Quantifying the Availability of Synchronized and Non-Synchronized Generating Units When Needed

Imagine the power grid as a massive, invisible orchestra playing a song that never stops. Every time you flip a light switch, the conductor (the grid operator) needs to make sure there are enough musicians ready to play the next note perfectly. Some musicians are already standing on stage, instruments in hand, eyes on the sheet music, waiting for the cue. These are the "synchronized" units; they are humming along with the grid, ready to instantly boost the volume if the song gets loud. Other musicians are in the green room, sleeping or tuning their instruments. They are the "non-synchronized" units. If the conductor suddenly needs them, they have to wake up, grab their gear, walk onto the stage, and get in sync before they can play a single note.

For a long time, everyone in the power industry just felt that the musicians on stage were more reliable than the ones in the green room. It seemed obvious: if you're already playing, you're less likely to trip over your own feet than if you have to run onto the stage from the dressing room. But in the world of engineering, "feeling" isn't enough. You need hard numbers. You need to know exactly how much more likely the green-room musician is to trip, and how much that risk changes the safety of the whole concert. This is where the paper by Yufan Zhang and Feng Zhao steps in. They wanted to turn that gut feeling into a precise mathematical score, creating a way to measure exactly how often these two types of "musicians" fail to show up when the conductor raises their baton.

The authors of this paper decided to stop guessing and start measuring. They created two new "failure scores" to compare the reliability of these two groups. Let's call them the "On-Stage Score" (for synchronized units) and the "Green-Room Score" (for non-synchronized units). The paper asks a simple question: If the grid needs power right now, what is the chance that a unit already running will suddenly stop, versus the chance that a unit sitting idle will fail to start and get ready in time?

To figure this out, the researchers used a mathematical model that tracks the "mood swings" of a power generator. Think of a generator as having four possible moods: it's either chilling in the green room (reserve shutdown), it's having a bad day and hiding in the corner (forced outage when not needed), it's performing perfectly on stage (in-service), or it's having a bad day while the audience is watching (forced outage when needed). By tracking how often generators jump between these moods, the team calculated the odds of failure for both types.

Their findings confirmed what the grid operators suspected, but now with a ruler and a calculator. They found that the "Green-Room Score" is generally higher and more unpredictable than the "On-Stage Score." In plain English, units that have to start up from scratch are more likely to fail when needed than units that are already running. The paper suggests that for synchronized units, the main risk is simply breaking down while they are already working (like a violin string snapping during a solo). However, for non-synchronized units, the biggest risk is the act of starting itself (like a musician forgetting their sheet music or tripping over the curtain while running to the stage).

The researchers tested their new formulas using real data from 222 power plants in the New England area, looking at records from 2016 to 2025. They set a specific time limit for the test: 15 minutes. This is the window of time the grid has to decide if a unit can be counted on to help. When they crunched the numbers, the results were clear. For most types of power plants—like the big steam turbines and the gas-fired ones—the non-synchronized units had a higher chance of failing to be ready. For example, the average failure probability for non-synchronized steam units was about 0.0178, while their synchronized counterparts were much lower at 0.0008.

There were a few interesting exceptions, though. Nuclear units, for instance, showed a zero failure rate for starting up in the data they looked at, simply because they were rarely asked to start from a cold stop during that time, and every time they did, they succeeded. Diesel units were a bit of a mystery, likely because they are used so rarely that it's hard to get a clear picture of their reliability. But overall, the data painted a consistent picture: being already synchronized gives you a significant reliability boost, supporting the operational intuition that non-synchronized units are generally less reliable.

The authors are careful to note that their current "score" only counts total failures, like a musician completely dropping out of the orchestra. They didn't count the times a musician played but was a little out of tune (partial power loss), which is a limitation they plan to fix in future studies. But for now, they have successfully turned a vague belief into a concrete number. They showed that while the green room is necessary for having enough musicians, the ones already on stage are statistically the most reliable when the music needs to get louder, faster. This helps grid operators make smarter decisions about how much "spare" power they need to keep in reserve, ensuring that when the lights go on, the orchestra never misses a beat.