Case Study

We built a digital twin of an enameling process to simulate the furnace's operations under different conditions. Through ten experiments, we investigated the effects of using pure oxygen as an oxidizer, using inverters in the air supply system, dynamically controlling the set temperature point, and controlling the stoichiometric point of the fuel mixture. We measured fuel consumption, electricity consumption, harmful emissions, air temperature in the furnace, load temperature, and flame temperature for each experiment.

​

The goal of this project is to develop an accurate model of the enameling furnace and discover opportunities to reduce fuel costs by using intelligent control algorithms based on Reinforcement Learning (RL). RL is used to quickly find an optimal automated control scheme for the furnace under different conditions. An agent created using RL can determine the optimal operating parameters of the furnace in real-time based on the size, mass, and material of the load and the current state of the furnace.

​

In the short term, installing an intelligent control algorithm for dynamic control of the set temperature point proved to be the fastest solution. During tests, this algorithm reduced fuel consumption by 14.7% and electricity consumption by 34%. Thanks to the algorithm's ability to predict the future need for thermal energy based on the load properties and current furnace state, the algorithm was able to reduce the amplitude of the error correction of the PI controller.

Key concepts

• Digital Twins

• Energy-efficiency

• Artificial Intelligence

• Enameling process​

Results

• 14% reduction in gas consumption

• 34% reduction in electricity consumption

• Digitalized process