Int. J. Nav. Archit. Ocean Eng.

(2015) 7:409~425
[Link]
ⓒ SNAK, 2015 pISSN: 2092-6782, eISSN: 2092-6790

Dynamics modeling of a semi-submersible autonomous

underwater vehicle with a towfish towed by a cable

Jinmo Park and Nakwan Kim

Department of Naval Architecture and Ocean Engineering, Seoul National University

ABSTRACT: In this paper, we employ a dynamics modeling method for investigating a multi-body dynamics system of
semi-submersible autonomous underwater vehicles consisting of a towing vehicle operated near the water surface, a
tow cable, and a towfish. The towfish, which is towed by a marine cable for the purposes of exploration or mine hunting,
is modeled with a Six-Degree-of-Freedom (6-DOF) equation of motion that reflects its hydrodynamics characteristics.
The towing cable, which can experience large displacements and deformations, is modeled using an absolute nodal
coordinate formulation. To reflect the hydrodynamic characteristics of the cable during motion, the hydrodynamic force
due to added mass and the drag force are imposed. To verify the completeness of the modeling, a few simple numerical
simulations were conducted, and the results confirm the physical plausibility of the model.

KEY WORDS: Absolute nodal coordinate formulation; Flexible cable dynamics; Multi-body dynamics; Large deforma-
tion; Semi-submersible AUV; Towed system.

INTORODUCTION
Because the need for subsea exploration for discovery of subsea resources and for military purposes such as mine hunting is
increasing, studies of marine cable motion and the use of towfish have been performed by many researchers. Most previous stu-
dies on these subjects were focused on a system that consisted of a surface vessel, a marine cable, and a towfish (Grosenbaugh,
2007; Huang, 1994; Park et al., 2003; Vaz and Patel, 1995). Because a marine cable exhibits highly nonlinear characteristics
under water, analysis of marine cable dynamics has typically relied on numerical methods. In addition, the highly nonlinear
motion of a marine cable can cause coupled motion of the towfish, add to the towfish’s own nonlinear motion. A few studies
have examined ways to control the motion of towfish, using devices such as a nonlinear adaptive controller (Curado et al.,
2010). For analysis of the motion of marine cables, the lumped mass approximation method (Buckham et al., 2003; Kamman
and Nguyen, 1990; Kamman and Huston, 2001) and classical cable theory based on the finite element method (Park et al., 2003;
Wu and Chwang, 2001; Yuan et al., 2013) have usually been employed. The lumped mass approximation method assumes that
the mass is concentrated at a nodal point and that the cable segment is a bar element. The great advantage of this method is its
ease of implementation. The disadvantage of this method is that it cannot take into consideration bending deformation of the
cable element. To address this limitation of the lumped mass approximation method, classical cable theory based on the finite
element method can be applied. This approach can reveal a cable during motion. However, this approach requires dividing the
cable into many segments to obtain accurate simulation result (Park et al., 2003). Another method for modeling cable dynamics

Corresponding author: Nakwan Kim, e-mail: nwkim@[Link]

This is an Open-Access article distributed under the terms of the Creative Commons
Attribution Non-Commercial License ([Link]
which permits unrestricted non-commercial use, distribution, and reproduction in any
medium, provided the original work is properly cited.

Unauthenticated
Download Date | 9/24/15 11:42 PM
410 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

is the Absolute Nodal Coordinate Formulation (ANCF) method (Berzeri and Shabana, 2000; Gerstmayr and Shabana, 2006;
Gerstmayr et al., 2013; Shabana and Yakoub, 2001; Shabana et al., 1998). Only a few studies have applied this method to the
analysis of underwater cable dynamics (Kim et al., 2012; Takehara et al., 2011). This method was first proposed by Shabana et
al (1998) and has since been developed further by other researchers (Yakoub and Shabana, 1999; 2001). The great advantage of
the ANCF method is that it uses a constant mass matrix in the equation formulation. Furthermore, it can produce accurate
results with fewer cable segments than can be produced using classical cable theory (Berzeri and Shabana, 2000).
However, in the systems previously studied, the motion of the surface vessel dominated the whole system. Therefore, it is
not necessary to consider the interaction force between the mother ship and the subsea system-in other words, the interaction
force is negligible. However, in the case of a semi-submersible Autonomous Underwater Vehicle (AUV) system, the towing
vehicle and the towfish are comparable in size, and the motion of the cable can affect the whole system, so we need to model
the motion of the system as a multi-body system (Buckham et al, 2003). Semi-submersible AUV systems usually consist of
three components: the towing vehicle, the towing cable, and the towfish (see Fig. 1). The towing vehicle is usually operated
near the water surface (at a depth of approximately 3 to 5 meters), and 70-80% of the body is submerged in water. The main
mission of the towing vehicle is to drive the towing cable and towfish. The towing cable connects the two vehicles, and the
towfish explores the underwater environment. Because the towing vehicle is powered by an internal combustion engine, it is
constrained to operate at depths that do not exceed 5 meters. Because the towfish has no self-propulsion, analyzing the cable
motion is very important to controlling the towfish motion. Buckham et al. (2003) conducted dynamics modeling of such a
system. However, in the proposed model, the cable dynamics were based on lumped mass approximation, which may lead to
considerable modeling uncertainty.

