IndonesiaEnergy and utilities2025 – ongoing

Vale Indonesia

Machine learning to adjust the slag furnace Silica/Magnesia ratio

Vale Indonesia's machine learning model lifted slag furnace ratio compliance from 89% to 95%

WorkingCompany's own claim

Problem

Operators at Vale Indonesia's nickel smelter adjusted the Silica/Magnesia (S/M) ratio of the slag furnace by hand, around the clock. Adjustments could be delayed by 30 to 150 minutes.

Approach

Vale built a machine learning system that adjusts the S/M ratio automatically. The project won a Gold award at OPEXCON 2025.

Results

89% to 95%average S/M compliance, as reported by the Indonesian Mining Association siteCompany's own claim[1]

30-150 minutes to 0adjustment delay, as reported by the Indonesian Mining Association siteCompany's own claim[1]

about 182 tons per yearpotential nickel recovery, as reported by the Indonesian Mining Association siteCompany's own claim[1]

up to 6,000 t CO2eq per yearpotential emissions reduction, as reported by the Indonesian Mining Association siteCompany's own claim[1]

Takeaway

Removing the adjustment delay was the main change. The nickel and emissions figures are potential annual values stated by the company.

Sources

Inovasi Machine Learning PT Vale raih emas di OPEXCON 2025
Indonesian Mining Association (ima-api.org)13 Nov 2025News article

Source summaries

[1]Indonesian Mining Association (ima-api.org)13 Nov 2025

The article reports that PT Vale Indonesia's machine learning system automates S/M ratio adjustment in the slag furnace and won Gold at OPEXCON 2025. It gives compliance rising from 89% to 95%, adjustment delay falling to zero, and potential savings of about 182 t of nickel and up to 6,000 t CO2eq a year.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026