Professional Documents
Culture Documents
25 - 01 - 2019 - Control of
25 - 01 - 2019 - Control of
To cite this article: Reza Dezvareh, Khosrow Bargi & Seyed Amin Mousavi (2016) Control
of wind/wave-induced vibrations of jacket-type offshore wind turbines through tuned liquid
column gas dampers, Structure and Infrastructure Engineering, 12:3, 312-326, DOI:
10.1080/15732479.2015.1011169
Wind/wave-induced vibrations of jacket-type offshore wind turbines (JOWTs) are suppressed by placing a passive vibration
absorber, called tuned liquid column gas damper (TLCGD), on the turbine nacelle. Adopting an ensemble of 75 wind/wave
combinations, three different JOWTs and a SimuLink-based nonlinear model in time domain, main parameters of the
TLCGD are optimised to reach the minimum standard deviation of nacelle displacement. Obtained results indicate that
contribution of the proposed TLCGD is more pronounced in the case of regular excitations such as those from sea waves and
less turbulent winds. Depending on the wind/wave combination, TLCGD can result in reductions up to 45% and 51% in
nacelle displacement standard deviation and maximum acceleration, respectively. As a result, TLCGD deemed to be well
suited to protect fatigue critical JOWTs as well as acceleration-sensitive devices of the nacelle.
Keywords: wind/wave excitation; offshore wind turbines; passive control; tuned liquid column gas dampers
F hydro ¼ F Morison ¼ F D þ F I
_ jv 2 u_ j
¼ 0:5rw C D Aðv 2 uÞ ð7Þ
€
þ rw BðC M v_ 2 ðC M 2 1ÞuÞ:
where, M, C and K stand for mass, inherent damping and lumped at different levels. Mass, stiffness, area and
lateral stiffness matrices, respectively. Moreover, u denote volume matrices are obtained from physical character-
displacement vector and dot operator indicates differen- istics, while damping matrices are obtained from a
tiation with respect to time. Aerodynamic force vector is frequency-independent formulation in which damping
represented by FAero and FHydro is the hydrodynamic force ratio in all modes regulated to be 2%. Detailed discussion
vector. Besides, FTLCGD is the force vector corresponding about frequency-independent damping has been provided
to the TLCGD forces. Further details about aerodynamic by Clough and Penzien (1993).
and hydrodynamic forces are provided in Appendix A. In order to verify the accuracy of the used lumped mass
It should be pointed out that many terms in Equation models in SimuLink, obtained modal properties are
(14) are coupled with each other and conventional finite compared with those of SAP 2000 as presented in
element methods would fail to provide fast computational Tables 3 –5 and illustrated in Figure 4. It is interesting to
tools for these nonlinear-correlated equations. MATLAB note that, effective modal mass at the second mode is
(2008) proved to be a well-suited tool for such computations higher than that of the first mode in all considered JOWTs.
and very large number of analysis (in the order of 104 This is due to the specific stiffness/mass distribution along
analyses or more) can be carried out with reasonable speed the height of the JOWTs. The so-called equivalent static
and without parallel techniques. As a result, in the present procedure is not applicable as response of the structure is
study, Equation (14) is solved through a SimuLink based not well governed by its first mode. As a result, dynamic
model as briefly introduced in Appendix B. procedures, either spectral or time history, are required for
seismic design of JOWTs. Further discussion about this
interesting topic is out of scope of this study.
It can be seen that the lumped mass model is in good
4.2. Verification of the discrete model agreement with its 3D finite element counterpart. This is
As previously described, considered JOWTs include a mainly due to the fact that used stiffness matrices are
5 MW wind turbine with the previously presented obtained from flexibility matrices. Note that offshore wind
characteristics and four-legged support structures which turbines are flexible structures with noticeable rotational
are in different water depths of 50 m, 70 m and 100 m. All responses. Accordingly, stiffness matrix should be
JOWTs are modelled with 11 degrees-of-freedom (DOFs) estimated from the flexibility matrix to account for
Table 3. Modal information obtained from SAP and SimuLink models for JOWT-A (first five modes).
