BiObjective Flowshop (makespan-resources)

Description

This dataset contains the benchmark instances, non-dominated sets (Pareto fronts), and computational performance metrics associated with the research paper "A bi-objective permutation flowshop with additional resources during setups: exact algorithms and metaheuristics". Problem Description: The data corresponds to a bi-objective permutation flowshop scheduling problem that considers sequence-dependent setup times and additional resource consumption during those setups. The two objectives to be minimized simultaneously are the makespan and the maximum number of additional resources required. Dataset Contents: The repository is divided into three main components: Benchmark Instances: Text files containing the problem parameters (number of jobs/machines, processing times, sequence-dependent setup times matrices, and sequence-dependent resource consumption matrices) for both Small and Large instances. Pareto Fronts: Text files containing the non-dominated solutions in the objective space [Makespan, Resources]. It includes the exact Pareto fronts obtained via the Epsilon-Constraint method (using CP and MILP models) for small instances, and the approximated fronts generated by the proposed metaheuristic and the baseline MOEA for larger instances. Performance Metrics: Excel spreadsheets summarizing the Hypervolume and Unary Epsilon Indicator results across all tested configurations. Please refer to the included readme.txt file for detailed instructions on the data structure.

Steps to reproduce

Download and extract the benchmark instance files from the "Small_Instances.zip" and "LargeInstances.zip" archives. Please follow the instructions detailed in the readme.txt file, paying special attention to the non-zero diagonals in the setup time and resource consumption matrices (which denote the initial setup values). Implement and execute your bi-objective scheduling algorithms (exact methods or metaheuristics) using these parameters to simultaneously minimize the makespan and the maximum number of additional resources. Record the non-dominated solutions (Pareto fronts) obtained by your algorithms. To benchmark your results, compute the performance metrics (e.g., Hypervolume and Unary Epsilon Indicator) of your generated fronts. Compare your fronts and metrics against the baseline exact solutions (provided for small instances) and the approximated fronts generated by the MOIG-RLS and MOEA algorithms provided in this repository.

Institutions

• Universidad de Valladolid

  Castille and León, Valladolid

Categories

Licence