High-voltage engineering and insulation

In short

High-voltage engineering studies how insulation behaves under extreme electric stress, how high voltages are generated and measured in the laboratory, and how equipment is tested so that it survives lightning and switching surges in service. It decides the clearances, bushing lengths, oil quality and insulator strings used on India's 33 kV to 765 kV networks and on ±800 kV HVDC links. It is examined in university power-systems papers, carries regular GATE (EE) questions on breakdown mechanisms, impulse generators and insulation coordination, and is the core skill in test, commissioning and R&D roles at CPRI Bengaluru, ERDA Vadodara, POWERGRID, NTPC, state transcos, BHEL, Siemens, Hitachi Energy, CG Power and Apar/Polycab. The governing documents are IS 2071 / IEC 60060 (test techniques), IS 2165 / IEC 60071 (insulation coordination) and CEA Technical Standards under the Electricity Act, 2003.

Explained

Dielectrics and electric stress

• A dielectric has no free charge carriers; an applied field only polarises it (electronic, ionic, orientational/dipolar, interfacial). Permittivity ε = ε₀εr, with ε₀ = 8.854 × 10⁻¹² F/m.
• Dielectric strength is the field a material withstands before breakdown (kV/mm or kV/cm): air ≈ 30 kV/cm peak at NTP, SF₆ ≈ 89 kV/cm at 1 bar, transformer oil 120–200 kV/cm, impregnated paper 300–500 kV/cm, XLPE ≈ 400 kV/cm (design stress far lower).
• Stress is non-uniform at curved and sharp electrodes; the utilisation (Schwaiger) factor η = E_mean/E_max measures how efficiently a gap is used — uniform-field Rogowski and Bruce profiles give η ≈ 1, while sphere-plane and point-plane gaps are far lower.
• In a series (layered) dielectric the field divides inversely with permittivity, so the weakest low-ε layer (an air gap or void) carries the highest stress and fails first — the reason voids cause partial discharge.
• Stress control: capacitance grading (condenser bushings with foil layers), intersheath grading, geometric grading (stress cones in cable terminations), corona rings and grading rings on insulator strings.
• Composite insulation is normal practice: oil + pressboard in transformers, SF₆ + epoxy spacers in GIS, silicone rubber + FRP core in polymeric insulators.

Breakdown in gases

• Mechanisms begin with electron avalanche: a free electron gains energy over a mean free path and ionises by collision; α is Townsend's primary ionisation coefficient (ionisations per cm of drift), γ the secondary emission coefficient at the cathode.
• Townsend criterion: γ(e^{αd} − 1) = 1 — self-sustaining discharge. It explains breakdown in short, uniform gaps at low pressure, where secondary cathode processes matter.
• Streamer (Raether–Meek) theory: at high pd, the avalanche's own space charge distorts the field; when αd ≈ 18–20 (about 10⁸ electrons) a self-propagating streamer bridges the gap. This explains breakdown in long gaps and at atmospheric pressure, and the filamentary, cathode-independent nature of sparks.
• Paschen's law: breakdown voltage is a function of the product (pressure × gap), V_b = f(pd), not of p and d separately. The curve has a minimum (about 330 V for air at pd ≈ 0.57 torr·cm); both vacuum and high pressure improve strength.
• Corona: partial self-sustained discharge at a sharp or thin electrode where local stress exceeds the critical value while the gap as a whole does not break down. It causes power loss, audible noise, radio interference and ozone; reduced by larger conductor diameter and bundled conductors on EHV lines.
• Electronegative gases: SF₆ attaches electrons (reduces effective α), giving 2.5–3 times the strength of air at the same pressure and excellent arc quenching — the basis of GIS and SF₆ breakers. Drawbacks: liquefaction at low temperature, sensitivity to particles and surface roughness, toxic arc by-products (SOF₂, SO₂F₂) and a global warming potential of about 23,900, so recovery and handling are mandatory; C4-FN/CO₂ and clean-air alternatives are entering Indian substations.
• Vacuum breakdown is governed by electrode surface effects (field emission, clump and microparticle theories), not gas ionisation — exploited in vacuum circuit breakers up to 36 kV (and higher in newer designs).
• Surface flashover along a solid-gas interface occurs well below the gas's bulk strength; pollution plus moisture on insulators forms conducting layers, dry bands and arcing → pollution flashover.

