Detecting Water system problems before...

[KurtEck]

... You run out of water!

Every few years, I keep thinking of doing something about this. Usually after the water stops flowing out of the tap

Yesterday was one such day. The water pressure dropped off, so I walked up to our well house (which is maybe 50 yards from the house, tapped on the 2800 gallon water tank and it was empty. Luckily float valve in tank keeps pressure pump from running... Checked our two wells. One of them turned on for a few minutes and cycled off (no water, needs time to recharge), and the main one did not turn on at all... Luckily called well pump company who was out in a couple hours and found the well controller was dead, which they replaced (actually replaced both) and now we have water again

Each time something like this happens, I think I should do a project to monitor the system.

Some of the things I should probably monitor:

1) Tank level - Not sure best way to measure. Ultrasonic, Add a few more floats...

2) Know when and how long each Well is running and likewise maybe likewise pressure pump. Wonder if maybe hooking up a current sensor like: https://www.sparkfun.com/products/11005 to at least one of the legs for each of these (all three of these are on separate circuit breakers. Should be able to detect when each turns on and off, and potentially maybe detect information about the pumps... Not sure if I would get any more meaningful information if I measured both legs going to each of these pumps.

If this time around I actually do something about this. I can imagine breaking this down into a few different parts.

Collect the data: Maybe start off with a log of: Which motor, turn on time, turn off time. And maybe tank level. Could add other things like: temp in well house... Trying to decide what processor to use here (Teensy? RPI...)

Maybe send the data to the house: Not sure best way. Again maybe 50 yards. there are trees and the like, no wires running from Well to house (or barn). Pump house is on own Meter which is on different Transformer... Probably too far normal XBees, don't want wifi to go out that far...

Some way to alert me. That is once I run into problem like this, for the next few weeks, I know I will be good and walk to pump house at least once a day to check the tank, but soon that may drop off to once every few days, then maybe once a week... So probably need some way to alert me when things don't appear normal...

Anyway we will see if this time I actually do something

Suggestions?

Kurt

[tician]

Perhaps a pair of these with its Arduino Zero, LiPo battery charger, and 2~20km range (depending on settings and antennae)? The cheaper, shorter range (500m) versions using the RFM96 are all out of stock.

A single current sensor on each hot wire should be enough to tell when the pumps are running, although it does not necessarily mean they are doing useful work. Sending both the hot and neutral through a single current sensor is the basis of ground fault interrupters: current measured should be zero with both wires carrying exact same current in opposite directions so the fields cancel out; any leakage will create an imbalance in the currents resulting in a detectable field.

If the tank were smaller, then inexpensive load sensors at the mount points would give a very accurate measure of the water remaining in the tank. Sparkfun carries some rather cool purely optoelectronic level (on/off) sensors that can mount in the walls of tanks, although they are a bit pricey. Accuracy of any sensor will decrease with increased sediment accumulation in the tank. Know someone with a well that dumps so much clay sediment into the water flow that it only takes a couple weeks to clog the aerators/filters on all the taps and appliances.

[KurtEck]

Thanks,

They do look pretty interesting with the longer range, which probably may imply might be able to get the signal through the trees and maybe walls...

As you said probably one current sensor per motor would be sufficient. I read in some uses that sometimes helps to have one on each power leg (240v), to see about uneven power usage...

Also potentially recording the actual current usage may be useful, to see if something appears to be changing over time. Not sure how much in my case.

Tank level. Potentially could use a couple of the spark fun ones maybe sealed into a pipe that is lowered in from the top of the tank. So it can potentially tell me some key information, like the tank is half empty.

Maybe the first step would simply know if the tank is or has been full recently. I am thinking maybe with the same current sensors, I may be able to deduce this. That is, when the tank is not full, it turns on power to the wells. The wells will turn off one of two ways.

If the tank fills, the float valve will turn on and the power is shutoff to the wells. So the current should in theory drop straight off at the end. However if instead the well goes dry, the pump will start to use less power at which time the pump saver circuit detects this and shuts off the pump and sets up a timer to restart the pump after some delay like 100 minutes. So I am guessing that my current sensor should detect the drop as well and I can deduce (Not full).

Maybe sounds like I need to order some parts to try it out.

[tician]

I was assuming single-phase (120V or 230V). For split-phase pumps (240V or 460V) it would not hurt to have a sensor on each hot wire. For two or three-phase systems, a current sensor is very much recommended on each hot wire since running the motor after any phases cut out can potentially cause significant damage. With the Feather M0 having 10 ADC inputs there should not be any trouble checking them if they simply get added later.

[Brooks]

1) Deep-discharge 12v battery on float charger
2) At least two separate floats plumbed to signal when tank is half full
3) Car horn. Or bury light-duty burial-able wire to a 12v light in the kitchen. Your neighbors might get annoyed by the horn
4) Arduino (you knew I'd get there eventually!) to toot horn or flash light for a 2-second burst every 30 seconds
5) Alert if power is lost or either float shows tank is half-empty.
- For me, I'd put the light in the bathroom. It's more likely I'd see it...

[KurtEck]

Thanks,

I ordered Two of the feathers, also I ordered 2 Teensy 3.2s and 2 RFM69 breakout boards from 3 of the current sensors Sparkfun (So I will have a couple of options to try out).

I am also thinking about maybe getting one of their liquid sensors, but did not see anywhere that mentioned if they were safe for potable water systems...

Brooks, I have also thought about a low tech solution as well, Could be as simple as another Float sensor, mounted for half way, which turns on a light probably mounted on exterior of pump house.

[jwatte]

Another option is ultrasonic at the top of the tank, aimed down.
If you will fill the tank to 100% at times, it needs to be watertight and potable-safe, which narrows the field, but if you never fill it 100%, you can probably get away with an automotive sensor like the ones they have in bumpers.

[Brooks]

Low-tech is good sometimes! If you lose power, or take a close lightening strike, batteries become your friend! And I like your mid-tank sensor idea. That and a float-charged battery will cover you through almost anything but the kind of disaster where dead pumps are the least of your problems!

[KurtEck]

Quick update: Yesterday shipment arrived from Adafruit with two Feather M0 with LoRa 900mhz chips. So far I have done a little testing, by soldering on 3" wire antenna and some test programs that allow the two to talk to each other. Next up will to see how far I can get these two to talk.

Tomorrow shipment from Sparkfun with the current sensors and other stuff arrive, so I can start to work on program to collect data and hopefully be able to send it...

However today I have another fun interruption. I received one of the first beta boards for new Teensy. The first board is much larger than the final one will be, which is one inch longer than current T3.2. No pictures allowed, but, these boards should be fun. (single floating point, more SPI, I2C, USarts... Also SDIO and host usb... More details up on beta thread up on pjrc forums.

[Slugman]

Sorry for joining this late - Have been away on holidays & sporty stuff, but water is something I deal with a lot at work...
If it's just 50 yards away, how about a float inside, connected to a weight & pully with a reverse measuring gauge on the outside of the tank. When the weight/marker is at the bottom of the gauge, the tank is full, & when it's near the top (& likely more visible), then the tank is near empty?
Pro - It's low tech with one each of a float, cable, pully, weight & measuring plate.
Con - No electronics required.