North America is windy. If the U.S. and Canada had enough wind turbines, they could produce all the electricity they need, and then some, from wind alone. The same is true of solar energy, with even bigger power surpluses. The U.S. Southwest's deserts get enough sunlight to sustain the country's thirst for electricity—20 times over. But both these sources are inherently erratic: winds wane and clouds show up with little notice. Wind, moreover, tends to blow harder at night, when demand for electricity is at its lowest.

According to the U.S. Department of Energy, when intermittent sources such as solar or wind reach about 20 percent of a region’s total energy production, balancing supply and demand becomes extremely challenging: rolling blackouts can sometimes become inevitable. The same problem exists elsewhere, notably in Germany, where a vast photovoltaic capacity has sprung up thanks to generous subsidies.

Burton Richter, a physics Nobel laureate who was on a recent panel that studied California's power supply situation, told The New York Times blogger Andrew Revkin that because of intermittency, utilities would need to keep fossil fuel–burning plants as a backup that can quickly ramp up generation as need be. This large-scale load following, as it is called, "can only be done with natural gas," Richter told Revkin.

Of course, if a statewide lull in wind were to last several days, California could buy electricity on the interstate market, as it often does. Proposals for a continent-wide supergrid that can ferry large amounts of power coast to coast are in part predicated on the idea that "the wind always blows somewhere." Also helpful would be to make the grid smarter, so that users, prompted by dynamic pricing, would buy the energy they need when it is abundant and dial down their use when the cost goes up.

Experts, however, are increasingly skeptical that even supergrids or smart grids would suffice to cover the intermittency of wind and solar power. The solution, they say, must include storing massive amounts of energy for later use. Ideally, the U.S. could build a really gigantic battery and be done with it. But we are talking gigawatts (billions of watts) of power, and such a battery would be prohibitively expensive, at least with current technology.

Fortunately, two types of large-scale energy storage exist that are already mature and economically feasible—and some more futuristic ones are also promising—as I describe in the article "Gather the Wind," in the March 2012 issue of Scientific American. These two technologies are less flashy than the ones the media usually likes to cover—they are decidedly "more Flintstones than Jetsons," as one online trade journal put it. One involves pumping water uphill, the other one involves compressing air.

A pumped-hydro facility consists of two reservoirs with a substantial drop in height between them. When there is excess electricity to go around, electric pumps move water from the lower reservoir into the upper one, thereby storing energy in the form of gravitational potential energy. When wind and solar wane or simply cannot keep up with demand, operators let water flow down and through turbines, generating electricity. In compressed-air facilities, excess electricity pumps air into underground caverns, and it is later released at high pressure to turn turbines.

Pumped hydro has been used for decades to balance the load on large U.S. grids. About 2.5 percent of the electricity used by U.S. consumers has cycled through one of these plants. In Europe the amount is 4 percent and in Japan 10 percent. But how much storage capacity would the North American grid need to stay reliable as more wind and solar comes online?

That question is surprisingly hard to answer exactly. "It is very complicated," says Rick Miller, a senior vice president at HDR Engineering, Inc., a company that builds pumped-hydro facilities. Different mixes of renewable sources can be vastly different in their output profiles, and some are easier to match with demand, others harder. For example, photovoltaic panels in New Mexico will reliably pump out electrons right at the time when nearby towns need it the most—that is, when the air conditioners are on. With wind, it's trickier.

Haresh Kamath of the Electric Power Research Institute in Palo Alto, Calif., says many factors will determine how much storage is needed. What penetration of renewables do people want? What level are they willing to pay for? How much transmission are they willing to live with? What kind of loads are they likely to use? And what level of reliability are they prepared to live with? "There are a great many people with different ideas of how to answer these questions," Kamath says, "and we are likely to come up with different answers to this question in different parts of the world at different times."

It helps to get an idea of the scale of the problem. The U.S. consumes, on average, about 500 gigawatts of electrical power at any moment—roughly the equivalent of 500 million toasters or hair-dryers plugged in at all times. A bit less than 20 percent of that energy comes from nuclear power, which is already virtually carbon-free, and another few percent is from traditional hydropower, which in addition to being green does not have an intermittency problem. The portion of the power supply that needs to be decarbonized (think: coal plants) and that could be end up being intermittent is, then, the remaining 80 percent, or roughly 400 gigawatts.

If all of that power came from wind, and if the wind went down everywhere at once, the country would in principle need 400 gigawatts of backup. But such a worst-case scenario seems unlikely to say the least.

Instead, according to Imre Gyuk, who heads the energy storage program at the DOE, many grid operators and utilities agree that a good rule of thumb is that a typical portfolio of renewables will need about a 20 percent storage backup. For our average 400 gigawatts of renewable power, that 20 percent would amount to 80 gigawatts. That is a lot, comparable to 80 nuclear power plants, but perhaps not unattainable. The U.S. already has more than 20 gigawatts of pumped hydro-capacity, and the industry is considering proposals that would double that number.

Generating capacity is, however, only one side of the story. Storage systems are rated not only by their power, or how fast they can crank out energy (measured in gigawatts), but also by the total amount of energy they store (measured in gigawatt-hours). A facility with an energy capacity of one-gigawatt that can only supply electricity for 10 minutes would not be very helpful; in an ideal world it could do so for, say, 100 hours, thus storing 100 gigawatt-hours. Building up new pumped hydro-facilities similar to existing ones would probably help in all but the most disastrously long of wind lulls. For those worst-case scenarios, we might still have to brace for rolling blackouts.

Of course, this simple calculation also assumes current consumption levels. How would we power all those electric cars that we're supposed to be driving in the future?

The electrification of parts of the economy that are now fossil fuel–driven—notably, if hybrid or all-electric cars become the main means of transportation—would indeed drastically increase electricity consumption. Fortunately, for the purpose of balancing the load, that problem would in part take care of itself, because most people would recharge their cars at night, and would use smart meters that would enable them to draw power at times when it's abundant and cheap.

The intermittency of renewable sources of energy is emerging as the single biggest obstacle to decarbonizing the power grid. Fortunately, the problem does seem solvable.