Accepted Manuscript Multi-body interaction effect on the energy harvesting performance of a flapping hydrofoil Mohsen Lahooti, Daegyoum Kim PII:
S0960-1481(18)30699-2
DOI:
10.1016/j.renene.2018.06.054
Reference:
RENE 10210
To appear in:
Renewable Energy
Received Date: 5 September 2017 Revised Date:
12 June 2018
Accepted Date: 13 June 2018
Please cite this article as: Lahooti M, Kim D, Multi-body interaction effect on the energy harvesting performance of a flapping hydrofoil, Renewable Energy (2018), doi: 10.1016/j.renene.2018.06.054. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
2
Multi-body interaction effect on the energy harvesting performance of a flapping hydrofoil
RI PT
1
Mohsen Lahooti, Daegyoum Kim∗
4
Department of Mechanical Engineering, KAIST, Daejeon 34141, Republic of Korea
SC
3
Abstract
6
The effect of an upstream bluff body on energy harvesting performance of
7
a heaving and pitching hydrofoil is investigated numerically using a two-
8
dimensional immersed boundary method at Re = 1000. The presence of the
9
upstream body changes flow structure around the hydrofoil and enhances
10
efficiency significantly by two mechanisms. Mutual interaction of the vortex
11
shed from the upstream body and the leading-edge vortex of the hydrofoil
12
precipitates the separation of the leading-edge vortex from the hydrofoil and
13
its streamwise transport. The incoming flow deflected by the upstream body
14
changes the effective angle of attack for the hydrofoil. These phenomena
15
significantly increase heaving force and pitching moment during stroke re-
16
versal, and major contribution to efficiency enhancement is from the change
17
in pitching moment. 30% increase in efficiency, relative to a hydrofoil without
18
an upstream body, can be achieved for same kinematics. However, the up-
19
stream body may be disadvantageous in some configurations. If the hydrofoil
20
is placed closely to the body in transverse direction, the leading-edge vor-
21
tex formation is suppressed after stroke reversal. When flapping frequency
22
does not match with vortex shedding frequency of the upstream body, non-
AC C
EP
TE
D
M AN U
5
Corresponding author: D. Kim Emailsubmitted addresses: (Mohsen Lahooti), Preprint to
[email protected] Renewable Energy
[email protected] (Daegyoum Kim) ∗
June 14, 2018
ACCEPTED MANUSCRIPT
periodic flow structure formed around the hydrofoil can cause efficiency drop
24
and irregular power generation.
25
Keywords: Energy harvesting, Ocean energy, Hydrodynamics, Flapping
26
hydrofoil, Bluff body, Leading-edge vortex
27
1. Introduction
SC
RI PT
23
Recently, a flapping hydrofoil has been widely studied as a novel method
29
to extract energy from fluid flows. The flapping hydrofoil can perform with-
30
out losing efficiency noticeably in unsteady flow environment. The hydrofoil
31
is also better fitted for shallow water installation because of their rectangular
32
sweeping area[1–3]. The idea of using the flapping hydrofoil was first intro-
33
duced by McKinney and DeLaurier[4]. Motivated by their work, Jones and
34
Platzer [5] showed that, for the hydrofoil of combined heaving and pitching
35
motions, one could change from a propulsion regime to an energy harvesting
36
regime if the pitching amplitude exceeds the heaving-based angle of attack.
TE
D
M AN U
28
The basic mechanism of efficient energy extraction from a flapping foil
38
(from now on the “hydrofoil” is referred to as “foil” for simplicity) can be
39
explained with vortices generated by the foil. The heaving motion of the foil
AC C
EP
37
40
with an appropriate angle of attack forms a strong leading-edge vortex near
41
the surface of the foil and produces vertical force due to the low pressure
42
region inside the vortex. However, as the leading-edge vortex separates and
43
washes away from the surface, the vertical force decreases. To maintain large
44
power generation, the foil should rotate and change its heaving direction to
45
form another leading-edge vortex near the opposite surface of the foil [1, 6–8].
46
Many studies on the energy harvesting foil have addressed the role of 2
ACCEPTED MANUSCRIPT
motion parameters and foil geometry on power generation. Among the earli-
48
est works is a numerical and experimental study conducted by Linsey [9] to
49
determine the feasibility of energy extraction. Another comprehensive work
50
was done by Dumas and Kinsey [10] and Kinsey and Dumas [7] in which they
51
presented an efficiency map as a function of pitching amplitude and reduced
52
frequency. Their map shows the efficiency as high as 34% can be reached for
53
a single foil and the best performance is around f ∗ = 0.15 which is consis-
54
tent with the analysis on optimal frequency for energy harvesting [8]. The
55
foil shape is known to have minor impact on efficiency for thin foils [6, 7].
56
It was also claimed that the increase in Reynolds number led to the better
57
performance [7]. However, due to complication of turbulent flows, solid con-
58
clusion on the Reynolds number effect cannot be established, and it needs
59
more investigation [1]. The studies on a foil with finite span reported that
60
three-dimensional effects caused the reduction in efficiency, compared to its
61
two-dimensional counterpart [6, 11, 12].
TE
D
M AN U
SC
RI PT
47
Several studies attempted non-sinusoidal pitching and heaving motions
63
and achieved higher efficiency than that of sinusoidal motions for some spe-
64
cific conditions [13, 14]. The flexible aero/hydrofoils motivated by the com-
AC C
EP
62
65
pliant propulsors of flying and swimming animals were also investigated, in
66
which the flexible aero/hydrofoil showed efficiency improvement although
67
the prescribed deformation of the aero/hydrofoil was considered instead of
68
its passive deformation [15]. It is worth mentioning that the effect of a flex-
69
ible foil on energy harvesting has not been fully understood, and in-depth
70
research is required in this area [1].
71
Another bio-inspired idea to improve the harvesting performance is from
3
ACCEPTED MANUSCRIPT
multi-body interaction. This idea was motivated by aquatic animals and
73
birds travelling as a group to benefit from the flow induced by their neigh-
74
bors [16–18]. One of the earliest studies on multiple foils for energy harvesting
75
is the work by Jones et al. [19], in which they experimentally investigated
76
the efficiency of twin foils in a tandem configuration with small clearance
77
and phase lag of 90◦ between them. Since the idea of using tandem foils
78
is to extract the remaining energy from the vortices generated by the front
79
foil, both phase lag and relative distance between front and rear foils are im-
80
portant parameters to strongly determine the overall performance [13, 20].
81
The tandem configuration seems disadvantageous because the rear foil posi-
82
tioned in the wake of the front foil is exposed to relatively lower flow velocity
83
and fluid kinetic energy. Nevertheless, we are able to gain some benefit by
84
properly positioning the rear foil for positive interaction with the vortices
85
shed from the front foil [2]. Indeed, using the tandem configuration, higher
86
efficiency can be achieved in some specific conditions (e.g., the phase lag of
87
180◦ ) [21–23].
EP
TE
D
M AN U
SC
RI PT
72
In addition to the tandem arrangement, the parallel foil arrangement
89
was investigated experimentally or numerically for both in-phase and out-of-
AC C
88
90
phase modes [24–26]. For the in-phase mode, the per foil efficiency decreases
91
as the distance between the two foils decreases. However, for the out-of-phase
92
mode, the reduction in the gap distance produces larger overall output.
93
Most, if not all, of previous works on multi-body effects focused on diverse
94
arrangements of multiple foils with the same geometry. Instead of multiple
95
foils, mutual interaction of an upstream bluff body and a downstream energy
96
harvester is considered to improve the efficiency in our work. The stationary
4
ACCEPTED MANUSCRIPT
upstream body deflects uniform flow and increases the flow velocity encoun-
98
tered by the downstream energy harvester, which is able to contribute to
99
larger power generation. This idea was applied to a vertical-axis rotary tur-
100
bine with an upstream bluff body, and the increase in efficiency was reported
101
[27, 28]. In the application of this idea to the heaving and pitching foil, un-
102
steady motion of the foil and its interaction with the vortex shed from an
103
upstream body make flow dynamics too complicated to predict the overall
104
effect based on available data reported in the literature. This difficulty moti-
105
vates us to investigate the effect of an upstream body on the power generation
106
of the downstream flapping foil and identify flow phenomena responsible for
107
noticeable changes in power generation. We believe that this study will lead
108
us to design an energy harvesting hydrofoil system with better performance.
109
The rest of paper is organized as follows. In Sec. 2, a foil model and
110
variables investigated in this study are described. The method of numerical
111
simulation and the validation of our numerical code are explained in Sec. 3.
