Comparison of evolutionary algorithms: A case study on the multi-objective carbon-aware mine planning

• Main Authors: NURUL ASYIKEEN BINTE AZHAR, GUNAWAN, Aldy, CHENG, Shih-Fen, LEONARDI, Erwin
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2024
• Subjects: Genetic algorithms; Pareto optimization; production planning; environmental economics; Operations Research, Systems Engineering and Industrial Engineering; Theory and Algorithms
• Online Access: https://ink.library.smu.edu.sg/sis_research/9495; https://ink.library.smu.edu.sg/context/sis_research/article/10495/viewcontent/Final_CASE2024_Comparison_of_EAs_av.pdf
• Institution: Singapore Management University

The NP-hard precedence-constrained production scheduling problem (PCPSP) for mine planning chooses the ordered removal of materials from the mine pit and the next processing steps based on resource, geological, and geometrical constraints. Traditionally, it prioritizes the net present value (NPV) of profits across the lifespan of the mine. Yet, the growing shift in environmental concerns also requires shifts to more carbon-aware practices. In this paper, we use the enhanced multi-objective version of the generic PCPSP formulation by adding the NPV of carbon costs as another objective. We then compare how the Non-dominated Sorting Genetic Algorithm II (NSGA-II) and the Pareto Envelope-based Selection Algorithm II (PESA-II) solve several real-world inspired datasets, after experimenting with the selection pressure parameter of PESA-II. The outcome reveals that PESA-II runs faster for 75% of the datasets and gives sets of solutions that are more distributed. Meanwhile, NSGA-II consistently produces non-dominated solutions even when the apportionment of a decision variable is varied. Moreover, we assess how the uncertainty of ore tonnage at the mine site modifies the Pareto front via sensitivity analysis. We show that deviations above 15% induce a larger gap from the original.