Model development for shear sonic velocity using geophysical log data: Sensitivity analysis and statistical assessment

Abstract

Reservoir geomechanical parameters play a vital role in evaluating sanding potential, wellbore stability, and drilling performance for making decisions on development strategies of oil and gas fields. Compared to expensive experimental/laboratory investigations, the petrophysical log data-driven shear sonic velocity models are proven to be cost effective and quick to estimate geomechanical formation properties. The objective of this study is to introduce a new deterministic model using the most contributing predictor variables and to rank the variables according to their relative importance while predicting dynamic shear sonic wave velocity for clastic sedimentary rocks. The least-squares support vector machine with a global optimization technique is used to construct a data-driven model for shear sonic velocity estimation. A new regressive correlation is also proposed. The model performance is assessed using statistical parameters to ensure the model accuracy and reliability. Based on the relative contribution of log variables, the influential variables are ranked (higher to lower order) as follows: acoustic compressional sonic velocity, porosity, and bulk density. The developed model is validated; the predictions are compared with real data and existing log-based correlations for clastic sedimentary rocks using data from two real fields, namely the North Sea and the Niger Delta basins. The developed shear wave velocity model will assist geomechanists and drilling engineers in obtaining accurate formation elastic properties and rock strength, which are required for wellbore failure analysis as well as assessment of sanding occurrence during exploration and drilling.