As part of intermodal transportation, drayage operations are movements of empty and full containers in the hinterland area of an intermodal terminal. We consider a truck company owning a fleet of trucks and that receives a set of import and export missions from different clients to fulfill. The trucks start their journey from the truck company yard and execute the assigned missions by moving containers between different locations (depots where containers are stored, industrial customers and terminals) while respecting the associated time windows. This problem is known as the Daily Drayage Problem with Time Windows (DDPTW). The problem can be seen as a generalization of the Pickup and Delivery Problem with Time Windows (PDPTW). In addition to the traditional PDP constraints, composed requests introduce precedences between pairs of pickup and delivery requests, as well as related time constraints. Since solving the Mixed Integer Linear Programming (MIP) formulation of the DDPTW using a standard solver cannot scale on large instances, we propose a Large Neighborhood Search (LNS) heuristic.