Extended instance sets for the parallel serial-batch scheduling problem with incompatible job families, sequence-dependent setup times, and arbitrary sizes

Published: 22 December 2022| Version 2 | DOI: 10.17632/rxc695hj2k.2
Contributor:
Christian Gahm

Description

The extended data set (UGWT) for the parallel serial-batch scheduling problem with incompatible job families, sequence-dependent setup times, and arbitrary sizes contains instances with up to 3,200 jobs and different types of setup times (symmetric Euclidean - SE, asymmetric Euclidean - AE, and arbitrary - AR). The instances are separated into three classes: small (S), large (L), and very large (XL). Main characteristics of the small instances S: number of jobs n = 15, 30, or 60; number of machines m = 1, 3, 4, or 5; number of incompatible job families q= 3, 5, or 10; Main characteristics of the large instances L: number of jobs n = 100, 200, or 300; number of machines m = 1, 3, 4, 5, or 10; number of incompatible job families q= 3, 5, 10, or 20; Main characteristics of the large instances XL: number of jobs n = 800, 1,600, or 3,200; number of machines m = 5, 10, or 20; number of incompatible job families q= 10, 20, or 40; Further characteristics: - The maximum batch capacity (bc) is set to 50 for all instances. - Capacity requirement scenarios defining job sizes: crs = CRS1 with [1, 12], CRS2 with [1, 25], CRS3 with [1, 50], or CRS4 with [13, 38]. - Job processing times and weights are drawn from discrete uniform distributions with the parameters [1, 100] and [1, 10]. - Distribution of jobs to families: jtfam = UD, discrete uniform distribution with {1,…, q} and jtfam = ND, normal distribution with the parameters q/2 and 0.2q (with discretization); Note that due to the assignment procedure, the actual number of families of an instance can be different from q (the q value used for instance generation is depicted in the filename, whereas the actual number of families is depicted within the files). - Setup time type: st = SE, AE, or AR - Setup time severity factor: eta = 0.25 or 0.75 - Tardiness factor: tf = 0.3 or 0.6 - Due date range factor: rdd = 0.5 or 2.5 Because not all instance characteristics combinations are reasonable regarding batch capacity requirement scenarios, instance sets are restricted to those combinations with a sufficiently large, approximated number of batches per machine. Altogether, the three instance sets S, L, and XL contain 93,360 instances in total: |S|=22,320, |L|=57,600, and |XL|=13,440. All the scheduling relevant data is given within the files, whereby each file represents one problem instance. A unique identifier is given by the filename and also in the text file itself. In addition, best-known objective values (and computation times) for several solution methods are included. The last value of the folder names indicates the number of instances with the corresponding characteristics). More details on the instances can be found in "Learning-augmented heuristics for scheduling parallel serial-batch processing machines" (Computers & Operations Research,10.1016/j.cor.2022.106122).

Files

Institutions

Universitat Augsburg

Categories

Batch Scheduling Scheduling

Licence