A modified one-to-one algorithm for optimizing sustainable lot sizing multi-item models

Abstract

Multi-item inventory management in modern production environments faces major challenges related to stochastic demand, transportation costs, and carbon emissions. This study aims to develop a sustainable lot sizing model that integrates economic and environmental aspects, and proposes the one-to-one based optimization (OOBO) algorithm as a problem-solving approach. The methodology used includes non-linear programming (NLP) formulation that considers stochastic demand, ordering and storage costs, carbon emissions, energy consumption, and vehicle capacity constraints. The model is then optimized using OOBO and compared with the Aquila, particle swarm optimization (PSO), and genetic algorithm (GA) algorithms in three case scale scenarios (6, 30, and 50 items). The experimental results show that OOBO consistently outperforms the comparison algorithms, with cost savings of up to 40.9% in the 50-item case. OOBO also demonstrated high exploration resilience without premature convergence and competitive computational time efficiency. These findings confirm that OOBO is effective in simultaneously optimizing total costs and carbon emissions, making it an adaptive solution for sustainable supply chain management. The theoretical implications include the expansion of OOBO's application to multidimensional stochastic systems, while in practical terms, this model supports decision-makers in formulating environmentally friendly and efficient inventory policies.