Design, performance, and demand for a novel in-line chlorine doser to increase safe water access

Globally, 2.1 billion people still lack access to safe water at home. While most people now use improved sources, many sources do not meet SDG criteria for safely managed water and can be contaminated due to distribution system failures, intermittent supply, and recontamination. Point-of-use (POU) chlorination can improve water quality and reduce diarrheal disease but often has low sustained uptake because it requires consistent household behavior change. Point-of-collection (POC) chlorination at shared water points can shift the treatment burden from households to service providers. Automatic, in-line chlorinators offer advantages over manual dosers by not requiring user action and allowing any volume to be collected, but many available devices rely on solid chlorine tablets, which can face supply constraints. Building on prior pilot studies in Bangladesh that showed higher chlorination levels with in-line dosing compared with household tablets and highlighted issues like increased wait times with batch systems, the study aimed to develop and evaluate a low-cost, electricity-independent liquid-chlorine Venturi doser. The research questions were: (1) Does the device consistently deliver target chlorine residuals at kiosks over time? (2) Do customers choose chlorinated water when given a side-by-side option? (3) Are kiosk operators willing and able to pay for the device via service packages?

Evidence shows that improved sources are not always microbiologically safe, with an estimated 1.4 billion people using contaminated water from improved sources. POU chlorination is effective and low-cost but suffers from adherence challenges that undermine health impacts. POC treatment can increase adoption through social pressure with manual dispensers, yet it still requires user action and uptake varies. In-line chlorinators remove the need for behavior change and permit variable volumes but often use solid tablets that may be difficult to source reliably in some settings. Prior field trials of automated chlorination systems in Bangladesh reported higher chlorination at the POC versus household tablets and notable reductions in E. coli and child diarrhea at similar dosing levels. Automated batch systems like ZIMBA achieved consistent dosing but increased collection wait times. Emergency and development context studies of automatic chlorinators have reported detectable free chlorine in 77–100% of samples at POC but often involved free provision of devices or limited fee collection, leaving questions about long-term financial sustainability.

Study setting and design: The study was conducted at water kiosks in Kisumu County, Kenya (urban, peri-urban, rural) from August 2017 to May 2018. The implementing partner, Safe Water and AIDS Project (SWAP), acted as a distributor, performing sales pitches, installations, payment collection, weekly monitoring, and surveys. Eligibility criteria for kiosks included location in Kisumu County, reported daily operation, not currently chlorinated (KIWASCO-supplied kiosks excluded), and sufficient tap height (≥ ~65 cm) to fit the device above 20-L containers. Device and laboratory calibration: The Stanford-MSR Venturi is an in-line, electricity-independent chlorinator designed to attach at the tap outlet and function across flows of approximately 6–60 L/min. It includes a 1-L chlorine storage tank, float (constant head) tank, adjustable needle valve, sight tube, and Venturi injector. Key design parameters optimized via laboratory testing included a 12.7 mm Venturi throat diameter (balancing backpressure and accurate dosing across flows), 63.5 mm outlet-to-injection height (boosts low-flow dosing), and 2.0 mm float tank height relative to the injection point (prevents dosing delays and controls concentration). Design refinements addressed non-dosing at low flows by modifying flow area and vertical drop, and mitigated airlocks by repositioning the needle valve, upgrading to compression fittings and fluorocarbon tubing, and removing service loops. A security cover was added per user feedback. Dosing control: The needle valve (settings 0–10) controls dose. Prior to installation, each device was calibrated in the lab using 1.2% sodium hypochlorite solution (e.g., WaterGuard). Dechlorinated water flowed 30–60 s before sampling; flow was measured with a Dwyer VFC II, and free chlorine was measured immediately using a Hach Pocket Colorimeter II. Installation settings were selected to yield 0.5–1.5 mg/L free chlorine over ~6–38 L/min. Field target dosing was 0.2–1.2 mg/L to balance taste acceptability and WHO minimum residual guidance. Intervention and service packages: After baseline surveys with kiosk managers/operators (sales and management practices), kiosk owners received a sales pitch and chose among four 6‑month service packages: (1) Lease ($15/mo; option to purchase at end for lump sum; total $260), (2) Lease + chlorine delivery ($18/mo plus chlorine; lump-sum $170; total $278), (3) Lease-to-own ($42/mo; total $252), and (4) Lease-to-own + chlorine delivery ($46/mo; lump-sum $170; total $276). Prices were designed to cover an estimated $150 device sale price plus maintenance and chlorine. Payments were via M-PESA; installations occurred within 7 days of first payment. A limited number of devices (eight) were available. Follow-up measurements: SWAP staff visited kiosks weekly. They interviewed operators on daily volumes sold (chlorinated/unchlorinated) and prices for the prior week (usually from logbooks), satisfaction, and maintenance. Sales were self-reported (no meters). For dosing performance, staff collected three samples directly from the Venturi outlet upon arrival and measured free chlorine residual with the Hach colorimeter; if residuals were near or outside 0.2–1.2 mg/L, repeat testing was performed and the dosing dial adjusted at staff discretion, with post-adjustment samples collected. Staff also reminded kiosks about payment schedules. Post-intervention surveys were conducted with customers (≥10 per kiosk, 2–6 days) at 3–6 months post-installation, and with kiosk operators/managers at 2–5 months post-installation. Ethics and analysis: Written informed consent was obtained (Luo, Swahili, or English). Approvals: Maseno University Ethics Review Committee and Stanford University IRB (protocol 40689). For analysis, the mean of the first three chlorine measurements per visit was used; detectable residual defined as >0.05 mg/L (instrument range 0.02–2.00 mg/L). Data were aggregated by months since installation. Because kiosks often closed (e.g., rain, management issues, election-related violence), missing or closed days were treated as zero sales. Analyses were conducted in Stata/SE 14; figures in R.

