Home / Technical Articles / Ground faults in ungrounded systems (risks & detection)

Where are ungrounded systems applied

Ungrounded systems are power systems with no intentionally applied grounding. However, they are grounded by the natural capacitance of the system to ground. Thus, the fault current level is very low, such that equipment damage is minimal.

Catching the ground faults in ungrounded systems
Catching the ground faults in ungrounded systems

It is not necessarily essential that the faulted area be rapidly isolated. This is an advantage and therefore, it is sometimes used in industrial plant systems where a high continuity of service is important to minimize interruptions of expensive production processes.

However, ungrounded systems are subject to high and destructive transient overvoltages and, consequently, are always potential hazards to equipment and personnel.

Thus, they are generally not recommended, even though they are normally used.


  1. Faults in ungrounded systems
  2. Transient overvoltages (as a consequence)
  3. Grounded-detection methods for ungrounded systems
    1. Three-voltage transformers
    2. Single-voltage transformers
  4. Ungrounded System – easily explained (VIDEO)

1. Faults in ungrounded systems

Phase-to-ground faults on an ungrounded system essentially shift the normal balanced voltage triangle, as shown in Figure 1. The small currents flowing through the series phase impedances will cause a very slight distortion of the voltage triangle, but practically, it is as shown in Figure 1b.

Voltage shift for a phase-a-to ground fault on an ungrounded system: (a) normal balanced system; (b) phase a solidly grounded
Figure 1 – Voltage shift for a phase-a-to ground fault on an ungrounded system: (a) normal balanced system; (b) phase a solidly grounded

A typical circuit is illustrated in Figure 2 showing the current flow.

The sequence networks are shown in Figure 3. The distributed capacitive reactance values X1C, X2C, and X0C are very large, whereas the series reactance (or impedance) values X1S, XT, X1L, X0L and so on, are relatively very small. Thus, practically, X1C is shorted out by X1S and XT in the positive-sequence network, and similarly for the negative-sequence network.

Because these series impedances are very low, X1 and X2 approach zero, in relation to the large value of X0C.

Sequence networks and interconnections for a phase-a-to-ground fault on an ungrounded system
Figure 3 – Sequence networks and interconnections for a phase-a-to-ground fault on an ungrounded system


I1 = I2 = I0 = Vs/X0c (equation 1)


Ia = 3I0 = 3Vs/X0c (equation 2)

This calculation can be made in per unit (pu) or amperes (A), remembering that VS and all the reactances (impedances) are line-to-neutral quantities.

The unfaulted phase b and c currents will be zero when determined from the sequence currents of Equation 1. This is correct for the fault itself.

However, throughout the system the distributed capacitance X1C and X2C is actually paralleled with the series reactances X1S, XT, and so on, so that in the system I1 and I2 are not quite equal to I0. Thus, Ib and Ic exist and are small, but they are necessary as the return paths for Ia fault current.

This is shown in Figure 2.

Phase-to-ground fault on an ungrounded system
Figure 2 – Phase-to-ground fault on an ungrounded system

If Ia = −1 pu, then Ib = 0.577 ∠+30° and Ic = 0.577 ∠−30°.

In industrial applications where ungrounded systems might be used, the X0C is equal practically to X1C = X2C and is equivalent to the charging capacitance of the transformers, cables, motors, surge-suppression capacitors, local generators, and so on, in the ungrounded circuit area.

Various reference sources provide tables and curves for typical charging capacitances per phase of the power system components. In an existing system the total capacitance can be determined by dividing the measured phase-charging current into the line-to-neutral voltage.

Note that as faults occur in different parts of the ungrounded system, X0C does not change significantly. Because the series impedances are quite small in comparison, the fault currents are the same practically and independent of the fault location. This makes it impractical for selective location of faults on these systems by the protective relays.

When a phase-to-ground fault positively occurs, the unfaulted phase-to-ground voltages are increased particularly by √3 (see Figure 1b). Thus, these systems require line-to-line voltage insulation.

In the normal-balanced system (see Figure 1a) Van = Vag, Vbn = Vbg, and Vcn = Vcg. When a ground fault occurs, the phase-to-neutral voltages and the phase-to-ground voltages are quite different.

