Application of Discriminant Analysis in the Interpretation of Well-Logging Data

Abstract

AbstractDuring the geophysical data interpretation of oil and gas wells, situations arise when, for various reasons, there is a lack of necessary information. Information—for carrying out traditional complex geological interpretation. The main typical tasks are the identification of reservoir rocks in the sections of wells, the determination of their characteristics, and the determination of the nature of the rock saturation. The lack of information can be compensated for by using pattern recognition methods. Here, even several curves of different logging methods, addition the combined data, allow solving the tasks of oil and gas geology with a certain probability. Methods, using artificial neural networks and, much less often, methods of discriminant analysis are quite popular at the present stage of geophysical research. The discriminant analysis method is quite easy to use and understandable from a mathematical standpoint. It is undeservedly little used in geophysical research. Below we will consider examples of using this statistical method to solve the tasks of assessing the probability of the presence of gas-saturated and water-saturated rocks in the productive sections of the wells of the Precarpathian trough gas field. The lithological features of the rocks of gas fields in the study area are the frequent alternation of clays, sandstones and siltstones in the sections of wells. As a result, the curves of geophysical parameters are monotonic, poorly differentiated lines. This significantly complicates the quantitative and qualitative geological interpretation of well logging data. Pattern recognition methods can be of great help in solving tasks associated with determining the intervals represented by gas-saturated reservoir rocks in such cases.KeywordsWell loggingGas saturationRocksReservoir rockDiscriminant analysisInterpretation