Development and implementation of a mathematical model for predicting zinc coating mass in continuous hot-dip galvanizing plants

Continuous galvanizing lines require precise control of zinc coating mass (g/m 2 ) to ensure product quality, avoid overcoating, and reduce costs. However, the physical distance between the air knives and the on-line coating gauge introduces a measurement delay that hinders real-time adjustments. To address this limitation, a multiple linear regression model was developed for Continuous Galvanizing Line 1 (LZC1) at CSN. A total of 343,573 operational data records, sampled at 10-second intervals, were collected. After data cleaning and preparation, 156,201 observations were retained. Exploratory data analysis and a stepwise variable selection procedure identified the most significant predictors, including line speed, air knife pressure and knife-to-strip distance. The ordinary least squares (OLS) model, validated through 10-fold cross-validation, achieved a coefficient of determination (R 2 ) of approximately 0.78. The fitted model was implemented in a dedicated function block within the line’s programmable logic controller (PLC), where it continuously estimates coating mass in real time. A dynamic offset mechanism, frozen at the onset of each coating transition, compensates for steady-state error and accelerates convergence to the target value. Historical records confirm reduced coating variability, decreased transient overcoating, and lower zinc consumption, yielding both economic and environmental benefits.