Fig. 1 Illustration of semi-submersible AUV.

This paper presents an approach to multi-body dynamics modeling of a semi-submersible system. The towing vehicle and
towfish are modeled using 6-DOF rigid body dynamics. Because the towing cable can experience large deformations and
displacements, its dynamics are modeled using the ANCF method. Considering the fact that the cable element has no signify-
cant effect on torsional motion, we can formulate the cable using a lower-order cable element (Gerstmayr and Shabana, 2006),
and doing so contributes to the computational efficiency of the model. This paper is organized as follows. The first section
presents a brief review of the ANCF method and describes how an external force, including a hydrodynamic force, can be
imposed on the cable dynamics. The second section presents the vehicle dynamics and the interaction force between the ve-
hicles and the cable. The third section explains the computer implementation scheme used to solve the nonlinear equation for
the system dynamics. The fourth section presents the steps in the numerical simulation process. The fifth section presents a few
simple numerical simulations and discussion of the results obtained. Finally, the sixth section presents a summary of the
research and the main conclusions drawn from the results.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 411

CABLE MODELING
In a semi-submersible AUV system, the towing cable connects the towing vehicle and towfish and tows the towfish.
Because the cable is a flexible body and can experience large deformations, depending on the motion of the towing vehicle and
towfish, it is reasonable to model the cable using ANCF. This modeling approach can model large deformations and large
rotations. In this formulation, the cable element is represented by a global position vector and a slope vector in absolute nodal
coordinates, not a Euler angle, taking advantage of a constant mass matrix. In many cable element problems, the torsional
stiffness has no significant effect (Gerstmayr and Shabana, 2006), and it is reasonable to model the cable as a lower-order cable
element and thereby improve the calculation time. In a lower-order cable element based on ANCF, the nodal point of the
element is composed of one position vector and one slope vector, obtained by differentiation with respect to the element center
line. Fig. 2 illustrates the cable element representation.

Fig. 2 Cable element representation.

Governing equation of motion of towing cable

As mentioned in the previous section, the nodal point of the cable element in three-dimensional space is defined by
position vector and slope vector, as shown in Eq. (1):


e = [e1 , e2 , e3 , e4 , e5 , e6 , e7 , e8 , e9 , e10 , e11 , e12 ]
∂rX ∂rY ∂rZ
=e1 rX=
= , e2 rY=
|x 0= , e3 rZ=
|x 0= |x 0 , e4 = , e5 = , e6
=∂ x x 0= ∂ x x 0= ∂x x 0 (1)
∂rX ∂rY ∂rZ
=e7 rX=
= , e8 rY=
|x l = e9 rZ=
|x l ,= |x l , e10 = , e11 = , e12
= ∂x x l = ∂x x l = ∂x x l

where ri (i = X , Y, Z) is the position vector of the nodal point and other components of the partial derivative of each
positional vector with respect to the cable center line, l is the length of cable segment, and x is the arc-length coordinate of
an arbitrary point on the undeformed cable element. The global position vector in an arbitrary cable element can be expressed as
shown in Eq. (2):

 
=r [r=
X rY rZ ]
T
Se (2)

where S is a global shape function that is based on a cubic spline interpolation function and is an isoparametric function. This
shape function can be expressed as shown in Eq. (3):

S = [ S1 I S 2 I S3 I S 4 I ] (3)

where I is a 3 × 3 identity matrix and the polynomial function element Si (i = 1, 2,3, 4) is as shown in Eq. (4):

Unauthenticated
Download Date | 9/24/15 11:42 PM
412 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

1 − 3ξ 2 + 2ξ 3
S1 = l (ξ − 2ξ 2 + ξ 3 )
S2 =
(4)
3ξ 2 − 2ξ 3
S3 = l (ξ 3 − ξ 2 )
S4 =

where ξ is a nondimensional quantity defined as ξ = x / l . The value of this quantity ranges between zero and one.
The generalized mass matrix of the cable element can be derived from a kinetic energy formulation. The kinetic energy of
 
∫
the cable element is expressed as T = 1/ 2 r r T rdV
V
. Because the positional vector is expressed as shown in Eq. (2), the
generalized velocity vector can be expressed as shown in Eq. (5)

 
r = Se (5)

Therefore, the kinetic energy of the cable element can be expressed as shown in Eq. (6):

1   1  T  1  T 
2 ∫V
=T = r r T rdV e (∫ = r S T SdV )e e Me (6)
2 V 2

The generalized mass matrix of the cable element is as shown in Eq. (7):

M = ∫ ρ S T SdV (7)
V

where r is the density of the cable element and V is the volume of the cable element. Note that, in this formulation, the mass
matrix remains constant.

Internal force
Because the cable element can experience large deformations and displacement, we need to consider elastic forces. As
discussed in Section 2, in the case of a cable element, the torsional effect is not significant, and we need only consider elastic
axial and bending force effects. These elastic forces can be derived from the strain energy of the cable element, which is
expressed as shown in Eq. (8):

1
2 ∫l
U = U a + Ub = ( EAε 2 + EI κ 2 )dx (8)

where U a is the axial strain energy, U b is the bending strain energy, ε is the axial strain, and κ is the curvature due to
bending of the cable. Because the elastic force can be derived from the partial derivative of the strain energy with respect to the
position vector, the elastic force component can be written as shown in Eq. (9):

∂ ∂  
Felastic = Fa + Fb =  ∫ EAe 2 dx +  ∫ EI κ 2 dx = K a e + K b e (9)
∂e l ∂e l

where K a is the axial stiffness matrix and K b is the bending stiffness matrix. We then need to express ε and κ as func-
tions of absolute nodal coordinates. Using Green’s strain tensor, ε can be expressed as shown in Eq. (10):

T 
1 ∂r ∂r
=ε ( − 1) (10)
2 ∂x ∂x

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 413

1 T ∂S ∂S  
We also define the following matrix: = Sl S= 'T S ' Sξ Sξ , where = S' = , Sξ . Since r = Se and
l2 ∂x ∂ξ
 
∂r / ∂x = (∂S / ∂x)e=, then ε 1 / 2(eT Sl e − 1) . Therefore, the axial stiffness matrix can be expressed as shown in Eq. (11):

1  
=Ka ∫=
EAe S dx
l l l
2 ∫l
EA(eSl e − 1) Sl dx (11)


Note that stiffness matrix of the axial effect is composed of nodal coordinates e , and the nonlinear effect arises.
The stiffness matrix for the bending effect can be derived from the curvature definition. The curvature is defined as the
magnitude of the normal vector of the deformed shape, κ =| rss | . By the chain rule, rs = ∂r / ∂s =∂r / ∂x ⋅ ∂x / ∂s , and
rs = rx (∂x / ∂s ) = rx / | rx | . The curvature corresponding to small axial deformation can be expressed as shown in Eq. (12):

∂r  
=κ = e T S ′′T S ′′e (12)
∂x 2

Therefore, the stiffness matrix for bending effect is as shown in Eq. (13):

K b = ∫ EIS ′′T S ′′dx

l (13)

External force

The external forces acting on cable during its motion are the gravity force, buoyancy force, drag force, and force due to
added mass. The external force can be calculated from the virtual work principle, as shown in Eq. (14):

F T δ r F=
= T
Sδ e Qeδ e
(14)
∴ Qe =
FT S

where Qe is the generalized external force vector in ANCF and F is the force vector in global coordinates. Using this
formulation, the generalized gravity force can be expressed as shown in Eq. (15):

QGravity = ∫ [0 0 rc g ]SdV (15)

V

where r c is the density of the cable material and g is gravitational acceleration. The buoyancy force can be derived as in
Eq. (16):

QBuoyancy = ∫ [0 0 -ρ g ]SdV (16)

