Enhancing operational efficiency in a voluntary recycling project through data-driven waste collection optimization

Recycling in developing countries is often driven by voluntary initiatives, typically led by the private sector. While commendable, these efforts face significant challenges, particularly in ensuring operational efficiency due to their non-profit nature. The logistics process involves collecting recyclable plastics from collection points and delivering them to a recycling facility. This aspect is crucial, as it represents the largest cost component, making optimization essential. Traditional approaches, such as the Traveling Salesman Problem and its meta-heuristic variants, are time-consuming for practical applications. To address these challenges, we propose a three-step data-driven approach designed to optimize waste collection within the constraints of non-profit projects. The first step uses K-means clustering to group collection points geographically, reducing the complexity of subsequent optimization stages. The optimization models in the second and third steps aim to maximize the amount of recyclable plastic per trip and determine the most efficient collection route. Real-time data on waste volume at each point and live traffic conditions, retrieved via the Intelligent Traffic Information Center, are integrated into these models, making it possible to achieve a high level of practicality and accuracy. The efficacy of this approach is demonstrated through a case study of the Won Project, involving 152 plastic waste collection points. The results show a significant daily increase in the average amount of plastic collected and reduction in the average distance traveled. The proposed method can produce prompt, reliable solutions for daily operations using open-source software, successfully addressing the challenges of voluntary waste collection projects.