This study uses Artificial Neural Network (ANN) to develop correlations for estimating critical oil rate in multi-fractured horizontal wells (MFHWs) with water cresting issue.
A programming software was coupled with a commercial simulator to develop the correlation using a black oil model with typical rock, fluid and reservoir properties. The effects of reservoir fluid properties, fracture properties, water oil contact, horizontal and vertical permeability and length of the fractures were investigated in comparison with the base case and the results were converged to form empirical equations for calculating critical oil rate.
The proposed correlation is dimensionless and can be used for any multi-frac horizontal well and are function of fracture height, permeabilities, fracture conductivities and fluids' densities. All observations of this correlation complies well with the general trends e.g. low anisotropy ratio, lesser fracture conductivity and high viscosity delayed the water breakthrough. Higher anisotropy ratio and low viscosity increased critical oil rate. The results obtained from this correlation when compared with simulated data set had an absolute percentage error of and standard deviation of for critical oil rate correlation, which is quite satisfactory.
Most of the work in literature has been done for water-coning in naturally fractured reservoirs and in some limited aspects of hydraulically-fractured reservoirs, but there is a need to estimate water cresting in MFHWs in order to support early production strategies. This correlation can add value to oil and gas industry in terms of time and money.