A Hybrid Data-Driven/Physics-Based Modeling Approach for Hydraulic Fracture Initiation and Early-Phase Propagation in Shale

Abstract

Optimal hydraulic fracture (HF) initiation and early-phase propagation results in minimal near-wellbore tortuosity, decreasing the likelihood of screenouts and maximizing the resultant well productivity. While most predictive models for the HF geometry produced in a treatment consider the far-field region, the near-well vicinity should be an integral part of a properly-engineered reservoir stimulation strategy. In this work, a novel hybrid data-driven/physics-based approach is elaborated for modeling HF initiation and early-phase propagation from perforated horizontal wells.A treatment-optimization scheme via oriented perforating is presented using the developed hybrid model, considering the orientation of the induced HF initiation (longitudinal or transverse with respect to a well drilled along the minimum horizontal in-situ principal stress) and the resultant formation breakdown pressure (FBP). Transverse HF initiation and early-phase propagation is ideal for wells drilled in low-permeability “tight” formations, while FBP minimization suppresses the overall on-site hydraulic horsepower (HHP) requirements. The demonstrated optimization scheme is applied separately to the in-situ stress states of seven prolific shale plays from the U.S. and Argentina, suggesting oriented-perforating strategies that target the promotion of transverse HF initiation in two of these (Barnett and Marcellus), while suggesting oriented-perforating strategies that target FBP minimization in the remaining five (Bakken, Fayetteville, Haynesville, Niobrara, and Vaca Muerta). The effectiveness of oriented-perforating can potentially be compromised by fracturing fluid leakages around the casing’s circumference, which is shown to hinder transverse HF initiation.The hybrid-modeling approach is also used to estimate fracture initiation pressure (FIP) values for the seven shale plays studied, indicating significant discrepancies with analytical expressions traditionally used to approximate these FIPs in modern HF computational simulations. This sets the basis for expanding this hybrid-modeling approach over a range of in-situ stress states, incorporating aggregate data-driven (numerically-derived) correction factors in order to compensate for inaccuracies in the analytical approximations, which comprise the physics-based core of the proposed hybrid model.