Returnable transport items (RTI) are used for the handling and transportation of products in the supply chain. Examples of RTIs include plastic polyboxes, stillages or pallets. We consider a network where RTIs are used by multiple suppliers to deliver parts packed in RTIs to multiple customers. We address the short-term planning of empty-RTI flows (i.e., reverse flows) which consists of optimizing the transportation routes used to return empty RTIs from customers to suppliers. A transportation route consists of one or several trucks traveling from a customer to a supplier at a given frequency. The RTI short-term planning problem is critical because it impacts the continuity of loaded-RTI flows and affects the transportation and shortage costs of empty RTIs incurred at the very-short-term. We study a heterogeneous fleet of automotive parts RTIs, under two configurations: pool RTIs, which are standard RTIs shared between suppliers, and dedicated RTIs that are specific to each supplier. To solve the short-term planning problem, we develop a two-step approach using mixed-integer linear programming (MILP) and a greedy heuristic. For pool RTIs, our models enable a reduction of 30% in the number of trucks used and 20% in the distance traveled. Furthermore, if dedicated and pool RTIs are jointly planned, this would enable a 9% gain in terms of transportation costs.