In short. A 2N generator plant has two complete, independent systems, each able to carry the entire critical load alone, with separate switchgear, fuel, controls and distribution. N+N is the same idea seen from the load side: two feeds, A and B, each sized for the full demand. The value lies in independence, not doubled equipment. If the two systems share a fuel tank, a controller or a room, the plant is two halves of one system and is not 2N.
From N+1 to 2N
N+1 adds one spare unit to a shared bus and protects against one failure in a generator. It does nothing for a failure of the bus itself, the paralleling controller or the distribution behind it, and it leaves the plant unprotected while a set is out for maintenance. 2N answers all three weaknesses by duplicating the complete path from engine to load. System A and system B are each built as a full N plant with their own generators, switchgear, fuel supply, controls and feeders, and each load is either fed from both through a static transfer switch or dual power supply, or is split so that losing either system removes only the loads that can tolerate it. The N+1 power engineering page explains the single-bus approach; the N, N+1, N+2, 2N and distributed redundant page sets out the full family.
Comparing the architectures
| Architecture | Survives one generator failure | Survives failure during maintenance | Survives a bus or controller failure | Installed capacity for a 1,000 kVA load with 500 kVA sets |
|---|---|---|---|---|
| N | No | No | No | 2 sets |
| N+1 | Yes | No | No | 3 sets |
| N+2 | Yes | Yes | No | 4 sets |
| 2N | Yes | Yes, if the other system is healthy | Yes | 4 sets on two buses |
| 2(N+1) | Yes | Yes | Yes | 6 sets on two buses |
The capacity column shows why 2N is a decision about the cost of downtime rather than about engineering elegance. The same four sets arranged as N+2 on one bus protect against two generator failures but not against a bus fault; arranged as 2N on two buses they protect against a bus fault but only one generator failure per side. Which exposure matters more depends on the facility, and the answer should be written into the design basis.
What independence actually requires
Drawing two buses on a single-line diagram is the easy part. Independence has to be carried through every supporting system, and each shared item should be listed and either separated or accepted with a reason. The questions to ask about systems A and B are whether they draw fuel from different tanks with different pumps, whether their master controllers can run without each other and without any shared network, whether their switchgear is in different rooms or at least different fire compartments, whether their cables leave the generator yard by different routes, whether their cooling air and exhaust are separate, whether a single earth fault can affect both through a shared neutral, and whether one operator error such as a wrong breaker operation can take both down. A shared bus-tie breaker between A and B is common and useful for maintenance, but it must be normally open with interlocks that prevent the two systems from being paralleled by accident. The common-mode and common-cause failure page provides a checklist for this review.
Software and settings are shared infrastructure too
Two controllers of the same make running the same firmware with the same settings will respond to the same unforeseen condition in the same way. Where this is accepted, it should be a conscious decision, and both systems should at least have a manual fallback that lets an operator start a set and close a breaker without the automation.
The load side: dual-cord and single-cord equipment
2N at the generators is wasted if the load can only accept one feed. Modern servers, UPS systems and medical imaging equipment are often dual-corded and can be fed from A and B simultaneously, drawing from both and continuing on either. Single-corded equipment such as chillers, pumps and most industrial machinery needs a transfer device, usually a static transfer switch for electronic loads or an automatic transfer switch for motors, to move it between systems. The transfer device then becomes a shared element on that load and must be rated, maintained and bypassable. The dual-cord and single-cord critical loads page explains how to classify the loads, and the A and B power distribution architecture page follows the two paths through the building.
Operating and testing a 2N plant
A 2N plant is normally operated with both systems live and each carrying roughly half the load, so that the failure of either produces a load step of about 50 percent on the survivor rather than a cold start from zero. Each system must therefore be able to accept that step, which should be confirmed by a transient calculation and then by test. Commissioning should include deliberate loss of each system in turn with the load running, loss of a single set within a system, loss of the master controller of each system, and the maintenance configuration with one system entirely isolated and the other carrying the full load for several hours. These tests should be repeated annually, and the results compared with the original acceptance records. The integrated systems testing page describes how such a programme is structured.
Cost and where 2N is justified
2N roughly doubles the generator plant, switchgear, fuel storage, civil works and maintenance cost compared with N, and adds perhaps a third over N+1 once the second bus and controls are included. Each system also runs at half load most of the time, which raises diesel consumption per kWh and, on older engines, invites wet stacking. Data centres, large hospitals, payment processing sites and continuous pharmaceutical processes routinely justify this on the cost of an outage. Most manufacturing plants, commercial buildings and educational campuses do not, and are better served by N+1 with attention to the shared dependencies. We advise clients to price an hour and a day of lost production before choosing, and to remember that 2N on paper and 2N in operation are separated by the independence review above.
Common mistakes
- Building two generator buses that share one fuel tank and transfer pump, so a contaminated fuel delivery takes both systems down within the same hour.
- Running systems A and B through one master controller or one network switch, so a control failure stops both.
- Feeding single-corded loads from one system only, which leaves half the facility exposed to a single failure despite the 2N plant behind it.
- Leaving the bus-tie closed for convenience, which quietly converts the 2N plant into a single N+N bus with no fault separation.
- Testing each system alone but never testing the transfer of the full load onto one system while the other is isolated.
What to send us
Send the critical load list with each load marked as dual-corded or single-corded, the proposed or existing single-line diagram for both systems, the fuel storage and transfer layout, the location of switchgear rooms and cable routes, and the tolerable outage duration for the facility. Nameplate photographs of existing sets and UPS systems help. Use the application engineering enquiry and we will review the independence of the two paths and the sizing of each.
Questions engineers ask
What is the difference between 2N and N+N?
They describe the same architecture from different directions. 2N refers to two complete generator systems, each sized for the full load. N+N is used at the distribution level to describe two feeds, A and B, each sized for the full demand. A facility can be N+N at the UPS output while being only N+1 at the generators.
Is 2N better than N+2?
Neither is better in general. N+2 on one bus tolerates two generator failures or one failure during maintenance, but a bus or controller fault removes everything. 2N tolerates a complete system failure including its bus, but only one generator failure per side. Choose by the failure you most need to survive.
Can systems A and B share a bus-tie breaker?
Yes, and it is common for maintenance flexibility. The tie should be normally open, interlocked so that closing it requires a deliberate sequence, and included in the testing programme. A tie that is left closed makes the two systems one.
At what load should each system run in normal operation?
Usually about half the critical load each, so that losing one produces a 50 percent step on the other rather than a start from zero. Each system must be shown to accept that step within the facility’s voltage and frequency limits, and low-load running should be managed with periodic load bank runs.
Technical content reviewed by Srinivas Reddy, Managing Partner, Reddy Generators. Last updated 19 October 2026.