An In-Field Assessment of the P.ALP Device in Four Different Real Working Conditions: A Performance Evaluation in Particulate Matter Monitoring

Abstract

This study aimed to assess the performance, in terms of precision and accuracy, of a prototype (called "P.ALP"-Ph.D. Air Quality Low-cost Project) developed for monitoring PM_2.5 concentration levels. Four prototypes were co-located with reference instrumentation in four different microenvironments simulating real-world and working conditions, namely (i) office, (ii) home, (iii) outdoor, and (iv) occupational environments. The devices were evaluated for a total of 20 monitoring days (approximately 168 h) under a wide range of PM_2.5 concentrations. The performances of the prototypes (based on the light-scattering working principle) were tested through different statistical methods. After the data acquisition and data cleaning processes, a linear regression analysis was performed to assess the precision (by comparing all possible pairs of devices) and the accuracy (by comparing the prototypes against the reference instrumentation) of the P.ALP. Moreover, the United States Environmental Protection Agency (US EPA) criteria were applied to assess the possible usage of this instrumentation, and to evaluate the eventual error trends of the P.ALP in the data storage process, Bland-Altman plots were also adopted. The outcomes of this study underlined that the P.ALP performed differently depending on the microenvironment in which it was tested and, consequently, on the PM_2.5 concentrations. The device can monitor PM_2.5 variations with acceptable results, but the performance cannot be considered satisfactory at extremely low and remarkably high PM_2.5 concentrations. Thanks to modular components and open-source software, the tested device has the potential to be customized and adapted to better fit specific study design needs, but it must be implemented with ad hoc calibration factors depending on the application before being used in field.

Keywords: air pollution; air quality; exposure assessment; low-cost monitor; microenvironment; miniaturized monitors.

Figure 1

A box plot chart reporting the concentration values, expressed in µg/m³, of the reference instrument (Aerocet) and the four P.ALP prototypes considering the complete set of data acquired during the whole study. The central black mark is the median, the edges of the box are the 25th and the 75th percentiles, and the error bars show the extent of the most extreme data points that are not considered outliers.

Figure 2

The Bland–Altman plots of the data acquired from the four P.ALPs plotted against the reference instrument (Aerocet); both the X and Y axes are expressed in µg/m³. The data referring to the office microenvironment are highlighted in blue, the data referring to the home microenvironment are shown in green, the data referring to the outdoor microenvironment are shown in red, and the data referring to the occupational microenvironment are highlighted in purple. The dotted black line indicates the theoretical perfect agreement between the two compared instruments (P.ALP and Aerocet). The solid red line represents the mean error between the compared techniques, and the two solid black lines represent the upper and lower 95% confidence intervals, respectively.