A reading is ambiguous when two different bank states give the same value at the relay. According to IEEE C37.99-2000, this happens in double wye, H-bridge and mid-point tap banks, when faults on opposite sides cancel each other. The alarm must therefore be very sensitive, and in some banks certain rack-to-rack faults produce no signal if the equipotential bonding is wrong.
Short answer
An ambiguous indication is one that can come from two different bank states with exactly the same reading at the unbalance relay. The classic case is the current between the neutrals of a double wye bank: it is almost zero with the bank healthy, and it is also almost zero if an equal number of units are out of service in the same phase of both wyes. IEEE C37.99-2000 (clause 8.2.1) calls for a very sensitive alarm in these schemes and for allowing for cancelling faults when setting the trip. In addition, clause 8.2.2 describes faults that certain schemes simply do not see.
What an ambiguous indication is
Unbalance protection measures a single value, such as a current or a voltage, and infers the state of the bank from it. When the bank is balanced, that value is close to zero, which is what you want. The problem is that other bank conditions can give the same zero.
The standard gives the example of an ungrounded double wye bank with a CT between the two neutrals. With the bank healthy, the current through that CT is negligible. If the same number of units or elements of the same phase are removed on both sides, the current is also negligible. The second condition is undesirable, because the remaining units carry a continuous overvoltage, and the relay reads it as normal.
Fault cancellation occurs readily in H-bridge unbalance and in double wye banks.
Where it appears
The guide identifies three common schemes with this ambiguity:
| Scheme | How the cancellation happens | Standard figure |
|---|---|---|
| Double wye | Fuse operations or shorted elements in one wye cancel the signal produced in the other | Figure 7 |
| H-bridge | Faults in one leg of the H cancel the signal of another leg | Figure 8 |
| Bank with a mid-point tap | Faults above the tap cancel the signal of those below | Figure 9 |
In all three cases the relay does not measure the cause but the net result. If two faults contribute signals of opposite sign, they subtract.
What to do about possible cancellation
The standard recommends two design and setting measures, without giving values.
- Sensitive alarm. Where ambiguity is possible, the alarm should operate for a single failure: one fuse in banks with external or internal fuses, or one element in fuseless banks. The first failure is then dealt with before another one accumulates and offsets it. The effect of a late alarm is to run for an indefinite time with permanent overvoltages on the remaining units.
- Trip that allows for offset faults. It may also be desirable to set the trip level from an estimated number of faults that cancel, so as not to expose the units to damaging voltages or force the fuses to operate above their voltage rating.
Both measures rest on a study of the bank and not on a general rule. On the relay side, the alarm setting must also take the inherent unbalance of the bank into account, because a very sensitive alarm without compensation would cause false operations.
Faults that some schemes do not detect
Clause 8.2.2 lists failure modes that, for certain configurations, produce no unbalance signal:
- Rack-to-rack fault in banks with two series groups connected phase over phase, which use neutral voltage or current for unbalance.
- Rack-to-rack fault in certain H-bridge connections.
The underlying explanation is in the guide on external arcs and rack faults: a phase-to-phase arc at mid-height does not shift the neutral. What matters here is what the standard adds for phase-over-phase configurations in double wye and H-bridge: the correct connection of the racks makes the unbalance protection respond to a flashover between racks.
The extreme case is the external arc and rack fault. How the signal of the healthy bank masks small ones is explained in inherent unbalance and measurement error.
Equipotential bonding of the racks
Figure 10 of the standard compares two ways of connecting the groups of an ungrounded double wye to the rack:
| Connection | What happens with a flashover between racks |
|---|---|
| Wrong: all groups connected on the same side | No current flows through the CT between neutrals; the relay does not see it |
| Correct: groups connected on alternating sides | Current flows through the CT between neutrals; the relay can operate |
The note to the figure states that the connection is similar in a grounded double wye bank with a phase-over-phase design. It is an installation detail that does not change the cost and does change the protection coverage. It is verified during commissioning, before energization, by checking the drawing and the rack.
The racks and the bank grounding are covered in bank grounding, neutral and grid. The function is unbalance protection (ANSI 60).
How to verify it in the field
The standard does not prescribe a procedure, but design common sense allows it to be summarized:
- Compare the installation with the drawing of the equipotential bonding of the racks.
- Confirm that the alarm is at the single-failure level and not at a coarser level.
- Check the capacitance or the phase current periodically: a cancellation shows itself because the phase currents are no longer equal even though the unbalance relay does not alarm.
What not to do
- Do not read “zero” as “healthy” in schemes with ambiguity.
- Do not delay acting on a single-fuse alarm: the second failure can offset the first.
- Do not change the side on which the groups connect to the rack without checking the effect on the protection.
- Do not adjust the compensation without first checking the phase current and the capacitance of each unit.
When to contact us
If your bank is a double wye, an H-bridge or has a mid-point tap and you are not sure which faults your protection sees, at Tiza Group we analyze it and leave it documented. We help with remote relay configuration and quote the work according to its scope. Write to us.
Related guides
- IEEE C37.99: capacitor bank protection guide
- External arc and rack faults in a capacitor bank
- Inherent unbalance of a bank: why to compensate it
- Double wye unbalance: current between neutrals
- H-bridge protection in a capacitor bank
References
- IEEE C37.99-2000, IEEE Guide for the Protection of Shunt Capacitor Banks, clauses 8.2.1 and 8.2.2, figures 7, 8, 9 and 10.
Frequently asked questions
What does it mean for an unbalance reading to be ambiguous?
The same reading can correspond to a healthy bank or to one with units out of service. For example, the current between the neutrals of a double wye is almost zero both when the bank is balanced and when the same number of units of the same phase are out in both wyes. The relay cannot tell the two cases apart.
How is the risk of cancelling faults reduced?
With an alarm sensitive enough to operate for a single fuse in fused banks, or a single element in fuseless banks. The first failure is then dealt with before a second one appears that offsets it. The standard also suggests allowing for cancelling faults when setting the trip level.
Why does it matter which side of the rack the groups are connected to in a double wye?
Because it decides whether an arc between racks flows through the CT between neutrals. With the wrong connection, all groups on the same side, an insulator flashover between racks produces no neutral current and the relay does not see it. With the alternating connection, current does flow.