R&D for OWT Foundation Design

Transcription

1 R&D for OWT Foundation Design Geotechnical Engineering for Offshore Wind Infrastructure Workshop organized by HDEC and NGI Shanghai, China, 31 May, 2018 Youhu Zhang, PhD Technical Lead Offshore Geotechnics, NGI

2 Agenda Introduction Current research topics Partial drainage Robust foundation models for fatigue assessment Soil reaction curves for monopiles Loads on offshore wind turbines and effect of irregular cyclic loading Frontier and challenges Ground model and artificial intelligence Long term deformation of offshore wind turbine foundations Reliability based design Specific challenges in China Opportunities for collaboration NGI-China

3 Introduction: research and development at NGI NGI is a Research Institute 75% of the staff hold a MSc degree or above (26% hold a PhD) NGI receives an annual base funding from Research Council of Norway (RCN), which are used to fund small scale research projects (GBV) that everyone at NGI can apply and also larger multi-year strategic projects. NGI is actively involved in research projects funded through national, EU and other international funding bodies NGI carries out numerous industry sponsored research, such as jointindustry-projects (JIPs) NGI Strategy 2018 to 2021 World class research, development and innovation. At least 15% of revenue from research Technology for the future

4 Introduction: OWT and its foundations OWTs are light-weight and slender structures Low vertical load, large horizontal and overturning moment Dynamic sensitive structures under different excitation sources Foundation stiffness must be engineered precisely. Soil damping has an important impact on structural fatigue life Fatigue can often be the design driver Stringent serviceability limit Accumulated displacement/rotation after extreme loading event or during life time operation should be constrained to certain limits Mass production: large number of OWTs in a wind farm Requires efficient and streamlined design approaches.

5 Partial drainage General background Cyclic loading on OWT foundations at almost all time due to environmental loading and structural vibrations In sandy soils, drainage occurs simultaneously with generation of excess pore pressure, and even during a single load cycle Strength and stiffness of the soil is heavily dependent on the drainage condition during shearing Is accumulated pore pressure a good memory of cyclic effect? A big focus at NGI in the past few years Work carried out through an NGI internal strategic research project,

6 Partial drainage In-situ permeability measurement tool Permeability is a key input parameter for foundation design in sandy soils Difficult to obtain undisturbed sand samples In-situ measurement of the permeability is preferred An in-situ testing tool is developed at NGI, which features: A conventional CPT set-up, but with a filter section above the cone Pumping water in or out under constant pressure head through a plunger actuator In-situ permeability back-figured through design charts established by finite element analyses A conceptual illustration of an in-situ permeability measurement device under development at NGI

7 Partial drainage Laboratory triaxial testing with different level of drainage Mechanical behaviour of soil, in general, depends on the drainage condition when it is sheared. Particularly relevant for silty to sandy soils as the soil response under cyclic loading can be either drained, partially drained or fully undrained during a single load cycle A special laboratory triaxial test setup is developed at NGI to carry out tests under partial drainage in a controlled manner. undrained drained Five triaxial compression tests under varying degrees of drainage during shearing

8 Partial drainage Drainage criterion for monopiles in sand Normalised excess pore pressure at peak load, P Pile T p = t p c v 2 D Tp=48.5 (drained) Tp=24.3 Tp=9.7 Tp=4.9 Tp=2.4 Tp=1 Tp=0.5 Tp=0.05 (undrained) Normalised distance, L/D Dr. Shuzhao Li from CNOOC Research Institute carried out this work during her visit to NGI in 2017 Need a convenient criterion to establish the drainage condition under cyclic loading for monopile design For typical monopile size (5-10 m in diameter), essentially undrained soil response is expected during a single load cycle State-of-practice for monopiles in sand use p-y curves that were developed for drained soil response. This differs from the underlying assumptions for the p-y curves used for design, but what is the implications?

9 Partial drainage Drainage criterion for monopiles in sand Horizontal force, MN Horizontal force, MN Undrained 20 Tp= Tp= Tp= Tp= Tp=9 300 Tp=18 5 Tp= Drained Horizontal displacement at mudline, m D=6m L=30m D r =80% D=6m L=30m D r =80% 5 Drained 150 Undrained Rotation at mudline, Overturning moment at mudline, MNm Overturning moment at mudline, MNm Stiffer global load displacement/ration response with undrained conditions at high load levels than drained conditions due to dilation However, monopiles are designed with stringent SLS limit. DNV GL (2016) suggest to limit the mudline rotation due to environmental loading to 0.25, in addition to an installation tolerance of Within the rotation limit, almost identical global pile response between drained and undrained. Implication for practical design is insignificant.

