A Physics Inspired Data-Driven Model for Candidate Selection and Performance Assessment of Refracturing in Unconventional Wells

This study presents a physics-inspired, data-driven framework to improve refracturing candidate selection and assess post-job performance in unconventional reservoirs. By integrating advanced reservoir and well diagnostics with data-driven techniques, the proposed approach enhances the ability to identify underperforming wells with high remaining potential. A key innovation is the introduction of the Refrac Well Potential (RWP) factor, which combines Reservoir Potential (RP) and Completion Quality (CQ) to rank candidate wells. RP is derived from dynamic drainage volume (DDV), reservoir flow regimes, well spacing, average reservoir pressure, and injected fluid volumes, while CQ is based on productivity index decline trends and cumulative production. To evaluate complex production behavior and refracturing impact, advanced modeling techniques such as Sparse Identification of Nonlinear Dynamics (SINDy) and multi-segmentation analysis are applied. Field validation shows strong alignment between pre-job RWP rankings and post-job production outcomes, with high-RWP wells consistently delivering greater incremental production. Additionally, the framework enables performance comparisons across different refracturing strategies, supporting more capital-efficient decisions. This methodology offers a practical and validated workflow to optimize refracturing investments and maximize recovery in unconventional plays.