Вы находитесь на странице: 1из 8

International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 234

Effect of Initial Geometrical Imperfections on Buckling Strength and Design of Offshore


Wind Turbine Tower
Nu Nu Lwin1 and Bo Song2
1
School of Civil and Environmental Engineering, Univ. of Science and Technology Beijing,
China, P.O. Box 100083. E-mail: nunulwinjuly@gmail.com
2
School of Civil and Environmental Engineering, Univ. of Science and Technology Beijing,
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

China, P.O. Box 100083. E-mail: songbo@ces.ustb.edu.cn

ABSTRACT
Buckling of a wind turbine tower with different imperfections is studied through linear and
nonlinear numerical analysis, and the effect of the initial imperfections of the wind turbine tower
on the buckling capability under lateral loads is also investigated. The maximum initial
imperfections which are derived from EN 1993-1-6 and DNV-OS-C 401 are added into the
nonlinear analysis through the first buckling mode. It is shown that these initial deformations
significantly reduce the load carrying capability of the structure while linear buckling analysis
gives no significant difference in the buckling resistance under the different types of
imperfections. The design checks are also performed for four different classes of imperfections
based on nonlinear analysis results. It can be concluded that the incorporation of the initial
imperfections and the lateral loads into the tower models which form the basis for design criteria
should be emphasized.

INTRODUCTION
The failure of a wind turbine tower is usually due to buckling, because tower is a long
slender structure and there is wind load in the lateral direction. Evidence of this type of damage
of wind turbine tower was compiled by in (Lee, 2012). Because of the catastrophic buckling
failure, the reliable prediction of buckling strength of wind turbine towers is important.
From perspective of engineering, imperfections such as geometrical and structural
imperfections are unavoidable, which affect the relationship between the applied loads and
deflections (Prabu et al. 2010). Silva presented that the nonlinear buckling analysis should be
performed by considering the imperfect geometry (Silva, 2006).
It is presented in (Shahzad et al. 2007) that geometric imperfection has major effect on
buckling strength and behaviour of shells under different loading conditions of compression,
bending, torsion and pressure. Guo (Guo et al. 2011) studied the influence of initial geometry
imperfections of wind turbine tower by using pushover analysis and it was concluded that the
initial geometry deformation has a little influence on the bearing capacity and ductility.
In present paper, linear buckling analyses of different types of imperfections are carried out
to determine the worst buckling mode which would be served as an imperfect model for
nonlinear analysis. The first eigenmode multiplied by a factor is chosen to introduce to the
nonlinear analysis as maximum equivalent imperfections whereas the design checks are
performed for imperfect towers based on the results of nonlinear analysis.
The purpose of this research is to study how imperfections affect buckling capability of the
offshore wind turbine tower under code provisions (Eurocodes and DNV). It is also expected that
this study will verify that the initial geometrical imperfections should be incorporated carefully
in the design of offshore wind turbine tower.

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 235

DESCRIPTION OF STRUCTURE AND FINITE ELEMENT MODEL


The 86 m vertical steel tower is assembled by thin-wall cylindrical and conical pieces with
varying diameters and wall thicknesses (20 mm at the top and 60 mm at the base) and it is made
up of Q345 steel. The technical data and finite element models can be seen in Table 1 and Figure
1, respectively.

Table 1. Technical data of wind turbine supported structure.


Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

Hub Rotor Tower top Tower base Weight (kg) Total weight
height (m) diameter (m) diameter (m) diameter (m) Nacelle and rotor Tower (kg)
86 115.0 3.07 5.0 200,360 322,947 523,307

ANSYS is utilized to perform the evaluation of buckling strength of offshore wind turbine
tower (OWT). The geometry model of tower and blades is a 4-node elastic shell (SHELL 181)
which has 6 DOF and the nacelle model is SOLID45.
It is common practice to treat the tower like a structure ‘fixed’ at the base with no
consideration of the soil-foundation interaction. In present paper, the wind turbine tower has an
end fixed condition at the mud line.

Figure 1. Offshore wind turbine tower model and loading condition.


