This research explores estimating off-shore oily waste, considering waste-waste compatibility due to the heterogeneous nature of oily waste. Firstly, hyperparameters for Artificial Neural Networks (ANN), Support Vector Regression (SVR), and Improved Random Forest (IRF) models are optimized to develop a comprehensive oily waste estimation model incorporating liquid, solid, and total waste types. The results show that IRF is the most accurate model, with the lowest error indices and a higher correlation coefficient compared to ANN and SVR. This study then takes a step further to propose a waste allocation framework, which is tested using information on the Bella Bella oil spill incident in British Columbia. Incorporating treatment and receiving facilities' details, such as their location and capacity, the framework distinguishes all possible waste pathways for handling the waste from source to landfill. Genetic Algorithm (GA) is introduced to optimize waste transfer processes and successfully minimize transportation costs. The results show that the model can find the most optimized path to reduce transportation costs. The model's high customization, adaptability, and capacity to consider multiple nodes make it suitable for complex waste transfer networks, demonstrating its practicality in emergency situations. Efficiently allocating resources and ensuring cost-effective waste transportation while considering facility capacities and waste compatibility, the study holds practical implications for waste management practitioners, environmental authorities, and response teams.
