Geotechnical earthquake design of foundations for OWTs

Geotechnical earthquake design of foundations for OWTsGeotechnical Engineering for Offshore Wind InfrastructureWorkshop organized by HDEC and NGIShanghai, China, 31 May, 2018

Amir M. KayniaTechnical Expert, Vibration and Earthquake Engineering, NGIAdjunct Professor, Norwegian University of Science and Technology, NTNUAdjunct Professor, Zhejiang University (ZJU)

ContentIntroduction Earthquake hazard and definitionsEarthquake design criteriaEstimating accelerations and loads Limit States and performance criteriaPerformance of different foundation typesLiquefactionIssues in response under horizontal and vertical earthquake excitation

Type of support structures addressed

Mono-piles (most common type): Water depths ∼ 30 m, D ∼ 6 m, L/D ∼ 5 (next generation: water depth ∼ 50 m, D ∼ 11 m)Gravity-based foundation Monopod: D ∼ 15 m Steel jacket on piles Tripods on suction caissons Floating turbines with anchorsFoundations constitute ∼ 25% of costs

Basic concepts in seismic hazard Ground Acceleration, Magnitude, Intensity

Key parameters in earthquake design are peak ground acceleration (PGA) and response spectra Intensity is a qualitative means to express the level of shaking at a given location. It cannot be used in design unless it is converted to acceleration Magnitude (M) is used by seismologists to characterize earthquake energy at source. Again, to be useful, it should be used to estimate the acceleration at the site.

Seismic HazardInformation about accelerations are obtained from a Probabilistic Seismic Hazard Analysis (PSHA) which gives probability of exceeding PGA in a given time period.In most design codes it is probability of 10% exceedance in 50 years – in simple terms: earthquakes with Return Period 500 years. Typical values of PGA for 500-yr return period: Oslo: 5% g; Japan ∼ 40% g, Shanghai: 8% g

PGA

Overview of PSHAa) Identify all relevant earthquake sources.b) Characterize rates at which earthquakes

of various magnitudes (M) are expected to occur for each source.

c) Characterize distribution of source-to-site distances (R) for each source.

d) Predict the chosen intensity measure for all combinations of magnitude, distance and ε (number of standard deviations of ground motion model used to estimate the intensity measure) for each source.

e) Calculate the hazard curve at each spectral period.

Tectonic setting (example)

From Wang et al. (2014)

From Carlton et al. (2018)

Ground Motion Prediction Equation (GMPE)

Models that predict the expected range of an earthquake intensity measure (IM) at a site from a given earthquake scenario based on source, path, and site effects. Simplest models represent source effects by moment magnitude (Mw), path effects by the distance from the rupture zone to the site (RRUP) and site effects by the average shear wave velocity over the top 30 meters (Vs30)Local GMPEs are best, otherwise have to use GMPEs for similar seismic regions:─ Shallow crustal in active tectonic regions (e.g. California, Italy, Turkey, Greece)─ Shallow crustal in stable continental regions (e.g. Northern Europe, Eastern N.A.)─ Subduction zone (e.g. Japan, Chile, New Zealand, Alaska)

Results of PSHA:Hazard curveDe-aggregationUniform Hazard Spectrum (UHS)Conditional Mean Spectrum (CMS)

Provide input to:Selection of ground motion time histories and matchingSeismic site response analysisLiquefaction analysisDynamic slope stability

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

0.01 0.10 1.00 10.00

Res

po

nse

Acc

elel

erat

ion

(mm

/s2

)

Period (secs)

( a p g 5%, o o ta ot o , s30 500 /s 5 0)

4975 year

3500 year

2475 year

1000 year

475 year

300 year

100 year

0.1g

1g

Site response (amplification of ground motion)Soil response changes frequency content of earthquake motionGenerally soft soils amplify ground motion => Case of Mexico City (M = 8.1)

Large amplification ground motion around natural period of site (2 sec. here)

Global map of PGA on bedrock for return period of 500 years

Global Earthquake Hazard onshore

with typical soil amplification of ~2 on soft soil in Shanghai area, one could expect PGA ~ 0.15 g on seabed

Earthquake Hazard: offshore (ISO)

PGA = 0.1 g on bedrock for 1000 year return period, East China Sea

Using amplification factor 2 for soft soil, and converting to 500 year return period, gives PGA ∼ 0.15 g on seabed (this is not a low value!)

