Data Driven Shear Wave Velocity Prediction in a Carbonate Reservoir Using Conventional Well Logs

Abstract

ABSTRACT Geomechanical characteristics such as Young's modulus, Poisson's ratio, and Lamé parameters may be estimated using a combination of compressive and shear velocities. The static characteristics of formations and in-situ stresses may be estimated using these parameters. Sonic logs aren't typically available, especially for older wellbores. Additionally, on several occasions where the sonic logs have been made accessible, they are usually P-waves, and shear wave velocities are not present. Unfortunately, because of the significant expense associated with this log, it is not extensively utilized. Obtaining the shear wave velocity of rocks in a more efficient and cost-effective manner is necessary. Since well logs are vital in the reservoir characterization and development process, methods for approximating missing log information are essential. According to the literature, there is no specific straightforward correlation that could be utilized to accurately determine shear wave travel times based on well log data. The primary objective of this study is to predict shear wave velocity logs from conventional wire-line logs at the Wellington oil field in south central Kansas. This leads to the study's main idea. This research focuses on establishing a meaningful correlation between conventional logs and shear wave velocity in a cost-effective manner utilizing various approaches and data mining. The Wellington oil field, located in south central Kansas, is the subject of this research. Empirical correlations, linear regression, and neural networks were used initially to predict shear wave velocity. Ultimately, the data-driven model developed using standard well logs accurately predicts shear wave velocity in the Wellington oil field, demonstrating its reliability and effectiveness. Therefore, while direct measurement of shear wave sonic velocities is preferable, derived values can still provide useful insights when direct measurements are not possible or practical. INTRODUCTION Characterizing variations in petrophysical parameters over an oil and gas field's sedimentary interval is essential to its development (Archie, 1950). Consequently, laboratory measurements on core plugs, analysis of petrophysical well logs, and seismic inversion all contribute to the development of valuable estimations of reservoir physical parameters. The optimum way for determining uncertainty in predictions is to combine these several approaches, which have direct consequences for risk control in drilling operations (Pennington, 2001).