Reliable Estimation of Horizontal Stress Magnitudes from Borehole Breakout Data

Abstract

As mining depths increase to meet the global demand for minerals and conditions become more arduous at many underground mines, understanding of the state of in-situ stresses will be increasingly important to ensure the workplace safety. However, current stress measurement techniques experience difficulties in obtaining reliable results at low costs. Borehole breakout is a drilling-induced phenomenon due to the local stress concentration and its geometries are dependent on in-situ stresses. In Australia, breakout data is freely accessible as geological and geophysical logging is mandatory. Therefore, developing a horizontal stress estimation technique using borehole breakout is not only pivotal for the safety of underground operations, but also economically beneficial to the mining and petroleum industries. Breakout tests conducted in this study revealed that breakout geometries are influenced by three principal stresses, indicating the intermediate stress should be considered in horizontal stress magnitude estimation. The investigation of laboratory data and analytical solutions suggested breakout geometries are not dependent factors, so that estimation of both horizontal stress magnitudes is viable. Results also showed that the borehole size has significant influence on breakout geometries and an investigation was carried out using normal compression tests. An empirical relationship was proposed to predict the breakout initiation stress at various borehole sizes. Numerical simulation using Particle Flow Code was conducted to study breakout development and borehole size effect. The modelling results revealed that breakout angular span forms at the early stage of breakout, followed by substantial breakout elongation. The borehole size simulation indicated breakout initiated at the borehole wall with stress re-distribution from micro-cracking. The study also showed that breakout initiation stress decreases with increasing temperature, and larger breakouts were observed under high stress conditions when the temperature was over 300 ˚C. An estimation technique for horizontal stress magnitudes was proposed based on an Artificial Neural Network model and Mogi–Coulomb failure criterion. Using 26 field data, the technique estimated the minimum and maximum horizontal stress magnitudes at average error rates of 15.05% and 7.62%. Considering its reliability, simplicity and cost, this model is valuable for in-situ stress evaluation and can help improve safety in underground operations.