Studying Oil Spill Transport by Exploring Oil-Mineral-Aggregates Characteristics and Tidal Dispersion Properties

Abstract

Oil spill has been widely recognized as a major marine environmental issue, that could cause a profound and long-term impact on the environment, ecology, and socioeconomics. In particular, when the oil spill happens in coastal areas, the spilled oil can interact with the suspended sediments to form oil-mineral-aggregates (OMA), and its settling and trajectory are crucial for oil spill modelling. We started by identifying the most influential variables during OMA formation through a statistical method (Screening Design). Time was the most important factor for OMA median diameter (D50), followed by temperature and oil/clay ratio. The influence of time was highly dependent on the temperature and oil/clay ratio applied. We also assembled a high-speed camera and magnifying lens to measure the settling velocity of OMA precisely and explored how process variables influence the settling velocity. Principal component analysis revealed that increasing clay concentration in the water environment significantly promoted OMA settling velocity. However, the presence of dispersant and water density increment led to settling velocity reduction. The traditional empirical numerical modelling for OMA D50 and settling velocity prediction was based on collision-theory and Stokes’ law, respectively. The empirical predictions were conducted in this study, and the R2 of 0.62 and 0.39 were achieved. Machine learning algorithms were, for the first time, employed to predict OMA D50 and settling velocity. Adaboost algorithm and Gradient Boost Regression algorithm was identified to be the most satisfying one for predicting D50 (R2 of 0.74) and settling velocity (R2 of 0.61) respectively, which had correspondingly higher R2 than traditional empirical numerical modellings. As for the OMA transport, the particle tracking model was applied for simulation. The sensitivity of OMA transport to prediction accuracy was evaluated, and a high sensitivity was observed. In addition to oil spill modelling, the feasibility of using the Finite-Time Lyapunov Exponent (FTLE) to analyze the tidal dispersion properties and oil spill trajectory was assessed. The FTLE results were compared with a real oil spill in Burrard Inlet in 2015. The results indicated that FTLE could reasonably explain the spilled oil’s trajectories, suggesting FTLE could be a valuable addition to practical oil spill transport.