The Grid: Electrical Infrastructure for a New Era

by Gretchen Bakke

A number of companies got in the game, none more infamously than Enron. One of the first companies to operationalize online trading of energy futures as well as the exploitation of spot (just-in-time) markets, Enron found some resoundingly killer apps indeed. These included playfully named schemes like Death Star, Ricochet, Fat Boy, and Bigfoot as well more straightforwardly named, equally new, procedures like “megawatt laundering.”

This beckoning, called a “sink,” says to all the current on the grid: “look here, this is the easy way.” To get electricity to where we need it we make little sinks, like flipping on a light switch, and big ones like firing up a paper mill.

To protect themselves and their machines, operators in Albany started shutting down generating plants and isolating transmission lines across the Northeast. As in New York, so also in Pennsylvania, the western half of Michigan, Ontario, all trying to “island” themselves from Ohio and its great energy sink. This process of islanding, or decoupling one’s local grid from the system as a whole, is not easy given the

Two minutes later, FirstEnergy got an even more worrisome call, this one from the organization charged with monitoring and regulating all the power in the Midwest as it moves across state lines—the Midwest Independent Transmission System, or MISO. When MISO operator Don Hunter called in just after 3:30 P.M. he got a befuddled Jerry Snickey, an engineer with FirstEnergy.

To their credit, the control rooms of power companies are full of screens, and these are full of data, charts and colors and rows of text, and control room operators are highly dependent on alarm systems that are either audible or flash on their screen, calling attention to critical changes in the flow of information. Without these alarms it’s questionable whether any normal human being is capable of noticing critical shifts in data patterns or laggardly mechanical behavior. And the alarms that day in FirstEnergy’s system didn’t work. Neither the onscreen alarms nor the speaker-blasted ones. There was no reason for anyone to have paid attention to the fact that their screens weren’t refreshing. Those screens said everything was A-OK.

All the while the bug was there. Not a bit of malware, not a virus, not a worm. Just a glitch, more of a programming error than anything else. In effect what this bug, called XA/21, did was cause a machine to respond with a busy signal when multiple systems tried to access it simultaneously rather than prioritizing these requests and then taking each of the “calls” in turn. As more and more data points were rebuffed, they started to stack up rather than being logged and then deleted. Like silt in a water purification system or cholesterol in your artery, all these tiny bits of retained information started to stanch the free flow of all information, slowing everything down, way down, and eventually crashing the main server. All the accumulated unprocessed events were then transferred to the backup server, which was no better equipped than its predecessor to handle this vast backlog and so it, too, failed.

In the case of the 2003 blackout the error on the grid took the form of overgrown trees and the error on the computers took the form of a line of code that disallowed simultaneous incoming data reports. Each error had small but significant effects on the actual physical infrastructure they came into contact with, while the way in which this physical infrastructure was designed—its logic—allowed for stochastic propagation of negative effects. Which is to say: the cascade and then the blackout.

A single line of code, in a program comprised of a million such lines, a fluke lasting little more than a microsecond, causes an alarm-event application to go into a loop. It just keeps spinning, spitting that bit of data back at the server, getting a busy signal, spinning and spitting, as the data not making it through mushroom from bits to bytes. This spinning manifests itself in slow computers, frozen screens displaying perfectly acceptable conditions as the grid outside the window is degrading ever more quickly. It becomes alarms that don’t sound, which become systems operators that look at their screens and tell worried callers that everything is fine. Until there are enough worried callers, until the data glut has gotten so severe that the servers have crashed. Until it is too late to do anything about it. Within twenty minutes of that last phone call, FirstEnergy’s own command and control center had blacked out.

But what most students of industrial accidents recognize is that perfect knowledge of complex systems is not actually the best way to make these systems safe and reliable. In part because perfect real-time knowledge is extremely difficult to come by, not only for the grid but for other dangerous yet necessary elements of modern life—like airplanes and nuclear power plants.

Rather than attempting the impossible feat of perfect control grounded in perfect information, complex industrial undertakings have for decades been veering toward another model for avoiding serious disaster. This would also seem to be the right approach for the grid, as its premise is that imperfect knowledge should not impede safe, steady functioning. The so-called Swiss Cheese Model of Industrial Accidents assumes glitches all over the place, tiny little failures or unpredicted oddities as a normal side effect of complexity. Rather than trying to “know and control” systems designers attempt to build, manage, and regulate complexity in such a way that small things are significantly impeded on their path to becoming catastrophically massive things. Three trees and a bug shouldn’t black out half the country.

Or, to put it differently, one doesn’t try to eliminate the holes in the Swiss cheese—by compacting all cheese into cheddar, for example—but rather to keep the holes in the cheese from lining up, from becoming one big hole that runs through the entirety of the loaf. The holes will be there (trees will be too tall, budgets will be tampered with, regulators will be put off, computer bugs will worm their way in), but they won’t snowball into total systems collapse the way they did in August 2003, and the way they very nearly did at Davis-Besse a year earlier.

