A promising new approach has been developed by which changes in rock stress can be measured directly. The measurement of stress change depends on the reaction of a slender cell to the transient variations of rock stress about the cell. The change in pressure in the cell becomes equivalent to the change in rock stress in the direction perpendicular to the cell if the length-to-thickness ratio is very large and the compressibility contrast between the liquid and rock is very small. A stress meter, satisfying these conditions, has been operational for two years in a granitic stock near Salt Lake City, Utah. The performance of the buried device is adequate to detect earthtide signals of the order of 103 N/m2 and stress anomalies premonitory to rock bursts and earthquakes.
Observations of time-dependent anomalies in a variety of crustal rock properties (i.e., Vp/Vs, resistivity, seismicity, radon emissions) prior to earthquakes and rock bursts have led to the development of two categories of predictive models (Mjachkin, et al., 1975; Press, 1975). Although both models assume dilatant crack growth under sustained shear stress to govern the onset of these anomalies, they differ significantly in their assumptions about the stress histories during the second half of the anomalous periods. The dilatancy- diffusion model (Nur, 1972; Scholz, et al., 1973; Anderson and Whircomb, 1973), for example, assumes that the shear stresses in the vicinity of the focal region monotonically increase until rupture occurs. To cause the disappearance of the anomalies diffusion of water into the dilatant volume is required. The second class of models (Mjachkin, et al., 1975; Stuart, 1974; Brady, 1975; Dieterich, 1975), on the other hand, do not require the diffusion of water in and out of the focal region. Rather, these models assume that, following a period of crack growth and coalescence, the shear stresses in the focal region decrease causing cracks to close and anomalies to vanish. To gain the necessary insight into the stress histories that preceed sudden rupture of crustal rocks, there is no substitute for using devices that are capable of monitoring transient stresses for extended periods of time. A large number of these so-called stress meters have evolved over the years for this and other purposes (Panek and Stock, 1964; Fairhurst, 1968). In this paper an improved variation of the "flat-slot" method of stress measurement will be introduced. The description of the device begins with a brief outline of the operating principle. This is followed by the presentation and discussion of field data and their usefulness in the prediction of earthquakes and rock bursts.
PRINCIPLE OF OPERATION
The stress meter, in essence, consists of a very slender, liquid-filled cell firmly grouted in a narrow or flat slot cut into rock. As the stresses in the surrounding rock change, the cell deforms, causing the volume and the pressure in the cell to change. The ratio of change in internal pressure to change in external stress normal to the cell approaches unity only if two conditions are satisfied (Walsh, 1972; Barker, 1975): 1) the length-to-thickness ratio of the cell is very much larger than unity and 2) the compressibility of the contained liquid approaches the compressibility of the host rock.