Powerlessness

It’s been two weeks since the storm called Sandy hit the East Coast, with devastating effects on both people and infrastructure. I have some remarks to make about just one aspect of the damage: the prolonged power outage in parts of New Jersey, New York and a few other states. At the nadir, on October 30th, 8.2 million utility customers were without electricity. As of this morning, almost 99 percent of them have finally been reconnected—but that still leaves 89,000 in the dark, quite a sizeable number under ordinary circumstances.

I am outside the blackout area, but I have been following the story with particular care and interest because my daughter and her family are among those who have been living in pre-Edisonian gloom. She finally had service restored last night, after a wait 13 days.

The numbers charted above come from the U.S. Departmenrt of Energy, Office of Electricity Delivery and Energy Reliability. Note that they tally the number of customers—or in essence the number of electric meters—not the number of people. Carl Bialik, writing in the Wall Street Journal, quotes an estimate by Brian Wolff of the Edison Electric Institute: The peak outage of 8.2 million customers may have represented 60 million people in the dark. Over the two-week period, the outages sum up to about 32 million customer-days.

A few days into the ordeal, when my daughter was seeing no sign of power-company line crews, I offered this optimistic hypothesis: Maybe there’s a lot of work going on behind the scenes that will allow many thousands of customers to be switched on all at once. In other words, the plot of outage numbers would be concave-downward, with an initial plateau where there’s little change from day to day, then a steep plunge when all the circuits are finally re-energized. A glance at the actual plot shows that my theory was quite wrong, a product of mere wishful thinking. The shape of the curve is definitely concave-upward. As a matter of fact, over the first nine days, the progress of restoration is modeled fairly well by a simple exponential function: Each day, about 25 percent of the remaining outage victims get their lights back. Thus the number of restorations per day steadily declines. If we suppose that the work crews are putting in constant effort, this pattern of diminishing returns suggests that the work required to restore a single customer doubles every two or three days.

Sandy is the third storm in a little over a year to cause very widespread and persistent power-system disruptions in the Northeast and mid-Atlantic states. Hurricane Irene, in August of 2011, left 6.7 million customers without electricity for about a week. Then, a Halloween-weekend snowstorm knocked out power for 3 million customers, many of whom again had to wait several days for repair crews to show up. (My daughter’s family was among the unlucky ones in both of those events, too. They’ve spent roughly one full month out of the last 15 months off the grid.)

It’s my impression, based on nothing more reliable than personal recollection, that power losses of this scale and duration are a new phenomenon. Until 1996, I don’t think I ever experienced a power failure of more than 24 hours. What happened in 1996? I was in North Carolina at the time, and Hurricane Fran had me living by candlelight and charcoal grill for a week. It seemed like quite a novelty at the time.

We’ve certainly seen other large-scale blackouts over the decades. A vast swath of the Northeast went dark in November of 1965. New York City had a particularly traumatic night in 1977. In the summer of 2003 a cascading outage began in Ohio and spread east to the coast and north into Ontario. But these three events differ from the recent storm-associated outages in at least two important ways. First, they didn’t last as long: Few customers were cut off for as much as 24 hours. Second, the source of the problems was at the “wholesale” level of the electric-power grid, with generating plants and high-voltage transmission lines. A single fault in this part of the network can knock out power to millions of people; on the other hand, a single repair can bring it back again.

The power-grid damage caused by Sandy and other recent storms was mainly at the “retail” level, with hundreds or thousands of local distribution lines out of commission. Those snagged lines and broken poles have to be fixed one by one. (Sandy brought at least one exception to this “retail” pattern: The power failure in lower Manhattan was caused by flooding of a major substation.)

•     •     •

If it’s true that storm-damage outages have gotten bigger and more prolonged in the past couple of decades, what might account for the change? I don’t have an answer, but I have a bunch of speculative hypotheses to offer. I’ll let other people shoot these ideas down.

• It’s the climate, stupid. Intense storms are more frequent than they used to be. The problem is only going to get worse. Get used to it.
• Sprawl. Look at the hard-hit areas of New Jersey and Long Island, where millions of suburban and exurban residents have been suffering these past two weeks. Until recent years all those acres were farmland, occupied by cows and a few flinty farm families who knew how to light a kerosene lamp. When the power failed out in the sticks, only a few people were directly affected, and the rest of us never heard about it. Now those cow pastures are full of McMansions, whose occupants expect a much higher level of service.
• Sharpen the axe. When all that rural land was farmed, it was also cleared of trees. In the past 50 years or so, much of the northeast has been reforested at the same time it has been suburbanized. The roadside Norway maples planted circa 1960 are mature now—ready to fall on the nearest power line. If the power companies want to trim them, homeowners and municipalities won’t let them.
• Deregulation, privatization, fragmentation. In the good old days of my youth, the entire electric power industry was regulated by the Federal Power Commission and by state utility commissions. The companies themselves were vertically integrated, handling all aspects of electrical technology from generation to transmission to distribution to retail sales. There was an entrenched engineering culture that valued reliability above all else. After the 1965 blackout, the industry formed regional reliability councils to develop strategies and practices to avoid repetitions of that event. Now, in the post-Enron age, generating electricity, transmitting it and distributing it are economically separate activities. Organizations that were once cooperating divisions of the same parent company have become cut-throat competitors. Regulation has all but disappeared. Maintenance of right of way is seen as a cost and a threat to quarterly earnings, not as a prudent investment. No one corporate entity can be held responsible for keeping the lights on.
• Shoot the messenger. Most utility poles are owned by electric companies, but the owners also rent out the space beneath the power lines. The main tenants are telephone companies and cable TV. With the proliferation of communication circuits, poles carry more weight than they used to. Perhaps more important, fiber-optic telephone and cable-TV lines are often suspended from a very stout steel cable called a messenger wire. The poles along a suburban road could have two or three of these messenger wires. It used to be that a tree falling onto copper or aluminum wires would just snap them off; the repair was a quick splicing job. Now, the steel messenger wires can hold the weight of a fallen tree, and it’s the poles that give way. Replacing a snapped pole takes a lot longer than splicing a few wires. (During the Sandy recovery, there were rumors of pole shortages.)

Following the big blackouts of 1965, 1977 and 2003, boards of inquiry were formed to root out the causes and suggest remedies. I’m hoping we see the same kind of thoughtful inquiry this time.