2018 15th International Conference on Ubiquitous Robots (UR)

Hawaii Convention Center, Hawai'i, USA, June 27-30, 2018

A Study on Force Generated by Gymnotiform Undulating Fin

Van Hien Nguyen, Canh An Tien Pham, Van Dong Nguyen, Dae Hwan Kim, and Tan Tien nguyen

 𝑠 is the coordinate of fin-ray module along the fin,



Abstract- Underwater vehicles inspired by the motion of fish

can generate a highly efficient force. This paper considers the 𝑠 = [1, 𝑆]
gymnotiform undulating fin which imitates the motion of South
American black ghost knifefish. The equations of the force  𝑆 is the length of fin membrane.
generated by both continuous and discontinuous sine wave are  𝑙 is the coordinate length of fin-ray 𝑙 ∈ [0, 𝐿𝑞 ].
established. Using these equations, the factors influencing this
force are determined. The results in computational method,  𝜃(𝑠, 𝑡) is the motion equation of fin-ray, which is
experimental method and 2D simulation method are matched defined by:
and suggest the set of three prevalent factors, namely frequency, 2𝜋𝑠
amplitude and wavelength.
𝜃(𝑠, 𝑡) = 𝜃𝑚𝑎𝑥 sin (2𝜋𝑓𝑡 + 𝜙0 − ) 
𝜆

I. INTRODUCTION where,
 𝜃𝑚𝑎𝑥 is the angle amplitude of fin-ray.
Gymnotiform is a type of undulating fin motion which
allows fish to move quickly underwater with high efficiency.  𝑓 is the frequency.
According to research [1], this swimming pattern promises a  𝜙0 −
2𝜋𝑠
is the initial phase of fin-ray, and it
flexible and stable motion. This case study attracts a lot of 𝜆
researches around the world [2]–[8], and previously influences the profile of fin membrane. 𝜙0 is the initial
constructed bionic robots have given an initial view about its phase of the first fin-ray, 𝜆 is wavelength.
potential [9]–[13]. Many an approach are used to examine the From these parameters, we can generate the fin
force generated, the majority of which are based on
experiment and simulation. This paper specifically focuses on 15
three methods: computational, experimental and 2d
simulation, with an additional account for the discretization of 10

the continuous fin. y

5
0.3
10
II. COMPUTATIONAL METHOD 0
5 0.2
x
-10 0 z
A. Continuous fin 0 z -5
0.1
x 0.2
First, we will survey the dynamics of continuous fin. 10
0
0.1 -10 0

Based on the gymnotiform undulating fin model, an ideal fin

a. Fin profile (𝑓 = 1𝐻𝑧, 𝜙0 = 0, 𝜃𝑚𝑎𝑥 = 300 , 𝜆 = 𝑆)
including one full sine wave is considered. This model was
introduced by Hanbin Xie and Lincheng Shen [11]
x 15
Lq
θ 10

y 5
y 0.3
10
0 5
Figure 1. Ideal continuous fin x
0.2
10 0 z
The motion of the fin can be defined by 0
-5
0.1
z x 0.2
-10
𝑥(𝑠, 𝑙, 𝑡) = 𝑠 -10
0
0.1 0

{𝑦(𝑠, 𝑙, 𝑡) = 𝑙𝑐𝑜𝑠[𝜃(𝑠, 𝑡)] 𝑠 ∈ [0, 𝑆], 𝑙 ∈ [0, 𝐿𝑞 ]  b. Fin profile (𝑓 = 1𝐻𝑧, 𝜙0 = 0, 𝜃𝑚𝑎𝑥 = 300 , 𝜆 =
𝑧(𝑠, 𝑙, 𝑡) = 𝑙𝑠𝑖𝑛[𝜃(𝑠, 𝑡)] 0.5𝑆)
where, Figure 2. Fin profiles correspond to input parameters.

