How the Lights Stay On

Chapter 12: The Grid Grows a Brain

On a hot July afternoon in California, the grid is approaching its limits.

It is 4 PM. The temperature is in the high 90s across most of the state. Air conditioners are running flat out. Solar panels, after a long bright day, are beginning to lose their output as the sun drops lower in the western sky. The duck curve from Chapter 6 is rising into its evening neck. Power plants across the state are ramping up. Reserves are getting thin. The grid operators at CAISO are watching their screens carefully.

Twenty years ago, this would have been the moment they made an urgent phone call. Bring everything you have online. Reserve gas plants would have started spinning. If those were not enough, the operators would have started preparing rolling blackouts.

Today, something different happens. Across California, in homes the operators will never visit, signals begin to fan out through the internet.

A few thousand Nest thermostats receive a message to nudge their air-conditioner set point up by two degrees. Most homeowners do not notice. The houses stay comfortable.

About twenty thousand Tesla Powerwalls, sitting on suburban garages, begin discharging the solar electricity they have been storing all day. Each one is pushing about three or four kilowatts back onto the grid. The combined output, just from those Powerwalls, is roughly equivalent to a small natural gas plant.

Several hundred large industrial customers — manufacturing plants, cold storage warehouses, water treatment facilities — automatically reduce their non-critical loads in response to a pricing signal that just hit their energy management systems.

A few thousand electric cars, plugged in at homes and offices across the state, briefly stop charging. They will resume in an hour or two when the evening peak has passed.

The grid balances. The lights stay on. No emergency. No blackout. No one in the affected homes had to do anything except, possibly, glance at the thermostat and wonder why it suddenly said 76 instead of 74.

This is the smart grid. It is real. It is partially deployed. It is one of the most important shifts in how electricity works since the invention of the grid itself.

To understand why the smart grid matters, you have to understand how dumb the old grid was.

For most of the twentieth century, the electrical grid was a one-way pipe.

Power flowed in one direction: from generators (coal, gas, nuclear, hydro) through transmission lines, through substations, through distribution wires, into homes and businesses. The homes and businesses were passive consumers. They drew power when they wanted it. The grid had no way to know who was drawing what, except by reading meters once a month with a person walking around in a uniform.

The grid had almost no awareness of itself. The operators in the control room from Chapter 2 could see what their own generators were doing, and roughly what was happening at the major substations. They could not see, in real time, what was happening at the level of neighborhoods or individual customers.

When the lights went out, the utility usually found out because someone called.

This was not because the engineers were lazy. It was because the technology to do anything else did not exist for most of the grid’s history. You cannot have a smart grid without computers, sensors, GPS time-synchronization, fast wireless communication, and electricity-cheap microchips. None of those things existed in 1930. Only some of them existed in 1990. All of them exist now.

The smart grid is what happens when you put modern computing, communications, and sensing onto a system that was originally built without any of those things.

The first major piece of the smart grid was the phasor measurement unit, or PMU.

A PMU is a sensor about the size of a shoebox. It sits inside a substation or at a key point on the transmission grid. It measures the voltage and current of the grid up to 120 times per second. And — this is the key part — every measurement is tagged with a GPS time stamp accurate to the microsecond.

GPS time stamps are what make PMUs powerful. With them, a PMU in California and a PMU in New York can compare their measurements down to the millisecond. Operators can see, in real time, exactly how the grid is behaving across the whole country. They can detect wobbles in frequency, voltage sags, oscillations that did not exist before — all of which used to happen invisibly, in the seconds before a cascade, with no one watching.

The push to install PMUs began seriously after the August 2003 blackout (Chapter 4). One of the lessons of that disaster was that the grid had been blind. The operators in Ohio could not see what was happening in their own system, let alone across state lines. A US Department of Energy program began funding PMU deployment in the late 2000s. By 2026, a typical American electric utility has dozens to hundreds of PMUs streaming data into its control room, every minute of every day.

If you remember Chapter 2’s comparison of the grid to a heartbeat, PMUs are what allow operators to see the heartbeat clearly for the first time. Where the older system gave them x-rays — useful but slow — the PMUs give them an electrocardiogram, updating in real time.

The second major piece of the smart grid is the smart meter.

All 18 chapters of How the Lights Stay On · Library · All courses