OTC-29557-MS ApplicationoftheFindingsofthePiSAJointIndustryProjectintheDesign ofMonopileFoundationsforaNorthSeaWindFarm Sebastien Manceau Robert McLean Anna Sia and Marisa Soares SNC Lavalin Atkins Copyright 2019 Ofshore Technology Conference This paper was prepared for presentation at the Offshore Technology Conference held in Houston Texas USA 6 9 May 2019. This paperw seltdfrprentatnbyanTC prgammtefolngrevfimnntaind inanabract smitdbe athr(s)Cnntsf thepaerhaventbeen reviewed by the OfhoreTecholyConfrene and are suject to etion bythe auhor(s)Themaleril doesnoteessrly reflectay position of the Ofshore Technology Conferenoe s offiers ormembers. Electronic reproduction distibution or strage of any part of this paper without he wrlien not be copied The abstract must ontain cnspiuous acknowedment of OTC copyrigh. consentof the Ofshore Technlogy Conference is prohibite. Pemissin teproduen printis restricted t anbstract ofnot mre thn 30 words; ilustrations my Abstract Monopiles are the most mon foundation type in the offshore wind industry. Their design is largely dependent on the ability to accurately model the soil-structure response of the foundation with more refined modelling approaches enabling significant reductions in required embedment depth fabrication cost and installation risk. The PISA joint industry project (JIP) has been pleted in recent years with the objectives of developing a more refined soil-structure response modelling method pared to other available methods such as the API p-y curve approach. The scope of this paper is to detail how the PISA remendations have been implemented on a real offshore wind farm project located in the UK North Sea identifying how the findings can be incorporated into a bined geotechnical and structural analysis approach to enable efficient serial design of multiple foundations for wind turbines. The paper presents how existing design processes and criteria can be modified to take into account the remendations of the PISA JIP for use in design. Discussion will be provided on the following procedures: calibration of the PISA 1-D soil response formulations to site specific conditions; the bination of the homogeneous sand and clay formulations to accurately model soil-structure response in layered soil profiles; and consideration of the effects of cyclic loading in conjunction with the use of the PISA monotonic soil response formulations. Results will be presented to demonstrate the calibration of the PISA 1-D soil response formulations to a layered soil site. Discussion will also be provided on the significant monopile lengths savings achieved when using a PISA approach pared to an API p-y curve approach. The monopile mass reduction will be illustrated against trends derived from installed monopiles. Observations will be provided on how the use of a PISA based approach can affect the goverming design cases and how this is likely to impact on monopile design for future projects. Discussions and conclusions will also be presented on the challenges of implementing the PISA remendations in monopile design for real projects and what additional work The PISA JIP remendations are the cutting edge in monopile foundation design. The paper will provide discussion on how these remendations can be effectively implemented in design based on experience from the foundation design for a real offshore wind farm. The wind farm in question will be one
OTC-29557-MS of the first constructed for which foundations have been designed using a PISA based method demonstrating the significant CAPEX savings possible using the PISA approach. Introduction Offshore wind is the most scalable of the renewable technologies and has a major role to play in decarbonising energy infrastructure and helping mitigate climate change. In Europe the installed and operational capacity was around 15.8GW at the end of 2017 (Wind Europe 2018) and there are ambitious growth targets to 2030. The increase in offshore wind capacity has been supported by a dramatic reduction of the levelised cost of energy. For example in the United Kingdom the 2017 Contract for Difference (CfD) saw strike prices of f74.75/MWh and f57.50/MWh for delivery year 2021/22 and 2022/23 respectively (a reduction of 38% and 50% respectively pared to the 2015 CfD round) making offshore wind cheaper than new nuclear. Similar trends have occurred in mature European markets with even the first subsidy free auction bid recorded in Germany. Driving down the cost of foundations which remains a significant proportion of the overall cost is a key objective to support continued investment in offshore wind worldwide. Monopiles are the dominant foundation type representing 87% ofinstalled