UPS—What’s New and What Works

In order to fulfill its primary function—delivering power from a few seconds to as long as 10 minutes or more—an uninterruptible power supply (UPS) must continually monitor the power coming into a facility. Ideally, when the line power goes out, even for 1/120th of a second, which is half a cycle of 60-Hertz AC power, the UPS takes over and the facility's equipment doesn't even notice.
By Charles R. Bourn, Jr., P.E., Carter & Burgess Inc. Fort Worth, Texas December 1, 2001

In order to fulfill its primary function—delivering power from a few seconds to as long as 10 minutes or more—an uninterruptible power supply (UPS) must continually monitor the power coming into a facility. Ideally, when the line power goes out, even for 1/120th of a second, which is half a cycle of 60-Hertz AC power, the UPS takes over and the facility’s equipment doesn’t even notice.

The size, type and technology of the UPS needed to perform this task depend on the type of facility, as well as the application and the load to be served.

In virtually all facilities, the UPS must maintain power long enough to buy time to store data and shut down computers. In critical applications, where data loss and downtime are not tolerable, the UPS must provide steady, reliable power until an emergency backup power generation system can be started and brought on-line.

Where the average office building or call center might need 7 to 10 watts per square foot, a telecom or data center might require at least 50 watts per square foot when it first opens. In a few years, when facility reaches maximum capacity, power requirements may increase to as much as 200 watts per square foot.

In other words, every facility has unique demands, which must be carefully scrutinized to determine the proper UPS solution. Even facilities of the same type can differ greatly with respect to UPS requirements.

Static Systems

Some of the older UPS systems consist of arrays of lead-acid batteries, similar to those found under the hood of a car. They must be located in well-ventilated areas with spill containment designed into the floor to protect against accidental releases of the electrolyte.

In contrast, newer systems use valve-regulated batteries with minimal maintenance requirements. They are sealed, and the electrolyte is suspended either in a gel or glass mats that virtually eliminate spills. Known as a static system, this type of UPS uses rectifiers and inverters—instead of moving parts—to accomplish the power conversion.

Static systems use reliable, proven technology, but they can also pose some problems. For example, they are heavy and consume a lot of floor space. Floors must be strong enough to support the batteries, and this can add significantly to a structure’s costs over its life span, especially for a data center that might cost more than $300 per square foot to construct.

Battery arrays also present maintenance issues that must be addressed during the design phase. Batteries must be checked and tested regularly to ensure they will operate properly.

Important to remember is that a battery system’s peak capacity is at installation. After a few years, a system designed for five to 10 minutes of backup power might provide only three to four minutes, depending on how well it is maintained and how often it is charged and discharged. If the load being served has grown, the shortfall will be even worse.

UPS Options Today

Static standby UPS , the most economical type, are typically used in small systems. It allows AC power to serve the load directly, with batteries constantly kept at charge by incoming line power. In a power failure, direct-current (DC) power from the battery array is converted by an inverter into AC power, which can be used by electronic equipment. When power has been restored—or a backup generator has come on-line—the UPS system changes back to normal AC power and begins recharging the batteries.

One disadvantage with this system is that if it isn’t on-line all the time, there is a short delay on transfer. Additionally, the unit does not filter out noise as well as an on-line UPS.

Central UPS systems are on-line all the time and act as a filter and buffer between line power and the load. These are called double-conversion units, because the AC power is first filtered and converted into DC power by a rectifier, and then reconverted back to AC power to serve the equipment load. They provide cleaner power on a continuous basis, with no transfer time if outside power is lost.

Rotary UPS systems use standard AC current to operate a motor-generator that provides clean AC power for the equipment. In a power failure, batteries provide current to the motor/generator. Rotary machines totally isolate the served equipment from the incoming power but are more expensive and noisier than other types.

Flywheel-based UPS systems have won a niche by eliminating batteries. Battery-based UPS systems account for a high percentage of UPS installations, but technological advancements have made this battery-free option available.

Flywheel systems use better bearings and other improvements in technology to provide short-duration backup—also known as ride-through—without the cost, environmental concerns and maintenance issues posed by batteries. These stored-energy devices are especially suited to provide short-term power until a backup generator can kick in or when there is noise in a power line, caused by lightning, tree branches falling on a power line or switching of large power loads. Even without these outside influences, utility power can fluctuate, damaging sensitive equipment and corrupting vital data.

A flywheel UPS is a hybrid system, with the incoming power converted into DC and then back to AC, as in a static system. Under normal conditions, the DC bus of the UPS powers a motor/generator, which moves the flywheel. The flywheel stores kinetic energy, so that if the power fails the flywheel continues spinning. As long as the flywheel operates, the generator provides power to the DC bus, similar to the batteries in a static unit.

A major challenge for flywheel systems is the length of time during which they can provide ride-through capability. When power is interrupted, the flywheel continues to spin fast enough to meet the demands of the load for 10 to 40 seconds. At this point, the emergency backup system should be on-line. While flywheel systems can be engineered to supply power for longer periods, costs are prohibitive for most applications.

