My home started seeing high water bills recently. My water company doesn't provide live water meter readings. There is a water meter on the street which I can monitor while water is being used but I would like to build a more comprehensive system which can present live water usage on some screen. The end goal is to show how much water is being consumed at any given time, live.
I might be able to buy fancy smart water from the market and call it done but I would like to build the system myself. I am relatively skilled in electrical/electronics/computer engineering (my profession) and I have some basic knowledge of water pressure and related physics concepts.
I thought of implementing the system this way.
I will install a high resolution water pressure sensor on one of the taps inside the house. This will show me live instantaneous pressure into my house. Assuming no water is being consumed, this water pressure shows the water pressure level from the city to my house. Whenever there is some water use, I imagine that pressure will drop from a relatively stable baseline by some amount proportional to the flow rate. It wouldn't matter which tap in the house is draining water; I expect the sensor to show some pressure drop.
After the pressure is dropped, the city's main water line will attempt to recover the pressure to some extent (either fully or partially). This will bring the pressure to earlier baseline and even beyond the baseline because there is damping component that exists in the house piping before the water pressure can stabilize. I expect to then see some damped oscillations for this event; first starting with a pressure drop and then reducing oscillations until stability is obtained. The opposite scenario plays out then the draining tap is closed. This cause the water pressure sensor to register a pressure increase, which again after oscillations, gets to a baseline. I did this experiment, gathered the data and this is exactly what I see. I captured some graphs which you can see on this post (for a related albeit a different question: https://dsp.stackexchange.com/questions/68328/how-to-detect-a-signal-while-ignoring-another-signal-which-is-a-phase-shifted-ve).
I am aware that city's incoming water pressure is not constant (neighbors share the water line) but that water pressure disturbance would be high frequency random noise (many people using at any given time randomly). My problem now is how do I translate this data to a flow rate for my home?
I know that flow rate is proportional to square root of differential water pressure. I cannot install water pressure sensor before the city meter to get differential water pressure reading so I am looking for other creative ideas. Perhaps I can characterize the system (measures the impulse response) and then predict flow rate?