foundations at the end of2017 (Wind Europe 2018). The loading from the wind on the turbine and waves and currents on the monopile (and transition piece) is predominantly lateral and generates a large overturning moment at mudline. As water depths and turbine sizes increased so did the loading and the sizes of installed monopiles (diameter D and embedment L). The soil-structure interaction is key to the design of monopile foundations yet available guidance in offshore codes and other traditional approaches for modelling the soil response under lateral loading have severe limitations. The recently pleted PISA JIP offers an improved method for the assessment of the soil response. The practical application of the PISA JIP remendations to the design of a real offshore wind farm is presented and the resulting significant reductions in required embedment depth and monopile mass (with associated fabrication cost and installation risk reductions) are discussed. Monopilesoilresponse Traditional approaches Monopiles have traditionally been designed using a Winkler approach with the monopile modelled as a beam supported on non-linear p-y curves representing the relationship between lateral soil reaction and displacement. The p-y formulations presented in the offshore design codes (API 2011 DNVGL 2016) have been extensively applied in the oil and gas industry. They originate from limited pile lateral load tests on long slender piles undertaken in the 1950s through to 1970s and are aimed principally at the prevention of collapse. Their limitations for the design of large diameter rigid monopile foundations governed by consideration of natural frequency and fatigue have been well-documented. Efforts to derive alternative p-y curves for use in monopile design through finite element analyses have been hampered by the lack of pile load tests with representative conditions (L/D ratios and loading type) for calibration. Evidence from monitoring of installed turbines on monopiles (e.g. Kallehave et al. 2012) indicates that the conventional design approaches lead to an under-prediction of the natural frequency (i.e. an under-prediction of the foundation response stiffness). PISA approach The Pile Soil Analysis (PISA) joint industry project was set-up to address the shortings of conventional design methods. An overview of the project is presented in Byrne et al. (2017).
OTC-29557-MS The PISA project involved large diameter lateral pile load tests at two sites (Cowden UK for stiffclay and Dunkirk France for dense sand) state of the art finite element numerical modelling and the development of a one-dimensional (1-D) design approach using a Winkler-type approach extended to account for four ponents of soil response. Figure 1 (after Byrne et al. 2017) illustrates the four ponents of resistance and the associated 1-D soil rection curves: p-v curves for lateral soil reaction along the pile embedment ● m- curves for distributed moment along the pile embedment ●Hg-v curve for base shear at the pile tip ●Mg- curve for base moment at the pile tip Vetical shear (A'z)s Serloroead Figure 1Components of resistance considered in PISA 1-D formulations (after Byrne et al. 2017) The PISA project has defined two design approaches: ‘Rule-based’ approach. This uses 1-D soil reaction curves generated using pre-defined mathematical functions with simple soil parameters including undrained shear strength Ss and small strain shear modulus G for clays and initial vertical effective stress o's and G for sands. The formulations ofthe 1-D soil reaction curves established in the PISA report are based on specific soil profiles (idealised clay till profile and idealised dense sand profile) and a range of monopile geometries and loading regimes. The rule-based method can be adopted for preliminary design activities. ●“Numerical-based’ approach. This approach uses 3-D numerical modelling to establish bespoke soil reaction curves (for use in 1-D models) for site specific ground conditions (as well as monopile geometry and loading regime). The numerical-based method can be adopted in detailed analyses. Applicationtorealproject The project The remendations from the PISA JIP have been applied in the design of the foundations for a wind farm in the UK North Sea. The wind farm prises 90 MVOW v164-9.5MW wind turbine generators (WTGs) and two offshore substation platforms (OSPs) all the structures are supported on monopiles. The
