It is an axiom that mission-critical facilities must run continuously. Any interruption can disrupt business operations and result in significant financial losses. Almost all data centers today are designed with systems and redundancies to minimize loss of power. Just as important to mission-critical operations, however, is the quality of the power.
Power quality problems don’t make headlines the way power reliability problems do — for example, the blackout of Aug. 14, 2003. And power quality problems can be more difficult to understand, analyze and solve than reliability problems. But that doesn’t mean power quality should be ignored. Consider these statistics:
Clearly, mission-critical facilities are vulnerable to interruptions in general and certainly to those caused by power quality problems. Most often, vulnerability is reduced or negated using power conditioning equipment in redundant configurations from the point of common coupling downstream to the critical load.
Generally speaking, there are seven major categories of power quality phenomena:
1. Transients are undesirable but momentary events. Root causes include lightning strikes and transformer inrush.
2. Short-duration variations are related to voltage dips and surges and short interruptions. The root causes are fault conditions, “walk-in” of large loads, and improper wiring and grounding. Wiring and grounding problems account for 80 to 90 percent of power quality problems.
3. Long-duration variations can be caused by an over-voltage or under-voltage situation. The root cause is often attributable to load variations and switching operations. The latter leads some facility executives to cut maintenance budgets, justifying the cuts not only for economic reasons but also claiming that a high percentage of data center failures occur during switching maneuvers. Other executives will conclude the opposite, claiming that preventive maintenance is key to operational readiness.
4. Voltage unbalance is most often caused by having single-phase loads in three-phase circuit distribution.
5. Waveform distortion phenomena include noise, harmonics, notching and others found in waveform distortion. The root cause is often data center nonlinear load characteristics.
6. Voltage fluctuations are just what the name implies; flickers are a good example of this category. One example of a root cause is arc furnace operations. Data centers cannot risk being around equipment that can cause voltage fluctuations.
7. Power frequency variations involve changes to the fundamental frequency, 50 or 60 Hz. The main root cause is the source supplying the system.
Only two decades ago, most power quality phenomena were a mystery to engineers. Since then, problems have become more understandable as electrical engineers began to analyze specific events. This led to significant breakthroughs in the power quality field, including the design of sophisticated power-monitoring tools. Armed with these new tools, mission-critical facility operators need no longer be ignorant of the impact power quality problems cause data centers. These monitoring tools are now actively helping mission-critical facility operators to first recognize they have a power quality problem and then to choose the right solution to mitigate them.
Power quality standards became the initial conditions that must exist when designing power system architecture. For example, the ITIC (Information Technology Industries Council) curve, or the “renovated” CBEMA curve, is historically recognized and still used by some in the field. Developed in 1994, this curve defines the limits of the acceptable zones for computer voltage by a tolerance envelope. All data center conditioning equipment must meet this important industry standard.
Another important standard is IEEE 519, Standard Practices and Requirements for Harmonic Control in Electrical Power Systems. This standard defines total harmonic distortion of voltage and total demand distortion; those definitions are an important part of all technical decisions related to power quality.
The latest release of IEEE 1100, Recommended Practice for Powering and Grounding Electronic Equipment played an important role in the improvement of defining requirements for data center power quality. In fact, a popular trend began shortly after its release; IEEE 1100 compatibility tests are being created that help users better understand the chapter on wiring and grounding.
Power quality standards led to the creation of new requirements that helped ensure that data centers remain operational at all times. A wide variety of equipment is designed to meet these new standards and requirements, both in new system designs and upgrades.
Equipment can play an important role in addressing power quality problems. Some of the issues that might have to be solved with equipment are harmonics, and the number of power sources.
Harmonics are present in the system because of switch-mode power supplies. Care must be taken so that harmonics don’t harm the critical process. The harmonic signature of the system is different from point to point and changes with time in most of applications. Monitoring is imperative. Once problems occur, the solutions are various, such as using special transformers that cancel several harmonics, passive filters and active filters. This equipment must be deployed carefully to avoid starting a chain reaction of problems.
For example, implementing passive filters in a dynamic, changing facility might cause the appearance of resonant flows that are dangerous and hard to find. Implementing active harmonic cancelators is easy and will take care of most problems, but they are expensive and care must be taken in deploying them. Because this kind of equipment injects current directionally, it is very important to engineer the correct installation of current transformers. Sometimes a hybrid implementation of active and passive filters might be the correct solution.
When critical loads have just one power source, the only way to increase availability is to use automatic static transfer switches. An automatic static transfer switch can take advantage of a dual-path power supply downstream to the critical load under the ITIC standard constraints. The best solution, however, is to ensure that critical equipment has two or more separate, independent power sources.
While there are sound, cost-effective solutions to many data center power quality issues, other problems — like ones related to wiring and grounding — are not only expensive to solve but also complex. Some examples include loop grounds, ground faults and touching voltages.
Loop grounds, which are caused by incorrect wiring or grounding, can result in equipment malfunctions. These are not easy to find, but once located they can be eradicated.
Ground faults are dangerous to people and equipment. Sometimes ground faults are invisible, hidden by grounding practices, incorrect protection setting or the lack of ground fault protection equipment. Some equipment might interfere with operational targets, so careful consideration must always be given to ground fault detection and resolution.
Touching voltages is another issue to be addressed. Resolving static problems — by using dust control, for example — might not be enough to eliminate the problem, which is dangerous to people and equipment. This problem can be the result of high frequency drainage currents from switched mode power supplies of data center equipment. Due to their radio frequency behavior they do not obey Ohm’s Law, but rather electromagnetic wave propagation laws, resulting in the need for a signal reference grounding system. IEEE 1100 has recommendations that complete NEC requirements.
Power quality is a vulnerability any system design must carefully consider. It is a highly technical and complicated subject. Nevertheless, facility executives must be proactive in addressing power quality causes and effects.
K.L. Godrich is a senior electrical engineer with EYP Mission Critical Facilities in New York City.
E-mail comments to edward.sullivan@tradepress.com.
