Vous êtes sur la page 1sur 17

125

2016,28(1):125-141

DOI: 10.1016/S1001-6058(16)60614-5

A new biomimicry marine current turbine: Study of hydrodynamic performance and wake using software OpenFOAM*
Yung-Jeh CHU
Department of Civil Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur 50603, Malaysia,
E-mail: chujeh2022@gmail.com
(Received March 13, 2014, Revised August 20, 2015)
Abstract: Inspired by Dryobalanops aromatica seed, a new biomimicry marine current turbine is proposed. Hydrodynamic performance and wake properties are two key factors determining whether a new marine current turbine design is practical or not. Thus, a
study of hydrodynamic performance and wake of the proposed biomimicry turbine is conducted. The computational fluid dynamics
(CFD) software, OpenFOAM is used to generate the required results for the mentioned study. The hydrodynamic performance and
wake properties of the proposed biomimicry turbine is compared to two conventional turbines of Bahaj et al. and Pinon et al. respectively. The simulation results showed that the proposed biomimicry marine current turbine gives optimum power output with its
power coefficient, CP 0.376 at the tip speed ratio (TSR) of 1.5. Under the same boundary conditions, the maximum torque produced by the proposed biomimicry turbine at zero rotational speed is 38.71 Nm which is 1110% greater than the torque generated by
the turbine of Bahaj et al.. The recovery distance for the wake of the biomimicry turbine is predicted to be 10.6% shorter than that of
IFREMER-LOMC turbine. The above-mentioned results confirm the potential application of the proposed biomimicry marine current turbine in the renewable energy industry.
Key words: marine current turbine, wake, biomimicry, computational fluid dynamics (CFD), OpenFOAM

Introduction
The hydrodynamic performance and wake properties are two key factors to determine whether a new
marine current turbine design is applicable in industries or not. The torque and thrust generated by a rotating marine current turbine will determine its hydrodynamic performance during operation. On the other
hand, the wake property such as the wake recovery
distance is important in predicting the minimum spacing required between turbines in array. Thus, it is a
must to conduct hydrodynamic performance and wake
study through recognized methods in order to prove
that the new marine current turbine design is worth investing. There are a few recognized methods which
have been used to measure or predict the hydrodynamic performance and wake properties of marine current turbines, among them are experiments and numeri0F

* Biography: Yung-Jeh CHU (1987-), Male,


Graduate Civil Engineer

cal simulations.
There were few examples of successful experiments conducted in measuring marine current turbine
torque, thrust and wake properties such as those studied in Refs.[1-10]. A typical experiment consists of a
model scaled turbine which immersed in a flume tank
or a flow channel with supplies of inflow water in
close or open circuits. Sensors are used to measure the
torque and thrust on the turbine. The velocity at downstream flow at various positions is usually measured
with calibrated flow velocity measuring device such
as acoustic Doppler or laser Doppler velocimeter. The
experimental results of the turbine wake can be reliable if the scaled model of the turbine is sufficiently
large. In spite of the reliable results, many researchers
chose to predict the hydrodynamic performance and
turbine wake properties by using numerical simulations due to various restrictions on experiments such as
expensive set-up and long preparation time.
The CFD gains in popularity among researchers
as a tool for turbine research as shown in Refs.[11-15].
Several CFD software such as ANSYS-FLUENT and
OpenFOAM were available free or with affordable li-

126

cense price. Due to the fast growth of technology


nowadays most researchers are able to own at least a
high performance workstation for research study. This
enables accurate prediction of turbine wake properties
to be obtained with shorter period of time compared to
experimental methods. However, most of the predicted simulation results still need to be validated by experimental results due to the errors induced by the numerical method itself such as truncation errors.
The objective of this study is to predict the hydrodynamic performance and wake properties of a
proposed biomimicry horizontal axis marine current
turbine by using the open sourceCFD software,
OpenFOAM. The predicted hydrodynamic performance of the biomimicry marine current turbine is compared to Ref.[4] while the predicted wake properties are
compared to Ref.[11]. The results from the comparisons will give preliminary idea on the possible applications of the proposed biomimicry marine current turbine in renewable energy industry.

1. Background of the proposed biomimicry marine


current turbine
The idea of mimicking the Dryobalanops aromatica seeds in turbine design was proposed by Chu during his undergraduate, when Dryobalanops aromatica
seeds were found at the Geology Faculty in University
of Malaya. The inspiration came when one of the
Dryobalanops aromatica seed was seen rotating stably
in the air during a visit to the Geology Faculty. The
unique shape of the wings of Dyrobalanops aromatica
seeds might suggest a solution in optimum turbine
blade design. Thus, samples of Dryobalanops aromatica seeds were gathered, observed and photographed
for study use. Figure 1 shows the snapshots of a
Dryobalanops aromatica seeds found at Geology Faculty, University of Malaya.
Although the Dryobalanops aromatica seed uses
wind as the media of propagation, it is supposed that
its physical characteristics and flight mechanism can
be applied in the design of marine current turbine.
Thus, the biomimicry marine current turbine is proposed to have turbine blades similar to the shape of the
wings of Dryobalanops aromatica seed, three-bladed
and a downstream type turbine where the rotor is located behind the nacelle. Figure 2 further illustrates the
components of the proposed biomimicry turbine.

