An Intelligent Approach For Obtaining True Resistivity (𝑅𝑇) From Rock Acoustic Data : A Laboratory Verification

Abstract

Rock true resistivity (Rt) is known as more sensitive than compressional-wave velocity (Vp), the principal output of a seismic survey, to variation in water saturation. Therefore, it would be of a great value if there were a way to predict resistivity distribution from seismic signals. This study is essentially an effort to see the possibility of predicting Rt from Vp through a pattern recognition approach. For the purpose, a series of laboratory tests were performed on some Central Sumatran clay-free sandstone samples of various porosity values and at various water saturation levels. For studying the pattern of relationship, artificial neural networks (ANNs) were applied. From the ‘training’ (i.e.pattern recognition) activity performed using the ANNs, it has been show between Vp and Rt in the following ‘blind test’, it has also been shown that the trained relationship can be used to estimate Rt reliably using other data as input. Comparisons between estimated and observed Rt data have indicated good agreement implying the success of the approach taken in the study. This has laid the foundation and justification for further application of the approach on seismic and well-log data.