Search

Some of the important rock physics parameters, such as the shear-wave velocity and Poisson's ratio, are conventionally calculated from compressional and shear sonic well logs. Although these parameters are vital for geomechanical purposes, these types of shear sonic logs are rarely recorded for most wells. Therefore, this study aims to use ordinary well log and seismic data to predict the Poisson's ratio using some of the machine learning algorithms that are based on a proposed model calculated from a modified version of the Random Vector Functional Link (RVFL) using the Wild Geese Algorithm (WGA). This is applied as a case study in the Scarab gas field in the West Delta Deep Marine (WDDM) concession, Egypt. The main aim of using WGA is to determine the best configuration from the parameters of RVFL to enhance the process of prediction. The rock physics templates are used for interpreting the lithology and pore-fluid from well log data and RVFL-WGA. This is achieved using the cross-plot of P-impedance versus Poisson's ratio, Lambda-Rho versus Mu-Rho, Poisson's ratio versus bulk modulus and P-impedance versus Vp/Vs ratio from both methods. All cross plots are color-coded by the shale volume and hydrocarbon saturation. • The WGA method is used as a novel technique for computing the Poisson's ratio. • Computing Poisson's ratio from seismic data in the absence of some well log data. • Use of well log and seismic data to predict Poisson's ratio using machine learning. • The RVFL-WGA used to increase the validity of the prediction of Poisson's ratio. • Good matching between logs shows that machine learning tool is a reliable method. [ABSTRACT FROM AUTHOR]