Abstract: Moment tensor inversion theory is an effective method to study the rock failure mechanism. However, the inversion results are prone to large errors, which can mislead the understanding of fracture mechanism. In order to achieve a better understanding of the generation process and the mechanism of the rock macro-fracture surface, we perform a location analysis of source events based on the uniaxial compression test of a granite sample, with the help of ultrasonic testing and acoustic emission monitoring. The events near the macro-fracture with small location errors are selected for moment tensor inversion. Then, we use the network calibration method to calibrate the sensors so that more accurate moment tensors can be obtained. The results show that the source event locations are in good agreement with the locations of the specimen's macro-fracture. After the sensors having been calibrated, there are a few noticeable observations. The inversion root-mean-square (RMS) errors of moment tensors reduce significantly. The distributions of events on the T-k plot and P/T axis plot become more concentrated. The shear component and the proportion of different types of events change accordingly. The distributions of strike, dip and rake angles of the events become more concentrated, which are in general consistent with the macro-fracture of the specimen. The tensile angles of some events change from negative to positive. The above results reasonably explain the failure process and mechanism of the specimen and highlight the importance of sensor calibration for the moment tensor inversion, which can be a useful tool to provide guidance and reference for a deeper understanding of rock failure mechanism.

Key words: moment tensor inversion, failure mechanism, uniaxial compression, calibration, location