You are on page 1of 56

This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18.

No further distribution shall be made. Requires a valid subscription to be used.


RECOMMENDED PRACTICE

DNV-RP-0585 Edition August 2021

Seismic design of wind power plants

The PDF electronic version of this document available at the DNV website dnv.com is the official version. If there
are any inconsistencies between the PDF version and any other available version, the PDF version shall prevail.

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
FOREWORD

DNV recommended practices contain sound engineering practice and guidance.

© DNV AS August 2021

Any comments may be sent by e-mail to rules@dnv.com

This service document has been prepared based on available knowledge, technology and/or information at the time of issuance of this
document. The use of this document by other parties than DNV is at the user's sole risk. DNV does not accept any liability or responsibility
for loss or damages resulting from any use of this document.
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
CHANGES – CURRENT

Changes - current
This is a new document.

Topic Reference Description

Rebranding to DNV, cross- All Some of the documents referred to may not yet have been
references rebranded. If so, please see the relevant DNV GL document.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 3


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Acknowledgements

Changes - current
This recommended practice is based on a former joint industry project called 'ACE' (Alleviating Cyclone and
Earthquake Challenges for Wind Farms).

Recommended practice — DNV-RP-0585. Edition August 2021 Page 4


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
CONTENTS

Contents
Changes – current.................................................................................................. 3
Acknowledgements................................................................................. 4

Section 1 General.................................................................................................... 7
1.1 Introduction......................................................................................7
1.2 Objective...........................................................................................7
1.3 Scope................................................................................................ 7
1.4 Application........................................................................................ 7
1.5 References........................................................................................ 7
1.6 Definitions and abbreviations........................................................... 9

Section 2 Design principles for wind power plant assets...................................... 17


2.1 General........................................................................................... 17
2.2 Wind turbine, support structure incl. tower....................................17
2.3 Wind turbine, blades.......................................................................19
2.4 Wind turbine, machinery................................................................ 20
2.5 Offshore substations.......................................................................20
2.6 Offshore power cables.................................................................... 21
2.7 Installation vessels.........................................................................22
2.8 Other assets................................................................................... 22
2.9 Seismic analysis flowchart for wind turbines..................................22

Section 3 Seismic loading on wind power plants.................................................. 25


3.1 General........................................................................................... 25
3.2 Site-specific seismic hazard analysis.............................................. 25
3.3 Selection of seismic ground motions for time domain analysis........29
3.4 Seismic site response analysis........................................................29
3.5 Ground motion input to seismic design.......................................... 31
3.6 Seismic design load cases (DLCs)...................................................32

Section 4 Modelling approaches in seismic design................................................ 35


4.1 General........................................................................................... 35
4.2 Time domain analysis..................................................................... 35
4.3 Response spectrum analysis method (frequency domain).............. 42

Section 5 Geotechnical aspects............................................................................. 44


5.1 Liquefaction.................................................................................... 44
5.2 Damping in seismic design............................................................. 46

Recommended practice — DNV-RP-0585. Edition August 2021 Page 5


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
5.3 Soil-structure interaction modelling............................................... 52

Contents
Section 6 Post-earthquake considerations............................................................ 53
6.1 General........................................................................................... 53
6.2 Aftershocks..................................................................................... 53
6.3 Persistent liquefaction.................................................................... 53
6.4 Mitigation measures....................................................................... 54

Changes – historic................................................................................................ 55

Recommended practice — DNV-RP-0585. Edition August 2021 Page 6


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 1 GENERAL

1.1 Introduction
This recommended practice (RP) provides principles, technical recommendations and guidance for the seismic
design of wind power plants.
The emphasis is put on the best practice for designing to safeguard against complex natural phenomena,
such as earthquakes. This RP has been created because it is recognized that earlier design guidance and
standards are insufficient for seismic design or could be interpreted in different ways.
This recommended practice supplements DNV-ST-0126, DNV-ST-0145 and DNV-ST-0437.
This recommended practice has been written for worldwide application. National and governmental
regulations may include requirements in excess of the provisions stated in this recommended practice
depending on the size, type, location and intended service of the wind power plant.

1.2 Objective
The objectives of this recommended practice are to:
— provide an internationally acceptable level of safety by defining criteria for seismic designs, in combination
with other referenced standards, recommended practices, and guidelines
— serve as a reference document between suppliers and purchasers related to design, construction, and
installation
— serve as a guideline for designers, suppliers, purchasers, and regulators
— recommend procedures and criteria for seismic analysis and seismic design
— serve as a basis for verification and certification of seismic design.

1.3 Scope
This recommended practice gives recommendations for the following:
— seismic design return periods
— determination of seismic loading
— geotechnical considerations
— seismic analysis methods and procedures
— post-earthquake actions and mitigation.

1.4 Application
The recommended practice is applicable to both onshore and offshore wind power plants and their related
assets.
The recommended practice is applicable to the design of complete structures, including towers, substructures
and foundations, and other wind turbine (WT) components such as nacelles and rotors as well as offshore
substations, power cables and installation vessels.

1.5 References

1.5.1 General
Table 1-1 lists normative references used in this document.
The standards in Table 1-1 include provisions which, through reference in this text, constitute provisions of
this recommended practice. Current editions/revisions of the documents shall apply.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 7


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Table 1-1 Normative standards and guidelines

Document code Title

DNV-RP-0360 Subsea power cables in shallow water

DNV-RP-C212 Offshore soil mechanics and geotechnical engineering

DNV-ST-0126 Support structures for wind turbines

DNV-ST-0145 Offshore substations

DNV-ST-0359 Subsea power cables for wind power plants

DNV-ST-0376 Rotor blades for wind turbines

DNV-ST-0437 Loads and site conditions for wind turbines

DNV-ST-N001 Marine operations and marine warranty

IEC 61400-1 Wind energy generation systems - Part 1: Design Requirements

IEC 61400-3-1 Wind energy generation systems - Part 3-1: Design requirements for fixed
offshore wind turbines

IEC TS 61400-3-2 Wind energy generation systems - Part 3-2: Design requirements for floating
offshore wind turbines

IEC 61400-4 Wind turbines - Part 4: Design requirements for wind turbine gearboxes

IEC 61400-5 Wind energy generation systems - Part 5: Wind turbine blades

IEC 61400-6 Wind energy generation systems - Part 6: Tower and foundation design
requirements

IEC PT 61400-8 Wind energy generation systems - Part 8: Design of wind turbine structural
components

1.5.2 Informative references


Table 1-2 lists other references used in this document.
The documents in Table 1-2 include acceptable methods for fulfilling the requirements in the standards. Other
recognized codes or standards may be applied provided it is shown that they meet or exceed the level of
safety of the normative standards and guidelines, and this recommended practice.

Table 1-2 Informative standards and guidelines

Document code Title

ASCE 7 Minimum Design Loads and Associated Criteria for Buildings and Other
Structures

CNS 15176-1 Wind turbines - Part 1: Design requirements

DNV-RP-C207 Statistical representation of soil data

DNV-ST-C501 Composite components

EN 1998-1 Eurocode 8: Design of structures for earthquake resistance - Part 1: General


rules, seismic actions, and rules for buildings

Recommended practice — DNV-RP-0585. Edition August 2021 Page 8


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Document code Title

EN 1998-5 Eurocode 8: Design of structures for earthquake resistance Part 5: Foundations,


retaining structures and geotechnical aspects

ISO 19901-2 Petroleum and natural gas industries - Specific requirements for offshore
structures - Part 2: Seismic design procedures and criteria

ISO 19902 Petroleum and natural gas industries - Fixed steel offshore structures

ISO 19905-1 Petroleum and natural gas industries - Site-specific assessment of mobile
offshore units - Part 1: Jack-ups

ISO 2394 General principles on reliability for structures

1.6 Definitions and abbreviations

1.6.1 Definition of verbal forms


The verbal forms defined in Table 1-3 are used in this document.

Table 1-3 Definition of verbal forms

Term Definition

shall verbal form used to indicate requirements strictly to be followed in order to conform to the
document

should verbal form used to indicate that among several possibilities one is recommended as
particularly suitable, without mentioning or excluding others

may verbal form used to indicate a course of action permissible within the limits of the document

1.6.2 Definition of terms


The terms defined in Table 1-4 are used in this document.

Table 1-4 Definition of terms

Term Definition

accidental limit states ensure that the structure resists accidental loads and maintains its integrity and performance
if subject to local damage or flooding

Atterberg limits basic measure of the critical water contents of a fine-grained soil: its shrinkage limit, plastic
limit, and liquid limit

basic design standard standard(s) from Table 1-2 selected as a basis for the design in combination with this
standard

characteristic resistance reference value of a structural strength to be used in the determination of the design
resistance
The characteristic resistance is normally based upon a 5% quantile in the lower tail of the
distribution function for resistance.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 9


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Term Definition

characteristic value representative value of a load variable or resistance variable


For a load variable, the characteristic value is a high but measurable value with a prescribed
probability of not being unfavourably exceeded during some reference period.
For a resistance variable, it is a measurable value with a prescribed probability of being
favourably exceeded.

classification notes cover proven technology and solutions which are found to represent good practice by DNV,
and which represent one alternative for satisfying the requirements stipulated in the DNV
rules or other codes and standards cited by DNV

de-aggregation process in the probabilistic seismic hazard assessment to identify the relative contributions to
the overall seismic hazard from earthquakes of various distances and magnitudes

design basis document defining the owner's requirements and conditions to be taken into account in the
design and in which any requirements in excess of this standard should be given

design life period of time over which the structure in question is designed to provide an acceptable
minimum level of safety, i.e. the period of time over which the structure is designed to meet
the requirements set forth in this standard

design value value to be used in the deterministic design procedure, i.e. characteristic value modified by
the resistance factor or load factor, whichever is applicable

ductility ability of a structure system or its component to deform into the non-linear range without
sudden failure

expected value mean value, e.g. the mean value of a load during a specified time period

fatigue degradation of the material caused by cyclic loading

fatigue limit states related to the possibility of failure due to the cumulative damage effect of cyclic loading

foundation foundation of a support structure for a wind turbine is in this document reckoned as a
structural or geotechnical component, or both, extending from the seabed downwards

guidance note information in the standards in order to increase the understanding of the requirements

limit state state beyond which the structure no longer satisfies the requirements

liquefaction phenomenon of soil losing its shear strength due to pore pressure build-up from earthquake
excitations

magnitude measure of the earthquake strength

mean statistical mean over observation period

mean water level mean level between highest astronomical tide and lowest astronomical tide

metocean abbreviation of meteorological and oceanographic

offshore wind turbine structural system consisting of a support structure for an offshore wind turbine and a
structure foundation for the support structure

partial safety factor method associated with load and resistance factor design (LRFD) where uncertainties in loads
method are represented by a load factor, and uncertainties in strengths are represented by a material
or resistance factor

pile head position along a foundation pile that is level with the seabed
This definition applies regardless of whether the pile extends above the seabed.

primary steel structural parts where failure has substantial consequences (e.g. tower, monopile, flanges)

Recommended practice — DNV-RP-0585. Edition August 2021 Page 10


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Term Definition

probabilistic seismic system of methodology based on probability and seismology to derive the design ground
hazard analysis motion intensity at a site

purchaser owner or another party acting on the owner's behalf that is responsible for procuring
materials, components or services intended for the design, construction, or modification of a
structure

redundancy ability of a component or system to maintain or restore its function when a failure of a
member or connection has occurred
Redundancy may be achieved for instance by strengthening or introducing alternative load
paths.

reliability ability of a component or a system to perform its required function without failure during a
specified time interval

response spectrum ground motion intensity representation in the frequency domain, obtained from a series of
single degree of freedom structural dynamic analysis of different periods

response spectrum structural dynamic analysis using modal superposition techniques


analysis method

risk qualitative or quantitative likelihood of an accidental or unplanned event occurring considered


in conjunction with the potential consequences of such a failure
In quantitative terms, risk is the quantified probability of a defined failure mode times its
quantified consequence.

rotor-nacelle assembly part of wind turbine carried by the support structure

secondary steel structural parts where failure will be without significant consequence (e.g. ladders, platforms,
railings, boat landings and J-tubes)

serviceability limit states imply deformations in excess of tolerance without exceeding the load-carrying capacity, i.e.,
they correspond to tolerance criteria applicable to normal use

seismic site response analysis to propagate earthquake time histories from the bed rock vertically upward to the
analysis ground surface

superelement numerical tool to condense mass, stiffness, damping and load information in order to
facilitate separate analyses by the designers of the foundation and turbines

support structure of wind structure below the yaw system of the rotor-nacelle assembly which includes the tower
turbine structure, sub-structure, and foundation, see also Figure 3-2

target safety level nominal acceptable probability of structural failure

time domain analysis structural dynamic analysis using earthquake time histories in the time domain
method

tower structural component, which forms a part of the support structure for a wind turbine,
usually extending from somewhere above the still water level to just below the rotor-nacelle
assembly of the wind turbine

ultimate limit states correspond to the limit of the capacity, i.e. to the maximum load-carrying resistance

Figure 1-1 summarizes the various wind power plant assets and their components, while Figure 1-2 provides
a brief definition of wind turbine components.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 11


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.