Breakdown in liquid and solid insulation

• Pure liquid theories: electronic (avalanche in the liquid) and cavitation/bubble breakdown. Practical oils fail by suspended solid particles (aligning into a bridge), moisture globules and dissolved gas — so cleanliness dominates.
• Transformer oil is mineral oil to IS 335; breakdown voltage (BDV) is tested per IS 6792 with 2.5 mm spherical electrodes — new filtered oil typically ≥ 60 kV, with lower in-service limits by voltage class in IS 1866/IEC 60422. Moisture must be a few tens of ppm or less; a few ppm of water can halve BDV.
• Other liquids: synthetic esters and natural (vegetable) esters for fire-safe and green transformers, silicone oil for traction; askarels (PCB) are banned.
• Solid dielectric breakdown mechanisms (lowest voltage mechanism wins in service):
  1. Intrinsic — electronic, at very short times and high purity; the material's theoretical limit (MV/cm).
  2. Electromechanical — electrostatic compressive force crushes the sample.
  3. Thermal — dielectric loss heat exceeds heat removal; worst at high temperature and high frequency (loss rises with f), so AC thermal strength is below DC; it sets the practical long-term limit for many polymers.
  4. Streamer/discharge (void) breakdown — partial discharges in internal voids erode the polymer, producing electrical trees; with moisture, water trees in XLPE cables.
  5. Tracking and treeing on surfaces — carbonised conducting paths from surface discharge; measured as CTI (comparative tracking index).
  6. Electrochemical / ageing — oxidation, hydrolysis, contaminants; strength falls with time (the life–stress law Vⁿ t = constant).
• Thermal classes of solid insulation (IEC 60085 / IS 1271): Y 90, A 105, E 120, B 130, F 155, H 180 °C; the 10 °C rule — every 10 °C rise roughly halves insulation life.
• Creepage: specific creepage distance about 16 mm/kV for light pollution rising to roughly 25 mm/kV (heavy) and 31 mm/kV (very heavy) per IS/IEC 60815; coastal and industrial Indian sites use longer strings, RTV silicone coating or polymeric insulators.

Generation of high voltages and currents

• HVAC: a cascade of testing transformers (units stacked with insulated tanks, each feeding the next) gives MV-level output with modest unit insulation; series resonant sets (reactor + test object capacitance) are standard for cable and GIS testing because they draw only the loss current and fail safe on flashover.
• A Tesla coil (air-cored, damped high-frequency) is used for HF testing and demonstrations.
• HVDC: half-wave rectifier with smoothing capacitor; Cockcroft–Walton voltage multiplier (n stages → ideally 2nV_max) for MV DC at low current; Van de Graaff and electrostatic generators for very low current, very high voltage.
• Impulse voltages: single-stage RLC circuit — C1 (generator capacitance) discharges through series resistance R1 into load C2, with R2 shaping the tail. Marx (multi-stage) generator: n capacitors charged in parallel through resistors and discharged in series through spark gaps, output ≈ n × V_stage.
• Standard waveshapes: lightning impulse 1.2/50 µs (front 1.2 µs ±30%, time to half value 50 µs ±20%); switching impulse 250/2500 µs; chopped-wave and steep-front impulses for special tests; impulse current 8/20 µs (and 4/10 µs) for arresters.
• Tripping and control: triggered (trigatron) gaps, sphere-gap spacing set by charging voltage; a chopping gap truncates the wave for transformer winding tests.
• Impulse currents (kA) come from a low-inductance capacitor bank through a shaping circuit; used for arrester and earthing tests.