112
Sec. 4 discusses our results on heaving force/pitching moment generation and
113
overall efficiency, based on the physical interpretation of vortex dynamics,
114
which is followed by concluding remarks in Sec. 5.
AC C
EP
TE
D
M AN U
SC
RI PT
97
115
2. Problem description
116
Our model consists of a pitching and heaving foil and a stationary up-
117
stream bluff body, a thin vertical plate, positioned upstream of the flapping
118
foil (Fig. 1(a)). As a foil model, we used the symmetric Joukowski foil
119
mathematically described as [8]: x=z+
λ2 + d − e, z−e 5
(1)
ACCEPTED MANUSCRIPT
U∞
(b)
c
αe
h0
θ
.
h−Vˆ
Ly
RI PT
(a)
Û
1/3c
SC
M
y
H
h
M AN U
Lx
c
x
Figure 1: (a) Schematic of the flapping foil with an upstream body and (b) the Joukowski foil. In (a), the thick vertical line is the upstream body.
where x = x + iy and z = ξ + iζ are coordinates in (x, y) and (ξ, ζ)-
121
planes respectively. The Joukowski foil is the mapping of a circle with radius
122
r = λ + e + s from the (ξ, ζ)-plane to the (x, y)-plane. e and s are the
123
parameters characterizing foil thickness and trailing-edge sharpness. d is
124
defined as d = (λ+2e+s)+λ2 /(λ+2e+s)−(1/2+b). The parameters used for
125
the Joukowski foil in this work are [b, λ, e, s] = [−0.167, 0.247, 0.027, 0.0074].
126
With these parameters, the foil has a unit chord length with the maximum
AC C
EP
TE
D
120
127
thickness of about 0.15 and the pitching axis located at one third of the chord
128
from the leading edge (Fig. 1(b)). The kinematics of the foil are the combination of sinusoidal heaving and
pitching motions: h(t) = h0 sin(2πf t)
(2a)
θ(t) = −θ0 sin(2πf t + φ),
(2b)
6
ACCEPTED MANUSCRIPT
RI PT
where h0 and θ0 are heaving and pitching amplitudes, f is flapping frequency, and φ = −π/2 is phase difference. Instantaneous heaving efficiency ηh (t), instantaneous pitching efficiency ηp (t), and time-averaged total efficiency η are defined as follows: h˙ 2Fy h˙ = C (t) h 3 A ρU∞ U∞ s ˙ ˙ 2M θ θc ηp (t) = = C (t) p 3 A ρU∞ U∞ s Z t+T 1 [ηh (t) + ηp (t)] dt, η= T t
M AN U
SC
ηh (t) =
(3a)
(3b) (3c)
where As is the frontal area swept by the foil during its periodic motion
130
2 2 and U∞ is free-stream velocity. Ch (= 2Fy /ρU∞ As ) and Cp (= 2M/ρU∞ cAs )
131
are heaving force and pitching moment coefficients respectively. Pitching
132
moment M and angle θ are positive in a clockwise direction (Fig. 1(b)). The
133
averaged efficiency η was computed over ten cycles for all cases after initial
134
transient effects disappeared. The efficiency indicates how much energy can
135
be transformed to mechanical energy of the foil from fluid kinetic energy
136
available for the swept area of the foil. In addition to these parameters, a
137
parameter frequently used here is relative efficiency η ∗ defined as the ratio of
138
the efficiency for a foil with an upstream body η to that of a foil without an
139
upstream body (a single foil ) η0 ; η ∗ = η/η0 .
AC C
EP
TE
D
129
140
In this study, non-dimensional variables we mainly considered are hor-
141
izontal distance between the edge of the upstream plate and the pitching
142
axis of the foil, L∗x ∈ [0.5, 0.75, 1.0], vertical distance between the top edge of
143
the upstream plate and the middle of the sweeping area (the location of the
144
pitching axis in the middle of a stroke), L∗y ∈ [0.5, 1.0, 1.5], and the height
145
the upstream plate, H ∗ ∈ [1.0, 1.5]; L∗x = Lx /c, L∗y = Ly /c, and H ∗ = H/c 7
ACCEPTED MANUSCRIPT
(Fig. 1(a)).
RI PT
146
As mentioned in Sec. 1, this study focuses on finding the effects of an
148
upstream body on energy harvesting performance and in particular physical
149
mechanisms which cause the change in the performance, instead of obtaining
150
optimal model conditions. Therefore, the heaving and pitching motions of the
151
foil were fixed for all of our numerical simulations. The specific kinematics of
152
the foil were chosen such that they produced high efficiency (not the highest,
153
though) for a single foil without an upstream body. According to previous
154
researches with the single foil, for high efficiency, the heaving and pitching
155
amplitudes should be usually in the range of 0.5c < h0 < c and 70◦ < θ0 <
156
80◦ with the reduced frequency f ∗ (= f c/U∞ ) around 0.15 [6–8]. In our
157
simulations, the heaving and pitching amplitudes and the reduced frequency
158
were fixed as h0 = 0.5c, θ0 = 76.33◦, and f ∗ = 0.14. The Reynolds number
159
based on the free-stream velocity U∞ and the chord c is 1000.
160
3. Numerical method
EP
TE
D
M AN U
SC
147
Governing equations of a fluid domain are as follows for incompressible
AC C
isothermal laminar flow:
∇·u = 0
(4a)
1 ∂u + u · ∇u = − ∇p + ν∇2 u, ∂t ρ
(4b)
161
where ρ, ν, and p are fluid density, kinematic viscosity, and pressure, re-
162
spectively, and u is a velocity vector. The simulation of the flow around a
163
flapping foil is numerically a challenging problem due to the large-amplitude
164
motion of the foil. Several numerical approaches have been reported in lit-
165
erature including the panel method [9], the approach using moving meshes 8
ACCEPTED MANUSCRIPT
in combination with a translational reference frame [7, 12, 13, 22], the finite
167
difference method using mapping between physical and computational spaces
168
[8], the compressible flow solver with low Mach number [14], to name a few.
169
In the present work, a different approach based on the Immersed Boundary
170
Method (IBM) [29, 30] was employed, which enables us to arbitrarily move
171
an object of any shape in stationary grids of a fluid domain and realize the
172
relative motion of two objects, a moving foil and a fixed upstream body.
M AN U
SC
RI PT
166
An in-house IBM code of the finite different method was implemented
174
for two-dimensional flow. A moving rigid foil was treated by the Hybrid
175
Cartesian/Immersed Boundary (HCIB) method proposed by Glimanov and
176
Sotiropoulos [31]. The HCIB method benefits from the sharp interface char-
177
acteristic of Cartesian methods, rather than diffusing the interface over sev-
178
eral cells. In addition, although the characteristics of the Cartesian methods
179
area adopted, the HCIB method dose not require complex cell reconstruction
180
and grid modification near immersed interfaces [31]. Therefore, it takes the
181
advantages of sharp interface capturing capability of Cartesian methods and
182
simpleness and robustness of immersed boundary methods in treating moving
183
bodies inside flow, which makes it an appropriate choice for our problem. Fol-
AC C
EP
TE
D
173
184
lowing Ge and Sotiropoulos [32], governing equations were discretized with
185
the implicit second-order method. The QUICK scheme was used for the dis-
186
cretization of convective terms, and viscous terms and pressure gradient was
187
discretized using the standard central difference. For pressure-velocity cou-
188
pling, the fractional step method of Van Kan [33] was employed. Summary
189
of discretization schemes is provided in Table 1. The implicit second-order
190
approach produces a nonlinear system of equations which can be solved with
9
ACCEPTED MANUSCRIPT
Discretization
RI PT
Table 1: Summary of discretization schemes
implicit second-order
convective term
QUICK method
pressure gradient term
second-order central difference
viscous term
second-order central difference
pressure-velocity coupling
M AN U
Coupling
SC
temporal term
fractional step method of Van Kan [33]
the Newton-Krylov subspace methods such as GMRES. The linear system
192
of equations is computed during nonlinear iterations, and linear system also
193
arises for pressure equation. Convergence criteria for both nonlinear and
194
linear systems are to reduce the relative L2 -norm of error by eight order of
195
magnitude; i.e. ||ǫrel || < 10−8 where ||ǫrel || = ||ǫ||/||ǫ0 || and ||ǫ0 || is the error
196
norm at start of solving the system of equations. For the complete solution
197
procedure of the HCIB method, refer to Gilmanov and Sotiropoulos [31] and
198
Ge and Sotiropoulos [32].
EP
TE
D
191
For our simulations, a rectangular fluid domain of [−15c, 25c]×[−15c, 15c]
200
size was constructed with the pitching axis in the middle of a stroke at origin.
