A new modeling approach for microplastic drag and settling velocity

Area of Science:

• Environmental Science
• Fluid Dynamics
• Computational Science

Background:

• Microplastic (MP) transport and settling in aquatic environments are critical for environmental management.
• Existing models often lack accuracy and efficiency in predicting MP behavior.
• Novel computational approaches are needed to improve our understanding of MP dynamics.

Purpose of the Study:

• To develop a novel machine learning (ML) framework for accurate and interpretable drag and velocity models of microplastics (MPs).
• To enhance the prediction of MP settling behaviors across diverse MP types (1D, 2D, 3D, and mixed).
• To provide a more efficient and accurate alternative to traditional modeling methods.

Main Methods:

• Utilized machine learning techniques to create drag and velocity models for MPs.
• Validated the framework's predictive accuracy across various MP shapes and types.
• Performed sensitivity analysis to identify key parameters influencing MP settling.

Main Results:

• Achieved high predictive accuracy with R-squared values of 0.86-0.95 for drag and 0.92-0.95 for velocity models.
• Demonstrated significant error reductions compared to empirical approaches (59% RMSE for drag) and symbolic regression (18%-27%).
• Identified relative density difference and dimensionless diameter as key settling predictors, with shape parameters varying by MP type.

Conclusions:

• The ML framework provides accurate and efficient predictions of microplastic settling dynamics.
• This improved understanding can inform targeted mitigation strategies to reduce environmental MP impacts.
• The study highlights the importance of specific physical parameters in MP transport modeling.