Leveraging Deep Learning and Advanced Statistical Methods for Investigating the Fate and Transport of Microplastics in a Waste Water Treatment Plant
dc.contributor.author | Zhu, Frank | |
dc.date.accessioned | 2023-07-26T20:10:07Z | |
dc.date.available | 2024-07-26T04:50:06Z | |
dc.date.issued | 2023-07-26 | |
dc.date.submitted | 2023-07-26 | |
dc.description.abstract | Microplastics (MPs), defined as plastic particulates smaller than 5mm, pose significant environmental concerns due to their widespread distribution, adverse ecological impacts, and potential human health implications. The understanding of MP fate and transport in wastewater treatment plants (WWTPs) is crucial for effective management and mitigation strategies. This PhD thesis addresses key challenges associated with MP characterization in WWTPs and aims to enhance the understanding of MP fate and transport through improved MP identification, reduced data uncertainty, and the development of reliable count and mass balance models. The accurate identification of MPs in samples using spectroscopic methods presents a significant hurdle. The first study introduces PlasticNet, a deep learning convolutional neural network developed to overcome the challenges in identifying MPs in environmental samples through spectral classification. PlasticNet was trained, validated, and tested using complex spectra and proved to be capable of accurately classifying multiple types of common plastics. Its capability extends to recognizing spectra affected by factors such as the presence of additives and weathering, and variations in MP thickness. Compared with conventional library search routines, PlasticNet exhibits significant improvements, thus highlighting its potential as a standard automatic recognition tool for MPs in environmental samples analyzed by FPA FT-IR imaging. Uncertainty quantification in MP enumeration data is another challenge in MP research. The second study investigates and models the sources of error intrinsic to each step of MP analysis, including concentration heterogeneity, random errors during sample collection, and the loss of MPs during extraction. The impact of operational factors such as the number of replicates, MP size and shape variations, and differential recovery on data uncertainty is comprehensively examined. Bayesian uncertainty analysis, implemented through Markov Chain Monte Carlo (MCMC) methods, is utilized to quantify uncertainties and develop guidelines for reducing data uncertainty in MP enumeration. These findings provide valuable insights towards the development of standardized guidelines for analytical procedures quantifying MPs in wastewater and sludge samples. Furthermore, the third study addresses the establishment of mass balance models for MPs in WWTPs. Mass balance models offer valuable insights into fragmentation dynamics, weight-based ecological ramifications, and potential standardization across different studies and environmental contexts. However, accurate MP mass quantification remains challenging. The thesis explores innovative strategies for MP mass quantification and applies these techniques to develop reliable mass balance models in the primary treatment process at a selected WWTP. The third study also reveals the significant influence of MP size and sample type on recovery, hence the need for considering these variations in concentration estimations. The application of differential recovery resulted in significant improvement in the mass balance model, which underscores the importance of incorporating differential recovery into balance models studies. In sum, the third study shows the complementary nature of count and mass balance approaches in understanding the fate and transport of MPs in WWTPs. Overall, the enhanced MP identification, reduced data uncertainty, and reliable mass balance models developed in this thesis significantly contribute to a comprehensive understanding of MP fate and transport in wastewater treatment plants (WWTPs). These advancements offer multiple benefits, including accurate MP identification, precise MP enumeration, and deeper insights into MP removal processes. The optimization of processes and methodologies driven by this research aids in mitigating MP pollution, reducing ecological risks, and fostering standardization and comparative analyses across diverse studies and environments. Overall, this research enhances the robustness of MP research in WWTP contexts and supports the development of effective strategies for managing MP pollution and promoting sustainable practices. | en |
dc.identifier.uri | http://hdl.handle.net/10012/19634 | |
dc.language.iso | en | en |
dc.pending | false | |
dc.publisher | University of Waterloo | en |
dc.title | Leveraging Deep Learning and Advanced Statistical Methods for Investigating the Fate and Transport of Microplastics in a Waste Water Treatment Plant | en |
dc.type | Doctoral Thesis | en |
uws-etd.degree | Doctor of Philosophy | en |
uws-etd.degree.department | Civil and Environmental Engineering | en |
uws-etd.degree.discipline | Civil Engineering (Water) | en |
uws-etd.degree.grantor | University of Waterloo | en |
uws-etd.embargo.terms | 1 year | en |
uws.contributor.advisor | Parker, Wayne | |
uws.contributor.affiliation1 | Faculty of Engineering | en |
uws.peerReviewStatus | Unreviewed | en |
uws.published.city | Waterloo | en |
uws.published.country | Canada | en |
uws.published.province | Ontario | en |
uws.scholarLevel | Graduate | en |
uws.typeOfResource | Text | en |