OPSEARCH Управленческие информационные системы

Data-driven optimization of ferro alloys usage for enhanced efficiency and cost reduction in steelmaking, JSW steel

Padhy, Janaki K., Agarwal, Pradeep, Ramar, Ravindran, Bhange, Gaurav
Springer Nature · 2026-10-21 · DOI 10.1007/s12597-026-01297-6

Аннотация

Achieving precise chemical composition in the manufacturing of steel is critical for meeting quality standards and customer specifications. During the steelmaking process, Ferro alloys are added to adjust levels of elements like carbon, manganese, and silicon to reach the target composition. Previously, operators relied on experience or basic spreadsheet-based solvers, leading to inconsistent final chemistry, excessive use of costlier alloys, and longer production times. This lack of standardization caused variability in product quality, increasing costs and risking customer satisfaction. The issue impacted production efficiency significantly, making it essential to standardize and optimize the Ferro alloy addition process for cost-effective steel production. To minimize Ferro alloy costs while achieving desired chemistry, a Data Science model was implemented in 4 months’ time frame. This model determines optimal alloy quantities based on recovery rates, chemistry specifications, and alloy prices. A Python-based application built with Flask (web-based application) integrates this solver for real-time analysis and recommendations on live heats deployed in the operator control areas. The solution employs a linear optimization solver to calculate precise Ferro alloy quantities, achieving target chemistry efficiently with minimal adjustments. A Dashboard on Power BI using Azure services provides insights and monitoring into alloy consumption, costs, and chemistry compliance. The cumulative savings resulted from this project is 133.64 Crores from December 2023 till October 2024 across several steel plants (cost saving of 5–7% of total Ferro alloy cost in steel making) by investing 22 Lakhs in license and Azure enterprise setup, with a projected return on investment (ROI) of less than one month.

Читать у издателя →