Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Ocean Engineering 172 (2019) 446–455

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Moment loads acting on a blade of an ocean current turbine in shear flow T


Kazuomi Yahagi, Ken Takagi ∗

Dept. of Ocean Technology, Policy, and Environment, The University of Tokyo, Kashiwa, Japan

ARTICLE INFO ABSTRACT

Keywords: The Kuroshio flowing near Japan is one of the largest ocean currents and it is considered that this energy possibly
Current turbine be an important energy source for Japan. A full-scale generator with twin 40m turbine rotors is planned for at
Moment load the rated flow speed of 1.5 m/s. Estimation of the blade loads are crucial for such big turbines. We focus on the
Blade moment in the shear flow effect of shear flow on an ocean current turbine and highlight the periodic fluctuation of moment loads at the
CFD of turbine blade
blade root induced by the shear flow. Shear flow experiments and numerical computation with a commercial
code were conducted for estimating the moment loads on a blade. The numerical computation was verified
through the comparisons with the measured Cp and Ct in uniform flow. The numerical results of moment load in
the shear flow are validated through the comparison with the experimental results. It is concluded that the
moment loads are well estimated by the present CFD method. Using the same CFD method, some applications are
shown.

1. Introduction et al., 2014) measured the vertical profile of the Kuroshio near Miyake
Island, and they reported that there is a strong shear flow near the free
The Kuroshio flowing near Japan is one of the largest ocean currents surface. Thus, we focus on the shear flow effect in this paper, and the
and it is considered that this energy possibly be an important energy periodic fluctuation of moment loads at the blade root induced by the
source for Japan. The New Energy and Industrial Technology shear flow is highlighted because the bending moment is one of the
Development Organization (NEDO) and IHI corporation have devel- important parameters for the structural design of a blade.
oped a 100 kW class ocean current turbine system named “Kairyu” and In the field of wind turbines, the National Renewable Energy
successfully conducted a sea trial at the Tokara Islands in 2017 (IHI, Laboratory (NREL) summarized test results of the blade load due to the
2017). Although the turbine diameter for 100 kW system is about 11m, non-uniform wind (Hand et al., 2001), and NREL also provide a numer-
it is expected that the diameter will be 40m for the 1 MW full scale ical model for estimation of blade loads (Jonkman and Buhl, 2005). Lit-
generator at the rated flow speed of 1.5 m/s (Takagi et al., 2012). eratures on various numerical methods including the CFD are also
Estimation of the blade loads are crucial for the design of such big available (for example (Hansen et al., 2006)) References in the literature
turbines. Especially the periodic fluctuation of the blade load is im- dealing with non-uniform effects on tidal turbines also can be found. For
portant for fatigue design. Blade load fluctuations are supposed to be example, Milne et al. (2013) measured blade loads in planar oscillatory
induced by the turbulence, wave orbital motion and vertical profile of flow and in spatially uniform unsteady flow (Milne et al., 2015). Faudota
the inflow. In the case of tidal currents, a simple two-dimensional et al. (Faudota and Dahlhauga, 2012) conducted a wave loads measure-
boundary layer shape based on the assumption of shallow water flow is ments. Blackmore et al. (2016) and Mycek et al. (2014a) (Mycek et al.,
often used to represent the vertical profile of flows (DNV, 2015). On the 2014b) studied effects of turbulence. Gaurier et al. (2013) measured a
contrary, variation of profiles in the inflow for the ocean current tur- turbine blade strain under current and wave loading. On the other hand,
bine is supposed to be complex, since the water depth at the sea area numerical approaches are also found. Ahmed et al. (2017) made a com-
where the current turbines will be deployed is deep. parison of CFD with field data on fluctuating loads due to velocity shear
Thus, a simple two-dimensional boundary layer shape based on the and turbulence. Parkinson et al. (Parkinson and Collier, 2016) conducted
assumption of shallow water flow may not be applicable. In addition, a field measurement and compared with results by a commercial soft-
the influence of internal waves could be expected, though this is still ware. O'Rourke et al. (O'Rourke et al., 2015) proposed a mathematical
under investigation. For example, Kodaira et al. (2014) investigated model for the Blade Element Method (BEM) to predict the hydrodynamic
nonlinear internal waves in the Kuroshio. Kiyomatsu et al. (Kiyomatsu performance of a tidal turbine in non-uniform inflow conditions.


