UC Davis

The emergence of the novel bio-cementation techniques for liquefaction mitigation application has received more attention on the experimental level than the numerical level. Few studies have aimed at modeling the behavior of bio-cemented sands under various loading conditions, but their use can be restricted by the obtainability of their input parameters and the lack of a cumulative body of data to support their validation. This Dissertation presents the extension of a plasticity model for sands to bio-cemented sands (PM4SandC Version1). The motivation for this work is twofold: (1) to have a usable constitutive model which is applicable to bio-cemented sands (and by extension naturally cemented sands) under various initial and loading conditions, and (2) to advance the deployment of bio-cementation in the field as an alternative ground improvement method for liquefaction mitigation.The extension of the original constitutive model formulation to bio-cemented sands warrants an understanding of the main behaviors of interest. An extensive and critical review of the experimental studies on bio-cemented sands is performed first to collect the available knowledge on these geomaterials and identify the remaining gaps in this relatively new research field. This critical review of the mechanical testing is supplemented with one on the numerical studies on bio-cemented sands and other relevant cemented soils and weak rocks. The critical review provides insights into the mechanical behavior of bio-cemented sands and later informs the modifications to the original formulation of the constitutive model, and gathers the approaches used by other research to achieve this goal. In order to address the usability for the model by means of reasonable and attainable input parameters, a relationship between a field parameter, the cone tip resistance, and a constitutive parameter, the apparent cohesion, is developed to provide an estimation of the parameter most contributing to the cementation-induced changes. Readily available cone penetration measurements in a large tank experiment and a centrifuge test on bio-cemented sands in conjunction with synthesized cone penetration data from an axisymmetric penetration model using the Mohr-Coulomb constitutive model enable the development of such a relationship as a function of confining stress. The axisymmetric cone penetration model and its input parameters are described, simulations results are validated against experimental results, and simulations are extended to higher confining stresses before fitting a simple linear relationship. This relationship is later used in the estimation of cohesion for a system-level analysis using the PM4SandC model. Once the behaviors to be prioritized in the model extension and the input parameters to characterize cementation are known, the intended modifications are implemented in the original formulation of the sand model PM4Sand Version 3.2. Modifications in PM4SandC relative to the PM4Sand model include: (1) the introduction of input parameters quantifying the effect of cementation on the shear stiffness, the peak strength and the volumetric behavior, (2) the incorporation of an additional cementation-induced strength to the mean effective stress and hence the shift of the constitutive space to reflect the enhanced tensile strength of bio-cemented sands, (3) the adjustment of the dependence of the cemented shear modulus on confining stress, and (4) the addition of evolution laws to degrade/alter the cementation parameters as a function of damage accumulation. A generalized calibration demonstrating the performance of the model under various cementation levels, confining stresses, drainage, and loading conditions is performed in FLAC 8.1 and some guidance is provided to aid users in the calibration process. The extended model is then validated against a body of experimental data on bio-cemented sands. Overall, the model is able to qualitatively predict the trends seen in bio-cemented sands with minimal calibration effort by means of input parameters physically meaningful and obtainable from the field. The validation of the PM4SandC model at the element level is followed by its validation at the system level. Two one-dimensional site response analyses on bio-cemented columns are performed in FLAC 8.1 to (1) demonstrate the ability of the model to predict the free-field dynamic response of a treated site, (2) present a first-of-a-kind numerical effort to simulate site response analyses in bio-cemented sands. The PM4SandC constitutive model is assigned to the bio-cemented sand columns. The numerical analysis and its input parameters as well as the calibration process are described in detail. The dynamic response from the simulations in terms of accelerations, shear strains, cyclic demand, pore pressure generation, and response spectra is compared to the measured response in centrifuge models scaled to real site conditions. The constitutive model is found to reasonably approximate the dynamic responses from the centrifuge tests. A parametric investigation is also performed to provide insights into the effects of parameter uncertainties to PM4SandC on the overall dynamic response.