Page 12
Figure 1-1 Wind power plant assets and their components

DNV AS
Recommended practice — DNV-RP-0585. Edition August 2021
Seismic design of wind power plants
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Figure 1-2 Definition of wind turbine components

1.6.3 Symbols
1.6.3.1 Latin characters
C1 = consequence category (ISO 19902)
CR = seismic reserve capacity factor
H = wave height
Hmax = maximum wave height
HS = significant wave height
L1 = exposure level (ISO 19902)
Pf = nominal annualized probability of failure
su = undrained soil strength
V = volume
V1 = reference wind speed with one-year recurrence period
Vref = reference wind speed with 50-year recurrence period
Vin = cut-in wind speed
Vout = cut-out wind speed
Vhub = 10-minute average wind speed at hub height
W = energy dissipation in a loading cycle.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 13


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
1.6.3.2 Greek characters
= log decrement damping ratio
= shear strain
= strain
= material factor
= damping ratio.

1.6.3.3 Subscripts
c = characteristic value
d = design value
k = characteristic value
p = plastic
y = yield.

1.6.4 Abbreviations
The abbreviations described in Table 1-5 are used in this document.

Table 1-5 Abbreviations

Abbreviation Description

1-D one dimensional

2-D two dimensional

3-D three dimensional

ACE alleviating cyclone and earthquake

ASCE American Society of Civil Engineers

ALARP as low as reasonably practicable

ALE abnormal level earthquake

ALS accidental limit state

API American Petroleum Institute

BE bender element

CNS Chinese National Standards

COD co-directional

CPT cone penetration test

CRR cyclic resistance ratio

CSR cyclic stress ratio

DLC design load case

DSHA deterministic seismic hazard analysis

EC eurocode

ELE extreme level earthquake

Recommended practice — DNV-RP-0585. Edition August 2021 Page 14


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Abbreviation Description

FLS fatigue limit state

GMPE ground motion prediction equation

IEC International Electrotechnical Commission

ISO International Organization for Standardization

JIP joint industry project

LAT lowest astronomical tide

LRFD load and resistance factor design

MCE maximum considered earthquake

MUL multi-directional

MWL mean water level

NCM normal current model

NTM normal turbulence model

NWLR normal water level range

NWP normal wind profile model

OWT offshore wind turbine

OSS offshore substation

PGA peak ground acceleration

PSHA probabilistic seismic hazard analysis

PSM partial safety factor for material

RC resonant column

RNA rotor-nacelle assembly

RP recommended practice

RSM response spectrum analysis method

SA spectral acceleration

SCADA supervisory control and data acquisition

SDOF single degree of freedom

SE superelement

SLS serviceability limit state

SPT standard penetration test

SRC seismic risk category

SSRA seismic site response analysis

UHS uniform hazard spectrum

ULS ultimate limit state

Recommended practice — DNV-RP-0585. Edition August 2021 Page 15


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Abbreviation Description

UNI uni-direction

UP ultrasonic pulse

WT wind turbine

Recommended practice — DNV-RP-0585. Edition August 2021 Page 16


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 2 DESIGN PRINCIPLES FOR WIND POWER PLANT ASSETS

2.1 General
The seismic design of wind power plants should follow a limit-state-based approach, see [1.6] for a definition
of the limit state term. This recommended practice considers the following four limit state groups:
— Fatigue limit state (FLS), which corresponds to the lifetime capacity of an asset and the accumulated
effect of the repetitive actions that the asset would likely experience.
— Serviceability limit state (SLS), which is determined by various limiting values oriented towards the
normally envisaged use of the wind power plant asset, such as deformations or movements, motions,
vibration amplitudes and accelerations, crack widths and water tightness.
— Ultimate limit state (ULS), which corresponds to the maximum load-carrying resistance of an asset and is
often associated with collapse or other similar forms of structural failure.
— Accidental limit state (ALS), which is associated with events of short duration and significant magnitude
that are unlikely to occur on a given asset during its design life. This is often subdivided into the structural
damage caused by accidental loads, and the resistance and integrity of damaged components.
Guidance note:
Although IEC 61400-1 Annex B does not specify a limit state designation for the seismic design scenarios in the main text, the two
earthquake-related DLCs, 1.8 and 6.7, are associated with analysis type 'U' and partial safety factors 'N', indicating that both DLCs
are treated as ULS. It specifies a single 475-year return period earthquake with a partial load factor of 1.0 and partial material
factor for steel of 1.0.
In the seismic design of wind power plants, it is important to acknowledge the trade-off between risk and economic costs. This
implies that not all design philosophies and requirements from standards intended for other industries are applicable to wind power
plant design.
In IEC 61400-1, there is a term called 'engineering integrity' used for the seismic events only. The term 'engineering integrity'
has been mainly introduced to distinguish between active systems (such as wind turbines) and passive systems (such as civil
engineering structures typically prescribed in building codes). Wind turbines are in essence much more complex dynamic
structures than simple buildings because of the many mechanical components that could be affected. The correct response to a
seismic event should be ensured, considering personnel safety as the most important issue. 'Engineering integrity' is not to be
confused with 'integrity engineering' which is a well-defined term in other disciplines.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
[2.2] to [2.8] describe the recommended seismic return periods and associated limit state considerations for
various wind power plant assets and wind turbine components.

2.2 Wind turbine, support structure incl. tower


— FLS: the fatigue resistance of the support structure is normally not required to be verified under the
seismic design situation. However, in the case of plastification under ULS conditions, investigations into
the potential impact on fatigue resistance should be performed following the guidance in DNV-ST-0126.
— SLS: the accumulated displacement/tilt is normally a geotechnical consideration for WT support
structures. The general design criteria may follow the damage limitation state considerations in EN 1998-1
and EN 1998-5. An earthquake return period of 95-year may be used as the default case. If applicable
national or regional design codes prescribe specific SLS return period(s) to consider in the design, they
shall be considered as the governing value. Characteristic material and best estimate soil properties may
be used for this SLS check, and all partial load and material factor values should be taken as 1.0. Potential
earthquake-induced liquefaction effects under the SLS earthquake return period shall be considered.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 17


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Guidance note 1:
The 95-year return period defined in the Eurocode 8 series is consistent with the typical design life of buildings, which often
exceeds 50 years. Support structures for wind turbines, on the other hand, tend to have a design life of 30 years or less.
Therefore, the project may consider a lower design earthquake return period for SLS in relation to the actual design life of the
structure.
In addition to the geotechnical displacement/tilt consideration mentioned above, the SLS may also be interpreted as the 'damage
limitation' requirement for structures stated in EN 1998-1 at 95-year return period. For EN 1998-1 and other similar onshore
building standards, the 'design earthquake' is typically defined at 475-year return period. At this level, significant plasticity/
buckling is allowed for targeted structural components (for example X-braces) but other structural components (for example
columns) should remain linearly elastic. In addition to this ductility design requirement at 475-year return period, EN 1998-1
requires no significant plasticity/buckling to be allowed on any part of the structure at 95-year return period.
If the SLS design criteria can be fulfilled during the ULS check using an earthquake return period of 475-year, it will certainly pass
the same check using an earthquake return period of 95 years or less.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

— ULS: to be consistent with IEC 61400-1, an earthquake with a return period of 475-year should be used,
along with a partial load factor of 1.0 and partial material factor for steel of 1.0.
Guidance note 2:
For wind power plants, the tower and most primary support structure members may fail by buckling and hence have little or no
reserve capacity. Hence, their design principles should resemble those of columns rather than of X-braces when comparing with EN
1998-1. Linear behaviour is expected at the 475-year level, and main structural members are to be checked to ensure they meet
the strength and stability criteria without the occurrence of large plastification.
Typically, characteristic strength parameters are selected as the 95% fractile. As a result, a structure designed for a 475-year
-4 -3
seismic event using ULS has a nominal annualized probability of failure, , in the range of between 10 and 10 .

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
For a geotechnical design that uses best estimate (or other statistically similar) soil properties, ULS material
factors for soil strengths may be used for foundation sizing. Alternatively, an appropriate statistical evaluation
may be adopted to derive project-specific material factor values following the principles outlined in DNV-
RP-C207. The statistical evaluation should target an appropriate quantile level so that, when combined with
-4
the design earthquake, the design is consistent with an overall pf not exceeding 5x10 . When using the
statistical approach, for each soil layer contributing to overall soil resistance, the statistical distribution of its
derived characteristic properties, e.g. undrained shear strengths and friction angles, should be assessed for
their variance. The seismic foundation capacity design should be based on location-specific soil profiles but
adjusted for the observed statistical variance in each soil layer or unit. The same statistical approach may
also be applied for any other material whose characteristic strength does not conform to the standard 95%
survival probability as per IEC 61400-1.
Guidance note 3:
-4
DNV-ST-0126 specifies a pf of 10 as the target safety level for the normal safety class. It further specifies a maximum acceptable
-4
pf of 5x10 . These targets are consistent with the principles stated in IEC 61400-1 and ISO 2394.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
For a geotechnical design, potential earthquake-induced liquefaction effects under the ULS earthquake return
period shall be considered in combination with the appropriate material partial factors.
Guidance note 4:
While considering liquefaction effects is conservative for seismic foundation capacity design under ULS conditions, the same is not
true for ULS loading calculations, where the non-liquefied scenario tends to produce higher loads. Therefore, both liquefied and
non-liquefied scenarios should always be considered. Further guidance on formulating soil reaction springs is included in [5.3].
Potential earthquake-induced liquefaction effects should also be considered for SLS and ALS designs.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

Recommended practice — DNV-RP-0585. Edition August 2021 Page 18


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
— ALS: this is not normally required for WT support structures except in specific cases where an increased
safety demand is deemed necessary, see the guidance note below.
Guidance note 5:
The criteria to define whether an increased safety demand applies may be, but are not limited to:

— the WT is expected to be manned regularly for a significant period


— the WT is in close vicinity to sensitive infrastructure (e.g. a motorway or refinery)
— the wind farm is considered critical for supporting grid stability after a major earthquake.

When an increased safety demand is necessary, an ALS check should be performed to establish that:

— the WT does not globally collapse during the ALS earthquake


— the structural integrity of potentially manned areas is maintained during the ALS earthquake
— the state of the WT during and after the ALS earthquake does not pose a safety threat to the general public.

