Uma abordagem para diminuição de custos na logística reversa com o auxílio de gêmeos digitais e otimização multiobjetivo

Abstract

The DAI (Academic Doctorate for Innovation) is an initiative by CNPq with the objective of enabling Graduate Programs to foster interaction projects with companies through theses, dissertations, and final course projects. The partner company in this work is a multinational that produces electronics (smartphones, printers, laptops, monitors) for major brands such as HP and Motorola. The company is deeply concerned with the environment and the increasingly scarce natural resources. This concern has been growing recently among various companies, spurred by new laws that require them to be increasingly attentive to the problem of resource scarcity. Waste management is one of the most current ways to support the reuse of these finite natural resources, and one of the main techniques used for this management is Reverse Logistics (RL). This primarily involves the return of products at the end of their life cycle, from the user back to the company, followed by the separation and selection of reusable materials. One of the challenges about routing in Reverse Logistics concerns the vehicle used for collections, and although these routing issues are widely discussed in the literature, a new approach is emerging. This approach allows for studies that were previously difficult to conduct, given the challenge of analyzing RL outside a real-world context, i.e., without dispatching vehicles to perform the routes and analyze the results. To address this challenge, this work seeks to use Digital Twins (DTs), a virtual representation of a real-world model, as a way to virtually simulate this system. In addition to DTs, Clusters are used in conjunction with multi-objective optimization tools, such as Genetic Algorithms (GAs), to group nearby collections and optimize them. This approach made it possible to minimize economic costs and CO2 emissions relative to the historical data provided by the partner company, while simultaneously maximizing the number of collected products. To test the developed tool, DTs were used in two case studies with real data. As a result, it was possible to determine the best collections to be performed on a daily basis, enabling the company to better calculate costs and reduce pollutant gas emissions—two increasingly important goals for the company. A Framework was created that can group collections, perform calculations, and assist decision-makers in identifying the best collections to undertake. This differs from the company’s current model, where the decision-making responsibility lies with the carrier responsible for collecting electronic waste. The case studies showed that potential savings could reach up to 80% compared to the current amount paid by the company, demonstrating that the developed Framework has significant potential for implementation and can be a powerful decision-support tool.