1 INTRODUCTION
Experiments involving the concurrent use of active and passive seismic techniques have been performed to help study the processes leading to failure as rock samples are loaded by various techniques. This paper describes the preliminary results of experiments combining the techniques of acoustic emission analyses and ultrasonic tomographic imaging applied to Brazilian disk tests. Acoustic emissions (AE) have been studied in non-destructive evaluation applications and at a larger scale in geoseismic and earthquake studies (Hardy, 1981). It has become evident that full waveform recording and analyses, using an array of sensors, are needed if meaningful information other than the emission rate is to be obtained. The objective of the analyses, performed in this experiment, is to study the changing mechanisms, source characteristics, and temporal and spatial distribution of AE events with respect to the loading history of rocks. Because each AE event is associated with some degree of deformation, studies of the type and amount of deformation associated with each event are expected to yield information on the deformation process as a whole. Simultaneous digital full wave-form recording with a bandwidth of 10 MHz has been realized on four channels. Four channels are the minimum required for accurate source location, although this number is insufficient for detailed study of source mechanisms. Future experiments are planned using additional channels. The sampling frequency of 10 MHz is sufficient for the detection of acoustic emissions in rocks such as granite, which rapidly attenuate most of the energy at frequencies above 1 MHz. Locations of acoustic emission sources can be determined by the first motion arrival times of seismic waves recorded on multiple sensors. Possible source mechanisms and other source parameters can be determined by examining various characteristics of full waveform AE data. Acoustic, or ultrasonic tomographic imaging is a method of deter- mining the velocity distribution over a plane within the sample. In this study an algebraic reconstruction tomography (ART) algorithm was used (Censor, 1983). Ultrasonic pulses are passed through an object via many transmitter-receiver pairs of transducers. The travel time information thus obtained, gives us the average velocity field along the trajectory. Furthermore, the first arrival rise time can be used to estimate an average quality factor (Q) along the trajectory. By taking multiple trajectories spatially varied throughout the object, one can reconstruct the local variations in the velocity fields and Q values. Because seismic velocity is related to the state of stress within a rock, the velocity distribution gives a reflection of the stress fields. To date, inversions of rise time data have not been used to estimate the variation in Q within the samples. The transducers used for the tomographic imaging are resonant, and thus distort the shape of the wave-form. Using transducers with a flatter frequency response will overcome this problem. It is anticipated that these Q determinations may give a better representation of the changing stress and fracture conditions in the rocks than the velocity images.
2 THE EXPERIMENT
Samples of granite from the Atomic Energy of Canada underground testing facility at Pinawa, Manitoba have been used for this study.