V

Because the towing cable is moving in water, hydrodynamic forces need to be considered. In this case, we consider the drag
force and the force due to added mass. The drag force is composed of two components: the normal-direction drag force and the
tangential drag force.
The drag forces in the normal direction and the tangential direction at an arbitrary point on the cable element can be written
as shown in Eq. (17):

Unauthenticated
Download Date | 9/24/15 11:42 PM
414 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

1 CC
FDn = − CDn ρ w d | Vn | Vn
2
(17)
sc C
FDt = − µVt Nu ( Rn)
d

where FDn is the normal-direction drag force per unit length; FDt is the tangential-direction drag force per unit length; CDn is
the drag coefficient in the normal direction; Nu ( Rn) is the Nusselt number, which is a function of the Reynolds number; r w

is the density of water; d is the diameter of the cable element; sc is the circumference of the cable cross section; Vn is the

normal-direction velocity; Vt is the tangential velocity; and µ is the dynamic viscosity of water. The drag coefficient in the
normal direction (Choc and Casarella, 1971) is expressed in terms of the Reynolds number, as shown in Eq. (18):

8π
C=
Dn (1 − 0.87 S −2 ) (0 < Rn ≤ 1)
RnS d
(18)
CDn = 1.45 + 8.55 Rn −0.9 (1 < Rn < 30)
CDn =1.1 + 4 Rn −0.5 (30 ≤ Rn ≤ 105 )

where Rn is the Reynolds number and S d is defined as S d = −0.07721 + ln(8 Rn −1 ) .

For the tangential drag force, the relation between the average Nusselt number and the Reynolds number given by (Choc
and Casarella, 1971) is defined as shown in Eq. (19):

=
Nu ( Rn) 0.55 Rn1/ 2 + 0.084 Rn 2/3 (19)

Using the vector relation illustrated in Fig. 3, the tangential velocity and the normal velocity vector can be calculated as
show in Eq. (20):

    
Vn= V − (V �a )a
    (20)
Vt = (V �a )a

 
where V is the total velocity vector of the cable element and a is the normalized tangential vector at an arbitrary point on the
cable element. The drag force on the cable can be formulated as shown in Eq. (21):

=
QDrag ∫V
FDnT SdV + ∫ FDt T SdV
V
(21)

The last component of external force on the cable is the force due to added mass. We assume the force acts in the normal
direction. The acceleration in the normal direction can be calculated as shown in Eq. (22):

    
Vn= V − (V �a )a (22)


where V is the total acceleration. The force due to added mass can be written as shown in Eq. (23):


QAd = − ∫ ρ wVn SdV (24)
V

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 415

Fig. 3 Drag force description.

Constraints

In this system, we assume that the connecting parts between the vehicles and the cable are revolute joints. This type of joint
only constrains the translational position. Therefore, the constraint equation for the connections between the vehicles and the
cable can be written as shown below:

Φ (e m , ec1 ) = r m − r c1 = S m e m − S c1ec1 = 0
(25)
Φ (et , ecn ) =r t − r cn =S t et − S cn ecn =0

The first line of Eq. (25) is the constraint equation for the connection between the towing vehicle and the tow cable, and the
second line is the constraint equation for the connection between the towfish and the towing cable. In these equations, e m is the
position of the towing vehicle in the global coordinate system, ec1 is the position vector of the first segment of the towing cable,
et is the position of the towfish, and ecn is the position vector of the nth element (the cable is divided into n segments). This
constraint can be interpreted as the boundary condition.

Overall equation of motion of cable

Integrating the force effects on cable, the overall equation of motion can be expressed as shown in Eq. (26):

  
Me + K (e )e = QDrag + QAd + QGravity + QBuoyancy (26)

where QBuoyancy is the buoyancy force. To impose the constraint equation and the interaction forces between the vehicle and
cable, we employ a Lagrange multiplier. The overall equation of motion is as shown in Eq. (27):


 M Φ eT   e  Qq 
   =   (27)
Φ e 0  λ   γ 

where λ is the Lagrange multiplier, which can be interpreted as the interaction between the cable and a vehicle; Φ e is
the Jacobian matrix of constraints; γ is the matrix of acceleration-level kinematic constraints, which is defined as
  
Φ e = −Φ − (Φ e ) e − 2Φ e ≡ γ ; and Q is the matrix of elastic forces and external forces. To obtain a stable solution,
e tt e e et q

this algebraic equation is solved using the implicit Newmark integration method.

