Investigation into a new CPT-based design method for large diameter monopiles in sand

More Info
expand_more

Abstract

The API RP 2GEO (2011) p-y method is an industry standard approach to predict the lateral pile behaviour of slender oil platform piles with length (L) over diameter (D) ratios > 10. However, offshore wind turbines are founded on large diameter monopiles with an L/D ratio < 8, which are rather rigid piles. There are concerns about the applicability of the standard p-y approach to piles that have low L/D ratios and it is expected that other soil reaction terms such as base shear, base moment and distributed moment also play a role in rigid pile behaviour (Davidson & Donovan, 1983).
Due to the increasing demand for renewable energy, the offshore industry is taking a leap when it comes to the development of offshore wind parks. Because power generated by a wind turbine is directly related to rotor diameter and thus to monopile dimensions, it is expected that monopiles with even lower L/D ratios will become standard. With the prospect that 27 % of European energy should be generated by renewable energy sources by 2030 (EUCO 169/14, 2014), efficient design methods for offshore wind turbine foundations are desired. Therefore, in this thesis, the performance of the current design approach is analysed and an investigation is conducted into the development of a new design method for large diameter piles in sand that are tailored to the offshore wind industry.
In 2013, the PISA (PIle Soil Analysis) project, a joint industry research project, has been established to develop a new design method for large diameter monopiles under lateral loading. As part of the project, pile load tests (PLTs) and cone penetration tests (CPTs) were performed in Dunkirk, in sand. This new (unpublished) design method includes the derivation of soil springs, that represent the additional soil reaction terms, by using complicated Finite Element Method (FEM) calculations which require input from advanced laboratory tests. Even though these calculations are accurate, they are also time-consuming. In the past two decades, several cone penetration test (CPT)-based p-y methods have been published. With these methods, p-y soil springs are derived directly from CPT data and used as input in a pile-response model, avoiding the necessity of advanced laboratory testing and FEM calculations.
In this thesis, the performance of CPT-based p-y methods, compared to the current approach is evaluated by using the API RP 2GEO p-y method (2011) together with CPT-based p-y methods published by Novello (1999), Dyson & Randolph (2001), Li, Igoe & Gavin (2014) and Suryasentana & Lehane (2016) to simulate the pile behaviour of short (L/D=3), medium (L/D=5.25) and large (L/D=8) piles that were tested in Dunkirk during the PISA project. A pile-response model is built in MATLAB to perform these calculations quickly and accurately. It is found that the API p-y method generally overestimates the pile response compared to results of CPT-based p-y methods and PLT measurements, especially for the short pile. Therefore it can be concluded that the current API p-y approach is not sufficient for predicting short pile behaviour. The CPT-based p-y methods predict stiffer responses, yet not stiff enough to match PLT measurements of the short pile. This strengthens the expectations that for short pile behaviour, apart from the p-y term, also other soil reaction terms play a role.
Still, the largest contribution to pile displacement y is determined by the lateral soil pressure p (Byrne, et al., 2015a). Therefore, the development of a new p-y method based on the longest pile (L/D=8) that was available from the PISA PLT data set is investigated first. This investigation involves processing PISA PLT data into pile displacement- and soil pressure profiles. During this process, it is found that the direct differentiation of raw PISA measurement data comes with problems and results into wiggly soil pressure profiles. Curve fitting the moment profile has been applied in an attempt to overcome this issue. In this thesis, all ‘simple’ MATLAB curve fitting techniques have been analysed for the pile with L/D=8. The curve fitting techniques that have been investigated can be separated into four different types: global polynomials, interpolation splines, smoothing splines and least-squares splines. The ‘simple’ MATLAB curve fitting techniques in combination with the PISA PLT data did not result in a trustworthy pressure profile. Hence it was not feasible to continue the research into a new CPT-based p-y method based on the ‘simple’ MATLAB curve fitting techniques and the PISA data that was available.
However, a future work approach is presented that can be used as a guideline for further research into the development of a new CPT-based design method for large diameter monopiles. A step-by-step approach is given for the complete derivation of CPT-based p-y curves from PLT data, together with the first steps for deriving springs for the additional soil reaction terms. The future work approach elaborates on data gathering, PLT testing, PLT data processing, curve fitting, how to the link p-y curves to qc data, and what formulas can be used as a preliminary basis for the derivation of the base shear, base moment and distributed moment. Following the future work approach created in this thesis, an accurate and cost efficient design method for offshore wind turbines could be developed that positively contributes to the energy transition and a sustainable environment.