Skip to content

Ortec2022 (VRPTW)

Ortec2022 is the MAMUT-routing curation of the static VRPTW instances from the EURO Meets NeurIPS 2022 Vehicle Routing Competition, organized with direct support from ORTEC. This competition followed the DIMACS VRPTW track by using a cost-minimization contract and integer travel costs, but it introduced a qualitatively different instance family.

Unlike Solomon and Gehring--Homberger, the ORTEC instances are not Euclidean synthetic benchmarks. They were derived from anonymized real grocery-delivery operations in the United States. Arc costs are explicit asymmetric travel-time matrices, service times and time windows come from operational data, and the coordinate fields are anonymized spatial references rather than the source of the objective. This makes the family one of the most visible modern VRPTW benchmarks based on realistic non-Euclidean travel data.

The competition included both static and dynamic VRPTW variants. MAMUT-routing currently curates the static VRPTW layer, because it matches the benchmark-as-contract goal of publishing fixed instances, machine-checkable solutions, and explicit objective metadata for the classic VRPTW. The original competition split the static instances into two subsets: a public set used during the first phase, and a final hidden set used to rank the 10 finalist teams.

The ORTEC contract differs from classical Solomon-style benchmarks in another important way: the number of vehicles is effectively unlimited such that constructing a feasible solution is always feasible and a trivial solution with one elementary route (depot -> customer -> depot) per customer is a valid Upper Bound. Like DIMACS, it uses a single-objective cost-minimization contract. In MAMUT-routing, ORTEC instances store num_vehicles = null to make this convention explicit.

The quickstart repository provides the instances and final-round competition results. MAMUT-routing converts the original TSPLIB-like text files into checker-compatible JSON, preserves license metadata (Creative Commons Attribution Non Commercial 4.0 International), and writes one MonoCost BKS per instance. For the final subset, BKS files come from the published finalist results. For the public subset, BKS files were computed separately with a custom HGS-based solver over multiple seeds and long time limits on Grid5000.

Licensing note: ORTEC instances and related redistributed BKS files in MAMUT-routing retain the original Creative Commons Attribution Non Commercial 4.0 International terms. This is a non-commercial license and therefore differs from the MIT License used for MAMUT-routing source code.

One public instance required curation beyond direct format conversion: ORTEC-VRPTW-ASYM-2e2ef021-d1-n210-k17 had one customer time window that made even the elementary route from the depot infeasible under the stored asymmetric matrix. The repaired MAMUT-routing instance records this explicitly in metadata.repair_note: customer 210 originally had [8400, 11700], repaired to [8400, 13378], where 13378 is the earliest depot-to-customer arrival time.

The family currently contains 350 static VRPTW instances: 250 public instances and 100 final instances, all with MonoCost BKS files. Note that for this family, size folders are buckets, not exact customer counts: for example, an instance under n=200 may have 258 customers.

Data

Instances and best-known solutions: /benchmarks/vrptw/ortec2022/ on the website. Repository: https://github.com/ANR-MAMUT/MAMUT-routing/tree/main/benchmarks/VRPTW/Ortec2022.

Repository README

Ortec2022 benchmark family

This directory contains the MAMUT-routing curation of the static VRPTW instances from the EURO Meets NeurIPS 2022 Vehicle Routing Competition.

License scope:

  • ORTEC instances and related redistributed BKS files retain the original Creative Commons Attribution-NonCommercial 4.0 International license (CC BY-NC 4.0).
  • MAMUT-routing conversions, repairs, metadata, and computed BKS records in this directory are distributed with this family under the same CC BY-NC 4.0 notice unless an individual file states otherwise.
  • The repository-root MIT license applies to MAMUT-routing source code, not to these ORTEC benchmark data files.

Attribution:

The source material comes from the EURO Meets NeurIPS 2022 Vehicle Routing Competition organized with ORTEC support:

https://euro-neurips-vrp-2022.challenges.ortec.com/

The license deed is available at:

https://creativecommons.org/licenses/by-nc/4.0/

Preserve this README, this directory's LICENSE file, and the repository NOTICE when redistributing this benchmark family.