Report 180 · Energy Storage
On 18 September, roughly 1,200 battery modules at the Moss Landing plant caught fire on their own, twenty months after the January 2025 blaze. A fire breaking out at a site that burned out almost two years ago sounds like a mystery. It isn't one. A fire does not use up a battery's stored energy, and getting that energy out of a wrecked module is a hard, slow, unfinished engineering problem.
Here is what happened, from the people on scene. Workers for Vistra, the plant's operator, spotted a flare-up early on Friday 18 September against the wall of the building that burned in January 2025. According to Chief Jess Cortez of the North County Fire Protection District, as reported by CalMatters, the cause was roughly 1,200 inaccessible battery modules, still charged, that ignited spontaneously. Monterey County issued a shelter-in-place order and lifted it that afternoon. Air monitors at the fence line had not detected unhealthy particulate levels as of that day's briefing.
The cause of the original fire is still unknown. The California Public Utilities Commission says its root cause analysis is still underway. This report is not about that question. It is about a narrower one that the second fire answers on its own: why a battery building that burned out twenty months ago was still able to burn.
A charged lithium-ion cell stores energy chemically, as lithium sitting in the negative electrode waiting to move. A fire that destroys the building around a module, or even some of the cells inside it, does not necessarily discharge the cells that survive. They keep their charge. They have also often been crushed, soaked, overheated or had their casings breached, which is exactly the kind of damage that can short a cell internally and start the self-heating chain called thermal runaway.
The National Transportation Safety Board wrote this down plainly in 2020, in a safety report on lithium-ion fires in electric vehicles. Its finding:
The NTSB concludes that the energy remaining in a damaged high-voltage lithium-ion battery, known as stranded energy, poses a risk of electric shock and creates the potential for thermal runaway that can result in battery reignition and fire.
The same report shows what that looks like in practice. In one of the cases it studied, an SUV crash in Mountain View, California, the wreck was towed to an impound yard. About 20 minutes after it arrived, an officer heard popping sounds from it. Firefighters came to watch it, and were called back within the hour because the battery had reignited. It reignited again five days after the crash, under a tarp, on a day when investigators had stood on parts of the wreck near the front of the battery case.
The NTSB's next sentence is the one I would want every battery owner to read: "Firefighters have no method of determining whether stranded energy is present in a damaged high-voltage lithium-ion battery or of removing energy from the battery pack." The board noted that engineers can check remaining voltage through the battery management system if it still works, and that some packs have discharge ports. In three of its four cases, crash damage to the battery system blocked access to the management system or the discharge ports. In the Mountain View case, manufacturer engineers found the discharge port full of water and debris.
Now scale that up. EPA says the Moss Landing building held about 100,000 lithium-ion modules, and about 55% were damaged in the January 2025 fire. In the first phase of cleanup, crews reached, removed, de-energized and shipped out nearly 40,000 intact batteries. Everything left was, in EPA's words, "previously inaccessible": the most fire- or water-damaged material, in parts of the building that have to be demolished to reach. My own arithmetic, assuming EPA's "batteries" and "modules" are the same unit: that leaves on the order of 60,000 modules' worth of material still on site when the final phase began, and the 1,200 that ignited would be about 2% of it.
Note what those remaining modules are. They are the ones nobody could get to. That means nobody could read their state of charge either, and for many of them there is no working battery management system left to ask.
EPA's July 2026 fact sheet for the cleanup says unstable batteries "will be submerged in a brine solution," and that crews use thermal imaging to watch for heating. Brine discharge is the standard field idea: the salt water is conductive, so it slowly shorts the terminals and bleeds the cell down. It does work to a degree. The research says it is far less tidy than the phrase suggests.
A 2024 open-access study in Sustainable Energy & Fuels by a team at Aalto University and the University of Turku in Finland is blunt about the state of the field. The authors write that they had tested the standard salt-solution claim and "observed rapid corrosion of the LIB poles in various aqueous salt solutions," and that the corrosion, especially in sodium chloride, "inhibits both the electron transfer and the accurate measurement of voltage, while the batteries remain charged." In other words, the terminals can dissolve before the cell is empty, and the voltage reading you would use to confirm it is empty becomes unreliable. They add that many recycling reviews accept salt-solution discharge "as a fact without any critical evaluation."
The second problem is rebound. Pull a cell out of the bath and its voltage climbs back up. In their own experiments, a cell in an iron-salt solution fell to about 2.5 V after 100 hours, then recovered to 2.8 V. Even their best new electrolyte, which took a cell down to 0.7 V in 96 hours, saw the voltage settle back at 2.4 V once it was removed. They cite a voltage below 2.5 V as the preferred level before a cell is opened or crushed.
One important limit on that study: it used new, undamaged 18650 cells from consumer electronics, not burned grid modules. I am not claiming EPA's contractors are doing it wrong, and I have not seen their protocol. The point is only that "de-energize" names a slow process with known failure modes, not a switch. Fire-damaged modules in collapsed racks are a harder version of the same problem, not an easier one.
It does not say every battery site is a standing hazard. Two experts quoted by CalMatters, Dustin Mulvaney of San Jose State University and Denise Grab of the Emmett Institute on Climate Change and the Environment, both pointed to Moss Landing's design: an older facility inside an enclosed building, with older battery technology packed closely together. Grab also noted that fossil fuel plants carry serious health risks of their own. Design matters, and I wrote about how the industry's fire rate has fallen in what the 97% drop in battery fires means.
What the second fire does show is that the hazard of a lithium-ion fire has a long tail. The flames end in days. The stored energy in the damaged cells stays until someone removes it, cell by cell, and that work can take years. EPA's current projection is that battery removal and demolition at Moss Landing will finish in mid-2027.
For anyone buying or running a battery system, the practical lesson is about access and information. A system you can monitor down to the module, isolate in sections, and physically reach after an incident is far easier to make safe than one you can only watch from outside the fence. That is part of what I work on. I help design the AI battery-cycling systems for a veteran-owned (HUBZone) energy-storage integrator, run by fellow veterans. I do not own that company and earn nothing from this link. Full policy here. Nothing in this report involves them or any product they sell.
I have not seen Vistra's or EPA's discharge protocol, so I make no claim about how the Moss Landing modules are being treated beyond what EPA's public fact sheet says.
I make no claim about what triggered the 18 September ignition. EPA announced the start of building demolition and removal of the remaining batteries on 8 September. The fire came ten days later. Timing alone is not a cause, and no official has linked the two.
The NTSB report is about electric vehicles, not grid storage. I use it for its finding on stranded energy in damaged lithium-ion batteries, which is about the cells, not the vehicle.
Burning does not empty a lithium-ion battery. The cells that survive keep their charge, often while badly damaged, and the NTSB found that responders have no method to measure or remove that stranded energy from a wrecked pack. Salt-water discharge, the standard fix, corrodes terminals and lets voltage rebound. At Moss Landing, the modules no one could reach were the modules no one could drain, and about 1,200 of them caught fire twenty months later. The fire was not a mystery. It was the leftover energy.
Scope note: the estimate of roughly 60,000 modules' worth of material remaining and the 2% share are the author's arithmetic from EPA's published counts, assuming EPA's "batteries" and "modules" refer to the same unit. Labeled as such in the text.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.