Fig.2 Sketch of the proposed 3-bladed biomimicry downstream


horizontal axis turbine

Fig.1 Snapshots of a Dryobalanops aromatica seeds found at


Geology Faculty, University of Malaya (Note: Due to the
near distance of photographing, the objects nearer to the
camera appear to be larger. The 0.01 m grid size is only
for rough estimation)

2. Numerical simulations
Two sets of numerical simulations were done in
terms of hydrodynamic performance and wake predictions. For prediction of hydrodynamic performance,
the geometrical properties of the proposed biomimicry
turbine such as the number of blades, rotor size, hub
size and boundary conditions were totally made the
same as those of the turbine reported[4]. For wake prediction, all the geometrical properties of the proposed
biomimicry turbine such as the number of blades,
rotor size, hub size and boundary conditions were also
made the same as those of the IFREMER-LOMC turbine from Ref.[11]. The power coefficient, CP was set
approximately the same in wake simulation for the
cases of IFREMER-LOMC and biomimicry turbine in

127

order to check the differences of their wake properties


when the power output of both turbines is roughly the
same. Grid independence test was conducted to determine the suitable mesh sizes for all simulation cases
on account of the limiting available computational
power. The numerical simulation was continued with
the preferable set of mesh sizes and number of iterations. The discussion includes the predicted results
such as the torque, thrust, velocity line plots, pressure
line plots, vorticity contour, velocity contour and particle streamlines.

ckness and pitch distribution presented in Table 1.

3. Computational hardware and software


A workstation with processor Intel Core i74770K CPU @ 3.50 GHz8, were used to run the simulations with Ubuntu 12.04 LTS as the operating
system. GetData and JavaFoil were used to generate
the coordinates required in turbine geometry modelling. GetData Graph Digitizer is a program for digitizing graphs and plots. JavaFoil is a relatively simple
program, which uses several traditional methods for
airfoil analysis. The geometry of the turbines was modelled by using SALOME. SALOME is an open-source software that provides a generic platform for Preand Post-Processing for numerical simulation. The
meshing and solving process was done in OpenFOAM.
OpenFOAM is a free, open source CFD software package developed by OpenCFD Ltd at ESI Group and
distributed by the OpenFOAM Foundation. ParaView
was used to view the simulation results. ParaView is
an open-source, multi-platform data analysis and
visualization application.

Fig.3 Conventional turbine geometry modelled according to the


details available from Ref.[4]

Table 1 Turbine blade section properties from Ref.[4]

r /m

c/R

Pitch/ o

t c 1 / %

0.08

0.125

15.0

24.0

0.12

0.116

9.5

20.7

0.16

0.106

6.1

18.7

0.20

0.097

3.9

17.6

0.24

0.088

2.4

16.6

0.28

0.078

1.5

15.6

0.32

0.069

0.9

14.6

0.36

0.059

0.4

13.6

0.40
0.050
0
12.6
Note: r is the radius of local blade element, c is the blade
chord, R is the blade tip radius and t is the section
thickness.

3.1 Bahaj et al.[4] turbine geometry modelling


The turbine from Ref.[4] consists of three turbine
blades. Its rotor diameter is 0.8 m, hub diameter is
0.1 m and the turbine blades were developed from the
profile shape of a NACA 63-8xx and with chord, thi-

Table 2 Blade section characteristics of the IFREMERLOMC turbine obtained from Ref.[11]

r/R

c/R

Pitch/ o

t c 1 / %

0.13

0.06

29.57

80

0.15

0.06

29.57

100

0.16

0.06

29.57

100

0.20

0.15

25.63

36

0.24

0.25

22.15

21

0.29

0.24

19.30

21

0.33

0.23

16.97

22

0.37

0.21

15.05

22

0.42

0.20

13.46

22

0.46

0.19

12.12

22

0.50

0.18

10.98

23

0.55

0.17

10.01

23

0.59

0.17

9.18

22

0.63

0.16

8.45

22

0.68

0.15

7.82

22

0.72

0.15

7.26

21

0.76

0.14

6.77

21

0.81

0.14

6.34

20

0.85

0.13

5.95

19

0.89

0.13

5.61

19

0.94

0.12

5.29

18

0.98

0.12

5.01

18

1.00
0.07
4.87
25
Note: r is the radial distance measured from the rotational
axis, R is the radius of turbine disk, c is the chord
length, t is the maximum thickness as a fraction of the
chord.

The blade set angle is the angle at the tip of the


blade, (i.e. the blade angles in Table 1 refer to a set

128

angle of 0o )[4]. The blade is designed such that maximum power would be produced at a 5o set angle[4].
Thus 5o set angle of the turbine from Ref.[4] is used
in this study as a benchmark to determine the hydrodynamic performance of the proposed biomimicry
marine current turbine. The coordinates of the turbine
blade surfaces in Ref.[4] were generated by using
JavaFoil according to the turbine blade section properties in Table 1. The generated coordinates of the turbine blade surfaces in Ref.[4] were then exported to
SALOME for turbine geometry modelling. The modelled turbine as that of Ref.[4] is shown in Fig.3.

Fig.7 Definition of rake, pitch and skew angles of the proposed


biomimicry turbine blade (viewed from the top of turbine)
in this study

Fig.4 The modelled IFREMER-LOMC turbine geometry using


SALOME

Fig.8 Power coefficient graph of the proposed biomimicry turbine with different combinations of turbine blades pitch
and skew angles e.g., 15p30s means 15o in pitch and

30o in skew

Fig.5 Photographs showing outlines (light grey) of a wing of


Dryobalanops aromatica seed used in biomimicry turbine
blade modelling. The left side of the photo shows concave surface while the right side shows the side view of a
wing of Dryobalanops aromatica seed

Fig.6 The modelled biomimicry turbine geometries inspired by


Dryobalanops aromatica seed

Fig.9 Power coefficient graph (zoomed in for the sake of clarification) of the proposed biomimicry turbine with different combinations of turbine blades pitch and skew angles
e.g. 15p30s means 15o in pitch and 30o in skew

129

Table 3 Grid independence test result for Ref.[4] turbine case with 5o set angle, TSR = 2 and U = 1.73 m / s
Number of cells

