MCCB in Data Center Power Distribution

How are MCCBs applied in data center power distribution? In a data center running 2N or N+1 topology, an MCCB sits at the main LV switchboard, the UPS input and output boards, and the PDU branch panels, and every one of those positions has a different fault signature and a different consequence if it trips the wrong breaker. A feeder fault that clears at the incomer instead of the branch takes down redundancy for every rack downstream on that path, which defeats the reason the second path exists. This article covers A/B feed placement and separation, prospective fault current near step-down transformers and UPS output, selectivity so a branch fault never lifts the incomer, trip settings across a 2N topology, and how Ekip, Micrologic, and ETU trip units expose branch metering to DCIM and BMS platforms.

Where MCCBs Sit in the A/B Power Chain

A typical 2N data center feed runs utility (or generator) into a step-down transformer, into a main LV switchboard, into the UPS input breaker, through the UPS, out to a UPS output board, then to a PDU, then to the rack. MCCBs cover the middle of that chain — UPS input/output, PDU incomers, and PDU branch circuits in the 100 A to 1600 A band. Air circuit breakers (ACBs) tend to take the main utility incomer position above roughly 1600 A or where Icw short-time withstand across a bus-tie matters; MCCBs generally have limited or no Icw rating, so they are not the right choice for a bus-tie breaker expected to ride through a fault for several cycles.

Every one of those MCCB positions belongs to either the A path or the B path, never both. Two power supplies per server draw from separate PDUs, separate branch breakers, and ideally separate busway runs back to separate UPS modules. The MCCB itself does not create this separation — the electrical design does — but the breaker's rating and settings have to match the fault duty of the specific point in the chain it protects, and that duty differs by position, not just by current rating.

Prospective Fault Current Near Transformers and UPS Output

Fault current close to a large step-down transformer is high and sustained. For a 1500 kVA transformer at 6% impedance and a 400 V secondary, three-phase prospective fault current works out to roughly 36 kA. Paralleled transformers on a common bus push that higher. This is the number the main LV board and the UPS input breaker have to be rated for, with margin for future transformer upsizing — see Icu vs Ics ratings for MCCBs for how that margin should size against ultimate versus service breaking capacity.

UPS output is a different animal. An inverter typically limits its fault contribution to roughly 1.5-3x rated current for a handful of cycles before current-limiting electronics pull it back, so a fault on the UPS output board sees far less energy than the same fault would see fed straight from the transformer. The complication: when the UPS drops to static bypass, the output board is suddenly fed by mains-quality fault current through the bypass path, closer to the transformer number. A breaker sized only for inverter-limited fault current will be undersized the moment bypass engages.

Selectivity: Keeping a Feeder Fault Off the Incomer

Selectivity (discrimination) between a PDU branch breaker and its upstream incomer is the single most consequence-heavy setting decision in a 2N board. If a branch MCCB and the incomer above it have overlapping instantaneous (Ii) settings, both can see the fault current and both can attempt to open — sometimes the incomer trips first or simultaneously, taking out every rack fed from that PDU rather than just the faulted branch.

What we see in the field: engineers often read LSIG time-current curves off a datasheet and assume selectivity holds because the curves "look separated" on a log-log plot. Manufacturer discrimination tables (ABB Tmax XT selectivity tables, Schneider's discrimination charts, Siemens 3VA coordination tables) are built from actual tested current levels, not curve overlays, and they account for the instantaneous region where curves alone can be misleading. Full selectivity up to the full fault-current rating is achievable between properly graded MCCB pairs; partial selectivity, valid only up to a stated current threshold, is common between an MCCB feeder and an ACB incomer and is often accepted where the fault probability above that threshold is judged low — this depends on the specific board's fault-current profile and risk tolerance, and reasonable engineers land on different answers here.

Practically, this means the incomer's short-time delay (Isd) setting needs a grading margin above the branch breaker's clearing time — commonly a step of roughly 100-150 ms between adjacent levels — so the branch has time to clear before the incomer's short-time protection would otherwise operate. Adding a third level (main incomer, sub-distribution, PDU branch) means adding a third grading step, which starts to matter for arc-flash energy at the main board if the incomer's delay stacks too high.

Trip Unit Settings Across a 2N Topology

Ir (long-time pickup), Isd (short-time pickup and delay), and Ii (instantaneous) each do a different job in a 2N board. Ir protects the cable and busway from sustained overload and should sit close to the actual connected load, not the breaker frame rating, so nuisance tripping on legitimate load growth doesn't force an unplanned failover to the B path (see how to set MCCB trip settings for the Ir/Isd/Ii procedure). Isd handles the grading described above. Ii is a last-resort, no-delay cutoff for close-in faults and is generally set high enough that it does not interfere with the graded Isd coordination below it.

On electronic trip units, these settings are configurable in the field rather than fixed by a thermal element, which is exactly why electronic trips (ABB Ekip, Schneider Micrologic, Siemens ETU) dominate at the UPS input/output and main board level in data centers — a commissioning team can re-grade the whole selectivity chain from a keypad or software tool as the load profile changes, without swapping breakers.

Electronic Trips as the Branch Metering Layer for DCIM

Past protection, the electronic trip unit is also the metering point most DCIM and BMS platforms rely on at the branch level, because running separate metering CTs on every PDU circuit is more hardware than most designs want. ABB's Ekip Touch and Hi-Touch units carry integrated current/voltage/power metering, an event log with time-stamped waveform capture, and Modbus RTU or TCP output. Schneider's Micrologic 5.3E/6.3E adds energy metering (kWh, kVAR, power factor) on top of LSIG protection and speaks Modbus into an EcoStruxure or third-party gateway. Siemens' ETU on the 3VA2 frame — from ETU350 up through the graphic ETU850 — exposes similar metering with a COM module for Modbus, PROFIBUS, or PROFINET (a fuller breakdown is in Ekip vs Micrologic vs ETU trip units compared).

A gateway polls these registers and passes current, voltage, power, and breaker status into the DCIM or BMS, where it feeds load trending, PUE calculation, and rack-level capacity planning. None of this replaces protection settings — it rides alongside them on the same trip unit.

Testing and Maintenance Without Losing Redundancy

Primary and secondary injection testing on a data center MCCB is normally scheduled offline, with load shifted to the redundant path first. Testing a live branch breaker under load is generally avoided; an unplanned trip mid-test on the path still carrying IT load defeats the entire purpose of running 2N in the first place. This is also why many operators schedule a planned A-to-B failover ahead of any maintenance on the A path — it confirms the B path actually carries the full load before anyone opens a breaker on A.

Not always straightforward, though. Some facilities run active-active rather than active-passive across A/B, where both paths carry partial load continuously; a failover test there has to confirm the surviving path can absorb the full load, not just that it's energized.

Conclusion

The core discipline in data center MCCB selection is matching every breaker's rating and settings to its exact position in the A/B chain — transformer-fed fault duty at the main board, bypass-fed fault duty at the UPS output, graded selectivity down through every level, and an electronic trip unit at each point worth metering. Get the breaking capacity, the grading, or the trip settings wrong at any one position and the 2N design stops behaving like 2N the moment it's actually tested by a fault. For frame selection and breaking-capacity classes across brands, see the MCCB engineering guide, and browse current stock in molded case circuit breakers.