Library of benchmark test sets for variants of the Capacitated Arc Routing Problem under Time Restrictions with Intermediate Facilities

Published: 17 January 2017| Version 3 | DOI: 10.17632/9x4vd92rcj.3
Contributors:
Elias J. Willemse,
Johan W. Joubert

Description

Please use the following article when siting the Cen-IF or Lpr-IF datasets: Willemse, E. J. and Joubert, J. W. (2016a). Benchmark dataset for undirected and mixed capacitated arc routing problems under time restrictions with intermediate facilities. Data in Brief, 8:972–977, DOI: 0.1016/j.dib.2016.06.067. Please use the following article when siting any of the other datasets: Willemse, E. J. and Joubert, J. W. (2016b). Constructive heuristics for the mixed capacity arc routing problem under time restrictions with intermediate facilities. Computers & Operations Research, 68:30–62. The sets are also extensively used and described in: Willemse, E. J. and Joubert, J. W. (2016c). Splitting procedures for the mixed capacitated arc routing problem under time restrictions with intermediate facilities. Operations Research Letters, 44(5):569–574. Willemse, E. J. (2016). Heuristics for large-scale Capacitated Arc Routing Problems on mixed networks. PhD thesis, University of Pretoria, Pretoria. Available online from http://hdl.handle.net/2263/57510 (Last viewed on 2017-01-16) The dataset contains 5 benchmark test sets for the Mixed Capacitated Arc Routing Problem under Time Restrictions with Intermediate Facilities (MCARPTIF), which is a generalisation of the Capacitated Arc Routing Problem. The problem closely represents waste collection routing and caters for mixed road networks and vehicles unloading their waste at Intermediate Facilities. The dataset also contains 5 CARPTIF sets with undirected networks that only have two-way streets. The benchmark instance files of each set was compressed into a single .zip file, available for download below. Each instance file is a raw .txt file. The format of each file is given by the MCAPRTIF_benchmark_format.pdf file, also available below. The Cen-IF-Full, Cen-IF, Cen-IF-part and Act-IF sets are based on real road networks requiring waste collection services. The Lpr-IF and mval-IF sets were derived from the sets available from http://www.uv.es/belengue/mcarp/. The bccm-IF, bccm-IF-3L, gdb-IF, and gdb-IF-3L sets were derived from sets available from http://www.uv.es/belengue/carp.html. The Cen-IF and Lpr-IF instances were first used by Willemse & Joubert (2016c). All the other sets, except for Cen-IF-Full and Cen-IF-part, were used by Willemse & Joubert (2016b). For the Cen-Full-IF, Cen-IF, Cen-Part-IF, Act-IF and Lpr-IF instances, cost data are given in seconds, and demand data in killograms. The units for the other benchmark instances are not known.

Files

Steps to reproduce

Act-IF: The Act-IF dataset was derived from a case study reported by Willemse & Joubert (2016b). The dataset was created using OpenStreetMap data and information supplied by the local waste collection authority. It is therefore not available for direct reproduction. Cen-Full-IF, Cen-IF and Cen-Part-IF: A Geospatial Information System (GIS) data set of the Centurion area, courtesy of Business Connexion, was used to create three benchmark instances. The data set accurately describes the network and includes a number of useful attributes. Accurate deadheading costs, service costs and waste quantities are not available for the network, so we inferred the arc-routing data using a similar approach to that of Belenguer et al. (2006). The original road network was obtained through Business Connexion is not available publicly, and therefore not available for direct reproduction. Lpr-IF and mval-IF-3L: The datasets were developed by extending the lpr and mval datasets (available from http://www.uv.es/belengue/mcarp/) introduced by Belenguer et al (2006). The original instances were randomly generated by the authors and can therefore not be directly reproduced. For the modification of Lpr to the Lpr-IF, a route duration restriction of 28800 seconds (8 hours) was imposed on each instance, and two IF facilities were located at floor(nVertices/2) and 2xfloor(nVertices/2). For the mval to mval-IF-3L conversion, IFs were located and route duration restrictions were imposed as shown in the mval-IF-3L raw data files. All the Lpr-IF and mval-IF-3L files can be reproduced by imposing the same restrictions and IF locations contained in the modified files on the the original mval and lpr instance files. gdb-IF and bccm-IF: The datasets were developed by extending the gdb and bccm datasets (available from http://www.uv.es/belengue/carp.html). All the instance files can be reproduced by imposing the same restrictions and IF locations contained in the modified files on the the original gdb and bccm instance files. gdb-IF-3L and bccm-IF-3L: The datasets are almost identical to the gdb-IF and bccm-IF sets, the only difference being that the route duration restriction was multiplied by a factor of 3. All the instance files can be reproduced by imposing the same restrictions and IF locations contained in the modified files on the the original gdb and bccm instance files. References: Belenguer, J., Benavent, E., Lacomme, P., and Prins, C. (2006). Lower and upper bounds for the mixed capacitated arc routing problem. Computers & Operations Research, 33(12):3363–3383.