Abstract:
The theory of soil arching can aid a study into the field of sinkhole development to enable researchers to understand possible mechanisms that may mobilise in the overburden material when existing underground cavities propagate to the surface to ultimately manifest as sinkholes. An understanding of these failure mechanisms could lead to improved estimations of the likely sinkhole diameter which is required for the design of infrastructure in sinkhole-prone environments. Current methods for the estimation of sinkhole size are very conservative, leading to an over-prediction of sinkhole size, often rendering sinkhole-prone land too costly to develop. A need exists for improved guidelines to assess probable sinkhole size which should eventually culminate in less stringent building regulations in sinkhole-prone environments.
Preliminary studies indicated that cavities propagate upwards in a near-vertical fashion, raising questions about the conical funnel-shape as suggested by current building regulation for dolomitic areas in South Africa. This prompted further investigation. Plane-strain deep trapdoor experiments were performed using two different grades of silica sand (a fine and coarse sand) in which active displacement of a trapdoor underneath the soil was modelled to simulate progressive failure of cavity walls and roof. These experiments were performed under dry and moist conditions with varying trapdoor widths and were carried out in a geotechnical centrifuge. Photographs of the models were analysed using Particle Image Velocimetry (PIV) to produce plots of displacement and strain that indicated the geometry of failure mechanisms in the overburden material as the trapdoor displacement increased. Surface settlements were also measured during the experiments using a combination of LVDT surface readings and PIV analyses. These results were compared to form an understanding of the influence of the said variables on failure mechanisms and surface settlement.
Zones of influence above trapdoors in all tests tended to propagate vertically upwards rather than in a funnel shape. Surface settlement initially tended to follow a Gaussian shape, but rapidly deepened once the influence zone above the trapdoor (failure mechanism) reached the soil surface so that the Gaussian shape was no longer accurate. The trapdoor size tended to have very little effect on the general failure mechanism, but the propagation of the zone of influence above the trapdoor did advance more rapidly towards the surface when considering surface settlement versus normalised trapdoor settlement. The spatial frequency of shear zone formation in the sand was found to be related to the trapdoor width, with narrow trapdoors resulting in a denser shear band spacing. Increased trapdoor widths resulted in more symmetric shear zone formation.