 𝑡 is time. Considering a surface area element 𝑑𝑆 on the fin

membrane, we can analyze the force generated by this
element.
V.H. Nguyen is with the PetroVietnam Camau Fertilizer Joint Stock
Company, (e-mail: [email protected]). Fluid apply a force on the membrane, the equation of the
C.A.T. Pham, V.D. Nguyen, T.T. Nguyen are with the Department of force is
Mechatronics, HCMUT, Vietnam (e-mail: [email protected], 𝑑𝐹⃗𝑛 1
[email protected], [email protected]). 𝑓⃗𝑛 = ⃗⃗⃗⃗
= − 𝜌𝐶𝑛 ‖𝑉 ⃗⃗⃗⃗
𝑛 ‖𝑉𝑛
𝑑𝑆 2
D.H. Kim is with the Department of Mechanical and Automotive {  
𝑑𝐹⃗𝜏 1
Engineering, Pukyong National University, Busan, Korea (e-mail: 𝑓⃗𝜏 = ⃗⃗⃗⃗𝜏 ‖𝑉
= − 𝜌𝐶𝜏 ‖𝑉 ⃗⃗⃗⃗𝜏
[email protected]). 𝑑𝑆 2

978-1-5386-6333-2/18/$31.00 ©2018 IEEE 247

where, Eventually, the total force is
2
 𝐹⃗𝑛 and 𝐹⃗𝜏 are respectively the force vectors at 𝑑𝑆 in
normal direction and tangential direction. 1 𝐿𝑆
𝜃𝑠̇
𝐹⃗ = ⃗⃗⃗⃗
𝐹𝑛 = − ∬ 𝜌𝐶𝑛 (𝑓𝜆)2 ×
 𝜌 is density of fluid. 00 2 2
√ ̇
 𝐶𝑛 , 𝐶𝜏 are respectively the drag coefficient and the lift ( 1 + (𝑙𝜃𝑠 ) )
coefficient. 𝐶𝑛 depends on shape geometry of fin and −𝑙𝜃𝑠̇
Reynolds Number. 𝑠𝑖𝑔𝑛(𝜃𝑠̇) [−𝑠𝑖𝑛[𝜃(𝑠, 𝑡)]] 𝑑𝑠𝑑𝑙 
⃗⃗⃗⃗
 𝑉 ⃗⃗⃗⃗ cos[𝜃(𝑠, 𝑡)]
𝑛 , 𝑉𝜏 are respectively normal and tangential velocity
vectors of 𝑑𝑆. As in (12), the total force mainly depends on the angle
amplitude of fin-ray, frequency, wavelength and the length of
12 fin. The size for each model of underwater vehicles is usually
10 ds fixed. Three factors, namely angle amplitude of fin-ray,
8
frequency and wavelength are examined.
y 6 dl
4 Using MATLAB to compute the total force. With
0
2
𝐶𝑛 = 2.8, 𝜌 = 998.1𝑘𝑔/𝑚3 as in the database of ANSYS
10
5
FLUENT, we use 𝑙 = 0.1𝑚 and 𝑆 = 0.3𝑚 to compare with
0 z
-5
x 0.2
0.3
the experiment. The total force generated is shown in Fig.4.
0.1
-10 -4
0
x 10
Figure 3. Element of fin membrane 6 2

Total force of the fin applied by fluid can be defined by 1

4
Fx (N)

Fy (N)
𝐹 = ∬ 𝑓⃗𝑛 𝑑𝑆
⃗⃗⃗⃗ 0
{ 𝑛  
𝐹𝜏 = ∬ 𝑓⃗𝜏 𝑑𝑆
⃗⃗⃗⃗ 2
-1
Because Reynolds Number is too large (104 ~ 108 in
0 -2
transient), 𝐶𝑛 far outweighs 𝐶𝜏 . Therefore, we can ignore the 0 1 2 3 0 1 2 3
tangential force. Frequency (Hz) Frequency (Hz)
-4
x 10
Considering element 𝑃 on the membrane with the 1 6
Fx
coordinate 𝑟⃗𝑜𝑝 , the force and velocity vectors at 𝑃 are 0.5 4 Fy
Fz (N)

F (N)

