Testing and modeling of sheet pile reinforced dikes on organic soils

Abstract

The Netherlands is inherently challenged by water, as a large part of the country lies below sea level and several major rivers in North-western Europe cross the country. The most prevalent form of protection from coastal and river floods in the Netherlands includes approximately 22,000 km of earthen dikes, of which 3800 km are primary flood defenses (i.e., the first line of defense against high water). Subsidence, sea level rise, and the increase of rain intensity and river discharge due to climate change further challenge existing flood defenses to maintain required levels of safety. To do so, the top elevation of existing earthen dikes is often incrementally raised over time. However, raising a dike requires an extension of its base, which is frequently restricted by the presence of existing buildings and other spatial constraints. These dikes can be reinforced by alternative means such as a sheet pile wall.
Another specific challenge regarding dike reinforcement in the Netherlands includes the presence of soft subsoil conditions at many dikes. These soft soils often consist of organic clays and peats. Such soils have a low stiffness and continue to deform over time; their strength is not well understood and often underestimated. Furthermore, organic soils are often not properly identified from Cone Penetration Tests (CPTs) using common interpretation methods, while CPT is the main testing method in the Netherlands.
This research focuses on improving two aspects of the global stability assessment of dikes in the Netherlands: the modeling challenges of organic soil and dike reinforcement using sheet piles. Chapter 2 of this dissertation addresses the empirical relations for organic soils. The CPT-based correlation to derive the soil unit weight (Lengkeek et al. 2018) is validated and improved. Furthermore, new CPT-based correlations for organic soils are obtained by relating the soil state parameters to the cone resistance and the unique soil type properties to the friction ratio. An adjustment to Robertson (2010) CPT-based classification is proposed. In the improved SBT classification, organic soils (SBT=2) are redefined and subdivided into peat, organic clay, and mineral clay with organic matter.

In chapter 3 the new Critical Stress Ratio (CSR) model is presented, which classifies as ‘Simplified Critical State Soil Mechanics’. The CSR model can be seen as a theoretical version of the SHANSEP equation, providing a link between effective stress parameters, obtained from common laboratory tests, and the undrained shear strength. The model can be implemented in LEM for ultimate limit state stability analysis.
The CSR model provides the state dependent undrained shear strength for each stress point. The CSR model does not require to determine the exact yield contour as in a constitutive FEM model, this is taken into account by a variable spacing ratio, called the ‘Critical Stress Ratio’. This parameter of the CSR model can be regarded as the over-consolidation ratio at which no net excess pore pressures occurs, a parameter which can be fitted based on a few CAUC tests. Furthermore, the CSR model contains methods to obtain other model parameters for existing constitutive models used in the finite element method, such as the Poisson’s ratio which determines the horizontal and isotropic stress in unloading.
Chapter 4 presents the set-up, results, and evaluation of the full-scale failure test. (In Dutch: ‘Eemdijk damwandproef’), initiated by the Dutch Flood Protection Programme. The Eemdijk full-scale failure tests involves separate tests on (1) sheet pile panels, (2) on a ground dike, as well as a combined test (3) on a ground dike with sheet pile reinforcement. The Eemdijk full-scale failure tests provides valuable insights through a detailed analysis of the deformations of dikes leading up to and beyond failure. Furthermore, the soil investigation is re-examined and parameters are determined for multiple constitutive models applied in FEM back-analyses. Finally, both the CPT-based methods, the CSR model and the SHANSEP-NGI-ADP model are validated at the Eemdijk test site.
The back-analysis of the pull-over tests (PO-test) confirmed that the cross-section class 2 sheet piles (AZ26) reached the full plastic bending moment capacity and the cross-section class 3 sheet piles (AZ13-700 and GU8N) reached at least the elastic bending moment capacity. Furthermore, from the analysis of the SAAF measurements it is concluded that the stiffness of only the side sheet piles of panels should be reduced due to edge effects.
The ground dike test (GD-test) and sheet pile reinforced dike test (SPD-test) allowed for a unique comparison and provided insight in the critical deformation rate prior to progressive failure. This criterion is useful in the assessment of unstable slopes. The GD-test illustrates the importance of high density of soil investigations and the importance of high quality CPTs (ISO class 1) and proper CPT based classification.
The sheet pile reinforced dike test (SDP-test) shows that a continuous sheet pile, with sufficient length and embedment, makes an important contribution to the robustness of the dike after failure. Even after structural failure due to a plastic hinge, all sheet piles remained intact and interlocked. The failed sheet piles functioned as a weir and ultimately prevented breaching.
Based on a careful examination of the Triaxial (CAUC) test it is recommended to use the 15% axial strain value as a basis for the ultimate value and to apply additional criteria to prevent unrealistic high or low values for the undrained shear strength, and to re-examine the applied geometrical corrections.
Based on the performed variation analysis it is recommended to use average stiffness parameters in a SLS or ULS dike design analysis, when performed with advanced constitutive models in FEM.
The alternative approaches to dike assessment presented in this research are expected to result in a more economic and better understood dike design and assessment based on improved field data interpretation (chapter 2,) undrained shear strength and modeling procedures (chapter 3) and takeaways from the full-scale tests and back analyses (chapter 4).