262 754

655 362

1 304 043

Average node interval size


on blades surfaces/m

0.0015

0.0010

0.0008

Number of iterations

CP

CT

500

0.0611

0.2721

5 000

0.0611

0.2722

10 000

0.0611

0.2722

500

0.0591

0.2680

5 000

0.0591

0.2683

10 000

0.0590

0.2684

500

0.0573

0.2649

5 000

0.0574

0.2658

10 000

0.0575

0.2660

Table 4 Grid independence test result for biomimicry turbine case with rake angle 40o , pitch angle 15o , skew angle 30o ,
TSR = 1 and U = 1.73 m / s
Number of cells

265 542

680 424

1 373 305

Average node interval size


on blades surfaces /m

Number of iterations

CP

CT

500

0.2715

0.8160

5 000

0.2729

0.8364

10 000

0.2728

0.8351

500

0.2735

0.8175

5 000

0.2730

0.8321

10 000

0.2747

0.8396

500

0.2744

0.8112

5 000

0.2752

0.8384

10 000

0.2747

0.8370

0.0015

0.0010

0.0008

3.2 IFREMER-LOMC turbine geometry modelling


The IFREMER-LOMC turbine geometry from
Ref.[11] consists of three turbine blades, 0.700 m
rotor diameter, 0.046 m hub radius and 0.720 m hub
length. The turbine blades were developed from the
profile shape details as shown in Table 2.
The background of the IFREMER-LOMC turbine is as shown in the following quotes extracted from
Ref.[11]:
The second set of blades used for validation
purposes was first designed by tidal generation limited
(TGL) and experimentally tested in the IFREMER
flume tank of Boulogne-sur-Mer (France). However,
this set of blades, presented in Maganga et al., is patented so a similar and open set of blades was developed for experimental validation at IFREMER and numerical trial at LOMC-Le Havre University (France).
In the present paper, this last open blades configuration will be named as the IFREMERLOMC configuration. The turbine blades are designed from a

NACA63418 profile.
JavaFoil and SALOME were used to produce the
geometry of the IFREMER-LOMC turbine. The modelled geometry was then exported to OpenFOAM for
simulation. Figure 4 shows the 3-D view of
IFREMER-LOMC turbine geometry modelled in this
study.
3.3 Biomimicry turbine blade geometry modelling
The photo of a wing of Dryobalanops aromatica
seed were used to generate the required blade surface
coordinates for the mentioned geometry modelling.
The coordinates of the outlines of the wing of
Dryobalanops aromatica seed is generated by using
GetData. The generated coordinates were then exported to SALOME for geometry modelling. In order to
have a fair comparison with the simulation cases of
Ref.[4] and IFREMER-LOMC turbine, the rotor diameter, hub and the number of turbine blades of the
two modelled biomimicry marine current turbines

130

were made the same as those in Refs.[4] and [11]. The


key difference between the biomimicry turbine and
the turbine from Refs.[4] and [11] is the shape of the
turbine blades. Figure 5 shows the photos of a wing of
Dryobalanops aromatica seed while Fig.6 shows the
two modelled biomimicry marine current turbines for
hydrodynamic performance and wake simulations.
3.4 Optimum biomimicry turbine geometry test
The definition of rake, pitch and skew angles of
the biomimicry turbine blade are as shown in Fig.7.
The rake angle was fixed to 40o while the skew and
pitch of the biomimicry turbine blades were altered in
order to see which combination will give optimum
power output. Lower mesh numbers (each case around
250 000) were applied in this test.
Simulation results from eight different combinations of pitch and skew angles of the biomimicry turbine blade in the optimum turbine geometry test are as
shown in Fig.8 and Fig.9. From the results shown, the
combinations of 15o pitch and 30o skew gives the
highest maximum power coefficient, CP which is
around 3.8 among other combinations. Thus, it was
decided that this combinations of 15o pitch and 30o
skew is to be implemented into the biomimicry turbine geometry for the hydrodynamic performance comparisons.
3.5 Boundary and initial conditions
For the simulation of hydrodynamic performance,
the free stream velocity, U = 1.73 m / s with initial
turbulence intensity = 5% were applied throughout the
mesh domain of all the turbine of Ref.[4] and biomimicry turbine. The rotational speed of the rotor was
altered according to the specific tip speed ratio (TSR)
for each simulation case. Standard k - wall function
was applied at the turbine blades surfaces.
On the other hand, the free stream velocity,
U = 0.8 m / s with initial turbulence intensity = 5%
were applied throughout the mesh domain of both the
IFREMER-LOMC turbine and biomimicry turbine
wake simulation cases. Standard k - wall function
was also applied at the turbine blades surfaces of both
simulation cases. Several trials with different rotational speed and tip speed ratio of rotor were done for
both simulation cases to obtain the same predicted
power coefficient, CP while it was not being concerned for the case of thrust coefficient, CT in this study.
Through trial and errors, it was found that the optimum predicted power coefficient, CP of both the
IFREMER-LOMC turbine and biomimicry turbine
cases were very near, 0.37 and 0.38, respectively.
Thus, the wake properties of the IFREMER-LOMC
turbine and biomimicry turbine simulation cases with

optimum TSR 3.67 and TSR 1.5 respectively were


compared.
The predicted power coefficient, CP and thrust
Coefficient CT in this study were calculated with
Eqs.(1)-(3).
CP =

2Q
U 3 R 2

(1)

CT =

2T
U 2 R 2

(2)

R
U

(3)

where CP is the power coefficient, CT is the thrust


coefficient, is the tip speed ratio, Q is the torque,

is the angular speed, is the density of fluid, U


is the free stream velocity, R is the radius of rotor
and T is the thrust.