𝐹⃗ = ⃗⃗⃗⃗ 𝐹𝜏 = ∬(𝑓⃗𝑛 + 𝑓⃗𝜏 )𝑑𝑆

𝐹𝑛 + ⃗⃗⃗⃗  0 2
Fz

⃗⃗⃗⃗⃗⃗
𝑉 ⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝑒𝑝 = 𝑉𝑒𝑞 + 𝑉𝑞𝑝   -0.5 0
𝑥(𝑠, ̇𝑙, 𝑡) 0 -1 -2
𝜕𝑃⃗⃗𝑞(𝑠,𝑙,𝑡)
⃗⃗⃗⃗⃗⃗
𝑉𝑞𝑝 = ̇ 𝑡)] = [−𝑙𝜃̇𝑡 𝑠𝑖𝑛[𝜃(𝑠, 𝑡)]]
= [𝑦(𝑠, 𝑙,  0 1 2 3 0 1 2 3
𝜕𝑡 Frequency (Hz) Frequency (Hz)
𝑧(𝑠, 𝑙,̇ 𝑡) 𝑙𝜃̇𝑡 𝑐𝑜𝑠[𝜃(𝑠, 𝑡)]
a. Relation between total force generated and frequency of fin-ray
We assume that the fin is fixed on the coordinate system, (𝜆 = 0.3𝑚, 𝜃𝑚𝑎𝑥 = 300 )
so ⃗⃗⃗⃗⃗⃗
𝑉𝑒𝑞 = 0, ⃗⃗⃗⃗⃗⃗⃗⃗
𝜔𝑒𝑞 = 0. The inertial coordinate origin on the x 10
-7
1.5 2
ground and coordinate origin on the robot are coincident. The
velocity vector of 𝑃 is
1 0
Fx (N)

Fy (N)

0
⃗⃗⃗⃗
𝑉𝑝 = [ −𝑙𝜃 ̇ 𝑡 𝑠𝑖𝑛[𝜃(𝑠, 𝑡)]]  0.5 -2
𝑙𝜃̇𝑡 𝑐𝑜𝑠[𝜃(𝑠, 𝑡)]
Then the force generated is simplified by 0
0 0.2 0.4 0.6 0.8
-4
0 0.2 0.4 0.6 0.8
Amplitude (rad) Amplitude (rad)
𝐿𝑆 1
𝐹⃗ = ⃗⃗⃗⃗
𝐹𝑛 = ⃗⃗⃗⃗
− ∬00 𝜌𝐶𝑛 ‖𝑉 ⃗⃗⃗⃗
𝑛 ‖𝑉𝑛 √1 + (𝑙𝜃̇𝑠 )2 𝑑𝑠𝑑𝑙   x 10
-6
2 1 1.5
Fx
The normal vector 𝑛⃗⃗ of 𝑑𝑆 is 1 Fy
0
Fz (N)

−𝑙𝜃𝑠̇ Fz
F (N)

𝜕𝑃(𝑠,ℎ,𝑡) 𝜕𝑃(𝑠,𝑙,𝑡) 0.5

𝑛⃗⃗ = 𝑥 = [−𝑠𝑖𝑛[𝜃(𝑠, 𝑡)]] 
𝜕𝑠 𝜕𝑙 -1
cos[𝜃(𝑠, 𝑡)] 0

The normal velocity vector of 𝑃 is -2 -0.5

0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
−𝑙𝜃𝑠̇ Amplitude (rad) Amplitude (rad)
𝑙𝜃𝑡̇
⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗𝑝 . 𝑛⃗⃗0 )𝑛⃗⃗0 =
𝑉𝑝𝑛 = (𝑉 
̇ 2 +1 [−𝑠𝑖𝑛[𝜃(𝑠, 𝑡)]]
b. Relation between total force generated and angle amplitude of fin-ray
(𝑙𝜃𝑠) (𝑓 = 1𝐻𝑧, 𝜆 = 0.3𝑚)
cos[𝜃(𝑠, 𝑡)]

