🏛️ Issues and Advocacy Training
Courses › Battery Energy Storage Systems: How They Work, Fail and Burn

Lesson 9 of 9 · about 19 minutes

Environmental Damage, Cleanup and What a County Can Require

You'll be able to

  • Describe the air, soil and water hazards from a battery fire with real measurements
  • Summarize the Moss Landing testing record fairly, including where results disagree
  • Explain how damaged and retired batteries are handled and why decommissioning plans and bonds matter
  • List the conditions an Idaho county can attach to a battery permit and the limits on that power

When a battery plant burns, the fire ends and the questions start: what was in the smoke, where it landed, what washed into the drains, who tests it and who pays to haul away tens of thousands of burned modules. Moss Landing is the only large case with a public, multi-agency record, so most of this lesson leans on it. The siting half then turns to what an Idaho county can require so that it never has to find out the hard way.

Air: what the smoke carries

Lesson 6 listed the gases. The one that drives most health concern is hydrogen fluoride, which comes from the fluorinated salt in the electrolyte. In Larsson and colleagues' 2017 Scientific Reports study of seven commercial lithium-ion cell types, burning cells released 20 to 200 milligrams of HF per watt-hour of battery capacity, plus 15 to 22 mg/Wh of phosphoryl fluoride in some tests.

Worked example: the HF ceiling for one container

Apply the lab range to an invented 5 MWh container (5,000,000 Wh) that burns completely.

Low end:  5,000,000 Wh x 20 mg/Wh  = 100,000,000 mg = 100 kg HF
High end: 5,000,000 Wh x 200 mg/Wh = 1,000,000,000 mg = 1,000 kg HF

A one-container fire at a 160-container site:
  1 of 160 = 0.6% of the site's energy

Moss Landing 300 at 1,200 MWh, about 56,000 of
100,000 modules burned (EPA): roughly 56% of the energy

These are ceilings from small-cell lab tests, not predictions of what reaches a neighbor. HF reacts, dilutes and deposits; how much travels depends on how hot and high the plume rises and on wind. The arithmetic shows why the number of units that burn matters so much, and why air monitoring belongs in every emergency plan.

Real measurements so far have been reassuring at a distance. EPA ran nine monitoring stations during the January 2025 Moss Landing fire, two at the facility, four just outside it and three in nearby communities, measuring particulate matter and HF, and said on January 18, 2025 that it had not detected any risk to public health based on its data. Near Melba in October 2023, KIVI reported that preliminary Canyon County testing found concentrations within normal levels for the area. Residents near Moss Landing nonetheless reported physical symptoms after the fire and told Santa Cruz Local they felt sidelined by officials saying there was no problem. Monitoring at fixed stations can miss what a person downwind breathed during the worst hours. A UC Santa Cruz toxicologist urged testing of indoor dust and residents' blood, urine and hair; this course found no published human exposure testing as of October 2026.

Soil: the Moss Landing record

The testing record after Moss Landing is a good lesson in how different methods give different headlines.

Who testedWhenWhat they reported
California DTSC and Monterey County environmental healthLate January 2025, 7 sitesResults did not indicate a human health threat; a county official said she would not sound "a total all-clear"
Never Again Moss Landing (residents' group), commissioned samplingJanuary and after the February flare-up, 16 sitesHeavy metals detected in some areas both times; a UC Santa Cruz toxicologist called some manganese, cobalt and nickel levels cause for concern
Vistra's contractor Terraphase, with DTSC oversight and split samplesMay 2025, 27 samplesSamples across northern Monterey County, with DTSC taking its own at about half the sites, for heavy metals and dioxins
Moss Landing Marine Laboratories (Aiello et al., Scientific Reports, Nov. 26, 2025)Before and after the fire, Hester Marsh in Elkhorn SloughSurface nickel up to about 10 times and cobalt about 5 times pre-fire levels; enrichment in the top few millimeters; peak 1 to 3 km downwind; nickel to cobalt ratio about 2 to 1, matching NMC cathode material

Early reports put the rise in marsh surface concentrations at "hundreds- to thousand-fold" over pre-fire levels. The peer-reviewed paper's bulk soil numbers are smaller, roughly fivefold to tenfold at maximum, and the authors list real limits: a field instrument that overestimated nickel, an estimated 2 to 5 mm affected layer resting on a weak regression, one site over about two months, and unknown fate downstream. They estimated about 25 metric tons of nickel, manganese and cobalt across the marsh, which they say should be interpreted with caution. Surface levels fell back near pre-fire values after rain and tides in early February, then nickel rose again after the February 18 flare-up. Most post-fire surface samples had nickel above 50 ppm, a level the authors associate with risk to plants and aquatic life.

Where the evidence points: a large NMC fire does deposit cathode metals downwind, measurably and in an identifiable fingerprint, mostly as a thin surface layer. Agency testing has not found a human health threat. No published study has measured what people absorbed. For farm country, that combination is the argument for baseline soil sampling before any project is built.

Water and runoff

Water used to cool batteries or exposures, and any electrolyte that leaks, can carry metals and fluoride compounds. At Moss Landing, EPA says all water from the burned building is collected on site and sampled to ensure proper disposal, and that stormwater captured on site, including any firefighting water from a flare-up, is contained and sampled before disposal. A battery site over an aquifer that feeds domestic and irrigation wells needs the same thinking before a fire: containment, a place for firefighting water to go, and a sampling plan.

Cleanup, recycling and end of life

EPA issued a Notice of Federal Response Action to Vistra on January 22, 2025, and on July 17, 2025 signed an Administrative Settlement Agreement and Order on Consent under CERCLA (Docket No. 09-2025-08), under which Vistra performs and pays for battery removal with EPA oversight. EPA called it the largest lithium-ion battery cleanup in agency history. Removal began September 29, 2025. By June 24, 2026, 33,943 intact batteries from the burned area had been de-energized and more than 33,000 sent off site for recycling, with about 5,000 left in areas unsafe to reach. Phase 2 removes the most damaged modules and demolishes the building; unstable batteries are submerged in brine, and a 24-hour private fire crew stands by. On September 18, 2026, about 1,300 damaged modules caught fire during that work, prompting a shelter-in-place order within one-third mile. Twenty months after the fire, the cleanup was still producing smoke.

Retired batteries raise their own issues. EPA says most lithium-ion batteries are likely hazardous waste when discarded, usually ignitable (D001) and reactive (D003), and that it plans new rules for managing end-of-life lithium batteries. Recycling recovers lithium, nickel, cobalt, manganese and graphite by shredding cells into "black mass" and then smelting or leaching. The lifecycle impacts lesson covers mining and recycling capacity.

The 2021 and later IFC require a decommissioning plan that covers removing both an intact system and one damaged by fire. Idaho's 2018 edition does not. Supplier risk is real, too: Powin, which Idaho Power selected for its first large-scale battery projects and which built the system that burned near Melba, filed for bankruptcy in 2025, and FlexGen bought its software, hardware and spare parts. Warranties and service promises are only as good as the company behind them, which is why counties ask for bonds held independently of the owner.

What a county can require

In Idaho, a battery project on private land usually comes before a county or city as a special or conditional use. Idaho Code 67-6512 sets the rules. Notice goes to owners within 300 feet and to "any additional area that may be substantially impacted," as the commission decides. Before granting the permit, the county may require studies of social, economic, fiscal and environmental effects. And it may attach conditions, including those:

  • minimizing adverse impact on other development;
  • controlling the sequence, timing and duration of development;
  • assuring that development is maintained properly;
  • designating the exact location and nature of development;
  • requiring on-site or off-site public facilities or services;
  • requiring more restrictive standards than those generally required in an ordinance; and
  • requiring mitigation of effects on service delivery by any political subdivision, which includes fire districts.
ConcernConditions counties use
Fire safetyComply with the 2026 NFPA 855; UL 9540 listing; UL 9540A and large-scale fire test reports for the exact product; hazard mitigation analysis; explosion control
Emergency responsePlan approved in writing by the fire district; training and gas-monitoring equipment at the owner's cost; named evacuation and shelter radius with a notification method; 24-hour contact
DistanceSetbacks from homes, schools, roads, wells and canals beyond the code's 10 and 100 feet
EnvironmentBaseline soil and water sampling; runoff containment; air, soil and water sampling after any incident at the owner's cost
OperationsIncident notification within hours; cap on MWh and review of augmentation; access to alarm and maintenance logs; annual inspection paid by the owner
End of lifeDecommissioning plan for intact and damaged systems; bond or other security sized to an engineer's removal estimate, adjusted over time

The limits are real. A county cannot set electric rates or wholesale market rules; those belong to the Idaho PUC and FERC. Under 67-6512(a), a denial or conditions unacceptable to the landowner may be subject to the regulatory taking analysis in Idaho Code 67-8003. Conditions hold up best when each one is tied to a specific impact in the record. The appeals and takings lesson covers that analysis, and the energy siting course takes an openly protective position on how far to go.

Developers argue that very large setbacks and moratoriums are not based on the codes, that modern outdoor designs keep fires to one unit, and that blocking storage keeps older peaking plants running and raises costs. Residents argue that rural fire districts are often volunteer, that the consequences of a fire land locally while the revenue goes elsewhere, and that Moss Landing's 20-month cleanup shows what "contained to the site" can still mean. A record that addresses both, with test data, a real emergency plan and money set aside, is what holds up.

Common mix-ups

  • Claim: Moss Landing poisoned the whole region. What the evidence shows: Agency testing found no human health threat; peer-reviewed sampling found cathode metals enriched mostly in a thin surface layer downwind, with levels of concern for plants and aquatic life in places.
  • Claim: Air monitors showed nothing, so there was no exposure. What the evidence shows: Fixed monitors show conditions where they sit. Deposition was measured downwind, and no published human exposure testing exists.
  • Claim: Retired batteries can go to the landfill. What the evidence shows: EPA says most are likely hazardous waste when discarded.

Reporting angle

  • Request the application's environmental and hydrology materials, stormwater plan and any baseline sampling under the Idaho Public Records Act; ask whether the county has required studies under 67-6512(e).
  • Ask whether the permit requires post-incident air, soil and water sampling, who does it, who pays, and whether results are public.
  • Ask for the decommissioning cost estimate, the bond amount, who holds it and how often it is updated.
  • Ask who supplied the batteries and integration, and whether that company is still solvent and supporting its warranties.
  • Watch the planning and zoning hearing and the board's findings; a condition not written into the decision does not exist.

Research angle

Pictures, Numbers and Infographics

Federal data show what U.S. grid batteries are actually used for and how they have grown from short, fast units into multi-hour plants. The Moss Landing and lab figures show the scale of what one large lithium-ion fire leaves behind.

Burned and blackened lithium-ion battery case from a Boeing 787 auxiliary power unit, photographed by federal investigators
The auxiliary power unit battery from a Japan Airlines Boeing 787 that caught fire on January 7, 2013 at Boston's Logan International Airport, as photographed by the National Transportation Safety Board. The box measures about 19 by 13.2 by 10.2 inches and weighed about 63 pounds new. It is a small cousin of a grid battery module, built from the same family of lithium-ion cells, and the charring shows how far damage can reach inside one sealed case. Credit: National Transportation Safety Board. Public domain (U.S. government work), via Wikimedia Commons.

By the numbers

What U.S. grid batteries were mainly used for, 2023 (MW) Arbitrage 6,010.5 Frequency regulation 4,416.8 Excess wind and solar 2,634.4 System peak shaving 1,305.8 Load following 450.5 Co-located renewable firming 380.9 Ramping or spinning reserve 316.7 Voltage or reactive support 302.8 Load mgmt, backup, T&D 180.7 Blue: market trading and fast services. Green: storing wind and solar. Orange: other.
Power capacity of U.S. large-scale batteries by the primary application each owner reported for 2023. Bar length is proportional to megawatts. The last bar combines load management (119.8 MW), backup power (47.4 MW) and transmission and distribution deferral (13.5 MW); about 53.8 MW listed no primary application. Source: EIA, Battery Storage in the United States, 2024 figure data, Figure 12.
Average hours of storage, by year installed 0.6 2015 1.3 2016 2.2 2017 2.3 2018 2.8 2019 1.9 2020 3.0 2021 2.8 2022 2.8 2023 Hours = energy (MWh) divided by power (MW) for batteries installed that year.
Average duration of U.S. large-scale batteries by the year they were installed, rounded to one decimal. Early batteries chased fast services that need power more than energy; later ones were built to shift solar into the evening and to meet capacity rules. Source: EIA, Battery Storage in the United States, 2024 figure data, Figure 3.

Check yourself

1. Larsson and colleagues measured how much hydrogen fluoride from burning lithium-ion cells?

2. What did the peer-reviewed Scientific Reports study of Hester Marsh find after the January 2025 Moss Landing fire?

3. What did EPA say about its air monitoring during the Moss Landing fire?

4. Under what legal arrangement is Vistra removing the Moss Landing batteries?

5. Which Idaho statute lets a county attach conditions, including more restrictive standards, to a special use permit?

6. How does EPA say most discarded lithium-ion batteries should be classified?

Terms in this lesson

Hydrogen fluoride (HF)
A toxic, corrosive gas formed when the fluorinated salt in lithium-ion electrolyte burns or breaks down.
Deposition
Particles from smoke settling onto soil, plants, water and surfaces downwind.
CERCLA
The federal Superfund law, under which EPA can oversee and require cleanup of hazardous substance releases.
Decommissioning bond
Financial security posted with a county so equipment removal and site restoration get paid for even if the owner fails.
Baseline sampling
Testing soil, water and air before a project starts, so later results can be compared against real local conditions.

Full glossary

Sources