Corresponding author.
E-mail address: takagi@k.u-tokyo.ac.jp (K. Takagi).

https://doi.org/10.1016/j.oceaneng.2018.12.026
Received 24 March 2018; Received in revised form 5 December 2018; Accepted 5 December 2018
Available online 13 December 2018
0029-8018/ © 2018 Elsevier Ltd. All rights reserved.
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

25.2
100.7
40

390 60

unit : mm Center of turbine rotor


Turbine axis
y z
Fig. 1. Plane view of the blade. x

Table 1 (a)
Summary of the numerical method.
Items Methods

Solution 1st order implicit


Turbulence model SST k-ω model
Velocity-pressure coupling SIMPLE method
Gradient calculation Least Squares Cell Based
Discretization Momentum 2nd order upwind
k 1st order upwind
ω 1st order upwind
y

Inlet
(b)
Fig. 3. Detail of the mesh. (a) Meshing on the blade surface. (b) Meshing near
the blade tip.
Outlet

65 152

Rotational region
390 unit: mm
W.L

900
z
120 550
120
y
Stational region
0.0 4.0 8.0(m) x
2.0 6.0
390
Fig. 2. Overall computational domain.

570
In the present work, experiments in a circulating water channel,
where a shear flow was created, and in an ocean engineering basin were Fig. 4. Principal dimensions of the turbine rotor.
conducted, and numerical estimations of the blade loads were also
carried out by a commercial CFD code and compared with the experi-
mental results. where the definition of r is found in Fig. 1. It is noted that two blades
are equipped as a turbine rotor because the Kairyu (IHI, 2017) uses two
2. Numerical computation of the blade loads bladed turbines for better performance in the installation and the
maintenance. It is also noted that, although the experimental model has
2.1. Blades and hub a boss at the center of rotor, the boss is ignored for the numerical
calculations to reduce the computational effort.
Since the aim of this study is to reveal the effect of shear flow, a It is noted that we focused on the blade moment load in this study as
popular wing section is used, and the twisting angle is presented by a mentioned above. The moment about the turbine axis and an axis
simple equation. The blade used in the present work is designed by perpendicular to the r-axis and turbine axis are defined as Mx and My
referring to NMRI's blade (Chujo et al., 2017). NACA0012 is used for respectively. Non-dimensional forms of Mx and My are defined as
the wing section and the plan view of the blade is a trapezoid as shown
Mx My
in Fig. 1, where the size of the blade is the same as that of experimental CMx = 1
, CMy = 1
U 2 AR U 2AR (2)
model which will be mentioned in section 3. 2 2
The twisting angle of the blade is represented as
where U is the inflow velocity at the turbine center, A is the projection
= tan 1 450
tan 1 1 area of the turbine rotor, R is the radius of the turbine rotor and is the
11r 11 (1) water density.

447
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

Length of test section=4.9m

Velocity meter

0.55m
Velocity meter

0.55m

0.55m

Depth = 1.2m
Inlet 0.6m 0.6m
1.5m
Turbine
0.8m
Screen×3
Side view 0.57m
1.5m
Fig. 5. Arrangement of the experiment in a circulating water channel.

0.45 0.035 Circulating water channel


Velocity meter1
0.030 Ocean Engineering Basin
0.30 Velocity meter2
0.025
0.15
0.020

CMx
Z [m]

0.00 0.015
-0.15 0.010
0.005
-0.30
0.000
-0.45 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
0.30 0.35 0.40 0.45 0.50 0.55 0.60 TSR
Velocity [m/s] (a) Mx
Fig. 6. Time averaged vertical profile of the shear flow in a circulating water 0.70
channel. The shear is created by three screens. Circulating water channel
0.60 Ocean Engineering Basin
0.40 0.50
0.35 0.40
CMy

0.30 0.30
0.20
0.25
0.10
CMy

0.20
0.00
0.15 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
TSR
0.10
(b) My
0.05 Fig. 8. Comparisons of time averaged (a) Mx and (b) My measured in the cir-
culating water channel and in the ocean engineering basin.
0.00
0.0 1.0 2.0 3.0 4.0 5.0 uj
p u 2 ui
Time[s] ( ui ) + ( u i uj ) = + [µ ( i + ij )]
t xi xi xi xi xi 3 xi
Fig. 7. Time history of My in the uniform flow. Measurement has been done in
the ocean engineering basin. + ( ui uj )
xj (4)