Sizing the System

Determining the size of the UPS system entails finding the balance between serving the current load at a reasonable cost and providing the flexibility to expand as the equipment load grows. Many designers think bigger is better. This philosophy may provide the largest cushion for operations, but it can add hundreds of thousands of dollars in cost, yet provide only marginal improvement in power reliability.

The bigger-is-better idea reigned during the dot-com boom of the 1990s. In today’s economy, however, companies are looking for ways to protect data systems without needless spending. By carefully planning to meet the current needs of the facility, and designing for modular expansion as demand grows and as equipment is added to the data center, the first-cost of a facility can be reduced because:

  • UPS equipment will be smaller and cost less.

  • Fewer batteries will be needed.

  • Air-conditioning equipment can be smaller, because a smaller UPS system generates less heat.

  • Ongoing operating costs for all systems will be lower.

The UPS does not need to supply power to all electrical systems in a facility, and it is not required to serve the complete load that is served by an emergency generating system. The UPS system only has to be powerful enough to give critical systems the time they need to be shut down safely or shifted to the emergency backup system. Designers should resist the temptation to design UPS systems to meet the ultimate, build-out load if, as sometimes is the case, demand will ramp up over time. Instead, it is useful to determine the load on Day One and then project load growth over the facility’s lifespan. Manufacturers package UPS systems into modules. With proper space planning, UPS modules can be added in the existing space as demand increases.

A system that is too large is not only more expensive than it has to be, but it also takes up valuable space that could be used for more productive applications. The electric room equipment—distribution equipment, power panels, transfer switches, UPS—in an Internet data center can take up 20% to 25% of the entire floor space.

In addition, redundancy levels must be considered when sizing the system. UPS systems are not perfect. Plan for a failure. Specify a maintenance bypass on all in-line modules so that if redundant units are not installed, the UPS system can be taken off-line and serviced. For example, in one application that could be served by three 750-kVa modules, four were installed. At any time, three served as the UPS while the fourth provided a standby if one failed.

Working Together

The UPS and the rest of the electrical system, including the normal power system and any backup system, must work as a single unit. Designers should coordinate UPS operating characteristics with the normal power system to make sure it is designed to withstand the available short-circuit current, and also to work with the emergency power system, if one is provided.

If utility power is disrupted, it can take several seconds for the emergency generating unit to go from a cold state to meeting all the facility’s needs, including lighting and HVAC. For these few seconds, demand on the UPS system is high.

Further, discharged batteries in the UPS may begin re-charging as soon as the power is restored by the emergency system. This adds to the electrical demand on the generator. Include the higher power requirements of the UPS during recharging when determining the load to be served by the emergency generator. If the emergency generator is sized to only meet the facility load and not the load of the recharging UPS, it may sense an overload and go off-line. The UPS system will detect this and pick up the load until it discharges the batteries.

As new advances in technology are introduced, UPS systems will continue to improve. Through careful planning and engineering, systems will be designed to take advantage of these capabilities to help keep facilities on-line.

From Pure Power, Winter 2001.

Calling Central

Central UPS systems that are designed to serve an entire building or facility not only provide a level of security above that of individual units but also simplify matters for the building owner and operator.

Implementing a central UPS system—rather than scattering battery units throughout a building—puts everything under the control of a trained operator. The UPS system and batteries can be maintained and serviced to ensure rapid, reliable operation whenever needed.

In addition to a device that stores energy, a UPS system incorporates surge protectors and filters that reduce the “noise” from the incoming and outgoing power signals. To minimize the chance of damage to the UPS, separate transient voltage surge protectors should be provided between the utility line and the UPS to protect it—and the load—from lightning strikes and other power events.

IGBT Update

Development of the insulated gate bipolar transistor (IGBT) is the innovation that has had the greatest impact on UPS technology. These devices have largely replaced silicon-controlled rectifiers (SCRs) in UPS inverter technology, though SCRs are still used on the rectifier and static switch parts of a UPS.

IGBTs are primarily understood as a means of eliminating the harmonics problems associated with SCRs. With a pulse-width modulated (PWM)-controlled IGBT UPS, harmonics can be greatly reduced. This, in turn, allows a much more economical electrical backup system overall. Generator size need only be 1.3% greater than the UPS rating, rather than the 200% required for a traditional UPS system with high harmonics.

But there are other advantages as well. IGBTs take up only about one-fifth the space of SCR components, allowing for a much smaller UPS unit—and considerable space savings.

UPS manufacturers have been working in recent years to convince clients that there may well be a payoff to swapping a traditional UPS unit that is still in good shape—with an expected 15 to 20 years of life—for a PWM-controlled IGBT UPS.

There have been two basic approaches to incorporating IGBT technology. Some UPS makers have opted to retrofit inverter sections in their UPS units with IGBTs. Others have chosen to create an entirely new IGBT-based UPS line.

No matter what type of IGBT UPS system chosen, it appears to be a good bet that facility owners can count on much increased efficiencies. In fact, manufacturers are boasting a 5% to 10% greater efficiency for IGBT compared with the older SCR units.