At the extreme wind speed Ve50 of 52.5 m/s (DNV-OS-J101), the rotor thrust of 1007.724 kN
is applied as a concentrated load at the elevation of hub height for the non-spinning rotor as our
turbine’s cut-out speed is reached before the wind speed reaches Ve50 and the wind pressure
distribution is also considered according to EN 1991-1-4 (2005).
Wind direction and wind pressure distributions on the circumference along the height of the
tower can be seen in Figure 1. Wave induced forces are applied to the tower between sea bed and
wave crest, which is 5.13 m height above the sea bed. Considering extreme high water level (50
years of recurrence), the total calculated force induced by waves is 250.65 kN.

LINEAR BUCKLING ANALYSIS OF PERFECT AND IMPERFECT TOWER MODELS


For real structures, linear buckling analyses (LBA) are the best way to study the effects of
various parameters. Linear buckling analysis is performed on four computational models, namely
perfect shell and three imperfect shells with out-of-roundness, accidental eccentricity and
dimples under extreme wind load combination (G+1.35W), G is gravity load and W is

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 236

combination of wind and wave loads.


Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

Figure 2. Different types of imperfections and first-eigenmode comparison of perfect and


imperfect models.
The introduction of the out-of-roundness imperfections to the FE model is achieved by
transforming the circular cross-sections into ellipse shapes with the maximum diameter
perpendicular to the load direction. Accidental eccentricity imperfections can be introduced by
inputting the eccentricity to the flange connection of perfect shell without changing other
geometry. In the case of dimple imperfection, the first buckling eigenmode is multiplied by a
factor, in order to accomplish the dimple tolerance parameter. (Veljkovic 2012)
As demonstrated in Figure 2 and Table 2, the buckling modes and eigenvalues of the perfect
and imperfect shell configurations are almost identical for out-of-roundness and eccentricity
imperfections while the dimple imperfections added via first buckling mode reduce the buckling
capacity significantly.
By investigation on these effects of different types of imperfections, the worse pattern of
geometry imperfections for buckling strength can be known. So, dimple imperfections produced
by factored first-eigenmode will be used in further investigations.

Table 2. Tolerance parameters and eigenvalues comparison of perfect and imperfect


towers.
Out-of-roundness Eccentricity Dimple
Class description Perfect
Ur=(dmax−dmin)/dnorm ea (m) δw0 (m)
Tolerance - 0.007 0.002 0.011
Eigenvalue, rRcr 9.695 9.592 9.688 9.182
Decreased % - 1.07 % 0.07 % 5.59 %

NONLINEAR BUCKLING ANALYSIS OF PERFECT MODEL


The nonlinear analysis is performed on the perfect structure to determine the perfect elastic-
plastic buckling resistance ratio rGMNA that is calculated by dividing the limit load by design load.
The arc-length method which cloud provide the nonlinear static equilibrium solutions for
unstable problems is adopted in this study.
For structural analysis, nonlinear effect of material will be considered in this model based on
bilinear stress-strain model. Stress-strain relations are shown in Figure 3 in which elastic
modulus is 206 GPa, yield strength is 345 MPa and tangent modulus is 20.6 GPa.

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 237
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

Figure 3. Stress and strain relation.

Figure 4. Load-deflection curve of perfect model exhibiting local buckling.


The characteristic buckling resistance ratio for the perfect shell estimated by Geometrically
and Materially Nonlinear Analysis (rGMNA) is 1.246 with the maximum load carrying capacity of
(G+2.69W) in which G is gravity load and W is a combination of wind and wave loads. As
described in Figure 4, the tower is in the elastic range until critical point (Pcr) is reached. Beyond
this critical point, increase of load leads to plasticity whereas the failure phase of tower begins
beyond limit load (Pmax). The von Mises stress and von Mises strain of the OWT at the limit load
are shown in Figure 5.

Figure 5. Von Mises stress and strain of wind turbine tower at limit load (Pmax).