Plasticity and damage to WT assets are allowed under this ALS check if the above three criteria are met. The analysis should
be performed in accordance with the ALE assessment procedures in ISO 19901-2 and ISO 19902. All partial load and material
factor values should be taken as 1.0. Potential earthquake-induced liquefaction effects should also be considered.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

2.3 Wind turbine, blades


The main design standards for wind turbine rotor blades are DNV-ST-0376 and IEC 61400-5.
Guidance note:

Although DNV-ST-0376 does not specifically cover any seismic design, it specifies for rotor blades, quoting DNV-

ST-C501, where relates to a composite structure whose failure implies a low risk of human injury and minor

environmental and economic consequences. Therefore, in principle, the seismic design of rotor blades should be no different to
that of WT support structures. However, rotor blades are repairable/replaceable, and localized blade failure does not automatically
pose a threat to the overall WT or to human life.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The following recommendations are made regarding the seismic design of rotor blades:
— FLS: same as that in [2.2].
— SLS: same as that in [2.2]. In addition, SLS acceptance criteria as per DNV-ST-0376 shall be followed.
— ULS: the default case is the same as for WT support structures and towers [2.2]. However, if
seismic design becomes governing and the cost of making the blades seismically robust becomes
disproportionately large, it is permissible to allow for some plasticity or other types of structural damage
under the 475-year ULS check. A lower return period level ULS check is permitted for blades. This level
of return period should be determined by the national or regional design codes if available, or by applying
the as low as reasonably practicable (ALARP) principle with 95 years as a recommended minimum.
When the ULS seismic design return period is lowered, the project should notify all stakeholders that the
nominal reliability of the blades has deviated from the base case, so that the cost, maintenance, and
operation models are adjusted correspondingly.
In addition, it shall be ensured that a localized failure of the blades does not negatively affect the reliability
of other WT components such as the tower, the support structure, and the rotor-nacelle assembly (RNA).
Furthermore, operational procedures should ensure that environmental consequences remain minor should a
blade failure occur. The assessment may be conducted in a risk identification and mitigation format.
— ALS: this is not normally required for rotor blades except in specific cases where an increased safety
demand is deemed necessary. See [2.2] for additional guidance on ALS considerations. In addition, when
an increased safety demand is considered necessary, machinery failure under ALS earthquake conditions
should not trigger a chain of events leading to the global collapse of the WT.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 19


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
2.4 Wind turbine, machinery
The main design standards for for wind turbine machinery are DNV-ST-0361, IEC 61400-4 and IEC PT
61400-8.
Guidance note:
The target reliability of DNV-ST-0361 is aligned with that of IEC. Machinery components and structures are highly important assets
within the WT and hence should, in principle, target a similar as that for WT support structures and rotor blades. However,
many machinery components are also repairable/replaceable like rotor blades in this regard.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The following recommendations are made regarding the seismic design of machinery components and
structures:
— FLS: same as that in [2.2].
— SLS: same as that in [2.2]. In addition, SLS checks as per DNV-ST-0361 shall be followed.
— ULS: the default case is the same as for the WT support structure and tower [2.2].
Similar to blades [2.3], if seismic design becomes governing and the cost of making the machinery
components and structures seismically robust becomes disproportionately large, it is permissible to
allow for some plasticity or other types of structural damage under the 475-year ULS check. A lower
return period level seismic ULS check is permitted for machinery. This level of return period should be
determined by the national or regional design codes if available, or by applying the as low as reasonably
practicable (ALARP) principle with 95 years as a recommended minimum.
When using this approach, the project should notify all stakeholders that the nominal reliability of the
machinery components and structures has deviated from the base case, so that the cost, maintenance,
and operation models are adjusted accordingly.
The design shall ensure that the failure of machinery components or structures does not negatively affect
the reliability of other WT components such as the tower, support structure, and blades.
In addition, operational procedures should ensure that the environmental consequences remain minor
should a machinery-related failure occur. The assessment may be conducted in a risk identification and
mitigation format.
— ALS: this is not usually required for machinery components and structures. In cases where an increased
safety demand is deemed necessary, this may be considered as an additional check to be performed.
If an increased safety demand is considered necessary, machinery failure under an abnormal level
earthquake should not trigger a chain of events leading to the collapse of the tower and/or support
structure.
See [2.2] for additional guidance for ALS considerations. In addition, when an increased safety demand is
being considered, machinery failure under ALS earthquake conditions should not trigger a chain of events
leading to the global collapse of the WT.

2.5 Offshore substations


For the earthquake design of offshore substations (OSS), the main applicable international standard is DNV-
ST-0145. The design process of OSS shares many similarities with that of WT support structures and towers,
except that the safety class high should be targeted.
DNV-ST-0145 [4.2.2] specifies that the target safety level for the design of a substation structure and its
foundation is , aimed at structures whose failures are ductile with reserve capacity.
Alternatively, it also allows the consequence category according ISO 19902, in which case the OSS should be
designed to high consequence category C1 and exposure level L1.
Regarding seismic design, DNV-ST-0145 categorizes earthquake loading as one type of environmental load
and requires that the structure and its foundation be designed to withstand the earthquake loads using
guidance in DNV-ST-0126, i.e. WT support structures.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 20


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Notwithstanding, key differences between the OSS and a WT foundation are that the OSS exists offshore
only, and could be permanently manned structures depending on the operational concept.
The following recommendations are made regarding the seismic design of OSS:
— FLS: same as that in [2.2].
— SLS: not applicable for seismic design, as the primary concern for offshore manned structures under
earthquake action is safety.
— ULS: the design performance criteria and process shall be based on ISO 19901-2 and ISO 19902
considering exposure level L1 structures during an extreme level earthquake (ELE).
Both the simplified and detailed action procedures within ISO 19901-2 are permissible, depending on the
seismic risk category (SRC). The simplified action procedure refers to the seismic zonation maps, whereas
the detailed action procedure refers to site-specific seismic hazard analysis.
In ISO 19901-2 and ISO 19902, the determination of the ELE return period depends on the implied
ductility level of the structure. These two standards should be used for such load and return period
determination.
Guidance note 1:
Experience shows that, for structures with the highest ductility considered in ISO 19902, the ELE return period is in the range
of 300 to 400 years. For structures with less ductility, for example tripod, non-X-bracing configuration, slender members, etc.,
the ELE return period is higher and approaching the ALE return period.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
— ALS: the design performance criteria and process shall be based on ISO 19901-2 and ISO 19902
considering exposure level L1 structures during an abnormal level earthquake (ALE).
An ALS check during abnormal level earthquakes typically involves non-linear time history analyses.
Jacket braces are allowed to buckle to absorb seismic energy. The ALS survival criterion should be that the
stability of the structure is maintained at the end of the time history analysis.
X-braces are the preferred jacket bracing configuration under seismic loads. Significant buckling and
plasticity in X-braces are allowed. For jacket legs and piles, no buckling should occur under ALE loads.
Only minimal plastification is allowed for legs and piles under ALE loads.
The minimum ALE return period shall be 2,500 years.
Guidance note 2:
Experience shows that, when ISO 19901-2 is followed, the ALE return period is typically in the range of 3,000 to 4,000 years.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

2.6 Offshore power cables


The main design standard and recommended practice for offshore power cables are DNV-ST-0359 and DNV-
RP-0360.
In both of these, an earthquake is classified as a hazard relevant for cable protection design. Experience from
both the cable and offshore pipeline industries shows that earthquake loadings, i.e. accelerations, pose little
threat to either buried or exposed cables or pipelines. Instead, earthquake-induced displacement poses a
valid threat in terms of:
— localized fault movements
— liquefaction-induced floatation or settlement of cables and/or cable protection systems
— liquefaction-induced differential settlement between the WT and cables
— slope instability and runout, submarine landslide and other geohazards.
These conditions should be assessed under the 475-year return period seismic load. Similar to [2.2], for
a geotechnical design that uses best estimate (or other statistically similar) soil properties, ULS material
factors on soil strengths may be used. Alternatively, an appropriate statistical evaluation may be adopted to
derive project-specific material factor values following the principles outlined in DNV-RP-C207.
The statistical evaluation should target an appropriate quantile level so that, when combined with the design
-4
earthquake, the design is consistent with an overall pf not exceeding 5x10 .

Recommended practice — DNV-RP-0585. Edition August 2021 Page 21


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
2.7 Installation vessels
The main standard for marine operations of installation vessels is DNV-ST-N001. Note that this standard
covers marine operations for a specific project, rather than for a specific installation vessel throughout its
operational life.
The recommendations provided in this subsection are more relevant to jack-up types of installation vessels.
There are no specific provisions for earthquakes in DNV-ST-N001. Therefore, a conservative approach is to
treat the installation vessel as an equivalent manned offshore installation, i.e. following the ISO 19901-2
approach assuming exposure level L1 and an effective in-service life of approximately 25 years. This
approach is highly conservative because:
— The construction period for an offshore wind power plant typically spans a few months within a typical
year.
— The same installation vessel does not necessarily work continuously in seismically active regions
throughout its operational life.
An alternative approach for installation vessels is based on ISO 19905-1, which mentions that jack-up
vessels shall be assessed for 1,000-year earthquake events. This is also conservative for the same reasons
as those mentioned above.
Guidance note 1:
In ISO 19905-1, the jackup is first checked for ULS under the 1,000-year return period. If the structure does not pass this ULS
check, non-linear time domain analysis may be used to demonstrate no collapse.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
To reduce over-conservatism, this document allows the performance-based seismic design of installation
vessels. The design principles may follow the ISO 19901-2 framework, assuming exposure level L1, but use
the maximum envisaged cumulative offshore construction window within the project instead of the vessel’s
service life to derive the applicable ALE and ELE return periods. However, the resulting ELE return period shall
not fall below 95 years or the design SLS return period of other key wind power plant assets, whichever is
lower.
Guidance note 2:
For example, if the cumulative offshore construction window is 12 months for one project, then in order to achieve the same life-
-4
time probability of exceedance for L1 structures (4x10 per year × 25 years = 1%), an ALE target return period of 100 years is
-4
applicable. The pf for the vessel would remain at 4x10 during the short exposure period within the project if the ISO 19901-2
framework is followed.
It is worth noting that most construction vessels, such as jack-ups, have little or no static reserve strength or ability to
sustain large non-linear deformations, meaning that the applicable seismic reserve capacity factor, , should take a value of
approximately 1.0. In this case, the ELE target return period becomes identical to that of ALE.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The approach described above balances the seismic risk between project owners and installation contractors.
All stakeholders should come together to reach an agreement on the applicable return period. Other criteria
may also be defined provided they are based on sound engineering judgement.

2.8 Other assets


For all other wind power plant assets, such as met masts that are not covered by [2.2] to [2.7], it is
recommended to involve the certification body and other relevant stakeholders early on to decide on an
appropriate set of earthquake performance criteria and certifiable design procedures.

2.9 Seismic analysis flowchart for wind turbines


To assist in making seismic design choices, a flowchart is illustrated in Figure 2-1. This flowchart is
based on the principle of comparing ULS earthquake load effects with those from other ULS loads, and

Recommended practice — DNV-RP-0585. Edition August 2021 Page 22


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
hence is applicable for WT components whose ability to resist external loads is not negatively affected by
earthquakes. For geotechnical foundation design, as a counter-example, earthquake-induced excess pore
pressures and loading cycles may severely impact the foundation capacity. This means that geotechnical
foundation sizing may still be governed by the seismic case, even though seismic loads are lower than other
ULS loads. Additional guidance is provided in Sec.5.
Guidance note:
When comparing the seismic load effects with those from other ULS loads, one may consider the bending moments and shear
forces experienced by the transition piece cross-section together with the base shear and overturning moments at the mudline
cross-section in order to judge whether the seismic risk is low, moderate or high.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
If the seismic risk is very low (box c), it is permitted to consider no further actions. This applies when the site
seismicity is lower than the required level in the local building code. The definition in ISO 19901-2 may also
be used.

Figure 2-1 A flowchart for seismic design considerations

For low seismicity, a simplified structural dynamic analysis method such as the response spectrum analysis
method (RSM) is permitted (box d). Note that for onshore wind turbines RSM is generally permitted. For
offshore wind turbines, the time domain analysis method is preferred, but RSM is permitted for low seismic
regions.
For moderate to high seismicity, the time domain analysis method should be considered (box f). This is
due to the higher accuracy of the time domain analysis method. Experience also shows that in general the
time domain analysis method produces less conservative results, especially when seven or more sets of
earthquakes are considered.
When the seismic load is higher than other load cases and becomes dominant (box h), the structural
component should now be checked under the 475-year seismic loads as the ULS condition. For tower and
foundation structural components, significant non-linear behaviour is not allowed.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 23


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
For structural components that do not pass the 475-year seismic check, it is permitted to consider a lower
return period (box i and box j) for the ULS check. This only applies to blades [2.3] and machinery [2.4].
In box j it is stated that the 'structural integrity' at a level 1 earthquake (i.e. 475-year) should be verified,
and non-linear behaviour is allowed for this check. The verification is such that the components under
consideration do not fail completely at the 475-year level. For the main structural components, a 475-year
ULS check shall be carried out as stated above.
Note that the illustrated flowchart in Figure 2-1 only applies to WT. For offshore substations [2.5], the ISO
19901-2 and ISO 19902 seismic design approach should be followed.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 24


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 3 SEISMIC LOADING ON WIND POWER PLANTS

3.1 General
Seismic analysis of structural/mechanical systems requires the seismic loading input to be defined, and the
modelling of these systems. Seismic loading is typically defined by the seismic response spectrum, which
represents the seismic environmental condition at the design location. This is analogous to the wind/wave
design load cases where wind/wave design spectra are required.
Guidance note:
The seismic environment is typically defined by a design response spectrum, which represents the maximum structural response of
a single degree of freedom (SDOF) system. The response spectrum is frequency dependent because the SDOF can vary from very
stiff (high frequency) to very soft (low frequency).
The definition of the response spectrum lends itself to the structural response spectrum analysis method, which efficiently
calculates the maximum structural responses of a multiple-degree-of-freedom system.
The response spectrum analysis method is a recognized structural dynamic method. However, its role in wind turbine seismic
analysis is limited in IEC 61400-3 and IEC 61400-6. These two standards encourage the use of the time domain simulation
method.
Note that even when time domain simulations are used for structural calculations, the ground response spectrum is still required
as a definition of the seismic design environment. For time domain simulations, multiple sets of time histories compatible with the
design response spectrum are selected/generated, and then used for time domain structural simulations.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The design seismic response spectra should refer to applicable local standards wherever possible. If
insufficient guidance is given in these standards, the preliminary design and screening may refer to available
seismic hazard maps covering the site or region of interest, such as those given in ISO 19901-2. For detailed
design, a site-specific seismic hazard analysis should be performed for sites with moderate seismicity or
above.

