Artificial Neural Network Modeling of High Pressure Descaling Operation in Hot Strip Rolling of Steels

Abstract

High pressure (HP) hydraulic descaling is usually used in hot strip mills to remove oxide scales from steel strip formed after the reheat furnace and during hot rolling. In the present work, an artificial neural network (ANN) model was developed to improve efficiency of the HP hydraulic descaling operation using flat spray nozzles. The model was trained based on the industrial data from the hot strip rolling mills of Mobarakeh Steel Complex. The spray angle, inclination angle, spray pressure, vertical spray height and water flow rate were all considered as the main input parameters of the HP descaling operation.The ANN model developed is able to predict spray impact, spray width and spray depth for any given spray nozzle system. The model can also analytically compute the spray overlap for a given spray nozzle arrangement. A sensitivity analysis was carried out using the ANN model. It was shown that, among all process parameters, the spray angle followed by the inclination angle are the most important parameters affecting the spray impact. The model developed can be used as a proper tool to improve the efficiency of the descaling system in terms of achieving the highest spray impact and an optimum spray overlap for any process condition.