Abstract View
Evaluation of a New Low-Cost Particle Sensor as an IoT Device for Indoor and Outdoor Particulate Matter Monitoring
ABI ROBERTS, Kathryn Van Valkinburgh, Christopher Post, John Pearce, Elena Mikhailova, Andrew Metcalf, Clemson University
Abstract Number: 318
Working Group: Instrumentation and Methods
Abstract
Low-cost particle sensors provide an opportunity to increase the spatial and temporal density of outdoor air quality measurements when integrated with an internet of things (IoT) system that is able to report sensor data in near real-time. Many low-cost particle sensors are currently available, but there are serious concerns about data accuracy and precision, sensor reliability, and suitability for outdoor deployment. We evaluate a newly available, low-cost particle sensor from Sensirion AG, which reports having high accuracy with other desirable capabilities, including low power consumption, for the ability for long-term sensor deployment, and measurement of particulate matter (PM) size speciation with number concentration, for less than $50/sensor.
This talk will discuss the testing of this new PM sensor in both a laboratory and ambient setting, efforts to develop an IoT system for both indoor and outdoor deployments, and the evaluation of this sensor to collect measurements of indoor and outdoor PM levels. In all tests, the new sensor was compared to measurements from a DustTrak-DRX Model 8533, and in some tests, to a condensation particle counter, scanning electrical mobility particle sizer, and Federal Reference and Equivalence Methods for PM measurements. Laboratory calibration used particles of known size (PM1, PM2.5, etc.) at various number concentrations (spanning 102 to 104cm-3). Outdoor testing demonstrates performance over a range of airborne PM levels in a relatively humid, rural environment. Time-series plots reveal general agreement in short-term PM variability; however, consistent differences in absolute values reveal that offsets may be required. Time-series analysis techniques are compared to identify similarities between tested sensor output, while spatial interpolation of averaged and event-specific data is used to build maps of PM levels.