Workreport 2016-52



Parametrisation of Fractures – PUSH Test Execution and Back-Analysis


Valli, J., Hakala, M., Suikkanen, J., Mattila, J., Heine, J., Simelius, C.



Page count:



An in situ push test of a fracture surface under constant normal stiffness (CNS) conditions was conducted at a depth of 437 meters in the ONKALO underground research facility. The objective of the experiment was to study the in situ shear strength of fracture surfaces at an easily accessible location, with a simple associated geometry, under low normal force and large shear displacement. The push test was conducted to assess the methodology of conducting an in situ shear test and to gain knowledge of the shear behaviour of fractures at the site.

Loose blocks surrounding the tunnel opening are formed when natural fracture surfaces intersect the excavation surface of the tunnel. The targeted block is limited by a steep, North-South-orientated Riedel-fracture related to the BFZ-300 geological feature in the ONKALO Parking Hall.

This working report describes the experiment procedure of the push test and analysis of the measured results. The displacement and rotation during the push test were monitored by LVDT-sensors, while the normal- and shear forces were measured using load cells and a hydraulic pressure gauge. Fracture parameters were initially estimated according to the Q-classification system along with JRC evaluation. The test response and these selected parameter values were then back analysed with iterative rock mechanics simulations using the three-dimensional distinct element code 3DEC.

Back-analysis using 3DEC was performed through a step-wise approach in order to avoid numerical convergence problems and to study the effect of geometry on results. The three fracture surfaces modelled were a planar fracture surface, a fully-mated coarse surface and an unmated coarse surface. The shear and normal stress measurements from the shear test experiment were compared to those from the different models run with different parameters, based on the originally estimated fracture surface area. The unmated model was found to yield an acceptable match using an initial friction angle of 47° and a residual friction angle of 40°, although underestimating the initial peak strength after 2 - 5 mm shear. This peak was most likely associated with upper side areas of the block broken off during the initial peak. These details were not included in the model geometry. Most importantly a low constant normal stiffness condition was found to increase shear resistance by over 30% from the initial peak to final shear displacement of 34 mm. The back-analysed friction angles are about 10° greater than those observed from small scale laboratory tests. This suggests that the scale effect where longer fractures exhibit a lower shear resistance normally assumed in practice, may not be valid for the repository rock mass under study.

Findings from this study revealed the following: (1) Displacement control should be used to prevent rotation of the block. (2) Anchor bolts should be fully grouted with proper setup using I-beams to eliminate any effect from the bending of the bolts. (3) Normal displacement should be measured in multiple locations. (4) Finally, the test setup should be able to generate high normal stress to mimic constant normal stiffness resulting from confined in situ conditions at 450 m depth.


ONKALO, fracture, in situ, shear test, stiffness, numerical, back analyses.


Parametrisation of Fractures – PUSH Test Execution and Back-Analysis (pdf) (14.2 MB)


Share article: