Ground-Fault Protection in MV Networks: Solid, Resistance, Isolated and Resonant Grounding Compared

The grounding method determines the magnitude, phase angle and duration of an MV earth fault—and therefore which protection can be selective. A pickup value that works on a solidly grounded bus can be blind or unstable on an isolated or Petersen-coil network. The correct design starts with the zero-sequence network, not a copied ANSI 50N/51N setting.

This guide compares solid, low-resistance, high-resistance, isolated and resonant grounding; maps each method to practical sensing and protection; and covers neutral equipment, transient overvoltage, coordination, commissioning and common failure modes. Numerical ranges are illustrative—system studies, applicable rules and the actual relay/transformer/resistor/reactor data govern the project.

Executive comparison

Grounding method	Single-line-to-earth fault behaviour	Primary protection tendency	Main engineering trade-off
Solid/effectively grounded	High current, usually dominated by source/transformer zero-sequence impedance	50N/51N, residual phase OC, differential/REF	Simple, fast detection versus high damage/arc energy and voltage depression
Low-resistance grounded (LRG)	Resistor limits current to a deliberately substantial level	50N/51N, neutral OC, feeder CBCT, directional where parallel paths exist	Selective clearing versus NGR thermal duty and planned outage on first fault
High-resistance grounded (HRG)	Low current, selected in relation to total system charging current	59N/neutral displacement, sensitive 51N/67N, first-fault alarm/location	Continuity on first fault versus fault-location and second-fault risk
Isolated/ungrounded	Current is chiefly capacitive; neutral shifts	59N plus sensitive directional/admittance/transient earth fault	Possible first-fault operation versus restrike/transient-overvoltage and selectivity difficulty
Resonant grounded	Petersen coil compensates network capacitive current; small residual remains	Wattmetric/admittance/transient/intermittent 67N, 59N	Arc self-extinction/continuity versus tuning, measurement and algorithm complexity

1. What system grounding actually changes

For a phase-to-earth fault represented by symmetrical components, a useful study expression is:

I_f = 3V₁ / (Z₁ + Z₂ + Z₀ + 3Z_f).

The grounding method strongly changes Z₀; fault resistance Z_f, cable capacitance, source topology and transformer winding connections can change the result just as materially. The protection study must retain magnitude and angle of 3I₀ and 3U₀, because directional earth-fault algorithms discriminate from their phasor or transient relationship.

Safety: touch/step potential and equipotential bonding are earthing-system questions; limiting neutral current does not remove the need for protective earthing.
Equipment stress: grounding changes arc damage, conductor/transformer thermal duty and temporary/transient overvoltage.
Continuity: some low-current systems may remain energised after a first earth fault, but only under an approved operating philosophy with prompt alarm/location.
Selectivity: current magnitude may identify the faulted feeder in solid/LRG systems; isolated/resonant systems often require direction, active power, admittance or transients.

2. Required study data

normal, contingency, bus-coupled, generator, islanded and maintenance topologies;
transformer winding connections, neutral availability and zero-sequence impedances;
source positive-, negative- and zero-sequence impedances and maximum/minimum fault level;
phase-to-earth capacitance of every cable, line, motor, transformer and surge device by switching state;
NGR resistance, voltage class, continuous/short-time current and thermal time; or coil inductance, control range and loss resistance;
fault resistance range and arcing/restriking fault behaviour;
CT, neutral CT, CBCT and VT/broken-delta ratios, errors, polarity, burden and transient range;
relay earth-fault algorithm, sign convention, sensitivity, frequency tracking, filtering and settings groups;
insulation level, surge protection, touch/step study, equipment-bonding and local regulatory constraints.

3. Solid or effectively grounded networks

A low-impedance neutral connection provides a strong zero-sequence source. Earth-fault current can approach or exceed three-phase-fault current depending on transformer/source impedances. Healthy phase-to-earth voltages are comparatively restrained during a fault, but the faulted equipment sees high electromagnetic, thermal and arc energy until clearing.

Protection: phase-CT residual 50N/51N can be adequate; CBCT gives better sensitivity. Neutral CT, transformer REF (64REF/87N) and differential protection provide source/zone-specific high speed.
Coordination: grade feeder earth overcurrent with downstream devices and upstream neutral/transformer protection using total-clearing tolerance bands.
Direction: 67N may be required with multiple grounded sources, parallel transformers, rings or embedded generation.
CT duty: high asymmetric through current can saturate unequal phase CTs and create residual spill; prove security dynamically.
Application caution: never infer touch safety from rapid overcurrent operation alone; complete the earthing grid and bonding study.

