Novel measurement techniques for monitoring particulate matter: development, application and evaluation of a low-cost IoT sensor network

Cowell, Nicole Hannah ORCID: 0000-0002-6270-0913 (2024). Novel measurement techniques for monitoring particulate matter: development, application and evaluation of a low-cost IoT sensor network. University of Birmingham. Ph.D.

Preview

Cowell2024PhD.pdf
Text - Accepted Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.
Download (6MB) | Preview

Abstract

Presented as a collection of articles, this thesis assesses the role of low-cost and Internet-of-Things (IoT) technology in monitoring of ambient particulate matter. Adapting an off-the-shelf low-cost optical particle sensor, an IoT enabled particulate matter sampler was developed, built, and tested in a range of environments. The sampler was designed with field deployment practicality in mind. Completely self-contained, AltasensePM was designed to use battery power and the latest IoT communications technology to enable easy installation onto existing street furniture. Although subject to limitations from covid-19 pandemic, this thesis is able to present exploratory analysis from the use-cases of low-cost sensors for particulate matter monitoring.
Field testing enabled the evaluation of AltasensePM device and the development of a site specific calibration method for tackling the effect of humidity on detected particle concentrations (Cowell et al., 2022). Sensors performed well in co-location with a reference instrument with average correlation coefficients (r2) between raw device data and reference instrumentation are 0.718, 0.703 and 0.543 for PM1, PM2.5 and PM10 respectively, and calibration improving performance. During this testing process, fluctuating performance over time was detected from the sensors. This insight into raw sensors data (which is not always shared when using commercial low-cost sensors), suggest that sensors are best suited to longer term averaging and stable environments. The raw data also suggested that the sensor is not independently monitoring PM10. AltasensePM devices were hosted on the Birmingham Urban Observatory, an open-access online platform for environmental data across the city of Birmingham, UK. The calibration method was automated as an online process, allowing the following work to utilise real-time air quality measurements from the AltasensePM alongside commercial air quality and meteorology sensors.
When deployed in a network, the low-cost PM samplers provided a novel opportunity to study particulate matter concentrations at previously unattainable spatial scales, in both indoor and ambient monitoring settings.
The indoor deployment took place in a typical UK suburban home, during the Covid-19 pandemic, and exposed how activity such as cooking elevated concentrations of PM1 and PM2.5 in rooms beyond the kitchen (Cowell et al., 2023a). The role of ventilation in reducing extreme peaks in concentrations was highlighted and when extrapolating data from this 7-week study to annual exposure levels, both 2021 and 2004 WHO Air Quality Guidelines were exceeded. The limitations of the sensor highlighted in chapter 2 were also reflected in the indoor performance of the sensor. Novel opportunities for automating indoor monitoring using IoT and Smart home devices could create unprecedented future insight into occupant activity and the impacts on PM, although there are security and privacy risks that will need to be resolved first.
For ambient monitoring, AltasensePM was integrated into an opportunistic nested network of sensors, bringing together commercial low-cost sensors deployed by the University and local authority and AltasensePM onto the Urban Observatory open-access platform. The live-stream of open access PM2.5 data from this network gave unprecedented, localised insight into PM concentrations across the city. The network performed very well in longer term regional averaging compared to the regulatory network at estimating city wide annual PM2.5 concentrations- with sensor network average for the period within 0.2µgm-3 of the regulatory average. It is unclear whether sensors are sensitive enough to detect roadside sources, or if other regional sources are more predominant in roadside locations. Sensor performance and local variability highlighted the importance of network design to ensure representative sampling of the city, and the opportunities that collaborations between various low-cost sensing users could provide.
Finally, following from this experience with sensor networks, an evaluation of the future role of IoT, low-cost sensors and live stream data in supporting the management of air quality is presented. Live-stream data from sensors such as the AltasensePM has been shown to have the potential to support digital twin technology, where real time decision making is enhanced by both model and live data streams simultaneously. However, the future of PM in countries such as the UK where there is a modal shift toward electric engine vehicles means that regional sources rather than local traffic sources are likely to become more important in air quality management. Any future digital twins will need to onboard national and international data sets to support such management. Here the best practice in managing data streams for air quality interventions via digital twins is discussed.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):