The neutral n or N is defined as ‘‘the point that has the same potential as the point of junction of a group (three for three-phase systems) of equal nonreactive resistances if connected at their free ends to the appropriate main terminals (phases of the power system)’’ (IEEE 100). This is the n shown in Figure 1b.

From this figure, the voltage drop around the right-hand triangle is:
Vbg − Vbn − Vng = 0 (equation 3)

and around the left triangle: Vcg − Vcn − Vng = 0  (equation 4)

In addition: Vng + Van = 0  (equation 5)

From the basic equations,

Vag + Vbg + Vcg = 3V0 (equation 6)
Van + Vbn + Vcn = 0  (equation 7)

Subtracting Equation 7 from Equation 6, substituting Equation 3 through Equation 5, and with Vag = 0:

Vag − Van + Vbg − Vbn + Vcg − Vcn = 3V0,
Vng + Vng + Vng = 3V0,
Vng = V0 (equation 8)

Thus the neutral shift is zero-sequence voltage. In the balanced system of Figure 1a, n = g, V0 is zero, and there is no neutral shift.

Go back to contents ↑

2. Transient overvoltages as a consequence

Restriking arcs after current interruption in the breaker or in the fault can result in large destructive overvoltages in ungrounded systems. This phenomenon is illustrated in Figure 4 below.

In the capacitive system the current leads the voltage by nearly 90°. When the current is interrupted or the arc extinguished at or near its zero value, the voltage will be at or near its maximum value. With the breaker open, this voltage remains on the capacitor to decay at a time constant of the capacitive system. In the source system, it continues as shown for VS.

Thus, in a half cycle, the voltage across the open contact is almost twice the normal peak value.

If a restrike occurs (switch closed in Figure 4), the basic +1 pu voltage of the capacitive system will shift to the system voltage of −1 pu, but because of the system inductance and inertia, it will overshoot to a maximum possibility of −3 pu.

Transient overvoltage on an ungrounded system
Figure 4 – Transient overvoltage on an ungrounded system

If the arc goes out again near current zero (switch open) but restrikes (switch closed) again, the system voltage will try to shift to +1 pu, but yet another time overshoot, this time to a potential maximum of +5 pu.

This could continue to −7 pu, but, meanwhile, the system insulation would no doubt break down, causing a major fault. Thus, ungrounded systems should be used with caution, and applied at the lower voltages (<13.8 kV), where the system insulation levels are higher.

If this system is used, prompt attention is important to locate and correct the ground fault. As the fault current is very low, it is easy to ignore the fault and continue operation.

However, with the fault the other phases are operating at basically 1.73 times the normal line-to-ground voltage. If an insulation deterioration caused the first ground fault, the higher voltages might accelerate the breakdown of the unfaulted phases, to result in a double line-to-ground or three-phase fault.

Then, high fault currents would result, requiring fast shutdown and instant loss of production.

In actual practice totally ungrounded systems do not exist. As soon as a fault detector is applied using one or three voltage transformers, the system is grounded through the high impedance of these devices. The resistance of the relays and associated ballast resistors help in limiting the transient overvoltages, so that very few cases of overvoltage essentially exist.

Go back to contents ↑

3. Grounded-detection methods for ungrounded systems

Voltage provides the best indication of a ground fault because the current is very low and, basically, does not change with the fault location. The two methods used are shown in Figure 5 and Figure 6.

These indicate that a ground fault exists but not where it is in the primary system.

Go back to contents ↑

3.1 Three-voltage transformers

Wye-grounded–broken-delta voltage transformer connections are preferred (see Figure 5).

Ballast resistors are used to reduce the shift of the neutral from either unbalanced excitation paths of the voltage transformers or from ferroresonance between the inductive reactance of the voltage transformers and relays and the capacitive system.