Measurement of high voltages and currents

• Sphere gap (IS 1876 / IEC 60052): an approved measuring device for peak AC, DC and impulse; accuracy ±3% if gap ≤ 0.5 D, surfaces clean and clearances respected. Needs air density and humidity correction.
• Electrostatic (attracted-disc) voltmeter: measures rms directly from electrostatic force, no loading, but low sensitivity.
• Resistive dividers for DC, capacitive dividers for AC, damped capacitive (mixed R-C) dividers for impulse — plus a coaxial cable and matching resistor to a digital recorder; the divider's response time and the lead inductance decide fidelity.
• Chubb–Fortescue method: peak AC voltage from the rectified capacitor current, I_avg = 2 f C V_max — simple and accurate for peak measurement.
• Capacitive voltage transformer (CVT) and electromagnetic VT/PT for service metering and protection; peak voltmeters (Rabus, digital) with a standard capacitor for calibration.
• Impulse / high currents: coaxial current shunt (low inductance), Rogowski coil (gives di/dt, integrated), Hall-effect and magneto-optic sensors; magnetic links for lightning current estimation on towers.
• Measuring high resistance / insulation resistance: megger at 500 V–5 kV, guard terminal to exclude surface leakage.

HV testing, diagnostics and standards

• Test classes: *type* tests (design proving, at CPRI/ERDA), *routine* tests (every unit), *acceptance* tests (buyer's sample), *commissioning* and *maintenance* tests (field).
• Power-frequency withstand test: 1 minute at a specified voltage; induced overvoltage test for transformer windings; DC withstand is used for cables only cautiously (space charge in XLPE makes DC unrepresentative — VLF 0.1 Hz or damped AC is now preferred).
• Impulse test on transformers: reduced full wave (typically 50–75%), full wave, chopped wave, full wave again — a change in the recorded neutral-current / voltage oscillogram between reduced and full waves indicates failure.
• Partial discharge test (IS 6209 / IEC 60270): apparent charge in pC; record inception and extinction voltage; typical acceptance for power transformers is in the range of a few hundred pC at specified voltage (limit depends on the standard and specification). Phase-resolved PD patterns identify void, surface or corona discharge; acoustic and UHF methods locate it.
• Dielectric loss (tan δ / Schering bridge) on bushings, CTs, cables and windings; rising tan δ or capacitance indicates moisture or ageing. Typical new-condenser-bushing tan δ ≈ 0.003–0.005.
• Insulation resistance, PI and DAR: polarisation index = IR at 10 min / IR at 1 min; > 2 generally acceptable for machine insulation, < 1.5 suspect (IEEE 43).
• Oil diagnostics: BDV, moisture (Karl Fischer), acidity, interfacial tension, dissolved gas analysis (sampling IS 9434 / IEC 60567; interpretation IS 10593 / IEC 60599) — C₂H₂ indicates arcing, CH₄/C₂H₄ overheating, H₂ partial discharge, CO/CO₂ paper degradation; furan analysis estimates paper ageing.
• Radio interference voltage (RIV) and corona-extinction tests on EHV hardware; artificial pollution tests on insulators (IS/IEC 60507).
• Indian framework: IS 2071 / IEC 60060 (techniques), IS 2165 / IEC 60071 (coordination), IS 15086 Part 4 / IEC 60099-4 (surge arresters; formerly IS 3070), IS 2099 / IS 12676 (bushings), IS 13118 / IEC 62271 (switchgear), CBIP manuals and CEA (Measures relating to Safety and Electric Supply) Regulations for clearances and testing under the Electricity Act, 2003.

