water · infrastructure · family: the wrong ruler
the sewagethat never arrives
South Africa's green drop audit scores sewage works on what they treat, not on the sewage that never arrives — so a city can lose a fifth of its wastewater to leaking sewers without its score moving
Problem statement
South Africa's Green Drop program audits the country's municipal wastewater treatment works (WWTWs; the paper's dataset covers 876) 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%. The Green Drop reports carry no measure of 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 streams and rivers untreated; the authors report that the inflow deficit at Darvill was matched by rising E. coli counts in the rivers of the Pietermaritzburg area, which they read as "pervasive, severe sewage pollution," 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%). The scoring can also shape behaviour: the paper says a works can reach a high score through acceptable financial and administrative management "while completely neglecting the infrastructure network," and that works "may actually be motivated to not collect sewage" if collecting it would push them over design capacity or cause treatment problems that are penalised. The data needed to close the gap are also thin — 38.9% of works reported no data on the share of their design capacity in use in 2021 (up from 33% in 2013), more than half lacked continuous capacity monitoring, 43% reported having no maintenance team in the 2023 report, 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. Its return showed that there had been considerable deterioration while it was halted, 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 failure is often not captured by it — the Darvill score moved from 79% to 78% between 2013 and 2021 despite what the authors call evidence of declining performance in its sewer network. The data that would expose the gap are patchy: 341 of the 876 works reported no figure for design capacity in use in 2021. Citizen-science tools (miniSASS stream assessment scores, clarity tubes) are put forward by the authors as first-order indications to add to routine monitoring, and river E. coli sampling detects the consequence, but neither supplies the missing-flow number; press coverage of the 2023/24 Green Drop release reports a further decline (works rated excellent/good fell from 14% to 8% between the 2021/22 and 2023/24 audit periods). The authors' own estimation method — projecting forward the historical relationship between inflow and population and comparing it with current inflow — was applied to one city as a case study; it is not a monitoring system.
What would unlock progress
The paper's own recommendations are that Green Drop technical site assessments need "a much stronger focus on sewage reticulation," that monitoring must show whether wastewater reaches the works at all, at finer resolution in time and space, and that penalties may be required because the incentive-based system appears inadequate. What follows is this write-up's proposal for acting on that; the paper does not state it. 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. A possible model, which the paper does not mention, is the non-revenue-water accounting used on the water-supply side (the IWA water balance); a wastewater "sewer balance" would be the equivalent for collection. Sewer level sensing and overflow detection could be tried in the same role.
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.
Genes
Structural cousins: stuck the same way, in other fields
Sources
Graham PM, Pattinson NB, Still D, "The state of wastewater management in South Africa: data gaps, missing wastewater, and Green Drop reporting," Water SA 51(2), April 2025, doi:10.17159/wsa/2025.v51.i2.4152, accessed 2026-08-17 go to source ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
The source is a peer-reviewed article in Water SA (the journal of South Africa's Water Research Commission) by South African researchers (GroundTruth; UKZN Centre for Water Resources Research; Partners in Development / Duzi-uMngeni Conservation Trust) analysing their own country's regulatory data; the article was read via a structured extract of the full text (methods, national comparison, Pietermaritzburg case, discussion). Figures used are as reported there: 431 works compared; ~5.52 million population increase; Darvill ~94.66 vs ~75.00 ML/day; 44% more sewer-connected flush toilets 2011–2022; E. coli >10,000 CFU/100 mL sites ~40%→~90% (about 2016 to the end of 2022); 38.9%/33.0% missing DCU data; 155 of 876 works site-assessed in 2022; 30/70 score weighting; Johannesburg −18% and eThekwini −50% volume examples. The 14%→8% excellent/good figure comes from press coverage of the 2023/24 Green Drop release (Engineering News, 17 April 2026: 118 → 66 systems; 848 municipal systems audited; https://www.engineeringnews.co.za/article/new-drop-reports-shows-deterioration-of-south-africas-wastewater-systems-2026-04-17) and was not read in the primary report — treat as secondary; the verifier confirmed the same numbers in that coverage on 2026-08-17. `failure:ignored-context` is applied in its "assessment as binding constraint" sub-pattern (the evaluation system shapes behaviour more than the policy). `failure:wrong-problem` was considered and ruled out: the Green Drop objective (safe treatment and competent management) is right; the metric set is incomplete, which is a context/design omission rather than a mis-specified goal. `temporal:worsening` passes on the E. coli trajectory, the volume-treated decline against population growth, and the named mechanism (reticulation decay compounding under-maintenance). `stakeholders:multi-institution` was considered (DWS regulator, municipalities, water boards) but the binding constraint is the absent metric and data, not the institutional boundary. Related collection briefs: `water-intermittent-supply-nrw-performance-indicators` (staging; supply-side loss indicators — the wastewater analogue proposed here is distinct), `water-onsite-sanitation-emptying-treatment-accounting-gap` (staging; non-sewered accounting), and `water-aging-pipe-network-failure-prediction`.
