A fluid catalytic cracking (FCC) unit has been simulated by integrating FCC riser reactor and regenerator models. This simulation uses a new10-lump riser reactor kinetic model developed in-house. The lumping scheme and reactions are based on more detailed description of the feed in terms of PNA (paraffins, naphthenes, and aromatics) in both light and heavy fractions. An artificial neural network (ANN) model, also developed in-house, relates routinely measured properties such as specific gravity, ASTM temperatures, and so on to the detailed feed composition needed for the kinetic model development. The conversion and product yields obtained by integrating the model equations were found to be in close agreement with those measured in the plant in all the cases investigated. Simulation results using the present model, when compared with results from a conventional 5-lump model, clearly brought out the improvement in prediction because of detailed feed description calculated from ANN models. A parametric sensitivity study was undertaken with respect to operating conditions such as effects of feed preheat temperature, feed flow rate, and reactor outlet temperature (independent variables) on the performance of the FCC unit, and the results have been discussed.