2.2. Solver and mesh where, denotes the pressure, µ the viscosity, ij the Kronecker delta, ui
the turbulent component and the over bar denotes the time average.
ANSYS Fluent ver.18.0 is used for the numerical computation. Since The last term is obtained by using a turbulence model. In this paper, the
the flow is incompressible, the conservation of mass is satisfied as SST k-ω turbulence model (Menter, 1993) is used. Other important
information on the solver is shown in Table 1.
The Reynolds number is 2.2×105 at the tip of the blade when the tip
+ ( ui ) = 0
t xi (3) speed ratio (TSR) is 11.0 and the inflow velocity is 0.5 [ms−1]. Based
on this value, the numerical mesh is constructed. Fig. 2 shows the size
where, the Cartesian-coordinate system x i is used and ui denotes the of overall computational domain.Where D denotes the diameter of
fluid velocity. The Reynolds-averaged Navier-Stokes equations (RANS rotor. It is noted that the sliding mesh is mainly used in the numerical
equations) is solved by the Fluent. The RANS equation is represented as computation. On the other hand, the Multiple Reference Frame (MRF)

448
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

0.45 0.40
0.40
0.35
0.35
0.30 0.30
0.25
Cp

0.20 0.25
CFD -2 deg. (MRF)
0.15 CFD 0 deg. (MRF)

CMy
0.20 CFD(TSR7.2)
0.10 CFD -2 deg. (Sliding Mesh)
Experiment -2 deg. Experiment(TSR7.2)
0.05 0.15
Experiment 0 deg.
0.00
0.10
3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
TSR
0.05
(a) Cp
0.00
1.20 0.0 1.0 2.0 3.0 4.0 5.0
1.00 TIme[s]

0.80
(a) Time history
0.60
Ct

CFD -2 deg. (MRF) 0.003


0.40 CFD 0 deg. (MRF) CFD
CFD -2deg. (Sliding Mesh)
0.20 Experiment -2 deg. Experiment
CMy Periodogram [(N*m)2]/Hz

Experiment 0 deg. 0.002


0.00
3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
TSR 0.002
(b) Ct
Fig. 9. Measured and computed results of (a) Cp and (b) Ct in the uniform flow. 0.001
Blade pitch angles are −2 and 0°.

model is used only for the uniform flow case and compared with the 0.001
result of the sliding mesh computation. Fig. 3 shows the detail of
meshing near the blade. Height of the wall cell is set to be smaller than
y+ = 1. The time step is determined so that each time step corresponds 0.000
to 1 degree of the turbine rotation. Total number of the mesh is 0.0 0.5 1.0 1.5 2.0 2.5 3.0
4,801,511 for the following calculations. In order to verify the con- Frequency [Hz]
vergence of results, a finer mesh calculation has also performed with
12,133,638 meshes. The difference of the power coefficient is 0.88%
and 0.1% for the thrust coefficient. Thus, we assumed that the nu- (b) Periodogram
merical results are converged with 4,801,511 meshes.
It is noted that a linear variation in z-direction is used for a math- Fig. 10. An example of measured My in the circulating water channel for the
ematical model of the shear flow at the inlet boundary in the shear flow blade pitch angle of −2°. (a) Time history and (b) Periodogram of the time
history.
computation. The coefficient of the linear equation is determined based
on the measurement of shear flow in the circulating tank as
moments are measured by a moment load sensor which is located at the
u= 0.416 + 0.045z [ms−1](5)
root of one blade. The experimental turbine rotor has a boss of 0.12m
diameter and a sensor and a dummy sensor are fixed at both sides of the
boss, while the numerical model ignores the boss and the censors. The
3. Experiment thrust and moment of the turbine are also measured to check the power
coefficient Cp and the thrust coefficient Ct.
3.1. Experimental set up Xia and Takagi (Xia and Takagi, 2016) created a shear flow in the
circulating water channel using screens. The screen is a lattice of 15
Principal dimensions of the turbine is shown in Fig. 4. The shape horizontal bars with diameter of 0.0272m and the bars are extended to
and twisting angle are chosen so that the blade shape is represented by the channel wall. The same screen is used for creating a shear flow in
simple equations for the convenience of generating the numerical mesh the circulating water channel. Fig. 5 shows the experimental arrange-
as mentioned above. The diameter is chosen to guarantee the accuracy ment of screens, velocity meters, and the turbine. The test section of the
of the moment measurement. As a result, a free surface effect is ob- circulating water channel has dimensions of 4.9m (length) x 2.0m
served that will be explained later. The rotation of the turbine rotor is (width) x 1.2m (water depth). Three screens are used to create a shear
controlled to keep the constant rate from TSR = 5 to 14. The blade flow with 10% gradient. An electro-magnetic velocity meter is installed

449
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

0.45 0.045
0.040 CFD
0.40
Experiment
0.35 0.035

CMx average
0.30 0.030
0.25 0.025
Cp

0.20 0.020
0.015
0.15 CFD Uniform Flow
0.010
0.10 CFD Shear Flow
Exp Uniform Flow 0.005
0.05
Exp Shear Flow 0.000
0.00 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 TSR
TSR
(a) Cp (a) Mx