Fig.10 Mesh domain slice sections

3.6 Solver, libraries and turbulence model


All the simulation cases in this study were solved
by using the OpenFOAM solver named
MRFSimpleFOAM. MRFSimpleFOAM is a steadystate solver for incompressible, turbulent flow of fluids with multiple reference frame (MRF) regions. The

131

Table 5 Mesh properties of simulation case for Biomimicry and IFREMER-LOMC turbine

IFREMER-LOMC
turbine case

Biomimicry
turbine case

Number of mesh cells

Blade surface mesh size/m

Rotor mesh size/m

Stator mesh size/m

525 982

0.0031

0.0250

0.050

1 610 174

0.0020

0.0165

0.033

3 667 938

0.0015

0.0125

0.025

592 672

0.0030

0.0250

0.050

1 765 874

0.0021

0.0165

0.033

3 937 105

0.0016

0.0125

0.025

SST k - turbulence model was adopted for the simulations in this study. The torque and thrust of the
turbine was calculated by the tool from a post-processing library named forces. The forces post-processing library provides tools for post-processing force/
lift/drag data with function objects.
3.7 Grid independence test for hydrodynamic performance simulation case
Grid independence test was conducted for the
Ref.[4] and biomimicry turbine case to ensure the
mesh quality is enough to achieve accurate results. Different number of cells and number of iterations were
applied in the test and the results are as shown in
Table 3 and Table 4.
From the result of the grid independence test, the
CP and CT values converged at 10 000 iterations and
1 304 042 mesh cells for the turbine of Ref.[4]. For
the biomimicry turbine, the CP and CT values converged at 10 000 iterations and 1 373 305 mesh cells.
Thus, all the hydrodynamic performance simulations
were run in 10 000 iterations and the mesh with
1 304 043 cells is chosen for simulations of turbine
from Ref.[4] while the mesh with 1 373 305 cells is
chosen for the biomimicry turbine simulations. Figure
10 shows the mesh domain slice sections for the simulation cases of Ref.[4] and biomimicry turbine.
The quantitative uncertainties for the grids were
estimated using the factor of safety method and they
are referred to [16-19]. The uncertainties of CP and
CT were 2.5%S and 15.7%S (where %S is percentage of fine grid solution) of the 1 304 042 mesh
cells solution for the turbine of Ref.[4] while it is
5.1%S and 10.2%S of the 1 373 305 mesh cells solution respectively for the biomimicry turbine.

3.8 Grid independence test for wake simulation case


Grid independence test was also conducted for
both IFREMER-LOMC turbine from Ref.[11] and
biomimicry turbine to ensure good mesh quality. Simulation of cases with different mesh cell numbers

were run in the test and their mesh properties are as


shown in Table 5.
The difference in mesh size and number of cells
between the IFREMER-LOMC turbine and Biomimicry turbine were adjusted to be as small as possible to
avoid different influence due to different grid size.
The velocity distribution located at a distance of 1.2D
(0.84 m) from the origin was adopted for the test. The
results from the test show that the influence of the grid
sizes was of minimum when the number of cells reaches 3 667 938 for IFREMER-LOMC turbine while it
was 3 937 105 for biomimicry turbine. The optimum
number of iterations for the best results is about
10 000. The velocity distribution plots were as shown
in Fig.11 while the slice sections of the mesh domains
are presented in Fig.12 and Fig.13.
Similar to grid independence test for hydrodynamic performance, the uncertainties for the grids were
estimated using the factor of safety method. The uncertainty of the axial velocity at lateral distance = 0
is 12.4%S of the 366 7938 mesh cells solution for
IFREMER-LOMC turbine while it is 14.2%S of the
3 937 105 mesh cells solution for biomimicry turbine.
4. Results and discussion
For the case of hydrodynamic performance, discussion is focused on the predicted hydrodynamic torque and thrust from the simulation results for the turbine of Ref.[4] and biomimicry marine current turbine.
Comparisons on power coefficient, CP and thrust coefficient, CT for each simulated turbines were also presented and discussed in this section. For the case of
wake predictions, several aspects of the turbine wake
properties of the IFREMER-LOMC turbine and the
proposed biomimicry turbine were discussed in detail.
The discussed aspects were the velocity distribution,
vorticity, pressure distribution, wake recovery distance and the particle streamlines from the predicted results. Through the comparisons of the results from
these aspects, the feasibility of the proposed biomimicry turbine to be adopted in turbine farm design can

132

of the IFREMER-LOMC turbine and biomimicry turbine.

Fig.11 Velocity line plots at distance of 1.2D from origin with


different mesh numbers. Velocity line plots at distance
of 1.2D from origin with different number of iterations

be predicted. Table 6 shows the summary of the conditions and predicted results for the wake simulation

4.1 Torque and thrust prediction


The hydrodynamic performance of the proposed
biomimicry marine current turbine outweighs the turbine of Ref.[4] in terms of torque. Referred to Table 7,
the hydrodynamic torque produced by the biomimicry
turbine is greater than that of Ref.[4] from TSR = 0
to TSR = 2.5 . The predicted maximum torque produced by the proposed biomimicry marine current turbine at zero rotational speed is 38.71 Nm which is
1 110% greater than the torque generated by the turbine of Ref.[4]. The high torque generated by the biomimicry turbine can overcome resisting torque (e.g. caused by friction) from the nacelle effectively and thus
avoids rotor jamming problems especially during initial stage when the turbine starts to rotate. Table 7
shows the hydrodynamic torque comparison between
turbine of Ref.[4] and biomimicry turbine.
The high thrust produced by the biomimicry turbine as shown in Table 7 is detrimental to the turbine
blade root due to the induced bending moment. The
required rigidity and elastic properties of the turbine
blade will be determined by the thrust produced by the
marine current turbine during its operation. Higher generated thrust will lead to more amount of or expensive materials needed to strengthen the turbine blade
root. However, the blade root bending moment might
be able to be alleviated by introducing flexible turbine
blade. Further study will be conducted in the future to
investigate the efficiency of flexible turbine blade in
relieving blade root bending moment. Despite the
blade rigidity issue, the thrust produced by the biomimicry turbine at optimum TSR, TSR = 1.5 is only
13.24% greater than the thrust produced by the turbine
of Ref.[4] at its optimum Tip Speed Ratio, TSR = 6.0 .
This extra 13.24% of thrust can be overcome by strengthening the blade root with stronger material and
proper reinforcement. The suitable blade thickness
and materials to strengthen the biomimicry blade will
also be investigated in future study.
The power coefficients, CP predicted in this
study show that the biomimicry marine current turbine
operates at lower, TSR. This means that with the same
rotor size and a given free stream velocity, the biomimicry turbine will rotates slower than conventional
turbine e.g. when compared to the turbine of Ref.[4].
This slow operating rotational speed makes the biomimicry marine current turbine environmental friendly
as the collision risk is expected to be low compared to
other conventional marine current turbines. Figure 14
and Fig.15 show the predicted power coefficient and
thrust coefficient with the results adopted from
Ref.[4].

