALM). The sensitivity
analysis showed low sensitivity with a blade sweep ratio (BSR) less than 1.0, though
a possible connection between the maximum BSR and the non dimensional parameter
ϵ/∆ is discussed. The classical ALM shows low sensitivity with a smoothing-ratio of
0.85 or greater. However, this is increased to 300 actuator points for ALM with the
addition of the filtered-lifting line correction (FLLC). The FLLC significantly improved
the accuracy compared to the classical ALM in both fixed and chord-varying ϵ configurations. The latter could achieve similar accuracy, but grid refinement limitations, due to computational cost, were mitigated by using the FLLC. A static force balance
model was used to explore mechanisms driving time variations in the MB force vector.
The analysis showed the modified out-of-plane bending moment vector drives the time
variations in the MB force vector, due to the contribution from the hub forces being 1-3
order of magnitude smaller in comparison. Furthermore, rotor weight contributes only
to the average MB force vector and not to the time variations, in a rigid rotor configuration. It is demonstrated that asymmetry in the velocity field over the rotor disk drives
variations in the MB force. Comparing the out-of-plane bending moment generated by
ABL turbulence against a steady shear inflow, atmospheric turbulence generates high
levels of fluctuations at the low and high frequency content and modulates the time
variations in the 3P frequency content. Leading to the classification of three distinct
frequency ranges (low, 3P, high). Specific periods of high-frequency and 3P activity
were identified and found to qualitatively align with low-frequency peaks in MB force.
However, limited overlap between peak ”bursting” events across frequencies suggests
distinct underlying mechanisms. A novel blade asymmetry vector was introduced, revealing that blade asymmetry is the dominant driver of MB force variability across all
frequency bands. But there are subtle differences in the way asymmetry drives the time
variations over the three frequency ranges. Using a novel methodology high-frequency
fluctuations were shown to cause sub-second force jumps comparable in magnitude to
rotor weight, posing a risk for edge loading, flange impact, and reduced lubrication
film thickness. These extreme transients may be contributing to surface-initiated failure mechanisms. Further analysis demonstrated that LSS contribute more to rotor
asymmetry and large force jumps than HSR, likely due to their higher coherence and
internal velocity gradients. Finally, the influence of blade flexibility was assessed by
re-running simulations with deformable blades. A less than 10% change between the
calculations indicated that the core findings from the rigid rotor analysis remain valid.
High-frequency loading events persist, suggesting that ABL turbulence, particularly
in the form of coherent eddies like LSS, plays a key role in triggering dynamic load
variations that may initiate MB failure.
| Date of Award | 17 Jun 2025 |
|---|
| Original language | English |
|---|
| Awarding Institution | - University Of Strathclyde
|
|---|
| Sponsors | University of Strathclyde & EPSRC (Engineering and Physical Sciences Research Council) |
|---|
| Supervisor | Edward John Hart (Supervisor) & Adam Stock (Supervisor) |
|---|