Effects of anisotropy, strain softening and shearing rate in numerical simulations of T-bar penetration in Onsøy clay

Abstract

The material behavior of clay includes strength anisotropy, where the active undrained shear strength is larger than the direct simple shear strength and the passive strength. Clay also experience strain softening, which is a reduction of the strength with strain after the peak strength is reached. Clay is a rate dependent material, and an increase of the strain rate increases the peak strength.

T‐bar is a full flow penetrometer which is used in‐situ to determine the undrained shear strength of clay. A horizontal cylinder is pressed vertically into the soil, and the resistance of the plastic flow around the T‐bar is measured as the penetration resistance. The bearing capacity factor NT‐bar is expressed as the penetration resistance divided by the average undrained shear strength. Empirically, the range of NT‐barfactor is 10 to 13, and with T‐bar penetration tests the undrained shear strength of a clay deposit can be determined. There have been done T‐bar field tests at Onsøy, Norway, and high quality block samples have been taken from the site and used in laboratory tests.

The T‐bar penetration resistance is affected by different aspects of clay material behavior. Numerical simulations have been done to investigate the effects of anisotropy, strain softening and rate dependency on the NT‐bar factor. Input parameters to the simulations were found from back‐calculating results from triaxial, DSS and CRS tests.

The T‐bar penetration was modeled as plane strain in Plaxis, using 560 15‐noded elements. The updated mesh setting was used for large deformation analyses. During penetration a continuous circular failure mode developed around the T‐bar.

The material models used were the total stress based NGI‐ADPsoft and the viscoplastic n‐SAC. While the former has the over non‐local strain method implemented to control the shear band thickness, the latter has no regularization technique to prevent localization. In the n‐SAC simulations with strain softening the strain concentrated in thin shear bands, and a continuous failure mode failed to develop.

In the simulations, anisotropy of the peak undrained shear strength had no significant effect on the NT‐barfactor. The NT‐bar factor was however greatly affected by strain softening. The residual strength was important for the penetration resistance. Simulations with ADP where the sensitivity was 3 and 5, had resistances which were 39 % and 25 % respectively of the resistance without softening.

It was found that for clay modeled as rate independent, the NT‐bar factor should be increased as much as its peak strength increases in a triaxial test with a 106 higher strain rate. Based on the simulations, the shearing rate during T‐bar penetration corresponds on average to a rate six orders of magnitude higher than in standard triaxal test.

It is believed that the rate and softening effects partly compensate for each other. The former increases the penetration resistance while the latter decreases it. Because the degree of softening was modeled unrealistically high, the values of the NT‐bar factors of the simulations were lower than 11.0 from the field investigations.