The problem with the Swiss cheese model, which is otherwise remarkably robust (we have it to thank for the past thirty years in which flying has been consistently safer than driving; even frequent fliers are more likely to be killed in a lawn mower accident than an airplane crash), is that the grid, like any complex mechanical system, is not just a machine but also the regulatory, business, cultural, and natural environments within which this machine functions. The grid is the physics, mechanics, engineering, construction, management, upkeep, and use internal to itself. It is also the storms, earthquakes, laws, hatreds (and other personal opinions), and profit motives that surround the mechanism, that change over time, and that can differ drastically between one state,...

While the grid, electricity’s infrastructure, was reconceptualized in law to something rather like a box or shipping container. Conceptually if not actually, electricity now is made and sold in units to whoever pays the best price and then shipped to them by means of the grid.

The Energy Policy Act did nothing to upgrade the lines. The result was a simple one. The free trade of electricity meant that there was too much electricity traveling too far. Lines were getting overburdened, heating up, sagging, shorting out, arcing, and filling with harmonic resonances. All of which are both very bad for reliability and wasteful, and thus bad for efforts at conservation.

There was just too much electricity moving too far on the grid for it to behave in the same, somewhat predictable ways it had when power production and transmission were both more local phenomena.

“Predictability” is an awkward term when speaking of electricity, because strictly speaking it isn’t. Previously, predictability had been premised upon an algorithmic sense of user behavior and existing patterns of generation, transmission, and consumption that follow from population density, the season, the workday, and other fairly stable indicators of when and how much power might be used.

Long-distance wheeling and the unlinking of generation from local consumption habits has meant that a lot more data is now needed to manage the electricity moving on the system at any given moment.

Enron’s market-manipulating machinations to the side, almost all contemporary electricity traders engage in forms of electricity arbitrage (buying cheap here, selling dear there) made possible by the critical combination of the Internet and the Energy Policy Act.

Since 2000, however, not only is there far more information pouring into utility databases—a product of the widespread introduction of computerization at many points in the system, of which digital “smart” meters are the most infamous—but that information is treated as proprietary. Electricity trading, in this way, fails to be laissez-faire capitalism because nobody quite has access to the information necessary to make real-time decisions about how to manage their systems, and thus also their capital. Not even, oddly, the utilities themselves, since they now have so much information to contend with that most of it sits unprocessed in giant servers called “historians.” Big data has become just another modern way to use up electricity.

Historically, utilities made money when people used electricity; the more we used the more money they made. Now they don’t. Today’s utilities make money by transporting power and by trading it as a commodity.

It is incontrovertible that our wires were underprepared for long-distance wheeling, just as they were unprepared for more predictable things like population growth, the veritable explosion of pluginable devices that has come with the revolution in telecommunications technology, and our seemingly insatiable hunger for air conditioners. Not to mention the worrisome possibility that our cars, too, might be electric one day soon.

From the customer surprised to discover that his electricity usage actually increased during a six-hour blackout to apartment dwellers who found themselves quite suddenly in the position of paying more for their electric bill than for their rent, smart meters looked like and felt like a cash grab. And though there has been a lot of quibbling as to why, nobody argues with the fact that with the new technology’s arrival, monthly electric bills doubled, at times trebled. This happened in 2010, when Bakersfield had an unemployment rate of 16 percent and an underemployment rate double that. Nobody had $1,000 to pay a July electric bill that had cost $300 in June (not to mention $300 the previous July) with no appreciable change in consumer behavior. If both the old meters and the new meters were accurate, as the utility claimed, then the only viable explanation for the people of Bakersfield was highway robbery.

Meanwhile, in rural Maine, residents successfully pushed through an opt-out program because “safety standards for peak exposure limits to radio frequency have not been developed to take into account the particular sensitivity to eyes, testes and other ball-shaped organs.”

Energy Futures failed not because it was crooked; its bankruptcy was run-of-the-mill, utterly unlike Enron’s. It made a few too many bad business decisions over the years. What is interesting about its collapse, however, is how evident it became as the business was stripped down for public viewing that even things that look like utilities aren’t really utilities anymore. What’s really there after deregulation, restructuring, and the incorporation of the free market is an investment company running a bunch of endeavors that when taken together were, once upon a time, known by the name “utility.” Even if the “utility death spiral” doesn’t get them, the market, it seems, already has.

From the Lovinses’ point of view, the power outages that occur regularly in the United States regardless of their cause—whether big storms or changing legislation or computer bugs or terrorist hackers—are a natural and utterly predictable side effect of having such a big, centralized electrical grid. “The size, complexity, pattern, and control structure of these electrical machines,” they wrote, “rotating in exact synchrony across half a continent, and strung together by an easily severed network of aerial arteries whose failure is instantly disruptive … make them inherently vulnerable to large-scale failures.”

