Don't Pardon the Interruption
A centralized UPS scheme offers reliability against emergencies before they happen
The proliferation of electronic data processing, along with an increasing reliance on electromechanical tools and equipment, has made uninterruptible power supply systems (UPS) an essential requirement for almost any type of facility. The question is no longer whether such a system is a justifiable expense, but rather which UPS system will be most reliable and economical, and whether a centralized or decentralized strategy should be employed.
The bottom line is that a loss of power—even for a typical office building, let alone a major data center or a hospital—can be devastating. Even the most sophisticated computer networks are susceptible to harm from short interruptions in electricity. For some enterprises, these small outages can temporarily put them out of business, with considerable financial loss. A permanent loss of data might prove incalculable.
In many cases, utility outages may only last a matter of seconds—the time it takes to flick a computer switch on and off. Such a loss is normally covered by stand-alone UPS units, which are usually found on today's essential computers. These built-in, backup battery systems may be sufficient for short-term outages, but as electrical loads for equipment requiring uninterrupted service increase, a better solution may be warranted.
A computer fed by its own UPS will stay online as long as the battery can sustain the computer load. But greater reliability depends on the answer to an important design question: Should equipment that requires a UPS have its own dedicated unit, which must be tested and changed out on a per-unit basis, or should all essential equipment be fed from a central "house" system?
Opting for the latter presents the dilemma of putting all the eggs in one basket, and of course, keeping a constant watch on the basket. On the other hand, using multiple stand-alone UPS units means every basket must be watched—a difficult, if not impractical, task that often results in a poor maintenance program. Furthermore, only in emergencies will building staff become aware that units are not working. By then, it's too late.
All things considered, for most applications it is more manageable to take care of a single system. But this is a decision that needs to be made carefully by each business owner, based on individual needs and preferences. Centrally configured
A centralized UPS can provide reliable service in a number of configurations, each of which can provide a desired level of power-quality protection as defined by the International Electrotechnical Commission (IEC) standard 62040-3 (see "Three Levels of Power Quality," p. 43). All UPS systems require a source of direct current (see "The Instant DC Source"). The three basic UPS configurations consist of the following:
Standalone. In this configuration—with either batteries or a rotary flywheel as the direct-current source—emergency power is supplied continuously to the load as long as the DC source has the capacity to supply power (figure 1).
Battery/flywheel-backed. With this option—typically including a diesel-engine-driven generator—the DC energy source to the UPS provides power to the load until the generator is on-line (figure 2, p. 42). The generator may be able to support the load in as little as 10 seconds. A rotary flywheel source generally is sized to the load and can provide the power required until the generator assumes the load. Based on the types of loads, a battery system backed up by a genset may be smaller, since the time the batteries are expected to provide power is only seconds, as opposed to hours.
Combined DC energy sources. Combining a rotary flywheel and batteries prolongs battery life, because the batteries are not required to support the large demand of initial current requirements, as the flywheel supplies this energy. Such a configuration will also minimize deep discharges of the battery and frequency of their use, which will add considerably to the life of the batteries (figure 3).
The configurations in figures 2 or 3 may prove more advantageous in terms of providing a reliable source of energy for the electrical loads, because they combine the advantages of two systems. Moreover, UPS units can be configured in series- or parallel-redundant systems for even greater protection, and multiple gensets can also be configured in parallel arrangements. The inclusion of diesel generators in any UPS system—depending on the amount of fuel available—can add hours of off-line operation, particularly by sequencing them in parallel. Of course, any of these overlapping configurations has an associated cost, and the decision must really be based on the desired system reliability—how much capacity is needed now, and how much may be required in the future. A matter of distribution
The benefit of decentralized units is that the UPS is local, with no additional wire required. In contrast, a centralized system requires, by definition, a distribution plan.
Typically, feeders from a distribution panel in the UPS room will feed panels on each floor (figure 4), and 120-volt receptacles are fed out of each panel. The receptacle cover plate can be a different color—often yellow—to distinguish that only a UPS load should be fed from that receptacle.
