Generator Insulation Testing: A Field Engineer’s Guide

A generator failure during an emergency is the worst possible time for a generator to fail. Data centers, hospitals, water treatment plants, and utility backup systems all depend on generators that sit idle for months and then must start and carry full load within seconds. Whether they do depends largely on the condition of their insulation — and whether that condition has been verified recently.

This guide covers insulation testing for generators specifically, which is different in important ways from motor testing. A generator has two separate winding systems (stator and rotor) with very different insulation characteristics and test requirements. There are hydrogen-cooling considerations, water-cooled stator bar issues, and the unique challenge that the most critical generators are tested least often because they run rarely.

All values come from IEEE 43-2013, IEEE 56, and IEEE 1248, cross-referenced with manufacturer practice.

Why Generator Testing Is Different from Motor Testing

A motor converts electrical energy to mechanical energy. A generator does the opposite. From a physics standpoint, they’re nearly identical — both have stator windings, rotor windings, and the same insulation considerations.

But operationally, generators have characteristics that change how you test them:

They’re usually larger than motors. Industrial motors are commonly under 500 kW. Generators in serious applications are often 1 MW, 10 MW, or larger. Size matters for insulation testing — bigger machines have more insulation surface area, which affects expected IR values and test procedures.

They run at higher voltages. A typical industrial motor runs at 400 V or 690 V. Utility and industrial generators commonly operate at 6.9 kV, 13.8 kV, or higher. Test voltages scale up accordingly.

They have separately excited field (rotor) windings. Motors (induction, mostly) don’t have a separately excited rotor. Generators do — and the rotor field winding has its own insulation system that needs separate testing.

They run intermittently. Emergency backup generators may sit idle 99% of the time. That idle time is when moisture absorption and contamination accumulate. The “test before every critical load” approach doesn’t exist for generators — you test on a schedule and trust the results.

They’re often hydrogen-cooled. Large generators (typically 40 MW+) use hydrogen gas for cooling. This changes testing procedure dramatically — you can’t test with the machine depressurized in some cases, and some tests have to wait for scheduled outages.

Water-cooled stator bars add complexity. The largest generators use water-cooled stator windings. The cooling water path creates a conductive path through the winding that invalidates a standard IR test unless the water is drained first.

Stator vs Rotor: Two Different Insulation Systems

Understanding the difference between stator and rotor insulation is the key to interpreting generator test results correctly.

Stator windings

The stator is the large stationary winding where the electrical output is generated. In most generators:

• Voltage: Up to the generator’s output voltage (commonly 6.9 kV–27 kV)
• Insulation: Extensive — typically epoxy-mica form-wound coils with multiple layers of ground wall insulation
• Insulation volume: Very large. Absorption current decays slowly over 10+ minutes
• PI behavior: The PI test is highly relevant. A healthy stator shows a rising resistance curve for 10 minutes, producing PI ≥ 2.0

Rotor (field) windings

The rotor carries DC current that creates the magnetic field (in synchronous generators). In most generators:

• Voltage: DC, typically 125 V, 250 V, or 375 V (exciter output)
• Insulation: Minimal compared to stator — windings are essentially exposed to the environment, with insulation mostly on the conductors themselves
• Insulation volume: Much smaller. Absorption current decays rapidly
• PI behavior: A healthy rotor often shows a flat PI near 1.0 or a modest PI of 1.2–1.5. IEEE 43-2013 recognizes that rotor PI typically runs lower than stator PI due to the exposed insulation system.

Why this matters

Applying stator PI criteria (≥ 2.0) to a rotor will fail healthy rotors. You need separate interpretation tables for each.

Some technicians call the rotor the “field.” Both terms are correct. In brushless exciter systems, the terminology gets confusing because the rotating component can be either the field or the armature depending on the design. For this article, “rotor” and “field” refer to the DC-excited winding — whichever physical component that is on your specific generator.

Applicable Standards

Standard	Role
IEEE 43-2013	Primary standard for insulation resistance and PI testing of rotating machinery — applies to both motors and generators
IEEE 56	Guide for insulation maintenance of large AC rotating machinery (10,000 kVA and larger)
IEEE 1248	Guide for commissioning large generators
IEEE 95-2002	High DC voltage testing for form-wound machines rated 2,300 V and above
NETA MTS	Maintenance testing specifications for field personnel

For a small generator (below 1 MW), IEEE 43-2013 is the primary reference. For large generators, IEEE 56 adds guidance specific to large-machine maintenance, and IEEE 1248 covers commissioning. For very high-voltage step testing, IEEE 95 applies.