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 238

EFFECTS OF INITIAL GEOMETRIC IMPERFECTIONS UNDER LATERAL LOADS


In order to clarify this effect, the initial imperfections will be added into nonlinear analysis
via small changes in geometry in which the first eigenvalue mode shown in Figure 2 is used. The
maximum imperfections are taken into account according to EN 1993-1-6 (2007) and DNV-OS-
C 401 (2010).
Imperfection (Class A, B and C): The EN 1993-1-6 proposes the imperfection which can be
introduced to the finite element model by equivalent geometrical imperfection in the position
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

where the first eigenmode occurs. The amplitude of equivalent geometrical imperfection (δmax) is
taken as the larger value of
δeq,1  lg Un,1 or δeq,2  n i tUn,2 (1)
where lg is the gauge length (seen in Figure 6); t is thickness; ni is the multiplier for appropriate
tolerance level (taken as 25); and Un,1 and Un,2 are dimple imperfections parameters which were
taken as 0.01, 0.016 and 0.025 for the fabrication tolerance quality class A, B and C,
respectively.

Figure 6. Measurement of equivalent imperfection.


After determining δmax, the first buckling eigenmode is multiplied by a factor, in order to
accomplish the dimple imperfection with this equivalent imperfection amplitude.
Imperfection (DNV): It is noted in DNV-OS-C-401 that the maximum imperfection should
be within the following tolerance:
0.01 lg
δ max  (2)
1  lg / r
where r is the radius, lg is the gauge length which can be taken the smaller of 1.15(l)1/2(rt)1/4 or
πr/2 in which l is the distance between rings or bulkhead and t is the thickness. The calculated
δmax is introduced to the FE model by multiplying the first buckling eigenmode by a
corresponding factor.
Effect of imperfections on buckling strength and design: Eurocodes propose three
different classes of imperfection (Class A, B and C) while DNV defines only one class of
geometrical imperfection which is almost similar to class A of Eurocodes. The amplitudes of the
equivalent geometrical imperfection calculated by Equations 1 and 2 are imbedded in the FE
model for each class via first-eigenmode multiplied by a factor.
Nonlinear effect of material is also considered in the model. In these analyses, the design
load (G+1.35W) progressively increased until the limit load, beyond which the failure phase
begins, is reached. This limit load ratio rGMNIA can be calculated by dividing the limit load by
design load.
As shown in Figure 7, the limit loads varies according to different imperfection classes,
excluding Class A and DNV. At the limit load of each imperfection class, the von Mises stresses
are quite the same in the middle portion and top of the tower although those of the upper and
lower portions are significantly different (seen in Figure 8).

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 239
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

Figure 7. Load-deflection curves of different imperfections.

Figure 8. Stress distribution along the height of tower.


However, the geometrical imperfections of Class A, B, C and DNV reduce significantly the
buckling strength of perfect model by approximately 16%, 20%, 25% and 18 % respectively.
The comparison of limit load ratios of perfect, Class A and DNV are shown in Figure 9.

Figure 9. Comparison of limit loads of perfect and imperfect tower.


To examine the effect of each imperfection, the design verification for each imperfection
class is carried out according to EN 1993-1-6, i.e.

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 240

k GMNIA  rGMNIA
rRd   1.0 (3)
γ ml
where rRd is the design buckling resistance ratio, the factor of safety γm1 should not have a value
lower than 1.1 and kGMNIA is the calibration factor depending on the results of reliability check.
So, the factor kGMNIA = 0.95 which was derived from the results of buckling and post-buckling
analysis in ANSYS mechanical APDL technology demonstration guide is adopted in Equation 3
(ANSYS, 2013). The design verifications for four imperfection classes are shown in Table 3.
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

Table 3. Effect of imperfection on the buckling strength and the design of OWT.
Class δmax (m) Limit load rGMNIA rGMNIA/rGMNA rRd Effect on design
A 0.018 G+2.24W 1.164 0.93 1.01 Satisfied
B 0.028 G+2.14W 1.145 0.92 0.99 Not satisfied
C 0.044 G+2.08W 1.123 0.90 0.97 Not satisfied
DNV 0.021 G+2.21W 1.158 0.93 1.00 Satisfied

Based on the results obtained, it can be clearly seen that the effect of imperfections should be
emphasized for determining buckling strength and design. Moreover, since the ratios of rGMNIA to
rGMNA for all classes of imperfections is much larger than the elastic imperfection reduction
factor (α=0.5), it can be said that the shell has entered into plastic state, before reaching limit
load.

