Construção de banco de dados petrofísicos para determinação automática de litologias

Abstract

In this work we develop a solution for the classic problem of automatic lithology determination from wireline logs. A data bank generated using the discriminant analysis technique, applied over cored intervals of logged wells from a deep water field situated in the Campos basin. The clustering technique, although preferred by many workers in this field, was found to give inferior results compared to discriminant analysis. The log data, after going through environmental corrections, were placed on depth with the cored data, and a discriminant function was computed. This function classifies the log response according to the facies described from the cores. Thus we obtain a data bank, whitch allow us to perform lithologic discrimination on cored or uncored wells, in the same area. We show that the litology has to be clustered into four facies in order to get optimum results. The final product is a discrete (discriminated) curve which reproduces, reasonably well the subsurface litology. As an aplication of this method, we use the discrete (discriminated) curves to correlate three wells using an automatic correlation algorithm.