Denman Island, British Columbia
Well Depth Sensor |
The objective of this project was to provide a means of taking frequent. accurate depth measurements of a water well to determine usage patterns and recharge rates. The design is pneumatic. The princple is to blow bubbles from a pipe at the bottom of the well. The air pressure required to fill the pipe with air is equal to the water pressure at the bottom of the well, which directly proportional to the depth of the well.
The sensor consists of hardware, electronic and software components. The schematic diagram above shows the hardware and electronic components.
The hardware consists of a small 12-volt air pump (designed for inflating car tires), a pressure gauge to provide manual readings for calibration, an electric pressure sensor (designed for automotive oil pressure gauges) and a pipe leading to the bottom of the well. The pipe down the well is 1/8" polyethylene tubing, with a brass weight attached to the end to keep it from floating.
The electronic components consist of two OneWire® modules: an analog-digital converter and a relay. The analog-digital converter senses the voltage from the pressure sensor and converts it to a numeric value. The relay is used to activate the pump. Both electronic devices are connected to a computer using the OneWire® network. Custom software on the computer controls the relay and pump and reads the pressure data.
The software component of the system uses an application that I wrote originally for making weather observations. The weather instruments use the same OneWire® network as the well depth sensor, and the data recording function is identical, so I just added the sensor as though it were another weather instrument.
General view of well depth sensor components, showing air pump (left rear), air manifold and pressure sensor (left foreground), power supply (right lower) and the box containing the electronic components (right upper).
The tubing leading to the well leaves the picture to the left. For its run across the ground to the well head, it is enclosed in 3/4" poly pipe for physical protection.
This view shows the inside of the electronics box. The relay module is at the top. A terminal strip is at the right. The module at the bottom was purchased as a solar radiation sensor. Hobby Boards sells several variants of the DS2438 board, reading temperature, humidity and solar radiation. The solar version is the cheapest. The well depth sensor is connected to the unused ADC input that is intended for the humidity sensor.
The pipe to the bottom of the well must be kept pressurized in order to obtain accurate readings. However, the pump need not run continuously. It only needs to operate periodically to make up for leakage and usage losses. With good plumbing technique, leakage losses are negligible.
Air is lost from the pipe when water is drawn from the well. As the well level drops, the water pressure at the base of the well drops proportionately. The air pressure in the pipe is then too high to balance the water pressure, and air leaks out to restore the balance. When the water level in the well recovers, the increased water pressure compresses the air in the pipe and water enters the bottom of the pipe. Because there is now water in the pipe, less air pressure is required to achieve equillibrium, and the pressure gauge will read low for the actual water depth.
To replace air lost by water usage as described above, the pump must be operated from time to time. The current schedule calls for it to be run three times per day. To operate the pump, the software activates the relay module, which applies power to the air pump. Ten or twenty seconds of operation is usually more than sufficient to refill the pipe.
Ideally, one would simply turn on the pump and read the pressure. However, frictional forces combine to complicate the measurement. When the pump is first turned on, there is usually water in the pipe. As the water is forced out, friction between the moving water and the pipe requires a higher air pressure than the water depth alone would suggest.
Once all the water is forced out and bubbles are forming at the bottom of the pipe, the friction from the water is eliminated, and the pressure drops. However, air is still moving, and it contributes its own friction. The pressure is therefore still higher than would be required for a static equillibrium. It is only when the pump is shut off that a static equillibrium forms and one is able to take a meaningful pressure reading that is proportional to depth.
When taking readings manually, it is important to watch for the initial pressure drop as the last water is forced out of the pipe, and then to shut off the pump before taking a pressure reading. When the software is operating the pump, it simply runs it for a predetermined length of time.
The effects of friction described above could be reduced by using a larger diameter pipe. I used 1/8" tubing because it was readily available. Tubing that is 1/4" or larger would be preferable. On the other hand, a larger pipe will require more air, necessitating either a bigger pump or operating it for longer cycles.
The automotive oil pressure sensor provides a variable resistance. Because it is designed for a meter to read a high current when the pressure is high and a low current when it is low, the resistance is (approximately) inversely proportional to pressure: a low resistance corresponds to a high pressure and vice versa. In order to provide the analog-digital converter with a voltage that varies (approximately) directly with pressure, and to keep the range of voltages within the input range of the converter, ADC measures the voltage across a fixed resister in series with the sensor's variable resistor.
Calibration was performed by starting with an estimated pressure-resistance response curve for the pressure sensor, then adjusting it to obtain the correct known depth. The full response curve will be calibrated by pressurizing the pipe to different pressures and reading the actual voltage at the ADC input.
Once the air pressure in the pipe is known, converting it to water depth is a simple calculation: the pressure in pounds per square inch, multiplied by 2.31, gives the depth of water in feet.
The software takes a pressure reading every five minutes, around the clock, at the same time as it takes weather readings. These are recorded in a permanent database.
I have not yet operated the automatic well depth sensor for a full season. However, I have operated its manual precursor sporadically for three summers. The manual observations show that the depth in our well drops dramatically in the spring and does not recover until the fall. However, the manual observations provide data that is too coarse for drawing interesting conclusions from.
The results of the observations are shown graphically below. Data from 2007 to February 2010 are from manual observations. Data from March 2010 onwards are automatic observations. Calibration is ongoing and will continue until the well reaches its lowest level in late summer.
Things I hope to learn from running the sensor automatically, with its finer temporal resolution, are:
Home
Copyright © 2010 Keith Walker
Last modified: 30-Apr-2010