Single-Point Pressure Sensor for Household Water Leak and Tap Detection

The problem: water is scarce, and homes cannot see where it goes

Cities across Pakistan and South Asia face growing water stress. Households pay for tanker deliveries, run pumps for hours, and still do not know where their water goes. A running toilet cistern or a leaking pipe inside a wall can waste a large share of a household's supply without anyone noticing until the tank runs dry or a wall shows damp.

Installing a flow meter on every fixture is impractical. A single sensor that can tell taps, showers and toilets apart, and spot leaks, would change that. Research has shown this is possible on pressurised plumbing. Whether it works on the gravity-fed rooftop tank systems used in most South Asian homes is an open question, and an excellent one for a final year project.

What already exists and the gap

• Phyn Plus measures water pressure at high frequency on the main supply line, fingerprints fixtures and detects leaks, with an automatic shut-off valve. It is designed for pressurised North American plumbing and installed by a plumber.
• Flo by Moen monitors pressure, flow and temperature on the main line and runs periodic pressure tests to find small leaks.
• Tank level sensors sold locally show how full the tank is, not which fixture used the water.

The gap is a low-cost, install-it-yourself sensor that has been tested on South Asian tank-fed plumbing, with honest results about where it works and where it does not.

Inspiration

Froehlich and colleagues introduced HydroSense at UbiComp 2009: a single pressure sensor attached to any accessible tap, which identified activity at individual fixtures from the pressure transients created when valves open and close. They reported fixture identification with 97.9 percent aggregate accuracy in their test homes. Those homes had pressurised supply. Your project reproduces the method and tests it in a very different plumbing setting.

What you build

• The sensor unit threads onto a spare tap or tank outlet with an adapter, measures pressure at high rate and runs event detection on the ESP32.
• The data pipeline stores event windows with timestamps and lets you label them during data collection sessions.
• The models segment events, classify fixtures and look for leak signatures.
• The household sees daily water use by fixture and receives an alert if a leak is suspected.

The technically interesting part

• Low pressure, small signals. A rooftop tank one or two storeys up gives far less pressure than a mains supply, so valve transients are smaller and noisier. Choose a sensor range and ADC resolution to match, measure the noise floor, and find out which fixtures are still distinguishable.
• Sampling and events. Water hammer transients last tens of milliseconds. Sample fast enough to capture them, detect events on the device and send only short windows around them to save bandwidth.
• Features and classification. Extract features from the open and close transients and the steady pressure drop, then train a classifier. Report accuracy on homes not used for training, because plumbing differs from house to house.
• Leaks. A slow leak shows as a small, steady pressure drop or a system that never returns to its quiet baseline overnight. Simulate leaks with a slightly open tap and measure the smallest leak you can detect.
• Pumps. Many homes use a pressure pump on top of the tank. Its cycling dominates the signal, so detect and separate pump events.

How to evaluate it

• Fixture identification: accuracy per fixture type, within a home and across homes.
• Plumbing comparison: the same metrics on gravity-fed and pump-pressurised systems, side by side.
• Leak detection: smallest detectable leak rate and false alarm rate over several quiet nights.
• Practicality: installation time by a non-expert following your guide.

Suggested plan

• Weeks 1 to 4: sensor selection, test rig with a raised tank and several fixtures, firmware for high-rate sampling.
• Weeks 5 to 9: event detection, labelling tool, first data from the rig.
• Weeks 10 to 14: data collection in real homes, features and classifier.
• Second semester: leak detection, pump handling, cross-home evaluation, dashboard and report.

Path to a product

Water utilities, housing societies, apartment buildings and environmentally minded households all want to cut water waste. A low-cost sensor with a clear leak alert is a credible product for South Asian and Middle Eastern markets, where tank-fed plumbing is common and existing products do not fit. The research result alone, how pressure sensing behaves on tank systems, is publishable.

Scope warning

Do not modify household plumbing beyond attaching to an existing tap with a proper adapter, and get the owner's permission. Do not install shut-off valves in the first version. Report negative results honestly; finding that some fixtures cannot be separated on gravity systems is a valid outcome.

Learn what you need on Sefism

• Embedded systems and IoT roadmap: ADCs, sampling and firmware design.
• Embedded systems and IoT cheat sheet: interrupts, timers and data buffering.
• Introduction to RTOS (video series): structuring high-rate sampling with tasks and queues.
• Machine learning roadmap: features, classification and cross-home evaluation.
• Making Embedded Systems, 2nd Edition: designing robust embedded software.

Get help with this idea

The real research question is whether the signal survives low tank pressure. Tauseef can help you design the test rig, the labelling sessions and an evaluation that is honest about what works. If you are new here, create a free account first.

• Discuss this idea with a mentor
• Send a query to Tauseef
• Book a 1:1 session