This study, conducted at "The University of Texas at Austin" thanks to a joint research project with the Turin Polytechnic, concerns the effect of dilatancy on the stability of keyblocks in rock tunnels. This topic has been addressed from two point of view: - an experimental part to analyzed and validate a modified Barton's model to supplement some weakness points in the original Barton's model; - an analytic part to apply the modified Barton's model to evaluate the stability of keyblocks in rock tunnels. The advantages achieved through the adjustments of the Barton's model are validate from laboratory tests on four different rock types: limestone, sandstone, granite and metamorphosed limestone. The laboratory tests are Point Load Test, Schmidt hummer test, Uniaxial Compression Test, Direct Shear Test on sawed and rough artificial discontinuities. The elaborate continues with numerical simulations, conducted using an algorithm called "BS3D" (Tonon & Asadollahi, 2008) to evaluate the stability of keyblocks of different shapes at the rock tunnels boundary. The final considerations indicate that the modified Barton's model guaranties better approximations of the stress vs horizontal displacement graphs while the vertical displacement vs horizontal displacement graphs are predicted in an approximate way either through the Barton's model and the modified model. The modified Barton's model guaranties the best estimates of the peak shear displacement. At the end the numerical analysis show that the in situ stress is the geomechanical parameter that greater influences on the factor of safety.
