Here you find instance definitions and the best known solutions (to our knowledge) for the (approximately) 100, 200, 400, 600, 800, and 1000 tasks instances of Li & Lim's PDPTW benchmark. Each task is either a pickup or a delivery, and has a sibling (a delivery or a pickup). The version reported here has a hierarchical objective: 1) Minimize number of vehicles 2) Minimize total distance. Distance is Euclidean, and the value of travel time between two nodes is equal to the value of distance. Both should be calculated with double precision. Total distance results are rounded to two decimals in the tables below. Exact methods typically use a total distance objective only and use integral or low precision distance and time calculations. Hence, results are not directly comparable.
The Li & Lim PDPTW instance definitions are found here (zip files):
The pdp 100-cases (56 instances)The pdp 200-cases (60 instances) The pdp 400-cases (60 instances)The pdp 600-cases (60 instances)The pdp 800-cases (60 instances) The pdp 1000-cases (58 instances*)
*The two 1000-task instances LRC2_10_8 and LRC2_10_9 were not present in the original data sent to us by Li & Lim. Supposedly they forgot to include them. If you happen to have these definitions, we would very much appreciate if you could forward them to firstname.lastname@example.org .
Published April 18, 2008