Application of extreme gradient boosting method for evaluating the properties of episodic failure of borehole breakout

Abstract

In petroleum engineering, determining in situ stresses and breakout dimensions are crucial. Nowadays, hydraulic fracturing is a frequently used method for identifying in stresses and breakout dimensions. Regrettably, the hydraulic fracturing procedure is costly and time-consuming. In petroleum engineering, it is very appealing to use the in-situ stresses and properties of the stone to calculate the breakout dimensions (rB/a, and θB). However, since the rB/a, and θB are so complicated prediction is challenging. In this case, impactful algorithms were established utilizing the hybridized extreme gradient boosting (XGB) strategy on collected data in order to accurately anticipate the rB/a, and θB. In addition, a number of recently developed optimization algorithms, including Fire Hawk Optimization (FHO), Artificial Hummingbird Algorithm (AHA), Flow Direction Algorithm (FDA), and COOT Optimization Algorithm (COA), were used to tune the XGB's hyperparameters. By analyzing XGB-based developed models considering various aspects of analysis, the suggested FHO−XGB could be identified as the superior model in the prediction of rB/a, and θB, which can be utilized in order to reduce the hydraulic fracture attempts.