Predictive models and feature ranking in reservoir geomechanics: A critical review and research guidelines.

Comprehensive investigation and accurate models of geo-mechanical properties are crucial to maintain wellbore stability and optimize the hydraulic fracturing process. This review attempts to revisit the existing correlations/models and also critically summarizes the research gaps related to acoustic waves, elastic constants, and rock strength models in geomechanics. In addition, it highlights the present status of predictive models and features ranking for uniaxial compressive strength in reservoir geomechanics. It is found that rock strength models have a distinct relationship with sonic velocities and petrophysical properties of rocks. For instance, it also explains the novel implementation of machine learning-based hybrid connectionist model and variable ranking strategies to obtain accurate geomechanical properties using core and log data. It is anticipated that the proposed research approaches will enable researchers to estimate accurate rock mechanical properties and variable ranking for sanding potentiality and wellbore failure analysis in a cost-effective through a timely manner. • Existing correlations/models of sonic acoustic velocity and rock strength parameters are critically reviewed. • Data-driven models critically analyzed to find the research gaps in rock mechanical model. • Introduced novel implementation of LSSVM-CSA based hybrid model to predict geomecahnical properties. • A feature ranking strategy explained for formation evaluation and geomechanically studies. [ABSTRACT FROM AUTHOR]