ABSTRACT
The need to solve complex phenomena that still do not have a closed form analytical solution is one of the challenges of current practice, including the susceptibility estimation of cavity blowout during drilling of horizontal cavities. Computational tools and high frequency data acquisition in conjunction with machine learning open up new opportunities to create tools that can aid in the understanding and design of these problems. We propose the use of a predictor function, calibrated with a pool of numerical simulations that can predict the susceptibility of blowout of a cavity, given a fixed geometrical and stress configuration based on the mechanical parameters of the soil the cavity is embedded in; results using k-means clustering (for classification) and support vector machines (SVM) to create the predictor function, showed and accuracy of about 87% in predicting the blowout susceptibility.
