Weekly-demand instances and generator for rack-loading template library design in sterilization

This dataset contains the 105 weekly-demand instances used in "Designing a compact library of validated rack-loading templates to minimize sterilization cycles" (submitted), together with the Python generator that produces them. Item families and dimensions come from the retail stockroom catalogue of Hoare and Beasley (2001), distributed through the OR-Library. Weekly quantities are generated synthetically. Each instance is a series of 104 weeks, stored as one CSV file per week (qty;length;width;height, one row per item family, dimensions in mm). The 105 instances cross five industry profiles (small, large, tall, mixed, extreme heterogeneity), three core fractions (α = 0.3, 0.6, 0.9) and seven nominal workloads (1, 5, 10, 15, 20, 25 and 30). One nominal unit is a target item footprint derived from the rack volume and a representative shelf height. The actual number of cycles depends on packing. Instances are stored in instances/cycles_alpha_<alpha>/cycles_<load>/cluster_<profile>/. The generator (src/tlbench, Python 3.11+) reproduces the benchmark and can also generate new series. The source catalogue is included in data/catalogue/. docs/instance-generator.pdf documents the generator: the size clustering of the catalogue, the workload sizing, the industry profiles and the weekly demand model. It also indicates the module implementing each step. The article refers to this note for these details. The size of all files when uncompressed is around 310 MB.