The present work aims to solve a 2-Echelon Multiple-Trip Vehicle Routing Problem jointly considering Capacitated Satellites and Reverse Flows (2E-MTVRP-CSRF). It fills two gaps in the two-echelon VRP literature: First, to the best of our knowledge, only inbound distributions activities were addressed in 2E-VRP. Second, the capacity at intermediate facilities (also called satellites) is generally considered either unbounded or null.

We propose a decomposition-based matheuristic. It iteratively optimizes the first and the second echelon, by fixing the other echelon, and creates new complete solutions by solving a column-based MILP formulation with the set of generated trips.
The optimization of each echelon is performed with tailored Small and Large Neighborhood Search heuristics.

Our numerical experiments highlight that using medium-size satellites achieves a good trade-off between the routing cost and the cost of storage spaces in city-cores. In addition, we found that the integration of forward and backward flows in this type of setting reduces the routing costs by $34\%$.