Loads on OWTsMain loads:─ Wind (mean and turbulent) ─ Wave loads─ Harmonic load in connection with rotor rotation,

“1P load”─ Harmonic load due to blade passing/shadowing,

“3P load”─ Other loads, like earthquake, ship impact, ice, . .

Earthquake loads are in most regions not governing due to their long natural period , however other aspects such as liquefaction of loose-medium dense sands and vertical earthquake motions are important issues Mean and turbulent

wind profiles

Load frequencies and dynamic criteria

The loads have different frequency bands─ Wind about 0.01 Hz (100 s.) ─ Wave typically 0.08-0.15 Hz, ─ 1P: depending on turbine, 0.13-0.3 Hz─ 3P: three times above, 0.4-0.9 Hz─ Earthquake: horizontal 0.5-2.0 Hz─ Earthquake: vertical 2.0-8.0 Hz

Vestas’s V90 3.0 MN turbine

To avoid resonance, natural frequency of turbines should lie in the 1P – 3P band

f1 = 0.31 Hz

Natural frequency of OWTs is typically in the range 0.3 – 0.4 Hz

Some design differences to other structuresFor design earthquake (475-yr), foundation should not experience permanent tilt (more than 0.50°) due to strict performance criteria of turbines.Earthquake is considered simultaneously with other environmental loads (wind and wave) representing operational conditions.There is very little damping in tower structure (as low as 0.5%) in side-side direction, and equally low in fore-aft direction in stand-still condition. Mono-piles are much larger than traditional piles used in other structures; therefore, classical solutions, such as p-y curves are not valid.The response of piles to liquefaction has not been adequately studied and it is not well understood. Kinematic pile interaction will result in larger rotations at pile head than in traditional (smaller diameter) piles.

Limit States and performance requirementsFour limit states are to be satisfied─ Ultimate Limit State (ULS): structural strength and stability of members and joints ─ Serviceability Limit State (SLS): maximum deformations of structure during operation─ Accidental Limit State (ALS): for example, effect of impacts due to ship collision ─ Fatigue Limit State (FLS): structure/pile to withstand accumulated damage in design life

Four acceptance/performance criteria for monopoles (DNV)─ vertical tangent criterion or “zero-toe-kick” criterion at monopile’s deflection curve ─ maximum lateral deflection at mudline 120 mm ─ maximum lateral deflection at pile toe 20 mm ─ maximum rotation at mudline of 0.50° (this includes installation imperfection of 0.25°)─ At present, there is no additional criteria for earthquake loading. It is expected that new

requirements will appear in standards, especially for vertical acceleration of turbine.

Load combinations for foundation designDNV – 5 load combinations for ULS limit sateMost codes do not specify how to combine earthquake loads with ULS loads. Germanischer Lloyd (GL, 2010) suggests using earthquake load from Eurocode 8 or API with return period of 475 years. Resulting earthquake load with a load factor of 1.0 to be combined with design wind load cases during operation (normal wind profile and turbulent model) + an additional load case in which earthquake load is combined with 80% of reference wind speed for parked (standstill) turbine.

Earthquake return periodsThere appears to be consensus on use of earthquakes with return period 500 years under ULS limit sate for OWTs. This is consistent with the two-tier design approach from ISO 19901-2 for earthquake analysis of offshore structures.However, collapse of accommodation platforms could involve loss of life, and should therefore also be designed for earthquakes with higher return periods (typically 2500 years) under ALS conditions following ISO.

DanTysk (Vattenfall)Horns Rev 2Alfa Ventus

Influence of soil/foundation on earthquake response

Response of OWT is strongly dependent on soil/foundation behavior Often advanced foundation models are necessary to predict dynamic and nonlinear response of OWTs => performance-based designThe following are cases where moderate-strong earthquakes could impact design. Two examples are presented to highlight this: ─ Determination of permanent tilt, especially jackets, monopods and tripods

NB: permanent foundation tilt due to loads 0.25° => performance-based design

─ Liquefaction effect on large-diameter piles─ Liquefaction effect on anchors of floating OWTs ─ Vertical shaking

Liquefaction and consequences

Liquefaction is the condition of pore pressure reaching total stress due to cyclic loading.