3.2 Site-specific seismic hazard analysis


For the design of onshore wind turbine generator (WTG), the seismic hazard is typically given in local building
codes and no site-specific seismic hazard analysis is required. For offshore WTG, in general the seismic
design requires site-specific input.
There are two broad methods for performing site-specific seismic hazard analysis: deterministic and
probabilistic. Deterministic seismic hazard analysis (DSHA) looks for a reasonable worst-case ground motion
event near the site of interest, whereas the probabilistic seismic hazard analysis (PSHA) incorporates seismic
source and ground motion uncertainties quantitatively. Either method may be adopted if permitted by the
applicable local standards. If insufficient guidance is given in these standards, PSHA is preferred.
At its most basic level, PSHA is composed of five steps, as illustrated in Figure 3-1:
1) Identify all earthquake sources capable of producing damaging ground motions.
2) Characterize the distribution of earthquake magnitudes (the rates at which earthquakes of various
magnitudes are expected to occur).
3) Characterize the distribution of source-to-site distances associated with potential earthquakes.
4) Predict the resulting distribution of ground motion intensity as a function of earthquake magnitude,
distance, etc.
5) Combine uncertainties in earthquake size, location, and ground motion intensity using total probability
theorem calculations.
The result of these calculations is a full distribution of levels of ground shaking intensity, and their associated
rates of exceedance.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 25


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Figure 3-1 Schematic illustration of the basic five steps in probabilistic seismic hazard analysis

To fully utilize the information provided by those calculations, several extensions are often used:
1) Perform de-aggregation to identify the relative contributions from earthquakes of various distances and
magnitudes, illustrated in Figure 3-2. De-aggregation is important when selecting the appropriate time
histories with the relevant magnitude and distance for structural analysis purposes.
2) Generate the design response spectrum associated with a target return period. This spectrum is
developed by first performing the above PSHA calculations for spectral accelerations (SA) for a range of
periods. A target rate of exceedance is chosen, and the spectral acceleration amplitude corresponding to
that exceedance rate is identified. This process is illustrated in Figure 3-3.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 26


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.

Page 27
Figure 3-2 Example of de-aggregation for SA (1.0s) at a site

DNV AS
Recommended practice — DNV-RP-0585. Edition August 2021
Seismic design of wind power plants
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Figure 3-3 Combining hazard curves from individual periods to generate a uniform hazard
-4
spectrum (UHS) with a 4×10 annualized rate of exceedance

The PSHA should be performed using sound engineering judgement, and adopt a reasonable and justifiable
set of inputs and methods, including:
— reputable and reliable data sources for historical earthquake records
— an appropriate coverage 'radius' of historical earthquake records around the wind farm site
— a sufficient set of ground motion prediction equations (GMPEs) appropriate for the site
— consideration of variations across the site
— derivation of the vertical and horizontal design response spectra.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 28


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Guidance note:
The coverage 'radius' of historical earthquake records should be where expanding the coverage only marginally increases
the resulting spectral accelerations. Provisions stipulated in local standards should be followed wherever possible. If no such
information is available, a minimum 'radius' of 200 km may be considered as a starting point.
The PSHA should be performed so that the resulting design response spectra are conservative for the entire wind farm site,
including the export cable corridor. In practice, five locations within the wind farm site may be considered as a starting point: the
centroid (1 point) and the extremities in four directions (4 points). If the resulting spectral acceleration difference, at a one-second
period is less than 0.05 g, the enveloping spectral accelerations from the five points considered should be taken as the design
spectral accelerations. Otherwise, zoning the site may be considered, with the same process performed for each zone.
Currently there is no standardized approach to how a PSHA should be performed. Instead, a high-level description of the overall
approach is available in ISO 19901-2. More detailed requirements are available in regional standards, such as CNS 15176-1 for
Taiwan, Technical Standards and Commentaries for Port and Harbour Facilities for Japan, and ASCE 7 for the United States.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

3.3 Selection of seismic ground motions for time domain analysis


For time domain analysis, earthquake ground motion time histories are selected/generated to represent the
engineering bedrock motion underneath the wind farm site, based on the design seismic response spectra.
Multiple sets of input ground motions are required. The selection of earthquake time histories shall be
consistent with the de-aggregation results from the PSHA. The following criteria should be considered in the
selection:
— The chosen earthquake magnitudes should be within ±0.5 magnitude units of the target magnitude.
— The distances to the fault should be similar, in particular for near-fault wind farm sites.
— The ground conditions of the selected earthquake motion records should be broadly similar to those of the
wind farm site.
— Near-source effects such as earthquake directivity should be considered if applicable for the wind farm
site.
— Other relevant attributes such as non-stationary characteristics, earthquake duration, peak ground
acceleration magnitude and spectral content and shape should also be considered.
Artificial time histories compatible with the seismic design spectrum are allowed, but their role should
be limited to complementing the selected historical earthquake motion data, rather than being the main
earthquake motion inputs for the design.

3.4 Seismic site response analysis


The selected bedrock earthquake acceleration time histories travel from the earthquake source and
propagate through the soil to reach the ground surface or mudline. Observations from historical earthquakes
show that local site conditions significantly influence the attenuated ground motion characteristics during
earthquakes. These site effects should be considered in the seismic design of wind power plants.
This propagation process may be simulated through the seismic site response analysis (SSRA). SSRA
provides near-surface acceleration time series, acceleration response spectra, and spectral amplification
factors based on the dynamic response of local soil conditions.
Guidance note 1:
If site-specific data is not available or only a limited amount of information is accessible, several seismic codes (e.g. EN 1998-5,
ISO 19901-2) account for site effects using response spectra with a modified spectral shape based on soil categories defined
according to a 1-D proxy.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The one-dimensional (1-D) SSRA propagation models may be used in the design if 2-D or 3-D effects are not
significant for the site of interest. Otherwise, a more advanced analysis considering detailed topology wider
than the site of interest is required. Notable 2-D and 3-D effects include the following:
— bedrock slopes, presence of faults in the vicinity, and other topographical irregularities

Recommended practice — DNV-RP-0585. Edition August 2021 Page 29


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
— complex/irregular embedded soil lenses
— sloping soil layers
— presence of canyons, hills, cliffs, ridges, and basin edges in the vicinity
— potential interaction with other large embedded structures.
In 1-D SSRA, frequency domain equivalent-linear and time domain non-linear analyses are the most common
approaches. The dynamic responses computed via these methods can vary considerably because of the
inherent differences in the numerical approaches and differences in how non-linear soil response is modelled.
Time domain non-linear SSRA propagates the input ground motion through the soil deposit in the time
domain and varies soil properties with time simultaneously. This approach allows more realistic modelling
of non-linear soil response than equivalent-linear methods, which only approximate transient non-linear
behaviour as a strain compatible parameter. Therefore, the non-linear method generally provides more
accurate site response results, especially for high-intensity input motions. This method requires more
detailed measurement, testing and interpretation of dynamic soil properties than the equivalent-linear
method. It should be performed by trained specialists and well documented with underlying assumptions.
In comparison, equivalent-linear 1-D SSRA procedures require relatively few parameters and are generally
well defined. Uncertainties often reside with the input (e.g. the fitted 'backbone' curves), more than the
method or software. When limited input data is available, or when there is significant scatter within the strain
range of interest, one should consider the potential uncertainties through sensitivity studies, or investigate
whether such uncertainties have already been covered elsewhere. If input data is too scarce, then expansion
of the laboratory testing programme is recommended.
Guidance note 2:
For capturing energy dissipation, time domain analysis sometimes uses a frequency-dependent Rayleigh damping formulation. This
contradicts the generally known frequency-independent behaviour of soil damping. Such a damping formulation may negatively
impact the accuracy of the non-linear analysis result, particularly for low-intensity input motions.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
Regardless of the method chosen, i.e. non-linear or equivalent-linear, the maximum soil layer thickness
in a 1-D SSRA should be chosen so that most of the earthquake energy (and hence the majority of the
frequencies) will not be artificially filtered out through the modelling process.
By default, SSRA should be considered for all WT locations within the wind farm, unless it can be
demonstrated that certain 'worst location(s)' or 'worst-case response envelope(s)' exist for the wind farm.
One possible approach is to group WTs into 'mini-clusters' where the following aspects are consistent or
broadly similar:
— Input seismic motion and engineering bedrock properties: within each 'mini-cluster', the depth of the
engineering bedrock (and associated V_(s,30) definition) from the PSHA, and the input seismic motion
derived from spectral matching.
— Type of support structure: each 'mini-cluster' should feature the same type of foundation and support
structure.
— Water depth: each 'mini-cluster' should not cross a major water depth change.
— Soil stratification: all positions grouped into a 'mini-cluster' should feature consistent soil layering and
broadly similar stratification depths.
— Soil parameters: the main input parameters for each soil layer should be broadly similar, ideally to the
point where all positions within each 'mini-cluster' produce comparable results.
The above criteria may be reviewed so that one or multiple sets of SSRA (lower bound and upper bound)
cover all positions within each 'mini-cluster'.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 30


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Guidance note 3:
If no seismic 'mini-clustering' exercise is performed, the variation in geotechnical site conditions across any conventional 'cluster'
may lead to very different SSRA results. A common scenario is illustrated in Figure 3-4, where A1 and A2 are two different WT
positions within the same wind farm site.

Figure 3-4 Hypothetical SSRA results for two WTs (A1 and A2) within one wind farm site

Given the depth-variant nature of the SSRA results and their dynamic and varying influence on different components within the
WT, it is often impractical to arrive at a so-called 'worst case' WT position.
A potentially viable approach may be to identify and consider the softest and stiffest locations within a cluster, and possibly
a few positions in between where concurrence between SSRA peaks and the WT's natural periods is observed. However, the
definition of the 'softest' or 'stiffest' location within a large cluster as input for SSRA is not straightforward, as it can be a function
of the soil column's natural periods, the expected pile displacements, liquefaction depths and so on. This approach is also not
guaranteed to fully capture variability across the cluster or site. Hence WT-specific or the 'mini-clustering'-based approaches are
still recommended.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

3.5 Ground motion input to seismic design


The depth-varying earthquake motion signals between the engineering bedrock level and the ground surface
are obtained from SSRA. If the foundation of the wind power plant extends substantially into the soil, e.g.
jacket piles and monopiles, earthquake motions can vary significantly across the foundation depth.
In subsequent seismic modelling and analysis, when only one set of ground motions (two horizontal direction
and one vertical direction inputs) is selected and applied uniformly throughout the foundation depth or
at the node(s) between the foundation and the support structure above it, this is called 'uniform ground
motion input'. This may be necessary if the foundation is shallow or not explicitly modelled. In contrast, if
the foundation is explicitly modelled and the depth-varying ground motion inputs are implemented along the
foundation depth, this more rigorous method is termed the 'depth-varying ground motion input'.
The 'depth-varying ground motion input' method is physically more accurate, and theoretically more robust.
This is because SSRA does not consider the interaction between the soil and the foundation structure, and
the effect of difference in stiffness between soil and structure during an earthquake (kinematic interaction
effects) can only be accounted for by applying depth-varying ground motions to a structural foundation
model. This method is applicable for all deep or embedded foundation types and is the preferred method for
analysing pile-supported wind power plants. The depth-varying ground motion inputs should be based on the
soil layering and selected at frequent, reasonably regular spacings.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 31