VEHICLE MODELING

The vehicles (the towing vehicle and the towfish) experience little deformation. Therefore, we formulate the motion
of these vehicles using a 6-DOF rigid body model. To describe rigid body motion, two coordinate systems need to be

Unauthenticated
Download Date | 9/24/15 11:42 PM
416 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

defined: the global coordinate system and the body fixed frame. Fig. 4 shows the coordinate systems used for the vehi-
cle modeling.

(a) Towing vehicle. (b) Tow-fish.

Fig. 4 Coordinate systems of vehicles.

6-DOF equation of motion of vehicle

Based on Newton’s second law, Kirchhoff’s equation of motion in the body fixed frame of the towing vehicle and the
towfish can be written as shown in Eqs. (28) and (29):

mm [u − vr + wq − xGm (q 2 + r 2 ) + yGm ( pq − r) + zGm ( pr + q )]= X hyd + X prop + X int

mm [v − wp + ur − yGm (r 2 + p 2 ) + zGm (qr − p ) + xGm ( pq + r)] = Yhyd + Yprop + Yint
mm [ w − uq + vp − zGm ( p 2 + q 2 ) + xGm (rp + q ) + yGm (rq + p )]= Z hyd + Z prop + Z int
(28)
I xm p + ( I zm − I ym )qr + mm [ yGm ( w − uq + vp ) − zGm (v − wp + ur )]= K hyd + K prop + K int
I ym q + ( I xm − I zm )rp + mm [ zGm (u − vr + wq ) − xGm ( w − uq + vp )]= M hyd + M prop + M int
I zm r + ( I ym − I xm ) pq + mm [ xGm (v − wp + ur ) − yGm (u − vr + wp )]= N hyd + N prop + N int

mt [u − vr + wq − xGt (q 2 + r 2 ) + yGt ( pq − r) + zGt ( pr + q )]= X hyd + X int

mt [v − wp + ur − yGt (r 2 + p 2 ) + zGt (qr − p ) + xGt ( pq + r)] = Yhyd + Yint
mt [ w − uq + vp − zGt ( p 2 + q 2 ) + xGt (rp + q ) + yGt (rq + p )]= Z hyd + Z int
(29)
I xt p + ( I zt − I yt )qr + mt [ yGt ( w − uq + vp ) − zGt (v − wp + ur )] = K hyd + K int
I yt q + ( I xt − I zt )rp + mt [ zGt (u − vr + wq ) − xGt ( w − uq + vp )]= M hyd + M int
I zt r + ( I yt − I xt ) pq + mt [ xGt (v − wp + ur ) − yGt (u − vr + wp )]= N hyd + N int

Eq. (28) is the equation of motion of the towing vehicle, and Eq. (29) is the equation of motion of the towfish. In Eq. (28)
and (29), [u , v, w]T is surge, sway and heave directional linear velocity in body fixed frame, and [ p, q, r ]T is roll, pitch and
yaw angular velocity in body fixed frame. [ xGm , yGm , zGm ]T and [ xGt , yGt , zGt ]T are the position of center of gravity of towing
vehicle and towfish in its body fixed frame. The subscript int refers to the interaction force, the subscript hyd refers to the
hydrodynamic force, and the subscript prop refers to the propulsion force. Because the towfish has no self-propulsion, the
propulsion force term is absent in the right-hand side of the equation of motion. As mentioned earlier, the equation of motion of
vehicles is described in the fixed-body frame. Therefore, we need to transform the frame.

Interaction force and moment between towing vehicle and towing cable
The interaction between the cable and the towing vehicle is assumed to affect the thrust force of the towing vehicle. As Fig.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 417

5 shows, once the equation of the towing vehicle is solved, the boundary condition of the cable is determined. The interaction
force acts on the motion of the towing vehicle. As shown in Eq. (27), λ , the Lagrange multiplier, can be interpreted as the
interaction force. Therefore, the thrust force and moment of towing vehicle can be written as shown in Eq. (30):

 X prop + X int   X prop   λ1 

    λ 
 Yprop + Yint =  Yprop  + RIV
T
 2
 Z prop + Z int   Z prop   λ3 
   
(30)
 K prop + K int   K prop   λ1 
    λ 
 M prop + M int =  M prop  + rattatch × RIV
T
  2
 N prop + N int   N prop   λ3 
   

where RIV T is the transformation matrix between the fixed-body coordinates and the global coordinates for the towing vehicle,
the subscript prop refers to the thrust force and moment without the cable, and rattatch is the position vector of the cable
attached to the towing vehicle in the body fixed coordinate. When rattatch is not zero, we need to consider the moment due to
the cable, expressed as the cross product of rattatch and the interaction force.

