The Slow Explosion Problem: Why Grid-Scale Battery Storage Is Misclassified
Battery Plant Explosions in California | Here's What's Happening - YouTube
What Moss Landing Actually Revealed
The January 2025 fire at Vistra Energy's Moss Landing facility was the largest battery storage incident in U.S. history. A recent podcast explored its consequences — sometimes accurately, often not. But the most important failure may belong to the regulatory framework itself: treating a self-oxidizing slow explosion as an ordinary fire.
The Moss Landing battery fire of January 16, 2025 was a genuine industrial disaster: peer-reviewed science confirms substantial heavy-metal contamination of adjacent wetlands and farmland, hundreds of residents reported acute health effects, and long-term carcinogenic risk from cobalt and nickel nanoparticles cannot be dismissed. A California Insider podcast explored these harms — but embedded them in a pattern of significant factual errors, including a fabricated incident count, a false claim that wind and solar have the worst lifecycle greenhouse-gas emissions of any energy source, and an assertion that offshore wind has killed over 5,900 whales that is contradicted by every federal agency responsible for marine mammal protection.
The legitimate scientific story is serious enough without embellishment. More importantly: the central technical insight that most commentary has missed — including initial regulatory framing — is that large-format lithium-ion thermal runaway is not a fire. It is a slow explosion: self-oxidizing, chemically unstoppable, and categorically different from any hazard addressed by current fire codes or emergency response protocols. That misclassification has consequences for siting, evacuation radii, first responder safety, agricultural protection, and the RO-RO shipping crisis now unfolding in maritime transport.
What Actually Happened at Moss Landing
At 3:00 p.m. on January 16, 2025, thermal runaway initiated in one or more battery modules inside the Phase 1 building of Vistra Energy's Moss Landing Power Plant — a 300-megawatt / 1,200-megawatt-hour lithium-ion battery energy storage system (BESS) built inside the repurposed shell of a 1950s PG&E turbine hall. Within hours, 75–80% of the approximately 100,000 battery modules in that building had undergone cascading thermal runaway. Approximately 1,200 nearby residents were evacuated. The event burned for the better part of three weeks, with recurring flare-ups as successive cell groups initiated. First responders correctly stood down from active suppression. The cause remains officially undetermined as of this writing.
The facility used nickel-manganese-cobalt (NMC) lithium-ion chemistry — an older, higher-energy-density formulation with a lower thermal runaway onset temperature than the lithium iron phosphate (LFP) chemistry now standard in most new grid-scale installations. Its indoor, non-compartmentalized configuration meant that once initiation occurred, heat transfer to adjacent racks was effectively unimpeded. EPA monitoring detected no hydrogen fluoride above California's ambient standards at the perimeter. However, subsequent peer-reviewed research published in Scientific Reports documented approximately 25 metric tons of heavy-metal particles — cobalt, nickel, and manganese in ratios precisely matching NMC cathode chemistry — deposited across roughly half a square mile of adjacent Elkhorn Slough wetlands. Particle sizes were sub-100 microns, well within the range capable of deep pulmonary penetration. Vistra wrote off approximately $400 million in plant value. EPA entered a formal cleanup agreement with Vistra in July 2025; battery removal continued through late 2025.
The Central Analytical Error: Calling It a Fire
Every official document — Vistra's statements, EPA releases, Monterey County communications, peer-reviewed studies — describes the Moss Landing event as a "fire." The California Insider podcast guest, Prof. Michael Hogan, insisted it was an "explosion." Both characterizations miss the most important technical point.
Lithium-ion thermal runaway is neither a conventional fire nor a conventional explosion. It is a slow explosion — a term that maps precisely onto established energetics taxonomy and carries regulatory consequences that neither "fire" nor "explosion" alone fully captures.