1.20 0.45
0.40
1.00 0.35

CMy average
0.30
0.80
0.25
0.60 0.20
Ct

CFD Uniform Flow 0.15


0.40 CFD Shear Flow 0.10 CFD
Exp Uniform Flow 0.05 Experiment
0.20
Exp Shear Flow 0.00
0.00 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 TSR
TSR (b) My
(b) Ct Fig. 12. Comparison between experimental results and computational results
on the time average of Mx and My in the shear flow for the blade pitch angle of
Fig. 11. Measured and computed results of (a) Cp and (b) Ct in the uniform
−2°.
flow and in the sea flow. Blade pitch angle is −2°.

at the center line of the channel to measure the reference inflow velo- other hand, the free surface effect is more serious for the magnitude of
city and another one is installed 0.35m off-center. Both meters are lo- fluctuations in the shear flow, because the amplitude of periodic fluc-
cated 0.6m in front of the turbine. tuation appearing in Fig. 7 has almost the same order as the fluctuation
A measurement of vertical profile of the flow was conducted before in the shear flow. Thus, the fluctuations of Mx and My in the shear flow
installing the turbine. Fig. 6 shows the measured vertical profile of are corrected by subtracting the fluctuation appeared in uniform flow.
shear flow where the velocity is averaged in time. Although the vertical Fig. 8 shows comparisons of time averaged (a) Mx and (b) My
profile is not perfectly straight, the screens induce a linear shear flow measured in the circulating water channel and in the ocean engineering
with the gradient of 10.8%. We also measured the turbulence intensity, basin. Apparent blockage effect is observed. The measured value in the
and it is found that the turbulence intensity for the shear flow is 4–5% circulating tank is corrected using two values as the blockage effect
while it is less than 2% for the uniform flow. correction. In addition to this, the free surface correction is made for the
fluctuated part of the measured value in the shear flow as mentioned
previously. The same blockage correction is made for Cp and Ct.
3.2. Blockage effects

Since the width of the test section is only 2.0m, a blockage effect is 4. Results and discussion
expected. Thus, the same experiment is conducted in the ocean en-
gineering basin of the University of Tokyo. The basin has dimensions of 4.1. Comparison of the CFD with the experiment
50m (length) x 10m (width) x 5m (water depth). We used the same
experimental set up, while the turbine was towed by a towing carriage. In order to verify the numerical procedure, comparisons of Cp and
It is mentioned that, even in the ocean engineering tank, the free Ct are performed. Fig. 9 shows the numerical results and the measured
surface effect cannot be removed, though it is supposed to be smaller results of Cp and Ct. The numerical computations for the pitch angle of
than the wall effect. Fig. 7 shows an example of time history of My −2° are performed both by the MRF model and the sliding mesh model.
which has a periodic fluctuation. It has the same period as the rotation The MRF results are almost identical to the sliding mesh results, and the
of the turbine, and it is supposed to be a free surface effect. The am- MRF computation and the sliding mesh computation both show a fairly
plitude of fluctuation is 2.5% of the average value. Therefore, the good agreement with experimental results. Thus, we concluded that the
average value is supposed to have an inaccuracy of the same order due numerical procedure is appropriate, and the numerical results have
to the miller effect of the free surface. However, this inaccuracy is ig- enough accuracy for the further discussion, though the experimental
nored in our experimental results since the inaccuracy is small. On the value could have small errors due to the free surface effect as

450
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

0.010 0.0
CFD
-10.0
0.008 Experiment
-20.0
CMx amplitude

CMx phase
0.006 -30.0

0.004 -40.0
-50.0
0.002 CFD
-60.0 Experiment
0.000 -70.0
4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
TSR TSR

(a) Mx (a) Mx
0.020 0.0
CFD
Experiment -10.0
0.015
CMy amplitude

-20.0

CMy phase
0.010 -30.0
-40.0
0.005 -50.0
-60.0 CFD
0.000 Experiment
-70.0
4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
TSR 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
TSR
(b) My
(b) My
Fig. 13. Comparison between experimental results and computational results
on the amplitude of the first harmonic component in the time history of Mx and Fig. 14. Comparison between experimental results and computational results
My in the shear flow for the blade pitch angle of −2°. on the phase of the first harmonic component in the time history of Mx and My
in the shear flow for the blade pitch angle of −2°.