Source type: Self-articulated (South African water-research community, published in the WRC's journal, articulating a blind spot in its own national regulator's audit)
Verified at intake 2026-08-17: gate (net) + adversarial source check + contested-tag second coding. Verifier note: SciELO full text fetched; 431 of 876 works, ~5.52 million, 94.66 vs 75.00 ML/day (deficit ~19.66), 44% flush-toilet increase, ~40%→~90% sites >10,000 CFU, 38.9%/33.0%, 155 of 876, 30/70 weighting, 'small and neglected', 'motivated to not collect sewage', 'pervasive, severe sewage pollution', Johannesburg −18% / eThekwini −50%, Darvill 79%→78%, 43% no maintenance team all confirmed.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against its sources; a second reader went over the repairs on the same day. The article (Graham, Pattinson and Still, Water SA 51(2), 2025) was fetched on 2026-10-01 directly from https://www.scielo.org.za/scielo.php?script=sci_arttext&pid=S1816-79502025000200001 and read in full (abstract through conclusions) by both readers; its citation was compared with the Crossref record for the DOI. The second source this page names, press coverage in Engineering News (17 April 2026), could not be opened by the first reader; the second reader opened it directly and through the Internet Archive copy of 2026-09-04, and its link is now in the notes above. (1) The text said Green Drop audits "roughly 850–900 municipal" treatment works. Neither source gives that range: the paper's dataset covers 876 works and the press report says 848 municipal systems were audited in 2023/24. The sentence now says the country's municipal works, giving the paper's 876; "municipal" is the press report's word. (2) "Nobody has a metric or a monitoring system" for the missing wastewater went further than the paper, which says this information is absent from the Green Drop reports; the sentence now says that. (3) Sewage going "into streets" is not in the paper, which speaks of water systems; "streets" was removed. (4) The text said the authors link the Darvill deficit "directly" to pollution "in the uMsunduzi system." The paper says the deficit was matched by rising E. coli counts in the rivers of the Pietermaritzburg area and does not name the river system; the sentence now follows the paper. (5) The sentence that the scorecard "is what municipalities manage to" and so becomes "a blind spot in budgets and maintenance" is not in the paper. It now gives the paper's own statement, that a works can score well on financial and administrative management "while completely neglecting the infrastructure network." (6) "No design-capacity data" was imprecise: the 38.9% and 33% figures count works that reported no data on the share of design capacity in use. "More than half lacked continuous capacity monitoring" is the paper's own statement (more than 50% of works did not have continuous monitoring of that share) and stays as written. The 43% with no maintenance team comes from the 2023 report, and the text now says so. (7) "Improved transparency about treatment" is not in the paper; the sentence now says the programme's return showed that there had been considerable deterioration while it was halted, which is what the paper says. (8) Reticulation collapse being "invisible" to the audit and the Darvill network having "visibly failed" were stronger than the paper, which says such failures are often not captured and cites evidence of declining performance in the network; both were softened. (9) "Flow metering ... is patchy" and "metering the collection network upstream of the works is rarer still" are not in the paper, which reports only that 341 works gave no figure for design capacity in use in 2021; the sentence now says that. (10) "Macroinvertebrate" was removed from the description of miniSASS (the paper does not describe it that way), and "not tied into the audit" was replaced with what the paper says: the authors put these tools forward as first-order indications to add to routine monitoring. (11) The 14% to 8% figure was credited to "independent watchdog reports"; it comes from press coverage of the government's own 2023/24 Green Drop release, and the sentence now says that. The figure itself (118 to 66 systems; 848 audited) is confirmed in the Engineering News article. (12) The authors' method was described as "regressing historical inflow on population" and as "a workaround built for one city." The paper fits inflow and population each against year and projects their earlier relationship forward, as a case study; the sentence now says that. (13) In "What Would Unlock Progress," the paper's own recommendations were added and the rest is now marked as this write-up's proposal. Three statements not in the paper were cut: that water utilities have "decades" of non-revenue-water accounting, that smart-sewer sensors were "developed for combined-sewer cities," and that South African networks have "intermittent power." The comparison with the IWA water balance is kept, marked as not coming from the paper. (14) In the notes above, the E. coli rise from about 40% to about 90% of sites was dated 2010–2022; the paper dates it from about 2016 to the end of 2022, and the note now says so. Confirmed in the sources and left as written: the Source line; the 30% and 70% score weighting; the quoted phrases "a small and neglected component," "pervasive, severe sewage pollution," "may actually be motivated to not collect sewage," "while completely neglecting the infrastructure network" and "a much stronger focus on sewage reticulation"; 431 works with data in both years; the population rise of about 5.5 million (10%); Darvill at about 75 against about 95 ML/day expected; the 44% rise in sewer-connected flush toilets from 2011 to 2022; no penalty for operating below expected intake; Johannesburg Northern Works (18% less, score unchanged) and eThekwini Northern Works (50% less, score 2% lower); Darvill 79% to 78%; 155 of 876 works site-assessed in 2022; Water SA as a publication of the Water Research Commission. No title changed; note that the long title's "without its score moving" stands for a one-point change (79% to 78%) that the authors call "essentially unchanged," and "leaking sewers" is this page's wording for what the paper calls reticulation failures. Genome tags untouched.