LPG Cylinder Gas Level and Leak Monitor for Restaurant Kitchens

The problem kitchens have with LPG cylinders

Restaurants, roadside dhabas, hostels and many homes cook on LPG cylinders. Two things go wrong. First, nobody knows how much gas is left, so the cylinder runs out in the middle of lunch service and someone rushes to buy a refill at whatever price is on offer. Second, leaks from worn regulators and pipes cause fires, and a leak at night in a closed kitchen is especially dangerous. Owners solve the first problem by shaking the cylinder and the second by hoping.

What already exists and the gap

• Mopeka Pro Check uses an ultrasonic sensor stuck to the bottom of a steel cylinder and reports level over Bluetooth. It is designed for barbecue and RV propane tanks and is priced for that market.
• Commercial kitchen gas detectors from suppliers such as Flowtech combine an LPG detector with an automatic shut-off valve, but they are alarm panels only: no gas level, no history, no phone app.
• Cheap MQ-2 or MQ-6 leak alarm modules are sold everywhere, drift over time, and false-alarm often enough that people unplug them.

The gap is one affordable device that answers both questions (how much gas is left, is anything leaking), closes the valve on a real leak, and keeps a record the owner can show a landlord or insurer.

What you build

• Weighing platform: a flat plate on four load cells feeding an HX711 amplifier, read by an ESP32. The owner enters empty (tare) and full weights once.
• Leak node: an LPG-sensitive MQ sensor mounted low (LPG is heavier than air), a siren, and a relay that drives a normally closed solenoid valve on the gas line.
• App: shows gas left, usage per day and the predicted empty date, plus alarm history.
• Backend: stores readings and sends push notifications.

Main flows: daily usage learning and refill alerts; a leak raising the local alarm, closing the valve and notifying the owner; and a manual reset that requires someone to be physically present.

The technically interesting part

Both sensors are noisy and slow, and the safety decision must be local. Load cells drift with temperature and creep under constant load, so you will need a calibration routine, filtering and a way to ignore the spikes when someone bumps or swaps the cylinder. MQ sensors need a warm-up period and a moving baseline, and their readings are affected by humidity and cooking fumes. Your job is a leak detector that closes the valve fast on a real leak but does not trip every time someone fries onions. Show evaluators the data you collected, the thresholds you chose and the false alarm rate over a week in a real kitchen.

How to evaluate it

• Weight accuracy at five known weights, and drift over 24 hours with a constant load.
• Refill prediction error: predicted empty date against the actual date across at least two cylinders.
• Leak alarm response time from a controlled gas source at a fixed distance, repeated ten times, done safely outdoors with a supervisor.
• False alarm count over one week of normal cooking in a real kitchen.

Suggested plan

1. Weeks 1 to 3: load cell platform, calibration and filtering.
2. Weeks 4 to 6: gas sensor characterisation, baseline logic and siren.
3. Weeks 7 to 8: valve control, fail-safe reset flow and tests.
4. Weeks 9 to 11: backend, app and refill prediction.
5. Weeks 12 to 15: kitchen trial, measurements and report.

Path to a product

The first customer is a small restaurant or hostel mess that has run out of gas during service or had a leak scare. Sell the device for a one-time price with installation by a gas fitter, plus an optional monthly fee for the app and alerts. Gas distributors are a strong channel: they get refill orders from the app and could bundle the device. Scale by selling to hostel chains and university kitchens.

Scope warning

Do not claim the product is a certified safety device; it is a prototype and you should say so. Do not build your own gas sensor, supplier marketplace or delivery tracking. Do not modify the cylinder or regulator itself; install the valve only on a pipe section fitted by a qualified person.

Learn what you need on Sefism

• Programming cheat sheet : quick reminders for the C++ you will write for filtering and state logic.
• Mobile roadmap : building the owner app with push notifications.
• Firebase Documentation : storage and push notifications without running your own servers.
• Supabase Documentation : a Postgres-based alternative if you want SQL for usage history.
• How to get better at debugging : sensor bugs rarely show up in a stack trace, so method matters.

Get help with this idea

The safety logic deserves careful thought, so plan how you will test it before you order parts. If you want a second opinion on scope, hardware choices or how to set up the evaluation before you commit, bring your plan and your bill of materials and we will pressure-test it together. New here? Create a free account first.

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