the sewage

that never arrives

Problem statement

South Africa's Green Drop program audits its roughly 850–900 municipal wastewater treatment works (WWTWs) and gives each a score, 30% of which reflects effluent quality and 70% capacity, environmental, financial and technical management. What the score does not capture is whether the sewage a town generates ever reaches the works: reticulation (sewer network) performance is "a small and neglected component" of the evaluation. Researchers at GroundTruth and the University of KwaZulu-Natal's Centre for Water Resources Research compared Green Drop data for 431 works with records in both 2013 and 2021 and found daily volumes treated fell on average even as the population grew by about 5.5 million (10%); in Pietermaritzburg the Darvill works received a median inflow of ~75 ML/day in 2022 against ~95 ML/day expected from the historical population–inflow relationship — a ~20 ML/day shortfall — despite a 44% increase in sewer-connected flush toilets between 2011 and 2022, while the share of river monitoring sites with E. coli above 10,000 CFU/100 mL rose from ~40% to ~90%. Nobody has a metric or a monitoring system for this "missing wastewater," and the audit's design gives works no penalty for operating below expected intake.

Why this matters

Sewage that does not reach a treatment works goes into streets, streams and rivers untreated; the authors link the inflow deficit at Darvill directly to "pervasive, severe sewage pollution" in the uMsunduzi system, and the same pattern (declining volumes treated with unchanged or barely changed scores) appears at Johannesburg's Northern Works (18% less sewage, unchanged score) and eThekwini's Northern Works (50% less sewage, score down 2%). Because the regulator's scorecard is what municipalities manage to, a blind spot in the scorecard becomes a blind spot in budgets and maintenance: the paper notes works "may actually be motivated to not collect sewage" if collecting it would push them over design capacity or hurt effluent compliance. The data needed to close the gap are also thin — 38.9% of works reported no design-capacity data in 2021 (up from 33% in 2013), more than half lacked continuous capacity monitoring, 43% reported having no maintenance team, and only 155 of 876 works received a technical site assessment in 2022.

What’s been tried and why it hasn’t worked

Green Drop itself is the attempt: an incentive-based national audit revived in 2021/22 after a hiatus, with published scores and technical site assessments. It has improved transparency about treatment, but its weighting rewards financial and administrative management, its site assessments cover a minority of works, and it has no requirement to reconcile expected sewage generation (from population and connections) with measured inflow, so reticulation collapse is invisible to it — the Darvill score moved from 79% to 78% between 2013 and 2021 while the network visibly failed. Flow metering that would expose the gap is patchy: many works cannot report design capacity utilisation at all, and metering the collection network upstream of the works is rarer still. Citizen-science indicators (miniSASS macroinvertebrate scores, clarity tubes) and river E. coli sampling detect the consequence but are not tied into the audit, and independent watchdog reports document the decline (works rated excellent/good fell from 14% to 8% between the 2021/22 and 2023/24 audit periods) without supplying the missing-flow number. The authors' own estimation method — regressing historical inflow on population and comparing with current inflow — is a workaround built for one city, not a monitoring system.

What would unlock progress

The core need is an operational "expected-versus-actual inflow" metric per works and per catchment, cheap enough to compute nationally from census connections, water-sales data and works inflow records, and hard enough to be written into the Green Drop framework as a scored, penalisable item — plus low-cost flow or level sensing at key manholes and pump stations to localise where the deficit arises. Adjacent precedent: water-supply utilities have decades of non-revenue-water accounting (the IWA water balance) that reconciles input with billed consumption; wastewater lacks the equivalent "sewer balance." Smart-sewer level sensors and sewer-overflow detection developed for combined-sewer cities could be adapted for South African gravity networks with intermittent power.

Entry points for student teams

A data team could build the missing-wastewater estimator: for a chosen municipality, combine census toilet-connection counts, per-capita water use, treated-water sales and works inflow records to produce a monthly expected-vs-actual inflow series with uncertainty, and publish it as a reproducible dashboard the regulator could adopt. A sensing team could prototype a battery-powered ultrasonic level logger with SMS/LoRa reporting for manholes and pump stations, and test whether a small number of nodes can localise a chronic loss. A policy team could draft the Green Drop amendment — the definition, data sources and scoring weight for a reticulation-performance criterion — and stress-test it against the perverse-incentive scenario the authors describe. Relevant skills: data engineering and statistics, hydraulic and sanitation engineering, embedded sensing, and regulatory design.