HFSR_instances

This dataset contains hybrid flow shop (HFS) rescheduling instances derived
from OR-Library benchmarks. It covers three numbers of stages K = {2, 5, 8}
and includes two blocking settings: (i) no blocking constraints and
(ii) mixed blocking constraints.

For each configuration, instances are generated for five batch sizes
N = {5, 10, 20, 50, 100} and 10 datasets defined by:
- Processing times p_ik ~ DU[1,100]
- Job weights w_i ~ DU[1,5]
- Machines per stage m_k ~ DU[1,5]

Each instance is stored as a JSON file containing three disruption
scenarios: mild, medium, and severe. For each disruption level, the file
specifies:
- The full data of newly arriving jobs (release times, processing times,
and weights),
- The identifiers of cancelled jobs, if there is any.

Each JSON file also includes the baseline schedule obtained in the
predictive phase using a genetic algorithm minimizing the Total
Weighted Waiting Time (TWWT). The baseline schedule contains:
- The best job permutation found by the GA
- Starting times of all operations
- Completion times of all operations
- The corresponding TWWT value