DEVELOPMENT OF CRITICAL TOTAL DRAWDOWN PRESSURE AND GEOMECHANICAL PROPERTIES MODELS: A DATA-DRIVEN APPROACH

Critical total drawdown (CTD) along with the sand production index method (based on static Young’s modulus (Es) and static Poisson's Ratio (????s)) are commonly used to detect sand production. Some CTD, Es, and ????s models are used in the literature. However, these published models have limitations: lack of accuracy, limited data ranges, and lack of proving the relationships between the inputs and outputs. Moreover, the previous Es and ????s models which are used to determine the sand production prediction fail to detect the accurate sand onset tendency and rock’s types. This study aims to apply data-driven approaches to accurately and reliably predict the CTD, Es, and ????s for accurate evaluation of the sand rock consolidation and sand onset tendency. Different data-driven models, namely the Adaptive-Neuro-Fuzzy-Inference-System (ANFIS) and Gaussian-process-regression (GPR) were developed based on 23 wells from the Adriatic Sea for the CTD, 1853 and 1691 datasets from the United States, Malaysia, India, Saudi Arabia, and Venezuela for the Es and ????s.