133

Fig.12 Slice sections of mesh domains showing the turbine, rotor and stator

Fig.13 Enlarged view of the slice sections of mesh domain, showing the origin for downstream distance measurements

Similar to the hydrodynamic performance comparisons with Ref.[4], the power coefficient, CP and
thrust coefficient, CT of the biomimicry marine cu-

rrent turbine were also compared to IFREMERLOMC turbine from Ref.[11]. Referred to the
OpenFOAM results, the optimum TSR for the biomimicry marine current turbine is 1.5 and the maximum
power coefficient, CP is approximately 0.38 while the
optimum Tip Speed Ratio, TSR for the IFREMERLOMC marine current turbine is 3.67 and the maximum power coefficient, CP is approximately 0.37.
These approximately the same maximum power coefficients (0.37 and 0.38) of the IFREMER-LOMC and
biomimicry turbine will give valid wake properties
comparisons as shown in the next sections. Figure 16
and Fig.17 show the predicted power coefficient and
thrust coefficient graphs with the results adopted from
Ref.[4].
The predicted power coefficient, CP of the biomimicry marine current turbine decrease abruptly
once the TSR is larger than 2 which is much lower
than the two conventional marine current turbines according to Fig.14 and Fig.16. The range of operative
TSR (assuming preferable CP > 0.3 ) for the biomimicry marine current turbine is limited from 1 to 2.5
while it is wider for the case of the other two conventional marine current turbines. This limited range of
operative TSR will cause the necessity to control the
rotational speed of the turbine based on incoming flow
velocity in order to achieve optimum power extraction.
The mentioned control also applies to the limited operative TSR range for the biomimicry marine current
turbine in terms of the thrust coefficient, CT according to Fig.15 and Fig.17. Improvements on the design

134

Table 6 Summary of conditions and predicted results of wake simulation for IFREMER-LOMC and biomimicry turbine
Incoming
flow
velocity,

Rotor
diameter/
m

Number
of
blades

Hub
radius/
m

Hub
length/
m

Optimum
TSR

Rotational
speed/
rads1

Predicted
torque/
Nm

Predicted
thrust/N

CP

CT

IFREMERLOMC

0.8

0.7

0.046

0.720

3.670

8.390

4.360

82.980

0.370

0.670

Biomi-micry

0.8

0.7

0.046

0.720

1.500

3.430

10.920

125.450

0.380

1.018

U /ms1

Table 7 Hydrodynamic torque comparison between the turbine of Ref.[4] and biomimicry turbine
Bahaj et al.[4] turbine case

Biomimicry turbine case

TSR

Torque/
Nm

Thrust/
N

TSR

Torque/
Nm

Thrust/
N

3.20

95.41

38.71

373.72

0.5

3.28

105.20

0.5

43.03

418.18

1.0

4.23

128.87

1.0

82.67

629.61

2.0

8.66

200.15

1.5

75.39

767.99

4.0

25.59

448.31

2.0

52.14

840.54

6.0

25.99

678.18

2.5

33.13

972.73

8.0

19.55

798.37

3.0

1.48

1 004.85

10.0

11.93

870.14

3.5

34.41

1 039.75

4.0

74.71

1 079.62

Fig.14 Power coefficient graph of the turbines resulted from cavitation tunnel, towing tank, SERG-Tidal, GH-Tidal
Bladed and OpenFOAM simulations. The cavitation tunnel, towing tank, SERG-Tidal and GH-Tidal Bladed
results were adopted from Ref.[4]

Fig.15 Thrust coefficient graph of the turbines resulted from


cavitation tunnel, towing tank, SERG-Tidal, GH-Tidal
Bladed and OpenFOAM simulations. The cavitation tunnel, towing tank, SERG-Tidal and GH-Tidal Bladed
results were adopted from Ref.[4]

of this biomimicry marine current turbine with respect


to the stated limitation will be discussed in future
study.
Validation of the OpenFOAM predicted CP and

(where %E is percentage of experimental data) while


it is 15.38%E and 17.59%E respectively for the
case of IFREMER-LOMC turbine. These large average comparison error may be due to the limitations of
the implemented turbulence model or solver in predicting accurate hydrodynamic forces especially in high
pressure gradient region e.g. blade surfaces and near
wake region. The cause of this large average comparison error will be studied in future by comparing the

CT were done by referring to the experimental results


from Ref.[4] and also Ref.[11]. The average comparison error of the OpenFOAM predicted CP and CT
with Ref.[4] turbine is 20.63%E and 10.65%E

135

results from implementing different turbulence models


and solvers to simulation of the turbine of Ref.[4] and
IFREMER-LOMC marine current turbine e.g. other
Low-Reynolds turbulence model with transient solver.

Fig.16 Power coefficient graph of the turbines resulted from


experiment, particle method and OpenFOAM simulations. The IFREMER-LOMC experiment and particle method results were adopted from Ref.[11]. represents
smoothing parameter in particle method prediction by
Ref.[11].