A Practical Taxonomy of Explosive Events
- HIGH EXPLOSION detonation - Supersonic reaction front. The detonation wave outruns the pressure wave it creates, producing an instantaneous shock. Surrounding material has no time to respond mechanically — it disintegrates. Energy release measured in microseconds. Conventional suppression is meaningless; the event completes before any response is possible. TNT · RDX · PETN · ANFO · shaped charge
- LOW EXPLOSION - deflagration - Subsonic combustion front. Reaction propagates below the speed of sound, producing rapid but not instantaneous pressure rise. No shock wave, but still violent enough to destroy structures. Energy release measured in milliseconds. Can sometimes be vented or suppressed if the geometry is favorable. Black powder · vapor cloud deflagrations · some dust explosions
- SLOW EXPLOSION - thermal runaway - Self-oxidizing, self-sustaining exothermic cascade. The oxidizer is internal to the reacting material — it cannot be starved of oxygen. Total energy release is thermodynamically fixed and complete. Individual pressure-release events (cell venting, electrolyte deflagration) recur throughout the cascade. The temporal envelope is seconds to hours, which superficially resembles a fire — but the chemistry is closer to a solid-rocket propellant burn than to hydrocarbon combustion. Conventional suppression is physically impossible, not merely ineffective. Li-ion thermal runaway · BLEVE (partial analog) · some runaway chemical reactions · smoldering dust events transitioning to deflagration
The "slow" qualifier is doing critical work. It distinguishes the millisecond timescale of a deflagration from the minute-to-hour timescale of a cascading BESS — but the violence per unit time during an individual cell's runaway event is genuinely explosive in character. More importantly, the defining characteristic shared with high and low explosives — that the energy release is complete, self-sustaining, and cannot be chemically interrupted — is fully present. A NMC cell in thermal runaway will release its full electrochemical energy inventory regardless of any external intervention. The only viable strategy is thermal isolation of adjacent cells, not suppression of the initiating cell.
The total stored energy of Moss Landing's Phase 1 system — 1,200 megawatt-hours — is approximately 4.3 petajoules. The fraction that released during the event, distributed over roughly three weeks of cascading activity, averages to enormous instantaneous power densities during each individual runaway event, even though the aggregate temporal envelope is long.
Why Water and Foam Fail: The Chemistry
Understanding why conventional suppression is futile requires a brief account of what is actually happening inside a cell in thermal runaway. The cathode material in NMC batteries is a layered oxide (LiNiₓMnᵧCoₙO₂). As it heats above approximately 130–180°C, the crystal structure destabilizes and begins releasing oxygen directly from the lattice — independent of any atmospheric oxygen supply. This oxygen reacts exothermically with the organic electrolyte and the carbonaceous anode, generating more heat, which drives further cathode decomposition, releasing more oxygen. The feedback loop is self-amplifying and cannot be broken by removing atmospheric oxygen.
Water application does not extinguish the reaction. It produces steam, which can accelerate electrolyte hydrolysis, generating additional hydrogen fluoride. It can create hydrogen gas through reaction with residual lithium. It can thermally shock adjacent cells and potentially initiate additional runaway events. The appropriate use of water in a large BESS incident is not suppression of burning cells but thermal management of adjacent cells — keeping them below onset temperature to prevent cascade propagation. This requires sustained high-volume application, often for days, and represents total loss of the initiating cell regardless.
Thermal Runaway Off-Gas Profile — Respiratory Hazards
During thermal runaway, a cell vents a complex mixture requiring supplied-air respiratory protection (SCBA/OBA) for all personnel in enclosed spaces or within the plume. Filter respirators are inadequate. Key species include:
- Hydrogen fluoride (HF) — from LiPF₆ electrolyte decomposition; IDLH 30 ppm; penetrates skin as well as lungs; extremely dangerous
- Carbon monoxide & CO₂ — from organic electrolyte combustion; classic asphyxiant
- Vinylidene fluoride and organic fluorides — from PVDF binder decomposition; pulmonary irritants
- Metal oxide nanoparticles (Co, Ni, Mn oxides) — sub-100 micron; penetrate to alveolar level; cobalt classified IARC Group 2A carcinogen; particles are jagged (fracture morphology), not rounded (combustion morphology), consistent with mechanical rupture events
- Hydrogen cyanide — from nitrogen-containing electrolyte additives in some formulations
- Hydrogen gas — flammable; contributes to secondary deflagration events
The RO-RO Ship Crisis: Slow Explosions at Sea
The most instructive case studies for understanding large-scale lithium-ion thermal runaway events are not stationary BESS facilities but roll-on/roll-off (RO-RO) car carrier vessels transporting battery electric vehicles. These incidents have demonstrated in stark, irreversible terms the futility of conventional maritime firefighting against self-oxidizing chemistry.
