3.5 Insulation Degradation & Dielectric Theory

The Four Insulation Degradation Modes

Quick Answer: Insulation fails by thermal, electrical, mechanical, and environmental mechanisms. Most real failures combine two or more — a thermally aged winding absorbs moisture, tracks, and then breaks down electrically.

• Thermal: Every 10°C above the insulation class rating roughly halves insulation life (the Dakin-Arrhenius relationship). Overload, blocked ventilation, and cyclic loading embrittle the resin and delaminate layers. Most common long-term degradation.
• Electrical: Partial discharge (PD) in voids erodes resin via ion bombardment and ozone. Corona is high-voltage PD in air. Treeing (electrical trees in solid insulation, water trees in XLPE cables) is slow dendritic breakdown driven by voltage stress and moisture. Tracking is carbonized surface conduction paths from contamination and leakage.
• Mechanical: Vibration loosens windings in slots, thermal cycling expands and contracts conductors, and short-circuit magnetic forces deform coils. Abrasion creates voids that invite PD.
• Environmental: Moisture lowers IR and promotes water treeing; dust, oil, and chemical contamination create surface leakage paths. Motors stored without space heaters absorb moisture and fail on first energization.

IR, PI, DAR, surge, PD, tan-delta, and VLF cable tests each detect different combinations of these modes — which is why NETA acceptance packages use several methods.

Dielectric Theory: Three Currents in an IR Test

When you apply DC test voltage (e.g., 1,000 or 5,000 V with a megohmmeter), the total current is the sum of three components with different time constants:

1. Capacitive charging current (Ic): high initially, decays in seconds as the geometric capacitance charges — usually gone within the first minute.
2. Polarization / absorption current (Ia): decays over minutes to tens of minutes as dipolar molecules align with the field and interfacial (Maxwell-Wagner) polarization develops. This is the current PI and DAR measure.
3. Conduction leakage current (Ig): steady-state current through and over the insulation — what the final IR reflects.

The megohmmeter reads IR = Vapplied / (Ic + Ia + Ig). As Ic and Ia decay, measured current drops and IR rises. On good, dry insulation IR keeps climbing for 10 minutes; on wet or contaminated insulation, conduction leakage dominates and IR flattens within a minute.

Why IR, PI, and DAR Work

• Insulation Resistance (IR): the 1-minute value, IR₁ — IEEE 43 references this for minimum acceptance.
• Dielectric Absorption Ratio (DAR): DAR = IR(60 s) / IR(30 s). Good dry insulation shows DAR ≥ 1.4; below 1.25 suggests moist or contaminated insulation. Used where a 10-minute test is impractical.
• Polarization Index (PI): PI = IR(10 min) / IR(1 min). Good insulation shows PI ≥ 2.0 (class A) or ≥ 2.5-4.0 for modern epoxy; PI < 1.0 is unacceptable. PI is temperature-independent (both readings at the same temperature) and the most sensitive of the three to absorption behavior.

A low PI does not always mean low IR — a winding can have a high 1-minute IR but flatten out, indicating absorption is absent (moisture-saturated or contaminated). PI is the trending tool; IR is the acceptance tool.

IEEE 43-2013 Minimum IR (at 40°C)

IEEE 43-2013 Clause 12.3 gives three minimum IR₁ values:

The 'kV' is the rated line-to-line RMS voltage in kilovolts for three-phase AC machines, line-to-ground for single-phase, and rated DC voltage for DC machines or field windings. A 13.8 kV machine has a kV + 1 minimum of 14.8 MΩ. For modern post-1970 form-wound stators the 100 MΩ floor is usually the controlling minimum (because 100 > kV + 1 for typical machines).

Temperature Correction

Insulation resistance varies inversely on an exponential basis with temperature — higher temperature means lower IR. IEEE 43-2013 corrects readings to 40°C using Rc = KT × RT. The simple field approximation for thermoplastic insulation is the '2× per 10°C' rule: IR halves for every 10°C rise, doubles for every 10°C drop. Modern thermosetting (epoxy/polyester) insulation is less sensitive and uses a different formula.

Always record winding temperature at every test — without it, the reading is useless for trending. PI and DAR do not need temperature correction (both readings are at the same temperature, so KT cancels). Do not apply the thermoplastic correction to modern epoxy windings — it over-corrects. Moisture invalidates the correction below the dew point; dry the winding (space heaters, blanked-off motor with low-current heat) before taking acceptance readings.

Insulation Coordination: BIL and Clearances

Insulation coordination (IEEE C62.82) matches equipment insulation strength to the protective level of surge arresters so switching and lightning surges are clipped below the equipment's withstand. Key terms:

• BIL (Basic Insulation Level): the standard lightning-impulse withstand voltage, in kV crest of a 1.2/50 μs wave. A 15 kV class switchgear has a typical BIL of 95 or 110 kV.
• CFO (Critical Flashover): the impulse voltage at which 50% of impulses cause flashover.
• Surge arrester protective level: the residual voltage across the arrester when it conducts the expected surge current. The margin between BIL and arrester protective level is the coordination margin (typically ≥20%).
• Clearances: minimum air distances between live parts and ground, set by BIL and voltage class. NETA acceptance tests verify clearances during visual inspection.

Poor coordination lets surges reach the insulation, causing puncture or flashover — exactly what IR, PI, hipot, and VLF tests try to detect after the fact.

Exam Trap

Do not apply a generic '100 MΩ minimum' to every machine. IEEE 43-2013 sets three different minimums depending on insulation age and winding type. A pre-1970 winding or any field winding uses kV + 1 MΩ. A modern form-wound stator uses 100 MΩ. A random-wound stator or sub-1 kV coil uses 5 MΩ. A 13.8 kV machine under the kV + 1 rule only needs 14.8 MΩ — well below 100. If the exam names the machine age or winding type, pick the right row of the table.