Water quality targets and regulations

A year after the preliminary studies of CIV-280, Willowbend Environmental in Sioux Falls, South Dakota, has won the detailed design of two plants for the city. The first is a new drinking water treatment plant that will take water from the Big Sioux River and an off-stream reservoir, sized for 10.5 million gallons per day (MGD) on an average day and 16 MGD on a peak day. The second is the upgrade of a city wastewater treatment plant, which treats about 7.9 MGD and must now meet tighter limits on phosphorus and ammonia. Nadia Okoro, now in her second year as an engineer intern, leads the process calculations under Martin Lefebvre, the firm's senior process engineer, with Joy Kwan running the laboratory and pilot work.

This course follows both plants through their unit processes, one lesson at a time. Before any tank is sized, though, the design team needs to know what the plants must achieve. Every process choice, from the coagulant dose to the length of time sludge stays in an aeration tank, traces back to a number in a guideline, a regulation or a permit. This lesson sets out where those numbers come from, how they are enforced, and how a designer turns them into design targets with a margin of safety.

Treatment as a series of barriers

Modern drinking water practice rests on the multi-barrier approach: no single step is relied on to protect public health. Source water protection keeps contaminants out of the river and the reservoir; treatment removes or inactivates what gets through; a disinfectant residual protects the water in the distribution system; and monitoring, trained operators and emergency plans catch failures before they reach consumers. The same idea applies, in reverse, to wastewater: collection, treatment, disinfection where needed and monitoring together protect the river downstream.

Two Canadian tragedies made the approach concrete. In Walkerton, Ontario, in 2000, manure-borne E. coli O157:H7 and Campylobacter entered a shallow well after heavy rain; chlorination was inadequate and monitoring was falsified, and seven people died. In North Battleford, Saskatchewan, in 2001, Cryptosporidium passed through a poorly operated surface water plant and thousands fell ill. In the United States, the 1993 Milwaukee Cryptosporidium outbreak sickened an estimated 400,000 people, and the Flint crisis of 2014 to 2015 showed how a change in source water and corrosion control can release lead from pipes. Each event led to stronger rules, and each is a reminder that regulations are written in response to real failures.

Guidelines, regulations and permits

A guideline is a recommended value based on health or aesthetic considerations; it becomes enforceable only when a law adopts it. A regulation is a legal requirement that applies to all systems of a given type. A license or permit is a legal instrument issued to one facility, setting limits and monitoring requirements specific to that plant and its receiving water. A designer has to work with all three.

In the United States, the Safe Drinking Water Act gives the Environmental Protection Agency (EPA) authority to set National Primary Drinking Water Regulations. Each regulation sets a maximum contaminant level goal (MCLG), a non-enforceable health goal, and either an enforceable maximum contaminant level (MCL) or a required treatment technique where measurement is impractical. Secondary standards cover taste, odor and color and are not federally enforceable. States with primacy enforce the rules; in South Dakota that is the Department of Agriculture and Natural Resources, which reviews plant designs, issues construction approvals and requires certified operators. Many states also follow the Recommended Standards for Water Works, known as the Ten States Standards, for design details.

Wastewater effluent in the United States is regulated under the Clean Water Act. Every point discharge needs a National Pollutant Discharge Elimination System (NPDES) permit, issued in South Dakota by the state. Municipal plants must at least meet the federal secondary treatment standards in 40 CFR Part 133: a 30-day average BOD₅ and suspended solids of 30 mg/L each, 7-day averages of 45 mg/L, at least 85 % removal, and pH between 6.0 and 9.0. Permits then add water-quality-based limits where the receiving water needs more protection, such as limits on ammonia, E. coli in the recreation season, total residual chlorine and, increasingly, phosphorus and nitrogen. The Big Sioux flows to the Missouri and then the Mississippi, which carries the nutrient load that causes the hypoxic zone in the Gulf of Mexico; state nutrient reduction strategies now push upstream plants to remove more phosphorus and nitrogen.

Key drinking water parameters

The table below lists values used throughout this course. They are summarized here for design; the current official documents should always be checked, because values change as health evidence develops.

Two rules are changing as this course is written. For PFAS, EPA set MCLs in 2024 of 4.0 ng/L for PFOA and PFOS, plus limits for PFHxS, PFNA, HFPO-DA and a hazard index for mixtures, with compliance due in 2029. In May 2026 EPA proposed to keep the PFOA and PFOS limits, rescind the others and allow systems to apply for two more years, to 2031; challenges to the 2024 rule are also before the courts. For lead, the Lead and Copper Rule Improvements (LCRI) of 2024 lower the action level to 0.010 mg/L, require most lead service lines to be replaced within ten years and take effect on November 1, 2027; the rule faces litigation but EPA is defending it. A designer facing this uncertainty plans for the stricter outcome where the cost of doing so later would be high.

From standard to design target

A plant that is designed to just meet a limit will exceed it. Influent quality varies, equipment fails, operators change and analytical results scatter. Designers therefore set design targets well inside the limits. Common practice for drinking water is to aim for a filtered water turbidity of 0.1 NTU or less, a third of the regulatory value, and for wastewater to design for an average effluent concentration around half to two thirds of a monthly permit limit, so that the variable day-to-day performance still keeps the monthly average compliant.

The statistical form of each limit matters as much as the number. A maximum applies to every sample. A running annual average, used for disinfection by-products, smooths seasonal peaks, so a plant can exceed the value in one quarter and still comply if the yearly average stays below the limit. A percentile, used for turbidity and for lead, allows a small share of results to be high. A rolling 30-day average, used for Manitoba's phosphorus standard, responds to a short upset within weeks. Nadia's first task is to build a table that lists, for each parameter, the limit, its statistical form, the sampling location and the design target that the process will be sized to achieve.