248
 0.8 0.1 In Figure 5, the coordinate of q is expressed by 𝑟⃗𝑞𝑖 and
0.6 0 𝑟⃗𝑞𝑖+1
Fx (N)

Fy (N)
0.4 -0.1 𝑟⃗𝑞(𝑖+1) − 𝑟⃗𝑞𝑖
𝑟⃗𝑞 = 𝑟⃗𝑞𝑖 + 𝑤 (14)
0.2 -0.2 ‖𝑟⃗𝑞(𝑖+1) − 𝑟⃗𝑞𝑖 ‖
0 -0.3 𝐷𝑖
0 0.5 1 0 0.5 1
Wavelength (m) Wavelength (m) 𝑟⃗𝑞 = [ ℎ𝑠𝑖𝑛[𝜃(𝑖, 𝑡)] ] + 𝑀 ∗ 𝑤 
Fx ℎ𝑐𝑜𝑠[𝜃(𝑖, 𝑡)]
0.5 1
Fy
Fz
where,
0.5
𝐷
Fz (N)

F (N)

0
0 𝑀=[ ℎ[𝑠𝑖𝑛(𝜃(𝑖 + 1, 𝑡)) − sin(𝜃(𝑖, 𝑡))) ] ×
-0.5
-0.5 ℎ[𝑐𝑜𝑠(𝜃(𝑖 + 1, 𝑡)) − cos(𝜃(𝑖, 𝑡))]
1
-1 -1  
0 0.5 1 0 0.5 1 √𝐷 2 +2ℎ2 −2ℎ2 cos[𝜃(𝑖,𝑡)−𝜃(𝑖+1,𝑡)]
Wavelength (m) Wavelength (m)
c. Relation between total force generated and number of waves Set
(𝑓 = 1𝐻𝑧, 𝜃𝑚𝑎𝑥 = 30 ) 0
𝐶 = √𝐷2 + 2ℎ2 − 2ℎ2 cos[𝜃(𝑖, 𝑡) − 𝜃(𝑖 + 1, 𝑡)] 
Figure 4. Influence of three factors on total force generated.
As we can see in those figures, the total force increases Using the same equation of force generated by an area 𝑑𝑆 in
remarkably when the frequency and angle amplitude increase. previous section
But the total force fluctuates upward if the wavelength is less 1
than 0.3 m, in other words, less than the length of the fin. This 𝐹⃗ = ⃗⃗⃗⃗ ⃗⃗⃗⃗
𝐹𝑛 = − ∬ 𝜌𝐶𝑛 ‖𝑉 ⃗⃗⃗⃗
𝑛 ‖𝑉𝑛 𝑑𝑆 
2
means the number of waves on the fin is larger than 1. In where,
addition, 𝐹𝑦 and 𝐹𝑧 are insignificant, which helps stabilize the
fin when it undulates. In contrast, the magnitude of 𝐹𝑥 , 𝐹𝑦 , 𝐹𝑧 𝑉𝑛 = (𝑛⃗⃗0 . 𝑟̇⃗𝑞 )𝑛⃗⃗0
⃗⃗⃗⃗ 
⃗⃗𝑞 𝜕𝑟
𝜕𝑟 ⃗⃗𝑞
rises dramatically when the wavelength is greater than 0.3m.
