PDUs, RPPs and transformers.
This is where the electrical room walks out onto the data hall floor and starts working alongside people who have never heard of an approach boundary. The equipment is ordinary. The population around it is not.
The panelboard rows, and the exits that matter at a PDU
Under the table method, panelboards and equipment rated 240 V and below are Category 1 with a 19 in. arc flash boundary; above 240 V through 600 V, Category 2 at 3 ft — both rows carrying parameters of 25 kA available fault current and 0.03 s clearing. A PDU is a transformer bolted to distribution: on its 480 V primary side, sitting close to stiff bus, the 25 kA parameter is not a formality. Where the parameters are exceeded, the table method is off the table and 130.5(G)'s incident energy analysis is mandatory — which is why the PDU fleet is usually the strongest case for extending the arc flash study past the main rooms.
Opening a hinged cover that exposes bare energized parts is a Yes on the likelihood table, full stop. The RPP's tidy dead-front does not change what is behind it.
Breaker operations in front of an audience
Operating a breaker on healthy equipment under a normal operating condition is a No for likelihood — that row is what lets facilities techs work RPP schedules all day without an arc flash suit. But the row is conditional on all seven conditions of the normal operating condition, and two of them do the real gatekeeping on a data hall floor: condition of maintenance, and no evidence of impending failure. A breaker that just tripped is not a normal operation candidate — the reset after a trip is the abnormal-condition row, a Yes, until the cause is known. The difference between those two rows is the difference between a routine keystroke and an energized-work decision, and it turns entirely on why you are touching the handle.
Control circuits at 125 V and below get their own No row — provided no other equipment above 125 V is exposed in the interaction. An open PDU cabinet next to the control section voids the relief.
Transformers: the inrush moment
Energizing the PDU transformer — first time after installation, or first time after any maintenance — is exactly the likelihood table's Yes row for first operation after service. Commissioning sequences that energize dozens of PDUs in a week should treat each first energization as its own event with the boundary cleared, not as a checklist line. The 2027 emphasis on condition of maintenance in 130.5(B) points the same direction: the estimate of likelihood is required to consider the operating time and condition of the overcurrent device protecting the circuit, and a never-yet-operated breaker has no track record at all.
RPP branch breakers operated hundreds of times a year by IT staff under a facilities SOP that has never once referenced the seven conditions — meaning nobody is empowered to notice the one morning the panel smells hot and the operation is no longer normal. The SOP needs a single added line: any evidence of impending failure stops the operation and calls electrical.
About to reset a tripped breaker? That is the abnormal-operation row. Walk it — the answer is different from the morning rounds.
This is the material of the 8- and 16-hour NFPA 70E classes — taught on the 2027 edition, against data center topology, about 55 times a year. Private onsite days run against your one-lines.