Isabel Gomes

In this work a two-stage scenario-based modeling approach is proposed in order to simultaneously deal with the design and planning decisions in supply chain networks, where both forward and reverse flows are considered (closed-loop supply chains) subject to uncertain conditions. A mixed integer linear programming (MILP) approach is developed with the underlying objective of profit maximization. Planning takes into account raw material acquisition and processing, storage and distribution of several products flowing through the network. Uncertainty is associated to the quantity and quality of the flow of products of the reverse network, which are directly affected by customers and sorting centers, respectively. The approach considers, by means of scenarios, the simultaneous integration of two important uncertainty sources, thus presenting an important modeling advantage, i.e. that of providing a better understanding of the CLSCs. The applicability of the formulation is tested with a real sized example of a Portuguese glass company.

The Portuguese directive 230/2004 established December 2006 as the dead-line for the legal targets of EEE recovery to be achieved. The EEE producers working in Portugal set up an organization whose mission is to design and manage this recovery network. A MILP model was previously developed to optimize the network design where, by considering EEE sources and recycling facilities locations as known, sorting centre locations were determined together with the planning of collection and sorting activities. In the present paper, the network planned by the Portuguese EEE organization is compared with the results provided by this early model. In addition, and taking into account new organizational objectives, where collecting and sorting entities should independently perform the associated activities, the developed model is further extended, in order to account for a new network level related to the collection and sorting activities, and the previous network is retrofitted.

Packaging waste collection systems are responsible to collect, within a geographic area, three types of packaging materials (paper, glass and plastic/metal) that are disposed by the final consumer into special bins. Those systems are often characterized by having a network with multiple depots that act as transfer and sorting stations, and where the vehicle fleet is based. However, each depot is often managed independently and not as a part of a unique system. In this work, four current tactical/operational practices that contribute to the independent management of each depot are analysed. The change of such practices is investigated and their impact assessed on the total collection cost. A solution methodology based on mathematical formulations is developed to plan service areas, vehicle routes and vehicle schedules taking into account new alternative solutions in managing the system as a whole. Such methodology is applied to a real case study of a company responsible for the collection of the packaging waste in 7 municipalities in mainland Portugal. New service areas, collection routes and vehicle schedules are defined and significant savings are obtained in terms of the total distance travelled as well as in terms of the number of required vehicles, resulting in a decreasing of the total system cost.