201
The domain was divided into non-uniform rectangular grids with the mini-
202
mum spacing of dx = dy = 0.005c near the foil and upstream body. Since the
203
foil moves through the domain, in order to have enough grid resolution in the
204
vicinity of the foil and upstream body, uniform grids with dx = dy = 0.005c
205
were used in the region with approximate size of [−1c, 1.2c] × [−2.5c, 1.2c]
AC C
199
10
ACCEPTED MANUSCRIPT
containing both upstream body and entire swept area of the foil. Outside
207
the region of uniform grids, the grid size increases smoothly with the length
208
ratio of r = 1.05 until the whole computational domain is covered. Accord-
209
ingly, the total number of grids is Nx × Ny = 628 × 755. More details on
210
the procedure of choosing grid size and grid convergence study are provided
211
in Sec. 3.1. Time steps were determined by the Courant-Fredrisch-Levwy
212
(CFL) criterion, which restricts a Lagrangian point on the foil to pass less
213
than one grid of the fluid domain in each time step. The CFL number used
214
in this work is CF L = 0.5 which results in the time steps size of typically
215
2800 − 3000 steps per cycle (T /∆t = 2800). Since the Reynolds number of
216
our model is 1000, actual flow is three-dimensional. However, in this study,
217
based on two-dimensional simulation, we focus on obtaining insights into how
218
the efficiency of the foil can be improved by placing an upstream body rather
219
than simulating a realistic three-dimensional flow.
220
3.1. Code validation
TE
D
M AN U
SC
RI PT
206
To validate the accuracy of our in-house IBM code, two sets of tests
222
were conducted. The classic problem of a uniform flow around a stationary
223
circular cylinder with diameter d was first considered. A fluid domain is
224
30d in both x and y directions, and the cylinder is placed in the middle
225
of the domain. Non-uniform grids were used with the minimum spacing of
226
0.01d near the cylinder; the total number of grids is Nx × Ny = 386 × 300.
227
Drag and lift coefficients are compared with other references in Table 2 for
228
Re = 200 and 1000. In addition, Fig. 2 compares Strouhal number St as a
229
function of Re. For Re < 49, the flow around the cylinder is steady laminar
230
with symmetric wake vortices, and, from Re = 50 to Re = 140 − 194, the
AC C
EP
221
11
ACCEPTED MANUSCRIPT
around a circular cylinder.
RI PT
Table 2: Comparison of drag coefficient CD and lift coefficient CL for the uniform flow
CL
CD Re = 200
±0.67
1.34 ± 0.04
Linnick and Fasel [35]
1.34 ± 0.04
±0.69
Taria and Colonius [36]
1.36 ± 0.04
±0.69
1.49 ± 0.2
±1.36
Present work Henderson [37]
1.51
1.45 ± 0.2
±1.38
D
Mittal et al. [38]
M AN U
Re = 1000
SC
Present work
flow is in laminar vortex-shedding regime. Increasing Re from 180 to 260,
232
the flow is in three-dimensional wake-transition regime in which the near-
233
cylinder wake develops vortices of smaller scale and the wake becomes three-
234
dimensional. Increasing Re from 260 to 1000 leads to increase in disorder
235
of fine-scale three-dimensionalities [34]. Although a flow over the circular
EP
TE
231
cylinder becomes three-dimensional from about Re = 180, the purpose of
237
this two-dimensional simulation is solely to validate the accuracy of our code
238
against available two-dimensional simulations reported in literature. The
239
results of our simulations are found to be in excellent agreement with those
240
of other references.
AC C
236
241
Next, for the problem including a moving object, we compared our results
242
of a single flapping foil to those of Kinsey and Dumas in order to validate our
12
ACCEPTED MANUSCRIPT
RI PT
0.25
0.20
SC
St 0.15
0
M AN U
0.10
Present work Mittal et al. 2008 Williamson 1992 Zang et al. 1994
250
500
750
1000
Re
Figure 2: Strouhal number St as a function of Reynolds number Re for the circular cylinder. Filled circle: present work, filled triangle: Mittal et al.[38], hollow square:
D
Williamson[39], and hollow diamond: Zang et al.[40].
code [7]; h0 = c, θ0 = 76.33◦, and f ∗ = 0.14 at Reynolds number Re = 1100.
244
Since the Joukowski foil was used in our study, the foil geometry parameters
245
were chosen such that it has the most resemblance to the NACA0015 foil used
246
in Kinsey and Dumas. For this simulation, a fluid domain with 30c in both
247
x and y directions was discretized non-uniformly. To select appropriate grid
AC C
EP
TE
243
248
size, a series of tests was conducted with different grid numbers. For all these
249
meshes, grid size is distributed non-uniformly and grows smoothly to domain
250
boundaries. However, in the region covering the foil swept area, uniform grids
251
with the minimum grid size was used. Table 3 compares average efficiency in
252
one period for these grids. As evident in Table 3, the solution converges in
253
grid no. 3; the difference between no. 3 and no. 4 is negligible. Hence, the
254
grid no. 3, i.e. the mesh with the minimum grid size of dx = dy = 0.005c
13
ACCEPTED MANUSCRIPT
RI PT
Table 3: Comparison of time-averaged total efficiency η (Eq. 3c) for simulations with different grids.
No.
Nx × Ny
dxmin = dymin
η
1
235 × 255
0.02c
0.283
2
347 × 415
0.01c
3
547 × 755
0.005c
0.324
4
913 × 1262
0.0025c
0.325
M AN U
SC
0.313
Table 4: Comparison of time-averaged total efficiency η (Eq. 3c) between our simulation and previous studies.
η
0.324
D
Present work
0.337
Zhu [8]
0.315
EP
TE
Kinsey and Dumas [7]
255
near the foil and the total grid number of Nx × Ny = 547 × 755, was chosen
256
for simulation of a single foil. Based on this grid convergence study, the same grid resolution, dx = dy = 0.005c near the foil and upstream plate, was
258
chosen for our simulations with an upstream body.
AC C
257
259
Table 4 compares the averaged efficiency of our simulation with those of
260
Kinsey and Dumas [7] and Zhu [8]. In addition, Fig. 3 compares drag and
261
heaving force coefficients with those of Kinsey and Dumas[7]. Table 4 and
262
Fig. 3 validate that our code produces reliable data in both time-averaged
263
and time-resolved approaches.
14
SC
RI PT
ACCEPTED MANUSCRIPT
(b)
M AN U
(a) 4
2 1
3
Ch(t*) 0
Cx(t*) 2
−1
0 0
TE
D
1
0.25
0.50
0.75
–2 1
0
0.50
0.75
t*
EP
t*
0.25
2 Figure 3: Comparison of the time history of (a) drag coefficient Cx (= 2Fx /ρU∞ As ) and
AC C
(b) heaving force coefficient Ch (Eq. 3a) between our simulation (solid line) and Kinsey and Dumas[7] (circle dots). t∗ = t/T , and T is flapping period.
15
1
ACCEPTED MANUSCRIPT
4. Results and discussion
265
4.1. Physical principles of efficiency improvement with an upstream body
RI PT
264
First, we describe the change in flow structure of a flapping foil with the
267
presence of an upstream plate and address their effects on energy harvesting
268
performance. In this section, we will consider only a single case of L∗x = 0.75,
269
L∗y = 1.5, and H ∗ = 1.0 (Fig. 4). This particular case was chosen because
270
the foil was located at the region of higher power output than the single-
271
foil case and the characteristics of flow structure and force/moment trends
272
shown in this case can be generalized to other cases which exhibit comparable
273
improvement in efficiency over the sing-foil case.
M AN U
SC
266
Figs. 4(a) and (b) show the time history of heaving force coefficient Ch
275
and pitching moment coefficient Cp for one cycle. t∗ (= t/T ) = 0.0 (and
276
1.0) and 0.5 correspond to the uppermost and lowermost positions of the
277
foil respectively, where T is the flapping period; downstroke at 0 < t∗ < 0.5
278
and upstroke at 0.5 < t∗ < 1.0. During downstroke motion, as long as the
279
foil is far enough from the upstream plate (t∗ < 0.35), the influence of an
280
upstream plate is negligible, and the heaving force and pitching moment are
281
almost identical to those of the single foil. When the foil moves upward
282
away from the upstream body (0.60 < t∗ < 1.0), the effect of the upstream
283
body becomes insignificant again, and the heaving force and pitching moment
284
become similar to those of the single foil.
AC C
EP
TE
D
274
285
However, at the late phase of the downstroke (0.35 < t∗ < 0.50) and
286
at the early phase of the upstroke (0.50 < t∗ < 0.60), the heaving force
287
and pitching moment increase noticeably compared to the single foil (Figs.