Stator Testing Procedure

Before starting

1. Isolate the generator. Trip and lock open the generator breaker. Disconnect from the grid. If connected to a step-up transformer, open the transformer-side disconnect.
2. Depressurize hydrogen. For hydrogen-cooled machines, purge the hydrogen and refill with air or inert gas before opening any terminal covers. Follow the OEM procedure exactly — mishandled hydrogen purges are life-safety events.
3. Drain water. For water-cooled stator bars, drain and vacuum the cooling water. Testing with water in the bars will give meaningless results and can damage the insulation.
4. Disconnect auxiliaries. Disconnect surge capacitors at the generator terminals, neutral grounding equipment, CT and VT connections, and RTDs (ground them or isolate them).

Test connections

For a three-phase generator with all six leads brought out (both main leads and neutral leads accessible):

• Disconnect both ends of each phase. This allows true phase-by-phase testing.
• Connect copper bonding jumpers at both ends of the winding during the test. This prevents transient voltages if the test is interrupted.

For a generator with only three terminals brought out (neutral internally connected):

• You must test all three phases simultaneously (bulk testing), which gives only insulation-to-ground data.

Running the test — stator IR and PI

1. Select test voltage per IEEE 43-2013 Table 1:

Stator Rated Voltage	DC Test Voltage
Up to 1 kV	500 V
1–2.5 kV	500–1,000 V
2.5–5 kV	1,000–2,500 V
5–12 kV	2,500–5,000 V
Above 12 kV	5,000–10,000 V

For a typical 13.8 kV generator, use 5,000 V DC.

2. Test each phase individually (if possible), with the other two phases grounded. Ground all three phases to the same point as the stator core to prevent ground circuit losses.
3. Apply test voltage. Record the reading at 30 seconds, 1 minute, 5 minutes, and 10 minutes.
4. Calculate:
   • IR (spot reading): The 60-second reading
   • DAR: IR₆₀ ÷ IR₃₀
   • PI: IR₆₀₀ ÷ IR₆₀
5. Repeat for each phase with a full discharge between tests.

After the test

• Discharge for at least 40 minutes after a 10-minute PI test. Large stators store significant charge.
• Verify voltage below 50 V with a voltmeter before touching any conductor.
• Reconnect all auxiliaries in reverse order.
• Refill hydrogen per OEM procedure.

Rotor (Field) Testing Procedure

Key differences from stator testing

• Test voltage is fixed: 500 V DC regardless of the exciter’s operating DC voltage. The rotor’s exposed insulation system doesn’t require higher test voltages.
• No phase-by-phase testing: The rotor has a single winding, not three phases.
• Lower PI expectations: 1.2–1.5 is typical for healthy rotors. Applying stator criteria (PI ≥ 2.0) will fail healthy machines.
• Brush rigging considerations: For brush-type excitation systems, the test connections are at the slip rings. For brushless exciters, access may require disassembly.

Procedure

1. Isolate the excitation system. Open the field disconnect. For brushless systems, remove the exciter armature connections.
2. Short the field. Connect a grounding jumper across the rotor winding to discharge any stored energy.
3. Connect the megger:
   • LINE lead to the positive slip ring (or field positive terminal for brushless systems)
   • EARTH lead to the rotor shaft (ground)
4. Apply 500 V DC for 60 seconds minimum. If you want PI data, extend to 10 minutes.
5. Record the 30-second, 60-second, and 10-minute readings.
6. Discharge thoroughly through a grounding jumper for at least 40 minutes after a PI test.

Interpreting Stator Readings

Minimum IR for generator stators (IEEE 43-2013, Clause 12.1)

Same values as motors:

Stator Type	Minimum IR at 40°C
Form-wound (post-1970)	100 MΩ
Random-wound and form-wound <1 kV	5 MΩ
Pre-1970 windings	(kV + 1) MΩ

For a 13.8 kV generator with modern form-wound insulation: minimum 100 MΩ at 40°C.

PI interpretation for stators (IEEE 43-2013, Table 3)

Insulation Class	Minimum PI
Class A	1.5
Class B, F, H	2.0

Most modern generator stators are Class F — minimum PI = 2.0.

When to investigate

• IR below minimum: Investigate. Common causes are moisture (most common), contamination, internal damage.
• PI below 2.0 (Class F/H): The insulation is not absorbing charge normally. Often indicates moisture saturation.
• PI above 8: IEEE 43-2013 notes that very high PI values on older insulation types (varnished cambric, shellac mica-folium, asphaltic) can indicate that the insulation has aged thermally and become brittle. For modern epoxy-mica insulation, this concern doesn’t apply.
• Per-phase imbalance: One phase reading significantly below the other two indicates a localized problem in that phase — usually moisture ingress, physical damage, or end-winding contamination.