Table 4. Modal information obtained from SAP and SimuLink models for JOWT-B (first five modes).
Table 5. Modal information obtained from SAP and Simulink models for JOWT-C (first five modes).
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u N
u1 X 1X N
s¼t ðxi 2 x Þ2 where x ¼ xi ; ð15Þ
N i¼1 N i¼1
b ¼ 16 m; h ¼ 3 m; Dh ¼ 1 m; Db ¼ 1:67 m;
dL ¼ 0:5:
Figure 4. SAP and lumped mass first three mode shapes of (a) It should be pointed out that the head loss coefficient
JOWT-A, (b) JOWT-B and (c) JOWT-C. of the TLCGD (dL) can also have noticeable effect on
the TLCGD efficiency. However, as reported by Mousavi
rotational DOFs as well. Obtained stiffness matrix is a et al. (2013), optimum head loss coefficients of 0.5– 1
condensed matrix with only translational DOFs (Chopra, result in good efficiency in most cases. As a result a
1995). It should be clarified that adopted discrete model constant value of 0.5 is adopted herein. Note that the
has only lateral translational DOF in each level and frequency ratio is defined to be the ratio of TLCGD
considered wind/wave excitations has the same direction, frequency to the JOWT fundamental (first mode)
as proposed in IEC61400-3 (2009). In addition, considered frequency. Besides, the mass ratio is the ratio of TLCGD
external loads assumed to be perpendicular to the rotor of fluid mass to total mass of the JOWT. Optimisation is
the wind turbine due to the yaw control mechanism of the achieved through a comprehensive parametric study in
nacelle and effect of the aerodynamic force on the tower of terms of frequency ratio (a) and mass ratio (m) with the
the wind turbine is neglected. following lower/upper bands and increments.
0:1 Spaced
a ¼ 0:5 ! 1:4
5. Numerical analysis
TLCGD is a rather new vibration absorber and its major
parameters, namely mass ratio and frequency ratio, are 2% Spaced
m ¼ 1%; 2% ! 10%:
not yet optimised for offshore wind turbines under wind/
wave excitations. To assess TLCGD performance in the
time domain, a criterion beyond merely the maximum An ensemble of 75 different wind/wave combinations
response has to be considered, as the maximum response is considered to cover a broad range of possible loading
8 Structure and Infrastructure Engineering 319
states:
4 Spaced
T pwave ¼ 8 s ! 16 s:
m 4 Spaced m
uwind ¼8 ! 24
s s Note that uwind is the mean wind speed at the nacelle
elevation, Hswave is the significant wave height and
2 Spaced Tpwave is the wave peak period. Obtained results are
H swave ¼ 2 m ! 10 m
summarised in Figure 5 in which as the mass ratio
Figure 5. Mean S.D. Displacement reduction versus a for mass ratios of (a) 1%, (b) 2%, (c) 4%, (d) 6%, (e) 8% and (f) 10%.
320 R. Dezvareh et al. 9
increases, the optimum frequency ratio (the peaks of the 5.2. Contribution of the TLCGD under wind and
presented curves) decreases. Moreover, Figure 5 indi- wave excitations
cates that the higher the mass ratio, the more sensitive 5.2.1. Wind load
the TLCGD’s performance to its frequency. In other
Winds with mean speeds varying from 8 to 24 m/s are
words, as the liquid mass increases, the frequency span
adopted in this section. Note that considered wind speeds
in which a favourable performance was achieved (the
are larger than cut-in and smaller than cut-out speeds.
positive side of the vertical axis in Figure 5), would be
As a result, maximum possible aerodynamic forces are
narrowed. Both of the above-mentioned features are also
expected to occur during the carried out simulations.
the case for TLCGD under seismic excitations (Mousavi
Obtained results in terms displacement reduction at the
et al., 2013).