mentioned above. We also confirmed that the MRF model gives good
approximation for Cp and Ct and saves the computational time, though the shear-flow case is not apparent, though the forces in the shear flow
the model is mathematically not correct for the transient problem. It is is theoretically expected to be larger than it in the uniform flow. On the
mentioned that from here on out we discuss on the case of blade pitch contrary, the CFD shows that Ct in the shear flow is slightly larger than
angle of −2°, since Ct is larger than the case of 0°. it in the uniform flow, though this tendency is not clear in the case of
Fig. 10-(a) shows an example of time history of My in the shear large TSR. This tendency is not simple for Cp, since the shear flow case
flow. Since the turbulent intensity of the inflow is about 4%, high fre- is larger only in the middle TSR. Thus, it is concluded that the influence
quency fluctuations are observed in the experimental time history. of the shear flow is small if the shear is not strong, and the tendency
However, an apparent periodic variation is recognized and its fre- depends on TSR. A large shear case will be shown and discussed in the
quency is equal to the rate of turbine rotation. On the contrary, the CFD next subsection as an example of the CFD application.
result shows an apparent periodicity, since the inflow of the CFD does Fig. 12 shows a comparison between experimental results and
not have a turbulent component. computational results on the time average of Mx and My, where the
In order to clarify this frequency, we obtained a periodogram of the sliding mesh is used for numerical computation to represent the shear
time history which is shown in Fig. 10-(b). A sharp peek is appearing flow. The experimental results are corrected for blockage effects. The
both for the experiment and the CFD at the frequency of 1.06 Hz which agreement between the experiment and the CFD is the same as that of
is equal to the rotation rate 1.06 rps of the turbine. Another small peak Cp and Ct. However, the agreement for smaller TSR is not good.
is also observed in the experiment at the frequency of 2.12 Hz which is Especially Mx at TSR = 5, the discrepancy between the CFD and the
supposed to be the second harmonic, while the CFD result has no experiment is big. The reason is supposed to be a separation of the flow
second peek. It is concluded from these results that the first harmonic on the blade. The influence of the separation is discussed in the next
component of the Fourier series expansion is useful for the discussion subsection.
on the fluctuations of the load acting on the blade, though the inflow Fig. 13 shows the amplitude of the first harmonic component of
contains a turbulent component and its influence is observed in the time fluctuation in the time history of (a) Mx and (b) My in the shear flow,
history. and the phase is shown in Fig. 14. Where, the amplitude and phase are
Fig. 11 shows measured and computed results of Cp and Ct in the obtained as the first harmonic component of the Fourier expansion. The
uniform flow and in the shear flow, where the sliding mesh is used for amplitude and the phase are corrected for the blockage effect and the
the computation. Since the generated shear in the circulating water free surface effect as mentioned above. It is apparent from these figures
channel is not strong, the difference between the uniform-flow case and that the CFD results are in fairly good agreement with experimental

451
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

0.450 0.40
Shear flow (40 )
0.35 Shear flow (10%)
0.425
Uniform flow
0.30
0.400 0.25

CMx
0.375 0.20
Cp

0.15
0.350
Shear flow (40%) 0.10
0.325 Shear flow (10%) 0.05
Uniform flow 0.00
0.300
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time [s] Time [s]
(a) Mx
(a) Cp
0.40
1.00
Shear flow (40%)
0.35
0.98
Shear flow (10%) 0.30
0.95 Uniform flow 0.25
0.93

CMy
0.20
0.90
Ct

0.15
0.88
0.10 Shear flow (40%)
0.85 Shear flow (10%)
0.05
0.83 Uniform flow
0.00
0.80
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 Time [s]
Time [s]
(b) My
(b) Ct
Fig. 16. Time histories of (a) Mx and (b) My for the blade pitch angle of −2°
Fig. 15. Instantaneous value of (a) Cp and (b) Ct for the blade pitch angle of and TSR = 7 (Period of turbine rotation = 0.81 s) in the shear flow.
−2° and TSR = 7 (Period of turbine rotation = 0.81 s) in the shear flow.