CONCLUSIONS
Linear buckling and nonlinear buckling analyses are carried out on offshore wind turbine
tower, considering geometrical imperfections proposed by Eurocodes and DNV. The results of
this study emphasize the need for incorporation of the initial imperfection under lateral loads for
design criteria.
(1) Based on linear buckling analysis of imperfect shells (out-of-roundness, accidental
eccentricity and dimples), the results indicate that the influences of out-of-roundness and
eccentricity imperfections are insignificant to the buckling resistance of the tower.
(2) Compared to other types of imperfections, the dimple imperfection is the worse pattern
for the buckling strength and this imperfection is introduced to OWT model in nonlinear
buckling analysis.
(3) For geometrically and materially nonlinear imperfection analysis (GMNIA), the
introducing of imperfection is straightforward by multiplying first-eigenmode by a factor,
in order to accomplish the equivalent imperfections proposed in Eurocodes (imperfection
Class A, B and C) and DNV-OS-C 401(DNV).
(4) Based on nonlinear buckling analysis, the significant differences are found in the limit
loads for different imperfection classes. It can be said that the buckling strength of perfect
model was significantly reduced due to equivalent imperfections.
(5) The limit loads and failure patterns of the tower with imperfections (Class A and DNV)
are almost identical since the maximum imperfections (δmax) of Class A (Eurocode) and
DNV are closely similar and their design buckling resistance ratios rRd satisfy the design
requirement under maximum loading conditions.
(6) The other imperfection classes (Class B and Class C) should also be paid attention
whether these imperfections fulfill the design requirements. If not, measure like
straightening should be taken to improve the imperfections to Class A carefully.

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016


International Collaboration in Lifeline Earthquake Engineering 2016 IRP 1 241

Through this study, it can be concluded that initial imperfections should be of great concern
to the designers.

ACKNOWLEDGEMENTS
This work is supported by the Overseas Famous Teachers Project of Education Department
(MS2011BJKJ005).
Downloaded from ascelibrary.org by University of Sunshine Coast on 09/16/17. Copyright ASCE. For personal use only; all rights reserved.

REFERENCES
ANSYS, (2013). Inc. ANSYS mechanical APDL technology demonstration guide, USA.
DNV-OS-C 401, (2010). Fabrication and testing of offshore structures, Det Norske Veritas:
Offshore Standard.
DNV-OS-J 101, (2013). Design of Offshore Wind Turbine Structures, Det Norske Veritas:
Offshore Standard.
EN 1991-1-4, (2005). Eurocode 1: Actions on structures. Part 1–4 General actions: Wind
actions, European Committee for standardization.
EN 1993-1-6, (2007). Eurocode 3: Design of steel structures. Part 1–6 Strength and stability of
shell structures, European Committee for standardization.
Guo, L. et al., (2011). “Pushover analysis of a 53 m high wind turbine tower.” Advanced Science
Letters, 4: 1–7.
Kang-Su Lee and Hyung-Joon Bang (2012). “A Study on the Prediction of Lateral Buckling
Load for Wind Turbine Tower Structures”. International Journal of Precision Engineering
and Manufacturing, 13(10): 1829–1836.
Prabu, B. et al., (2010). “Parametric study on buckling behaviour of thin stainless steel
cylindrical shells for circular dent dimensional variations under uniform axial compression”.
International Journal of Engineering Science and Technology, 2: 134–149.
Shahzad, K. et al., (2007). “Effect of geometric imperfection on buckling strength of cylindrical
shells”. Failure of Engineering Materials and Structures, Code 40.
Silva, V.D., (2006). Mechanics and Strength of Materials, In Springer, Netherland.
Veljkovic, M. et al., (2012). High-strength tower in steel for wind turbines (HISTWIN),
Directorate-General for Research and Innovation, European Union.

© ASCE

International Collaboration in Lifeline Earthquake Engineering 2016

Вам также может понравиться