Figure 6. Mean S.D. Displacement reduction percent under different wind speeds.
10 Structure and Infrastructure Engineering 321
nacelle are illustrated in Figure 6. Note that aforemen- In other words, the waves with height higher than 10 m are
tioned sea states (Hs of 2, 4, 6, 8 and 10 m and Tp of 8, 12 less frequent during annual periods and are considered as
and 16 s) are also considered in this section and each point an ultimate level wave.
on the depicted curves of Figure 6 represents averaged Figure 7 indicates that as the wave height increases the
result of different sea states. efficiency of the placed TLCGD would also increase. This is
due to the fact that when compared with aerodynamic forces,
hydrodynamic forces are rather regular with narrower
5.2.2. Wave load frequency band. As a result, an increase in the hydrodynamic
To investigate wave effects on the TLCGD performance, force leads to more regularity in the structural response and
waves with significant heights of 2, 4, 6, 8 and 10 m and consequently improves damper performance. Note that each
period of 8 s are considered. This range is in accordance point in the represented curves of Figure 7 denotes averaged
with the annual statistics of waves in different seas. result of different wind mean speeds.
Figure 7. Mean S.D. Displacement reduction percent versus different wave heights.
322 R. Dezvareh et al. 11
Figure 8. (a) Periodic and (b) irregular aerodynamic force for the mean speed of 12 m/s.
5.2.3. Effect of wind turbulence intensity contribution of the TLCGD is more pronounced in the
In the earlier sections, aerodynamic forces were case of periodic aerodynamic forces which lead to more
accompanied by their inherent turbulences and it is regular structural vibrations compared with those of
observed that higher irregularities in structural vibration turbulent aerodynamic forces. This is in agreement with
results in lower efficiency of the placed TLCGD. Note that obtained results in Figure 7 in which TLCGD performed
all turbulence intensities were considered based on more effectively in the cases of higher wave heights
IEC61400-3 (2009) in which turbulence intensity varies (higher hydrodynamic forces). This behaviour deemed to
with mean wind velocity. be due to the irregularity compensation made by higher
To achieve a better understanding about sensitivity waves which are more regular.
of the TLCGD to wind turbulence, another parametric From Tables 6 –8, it can be seen that reduction in
study is carried out considering both periodic and displacement standard deviation is more encouraging
turbulent aerodynamic forces. The authors clarify that when compared with that of maximum displacement.
periodic aerodynamic forces came from wind shear, Therefore, TLCGD seems to be more effective in the
rotor spinning, non-axial winds and tower shadow. case of fatigue dominated offshore wind turbines.
Simulated in the FAST code, Figure 8 illustrates periodic Moreover, regardless of wind turbulence and wave
and irregular turbulent aerodynamic forces for a sample height, maximum nacelle acceleration was substantially
wind with the mean speed of 12 m/s. In this section, reduced in the case of wind turbine with TLCGD. The
considered TLCGD has a mass ratio of 6% and aforementioned reduction roughly ranges from 10% to
frequency ratio of 0.8. 50% depending on the loading state. As a result,
The computed aerodynamic forces are imposed on the TLCGD is a well-suited controller in order to improve
SimuLink model, and Figures 9 and 10, respectively, serviceability performance of the nacelle mechanical/
illustrate nacelle displacement and acceleration. Besides, electrical devices.
Tables 6 –8 represent the contribution of the TLCGD in From the obtained results, it is turned out that
terms of reduction percentage of the nacelle response. efficiency of a pre-regulated TLCGD has minor sensitivity
Note that only JOWT-C is considered in this section, and to the height and flexibility of its JOWT. The same
adopted wave heights are 2, 6 and 10 m. Again, conclusion has been made earlier by Mousavi et al. (2013)
12 Structure and Infrastructure Engineering 323
Figure 9. Nacelle displacement under mean wind speed of 12 m/s. (a) Irregular force, Hs ¼ 2 m, (b) periodic force, Hs ¼ 2 m,
(c) irregular force, Hs ¼ 6 m, (d) periodic force, Hs ¼ 6 m, (e) irregular force, Hs ¼ 10 m, and (f) periodic force, Hs ¼ 10 m.