Fig.17 Thrust coefficient graph of the turbines resulted from


experiment, particle method and OpenFOAM simulations. The IFREMER-LOMC experiment and particle method results were adopted from Ref.[11]

4.2 Velocity, vorticity and pressure distribution


The axial velocity contour plots and velocity line
plots of the wake of IFREMER-LOMC turbine and
biomimicry turbine were discussed in this study. The
axial velocity contour plots show no significant difference of the turbine wake pattern at the far wake region while the degree of fluctuations of axial velocity
is very much different at the near wake region. The
near wake region is defined as the wake region within
distances of 4D from origin while the far wake region
is farther than 4D away in the downstream direction
according to Ref.[11]. Figure 18 shows the simulated
velocity contour plots along the downstream direction
for the IFREMER-LOMC turbine and biomimicry turbine.
To go in depth on the turbine wake properties,

the velocity line plots are discussed. This discussion


will focus on the difference of OpenFOAM results on
turbine wake between the IFREMER-LOMC turbine
and biomimicry turbine. The purpose of including the
experimental and particle method results from Ref.[11]
is to validate the results for IFREMER-LOMC turbine
predicted by OpenFOAM. Since the velocity line plots
of the IFREMER-LOMC turbine predicted by
OpenFOAM lies within the ranges of the experimental
and particle method results from Ref.[11] as shown in
Figure 19 and Figure 20, the validation is considered
done for this study. Figure 19 shows the near wake
while Figure 20 shows the far wake of the axial velocity distribution line plots for IFREMER-LOMC turbine and Biomimicry turbine.
For the case of near wake, the velocity distribution of the biomimicry turbine wake predicted by
OpenFOAM shows lower axial velocity values at the
axis of rotation, z / D = 0 compared to IFREMERLOMC turbine. This low axial velocity suggests that
the larger blade area of the biomimicry turbine captures more incoming flow near the hub region compared
to IFREMER-LOMC turbine. This is different at far
wake region where the axial velocity on the axis of
rotation of the biomimicry turbine wake is higher
compared to IFREMER-LOMC turbine. This might
means that the wake recovery of the biomimicry turbine wake is shorter than the IFREMER-LOMC turbine
wake. The axial velocity distribution for the biomimicry turbine and IFREMER-LOMC turbine were of no
much difference at the wake region outside the rotor
area, for z / D > 0.5 and z / D < 0.5 . This suggests
that the wake expansion for the biomimicry turbine
wake is the same as the IFREMER-LOMC turbine
wake throughout the downstream direction. This result
also implies that the spacing of a biomimicry turbine
farm in lateral can be the same as a turbine farm from
IFREMER-LOMC turbines.
There are significant differences observed in the
vorticity contour plots of the predicted wake of
IFREMER-LOMC turbine and biomimicry turbine.
More vortices formed in a turbine wake means more
translational flow being converted to local rotational
flow. If more translational flow is being converted to
local rotational flow, then there will be less energy left
for extraction in the downstream wake. For the case of
IFREMER-LOMC turbine wake, the flow with high
vorticity magnitude (Vorticity magnitude 1 ) is distributed further downstream compared to the biomimicry turbine wake. The high vorticity regions are distributed at the tip of the IFREMER-LOMC turbine and
also behind its hub. The biomimicry turbine wake vorticity is observed to be concentrated nearer to the turbine. These differences further support that the biomimicry turbine wake recovery distance is shorter compared to IFREMER-LOMC turbine wake. Figure 21

136

Fig.18 Velocity contour plots

Fig.19 Velocity line plots showing axial velocity distribution at different locations in near wake at downstream direction for
IFREMER-LOMC and Biomimicry turbine cases from this study and Ref.[11]. Note: The letter D denotes measurement of
distance in diameters of rotor, in this case 1.0D is 0.7 m. e.g. 1.2D means 1.2 times diameter of rotor measured from
origin. The lateral bars for the experimental data represent the standard deviation around the mean velocity value

137

Fig.20 Velocity line plots showing axial velocity distribution at different locations in far wake for IFREMER-LOMC turbine and
Biomimicry turbine from this study and Ref.[11]

Fig.21 Vorticity contour plots of simulation

138

Fig.22 Pressure distribution line plots at different locations in


near wake in downstream direction for IFREMERLOMC turbine and Biomimicry turbine simulated by
OpenFOAM

Fig.23 Pressure distribution line plots at different locations in


far wake in downstream direction for IFREMERLOMC turbine and Biomimicry turbine simulated by
OpenFOAM

further illustrates the vorticity of the IFREMERLOMC and biomimicry turbine.

The wakes of IFREMER-LOMC turbine and biomimicry turbine show significant reduction of pressu-

139

re at the axial region ( z = 0) . This reduction of pressure indicates that the flow is disturbed and slowed
down by the rotating turbine. It is observed that the
dip of the pressure distribution of the biomimicry turbine at the near wake is lower than the IFREMERLOMC turbine (at y = 1.2 D , 2.0D , 3.0D and 4.0D ).
The dip of the pressure distribution of the biomimicry
turbine wake becomes higher than the dip of the pressure distribution of the IFREMER-LOMC turbine
wake further downstream from the location at y =
6.0D . This observation also strengthens the idea that
the wake recovery distance for the biomimicry turbine
is shorter than the wake recovery distance of
IFREMER-LOMC turbine wake. Figure 22 shows the
pressure distribution plot of the IFREMER-LOMC
and biomimicry turbine at near wake while Fig.23
shows the pressure distribution plot of the IFREMERLOMC and biomimicry turbine at far wake.
Validation of the OpenFOAM predicted turbine
wake axial velocity was done by referring to the experimental results from Ref.[11]. The average comparison error of the OpenFOAM predicted axial velocities
with Ref.[11] turbine ranges from 8.83%E to
3.30%E with a decreasing magnitude of %E towards downstream direction.

