Predicting the deformation of compacted loess used for land creation based on the field monitoring with fiber-optic technology

Abstract

Deformation Prediction of compacted loess is important for the constructions on the land created with such soil. In this study, in-situ fiber-optic monitoring was employed through a 77 m-deep borehole to obtain the variations of deformation and water content in both compacted and natural loess. The monitoring results were analysed together with the sampled soil properties. Initial field studies indicated that natural loess experienced much smaller deformation than compacted loess because of its cementation and different structure. Further analysis showed that the variations of stress and water content increment are the most significant factors in compacted loess, exhibiting a logarithmic relationship with loess deformation. Dry density is also important but needs to be considered in conjunction with stress in the vertical direction, while initial volumetric water content and clay-size fraction exhibit insignificant effects on the time-dependent deformation of compacted loess. Based on the analysis of these factors, a time-dependent model was developed, allowing the compacted loess deformation to be estimated. The proposed model was further verified by a numerical analysis using the artificial neural network (ANN) method, showing the relevance of the factors identified from the field data analysis and adopted in the proposed model.