Fibre optic strain monitoring in spatially variable deep mixed soil

The reliability of fibre optic strain monitoring depends on an efficient strain transfer from soil to sensor. Previous studies on this soil–sensor strains transfer have relied on shear lag theory implicitly assuming homogeneous soils, while strain transfer effects in heterogeneous soils have not yet been studied. In this paper, the implications of spatial variability in deep mixed columns on strain measurement reliability are investigated. Governing strain transfer equations are analysed by global variance-based sensitivity analysis and synthetic random fields, and it is found that interface stiffness and interface strength are the most important soil–sensor interface parameters. A novel, near-element-scale interface testing procedure is developed to accurately determine these parameters. The findings suggest that boundary effects are negligible for field conditions, but that the heterogeneity in the deep mixed soil results in a requirement for accurate determination of the interface stiffness. Furthermore, it is shown that pre-peak interface behaviour is likely to be achieved by proper sensor selection. A generic framework to analyse vertical strain monitoring data is introduced and exemplified with field monitoring data. While the paper focuses on DM ground improvement, the framework is generally applicable to other spatially variable soil conditions.