Premium Membership ♕

Save 50% on all EEP Academy courses with Enterprise Membership Plan and study specialized LV/MV/HV technical articles & guides.

Home / Technical Articles / Why Ground Fault Protection Matters and Which Scheme For Sensing Ground Faults To Choose

Ground Fault Occurrence

A ground fault normally occurs in one of two ways: by accidental contact of an energized conductor with normally grounded metal, or as a result of an insulation failure of an energized conductor. When an insulation failure occurs, the energized conductor contacts normally non-current-carrying metal, which is bonded to a part of the equipment-grounding conductor.

Why Ground Fault Protection Matters and Which Scheme For Sensing Ground Faults To Choose
Why Ground Fault Protection Matters and Which Scheme For Sensing Ground Faults To Choose

In a solidly grounded system, the fault current returns to the source primarily along the equipment-grounding conductors, with a small part using parallel paths such as building steel or piping.

If the ground return impedance were as low as that of the circuit conductors, ground fault currents would be high, and the normal phase-overcurrent protection would clear them with little damage.

Unfortunately, the impedance of the ground return path is usually higher. The fault itself is usually arcing; and the impedance of the arc further reduces the fault current.

In a 480Y/277-V system, the voltage drop across the arc can be from 70 to 140 V. The resulting ground fault current is rarely enough to cause the phase overcurrent protection device to open instantaneously and prevent damage.

Sometimes, the ground fault is below the trip setting of the protective device and it does not trip at all until the fault escalates and extensive damage is done.

For these reasons, low-level ground protection devices with minimum time-delay settings are required to rapidly clear ground faults.

This is emphasized by the NEC requirement that a ground fault relay on a service shall have a maximum delay of 1 s for faults of 3000 A or more.

The NEC (Article 230.95) requires that ground fault protection, set at no more than 1200 A, be provided for each service-disconnecting means rated 1000 A or more on solidly grounded wye services of more than 150 V to ground, but not exceeding 600 V phase-to-phase.

Practically, this makes ground fault protection mandatory on 480Y/277-V services, but not on 208Y/120-V services. On a 208-V system, the voltage to ground is 120 V. If a ground fault occurs, the arc will extinguish at current zero, and the voltage to ground is often too low to cause it to restrike.

Therefore, arcing ground faults on 208-V systems tend to be self-extinguishing. On a 480-V system, with 277 V to ground, restrike usually takes place after current zero, and the arc tends to be self-sustaining, causing severe and increasing damage, until the fault is cleared by a protective device.

The NEC requires ground fault protection only on the service disconnecting means. This protection works so fast that for ground faults on feeders, or even branch circuits, it will often open the service disconnect before the feeder or branch overcurrent device can operate.

This is highly undesirable, and in the NEC (Article 230.95) a fine-print note (FPN) states that additional ground fault–protective equipment will be needed on feeders and branch circuits where maximum continuity of electric service is necessary. Unless it is acceptable to disconnect the entire service on a ground fault almost anywhere in the system, such additional stages of ground fault protection must be provided.

At least two stages of ground fault protection are mandatory in health care facilities (NEC Article 517.17).

IMPORTANT! – Overcurrent protection is designed to protect conductors and equipment against currents that exceed their ampacity or rating under prescribed time values. An overcurrent can result from an overload, short circuit, or high-level ground fault condition.

When currents flow outside the normal current path to ground, supplementary ground fault protection equipment will be required to sense low-level ground fault currents and initiate the protection required.

Normal phase-overcurrent protection devices provide no protection against low-level ground faults.

Sensing ground faults

There are three basic means of sensing ground faults:

  1. Ground Return Method
  2. Zero Sequence Sensing Method
  3. Residual Sensing Method

Method #1 – Ground Return Method

The most simple and direct method is the ground return method as illustrated in Figure 1. This sensing method is based on the fact that all currents supplied by a transformer must return to that transformer.

When an energized conductor faults to grounded metal, the fault current returns along the ground return path to the neutral of the source transformer. This path includes the grounding electrode conductor – sometimes called the ground strap as shown in Figure 1.

A current sensor on this conductor (which can be a conventional bar-type or window-type CT) will respond to ground fault currents only. Normal neutral currents resulting from unbalanced loads will return along the neutral conductor and will not be detected by the ground return sensor.

This is an inexpensive method of sensing ground faults in which only minimum protection per NEC Article 230.95 is desired.

For it to operate properly, the neutral must be grounded in only one location (as indicated in Figure 1).

Ground return sensing method
Figure 1 – Ground return sensing method

In many installations, the servicing utility grounds the neutral at the transformer, and additional grounding is required in the service equipment. In such cases and others, including multiple source with multiple interconnected neutral ground points, residual or zero-sequence sensing methods should be employed.

Go back to contents ↑

Method #2 – Zero Sequence Sensing Method

A second method of detecting ground faults is the use of a zero sequence sensing method as illustrated in Figure 2. This sensing method requires a single, specially designed sensor, either of a toroidal or rectangular-shaped configuration.

