Evaluation of sheet pile behavior based on Eemdijk monitoring data

Abstract

From the 3500 kilometers of primary flood defenses in the Netherlands, a third does not comply with the current standards. As a result, over 700 kilometers will be strengthened in the upcoming years. Traditional strengthening of dikes consists of strengthening with soil and requires space that is often not available. Further strengthening of dikes, therefore, requires smart solutions, which increase the strength, without changing the dikes cross-sectional area. One of these solutions is the use of sheet pile walls as stability screens. However, in contradiction to all commonly used design codes, including the Eurocode, the current design approach for stability screens in dikes does not allow for plastic calculation. Furthermore, a model factor of 1.1 is applied to the forces and moments, when the sheet pile is applied as a panel. These two aspects make the application of sheet piles as stability screens expensive to apply. A better understanding of the failure behavior is needed to clarify whether the current design approach is conservative. Within the scope of the Flood Protection Program, a cross-project exploration was formed between water boards and the national government at the end of 2014. This collaboration is called POV and the aim is to innovate dike reinforcement which will result in a better, faster and cheaper process. A separate branch of this organization, POVM, is focusing on macro-stability. Within this scope, the use of sheet pile walls as stability screens is investigated. In this context, the Eemdijk test has been performed. This consisted of two full-scale tests up to failure. Although the test was successful and led to many new insights, the reliability of strain measurement data is questionable. The strain distribution, however, is essential in understanding the failure behavior. That is why this thesis addresses the reliability of the measurement data. This is done by means of comparing measurements of different sensor types. Each sensor is then classified using four classes that indicate the reliability. Subsequently, reliable data is used to obtain a curvature distribution. In order to convert strains into moments, a moment-curvature diagram is required, which relates curvatures to moments in the sheet pile. This relationship is obtained by means of finite element calculation of a 4-point bending test. Subsequently, the chosen model is calibrated by means of reproducing the measurement results of three real documented 4-point bending tests. Finally, the model is applied to the sheet pile profiles used in the Eemdijk test. With the obtained moment distributions, conclusions are drawn with respect to the current design approach. Furthermore, recommendations are made with regards to the applied monitoring program and the influence of soil on local buckling.