The paper presents the architecture of a submerged arc furnace process model. It utilises the robust and simplistic solving capabilities of a time dependent, dynamic modelling by employing flowrate and composition data. Next, it employs continuously updated electrical data in order to optimise the model's predictive ability, with zero-order hold elements enforced on flowrate and composition data until these variables are resampled. The addition of a set of fundamentally derived variables generated through the online use of a vectorised, look-up table, is also proposed. Such a look-up table is created offline by utilising computational fluid dynamic (CFD) modelling techniques. Promising results indicate that this pragmatic methodology allows an intermediate version of this model to reasonably predict the metallurgy of the subsequent tap within the inherent accuracy of the data. The model is being tested on industrial data. (c) 2005 Elsevier Ltd. All rights reserved.