Figure 10. Nacelle acceleration under mean wind speed of 12 m/s. (a) Irregular force, Hs ¼ 2 m, (b) periodic force, Hs ¼ 2 m,
(c) irregular force, Hs ¼ 6 m, (d) periodic force, Hs ¼ 6 m, (e) irregular force, Hs ¼ 10 m, (f) periodic force, Hs ¼ 10 m.
324 R. Dezvareh et al. 13
Table 6. Nacelle displacement and acceleration of JOWT-C, under mean wind speed of 12 m/s and Hs ¼ 2 m.
Table 7. Nacelle displacement and acceleration of JOWT-C, under mean wind speed of 12 m/s and Hs ¼ 6 m.
Table 8. Nacelle displacement and acceleration of JOWT-C, under mean wind speed of 12 m/s and Hs ¼ 10m.
for offshore jacket platforms under seismic excitations. this study, is the ratio of TLCGD mass to total mass of the
As a result, the authors believe that reported results of this JOWT. In general, as the mass ratio increases, the
study can be generalised for other JOWTs as well. Based optimum frequency ratio decreases which is in agreement
on this study, to achieve a reliable and robust TLCGD, with earlier studies. Moreover, TLCGDs with higher mass
a mass ratio of 4– 6% with frequency ratio of 0.8– 0.85 ratios are more sensitive to their frequency ratios such that
can be proposed. they can impose destructive effects in the cases of ill-
regulated frequencies. It should be pointed out that
observed optimum frequency ratios are smaller than those
commonly used in other structures. This can be attributed
6. Conclusions to the turbulent nature of the considered excitations as well
This study attempts to reduce vibrations of JOWTs under as pitch/stall regulations of the turbine which lead to
wave and wind excitation through a rather new vibration further distraction in the wind thrust.
absorber called TLCGD. Considering water – structure The comparative study revealed that contribution of
interaction and nonlinear damping mechanism of the TLCGD would be more pronounced in the case of regular
TLCGD, a SimuLink model is developed based on the excitations such as sea waves and winds with low
correlated equations of motions of the JOWT –TLCGD turbulence intensity. Accordingly, TLCGD is more
system. Considering different wind/wave combinations, suitable in offshore area with higher wave heights and
frequency and fluid mass of the TLCGD are optimised lower wind turbulence. Finally the authors point out that
through a parametric study for three JOWTs with different TLCGD is well suited for fatigue critical JOWTs as it
heights. leads to more reduction in standard deviation of
The obtained results indicate that TLCGD would displacements compared with the maximum displace-
perform better in the case of higher mass ratios. However, ments. For example, obtained results indicated that
no noticeable improvement would be achieved for mass TLCGD could reduce standard deviation of nacelle
ratios more than 4%. Note that the defined mass ratio, in displacement up to 45%, while the corresponding
14 Structure and Infrastructure Engineering 325
reduction in maximum displacement was 18%. Observed Lackner, M.A., & Rotea, M. (2011a). Passive structural control
reductions are highly sensitive to the imposed wind/wave of offshore wind turbines. Wind Energy, 14, 373– 388.
doi: 10.1002/we.426
combination. Moreover, maximum nacelle accelerations
Lackner, M.A., & Rotea, M. (2011b). Structural control of
can be dramatically reduced (up to 50%) by the TLCGD. floating wind turbines. Mechatronics, 21, 704 – 719.
