🕐10 min read
In This Article
Control Room B, Unit 3
I’m a licensed reactor operator at Hadleigh Station, which is a pressurized water reactor facility in western Massachusetts. I’ve had my license for nine years. Before that I was a non-licensed operator for four years, and before that I was Navy nuclear — six years on submarines, two deployments. I have been in and around nuclear power plants for most of my adult life, which is to say: I am the kind of person who is made less anxious by complex technical systems, not more. I find the instrumentation reassuring. I find the procedures reassuring. The plant makes sense to me in a way that most of the world doesn’t.
I want to establish that before I tell you the rest of this, because I know how it sounds.
The night shift at Hadleigh runs from midnight to eight AM. Control Room B handles Unit 3, which is one of two units at the site. A typical overnight watch consists of two licensed operators, a shift supervisor, a non-licensed auxiliary operator, and a security roving patrol. The control room itself is a significant space — a curved panel arrangement, maybe forty feet wide, with hundreds of indicators covering temperature, pressure, flow, valve position, chemistry, electrical output. You learn to read the board the way you learn to read a face — pattern recognition, knowing what normal looks like so that abnormal stands out immediately.
The abnormal readings started appearing in March of my second year on nights. They were in a section of the board I’ll call the Station Service panel — a set of instrumentation covering auxiliary building systems, not the reactor directly. Temperature indicators, pressure gauges, that kind of thing. Mundane plant systems. For more listening ideas, check out our creepy audiobooks to listen to.
The First Reading
I was doing my hourly panel scan at 2:15 AM when I noticed a temperature indicator that was reading 94 degrees Fahrenheit. The indicator was for a space designated in our system as AUX-7B — Auxiliary Building Section 7B, which our documentation listed as a “mechanical chase area, steam supply.” Normal temperature for an unconditioned mechanical space in March would be ambient outdoor plus a few degrees from pipe radiation — call it maybe forty-five degrees. It was reading 94.
My first thought was a failed instrument. Instruments fail. Temperature probes short out, signal conditioning cards go bad. We log it, we tag it for calibration, we move on. I logged the reading and left a note in the shift log for the maintenance supervisor.
Maintenance came the next day and calibrated the instrument. It checked out fine — no fault found in the thermocouple or the signal conditioning. They verified the reading at the time of calibration: 68 degrees, ambient daytime temperature. Normal.
That night, the same indicator read 91 degrees at 2:20 AM.
Over the next two weeks, the readings came in on most overnight shifts, always between about 2 AM and 4 AM, always on the AUX-7B instruments. Not just temperature — a pressure indicator for the same section showed readings that were, technically speaking, impossible: negative pressure in a non-sealed space, which meant either the gauge was reading below the range of physical reality or something was producing a vacuum in an area with no vacuum-producing equipment. A flow indicator showed readings for a pipe that was, according to our latest system diagrams, decommissioned — valves closed and locked, no flow possible.
Three different maintenance teams looked at it over two weeks. Every instrument checked out fine. No faults, no wiring issues, no calibration errors. The instruments were reading correctly. What they were reading was the problem.
What Yevgeny Told Me
The shift supervisor for the mid-week overnight at that time was Yevgeny, who has been at Hadleigh Station since 1993 — he was here when Unit 3 came online. He’s seen every iteration of the plant’s instrumentation systems, multiple refueling outages, a significant flood event in 2008 that took Unit 2 offline for two months, and every other thing that can happen to a plant over thirty years. He is the calmest person I have ever worked with in a professional environment and he is specifically calm about technical anomalies, which he regards with methodical skepticism.
I brought my log of the AUX-7B readings to Yevgeny in the third week. I laid out the dates and times and values. He looked at the log for a while. Then he said: “AUX-7B.”
“Yes,” I said.
He was quiet for another moment. Then he said: “Come with me.”
He took me to the site library — a room off the administrative corridor with bound copies of every plant modification package, every drawing revision, every system change since the plant was licensed. He found a binder from 2010 and opened it to a package designated Mod 10-047. The cover page described a site modification: demolition of Auxiliary Building Section 7, comprising the original steam supply chase and associated mechanical space, due to obsolescence and site reconfiguration. The modification had been completed in 2011. The space no longer existed.
He showed me the drawing — a site plan from 2011, marked “AS-BUILT POST-MOD 10-047.” The area where AUX-7B had been was now a paved equipment laydown area on the north side of the turbine building. There was a parking area for a mobile crane. There was nothing there.
He turned to me. He said: “The instruments are reading a building that we tore down fifteen years ago.”
I looked at the drawings. I looked at my log. I asked him how that was possible.