Renewable energy resources don’t need risky supply chains—like the chance arrival of a fuel truck in the middle of a hurricane—in order to function. Because it is more difficult (though far from impossible) to disrupt their functioning they are by their nature more secure.

Never just diesel; not just natural gas or solar plus wind, but wind plus natural gas plus battery storage plus conservation and energy efficiency measures all taken as one. Diversification, as any investor knows, is the essence of a robust portfolio. The same, they argue, is true of power production, transmission, and distribution systems.

From a study of biological systems rather than technological ones, the Lovinses argued that an organism’s longevity consistently relies upon “local back-up, local autonomy, and a preference for small over large scale and for diversity over homogeneity.” All of these things increase resilience in all cases.

Because of the need for resiliency—which is to say, many ways of making power, many ways of delivering it, and many ways of using it, all of which can be adjusted in a pinch—almost all microgrids incorporate some form of renewable generation.

A secure, ample energy system also includes good employment policies that lead to peaceful labor relations and good county, state, and federal regulations that ensure both the affordability and the long-term stability of extractive technologies.

And, as they search for the right guys to make their sketches into reality, other holy grails rise and fall from favor.

“In a symphony,” says Clay Stranger, “no instrument plays all the time, but the ensemble continuously produces beautiful music.” So too, he suggests, might we learn to conduct our grid, more artfully and with less of a lead foot.

Grid-scale storage, an element we need in order to integrate significant variable generation into the grid and to deal with our market proclivities for selling and shipping electricity as if it were a regular commodity, is limited to this: some artificial lakes, one compressed air plant, three molten salt towers, eight solar trough plants, and a lot of dreams about batteries.

As the anthropologist Akhil Gupta reminds us: “We need to reimagine electricity use in the future that does not simply seek to extend patterns in the present.”

Batteries, despite their ability to produce electricity on call, don’t actually have electricity inside them, instead they are full of chemicals. Under the right conditions these chemicals can be coaxed into a reaction that causes chemistry to produce electricity.

Every battery on the market is a compromise between price, toxicity, reactivity, and weight.

Infrastructure should, according to the design guru Donald Norman, fade into invisibility. It should be made to disappear from sight and equally to disappear from consciousness. It should be quiet, task-specific, and unobtrusive. We shouldn’t notice it, we shouldn’t think about it, and we shouldn’t seek it.

Systemic reform under these conditions is not easily accomplished: anything we add to the grid must have the capacity to interface effectively with everything that was there before, while everything we subtract from it must not disrupt the flow of power that we are so reliant upon.

Unless we make trouble, with guns, air conditioners, or home solar systems, we remain a relatively lumpen mass lacking even basic demographic nuances one might expect any twenty-first century business to employ. For example, in all my research into the grid I have never heard a utility customer referred to in the feminine. When speaking of the users of electricity, the (mostly) men who make the system work, and the (mostly) men who push at it and try to invent a way beyond it, imagine a nation of users who are also men.

There are thus three problems of different kinds that meet at the grid and get stuck there: How to deal with the combined interests of many different players—which does, and should, include global warming. How to deal with the legacy technology, which is to say the grid we’ve got. And how to deal with the fact that it’s made and run by humans, who are by their nature rather squirrelly and shortsighted.

In the high-stakes, low-sex-appeal battle to ramp up the interoperability of the grid’s thousands upon thousands of subsystems, the most boring, if heated, conversations behind the scenes focus on standardization. A platform is not just a software problem. We actually need the technologies that currently constitute our grid to be able to work with, and communicate with, newer components and newer ways of doing things.

The grid, as should be clear by now, is not a technological system. It is also a legal one, a business one, a political one, a cultural one, and a weather-driven one, and the ebbs and flows in each domain affect the very possibility of success of any plan for its improvement. If the integration of systems across domains, especially the irritating bits, cannot be made to flourish, the problem will be not with the machinery we use or the technology we govern, but with us.

Such interconnections between resources allows us to keep the idea of a power plant without our necessarily needing the power plant itself. A virtual power plant is thus primarily an organizational tool that uses information about electricity transmitted by electricity (digital smart meters most especially) to respond to the ebb and flow of power on the grid with a degree of timeliness and nuance that a human simply cannot match. We have always been too slow for electricity, but with smart meters, thousands upon thousands of distributed microsensors and the right computer programs, we can at long last do something about it.

Americans don’t like dealing with remnants or with imperfect things. We don’t want to see that what now serves us perfectly well was once broken or damaged. We want youth and vigor, not old age and a storied life. Repair is not a cultural value. Replacement is.

If we want to keep a grid for all, we might be wise to mend our grid like a Japanese pot. The most valuable bits, the golden threads, the tiny machines, we rub into all the seams—the glue would matter most.