A Low-Cost Water Depth and Electrical Conductivity Sensor for Detecting Inputs into Urban Stormwater Networks

Abstract

High-resolution data collection of the urban stormwater network is crucial for future asset management and illicit discharge detection, but often too expensive as sensors and ongoing frequent maintenance works are not affordable. We developed an integrated water depth, electrical conductivity (EC), and temperature sensor that is inexpensive (USD 25), low power, and easily implemented in urban drainage networks. Our low-cost sensor reliably measures the rate-of-change of water level without any re-calibration by comparing with industry-standard instruments such as HACH and HORIBA's probes. To overcome the observed drift of level sensors, we developed an automated re-calibration approach, which significantly improved its accuracy. For applications like monitoring stormwater drains, such an approach will make higher-resolution sensing feasible from the budget control considerations, since the regular sensor re-calibration will no longer be required. For other applications like monitoring wetlands or wastewater networks, a manual re-calibration every two weeks is required to limit the sensor's inaccuracies to ±10 mm. Apart from only being used as a calibrator for the level sensor, the conductivity sensor in this study adequately monitored EC between 0 and 10 mS/cm with a 17% relative uncertainty, which is sufficient for stormwater monitoring, especially for real-time detection of poor stormwater quality inputs. Overall, our proposed sensor can be rapidly and densely deployed in the urban drainage network for revolutionised high-density monitoring that cannot be achieved before with high-end loggers and sensors.

Keywords: distributed sensing; electric conductivity; illegal discharge detection; low cost; low power; real-time environmental monitoring; water IoT; water level measurement.

Figure A1

(a) The base on the designed 3D sensor covers for easy and quick installation in the field; (b) how the sensor board, EC pins, and the sensor holder are placed into the sensor case; (c) the sensor case lid to push the sensor board down and forms as physical protection of the sensor; and (d) injecting the potting compound gel for waterproofing purpose.

Figure A2

Depth sensor calibration coefficients throughout the monitoring period; red dot—Automatic sensor re-calibration when the HACH probe at its lowest detection limit 9 mm; black dashed line—The linear interpolated calibration coefficient between two automatically generated calibration point; the calibration coefficient is assumed to be constant for the period before the sensor dead.

Figure A3

(a) The correlation between low-cost EC sensor readings and HORIBA measurement at Site B; (b) the correlation comparison at Site C.

Figure 1

(a) The circuit design of MS5803-01BA sensor with I²C communication: +3.3 V for the power supply, GND for ground connection, and SDA and SCL for the I²C bus; (b) a simple voltage divider circuit to measure the conductivity of water-based on 5 V power supply: R1 is the selected 100 Ω resistor in this study, TP1 and TP2 are the two screw terminals on the PCB circuit to connect with the stainless steel rod, and V_OUT connects to one of the analog pins on the MCU to measure the voltage drop across R1; (c) the circuit diagram of a digital temperature sensor DS18B20: DATA_TEMP wire connects to one of the digital pins on the MCU to measure the environment temperature; and (d) sensor housing design with a vertical wall at the back of the sensor for easier installation.

Figure 2

(a) Lab experiment setup in a water column for testing the proposed water depth sensor: water level is controlled through a tap and the level was changed between 0 and 850 mm; (b) Field installation layout to compare the performance between the low-cost and HACH submerged probe; the HACH sensor is 93 mm above the invert of the pipe and the low-cost sensor is 17 mm above the invert.

Figure 3

The processes of manual and automated sensor re-calibration.

Figure 4

Linear regression curves of water depth between developed low-cost sensors and manual measurements: (a) IN-WATER 1, (b) IN-WATER 2, and (c) IN-WATER 3; (d) SWITCH 1, (e) SWITCH 2, (f) SWITCH 3 (one offset calibration before installation); (g–i) SWITCH 1–3 sensors after continuous manual re-calibration.

Figure 5

Time series plots of the absolute level difference between low-cost sensor measurements and the interpolated true measurements of water depth for (a) IN-WATER 1–3 and (b) SWITCH 1–3 sensors; For reference, the true measurement points and interpolated data are shown at the top.

Figure 6

Comparing the depth of water above the drain’s invert from the low-cost depth sensor and the HACH submerged probe: (a) raw low-cost sensor readings without conduction continuous automatic re-calibration; (b) low-cost sensor conducting re-calibration when HACH reached its lowest detection limit—9 mm.

Figure 7

(a) averaged daily difference between low-cost water depth sensor and HACH submerged probe for both low-cost Sensor 1 and Sensor 2—The positive difference represents the low-cost water depth is higher than that of HACH probe; (b) the measured water depth of low-cost Sensor 2 and HACH submerged probe during the red-shaded period.

Figure 8

The percentage error between the low-cost and HACH sensors at different water levels: (a) low-cost sensor measurements with one offset calibration only; and (b) low-cost data with automated sensor re-calibration.

Figure 9

Comparing the rate of depth change (trend) between low-cost and HACH probes: (a) with only one offset calibration before installation; and (b) by using the low-cost sensor data after automated re-calibration.

Figure 10

(a) correlation between the low-cost conductivity sensor and HANNA’s measurement results in standard EC solutions ranging from 0 to10 mS/cm; and (b) correlation in solutions between 0 and 60 mS/cm.

Figure 11

Low-cost sensor field readings compared with HORIBA measurement at Site A and D.