N+1 and N+2: How Much Generator Redundancy Is Enough?

Every data-centre specification eventually contains the phrase "N+1". It is used to mean at least three different things in practice, and the difference shows up on the worst day of the year. Here is what the notation really describes, and how it translates into generator capacity.

What N means

N is the capacity needed to carry the design load. If the IT load plus mechanical cooling and losses comes to 2.4 MW, then N is 2.4 MW of generation. Everything else is arithmetic on top of that number.

N+1 adds one redundant unit of the same size as the largest installed module. Four 600 kW sets carrying 2.4 MW is N; five is N+1, because any single set can fail and the remaining four still carry the load.

N+2 adds two. It covers two simultaneous failures, or a failure during maintenance when a unit is already out of service — which is the more realistic scenario.

The part the notation hides: can the remaining sets actually take the load?

Redundancy is only real if the surviving units can share what is left. That requires a paralleling system, load-sharing governors, and a bus arrangement that can isolate a failed unit without dropping the load. It also requires that the sets were commissioned together: four machines that have never run in parallel do not become a paralleled plant on demand.

Fuel autonomy is part of the redundancy

A site with N+2 generation and eight hours of fuel is an eight-hour site. Autonomy is usually specified separately — 24, 48 or 72 hours — and it interacts with the generator count: more sets running means more litres per hour, so the tank that was adequate for N may not be for N+2 under continuous duty. Fuel polishing and a tested refuelling contract matter as much as the extra set.

Testing a redundant plant

Redundancy that is never exercised degrades into a set of machines with unknown condition. A workable regime is:

• Monthly no-load start and 30-minute run for every unit, logged automatically.
• Quarterly load test at 50–80 % on each unit, one at a time.
• Annual full-load test of the plant in its redundant configuration: trip one unit under load and verify the remainder picks up the block within the transient limits.
• Annual black-building test if the design includes a full-site transition.

What this means for the generator package

For a redundant plant, the specification of each set matters less than the specification of the system. Identical models, identical controllers, matched alternators, a defined neutral arrangement and a paralleling scheme that has been tested as a system. Cunmi supplies containerised packages for campus projects with paralleling controllers, remote monitoring and load-bank-tested units, and we prefer to see the single-line diagram before quoting — because redundancy is a system property, not a set property.

FAQ

Is N+1 the same as 2N?

No. 2N duplicates the entire capacity — two independent plants, each able to carry the full load. It is far more expensive and is normally reserved for the most critical facilities. N+1 is one extra module, not a complete second plant.

Can I add a redundant set later?

Mechanically yes, but the switchgear, busbar and control system must have been designed for it. Retrofitting paralleling into a plant built for single-set operation is often more expensive than doing it at the start.

Does N+1 apply to the fuel system too?

It should. A single fuel line, a single transfer pump or a single day tank is a single point of failure that undoes the generator redundancy.

See our data-centre power packages