A HYBRID ADAPTIVE LARGE NEIGHBOURHOOD SEARCH ALGORITHM FOR ROUTING LESS-THAN-TRUCKLOAD PICKUPS WITH TIME WINDOWS AND TWO-DIMENSIONAL CAPACITY

Abstract

Less-than-truckload (LTL) carriers collect numerous small shipments from different shippers and consolidate them at a cross-dock terminal. Consequently, the efficiency of local pickup routes directly affects transportation costs and truck utilisation. This paper formulates the LTL pickup routing problem with time windows, weight and volume capacity constraints, and a fixed cost per truck, and proposes a hybrid adaptive large neighbourhood search (H-ALNS) algorithm. The algorithm integrates a regret-2 construction heuristic, four destroy and three repair operators selected using adaptive weights, a simulated annealing acceptance mechanism, and an O(1)-time time-window feasibility check based on cached insertion costs. Computational experiments conducted on 15 synthetically generated instances with 50, 100, and 200 requests show that H-ALNS reduces the objective value by 18.1–24.0% and the total travel distance by 24.8–35.1% compared with the best constructive heuristic. Furthermore, it outperforms a non-adaptive large neighbourhood search by 2.3–5.2%, with computation times ranging from 1.1 to 8.6 seconds.