RE: [cdn-nucl-l] "How the blackout came to life", Steven Strogatz

["Dukelow, James S Jr" <jim.dukelow@pnl.gov>]

Much as I admire Strogatz' marvelous book on chaotic dynamic systems, I have some quibbles with his appended essay.

The plant protection responses he describes DID serve the best interest of the grid (in the rather primitive context of the Midwest Independent System Operator), by preventing the frying of costly plant and network components that require long lead times (months and years) to replace.  Significantly, one of the more sensible post-blackout recommendations is the establishment of a national stockpile of such components.

The Midwest grid would be much more reliable if the grid manager (the ISO) simply had the ability to require load shedding from the 23 individual utilies that own the equipment, rather than being forced to "ask" them, not a very realistic process when you realize that the grid came down in nine seconds.  That 9 seconds was preceded by about an hour of network disruption, as it responded dynamically to the loss of three transmission lines in Ohio, probably time enough to take appropiate protective action, even with the relatively primitive Midwest grid.

I noted with interest that investigators are having difficulty piecing together the sequence of events during the blackout because individual utility "clocks" were not synchronized.  Next to my bed sits a $30 travel alarm clock that, like ET, phones home to the national atomic clock in Boulder, CO about six times a day. If each utility bought the PC-card version of that clock, they would be synchronized to within about a thousanths of a second all the time.  The utility industry has a long-standing reputation of being technologically backward, but really ...

As another potential quibble with Strogatz' prescription, I would note that writing reliable software for large, distributed control systems is an art (perhaps a black art) and not a science.

Best regards.

Jim Dukelow
Pacific Northwest National Laboratory
Richland, WA
jim.dukelow@pnl.gov

These comments are mine and have not been reviewed and/or approved by my management or by the U.S. Department of Energy.


-----Original Message-----
From:	Jerry Cuttler [mailto:jerrycuttler@rogers.com]
Sent:	Sun 8/31/2003 8:55 AM
To:	cdn-nucl-l (E-mail); ANS Member Exchange Listserv
Cc:	
Subject:	[cdn-nucl-l] "How the blackout came to life", Steven Strogatz

FYI
========================

How the Blackout Came to Life
By STEVEN STROGATZ 

With so much focus on the Ohio energy firm whose lapses may have triggered the blackout of 2003, it's been hard to remember that the real question is not how it started, but why it spread so far and so fast. Rather than tackle that question head-on, most commentators have reached for the usual metaphors: it was a chain reaction, a cascading failure, a domino effect. All of these are borrowed from the physical sciences. Maybe a better way to look at it is in biological terms.

We already use the language of epidemiology when we speak of "viruses" propagating across the Internet, "infecting" our computers. Likewise, it's tempting to view the blackout, spreading from link to link along the power grid, as a pernicious kind of electrical contagion. But that's not quite the right metaphor, either. The blackout was not caused by an infectious electrical disease; it was caused by the grid's immune response to the threat of such a disease. In other words, the grid suffered a violent allergic reaction, a sort of anaphylactic shock.

Just as the symptoms of a severe allergic reaction are caused not by the offending bee sting itself but by the overzealous response of the body's immune system to it, so the blackout was aggravated by the grid's attempt to defend itself, one power station at a time. Threatened by a torrent of electrical energy gone berserk, or overwhelmed by the sudden loads placed on it, each power plant in turn tripped its circuit breakers, detaching itself from the grid. Though this strategy achieved its desired aim -- saving each plant's generator from being damaged -- it was too myopic to serve the best interests of the grid as a whole.

What is needed is a more subtle, coordinated mode of response. When our own immune systems are performing at their best, they orchestrate their defenses through countless chemical conversations among T-cells and antibodies, enabling these defenders to calibrate their response to pathogens. In the same way, the thousands of power plants and substations in the grid need to be able to communicate with one another when any part of the system is breached, so they can collectively decide which circuit breakers should be tripped and which can safely remain intact.

The technology necessary to achieve this has existed for about a decade. It relies on computers, sensors and protective devices tied together by optical fiber so that all parts of the grid would be able to talk to one another at the speed of light -- fast enough to get ahead of an onrushing blackout and confine it. 

The sensors would continuously monitor the voltage, frequency and other important characteristics of the electricity coursing through the transmission lines. When a line appeared at risk of being overloaded, a computer would decide whether to switch on a protective device.  At present, such decisions are made purely parochially. Power plants defend themselves first, and don't worry about the consequences for neighboring plants on the grid. Nor do they consider any potentially helpful or harmful actions that those neighbors might be taking at the same time.

In the new approach, each plant would have nearly instantaneous information about all the other plants and power lines in its extended neighborhood. Everyone would know what everyone else was doing and thinking.  As threats arose (either from random failures or malicious attacks), the sensors would fire a flurry of warning signals down the optical fibers, and the networked computers would decide which protective devices to activate to contain the threat most effectively.  The grid would then be responding as an integrated entity, not as a ragtag collection of selfish units.  It would look a lot like an organism defending itself.

Granted, such a distributed control system would cost billions of dollars and, in this era of deregulation, there would be little incentive for energy companies to join forces and build it, especially when the big money is in power generation. But the construction of a systemwide immune network would be well worth the cost. Without it, our overburdened grid is likely to fail more and more often, and might even collapse, with costs that would be incalculable, both economically and in terms of national security. State and federal governments need to step in and provide incentives for utilities to do the right thing.

Of course, even if this new kind of smart defense system were to be built someday, one can already imagine an insidious disorder that might eventually outsmart it and afflict it, a catastrophic disruption of the immune system itself, rather than the grid it's supposed to protect. Such a thing would be the technological analog of AIDS.

A grim prospect, perhaps, but a realistic one.  We need to stop pretending that the grid is ever going to be a perfectible machine. Just as bacteria eventually develop resistance to the antibiotics  used to kill them, the defense of the grid will require ever-more inventive strategies on our part. We should recognize  that the power grid  needs to evolve and adapt, just like any other successful living creature.

END

Steven Strogatz, professor of applied mathematics at Cornell, is author of "Sync: The Emerging Science of Spontaneous Order."