Benchmark dataset for production planning and detailed scheduling in the chemical process industry with reinforcement learning

This dataset is intended for research on production planning (PP) and detailed scheduling (DS) in the chemical process industry (CPI). It is derived from the Westenberger-Kallrath (WK) problem and enhanced with additional data regarding customer due dates, sales prices, and costs factors for production, storage and raw material purchase. The size of the problem instances is determined by the customer order count. The folder "generated-dataset" contains all generated problem instances. Each file name contains the number of customer orders and a reference if a low-margin or a high-margin price profile was applied. The folder "source-data" contains all primary data from other scientific resources, which were used in data generation process. The folder "task1-no-prices" contains two special problem instances with 40 customer orders, following the pattern of "kallrath2002-instance-task1.txt". For one instance, the divergent material flow is de-activated, hence, it contains only linear and convergent material flows.