Prediction of optimum injection rate for carbonate acidizing using machine learning

In stimulation, acid is pumped into the formation at a certain rate to penetrate deep into the formation. The optimum injection rate is one of the of the key parameters that is sought after. By definition, the optimum injection rate happens when the acid penetrates deep with the minimum volume. In a linear coreflood experiment, this translates into a dominant wormhole dissolution pattern. Outside the range of such rate, the pattern could be face dissolution, conical wormhole, ramified wormhole, or uniform dissolution. Researchers have been trying to develop models to predict the optimum injection rate. However, such models require some reaction kinetic parameters to obtain the optimal injection rate. This paper aims to investigate the dependence of different parameters on the optimum injection rate for matrix stimulation. This study incorporates a wide range of data points obtained from linear coreflood and develops a new, robust, easy-to-use model using machine learning. This model predicts the optimal injection rate with the help of three arguments, namely, rock properties, HCl properties, and experimental conditions. The data used in this study are composed of 170 experimental data points. The data were collected from Indiana, Desert Rose, and Lavoux limestone. The results show that the newly developed model is capable of predicting the pore-volume-to-breakthrough (PVBT) along with the optimal injection rate with 90% accuracy. Furthermore, the model only requires the following parameters: type, length, diameter, porosity, and permeability of the core; HCl strength; temperature; and pressure. Such model will narrow the search for the optimal injection rate and will significantly reduce the number of laboratory experiments needed to obtain the PVBT curve. Furthermore, the results from this model can be upscaled to field scale. The novelty of this work lies in providing a simple, yet powerful, tool that is easy to use to predict the optimum injection rate. This will help production engineers to efficiently design a successful acid job.