Is power protection costing more than it should?

All electronics in an organisation - from desktops to data centres - require continuous, clean power. Anything less puts businesses at risk of data corruption, equipment damage and unplanned downtime. Most organisations meet this requirement by having a central UPS that serves an entire building/data centre, or distributed UPSs placed closer to the equipment being protected, or both.

If a UPS in an organisation is five years old, it’s costing more than it should. If it’s 10 years old, it could be costing thousands.

In a 1 MW data centre, a 10-year-old UPS could be wasting 150 kW and dissipating lots of heat. Replacing it with new, high-efficiency UPSs can free up about 120 kW of that power to support new equipment and reduce the burden on cooling systems. Replacing a 550 kW UPS from a redundant-UPS configuration with a higher-efficiency model could annually save over $40,000 in power and cooling bills, while eliminating 190 tonnes of CO₂ emissions.

Changing the ‘game’

Until recently, 96-97% energy efficiency was the upper limit in double-conversion UPSs. For every dollar spent on power, three to seven cents was used or dissipated as heat by the UPS. That doesn’t sound like much, but it adds up, as we’ll see later.

Energy-saving UPSs can deliver up to 99% efficiency, achieved by intelligently adapting to the quality of the utility power and operating in ‘energy-saving’ mode most of the time. In contrast, traditional double-conversion UPSs continuously process utility power through an inverter and rectifier, dissipating heat and wasting power at every stage.

Trade-off with high-efficiency UPSs

Until recently, there were significant trade-offs to increasing energy efficiency. To offer the highest efficiency, UPSs had to expose downstream equipment to potentially harmful surges and could create other problems due to slow reaction times. New advances in high-efficiency UPSs eliminate these sacrifices. This was the relief data centre managers sought when facing intense pressure to reduce energy consumption and meet environmental regulations, without compromising uptime.

The less efficient that power-protection systems are, the more utility power has to be bought to run a facility. Even slight gains deliver significant savings in power and cooling. For example, on a 250 kW load, about $4000 can be saved per percentage point of efficiency gain - enough to pay for the UPS in 3-5 years.

There’s been plenty of hype over the years about high-efficiency ‘eco-mode’ UPSs that switch between modes to improve efficiency. However, these multimode UPSs have limitations. They can be ineffective against many types of power problems, slow to respond to transient power conditions and vulnerable to high surges, downstream shorts and other electrical system conditions.

Despite that history, multimode UPSs can be entrusted to power mission-critical data centres, if one chooses wisely and asks the right questions of a UPS vendor. There are vast differences in the way these UPSs achieve their high efficiency - and the degree of protection they afford.

Five questions to ask before selecting a UPS

Does the UPS sacrifice protection to gain high-efficiency?

Some high-efficiency UPSs force a trade-off between power savings and power quality. Internal design profoundly affects efficiency and protection levels. Line-interactive UPSs are efficient, but they only offer limited voltage regulation, surge suppression and battery backup. Premium, double-conversion UPSs do the most processing to deliver clean power, but at a cost to efficiency.

New multimode UPSs offer the best of multiple topologies in one UPS. They flash between different modes to match the conditions of the moment. Under normal conditions, it’s in a high-efficiency mode that includes surge suppression and voltage regulation. When input power is poor, it uses double-conversion technology to deliver cleanest output power. When power quality is very bad or goes out altogether, it draws on internal or external batteries or a standby generator as needed, like a typical UPS.

How does the UPS achieve its high efficiency?

Models vary in how they switch between operating modes. Conventional multimode or ‘eco-mode’ UPSs usually operate in either:

• ‘Standby’ mode under normal conditions, powering the load from a utility input source that bypasses the internal circuitry of the UPS. Whenever major power disturbances occur, it has to start up and charge those internal components, synchronise the electrical waveform and then transfer to double-conversion mode. During short power disturbances, critical loads are left exposed to potentially damaging conditions. Or,
• ‘Line-interactive’ mode, with an inverter running continuously to provide some surge suppression and voltage regulation when in high-efficiency mode and enable faster switchovers to double-conversion mode.

