Article content will be supplied from the Insights CMS collection.
Generator rotor winding failures are among the more consequential failure categories in large turbine-generators — not primarily because the rotor winding is especially fragile, but because the failure modes are progressive, the monitoring signals are indirect, and the decision about whether to rewind at a given outage is made with imperfect information under schedule pressure.
This article covers how rotor winding failures develop, how they manifest in the data available before and during an outage, and what pre-outage analysis — including RSO testing and rotor flux trending — adds to the decision about the appropriate outage scope.
The Rotor Winding: What It Is and Why It Fails The rotor winding of a synchronous generator carries DC excitation current through copper coils wound in slots machined into the rotor forging. The current produces the magnetic flux that, rotating with the rotor, induces voltage in the stator winding. Adequate excitation is fundamental to the generator's ability to produce real and reactive power.
The rotor winding operates in a mechanically demanding environment. The rotor rotates at synchronous speed (3,600 RPM for 60 Hz two-pole machines), which subjects the windings to centrifugal loads that increase with the square of speed. The copper conductors undergo thermal cycling as the machine loads and unloads. The turn insulation — the insulation between adjacent turns of the same coil — is the component most susceptible to degradation over the machine's life.
Turn-to-turn insulation. Turn insulation in large generators is typically a composite of glass fiber and resin, applied as a groundwall-style tape or as slot liners. Unlike stator insulation, which operates at high voltage, rotor turn insulation sees only the inter-turn voltage — typically a few volts across each turn. The failure concern is not voltage breakdown but mechanical wear: the relative motion between conductors during thermal cycling and under centrifugal load abrades the turn insulation and eventually creates contact between turns.
Why failures develop gradually. A single turn-to-turn short in a large rotor winding does not immediately produce a catastrophic failure. A short circuit between two turns reduces the effective turns count in that coil, creates a localized current imbalance, and produces a thermal asymmetry. The machine can often operate with one or a small number of turn shorts — at reduced reactive capability and with characteristic vibration signatures. But the condition is progressive. The additional heating at the fault location damages adjacent insulation, additional turns fail, and the fault escalates.
Common Failure Modes Turn-to-Turn Shorts
Turn-to-turn shorts are the most common rotor winding failure mode in service-aged generators. They are created by mechanical wear on turn insulation, typically in the slot or in the end-turn region where conductor motion during thermal cycling is greatest. Contamination — oil, moisture, metallic particles — can accelerate insulation degradation and create low-impedance paths between turns before mechanical wear alone would produce contact.
Operational consequence: A turn short reduces the magnetic symmetry of the rotor. The flux imbalance creates a once-per-revolution force on the rotor — manifesting as a synchronous vibration component. Vibration data is often the first indication that a turn short has developed, though it is not definitive: changes in balance condition, bearing wear, or alignment issues produce the same vibration signature.
Rotor Ground Faults
A ground fault occurs when the rotor winding makes electrical contact with the rotor body. The rotor is typically operated floating — the winding is not intentionally grounded, and a first ground fault creates no current path. A single ground fault in a healthy rotor is often deferred, monitored, and addressed at the next planned outage. The risk is the second ground fault. If a second point in the winding contacts the rotor body while the first remains, a current path is created through the rotor forging between the two fault locations. The resulting current produces intense localized heating and can damage the rotor forging itself — a significantly more consequential and expensive failure than the winding fault that initiated it.
Monitoring practice: Most rotor ground fault detection systems operate on the first-fault alarm-only basis — they detect the first fault but do not trip the unit. The appropriate response to a first ground fault indication is not always immediate shutdown, but it does warrant accelerated monitoring and engineering review of the fault location's stability and the second-fault probability.
Thermal Sensitivity
Some generators exhibit rotor winding turn shorts that are intermittent — present at operating temperature and absent when the machine is cold. This phenomenon, called thermal sensitivity, occurs because thermal expansion of the conductors creates contact between turns that are nearly touching in the cold condition. A generator with thermal sensitivity will show inconsistent RSO test results between cold and hot testing, and may show vibration changes that track with excitation level or thermal load.
