On any ordinary day, electric power companies plan how much electricity to generate on the next day. They try to predict what customers will do, mainly by reading historical records of usage on the same day of the previous year. Then they adjust those figures to the current weather forecast for the following day.
"It's impossible to exactly predict what the demand for power will be at a given moment," says John Boyes, who manages the Energy Storage Program at Sandia National Laboratories. This scenario sets utilities up to make more or less electricity than customers use. The mismatch sends ripples through the grid, including variations in AC frequency, which, if not controlled, can damage electronics. Regional electricity managers, or independent system operators (ISOs), swoop in and try to close the gap by asking some power plants to change how much electricity they generate. But nuclear and fossil fuel plants can't do that quickly. Their slowness worsens the mismatch between electricity supply and demand.
Now, consider what happens on a sweltering day in Los Angeles when people citywide are running their air conditioners. These are peak demand conditions, when the most customers use the most electricity, which happens for a few hours on five to 10 days each year. On these days, facilities known as peaker plants are called into action. These expensive fossil-fuel plants sit idle all year and can emit more air pollution than a large coal-fired plant. "We wouldn't like to do it in a [smoggy] city like Los Angeles, but we do it anyway," says Imre Gyuk. If the peaker plants fall short, utilities pay large customers like aluminum smelters to use less electricity. "If nothing works, you have brownouts and rolling outages," says Gyuk.
Meanwhile, old substations are overloading. They're carrying more current than they're meant to handle, and the metal structures heat. "That's not recommended practice," says Boyes.
If the electric grid sounds stressed, you haven't seen anything yet. Read on.