Damage and failure in heterogeneous media is a problem of great complexity and difficulty in mechanics, and a series of fundamental questions still require answering. For instance, the behavior around the catastrophic rupture is essentially different from that of stable accumulation of damage, and it is difficult to be dealt with using the traditional mechanics of homogeneous continuous media. Acoustic emission is an important way to study the damage evolution in solid materials due to its capability of getting real-time information of the internal damages in materials continuously. Furthermore, larger size samples can improve the relative precision of acoustic emission location. Hence the fracture experiment of large rock samples with acoustic emission has been an important measurement to study the seismogenic process.
Supported by the Key Program of the National Natural Science Fund, the research group of LNM, Institute of Mechanics, CAS led by Prof. Yin Xiangchu carried out the experiment of large-scaled rock samples in May and June 2001.
    The experiment was part of the ACES (APEC Cooperation for Earthquake Simulation) project and more than twenty researchers from LNM, Ioffe Physical Technique Institute of Russian Academy of Sciences, Interunis Ltd of Moscow, Ibaraki University of Japan, Center for Analysis and Prediction of the Chinese Seismological Bureau, Peking University and Yunnan Seismic Engineering Institute were involved.
    The size of tested samples is of 105×40×20 cm³, made from gneiss, marble and sandstone respectively. Totally nine samples were tested. The acoustic emissions were recorded continuously during the whole experimental process, thus the obtained data of damage evolution were fairly abundant. The number of recorded acoustic emission events (every one corresponding to a damage) was more than one million, providing an important foundation for further theoretical study, numerical simulation and the improvement of time, location and magnitude precision of earthquake prediction. In order to simulate the seismogenic process more actually, the samples were loaded in two directions to create a tri-axial compression stress state. In addition, cycling stress was superposed to the monotonic load in order to simulate both tidal force and tectonic loading. Such a large sample acoustic emission experiment with tri-axial stress state is the first one in the world. The experiment results illustrate that the evolution of the system behaves great complexity, e.g. the sample-specificity, the critical sensitivity and the increasing load/unload response ratio, etc., prior to the catastrophic rupture. The violation of modulating relations between the acoustic emission signals and loading signals were found during the approaching to collapse for all samples, which may have the same mechanism with the phenomena that materials are less sensible to the external load in self-drive stage and LURR decreases when an earthquake is impending. This finding may arouse new ideas of the short period-forecast of earthquakes.
    The experimental data are still in process. Some rudimentary results aroused great interests and attention during the presentations in the ACES workshop in USA and the symposium at ACES headquarter, Australia.

    The photo shows researchers from China, Russia and Japan discussing ardently by the experimental equipment.