Maximum Power

Although long-term power outages happen only a few times a year, in most parts of the country, power fluctuations occur at least once a day. One daily fluctuation is enough to cause system and drive crashes, data transmission errors or shutdowns, server shutdowns and reboots, even premature component failure.

Because high-speed technologies across networks of every size are becoming the norm, higher sensitivity levels are creeping back into network hardware.

The need for high availability is increasing at incredible rates. Availability of 99.9% is no longer enough. You need 100% availability. If 100% availability seems too high for your system, consider what 99.9% availability translates into across user applications. When the system goes down, the network slows to a crawl or stops altogether. When competing with landline carriers for minutes of use, you need the maximum availability.

Although network managers must guarantee an unprecedented level of uptime, application platforms are more vulnerable than ever. Often, they contain a mix of leading-edge technology and aging technology used far from protected environments.

Despite these circumstances, distributed redundancy can help you ensure 100% power and improve availability. Unprecedented power-protection levels are possible regardless of platform, application criticality or downtime sensitivity. Distributed redundancy allows you to perform equipment maintenance easily and can fit within your investment/liability budget ratio.

HOW MUCH IS ENOUGH?According to Contingency Planning Research, an hour-long network outage can cost a wireless carrier 38,000 to 44,000 in service-activation losses. To make sure this doesn't happen to you, you must first deter mine an acceptable fault-tolerance level.

Fault tolerance encompasses not only how quickly the network can recover from a power failure, but also how much protection it needs against slow and fast power failures. Two elements determine fault tolerance. The most common is the acceptable mean time between failure (MTBF), or the average time before a failure. Perhaps a more relevant way to determine acceptable fault tolerance is through redundancy objectives as they apply to protecting the network.

The MTBF for critical support systems, such as uninterruptible power supplies (UPSs), should be as high as 10 times the load MTBF. The most basic MTBF calculations assume normal data distribution. Therefore, you should not use the MTBF characteristic alone to measure availability, predict risk failure prior to the calculated MTBF elapsed time or to value one particular MTBF number over another.

Redundancy objectives can predict your level of fault tolerance more realistically. These objectives are formed around the question: How much protection is enough? Redundancy objectives should have an operational goal of 24x7x365, adhere to a flexible power distribution and plan for maintenance and upgrades without shutdown.

THE OPTIONSThere are several traditional configurations designed for power reliability and availability. However, many factors affect them.

Single-module systems are perhaps the best known. They provide adequate continuous power protection under certain conditions. The UPS must be a true on-line topology. A static bypass switch should be in place to provide the utility to the load when a UPS transfers from its inverter. The entire critical load demand for the facility should not exceed 1000kVA.

There are several disadvantages sometimes hidden among the statistics of an average MTBF of ~100,000 hours. For example, there is only a 37% chance of achieving this MTBF goal. One reason is the UPS itself. If a UPS is not factory- or field-tested, it can reduce expected reliability by failing soon after installation.

Another challenge with this configuration: You have to take the UPS off-line for maintenance, service and IEEE battery testing. Therefore, it is not always available to the load.

An alternate UPS configuration is an isolated redundant configuration. In this configuration, a reserve UPS module supplies the bypass input of the primary UPS device. This protects against the primary module's failure. This happens when the primary bypass power is not available and during primary UPS module maintenance. However, a disadvantage of isolated redundancy is the same as its strength: The entire load resides on a primary UPS until the moment when it is needed most, at the point of power failure. This creates many potential risks. First, the standby UPS must accept up to a 100% step load. Then, not only must the primary module's static bypass switches operate properly to obtain power from the reserve module, but both static switches must detect the situation properly and supply current from the utility source.

Variations on the isolated redundancy configuration have included a single reserve module as backup to several independent primary UPSs. Generally, the reserve UPS is sized to support only one of the primary modules. This requires such complexity in the switch-gear configurations and associated controls that it usually offsets any reliability gains.

Another redundancy approach is the use of parallel redundant UPS modules with a static bypass switch. Although this is the most widely applied approach, you must apply UPS modules and their configurations carefully.

Reliability principles dictate the fewer parallel modules required for the load capacity, the better. Consider that the calculated system MTBF of two parallel redundant modules is 2.27 million hours. The MTBF with three modules used in the configuration drops to 757,000 hours. When six of the same UPS modules are required with one redundant module, the MTBF is reduced to only 151,000 hours. A variation of the parallel redundant system is the one+one configuration. Two module UPSs with internal static switches are connected in parallel.

Some form of system-level control still is necessary to allow the modules to share the load and control transfers to the bypass source. Multiple static bypass switches must operate in parallel for proper reaction transfers and overload conditions. In addition, system-level maintenance bypass still is necessary for system-level maintenance.

The distributed redundant configuration requires a complete change in the large UPS design approach. This change is best reflected in a recent Uptime Institute survey of large data-processing-center downtime. According to survey results, 79% of electrical infrastructure failures that interrupted critical load operation occurred between the UPS output bus and the critical load. The emphasis of critical-power-system designers needs to shift from building a bullet-proof UPS system to creating a fault-tolerant UPS system. This transfers the importance of power maintenance from the output of the UPS to the input terminals of the load equipment.

Distributed redundancy means creating dual, full-capacity UPS-system busses and redundant power-distributed systems. This eliminates as many single points of failure as possible, all the way up to the load equipment's input terminals. To provide fault tolerance, you must provide some method of allowing the load equipment to receive power from both UPS power busses. Protecting against fast power-system failures such as circuit-breaker trips or a power-system fault requires a fast switching method. Applying static transfer switches creates fast break-before-make transfers between two ac power sources. You should design the two ac-power sources to be as independent as possible to eliminate common failures. Switching between the two power sources needs to be break-before-make for the same reason.

You can devise several distributed-redundancy power configurations. Keep in mind, however, that redundancy needs to be as close to the load as possible to achieve its goal -- keeping power available at the load-equipment level. The ultimate distributed-redundancy configuration would feature two independent UPS power-distribution systems with dual-input load equipment as redundant ac power is provided up to and inside the load equipment. Some carriers may consider dual-input load equipment in this ultimate implementation as a drawback. But more information-technology equipment is being designed with dual-input capability. So distributed redundancy not only provides the best assurance of power reliability and availability, it also paves the way for an easy migration path as you deploy more dual-bus loads.

THE BUSINESS PLANIf you require high reliability, high availability and high probability of achieving the predicted MTBF, then redundancy is a must. For ultra-critical loads, your power system needs to be about 10 times more reliable than the load -- and redundant -- to avoid compromising the initial investment as well as the overall business plan. This almost certainly points to a distributed-redundancy configuration.

Facility-support systems cost little, just as critical loads cost little. Both are investments as opposed to overhead costs. The performance of the investment is the issue, not protection. Therefore, you should consider all components of the distributed-redundancy configuration part of a performance package, not optional functions.