Predicting the size of stick-slip events, an in-depth look into the dynamics of a spring slider system

Abstract

Spring slider experiments are performed with a varying normal load, spring stiffness, load point velocity and grain size to determine the influence of these parameters on the displacement, force drop, maximum velocity and the reoccurrence time of a slip event and to validate the numerical model QDYN (Quasi DYNamic earthquake cycle simulator). The friction volume used, consist of spherical glass beads with a uniform grain size to create reproducible slip events (i.e. events with similar stress drops and displacements). The experimental results are used to validate QDYN, a numerical model based on rate-and-state friction laws. The rate-and-state friction parameters used by QDYN are fitted to the experimental data and the effect of changing these parameters on the displacement, force drop and maximum velocity is investigated. Using spherical glass beads as a friction volume the following trends where observed: An increase in the normal load results in a linear increase of the displacement, force drop and maximum velocity; An increase of the spring stiffness results in a negative power law relation with the measured variables and an increase of the load point velocity results in the increase of the measured variables according to a negative logarithmic law. It is shown that Dc cannot be used as a substitute for the grain size in QDYN. The spherical glass beads were substituted with angular quartz sand to determine what the effect of a different material is on rate-and-state parameters. The results show similar trends with the glass beads experiment. To obtain these results the RSF parameters used for the glass beads had to be adjusted: (a-b) increased, Dc decreased. We observed dependencies of dx, dF and Vmax on (a-b) and Dc using QDYN. Due to the quasi-dynamic nature of QDYN, the model is not capable of a precise prediction of the maximum size of an earthquake event, but can provide an approximation. The results of this study show similar relations to events found in nature and it is perhaps possible to extract RSF parameters of these events using large-scale simulation models such as QDYN.