4. Low-resistance grounding

An NGR deliberately limits neutral current while retaining a level that conventional selective protection can measure. When resistor impedance dominates, the planning approximation is I_NGR ≈ V_LN/R_N. The complete study must include transformer/source zero-sequence impedance, NGR tolerance and fault resistance.

Select NGR current from equipment damage, protection sensitivity/selectivity, charging current, arc-flash and operational criteria—not a generic “standard” ampere value.
Rate voltage for the neutral duty and specify time rating consistent with protection plus breaker-failure backup. A short-time NGR must not be exposed by disabled trips or repeated reclosing.
If grounding is through a zig-zag or wye-delta earthing transformer, include its zero-sequence impedance, thermal/dynamic duty, protection and loss-of-ground-source alarm.
Use feeder CBCTs for sensitive/selective 51N. A source-neutral CT sees the sum and normally provides backup, not feeder identity.
For two parallel grounding sources, model current division and switching. Uncontrolled parallel NGRs can change both fault current and grading.

5. High-resistance grounding

HRG restricts a first earth fault to a low value. Its design is tied to total phase-to-earth capacitance: resistor current must be selected with adequate margin relative to system capacitive charging behaviour so the neutral is controlled and damaging transient overvoltage is not encouraged. This is a study result, not a universal “10 A” rule.

Detection: neutral displacement 59N, neutral-current measurement and sensitive feeder 67N/CBCT methods are typical.
Operation: a first-fault alarm may permit controlled continuity only where codes, process risk and equipment insulation allow it. Define maximum permitted duration, responsibility and fault-location method.
Location: pulsing resistor systems, portable detectors, feeder residual trending or selective relays may be used; prove the method at minimum network capacitance and high fault resistance.
Second fault: an earth fault on another phase can become a phase-to-phase fault through the earth network. Alarm response and isolation must prevent indefinite first-fault operation.
Monitoring: supervise NGR continuity, neutral connection, auxiliary power, temperature and sensing circuit. A broken resistor/neutral can silently turn HRG into isolated operation.

6. Isolated (ungrounded) networks

“Ungrounded” means no intentional neutral-to-earth connection; the network is still coupled to earth through distributed capacitance and insulation leakage. On a single earth fault, the neutral shifts and the healthy phase-to-earth voltages can rise approximately from phase voltage toward line voltage. Equipment insulation and surge arresters must be selected for the actual temporary-overvoltage duty.

If C₀ denotes total phase-to-earth capacitance per phase, a planning approximation for uncompensated charging current is I_C ≈ 3ωC₀V_phase. Confirm the project definition of capacitance and use a frequency-dependent network model where harmonics/transients matter.

Magnitude-only feeder overcurrent may not distinguish the faulted feeder because healthy feeders supply capacitive current toward the fault.
59N detects neutral displacement but normally does not identify the feeder.
Directional 67N, admittance/conductance, transient and intermittent-earth-fault algorithms exploit phase/energy differences.
Arcing faults can extinguish and restrike, producing severe transient overvoltage and irregular residual quantities.
Topology changes alter total capacitance and protection sensitivity. Recalculate after major cable additions or sectionalisation changes.

7. Resonant grounding and the Petersen coil

An arc-suppression coil supplies inductive neutral current to counter the network capacitive earth-fault current. Using consistent per-phase definitions, I_L = V_phase/(ωL) and approximate resonance occurs when I_L matches I_C, giving L ≈ 1/(3ω²C₀). Actual residual current includes coil/network losses, detuning, harmonics, asymmetry and fault resistance.

Benefit: low residual current promotes extinction of transient earth arcs, reduces damage and can support service continuity.
Cost: the small active residual component is hard to measure selectively; basic 51N is often insufficient.
Tuning: cable/feeder switching changes capacitance. Automatic tuning must have range, stable neutral-displacement measurement, alarms and defined behaviour on controller/drive failure.
Protection: wattmetric/conductance, admittance, transient and dedicated intermittent-earth-fault functions are common; settings depend on relay convention and measured network damping/detuning.
Overvoltage: evaluate neutral displacement, resonance/detuning and ferroresonance interactions with VTs; do not assume compensation removes insulation stress.

8. Measurement architecture

Signal	Implementation	Critical checks
3I₀	Residual of three phase CTs, CBCT or neutral CT	Polarity, ratio, phase error, saturation, cable screen/earth routing and standing residual
3U₀	Three VTs residual calculation or open/broken-delta secondary	Primary connection, fuse failure, ferroresonance, polarity, ratio and residual-voltage rating
Directional quantity	3U₀/3I₀, residual power/admittance or negative-sequence quantities	Forward convention, characteristic sector, minimum polarising threshold and topology
Neutral equipment	NGR/coil current, temperature, tap/position, continuity	Independent supervision and fail alarm; do not depend solely on the same measurement chain

For a CBCT, all phase conductors—and the neutral if it belongs to the protected circuit—must pass through the aperture consistently. Cable screens/earth conductors must be routed so normal screen current cancels as intended; an external return through the window can mask or create residual current. Record the physical routing on the as-built drawing.

9. Protection selection matrix

Function	Best fit	Limitation
50N/51N magnitude	Solid/LRG; source backup	Limited selectivity/sensitivity in isolated or resonant networks
59N neutral displacement	HRG/isolated/resonant fault presence	Usually cannot locate the feeder
67N directional OC	Multiple ground sources/ring; some isolated systems	Needs valid polarisation and correct angle/convention
Wattmetric/conductance	Resonant grounded networks	Very small active quantity; CT/VT phase error and coil losses matter
Admittance/multicriteria	Isolated/resonant, variable topology	Requires validated network model and vendor-specific settings
Transient/intermittent EF	Restriking/high-resistance faults	Must be tested with realistic waveforms/noise/switching events
REF/differential	Transformer/generator restricted zone	Not feeder-wide; CT matching/stability and zone boundaries govern

10. Worked planning example: resistance value is not the final answer

Illustrative only. A nominal 11 kV network has V_LN = 11/√3 = 6.35 kV. A planning objective of 400 A resistive earth current gives R_N ≈ 6,350/400 = 15.9 Ω. That arithmetic does not specify the NGR.

Recalculate actual fault current with transformer/source zero-sequence impedance, resistor tolerance and minimum/maximum voltage.
Confirm NGR voltage insulation, temperature coefficient, short-time current/duration and energy I²Rt across protection and BF clearing.
Calculate total charging current; confirm the selected resistive component provides the intended grounding behaviour.
Check minimum fault with arc resistance remains above feeder pickup plus errors.
Verify maximum fault does not exceed NGR/earthing-transformer/switchgear duty.
Coordinate feeder trip, source backup and NGR thermal alarm/trip, including stuck breaker and failed DC.

11. Commissioning and periodic test plan

Verify transformer vector group, neutral links, earthing transformer/NGR/coil nameplates and one-line against the study.
Measure NGR resistance or coil inductance/tap/drive over the applicable range; correct for temperature and instrument method.
Prove neutral-conductor continuity, independent supervision, temperature sensors and alarms.
Test CT/CBCT/VT ratios, polarity and secondary earthing; inspect cable-screen paths through CBCTs.
Secondary-inject 50N/51N/59N pickup, dropout, time and directional boundary with actual settings groups.
Use dynamic waveforms for forward/reverse, high-resistance, restriking/intermittent, detuned and CT-saturation cases.
Switch representative network sections and demonstrate coil tuning or settings-group/adaptive response.
Primary-inject/prove the complete sensor-to-trip path where practical, including breaker failure and source backup.
On live load, record healthy 3I₀, 3U₀, phasors, coil position and alarms as a baseline.
Archive study revision, settings/checksum, test files, COMTRADE and as-built neutral/screen routing.

12. Failure modes that must be designed explicitly

open NGR/neutral connection converts the network to isolated operation;
shorted NGR or bypass link raises earth-fault current and equipment duty;
earthing-transformer outage removes the only zero-sequence source;
parallel sources create unplanned neutral-current sharing;
Petersen-coil control/drive/tap failure leaves dangerous detuning;
VT fuse failure corrupts 3U₀ and direction;
CT mismatch/saturation creates false residual during phase faults;
CBCT screen routing creates cancellation or standing current;
network expansion increases charging current beyond NGR/coil/protection assumptions;
first-fault alarm is ignored until a second earth fault produces severe damage.

13. Design-release checklist

Grounding objective, permitted first-fault duration and trip/alarm philosophy approved?
All zero-sequence sources and switching states modelled?
Minimum/maximum fault, fault resistance and total charging current calculated?
NGR/coil/earthing transformer electrical, insulation and thermal duty demonstrated?
Healthy phase-to-earth temporary/transient overvoltage and arrester/VT duty checked?
Sensor sensitivity, accuracy, polarity, saturation and routing verified?
Protection algorithm matches grounding physics and is selective in every topology?
Directional sign/sector and VT-failure behaviour documented?
Breaker/BF clearing time is inside neutral-equipment rating?
Failure alarms, operating response and maintenance tests defined?
Dynamic and complete-path commissioning evidence accepted?
Expansion/change triggers for capacitance and grounding-study review established?

References and further reading

Engineering note: “Grounding” of the power-system neutral and “earthing/bonding” of exposed conductive parts are related but different designs. Both must satisfy the applicable standards, utility rules and local law.