Order first split second heuristic for alternative routing strategy for freight railways

Abstract

The rake movements in rail freight transportation are executed by dynamically allocating empty rakes to observed demands. This work proposes an alternate cycle planning strategy based on the average demand data. The system is modeled as the Pickup and Delivery Problem with customer specified origin and destination and multiple full truckload demands by each customer. Initially, an Exact Formulation of the problem that gives optimal cycles is presented. A modified Order First Split Second Heuristic (OFSS) is proposed to solve the deterministic version of this problem where a giant tour is constructed by assigning a value to each node based on the greedy path from that location. The splitting algorithm forms multiple cycles satisfying maximum cycle time constraint from the Giant Tour. A heuristic solution obtained by splitting optimal giant tour is also determined to study the performance of proposed heuristic. Considering the possibility of deviation of actual demands from average, an adaptive algorithm that modifies the planned cycles to accommodate the changes in demand is presented. Computations show that the heuristic gives near-optimal solution in terms of net revenue and the number of vehicles required. Testing this algorithm on dynamic demands show that the initial cycle plan can be followed with minor modification if the Degree of Dynamism of the system is small. The proposed algorithm is tested on Indian freight rail data and the improvements in rake utilization are noted.