New Orleans Drinks the Mississippi: The Water Treatment Chemistry Behind the Tap at WEFTEC 2026

This week about 23,000 water professionals are in New Orleans for WEFTEC 2026, and every one of them is drinking the Mississippi River. The glass of water at the convention center left a treatment plant a few miles upriver at Carrollton, and before that it was rain on a farm field somewhere between Montana and Pennsylvania. It is one of the hardest raw waters any big American utility has to work with: hard, alkaline, full of silt, carrying organics and farm runoff from 31 states, and every few years the Gulf of Mexico pushes up the riverbed toward the intakes. What follows is how the city makes that river drinkable, why salt is the one thing its plants can't take out, and the coagulant change it is planning next.

What comes down the river

The Mississippi and its tributaries drain about 1,245,000 square miles: 31 states, two Canadian provinces, 41% of the contiguous United States. At New Orleans the river averages around 600,000 cubic feet per second and moves at roughly 3 mph. The National Park Service estimates a drop of water takes about three months to travel from Lake Itasca to the Gulf, and more than 50 cities drink from the river along the way. New Orleans is near the end of that line.

The USGS has sampled the river at Belle Chasse, a few miles downstream of the city, since the 1970s. Those records describe the raw water SWBNO has to work with:

Put 137 million gallons a day through that and you get a lot of mud. SWBNO says its two plants remove about 18,000 tons of solids a year. That is the right order of magnitude: SWBNO's 54 billion gallons a year at the river's median suspended sediment would carry more than 25,000 metric tonnes, and not every milligram settles out in the basins. The river is also a lot less muddy than it used to be. USGS researchers estimate it carried about 400 million metric tons of sediment a year to Louisiana before 1900, against an average of 145 million from 1987 to 2006, Dams trap 100 to 150 million tons a year, which Meade and Moody credit with about half of the decline; river cutoffs, bank revetments and soil conservation account for the rest.

The Carrollton treatment train, one step at a time

Carrollton was built between 1905 and 1909 and has run continuously since. SWBNO's 2026 Water Quality Master Plan says the plant was "originally constructed in the early 1900s" and has had no significant equipment upgrades "since the 1950s." It treated an average of 137 million gallons a day in 2025. Algiers, on the West Bank, treated about 13 MGD. Both take their water from the Mississippi, and both run the same sequence:

Coagulation: why iron works on river silt

River clay and silt carry a negative surface charge. In the water column that charge keeps particles apart, so they stay suspended forever. A trivalent metal salt breaks that stability two ways. Fe³⁺ ions and their first hydrolysis products neutralize the surface charge, and then iron hydroxide precipitates as a gelatinous floc that sweeps particles into it as it forms. In bicarbonate-rich river water, the overall reaction for ferric sulfate is:

Two things follow from that equation. First, iron hydroxide is the floc, and it ends up in the plant's residuals: about 0.53 mg of Fe(OH)₃ for every mg of anhydrous ferric sulfate dosed. Second, the reaction consumes alkalinity and releases CO₂, which pushes pH down. Six bicarbonates are used up per mole of ferric sulfate, which is ≈0.75 mg/L alkalinity as CaCO₃ per mg/L Fe₂(SO₄)₃ (our arithmetic from molecular weights). Every coagulant does this to some degree:

Figures are stoichiometric, calculated by Alliance Chemical from molecular weights on an anhydrous basis for the iron salts and ACH. Real plants dose more than stoichiometry, and actual demand depends on raw water, pH target and dose.

Lime: putting the alkalinity back

If the coagulant eats alkalinity and drives pH down, something has to bring it back up, and water leaving the plant has to be non-aggressive toward pipes. Carrollton uses lime for that. SWBNO's own materials describe the lime as minimizing pipe corrosion, with a small polyphosphate dose to keep the lime in solution. Lime is cheap and each milligram of calcium hydroxide supplies about 1.35 mg of alkalinity as CaCO₃. The drawback is solubility. Calcium hydroxide only dissolves to about 1.7 g/L at room temperature, so lime goes in as a slurry, and whatever doesn't dissolve, plus the grit that comes with commercial lime, settles wherever the water slows down.

At Carrollton, that place is the contact basins. The 2026 master plan says they are "hydraulically constrained because of the accumulation of settled lime solids associated with existing pretreatment chemistry." The basins where chlorine and ammonia are meant to have contact time are filling with lime.

The fix the city is planning: ACH and caustic

The master plan's answer starts with chemistry. Bench testing evaluated "aluminum chlorohydrate (ACH) and caustic to replace the current ferric sulfate and lime processes," and the plan recommends a full-scale pilot at Carrollton. According to the plan, jar tests "indicate that ACH will reduce the lime concentration and subsequent buildup of lime deposition in the contact chambers."

The coagulant table above shows the logic. ACH is aluminum that has already been partly neutralized with hydroxide during manufacture, so it takes much less alkalinity out of the water than ferric sulfate does, and the plant needs less base to bring the pH back. The base it does add would be caustic soda instead of lime. Sodium hydroxide is fully soluble, so there's no slurry and nothing left undissolved to settle. One mg/L of NaOH supplies about 1.25 mg/L of alkalinity as CaCO₃, close to lime per unit mass, but all of it stays in solution.