Without a full wavelength on the fin, the fin will be subjected 𝑛⃗⃗0 = 𝜕ℎ 𝜕𝑤
⃗⃗
𝜕𝑟𝑞 𝜕𝑟⃗⃗𝑞  
‖ ‖
to forces in the y- and z-direction, which alters its direction. 𝜕ℎ 𝜕𝑤
𝜕𝐶
This can explain the process of changing direction of fish in −𝐷𝑤
𝜕ℎ
nature. 𝐶2
𝜕𝑟⃗𝑞 𝐶
B. Fin discretization = 𝑠𝑖𝑛(𝜃𝑖 ) + 𝑤(𝑠𝑖𝑛(𝜃𝑖+1 ) − 𝑠𝑖𝑛(𝜃𝑖 )) 𝐶−ℎ𝜕ℎ  
𝜕ℎ 2 𝐶
For robotic fabrication, the continuous fin is not 𝜕𝐶
𝐶−ℎ
approachable. Michael Sfakiotakis presented a model of [𝑐𝑜𝑠(𝜃𝑖 ) + 𝑤(𝑐𝑜𝑠(𝜃𝑖+1 ) − 𝑐𝑜𝑠(𝜃𝑖 ))
𝜕ℎ
]
𝐶2
discontinuous fin [1]. The fin is formed by equidistant fin- 𝐷
rays. Its undulation is defined by 𝐶
𝜕𝑟⃗𝑞 ℎ
2𝜋𝑛 𝐿 = [𝑠𝑖𝑛(𝜃𝑖+1 ) − 𝑠𝑖𝑛(𝜃𝑖 )]  
𝜃(𝑛, 𝑡) = 𝜃𝑚𝑎𝑥 sin (2𝜋𝑓𝑡 + 𝜙0 − ) (13) 𝜕𝑤 𝐶
𝑁 − 1𝜆 ℎ
[𝐶 [𝑐𝑜𝑠(𝜃𝑖+1 ) − 𝑐𝑜𝑠(𝜃𝑖 )]]
where,
The normal vector of 𝑑𝑆 is calculated by
 𝜃(𝑛, 𝑡) is the angle of n-th ray at moment 𝑡.
−ℎ𝐶𝑠𝑖𝑛(𝜃𝑖+1 − 𝜃𝑖 )
 𝜙0 is initial phase, in this paper, 𝜙0 = 0. 1
𝑛⃗⃗ = 2 [ 𝐷𝑐𝑜𝑠(𝜃𝑖 )𝐶 + 𝐷𝑤(cos(𝜃𝑖+1 ) − cos(𝜃𝑖 )) ] 
𝐶
 𝜃𝑚𝑎𝑥 is amplitude of n-th ray. −𝐷𝑠𝑖𝑛(𝜃𝑖 )𝐶 − 𝐷𝑤(sin(𝜃𝑖+1 ) − sin(𝜃𝑖 ))
 𝜆 is wavelength, 𝐿 is the length of fin. Because 𝐶 is the maximum distance of two fin-rays when
they undulate with the length of ray ℎ at moment 𝑡, 𝐶 =
{O}
x
D
𝑤𝑚𝑎𝑥 . The normal unit vector can be described by
ℎ
x
y h
− sin(𝜃𝑖+1 − 𝜃𝑖 )
{P} 𝐷 𝐷
r q(i+1)
𝑛⃗⃗0 = [ cos(𝜃𝑖 ) + 𝑤
̅(cos(𝜃𝑖+1 ) − cos(𝜃𝑖 )) ] 
r q 𝜍
rq(i)
y
z
w
−sin(𝜃𝑖 ) − 𝑤̅(sin(𝜃𝑖+1 ) − sin(𝜃𝑖 ))
z
where,

