The sponsor company, a global consumer goods enterprise known for high-quality products and customer service, has been on a digital transformation journey over the past few years to make its supply chain more responsive. Currently, the company manages its end-to-end supply chain planning using Mixed Integer Linear Programming-based (MILP) software. This process takes approximately two hours, limiting the company's ability to rapidly update production and distribution plans in response to sudden changes in supply or demand.
Our capstone project proposes the use of metaheuristic models as an alternative to their existing planning software, with the goal of reducing planning time while minimizing total relevant costs. Specifically, we identified the conditions in which the company currently operates and developed a model configurator to optimize the end-to-end supply chain planning. The application of the configurator, based on Genetic Algorithm and Particle Swarm Optimization metaheuristics, was demonstrated across three representative demand scenarios and proved successful in reducing planning time by approximately 85%.
Additionally, the proposed solution maintained the same quality as the current solution of the company, achieving a 2% and 13% cost reduction in production and distribution respectively, while accounting for penalties related to unmet inventory targets. This improvement is significant, as it enables the company to become more responsive to internal and external changes, improving its ability to adapt quickly to dynamic supply and demand conditions.