Thermal sensitivity is significant for outage planning because a winding that tests acceptably cold may have a different condition than the test results suggest.
How Rotor Winding Faults Manifest Before the Outage Excitation Current and Field Voltage Trending
A turn-to-turn short reduces the effective resistance of the rotor winding. If the field current required to maintain rated voltage at a given operating point increases over time — holding all other variables constant — the change indicates either a change in the rotor circuit resistance or a change in the machine's magnetic characteristics. Trending the ratio of field current to reactive power output, with appropriate normalization for operating conditions, provides an indirect indicator of rotor winding condition over time.
This is not a high-sensitivity indicator — significant turns can fail before the change is detectable in the operating data — but it is available without additional testing and can provide trend context for the RSO results obtained at outage.
Vibration Trending
Synchronous vibration changes (1X component) that track with changes in excitation level are characteristic of magnetic asymmetry from a turn short. The distinguishing feature is the thermal dependency: the vibration change occurs as the machine comes up to temperature or as excitation level changes, rather than tracking purely with speed or mechanical unbalance. Identifying this pattern in historical vibration data before the outage allows the outage team to enter the rotor inspection with a working hypothesis about fault location.
Flux Probe Testing (Online)
Some facilities have rotor flux probes installed — sensors positioned in the stator bore that detect the flux pattern produced by the rotating rotor. Flux probe data can identify the pole and approximate slot location of a turn short without requiring the rotor to be removed. This information is most useful as an input to outage scope planning: knowing which pole and slot is affected allows the outage team to prioritize that region during visual inspection and to assess whether the fault location is accessible for repair.
RSO Testing: What It Tells You and What It Doesn't Recurrent Surge Oscillation (RSO) testing is the primary offline electrical test for rotor winding turn-to-turn shorts. The test applies a voltage impulse to each end of the rotor winding and compares the resulting waveform from both ends. In an undamaged winding, the two waveforms are mirror images of each other; a turn short produces a detectable asymmetry.
What RSO testing tells you. RSO testing is effective at detecting turn-to-turn shorts that exist at the time of testing. It provides a relative indication of fault severity — a large waveform asymmetry indicates more faults than a small asymmetry — and when trended across multiple outages, it provides evidence of whether the condition is stable or progressing.
What RSO testing does not tell you. RSO testing does not reliably detect thermally sensitive faults in a cold winding. It does not localize the fault to a specific slot without additional analysis. And a single RSO test result without historical comparison provides limited basis for a decision about rewind versus return to service — the baseline established at prior outages is what makes the current result interpretable.
Establishing baseline. For generators without historical RSO records, the first RSO test produces a baseline, not an assessment. The assessment comes at the second outage, when the baseline can be compared. This is a common situation with units whose RSO testing has been intermittent — a current result that shows asymmetry provides no information about whether the condition is new or stable without a prior result to compare it against. The Rewind Decision Deciding whether a rotor winding requires rewind at a given outage — versus repair of accessible faults or return to service with monitoring — is one of the more consequential engineering decisions made during a turbine-generator outage.
Repair vs. rewind. Not all rotor winding faults require a full rewind. A single turn short in an accessible location in the end-turn region may be repairable by replacing the affected turn insulation. Slot faults require slot bar removal, which is more extensive but still short of a full rewind. The decision between repair and rewind depends on fault count, fault locations, the general condition of the turn insulation as assessed during the outage, and the machine's operating profile going forward.
The risk of deferral. Deferring rotor winding work with an identified fault and no monitoring plan converts an engineering judgment into an undocumented risk accumulation. If deferral is the right decision — and it sometimes is — the basis for deferral should be documented: the fault count, the RSO trend, the vibration trend, the monitoring plan for the next operating period, and the escalation criteria that would change the deferral decision.