Predicting mineralogy by integrating core and well log data using a deep neural network

Abstract

Mineralogy is an essential parameter for calculating the rock properties of reservoir formations. To evaluate mineralogy, mineralogy logs and core analysis results are generally used. Additional logs can provide information on mineralogy at all depth intervals of the borehole, but the provided information is acquired from indirect measurement with increasing costs. Core data can give direct measurements, but it is often uneconomical to acquire cores for all depth intervals. A deep neural network (DNN) can be used to extract relationships among multiple nonlinear features. Due to this advantage, DNNs have been actively applied to geophysical problems, which are solved through nonlinear equations. In this study, we propose a DNN model to predict the weight fractions of minerals from integrated conventional log data and X-ray diffraction (XRD) results analyzed using core data. To compensate for the limited XRD results, Fancy principal component analysis was adopted to augment the XRD results before training the DNN. We conducted blind tests to verify the effectiveness of the trained DNN model using field data. The trained DNN model is reliable and cost effective, demonstrating applicability to the prediction of mineralogy.