Novel Diagnostic Method for Accurate Identification of Water Coning and Channeling in Oil Wells

Reservoir heterogeneity and suboptimal production practices frequently lead to premature water breakthroughs in oil wells through channeling and coning, resulting in accelerated production decline and reduced recovery efficiency. Reliable identification of the governing water production mechanism is therefore essential for effective reservoir management and timely intervention. The conventional diagnostic approach, based on the derivative of the water-oil ratio (WOR') with respect to time (Chan plot), is widely used but is highly sensitive to production fluctuations arising from operational practices, such as choke optimization. This often introduces ambiguity in distinguishing between channeling and coning, necessitating the use of costly downhole diagnostic tools, including production logging and temperature/noise logs. Existing alternatives, such as machine learning and nonparametric methods, are often data-intensive and lack field-level simplicity. This study introduces a novel diagnostic method based on the time derivative of the cumulative water-oil ratio [d/dt(CumWOR)]. The proposed approach generates distinct and stable signatures for identifying water production mechanisms, including channeling, coning, and normal water breakthroughs, while demonstrating improved robustness to data variability. The methodology is validated using production data from 140 wells across multiple fields. In 93% of the cases, the proposed technique exhibits reduced data scatter and clearer trend identification, with results consistent with independent diagnostics from multirate tests and downhole logs. In contrast, conventional WOR' analysis frequently yields ambiguous trends, increasing diagnostic uncertainty. A mathematical formulation of the method is presented, highlighting its enhanced noise attenuation capability relative to traditional approaches, thus better signal capture of the water production trend. The proposed technique is computationally simple, relies solely on routinely acquired production data, and is readily deployable in field applications. It provides a cost-effective and reliable tool for real-time water production diagnostics, enabling early detection of water breakthrough and reducing dependence on expensive downhole surveillance.