Seismic Load Evaluation of Wind Turbine Support Structures with

PO. 167 Consideration of Uncertainty in Response Spectrum and Higher Modes

T. Ishihara, K. Takamoto, M. W. Sarwar

Department of Civil Engineering, The University of Tokyo

1

Objectives that are significantly low damped, this model fails to 400kW
& 2MW
k +1
estimate large fluctuations of the spectral acceleration  zi  3
γ j X ij = ∑ c jk  
0.8

Rapid development of wind energy in seismically active in order to establish reliable design spectrum. k =1 H 0.6

regions like Japan requires evaluation of design seismic

load of wind turbine support structures to ensure Model for Damping Correction Factor Coefficients and time period ratio 0.4

c jk
Tj / T1
structural integrity. Analytical estimation of the design j k 1 2 3 1st Mode
To account for excessive fluctuations in the response 1 1.1 0 0 1
0.2
2nd Mode
loads is usually carried out by the response spectrum 2 5.87 -6.00 0 0.127 3rd Mode
spectrum of low damped systems, damping correction 3 14.26 -38.20 24.00 0.043
Equation

method that encounters two problems when used for 0

-1.2 -0.8 -0.4 0 0.4 0.8 1.2
factor is proposed as a function of spectral uncertainty (γ),
load analysis of wind turbine support structures [1]. Fig. Model participation function
natural period (T) and damping ratio(ζ) so that,
These support structures are extremely low damped Significant contribution to the design load by 2nd and 3rd
that experience a wide range of frequencies when α
 
Fζ (ζ ,T ,γ ) f (T , γ )
7 modes is observed for 2MW turbine as shown below.
= =  ,...α
subjected to seismic activities. Response spectrum for  2 + 100ζ  SRSS method is used for superposition of these modes.
 
such low damped structures show excessive fluctuation
and such uncertainty in response spectrum can not be A set of 35 seismic waves, with observed and random 1

Seismic load for each mode: 2MW

captured by existing models of the damping correction phases, is used to evaluate uncertainty in acceleration n n

∑ ∑γ
0.8

response spectrum for damping ratios ranging from 0.5 =

Qij = F j X kj S a (T j , ζ )mk

Height Ratio
factors defined in Eurocode[2] and BSL[3]. In addition, kj
=k i=
use of the simplified SDOF model suggested by IEC[4] to 5%. k i
n
0.6

results in linear vertical load profiles. However, vertical 5000 1

=M ij ∑F
k =i
kj ( z k − zi )
0.4
Hachinohe
distribution of the seismic loads is found to be largely Taft

Cumulative Relative Frequency

SRSS mode superposition: 1st Mode
affected by the higher modes [1] of wind turbines. 4000 Elcentro
Kobe
0.8 0.2
2nd Mode
n n 3rd Mode
Therefore simplified but accurate analysis method to 3000 0.6
=Qi = ∑
Qij 2 , Mi ∑ M ij 2 0
Acc [gal]

estimate design load profiles is desired. =j 1 =j 1

-1500 -1000 -500 0 500 1000 1500

In this research, model for damping correction factor 2000 ζ=0.5% 0.4
Shear Force (KN)

that accounts for uncertainty in response spectrum, and Section IA 1 1

1000 0.2
modal participation functions encompassing complex IB
Section IB
Section IC
Log Normal Dist 0.8
vertical distribution of seismic loads are proposed. The 0
IA IC
0
0.8

1000 104
accuracy and reliability of the proposed method for 0.1 1

Height Ratio

Height Ratio
Time Period T[s] 0.6 0.6

evaluation of seismic design loads is examined against Fig. Segments for Statistical analysis Fig. CRF of acc. for each section
time history analysis and current design codes. 0.4 0.4

The log normal distribution is found to have well defined 0.2 0.2

Response Spectrum Method uncertainty in all sections of the acceleration response Previous Model
Proposed Model
Previous Model
Proposed Model
Time History Analysis Time History Analysis
spectrum as shown in above figure. Investigations to 0
0 0.2 0.4 0.6 0.8 1
0
0 0.2 0.4 0.6 0.8 1

Eqation of motion for jth mode of a MDOF system is, identify exponent α has shown linear relationship with Shear Force Ratio Moment Ratio
quantile (γ) and natural period (T) that lead to following:
q j + 2ζ jω j q j + ω j q j =
 2
−γ j 
xg Distribution of calculated load has successfully captured
α
 
