Leading Steel Manufacturer Saves $14M Annually in Energy Costs with AI Energy Forecasting

A top global crude steel producer spending nearly a quarter of production costs on energy had no predictive capability — only an internal monitoring tool providing plant-level energy analytics after the fact. This forced frequent unplanned purchases of external grid power, triggering demand charges and energy cost surges that sometimes led to abrupt production halts, directly impacting throughput and revenue.

In a 5-month deployment, C3 AI configured C3 AI Energy Management for a hot roll mill at the company's largest plant, ingesting 1 year of operational and energy data including 180,000 manufactured steel rolls. Three ML models were configured to disaggregate and forecast energy use at facility and equipment levels, with a deep learning model to optimize production schedules based on time-of-use rates and demand charge avoidance.

The solution generated $14 million in annual energy cost savings at one steel mill from increased onsite power use and reduced demand charges, plus $8 million in additional annual revenue from a modest increase in mill throughput. Onsite power use increased modestly, and utility demand charges were reduced by 40 MW per month.

• Forecasting energy demand at the equipment level — not just plant level — enables precision scheduling that avoids demand charge peaks without disrupting throughput. • Modeling physical product movement through production systems is a key differentiator that allows energy forecasts to be tied directly to planned production schedules. • A $14M energy saving at a single mill creates a compelling business case for rapid rollout to multiple facilities in a multi-plant steel operation.