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
In contrast, the 'uniform ground motion input' method is simpler and may be less demanding
computationally. Strictly speaking, it is only valid for shallow foundation types, such as gravity-based
structures, if the input ground motion comes directly from the SSRA.
However, the 'depth-varying ground motion input' method is not always a viable option for wind power
plants with deeper foundations. In such cases, a possible compromise is to estimate an 'equivalent depth'
of free field motion at which a single set of corresponding earthquake motions may represent the seismic
event. Rigorous determination and validation of this 'equivalent depth' require a kinematically-coupled soil-
foundation interaction analysis where both soil and pile(s) are modelled and the motion at the pile head (the
kinematic motion or effective input motion) is extracted. An equivalent depth may then be estimated based
on the maximum shear force transfer between the soil and foundation. Alternatively, the 'equivalent depth'
may be chosen so that the resulting earthquake motions are conservative across the frequency range of
interest for all wind power plant assets. However, this is often difficult in practice and sometimes impossible,
as different frequency components of the same earthquake signal may be amplified at different depth ranges.
As a starting point, when considering an 'equivalent depth' for horizontal seismic motions of jacket piles in
soft soils, an initial attempt may be taken as between four and seven times the outer diameter of the pile.
For monopiles, a starting point at one third of the pile length may be used. Note that for vertical seismic
motions, it is usually appropriate to select the 'equivalent depth' at the pile tip or foundation base level.
In all cases, the chosen 'equivalent depth' should include detailed justification of its appropriateness and
conservatism, considering the sensitivity of various wind power plant assets to different frequency ranges
within the earthquake ground motions. Extensive sensitivity analyses are usually required in this regard.
If the wind power plant's foundations are located in liquefiable and/or extremely soft soil layers, or
alternating stiff-soft soil layers, kinematic soil-foundation interaction analyses should always be performed to
ensure the integrity of both the support structure and foundation.
Guidance note:
In the oil and gas industry, it is common practice to use uniform ground motion input for simplicity and to circumvent
computational resource constraints. There are no reported incidences of the pile/structure being under-designed. However, this
should be taken with a pinch of caution because only limited offshore structures exist in the seismic zones. For a jacket seismic
analysis in the oil and gas industry, a popular rule of thumb (in the US region) is to use the SSRA motions obtained at 40 feet (12
metres) below the mudline. For typical jacket pile sizes used in the oil and gas industry, this 40 feet rule in general falls in-between
the four to seven times outer diameter range. ISO 19901-2 indicates that the horizontal input motion at the 1/3 pile length below
the mudline may be used.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

3.6 Seismic design load cases (DLCs)


According to DNV-ST-0437, the default seismic DLCs are 11.1, 11.2 and 11.3 as shown in Table 3-1.

Table 3-1 Seismic DLCs from DNV-ST-0437

Marine Condition
Design Wind Wind and Earthquake
DLC
situation condition Waves wave Sea currents Water level directionality
directionality

Earthquake NTM
(power 11.1 Vin ≤ Vhub ≤ H=Hs(V) COD, UNI NCM NWLR MUL
production) Vout

Recommended practice — DNV-RP-0585. Edition August 2021 Page 32


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Marine Condition
Design Wind Wind and Earthquake
DLC
situation condition Waves wave Sea currents Water level directionality
directionality

Earthquake
plus NWP
shut down 11.2 Vin ≤ Vhub ≤ H=Hs(V) COD, UNI NCM NWLR MUL
(power Vout
production)

Earthquake
(parked, NWP
11.3 H=Hs(V) COD, UNI NCM NWLR MUL
standing Vhub = V1
still or idling)

Additional guidance is included below to clarify their application in seismic design practice for WTs.
DLC 11.1 assumes the occurrence of an earthquake during normal operation. In this case, the main driving
variable action should be seismic rather than other environmental loads. Therefore, the project may use
operational wind and marine conditions along with the seismic input for this DLC. In other words, instead of
using the default wind speed range of V_in ≤ V_hub ≤ V_out, V_hub = V_r may be used. The basis of such
simplifications can be found in IEC 61400-1. Similarly, NWP may be used instead of NTM for DLC 11.1.
Guidance note 1:
When the major loading is the seismic condition, the average value of other loadings may be used in combination. This has been
the practice for the seismic analysis of offshore oil and gas platforms.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
DLC 11.2 comprises a superposition of the earthquake and a shut-down procedure. Again, instead of using
the default wind speed range of V_in ≤ V_hub ≤ V_out, V_hub = V_r may also be used for this DLC. The
shut-down could be due to either internal faults (e.g. excessive vibration) or external faults (e.g. grid loss).
In this regard, the local grid requirements and the grid operational requirements should be considered. If
the shut-down is due to an internal fault, the timing of the shut-down should be chosen in such a way as to
obtain conservative loads.
Guidance note 2:
It is usually sufficient if the timing of the shut-down is combined with the 'strong motion' part of the seismic signal. 'Strong motion'
is ground motion of sufficient amplitude and duration to be potentially damaging to the wind power plant asset. It may be defined
as the portion of the earthquake input signal between 5% and 75% of the cumulative energy transferred.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
DLC 11.3 considers a superposition of the earthquake and a previously occurred grid loss. The variations in
the rotor azimuth position should be considered in order to determine the loading on the rotor blades and
tower.
Guidance note 3:
The probability of concurrence between a ULS earthquake action and the worst-case rotor azimuth position is rather low. Instead,
the design should consider a continuous range covering at least one third of the possible azimuth positions, and the seismic load
should correspond to a reasonably conservative value from this range.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
For each DLC, as default, seven sets of different time histories should be considered in the SSRA. This is
to adequately capture the potentially wide range of dynamic ground responses to different excitations at
engineering bedrock. Each set of time histories consists of two horizontal components and one vertical
component. Correspondingly, seven sets of ground responses should be obtained, and each set used as input
for analysing the seismic load and structural response respectively.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 33


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
The overall design should be demonstrated to be satisfactory for at least half of the time history analyses
performed. This means that all WT components should pass the same four sets of time history analyses out
of the seven considered. The results from the top three sets may be discarded.
When fewer than seven sets of different time histories are used, the design should be demonstrated as
satisfactory for a higher percentile of the time history analyses performed, but no fewer than four sets of
time histories.
Guidance note 4:
The approach outlined above is consistent with ISO 19902. When seven or more earthquakes are considered, the design value of
the load action may be taken from the average of the response quantities in lieu of the 'four out of seven' criterion stated above.
Alternatively, IEC 61400-6 Annex O may also be followed, which suggests that the design seismic load should be evaluated
according to either of the following:

— The maximum load based on at least three different sets of time histories.
th
— The 85 percentile value of the load based on at least six different sets of time histories.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
Typically, the seismic signals derived at the design location are three dimensional. When multiple sets (seven
or more) of earthquake time histories are used in the structural assessment, the directional variability is
inherently considered in each of the chosen records. For sites that do not exhibit strong seismic directionality,
the predominant overall horizontal direction of each earthquake motion will be unique, varying in the
two-dimensional horizontal plane over time and with different single degree of freedom (SDOF) system
characteristics. In this case, the multidirectionality of the seismic signals need not be considered as
an additional variable. However, if the number of earthquake records chosen is fewer than seven, the
multidirectionality should be considered. The seismic signals should be obtained from the process as
described in this chapter.
The guidance above is intended for the time domain analysis as per DNV-ST-0437. If the RSM method is
used, the provisions within IEC 61400-1 shall be followed, as well as any additional requirements within
the applicable local standard(s). The RSM method often requires the most onerous directional combination
of earthquake loads and other loads. Therefore, multidirectionality shall always be considered. When the
applicable standard does not explicitly define what loads to combine with the design seismic action, the
principles of the load case table in Table 3-1 may be used.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 34


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 4 MODELLING APPROACHES IN SEISMIC DESIGN

4.1 General
A model of a physical system takes into consideration the mass, stiffness, and damping. IEC 61400-3 and
IEC 61400-6 require the use of time domain analysis. In IEC 61400-1, the use of frequency domain analysis
(i.e. the response spectrum analysis method) is permitted. With respect to seismic analysis in general, see
DNV-ST-0437 regarding the loads and site conditions and DNV-ST-0126 regarding support structures for
wind turbines.
Guidance note 1:
IEC 61400-6 states that the time domain analysis method must be used if non-linear interaction exists between the soil and
foundation. In IEC 61400-1, the response spectrum analysis method is elaborated in Annex D. For onshore wind turbines, where
foundations are simpler, the response spectrum analysis method is permitted. IEC 61400-3, which applies specifically to offshore
applications, requires the use of the time domain method.
It is also noted that, in the oil and gas industry, the response spectrum analysis method is routinely used for offshore piled jackets
under the ELE load level. The ELE load level return period as defined in ISO 19901-2 is typically around a few hundred years. For
ALE load level analysis, where the load level return period is a few thousand years, time domain analysis is typically used.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
In areas with low to moderate seismicity, wind and other ULS load cases are generally dominant. For
seismically active geographical areas, such as the Asia Pacific, seismic load cases may be dominant. For
definitions of low and moderate seismicity, consult ISO 19901-2.
Guidance note 2:
IEC 61400-1 states that no earthquake assessment analysis is required for sites already excluded by the applicable local seismic
code due to their weak seismic action. DNV's ACE JIP project shows that, for a 90 m (rotor diameter) onshore wind turbine, the
seismic and ULS load levels are in the same order of magnitude for tower top and tower bottom bending moments when the PGA is
in the order of 1.8 m/s². This value is for information and screening purposes only and should not be used to justify the omission
of seismic analysis.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
While time domain analysis is the required analysis method for offshore wind turbine structures, for onshore
wind turbines, frequency domain analysis, such as the response spectrum analysis method, is generally
permitted for onshore wind turbines, for details see [4.2.3]. For the applicability of the time domain analysis
and frequency domain analysis, see the seismic analysis flowchart [2.9].
Other rational analysis modelling procedures, that are different from the recommendations made in this RP,
may be used if they can be shown to have similar accuracies.

4.2 Time domain analysis

4.2.1 General
The preferred seismic analysis method for wind turbine structures is the time domain analysis method, also
known as the time history analysis method. To carry out a time domain analysis, realistic or synthetic time
history records are required. In a time domain analysis, different modelling approaches may be chosen from
among fully integrated modelling and different superelement modelling methods.
The following modelling approaches for seismic time domain analyses are covered in more detail in [4.2.2]
and [4.2.3] of this RP:
— fully integrated analysis approach
— normal superelement approach
— seismic superelement approach.
The workflow for both the fully integrated approach and the superelement approach is shown in Figure 4-1.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 35


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Figure 4-1 Workflow for the fully integrated and superelement analysis approaches

4.2.2 Fully integrated analysis approach (time domain)


4.2.2.1 Introduction
The fully integrated analysis approach is defined as the modelling approach where the entire structure of
the RNA, tower, sub-structure, and foundation are modelled within the wind turbine load calculation code, as
shown in Figure 4-2.

4.2.2.2 Minimum model requirements


The complete model of the RNA, tower and support structure, equivalent or depth-varying earthquake input
signals corresponding to the SSRA result, soil springs, soil damping characteristics, Sec.5, wind/wave loads in
accordance with [3.6], etc. are required to set up the model.

4.2.2.3 Support structure modelling approach


According to the fully integrated analysis approach, the support structure is explicitly and completely
modelled in the wind turbine load calculation code.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 36


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.

Figure 4-2 Example of the use of the fully integrated modelling approach (embedded piles
included)

Recommended practice — DNV-RP-0585. Edition August 2021 Page 37


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
4.2.2.4 Seismic signals
The fully integrated modelling approach allows the application of different types of ground motion, partially
depending on whether or not the embedded pile(s) model is included, as well as the wind turbine load
calculation code.
For guidance on the ground motion input to the structural model, see [3.5].

4.2.2.5 Soil-pile interaction properties


The soil reaction spring elements may be either linear or non-linear. Fully non-linear soil springs should be
included if possible. Guidance should be sought in [5.3].
When an embedded piled foundation is modelled for the pile-soil interaction, the distance between two
springs should be judiciously chosen. Equidistant spring distances or definitions used for calculating wind/
wave loads may not suffice for calculating seismic loads.
Likewise, guidance on the inclusion of soil damping should be sought in [5.2].

4.2.2.6 Wind turbine model


The wind turbine model used for type certification or component certification is generally sufficient for
calculating the seismic loads.
Rotor blade seismic loads should be analysed along the entire length up to the tip. As per DNV-ST-0376, the
loads analysed should be at the same point along the blade as those used in structural analysis in other load
cases.
The blade model shall include a sufficient number of blade modes and eigenfrequencies so that any
resonance effects caused by the seismic excitation are captured. This may require a more detailed model
than that required for the standard type certificate for the turbine model, see also guidance note 3 in [3.6].

