Modifications to the Paterson triaxial rock deformation apparatus to allow combined stress testing

Almost all rock deformation experiments are performed in pure shear (axial compression or extension) or simple shear (torsion). However, in general, natural deformation can be expected to occur under some combination of these end member loading geometries. One of the most widely used apparatus for deforming geological samples at elevated temperatures and confining pressures is the Paterson triaxial rock deformation apparatus which is now installed in several experimental rock deformation facilities worldwide. In basic design this apparatus has the capacity for deforming samples under simultaneously applied axial loads and torques but modifications are required to the way in which axial load and torque are measured during such experiments if the mechanical data acquired are to be meaningful. Two design complications in particular arise. Firstly, at present axial load and torque are measured by a single slotted elastic element which undergoes measurable (and hence able to be calibrated) elastic distortions in response to applied axial loads and torques. The use of a single element presents no difficulties if either axial loads or torques are applied but when they are applied together the torque leads to an apparent but not real axial load and vice versa. Secondly, in a torsion test it is important to be able to detect the point during twisting at which all the slack within the rig - needed to allow test assembly - has been taken up and initial torque is applied to the sample. In a pure torsion test this is achieved by having a gap between the axial ram (containing the load cell) and the sample assembly but when there is a simultaneously applied axial load this gap is closed and the resulting friction at this surface means that torque is transferred to the load cell from the onset of twisting. In this contribution we show how a low-friction thrust bearing assembly located between the axial ram and sample assembly can be used to provide a relatively easy and inexpensive modification that addresses the second of these difficulties. We then use this modification to illustrate the nature and significance of the first issue and conclude that it presents few problems for torsion experiments performed under a constant axial compressive force. This, in itself, considerably extends the range of loading states accessible to the apparatus. To extend further its capabilities to other loading configurations, such as axial compression under constant torque or some combination of simultaneously imposed constant axial and shear strain rate, almost certainly would require a redesign of the load cell into two separate elastic elements.