4.3 Generators, Transfer & Distribution Redundancy

Standby Generators and Fuel

When a utility outage outlasts the UPS batteries, the standby generator carries the site. The CDCP sequence is: utility fails, UPS batteries ride through (5-15 min), generator starts, synchronises, and accepts load (typically 30-90 s), and the generator then carries full load until utility returns. A properly sized generator supports not just the IT load but also cooling and critical mechanical plant - a data centre that keeps servers running but loses cooling will overheat within minutes.

Fuel and sizing

Most standby generators are diesel; some large or emissions-constrained sites use gas turbines or natural-gas engines. Key exam points:

• On-site fuel storage is sized to a replenishment horizon, commonly 24-72 hours at full load, with contracted refuelling for longer events.
• NFPA 110 classifies emergency and standby power systems (for example Level 1, Type 10 means power available within 10 seconds).
• Fuel must be maintained (polishing and filtering) because diesel degrades and grows microbial contamination in long storage.

ATS, STS, and bypass

Transferring load between sources is done by transfer switches - and the exam loves the ATS versus STS distinction:

An ATS senses utility loss and transfers the load to the generator. An open-transition ATS breaks before it makes (a brief gap the UPS covers); a closed-transition ATS briefly parallels the sources to avoid the break. An STS switches between two synchronised AC sources fast enough that IT power supplies never notice - used in 2N designs so a single-corded load can be fed from either of two independent buses. A maintenance (manual) bypass lets technicians isolate and service a UPS or STS while the load runs on the utility or an alternate path, which is essential for concurrent maintainability.

Distribution devices recap

Downstream distribution uses PDUs, RPPs, and busway. Busway (bus duct) is an overhead track of busbars with plug-in tap-off boxes; it is popular in modern halls because circuits can be added or moved without rewiring conduit - flexible for changing rack densities. The PDU and RPP introduced in Section 4.1 break UPS output into the branch circuits that feed the racks.

Redundancy configurations

This is the highest-yield topic of the whole power domain. Memorise the ladder:

• N is baseline - no spare; any failure drops the load.
• N+1 adds one redundant module (for example a fourth UPS where three are needed) so one failure is absorbed, but a single shared distribution path is still a single point of failure.
• 2N runs two completely independent systems, each able to carry the full load alone, eliminating single points of failure across the whole chain - required for Tier IV fault tolerance.
• 2(N+1) goes further: each of the two systems has its own spare, tolerating one failure on each side simultaneously.

A/B feeds and dual-corded loads

Tier IV / 2N sites deliver two independent power paths - the 'A' and 'B' feeds - all the way to the cabinet. Modern servers ship with dual power supplies (dual-corded), one plugged into the A rack PDU and one into the B, so either whole path can fail or be maintained with zero impact. A critical operating discipline: each path must be able to carry the full load, so neither A nor B should be loaded beyond roughly 40-50% in normal operation. If either feed exceeds 50% and its partner fails, the survivor overloads and trips. Single-corded equipment on a 2N floor must be fed through an STS so it can draw from either feed.

Worked example: the 50% loading rule

Consider a cabinet drawing 8 kW from dual-corded servers, split evenly across the A and B feeds at 4 kW each. If the B feed fails, all 8 kW instantly shifts onto A. If the A feed's rack PDU and upstream breaker were rated for, say, 7.4 kW, the survivor now sees 8 kW and trips - taking down the cabinet even though a full redundant path existed. This is why operators enforce the balanced, sub-50% rule and why capacity-management alarms in DCIM flag any feed creeping past its safe threshold. The redundancy on the single-line diagram is only real if the operating load respects it.

Common mistake: N+1 shared-bus blind spot

A frequent exam trap is to assume N+1 equals fault tolerance. It does not: N+1 protects against the failure of one component (a UPS module, a chiller) but the units usually share a single distribution path or common bus, which remains a single point of failure. Only 2N provides two fully independent paths. Similarly, concurrent maintainability (Tier III) means you can take a path out on purpose, whereas fault tolerance (Tier IV) means the site survives a path failing by surprise - the maintenance bypass and 2N distribution are what make each promise real.

Emergency Power Off (EPO)

Finally, NFPA 70 and NFPA 75 require an Emergency Power Off (EPO) capability that instantly cuts white-space power to protect life during a fire or electrocution event. The exam's twist is that accidental EPO activation is a leading cause of outages, so good design guards the button (lift-covers, dual-action, alarms) against misoperation.