“The electrical grid goes practically everywhere. It reaches into your home, your bedroom, and climbs right up into the lamp next to your pillow. It’s there while you sleep, and it’s waiting for you in the morning. Taken in its entirety, the grid is a machine, the most complex machine ever made.” —Philip Schewe in his latest book, The Grid.

Think about the pervasiveness of the global electrical network. It is a truly remarkable achievement. Like a gigantic octopus it envelops the whole earth and through its uncountable number of tentacles delivers, just in the United States, over 4 trillion KWh of electricity per year.

Not only does this network deliver such a tremendous amount of energy, it delivers it quite reliably, with an outage of only 8 hours total per year. In technical language, that is approximately 99.9% of uptime per year, more commonly referred to as “three nines” reliability (see the table). For an average household, such availability is adequate, but still results in nuisance outages. The food in the refrigerators will not spoil, but you might have to reset some of the clocks that are not battery backed.

While three nines availability results in just eight hours of outage per year, the grid experiences numerous disruptions due to electrical storms, electromechanical arcing, motor starts, and electrical welders, etc. These perturbances are almost impossible to prevent. For the commercial industry, such availability and line perturbances are simply unacceptable.

While exact calculation of the cost of a power outage varies (and varies for each segment of the industry), hospitals, airport and military installations cannot live with the three nines presently supplied by the power industry and have taken various measures to increase the uptime. Requirements in the telecom and datacom industry start at six nines reliability. However, the present smart chips can crash with power flickers of less than a millisecond. Preventing such fleeting disturbances requires greater than ten nines reliability.



Maximum Downtime per Year

Nine Nines


31.5 milliseconds

Eight Nines


315 milliseconds

Seven Nines


3.15 seconds

Six Nines


31.5 seconds

Five Nines


5 minutes 35 seconds

Four Nines


52 minutes 33 seconds

Three Nines


8 hours 46 minutes

Two Nines


87 hours 36 minutes

One Nine


36 days 12 hours

Table. Uptime and Maximum Downtime per Year

The cost increase associated with each higher level of uptime is not linear and it becomes increasingly expensive to achieve each additional nine of power availability. Each industry is willing to justify such cost based on their calculation of the business loss for each outage.
A retail store could probably live with three nines, since there is always the option to take a manual impression, instead of electronically scanning the credit cards. In this case, the sales loss would be minimal, unless you count the possibility of charges with a blocked or expired credit card that cannot be electronically verified with the bank. But some studies indicate that losses for the telecom, manufacturing and finance industries can be several million dollars per hour.

So, what are some of the ways to increase the uptime?

Since early inception, the telephone industry used batteries, and eventually incorporated battery chargers or rectifiers to recharge them. In today’s telecom systems rectifiers provide the -48 V for the telecom circuitry as they charge the backup batteries that provide backup power during power outages. The system bus voltages in such systems typically vary between a low of -36 V and a high of -76 V, which is required for charging and maintenance of the batteries.

If the projected down time is longer than the battery charge such systems use a backup generator that takes over when the battery power decreases to a certain level. Obviously, the generators need to start before the batteries fully discharge. Systems that do not require batteries for backup typically use front ends that supply +48 V with a narrower bus voltage range of +42 V to +53 V.

Batteries work well when at low and stable temperatures such as when banks of batteries are stored in concrete air-conditioned buildings. Unfortunately, with the remote installations of today, the enclosures mounted on desert mountaintops get very hot during the summer, resulting in very poor battery life. Replacement of the batteries in the field is also very costly and difficult, so there are various solutions to extend the battery life or to replace batteries altogether.

One of the obvious solutions is to use air-conditioning to stabilize the remote location temperature. However, such approaches use extra energy to cool the ambient and require backup generators to provide power to the cooling during power outages. Another approach is to use a flywheel to provide the power until a backup generator can pick up the slack.

There are essentially two types of flywheels: one, where a small volume flywheel spins very fast, and the other that uses a large-mass flywheel that spins slower. There are advantages and disadvantages for each method. The fast-spinning flywheel takes less room, however it needs to be encased in a shatter-proof enclosure to prevent shrapnel from flying in all directions, should the fast spinning flywheel shatter. The slow-spinning flywheel, on the other hand, takes up too much room.

In the datacom industry, where routers and switches are housed in large air-conditioned buildings, additional uptime is achieved by having redundant systems. Such systems may be simple dual systems that transparently switch from one to the other should one of them fail. Or they may consist of a multitude of smaller power systems that can fully power the datacom system when only one of the small power sources fails. Such power supply arrangement is commonly known as an n+1 redundant system, where n power supplies are enough to power the whole system with the additional supply taking over should any of the n power supplies fail.

Today, the widespread need for greater reliability is further complicated by growing demands for electrical energy and the pressures to reduce energy costs, particularly in the datacom industry. Recent developments in datacom, automotive, and other fields suggest that the user’s relationship with the power grid is changing in fundamental ways. The power supply industry will not be left unscathed by these changes. In next month’s column, I’ll look at some of the developments that threaten to reshape the grid and challenge power system designers. (read: Part Two: The Electric Grid—Now and in the Future)