10 Partial drainage Vertical pull-out capacity of suction bucket OWT has a light weight, but large environmental loading For a multipod jacket structure supporting an OWT, the windward leg can be subjected to tension loading during operation and extreme loading event Holding capacity is a function of loading rate Dr. Kanmin Shen, formerly Zhejiang University, now Huadong, visited NGI in 2016 and investigated the rate effect on holding capacity of suction buckets A advanced two surface plasticity model SANISAND was used in the simulation Shen et al. (2017) SUT

11 Partial drainage Vertical pull-out capacity of suction bucket

12 Robust foundation models for fatigue assessment General background Unsteady aerodynamic effects OWTs are dynamically sensitive structures Fatigue is often a design-driver Complex aero-hydro-servo-elastic modelling Time-domain simulations due to non-linearities Worked carried out at NGI in the last four years through a large project REDWIN sponsored by the RCN and industrial partners (Statoil, Vattenfall, NGI, IFE, NTNU, OlavOlsen) 1.5 M over four years Turbulent wind Current Irregular waves Soil-pile interaction

13 Robust foundation models for fatigue assessment Modelling approach Macro-element models

14 Robust foundation models for fatigue assessment Macro-element concept Provides load-displacement response of the foundation + soil at one interface point Different stiffness after load reversal Based on multi-surface plasticity Includes coupling between loads (similar to interaction diagrams) Hysteretic foundation damping

15 Robust foundation models for fatigue assessment REDWIN models - Flexible piles Foundation and substructre Model applicable Loading regime Redwin model 1 p, y p Soil support model y Redwin model 2 Distributed 1D model to be applied to any DOF. HM-loading Redwin model 3 VHM-loading Redwin model 3 1D p-y model with hysteretic damping Foundation structure interface VHM-loading V M H u θ u v u h Foundation model

16 Robust foundation models for fatigue assessment REDWIN models - Monopiles Foundation and substructre Model applicable Loading regime Redwin model 1 p, y p Soil support model y Redwin model 2 Distributed 1D model to be applied to any DOF. HM-loading Redwin model 3 VHM-loading Redwin model 3 Includes the coupling between x- and y-directions V M H u θ u v Foundation structure interface Foundation model u h VHM-loading

17 Robust foundation models for fatigue assessment REDWIN models gravity-based foundations Foundation and substructre Model applicable Loading regime Redwin model 1 p, y p Soil support model y Redwin model 2 Distributed 1D model to be applied to any DOF. HM-loading Redwin model 3 VHM-loading Redwin model 3 Foundation structure interface VHM-loading Includes the effect of the vertical load V M H u θ u v u h Foundation model

18 Robust foundation models for fatigue assessment REDWIN models bucket foundations Foundation and substructre Model applicable Loading regime Redwin model 1 p, y p Soil support model y Redwin model 2 Distributed 1D model to be applied to any DOF. HM-loading Redwin model 3 VHM-loading Redwin model 3 Foundation structure interface VHM-loading Includes the effect of the vertical load V M H u θ u v u h Foundation model

19 Robust foundation models for fatigue assessment Validation Natural frequencies Fatigue Damage Equivalent Loads

20 Robust foundation models for fatigue assessment Validation Natural frequencies Design prediction range Normlized PSD Revised analyses with Redwin model 2 (for monopiles) 1E f / f n,measured

21 Robust foundation models for fatigue assessment Validation Fatigue REDWIN Damage models Equivalent calibrated Loads to FEA (DEL) Simulated DEL Measured DEL agree well REDWIN with measured model 2 natural API p-y curves frequencies and DEL Difference 7 % is foundation damping 93 % is foundation stiffness

22 p-y curves for monopile design General background State-of-practice for monopile design uses p-y curves that were developed for conventional slender piles used for the oil and gas industry Monopiles: large diameter, small L/D ratio Lateral response dominated by the wedge mechanism The flow-around mechanism may not be relevant at all Pile tip resistance is important Possibly also the vertical skin frictions Calls for new models for monopile design Lateral loading Wedge Flow around Soil mechanism for a slender pile under lateral loading

23 p-y curves for monopile design General background The PISA project: 1) The PISA project is a large joint industry project 2) Large scale model testing in both clay and sand 3) Numerical simulations 4) The model provides a good framework for modelling the monopile response 5) Requires site specific numerical simulation to calibrate the model At NGI, work is also performed to develop a model that has a Byrne et al. (2017) OSIG SUT conference similar framework to PISA, but fundamentally based on soil stress-strain behaviour measured directly in lab testing.