4.2.3 Superelement approach (time domain)


4.2.3.1 Introduction
Two types of superelement modelling approaches are covered in this section - the normal superelement
approach and the seismic superelement approach.
Guidance note:
In this RP the term 'superelement' is used to describe a set of reduced matrices (and corresponding load vectors) that is able to
represent the response of the sub-structure (for example jacket or monopile), either with or without the inclusion of the embedded
pile(s) model, at the selected interface node(s). This may then be used in a wind turbine load calculation code, where the model
for the turbine RNA and tower is added and connected to the superelement.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
The superelement approach is considered common industry practice for offshore wind turbine design, where
superelements have been an important technique. This is due to the fact that in a commercial project
the responsibilities are split between different project parties, and the superelement approach allows the
responsibilities and intellectual property to be kept separate between the different parties.
The analysis is typically achieved through several iterations between the turbine and foundation designers.
The foundation designer provides the sub-structure and foundation with the corresponding loading as
superelement and load files to the turbine manufacturer, and the turbine manufacturer then provides wind
turbine interface loads back to the foundation designer. The interface loads are used for re-tracking, to obtain
the detailed loads within the sub-structure and foundation.
If more superelement interface nodes are used to allow earthquake time histories to be explicitly applied
in the wind turbine load calculation code, the rigid body motion component, due to the seismic input, is
automatically included in the calculation. If the wind turbine load calculation code accepts superelements with
multiple interface nodes, the seismic superelement may be used in the analysis, see [4.2.3.3].
The normal superelement may be used in wind turbine load calculation codes, whether or not they accept
superelements with multiple interface nodes, see [4.2.3.2]. If using a normal superelement, the response
due to rigid body motion should be included in the load transfer file. The superelement creation procedure
shall be documented for quality assurance.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 38


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
For the seismic design of wind turbine structures all superelement approaches are generally allowed, while
care should be taken to properly account for linear and non-linear behaviour.
Non-linear behaviour, resulting from either the pile-soil interaction, the structure above or both, present
explicit modelling challenges to the superelement approach. This is because superelements are by definition
linear, as the structure/soil is defined by matrices. This creates difficulties for seismic analysis as non-linear
behaviour may be important, especially when the return period is raised to the ALS level, see [4.2].
On the other hand, superelements can model the sub-structure in a more detailed manner using a sub-
structure specific software. Typical commercial wind turbine load calculation software used in a fully
integrated approach lacks the level of sub-structure details obtained from using the superelement approach.

4.2.3.2 Normal superelement approach

4.2.3.2.1 General
The normal superelement approach, applied for the purpose of seismic analysis, is defined as the
superelement approach that uses a single interface node (at the tower bottom interface) and includes the
response due to seismic input in the superelement loads.

4.2.3.2.2 Minimum model requirements


In general, the minimum model requirements necessary for the fully integrated analysis above also apply for
superelement approaches.

4.2.3.2.3 Support structure modelling approach


In the normal superelement approach the sub-structure of the tower/transition piece interface and structure
below is converted into one superelement. This may either include or exclude the embedded piles. The
response due to seismic input is included in the superelement loads, see Figure 4-3. The superelement is
then imported into the wind turbine load calculation code.
When using the normal superelement approach for seismic analysis, the rigid body motion components
should be embedded as part of the load components in the load transfer at the interface node. This may be
inadvertently missed in a seismic analysis and result in an error.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 39


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Figure 4-3 Normal superelement modelling approach excluding (l) and including (r) embedded
piles

4.2.3.2.4 Seismic signals


For the normal superelement modelling approach, an equivalent seismic signal at the node(s) at the bottom
of the sub-structure, an equivalent signal along the embedded pile(s), or a depth-varying seismic signal
along the embedded pile(s) may be included in the superelement creation process.
For guidance on the ground motion input to the structural model, see [3.5].

4.2.3.2.5 Soil-pile interaction properties


If the superelement technique is used, this inherently assumes the structure and foundation behave linearly
elastically. In this case linear soil springs are used. The soil spring stiffness should be appropriately linearized
based on the seismic load level considered. Guidance should be sought in [5.3]. The normal superelement
modelling approach may include soil-pile interaction along the embedded pile (if modelled) at multiple
locations. In this regard, it is similar to the fully integrated analysis approach, and the same provisions apply.
The key difference is that linearized rather than fully non-linear soil springs are used. Guidance should be
sought in [5.3], especially regarding the linearization process. Again, the distance between springs should be
judiciously chosen. Equidistant spring distances or definitions used for calculating wind/wave loads may not
suffice for calculating seismic loads.
Likewise, guidance on the inclusion of foundation damping should be sought in [5.2].

Recommended practice — DNV-RP-0585. Edition August 2021 Page 40


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
4.2.3.2.6 Wind turbine model
The requirements for the wind turbine model are identical to those for the fully integrated analysis approach.

4.2.3.3 Seismic superelement approach

4.2.3.3.1 General
The seismic superelement approach is defined as the superelement approach that uses multiple interface
nodes (one interface node at tower bottom interface, and seismic interface nodes for the application of
seismic input) and that does not include the response due to seismic input in the superelement loads.
Instead, the seismic input is directly applied to the seismic interface nodes of the superelement in the wind
turbine load calculation code.

4.2.3.3.2 Minimum model requirements


In general, the minimum model requirements for the fully integrated analysis above also apply for
superelement approaches.

4.2.3.3.3 Support structure modelling approach


According to the seismic superelement approach the support structure from the tower/transition piece
interface, typically until the mudline or below, is converted into a superelement. The response due to seismic
input is not included in the superelement loads in this approach, see Figure 4-4. The superelement is then
imported into the wind turbine load calculation code.

Figure 4-4 Seismic superelement modelling approach

4.2.3.3.4 Seismic signals


The seismic superelement modelling approach allows for an equivalent seismic signal at the seismic interface
node(s). Contrary to the normal superelement, the seismic superelement load file does not include the
seismic input. Instead, the seismic input is directly applied to the seismic interface nodes of the superelement
in the wind turbine load calculation code.
For guidance on the ground motion input to the structural model, see [3.5].

Recommended practice — DNV-RP-0585. Edition August 2021 Page 41


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
4.2.3.3.5 Soil-pile interaction properties
The soil-pile interaction properties for the seismic superelement approach follow the same provisions as
stated for the normal superelement approach in [4.2.3.2].
Likewise, guidance on the inclusion of foundation damping should be sought in [5.2].

4.2.3.3.6 Wind turbine model


The requirements for the wind turbine model are identical to those for the fully integrated analysis approach.

4.3 Response spectrum analysis method (frequency domain)

4.3.1 Frequency domain analysis


4.3.1.1 General
As stated in [3.4], the response spectrum analysis method is permitted for onshore wind turbine applications.
The analysis method is elaborated in IEC61400-1 Ed.4, Chapter 11.6. When comparing the response
spectrum analysis method to the time domain analysis method for onshore wind turbines (for cases with no
non-linear foundation behaviour), the following general trends apply:
— The frequency and time domain approaches generally generate comparable load results for the whole
elevation of the tower for the lumped mass RNA model.
— The calculated seismic loads of the frequency and time domain approaches converge to a high degree
when equivalent models and the same external conditions are applied.
— The influence of higher tower modes on the seismic load level increases with an increasing tower length.
— The seismic load level of the structure depends on the modal excitation of natural eigenfrequencies based
on the shape of the response spectrum.
— Free vibration mode shapes used within the RSM method were found to lead to better matching results
(compared to time domain results) than the use of the Craig-Bampton mode shapes.
The above in general confirms the applicability of the response spectrum analysis method for onshore
applications. However, recent research by DNV's ACE JIP project shows that tower top loads may be
underestimated by the response spectrum analysis method.
The ACE JIP project notes that, because of the tower top load differences, the time domain analysis method
is preferred for its accuracy. The response spectrum analysis method is best used for screening purposes and
preliminary design.

4.3.1.2 Minimum model requirements


The minimum model requirements for the response spectrum analysis method comprise a lumped mass RNA
model, tower/support structure geometry (section heights), mass distribution along the support structure,
mode shapes and eigenfrequencies, foundation stiffness, damping definition, earthquake input signal as a
response spectrum, and additional wind/wave loading if applied.

4.3.1.3 Support structure modelling approach


According to the response spectrum analysis method the entire support structure is modelled in terms of
geometry, mass distribution and the inclusion of structural mode shapes.

4.3.1.4 Seismic signals


The ground motion is defined by a response spectrum.
For guidance on the ground motion input, see Sec.3.

4.3.1.5 Wind turbine model


According to the response spectrum analysis method the wind turbine RNA model is included as a lumped
mass at the tower top, which either includes or excludes a description of the mass moment of inertia.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 42


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
As the full RNA is not modelled, when this modelling method is used, the seismic loads on the rotor blades
cannot be determined. The dynamic coupling between the flexible rotor blades and structure is also not
captured.
For WT structures with slender towers and flexible blades, the dynamic coupling may induce a higher load in
the tower top region.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 43


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 5 GEOTECHNICAL ASPECTS

5.1 Liquefaction

5.1.1 Introduction
Soil liquefaction occurs when the effective stress of soil is reduced to nearly zero. During earthquakes, this
is usually initiated by cyclic loading, i.e. repeated changes in the stress condition. Liquefaction may occur in
sandy or non-plastic silty soils, as well as some gravelly soils and clays. A soil in a saturated loose state is
most likely to liquefy during earthquakes as it has a tendency to compress when sheared, generating large
excess porewater pressure. As pore water pressure rises, the soil progressively loses strength as the effective
stress is reduced.
To address soil liquefaction for level-ground wind farm sites, it is important to consider the following three
aspects in sequence:
— Liquefaction susceptibility, i.e. whether the soil is inherently susceptible to liquefaction.
— Liquefaction initiation (or triggering), i.e. given that the soil is susceptible to liquefaction, whether it will
liquefy under the given earthquake ground motions.
— Liquefaction effects, i.e. given that the soil is susceptible to liquefaction and that it will liquefy under the
given earthquake ground motions, what the effects of the liquefaction phenomena will be.
The technical details involved are highly complex and should be examined by qualified geotechnical
specialists. Some general information is provided in the following sections as guidance.
It is important to note that whether or not a soil layer will liquefy is never 100% certain. In design
calculations, liquefactions and their likely impact on foundations/structures should be assessed. Their
sensitivities should be arranged in various design scenarios, i.e. the likely occurrences, including no-
liquefaction and liquefaction, should be considered to bound the design envelope. In addition, the design
shall always consider the impact of excess pore water pressures generated during an earthquake event.

5.1.2 Liquefaction susceptibility


Determining the liquefaction susceptibility of each relevant soil layer serves as a useful screening exercise.
The layers that are not susceptible may be discarded at an early stage to reduce the design effort required.
Assessment of a soil's liquefaction susceptibility does not require knowledge of the earthquake input itself.
Several criteria may be used, often in combination, to determine a soil layer's susceptibility to liquefaction:
— Historical criteria, i.e. soils that have liquefied in the past, can and probably will liquefy again in future
earthquakes. Information is sometimes available in the form of maps of areas where liquefaction has
occurred in the past and/or is expected to occur in the future. Some local government agencies have
prepared maps of sensitive areas, including areas susceptible to liquefaction. It is recommended to
consult local authorities on the availability of such maps.
— Geological criteria, i.e. the type of geologic process that created a soil deposit, has a strong influence
on the soil's liquefaction susceptibility. Saturated soil deposits that have been created by sedimentation
in rivers and lakes (fluvial or alluvial deposits), or the deposition of debris or eroded material (colluvial
deposits), or deposits formed by wind action (aeolian deposits) tend to be susceptible.
— Compositional criteria, i.e. the composition of the soil strongly influences the soil's liquefaction
susceptibility. Clayey soils, particularly sensitive soils, may exhibit strain-softening behaviour similar to
that of liquefied soils, but do not liquefy in the same manner as sandy soils. Soils composed of particles
that are all about the same size (poorly graded soils) are more susceptible to liquefaction than soils with a
wide range of particle sizes (well graded soils). Some empirical methods exist, based typically on the soil’s
particle size distribution, Atterberg limits, water content, and the soil behaviour type index.
— State criteria, i.e. the effective confining stress and density of the soil influences the soil's liquefaction
susceptibility. In essence, the criteria assume that 'loose' soils tend to liquefy, while 'dense' soils tend
to dilate, during shear. Differentiation between 'loose', or contractive, behaviour and 'dense', or dilative,

Recommended practice — DNV-RP-0585. Edition August 2021 Page 44


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
behaviour may be based on the soil type, its void ratio and the in situ effective stress. Pre-existing static
shear stress in a soil deposit is also an important factor.
A suitable combination of these criteria and associated simplified methods are usually sufficient for
liquefaction susceptibility screening.