Chlorine, then ammonia: why New Orleans uses chloramine

After coagulation, SWBNO adds chlorine and then ammonia. The two react to form monochloramine:

The ratio decides what you get. EPA's operational training material puts the best ratio for forming monochloramine at 4.5:1 to 5:1 Cl₂ : NH₃-N by weight. Push chlorine up toward the theoretical breakpoint of 7.6:1 (in real water, more like 9:1 or 10:1) and the chloramine is destroyed, first to dichloramine and nitrogen trichloride, which bring taste and odor complaints, and then to free chlorine. Run too little chlorine and you leave free ammonia in the pipes, which feeds nitrifying bacteria.

Utilities choose chloramine for two reasons that both apply to New Orleans. It lasts longer in a big, old distribution system, and it forms far fewer regulated disinfection byproducts than free chlorine does with river organics. The 2025 numbers show that working:

TOC removal is the number to watch. Algiers fell below the required ratio in 2022 (0.87 to 0.96), and the master plan calls for work to improve TOC removal there. TOC is the organic matter that turns into trihalomethanes and haloacetic acids when chlorine meets it, so removing it in the coagulation step is what keeps the byproduct numbers low downstream. The January 2025 freeze caused turbidity violations at both plants: 6.5% of Carrollton readings were above 0.349 NTU that month, and Algiers exceeded the 1.499 NTU single-sample maximum on January 22, which is where the 3.00 NTU in the table comes from.

The one thing the plant can't remove: the saltwater wedge

The riverbed of the lower Mississippi sits below sea level all the way from Natchez to the Gulf. Seawater is denser than river water, so when the river's flow drops, salt water creeps upstream along the bottom as a wedge under the fresh water. The Army Corps of Engineers puts the flow needed to keep salt out at around 300,000 cubic feet per second. In late September 2023 the river was running at about 150,000, half of that.

The Corps's answer is an underwater sill, a mound of dredged riverbed sediment across the channel at river mile 64 near Myrtle Grove that the dense bottom water has to climb over. It has been built in 1988, 1999, 2012, 2022, 2023 and 2024. In 2023 the first sill was built in July to 55 feet below the surface. The wedge overtopped it on September 20, and the Corps raised it to 30 feet below the surface, leaving a 620-foot-wide shipping lane at 55 feet. By September 25 the toe of the wedge was at river mile 69 near Jesuit Bend, sitting in a 120-foot-deep hole in the riverbed. It had retreated to mile 63.9 by October 9.

Plaquemines Parish, downriver, was hit hardest: intakes at Boothville, Port Sulphur and Pointe à la Hache needed reverse osmosis units and barged water, and residents were on bottled water from June 2023. The city's own intakes were in the forecasts too. In late September 2023 the Corps projected salt reaching the Algiers intake around October 22 and Carrollton around October 28, and SWBNO planned to barge 14 million gallons a day to Algiers if it had to. Revised forecasts pushed those dates back, and the wedge retreated before it arrived.

The USGS record shows how bad the river can get. On July 8, 1988, two samples taken one minute apart at Belle Chasse read 350 and 10,000 mg/L chloride, the second with a hardness of 3,660 mg/L and a conductivity of 27,300 µS/cm. Seawater runs about 19,000 mg/L chloride. The sample depths weren't recorded, but a gap that large one minute apart looks like surface water and wedge water.

Why coagulation can't touch chloride

SWBNO says it plainly: its plants "cannot remove salt," and salt "cannot be taken out through boiling or conventional filtration." The chemistry reason is that everything in the conventional train works on particles or on ions that form insoluble solids.

• Coagulation removes charged particles and colloids by precipitating metal hydroxide around them. Chloride is a single dissolved ion. It isn't a particle and it doesn't carry a surface charge that a floc can neutralize.
• Precipitation works when the ion forms an insoluble salt with something you can add. Chloride doesn't. Calcium chloride, sodium chloride, ferric chloride and aluminum chloride are all extremely soluble. Nothing you can dose in a rapid-mix basin will drop chloride out of solution.
• Filtration through sand and anthracite catches floc. A dissolved ion goes straight through.
• Boiling concentrates salt, because the water leaves and the chloride stays.

Removing chloride takes a membrane (reverse osmosis), distillation, or ion exchange. That is why RO units showed up in Plaquemines in 2023. For a plant the size of Carrollton, the master plan found RO, or blending with groundwater or upstream river water, to be "prohibitively high" in capital and operating cost.

