Simulation and AI-Driven Evaluation of Chemical Tracer Characteristics for Fracture Diagnostics and Interwell Communication in Sour Gas Reservoirs

Abstract

Accurate interpretation of chemical tracer data in sour, high-temperature unconventional reservoirs is fundamentally compromised by hydrogen sulfide (H₂S)-induced degradation and phase-partitioning effects. Using high-fidelity reactive transport simulations calibrated to Permian Basin conditions (Wolfcamp and Bone Spring formations), we quantify how common tracers—including fluorobenzoates and ethyl acetate—undergo 75–98% degradation within seven days at 140°C and 2000 ppm H₂S, rendering conventional interpretation protocols invalid. Neglecting these effects leads to systematic overestimation of fracture half-length by 40–60% and misinterpretation of interwell communication severity. Two field case studies from the Delaware Basin and Central Basin Platform demonstrate that substituting robust chemistries (naphthalene sulfonates, trimethylbenzene, perfluorocarbons) reduces degradation to below 30%, while degradation‑corrected analysis reduces estimated interwell communication volume from 89,000 to 62,000 barrels of oil equivalent—aligning with simulation truth (58,000 BOE) and preventing overly conservative infill drilling spacing. A trained XGBoost surrogate model (R²=0.973, prediction time <0.1 seconds) enables real‑time, degradation‑corrected field interpretation. These findings establish a validated workflow for tracer‑based fracture diagnostics in sour gas shales, directly linking chemical degradation kinetics to operational decision‑making.