Effects of soil strength on propagation mechanisms above deep trapdoors

Abstract

Approximately a quarter of South Africa’s Gauteng Province is underlain by dolomite, a carbonate rock which is susceptible to dissolution and therefore sinkhole formation. Current design codes for developing on dolomitic land are known to be conservative, resulting in areas of land which are uninhabitable due to the often-unfeasible construction costs associated with large sinkhole sizes. The Gauteng Province is the heart of South Africa’s economy and, as such, there is a need to investigate and understand the propagation mechanisms associated with sinkhole formation, as well as the factors that govern these mechanisms as cavities propagate from depth to the ground surface. A better understanding of cavity propagation mechanisms will enable design standards to be improved to optimise the utilisation of dolomitic land. Trapdoor experiments have been used extensively in the past to investigate soil arching and the associated material behaviour. More recently, cavity propagation associated with sinkhole formation has been investigated by means of deep trapdoor experiments. With the advent of Graphical Processing Unit (GPU)-based Discrete Element Method (DEM) codes, the DEM numerical method has become a popular tool for investigating the rheology of granular materials. This research study therefore focused on utilising physical deep trapdoor experiments, complemented by DEM, to investigate the relationship between the shear strength of a material and the soil deformation mechanisms during cavity propagation. Deep trapdoor experiments were undertaken at the University of Pretoria geotechnical centrifuge facility with three materials, namely a fine-grained sand, a coarse-grained sand and 2 mm glass beads. These tests were conducted to observe the different deformation mechanisms, at field stress conditions, associated with materials of distinctly different shear strengths. The trapdoor experiment with glass beads was further used to calibrate a numerical DEM model with spherical particles. The calibration procedure of the DEM trapdoor experiment included a sensitivity analysis of the numerical model input parameters, as well as validation by comparing displacement contours, particle displacement trajectories and maximum shear strain plots. Once satisfactorily calibrated, the model was used to investigate the effect of material strength, as a result of particle shape, on the material deformation mechanisms during trapdoor lowering, simulating cavity propagation. Particle shape was initially simulated by means of polyhedral particles. These simulations did not run successfully as the required time step at the calibrated particle stiffness was not feasible within the single precision DEM code for polyhedral particles. Furthermore, a reduction in the stiffness of the particles, yielding a greater required time step, was not suited to the high stresses within the material model. Varying rolling resistance and friction coefficients were therefore applied to the particles to simulate the effect of particle shape. Based on the maximum shear strain results from both the physical (centrifuge) and numerical (DEM) trapdoor results, it was found that an elongated ellipsoidal deformation mechanism governs behaviour for materials with a low shear strength. Whereas for materials with a comparatively high shear strength cavity propagation occurs between two vertical shear bands. The DEM models demonstrated that the greater the shear strength of the overburden material, the greater the tendency of the shear bands and cavity propagation towards verticality. It was further observed that for both sets of tests, the deformation mechanisms widen at the surface. The lower the material shear strength, the wider the zone of influence at the surface.