This feature would improve serviceability of mechanical/ doi: 10.1016/j.mechatronics.2010.11.007
electrical devices of the wind turbine which are commonly Laya, E.J., Connor, J.J., & Sunder, S.S. (1984). Hydrodynamic
lumped at the nacelle. Finally, it should be pointed out that forces on flexible offshore structures. Journal of Engineering
Mechanics, 110, 433– 448. doi: 10.1061/(ASCE)0733-9399
TLCGD is an evolving technology and more numerical,
(1984)110:3(433)
experimental and case studies in this field are welcome. Linder-Silvester, T., & Schneider, W. (2005). The moving
contact line with weak viscosity effects- an application and
evaluation of Shikhmurzaev’s model. Acta Mechanica, 176,
254– 258.
Disclosure statement
Mahadik, A.S., & Jangid, R.S. (2003). Active control of offshore
No potential conflict of interest was reported by the authors. jacket platforms. International Shipbuilding Progress, 50,
277– 295.
MATLAB (2008). User guide, Simulink version 7.6.0. Natick,
MA: MathWorks Inc..
Notes Mousavi, S.A., Bargi, K., & Zahrai, S.M. (2013). Optimum
1. Email: kbargi@ut.ac.ir mailto:rdezvareh@ut.ac.ir parameters of tuned liquid column-gas damper for mitigation
2. Email: s.a.mousavi@ut.ac.ir of seismic-induced vibrations of offshore jacket platforms.
Structural Control and Health Monitoring, 20, 422– 444. doi:
10.1002/stc.505
Mousavi, S.A., Zahrai, S.M., & Bargi, K. (2012). Optimum
References geometry of tuned liquid column-gas damper for control of
Chang, C.C., & Hsu, C.T. (1998). Control performance of liquid offshore jacket platform vibrations under seismic excitation.
column vibration absorbers. Engineering Structures, 20, Earthquake Engineering and Engineering Vibration, 11,
580– 586. doi: 10.1016/S0141-0296(97)00062-X 579– 592. doi: 10.1007/s11803-012-0143-z
Chopra, A.K. (1995). Dynamics of structures. Theory and Patil, K.C., & Jangid, R.S. (2005). Passive control of offshore
applications to earthquake engineering. Englewood Cliffs, jacket platforms. Ocean Engineering, 32, 1933 – 1949.
NJ: Prentice Hall. doi: 10.1016/j.oceaneng.2005.01.002
Clough, R.W., & Penzien, J. (1993). Dynamic of structures. Seidel, M. (2007). Jacket substructures for the REpower 5 M
New York, NY: McGraw-Hill. wind turbine. In Conference Proceedings European Offshore
Colwell, S., & Basu, B. (2009). Tuned liquid column dampers in Wind 2007. Brussels: European Wind Energy Association
offshore wind turbines for structural control. Engineering (EWEA).
Structures, 31, 358 – 368. doi: 10.1016/j.engstruct.2008. Soong, T.T., & Dargush, F. (1997). Passive energy dissipation
09.001 systems in structural engineering. Chichester: Wiley.
Dean, R.G., & Dalrymple, R.A. (1992). Water wave mechanics Soong, T.T., & Spencer, B.F. (2002). Supplemental energy
for engineering and scientists. Farrer Road, Singapore: dissipation: State-of-the-art and state-of-the-practice.
World Scientific Publishing Co.. Engineering Structures, 24, 243–259. doi: 10.1016/S0141-
Dong, W., Moan, T., & Gao, Z. (2011). Long-term fatigue 0296(01)00092-X
analysis of multi-planar tubular joints for jacket-type Stewart, G., & Lackner, M.A. (2011). The effect of actuator
offshore wind turbine in time domain. Engineering dynamics on active structural control of offshore wind
Structures, 33, 2002– 2014. doi: 10.1016/j.engstruct.2011. turbines. Engineering Structures, 33, 1807– 1816. doi: 10.
02.037 1016/j.engstruct.2011.02.020
Hansen, M.O.L. (2008). Aerodynamics of wind turbines. London: Vestergaard, J., Brandstrup, L., & Goddard, R.D. (2004). A brief
Etherscan. history of the wind turbine industries in Denmark and the
Hochrainer, M.J., & Ziegler, F. (2006). Control of tall building United States. In Academy of International Business
vibrations by sealed tuned liquid column dampers. Structural (Southeast USA Chapter) Conference Proceedings, Stock-
Control and Health Monitoring, 13, 980– 1002. doi: 10.1002/ holm (pp. 322 – 327). East Lansing, MI: Academy of
stc.90 International Business.