2𝐷2 𝑤 ̅ − 1)(1 − cos(𝜃𝑖+1 − 𝜃𝑖 )) +

̅(𝑤
𝜍=√   
ℎ2 𝑠𝑖𝑛2 (𝜃𝑖+1 − 𝜃𝑖 ) + 𝐷2
Figure 5. Coordinate systems on discontinuous fin. Velocity vector at 𝑞 is the positional derivative over time

249
𝑣⃗𝑞 = 𝑟̇𝑞   0.8 0.5

𝐷ℎ 0.6

Fy (N)
= ([𝑐𝑜𝑠(𝜃𝑖 )(1 − 𝑤
̅) + 𝑤
̅𝑐𝑜𝑠(𝜃𝑖+1 )] × 0

Fx(N)
𝜍 0.4
[cos(𝜃𝑖 )𝜃̇𝑖 + (𝑐𝑜𝑠(𝜃𝑖+1 )𝜃̇𝑖+1 − 𝑐𝑜𝑠(𝜃𝑖 )𝜃̇𝑖 )𝑤
̅] + 0.2
-0.5

[𝑠𝑖𝑛(𝜃𝑖 )(𝑤
̅ − 1) − 𝑤̅𝑠𝑖𝑛(𝜃𝑖+1 )] × 0 -1
0 0.5 1 0 0.5 1
[−sin(𝜃𝑖 )𝜃𝑖 + (𝑠𝑖𝑛(𝜃𝑖 )𝜃̇𝑖 − 𝑠𝑖𝑛(𝜃𝑖+1 )𝜃̇𝑖+1 )𝑤
̇ ̅]) Wavelength (m) Wavelength (m)
Fx
The force at 𝑞 is 0.1 1 Fy
1 1 2
𝑓⃗ = ⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗
𝑓𝑛𝑞 = − 𝜌𝐶𝑛 ‖𝑉 ⃗⃗⃗⃗ ⃗⃗⃗⃗ ⃗⃗0 𝑠𝑔𝑛(𝑉
𝑛 ‖𝑉𝑛 = − 𝜌𝐶𝑛 ‖𝑉𝑛 ‖ 𝑛
⃗⃗⃗⃗
𝑛 )  0 0.5
Fz
2 2

Fz (N)

F (N)
Calculate on the whole membrane with 𝑖 ∈ [0 𝑁 − 1], ℎ ∈ -0.1 0
[ℎ𝑚𝑖𝑛 ℎ𝑚𝑎𝑥 ], 𝑤 ∈ [𝑤𝑚𝑖𝑛 𝑤𝑚𝑎𝑥 ] -0.2 -0.5
ℎ𝑚𝑎𝑥 𝑤𝑚𝑎𝑥
𝐹⃗ = ⃗⃗⃗⃗
𝐹𝑛 = ∑𝑁−1
𝑖=0 ∫ℎ𝑚𝑖𝑛 ∫0
̇ 𝑑𝑤𝑑ℎ 
𝐹𝑛𝑞  -0.3 -1
0 0 0.5 1 0 0.5 1
Wavelength (m) Wavelength (m)
According to our previous study [14], the suitable number
of fin-rays is 16 for robotic manufacture. Thus, we calculate c. Relation between total force generated and number of waves
the total force in MATLAB with 𝑁 = 16 and other constants (𝑓 = 1𝐻𝑧, 𝜃𝑚𝑎𝑥 = 300 )
in the continuous fin. The results are shown in Figure 6. Figure 6. Influence of three factors on total force.
6 1 The values of 𝐹𝑥 in both continuous and discontinuous fin
models are similar. However, it can be observed that in the
0.5
latter, the values of 𝐹𝑦 and 𝐹𝑧 are greater than that of the
Fy (N)
Fx (N)

4
0 former. This leads to two conclusions: the efficiency of
2
-0.5
discontinuous fin is lower, and the movement in
discontinuous fin is less stable than that of continuous fin.
0 -1
0 1 2 3 0 1 2 3 III. EXPERIMENT
Frequency (Hz) Frequency (Hz)
A model of one discontinuous fin with 16 fin-rays is used
0.2
Fx to measure the force generated. This model includes a
0.1 4 F polyester membrane and a DC motor GA12 N20 for each fin-
6 y
Fz(N)

ray, which transmits torque to the fin-rays through the four

F (N)

Fz
0 2
bar linkage mechanism. A sensor is attached directly to this
-0.1 0 mechanism to determine the angle position of the fin-rays.
Only the results on the x-direction are measured because the
-0.2 -2
0 1 2 3 0 1 2 3 forces on the remaining directions are insignificant and thus,
Frequency (Hz) Frequency (Hz) difficult to measure.
a. Relation between total force generated and frequency of fin-ray
(𝜆 = 0.3𝑚, 𝜃𝑚𝑎𝑥 = 300 )
1.5 0

1 -0.05
Fx (N)

Fy (N)

0.5 -0.1

0 -0.15
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Amplitude (rad) Amplitude (rad)

0 1.5
Fx
-0.005 1 Fy
Fz
F (N)
Fz (N)

