INTRODUCTION
The Gjovik Olympic Mountain Hall, with a span of 61 m and a minimum cover of only 25 m, is claimed to be the largest underground cavern for public use. The requirement for a high standard of safety, compatible with its use by the public, necessitated careful attention to design and determination of parameters[I].
The stress field was determined by both overcoring and hydraulic fracturing methods. Preliminary measurements of rock stress using overcoting utilizing a single short hole drilled from a nearby existing tunnel indicated surprisingly high horizontal major principal stress of about 4 MPa in fractured and weathered gneiss. The nearness to the surface, limited intact rock and the limited depth of interest required special techniques for the hydraulic fracturing testing and its interpretation.
HYDRAULIC FRACTURE STRESS MEASUREMENTS
The equipment used was a simple wire line straddle packer of 40 mm diameter and an impression packer. Many closely spaced tests were made in the hope of f'mding some intact sections where the breakdown pressure could be measured and the data from tests which appeared to open existing fractures provided additional collaborative evidence using the fracture pressurization method[2,3].
Data was obtained from tests at 65 locations between 4 m and 55 m depth. Peak pressures that were attributed to breakdown were obtained at 23 of these locations. The shut-in pressure was interpreted using the tangent divergence technique[4]. The results are plotted on
Figure 1.
An impression packer and orientation device was used to determine the orientation of fractures at test locations. Impressions were attempted in test sections where the best defined peak pressures were obtained in the first pressurization cycle in the hope of finding hydraulically induced fractures which indicate the principal stress orientations directly. The shallow depth of tests, and therefore expected low levels of in situ stress, indicated that errors due to measurement of the peak pressure and shut-in pressure were likely to be significant. Errors were expected due to the distance between the transducers and the test section and the limited resolution and response time of the chart recorder used to record data. These were minimized by using slow rates of pumping. The size of errors is indicated by the wide range of shut-in pressure measured by performing several tests at one location, typically the measured shut-in pressure varied by 0.5 lVIPa