This core balance current transformer surrounds all the phase and neutral conductors in a typical three-phase, four-wire distribution system.

The sensing method is based on the fact that the vectorial sum of the phase and neutral currents in any distribution circuit will equal zero unless a ground fault condition exists downstream from the sensor.

All currents that flow only in the circuit conductors, including balanced or unbalanced phase-to-phase and phase-to-neutral normal or fault currents, and harmonic currents, will result in zero sensor output.

However, should any conductor become grounded, the fault current will return along the ground path – not the normal circuit conductors, and the sensor will have an unbalanced magnetic flux condition, and a sensor output will be generated to actuate the ground fault relay.

Zero-sequence sensing method
Figure 2 – Zero-sequence sensing method

Zero-sequence sensors are available with various window openings for circuits with small or large conductors, and even with large rectangular windows to fit over bus bars or multiple large-size conductors in parallel. Some sensors have split cores for installations over existing conductors without disturbing the connections.

This method of sensing ground faults can be employed on the main disconnect where minimum protection per NEC Article 230.95 is desired. It can also be employed in multitier systems where additional ground fault protection is desired for added service continuity.

Additional grounding points may be employed upstream of the sensor, but not on the load side.

Ground fault protection employing ground return or zero-sequence sensing methods can be accomplished by the use of separate ground fault relays (GFRs) and disconnects equipped with standard shunt trip devices or by circuit breakers with integral ground fault protection with external connections arranged for these modes of sensing.

Go back to contents ↑

Method #3 – Residual Sensing Method

The third basic method of detecting ground faults involves the use of multiple current sensors connected in a residual sensing method, as illustrated in Figure 3. This is a very common sensing method used with circuit breakers equipped with electronic trip units and integral ground fault protection.

The three-phase sensors are required for normal phase-overcurrent protection. Ground fault sensing is obtained with the addition of an identically rated sensor mounted on the neutral.

In a residual sensing scheme, the relationship of the polarity markings – as noted by the X on each sensor – is critical. Because the vectorial sum of the currents in all of the conductors will total zero under normal, non-ground-faulted conditions, it is imperative that proper polarity connections are employed to reflect this condition.

As with the zero-sequence sensing method, the resultant residual sensor output to the ground fault relay or integral ground fault tripping circuit will be zero if all currents flow only in the circuit conductors.

Residual sensing method
Figure 3 – Residual sensing method

Should a ground fault occur, the current from the faulted conductor will return along the ground path, rather than on the other circuit conductors, and the residual sum of the sensor outputs will not be zero. When the level of ground fault current exceeds the preset current and time-delay settings, a ground fault tripping action will be initiated.

This method of sensing ground faults can be economically applied on main-service disconnects, in which circuit breakers with integral ground fault protection are provided. It can be used in minimum-protection schemes per NEC Article 230.95 or in multitier schemes, in which additional levels of ground fault protection are desired for added service continuity.

Additional grounding points may be employed upstream of the residual sensors, but not on the load side. Both the zero-sequence and residual sensing methods have been commonly referred to as vectorial summation methods.

Most distribution systems can use any of the three sensing methods exclusively, or a combination of the sensing methods depending upon the complexity of the system and the degree of service continuity and selective coordination desired.

Different methods will be required depending upon the number of supply sources and the number and location of system-grounding points.

As an example, one of the more frequently used systems in which continuity of service to critical loads is a factor is the dual-source system illustrated in Figure 4. This system uses tie-point grounding.The use of this grounding method is limited to services that are dual-fed (doubleended) in a common enclosure or grouped together in separate enclosures and employing a secondary tie.

This system uses individual sensors connected in ground-return fashion. Under tie breaker–closed operating conditions, either the M1 sensor or M2 sensor could see neutral unbalance current and possibly initiate an improper tripping operation.

However, with the polarity arrangements of these two sensors, along with the tie breaker auxiliary switch (T/a) and the interconnections as shown, this possibility is eliminated. Selective ground fault tripping coordination between the tie breaker and the two main circuit breakers is achieved by preset current pickup and time-delay settings between devices GFR/1, GFR/2, and GFR/T.

The advantages of increased service continuity offered by this system can only be effectively used if additional levels of ground fault protection are added on each downstream feeder. Some users prefer individual grounding of the transformer neutrals.

In such cases, a partial differential ground fault scheme should be used for the mains and the tie breaker.

Dual-source system - single-point grounding
Figure 4 – Dual-source system – single-point grounding (click to expand schematics)

An infinite number of ground fault protection schemes can be developed depending upon the number of alternate sources, the number of grounding points, and system interconnections involved.

Depending upon the individual system configuration, either mode of sensing or a combination of all types may be employed to accomplish the desired end results.

Because the NEC Article 230.95 limits the maximum setting of the ground fault protection used on service equipment to 1200 A (or 3000 A for 1 s), to prevent tripping of the main-service disconnect on a feeder ground fault, ground fault protection must be provided on all the feeders.

To maintain maximum service continuity, more than two levels (zones) of ground fault protection will be required, so that ground fault outages can be localized and service interruption minimized. To retain selectivity between different levels of ground fault relays, time-delay settings should be employed with the GFR furthest downstream having the minimum time delay.