-0.01 0.5

-0.015 0 Figure 7. Module of one fin-ray

-0.02 -0.5 The entire fin is placed on a guide rail. As the fin
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Amplitude (rad) Amplitude (rad) undulates along the rail, a load-cell is applied to fixate its
b. Relation between total force generated and angle amplitude of fin-ray movement to the x-direction and measure the force generated
(𝑓 = 1𝐻𝑧, 𝜆 = 0.3𝑚) in this direction. A micro-controller will read the data in one
cycle, and then calculate the average force. The specifications
of the model are shown in TABLE I

250
 TABLE I. MODEL SPECIFICATIONS 2.5
Length of fin 0.3𝑚 Force measured at 40
0

Length of fin-ray 0.1𝑚 Force measured at 30

0

Angle amplitude 200 ÷ 400 2 Force measured at 20

0

Frequency 0.5𝐻𝑧 ÷ 2𝐻𝑧 0

Force calculated at 400
Number of fin-rays 16 Force calculated at 30
Mass 763𝑔 Force calculated at 20
0

(N)
Tank size 1.2𝑚 × 0.6𝑚 × 0.6𝑚 1.5

Fx
Linear
Controller
guide rail
box 1

0.5
DC
motor

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Fin Fin-ray Frequency (Hz)
membrane c. 𝜆 = 0.1𝑚
Figure 9. Relation between frequency and measured and calculated force in
Figure 8. Actual experiment model the x-direction at different angle amplitude and 𝜆.
Figure 9 show the results of both computational and It can be observed that with different values of 𝜆, the force
experimental method in the discontinuous fin case. generated by both methods are comparable. However, the
5 experimental results are lesser than that of computation. This
4.5 Force measured at 40
0 can be explained by accounting for the friction between the
Force measured at 30
0
guide rail and the model. In addition, the relatively small size
0
4 Force measured at 20 of the actual tank causes fluid waves to rebound upon impact
0

3.5
Force calculated at 400 with the tank wall and inflict significant force on the model,
Force calculated at 30 leading to the fluctuation in the results.
0
Fx (N)

3 Force calculated at 20
IV. SIMULATION METHOD
2.5
We employ ANSYS FLUENT 2D to create a model of
2 discontinuous fin profile. The amplitude 𝜑𝑚𝑎𝑥 of simulated
1.5 waves is determined by

1 𝜑𝑚𝑎𝑥 = 𝑙𝑠𝑖𝑛𝜃𝑚𝑎𝑥  

0.5 The length of the fin 𝐿 = 0.3𝑚 and the simulated fluid
area is equal to the tank cross-section.
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 The program employs the Finite Element Method (FEM)
Frequency (Hz)
to calculate on each element. Triangular meshing is used, and
a. 𝜆 = 0.3𝑚
the fluid is assumed to be transient and pressure-based.
3.5
0
Force measured at 40
3 Force measured at 30
0

0
Force measured at 20
0
2.5 Force calculated at 400
Force calculated at 30
(N)

0
Force calculated at 20
2
Fx

1.5

0.5
Figure 10. Model meshing
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 The Dynamic Mesh function, along with the UDF module,
Frequency (Hz)
is employed to define the fin motion. The fin motion process
b. 𝜆 = 0.15𝑚
is divided into 2 stages:

251
  Stage 1: develop from linear profile to sine wave force generated by both continuous and discontinuous sine
profile. wave are established. Both theoretical and experimental
 Stage 2: undulate along the sine wave. results have done to prove the effectiveness of the study. In
addition, the paper opens for further research into the fin’s
direction changing.