When a yellow UPS-protected receptacle is available, it's notable to see how many people suddenly feel that their computer loads also need to be on UPS. Strict enforcement may be needed so that non-critical equipment is not plugged into these receptacles, which could possibly overload the UPS system.
If the decision has been made to proceed with a centralized system, a number of items need to be determined. First, the cost of a UPS increases with its associated power requirement, along with the length of time the power is required. Therefore, it must be decided what equipment is critical; that is, what needs to be on the UPS system. Once the total electrical load of this critical equipment is determined, the time requirement can be factored in. Some equipment may require one hour, whereas a data center may want continuous operation for as long as it takes to restore utility power. The type of configuration will be determined by the answer to these questions. The maintenance factor
Finally, while a major advantage of a centralized UPS is single point of maintenance, the system is only as good as its maintenance. It is essential that a regular schedule be set up to check all components, including the UPS, batteries, rotary flywheel and generators. Trained personnel can bring their maintenance experience to the table, to test the equipment and identify potential problems. But maintenance contracts are available and should be considered. Also, a fresh fuel supply should not be overlooked where generators are involved.
Power is something building users often take for granted, but system engineers are responsible for making the right decisions to keep power running continuously when possible. In any case, when it comes to power, don't pardon an interruption.
The Instant DC Source
All UPS systems need a source of direct current (DC) to immediately replace the loss of utility power to the most sensitive systems. A power loss of just 10 seconds is an eternity to a computer that is down, to those left in the dark or to those hooked up to life-support equipment that is no longer functioning. There are three sources of DC current immediately available:
Sealed-cell batteries. This DC source enjoys widespread use because of its lower cost, even though they offer relatively less power than wet-cell type batteries. They have a life expectancy of three to five years and typically require maintenance two times per year.
Wet-cell, or "flooded" batteries. These batteries are relatively costly, but offer greater power capacity, making them the choice where higher amperage is necessary. Take, for example, a large motor where sufficient cranking power is needed for start up. Flooded-cell batteries are also more durable, with an average life expectancy of seven to 10 years. They are, however, larger than sealed-cell batteries—thus having a larger footprint—and maintenance must be performed on a quarterly basis.
Both types of battery require regular maintenance and share a testing problem that arises from the batteries being wired in series: An entire string can be tested for voltage capacity, but individual batteries cannot be tested for remaining life without breaking the string. This involves taking the group off line, or removing the DC source. In practice, a facility's staff only finds out that a battery is bad precisely at the time when power from the UPS is needed.
Moreover, each time a string is discharged, battery use-life is diminished. More uncertainty is added when two strings are specified, with one used more heavily than the other. Both strings will test the same, but one will fail more quickly. To assure that the UPS will have a battery source when required, the strings must be replaced regularly. This involves not only the cost of new batteries, but also disposal costs.
Rotary flywheel. This final option eliminates the need for batteries. It is a rotating mass that acts as a motor during normal operating times, and which continues to generate power during outages by use of its momentum. Typically, depending on the load, rotary UPS units only provide service for seconds or minutes. Before the usable energy created by the flywheel is expended, a backup generator takes over the electrical load. Once the generator is carrying the load, the rotary UPS system begins recharging for the next outage.
Three Levels of Power Quality
The International Electrotechnical Commission (IEC) established its standard IEC 62040-3 in November 1999. The intent is to more clearly describe the various UPS topologies and to indicate how different levels of power quality are corrected with each type of UPS. The three levels are:
Passive standby (Level 3).
Line-interactive (Level 5).
Double conversion (Level 9).
Level 3 protects against a total loss of utility power, as well as power sags and surges. Level 5 gives the same protection as Level 3, plus it guards against under-voltage and over-voltage. Level 9 provides all the protection of Levels 3 and 5 but further guards against electrical line noise, frequency variation, switching transients and harmonic distortion.
Level 9 represents the complete solution, based on these typical power-quality issues, by regenerating the sine wave serving the load. The output waveform is compared against a perfect digitized sine wave, and is sampled and corrected at multiple times per cycle.