Power protection basics

By Dr. James Kennedy, NCE, SRP, MBCI

It is getting harder and harder for utility companies to provide the type of clean, consistent, and continuous power required by today’s growing businesses. Buildings have aging electrical infrastructure designs, making it increasingly difficult to supply occupants with the clean, continuous power needed for sophisticated computing and network equipment needs. And the power utilities generation and distribution problems are not expected to improve in the near future. According to industry experts, it takes approximately a decade to get a new power plant online, and concerns about nuclear power and fossil fuels have stifled the construction of new generating facilities.

The ever-increasing deployment of networks, client/server computing platforms, and distributed computing systems have raised the potential adverse impact caused by sustained power outages. According to a Contingency Planning & Management survey, 45.3 percent of all computing and network equipment or data loss failures reported are attributed directly to power failures or surges.
The damage can be great. In just a typical example of a car hitting a utility pole, primary distribution lines are shorted, causing an electrical surge followed quickly by a complete loss of power. The surge travels instantaneously through the electrical distribution wiring into a server via the electrical outlet. The motherboard and its components are damaged, and data contained in the system memory is lost. After the surge passes and all power to the server is lost, unprotected hard disks can crash and critical data stored in the disk controller’s cache can be instantly lost. As a result, the business has suffered a damaging loss of critical data, which may or may not be recoverable.

The first step in protecting an enterprise from power failure is to understand and prepare for it. If an enterprise has not already experienced some type of power-related incident, it most likely will in the future.

Types of power failure
Typically there are six types of power anomalies and/or failure incidents experienced by enterprise computing and network users: total failures (also known as blackouts), severely reduced voltage (or brownouts), sags, spikes, surges, and noise.

Total failures, or blackouts, constitute a complete loss of electrical power to the networking or computing equipment. It can be total failure throughout an entire geographical location, a single building or group of buildings, or a single electrical panel within a building. These failures can cause system and network crashes, PC lockups, and corruption or loss of valuable data from servers and workstations. These blackouts are often caused by electrical storms, auto accidents involving utility poles, an electrical utility company’s inability to meet user demand (e.g., during sustained hot weather conditions), or simply by the inability of a buildings infrastructure to handle demand on an overloaded circuit.

Severely reduced voltage from the power utility, or a brownout, occurs when the utility company cannot meet customers demand. This usually occurs in summer months when the use of air conditioning during times of prolonged heat waves taxes the pool of available power. Sometimes these brownouts are planned by the utility and are directed at specific locations within their power grid. In this case they are called rolling brownouts. Today’s network and computing equipment is designed to operate using electrical power that falls within a specific acceptable range. Brownouts occur where the electrical supply voltage drops below this level. This lowered voltage places a strain on the electronic components contained within the computing and network equipment and can limit their operational life. It can also cause immediate failure of those electronic components.

Sags, spikes, and surges are power anomalies caused by electrical storms, extraordinary demands for power from the utility, and other electrical equipment. For example, the starting of an electrical motor, like an elevator, can sometimes cause momentary sag as the power source tries to meet demand, and then cause a spike once the motor has started and needs less electrical current to sustain operation. These anomalies are rapid, momentary decreases (sags) or increases (surges or spikes) in voltage levels at the electrical outlet. They can cause loss of data, total loss of a hard disk, and even catastrophic damage to the network or computing hardware in use.

Electrical noise is electromagnetic interference (EMI) caused by electrical storms, noisy electrical equipment (e.g., motors, welding equipment, etc.), fluorescent lighting, and even radio transmitters. It can cause events such as system lockups, temporary lapses in computing, circuit connection termination, data transmission errors, and even data corruption or loss.

Assessing risk
A complete assessment of power-related risks will help a company determine the appropriate level of protection needed and justify the cost of power-protection equipment. To determine the level of protection a company requires or can afford, the following questions should be considered:
* Does the business experience frequent power failures?
* Is there frequent construction going on near or inside of the company’s facilities?
* Are facilities located in areas where severe weather events occur, such as earthquakes, tornadoes, and snow/ice/hail storms?
* Is the distribution wiring in the facilities over 20 years old?
* Do the facilities have emergency generators, and do they supply power to critical systems and network equipment outlets? (Note: Oftentimes generators are in place, but they supply power only to emergency lighting, some elevators, and critical building infrastructure systems, not to tenants or occupants of the buildings systems.)
* When the business experiences downtime, is personnel time wasted or does the outage impact the revenue stream? Has the company performed a cost-of-downtime assessment?
* Does the company have mission-critical business networks, systems, and/or processes that must be operational 24/7?

Types of power protection
After reviewing the risks posed to an enterprise, the appropriate power protection devices can be selected based on the scope of interruption potentially faced.

Short-term interruptions
Short-term power anomalies such as spikes and surges lasting only a few millionths to a few thousandths of a second can be prevented using surge protection devices. Placed between the network or computing device and the power outlet it is plugged into, a surge protector can prevent these power anomalies from reaching the server, router, or hub.

