Physical and Numerical Modeling of Large Marine Microplastics Transport and Deposition

Abstract

The prevalence of micro-sized plastic pollution has raised significant concerns for aquatic ecosystems and their inhabitants, including humans, due to its distinct physical and toxicological characteristics. Large microplastics (MPs), primarily synthetic polymers measuring less than 5 mm, pose a considerable threat. According to the literature, rivers alone transport millions of tons of plastic into the ocean annually, regardless of other water bodies. During this transport, MPs undergo physical processes such as advection and diffusion, influenced by atmospheric and hydrological conditions. These processes are driven by larger-scale factors, including windinduced drag, wave drift, and buoyancy, all stemming from currents, geostrophic circulation, and turbulence in extensive water bodies. The inherent properties of MPs—such as size, shape, and density—along with fluid flow characteristics like velocity, depth, and pressure, play crucial roles in determining their transport and deposition patterns. Consequently, this thesis aimed to use physical and numerical modeling approaches to investigate MPs' transport and deposition under varying hydraulic parameters and bathymetric changes applied to the channel bed. The experimental component of this research involved reviewing existing studies and conducting physical modeling of MPs transport and deposition. Various experimental test scenarios were carried out in a tilting flume at the National Research Council Canada Ocean, Coastal, and River Engineering Research Centre (NRC-OCRE) physical testing facility in Ottawa. Data on MPs transport and deposition were gathered using different physical setups in a straight open-channel flume to assess the effects of flow velocity, water levels, channel geometries, and particle shapes. This study contributed to validating numerical models for MPs transport and deposition. The second part of the research, focusing on numerical modeling, followed three steps: (1) reviewing the literature, (2) using a machine learning (ML) approach to predict MPs transport and deposition dynamics for large spherical and cylindrical MPs, and (3) simulating MPs transport and deposition using a coupled computational fluid dynamics-discrete element method (CFD-DEM). For the first step, a comprehensive review was conducted on the structure of four well-known Lagrangian particle tracking models (D-WAQ PART, Ichthyop, TrackMPD, and CaMPSim-3D) in simulating MPs' fate and transport. Each model was examined in terms of its ability to account for key physical transport processes—such as advection, diffusion, windage, beaching, and washing off—as well as transformation processes like biofouling and degradation, which influence MPs' behavior in the marine environment. Additionally, the review evaluated the effects of MPs' physical properties (mainly size, diameter, and shape) on their fate and trajectories. This analysis identified gaps in particle tracking models, including insufficient consideration of homo- and hetero-aggregation, agglomeration, photodegradation, chemical and biological degradation, and additional advection caused by wave-induced drift. In the second step of the numerical research, an ML-based study was conducted to predict the deposition patterns of spherical and cylindrical MPs using Extreme Learning Machines (ELM) for the first time, based on laboratory datasets. Various test scenarios were performed to collect data on multiple input parameters, including flow velocity, channel bed deepening, channel bed slope, water level, particle shape (spherical or cylindrical), near-bed flow velocity, and particle velocity at deposition. Eleven models, utilizing different dimensionless input combinations derived from Buckingham’s theorem, were evaluated. The best-performing model incorporated all four dimensionless variables, achieving high accuracy for spherical (R=0.94) and cylindrical (R=0.90) MPs. Sensitivity analysis revealed that the ratio of water depth to particle diameter had the greatest impact on deposition patterns. This study highlighted the potential of ML in predicting MPs behavior, aiding environmental monitoring and management. In the final step, numerical results were validated against laboratory experiments. A numerical modeling approach was employed to simulate the transport and deposition of spherical MPs under different flow and channel geometry scenarios. The unresolved CFD-DEM method was used for simulations, with results successfully validated through experimental measurements. Additionally, the model effectively captured the influence of hydraulic parameters on MPs deposition.