A CRITICAL COMPARISON OF NEURAL NETWORKS AND DISCRIMINANT-ANALYSIS INLITHOFACIES, POROSITY AND PERMEABILITY PREDICTIONS

The application of a genetic reservoir characterisation concept to the calculation of petrophysical properties requires the prediction of li thofacies followed by the assignment of petrophysical properties, acco rding to the specific lithofacies predicted. Common classification met hods which fulfil this task include discriminant analysis and backprop agation neural networks. While discriminant analysis is a well-establi shed statistical classification method, backpropagation neural network s are relatively new, and their performance in predicting lithofacies, porosity and permeability, when compared to discriminant analysis, ha s not been widely studied. This work compares the performance of these two methods in prediction of reservoir properties by considering log and cove data from a shaly glauconitic reservoir. The neural network a pproach while subject to a degree of trial and error as regards the se lection of the optimum configuration of middle nodes, is shown to be c apable of excellent performance. In the example problem considered the neural network approach provided estimates superior to those based on a discriminant analysis approach. Further studies, on different forma tions, will be required to test the generality of this conclusion, and to refine the selection of neural network parameters.