Cost patterns learning trough logistic and sorting models integration in Waste Management

New advanced information systems, digital technologies and mathematical models are required to achieve the targets of sustainability and circular economy paradigms. Circular economy includes products, but also infrastructure, equipment and services ordered by waste recycling centers where materials are collected and then sorted to be converted in secondary raw materials. This scenario imposes a new view of operations with the aim of zero waste, in order to obtain this result it is critical to adopt an holistic approach and to optimize every step of production and logistics processes. These targets are addressed in this work by the development of an integrated models framework that is fueled by data and supported by two OR models. In the presented framework, the output of each model becomes an input of the following one. A mixed integer programming model is used to optimize the sorting operations while a pick-up and delivery routing model supports the logistic operations. Once both logistic and sorting processes have been optimized, a machine learning model performs process cost analysis. Decision makers can use these outcomes to support and verify management decisions, such as updating contracts of loss making customers or increasing service level in profitable locations. Validation of the approach is done with data of a real test case with promising preliminary results.