Modeling the hydrocracking process using artificial neural networks

Feed-forward neural networks that models the hydrocracking process of Arabi an light vacuum gas oil are presented. The input-output data to the neural networks was obtained from actual local refineries. Several network archite ctures were tried and the networks that best simulate the hydrocracking pro cess were retained. The networks are able to predict yields and properties of products of the hydrocracking unit (e.g. iC(4), nC(4), light and heavy n aphtha, light and heavy ATK, Diesel, etc.). The predictions of yields and p roperties of various desired and undesired products at different conditions are required by refineries for process optimization, control, design, cata lyst selection, and planning. The predictions of the prepared neural networ ks have been cross validated against data not originally used in the traini ng process. The networks compared well against this new set of data with an average percent error always less than 8.71 for the different products of the hydrocracking unit.