An approximate dynamic programming approach for

the vehicle routing problem with stochastic demands

Clara Novoa

Texas State University, Ingram School of Engineering,

Industrial Engineering Program

San Marcos, TX 78666, USA

Robert Storer

Industrial and Systems Engineering Department, Lehigh University,

Bethlehem, PA 18016, USA

This page contains detailed results for the paper above.

· Estimated expected routing cost for instances in Set 1

· Percentage improvement in routing costs for instances in Set 1

· Estimated expected routing cost for instances in Set 2

· Percentage improvement in routing costs for instances in Set 2

· Computational times in seconds for instances in Set 1

· Computational times in seconds for instances in Set 2

· Confidence intervals for the differences between the best rollout method and the perfect information case

· Comparison of computational times between the “original rollout” and the rollout method that computes the updated base sequences using Monte Carlo Simulation (MCS) for instances with 100 and 150 customers. Click here

· Comparison of routing costs (i.e. route lengths) between the “original rollout” and the rollout method that computes the updated base sequences using Monte Carlo Simulation (MCS) for instances with 100 and 150 customers. Click here