Disinfection Chemical Incidents at US Water Utilities

A comparative read of federal reporting data for chlorine gas and bulk sodium hypochlorite, 2010–2024

Ethan Brooke, cleanwater1 · May 19, 2026 · v1.1

1. Why a direct count is hard

Chlorine and sodium hypochlorite are tracked under different federal regimes. Chlorine is on the EPA Risk Management Program list_1. Under 40 CFR 68, any facility with at least 2,500 lb of Cl2 on site must register and report accidents on the RMP five-year history; those events feed the public RMP database and EPA-issued analyses_2. Bulk 12.5% sodium hypochlorite is exempt from the RMP threshold-quantity calculation_3. A sodium hypochlorite release only enters federal reporting when a CERCLA reportable quantity of available chlorine — 10 lb in 24 hours — leaves the facility, and even then it goes to the National Response Center rather than to RMP. The practical effect is that the same operational event has very different federal visibility depending on which chemical it involves: a 1,000-gallon sodium hypochlorite tanker spill at a water plant may never enter a federal aggregate, while a much smaller chlorine release at the same plant becomes an RMP accident-history entry.

This reporting asymmetry is itself a finding. The federal record makes bulk sodium hypochlorite appear quieter than chlorine because the reporting threshold is much higher, not because the underlying operational events are absent.

2. Chlorine — EPA RMP five-year accident history

The most recent EPA-published analyses of the RMP accident history come from the Safer Communities by Chemical Accident Prevention rulemaking docket_4. At water-supply and POTW facilities (NAICS 2213), the analyses identify approximately 8 to 9 RMP-reportable accidents per year. Industry-wide, RMP records show an annual average of roughly 145 reportable accidents across all chemicals and industries (2014–2023)_4. The Federal Register notice of the proposed RMP rule identifies the water-and-wastewater RMP universe as 1,449 government-owned and 519 non-government facilities₅ — roughly 1,968 chlorine-using water-sector RMP filers in total, producing on the order of 9 RMP-threshold accidents per year combined. Per-facility-year incidence at the RMP severity threshold is on the order of 0.5%.

3. National Response Center — both substances, fairer comparison

EPA's 2022 Federal Register notice on Clean Water Act Hazardous Substance Worst-Case Discharge Planning includes a distribution table of CWA hazardous-substance releases drawn from National Response Center reports, 2010–2019_6. The NRC captures both substances at comparable thresholds (CERCLA RQ of 10 lb in 24 hours), in the same database, over the same window.

Source: 87 Fed. Reg. 17,890 (March 28, 2022), Table 3, citing National Response Center reports, 2010–2019. Cross-industry; not segmented to water-sector reporters.

Two reads of these numbers are worth noting. Sodium hypochlorite is not an RMP-listed substance, and yet in the NRC database — at the threshold where both substances appear — its incident share is within one percentage point of chlorine's. For a chemical routinely positioned as the safer-by-design alternative, the operational incident frequency in federal data is comparable. The figures are cross-industry, not water-utility-only, but the threshold parity is the load-bearing methodological point.

4. The chemical-incompatibility pathway

Frequency alone does not capture the full hazard profile, because the most consequential bulk-sodium-hypochlorite events at and adjacent to water utilities have not been release-from-container events at all — they have been cross-contamination events where bulk NaOCl contacts an acid or an incompatible coagulant and generates a chlorine-gas release at the point of mixing.

MGPI Processing, Atchison KS (October 2016). A sulfuric-acid delivery truck was inadvertently connected to a bulk sodium hypochlorite fill line. The cross-contamination produced a chlorine-gas plume that triggered shelter-in-place orders for thousands of residents; more than 120 employees and members of the public sought medical attention. The CSB published its final investigation report in January 20187. The facility was a distilled-spirits plant, not a water utility, but the failure mechanism — bulk NaOCl plus an incoming acid at the same loading dock — is also the dominant NaOCl hazard at water plants, where coagulants and acids are routinely delivered alongside the disinfectant.

Birmingham Water Works, Shades Mountain WTP (February 2019). Bulk sodium hypochlorite and ferric sulfate were inadvertently combined during a delivery operation. The reaction off-gassed chlorine on site. Approximately 55 people were transported for medical evaluation (14–15 with direct exposure, the balance precautionary), and US Highway 280 adjacent to the plant was closed for over two hours_8. The event is sometimes catalogued externally as a chlorine release; the root chemicals on site were bulk NaOCl and an incompatible coagulant.