What the master plan says about the whole system

A few numbers from SWBNO's 2026 Water Quality Master Plan explain why the chemistry decisions matter to the city's budget:

• Nonrevenue water averaged 110 MGD, or 74% of production, from 2019 to 2022. This is water the plants produce that never gets billed. It includes leaks, but also unbilled use and metering losses. The plan places SWBNO at the 90th percentile among peer utilities. Every gallon lost still got coagulant, lime and chloramine.
• Rebuilding Carrollton is priced in three phases: $489.2 million, $802.8 million and $301.1 million. The rebuilt plant is planned around 240 MGD.
• Demand is expected to decline with population and efficiency, so the plan doesn't call for more capacity. The work is replacing what exists.
• The Carrollton plant itself is identified as the single greatest risk to level of service, with the Old River intake station, the sedimentation basins and the Claiborne filters among the highest-risk systems.

Power is part of the same system. The "Old City" drainage pumps still run on 25-cycle electricity, supplied by a mix of the Carrollton plant's Turbine 5 (1958), backup generators and frequency changers fed from the grid; the 1915 steam turbine T4 has been permanently retired. When power dips, pressure drops, and the East Bank has had boil-water advisories after a Mylar balloon hit a power line (August 2024), a power surge at Carrollton (January 2025) and a 48-inch main break (March 2026). We explain what those advisories mean chemically in what a boil water notice actually means.

The same system also drains the city. A. Baldwin Wood designed his screw pump in 1913, and the first four 12-foot pumps went in in 1915, each rated at 247,500 gallons per minute against a 7-foot lift. Some are still in service.

What WEFTEC 2026 is arguing about this week

WEFTEC runs September 26 to 30 at the Ernest N. Morial Convention Center, with the exhibit floor open September 28 to 30 and more than 950 exhibitors. Most of the chemistry on the agenda is the same chemistry described above, applied to different problems.

PFAS: the rule is still the 2024 rule

EPA's April 2024 rule set maximum contaminant levels of 4.0 ppt each for PFOA and PFOS, plus limits for PFHxS, PFNA, HFPO-DA and a mixture Hazard Index. In May 2026 EPA proposed two changes: extending PFOA and PFOS compliance by up to two years to April 2031, and rescinding the other four. Comment periods closed July 20, 2026. Neither proposal is final. The D.C. Circuit heard oral argument in AWWA v. EPA on September 18. Until something changes, the 2024 deadlines stand: initial monitoring by April 26, 2027 and compliance by April 26, 2029.

New Orleans is under the line. SWBNO's 2025 report shows PFOA averaging 1.7 ppt and PFOS 2.0 ppt on the East Bank and 2.2 ppt on the West Bank. For utilities that aren't, the treatment is mostly physical: granular activated carbon, ion exchange or reverse osmosis. Bulk chemicals support the membranes: citric acid and caustic for cleaning, bisulfite to dechlorinate ahead of RO, and acid for scale control. We laid out the full sequence in our PFAS treatment train guide.

PFAS in biosolids: comments due October 5

EPA's January 2025 draft risk assessment found that land-applying sludge at 1 ppb PFOA or PFOS could exceed its risk thresholds in some modeled farm scenarios. The draft guidance EPA released on July 1, 2026 is voluntary, says 1 ppb "was not intended to be interpreted as a 'safe level,'" and notes that about 60% of US sludge is land applied. Comments are due October 5, 2026 (docket EPA-HQ-OW-2026-2509). Thermal destruction is a live topic on the floor. The "Fighting PFAS With Fire" session on Wednesday covers advanced thermal conversion for biosolids.

Phosphorus: still a metal-salt problem for most plants

There's no new federal nutrient rule. Phosphorus limits still come through state criteria, TMDLs and NPDES permits, and chemical phosphorus removal still mostly means iron or aluminum salts. EPA's Nutrient Control Design Manual uses 1.0 mole of metal per mole of phosphorus as the stoichiometric starting point. It puts 1.5 to 2.0 Me:P at 80 to 98% soluble P removal, and says getting below 0.10 mg/L total P can take 6 to 7. Its worked example (2.0 mol Fe per mol P, 40% ferric chloride) comes to about 74.8 gallons per day at 1 MGD and 4 mg/L P. That ferric chloride also takes about 0.92 mg/L of alkalinity per mg/L dosed, which nitrifying plants can't spare: nitrification uses 7.14 g of alkalinity as CaCO₃ per gram of ammonia nitrogen oxidized. The session "Choose Your Own Adventure: Selecting Your Phosphorus Removal Method" is Wednesday morning. Our ferric chloride phosphorus removal guide works through dosing and sludge.

Lead, disinfection and supply

• Lead and Copper Rule Improvements: compliance begins November 1, 2027, with lead service line replacement within 10 years. New Orleans's 2025 90th-percentile lead was 6 ppb on both banks, with a single West Bank maximum of 100 ppb.
• Disinfection: Sunday's all-day workshop covered disinfection fundamentals, and Wednesday has a peracetic acid batch and pilot study for wastewater, one of the alternatives to hypochlorite that keeps coming up.
• Supply: EPA's 2023 supply-chain profiles rate every major treatment chemical as high criticality. They put hypochlorite and caustic soda at moderate-to-high disruption risk, ferric chloride and hydrochloric acid at moderate-to-low, and alum, sulfuric acid and hydrogen peroxide at low. EPA's profile also says about 80% of US ferric chloride goes to the water sector.