JolWatch Edge: Offline-First Water Intelligence and Deterministic Leak Detection for Low-Connectivity Facilities

1. Introduction & Problem Statement

Water distribution systems in schools, primary healthcare clinics, and community facilities in developing economies frequently operate under intermittent municipal supply or decentralized tube-well pumping. Community facilities frequently experience critical water, sanitation, and hygiene (WASH) infrastructure limitations (World Health Organization, 2019). Furthermore, achieving universal water access under Sustainable Development Goal 6 targets requires substantial capital investment and proactive loss mitigation (Hutton and Varughese, 2016). Facilities store water in overhead reservoirs to buffer against supply disruptions. However, piping infrastructure in these settings suffers from high rates of hidden leakage behind walls and underground, as well as mechanical float-valve failures that cause undetected overflows. Empirical plumbing end-use studies demonstrate that fixture leakage and unmonitored line failures represent significant fractions of aggregate municipal draw (Mayer et al., 2016).

Traditional monitoring relies on delayed utility billing, leaving leaks undetected for weeks. Modern commercial IoT monitoring solutions, while technologically advanced, are engineered for high-bandwidth cloud backends with recurring SaaS subscription fees. In resource-constrained facilities throughout the Global South, where internet outages are common, cloud-dependent alert pipelines fail to provide timely intervention.

2. Mathematical Formulation

2.1 Flow-Balance & Loss Rate Derivation

Consider a facility plumbing network with an instantaneous volumetric inflow \(Q_{in}(t)\) (L/min) and \(k\) monitored downstream sub-zones \(Q_{z,i}(t)\) (e.g., washrooms, kitchens, outdoor grounds, drinking points). The recorded aggregate demand \(Q_{zones}(t)\) is:

2.2 Decision Classification Rules

To maintain operational transparency, JolWatch Edge avoids opaque machine learning models in favor of five deterministic, explainable rules:

• Sensor Mismatch: If \((Q_{zones} - Q_{in}) > 2.0\text{ L/min}\), reject alert and flag sensor drift or meter misalignment.
• Possible Tank Overflow: If \(\text{TankLevel} \ge 95\%\), \(Q_{in} > 10\text{ L/min}\), and \(\text{LossRate} > 25\%\), trigger overflow alert to inspect shutoff valves.
• Possible Hidden Leak: If \(\text{LossRate} > 15\%\) and persists for \(\text{Duration} \ge 15\text{ minutes}\), trigger hidden leak alert.
• Monitor Warning: If \(\text{LossRate} > 8\%\), flag elevated imbalance requiring operator observation.
• Normal: Unexplained flow remains below operational action thresholds.

2.3 Essential Reserve Clock Projection

Given reservoir capacity \(C\) (L), level \(H(t)\) (%), and protected reserve fraction \(\beta\) (\(0.20\) for schools, \(0.30\) for clinics):

3. System Architecture & Edge Feasibility

The reference software implementation for JolWatch Edge (Khan, 2026) is designed for deployment on low-cost ESP32-class microcontrollers interfacing with hall-effect pulse flow sensors (YF-S201) and waterproof ultrasonic depth sensors (JSN-SR04T). The computation engine is fully self-contained in js/engine.js, requiring zero external runtime dependencies. Local persistence utilizes client-side storage with an exportable JSON schema, ensuring zero telemetry leakage.

4. Validation & Deterministic Benchmarking

5. Honest Disclosures & Research Roadmap

Current Prototype Status: This work represents a validated software implementation using synthetic benchmarks. Physical sensor rig deployment, empirical calibration of thresholds against physical pipeline friction, and long-term field validation remain subsequent milestones.

Water Potability Disclaimer: The tank maintenance attention indicator evaluates physical observational proxies (turbidity, TDS shift, temperature, inspection intervals). It cannot detect microbiological pathogens, viruses, or dissolved arsenic, and does not certify drinking water safety.

6. Conclusion

JolWatch Edge demonstrates that practical water conservation and leak mitigation in developing facilities do not require continuous cloud access or expensive proprietary hardware. By running transparent flow-balance heuristics on edge runtimes, facilities can safeguard essential water reserves, detect infrastructure failures early, and verify maintenance outcomes locally.