Phase-by-phase correction factor

IEEE Standard 43 notes that when megger-testing one phase at a time (with the other two phases grounded), the required megohm value should be multiplied by 2 unless guards are used. When guards are used, multiply by 3. Some practitioners apply a ×3 factor in both cases for conservative evaluation. In practice: don’t apply generic minimum values directly to phase-by-phase readings without considering this factor.

Interpreting Rotor Readings

Minimum IR for rotors

There’s no single published minimum for rotor IR because rotor insulation systems vary widely. General guidelines based on field practice:

Rotor IR at 500 V DC	Assessment
Above 100 MΩ	Excellent
10–100 MΩ	Good
1–10 MΩ	Monitor — still functional
Below 1 MΩ	Investigate — likely moisture or contamination
Below 0.1 MΩ	Critical — do not energize

PI for rotors: different expectations

The big difference from stators: rotor PI is typically much lower, even on healthy rotors, because the rotor insulation system is more exposed to the environment with less absorption capacity.

Rotor PI	Assessment
2.0+	Excellent — unusually good
1.2–2.0	Normal and healthy
1.0–1.2	Investigate if declining from previous readings
Below 1.0	Problem — resistance actively declining during test

The common mistake

Applying stator PI criteria (≥ 2.0) to rotors fails healthy machines. A rotor reading PI of 1.4 with IR of 50 MΩ is in good condition — not failing.

Hydrogen-Cooled Generators

Large generators (typically 40 MW+) use hydrogen gas in the machine frame for cooling. Hydrogen has about 14× the heat capacity of air and creates much less windage loss, allowing the machine to run cooler and more efficiently. But it creates testing complications.

Why hydrogen matters for insulation testing

Safety: Hydrogen is flammable and explosive at concentrations of 4–75% in air. Any spark during depressurization or purging can cause an explosion.

Procedural complexity: Testing the generator fully requires depressurizing the hydrogen, purging with CO₂ (to prevent hydrogen-air explosive mixtures), then filling with air. This process typically takes 24–48 hours and is done during scheduled outages only.

Routine tests can be done under hydrogen: Some tests, like megger testing of the rotor through the slip rings, can be done with the generator still pressurized with hydrogen — provided the test connections don’t require opening the generator frame.

What this means in practice

Routine (every 1–2 year) IR and PI testing of hydrogen-cooled generator stators is done during scheduled outages. During annual or semi-annual rotor tests, the machine often stays pressurized. The intervals are longer than for air-cooled machines because the procedural cost of accessing the machine is higher.

Critical safety rule

Never perform any insulation test that could produce a spark inside a hydrogen-pressurized machine. This includes tests that would require opening sealed terminals. The megger applying test voltage to external terminals doesn’t introduce sparks inside the machine — but physical access to terminals inside does.

Water-Cooled Stator Bars

The largest generators (typically 400 MW+) use hollow stator conductors with water circulating through them for cooling. Water-cooled stators are common on nuclear and large fossil generators.

The testing challenge

The cooling water inside the stator bars creates a conductive path through the winding. A standard IR test applied to a water-filled stator measures the combined resistance of the insulation and the water — which is meaningless.

Required procedure

Per IEEE practice:

1. Drain the cooling water from the stator bars using the OEM-specified drain procedure.
2. Apply vacuum to the stator bar water passages to evacuate residual moisture. Standard procedure is to pull vacuum to below 10 torr and hold for several hours.
3. Verify dryness before applying test voltage — some manufacturers require a low-voltage (250 V) pretest to confirm no conductive water path remains.
4. Perform the standard IR and PI test at the appropriate voltage for the stator rating.
5. Refill and vent the cooling water system after testing, following OEM procedure.

Practical reality

This procedure adds a day or two to a maintenance outage. For this reason, water-cooled stator testing is typically done at major overhauls (every 5–10 years), not as routine maintenance.

Emergency Backup Generators: The Neglected Asset

Emergency and standby generators are often the least-tested critical equipment in a facility. They sit idle most of the time, get started briefly during monthly exercises, and carry full load only during actual emergencies. The result: insulation problems that would be caught quickly on a continuously-operating machine go undetected for years.

Why this is dangerous

A hospital’s emergency generator is expected to start within 10 seconds of a utility outage and carry full load for hours. If the insulation has been quietly degrading during the 99% of the time the generator was idle, the emergency start can expose a fault that takes the generator offline exactly when it’s needed.

What a proper test program looks like

Monthly exercise runs are required by most codes (NFPA 110 for emergency systems). These verify the generator starts and runs, but they don’t verify insulation condition.

