The dilatant behaviour of sand–pile interface subjected to loading and stress relief

Abstract

Property and behaviour of sand–pile interface are crucial to shaft resistance of piles. Dilation or contraction of the interface soil induces change in normal stress, which in turn influences the shear stress mobilised at the interface. Although previous studies have demonstrated this mechanism by laboratory tests and numerical simulations, the interface responses are not analysed systematically in terms of soil state (i.e. density and stress level). The objective of this study is to understand and quantify any increase in normal stress of different pile–soil interfaces when they are subjected to loading and stress relief. Distinct element modelling was carried out. Input parameters and modelling procedure were verified by experimental data from laboratory element tests. Parametric simulations of shearbox tests were conducted under the constant normal stiffness, constant normal load and constant volume boundary conditions. Key parameters including initial normal stress ($\sigma_{{{\text{n}}0}}^{\prime }$), initial void ratio (e ₀), normal stiffness constraining the interface and loading–unloading stress history were investigated. It is shown that mobilised stress ratio ($\tau /\sigma_{\text{n}}^{\prime }$) and normal stress increment ($\Updelta \sigma_{\text{n}}^{\prime }$) on a given interface are governed by $\sigma_{{{\text{n}}0}}^{\prime }$ and e ₀. An increase in $\sigma_{{{\text{n}}0}}^{\prime }$ from 100 to 400 kPa leads to a 30 % reduction in $\Updelta \sigma_{\text{n}}^{\prime }$. An increase in e ₀ from 0.18 to 0.30 reduces $\Updelta \sigma_{\text{n}}^{\prime }$ by more than 90 %, and therefore, shaft resistance is much lower for piles in loose sands. A unique relationship between $\Updelta \sigma_{\text{n}}^{\prime }$ and normal stiffness is established for different soil states. It can be applied to assess the shaft resistance of piles in soils with different densities and subjected to loading and stress relief. Fairly good agreement is obtained between the calculated shaft resistance based on the proposed relationship and the measured results in centrifuge model tests.

Appendix: verification of DEM model parameters based on laboratory shearbox tests

Laboratory shearbox tests were performed to verify the input parameters and modelling procedure used in the DEM study. Computed results were compared with measurements from these tests under the CNL boundary conditions. Each laboratory test was performed on Toyoura sand sample, 70 mm diameter and 40 mm thick. Relative density of the sample was 65 %. Constant normal stresses of 100 and 400 kPa were adopted.

Figure 12a compares measured and computed relationships of stress ratio and shear strain. The computed peak stress ratio under normal stress of 100 kPa is 1.0 at a shear strain of about 4 %. Although the peak stress ratio is higher than that measured from laboratory test, the computed results capture the general trend of the experimental data. Considering the simplified model used in the DEM analysis, the computed and measured stress ratios agree fairly well. Figure 12b shows the comparisons of normal displacement from laboratory tests and the DEM study. The consistency between computed and measured results in both figures suggests that the input parameters and modelling procedure of DEM study adopted are reasonable.

Fig. 12

Comparison between computed results from DEM study and measurements from laboratory element tests under CNL condition: a mobilised stress ratio and b normal displacement