This will allow the GFR nearest the fault to operate first.

With several levels of protection, this will reduce the level of protection for faults within the GFR zones. Zone interlocking was developed for GFRs to overcome this problem.

Zone selective interlocking example
In ZSI example, the fault has occurred downstream of the feeder panelboard breaker located in Zone 3. Each of the circuit breakers with the blue downward arrow next to them establish that a fault has occurred. The feeder panelboard circuit breaker directly upstream of the fault sends an output blocking signal to the main switchboard feeder circuit breaker that feeds the panelboard confirming that the fault is actually downstream. A blocking signal is also sent to the main circuit breaker in the main switchboard. Since the feeder panelboard breaker does not receive a blocking signal from a lower zone breaker, this circuit breaker will interrupt the fault in 0.1 seconds, eliminating the need for any of the upstream circuit breakers to trip. If this breaker does not clear the fault, then the upstream breakers will trip according to their specific time delays (standard coordination without ZSI).

Ground fault relays (or circuit breakers with integral ground fault protection) with zone interlocking are coordinated in a system to operate in a time-delayed mode for ground faults occurring most remote from the source.

However, this time-delayed mode is only actuated when the GFR next upstream from the fault sends a restraining signal to the upstream GFRs. The absence of a restraining signal from a downstream GFR is an indication that any occurring ground fault is within the zone of the GFR next upstream from the fault and that device will operate instantaneously to clear the fault with minimum damage and maximum service continuity.

This operating mode permits all GFRs to operate instantaneously for a fault within their zone and to still provide complete selectivity between zones.

The National Electrical Manufacturers’ Association (NEMA) states, in their application guide for ground fault protection, that zone interlocking is necessary to minimize damage from ground faults.

A two-wire connection is required to carry the restraining signal from the GFRs in one zone to the GFRs in the next zone.

Circuit breakers with integral ground fault protection and standard circuit breakers with shunt trips activated by the ground fault relay are ideal for ground fault protection. Many fused switches over 1200 A, and some fusible switches in ratings from 400 to 1200 A, are listed by UL as suitable for ground fault protection. Fusible switches so listed must be equipped with a shunt trip and be able to open safely on faults up to 12 times their rating.

Power distribution systems differ widely from each other, depending on the requirements of each user, and total system overcurrent protection, including ground fault currents, must be individually designed to meet these needs.

Experienced and knowledgeable engineers must consider the power sources (utility and on-site), the effects of outages and downtime, safety for people and equipment, initial and life-cycle costs, and many other factors.

They must apply protective devices, analyzing the time-current characteristics, fault-interrupting capacity, and selectivity and coordination methods to provide the safest and most cost-effective distribution system.

Go back to contents ↑

Reference // Electrical Engineer’s Portable Handbook by Robert B. Hickey, P.E.

Premium Membership

Get access to premium HV/MV/LV technical articles, electrical engineering guides, research studies and much more! It helps you to shape up your technical skills in your everyday life as an electrical engineer.
More Information

Edvard Csanyi

Electrical engineer, programmer and founder of EEP. Highly specialized for design of LV/MV switchgears and LV high power busbar trunking (<6300A) in power substations, commercial buildings and industry facilities. Professional in AutoCAD programming.


  1. Brian M
    Jul 15, 2021

    Nice article

  2. Gary Olson
    Apr 06, 2018

    Edvard, nicely written article. I work primarily with emergency power systems and struggle regularly with ground fault sensing and protection issues on generator sets. Since it is highly unlikely that a ground fault will result in damage to a generator set, I believe ground fault protection should be applied to generator set feeders rather than the output of a generator, which is often done. (NEC is usually understood to require this.) This would minimize the potential of tripping on critical circuits by isolating the power interruption only to the affected circuits.

    Do you have any information that would seem to contradict this position (assuming generator set has required overload protection)?


    • Mario A.Meneses
      May 17, 2020

      if you transformer have a capacity of 100 KVA, for each phase there are 33.3 KVA if your unit is of 1000 KVA, there are 333 KVA
      I hope this information helps you.

    Mar 20, 2018

    Nice article! Thanks

  4. bernardo raquel
    Mar 20, 2018

    I would like to know in a three phase transformer with capacity of 100kva. How much kva for every line?
    Thank you.

Leave a Reply to RODOLPHO

Tell us what you're thinking. We care about your opinion! Please keep in mind that comments are moderated and rel="nofollow" is in use. So, please do not use a spammy keyword or a domain as your name, or it will be deleted. Let's have a professional and meaningful conversation instead. Thanks for dropping by!

fifty  ⁄    =  fifty

Learn How to Design Power Systems

Learn to design LV/MV/HV power systems through professional video courses. Lifetime access. Enjoy learning!

Subscribe to Weekly Newsletter

Subscribe to our Weekly Digest newsletter and receive free updates on new technical articles, video courses and guides (PDF).
EEP Academy Courses - A hand crafted cutting-edge electrical engineering knowledge