Analysis of Settlements of Test Embankments During 50 Years - A Comparison Between Field Measurements and Numerical Analysis

This report is dealing with time dependent settlements calculated with numerical methods and subsequent comparisons with field measurements. The numerical computations in this report have been performed using Plaxis, a two-dimensional numerical program based on the finite element method. Plaxis is very practical for solving complex geotechnical problems involving settlements or slope stability. The compared objects in this report are test embankments at Lilla Mellösa and Skå-Edeby. At Lilla Mellösa two test fills were constructed by SGI in 1945 – 1947 while at Skå-Edeby four test fills were constructed in 1957. The background to the building of test fills was the search of a place for a new international airport outside Stockholm. The soil profile in both areas consists of very compressible soil layers with large thickness. There are six different material models to choose between in Plaxis. The differ in models are how accurate they describe the mechanical behaviour of soils. The purpose of each model is to establish a relation between stresses and strains in the material. When modelling the test embankments, some different soil models of different complexity; Mohr Coulomb, Hardening soil (allows for the use of different deformation moduli for loading and reloading), and Soft soil creep (includes creep behaviour), respectively have been used. Because the work involves secondary compression for soft soil layers of clay, the use of Soft Soil Creep model has been considered reasonable. Mohr Coulomb and Hardening Soil have been used for other layers such as gravel, sand, dry crust and fills. Plaxis gives fairly good results compared to field measurements for all the cases for both drained and undrained conditions. Calculations show that the Soft Soil Creep model matches better the field measurements than the Soft Soil model. The calculated excess pore pressure distribution with time however shows a significantly different behaviour than the corresponding field measurements