The script below repeats the screen for the full set of results, computes the running annual average of trihalomethanes at one sampling location over five quarters, and checks a month of four-hourly combined filter turbidity readings against the 0.3 NTU, 95 % rule.

# Compliance screen for one month of treated-water results
UNITS = "US"
CA = UNITS == "SI"
# (limit, unit) pairs; None means no enforceable health-based maximum
if CA:
    limits = {"nitrate as N": (10.0, "mg/L"), "lead at the tap": (5.0, "ug/L"),
              "manganese": (120.0, "ug/L"), "total trihalomethanes": (100.0, "ug/L"),
              "HAA5": (80.0, "ug/L"), "sum of 25 PFAS": (30.0, "ng/L")}
else:
    limits = {"nitrate as N": (10.0, "mg/L"), "lead, 90th percentile": (10.0, "ug/L"),
              "manganese": (None, "ug/L"), "total trihalomethanes": (80.0, "ug/L"),
              "HAA5": (60.0, "ug/L"), "PFOA": (4.0, "ng/L"), "PFOS": (4.0, "ng/L")}
results = {"nitrate as N": 3.1, "lead at the tap": 6.2, "lead, 90th percentile": 6.2,
           "manganese": 35.0, "total trihalomethanes": 71.0, "HAA5": 44.0,
           "sum of 25 PFAS": 9.0, "PFOA": 2.1, "PFOS": 3.6}
for k, (lim, u) in limits.items():
    v = results[k]
    if lim is None:
        print(f"{k:24s} {v:6.1f} {u:5s} no federal health-based maximum")
        continue
    word = "action level" if k.startswith("lead, 90") else "limit"
    status = "ok" if v <= lim else "above the " + word
    print(f"{k:24s} {v:6.1f} {u:5s} {word} {lim:g}: {status}")
# running annual average of quarterly TTHM at one sampling location (ug/L)
q = [52, 94, 88, 61, 83]
for i in range(3, len(q)):
    avg = sum(q[i-3:i+1]) / 4
    print(f"TTHM running annual average to quarter {i+1}: {avg:.1f} ug/L")
# 4-hourly combined filter effluent turbidity for a 30-day month (NTU)
import random
random.seed(7)
turb = [round(abs(random.gauss(0.09, 0.05)), 3) for _ in range(180)]
turb[50] = 0.42; turb[51] = 0.35
share = sum(1 for x in turb if x <= 0.3) / len(turb)
print(f"turbidity: {len(turb)} readings, maximum {max(turb):.2f} NTU, {100*share:.1f} % at or below 0.3 NTU")

nitrate as N                3.1 mg/L  limit 10: ok
lead, 90th percentile       6.2 ug/L  action level 10: ok
manganese                  35.0 ug/L  no federal health-based maximum
total trihalomethanes      71.0 ug/L  limit 80: ok
HAA5                       44.0 ug/L  limit 60: ok
PFOA                        2.1 ng/L  limit 4: ok
PFOS                        3.6 ng/L  limit 4: ok
TTHM running annual average to quarter 4: 73.8 ug/L
TTHM running annual average to quarter 5: 81.5 ug/L
turbidity: 180 readings, maximum 0.42 NTU, 98.9 % at or below 0.3 NTU

The running annual average ends the fifth quarter at 81.5 µg/L, above the 80 µg/L MCL, so the system would be in violation at that location even though no single quarter set the trend on its own. The new plant's lower organic carbon must bring the summer peaks down, and the design target is set at 60 µg/L. The turbidity record shows two readings above 0.3 NTU in 180, or 98.9 % compliance, which meets the rule but signals a filter that needs attention.

Wastewater limits and the receiving water

Effluent limits are set either by technology, meaning what secondary treatment can reliably achieve, or by the receiving water, meaning what the river needs to meet its own water quality objectives. A technology-based limit is the same everywhere; a water-quality-based limit depends on the dilution available, the background concentration and the use of the river downstream. The Streeter-Phelps oxygen sag of CIV-280 is one tool for setting such limits; mass balances on ammonia, phosphorus and chloride are others. In a low-flow winter, the Assiniboine or the Big Sioux may provide only a few volumes of dilution for each volume of effluent, which is why ammonia limits are often stricter in winter even though nitrification is hardest then.

Un-ionized ammonia is toxic to fish at low concentrations, and its share of total ammonia rises steeply with pH and temperature. A plant that nitrifies converts ammonia to nitrate, removing the toxicity but adding nitrogen that feeds algae downstream. Removing that nitrate by denitrification, and removing phosphorus by biological or chemical means, is the subject of lesson 10. The city plant's new limits, and the costs of meeting them, are the reason for the upgrade studied in lesson 12.

Who designs, who approves, who operates

In South Dakota, the plant design must be prepared and sealed by a professional engineer licensed in the state. The water plant needs plan approval from the state drinking water program before construction, and the wastewater upgrade needs an amended NPDES permit and approval of the plans. Operators must be certified at the level of the facility's classification. Willowbend's drawings will be reviewed by state staff, often against the Ten States Standards, and the design report must show how each requirement is met.

Good regulation also relies on public reporting. Utilities publish annual reports of their results, and an exceedance usually triggers notification of the regulator within hours and of the public within days. Designers support this by providing enough sampling points, on-line analyzers and data systems to show compliance continuously rather than only on the day of a grab sample.