![]() | Only 14 pages are availabe for public view |
Abstract Producing undesirable phases like water from oil wells is a challenging problem in the oil industry. Water coning could be classified as one of the main reasons for that problem, as water coning is defined as a rate-sensitive phenomenon generally associated with high drawdown across the reservoirs to achieve high producing oil rates. Water coning phenomenon develops near-wellbore once the pressure forces drawing fluids toward the perforations overcome the buoyancy forces that segregate water from oil. This study implements numerical simulation to build different mechanistic models with different parameters known in the literature that affect water coning formation in oil reservoirs. Simulating water coning is very challenging due to the instabilities of matrices solvers in numerical simulators during solving severe saturation change nearwellbore unless very small-time steps and small grid sizes were used. The enormous number of simulation runs are used to quantify the effect of every parameter on the progress of water to form conning around the wellbore. Neural Network was built using datasets of the inputs and outputs parameters extracted from simulation cases to have an artificial neural network used for predicting the critical rate of production and how the uncertainty of the parameters would affect the progress of water coning and finally how this would affect ultimate oil recovery. |