Date of Award
Spring 5-15-2019
Degree Name
Doctor of Philosophy (PhD)
Degree Type
Dissertation
Abstract
Particulate matter (PM) has been monitored routinely due to its negative effects on human health and atmospheric visibility. Standard gravimetric measurements and current commercial instruments for field measurements are still expensive and laborious. The high cost of conventional instruments typically limits the number of monitoring sites, which in turn undermines the accuracy of real-time mapping of sources and hotspots of air pollutants with insufficient spatial resolution. The new trends of PM concentration measurement are personalized portable devices for individual customers and networking of large quantity sensors to meet the demand of Big Data. Therefore, low-cost PM sensors have been studied extensively due to their price advantage and compact size. These sensors have been considered as a good supplement of current monitoring sites for high spatial-temporal PM mapping. However, a large concern is the accuracy of these low-cost PM sensors.
Multiple types of low-cost PM sensors and monitors were calibrated against reference instruments. All these units demonstrated high linearity against reference instruments with high R2 values for different types of aerosols over a wide range of concentration levels. The question of whether low-cost PM monitors can be considered as a substituent of conventional instruments was discussed, together with how to qualitatively describe the improvement of data quality due to calibrations. A limitation of these sensors and monitors is that their outputs depended highly on particle composition and size, resulting in as high as 10 times difference in the sensor outputs.
Optical characterization of low-cost PM sensors (ensemble measurement) was conducted by combining experimental results with Mie scattering theory. The reasons for their dependence on the PM composition and size distribution were studied. To improve accuracy in estimation of mass concentration, an expression for K as a function of the geometric mean diameter, geometric standard deviation, and refractive index is proposed. To get rid of the influence of the refractive index, we propose a new design of a multi-wavelength sensor with a robust data inversion routine to estimate the PM size distribution and refractive index simultaneously.
The utility of the networked system with improved sensitivity was demonstrated by deploying it in a woodworking shop. Data collected by the networked system was utilized to construct spatiotemporal PM concentration distributions using an ordinary Kriging method and an Artificial Neural Network model to elucidate particle generation and ventilation processes. Furthermore, for the outdoor environment, data reported by low-cost sensors were compared against satellite data. The remote sensing data could provide a daily calibration of these low-cost sensors. On the other hand, low-cost PM sensors could provide better accuracy to demonstrate the microenvironment.
Language
English (en)
Chair
Pratim Biswas
Committee Members
Rajan Chakrabarty, Chenyang Lu, Pramod Kulkarni, Brent Williams,
Comments
Permanent URL: https://doi.org/7936/v9wd-dr27