Abstract

Smelters are under increasing pressure to boost operational efficiency, cut costs, and mitigate environmental impacts, especially through the reduction of off-gas emissions. These challenges are compounded by surging demand for critical minerals, driving the need for advanced, sustainable operations that support efficient production. Traditional smelter scheduling practices, particularly heuristic-based approaches, are often inadequate in meeting these complex demands, resulting in inefficiencies such as underutilized assets, inconsistent shift performance, prolonged operational delays, and emission levels that frequently exceed regulatory requirements. 

This paper proposes an integrated, AI enabled smelter production planning and scheduling framework that holistically considers emission management and operational efficiency. Through this approach, smelters gain enhanced visibility and control over their off-gas emissions, enabling timelier, data-driven responses to maintain environmental compliance, while meeting production requirements. The framework leverages real-time data integration, mathematical optimization, and generative AI models, providing a sophisticated solution that enables smelters to monitor production performance in real time and seamlessly adjust schedules and processes to maximize production and minimize environmental impact. 

This cloud-based Hatch solution is built to support comprehensive scheduling for furnace, crane, and converter aisle activities, optimizing the entire production chain. Through the integration of computer vision and generative AI, the solution enhances operational oversight, allowing precise monitoring of critical activities such as ladle movements, crane operations, and furnace tapping events. These technologies capture high-resolution timestamps and activity data; generates a continuous feedback loop, whereby any deviations in production performance, off-gas emissions or process inefficiencies can be identified and addressed immediately; and create an ongoing comparison between actual and planned schedules, enabling further analysis of performance gap against established benchmarks. The solution has demonstrated potential reductions of up to 50% in molten material waiting time, alongside substantial improvements in production, shift-to-shift performance consistency, and emission control. 

The paper concludes with a future vision of smelter digital transformation, emphasizing integrated business planning as a complement to operational scheduling. This integrated approach aims to optimize long-term monthly planning across the entire value chain, from concentrate blending to smelting and potential downstream casting. It accounts for complex trade-offs such as raw material availability, secondaries, process capacity, target copper grade, and fluctuating market prices. 

By leveraging prescriptive analytics, real-time data tracking, and advanced AI capabilities, the authors foresee a transformative path for smelters, which is expected to enhance operational resilience, sustainability, and profitability, enabling smelters to meet both production goals and environmental standards. As the industry faces increasing demands, this digital transformation will empower smelters to navigate challenges with greater agility while optimizing both economic and sustainability outcomes.