| dc.description.abstract | The project Kværnerdammen F1 in Oslo comprises a seven-story building at the toe of a slope, partly consisting of quick clay. Multiconsult is the geotechnical consultant on the project and designed a multi-anchored sheet pile wall (SPW) bolted to bedrock, securing the building pit. Lime cement columns (LCC) was used in front of the SPW to increase stability. This thesis compares results of different soil models in the finite element program PLAXIS 2D and the spring supported beam program GeoSuite Excavation (GS) to measured behavior of the SPW during construction of the building pit. Layers of silty clay and quick clay are modeled in PLAXIS with the constitutive models: Hardening Soil (HS), NGI-ADP and Mohr Coulomb (MC). In GS the clay layers are modeled using the Total Stress Automatic option. Using different models to calculate the same problem raises the question; Which of the models are the most accurate when compared to measured behavior?
A study was performed to gather necessary information about the project and on theory about the different models in each software. The soil parameters were tuned to a strength profile based on CPTu and triaxial tests from the area. A parametric study was performed primarily to examine the effects of LCC and roughness.
NGI-ADP gives the best fit to the measured displacements. When tuning models to a strength profile it easily goes at the expense of stiffness, especially for the HS model. HS is capable to model the long-term effect after consolidation, but may have difficulties in reproducing both short term and long term conditions using one parameter set. Comparing simulated and measured anchor forces was challenging since the load cells were placed at some distance from the chosen cross section. Still, all models seem to give an acceptable fit to the measured anchor forces, except for MC. The model gives too large displacements and the user should be careful when using this model on a multi-anchored wall as the anchor forces might not be representative. Inserting an unloading/reloading stiffness in MC by changing material parameters between phases, gives better results. In NGI-ADP, stiffness and strength relates through a stiffness ratio G/s$_u$ which seems to contribute in making the model fit better to measured behavior. Among the models used, the NGI-ADP is the one where changes in the input roughness number has the lowest influence on the results. Including LCC in the modeling seems to increase the accuracy of the results for most models.
It is difficult to find the balance between increasing the strength and maintaining the desired stiffness in GS. The installation of anchors will push the wall back towards the soil and active and passive zones in the profile may change during the phases. In GS, the roughness parameter used must be given a direction before calculating the project, and cannot be changed during the calculations. Nevertheless, modeling in GS produces surprisingly good results compared to the measured values, where including LCC with partial effect in the model were most accurate.
Obtaining realistic anchor forces when calculating by hand is challenging for this project. With several simplifications, too low forces were obtained from hand calculations, as pre-stressing of anchors and stiffness of the wall are not included. The use of numerical models is an advantage when calculating a multi-anchored SPW in sloping terrain. | |