288
4(a) and (b)). During the downstroke, the increase in the heaving force 16
ACCEPTED MANUSCRIPT
(a)
RI PT
(b) 1.5
4
1.0 2 0.5
Ch(t ) 0
Cp(t )
*
*
SC
0
−0.5
−2
−1.0
Foil with upstream body Single foil
0
0.25
−1.5
M AN U
−4
0.50
0.75
t*
(c)
1
0
0.25
0.50
0.75
1
0
0.25
0.50
0.75
1
t*
(d)
0.6
1.0
0.3
−0.3 −0.6
0.50
t*
0.75
−1.0 1
t*
(f) t*=1.0 (or 0.0) t*=0.75
AC C
1.2
0.25
0 −0.5
EP
0
(e)
ηp(t*)
D
0
TE
ηh(t*)
0.5
0.8 0.4
η(t ) *
Ly*=1.5
0
t*=0.5
−0.4
H*=1.0
−0.8 −1.2
0
0.25
0.50
t
*
0.75
1 Lx*=0.75
Figure 4: Time history of (a) Ch : heaving force coefficient, (b) Cp : pitching moment
17
coefficient, (c) ηh : heaving efficiency, (d) ηp : pitching efficiency, and (e) η(= ηh + ηp ): total efficiency for one cycle. (f): Schematic of the model. The foil with an upstream body of L∗x = 0.75, L∗y = 1.5, and H ∗ = 1.0 (solid line) and the single foil (dash-dot line). For the definition of output parameters, refer to Eqs. 3a−3c.
ACCEPTED MANUSCRIPT
is unfavorable because the heaving velocity h˙ is negative and the heaving
290
efficiency decreases accordingly (Eq. 3a); it is favorable at the early phase
291
of the upstroke, though. At 0.35 < t∗ < 0.50, heaving efficiency drops from
292
the single-foil case and reaches the minimum of −0.17 (Fig. 4(c)). Near the
293
end of the downstroke, the heaving velocity approaches to zero, and negative
294
effect of the heaving force on the heaving efficiency reduces.
SC
RI PT
289
Contrary to the heaving force that has both positive and negative effects
296
on the efficiency, the change in the pitching moment by the upstream plate
297
is almost favorable. During the stroke reversal (0.35 < t∗ < 0.60), the
298
pitching moment becomes higher than that of the single foil (Fig. 4(b)).
299
Since pitching angular velocity θ˙ is positive and reaches its maximum during
300
that time span, the augmentation of the pitching moment over the single foil
301
amplifies positive contribution to the efficiency (Fig. 4(d)). Near t∗ = 0.25
302
and 0.65, the pitching moment rather reduces from the single foil. However,
303
at these phases, the pitching velocity is near zero, which has little influence
304
on the pitching efficiency.
EP
TE
D
M AN U
295
The overall effects of the upstream body can be summarized as follows.
306
The noticeable rise in heaving force and pitching moment occur near the
AC C
305
307
stroke reversal from the downstroke to the upstroke. The change in heaving
308
force contributes positively to efficiency during the early upstroke. However,
309
its overall effect is to reduce the efficiency due to negative heaving velocity
310
during the downstroke. Meanwhile, in contrast to the heaving force, pitching
311
moment raises up the efficiency in both downstroke and upstroke. During
312
the stroke reversal, pitching angular velocity is near maximum while heaving
313
velocity is small. Therefore, major enhancement in total efficiency is due to
18
ACCEPTED MANUSCRIPT
the pitching moment (compare Figs. 4(c) and (d)). The combined effect of
315
both motions leads the gain in total efficiency at 0.4 < t∗ < 0.6 as evident
316
in Fig. 4(e). Consequently, with an upstream plate, the time-averaged total
317
efficiency η increases 31% from that of the single foil η0 ; i.e., η/η0 ≈ 1.3.
RI PT
314
To obtain more physical insight into the influence of the upstream body,
319
flow structures at several time steps are depicted in Fig. 5 for both the foils
320
without and with an upstream body. Compared to the single-foil case, the
321
local vorticity fields near the foil at t∗ = 0.15 (Figs. 5(a)-I and (b)-I) and
322
t∗ = 0.25 (Figs. 5(a)-II and (b)-II) show no significant difference, which was
323
expected since the foil is far away from the upstream body. Therefore, local
324
flow dynamics would be similar for both cases, so do the resulting force and
325
moment as discussed before. However, when the foil approaches more closely
326
to the body, the leading-edge vortex of the foil (red contour in Fig. 5) is
327
under the influence of the strong vortex created at the upper edge of the
328
upstream body (blue contour in Fig. 5). The flow induced by the vortex of
329
the upstream plate washes away the counter-rotating leading-edge vortex of
330
the foil in the x-direction, more quickly than the single foil (Figs. 5(a)-III∼V
331
and (b)-III∼V). At t∗ = 0.49 (Figs. 5 (a)-V and (b)-V), while the leading-
AC C
EP
TE
D
M AN U
SC
318
332
edge vortex is placed near the trailing edge for the single foil, it is already
333
behind the trailing edge for the foil with the body. Due to the separation
334
of the leading-edge vortex, the pressure near the lower surface of the foil
335
increases from the single-foil case, which results in the larger upward heaving
336
force. At the early phase of the upstroke (t∗ = 0.56), the leading-edge vortex
337
near the lower surface of the foil (red contour) becomes smaller and weaker
338
than that of the single foil due to the influence of the vortex of the upstream
19
ACCEPTED MANUSCRIPT
(II) t*=0.25
(V) t*=0.49
(VI) t*=0.56
(IV) t*=0.42
(III) t*=0.35
RI PT
(I) t*=0.15
SC
(a)
M AN U
(b)
(VIII) t*=0.75
AC C
(b)
EP
TE
D
(a)
(VII) t*=0.62
ωc/U∞
−20
−12
−4
4
12
20
Figure 5: Comparison of vorticity fields at several instances between (a) the foil with an upstream body (L∗x = 0.75, L∗y = 1.5, and H ∗ = 1.0) and (b) the single foil. Only the upper half of the upstream body is shown in each snapshot of (a).
20
ACCEPTED MANUSCRIPT
body (Figs. 5(a)-VI and (b)-VI), which keeps the upward heaving force
340
higher relative to the single foil. After t∗ = 0.56, the development of the
341
vortices around the foil show similar patterns between the two cases in spite
342
of some minor difference in vortex position and size (Figs. 5(a)-VII,VIII and
343
(b)-VII,VIII).
SC
RI PT
339
As described above, one of the mechanisms by which the upstream body
M AN U
affects the flow field near the foil is through the induced velocity caused by the strong vortex created at the edge of the upstream plate. To provide a clearer picture on how much this induced flow changes the location and shape of the leading-edge vortices, it is more convenient to compare directly the core and boundary of the vortices. Here the vortex identification method of Graftieaux et al. [41] is employed, in which two scalars Γ1 and Γ2 indicate the
TE
as follows:
D
core and boundary of the vortices, respectively. These scalars are computed
EP
Γ1 (xP ) =
Γ2 (xP ) =
1 X (xP M × uM ) · n N s |xP M | · |uM | 1 X [xP M × (uM − uP )] · n N
s
|xP M | · |(uM − uP )|
(5a) ,
(5b)
where xP is an arbitrary grid point in the computational domain for which
345
Γ1 and Γ2 are computed, and s is the area of 7 × 7 grid points, surrounding
346
xP , with a vector n normal to the area s. xP M is the vector from the point
347
xP to the point xM where xM presents one of the grid points lying on s.
348
uM is a velocity vector at xM , and uP is a velocity vector averaged on the
349
area s around xP . N is the number of points xM on s. Once Γ1 and Γ2 are
350
computed for the entire grids, their absolute values can be used to identify
351
vortex core and boundary respectively [41].
AC C
344
21
(b)
M AN U
SC
(a)
RI PT
ACCEPTED MANUSCRIPT
Figure 6: Comparison of vortex boundaries between the foil with an upstream body (blue contours) and the single foil (dashed red contours) at (a) t∗ = 0.42 and (b) t∗ = 0.56. In both (a) and (b), vortex centers (green contours) are shown as well.
Each subfigure of Fig. 6 superimposes vortex boundaries of the foils with
353
an upstream body (blue lines) and without an upstream body (red lines) in
354
downstroke (t∗ = 0.42) and upstroke (t∗ = 0.56). For the vortex boundary,
355
iso-contours of |Γ2 | = 0.65 (Eq. 5b) are plotted. For the identification of the
356
vortex center, the contours of |Γ1 | > 0.80 (green contours) are presented as
357
well (Eq. 5a). It should be noted that the vortex boundary may alter with
EP
TE
D
352
the change in the threshold value of |Γ2 |. Nevertheless, it is an acceptable
359
tool here to deliver how the upstream body would affect vortex dynamics.