Voltage ground-fault detection using three voltage transformers connected wye-grounded–broken-delta
Figure 5 – Voltage ground-fault detection using three voltage transformers connected wye-grounded–broken-delta

The voltage for the relay in Figure 5 from Figure 1b is:

  • Vpq = 3V0 = Vag + Vbg + Vcg
  • Vpq = (√3VLN cos30°) × 2 = 3VLN (equation 9)

Thus, the voltage available to the relay for a phase-to-ground fault on the ungrounded system is three times the line-to-neutral normal voltage.

Usually, the VT ratio of primary VLN: 69.3 V is used so that the maximum solid ground relay voltage would be 3 × 69.3 = 208 V. Because the relay will be used to send alarm, its continuous voltage rating should be greater than, or equal to this value. Otherwise, an auxiliary step-down voltage transformer must be used.

Figure 5 is simplified. Usually, the voltage transformer will be wye-grounded–wye-grounded and an auxiliary wye-grounded–broken-delta transformer will be used.

Sometimes the main voltage transformer will have a double secondary, one of which can be connected to the broken delta. Lamps can be connected across each broken-delta secondary winding to provide visual indications.

Typical resistance values across the secondary winding, derived from experience, are shown in Table 1.

Table 1 – Typical resistance values across secondary winding

Resistor R
OhmsWatts at 208V

Go back to contents ↑

3.2 Single-voltage transformers

The single-voltage transformer of Figure 6 is especially subject to possible ferroresonance, without adequate resistance in the secondary.

Voltage ground detection using single-voltage transformer
Figure 6 – Voltage ground detection using single-voltage transformer

Without this resistance Vbg is calculated (equation 10):

Single-voltage transformer formulae

If the distributed system capacitance XC divided by the transformer exciting reactance Xe equals 3, then theoretically, Vbg is infinite. Saturation of the voltage transformer would prevent this, but it is quite possible for the voltage triangle abc to have its ground point well outside this triangle.

This is called ‘‘neutral inversion’’, as illustrated in Figure 7.

Phasor diagram illustrating neutral inversion with unloaded voltage transformer connected to phase b
Figure 7 – Phasor diagram illustrating neutral inversion with unloaded voltage transformer connected to phase b as shown in Figure 6. Example with Xc = −j3
and Xe = j2. All values in per unit.

In this case, the ratio of X/ Xe is 1.5, therefore, in Equation 10 above, Vbg = 2.0 pu as shown in Figure 7. For simplicity, resistance neither in the system nor across the voltage transformer secondary isassumed.

Sustained phase-to-ground voltages almost four times higher have been experienced. Also, the interaction of the variable transformer-exciting impedance with system capacitance can produce ferroresonance, with very high and distorted waveforms. This application of the single VT is not recommended, but if used the secondary system should be loaded with resistance.

This ground detection scheme must be used with caution to avoid ‘‘neutral’’ inversion and ferroresonance, as outlined above. The voltage relay is set to have its contacts held open for the normal line-to-ground secondary voltage.

When a ground fault occurs on phase b, the voltage collapses and the voltage relay resets to close the undervoltage contacts. If a phase a or c ground fault occurs, the relay voltage increases by about 1.73 V to cause the relay to operate on overvoltage.

Either undervoltage or overvoltage operation usually sets off an alarm to alert the operators of a ground fault so that they can arrange an orderly or convenient shutdown.

Go back to contents ↑

4. Ungrounded System – easily explained (VIDEO)

This video-series consists of three clips – one explaining the advantages of an IT-system, the second describes what happens in an IT-system in the event of an insulation fault and the last one explains how to locate an insulation fault.

Part 1 – The advantages of an ungrounded system

Part 2 – What happens in the event of an insulation fault?

Part 3 – How do you find ground faults in the ungrounded systems?

Go back to contents ↑

Source // Protective Relaying Principles and Applications by J. Lewis Blackburn and Thomas J. Domin (Purchase hardcover from Amazon)

Premium Membership

Premium membership gives you an access to specialized technical articles and extra premium content (electrical guides and software).
Get Premium Now ⚡

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 fascilities. Professional in AutoCAD programming. Present on

Leave a Comment

Tell us what you're thinking... we care about your opinion!

Premium Membership

Get access to premium electrical guides, technical articles and much more!