Domicile Selection and Risk Pooling for Trucking Networks

Less-than-TruckLoad (LTL) carriers transport small to medium sized shipments through networks of terminals. One of the issues in designing an LTL network is to determine how drivers should be distributed among locations, with the goal of reducing total variability through risk pooling. By concentrating drivers at a limited number of terminals, the carrier has greater flexibility to respond to random variations in demand, because they can be more readily shifted among routes. The problem is challenging because the risk-pooling objective is concave/minimization, leading to multiple locally optimal solutions. Each solution is defined by a rank ordering of terminals according to assigned demand. When a tour visits multiple terminals, it is assigned to the terminal with the highest rank (i.e., the one with the largest total demand). Based on the rank-ordering concept, a heuristic algorithm and an exact (branch-and-bound) algorithm are developed and applied to test problems. The exact algorithm offers reasonable computation times for moderately sized networks, typical of center-to-center routes of LTL carriers. For large networks, incorporating smaller end-of-line terminals in addition to centers, early termination of the algorithm seems to be a reasonable heuristic.