The Felicity Ace, carrying approximately 4,000 Volkswagen Group vehicles including Porsche, Audi, and Bentley models, caught fire in the Atlantic in February 2022 and burned for two weeks before sinking — a total loss of vessel and cargo worth an estimated $400–500 million. The Fremantle Highway, carrying 3,783 vehicles, caught fire off the Dutch coast in July 2023, resulting in one crew fatality, 22 injuries, and the near-loss of the vessel. In both cases, the ship's built-in CO₂ flooding systems — standard suppression for conventional vehicle fires — were effectively useless against the self-oxidizing battery chemistry. The last resort in both incidents was massive seawater flooding of the vehicle holds, which takes days to execute and results in complete loss of the affected decks.
The International Maritime Organization (IMO) and classification societies including DNV and Lloyd's Register issued interim guidance in 2023–2024 on BEV transport by sea, but the fundamental problem — that no currently available suppression system can interrupt thermal runaway in a cell that has initiated — has no clean engineering solution at present battery technology. The IMO's Maritime Safety Committee has been working on updates to SOLAS Chapter II-2, but the pace of fleet BEV adoption has outrun regulatory adaptation in a pattern directly analogous to what occurred with ground-based BESS siting.
The Regulatory Misclassification and Its Consequences
The classification of lithium-ion battery assemblies as Class 9 dangerous goods — a "miscellaneous hazardous materials" catch-all — rather than as explosive-class materials is the most consequential regulatory gap in the current framework. Class 9 classification drives packaging requirements, transport restrictions, vessel and aircraft design standards, and emergency response protocols. It reflects typical-case behavior and does not adequately capture worst-case behavior.
The analogy to solid rocket propellant is instructive. Solid propellants are self-oxidizing materials that burn rather than detonate under normal conditions — yet they are classified as Class 1 explosives. That classification reflects worst-case behavior and the impossibility of conventional suppression, not typical burn rate. A principled argument exists that large-format lithium-ion assemblies, above some threshold energy density and pack size, warrant analogous classification logic — at minimum for transport, storage density limits, and built-environment siting requirements near occupied structures, agricultural land, and coastal ecosystems.
If BESS facilities were classified as slow-explosion hazards rather than fire hazards, the regulatory consequences would include substantially larger mandatory setback distances from occupied structures; blast-rated rather than merely fire-rated construction elements for containment structures; evacuation radii calculated on the basis of plume chemistry rather than smoke visibility; first responder protocols specifying OBA/SCBA as mandatory rather than optional; and agricultural soil protection requirements triggered by siting near food-producing land.
None of these requirements exist in current NFPA 855, IFC 2018, or California's SB 38 safety standards — the frameworks that govern essentially all new BESS development in the United States. The industry has improved dramatically since 2018: the global BESS failure incident rate dropped 97% between 2018 and 2023 according to EPRI/PNNL/TWAICE analysis, and the shift to LFP chemistry and containerized outdoor installation substantially reduces cascade risk. But the fundamental classification error — treating a slow explosion as a fire — remains embedded in the regulatory architecture.
Fact-Check of the California Insider Transcript
With the technical framework established, the specific claims made in the California Insider interview with Prof. Hogan can be assessed on their merits. Several are substantially accurate. Several others are not merely wrong but actively undermine the legitimate scientific concerns that deserve public and regulatory attention.