Enrollment and device installation: Of 26 kiosk owners pitched, 7 (27%) committed to a service package and had devices installed. Kiosks represented urban, peri-urban, and rural settings; all sources were boreholes except one spring. Technical performance: In the lab, all 71 samples from devices had detectable free chlorine; at selected needle valve settings, residuals were 0.91–1.48 mg/L over 6–38 L/min. In the field, 163/167 (97.6%) samples from kiosk taps had detectable residuals (>0.05 mg/L); 144/167 (88.0%) were ≥0.2 mg/L (WHO minimum), 164/167 (98.2%) were ≤1.2 mg/L (taste threshold), and 144/167 (86.2%) were within the 0.2–1.2 mg/L target. Mean residual was 0.55 mg/L (SD 0.29). Mean monthly residuals remained between 0.37 and 0.74 mg/L over six months. The dosing dial was adjusted on 5/167 visits (3.0%). The device functioned across observed flows of 2.2–63.2 L/min. One kiosk (F) experienced non-dosing due to iron-related sediment clogging after month 3; performance improved after device replacement at month 6. Payments and purchasing: Six of seven kiosks fulfilled all monthly payments during the study; one (D) defaulted and the device was uninstalled after 4 months. Three kiosks chose lease-to-own (C, E, G) and all completed ownership; among four lease kiosks, three paid in full (A, B, F), and two of these (A, F) purchased the device at end of term. Overall, five of seven kiosks ultimately purchased devices. Only 60% of payments were on time; average time to complete purchase payments was 7.5 months (vs 6 months planned). More than two-thirds of kiosks purchased with total payments exceeding $250 per device. Sales outcomes: Chlorinated water comprised 19.1% of total water sales across enrolled kiosks, with substantial variation by kiosk and season. Only one kiosk earned more monthly revenue from chlorinated water alone than its monthly service payment; however, five of seven kiosks earned more from combined chlorinated and unchlorinated sales than their monthly payment. Kiosks were fully open for all days in 69.2% of weeks with data, with 30.8% reporting at least one closure per week. Pricing strategies varied; three kiosks priced chlorinated and unchlorinated similarly, while four charged more for chlorinated water. Customer and operator perceptions: Customers reported purchasing a mean of 142 L/day (median 100 L) from kiosks; 94.4% purchased at least two days per week; average one-way walk time was 9.4 minutes. Despite using additional sources (91.6%), 66.2% reported the kiosk provided >60% of household water. Among customers, 66% reported buying chlorinated water; 52.0% bought both types, 14.1% only chlorinated, 33.8% only unchlorinated. For drinking, 89.4% of those collecting chlorinated water used it for drinking versus 24.6% for unchlorinated. Perceived safety was higher for chlorinated water (83% very safe) than unchlorinated (18% very safe). Mean prices were 4.2 KSH per 20-L for chlorinated and 3.0 KSH for unchlorinated. Satisfaction was high: 90.1% of customers were very or somewhat satisfied; 64.8% cited the device’s attractive/high-quality appearance; 76.1% reported no dislikes. All operators/managers reported satisfaction; 84.6% cited convenience of use as a positive. Most managers (87.5%) cited water sales profits as the primary payment source, though 75% found monthly payments somewhat or very difficult.

