Abstract
A realistic and efficient tactical model is developed that is able to optimize equipment and crew movements in long-haul trucking networks so that drivers are able to return home within a reasonable amount of time. A unique feature of the model is that driver, tractor, and trailer routes are simultaneously optimized. An underlying assumption is that routes remain in place for extended periods; though the trailers, tractors, and drivers may flow across these routes at variable rates. We formulate a static linear programming model for this tactical problem, and propose a column generation algorithm. The algorithm is tested with real data from a less-than-truckload carrier and randomly generated test data. Networks with up to 40 nodes (including the 30-node real-life example) are successfully optimized. For our sample problems, we found that static routes produced solutions within an average 0.22% of the optimum under dynamic conditions, provided that variable flows are permitted across the static routes.
Acknowledgements
This research was supported by the National Science Foundation. We also thank Consolidated Freightways for their participation in the project.
Notes
1Gap between the optimal cost using only the routes that optimize the static problem and the one without restricting the routes.
2Iteration count for optimizing the LP starting with the routes that optimize the static problem.
3CPU time for optimizing the LP starting with the routes that optimize the static problem.