Learning from intermittent water supply schedules: Visualizing equality, equity, and hydraulic capacity in Bengaluru and Delhi, India.

Intermittent distribution affects one in five piped water users, threatens water quality, and magnifies inequity. Research and regulations to improve intermittent systems are hindered by system complexity and missing data. We created four new methods to visually harness insights from intermittent supply schedules and demonstrate these methods in two of the world's most complicated intermittent systems. First, we created a new way to visualize the varieties of supply continuities (hours/week of supply) and supply frequencies (days between supplies) within complicated intermittent systems. We demonstrated using Delhi and Bengaluru, where 3278 water schedules vary from continuous to only 30 minutes/week. Second, we quantified equality based on how uniformly supply continuity and frequency were divided between neighbourhoods and cities. Delhi provides 45 % more supply continuity than Bengaluru, but with similar inequality. Bengaluru's infrequent schedules require consumers to store four times more water (for four times longer) than in Delhi, but Bengaluru's storage burden is more equally shared. Third, we considered supply inequitable where affluent neighbourhoods (using census data) received better service. Neighbourhood wealth was inequitably correlated with the percent of households with piped connections. In Bengaluru, supply continuity and required storage were also inequitably divided. Finally, we inferred hydraulic capacity from the coincidence of supply schedules. Delhi's highly coincident schedules result in city-wide peak flows 3.8 times their average - sufficient for continuous supply. Bengaluru's inconvenient nocturnal schedules may indicate upstream hydraulic limitations. Towards improved equity and quality, we provided four new methods to harness key insights from intermittent water supply schedules.

Effects of hydraulically disconnecting consumer pumps in an intermittent water supply.

We estimate 250 million people receive water using private pumps connected directly to intermittently pressurized distribution networks. Yet no previous studies have quantified the presumed effects of these pumps. In this paper, we investigate the effects of installing pressure-sustaining valves at consumer connections. These valves mimic pump disconnection by restricting flow. Installing these valves during the dry season at 94% of connections in an affluent neighborhood in Delhi, India, cut the prevalence of samples with turbidity > 4 NTU by two thirds. But considering the poor reputation of pumps, installed valves had surprisingly small average effects on turbidity (-8%; p<0.01) and free chlorine (+0.05 mg/L; p<0.001; N = 1,031). These effects were much smaller than the high variability in water quality supplied to both control and valve-installed neighborhoods. Site-specific responses to this variability could have confounded our results. At the study site, installed valves increased network pressure during 88% of the typical supply window; valves had a maximum pressure effect of +0.62 m (95% CI [0.54, 0.71]; a 40% increase vs. control). Further research is needed to generalize beyond our study site. Nevertheless, this paper provides unique evidence showing how the deployed valves mitigated pump effects, increased network pressure and improved water safety.

Embedding Usage Sensors in Point-of-Use Water Treatment Devices: Sensor Design and Application in Limpopo, South Africa.

Health benefits from point-of-use (POU) water treatment devices come only with consistent use. Embedded sensors can measure the consistency of POU-device use and can provide insights about improving it. We demonstrate both potentials with data from SmartSpouts: accelerometer-based sensors embedded in spigot handles that record the duration and timing of use. In the laboratory, most sensor readings correlated well (>0.98) with manually timed water withdrawals. In the field, SmartSpouts measured >60,000 water withdrawals across 232 households in Limpopo, South Africa. Sensors proved critical to understanding consistent use; surveys overestimated it by 53 percentage points. Sensor data showed when households use POU devices (evening peaks and delayed weekend routines) and user preferences (safe storage over filters). We demonstrate analytically and with data that (i) consistent use (e.g., 7 continuous days) is extremely sensitive to single-day use prevalence and (ii) use prevalence affects the performance of contact-time-based POU devices, exemplified with silver tablets. Deployed SmartSpouts had limitations, including memory overflows and confounding device relocation with water withdrawal. Nevertheless, SmartSpouts provided useful and objective data on the prevalence of single-day and consistent use. Considerably less expensive than alternatives, SmartSpouts enable an order of magnitude increase in how many POU-device sensors can be deployed.