A microstructural insight on a lime-stabilized loess: experimental evidences and pore size distribution modeling

Lime stabilization is a traditional method for improving foundation soils, and it also has potential applications for embankments and earth structures. In this study, several experimental techniques, including scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), and nuclear magnetic resonance (NMR), were used to provide a clear picture of the microstructural evolution of a lime-stabilized loess (LSL) from China. SEM micrographs were used not only to qualitatively highlight the dual porosity nature of the material, but also to provide quantitative information using Image-Pro Plus (IPP) 6.0 software. As the lime content increases, the pore area ratio decreases, the shape of the macropores and mesopores flattens, and the pore angle distribution becomes more uniform. The FTIR results show that the functional group strength of the LSL samples first increases and then decreases with increase in lime content, while the pore volume continues to decrease. A non-monotonic evolution of the strength with the lime content is then expected, as also confirmed by unconfined compression tests performed at different lime contents and curing times: at low lime contents, the reduction of the pore volume and the increase in the functional group strength imply an increase in the strength; at high lime contents, the competing effects of the reduction of the pore volume and the increase in the functional group strength lead to an overall decrease in the strength with the lime content. Then, as an intermediate step toward further quantitative predictions of the hydromechanical behavior of LSL, a pore size distribution model inspired by the proposal of Della Vecchia et al. (Int J Numer Anal Meth Geomech 39:702–723, 2015) was developed and used to reproduce NMR experimental data. The pore size distribution model proved to be able to reproduce the cumulative porosity curves for the whole range of lime content and curing time studied, with only four parameters kept constant for all the simulations. The predictive capabilities of the model were also confirmed by simulating experimental data from recent literature.