International Electrotechnical Commission (2009). Wind tur- Zhang, Z.L., Chen, J.B., & Li, J. (2014). Theoretical study and
bines-Part 3: Design requirements for offshore wind turbines. experimental verification of vibration control of offshore
Geneva: IEC (IEC International Standard 61400-3). wind turbines by a ball vibration absorber. Structure and
Jonkman, J.M., & Buhl, M.L. Jr (2005). FAST user’s guide Infrastructure Engineering, 10, 1087– 1100. doi: 10.1080/
(NREL/EL-500-38230). Golden, CO: National Renewable 15732479.2013.792098
Energy Laboratory.
Jonkman, J.M., Butterfield, S., Musial, W., & Scott, G. (2009). Appendix A
Definition of a 5-MW reference wind turbine for offshore
system development (NREL/TP-500-38060). Golden, CO: In order to achieve time domain descriptions of aerodynamic and
National Renewable Energy Laboratory. hydrodynamic forces exerted on the system, the wind velocity
Kelley, N.D., & Jonkman, B.J. (2007). Overview of the TurbSim and water level variations in time domains are required.
stochastic inflow turbulence simulator. Version 1.21 (NREL/ To calculate time history of wind speed, Kaimal spectrum is
TP-500-41137). Golden, CO: National Renewable Energy adopted which can be described by Equation (A-1). The Kaimal
Laboratory. spectrum is a function of frequency ( f), average wind velocity
326 R. Dezvareh et al. 15
(Vmean), length scale (l), and the turbulence intensity factor It of water particles in different depths are calculated and are used
(Hansen, 2008): in the Morrison formulation to obtain the corresponding
hydrodynamic forces:
I 2t V mean l
Sðf ÞKaimal ¼ 5 : ðA 1Þ
ð1 þ 1:5 V mean
fl
Þ3 5 H 2s 5 24
Sðf ÞPierson – Moskowitz ¼ exp 2 fT p : ðA 2Þ
16 T 4p f 5 4
In this study, using Turbsim (Kelley & Jonkman, 2007)
which is one of the subsets of FAST code, the turbulence
intensity factor, shear effect, and other contributing factors are Appendix B
considered so as to extract the time domain wind velocity for Figure A-1 illustrates the developed SimuLink model. This
different mean wind velocities. Note that Turbsim uses random model is built based upon the governing equation of motion of the
phases with uniform distribution to obtain wind speed time series. JOWT– TLCGD system, Equation (14). The derived model is
Having wind speeds in time domain, FAST is used to calculate indeed the MDOF model of the JOWT and TLCGD while taking
the aerodynamic force based on the so called BEM theory. into account the interaction between water and the
In order to calculate the time domain of water level, Pierson – support structure. In Figure A-1(a), a screen shot from the
Moskowitz spectrum, which is expressed in Equation (A-2), is model is depicted in which the main three blocks, called main
used. This spectrum is a function of the significant wave height structure, TLCGD and hydrodynamic, are represented. Details of
(Hs), the wave peak period (Tp), and frequency ( f). Using the each block are depicted in Figures A-1(b) – (d). Note that
above-mentioned spectrum and the Inverse Fast Fourier Trans- these blocks are correlated with each other through the closed
form (IFFT), the time history of water level variations would be loops. Direct inputs of the model are the time domain vectors of
obtained. Again, during the IFFT, random phases with uniform the aerodynamic force (obtained from FAST) as well as
distribution are adopted. Subsequently, adopting the so called water level variations (obtained from Pierson – Moskowitz
Airy theory (Dean & Dalrymple, 1992), velocity and acceleration spectrum).