How the Power Grid Really Works and Why Supply Must Match Demand Every Second

The biggest machine ever built has no off switch, almost no storage tank, and a rule that cannot be bent even for a second. Every time you flip a light, a power plant somewhere has to produce exactly one more light's worth of electricity at that same instant. Not in a minute. Not tomorrow. Now. It is a high-wire act performed by a whole continent, and the only reason you do not notice is that when it works, it works so quietly that you forget it is there.

I want to explain how it works, because you cannot argue sensibly about energy policy if you do not know what the grid demands. Listen to a debate on the radio and you will hear the words "clean," "affordable" and "green" tossed around like beach balls. You will almost never hear the word "dispatchable." That is the word that decides whether the lights are on at six o'clock on a January evening, and by the time you finish this post you should be able to explain it to your brother-in-law.

Four steps from the plant to the plug

Electricity travels a path with four stages.

1. Generation. A power plant turns some other form of energy into electricity. Coal, natural gas and nuclear plants boil water to spin turbines. Hydroelectric dams use falling water. Wind turbines catch wind, and solar panels convert sunlight directly.
2. Transmission. Step-up transformers raise the voltage to very high levels, anywhere from about 115,000 volts up to 765,000 volts, so power can travel long distances with less loss. These are the big towers you see marching across fields.
3. Distribution. At substations, step-down transformers lower the voltage, and the power moves onto the smaller lines along your street.
4. The meter. A final transformer, perhaps the gray can on a pole or the green box in a yard, drops the voltage to the 120 and 240 volts of your house.

Notice what is missing: a warehouse. Electricity is not stored in the system in any large amount. There are pumped hydro facilities and, increasingly, batteries, and they matter. But compared with the total demand of a continent, the amount held in reserve is small. For nearly all of the grid's history, electricity has been made on demand, the moment it is used.

The balance that cannot slip

Here is the physics, in plain terms. In North America the alternating current in the grid cycles back and forth 60 times a second. That is its frequency, 60 hertz. Think of the whole interconnected system as a gigantic set of spinning machines, all turning in lockstep at that rate.

When demand rises faster than generation, the machines slow down a little, and the frequency sags. When generation exceeds demand, they speed up, and the frequency rises. Equipment on the grid is built to tolerate only a narrow band around 60 hertz. If the frequency drifts too far, protective systems disconnect plants and lines to save the equipment. Each disconnection removes supply or load, which can push the frequency further and trigger the next disconnection. That is a cascade, and it is how a local problem becomes a regional blackout.

There was a famous one on August 14, 2003. A set of overgrown trees touched high-voltage lines in Ohio, an alarm system failed to alert the operators, and the problem spread until roughly 50 million people in the United States and Canada lost power. A tree and a software bug. That is how thin the margin can be, and why the discipline of balancing matters so much.

So the first law of the grid is: supply must equal demand at every instant. Operators do not meet that law by hoping. They meet it with a plan, a set of reserves, and a lot of people watching screens.

Who keeps the balance

The people responsible are called balancing authorities. Each one has an area and a duty to keep generation and load in balance within it, and to manage the flow of power to and from its neighbors. Many are utilities. Large parts of the country are covered by regional organizations known as regional transmission organizations or independent system operators, such as PJM in the mid-Atlantic and Midwest, MISO in the middle of the country, the Southwest Power Pool, and others. They run the wholesale markets that decide which plants run in what order.

How did these groups come about? In 1996, the Federal Energy Regulatory Commission issued Order 888, requiring utilities that owned transmission lines to open them to other sellers on fair terms. That helped create wholesale competition and the organizations needed to run it.

Above all of them sits a reliability body. After a huge blackout in the Northeast on November 9, 1965, the industry created the North American Electric Reliability Council in 1968. Under the Energy Policy Act of 2005, FERC designated the successor, the North American Electric Reliability Corporation, NERC, as the national reliability organization in 2006, with the power to set mandatory standards and impose penalties.

Three grids, plus a few

Now a surprise. There is not one American grid. There are three big ones in the Lower 48, called interconnections:

• The Eastern Interconnection covers most of the country from the Rocky Mountains east, except most of Texas.
• The Western Interconnection covers the area from the Rockies to the Pacific, including parts of Canada and Mexico.
• The Texas Interconnection, run by the Electric Reliability Council of Texas, or ERCOT.

Quebec and Alaska have systems of their own as well. Within each interconnection, everything is synchronized at 60 hertz. Between interconnections, there are only limited direct-current links that can move some power across the boundary without tying the frequencies together.

Why does it matter? Because a shortage in one interconnection cannot always be fixed by sending power from another. The size of the pipe between them limits how much can move.

Why Texas stands alone

Texas is the odd one, and the reason is a legal one. In 1935, Congress passed the Public Utility Act, whose second title became the modern Federal Power Act. It gave the federal government authority over the interstate transmission of electricity and wholesale sales in interstate commerce. The earlier law, the Federal Water Power Act of 1920, was renamed and expanded. The federal regulator then was the Federal Power Commission, which the Department of Energy Organization Act replaced in 1977 with the Federal Energy Regulatory Commission.

The usual account is that Texas utilities kept their lines within the state so as not to be subject to federal jurisdiction over interstate commerce. The Texas grid grew into its own system, and ERCOT was formed in 1970. The result is that the Texas market is overseen largely by the Public Utility Commission of Texas rather than FERC, apart from certain federal reliability rules.