Here，ωj，ζj and γj are natural frequency, damping ratio Fζ ζ ,T ,γ = 
7
(
 , α= −0.07T)+ 0.7γ + 0.5
non-linearity of the load profiles obtained by time
 2 + 100ζ 
 history analysis.
and mode participation factor of jth mode. Force for
each mode of vibration is calculated as follows: 4000
Determination of γ by Code Calibration Method
Fij = γ j X ij Sa (T j , ζ )mi γ=0.2
γ=0.5
γ=0.8
3000
Proposed Eq To determine suitable quantile (γ) for defining reliable
Seismic force depends upon: design spectrum, code calibration method[6] is adopted.
Acc [gal]

a. acceleration of response spectrum (Sa) of SDOF In this study, seismic waves for obtaining the structural
b. modal participation function (γjXij) 2000
ζ=0.5% design certification in Japan are used.
400KW 2MW
Design acceleration response spectrum[2] is defined as: 1 1

 T 1000 Hachinohe Hachinohe

a 0 ⋅ S ⋅ {1 + ⋅ ( β 0 ⋅ Fζ − 1)} (0 ≤ T ≤ TB ) 0.8
Taft
Elcentro 0.8
Taft
Elcentro
Kobe
 TB 
Kobe
γ=0.7

S a (T , ζ ) = a 0 ⋅ S ⋅ Fζ ⋅ β 0 (TB ≤ T ≤ TC )
Height Ratio
Height Ratio

0 0.6 0.6

 0.1 1

a 0 ⋅ S ⋅ Fζ ⋅ β 0  TC  Time Period T[s] 2MW

⋅  (TC ≤ T ) Fig. Variation of response spectrum with γ-values
0.4 0.4
 T 
0.2 0.2 400KW
Parameters for a return period of 500 years[5] Proposed damping correction factor shows good
agreement with analytical one for all quantiles. Also 0 0
0 500 1000 1500 2000 2500 0 2 104 4 104 6 104 8 104 1 105
proposed model performs well for both low and highly Shear Force (KN) Moment (KNm)
3.2 1.5 2.5 0.16 0.576 damped structures whereas EuroCode underestimates
A γ-value of 0.7, i.e., 70% quantile, is identified to obtain
where ao is design ground acceleration, S is soil amplification spectral acceleration for low damped systems.
factor and Fζ is damping correction factor. reliability level similar to that of the current design code.
3000 3000

Eurocode[2] defines the damping correction factor as a Eurocode Eq Eurocode Eq

function of damping ratio so that, Proposed Eq Proposed Eq

Analysis
Conclusions
Analysis

α
Acc [gal]

Acc [gal]

2000 2000 Wind Turbines

  This study proposes a model for the damping correction
Fζ (ζ )
 = 0.5%
7
=   , α = 0.5
 2 + 100ζ
 Buildings factor that accounts for uncertainty in the response
  = 4%

5000
1000 1000 spectrum and natural period of wind turbine. In addition,
ElCentro formula for analytical estimation of complex profile of
Taft

4000
Hachinohe
Kobe
seismic design loads are presented. Finally accuracy of
proposed formula is verified against time history
Acceleration [gal]

0 0
0.1 1 0.1 1
Time Period T[s] Time Period T[s]
3000
analysis and reliability level similar to that of current
Fig. Comparison of current and proposed model design code is demonstrated.
EuroCode
2000 ζ = 0.5.%

Formula for Vertical Profile of Seismic Loads References

1000 1. T. Ishihara, M. W. Sarwar, “Numerical and Theoretical Study on Seismic Response of Wind
EuroCode
ζ = 5%
Estimation of design loads require natural periods and Turbines”, Proc. of EWEC 2008, 2008.

modal participation function (γjXij) of the dominant modes.

2. Eurocode 8: Design of structure for earthquake resistance;Part1:Genersl rules, seismic actions
0 and rules for buildings,1998-1:2004.
0.05 0.1 1 5.0
Time Period T[s] Since first three modes accounted for modal mass of 3. BSL (2004), “The Building standard law of Japan”, The building centre of Japan. (in Japanese)
Fig. Response spectrum for Level II earthquake 4. IEC61400-1: Wind turbines, Part 1,Third edition.
85%[4], expression for calculating the respective model 5. JSCE (2007), “Guidelines for design of wind turbine support structures and foundations”, Japan
However, in case of the wind turbine support structures participation function and coefficients are listed below: 6.
society of civil engineers. (in Japanese)
M. Hoshiya, K. Ishii:, “Reliability based design for structures”, Kajima Press, 1986.

EWEA OFFSHORE 2011, 29 November – 1 December 2011, Amsterdam, The Netherlands