An availability target does not specify a generator redundancy level. A target such as 99.99% sets an allowable interruption budget; N, N+1 and 2N describe capacity arrangements. The missing link is the complete power system: generators, controls, fuel supply, transfer equipment and distribution paths. Adding generators can leave shared failure points intact or introduce new ones through more complex controls.
What an Availability Target Tells You
Start with the required availability percentage and the period over which it applies. The table below converts that percentage into an annual interruption budget. It does not assign a generator configuration or predict actual site performance.
For a continuously required service over a 365-day year:
[
D=\left(1-\frac{A}{100}\right)\times8{,}760
]
Here, (D) is allowable downtime in hours per year, and (A) is the availability target expressed as a percentage—for example, enter 99.99, not 0.9999.
| Availability target | Annual interruption budget | Calculation basis |
|---|---|---|
| 99% | 87.6 hours | Calculated using 8,760 hours/year |
| 99.9% | 8.76 hours | Calculated using 8,760 hours/year |
| 99.99% | 52.56 minutes | Calculated using 8,760 hours/year |
| 99.999% | 5.256 minutes | Calculated using 8,760 hours/year |
These figures are calculated budgets, not measured outage rates or values prescribed by a standard. Define the measurement point, what counts as an interruption, and whether planned maintenance is included before using them. Availability at a generator bus is not the same measurement as availability at the equipment being powered.
An annual total also hides event duration and frequency. A service that cannot tolerate a brief interruption needs that requirement stated separately; an annual percentage alone does not describe acceptable transfer time.
Generator Redundancy Levels and Equipment Counts
In the equal-size arrangements below, (N) is the minimum number of generator sets needed to support the defined load at the applicable site conditions and duty rating. The “+1” means one additional capacity unit—not a fixed percentage margin.
CoreSite’s redundancy explainer describes N, N+1, N+2 and 2N arrangements for infrastructure including generators. The table uses those capacity definitions without adopting its separate uptime claims.
| Arrangement | Equal-size generator count | Capacity retained under the stated condition | Equipment implication | Definition source and date |
|---|---|---|---|---|
| N | (N) | Full load with all required units available; no spare unit | Baseline generating capacity | CoreSite web explainer; no revision date displayed |
| N+1 | (N+1) | Full load with one unit unavailable, if the remaining units can serve the load | One additional generator and its associated connections and support equipment | CoreSite web explainer; no revision date displayed |
| N+2 | (N+2) | Full load with two units unavailable, subject to the same capacity and connection conditions | Two additional generators and their associated equipment | CoreSite web explainer; no revision date displayed |
| 2N | (2N), divided into two complete systems | Each system can support the full defined load independently | A second complete capacity system with independent distribution arrangements | CoreSite web explainer; no revision date displayed |
These are capacity descriptions, not guaranteed availability levels. N+1 does not automatically mean uninterrupted transfer, and 2N means more than placing twice as many generators on one shared bus. The ability to isolate failures and maintain the power path matters alongside installed capacity.
Equipment count is also not a price multiplier. Cummins’ Reliability Considerations in Simple Paralleling Applications, Power Topic #5590, revision 09/13, identifies foundations, space, cabling, commissioning, maintenance and switchgear as costs beyond the generator sets themselves. Its cost categories are useful planning considerations, not current project quotations.
Worked Example: A 99.99% Target and a 900 kW Load
Assume a facility requires 900 kW during utility outages. Each proposed generator has 500 kW of usable output under the specified site conditions and applicable operating rating. For this capacity-only example, assume the selected units can operate in parallel and that their kVA, load acceptance and other application limits will be checked separately.
The minimum equal-size unit count is:
[
N=\left\lceil\frac{P_{\text{load}}}{P_{\text{unit}}}\right\rceil
=\left\lceil\frac{900}{500}\right\rceil=2
]
Here, (P_{\text{load}}) is the defined load in kW, (P_{\text{unit}}) is usable output per generator in kW, and the ceiling symbol means round up to the next whole unit.
- N uses two generators, providing 1,000 kW with both available.
- N+1 uses three generators. With one unavailable, two remain and provide 1,000 kW.
- 2N uses two separate two-generator systems, four generators in total. Each system provides 1,000 kW independently.
The 99.99% target allows 52.56 minutes of interruption over the stated year. Neither equipment count proves that target will be met. A shared control failure or an unavailable transfer path could interrupt service even when sufficient generating capacity remains. Cummins specifically warns that centralized controls and their communication links can become single points of failure.
The maintenance state matters too: with one generator under maintenance, the three-unit N+1 plant has two available units. If another unit fails, only 500 kW remains—insufficient for the full 900 kW load. N+1 therefore does not provide capacity for both conditions simultaneously in this example.
This calculation establishes only a steady-state kW count. The actual model selection still needs manufacturer checks for kVA, power factor, motor starting, load steps, transient voltage and frequency response, site derating and permitted operating duty.
What to Check Before Choosing a Redundancy Level
Assess the arrangement against specific operating and failure conditions rather than assigning a configuration directly from the availability percentage:
- Can the remaining units carry the required load after the defined generator failure?
- Can a generator, breaker or control component be isolated for maintenance without interrupting the required service?
- Could shared switchgear, controls, fuel delivery or another common support system disable multiple units?
- Can the load tolerate the actual starting and transfer sequence?
These questions follow the system-level concerns in Cummins’ paralleling guidance, which emphasizes controls, isolation, load management and support for the complete installation.
Keep Data Center Tiers and NFPA Classifications Separate
Uptime Institute’s Tier Classification System evaluates infrastructure capabilities, not generator count alone. Tier III addresses concurrent maintainability; Tier IV adds fault tolerance. An N+1 generator arrangement does not, by itself, establish Tier III, and a 2N generator arrangement does not establish Tier IV.
NFPA 110 uses a different classification system. NFPA’s overview published January 23, 2023 explains Class as operating duration without refueling or recharging, Type as restoration time, and Level as the consequences of system failure. Level 1 is not another term for N+1 redundancy.
NFPA lists the 2025 edition of NFPA 110 as its current published edition. The edition adopted locally may differ. Applicable classification, installation and acceptance requirements must be checked against the adopted documents, equipment instructions and authority having jurisdiction by the responsible qualified professional—not inferred from the tables above.