The Implementation of Machine Learning Tools to Select Potential Depot Locations for the Supply Chain of Biomass Co-Firing
Coal is the second-largest source for electricity generation in the United States. However, the burning of coal produces Greenhouse Gas (GHG) emissions, such as carbon dioxide and sulfur dioxide. One alternative to coal combustion is biomass co-firing, where biomass is used as a partial substitute fuel, reducing emissions while maintaining current plant infrastructure. To establish biomass co-firing as a viable option, the optimization of its supply chain is essential. While most of the research conducted has focused on optimization models, the purpose of this thesis is to incorporate machine learning (ML) algorithms into an existing Mixed-Integer Linear Programming (MILP) model to decrease the total cost of the biomass supply chain (BSC) by selecting potential locations to serve as biomass storage depots. In addition, since optimization applications often involve NP-hard models, the purpose of incorporating ML is also to decrease the computational burden while obtaining high-quality solutions. In this thesis, we consider the level of moisture and level of ash in the biomass from each parcel location, the average expected biomass yield, and the distance from each parcel to the closest energy plant. The training labels - whether a potential depot location is beneficial or not - are obtained through the existing MILP model. Multiple ML algorithms are applied to a case study in the northeast area of the United States: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), and Multi-Layer Perceptron (MLP) Neural Network. It is found that the MLP outperforms the other algorithms in terms of selecting potential depots that decrease the total cost of the BSC and the computational burden of the MILP. The LR and the DT also perform well in terms of decreasing total cost.