REFERENCES
[1] M. Sfakiotakis and J. Fasoulas, “Development and experimental
validation of a model for the membrane restoring torques in undulatory
fin mechanisms,” in Control and Automation (MED), 2014 22nd
Mediterranean Conference of, 2014, pp. 1540–1546.
[2] S. Wang, X. Dong, and L.-J. Shang, “Thrust analysis of the undulating
ribbon-fin for biomimetic underwater robots,” in Intelligent Control
and Information Processing (ICICIP), 2011 2nd International
Conference on, 2011, vol. 1, pp. 335–340.
[3] M. Lamas, J. Rodríguez, C. Rodríguez, and P. González, “Three-
a. Stage 1 dimensional cfd analysis to study the thrust and efficiency of a
biologically-inspired marine propulsor,” Pol. Marit. Res., vol. 18, no.
1, Jan. 2011.
[4] A. A. Shirgaonkar, O. M. Curet, N. A. Patankar, and M. A. MacIver,
“The hydrodynamics of ribbon-fin propulsion during impulsive
motion,” J. Exp. Biol., vol. 211, no. 21, pp. 3490–3503, Nov. 2008.
[5] H. Liu and O. M. Curet, “Propulsive performance of an under-actuated
robotic ribbon fin,” Bioinspir. Biomim., vol. 12, no. 3, p. 036015, Jun.
2017.
[6] A. Willy and K. H. Low, “Initial experimental investigation of
undulating fin,” in Intelligent Robots and Systems, 2005.(IROS 2005).
2005 IEEE/RSJ International Conference on, 2005, pp. 1600–1605.
[7] F. Liu, K.-M. Lee, and C.-J. Yang, “Hydrodynamics of an Undulating
Fin for a Wave-Like Locomotion System Design,” IEEEASME Trans.
Mechatron., vol. 17, no. 3, pp. 554–562, Jun. 2012.
b. Stage 2 [8] J. Lighthill and R. Blake, “Biofluiddynamics of balistiform and
Figure 11. Fin motion process gymnotiform locomotion. Part 1. Biological background, and analysis
by elongated-body theory,” J. Fluid Mech., vol. 212, pp. 183–207,
The simulation results are displayed in figure 12, but will 1990.
not be compared with that of other methods due to the [9] T. Hu, L. Shen, L. Lin, and H. Xu, “Biological inspirations, kinematics
difference in dimension. modeling, mechanism design and experiments on an undulating robotic
fin inspired by Gymnarchus niloticus,” Mech. Mach. Theory, vol. 44,
30 no. 3, pp. 633–645, Mar. 2009.
[10] B. Peter, R. Ratnaweera, W. Fischer, C. Pradalier, and R. Y. Siegwart,
“Design and evaluation of a fin-based underwater propulsion system,”
25 0 in Robotics and Automation (ICRA), 2010 IEEE International
Force simulated at 40
Force simulated at 30
0 Conference on, 2010, pp. 3751–3756.
0 [11] H. Xie and L. Shen, “Dynamic analysis on the bionic propulsor
Force simulated at 20
(N/m)

20 imitating undulating fin of aquatic animals,” in Robotics and

Biomimetics, 2007. ROBIO 2007. IEEE International Conference on,
2007, pp. 268–273.
Fx

15 [12] M. Epstein, J. E. Colgate, and M. A. MacIver, “Generating thrust with

a biologically-inspired robotic ribbon fin,” in Intelligent Robots and
Systems, 2006 IEEE/RSJ International Conference on, 2006, pp. 2412–
10 2417.
[13] O. M. Curet, N. A. Patankar, G. V. Lauder, and M. A. MacIver,
“Mechanical properties of a bio-inspired robotic knifefish with an
5 undulatory propulsor,” Bioinspir. Biomim., vol. 6, no. 2, p. 026004,
Jun. 2011.
[14] Nguyen V.D., Phan D.K., Pham C.A.T., Kim D.H., Dinh V.T., Nguyen
0
T.T. (2018) Study on Determining the Number of Fin-Rays of a
0.5 1 1.5 2 Gymnotiform Undulating Fin Robot,Recent Advances in Electrical
Frequency (Hz) Engineering and Related Sciences: Theory and Application. AETA
Figure 12. Simulation results 2017. Lecture Notes in Electrical Engineering, vol 465. Springer

It can be observed that the lines above share the same

trend with those in the experiment method.

V. CONCLUSION
This paper presents the studied results on the influence of
frequency, amplitude and wavelength to the force generated
by the gymnotiform undulating fin. The equations of the

252