Tuesday, September 20, 2011

The Economist on Smart Grid

A former co-worker pointed out a Blog entry on The Economist website, titled "Difference Engine: Disaster waiting to Happen," and asked my opinion. I gave him a short answer in e-mail, but thought a longer answer was required.

I'm not sure who the author is, the byline just shows "N.V.". (Hmmmmmm. Envy? Of what?) I'm a bit puzzled by his (her?) unwillingness to attach a name to the posting. Regardless, it reads like a pretty reasonable assessment of the recent Southern California blackout, given that the failure analysis and determination of causes is in its opening stages. However, further thought (or, for that matter, scrolling through the Comments) identifies a number of factual inaccuracies (like the runway lights being out at the airport. Really, no backup generation?) and errors in thought.



It also discusses the economic and infrastructure issues that may have contributed to the problem. Some reasonable observations there, but I think the writer has to some degree over-generalized the California situation to the entire US. (S)he certainly doesn't discuss how the policy decisions by the State of California have contributed to the problem. Why aren't there baseload generators closer to population centers, and are there any other disincentives for SDE to install capacity? (NIMBY, anyone?)

The writer certainly makes some sweeping and inaccurate statements about the physical operation of the electric grid. (Sorry, regardless of how the markets operate, the voltage and current available at an outlet in California don't originate in West Virginia. Losses are too great, and electricity isn't packaged and shipped cross-country like produce.)

More amazing to me is the following pair of statements:

  • "All the smart grid does is add a communications layer to the local electricity-distribution network—so consumers can see at a glance how much electricity they are using at any time of the day, and how much it is costing them."
  • [After discussing the potential to remotely disconnect someone's home.] "But evil-doers from afar might not stop at that. Instead of switching off the power, they could run the voltage up and down to wreck sensitive electronic equipment, such as computers and television sets. And they could do that not just on single homes, but on whole communities and even to routers in substations—in an attempt to take transformers offline, if not actually fry them."
Pardon the pun, but does anyone else see the "disconnect" between these two statements? If "(a)ll the smart grid does is add a communications network" to distribution, to report usage and costs to consumers, how could a hacker practically "run the voltage up and down to wreck sensitive electronic equipment" and "fry" transformers?

The writer has simultaneously minimized smart grid's potential to benefit us, and maximized its risk. The risk/reward presented is utterly (intentionally?) biased.

The Comments range from blatant "MeeToo-isms" to actual, thoughtful discussions of the issues on both sides (and even some of the better comments have inaccuracies.)

Rather than picking apart the entire discussion (the better Comments already do a pretty decent job) let me focus on how a truly Smart Grid (not just "a communication layer" added to the distribution grid) might have made a difference in San Diego (and a host of other places) that day. The article states: "...a utility employee was "doing some work" on faulty equipment..." and inadvertently caused an entire substation to shut down. Assuming that this is the proximate cause (which is not entirely certain), an operating smart grid:
  • would have been tracking the actions of that employee while in the substation, and every action would have been coordinated with Operations.
  • would have had localized intelligence between the operator and the equipment, that would be programmed to trap and prevent most if not all errors. ("Doing this will take down the grid from here to Nevada. Are you sure?")
  • would have had available alternatives for self-protection at every station and generator, aside from just shutting down, likely reducing the extent and severity of the outage.
  • would have tracked and reported every event during the shutdown, allowing immediate, exact, identification of the cause and status, which would have enabled far more rapid restoration of service.
In short, an operational smart grid, based on the architectures actually under discussion, would have likely either prevented the outage, by trapping the error that started it, or made it a short term and far more limited problem.

In addition, an operational smart grid will be sufficiently diverse and distributed in its monitoring capabilities that the ability of a hacker to successfully intrude and do damage is likely to be less than exists now.

The outage in San Diego, like most outages, appears to have been a result of a single point of failure, cascading through the system. As the article correctly mentions, it makes no difference whether it is a technical error or a malicious act that creates that single failure. The automation that we have in place now, with its limited communication, monitoring and reporting capability, in all likelihood contributed to the extent of the San Diego outage, because it is only intelligent enough to save the equipment from damage by disconnecting it. Similarly, that same system is vulnerable to a hacker because it will respond in exactly the same way to someone triggering that single point of failure.

Of course, "a communications layer on the distribution grid" would have done nothing about a transmission level problem like the one described, but despite the author's assertions, that isn't what the smart grid is, or will be.

Yes, there are challenges to making the grid actively intelligent, and making the result safer, more reliable and secure. Some are physical, some are technical, and some are economic. I've talked about some of them elsewhere. Can we please honestly state the challenges, and try to understand them? Otherwise, we'll never solve them.

Overall, the article points up the need for, and challenge of, real education about the electrical grid. Part of making the grid smart is actually a process of making people smart about the grid.

Unfortunately, The Economist article does nothing to further, and instead actively hinders, that process.

No comments:

Post a Comment