ABSTRACT: A programme was set up to determine the response of a backfilled stope to seismic activity in a South African gold mine. Ground motion was measured at two triaxial geophones which were attached to the hangingwall of the stope in dip gullies. One of the dip gully geophone sites had backfilling on both sides of the gully and the other had backfilling on only one side. Twenty-two events, each with a magnitude, a location and recorded waveforms were selected from the data base for analysis. The findings were: In general ground motion is damped in the presence of backfill. Backfilled stopes experience higher dominant frequencies than unfilled stopes. With insufficient backfill to generate high stresses in the rock the seismic waveforms are out-of -phase. The zone of low and tensile stresses above the gully is restricted by the placement of backfill. The layout of backfill can cause local amplification of the waveforms. INTRODUCTION An understanding of the dynamic behaviour of the rockmass surrounding a stope during seismic events is important for designing support to control rockburst damage. As rock burst damage is caused by seismically induced ground motion which destroys the stability of the excavation, analysis of the response of mining excavations to non-damaging seismic events can provide invaluable insights into damage mechanisms. Observations of rockburst damage in stopes with filled and unfilled panels suggest that backfilling has a positive effect on maintaining the integrity of the stope (Castelyn 1988). Documentation of rockburst damage observations in backfilled stopes by Gay et al (1988), indicates that the backfill maintains the h.angingwall of the stope and contains damage in the stope faces and gullies. However, observations underground are subjective and limited, since not every part of a stope is visible, especially after severe seismic damage. It is also not always clear what damage is due purely to the rockburst and what is due to lack of cleaning after a blast. In addition, the variation in geology, fracturing and support and the error in event location relative to the stope may make comparisons of rockburst damage difficult. Spottiswoode and Churcher (1988) observed that ground motion in a deep-level conventionally supported stope during seismic events was profoundly different to ground motion measured at points in the rockmass remote from the stope. They predicted that backfill would have the following effect on stopes during seismic events: Reduce the loosely supported spans in the stope. Increase the resonance frequency of the hangingwall of the stope. Reduce absolute vibration levels by increasing the mass of the rock immediately adjacent to the stope. Reduce differential movements between the hanging wall and the footwall and thus reduce deformation of support units. The first three predictions formed the basis for an evaluation of seismic data collected at a backfill site established on a deep South African gold mine. The last prediction was not assessed because geophones were not positioned on the footwall of the stope.

ABSTRACT: The authors have been conducting research using AE for estimating initial stress and predicting failure of rock masses. It is suggested that there exists a time dependence in the Kaiser effect of AE, and in case of estimating initial stress using the Kaiser effect, it is necessary to perform testing as soon as possible after sampling rock in-situ. It was also found that there are great differences between rock types in AE frequency distribution and m-value distributions measured in processes up to failure, and this would be an effective means of predicting failure. 1 INTRODUCTION Acoustic Emission (hereinafter abbreviated to "AE") has been drawing attention in the field of civil engineering in recent years and many kinds of investigations have been made. Hardy (1981) put together a study of AE concerning rock masses. AE, which is called "Microseismic Activity" in the field of mining, is the phenomenon of a part of the energy stored inside a substance being released and propagated taking the form of elastic wave motion. This is a phenomenon which is generally observed in practically all solid materials in cluding metals. ceramics, concrete, glass, and ice. Ordinarily, the objects of investigation in AE research are faint signal waves in the inaudible range that cannot be heard as sound. When constructing underground structures such as tunnels and powerhouses, it would be possible for rational designs to be made if the initial stress of the ground were to be known. Consequently, numerous methods of estimating initial stress were devised in the past. At present, however, only the stress relief method (Merrill et al.. 1961) and the hydraulic fracturing method (Mizuta et al.. 1983) from among those methods are frequently used. A phenomenon called Kaiser effect (Kaiser, 1953) has been confirmed to exist in AE. This is a condition where in case a material to which a load had been applied in the past is reloaded, hardly any AE occurs up to the maximum stress produced by the load applied in the past. Goodman (1963) was the first to distinguish the Kaiser" effect in rock, following which studies concerning the Kaiser effect in rock were made in many Quarters (Yoshikawa et al. 1981). Hayashi et al. (1979) were the first to apply the Kaiser effect in estimating initial stress of a rock mass. The authors cited the various problems involved in estimating initial stresses of rock mass using the Kaiser effect and have carried out work to resolve these problems (1984, 1985, 1986). Meanwhile, since AE occurs accompanying formation of minute cracks and release of energy, it is conceivable to apply it in prediction of failure. Actually, in the field of seismology, the earthquake mechanism has been explained through model experiments on rocks (Mogi, 1962: Scholz, 1968). The authors have been thinking of applying AE to prediction of failure in underground caverns and evaluation of long-term stabilization. 2 EXPERIMENTAL APPARATUS An outline of the AE measurement system used in experiments is shown in Fig. 1.

