A stator winding is the largest single capacitance in most generating plants, and testing it means applying an alternating voltage to that capacitance for a defined period. Generator stator testing resonant AC methods exist because a conventional test set cannot economically supply the charging current of a large machine. What determines whether the test succeeds, though, is usually the preparation rather than the equipment.

The test window is short, the machine is out of service, and the set has to be mobilised, connected, tuned, run and removed within that window. Every item that has not been prepared before the window opens consumes time from the test itself, and the items that consume the most are the ones that are easiest to overlook.

Why stators are tested with resonant AC

Stator capacitance grows with machine rating and winding length, so at system frequency a large machine needs a test source of considerable apparent power that a site supply rarely provides. Why a resonant circuit solves that problem, and how it tunes to the winding, is explained in the site’s guide to why resonant AC testing is used for generator stators. This article picks up from there and covers what that explanation leaves out: how to plan and prepare the test window, what to do during the test, and how to read the result afterwards.

Two consequences of the physics matter for the planning that follows. The winding stores energy, which is why the discharge and grounding arrangements are part of the test rather than a step after it. And the test frequency is set by the tuned circuit rather than by the supply, so the frequency actually achieved has to be recorded rather than assumed.

Planning the test window in an outage

The window has to cover more than the test itself. Mobilising the set to site, assembling the reactor and the excitation transformer, making the high-voltage connections to the winding, tuning the circuit, running the test at the specified voltage for the specified duration, and then disassembling and removing everything.

The duration of that sequence depends on the machine and on the site. A machine in a power station with crane access and a laydown area presents a different task from a hydro unit in a confined powerhouse. Estimating the window from the test duration alone is the commonest reason a test is rushed at the end.

The plan should also account for what else has to happen while the machine is out of service. A stator test is often combined with other outage work, and the sequence has to be arranged so that the test is not competing for the same access as the other activities.

Preparation: isolation, cleanliness and grounding

Isolation and locking off come first, with the machine confirmed disconnected from all sources including the excitation system, the neutral connection and any auxiliary supply that could reach the winding. The earthing arrangement for the test is then established and the working area controlled.

Cleaning the stator end windings is the second preparation and it is not cosmetic. Surface contamination provides a leakage path that affects both the withstand result and the partial discharge measurement, and a winding that has accumulated dust, oil mist or conductive contamination behaves differently from a clean one. Cleaning before the test is what makes the measurement describe the insulation rather than the surface.

The instrumentation left connected to the winding has to be dealt with explicitly. Resistance temperature detectors, thermocouples and any monitoring equipment connected to the winding become part of the circuit during the test, and the plan should state which are disconnected, which are earthed and which are left connected on the basis of their insulation rating. Items that are forgotten are the source of both measurement errors and failures that are attributed to the winding.

Test voltage and duration by standard

The test voltage and the duration for a stator winding depend on the machine rated voltage and on the purpose of the test. A commissioning test on a new machine, a periodic maintenance test and a test after a winding repair may each use a different level, and the criteria are written in relation to the level applied.

The frequency used is that of the tuned circuit and depends on the winding capacitance and the reactor. The standard governing the test defines an acceptable frequency band, and a test performed outside that band produces a result that cannot be judged against the acceptance criteria derived for it. The frequency actually achieved should be recorded, not the intended one.

Partial discharge during the test

Measuring partial discharge during the withstand test is standard practice on machine stators because the winding’s failure mode is usually a slow insulation degradation rather than a sudden breakdown. The measurement indicates whether there is an active site at the test voltage, which the pass-or-fail outcome alone does not show.

The measurement has to be configured for the frequency being used. The rate of change of the applied voltage at a low test frequency differs from that at the power frequency, so the number and appearance of discharge events differ, and an instrument configured for power frequency measurement does not transfer directly. The method itself is defined in IEC 60270.

The noise environment of a powerhouse is hostile. The detection threshold achieved in that environment is the quantity that determines what the measurement can find, and it belongs in the record alongside the result. A discharge measurement reported without its threshold cannot be compared with a later test performed under different conditions.

Interpreting results against history

The most useful comparison is against the same machine at an earlier test, with the same voltage, duration and frequency. Where those conditions are held, the differences reflect the insulation rather than the test, and the trend of the loss and discharge values becomes interpretable.

Where no history exists, the first test establishes the baseline. That is the reason a first stator test is worth performing even when nothing suggests a problem: the value of the test years later depends on having something to compare against.

Comparisons against sister machines provide a weaker reference, because the windings differ in their construction and their history. The comparison is useful for identifying a machine that is markedly different from its group and less useful for detecting a small change.

Safety around a large capacitive load

A stator winding retains charge after the test voltage is removed, and the stored energy is large. The discharge procedure is therefore a critical part of the test, not an afterthought, and the plan should state how the winding is discharged, how long the discharge is maintained and how it is verified before anyone approaches the connections.

The resonant circuit adds its own considerations. The stored energy in the reactor and the winding together is substantial, and the circuit has to be de-energised in a controlled sequence. The grounding arrangement used during the test should be established before the voltage is applied and left in place until the winding has been discharged and earthed.

Access control during the test is the third element. The test voltage appears across parts of the machine that are normally accessible, and the working area has to be controlled for the duration. The general electrical safety guidance published by the UK Health and Safety Executive covers the principles that the site-specific procedure has to apply.

Reporting and follow-up actions

The report should state the test method, the achieved frequency, the applied voltage and duration, the connection and grounding arrangement, the current and voltage traces where recorded, and the discharge results with their detection threshold.

It should also state the comparison against the previous test on the same machine and the resulting interpretation. Where the test revealed activity, the report should name the further measurements that would characterise it rather than leaving the reader to decide.

Where the test was combined with other outage work, the report should reference that work, because a winding that has been cleaned, re-taped or repaired is a different test object from one that has not. Configuration guidance for resonant test arrangements is published by high-voltage test equipment suppliers such as the series resonance test system range and by Haefely, and machine insulation practice is coordinated through CIGRE study committees with the research background published by EPRI.

Setting the next test interval after the first stator test

The first test establishes the baseline and the interval decision follows from it. Where the results are clean and the machine duty is moderate, the next test can be set at the interval the operator’s standard specifies. Where the first test shows activity, loss characteristics above the expected level, or a value that differs from the sister machines, the next test should be set sooner.

The interval should also reflect the machine’s duty. A base-load machine and a peaking machine accumulate different thermal cycles, and thermal cycling is one of the mechanisms that degrades stator insulation. Where a machine’s duty changes, the interval deserves review.

The interval decision should be recorded with its reasoning, because it will be reviewed at the next test. A decision that reduced an interval on the basis of a specific observation is easy to revisit when that observation is either confirmed or not repeated. A decision with no recorded basis tends to be extended by default, which is how a baseline test becomes a formality. Machine and insulation test instruments are grouped on the generator detection testing hub and on the high-voltage insulation withstand test hub.