Use case: Steel


Machine Learning modeling for drifts detection of the gas consumption of a roll mill steel furnace.



Challenge

Our client, a leader in Steel production, wanted to reduce the gas consumption of a large Rolling mill gas furnace.

Solution

We analyzed a one-year history of consumption data and influence variables (1,600 time series in 1-second timestep). After a descriptive analytics of the available data, Energiency established a gas consumption model based on artificial intelligence algorithms, detecting and taking into account the most important influential variables. The baseline thus calculated is then compared every hour with the actual consumption, which makes it possible to quickly detect any abnormal consumption.

Benefits

The model identified a 2% natural gas savings potential. The predictive model is now being used in real time in the Energiency platform workshop to help furnace operators track performance improvements.