When you think of the fact that only $20 of $79 billion in annual cost is industry's share of national economic loss due to electric power interruptions (residential is $2 B and commercial is $57 B), does that make you think, "It's not my problem" or "sorry about that, dude?" This is a huge problem and getting bigger; a complex subject discussed at the Reliability Summit sponsored by Information Forecast Inc of Canoga Park, Calif., and held in Washington, D.C. the last days of September.
Electric power grid reliability for consumers means "no blackouts" but in the power sector it means "no cascading outages or uncontrolled separation of interconnected systems" and therein is a subtle difference. Blackouts are mostly distribution failures and local in nature, while the cascading outages can originate anywhere after the generating station but usually begin in the transmission network-the electricity pipelines. So grid reliability requires real-time, dynamic monitoring systems to measure conditions that describe reliability. The old way of measuring reliability and risk associated with failure was to account for peak load scenarios; that is, use static probabilities to assess failures and not account for system dynamics. New technology allows real-time feature measurements, which give moment-to-moment update of reliability and risk. These analyses no longer measure independent events, and even account for human error (such as failure to act or to execute incorrect actions in grid management) and assess protective and control system dynamics as significant contributors to system health and performance. What is emerging today is a more reasonable approach for industry users to shift suddenly into "power island mode" by disconnecting from the grid and becoming autonomous during times of high risk and declining reliability. This would have been a much better alternative for U.S. Steel Gary Works during the power outage of 14 August 2003 when they were caught with a furnace that froze into a load of metal that became a $250 million loss. Further, it was a loss not covered by insurance, because available technology could have prevented the problem.