Overvoltages, surge arresters and insulation coordination

• Overvoltage types: *temporary* (load rejection, Ferranti effect, resonance, earth fault — lasts seconds), *switching* (line energisation, current chopping, capacitor switching — hundreds of µs), *lightning* (direct strike or induced — µs), and *very fast transients* in GIS (ns).
• Lightning protection: shield (earth) wires with a protective angle, good tower footing resistance (ideally ≤ 10 Ω), counterpoise earthing, down-conductors and air terminations on buildings (IS/IEC 62305).
• Travelling waves: surges propagate at Z₀ = √(L/C) ≈ 300–400 Ω on overhead lines and 30–60 Ω on cables; at a junction, reflection coefficient ρ = (Z₂ − Z₁)/(Z₂ + Z₁) — a surge entering an open end doubles.
• Surge arrester: modern gapless metal-oxide (ZnO) arrester with an extremely non-linear I = kV^α (α ≈ 25–50) characteristic; key ratings are MCOV/COV, rated voltage, nominal discharge current (5/10/20 kA) and residual (discharge) voltage. Older types: expulsion gap, valve (SiC) with series gaps, rod gap (crude backup).
• Insulation coordination matches equipment withstand (BIL/LIWV and SIWV) to the arrester's protective level with a margin, after applying atmospheric and safety factors. Protective margin = (withstand − protective level)/protective level; IEEE/IEC practice expects ≥ 20%.
• Insulation is graded: line insulation may be allowed to flash over (self-restoring, external) while transformer internal insulation must not — so arresters are placed close to the transformer, with short, straight earth leads.
• Typical Indian standard levels (IS 2165/IEC 60071-1): 33 kV → 170 kV LIWV; 132 kV → 550/650 kV; 220 kV → 950/1050 kV; 400 kV → 1425 kV LIWV with 1050 kV SIWV.

Key points

• Dielectric strength – maximum stress before breakdown; depends on gap geometry, time, temperature, pressure and purity, not on the material alone.
• Townsend criterion – γ(e^{αd} − 1) = 1; applies to short, uniform, low-pd gaps.
• Streamer criterion – αd ≈ 18–20; explains atmospheric long-gap breakdown and makes breakdown cathode-independent.
• Paschen's law – V_b = f(pd), with a minimum near 330 V for air; vacuum and pressurised gas both insulate well.
• Corona – local discharge below full breakdown; losses, RIV, noise; cured by bigger/bundled conductors and grading rings.
• SF₆ – electron-attaching gas, 2.5–3× air at 1 bar, superb arc interrupter, but GWP ≈ 23,900 and particle-sensitive.
• Series dielectrics – stress divides inversely with εr; voids and air gaps take the highest stress and start PD.
• Oil breakdown – dominated by particles and moisture; BDV per IS 6792 at 2.5 mm gap, new oil typically ≥ 60 kV.
• Solid breakdown routes – intrinsic, electromechanical, thermal, discharge/treeing, tracking, electrochemical ageing.
• Thermal classes – Y/A/E/B/F/H at 90/105/120/130/155/180 °C; +10 °C roughly halves life.
• Marx generator – capacitors charged in parallel, discharged in series; output ≈ n × stage voltage; energy ½CV².
• Standard impulses – 1.2/50 µs lightning, 250/2500 µs switching, 8/20 µs current.
• Cockcroft–Walton – n-stage multiplier, ideal output 2nV_max; ripple and voltage drop rise steeply with n.
• Sphere gap – approved peak-voltage standard, ±3%, valid for gap ≤ 0.5D, with air-density and humidity correction.
• tan δ – ratio of loss current to charging current; loss P = V²ωC tan δ; a rising trend signals moisture or ageing.
• Partial discharge – measured in pC (IEC 60270); inception/extinction voltages and phase patterns diagnose the defect.
• BIL / insulation coordination – equipment withstand chosen above the arrester protective level with ≥ 20% margin (IS 2165/IEC 60071).
• ZnO arrester – gapless, highly non-linear I = kV^α; residual voltage at nominal discharge current is the protective level.

Formulas and examples

1. Cable/bushing stress and optimum radius. A single-core cable carries 60 kV to earth; sheath inner radius R = 4 cm, conductor radius r = 1 cm.