AC C
358
360
At t∗ = 0.42, for the foil with an upstream body, the boundary of the
361
leading-edge vortex near the lower surface of the foil translates more in the
362
x-direction; its core is displaced away in the x-direction about ∆x ≈ 0.23c
363
relative to the single foil (Fig. 6(a)). In fact, at this moment, the leading-edge
364
vortex of the foil is located just above the detached vortex of the upstream
22
(a)
(I) t*=0.42
(II) t*=0.56
(c) 90 60
SC
30
RI PT
ACCEPTED MANUSCRIPT
αe
(b)
0
M AN U
−30 −60
−90 0 |u | *
Foil with upstream body Single foil
0.25
0 0.8 1.6 2.4
0.50
t
0.75
*
D
Figure 7: Effect of the upstream body on free-stream deflection and the effective angle of
TE
attack. Streamlines are colored by velocity magnitude at t∗ = 0.42 and 0.56 for (a) the foil with an upstream body and (b) the single foil. (c) The effective angle of attack over
EP
one cycle for the foil with a body (square) and the single foil (circle).
body, and their centers are aligned vertically, which enhances the induced
366
motion of the leading-edge vortex along the x-axis. Flow acceleration in-
AC C
365
367
duced by the two counter-rotating vortices can be more clearly observed by
368
comparing velocity magnitude (Fig. 7). The region of high velocity mag-
369
nitude |u∗ |(= |u|/U∞) (red color), which is approximately 2.5 times of the
370
free-stream velocity, exists between the boundaries of the two vortices for the
371
foil with a body (Fig. 7(a)).
372
Besides mutual interaction of the vortices, the upstream plate itself affects
373
the flow field by changing streamlines of the incoming flow as illustrated in 23
1
ACCEPTED MANUSCRIPT
Figs. 7(a) and (b). Due to the presence of the upstream plate, streamlines
375
are deflected upward in front of the sweeping area of the foil while, for the
376
single foil, streamlines remain horizontally until they reach the foil. This flow
377
deflection caused by the upstream body directly changes the effective angle
378
of attack αe that the foil actually encounters from the incoming flow. Here,
SC
we define the effective angle of attack as follows (Fig. 1(b)): ! ˙ − Vˆ h αe = θ − tan−1 , Uˆ
M AN U
379
RI PT
374
(6)
where Uˆ and Vˆ represent incoming flow velocities in x and y directions re-
381
spectively. For the single foil, we have Uˆ = U∞ and Vˆ = 0. However, for the
382
ˆ and foil with an upstream body, special attention is required to quantify U
383
Vˆ . In the study, at each time instance, Uˆ and Vˆ were obtained by averaging
384
u and v velocities over ten sample points distributed on the vertical line of
385
one chord length in front of the foil. The vertical line for the sample points
386
has the horizontal distance ∆x = 1.5c from the pitching axis.
TE
D
380
Fig. 7(c) depicts the variation of the effective angle of attack during one
388
cycle for the foils with an upstream body (square symbols) and without an
389
upstream body (circle symbols). The effective angle of attack for the foil
390
with an upstream body is different from that of the single foil almost for
391
the entire cycle. For early downstroke and late upstroke where the foil is
392
distanced from the body, αe is closer to that of the single foil. However,
393
the difference amplifies as the foil moves near the upstream body during
394
0.4 < t∗ < 0.6 and reaches its maximum value of ∆αe ≈ 15◦ shortly after the
395
foil changes its stroke.
AC C
EP
387
396
The change in the effective angle of attack, when the upstream body ex24
ACCEPTED MANUSCRIPT
ists, has a direct impact on flow structure near the foil. At the late phase of
398
downstroke, as shown in Fig. 6(a), small vortices (blue contours) near the
399
lower surface of the foil are slightly closer to the foil and stretched along the
400
foil. This is due to the smaller magnitude of the effective angle of attack
401
|αe | (Fig. 7(c)), which aligns the incoming flow toward the chord of the foil
402
(Fig. 7(a)-I). The influence of the upstream plate is also clearly identified on
403
pressure distribution near the foil. Because of the decrease in |αe |, pressure
404
magnitude on the upper surface (pressure surface) of the foil reduces, com-
405
pared to the single-foil case (Fig. 8). For the foil with the upstream body,
406
pressure also increases on the lower side (suction surface) of the foil mainly
407
because of the detached leading-edge vortex (Fig. 6(a)). Therefore, the in-
408
crease in pressure on the lower side of the foil together with the decrease in
409
pressure on the upper side (relative to single foil) can explain noticeable rise
410
in the upward heaving force at the late phase of the downstroke (Fig. 4(a)).
411
With the upstream body, the pressure increase on the lower surface is more
412
distinct near the leading edge than the trailing edge, which leads to the rise
413
in the clockwise pitching moment (Figs. 4(b) and 8).
SC
M AN U
D
TE
EP
On the other hand, at the early phase of upstroke, the magnitude of the
AC C
414
RI PT
397
415
effective angle of attack |αe | becomes larger for the foil with an upstream
416
body (Fig. 7(c)). The incoming flow hits the lower surface (pressure surface)
417
of the foil with larger angle, especially near the leading edge (Fig. 7(a)-II).
418
Moreover, for the foil with an upstream body, the vortex near the lower
419
surface share almost the same position but considerably smaller in size, com-
420
pared to the single-foil case (Fig. 6(b). These changes increase pressure on
421
the lower side of the foil compare to the single-foil case, which causes the rise
25
(b)
TE
D
M AN U
(a)
SC
RI PT
ACCEPTED MANUSCRIPT
p*
−6.2
−4.2
EP
−8.2
−2.2
−0.2
1.8
2 Figure 8: Comparison of non-dimensional pressure (p∗ = 2(p − p∞ )/ρU∞ ) fields at t∗ =
AC C
0.42. (a) The foil with an upstream body (L∗x = 0.75, L∗y = 1.5, and H ∗ = 1.0) and (b) the single foil.
26
ACCEPTED MANUSCRIPT
in upward heaving force and clockwise pitching moment at the early phase
423
of the upstroke although the rise is not as large as at the late phase of the
424
downstroke (Figs. 4(a) and (b)).
RI PT
422
Although we have discussed two mechanisms for efficiency improvement,
426
the change in the angle of attack and the interaction with the vortex are
427
not mutually exclusive. For example, the vortex shedding from the upstream
428
body will changes the angle of attack for the foil near the body. Thus, the
429
effect of each mechanism cannot be quantified separately, and the overall
430
effects of the upstream body have been addressed instead in this section.
431
4.2. Effect of relative position of a foil
M AN U
SC
425
In Sec. 4.1, general physical mechanisms on the effect of an upstream
433
body on energy harvesting were described. In this section, the effect of the
434
location of the foil relative to the upstream body will be discussed. Fig. 9
435
presents relative averaged efficiency (η ∗ = η/η0 ) for several values of L∗x and
436
L∗y . When the upstream body is vertically located far enough (L∗y = 1.0
437
and 1.5), both vertical and horizontal distances are not a critical factor to
438
change the performance. The efficiency is almost the same for all distances;
439
the gain is about 30%, relative to the single foil. For these cases, vorticity
440
distributions and time histories of the heaving force and the pitching mo-
441
ment are also similar with minor difference, and the principles of efficiency
442
improvement described in Sec. 4.1 can be applied to all of them. In this
443
study, we limited L∗y upto 1.5. By further increase in L∗y beyond 1.5, the
444
efficiency would eventually approach the efficiency of a single foil since the
445
effect of an upstream body on the foil vanishes away by increase in vertical
446
distance.
AC C
EP
TE
D
432
27
ACCEPTED MANUSCRIPT
1.0
η*
0.5
0.75
1.0
M AN U
0
SC
Ly*=0.5 Ly*=1.0 Ly*=1.5
0.5
RI PT
1.5
Lx*
Figure 9: Relative averaged efficiency η ∗ = η/η0 for the foil with an upstream body (H ∗ = 1.0).
However, when the foil is placed too closely to the upstream body in a
448
vertical direction (say, L∗y = 0.5), the fluid dynamics to cause the change in
449
energy harvesting performance are different from those mentioned in Sec 4.1,
450
and they generally result in the loss of efficiency (Fig. 9). We first consider
451
a situation with the upstream body located at (L∗x , L∗y ) = (0.75, 0.5), which
452
has the worst averaged efficiency in Fig. 9 and compare it with the foil at
AC C
EP
TE
D
447
453
(L∗x , L∗y ) = (0.75, 1.5). The discussion on this case can be generalized to the
454
other cases of L∗y = 0.5.
