Brittleness Index Prediction From Conventional Well Logs in Unconventional Reservoirs Using Artificial Intelligence

Abstract

Abstract Brittleness index is one of the critical geomechanical properties of reservoir rocks. In fact, it is not possible to have accurate solutions to evaluate hydraulic fracturing effectiveness without having accurate brittleness value. Different methods were proposed to measure this parameter, which some of them need reservoir geomechanical parameters by tri-axil experiments, and some others such as mineral analysis are very expensive. Elastic based brittleness is commonly applied because it can provide continuous profile but S wave is generally not performed in all wells. Therefore, attempts have usually been carried out to use artificial intelligences for identification of the relationship between the well log data and core mineralogy based brittleness. This study proposed a new approach based on two common neural network approaches by conventional logging data. The mineral based brittleness database from one well in Jiaoshiba shale gas filed, China were established through QEMSCAN and X-ray diffraction (XRD) analysis. Construction of back propagation artificial neural network (BPANN) and least squares support vector regression (LS-SVR) for predicting formation brittleness is described. Moreover, an overall ranking of sensitivity degree is then provided to show the impacts of different well logs as inputs on models. Obtained results of both LS-SVR and BPANN model are competent for predicating brittleness. But, LS-SVR approach is more accurate than the BPANN method at same conditions. These artificial intelligence methods are first adopted in formation brittleness prediction and the integration of core mineralogy based brittleness and well logs is more practical and applicable than elasticity-based brittleness method using acoustic full waveform logging.