results, though the agreement is worse than that of averaged value. CFD. Fig. 15 shows instantaneous value of the Cp and Ct, where the
It is mentioned that the amplitude and the phase of Mx and My are period of turbine rotation is 0.81 s. Crests in the time history Cp cor-
not stationary in the experiment as seen in the time history shown in respond to instants when the blade is vertical. On the contrary, valleys
Fig. 10, since it is difficult to generate perfectly linear shear flow and correspond to the horizontal position. It is apparent that in the large
the inlet flow contains turbulent component. It is also noted that Mx is gradient case Cp has large fluctuations, while in the small gradient case
more sensitive than My, since Mx is much smaller than My. the fluctuations are small.
The phase lag appearing in the small TSR case is observed both in This tendency is explained by the attack angle of the local flow to
the experiment and CFD. This is one of important characteristics of the blade section. When the blade is vertical, the shear effect is added to
unsteady wing theory, though the phase lag is small. From these results, the average inflow of the turbine. The additional inflow to the upper
it is concluded that the CFD can predict the unsteady effects on the blade is positive while that to the lower blade is negative. Since the
turbine blades in the shear flow expect for the case of small TSR where magnitude of the blade force is proportional to the square of the inflow
the separation is expected. speed, the summation of upper and lower blade-forces is slightly larger
It is noted that, although the CFD is a powerful tool to estimate the in the vertical position compared to the blade-force when the blade
moment loads as mentioned above, it is too time consuming when we takes a horizontal position. The difference of the blade-force between
calculate the unsteady problem. Thus, in the simulation of the overall the vertical position and the horizontal position is supposed to be
turbine system which includes the floating body and the mooring proportional to the square of magnitude of the additional-inflow velo-
system, BEM is convenient for the estimation of blade force and mo- city. Thus, the fluctuation of the blade-force would be very small, if the
ment (Takagi, 2012). The present results are utilized for improving a attack angle of the flow does not change. However, the attack angle of
mathematical model which gives an unsteadiness correction to con- the flow changes linearly according to the variation of inflow speed. On
ventional BEM (O'Rourke et al., 2015). the other hand, if the attack angle increases Cp would increase and Ct
would decrease. As a result, instantaneous Cp has the large fluctuation,
4.2. Application of the CFD while the fluctuation of Ct is small. This is the case for the large gra-
dient of shear.
Although the shear gradient in the experiment is 10.8%, a more When the gradient of shear is small, the same thing happens.
severe gradient is reported in the measurement of the ocean current However, since the influence of attack angle variation due to the shear
(Kiyomatsu et al., 2014). Thus, a comparison between the small gra- is small compared to the attack angle without shear and the square of
dient case (10%) and a large gradient case (40%) was carried out by the the magnitude of additional inflow is small, the fluctuations on both Cp

452
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

Table 2
Summary of amplitude for different TSRs in the shear flow.
TSR = 6

Cp amplitude Ct amplitude CMx amplitude CMy amplitude

Shear flow (10%) 0.0003 0.0002 0.0022 0.0092


Shear flow (40%) 0.0077 0.0017 0.0126 0.0499

TSR = 7

Cp amplitude Ct amplitude CMx amplitude CMy amplitude

Shear flow (10%) 0.0002 0.0001 0.0021 0.0100


Shear flow (40%) 0.0106 0.0020 0.0142 0.0656

Shear gradient 10% Shear gradient 40% Uniform flow

0.60

0.45

0.30

0.15

0.00

Upper Lower Upper Lower Upper Lower

(a) TSR=5.0

Shear gradient 10% Shear gradient 40% Uniform flow

0.60

0.45

0.30

0.15

0.00

Upper Lower Upper Lower Upper Lower

(b) TSR=7.0
Fig. 17. Streamlines on the blade for the cases of (a) TSR = 5.0 and (b) TSR = 7.0 in different shear gradients of the flow. Blades are in the vertical position.

and Ct are very small. This is a qualitative explanation of this phe- histories apparently have a periodic fluctuation whose period is as the
nomenon. But, the quantitative estimation is difficult without the CFD, same as the period of turbine rotation. It is mentioned that the ampli-
because the unsteady effect is supposed to be considered. tude is not proportional to the shear gradient, and the ratio between
Fig. 16 shows time histories of Mx and My in the shear flow. Time amplitude and shear gradient depends on the TSR. The reason is

453
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

Skin friction coefficient


Shear gradient 10% Shear gradient 40% Uniform flow

Upper Lower Upper Lower Upper Lower

(a) TSR=5.0

Skin friction coefficient


Shear gradient 10% Shear gradient 40% Uniform flow

Upper Lower Upper Lower Upper Lower

(b) TSR=7.0
Fig. 18. Distribution of the shear stress on the blade for the cases of (a) TSR = 5.0 and (b) TSR = 7.0 in different shear gradients of the flow. Blades are in the vertical
position.

supposed to be the same as the cases of Cp and Ct. However, since the
influence of unsteadiness exists, again the quantitative estimation is
difficult without the CFD. Table 2 shows a summary of amplitude for
different TSRs. It seems that the larger gradient case experiences more
sever fluctuations of the load than a linearly expected value from the
10% case. This is important information for the fatigue design of the
blade.
One of advantages of the CFD is that we can observe the detail of
flow directly from visualized numerical results. Thus, two examples are
shown here. The fist example is relevant to the separation of the flow on
the blade which was mentioned in subsection 4.1. Figs. 12–14 show
that the agreement between the CFD and the experiment is not good
when TSR is small (TSR = 5.0), while the agreement is fairly good in
the case of higher TSR. Since the local attack angle of the blade is large
when TSR is small, it suggests that the separation on the blade happens.
Fig. 19. Contour plot of magnitude of vorticity distribution on the center ver- Fig. 17 shows stream lines on the blade in the cases of (a) TSR = 5.0
tical plane for the 40% gradient, blade pitch angle = −2° and TSR = 7.