( y ) = 100[1 u ( y )]

(5)

R* = R + r

(6)

where u ( y ) is the mean axial velocity, R is the radius of the disc representing the turbines area of influence, z is the lateral distance, u ( y, z ) is the axial
velocity, ( y ) is the mean axial velocity deficit, R
is the radius of the turbine disc and r is the assumed
radial increment due to turbines area of influence.
Pinon et al.[11] took r 0.05 m = 0.14 R . This
enlarged the integration interval to the two nearest experimental measurement nodes outside the rotor. The
whole velocity deficit was taken into account in this
way including the velocity deficit of the biomimicry
turbine wake. Figure 24 shows the mean axial velocity
deficit graphs between wakes IFREMER-LOMC turbine and biomimicry turbine.

Fig.25 Predicted wake recovery distance of IFREMER-LOMC


turbine and biomimicry turbine simulated by
OpenFOAM on the mean axial velocity deficit against
downstream distance graph. It is about 16.0D
(11.21 m) for biomimicry turbine while it is about
17.9D (12.54 m) for IFREMER-LOMC turbine

Fig.24 Mean axial velocity deficit against downstream distance


graph for experimental and numerical results on the
wake of IFREMER-LOMC turbine and biomimicry
tidal turbine

4.3 Wake recovery distances


The mean axial velocity deficits along the downstream direction of the turbine wake were calculated
by using the integrated axial velocity deficit equations
from Ref.[11]. Equations (4)-(6) as shown below were
the related equation used in obtaining the mean velocity deficits.
u ( y) =

1
R*2

R*

R*

z u ( y, z )d y

(4)

Fig.26 Particle streamlines at zero rotational speed (Note: Left


side image shows back view while right side image
shows side view of the mentioned particles streamlines)

140

Fig.27 Particle streamlines

The curves of the mean axial velocity deficit


graph for the IFREMER-LOMC turbine and biomimicry turbine simulated by OpenFOAM were extended
forward and their intercepts on the horizontal axis,
y / D show the predicted wake recovery distances. It
was assumed that the mean axial velocity deficit curves at further downstream to be represented by natural
log equations. Thus, a logarithmic trend line is applied
to the mean axial velocity deficit curves in MS Excel.
The intercepts show the wake recovery distances for
biomimicry turbine case is about 16D (11.21 m)
while it is about 17.9D (12.54 m) for IFREMERLOMC case. This made the wake recovery distance of
the proposed biomimicry turbine 10.6% shorter than
the case of IFREMER-LOMC turbine. This implies
that the spacing of a biomimicry turbine farm in downstream direction can be shorter than the spacing of
a turbine farm from IFREMER-LOMC turbines. This
reduction in turbine spacing by implementing the proposed biomimicry turbine will increase the performance of a turbine farm as more turbines can be included
in a specific area. Figure 25 shows the predicted wake
recovery distance of IFREMER-LOMC turbine and
biomimicry turbine simulated by OpenFOAM on an
enlarged scale of the mean axial velocity deficit graph.
Validation of the OpenFOAM predicted turbine
wake mean axial velocity deficit was done by referring to the experimental results from Ref.[11]. The
average comparison error of the OpenFOAM predicted mean axial velocity deficit with Ref.[11] turbine
is 19.46%E (where %E is percentage of experimental data).
4.4 Particle streamlines
The particle streamlines obtained from the simulation results further explain why the biomimicry turbine exhibits higher torque at zero or lower rotational
speed. Swirling is clearly seen in the particle streamlines behind the biomimicry turbine while it is less obvious for the turbine of Ref.[4]. The ability to deflect
more incoming water particles into the direction tangentially to the rotational axis enables the biomimicry
turbine to generate higher torque. The spoon like
shape of the biomimicry turbine blade causes the rea-

ction forces from the deflected water particles to be


more concentrated near the blade tip. Thus, this blade
property provides longer moment arm for higher torque generation. Figure 26 shows the particle streamlines around the turbine of Ref.[4] and biomimicry
turbine at zero rotational speed.
On the other hand, the predicted particle streamlines of the wakes of IFREMER-LOMC turbine and
biomimicry turbine show similar characteristics which
is helical. However, the pitch of the helical streamlines in the near wake region was different for both
cases. From the observations, the pitch of the predicted helical streamlines for the biomimicry turbine
wake is shorter compared to IFREMER-LOMC turbine wake. The water particles at the biomimicry turbine
wake rotate more than the water particles at the
IFREMER-LOMC turbine wake. This suggests that
the biomimicry turbine is able to convert the translational energy into rotational energy effectively despite
the low operating rotational speed applied. Figure 27
shows the particle streamlines of the IFREMERLOMC turbine and biomimicry turbine.
5. Conclusions
The preliminary results from the predictions by
OpenFOAM showed the proposed biomimicry marine
current turbine inspired by Dryobalanops aromatica
seed have high potential to be implemented in the renewable energy industry as an environmental friendly
and effective energy extractor. The following are the
research findings in this study:
(1) The hydrodynamic torque produced by the
biomimicry turbine is greater than the turbine of
Ref.[4] from at lower TSR which is from TSR = 0
to TSR = 2.5 . This high torque generated by the biomimicry turbine overcome the resisting torque from
the nacelle and thus avoids rotor jamming problems.
(2) The thrust produced by the biomimicry turbine at optimum TSR, TSR = 1.5 is 13.24% greater
than the thrust produced by the turbine of Ref.[4] at
optimum TSR, TSR = 6.0 . This extra 13.24% of thrust can be overcome by strengthening the blade root
with stronger material and proper reinforcement.