He said: “I don’t know. The same thing happened with the Unit 1 decommission in 2004 — we got readings on the old Unit 1 auxiliary systems for about a year after they were demolished. The instrument leads are still in the conduit runs, they just terminate now in sealed junction boxes buried in the concrete pour. My best theory is induction from adjacent wiring. But it doesn’t explain the pattern in the values.”
I asked what the pattern was.
“The temperatures and pressures we’re seeing on AUX-7B,” he said, “match normal operating conditions for that section as it existed when it was active. Like the instruments are reading conditions that were present when the space was in use.”
The Second Year of Readings
The AUX-7B readings continued for the better part of two years, always in the 2-to-4 AM window, always on overnight shifts. We documented everything. The readings appeared in our surveillance log as “anomalous legacy instrument behavior, Mod 10-047 interface, under investigation.” Every senior engineer who looked at the wiring traces concluded the same thing as Yevgeny: induction from adjacent leads, no physical fault.
What no one explained was the specificity of the values. Normal induction produces noise — random values, or values that track nearby systems in some proportional way. What we were seeing was internally consistent: temperatures that would make sense for a steam supply chase under normal load, pressures that would make sense for a condensate drain line at normal operating conditions. Not random noise. Conditions.
The non-licensed auxiliary operator on our shift, a young man named Darryl who’d been at the plant for less than a year, asked me once, during a quiet stretch, whether I thought the instruments were reading something real. I said I didn’t know what real meant in that context. He said: “Like, what if the space is still there somehow. Like, there’s a version of it.”
I told him I didn’t think that was a useful engineering hypothesis. He agreed it wasn’t. We went back to our panel scans.
But I’ve thought about it since. Other people who work nights in isolated technical facilities describe this kind of thing — readings that don’t correspond to current physical reality, instruments registering conditions in spaces that have been modified or demolished. I’ve heard similar accounts from people who work night shifts in industrial facilities about systems that behave as though the original building layout is still present. I don’t know what it means. I know the instruments were accurate. I know the space is gone.
As an Amazon Associate, we earn from qualifying purchases. This post contains affiliate links.
When the Readings Stopped
The AUX-7B readings stopped in November of the second year. No gradual decrease — just present one night and absent the next, and absent every night after that. Maintenance did a full wiring trace following the cessation and found that a junction box in the conduit run — one of the sealed termination boxes from the 2011 modification — had degraded and the leads inside had finally corroded through completely. They replaced the junction box and sealed it with the leads terminated. Possible induction path eliminated.
The readings didn’t return. The engineering team closed out the investigation as “resolved — legacy wiring induction, probable.” I noted in my shift log that the anomaly had ceased and that the wiring modification had been completed, and I left it at that.
I still work the overnight shift. The panel looks the same as it always has, curved rows of indicators, most of them green, some amber, occasionally a red that requires attention. I know the normal values for every indicator on that board. I know what the board looks like at 2 AM on a typical night.
The AUX-7B indicator cluster is still on the panel. Nobody removed the physical gauges — they’re part of the panel structure and removal would be a whole project. The faces read zero now, every one of them. Which is what they should read when there’s nothing connected.
Some nights I still glance at them. Old habit. They’re always at zero. A demolished space, reading as demolished.
I don’t know what they were reading before. I know what the logs say. I know what Yevgeny told me. I have my own hypothesis but it’s not an engineering hypothesis and I don’t put it in the shift log. Some things aren’t shift-log things.
More from the Night Shift
What’s a typical night shift like at a nuclear power plant?
A typical night shift at a nuclear power plant like Hadleigh Station runs from midnight to eight AM. The control room team includes two licensed operators, a shift supervisor, a non-licensed auxiliary operator, and a security roving patrol. The team monitors hundreds of indicators covering temperature, pressure, flow, and more to ensure the plant runs smoothly.
How do reactor operators learn to monitor the plant’s systems?
Reactor operators learn to monitor the plant’s systems through pattern recognition. They study the instrumentation and learn what normal readings look like, so they can quickly spot abnormal readings. It’s similar to learning to read a face – you know what’s normal, so you can tell when something’s off. This skill takes time and practice to develop.
What kind of training do reactor operators have?
Reactor operators have extensive training and experience. The author, for example, was Navy nuclear for six years, including two deployments on submarines. They also obtained their reactor operator license after four years as a non-licensed operator. This background helps them feel comfortable with complex technical systems and procedures.
What’s the most important thing for a reactor operator to notice during their shift?
The most important thing for a reactor operator to notice during their shift is any abnormal readings. Even small changes can indicate a larger issue. The author recalls noticing unusual temperature indicators on the Station Service panel, which prompted further investigation. Being vigilant and spotting anomalies is crucial to ensuring the plant runs safely and efficiently.
You Might Also Like
Stories From the Graveyard Shift
True stories from nurses, truckers, hotel clerks, and security guards who work while the world sleeps. Weekly dispatch.
No spam. Unsubscribe anytime.