WavebreakMediaMicro - Fotolia
Uninterruptible power supply selection starts with a plan and a business purpose. The hardware's goal is to maintain power when the utility service quits, but admins should determine for how long, how much redundancy is necessary, how big the supply must be, if it must eliminate power anomalies and how often the system must be online to provide backup power.
There are several uninterruptible power supply (UPS) options, but one size does not fit all. IT teams should figure out what they need and shouldn't overlook full bypass control requirements.
Reliable generators don't require long ride-through times, so admins may consider a flywheel UPS. It takes less space than other battery backup systems, has no expensive battery replacement and can cost less initially. Backup duration may be only 15 to 30 seconds, depending on the system's load size and configuration, but this is ample time for any good generator. Plus, a flywheel UPS doesn't require cooling.
Admins with exceptionally large data centers might consider a Diesel Rotary UPS, which combines the generator and UPS into one system. They're economical for larger facilities and can also save space and money but can have specific mechanical maintenance considerations.
For most organizations, a battery-based UPS is likely the ideal choice. Most options now have power factors of 0.95-1.0, meaning a 100 kVA UPS can actually deliver 95-100 kW of real power to any load.
Battery UPSes are now transformerless, which makes them more efficient than previous generations. It's common to see full, double-conversion UPSes with efficiencies above 97%, and efficiencies can remain at 96% or more with a rated load of 30% to 40%.
This efficiency is an advantage for 2N redundant systems, where each UPS must run at less than half load to maintain the redundancy. High-efficiency UPSes also minimize the effectiveness of any Eco Mode, which runs the UPS in Bypass Mode until an interruption occurs.
Eco Mode runs IT systems on utility power most of the time and only switches to full double conversion when necessary. Admins who minimize conversion losses over time may gain a percent in efficiency, but the risks of switching delays or failures means IT teams should seldom use it, especially with unstable power.
Big changes in battery technology
After admins decide on a UPS type, they must examine the available battery technologies. The market now has three main choices, and the options are all different.
Very large UPSes still use flooded lead acid batteries, or wet cells, which last at least 25 years. They're large, heavy and require professional maintenance, special rooms with hydrogen detection, exhaust fans, acid spill containment, showers and eyewash stations.
The main alternative has been valve-regulated lead-acid (VRLA) batteries, also known as sealed cells. Admins can place these almost anywhere without special precautions, but they require replacement every three to five years, especially if unstable power causes multiple discharge and recharge cycles. Long-life versions are more expensive and may last 10 years but might still require several costly replacements over the life of the UPS.
New Lithium-ion (Li-ion) batteries have longer lifespans than VRLA batteries. The industry is still learning about these batteries and discovering new chemical configurations. But unlike VRLA's, usage does not affect lifespan as much. Admins can partially discharge and recharge Li-ion cells many times without reducing lifespan, plus they're smaller and lighter weight. Most people still associate these batteries with consumer devices such as cellphones or tablets.
But UPS Li-Ion battery chemistry or packaging is not anything like the ones that phones or smaller devices use. They're safe and admins can install these batteries in most places. They have a higher initial cost than VRLA batteries and are not compatible with every UPS, but they do pay for themselves over the long term. There are even UPSes with Li-Ion batteries that work at the higher temperatures in line with ASHRAE TC 9.9 recommendations, which can save energy.
Modularity helps admins get UPS sizing right
The biggest cost mistake is oversizing -- both for UPS capacity and battery duration. Admins can use modularity to address this issue. IT teams no longer must buy a UPS sized for future growth.
Instead, organizations can buy system frame capacity for future growth but don't need to completely populate it; if the overall load grows, admins simply just add more capacity.
If an organization doesn't use the full system, then it hasn't spent unnecessary money. And if the load reduces, admins can remove modules and keep them as spares, so the UPS always runs at its optimum efficiency.
Moreover, admins can add redundancy at minimal cost with a modular setup. If organizations have a 100 kW UPS with five 20 kW modules, admins can add another module for N+1 redundancy without the need to buy another 100 kW UPS.
IT teams can also use modular batteries to extend available backup power, but the amount of battery power they should use can vary. With redundant generators, five to 10 minutes of backup power is plenty. A single generator may require more in cases of startup issues, but most UPSes run only 30 to 45 minutes without cooling; this means an hour of a UPS power is more than enough for orderly shutdowns.
For smaller data centers, admins can consider a distributed UPS. With the placement of modular UPS cabinets in each rack, organizations can save space and branch wiring cost. Admins can even install one modular unit, plus small, distributed rack-mounted units for truly critical systems. These small UPSes are just as reliable and efficient today as their bigger counterparts.