The two events together establish that the gas-chlorine catastrophic-release class is not eliminated by switching from bulk chlorine to bulk sodium hypochlorite. The release is relocated upstream — from the chlorine-gas storage and dosing operation to the sodium-hypochlorite receiving and storage operation. The mechanism does not depend on the kind of facility (distilled spirits, water treatment); it depends on the colocation of bulk NaOCl with an acid or an incompatible coagulant on the same delivery and storage path.

5. Synthesis

The federal databases do not support a clean head-to-head accidents-per-year count for the two substances. The asymmetry of reporting thresholds (RMP applies to chlorine, not to bulk sodium hypochlorite) means that any single-database count is structurally misleading. In the one database that catches both at comparable thresholds — the National Response Center — chlorine and sodium hypochlorite produce events at comparable frequency, 4% versus 3% of cross-industry CWA hazardous-substance discharges over 2010–2019.

What can be said with the data in hand: chlorine-gas events at water and wastewater RMP filers occur at roughly₉ reportable events per year across roughly 1,968 facilities; sodium-hypochlorite events occur at comparable frequency at the NRC threshold, with most lower-severity NaOCl events handled at the facility level and not appearing in federal aggregates. Severity per event differs materially — chlorine gas carries higher catastrophic-release potential in a vacuum — but the MGPI 2016 and Birmingham 2019 pathway shows that bulk sodium hypochlorite can produce chlorine-gas releases of comparable severity when it colocates with acids or incompatible coagulants. The safety advantage of bulk sodium hypochlorite over bulk chlorine — measured in actual operational incidents rather than regulatory classification — is narrower than the conventional framing suggests, and is absent on the receiving and storage step.

On-site generation of low-strength (0.8%) sodium hypochlorite from salt and water eliminates the upstream chain that produces both event classes: no bulk-chlorine cylinders or tankers to receive, no concentrated NaOCl tanker to mis-deliver, no concentrated NaOCl in bulk storage to cross-contaminate.

Limitations. RMP figures come from EPA-published analyses; the public RMP query tool was taken offline in early 2025. NRC percentages are cross-industry, not water-sector-only. DOT PHMSA per-UN-number counts (UN 1017, UN 1791) were not extractable within this scope and are a follow-up.

References
1. 40 CFR Part 68 — Chemical Accident Prevention Provisions. https://www.ecfr.gov/current/title-40/chapter-I/subchapter-C/part-68
2. US Environmental Protection Agency, Risk Management Program (RMP). https://www.epa.gov/rmp
3. US Environmental Protection Agency, Amounts of Chlorine Present in Sodium Hypochlorite (RMP applicability guidance). https://www.epa.gov/rmp/amounts-chlorine-present-sodium-hypochlorite
4. US Environmental Protection Agency, Safer Communities by Chemical Accident Prevention rulemaking docket EPA-HQ-OLEM-2025-0313 (RIA and RMP accident-data analysis). https://www.regulations.gov/docket/EPA-HQ-OLEM-2025-0313
5. Federal Register, Accidental Release Prevention Requirements: Risk Management Programs Under the Clean Air Act; Safer Communities by Chemical Accident Prevention, 87 Fed. Reg. 53556 (proposed rule, August 31, 2022). https://www.federalregister.gov/documents/2022/08/31/2022-18249
6. Federal Register, Clean Water Act Hazardous Substance Worst-Case Discharge Planning Regulations, 87 Fed. Reg. 17890 (proposed rule, March 28, 2022). https://www.federalregister.gov/documents/2022/03/28/2022-05505
7. US Chemical Safety and Hazard Investigation Board, Toxic Chemical Release at the MGPI Processing, Inc. Facility, Atchison, Kansas, October 21, 2016 (final report, January 2018). https://www.csb.gov/mgpi-processing-inc-toxic-chemical-release-/
8. Aggregated public-source reporting on the Birmingham Water Works Shades Mountain WTP chlorine release, February 27, 2019 (AL.com, Birmingham Times, ABC News, NBC News).