In short. Redundancy is a statement about capacity after a failure. N+1 means that with the largest unit out, the rest still carry the load. 2N means two independent systems each carry the whole load. Calculate the capacity left after the failure, not the number of spare units.
The definitions in capacity terms
| Architecture | Rule |
|---|---|
| N | Capacity of all units ≥ load |
| N+1 | Capacity with the largest unit out ≥ load |
| N+2 | Capacity with the two largest units out ≥ load |
| 2N | Each of two independent systems ≥ load |
| 2N+1 | Two systems, each with N+1 internally |
Counting with equal units
N+1 = N + 1 N+2 = N + 2 2N = 2 × N
Worked example (illustrative)
Critical load 2,400 kW, unit capacity 800 kW.
- N = ⌈2,400 ÷ 800⌉ = 3 units, 2,400 kW installed.
- N+1 = 4 units, 3,200 kW installed; with one out, 2,400 kW.
- N+2 = 5 units, 4,000 kW installed; with two out, 2,400 kW.
- 2N = 6 units in two groups of 3, 4,800 kW installed.
Now change the unit size to 1,200 kW. N = 2, N+1 = 3 (3,600 kW), 2N = 4 (4,800 kW). Unit size changes the cost of redundancy, especially for N+1: smaller units give less excess capacity.
Unequal units
With different sizes, remove the largest unit and check what remains. For example, units of 1,000, 1,000 and 600 kW for a 1,500 kW load: remove a 1,000 kW unit, 1,600 kW remains, so N+1 holds. Remove the largest in each check, since that is the worst case.
Check the usable capacity
Use capacity after derating and any loading limit, not the nameplate. See derating. Also check the largest motor start and load steps with one unit out. See load-step screening.
Where counting units fails
- Spare units share a fuel line, controller or breaker, so a single failure removes several.
- Load growth erodes the spare.
- The remaining units cannot take the step or the starting current.
- Distribution is not redundant even if generation is.
Capacity rules by level
| Level | Sets installed | Capacity available with one failed |
|---|---|---|
| N | N | (N−1) × P, below load |
| N+1 | N+1 | N × P, equal to load |
| N+2 | N+2 | (N+1) × P, with margin |
| 2N | 2 × N | N × P in the surviving system |
Worked example (illustrative)
Load 900 kW, unit rating 500 kW at site derating. N = 2. N+1 installs 3 units (1,500 kW), with one failure leaving 1,000 kW against 900 kW load (90 percent loading). N+2 installs 4 units; with two failed, 1,000 kW remains. 2N installs two independent systems each of two 500 kW units (4 in all), each system able to carry the load on its own only if loaded to 900 of 1,000 kW, so a failure in one system leaves the other at 90 percent with no internal spare.
Notes
- The loading margin on the surviving units after failure should be at least 10 percent.
- 2N with internal N+1 in each system provides the highest tier but costs the most.
- The count of sets alone does not give resilience; shared fuel, switchgear or controls undermine it.
Redundancy needs independent failure paths. See N to 2N architecture and redundancy in critical power.
Common mistakes
- Describing a design as N+1 without checking capacity after failure.
- Ignoring derating, so the actual spare is smaller than expected.
- Forgetting the UPS, transformers and cables are also part of the chain.
What to send us
Send the critical load, unit sizes under consideration and the redundancy target. Send your application details. See also redundancy architecture selection.
Questions engineers ask
Is N+1 always one extra generator?
With equal units, yes. With unequal units, test by removing the largest and checking the capacity left.
Why is 2N so expensive?
It doubles the capacity and infrastructure, since each system must carry the whole load.
Does N+1 cover maintenance?
It covers one unit being out, whether failed or in maintenance, but not a failure during maintenance.
Should the UPS and transformers be redundant too?
For true resilience, yes, because the weakest element sets the system.
Figures are illustrative. Redundancy decisions need a full system review.