Partially vindicated by the slow-explosion framework — but not in the way Hogan intended. His characterization is non-standard and would be rejected by most fire safety engineers as applied to this event. However, the deeper insight — that thermal runaway is categorically different from a hydrocarbon fire and that "fire" misrepresents the chemistry — is correct and important. The jagged particle morphology he cites is consistent with mechanical rupture events during cell venting, not uniquely diagnostic of an "explosion" versus a "fire." The correct framing is slow explosion: self-oxidizing, chemically unstoppable, involving repeated pressure-release events throughout a cascade that unfolds over hours to weeks.
Not supported by any available evidence. The EPRI BESS Failure Incident Database — the most comprehensive global record of grid-scale battery incidents — recorded approximately six total global incidents in calendar year 2024. The American Clean Power Association documented only approximately 20 fire-related incidents globally across the entire preceding decade, against a 25,000% increase in installed capacity. No independent source consulted for this review documents eight California battery plant incidents in any single year. The claim as stated appears to be fabricated or represents a severe misreading of the available data, and undermines the credible evidence for the Moss Landing incident's genuine severity.
Confirmed. There were prior smoke and thermal events at the MOSS300 building, including a 2021 incident during which some batteries began to smoke and required fire department response. The adjacent PG&E-owned Elkhorn Battery facility experienced a significant incident in September 2022 that required evacuation and a multi-month operational shutdown. The January 2025 event was, by a substantial margin, the most severe.
Confirmed by peer-reviewed science. The 2025 Scientific Reports study by Aiello et al. documented approximately 25 metric tons of NMC cathode metal particles deposited across roughly half a square mile of Elkhorn Slough wetlands, with nickel-to-cobalt ratios precisely matching battery cathode chemistry. Cobalt is an IARC Group 2A carcinogen. Sub-100-micron metallic particles are persistent in soil and can be taken up by root systems into food crops. The SJSU/EMBER research consortium continues monitoring for food-chain bioaccumulation. Hogan's concern is scientifically grounded, though some specific quantitative claims (century-long persistence, confirmed food uptake) remain uncertain pending further study.
No state subsidy of either figure has been documented. The $400–500 million figure in the public record refers to Vistra's insurance coverage — private, not public. PG&E holds long-term resource adequacy contracts with the facility funded through ratepayer revenues, but these are power purchase agreements, not capital grants. The claim that without a state subsidy "no private entity would have dared build it" is inconsistent with Vistra's status as a large publicly traded power corporation that sought standard regulatory approvals. This claim requires a documented source before it can be credited.
Directly and decisively contradicted by the scientific consensus. The NREL Life Cycle Assessment Harmonization Project — synthesizing thousands of published LCA studies — found lifecycle GHG emissions from solar, wind, and nuclear to be between 400 and 1,000 gCO₂e/kWh lower than coal and natural gas. A landmark study in Nature Energy (Pehl et al.) concluded that "the carbon footprint of solar, wind and nuclear power are many times lower than coal or gas." The IEA's 2024 Life Cycle Upstream Emission Factors database and IPCC AR6 concur. Manufacturing-phase emissions are real and are higher when components are produced with coal-fired electricity — but even under worst-case manufacturing assumptions, wind and solar lifecycle emissions are a fraction of any fossil fuel source. This claim is false without qualification.
No credible scientific or regulatory source supports this figure. NOAA Fisheries, BOEM, the Marine Mammal Commission, the U.S. Department of Energy, and the Government Accountability Office (April 2025) have all stated that no whale death has been causally linked to offshore wind development. The elevated whale mortality events on the U.S. East Coast are real — declared unusual mortality events for humpback whales (2016), minke whales (2017), and North Atlantic right whales (2017) — but necropsy data from 2017–2024 attribute 25 of 30 examined deaths to vessel strikes or fishing gear entanglement. Statistical analyses claiming a wind-survey correlation have not been peer-reviewed and are disputed by marine biologists. The assertion that the White House issued "kill permits for 400 whales" fundamentally misrepresents MMPA incidental harassment authorizations, which cover behavioral disturbance, not lethal taking.
