10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Design and Optimisation of Low-cost Air Quality Sensor Package (KOALA)
XIAOTING LIU, Matthew Dunbabin, Bryce Christensen, Rohan Jayaratne, Phong Thai, Lidia Morawska, Queensland University of Technology
Abstract Number: 1135 Working Group: Low-Cost and Portable Sensors
Abstract There is currently a large selection of low-cost air quality monitors available in the market for consideration, although it was determined to be unrealistic to test them all individually for the scope of this study given the diversity in the manufacturing quality, functionality and consistency of monitoring results. The aim of the study was to develop a modular sensor package which can reliably measure the air quality and individual pollutants within acceptable tolerances with consideration for the low-cost production costs of the devices. The processes for the design and optimisation of low cost air quality sensor packages will be demonstrated in this study.
This low-cost air quality sensor package (named KOALA), developed by QUT, was equipped with Plantower PM sensor PMS1003 and Alphasense CO-B4 sensor as well as temperature and humidity sensors. The Plantower PM sensor PMS1003 and Alphasense CO-B4 sensor were selected because of their consistent and reliable performances in our previous lab and field tests. The CO sensor was used to identify the combustion sources of air pollutants which was shown to be the main source for air pollutants in our study areas. With the combination of PM and CO sensor, the KOALA air quality sensor package was able to identify the difference between sources of air pollutants in the targeted area.
The KOALA air quality sensor package has been designed to be robust enough for remote field deployment and longevity, while still performing the required monitoring functions. The external components of the device include a custom designed hardened plastic casing, solar panel and mount, PM sensor intake, PM sensor fan exhaust, CO sensor rain guard, and the mounting plate for the device. Considerations were made for water resistance in the design of any external components, with all devices having silicone applied around external components before deployment.
The internal structure of the KOALA includes a programmable Arduino board, 3G Module, Antenna, microSD card, power switch, battery, CO sensor and Plantower PM Sensor. With the function of solar panel and rechargeable battery, the KOALA air quality sensor package was able to operate in the outdoor environments without the need for a dedicated mains power source. The software was using the Arduino software framework.
The improvement and optimisation of the sensor package has been completed through a staged internal lab testing and then external deployment campaign. The following problems were encountered and subsequently resolved during the continuous testing and operation of the air sensor package. 1)Achieving low-power management for relatively high-power sensors (e.g. Plantower) which could cause unexpected restarting of the sensor node; 2) Data delivery delays due to low-signal 3G connectivity issues; 3) power management issues causing inability to send data, and the automated switching of the unit to caretaker mode to protect the battery pending charge; 4) Incorrect clock settings due to power loss on the internal board battery. To solve these problems, optimisation of both hardware and software of the sensor package was completed and involved: 1) custom software development to overcome the hardware power management issues: 2)to prevent the node batteries to drain completely before deployment which may cause trouble to get them charged again (Lithium batteries), major software change has been done with regard to power management;3) For locations with limited sunlight for the solar system during the day, software was developed; 4) increase the solar panel size; 5) developed a deployment protocol for ensuring solar energy harvesting and suitable 3G signal quality for data transfer.
Data communication from the sensors is performed using a variety of low bandwidth technologies. The data is stored on the sensors SD card, which retains records in a queue to upload at a nominal interval, currently each 30 minutes. The data adheres to the lightweight MQTT messaging format, a condensed string format to reduce overhead during transmission. Using 3G, the data is sent to an IoT endpoint on Amazon Web Services (AWS). It is then parsed through a variety of AWS services to reside in a PostGreSQL database, which is accessed via a public web interface.
The website to date has stored over 400,000 records successfully during this testing phase, which around 2% purged due to poor date and time data. The website offers dynamic graphing of the incoming sensor readings including battery health, PNC readings, PM readings, temperature, humidity and CO information. The raw data is available in table format, or as a downloadable JSON format. Additional features will include extendable querying of the dataset and CSV format downloads for larger datasets, and a geo-spatial visualisation in preparation for the expansion of the monitoring network and introduction of mobile sensors.