Enhancing accuracy and precision of transparent synthetic soil modelling

Over recent years non-intrusive modelling techniques have been developed to investigate soil-structure interaction problems of increasingly complex geometry. This paper concerns the development of a small-scale, 1 g, modelling technique using a transparent analogue for soil with particle image velocimetry for internal displacement measurement. Larger model geometry achieved in this research using fine-grained transparent synthetic soils has led to an increased need for rigorous photogrammetric correction techniques. A correction framework, based upon a modified version of the pinhole camera model, is presented that corrects for lens and camera movement induced errors as well as scaling from image space to object space. An additional statistical approach is also developed to enhance the system precision, by minimising the impact of increased non-coplanarity between the photogrammetry control plane and the target plane. The enhanced data correction and statistical precision is demonstrated using a case study examining the failure mechanism around a double helical screw pile installed in transparent synthetic soil representative of a soft clay