Broadly correct and consistent with expert consensus. Fire safety engineers at EPRI and Pacific Northwest National Laboratory, industry analysts, and the EPRI/PNNL/TWAICE root-cause analysis all agree that the Moss Landing incident reflects design choices — NMC chemistry in a large non-compartmentalized indoor space without modern suppression systems — that are not representative of current best practice. The shift to LFP chemistry raises thermal runaway onset temperature significantly. Modular containerized outdoor installations with passive thermal barriers limit cascade propagation. Updated codes (NFPA 855 2026 edition, IFC 2018) impose requirements the original MOSS300 design predated. These improvements are real and meaningful, though they do not address the fundamental slow-explosion classification problem.
Verdict Summary
| Claim | Verdict | Key Evidence |
|---|---|---|
| "It was an explosion, not a fire" | PARTIAL | Slow explosion framework partially vindicates; framing is non-standard |
| 8 CA battery plants exploded last year | FALSE | EPRI records ~6 global incidents total in 2024 |
| Prior incidents at Moss Landing | ACCURATE | 2021 smoke event, 2022 Elkhorn incident documented |
| Heavy-metal contamination of soils and wetlands | ACCURATE | ~25 metric tons confirmed by Scientific Reports peer-reviewed study |
| Century-long persistence; confirmed food uptake | UNCERTAIN | Plausible but not yet documented; EMBER consortium monitoring ongoing |
| $500M–$1.5B state subsidy for Moss Landing | UNVERIFIED | $400–500M figure refers to Vistra's private insurance, not state subsidy |
| Wind/solar have worst lifecycle GHG emissions | FALSE | Contradicted by NREL, IPCC AR6, IEA, Nature Energy |
| Offshore wind killed 5,900+ whales | FALSE | No causal link: NOAA, BOEM, GAO, Marine Mammal Commission |
| Newer LFP/containerized BESS is substantially safer | ACCURATE | Broad expert consensus; EPRI/PNNL/TWAICE root-cause analysis |
| Water suppression is ineffective on thermal runaway | ACCURATE | Self-oxidizing chemistry; confirmed by fire safety engineering literature |
| SCBA/OBA required for first responders | ACCURATE | NFPA 855, EPRI 2023 guidance; HF and metal oxide off-gas profile |
What Sound Policy Should Address
The slow-explosion framework points to a specific set of regulatory gaps that neither the California Insider interview nor most mainstream policy commentary has identified cleanly.
First, the fundamental hazard classification of large-format BESS facilities should be reviewed by NFPA, ICC, and relevant federal agencies with explicit reference to the self-oxidizing nature of NMC and LFP thermal runaway. The question of whether current fire-code-derived frameworks adequately address a hazard whose suppression is physically impossible — not merely operationally difficult — deserves formal analysis.
Second, agricultural siting restrictions are scientifically warranted. The documented persistence of cobalt and nickel nanoparticles in soils and the potential for crop uptake create a class of harm that conventional fire code setback requirements were not designed to address. The Moss Landing case should drive specific soil contamination modeling requirements for BESS siting near food-producing land.
Third, the maritime sector requires urgent regulatory catch-up. The RO-RO ship incidents demonstrate that the slow-explosion problem is not confined to stationary BESS. IMO SOLAS updates need to grapple with the chemistry, not merely with improved detection and early suppression protocols that assume the fire can eventually be put out.
Fourth, first responder training and equipment standards should be updated to treat BESS incidents as chemical hazard events requiring supplied-air protection from the outset, not as building fires where SCBA is deployed reactively when conditions deteriorate.
None of these measures require halting battery storage development. The 97% reduction in global BESS incident rates since 2018 demonstrates that the technology can be deployed safely at scale. What is required is regulatory honesty about the nature of the hazard — and that begins with calling it what it is.
Verified Sources — Formal Citations
Comments
Post a Comment