DC Control, Breaker Failure (50BF) & 86 Lockouts

A relay doesn't open a breaker by itself. It closes a contact, or switches a solid-state output, that connects the substation's 125 V DC battery across the breaker's trip coil. The coil releases a stored-energy spring or pneumatic latch, the main contacts part, and the arc is extinguished in vacuum or SF₆. Any break in that chain (battery, fuse, wiring, coil, mechanism, interrupter) means the fault is still there after the relay has done its job correctly.

The drawing that shows this chain is the DC elementary (schematic). It is read like a ladder: positive rail on the left, negative rail on the right, and each horizontal rung a series circuit that is energized only when every contact in it is closed.

Breakers are reliable, but a substation has many of them and they are called on to operate rarely. Common causes of failure to interrupt:

• Mechanical: a jammed or seized operating mechanism, a broken linkage, a discharged spring.
• Electrical: an open trip coil, broken trip wiring, a blown DC fuse, or a depleted battery after an unnoticed charger failure.
• Interrupter: low SF₆ density, a lost vacuum, or welded contacts. The mechanism moves but the arc does not go out.

Remote backup (zone 2/3 distance and time-overcurrent elements at the far ends) will eventually clear the fault, but slowly and by tripping much more of the system than necessary. Local breaker-failure protection clears it faster and more selectively. It trips only the breakers that bound the failed one, plus the remote end of any line through DTT. Because total clearing time feeds directly into cable shield thermal duty, ground grid step and touch voltages, and stability (the critical clearing time), BF timing is a design quantity, not a default.

Every protective trip of the breaker also asserts BFI. While BFI is present and the supervision says the breaker is still closed, the 62BF timer runs. Most relays then hold (seal in) the initiate while the supervision is still picked up, so a short trip pulse can't reset the timer early; the lab models this. If the breaker opens normally, the supervision drops out and the timer resets. That is why the budget below is built from the supervision's dropout, not from the trip resetting. If the timer expires, the scheme concludes the breaker has failed and energizes the 86BF lockout.

If the timer is supervised by the 52a instead of (or as well as) the current detector, replace the detector reset with the time the auxiliary switch takes to change state after the main contacts part. That comes from the breaker's timing test and is usually longer than a modern detector reset. The lab uses a representative 20 ms (about 1 cycle). If retrip to TC2 is enabled, add its output delay (2 ms in the lab), because a TC2 clear opens that much later than a TC1 clear.

Modern numerical relays reset the current detector in well under a cycle using open-pole and subsidence-current detection, so most of the budget is the breaker itself. Older electromechanical detectors could take 1–2 cycles to drop out, especially with DC offset or CT subsidence current, and settings from that era often carry extra margin. IEEE C37.119 is the guide that covers these choices.

Worked example

The plant's 230 kV HV (gen-tie) breaker has a 3-cycle (50 ms) interrupting time. The relay trips in 1 cycle (17 ms), the 50BF detector resets in 10 ms, the 86 operates in 10 ms, and you want a 2-cycle (33 ms) margin. The bounding breakers (the 34.5 kV main and feeders on the plant side, and the utility breaker at the far end of the gen-tie) are also 3-cycle, and the DTT channel to the utility end takes 8 ms. All values are representative. Find the 62BF setting and the total BF clearing time at the remote end.

Generation collector substations break a hidden assumption of transmission BF logic, namely that a failed breaker carries fault current. That assumption still holds for a fault on a collector feeder: the feeder breaker carries the grid contribution through the MPT, typically thousands of amps, and only the inverters' own contribution is small (typically about 1.0–1.5 pu of rating, set by the OEM current limit, after a sub-cycle transient that can be higher; take the actual value from the OEM short-circuit model). The trips that genuinely happen with almost no current are these:

• 63 sudden-pressure and other mechanical transformer trips, which are not faults the breaker current can reveal.
• Low-level internal faults caught by sensitive 87T, REF (87N) or turn-to-turn protection.
• V/Hz (24), frequency (81), voltage (27/59) and anti-islanding trips of the main or HV breaker.
• Transfer-trip receive from the utility or from another station.
• Any trip issued at night or with the plant offline, when the breaker carries only charging or magnetizing current.

The usual answer is a second supervision path: 52a-supervised breaker failure for low-current trips, combined with the current detector in an OR. The 52a path brings its own risks, so treat it carefully:

• Initiate the 52a path only from the trip functions that need it. Don't initiate BF from elements that stay asserted while the unit is offline (a standing 27 on a dead bus), unless the logic also proves the breaker was closed when the trip was issued.
• Use 52a and 52b where possible, and treat disagreement as an alarm, not as a breaker-closed indication.
• Time the 52a path from the auxiliary-switch timing, which is slower than the main contacts, not from the current detector reset.
• A 52a/52b discrepancy check can only work once the 52b has changed state, so it relies on the 62BF delay being longer than the auxiliary-switch timing. In the lab, a 62BF of 70 ms or less (the interrupting time plus the auxiliary-switch lag) still false-trips the offline case even with the check enabled; the boundary moves with the interrupting-time slider.

When 62BF expires, the scheme normally issues an immediate retrip (if it hasn't already) and energizes a hand-reset 86BF lockout relay. In the classic architecture, one 86 operation does four things:

1. Trips every bounding breaker. For a failed 34.5 kV feeder breaker that is the rest of the collector bus: the other feeders plus the main or HV breaker. For the failed HV breaker modelled in the lab it is the 34.5 kV main on the plant side (many schemes also trip the feeders, which is more than strictly needed when a main exists) plus the utility breaker via DTT. The lockouts are often named 86BFM1 or similar on drawings.
2. Keys direct transfer trip to the remote terminal of any line or gen-tie whose local breaker failed (over fiber, Mirrored Bits or GOOSE; see Module 4.2).
3. Blocks closing through its b-contacts in every affected close circuit.
4. Stays latched until a person investigates and resets it by hand. Nobody should re-energize into a breaker that just failed to interrupt.

Many modern schemes split these jobs instead: the relay trips the bounding breakers directly from its own outputs (or over GOOSE), and the 86 is kept for latching and close blocking. Both architectures are common, so read the elementary to see which one you have.

Breaker failure is the last line of defence for a trip that didn't work. The DC design should make that rare in the first place:

• Trip-circuit monitoring (TCM): a small supervision current through each trip coil, alarmed when it stops, so an open coil or broken wire is found before it is needed.
• DC ground detection: a first ground on an ungrounded 125 V battery doesn't by itself cause a misoperation, but it removes the margin against the next one. A second ground can energize a trip coil or defeat a trip, so the first must be alarmed and found.
• Separation: feed TC1 and TC2, and redundant relay systems, from separately fused DC circuits (or separate batteries). For BES elements, the TPL-001 P5 single-point-of-failure studies look at exactly these shared components.
• Pole disagreement on independent-pole HV breakers, which trips all three poles if they end up in different positions.
• Generator-breaker flashover on a CTG: an open breaker can flash over just before synchronizing or just after the unit is tripped, when up to about 2 pu appears across its contacts, and C37.102 describes dedicated protection for it.