Comparing UPS System Design Configurations

Overview

The paper covers five principal UPS architectures: capacity (N), isolated redundant, parallel redundant (N+1), distributed redundant (catcher, with and without STS, tri-redundant) and system-plus-system 2N/2(N+1). Each gets a one-line description and pros and cons, backed by a Markov/RBD availability comparison.

Each architecture is related to Uptime Tier classes and to an indicative first cost per rack. The paper also shows how module size in N+1 systems sets per-module load and therefore efficiency: for a 240 kW load, 2 x 240 kW modules run at 50% and 5 x 60 kW modules at 80%.

Modelled availability is 99.92% for N, 99.93% for N+1, 99.9994% for distributed redundant and 99.99997% for 2(N+1). In the analysis, isolated redundant and N systems showed the same unreliability (about 1.8% per year). N systems need 2-4 hours a year of full shutdown for system-level maintenance unless a wrap-around bypass exists.

The conclusion is that 2(N+1) feeding dual-corded loads directly gives the highest availability. The selection guide points N+1 or isolated redundant at IT loads typically under 500 kW, distributed redundant above 1 MW, and 2N/2(N+1) at large multi-MW sites. The figures are comparative and should not be read as absolute values.

Key points

1. Modelled availability: N 99.92%; isolated redundant 99.93%; N+1 99.93%; distributed redundant catcher 99.9989%; distributed redundant 99.9994%; 2(N+1) 99.99997%.
2. Indicative capex: Tier I $13.5-18k/rack; Tier II $18-24k; Tier III $24-30k; Tier IV $36-42k (30 ft²/rack, 2.3-3.8 kW/rack basis).
3. N+1 module loading for a 240 kW load: 2x240 kW 50%; 3x120 66%; 4x80 75%; 5x60 80%.
4. Typical 3-phase module overload rating: 125% for 10 minutes before transfer to static bypass.
5. Isolated redundant and N systems had the same unreliability (about 1.8% per year) in the analysis.
6. N systems need 2-4 h/yr full shutdown for system-level maintenance unless a wrap-around bypass exists.
7. Selection guide: N+1 / isolated redundant typically under 500 kW IT; distributed redundant above 1 MW; 2N/2(N+1) for large multi-MW.

Using it in design

• Offer N, isolated redundant, N+1, distributed redundant (catcher/STS), tri-redundant, 2N and 2(N+1) in the topology library, and label availability values as comparative rather than absolute.
• Compute per-module load as load / (number of modules x module kW) and feed it to the UPS efficiency curve.
• Map each topology to a Tier hint and scale the per-Tier cost ranges to the current year in estimates.
• Show a warning that N systems need a full shutdown for maintenance unless a wrap-around bypass is provided.

Who should read it

Electrical designers choosing a UPS architecture and decision makers weighing redundancy against cost.

Related standards

• Uptime Tier Standard: Topology
• TIA-942
• EN 50600-2-2
• IEEE 3006.7

Publication details

Authors

Kevin McCarthy; Victor Avelar

Publisher

Schneider Electric

Year

2016

Document number

WP 75

Document language

English

Source type

Manufacturer

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation