Furnace Equipment Temperature Optimization: 4% LNG Savings

In the hot-dip galvanizing process, consistent furnace temperature control is crucial for producing uniform material. Although rule-based control scenarios exist, field engineers rely on their experience, leading to variations in quality and energy usage. A systematic, data-driven approach is needed to optimize control methods and ensure consistency.

A digital twin of the furnace equipment was implemented using accumulated operation data and a dynamics model. An objective function reflecting the expertise of experienced field engineers was defined. Model predictive control (MPC) based on the dynamics model was applied to identify optimal control logic, minimizing energy usage while achieving target material quality.

Data from PLCs was used to train the dynamics model on real-time furnace conditions. The highly accurate MPC control logic derived optimal temperature setpoints, reducing LNG usage by up to 4% while ensuring consistent target material quality. Additionally, an AI operational environment (MLOps) was established, enabling the dynamics model to continuously learn and quickly adapt to changes in equipment and the environment.