455
When the foil is located at L∗y = 0.5, even at the early part of downstroke
456
(t∗ < 0.25), the foil moves closely to the vortex of the upstream body. Due
457
to the larger incoming flow induced by the vortex of the body, the foil ex-
458
periences higher pressure at the upper surface of the foil, especially near the
459
leading edge (Fig. 10(c)-I). Thus, the foil at L∗y = 0.5 has larger (negative) 28
ACCEPTED MANUSCRIPT
heaving force and larger (negative) pitching moment than the foil at L∗y = 1.5
461
(Figs. 10(a) and (b)).
RI PT
460
After t∗ = 0.25, the foil starts to be submerged in the vortex of the
463
upstream body (Fig. 10(c)-II,III). The leading-edge vortex of the foil (red
464
contour) moves downstream more quickly, due to its interaction with the
465
vortex of the body, compared to its counterpart of L∗y = 1.5 (Fig. 5(a)-
466
II,III). As a result, the heaving force starts to increase sooner than that
467
of the foil at L∗y = 1.5 (Fig. 10(a)). The detached leading-edge vortex of
468
the foil moves upward closely to the foil and causes a low pressure region
469
near the trailing edge of the lower surface, which contributes to the increase
470
in the pitching moment near t∗ = 0.35 (Fig. 10(b)). The submerging foil
471
causes the increase in the heaving force and pitching moment earlier than
472
that of L∗y = 1.5 and experiences their peaks around t∗ = 0.35 instead of
473
around t∗ = 0.5. Compared to the foil of L∗y = 1.5, the earlier increase in the
474
heaving force has a negative effect on the efficiency because the magnitude
475
of negative heaving velocity is maximum near t∗ = 0.25. With a similar
476
argument, the earlier increase in the pitching moment is also undesirable
477
since the pitching angular velocity is maximum at t∗ = 0.5.
AC C
EP
TE
D
M AN U
SC
462
478
The most striking difference between L∗y = 0.5 and L∗y = 1.5 occurs
479
at the early phase of upstroke. Around t = 0.5, the foil is aligned in a
480
streamwise direction behind the upper edge of the upstream body. The
481
region just behind the upstream body has low pressure, especially below the
482
lower surface of the foil, and accordingly the heaving force becomes negative
483
(Fig. 10(a)). The leading-edge vortex generated during the downstroke (red
484
contour near the foil in Fig. 10(c)-III) sheds away from the trailing edge of
29
(a)
(b) 4
RI PT
ACCEPTED MANUSCRIPT
1.5 1.0
SC
2
Ch(t*) 0
Cp(t*) 0
−2
M AN U
0.5
−4
0
(I) t*=0.15
0.25
0.50
t*
(II) t*=0.25
0.75
−1.0 −1.5 0
1
(III) t*=0.35
0.25
(IV) t*=0.42
0.50
t*
0.75
(V) t*=0.49
AC C
EP
TE
D
(c)
−0.5
Ly*=0.5 Ly*=1.5
ωc/U∞
−20
−12
−4
4
12
20
Figure 10: (a,b) Comparison of the heaving force coefficient Ch and pitching moment coefficient Cp between the foils with an upstream body: (L∗x , L∗y ) = (0.75, 0.5) (dotted
line), (0.75, 1.5) (solid line). (c) Vorticity fields at several instances for (L∗x , L∗y ) = (0.75, 0.5).
30
1
(VI) t*=0.56
ACCEPTED MANUSCRIPT
the foil, and the leading-edge of the foil is under stronger influence of the
486
low pressure region just behind the upstream body. For these reasons, the
487
pitching moment also becomes negative near t = 0.5 (Fig. 10(b)). At the
488
start of the upstroke, since the incoming flow is blocked by the upstream
489
body, the formation of the leading-edge vortex on the upper surface of the
490
foil is delayed. Due to negative heaving force and pitching moment at t = 0.5
491
and delayed development of the leading-edge vortex, the force and moment
492
are noticeably reduced compared to the L∗y = 1.5 case.
M AN U
SC
RI PT
485
For L∗y = 1.0 and 1.5, main contribution to efficiency improvement is
494
from the increase in positive pitching moment near t∗ = 0.5 when positive
495
angular velocity is maximum (Sec. 4.1). However, when the upstream body is
496
positioned at L∗y = 0.5, the pitching moment is even negative near t∗ = 0.5.
497
This reduced pitching moment is the main reason that the averaged total
498
efficiency becomes much smaller than that of the foil at L∗y = 1.0 or L∗y = 1.5
499
and even smaller than that of the single foil (Fig. 9).
500
4.3. Regular and irregular formations of flow structure
EP
TE
D
493
In the previous section, it was shown that, if the height of the upstream
502
body is equal to the chord length (H ∗ = 1.0), averaged total efficiency η
503
is almost independent of the horizontal and vertical locations of the foil at
504
L∗y = 1.0 and 1.5. For H ∗ = 1.0, the dimensional vortex shedding frequency
505
of the upstream body is fv (= fv∗ U/H ≈ 0.14U/H) without a foil. This
506
vortex shedding frequency matches the flapping frequency of the foil f (=
507
f ∗ U/c = 0.14U/H) in our study. The match in the frequencies results in
508
the synchronization of the vortex shedding process of the upstream body
509
and the vortex formation of the foil and the production of periodic heaving
AC C
501
31
ACCEPTED MANUSCRIPT
(III) t*=0.35+2.00
(IV) t*=0.35+3.00
RI PT
(II) t*=0.35+1.00
M AN U
SC
(a)
(I) t*=0.35
TE
D
(b)
ωc/U∞
−12
−4
4
12
20
EP
−20
Figure 11: Vorticity fields of four successive cycles at the same phase of the foil. (a)
AC C
H ∗ = 1.0 and (b) H ∗ = 1.5. L∗x = 0.75 and L∗y = 1.5 for both (a) and (b).
510
force and pitching moment over cycles. This synchronization may also occur
511
even when the two frequencies have minor discrepancy, by tuning the vortex
512
shedding frequency fv to the flapping frequency f .
513
Then, a question arises whether this synchronization is possible even with
514
large discrepancy between the two frequencies. To answer this question, we
515
investigated a vortex formation pattern by increasing the height of the up-
516
stream body from H ∗ = 1.0 to H ∗ = 1.5 with the other parameters un32
ACCEPTED MANUSCRIPT
t*=0.56
t*=0.64
M AN U
SC
t*=0.50
RI PT
t*=0.43
t*=0.35
Figure 12: Vorticity contours for five time instances during the stroke reversal. L∗x = 0.75, L∗y = 1.5, and H ∗ = 1.5.
changed. Without a foil, the dimensional vortex shedding frequency fv in
518
H ∗ = 1.5 is reduced to 2/3 of fv in H ∗ = 1.0, from which we can expect
519
the mismatch of the vortex shedding frequency and the flapping frequency
520
in H ∗ = 1.5. Fig. 11 illustrates repeating and non-repeating patterns of
521
flow structure for both H ∗ = 1.0 and H ∗ = 1.5 respectively. For H ∗ = 1.0,
522
because of strong synchronization, vorticity distribution is almost the same
AC C
EP
TE
D
517
523
for all cycles (Fig. 11(a)). However, for H ∗ = 1.5, the synchronization is
524
lost, and vorticity distribution is not periodic at the same position of the foil
525
in each cycle (Fig. 11(b)). Especially the vortices detached from two edges
526
of the body exhibit stark difference in their size and position in each cycle.
527
For H ∗ = 1.5, the shear layers of the body become unstable and break into
528
small eddies whereas, for H ∗ = 1.0, periodic flow structure keeps the shear
529
layers from breaking into small eddies.
33
ACCEPTED MANUSCRIPT
As H ∗ increases from 1.0 to 1.5, the size of the vortex formed by the
531
body is also scaled with the height of the body. This implies that the body
532
vortex becomes more influential in the interaction with the foil. Due to
533
stronger interaction, the shear layers of the upstream body tend to become
534
more unstable in H ∗ = 1.5, and small eddies are formed more frequently, in
535
spite of the same kinematics and relative position of the foil as H ∗ = 1.0.
