| dc.description.abstract | Nowadays, a significant part of geotechnical analyses refers to settlement problem, due to its impact on designed constructions. The problem is more noticeable when displacements are large, what usually takes places when the soil profile consist of thick soft and organic layers. Unfortunately, the tools available for geotechnical engineers are not perfect. Indeed, they cannot predict the settlements fully correct but still can be used to evaluate soil behaviour which can be later compared to field measurements and evaluated.
In this study, a motorway site located in Gdansk (northern Poland) is investigated. Its soil profile consists of five main layers with different properties, which were decided to be improved by application of vertical drains and preloading embankments. Construction procedure, soil conditions and load history were modeled in Plaxis 2D in order to obtain settlement of surface and horizontal displacement at the toes of embankment. Later, the results are compared with the records obtained in the field.
To ensure reliable results, three different soil models were chosen. Linear elastic perfectly-plastic Mohr-Coulomb model was assigned to weak permeable layers to reduce error which could show up due to assumed parameters which were not given in the documentation. Soft, low permeable layers were simulated with Soft Soil Creep Model, which takes into account not only primary consolidation but also the creep after pore pressure dissipation. For sand bodies, the Hardening Soil Model was applied. However, some assumptions were necessary to model these layers, but as the sensitivity analysis had shown, the results were almost not affected by this action.
Furthermore, vertical displacement measurements are compared in five different cross-sections to evaluate the performance of Plaxis 2D. The analysis showed, that for chosen soil models, vertical deformation was estimated correct in four of five cross-sections (error less than 1%). Later, the analysis investigating progress of settlement along cross-sections, showed, that final settlement was modeled correctly in 3 out of 4 cases. Plots created after the consolidation of first consolidation vary from the data obtained from the field. Moreover, it seems that Plaxis underestimates deformation on the sides of the embankment.
In addition to the above, horizontal displacements were investigated. Three out of four curves present similar shapes to the ones obtained in the field. However, the displacement errors are significantly larger in comparison to the vertical deformation s. The same three graphs mentioned above, give the error bigger than 15% compering to the average value of field measurements.
The study shows that Plaxis with chosen soil models can fairly good estimate final settlement. However, its development in time and vertical displacement computation still need to be improved. | en |
