The question every campaign should start with

When a fleet study points at yaw calipers, the easy path is to replace the units, tick the boxes, and move on. The honest path starts with a different question: how will we know the intervention worked — in numbers, per turbine, before anyone orders a part? If that answer doesn't exist at quoting stage, the campaign is buying activity, not outcomes.

Baseline first — before anyone touches a turbine

Verification is impossible without a baseline. Six to eight weeks before the work window, we record each target turbine's behaviour: vibration RMS in the frequency bands associated with yaw activity, binned by wind speed; the yaw misalignment distribution from SCADA; and the yaw duty cycle — how often and how hard the drives and brakes are working to hold position.

The wind-speed binning matters. A turbine's vibration signature in 6–8 m/s says nothing about its behaviour in 12–14 m/s, where the loads that reveal a weak caliper actually appear. Compare like with like, or don't compare at all.

What we measure after the work

Post-work, the same measurements run again over an equivalent period and equivalent conditions. The comparison is done on three levels:

  • Vibration distribution shift — not one number, but the whole histogram moving down and tightening in the yaw-related bands.
  • Misalignment statistics — the frequency and depth of action-band excursions, measured against the same wind bins as the baseline.
  • Yaw duty cycle — a healthy braking system holds position with less hunting, and the drives work less to correct drift.

Reading the before/after honestly

Two false positives show up constantly. The first is seasonal: if the baseline was recorded in autumn winds and the verification in winter gales, the raw comparison is meaningless — which is exactly why every measurement is binned by wind speed first. The second is sensor drift: a misalignment sensor that drifted between campaigns will happily "improve" a turbine that didn't change. Cross-checking against power curve behaviour and vibration catches that.

The reverse mistake also exists: a genuine improvement that lands outside the measurement window gets missed, and a turbine that was actually fixed gets re-flagged. That's why the verification period is matched to the baseline in duration and conditions, not just run for a convenient fortnight.

Documentation that survives an audit

The close-out pack for a caliper campaign should stand on its own: as-found and as-left condition per unit, torque and adjustment records, the before/after verification plots with wind-bin context, and a plain-English statement of what changed and by how much. Anything less turns the campaign back into an assumption — an expensive one, this time with paperwork.

Takeaways

  • Define the verification metric at quoting stage, not at close-out.
  • Baseline each target turbine weeks before the work, binned by wind speed.
  • Judge distributions, not single points — and equivalent conditions, or nothing.
  • Watch for seasonality and sensor drift; both manufacture improvements that aren't there.
  • Close out with a pack that proves the outcome to someone who wasn't on site.