5.1.3 Liquefaction initiation


Just because a soil is susceptible to liquefaction does not mean that it will liquefy during an earthquake.
The extent of the given earthquake ground motions should be considered when combined with the soil's
properties, this combination determines whether liquefaction will be initiated (liquefaction initiation or
triggering). Liquefaction initiation depends on the stress state of the soil, the duration of the earthquake
loading, and the amplitude of the earthquake loading.
When using stress-based methods, the factor of safety against liquefaction initiation is usually defined as the
ratio between the cyclic resistance ratio (CRR) and the cyclic stress ratio (CSR). The CSR measures the cyclic
stress that is applied to the soil, i.e. the loading 'demand', whereas the CRR measures the soil's resistance to
liquefaction, i.e. the 'resistance'.
The most reliable way to derive the CSR is from the SSRA. Different ground motions will give different CSR
values due to their frequency components and respective amplitudes. The choice of which CSR value to select
should take into consideration the chosen design standards for any wind farm project.
CRR, on the other hand, may be obtained from laboratory tests such as cyclic simple shear, cyclic triaxial,
and other advanced laboratory soil testing methods. There are challenges regarding the validity of the
laboratory-test-based methods, the most notable of which being how to ensure that truly undisturbed
cohesionless soil samples can be obtained in the ground. Conventional techniques such as borehole sampling
inevitably disturb sandy soil samples that are the most relevant for liquefaction initiation assessment. An
additional challenge is that the number of loading cycles should also be pre-determined.
Because of these practical difficulties, a number of empirical methods have been proposed by various
authors over the past few decades to aid engineering practice. These methods intentionally move away from
laboratory-testing-based methods but rely on in-situ site investigation methods to evaluate the CRR and
approximate CSR using ground acceleration characteristics. Each method is correlated either to a certain site
investigation technique (e.g. standard penetration test, cone penetration test, or Becker penetration test), or
a certain parameter measured in-situ (e.g. shear wave velocity).
These simplified stress-based empirical methods are usually sufficient for depths up to 15 m. One should
apply several methods rather than relying on any particular method, unless there is good reason to do so.
Furthermore, parameter definitions from the original publications shall be strictly followed for each method,
as different methods may provide different definitions for parameters of the same name. Beyond the depth of
15 m, more advanced alternative approaches, such as those based on laboratory and/or physical modelling
tests, are recommended.
Guidance note:
If a simplified stress-based empirical method is used in the design without considering more advanced approaches, a minimum
safety factor of 1.25 against liquefaction should be considered. This implies that liquefaction is considered as initiated under level
ground conditions if the CSR reaches or exceeds 80% of the CRR at any depth within the top 15 m.
The 1.25 safety factor is not required if site-specific laboratory test data and associated correlations for liquefaction initiation have
been used.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
When considering liquefaction susceptibility based on laboratory testing, one should consider the relative
density, fines content, confinement pressure, possible sample disturbance, and other uncertainties. The CRR
for fine grained soils should always be based on advanced cyclic laboratory test data.
In addition to the stress-based methods that rely on the evaluation of the CRR and CSR, there are a growing
number of alternatives, such as the cyclic strain-based approach, energy-based approaches, laboratory
and physical model tests, field measurements of pore-pressure generation under dynamic loading, and
computational mechanics approaches.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 45


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
5.1.4 Liquefaction effects
The potential consequences of liquefaction associated with foundations include the loss of vertical bearing
capacity, loss of lateral stiffness and capacity, lateral loading due to lateral soil displacements, and downdrag
(i.e. negative skin friction) on piles due to the post-liquefaction reconsolidation of soil. These consequences
may be divided into inertial effects (i.e. those that impact the dynamic response of the pile foundation) and
kinematic effects (i.e. those induced by soil displacement relative to the foundation).
For deep foundations, such as piles, inertial and kinematic effects are often considered separately
because the available simplified design methods cannot account for them as coupled phenomena, and the
superposition principle is usually applied. Additionally, the peak inertial loads are likely to occur before the
soil liquefies and begins to 'flow'. The intensity of the ground motion subsequent to liquefaction will be
reduced and, thus, the peak inertial load may be considered separately from the kinematic loads due to
liquefaction-induced displacements.
The primary inertial interaction effect prior to liquefaction-induced displacements is a reduction in the lateral
stiffness of the foundation, which may be considered in the inertial loading analysis. The other effects cited
above will occur only after triggering and, therefore, may be assumed to be independent of the dynamic
response of the foundation system. Alternatively, an advanced computation model may be used to consider
all inertial and kinematic loads together.
Liquefaction affects the foundation, the support structure, tower, and turbines very differently. For example, a
layer of liquefied soil may significantly increase the load demands on the foundation, but the loads imparted
to the tower will be smaller because the seismic motions have been filtered out by this liquefied soil layer.
Furthermore, the residual strength of a liquefied soil layer should also be carefully considered. This depends
on many factors, including the soil's grain size distribution and initial density (or void ratio). The stresses in
the ground prior to the earthquake, and the presence of overlying confining layers, are also postulated to
affect the residual shear strength of a soil. The post-triggering strength may also change as consequences
develop due to effects such as particle rearrangement, excess pore pressure and stress redistribution, and
the intermixing of soils during deformation. In general, the effects of kinematic motion caused by liquefied
soil layers should not be neglected. Instead, a realistic evaluation of the structure's dynamic responses
should be based on the degraded soil strength and stiffness as a function of the pore pressure and relative
soil-structure displacement. When kinematic effects are likely to be significant, this implies that liquefaction
initiation and liquefaction effects should be assessed in conjunction by analysing the shear strain and excess
pore water pressure generation within the underlying liquefiable soil layers.

5.1.5 Liquefaction due to other non-seismic actions


It is also worth noting that an earthquake is not the only source of soil liquefaction. Other actions such as
tropical cyclones and impact pile-driving may also produce similar effects in the same soil. The former may
generate undulating waves in a previously flat seabed, whereas the latter may lead to significant risks of pile
run during impact pile driving. These risks should be adequately considered and mitigated during the design
and installation phases.

5.2 Damping in seismic design

5.2.1 General
The specific damping capacity of a system, also called the specific loss, is considered the most fundamental
measure of damping. It is defined as the ratio of where is the energy dissipated during a loading
cycle, and W is the maximum elastic energy stored during the cycle.
The damping ratio, , for the ideal stress-strain path shown in Figure 5-1, is calculated by:

Recommended practice — DNV-RP-0585. Edition August 2021 Page 46


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
where is the area of the shaded hysteresis loop, and is the area of the triangle OAB.
Due to damping, the response (displacement or strain) within an inelastic material lags behind the input
(force or stress) by a phase angle.

Figure 5-1 Hysteresis stress-strain loop for viscously-damped material

In seismic design practice, a certain percentage of critical damping is often used to indicate the damping
value inherent in the structural or foundation system. For example, design earthquake response spectra
are often presented in the form of 5% damping in many local and international standards. This 5% value
serves to align the processes of PSHA, SSRA and subsequent spectral matching. It is not a recommended
value, especially for WTs. Instead, the damping of wind power plant assets should be analysed and evaluated
carefully and on a project-specific basis.
Guidance note:
In the oil and gas industry, 5% damping is often used for the seismic analysis under ELE conditions. This 5% damping consists of
several components:

— steel structure, typically 1%


— hydrodynamic damping, typically 0.5% - 1%
— soil damping, typically approximately 3% - 4%.
For multi-legged, pile-supported oil and gas jacket structures, the damping ratio is typically assumed to be 5% as a combination
of the above components if a response spectrum analysis is performed. This experience is useful for the wind turbine structural
analysis, but care should be exercised for monopiles and other configurations that are not standard jacket types.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
In the turbine manufacturing industry, log decrement damping is frequently used. This damping is associated
with the successive free-vibration oscillatory decay of a single degree of freedom (SDOF) system, see Figure
5-2. The damping ratio and log decrement damping are interchangeable by the following equation:

where is the log decrement damping ratio. The above equation is valid for ≤ 10%. The correspondence
between these two damping ratios is shown in Table 5-1.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 47


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Table 5-1 Correspondence between damping and log decrement damping ratios

Damping ratio, Log decrement damping,

1% 6.3%

2% 12.6%

3% 18.8%

4% 25.1%

5% 31.4%

6% 37.7%

7% 44.0%

8% 50.3%

9% 56.5%

10% 62.8%

Figure 5-2 Free vibration of damped SDOF system

[5.2.2] to [5.2.4] focus on foundation damping and its implementation in the seismic design of wind power
plants.

5.2.2 Measurement of soil material damping


Soil material damping is attributed to many factors, including inter-particle sliding and friction, structure
rearrangement, and pore fluid viscosity. The damping ratio is commonly used in geotechnical engineering as
a measure of the energy dissipation during dynamic or cyclic loading. Strictly speaking, the term 'damping
ratio' only applies to an SDOF system but is often used in this context as an equivalent parameter.
Soil damping is an inherent material property and primarily internal. Internal damping is subdivided into
intrinsic damping and extrinsic damping. Intrinsic damping describes the energy losses at a specific point
within the material, whereas extrinsic damping characterizes the global energy loss within a finite volume.
When damping is independent of the strain or strain rate, both are equal. If the damping is a function of the

Recommended practice — DNV-RP-0585. Edition August 2021 Page 48


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
strain level and rate, the measured (extrinsic) damping will be an average of the intrinsic material damping
over the volume, , of the specimen:

where is the global (extrinsic) damping ratio and is the intrinsic damping ratio as a function of the
strain level and rate. As a result, cannot be easily inferred for a given . This shows the difficulty of
measuring the damping of highly non-linear materials, such as soils.
Soil damping is often measured together with other dynamic properties such as the shear wave velocity and
shear modulus. There are two broad categories of measurement methods: in-situ methods and laboratory-
based methods.
In-situ measurement methods are conducted in the field without soil sampling. They can be subdivided into
low-strain methods where the maximum shear strain is typically , and high-strain methods where
.
Basic forms of the low-strain in-situ tests include seismic reflection test, and seismic refraction test, which
can measure the seismic wave velocities, thicknesses, and profiles of superficial and lower layers. These
methods work well when soil strata become consistently denser with depth, when the terrain and sub-surface
soil strata are relatively flat, and when trying to map a high-contrast boundary such as between soil and
bedrock. They work less well when strong and weak soil strata occur intermittently, or when surface or
sub-surface strata show significant slopes. More advanced forms are often termed surface wave methods,
in that they involve the production and measurement of Rayleigh waves. They can develop complex soil
stratigraphy and 'see through' soft soil layers. Another group of methods that fall under this category involve
measurement using boreholes. The strain levels involved in all low-strain in-situ tests are generally small,
and hence most useful for low-strain applications such as SSRA or FLS analyses of WTs. For WTs under ULS
seismic loading, the shear strain values can be up to three orders of magnitude larger for the shallow soil
layers. Using the values determined from the low-strain in-situ tests for these layers is usually conservative.
High-strain in-situ methods, on the other hand, essentially refer to the standard penetration test (SPT) or the
cone penetration test (CPT). Direct damping measurement from these high-strain in-situ tests is not feasible.
Instead, it is correlated to available measurement data and other correlated soil properties. This brings a
multitude of uncertainties, which means other methods should be used to calibrate the proposed correlations.
In general, it is not recommended to infer soil damping properties from these high-strain in-situ methods
without calibration with laboratory testing or other direct measurement techniques.
Laboratory measurement methods are also subdivided into low-strain and high-strain methods.
Only a limited number of low-strain laboratory tests are available for determining the dynamic properties
of soils, including the damping levels. These include the resonant column (RC) test and the piezoelectric
bender element (BE) test. For both test types, the porewater pressure is rarely measured, and the dynamic
soil properties are usually measured at frequencies above those of most earthquake motions. Extracting
undisturbed soil samples is also a challenge, given the sensitivity of these dynamic soil properties to strain
histories. In addition, both BE and RC tests subject the soil to a significantly greater number of loading cycles
than in real earthquakes. A further limitation of the RC test is that water saturated specimens under large
strain amplitudes may liquefy. This is because high frequency shaking makes free drainage of pore water
impossible.
High-strain laboratory tests include the cyclic triaxial test, cyclic direct simple shear test, and cyclic torsional
shear test. At high shear strain amplitudes, soils generally exhibit tendencies to change volume. Under
drained loading conditions, these tendencies are allowed to manifest themselves in the form of volumetric
strain, but under undrained conditions they result in changes in pore pressure (and effective stress). Since
soil behaviour is governed by effective stresses, all methods of testing soils at high strain levels should be
capable of controlling porewater drainage from the specimen and measuring volume changes and/or pore
pressures accurately. The problem of system compliance (volume changes due to the testing apparatus
rather than the soil), which can lead to errors in volume change/pore pressure measurement, is important
in the interpretation of high-strain test results. Membrane penetration in coarse-grained soil is an important
contributor to system compliance.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 49


