Cone penetration test in a virtual calibration chamber

Cone penetration test (CPT) is a fast and reliable site investigation tool for exploring soils and soft ground. While the interpretation of the test results in clay has advanced considerably from a theoretical and numerical viewpoint that of tests in sands still relies largely on empirical correlations. A major source of such correlations comes from tests done in calibration chambers (CC), where soil state and properties might be tightly controlled. Calibration chambers are relatively large pieces of equipment, and calibration chamber testing is expensive and time consuming. Moreover, CC tests are performed on freshly reconstituted sands whose fabric may vary from that of natural sand deposits. Hence, correlations developed for one type of sand might not be suitable for another sand deposit. Numerical DEM-based calibration chambers might offer an interesting alternative to the more cumbersome physical tests. This study is the first attempt to perform a three-dimensional DEM-based simulation of cone penetration test. The three-dimensional commercial DEM code (PFC3D) is used to develop Virtual Calibration Chamber CPT (VCC CPT) model. To achieve that objective, several steps were necessary. First, calibration of an analogue discrete material to represent Ticino sand was performed using single-element tests. Afterwards, the mechanical response of the discrete material was further validated by performing additional triaxial tests with different initial conditions. The VCC CPT model was then constructed. Comprehensive dimensional analysis showed that the best option to balance computational efficiency and realism was to fill the chamber with a scaled-up calibrated discrete material. An original filtering technique was proposed to extract steady state cone resistances. A basic series of simulations was performed to explore the effect of initial stress and relative density in cone resistance. The results obtained from the simulations did fit closely the trends that had been previously established using physical chambers. That result was taken as a general validation of the proposed simulation approach. From the micromechanical point of view, the granular material is highly discontinuous and inhomogeneous. Obtaining a homogeneous initial state (especially in the zone of the penetrating cone) is crucial to obtain easily interpretable results. Specific procedures to assess initial state inhomogeneities were developed. DEM-based models can provide results at various level of resolution i.e. the microscale, the meso-scale and the macro-scale. A large series of VCC CPT has been performed. Simulations were performed for models with different horizontal servo-control walls, various sizes of chamber, cone and particles and two boundary conditions. The results were analyzed, focusing on aspects such as chamber size, particle size and boundary condition effects on steady state cone resistance values. A smaller number of tests have also been examined from the point of view of shaft resistance. Most trends and results obtained are shown to be in agreement with previous physical tests. When disagreements appear, the causes are identified: the most severe disagreements result from initial inhomogeneities in the discrete model. The work described in this thesis showed ease the burden of future CPT calibrations in granular materials.