where the sludge goes, nobody counts

Problem statement

Globally, about as many people rely on on-site sanitation (pit latrines and septic tanks) as on sewers, and on-site systems are growing faster than sewers even in cities. The Sustainable Development Goal indicator that matters — "safely managed" sanitation — asks whether the excreta from those pits and tanks is either safely disposed of in place or emptied, transported and actually treated. For sewered populations that question is answerable from wastewater-plant records; for on-site sanitation it mostly is not. The JMP reports that data on excreta "emptied and treated off-site" were available for just 1% of the global population (JMP Table 3; on-site sanitation was used by 43% of the global population in 2020), from only seven countries, all high-income. The failure is structural: households cannot say what happened to their sludge after the truck (or the man with a bucket) left; emptying is often informal and unrecorded; and where treatment-plant records exist they count trucks or cubic metres from undefined catchments, which cannot be reconciled with the population-based numbers that surveys produce. Nobody — not the country, not the city, not the plant — can produce a defensible number for how much faecal sludge is generated, where it goes, and how much is treated.

Why this matters

Without that number, unsafe emptying and dumping — the JMP calls it "a growing risk to public health and the environment" — is invisible in national statistics, so investment in emptying services and treatment capacity cannot be sized or justified, and progress toward SDG 6.2 cannot be tracked for the on-site half of humanity that includes "the world's poorest people." The same missing data cripples engineering: Eawag's Sandec group notes there has been "no reliable method to estimate total accumulated quantities and qualities" of faecal sludge because it "is often stored underground in containments, records frequently do not exist, and construction is not standardized," so treatment plants have been designed from literature values borrowed from other cities. In the JMP's absence-of-data rules, countries where sewers dominate are simply assumed to treat sludge as well as wastewater, and where on-site dominates no estimate is made unless the country has its own sludge-treatment data — meaning the countries with the most on-site sanitation are the ones most likely to have no safely-managed figure at all.

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

The JMP, with Gates Foundation support, ran SMOSS data-collection pilots in six countries (including Bangladesh n=3,149, Zambia n≈23,000, Serbia n=1,560, plus Indonesia, Ecuador and Kenya) from 2020, adding questions to national surveys and running dedicated household surveys with sanitary inspections. Household questions worked for containment and for whether a pit had ever been emptied, but hit a wall downstream: "household survey respondents can't reliably indicate if fecal sludge is treated once it leaves the property; additional information is needed from administrative sources." Administrative data proved the weak link — "few [countries] maintain centralized records of desludging services that would allow calculation of how much of the removed excreta is actually delivered for treatment," and when records exist "it is often in different units (i.e. number of trucks, volume or mass of waste) or from undefined catchment areas that make it difficult to align it with household survey data or determine the equivalent population served." Volumes delivered to plants are "not a direct function of the number of users but also depend on the containment size and emptying practices (partial or full emptying, adding water, etc.)." Where informal or private emptiers operate alongside regulated ones, "the data may only cover a portion of the emptying activities." Integrating questions into national surveys was constrained by "the limited number of questions that could be included, the low capacity of enumerators regarding technical sanitation aspects," and the fact that on-site sanitation is often regulated by different government departments than sewered services. Sandec's Q&Q method — predicting sludge quantities from spatially available demographic, environmental and technical data — is promising (1,206 data points; pit latrines ≈270 L/cap·year, septic tanks ≈280 L/cap·year in Kampala) but its authors describe it as "still a baby" needing context-specific adaptation.

What would unlock progress

The missing piece is a reconciliation layer that turns what is actually recorded — truck trips, plant intake volumes, emptier logbooks, GPS traces — into population-equivalent flows that survey data can be compared with: a "faecal-flow accounting" method analogous to the mass-balance and material-flow accounting used in solid-waste and energy statistics, or to the way remittance flows are estimated from partial formal records. Cheap instrumentation of the emptying step (a truck-mounted volume/GPS logger, or a mobile-money-linked emptying receipt) would make the informal chain legible without a census; predictive Q&Q models could fill the gaps between measurements. On the governance side, the JMP notes there are "no global assumptions regarding safety of different types of emptying or transport" — countries must define them — so a defensible, reusable classification scheme would itself be a contribution.

Entry points for student teams

A team could build and test, with one city's data (several African and Asian cities have published shit-flow diagrams and plant intake records), a reconciliation model that converts truck-load and plant-intake data into "population served" with explicit uncertainty, then compare it against household-survey emptying rates to expose the informal share. A hardware team could prototype a low-cost emptier logger (volume, timestamp, GPS, disposal-site geofence) and pilot it with a vacuum-truck operator. A data/GIS team could reproduce Sandec's SPA-DET approach for a new city from open demographic and containment data. Skills: environmental engineering, survey statistics, GIS, embedded systems, public-sector data systems.