INTRODUCTION
Considerable advances have been made in the measurement of in situ stress over the past 25 years; however, the interpretation of these measurements with respect to geological structures has received significant attention only over the past 10 years. The major limitation in the interpretation of stress measurements is the paucity of measurements at any one site. An exception is the Underground Research Laboratory (URL), where about 350 triaxial stress measurements have been made in a volume of rock about 100 m x 100 m x 500 m deep. This is more than twice the number of triaxial stress measurements for all other sites in the Canadian Shield combined. This paper examines the relationship between stress heterogeneity and geological structures at the URL.
IN SITU STRESS NEAR A THRUST FAULT
Excavation of the URL shaft intersected two major thrust faults that dip about 25 to 30 ø southeast (Figure 1). These faults are referred to as Fracture Zone 3 and Fracture Zone 2 and their splays as Fracture Zones 2.5 and 1.9. The fracture zones divide the rock mass into distinct stress domains [1]. The rock mass from the surface to Fracture Zone 2.5 contains two or more regular joint sets, which decrease in persistence with depth. Below Fracture Zone 2 the rock is massive with no jointing. The maximum horizontal stress orientation versus depth indicated two distinct stress domains at the URL [1]: one from the surface to Fracture Zone 2, where the maximum horizontal stress is oriented parallel to the major subvertical joint set, striking about 040 ø (Figure 1); and the second extending below Fracture Zone 2, where the maximum horizontal stress has rotated about 90 ø and is aligned with the dip direction (? 130 ø) of Fracture Zone 2 (Figure 1).
A numerical model was used to illustrate the mechanism of stress rotation across the fracture zone. At the URL, the horizontal stress at depth is in the dip direction of Fracture Zone 2 (see Figure 1) and the ratio between the maximum and minimum horizontal stress is about 1.2. This basic information was input into a UDEC numerical model as shown in Figure 2. The model was compressed in the horizontal direction and the Fracture Zones allowed to slip. The resulting horizontal stresses versus depth are shown in Figure 2. As is the case at the URL (see Figure 1), the maximum horizontal stress direction rotates from the dip direction below the fracture zone to the strike direction above the fracture zone. This type of stress release and associated stress rotation is commonly observed in modelling with constant-displacement boundary conditions, where the block of rock above the fault has lost its original load because of displacements above the fault. Two other points are also worth noting from the results of this simple model. First, the magnitudes of the maximum stress above Fracture Zone 2 are considerably less than the stress magnitudes below Fracture Zone 2. In the UDEC model no allowance is made for vertical fracturing in the proximity of Fracture Zones 2.5 and 3, which would tend to reduce the horizontal stresses even more and bring the model stresses closer to those measured (Figure 2). Second, the stress maguitudes below Fracture Zone 2 are fairly constant with depth (the model had a depth of 1.5 km). Although stress magnitudes at the URL have only been measured to a depth of 512 m, measurements and construction observations suggest that the maximum horizontal stress from Fracture Zone 2 to 512 In is 5drly constant at about 55 MPa (Figure 2).
The vertical stress magnitudes sampled from and around the shaft at the URL were compiled from triaxial overcore results, from hydraulic fracturing conducted in horizontal boreholes and from subhorizontal hydraulic frac