ABSTRACT: An experimental research program, designed to detect acoustic emissions (AE) produced during propagation of discrete Mode I and Mode II fractures in large rock specimens, is underway at the Rock Mechanics Laboratory at The University of Texas at Austin. Equipment developed, and signal processing techniques used, are described in detail. Preliminary results of testing granite and dolostone rock specimens are presented. Results indicate that very similar signals may be produced by very different rocks, suggesting a similarity of internal mechanism during crack propagation. 1 INTRODUCTION Acoustic emissions (AE) are elastic waves generated in conjunction with energy release during crack propagation and internal deformations in materials. This release of energy is manifested as transient stress waves which propagate from the locus of a structural change associated with material failure and changes in the local stress field. Micro-structural changes or displacements occur very rapidly and can be produced by a wide variety of material responses to stress changes, from small scale changes within a crystal lattice structure to growth of macro-cracks. Each AE event is a signature of an actual mechanism, a discrete event directly reflecting material response. Information about material properties and failure mechanism is contained in each AE waveform. The aim of this experimental study is to learn to "read" the AE signature, to identify characteristics and correlate with mechanisms, assisting in establishing the basis for systematic AE record interpretation. The study of AE has developed into an increasingly popular form of nondestructive testing. The technique has long been employed in pressure vessel testing, and an American Society for Testing and Materials (ASTM) method is widely used in the nuclear industry for this purpose. Metallurgists have successfully used AE to identify discrete molecular level events. AE events have been used, especially in Japan, to model earthquake activity. In the field of rock mechanics, AE methods have been implemented to address problems such as rock burst predictions, hydraulic fracturing, mine pillar stress and deformation, rock mass stability, and the velocity of groundwater movement. Detection and analysis of AE signals are made difficult for several reasons. Following each event, the originally produced signal is filtered as the emitted stress waves propagate through the surrounding material. The filtered stress waves are detected by observing and measuring material response only at an accessible surface. Such surface displacements related to AE events are extremely small, necessitating the use of elaborate electronic transducers and analytical devices. In addition, simple AE events are not instantaneous, and the duration of these events can be from microseconds to many milliseconds depending on the material, the loading, and the nature of the source. Each detected waveform may be very complicated and difficult to analyze, with some of the complexity introduced as energy is released over the duration of the event. In the near-field, wave modes are not easily defined, while in the far-field different wave modes can be discriminated. Local geometry and boundary condition effects will combine to introduce reflections which complicate the latter portions of detected signals.

INTRODUCTION Since 1962, the authors have been carrying out earthquake observations at the ground surface and underground at rocky ground in the mountains of the northern part of the Kanto Region, approximately 150 km north of Tokyo. Several papers have already been published regarding the results of these earthquake observations. There are several thin weathered layers interspersed in the bedrock at the observation site and it was found from observation results that earthquake ground motions were influenced by these layers, and this is discussed in the present paper. OUTLINES OF GEOLOGY AND SEISMOMETERS The geology of the observation site consists of green tuff formed from the Tertiary to Quaternary periods which is relatively homogeneous and with core recovery rate of roughly 100% at the bore hole used in observations. The site is located at a slope of approximately 19 deg. The condition of the ground as obtained from the results of boring is given in Table 1 with clay layers of thicknesses of 30 cm, 20 cm, and 10 cm produced by weathering of the green tuff found at depths of approximately 16 m, 36 m and 47 m, respectively, from the surface. Also, a number of cracks is found at depths of about 50 m and 52 m. There is a vertical shaft for access to an underground powerhouse at a point approximately 14 m distant from the bore hole, About 10 m of the surface layer is comparatively weathered at this point with clay layers at depths of 13.8 m and 14.25 m, while no core was recovered between the depths of 53.4 m and 54.8 m. Based on the above, it is surmised that ground layers run roughly parallel with the surface of the slope. As shown in Fig. 1, eight strain gage-type accelerometers have been installed horizontally in the E-W direction down to a depth of 67.2 m and in a manner to sandwich clay layers. The accelerometer frequency characteristics are more or less flat from 0 to 10 Hz, and the mutual differences between characteristics of accelerometers are slight. Five accelerometers have been installed in the vertical shaft. These are electro-magnetic-type seismometers of natural period of 0.3 sec having integrated frequency characteristics nearly flat in range of 1 to 24 Hz, and are located at intervals of approximately 17 m from the ground surface to a depth of 67.2 m. Measurements are made in the horizontal, E-W direction. Since these two varieties of seismometers have different frequency characteristics, dissimilarities are produced in waveforms at comparatively high frequency components, while there are also differences in absolute values. Recording is done by oscillographs activated by starters and paper speed is 10 cm/sec. Acceleration records obtained at the bore hole are studied in this paper. EARTHQUAKE RECORDS Approximately 340 earthquakes have been recorded up till the present, these earthquakes mainly having occurred in the Kanto Region, Tohoku Region and the neighboring sea area.