In short. N is the capacity needed. N+1 adds one spare unit, N+2 adds two, 2N duplicates the whole system and distributed redundancy shares spares across several systems. Each protects against a different set of failures. Choose by the failure you must survive and the cost you can accept, and count redundancy against the load, not the number of cabinets.
The family
| Architecture | What it survives | Weak spot |
|---|---|---|
| N | Nothing; capacity only | Any failure or maintenance outage |
| N+1 | One unit failure or one maintenance outage | Failure during maintenance; shared paths |
| N+2 | Failure during maintenance | Shared paths; cost |
| 2N | Loss of one entire system | Cost; each side loaded under 50 per cent |
| 2N+1 | Loss of a system plus a unit | Cost and complexity |
| Distributed redundant | Loss of one block, using shared spare | Complex control and transfer |
Counting capacity correctly
For generators: if the load is 1,800 kVA and each unit is 750 kVA, N is three units (2,250 kVA). N+1 is four units. With units of 1,000 kVA, N is two (2,000 kVA) and N+1 is three. Unit size matters: N+1 with two large units costs less capacity than with four small ones, but a larger share of the capacity is lost with each failure.
Redundancy exists only if the remaining units can carry the full load after the failure. Check derating, the largest motor start and load steps with the unit out of service. See N, N+1, N+2 and 2N capacity guide.
The 2N meaning
2N usually means two independent systems, each able to carry the whole load, with separate sources, distribution and controls. Dual-corded loads use both. The result survives the complete loss of one system, including maintenance, but costs about twice the capacity.
Distributed redundancy
For example, “3 to make 2”: three systems each carry two-thirds of their rating, so the loss of one transfers its load to the other two. It uses less spare capacity than 2N but needs more complex transfer and control. See fault-tolerant architecture.
Redundancy of what?
State redundancy per element, not for the site as a whole.
- Sources: generators, utility feeds, UPS modules.
- Distribution: transformers, switchboards, cables, transfer switches.
- Support: fuel, cooling, controls, DC supplies.
An N+1 generator plant behind an N distribution is N overall. The weakest element sets the system.
Choosing
- Define the failure to survive, including maintenance and common-cause events.
- Calculate the cost of the outage and the cost of each option.
- Check that independence is real, not only on paper.
- Choose the least complex architecture that meets the requirement.
Worked illustration
A facility needs 1,200 kW. A 2N power system would use two 1,200 kW plants, 2,400 kW installed. N+1 with three 600 kW generators gives 1,800 kW installed and survives one failure with the load carried by the other two. It does not survive a shared fuel failure, a bus fault or a failure during maintenance of another unit. The extra cost of 2N buys protection against those cases. The decision is about those cases, not the arithmetic.
The redundancy ladder
| Level | Installed | Tolerates | Weakness |
|---|---|---|---|
| N | Need only | No failure | Any failure is an outage |
| N+1 | Need plus one | One failure | Exposed during maintenance of another unit |
| N+2 | Need plus two | Two failures, or one while another is in maintenance | Cost |
| 2N | Two complete systems | Total failure of one system | Cost, space, complexity |
| Distributed redundant | Loads rotated across several systems, each with spare capacity (such as 3 to make 2) | One system failure | Design and control complexity |
Distributed redundancy illustration
Three UPS systems each rated 1 unit, load 2 units shared across them as two-thirds of capacity each. If one fails, the other two each carry 1 unit and remain at 100 percent. The installed capacity is 3 for a load of 2 (50 percent overhead), where 2N would install 4 (100 percent overhead).
Independence matters more than count
Redundancy needs independent failure paths. A shared input switchboard, shared battery room, shared fuel line or shared software update can fail all units together.
Choosing
- Cost of outage versus cost of added units.
- Maintenance plan: concurrent maintainability demands N+1 at least.
- Growth path and space.
See concurrent maintainability and N to 2N architecture.
Common mistakes
- Counting spare units but not the spare in distribution.
- Calling a design N+1 when load growth has eaten the spare.
- Assuming 2N without separating common elements.
What to send us
Send the load and the failures you must survive. Send your application details. See also redundancy architecture selection.
Questions engineers ask
Is 2N always better than N+1?
It survives more failure types but costs more. Choose by the risk you need to cover.
What does N mean exactly?
The number of units, or the amount of capacity, needed to carry the load with no spare.
How do I keep N+1 valid over time?
Track the load against the capacity with one unit out, and act before growth removes the spare.
Can I mix architectures?
Yes. Different load classes can use different levels, such as 2N for critical and N+1 for the rest.
This guide gives general principles. Architecture selection needs your actual loads and risk requirements.