That independence has advantages: Texas sets its own rules, and it has seen rapid growth in new plants. But it has a cost. In February 2021, a severe winter storm took a large amount of generation offline at the same time that demand surged, and millions of Texans lost power for days. Because the Texas grid has limited connections to its neighbors, it could not import much help. A tie to a bigger neighbor is like having a friend with a spare generator down the road. Texas had a few links and a long driveway.

Federal and state: who regulates what

Here is a clean way to remember how the Federal Power Act divides the work. Washington, through FERC, regulates wholesale sales and interstate transmission. States, through their public utility commissions, regulate retail rates, the distribution system, and for the most part the siting of power plants. That is federalism in action. Your electric bill is set by your state commission, which is why your state commissioners matter more to your pocketbook than you may realize. If you want to see how a different set of rural lights went on in the first place, read Light Came First, and What Rural Electrification Teaches About Broadband.

Capacity, reserves and the arithmetic of a hot day

Operators plan for the hardest hour of the year, not the average. Two ideas matter.

Capacity is the amount of power the system can produce at a given moment. Peak demand is the highest amount customers draw. The reserve margin is the cushion: capacity minus peak demand, divided by peak demand.

Here is a hypothetical with invented numbers. A region has 120 gigawatts of capacity and a forecast peak of 100 gigawatts.

• Reserve margin = (120 − 100) ÷ 100 = 20 percent.

Now plants break down. Suppose 15 gigawatts are unavailable on the hottest day, because of maintenance, forced outages or fuel problems.

• Available capacity = 120 − 15 = 105 gigawatts.
• Reserve margin = (105 − 100) ÷ 100 = 5 percent.

Then a heat wave pushes demand to 112 gigawatts.

• Shortfall = 112 − 105 = 7 gigawatts, which is 7 ÷ 112 = 6.25 percent of demand.

When there is a shortfall, operators ask people to conserve, and if that does not suffice they cut power to some customers in rotating outages. Someone's refrigerator and someone's oxygen machine go dark. The cushion is not a luxury. It is the difference between a hot day and a crisis.

Notice the other lesson: capacity is not the same as what a plant will actually deliver when you need it. A plant's nameplate rating is what it can produce under good conditions. What matters is what it can be counted on to produce during the worst hour.

Dispatchable, in plain English

Now the word I promised. A dispatchable power source is one the operator can turn up or down on command, within its limits. A natural gas plant can be started when demand rises. A hydroelectric dam with a reservoir can release more water. A coal plant can be ramped, more slowly. A nuclear plant is usually run steadily, but it is a firm source that is there when called upon, day or night.

A variable source, such as wind or solar, produces when the wind blows or the sun shines, not when the operator calls. That is not a moral failing. It is a feature of the resource. Solar panels produce plenty in the afternoon, and that is useful, because in many places demand peaks on summer afternoons. But on a winter evening after sunset, solar output is nil, and demand may be at its highest. System operators in California have a name for the shape this creates: the "duck curve," in which the need for other power sources climbs steeply as the sun goes down and household use rises.

Batteries are helping, and their role is growing. They are excellent at moving energy across a few hours. The harder question is a long stretch, a cold, still, cloudy spell lasting days. That is the case the system has to survive, and it is why dispatchable plants matter.

The strongest objection

Let me give the other side its best argument, because it is a good one. Grid operators already manage variable demand all day, and they are skilled at managing variable supply. Forecasting has improved. Wind and solar are now among the cheapest sources of new energy in many places. Batteries are falling in price. A larger, better-connected grid smooths out local weather. Demand can flex too, as some customers agree to cut use at peak times for a payment. A diverse mix is the cheapest and most reliable portfolio, they say, and warnings about reliability are scare tactics from people who want to protect old plants.

Much of that is true, and I will not deny it. I am not asking anyone to ban wind or solar. They are real and useful. But here is the answer. A portfolio is judged by its worst hour, not its average hour. The cost of a blackout does not average out. What you need to ask of any plan is a plain question: what keeps the lights on at six o'clock on the coldest night of the year, when the wind is calm? If the answer is "batteries that last four hours," ask what happens in hour five. If the answer is "imports from the neighbors," ask what the neighbors are doing that night too.

Why it matters more now

The new factor is demand. Large new customers, including data centers, can add a lot of demand in a short time, while a power plant or transmission line can take years to build. The faster demand grows, the more it matters that supply be planned, firm and affordable. If you want to follow where that goes for your own pocketbook, Trace Your Family's Energy Dollar From the Meter and Pump to the Grocery Bill walks through your bill. For the problems of building firm power, read Why America Stopped Building Reactors and What the NRC Actually Does and France Built Reactors, Germany Shut Them, and America Argued in a Side-by-Side. For a report card on each source of power, see All of the Above, Graded on What a Family Actually Needs From Power. And on who gets to decide how the rules are written, The Court Did Not Ban Clean Air. It Told EPA to Ask Congress First. covers the fight over EPA's reach into the grid.

The bottom line

• The grid has to balance supply and demand every second, at 60 hertz. When it fails, it can fail by cascade, as it did in the Northeast in August 2003.
• Operators are judged on the worst hour. Capacity, reserve margins and dispatchable power are the numbers that decide whether the lights stay on in a heat wave or a cold snap.
• States regulate your retail rates through public utility commissions, and Washington regulates interstate transmission and wholesale sales under the Federal Power Act of 1935. Attend a commission hearing, or at least read your commission's website.
• Ask any plan for energy a plain question: what happens on the coldest, stillest evening of the year? Then ask your representatives to back firm, reliable power alongside everything else.