The first mode takes too long to respond to power conditions. Damaging surges or transient-power problems could still reach equipment. The second mode is faster to jump into action but consumes more energy, tending to be no better than 96-98% efficient.

New multimode technology resolves both issues, where the inverter is continuously charged but not processing power. It remains connected, running all controls and synchronised with the input power, so the UPS can transition to double-conversion mode without delay and without compromising efficiency. Inverter filtering components are connected to the load all the time, providing surge conditioning comparable to ‘double-conversion’ mode.

How efficient is the UPS when lightly loaded?

Real-world efficiency can be different from stated efficiency. Manufacturers usually state UPS efficiency ratings at full load, but most UPSs are markedly less efficient under lighter loads, which is how they’re typically being used. Since so many IT systems use dual-bus architecture for redundancy, the typical data centre loads each power bus (and corresponding UPS) at less than 50% capacity, often only 20-40%.

It’s important to know UPS efficiency across the entire load range, not just under theoretical ‘ideal’ UPS operating conditions. While many UPSs drop off markedly in their efficiency under light loads, others can perform at 99% efficiency even when lightly loaded, as much as 15 percentage points better than traditional UPSs.

How quickly does the UPS detect and respond to power events?

Electronic equipment can only tolerate brief, sub-second interruptions. Look for UPSs that quickly detect and respond to power problems - ideally 2-4 milliseconds - for two reasons:

• Minimising inrush current. After even a brief disruption in power, equipment draws surges of energy to recharge their capacitors. The longer the interruption, the greater the inrush current. Even if disruptions are only 10-15 milliseconds, surge currents could be 10-times higher than normal draw. If hundreds of servers were all drawing inrush current, UPSs could be overloaded or circuits could trip. For this reason, choose a UPS that switches between modes with the least possible interruption in power, ideally 2 milliseconds or less.
• Preventing disruption to downstream static switches. If a data centre has A- and B-side power systems for redundancy, it probably has static switches in the power infrastructure to extend this A/B redundancy to single-corded loads. If an upstream UPS takes too long to change state - from ‘high-efficiency’ to ‘double-conversion’ mode or back - these downstream static switches could mistakenly perceive a power disruption and switch between A and B power sources. To prevent these unwanted transfers occurring, UPSs must have faster detection/transfer times than the static switch.

Look closely at any vendor’s claim that UPSs change modes in a stated number of milliseconds. The stated figure sometimes doesn’t include all the steps involved. Transition time is a function of two factors:

• How fast UPSs can turn on their inverters; and
• How fast they can turn off static switches that allow power to flow through the UPS’s internal circuitry.

Only when both activities have been accomplished are UPSs truly online in double-conversion mode.

What extras does the UPS offer for maximum protection?

New high-efficiency UPSs have proved reliable under prolonged and repeated power problems far greater than typical commercial sites experience. Even so, some facilities managers will feel more comfortable knowing the UPS is fully in double-conversion mode at times, such as during thunderstorms. Some multimode UPSs offer options for locking in double-conversion mode under user-specified conditions.

Look for UPSs that can differentiate between upstream and downstream faults and respond differently depending on the power problem origin.

Conclusion

Utility costs now account for 20-30% of data centre operating costs. According to IDC, for every dollar spent on new IT hardware, an additional 50 cents is spent on power and cooling - more than double the ratio of five years ago. The cost of electricity is already outpacing the cost of hardware. $1 million worth of servers that was purchased in 2009 will consume $1.2 million in electricity over a three-year operating life - a figure that will only increase with rising utility rates.

The good news is that manufacturers have dramatically improved the efficiency of power-protection systems - reducing the costs and environmental impact of powering the business. More efficient allocation of power with high-efficiency UPSs reduces operating costs; achieves more with available backup power and cooling systems; delays upgrades and expansions; and helps advance corporate sustainability initiatives.

By Pedro Robredo