1. Assessment2. Consequences (sinking/tilting): important to

estimate in platform design3. Mitigation

Turkey 1999

Niigata, 1964

NZ, 2011

Assessment of liquefaction susceptibilityUse of empirical methods, for example by Seed, Robertson, etc. for silica sand (for example, using CPT data) based on CRR and earthquake-induced CSR (Cyclic Stress Ratio). CSR is computed by simple formula from ground surface PGA:

Use of plasticity models - a few promising models have recently been proposed and implemented in FE/FD codes.For special soil, like carbonate sand where empirical methods do not exist, use soil test data to estimate pore pressure and assess its impact on structure

max v0CSR=0,65τ σ ′ max d v0Ha rg

τ σ= ( )d cos 0,04r z=

Compaction with vibratory probes

21

Installation of Field Drains

Liquefaction mitigation (land applications)

Dynamic Compaction Method Explosives Compaction

22

Liquefaction mitigation (land applications)

Impact of liquefaction on foundationsPile foundations─ They have the advantage that the pile can still be designed safely by ignoring the

pile segment in liquefiable soil. ─ If liquefaction is close to surface, the impact on design is not major, because the

soil support at shallow depth is relatively small. ─ In thick liquefiable layer, there is the additional potential of sinking (leading to

tilt in jackets).

Suction piles and anchors─ Same principle as in piles, except that due to shorter length of

anchors, the anchor might miss a large percentage of its capacity

Gravity based foundation─ The main problem is tilt during liquefaction. If base is stiff

enough, it could even out the pressure under the base and minimize the tilt.

Permanent tilt of foundations (demonstration of mechanism)

Rated power 3.5 MWH0 = 90 m, Rotor diameter = 20 m, mass = 220 tonsTwo caisson configurations in uniform soil profile with su = 120 kPa & Gmax/su = 600 D=20 m & L/D =0.5 ; D=25 m & L/D=0.2 Wind and wave loads treated as static loads, 1 MN and 2 MN, respectivelyVon Mises failure criterion with kinematic hardening model in AbaqusExcitation: Takatori, Kobe, scaled to PGA = 0.35 g

Permanent tilt of foundationSimultaneous action of static lateral loads and earthquake shaking could lead to permanent lateral tilt. Therefore, accurate prediction of nonlinear response is critical.Computed tilt is 0.12° (∼ half of allowed 0.25°). Sensitivity analyses on soil parameters and earthquake record could result in larger tilt.Simpler models, like foundation macro-elements, could also be used.

Modern, large OWTs have relatively high natural periods in lateral direction - typically 3-3.5 s; therefore, they are not expected to be very vulnerable to horizontal earthquake shaking in areas with minor to moderate seismicity. On the other hand, they have low natural periods in axial direction which could result in large vertical response under vertical earthquake shaking. Presently, the are no standards for earthquake analysis requirements and performance of OWTs, and engineers often ignore earthquake loading on the basis of above argument.

Eurocode 8 for different ground

Vertical earthquake shaking

Following traditional earthquake structural analysis (no SSI), one should expect large vertical acceleration and stresses in tower and turbine due to vertical earthquake shaking.

Vertical earthquake shaking

Recent study (Kjørlaug and Kaynia, 2015) has confirmed this, and has also pointed out importance of considering SSI, including radiation damping, in reducing earthquake response.

Fact

or o

f 3Fa

ctor

of 2

No radiation damping

Summary and conclusionsOffshore wind industry seems to have a bright future and major growth. OWTs use different foundation supports each with their own challenges and large impact on design/cost.SSI (stiffness and damping) important for design and cost.Small allowable permanent foundation tilt (as low as 0.25%) calls for detailed nonlinear dynamic analyses => Performance-Based Design. Liquefaction is important for foundation design, especially anchors for floating OWTs.Some traditional soil-foundation analyses, such as p-y curves, are not applicable for large-diameter piles in OWTs. Kinematic interaction is also more critical due to small aspect ratio of foundations.

Thank you for your attention

感谢您的关注

29

@infoNGI

NORWEGIAN GEOTECHNICAL INSTITUTENGI.NO