Optimization of a molten iron scheduling problem with uncertain processing time using variable neighborhood search...

Area of Science:

• Industrial Engineering
• Operations Research
• Materials Science

Background:

• Steel production scheduling is critical for cost efficiency.
• Uncertainty in pretreatment process times between iron and steel making stages poses a significant challenge.
• Existing scheduling models may not adequately address this inherent uncertainty.

Purpose of the Study:

• To develop a distributionally robust iron-steel allocation (DRISA) model for robust steel production scheduling.
• To minimize production objectives under uncertain pretreatment times while ensuring constraint satisfaction with a certain probability.
• To provide a computationally efficient method for solving large-scale scheduling problems.

Main Methods:

• Formulation of a distributionally robust optimization model for iron-steel allocation.
• Definition of an ambiguity set for pretreatment time distributions based on first and second moments.
• Development of a variable neighborhood algorithm for efficient, near-optimal solution generation.
• Validation using actual production data.

Main Results:

• The DRISA model generates robust production schedules that minimize costs.
• The developed variable neighborhood algorithm efficiently solves large-scale problems.
• Experimental results confirm the model's efficiency and the robustness of the derived schedules.
• Schedules from the DRISA model exhibit less adjustment and delay compared to nominal schedules.

Conclusions:

• The DRISA model effectively addresses uncertainty in steel production scheduling.
• Robust scheduling significantly reduces operational disruptions and associated costs.
• The proposed algorithm provides a practical solution for real-world steel manufacturing.