Railyard Optimization

Railyards serve as the nerve center for freight railroads in moving goods from their origins to their destinations. Railcars are switched and new trains are formed at yards either by using humps to build outbound railcar blocks using gravity or by manually ordering railcars as per their destinations. Inefficiencies in railyard operations impact congestions in the yard resulting in excessive delays, higher operational cost per railcar switching, and lower throughput. We have developed simulation-driven optimization algorithms which in real time decide (i) where to place an inbound train in the yard, (ii) which blocks should be built where in the yard, and (iii) in which order outbound trains should be built. These algorithms use metaheuristics including very large-scale neighborhood search for the optimization. Our algorithms have been validated for a very large yard for a railroad in North America demonstrating significant improvements in yard key performance indicators.