454
K. Yahagi, K. Takagi Ocean Engineering 172 (2019) 446–455

and (b) TSR = 7.0. Since the Cp, which represents the efficiency of the moment load are estimated with sufficient accuracy, though the
turbine, is high in the case of TSR = 7, this case is considered to re- agreement between the CFD and the experiment is worse than that of
present the normal operation-condition, while the case of TSR = 5.0 time averaged value. When TSR is small, the agreement between the
represents the torque-rich condition, i.e. low TSR. In the uniform flow, CFD and the experiment on Mx and My is not good, the reason con-
Fig. 17 shows that there are streamlines toward spanwise direction near sidered to be a separation happening on the blade. Flow visualizations
the tip of the blade when TSR = 5.0. These stream lines suggest the of the CFD result confirm detail of the separation.
flow separation for the case of low TSR. On the contrary, when TSR is Using the same CFD method, moment loads on a blade in the shear
high, streamlines toward the spanwise direction are not observed. This flow of 40% gradient were computed. It is found that the larger gra-
is confirmed in Fig. 18 which shows distributions of the skin friction dient case produces more severe fluctuations of the load than a linearly
coefficient. The skin friction coefficient is defined as expected value, and the wake behind the turbine is not symmetric.
w
These phenomena might be useful for the blade design as well as the
1
U2 design of a turbine array.
2 (6)
where, w denotes the shear stress. Acknowledgements
It is apparent in Fig. 18 that the distribution of skin friction in
uniform flow is not smooth when TSR is low, while it is smooth when This work is funded by the New Energy and Industrial Technology
TSR is high. These tendencies are basically the same in the cases of Development Organization, Japan (NEDO). The authors are grateful for
small and large shear gradient cases. However, when the shear gradient their support. The authors are also appreciated Akishima Laboratory
is large, directions of streamline are different from those in the case of Inc. for their support to the experiment in the circulating water channel.
uniform flow. The reason is supposed to be the unsteadiness of the flow,
since the two-dimensional unsteady wing theory suggests that the se- References
paration does not happen even for the large attack angle if the flow is
unsteady. But, the detail of flow is difficult to expect without the CFD. Ahmed, U., Apsley, D.D., Afgan, I., Stallard, T., Stansby, P.K., 2017. Fluctuating loads on a
Although we do not give examples of three-dimensional visualization, it tidal turbine due to velocity shear and Fluctuating loads on a tidal turbine due to
velocity shear and. Renew. Energy 112, 235–246.
may be more informative. Blackmore, T., Myers, L.E., Bahaj, A.S., 2016. Effects of turbulence on tidal turbines:
Another example is the wake flow behind the turbine which is very implications to performance, blade loads, and condition monitoring. Int. J. Mar.
important for the design of the farm, since the wake affects the inflow of Energy 14, 1–26.
Chujo, T., Matsui, R., Haneda, K., Nimura, T., Ishida, S., Inoue, S., 2017. Tank experiment
the turbine behind the front turbines. In addition, in the case of floating of floating horizontal Axis current turbine with twin rotor for safety assessment. In:
type ocean current turbine, the wake affects not only the efficiency but The 26th Ocean Engineering Symposium, Tokyo.
also the motion of floating devices in the second row. DNV, G.L., 2015. STANDARD DNVGL-ST-0164 Tidal Turbines. DNV GL AS.
Faudota, C., Dahlhauga, O.G., 2012. Prediction of wave loads on tidal turbine blades.
Fig. 19 shows a contour plot of magnitude of vorticity distribution on Energy Procedia 20, 116–133.
the center vertical plane in the shear flow. It is noted that the inflow Gaurier, Z.B., Davies, P., Deuff, A., Germain, G., 2013. Flume tank Characterization of
passes through the center of the turbine, since the boss is ignored in the marine current turbine blade behaviour under current and wave loading. Renew.
Energy 59, 1–12.
CFD computation. It is apparent that the wake after the turbine is not
Hand, M.M., Simms, D.A., Fingersh, L.J., JagerD, W., Cotrell, J.R., 2001. Unsteady