536
Fig. 12 provides the evolution of flow structure during stroke reversal. In
537
downstroke, the shear layer of the upper edge of the upstream body forms
538
smaller eddies (secondary vortices). Then, the small vortex near the edge is
539
dragged up by the upward motion of the foil after t∗ = 0.50, and is extended
540
to the lower surface of the foil. The dragged motion of the vortex toward the
541
foil in the upstroke is not observed at H ∗ = 1.0 (refer to Fig. 5). Here we
542
emphasize again that the flow pattern depicted in Fig. 12 does not repeat
543
regularly over cycles although it is observed frequently.
TE
D
M AN U
SC
RI PT
530
The non-periodic flow structure observed in H ∗ = 1.5 directly affects the
545
heaving force and pitching moment (Fig. 13). While Ch and Cp clearly show
546
regularly repeating patterns over cycles for H ∗ = 1.0 (Figs. 13(a) and (c)),
547
the repeatability is absent for H ∗ = 1.5, and the peaks of Ch and Cp fluctuate
AC C
EP
544
548
every cycle (Figs. 13(b) and (d)). In addition to the case of Figs. 11−13
549
(L∗x = 0.75, L∗y = 1.5, and H ∗ = 1.5), the aforementioned trend related with
550
asynchronization between the foil motion and the vortex originated from the
551
upstream plate is also observed at other relative positions of the foil with
552
H ∗ = 1.5. This finding gives some insight into designing the system of
553
an energy harvesting foil with an upstream body. To maintain stable and
554
predictable behaviors of the system, we should avoid the configuration which
34
(a)
SC
RI PT
ACCEPTED MANUSCRIPT
4 2
Ch(t*) 0
M AN U
−2 −4 4
(b)
2
Ch(t*) 0 −2 −4 2
(c)
1
Cp(t ) 0
D
*
−1
TE
−2 2
(d)
1
Cp(t ) 0 *
EP
−1
0
5
10
15
20
t*
AC C
−2
Figure 13: Time history of (a,b) heaving force coefficient Ch and (c,d) pitching moment coefficient Cp for H ∗ = 1.0 (dashed line) and H ∗ = 1.5 (solid line) over 20 cycles. L∗x = 0.75 and L∗y = 1.5.
35
ACCEPTED MANUSCRIPT
RI PT
1.5
1.0
0.5
0.50
0.75
1.00
M AN U
0
SC
η*
Lx*
Figure 14: Relative efficiency η ∗ for H ∗ = 1.0 (dashed lines) and H ∗ = 1.5 (solid lines). L∗y = 1.0 (square) and L∗y = 1.5 (circle).
causes irregular flow pattern and power generation.
D
555
For H ∗ = 1.5, the averaged total efficiency η becomes smaller than that of
557
H ∗ = 1.0 at the same relative positions, L∗x and L∗y , of the foil (Fig. 14). The
558
asynchronization of vortex formation not only causes the irregular generation
559
of power output, but also leads to the reduction in total efficiency. In spite of
560
irregular force and moment generation (Figs. 13(b) and (d)), in general, the
561
upstream body of H ∗ = 1.5 increases the heaving force particularly around
AC C
EP
TE
556
562
t∗ ≈ 0.25, at which the magnitude of the negative heaving velocity is around
563
its maximum. This results in efficiency loss compared to H ∗ = 1.0. Moreover,
564
pitching moment generally reduces from that of H ∗ = 1.0 during the stroke
565
reversal where pitching angular velocity is maximum, which is the major
566
reason of efficiency loss. Therefore, the upstream body with height greater
567
than chord length, i.e., H ∗ > 1.0, should be examined with extra care in
568
context of efficiency improvement as well as stable operation. 36
ACCEPTED MANUSCRIPT
5. Concluding remarks
RI PT
569
In this study, we have numerically investigated the effect of an upstream
571
bluff body on energy harvesting of an oscillating foil. Although we rely on
572
two-dimensional simulation at the Reynolds number much lower than that
573
of the actual application, we were able to find fluid-mechanical principles of
574
improving the efficiency significantly by using a simple additional structure.
575
The presence of the upstream body changes local flow structure around the
576
foil in two ways. First, it deflects an incoming flow into the sweeping area of
577
the foil, which results in the decrease in the effective angle of attack during
578
downstroke and its increase during upstroke. In addition, the vortex shed
579
from the upstream body induces faster separation of the leading-edge vor-
580
tex generated by the foil. These mechanisms change pressure distribution in
581
both surfaces of the foil and increase heaving force and pitching moment, es-
582
pecially during stroke reversal near the body, which contributes to improving
583
efficiency. By locating the foil at a proper location relative to the upstream
584
body, we are able to enhance overall power output by 30% from the single
585
foil.
M AN U
D
TE
EP
The installation of the upstream body does not always provide positive
AC C
586
SC
570
587
results. Instead, it may deteriorate the performance of the foil in several con-
588
ditions. When the foil is close to the upstream body in a transverse direction
589
(e.g., L∗y = 0.5), the development of the leading-edge vortex responsible for
590
power generation is delayed due to the blockage of the incoming flow during
591
the stroke reversal. If the vortex shedding from the upstream body is not
592
synchronized with the formation of the leading-edge vortex of the foil (e.g.,
593
H ∗ = 1.5), the flow structure becomes unstable, producing irregular power 37
ACCEPTED MANUSCRIPT
output over cycles.
RI PT
594
The strategy to increase power output of an energy harvesting system by
596
manipulating the distribution of an incoming flow was originally suggested for
597
conventional rotary turbines. Here, we conducted a proof-of-concept study
598
to apply this strategy to a foil periodically moving in a uniform stream. In
599
particular we emphasized the change in flow dynamics by the presence of
600
the upstream body and its correlation with power generation. In future,
601
extensive study should be pursued in order to find the optimal configuration
602
of the system and confirm its applicability to actual three-dimensional pilot
603
devices in high Reynolds number.
604
Acknowledgments
D
M AN U
SC
595
This study was supported by the Basic Science Research Program through
606
the National Research Foundation of Korea (NRF) funded by the Ministry
607
of Science, ICT & Future Planning (NRF-2015R1C1A1A02037111) and by a
608
grant [KCG-01-2016-04] through the Disaster and Safety Management Insti-
609
tute funded by Korea Coast Guard of Korean government.
AC C
EP
TE
605
38
ACCEPTED MANUSCRIPT
Frontal swept area of a foil
αe
Effective angle of attack
c
Foil chord
Γ1
Scalar quantity indicating vortex core
Ch (t)
2 A ) Heaving force coefficient (2Fy (t)/ρU∞ s
Γ2
Scalar quantity indicating vortex boundary
Cp (t)
2 Pitching moment coefficient (2M (t)/ρU∞ cAs )
η
Average total efficiency
f
Oscillation frequency
η0
Average total efficiency of a foil
f∗
Reduced frequency (f c/U∞ )
η∗
Relative efficiency (η/η0 )
Fy
Heaving force
η(t)
Total efficiency (ηh (t) + ηp (t))
h(t)
Heaving position
ηh (t)
3 ˙ Heaving efficiency (2Fy (t)h(t)/ρU ∞ As )
h0
Heaving amplitude
ηp (t)
3 ˙ Pitching efficiency (2M (t)θ(t)/ρU ∞ As )
H∗
Non-dimensional height of an upstream plate (H/c)
θ0
Pitching amplitude
L∗ x
Non-dimensional horizontal distance between the edge of
θ(t)
Pitching angle
an upstream plate and the pitching axis of a foil (Lx /c)
φ
Phase difference between heaving and pitching motions
M AN U
D
Non-dimensional vertical distance between the top edge of an upstream plate and the middle of the swept area (Ly /c) Pitching moment
T
Oscillation period
U∞
Non-dimensional time (t/T ) Free-stream velocity
EP
t
∗
TE
M
SC
As
L∗ y
Appendix A. Nomenclature
AC C
610
RI PT
Nomenclature
39
611
612
613
References
RI PT
ACCEPTED MANUSCRIPT
[1] Q. Xiao, Q. Zhu, A review on flow energy harvesters based on flapping foils, J. Fluid Struct. 46 (2014) 174–191.
[2] J. Young, J. C. Lai, M. F. Platzer, A review of progress and challenges
615
in flapping foil power generation, Prog. Aerosp. Sci. 67 (2014) 2–28.
616
[3] A. B. Rostami, M. Armandei, Renewable energy harvesting by vortex-
617
induced motions: review and benchmarking of technologies, Renew.
618
Sust. Energ. Rev. 70 (2017) 193–214.
621
622
M AN U
Energy 5 (1981) 109–115.
D
620
[4] W. McKinney, J. DeLaurier, Wingmill: an oscillating-wing windmill, J.
[5] K. D. Jones, M. F. Platzer, Numerical computation of flapping-wing propulsion and power extraction, AIAA J. 97 (1997) 826.
