Ernest T. Bass asked an insightful question as he vented his frustration over having to study arithmetic: “Nines, nines — what's so important about nines?” The “rock-throwing nut” from The Andy Griffith Show has some great quotes, and this line came to mind as I was reading the Stephens investment banking report (September 2000) on power electronics. According to the authors, Todd Cooper and Holman Harvey, only electricity available 99.9999999% of the time or greater is reliable enough for our digital semiconducting society. They suggest some of today's sensitive electronic systems need reliability to the ninth or tenth nine to avoid costly power disruptions. In other words, they claim highly sensitive customers can't afford disturbances of more than 32 milliseconds per year. Even if this is true, you can't adequately assess power reliability by this index alone. We need to understand what makes nines important.
In the power industry, the nines index is known as the average system availability index (ASAI). It's a measurement that utilities track; however, more meaningful indices are hidden in the ASAI called the customer average interruption duration index (CAIDI) and the customer average interruption frequency index (CAIFI). Because customer sensitivities are different, it's important to first know how long of an interruption the customer's system can handle before it causes them a problem. Then, you correlate the frequency and the duration of the events to decide what kind of uninterruptible system is appropriate.
In the report, the authors state that utilities are ill equipped to meet this high reliability need with their existing systems, which typically produce only three nines (99.9%) reliability. I agree. Because utility systems are exposed to numerous adverse environments they can't cost effectively deliver much better service. The preferred way to increase reliability from a design and cost perspective is to provide uninterruptible equipment or redundant systems at the customer site.
The authors estimate the cost of providing high reliability service, claiming six nines could cost $1000 per kWh and ten nines reliability more than $100,000 per kWh. It appears there might have been some confusion between kW and kWh, because these numbers are astronomical. Wholesale peak prices have never exceeded $10 per kWh and these have only existed a few days a year — otherwise the average is around 10 cents per kWh. Regardless, much of the report's premise is still valid: Suppliers of backup and UPS equipment depending highly on power electronics should grow significantly over the next few years. The report does cite another source (Gilder Group), stating approximately 20% of all electric customers need greater than 99.9% reliability. This is good news for power electronic suppliers considering that 20% of all electric customers spend about 40 billion on electricity. Cooper and Holman predict that, “in the not too distant future, the aggregate profits of the companies providing reliable power will exceed the aggregate profits of those companies/utilities providing 99.9% reliable electricity.” This section of the report concludes, “Over the next decade, one of the most robust investment sectors in the economy will be the electric industry.” This makes sense, presuming it's primarily addressing the equipment suppliers rather than electric utilities. Ultimately, the conclusion is sound — even though the profit prediction is off base with cost figures ($/kWh) several orders of magnitude too high.
Since the late 1980s, I've believed power quality was a market waiting to happen. Until now, it has only applied to customers with sophisticated systems. I think we're near the knee in the curve, where the proliferation of digital equipment is so widespread that most business customers will need uninterruptible power. The long awaited market for power quality has gained critical mass and is now beginning to gain the momentum many have projected. The electric utility industry will be advancing with this increased necessity for uninterruptible power — but it's not the driver, and it can't deliver the nines. Society's growing dependence on sensitive electronic systems is driving this need, understanding that only power electronics can deliver high nines reliability.