The Venturi doser achieved consistent, precise chlorination at point-of-collection kiosks over six months, addressing the need for reliable disinfection without requiring household behavior change. With 97.6% detectable free chlorine and 86.2% of samples within the 0.2–1.2 mg/L target, performance compares favorably with prior in-line chlorinator field trials. Minimal adjustments (<5% of visits) suggest low ongoing maintenance needs. A critical caveat is susceptibility to iron-related clogging; a prefilter or source screening for iron may be necessary in high-iron settings. Financially, the service packages demonstrated viability for many kiosks: five of seven ultimately purchased devices, indicating effective demand and perceived value among operators, though only 27% of pitched kiosks enrolled, suggesting selection by ability/willingness to pay. Payment timeliness was a challenge, and operators expressed a preference for longer or more flexible lease-to-own terms, potentially seasonalized to align with higher-demand dry seasons. Importantly, combined water sales generally covered monthly payments for most kiosks, even if chlorinated water alone did not, indicating that offering chlorinated water alongside unchlorinated can be financially sustainable within a kiosk’s broader revenue model. Customer behavior and perceptions were encouraging: two-thirds purchased chlorinated water when given a choice, most used chlorinated water for drinking, and perceived safety was substantially higher for chlorinated water. Pricing strategies influenced uptake, with some kiosks achieving higher chlorinated sales shares by keeping prices similar or employing innovative sales tactics (e.g., requiring small chlorinated purchases or moving the device to meet demand). Relative to existing technologies, the Venturi offers several advantages: no electricity requirement, minimal impact on collection time, compatibility with widely available liquid chlorine (e.g., WaterGuard), potential affordability with local hypochlorite production, and an estimated molded unit cost of $34 with a projected five-year lifespan. These features, together with demonstrated technical performance and operator demand, suggest substantial potential to expand safe water access in low-resource settings via POC chlorination, especially where centralized treatment is infeasible.

A novel, low-cost, electricity-independent Venturi-based in-line chlorinator consistently delivered target free chlorine residuals at Kenyan water kiosks with minimal maintenance and strong user and operator acceptance. Effective demand for the device was demonstrated: most participating kiosks completed payments and ultimately purchased devices, though payment schedules may need to be more flexible to improve timeliness and inclusivity. Customers frequently chose chlorinated water, particularly for drinking, and perceived it as much safer. These results indicate that point-of-collection in-line chlorination, delivered via financially sustainable service packages, can increase access to microbiologically safer water in low-income communities. Future work should evaluate long-term performance beyond six months, establish iron threshold screening and/or prefiltration needs, assess microbial outcomes and health impacts in Sub-Saharan Africa, and analyze business model profitability for real-world distributors with less intensive monitoring. Exploring extended and seasonalized lease-to-own options may enhance adoption and payment compliance.

The study included a small sample of kiosks (n=7 installed) with a six-month follow-up, limiting generalizability and the ability to assess long-term performance and maintenance needs. Kiosks self-selected into enrollment, potentially biasing results toward those more able/willing to pay. Payment collection and technical support were facilitated by an NGO with weekly visits, which may not reflect real-world distributor operations and costs. Water sales volumes were self-reported (no meters) and kiosks frequently closed, possibly underestimating sales potential. Iron concentrations were not measured, preventing establishment of a clogging risk threshold. Microbial water quality and health outcomes were not assessed in this cohort, though prior studies indicate expected benefits at similar dosing levels. The study could not evaluate profitability for a commercial service provider.