OTC-29557-MS water depth ranges from 15 to 21m LAT. The ground conditions vary across the site but typically prise high strength to very high strength overconsolidated clays and dense to very dense sands. The geotechnical properties of the materials were generally parable to those considered by the PISA JIP at the Cowden and Dunkirk test sites and in the development of formulations for the soil response curves for the 1-D rule- based approach. Implementation of PISA remendations in soil lateral response modelling The numerical-based approach was used; Figure 2 summarises the process followed to derive 1-D soil lateral response formulations with the following subsections presenting a brief summary of the key steps. 1-D soil response modelling Numerical modelling Development of design tools to implement PISA 1-D soil response Set up FLAC 3-D numerical model formulations PISA Validation of 1-D tools against Validation of FLAC 3-D numerical PISA results model against PISA results site 1-D soil response analyses for FLAC 3-D models for representative locations across representative locations across the site the site Calibration of 1-D soil response formulations (for use in geotechnical design) to numerical model results Derivation of modifiers on p-v and Note 'simplified approach’ for use Hy-vcurves to account for in structural design to increase moments effects effciency of analyses process Figure 2Numerical-based lateral response design method PISA validation. Tools for 1-D approach The 1-D assessment of the monopile lateral response was undertaken using the Oasys Alp 19.1 software (hereafter referred to as ALP). Limitations of the software had to be overe. Firstly ALP does not allow non-linear m-q and Mg- reaction curves to be modelled directly. To circumvent the issue the resistances from the m-w and Mg-y curves were modelled in ALP as resisting moments. Iterations were required to ensure that the values used corresponded to the predicted rotations (and in sands distributed load level). Secondly the non-linear p-v and Hg-v curves could only be defined by six points and iterations were required to ensure the curves discretisation was fine enough around the predicted displacements to limit inaccuracies (as the software interpolates between points). A tool was developed to generate non-linear soil reaction curves of the form remended by PISA for all four ponents of resistance. A second tool using an application programme interface was developed
OTC-29557-MS 5 to automatically import inputs into an ALP model run ALP export and interrogate the ALP outputs and iterate with the tool generating the soil reaction curves. Analyses were performed to validate the tools by reproducing results presented in the PISA report as illustrated on Figure 3. Figure 3 presents pile head displacement versus applied horizontal load for a monopile in clay; the plain and dash black lines are results from PISA finite element and 1-D analyses and the red line is the 1-D validation using the tools. The validation analyses confirmed that 1-D models with soil resistance curves of the form remended by PISA can successfully be implemented in ALP (and other traditional beam element software). 20.0 18.0 3DFE =-- 1D (parametric) API/DNV 40 0.1D! ZO Ple head dsplocemet % (I) 0.4 0.6 APSA cures ponerts Figure 31-D tool validation (clay) Numerical Modelling Numerical analyses were performed using the finite difference FLAC3D software version 5.01 (although alternative numerical modelling sofware could also have been used). The soils were modelled using an isotropic pre-failure non-linear elasto-plastic constitutive law assuming an associated Tresca failure criterion in clays and a non-associated Mohr-Coulomb failure criterion in sands. In this model the shear modulus degrades with increasing shear strain using a non-linear normalised shear modulus shear strain relationship applied through a FISH function. The numerical modelling approach was validated against results presented in the PISA report for finite element analysis of monopiles with geometries parable to those anticipated for the wind farm. In clay the FLAC3D model provided a very good match to the load-displacement response predicted by the PISA numerical modelling work up to load of circa 75% of the failure load (defined as the load causing ground level deflections of 0.1D) as illustrated on Figure 4. This means that the FLAC3D model provided a very good estimate of the monopile response over the loading range under consideration (the factored ULS load was less than 75% of the failure load). For higher load levels the FLAC3D model overpredicted the stiffness of the response owing to limitations of the Mohr Coulomb model. In sand the FLAC3D model provided a very good match to the load-displacement response predicted by the PISA numerical modelling work for small displacements and conservatively underpredicted the stiffness of the response at higher load levels.
PISA联合工业项目成果在北海风电场单桩基础设计中的应用.pdf
