The Grid We Need
Heyeck says that the existing grid resembles the nation’s congested and overloaded road network of the 1950s. President Eisenhower’s Interstate Highway System fixed that; now Heyeck is one of a growing number of people calling for the electricity equivalent—a network of power “superhighways” crisscrossing the continent. It would be an overlay on the existing high-voltage grid, just as Eisenhower’s blueprint laid a network of modern freeways over the existing patchwork of smaller roads.
Designs for a new interstate power grid exploit two extra-high-voltage technologies. The most efficient over long distances is high-voltage DC (HVDC), which employs converter stations analogous to the transfer stations that mediate power flow among the nation’s Eastern, Western, and Texas grids today. The difference is that in HVDC, the converters can be spaced farther apart—hundreds or even a few thousand miles away from each other. Power converted from AC to DC at one station is sent down a cable to be converted back to AC at another station at the end of the line. All that conversion is worth the trouble for long-distance transmission because DC travels with less resistance than AC. As a result, less energy is wasted heating up the cable and more energy reaches its destination.
Using HVDC to move renewable energy is hardly a new idea. Remote dams around the world already deliver hydropower via HVDC lines, including a 1,480-kilometer (920-mile) link that has carried energy from James Bay in northern Quebec to New England since 1990. But the technology is getting better and cheaper, making its use in a transcontinental grid seem more realistic.
One showcase for the new technology is the extra-high-voltage HVDC line that China is building to feed electricity from Xiangjiaba, one of its western hydropower megaprojects, to Shanghai. The 800-kilovolt line will carry three times as much power as Quebec’s, enough to satisfy the needs of roughly 31 million people on China’s teeming eastern seaboard, according to the company heading the project. (If the 1990-vintage technology had been used, transmission losses would have reduced the population served by about 1 million.) Similar grid upgrades could deliver wind power from America’s northern plains and Texas to the Atlantic and Pacific coasts. The downside of this approach involves the converter stations needed to add or remove power from the lines. The industrial-scale electronic switches behind AC-DC conversion are expensive. That cost limits the number of nodes that can be installed in an HVDC network, creating the transmission equivalent of an interstate highway with on- and off-ramps spaced hundreds of miles apart.
Heyeck sees another way to make the power highways more flexible while sticking with good old AC, albeit at an extra-high 765 kilovolts. Such lines would operate at more than twice the voltage at which most U.S. and European lines max out and carry up to six times as much power, since the capacity of a line is roughly proportional to the square of the voltage. (China just set up ultra-high-voltage 1,000-kilovolt AC lines that push this advantage even further.) This technology minimizes the need for converting back and forth between AC and DC. Moreover, a new network of extra-high-voltage AC lines with plenty of on-ramps would enable long-haul power flows to ride alongside the energy traveling over the existing AC grid.
The higher-voltage AC concept is so attractive that engineers at American Electric Power (AEP) and the American Wind Energy Association have sketched out a model hefty enough to supply 20 percent of U.S. power via wind energy alone (see map below). Their plan employs 19,000 miles of 765-kilovolt lines that would crisscross the country. At $60 billion the price looks high, but nationwide it would add just a few dollars per month to the average power bill.
Ultimately, Heyeck says, the interstate power networks are likely to blend the two technologies, favoring HVDC for the trunk lines that push power the longest distances and 765-kilovolt AC where flexibility counts most.
