Assessment of Post-blast Damage Zones in Tunneling Operations Through MWD and Machine Learning

The continuous automatization and digitalization of the different stages involved in any underground construction bring new possibilities to optimize the operation. One of the most important requirements for the development of these projects consist of a proper rock mass characterization for optimizing the stages of the drilling, blasting and support cycle. In the last few years the road to digitalization has gone through data science and big data, showing the importance of data gathered to describe the geological and geotechnical environment that surrounds the operations. Since rock characteristics have an important influence in the drilling response, the technological advances in the sensorization of drilling machinery such as the Measurement While Drilling system (MWD) allow to gather the necessary amount of information to be applied in data science processes to evaluate changes in the rock mass with higher resolution than conventional methods. In this paper, Machine Learning techniques have been used to combine data monitored from the drill rigwith scanner profiles of the excavated sections to develop an automatic geotechnical rock mass characterization model able to estimate potential zones of over- or under-break of the remaining rock mass in underground blasting. By comparison of scanner profiles of the excavated sections with the blasthole positions, an Excavated Mean Distance (EMD) between the contour blastholes and the excavated profile has been obtained, which may be considered as damage measure. Considering blasting factors as constant, the model combines the rotary, percussive, hydraulic flushing and the rate of advance of the drilling, and the confinement of the rock mass by depth, to detect zones with different rock properties. This provide an early information of the rock mass condition around the blastholes and the prediction of expected overbreak generated after blasting, being possible to design more efficient bastings, in which each borehole is filled with the right and necessary explosive charge. The analysis comprises data from 57 blasts (around 1700 contour blastholes) and more than 4000 excavated sections.