From the field · Case 03

Yaw Hunting — The Nacelle Oscillates While Turning

An Enercon E92. Instead of settling into the wind, the nacelle kept overshooting and coming back — swinging about ±18° at roughly 0.5 to 1 Hz. Yaw hunting is one of those faults where looking for a single culprit wastes your day, because it is usually several small things adding up.

Symptom

• Behaviour: the nacelle oscillates ±18° about the wind direction instead of settling
• Frequency: roughly 0.5–1.0 Hz
• Yaw motor current: elevated, because the drive is working almost continuously
• Production: below expectation, since the rotor spends much of its time off-axis
• Sound: a friction noise from the yaw deck during movement

First assessment

Yaw is a closed loop: a sensor reports the wind direction, a controller decides, a drive moves the nacelle, and the sensor reads the result. Hunting means that loop has become unstable. Instability comes from one of three places — the sensor is feeding in noise, the controller is reacting too hard, or the mechanics are not responding the way the controller expects.

The friction noise made me suspect the mechanical side, but I did not want to assume. With a control loop it pays to examine every element, because the answer is often a combination rather than a single fault.

Step-by-step diagnosis

1. Looked at the SCADA trend for the oscillation shape. A regular, sustained oscillation rather than a decaying one. That already suggests the controller is feeding the motion rather than damping it.
2. Drove the yaw manually in controlled steps. With the automatic loop disabled. The nacelle moved, but not smoothly — the resistance varied through the rotation.
3. Checked bearing drag. Measuring the current needed to start and sustain the movement showed it was higher than it should be, and it varied by position. The bearing was dragging.
4. Examined the vane sensor. Its mounting was slightly loose. A loose sensor produces small, fast fluctuations in the reported direction — jitter — which a controller will faithfully chase.
5. Checked the drive electronics. Motor windings and drive output were normal. Nothing to explain the behaviour on the electrical side.
6. Read the controller parameters. The proportional gain sat at 1.5 against a design value of 0.6. A gain that high makes the loop react aggressively to every small error, including sensor noise. This was the piece that turned a sluggish mechanism into an oscillation.

Root cause

Three factors compounding. The bearing was dragging, so the nacelle did not respond the way the controller expected. The vane sensor was loose, so the controller was being fed jitter. And the proportional gain had been left at 1.5 instead of 0.6, so the controller amplified both problems instead of absorbing them.

Any one of these alone would have produced a mildly imperfect yaw. Together they produced a loop that would not settle. This is the normal shape of a control-loop fault: no single part is dramatically broken, but the combination is unstable.

Fix

1. Controller parameter restored (30 minutes): proportional gain set back to the design value of 0.6.
2. Vane sensor remounted (1 hour): mounting tightened and alignment verified, removing the jitter at source.
3. Bearing lubrication (2 hours): the slewing bearing was lubricated and the lubrication delivery checked at each point.
4. Result: the oscillation narrowed to about ±2–3° within two weeks, which is normal operating behaviour, and the yaw motor current came back down.

Preventive maintenance

• Treat hunting as a loop problem, not a part problem. Check the sensor, the controller and the mechanics in turn; expect the answer to be a combination.
• Record controller parameters after every intervention. A gain left at a commissioning or troubleshooting value is a common and invisible cause.
• Check vane sensor mounting at every service. It takes a minute and a loose sensor poisons every downstream decision.
• Watch yaw motor current as a mechanical health signal. Rising current with no change in duty is the earliest sign of bearing drag.