Recent Projects
-
This project seeks to understand the relationships between historical financial metrics found on 10-K annual statements and future performance. To do so, statements from process industry companies (simplified to energy, manufacturing and pharmaceutical sectors) were compiled into a dataset of over 160 categories. PCA and clustering analyses were completed for initial data exploration revealing 17 components are required to minimally describe the relationships. A PLS model was then created based on the training data compiled. Finally, a neural network was set up using PyTorch using the same training data. The results were compared between PLS and the neural network, showing predictive ability from historical EBITDA, COGS, some revenue metrics, and some cashflow metrics. It should be noted that the nonlinear relationships between some of these metrics make the neural network considerably more accurate than the PLS model.
-
The global market for nickel sulphate hexahydrate (NSH), an important pre-cursor component needed to create cathodes used in lithium-ion batteries, is experiencing significant growth due to the increasing demand for electric vehicles. Traditional NSH production methods rely on resource-intensive processes that significantly impact the environment. This scoping study, a collaboration between McMaster Engineering and Hatch Ltd., proposes a novel, sustainable design for a 50,000 tonnes/year battery-grade NSH refinery based in Sudbury, Ontario. The design prioritizes minimizing environmental impact while maintaining economic viability. Utilizing Mixed Hydroxide Precipitate (MHP) feedstock, the proposed refinery employs acid leaching, solid-liquid separation, neutralization, solvent extraction, and crystallization to achieve >99.4% NSH purity, along with over 4,000 tonnes/year of 99.9% pure cobalt sulphate hexahydrate. A comprehensive process simulation, including mass and energy balances, a full mechanical equipment list, process flow diagram, and complete capital and operating costing analysis supports the design.