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
5.2.3 Modelling of soil damping in site specific response analysis
In 1-D SSRA (site specific response analysis), the primary mechanism for energy dissipation is soil material
damping. Many empirical relationships for estimating modulus reduction and damping curves (i.e. the
'backbone' curves) have been developed based primarily on laboratory test data. Each model features a
unique set of inputs, and some offer greater flexibility for curve fitting than others. There are no strict rules
in terms of which framework to adopt, and the choice should be governed by the quality of the resulting fit
and the availability of the required input parameters.
Guidance note 1:
DNV-RP-C212 provides guidance on shear modulus and damping curves for clay, sand, and silt. These may be used as default if
they are well-matched by site investigation and/or laboratory test data.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
One principle to uphold is that the same reference curve should be adopted for all soil layers at any given
site that share a similar geological history. Some commercial non-linear time domain site response analysis
software uses a hyperbolic model to represent the backbone response of the soil and the extended unload-
reload Masing rules to model hysteretic behaviour. This means that, at very small strains, the response is
nearly linear (hysteretic damping is nearly zero) and the use of velocity proportional viscous damping is
required. In this regard, frequency dependent Rayleigh damping is commonly employed and may result in
over or under damping depending on the loading frequency. At large strains, this formulation may result in
overestimation of the damping value. Careful selection and calibration of the damping values used in such
analyses is hence needed to minimize the impact of these inaccuracies.
Guidance note 2:
1
In Japan, the Ishihara-Yoshida method (Ishihara, et al., 1985 ) is often used to resolve the overestimation of the damping value
when the Masing rule is used.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---
In 2-D or 3-D SSRA, when earthquake energy is released from a fault below the ground surface, body
waves travel away from the source in all directions. The specific energy (elastic energy per unit volume) also
decreases as the wave travels through a material. Even though elastic energy is conserved (no conversion to
other forms of energy takes place), this reduction in amplitude due to the energy spreading over a greater
volume of material is often referred to as radiation damping (also as geometric damping and geometric
attenuation). It should be distinguished from material damping in which elastic energy dissipates by viscous,
hysteretic, or other mechanisms. If the rupture zone is represented as a point source, the wavefronts will
be spherical geometric attenuation that causes the amplitude to decrease at a rate of with being the
distance to the point source. It can also be shown that the geometric attenuation of surface waves causes
their amplitudes to decrease at a rate of ; in other words, surface waves attenuate (geometrically) much
slower than body waves. This explains the greater proportion of surface wave motion (relative to body wave
motion) that is commonly observed at large epicentral distances, and the advantages of the surface wave
magnitude, relative to body wave magnitude, for characterizing of distant earthquakes. Unless the seismic
source is very close to the ground surface, the effects of radiation damping often dominate those of material
damping in 2-D and 3-D SSRA.

5.2.4 Modelling of foundation damping in seismic analysis


Like 2-D and 3-D SSRA, soil-structure interaction seismic analysis should also consider the soil material
damping and radiation damping. While radiation damping is almost always present and may be taken
advantage of in the design, the treatment of soil material damping depends on the modelling approach
adopted.

1
Ishihara, K., Yoshida, N., Tsujino, S., (1985). Modelling of stress-strain relations of soils in cyclic
loading, Fifth International Conference on Numerical Methods in Geomechanics, p.373-380, Nagoya,
Japan.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 50


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
When using the time-domain fully-integrated analysis approach [4.2.2], the foundation damping may be
represented as a series of dashpots or damper elements connecting to the same nodes as the soil reaction
springs. Figure 5-3 shows an example of a monopile-supported WT. Each dashpot may be approximated as
the simple addition of material hysteretic damping and radiation damping:

Figure 5-3 Monopile foundation–supported wind turbine: (A) wind turbine; (B) Winkler model

Guidance note 1:
The Winkler model shown in Figure 5-3 is only applicable when the length-to-diameter ratio of the pile is larger than 10, with the
pile's outer diameter preferably less than 2 m. For pile foundations exceeding these limits, additional stiffness and damping terms
for the rotational degrees of freedom should be considered.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

This formulation ensures that the hysteretic material damping component ( ) is independent of the

loading frequency, whereas the radiation damping component ( ) is and starts to become significant for

frequencies higher than about 1Hz. One may choose to ignore the term to be on the conservative side.

However, if the implemented soil springs are fully non-linear and non-elastic, the term shall always be
omitted as it is already inherent in the non-linear soil reaction springs implemented in the analysis.
The time-domain superelement-based analysis approaches [4.2.3] may also include the above formulation
when the foundation structure is explicitly modelled and included in the superelement generation. The
difference is that only linearized damping and soil reaction springs may be included. If the foundation
damping is defined in the superelement definition, conservative estimates should be used.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 51


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
Guidance note 2:
For the conventional superelement approach, it is extremely challenging to obtain realistic damping values in the design stage that
are applicable for different limit state designs. Simplifications are hence often used, e.g. by assuming small-strain soil damping
values throughout the analysis. These simplifications can have a significant influence on the resulting seismic load calculations,
often leading to much higher load values compared to other more advanced methods and reality.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

5.3 Soil-structure interaction modelling

5.3.1 General
In general, DNV-RP-C212 should be followed when modelling dynamic soil-structure interaction.
For time-domain seismic analysis, it is most accurate to capture the stiffness and strength changes in the
soil via dedicated constitutive models and soil-structure interaction simulation. The soil springs should then
be calibrated as needed to salient time-steps within the earthquake event and implemented as a time-
dependent input to the seismic analysis model.
However, this rigorous approach is expensive and time consuming, and is often impractical for wind farm
projects. As a compromise, bounding scenarios may be chosen to represent the softest and stiffest soil
reactions, as well as intermediate values that can potentially cause resonance in the structure. This should
take liquefaction uncertainty into consideration. These analyses should be performed for the limit state that
governs the structural and foundation design, usually either ULS or SLS, see Sec.4.
It is preferable to implement non-linear soil reaction springs directly in the model. If this is not possible
and linearization shall be performed, it is important to acknowledge the interactive nature between the soil
spring stiffness and the resulting seismic loads. The linearized soil spring stiffness should reflect the strain
level and excess pore pressures associated with the load case under consideration. It is not appropriate,
for example, to apply the regular ULS or FLS soil springs for seismic DLCs. An iterative procedure should be
adopted to determine the equivalent linearized soil spring stiffness applicable for earthquake DLCs through
close collaboration between the turbine manufacturer and the foundation designer.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 52


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
SECTION 6 POST-EARTHQUAKE CONSIDERATIONS

6.1 General
Operators should plan the immediate actions to be taken after a significant earthquake event. This may be
in the form of an action flowchart, starting with an assessment of the intensity of the event relative to the
design level. Appropriate actions should then follow, including inspections and operational restart.
Additional comments and guidance on post-earthquake considerations are provided below.

6.2 Aftershocks
Major earthquakes typically feature significant aftershocks. If the seismic design process stated in this RP is
followed, then aftershocks do not have to be taken into consideration in the design. This is because:
— The main WT components and the OSS are designed to cope with a ULS earthquake with little to no
damage. The aftershocks are lower in magnitude and hence are considered automatically covered.
— An ALS earthquake does not normally have to be taken into consideration for WT components unless
personnel safety aspects are applicable, see Sec.4. If these are applicable, the design philosophy ensures
that there is adequate time for the evacuation of personnel, so that aftershocks should not pose a further
threat to personnel safety.
— The nature and location of the wind farms typically ensure negligible risk to the general public. The safety
logic also applies to the OSS.
— For SLS earthquakes, the differential settlement (and hence tilt) is primarily driven by the direction and
magnitude of the thrust force considered rather than the earthquake motion. The common practice is to
consider an onerous combination of the thrust force and SLS earthquake motion. It is highly unlikely that
the same thrust force will be experienced by the WT during the aftershock.
The same rationale also applies to foreshocks.
Guidance note:
The focus of the prevalent earthquake-related design standards, such as ISO 19901-2 and the Eurocode 8 series, is to ensure that
the design can sustain a single major earthquake event without compromising structural reliability, and there is adequate provision
for personnel safety/evacuation. Potential aftershocks are therefore not considered in these standards.
The regional standard ASCE-7 for the U.S.A. suggests that aftershocks should be considered for 'secondary containment systems'
to protect the general public, and states that two-thirds of the maximum considered earthquake (MCE) ground motion may be
used as the design aftershock magnitude.

---e-n-d---o-f---g-u-i-d-a-n-c-e---n-o-t-e---

6.3 Persistent liquefaction


Liquefaction effects may persist for a period of time after the initial earthquake, during which the soil
gradually regains its pre-earthquake strength. The impact of this residual liquefaction on WTs should be
assessed. However, dedicated load cases for the post-earthquake design scenario are not yet covered by any
existing design standards.
The key design objective for the post-earthquake scenario should be no different from that for other design
scenarios, i.e. to ensure the same consistent level of safety as the other scenarios within the same limit
state.
Considering the 475-year return period for the ULS scenario, any persistent liquefaction effects will contain
−3
an embedded probability of occurrence of 2.1×10 . Assuming that these effects will last for six months,
then combining such with a characteristic storm with a 10-year return period would lead to a resulting joint
-4
probability of occurrence of approximately 1×10 within this six-month period.
This implies that characteristic load and material/resistance values may be used for the post-seismic
assessment. In this regard, the proposed approach effectively treats the liquefied soil supporting the WTs as
the 'damaged' component which will repair itself over time. This means an equivalent ALS philosophy may

Recommended practice — DNV-RP-0585. Edition August 2021 Page 53


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
be applied, i.e. ALS type 2 in DNV-ST-0126, and the corresponding ALS design requirements in the same
standard also apply.
This avoids the difficult question of what partial load and material/resistance safety factor values are
applicable for this post-earthquake design scenario. If persistent liquefaction may last for longer than six
months, the return period of the characteristic storm shall be higher, and vice versa.
When applying the approach proposed above, one should not take into account the strength regain of the
soil over time, as the assumption is that the characteristic storm may hit at any time during this persistent
liquefaction period.

6.4 Mitigation measures


If an earthquake exceeding the SLS design assumptions occurs in a wind farm, the operator should have
procedures in place for performing 'health checks' on the wind power plant assets before starting full normal
operation again.
Specific attention should be given to blades and machinery components, especially if the design return period
level is lower than their respective defaults, see [2.3] and [2.4].
As a recommendation, information on the design seismic level and action plans should be properly
transferred from the designer to the wind farm operator. The transferred information may include, but is not
limited to, requirements on inspections of the critical components, and remote monitoring or observation of
the turbines' SCADA systems, e.g. by checking for frequency shifts or accelerations measured by sensors if
applicable.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 54


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
CHANGES – HISTORIC

Changes – historic
There are currently no historical changes for this document.

Recommended practice — DNV-RP-0585. Edition August 2021 Page 55


Seismic design of wind power plants

DNV AS
This copy of the document is intended for use by Minji Seo only. Downloaded 2023-01-18. No further distribution shall be made. Requires a valid subscription to be used.
About DNV
DNV is the independent expert in risk management and assurance, operating in more than 100
countries. Through its broad experience and deep expertise DNV advances safety and sustainable
performance, sets industry benchmarks, and inspires and invents solutions.

Whether assessing a new ship design, optimizing the performance of a wind farm, analyzing sensor
data from a gas pipeline or certifying a food company’s supply chain, DNV enables its customers and
their stakeholders to make critical decisions with confidence.

Driven by its purpose, to safeguard life, property, and the environment, DNV helps tackle the
challenges and global transformations facing its customers and the world today and is a trusted
voice for many of the world’s most successful and forward-thinking companies.

WHEN TRUST MATTERS

You might also like