• E_max = V/(r ln(R/r)) = 60/(1 × ln 4) = 60/1.3863 = 43.3 kV/cm at the conductor surface.
• At the sheath: E = 60/(4 × 1.3863) = 10.8 kV/cm — the dielectric is poorly utilised.
• Optimum radius r = R/e = 4/2.7183 = 1.47 cm → E_max = V/(r × 1) = 60/1.4715 = 40.8 kV/cm, a 6% reduction for the same sheath radius; grading or a larger conductor is the practical fix.

2. Marx impulse generator design. 10 stages, each capacitor 0.5 µF charged to 100 kV; load (test object + divider) C2 = 1 nF; target 1.2/50 µs.

• Series capacitance C1 = 0.5/10 = 0.05 µF; nominal output = 10 × 100 = 1000 kV.
• Stored energy W = ½ × (10 × 0.5 µF) × (100 kV)² = ½ × 5 × 10⁻⁶ × 10¹⁰ = 25 kJ (same as ½ × 0.05 µF × (1 MV)²).
• Efficiency η = C1/(C1 + C2) = 0.05/0.051 = 0.980 → usable peak ≈ 980 kV.
• Front resistor R1 ≈ T1/(3C2) = 1.2 × 10⁻⁶/(3 × 10⁻⁹) = 400 Ω.
• Tail resistor R2 ≈ T2/[0.7(C1 + C2)] = 50 × 10⁻⁶/(0.7 × 51 × 10⁻⁹) = 1.4 kΩ.

3. Bushing tan δ and loss. A 132 kV condenser bushing has C = 500 pF, tan δ = 0.005, 50 Hz, phase voltage 132/√3 = 76.2 kV.

• P = 2πf C V² tan δ = 314.16 × 500 × 10⁻¹² × (76.2 × 10³)² × 0.005 = 314.16 × 5 × 10⁻¹⁰ × 5.806 × 10⁹ × 0.005 ≈ 4.6 W.
• If ageing raises tan δ to 0.020, loss becomes 18.2 W — four times the heat input, which accelerates thermal ageing; this trend, not the absolute value, triggers replacement.

4. Insulation coordination at 220 kV. Transformer LIWV (BIL) = 1050 kV; the station-class ZnO arrester has a residual voltage of 560 kV at 10 kA, 8/20 µs.

• Protective margin = (1050 − 560)/560 × 100 = 490/560 × 100 = 87.5% — comfortably above the ≈ 20% minimum.
• If the arrester lead adds 30 kV of inductive drop (L·di/dt), the effective protective level is 590 kV → margin = (1050 − 590)/590 = 78%; long earth leads eat the margin, so leads are kept short and straight.

5. Sphere-gap correction. A 25 cm sphere gap set at 5 cm has a standard-condition breakdown of 137 kV peak (from the IS 1876 / IEC 60052 table). Test conditions: 960 hPa, 35 °C, normal humidity.

• Relative air density d = (960/1013) × (293/308) = 0.9477 × 0.9513 = 0.901.
• Expected breakdown ≈ 0.901 × 137 = 123 kV peak; conversely a spark at 123 kV corresponds to a standardised 137 kV. Ignoring this correction gives a 10% error — enough to fail a good test object.

Where it is used

• Testing and certification labs: CPRI Bengaluru/Bhopal, ERDA Vadodara and NABL-accredited works labs run type tests on transformers, breakers, insulators and arresters; roles include HV test engineer and lab in-charge.
• Transmission utilities: POWERGRID and state transcos specify BIL, clearances, creepage and arrester ratings for 132–765 kV AIS/GIS substations and ±800 kV HVDC terminals (for example Raigarh–Pugalur), and run condition-monitoring programmes (DGA, tan δ, online PD).
• Manufacturing: BHEL, CG Power, Hitachi Energy, Siemens Energy, Transformers & Rectifiers India, Apar, Polycab and Finolex design and routine-test transformers, bushings, cables and insulators; design engineers do stress and grading calculations in FEM tools.
• Industrial and generation plants: NTPC, Tata Power, refineries and steel works carry out annual megger/PI, tan δ and oil tests on generators, motors and transformers; oil filtration and dehydration are routine maintenance.
• Cable and GIS commissioning: on-site VLF or damped-AC withstand plus PD mapping on EHV XLPE cable circuits in metro and city networks (Delhi, Mumbai, Bengaluru).
• Lightning and surge protection: telecom towers, solar parks, railway OHE and data centres use arresters, earthing design and IS/IEC 62305-based LPS.