24 p-y curves for monopile design NGI model: ingredient 1: p-y curves for wedge failure 20D Analysis 1 p Analysis 2 Analysis 1 y p Analysis 2 Difference between analyses 1 and 2 y D Horizontal translating 10D

25 p-y curves for monopile design NGI model: ingredient 2: base shear spring 20D 5D footing 10D 2.5D L tip 2.5D 20D 10D Step 1 Step 2

26 p-y curves for monopile design Validation Some observations: 1) At very low load levels, the monopile exhibit flexible/semi-rigid pile behaviour, i.e. large proportion of steel deformation 2) At high load levels, the monopile exhibit rigid pile behaviour, little steel deformation 3) Soil mechanism divided into two zones: a y m /D = 0.2% y m /D = 31% wedge failure in the upper part and a rotational mechanism at lower part Essentially replace the scoop failure surface with a flat base shear plane 4) Sufficient to consider the base shear only as the rotation centre way above the pile tip. Base rotational moment negligible.

27 p-y curves for monopile design Conceptual model for application External loading 1) Above the rotation point, p-y curves for wedge failure. 2) Below the rotation point, where the pile Tension gap p-y curves for wedge mechanism kicks back, soil on both sides of the pile is forced to be engaged. Flow-around p-y curves. 3) At the pile tip, the base shear S-u model Rotation point S-u curve for pile tip p-y curves for flow-around mechanism

28 p-y curves for monopile design Example validation results D = 6 m Depth below mudline, m , 2473, 5569 kn Mudline Depth below mudline, m , 2473, 5569 kn Mudline Depth=30m Load eccentricity: 30 m above ground 20 Solid curve: FEA Dashed curve: Beam-column model Lateral displacement, m Bending moment, knm s u = 100 kpa

29 p-y curves for monopile design Laboratory p-y testing apparatus

30 p-y curves for monopile design Laboratory p-y testing apparatus

31 p-y curves for monopile design Laboratory p-y testing apparatus Norm. secant stiffness at steady state K sec,norm UB BF LB Test 1 Test 2 Test 3 Test 4 Test 5 Normalized damping ratio D norm Test 1 Test 2 Test 3 Test 4 Test 5 Damping Norm. lateral displacement ( y/d) Norm. lateral displacement ( y/d) Normalized stead-state stiffness Normalized steady state damping ratio

32 Tp, s Loads on offshore wind turbines and effect of irregular cyclic loading General background The size of offshore wind turbines is increasing Requires larger foundations There is need to reduce the uncertainties in design through: Hydrodynamic models for critical design loads Procedures accounting for irregular loads on soil Large research project from RCN: WAS-XL Total budget of 22 MNOK Partners: Sintef Ocean, NGI, NTNU 4 year duration Project start in 2018 Hs, m

33 Loads on offshore wind turbines and effect of irregular cyclic loading 80 0 h τ (kpa) Time (s) 0 τ h (kpa) Time (s)

34 Frontier and challenges Streamlined design flow for large number of turbine locations A wind farm consists of many turbines. Variations in soil layering, strengths, water depth, wind speed etc. Site specific design is typically required For efficient design, this requires a streamlined design flow and interpolation and extrapolation of ground conditions and soil reaction models (e.g. p-y springs)

35 Frontier and challenges Intelligent ground model and artificial intelligence Advanced ground models combining geophysical and geotechnical data are already in existence Such ground models allow for interpolation and extrapolations of soil conditions across a large area The ground model can be continuously updated with additional data from field and laboratory investigations One step further to combine Metocean data Load calculation models Foundation design tools Which allows for: Preliminary foundation sizing based on chosen foundation concept Continuously updating through the different phases of the project Ground model with CPT data

36 Frontier and challenges Reliability based design A wind turbine is a complex system and the design is an outcome of: Turbine structure Ocean environment Soil conditions Aero, hydro, structural, geotechnical analysis models A reliability assessment of each modelling components Study the sensitivity of design outcome to the various components and identify factors that require most efforts for optimised design.

37 Frontier and challenges Long term deformations of offshore wind turbine foundations Life time deformation for OWT foundations OWT foundations subjected to millions of cycles of loading during the design life Accumulated deformation during the design life of OWT should be estimated Robust calculation method is to be developed. In soft seabed, consolidation settlement, in particular uneven settlement, need to be considered and assessed.

38 Frontier and challenges Specific challenges in China General very soft ground conditions, soft clay, silt, silty clay, silty sand Perhaps some catch-up to be made on the site investigation practice? Relatively limited experience in the soil properties under cyclic loading? Long term settlement Seismic and liquefiable soil conditions Frequent typhoon events Suction assisted installation in silty soils Scour around foundations

39 Opportunities for collaboration between NGI-China Every year, NGI have many small GBV projects, perfect platforms for initiating collaboration EU projects: NGI can apply for EU research funding, co-funding mechanism available from National Science Foundation China (NSFC) China-Norway funding possibilities: in 2018, Research Council of Norway and NSFC announced NOK 80 million joint funding for China- Norway NGI hosts CSC PhD and academic fellows with relevant background NGI carry out numerous industry sponsored research

40 Thank you for your attention!

41 NORWEGIAN GEOTECHNICAL INSTITUTE NGI.NO #onsafeground