141

(3) The power coefficients predicted in this study


show that the biomimicry marine current turbine operates at lower TSR. This low operating rotational
speed makes the biomimicry marine current turbine
environmental friendly as the collision risk is expected
to be relatively low compared to other conventional
marine current turbines.
(4) The near wake region of the biomimicry marine current turbine showed significant reduction in
axial velocity. This result may indicate the effectiveness in capturing incoming flow due to larger blade
area of the proposed biomimicry turbine. The far wake
region of the biomimicry marine current turbine showed that the expansion of the wake is not much different from the IFREMER-LOMC turbine wake. This
result suggests that the spacing of the biomimicry turbine farm in lateral can be the same compared to a turbine farm from IFREMER-LOMC turbines.
(5) The predicted wake recovery distances is
about 16.0D (11.21 m) for biomimicry turbine case
while it is about 17.9D (12.54 m) for IFREMERLOMC turbine. This made the wake recovery distance
of the proposed biomimicry turbine 10.6% shorter
than that of IFREMER-LOMC turbine. This short
wake recovery distance suggests that the performance
of a turbine farm will be improved by implementing
the proposed biomimicry turbine as more turbines can
be included in a specific area.
(6) The predicted water flow at the wakes of biomimicry turbine and IFREMER-LOMC turbine is helical. The shorter pitch of the helix water particle pathway at biomimicry turbine wake suggest that the biomimicry turbine is able to convert the translational
energy into rotational energy effectively despite the
low operating rotational speed applied.
Experiments will be conducted in future study to
further validate the predicted results. Through these
studies, it is hope that the proposal of this biomimicry
marine current turbine will be able to bring advancement in marine current turbine industries.
Acknowledgements
The author wish to extend his gratitude to University of Malaya for the facilities and services provided in supporting this study and also the effort in preserving the Dryobalanops aromatica trees in the campus.
References
[1] MCADAM R. A., HOULSBY G. T. and OLDFIELD M. L.
G. Experimental measurements of the hydrodynamic performance and structural loading of the transverse horizontal axis water turbine: Part 2[J]. Renewable Energy,
2013, 59(6): 141-149.
[2] MCADAM R. A., HOULSBY G. T. and OLDFIELD M. L.
G. Experimental measurements of the hydrodynamic per-

formance and structural loading of the transverse horizontal axis water turbine: Part 3[J]. Renewable Energy,
2013, 59(6): 82-91.
[3] BAHAJ A. S., MOLLAND A. F. and CHAPLIN J. R. et al.
Power and thrust measurements of marine current turbines
under various hydrodynamic flow conditions in a cavitation tunnel and a towing tank, Renew[J]. Renewable Energy, 2007, 32(3): 407-426.
[4] BAHAJ A. S., BATTEN W. M. J. and MCCANN G. Experimental verifications of numerical predictions for the
hydrodynamic performance of horizontal axis marine current turbines[J]. Renewable Energy, 2007, 32(15): 24792490.
[5] MYERS L., BAHAJ A. S. Wake studies of a 1/30th scale
horizontal axis marine current turbine[J]. Ocean Engineering, 2007, 34(5-6): 758-762.
[6] MYCEK P., GAURIER B. and GERMAIN G. et al. Experimental study of the turbulence intensity effects on marine current turbines behaviour. Part I: One single turbine[J]. Renewable Energy, 2014, 68(7): 729-746.
[7] MYCEK P., GAURIER B. and GERMAIN G. et al. Experimental study of the turbulence intensity effects on marine current turbines behaviour. Part II: Two interacting
turbines[J]. Renewable Energy, 2014, 68(7): 876-892.
[8] MYCEK P., GAURIER B. and GERMAIN G. et al. Numerical and experimental study of the interaction between
two marine current turbines[J]. International Journal of
Marine Energy, 2013, 1: 70-83.
[9] TEDDS S. C., OWEN I. and POOLE R. J. Near-wake
characteristics of a model horizontal axis tidal stream turbine[J]. Renewable Energy, 2014, 63(1): 222-235.
[10] MAGANGA F., GERMAIN G. and KING J. et al. Experimental characterisation of flow effects on marine current
turbine behaviour and on its wake properties[J]. Iet Renewable Power Generation, 2010, 4(6): 498-509.
[11] PINON G., MYCEK P. and GERMAIN G. et al. Numerical simulation of the wake of marine current turbines with
a particle method[J]. Renewable Energy, 2012, 46(5):
111-126.
[12] ZHAO Guang, YANG Ran-sheng and LIU Yan et al. Hydrodynamic performance of a vertical-axis tidal-current
turbine with different preset angles of attack[J]. Journal
of Hydrodynamics, 2013, 25(2): 280-287.
[13] WANG Ji-feng, PIECHMA Janusz and MLLER Norbert
A novel design of composite water turbine using CFD[J].
Journal of Hydrodynamics, 2012, 24(1): 11-16.
[14] MAGANGA F., PINON G. and GERMAIN G. et al. Numerical simulation of the wake of marine current turbines
with a particle method (congress)[J]. World Renewable
Energy Congress X 2008. Glasgow, UK, 2008.
[15] KANG S., BORAZJANI I. and COLBY J. A. et al. Numerical simulation of 3D flow past a real-life marine hydrokinetic turbine[J]. Advances in Water Resources, 2012,
39: 33-43.
[16] OBERKAMPF W. L., ROY C. J. Verification and validation in scientific computing[M]. New York, USA:
Cambridge University Press, 2010.
[17] ROACHE P. J. Verification of codes and calculations[J].
AIAA Journal, 1998, 36(5): 696-702.
[18] XING T., STERN F. Closure to Discussion of Factors of
safety for Richardson extrapolation[J]. Journal of
Fluids Engineering, 2011, 133(11): 115501.
[19] XING T., STERN F. Factors of safety for Richardson extrapolation[J]. Journal of Fluids Engineering, 2010,
132(6): 061403.

Vous aimerez peut-être aussi