TE
619
SC
614
[6] D. Kim, B. Strom, S. Mandre, K. Breuer, Energy harvesting perfor-
624
mance and flow structure of an oscillating hydrofoil with finite span, J.
625
Fluid Struct. 70 (2017) 314–326.
AC C
EP
623
626
627
628
629
630
631
[7] T. Kinsey, G. Dumas, Parametric study of an oscillating airfoil in a power-extraction regime, AIAA J. 46 (2008) 1318–1330.
[8] Q. Zhu, Optimal frequency for flow energy harvesting of a flapping foil, J. Fluid Mech. 675 (2011) 495–517.
[9] K. Lindsey, A feasibility study of oscillating-wing power generators, Ph.D. thesis, Monterey, California. Naval Postgraduate School, 2002.
40
ACCEPTED MANUSCRIPT
633
[10] G. Dumas, T. Kinsey, Eulerian simulations of oscillating airfoils in power
RI PT
632
extraction regime, WIT Trans. Eng. Sci. 52 (2006).
[11] B. J. Simpson, S. Licht, F. S. Hover, M. S. Triantafyllou, Energy extrac-
635
tion through flapping foils, in: ASME 2008 27th International Confer-
636
ence on Offshore Mechanics and Arctic Engineering, American Society
637
of Mechanical Engineers, 2008, pp. 389–395.
639
M AN U
638
SC
634
[12] T. Kinsey, G. Dumas, Three-dimensional effects on an oscillating-foil hydrokinetic turbine, J. Fluid Eng. 134 (2012) 071105. [13] M. A. Ashraf, J. Young, J. S. Lai, M. F. Platzer, Numerical analysis of
641
an oscillating-wing wind and hydropower generator, AIAA J. 49 (2011)
642
1374–1386.
D
640
[14] Q. Xiao, W. Liao, S. Yang, Y. Peng, How motion trajectory affects
644
energy extraction performance of a biomimic energy generator with an
645
oscillating foil?, Renew Energ. 37 (2012) 61–75. [15] W. Liu, Q. Xiao, F. Cheng, A bio-inspired study on tidal energy extraction with flexible flapping wings, Bioinspir. Biomim. 8 (2013) 036011.
AC C
647
EP
646
TE
643
648
649
[16] V. Belyayev, G. Zuyev, Hydrodynamic hypothesis of school formation in fishes, Probl. Ichthyol. 9 (1969) 578–584.
650
[17] I. Akhtar, R. Mittal, G. V. Lauder, E. Drucker, Hydrodynamics of a
651
biologically inspired tandem flapping foil configuration, Theor. Comp.
652
Fluid Dyn. 21 (2007) 155–170.
41
ACCEPTED MANUSCRIPT
654
[18] F.-O. Lehmann, Wing–wake interaction reduces power consumption in
RI PT
653
insect tandem wings, Exp. Fluids 46 (2009) 765–775.
[19] K. D. Jones, K. Lindsey, M. F. Platzer, An investigation of the fluid-
656
structure interaction in an oscillating-wing micro-hydropower generator,
657
WIT Trans. Built. Env. 71 (2003).
[20] M. F. Platzer, M. A. Ashraf, J. Young, J. Lai,
Development of a
M AN U
658
SC
655
659
new oscillating-wing wind and hydropower generator, in: 47th AIAA
660
Aerospace Sciences Meeting including The New Horizons Forum and
661
Aerospace Exposition, 2009, p. 1211.
[21] T. Kinsey, G. Dumas, G. Lalande, J. Ruel, A. Mehut, P. Viarouge,
663
J. Lemay, Y. Jean, Prototype testing of a hydrokinetic turbine based
664
on oscillating hydrofoils, Renew Energ. 36 (2011) 1710–1718.
TE
D
662
[22] T. Kinsey, G. Dumas, Computational fluid dynamics analysis of a hy-
666
drokinetic turbine based on oscillating hydrofoils, J. Fluid Eng. 134
667
(2012) 021104.
[23] T. Kinsey, G. Dumas, Optimal tandem configuration for oscillating-foils
AC C
668
EP
665
669
670
671
hydrokinetic turbine, J. Fluid Eng. 134 (2012) 031103.
[24] H. Abiru, A. Yoshitake, Experimental study on a cascade flapping wing hydroelectric power generator, J. Energ. Power Eng. 6 (2012) 1429.
672
[25] K. Isogai, H. Abiru, Study of multi-wing configurations of elastically
673
supported flapping wing power generator, T. Jpn. Soc. Aeronaut. S. 55
674
(2012) 133–142. 42
ACCEPTED MANUSCRIPT
[26] K. Isogai, H. Abiru, Lifting-surface theory for multi-wing configurations
676
of elastically supported flapping wing power generator, T. Jpn. Soc.
677
Aeronaut. S. 55 (2012) 157–165.
RI PT
675
[27] D. Kim, M. Gharib, Efficiency improvement of straight-bladed vertical-
679
axis wind turbines with an upstream deflector, J. Wind Eng. Ind. Aero.
680
115 (2013) 48–52.
684
685
686
M AN U
683
interaction with an upstream deflector, Exp. Fluids 55 (2014) 1658. [29] F. Sotiropoulos, X. Yang, Immersed boundary methods for simulating fluid–structure interaction, Prog. in Aerospa. Sci. 65 (2014) 1–21. [30] R. Mittal, G. Iaccarino, Immersed boundary methods, Annu. Rev. Fluid
D
682
[28] D. Kim, M. Gharib, Unsteady loading of a vertical-axis turbine in the
Mech. 37 (2005) 239–261.
TE
681
SC
678
[31] A. Gilmanov, F. Sotiropoulos, A hybrid cartesian/immersed bound-
688
ary method for simulating flows with 3d geometrically complex moving
689
bodies, J. Comput. Phys. 207 (2005) 457–492.
EP
687
[32] L. Ge, F. Sotiropoulos, A numerical method for solving the 3d unsteady
691
incompressible navier–stokes equations in curvilinear domains with com-
692
plex immersed boundaries, J. Comput. Phys. 225 (2007) 1782–1809.
AC C
690
693
[33] J. Van Kan, A second-order accurate pressure-correction scheme for
694
viscous incompressible flow, SIAM J. Sci. Stat. Comp. 7 (1986) 870–
695
891.
43
ACCEPTED MANUSCRIPT
697
[34] C. H. Williamson, Vortex dynamics in the cylinder wake, Ann. Rev.
RI PT
696
Fluid Mech. 28 (1996) 477–539.
[35] M. N. Linnick, H. F. Fasel, A high-order immersed interface method
699
for simulating unsteady incompressible flows on irregular domains, J.
700
Comput. Phys. 204 (2005) 157–192.
702
703
704
[36] K. Taira, T. Colonius, The immersed boundary method: a projection
M AN U
701
SC
698
approach, J. Comput. Phys. 225 (2007) 2118–2137. [37] R. D. Henderson, Details of the drag curve near the onset of vortex shedding, Phys. Fluids 7 (1995) 2102–2104.
[38] R. Mittal, H. Dong, M. Bozkurttas, F. Najjar, A. Vargas, A. von
706
Loebbecke, A versatile sharp interface immersed boundary method for
707
incompressible flows with complex boundaries, J. Comput. Phys. 227
708
(2008) 4825–4852.
TE
D
705
[39] C. Williamson, The natural and forced formation of spot-like vortex
710
dislocation in the transition of a wake, J. Fluid Mech. 243 (1992) 393–
711
441.
AC C
EP
709
712
[40] Y. Zang, R. L. Street, J. R. Koseff, A non-staggered grid, fractional
713
step method for time-dependent incompressible navier-stokes equations
714
in curvilinear coordinates, J. Comput. Phys. 114 (1994) 18–33.
715
[41] L. Graftieaux, M. Michard, N. Grosjean, Combining piv, pod and vortex
716
identification algorithms for the study of unsteady turbulent swirling
717
flows, Meas. Sci. Technol. 12 (2001) 1422. 44
ACCEPTED MANUSCRIPT
• • •
AC C
EP
TE D
M AN U
SC
•
The effect of an upstream bluff body on the energy harvesting performance of a pitching and heaving hydrofoil is investigated. The upstream body can improve the efficiency of the hydrofoil by about 30%. Mutual interaction of the vortex shed from the upstream body and the leadingedge vortex of the hydrofoil is a main mechanism for efficiency improvement. The change in the effective angle of attack for the hydrofoil is another important mechanism for efficiency improvement. The upstream body placed in an improper position can cause efficiency drop and irregular power generation.
RI PT
•