Intermediate-term interruptions
Intermediate-term power failures, such as blackouts, brownouts, and electrical noise lasting from a few seconds to 30 minutes, are treatable through uninterruptible power supplies (UPSs). UPS devices contain batteries and an electronic inverter, which converts DC power from batteries to AC power required by the equipment being protected. Software used in conjunction with a UPS can detect a loss of power and, after a predefined period of time, cause servers to execute their shutdown procedures. This shutdown allows files to be closed and updated in a protected fashion, eliminating data loss and corruption. Once power is restored, the system is signalled to start back up. In the case of brownouts, the UPS detects this condition and can take over the job of supplying power in the proper operating range.

In general, there are three different types of UPS devices. With an offline standby power supply (SPS), power is usually derived directly from the power line until power fails, at which time a battery-powered inverter turns on to continue supplying power. The time it takes for the inverter to come on line varies by unit. The battery is charged when line power is available. Offline UPSs do not compensate for voltages within the 103-132VAC-input window, and therefore are not useful in dealing with minor fluctuations in voltage. They generally only protect systems from power spikes, and only protect against power sags and line noise when the battery is switched on due to a blackout. They are generally not suitable for servers, but because of their low cost they are often used in protecting desktop workstations.

When normal operating line power is present, a line-interactive UPS system conditions power using a Ferro resonant transformer. The UPS maintains a constant output voltage even with a varying input voltage and provides good protection against line noise. The UPS transformer also maintains output on its secondary briefly when a total outage occurs. As with the offline SPS, the line-interactive UPS switches to battery operation if it experiences a significant power surge or complete outage. It offers adequate protection as long as power sags or outages do not occur with a high degree of frequency in a short period of time, in which case the battery would switch on and off without having enough time to completely charge. This type of failure mode would reduce available battery time during brownout or blackout conditions.

With a true UPS system, power is supplied continuously from an inverter. There is no switchover time, and such systems provide the highest level of protection during normal operation. It can provide line conditioning and voltage regulation and protection. The true UPS system delivers power under all conditions and continually recharges its battery, making it always ready to respond if a severe power problem occurs. This type of UPS is most often used for mission critical networks and computing equipment requiring high availability.

The appropriate load size of a UPS can be calculated by adding the volt-amperes (VAs) of the devices to be connected to the UPS. (If a devices draw is specified in volts and amperes individually instead of VAs, the VA value can be obtained by multiplying the volts by the amperes. If the device only identifies the number of watts required, the VA value can be calculated by multiplying the number of watts by 1.4.) For safety, the UPS should have a VA of 40 to 50 percent more than the calculated VA value of the devices connected to it. This will generally provide a UPS with a battery capable of supplying power long enough to allow the protected devices to properly shut down under the maximum load size. UPS devices should be selected with enough battery time to power protected devices through brownouts periods. In most cases, UPSs with battery times of at least 15 to 20 minutes are selected.

Finally, operational tests and maintenance checks should be performed regularly. Testing for line failure recognition should be conducted monthly, and testing for full power failure shut down and restore capability should be conducted quarterly.

Long-term interruptions
Long-term power failures can last from a few hours to several days. In this case, alternate long-term power generation equipment is needed. This can come in the form of portable power generators or generators that have been installed as part of a buildings infrastructure. These generators are powered by diesel, gasoline, or propane and can continuously supply electricity to a building’s mission critical systems for up to a month.

Such standby power systems (SPSs) typically generate electricity in the 152000kW range. SPSs can be triggered into operation either manually or automatically. In the manual operation, a user starts up the SPS and, once it has reached speed, manually switches it into operation. Automatic SPSs switch into operation when the power failure has been detected for a specific period of time and switches out once power is restored. This switch-in time should be less than the UPS dropout time; otherwise, systems and networks will experience a short-term outage window.

Only experienced personnel, with the aid of electricians, building management personnel, engineering contractors, and equipment vendors should undertake the selection, installation, and testing of SPSs.

The inclusion of an SPS into an enterprise’s power management plan can significantly reduce problems associated with complete loss of power. The cost of an SPS typically is between $10,000 and $50,000. Emergency generators also can be rented in case of a prolonged outage; however, the need for a licensed electrician to connect the generator and approval of the buildings manager are all contingent on its use.

A growing need
The experience of a worst-case failure to reproduce or recreate lost data or equipment may far exceed the cost or time needed to implement a good power protection plan. Power failures are most often unpredictable, and are often disastrous when they do occur, so enterprises without power protection strategies in place need to implement them before it’s too late.

Dr. James Kennedy, NCE, SRP, MBCI is a Distinguished Member of Technical Staff at Lucent Technologies Inc. He can be contacted at 973-386-2223 or at jtkennedy@lucent.com

•Date: 7th November 2003 •Region: Worldwide •Type: Article •Topic: Power man.
Rate this article or make a comment - click here