Annual insulation testing should be a separate procedure:

• Complete IR and PI test on stator
• Rotor IR test
• Comparison to previous readings (trending is critical here)
• Temperature correction to 40°C for motor/generator comparison

After any extended storage (above 30 days of not running under load), perform a full IR test before returning to service. Moisture absorption during storage can cause IR to drop below minimum.

Humidity and temperature matter most here. Emergency generators in unconditioned rooms (common for rooftop or basement installations) experience larger humidity swings than generators in operating rooms. Insulation absorbs moisture during humid periods and dries out during dry periods. Test during both conditions if practical — you’ll see what the actual operating range is.

Common neglect patterns

• “We test it every month by running it, so we don’t need insulation tests.” Running tests verify the starter, governor, fuel system, and voltage regulator. They do not verify insulation condition.
• “The generator has never failed, so we assume it’s fine.” Latent insulation problems don’t announce themselves until the moment of failure.
• “Insulation testing is for maintenance, not commissioning.” Wrong. IEEE 1248 specifies insulation testing as part of generator commissioning, and baseline readings are essential for all future trending.

Test Intervals by Generator Type

Generator Type	Stator IR/PI	Rotor IR	After Extended Idle
Emergency/standby (< 1 MW)	Annually	Annually	Before return to service
Continuous-duty industrial (1–10 MW)	Every 2 years	Every 2 years	If idle >60 days
Utility generators (10–100 MW)	Every 3–5 years (with other overhaul work)	Annually at brush inspection	At every outage
Large hydrogen-cooled (>100 MW)	Every 5 years at major outage	Annually if access permits	N/A (continuous operation)
Water-cooled stator generators	Every 5–10 years at major overhaul	Same as rotor above	N/A

Note: These intervals are general guidelines. Specific requirements come from the OEM, the facility’s maintenance philosophy, and applicable codes (NFPA 110 for emergency systems, NERC standards for utility generators).

Common Mistakes

Applying stator PI criteria to rotors. A rotor PI of 1.3 is healthy. Applied to a stator, it would indicate a problem. Use the right interpretation for the right winding.

Testing water-cooled stators without draining the water. The reading is meaningless. Worse, some technicians interpret the resulting “low” reading as insulation failure and initiate unnecessary repairs.

Not depressurizing hydrogen before testing terminal connections. Safety hazard — potential hydrogen-air explosive mixture during terminal access.

Only testing at scheduled maintenance. Emergency generators need insulation testing annually regardless of maintenance schedule. Generators after extended storage need pre-service testing.

Not trending the data. A single reading tells you the generator passed. A trend tells you whether it’s getting worse. Without trending, you catch problems only when they exceed the minimum threshold — not before.

Assuming monthly runs equal health verification. Running the generator verifies starting and mechanical health. It doesn’t verify insulation condition. These are separate checks.

Over-testing with high voltage. Applying the highest possible test voltage during routine maintenance can stress aged insulation. For routine tests, use the lower end of the IEEE 43-2013 voltage range. Save high-voltage testing for commissioning and post-repair verification.

FAQ

Key Takeaways

• Stators and rotors need different testing procedures and different interpretation criteria. Don’t mix them.
• Stator test voltages follow IEEE 43-2013 Table 1. For a 13.8 kV stator, use 5,000 V DC. For rotors, always use 500 V DC regardless of operating voltage.
• Stator minimum PI: 2.0 for Class F/H insulation. Rotor PI of 1.2–1.5 is typically healthy — don’t apply stator criteria.
• Water-cooled stators require draining and vacuum drying before testing. Testing with water in the bars gives meaningless results.
• Hydrogen-cooled generators require careful procedure. Never open terminals in a hydrogen-pressurized machine.
• Emergency generators are the most-neglected critical equipment. Annual insulation testing is mandatory — monthly run exercises don’t verify insulation condition.
• Trending matters more for generators than for motors because of the longer intervals between tests. Record every reading with conditions.
• Test after extended idle periods (30+ days for emergency units). Moisture absorption during storage is the most common generator insulation issue.

Standards Referenced in This Article

Standard	Key Content
IEEE 43-2013	Test voltages (Table 1), minimum IR (Clause 12.1), minimum PI (Table 3), phase-by-phase correction factors, rotor PI guidance
IEEE 56	Insulation maintenance of large AC rotating machinery (10,000 kVA+)
IEEE 1248	Commissioning practices for generators
IEEE 95-2002	High DC voltage testing of form-wound machines ≥2,300 V
NFPA 110	Emergency and standby power systems — testing requirements
NETA MTS	Maintenance testing specifications