symmetry. The reason is supposed to be that the downwash induced by Aerodynamics Experiment Phase V: Test Configuration and Available Data
spiral trailing vortices is not uniform in a cylindrical region behind the Campaigns. NREL/TP-500-29491.
Hansen, M.O.L., Sørensen, J.N., Voutsinas, S., Sørensen, N., Madsen, H.A., 2006. State of
turbine. Stronger trailing vortex at the upper half decelerate the fluid
the art in wind turbine aerodynamics and aeroelasticity. Prog. Aero. Sci. 42 (4),
quicker than that in the lower half. As a result, the vertical velocity is 285–330.
induced. This effect might be important for the design of a turbine array. Demonstration of the World's Largest Ocean Current Power Generation System, vol 57
Technical Report of IHI Corporation 4.
Jonkman, Jason M., Buhl Jr., Marshall L., 2005. FAST User's Guide. Technical Report
5. Conclusions NREL/EL-500-38230.
Kiyomatsu, k., Kodaira, T., Kadomoto, Y., Waseda, T., Takagi, K., 2014. ADCP measure-
We focus on the shear flow effect on an ocean current turbine. The ments of ocean currents near Miyake Islands. J. Jpn. Soc. Nav. Archit. Ocean Eng. 20,
147–156.
periodic fluctuation of moment loads at a blade root induced by the Kodaira, T., Waseda, T., Miyazawa, Y., 2014. Nonlinear internal waves generated and
shear flow is highlighted because the bending moment is one of im- trapped upstream of islands in the Kuroshio. Geophys. Res. Lett. 41, 5091–5098.
portant parameters for the structural design of the blade. Experiments Menter, F.R., 1993. Zonal two equation k-ω turbulence models for aerodynamic flows. In:
23rd Fluid Dynamics, Plasmadynamics, and Lasers Conference , Orlando.
in a circulating water channel and a towing tank were conducted, and Milne, I.A., Day, A.H., Sharma, R.N., Flay, R.G.J., 2013. Blade loads on tidal turbines in
numerical estimations of the blade loads were also carried out by a planar oscillatory flow. Ocean Eng. 60, 163–174.
commercial CFD code and compared with the experimental results. Milne, I.A., Day, A.H., Sharma, R.N., Flay, R.G.J., 2015. Blade loading on tidal turbines
for uniform unsteady flow. Renew. Energy 77, 338–350.
A linear shear flow was created in the circulating water channel Mycek, P., Gaurier, B., Germain, G., Pinon, G., Rivoalen, E., 2014a. Experimental study of
with three screens, and moment loads at the root of one blade was the turbulence intensity effects on marine current turbines behaviour. Part I: one
successfully measured in this shear flow. The blockage effect and in- single turbine. Renew. Energy 66, 729–746.
Mycek, P., Gaurier, B., Germain, G., Pinon, G., Rivoalen, E., 2014b. Experimental study of
fluence of the free surface was removed from the measured data by
the turbulence intensity effects on marine current turbines behaviour. Part II: two
utilizing experimental results obtained in the ocean engineering basin. interacting turbines. Renew. Energy 68, 876–892.
ANSYS fluent was used for the computation and the numerical results O'Rourke, F., Boyle, F., Reynolds, A., Kennedy, D.M., 2015. Hydrodynamic performance
were compared with the experiment. prediction of a tidal current turbine operating in non-uniform inflow conditions.
Energy 93, 2483–2496.
The meshing, size of computational domain, and the turbulence Parkinson, S.G., Collier, W.J., 2016. Model validation of hydrodynamic loads and per-
model are verified through the comparisons between the numerical formance of a full-scale tidal turbine using Tidal Bladed. Int. J. Mar. Energy 16,
result and the experiment for the power coefficient and the thrust 279–297.
Takagi, K., 2012. Motion analysis of floating type current turbines. In: 31th International
coefficient in uniform flow. The numerical results of moment load in Conference on Ocean. Offshore and Arctic Engineering OMAE2012, Rio de Janeiro.
the shear flow are validated through the comparison with the experi- Takagi, K., Waseda, T., Nagaya, S., Niizeki, Y., Oda, Y., 2012. Development of a floating
mental results of Mx and My. It is concluded that the time averaged current turbine. In: OCEANS 2012, Virginia Beach.
Xia, Y., Takagi, K., 2016. Effect of shear flow on a marine current turbine. In: Proc. of
values of Mx and My are estimated by the present CFD method very OCEANS, Shanghai, 2016.
well, and the amplitude and phase of the periodic fluctuation of the

455

You might also like