Interview questions

• Q: Why does the Townsend mechanism fail to explain breakdown of long atmospheric gaps? A: At high pd the avalanche's own space charge distorts the field and a streamer propagates on its own, so breakdown becomes filamentary and independent of cathode secondary emission.
• Q: State Paschen's law and explain why both vacuum and high-pressure gas insulate well. A: Breakdown voltage depends on the product p·d; at very low pressure there are too few collisions to ionise, and at high pressure the mean free path is too short for electrons to gain ionising energy.
• Q: Why is SF₆ so much stronger than air? A: It is electronegative — it captures free electrons to form heavy negative ions, cutting the effective ionisation coefficient, and its large molecules shorten the mean free path.
• Q: What dominates breakdown in commercial transformer oil? A: Suspended solid particles and moisture, which align and bridge the gap; hence filtration, degassing and dehydration rather than any change in the oil chemistry.
• Q: Why does a void inside solid insulation fail first? A: The series field divides inversely with permittivity, so the low-ε void carries a much higher stress while having a far lower strength, giving partial discharge that erodes the solid into trees.
• Q: What are the standard lightning and switching impulse shapes and their tolerances? A: 1.2/50 µs (front ±30%, tail ±20%) and 250/2500 µs; peak tolerance ±3%.
• Q: In a Marx generator, how do R1 and R2 shape the wave? A: The series front resistor R1 with the load capacitance C2 sets the front time (≈ 3R1C2); the discharge resistor R2 with (C1 + C2) sets the tail (≈ 0.7R2(C1+C2)).
• Q: Why is impulse-generator efficiency less than 100%? A: The generator capacitance C1 shares charge with the load capacitance C2, so the peak is only C1/(C1 + C2) of the nominal charging voltage.
• Q: How does a gapless ZnO arrester conduct at surge and not at normal voltage? A: Its I–V curve is extremely non-linear (I ∝ V^α, α ≈ 25–50), so at MCOV it passes only microamperes but clamps a kiloamp surge at its residual voltage.
• Q: What is insulation coordination and what margin is expected? A: Selecting equipment withstand levels above the protective level of the arresters, with safety and atmospheric correction factors; a protective margin of at least about 20% is normal practice.
• Q: Why is DC withstand testing avoided on XLPE cables? A: DC injects space charge that distorts the field and can damage the insulation, and it does not reproduce the AC stress distribution; VLF (0.1 Hz) or damped AC testing with PD measurement is used instead.
• Q: Which gas in DGA indicates arcing, and why? A: Acetylene (C₂H₂) — it needs very high temperature (above about 700 °C) to form, so its presence points to an arc or high-energy discharge rather than normal overheating.

Common mistakes

• Treating dielectric strength as a fixed material constant and ignoring gap geometry, time of application, temperature and moisture.
• Applying the Townsend criterion to long atmospheric gaps, or quoting streamer theory for low-pressure short gaps.
• Reading Paschen's law as "breakdown voltage rises with pressure" and missing the minimum and the pd product.
• Forgetting air-density and humidity correction in sphere-gap measurements, then arguing about a 10% discrepancy.
• Using the nominal n × V_stage as the impulse-generator output and ignoring efficiency C1/(C1 + C2), or mixing µF with nF in R1 and R2 calculations.
• Confusing arrester rated voltage with its residual (protective) voltage, and neglecting the lead inductance drop when computing protective margin.
• Confusing clearance (through air) with creepage (along the surface), and quoting one when the standard asks for the other; also assuming the same insulation level for internal and external insulation.

Related topics

Transmission lines and cables · Overhead line design: sag, tension and insulators · Underground cables · Power system protection: relays and circuit breakers · HVDC transmission · Distribution systems and substations