Fig. 5 Illustration of interaction between towing vehicle and towing cable.

Interaction force and moment between towing cable and towfish

The interaction force between the towfish and the towing cable is almost same as that between the towing vehicle and the
towing cable. The one difference is that the towfish has no thrust of its own. The interaction force acting on the towfish is an
external force. Thus, we can write the thrust of the towfish as shown in Eq. (31):

 X int  λ4 
 Y  = R T λ 
 int  IV  5 

 Z int   λ6 
(31)
 K int  λ4 
 M=  λ 
 int  rattatch × RIV
T
 5
 N int   λ6 

Because the towfish has no self-propulsion force and moment, the position of the towing cable attached to the towfish plays
an important role in towfish motion.

Unauthenticated
Download Date | 9/24/15 11:42 PM
418 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

Fig. 6 Illustration of interaction between towfish and towing cable.

SIMULAITON PROCEDURE

Fig. 7 illustrates the simulation procedure. The steps in the simulation procedure are as follows:
Step 1. Initialize the parameters, position, velocity, acceleration, and interaction force.
Step 2. Solve the equation of motion for the towing vehicle and the towfish.
Step 3. Using the results of step 2 as boundary conditions for the cable equation, solve the cable dynamics equation using
the implicit Newmark method.
Step 4. Update state of vehicles and cable state.
Step 5. Iterate steps 2-4 until the simulation termination time is reached.

Fig. 7 Simulation procedure.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 419

NUMERICAL SIMULATION
To evaluate the validity of the proposed modeling approach, two numerical simulations are conducted. The first simulation
case is forward running with constant propeller thrust of the towing vehicle. The second case is a turning test with a 20 degree
rudder deflection angle of the towing vehicle at steady-state forward running. The first case allows us to examine the behavior
of the cable, the towfish, and the towing vehicle during the acceleration and the steady-state behavior of the system. The second
case allows us to examine the system’s behavior during a turning motion. In the simulation, the initial conditions of the towing
vehicle are zero velocity and acceleration and a 3.5 m depth. The cable length is set to 200 m, and the towfish has the same
velocity and acceleration as the towing vehicle and a 203.5 m depth. The cable is attached to the bow of the towfish to provide
stability in forward running, and on the towing vehicle, the cable is attached at the center of gravity. As shown in Fig. 1, the
towing vehicle has three fins such as two elevators and one rudder, and the tow fish has main wings, two elevators and one
rudder. The towing vehicle and the towfish have a torpedo-type body and employ the hydrodynamic coefficients of ISiMI (Jun
et al., 2009) for bodies and fins and the coefficients of SNU Glider (Seo, 2009) for the main wings of the towfish. The cable
material density is set to 1300 kg/m3, and the diameter of the circular cross section is set to 4.1 cm . The parameters used for
simulation are given in Table 1.

Table1 Parameters for computation.

Item Value Unit
Density of cable 1300 kg / m3

Length of cable 200 m

Young’s modulus of cable 7×108 N / m2
Diameter of cable 0.041 m

Length of towing vehicle 8.2 m

Mass of towing vehicle 6950 kg

Moment of inertia of towing vehicle [397.04, 3969.8, 3069.1] kg ⋅ m 4

Center of mass of towing vehicle [0,0,0.2] m

Length of towfish 3.3 m

Mass of towfish 416 kg

Moment of inertia of towfish [28.64, 286.35, 301.47] kg ⋅ m 4

Center of mass of towfish [1,0,0.12] m

Case 1: Forward running

Figs. 8-13 illustrate the results of the first simulation case over a time period of 100 seconds. Fig. 8 shows the cable
alignment every 15 seconds. In this figure, we can observe the cable shape during the transient interval. It is because the
velocity of cable near the towing vehicle is faster than that of lower part. This can be seen in Fig. 8. Also, the cable lower bound
is biased to rear. This is because the velocity of towfish is not in steady state and the cable end is biased to towfish position.
After the transient interval, the shape of the cable is almost a straight line. This cable shape indicates that the forward running
motion of the system entered a steady state. Figs. 9 and 10 show that the motion of system entered a steady state at approxima-
tely 50 seconds and that the steady-state surge velocity is 6 m/s. An interesting observation is that the towfish entered a steady
state approximately 30 seconds later than the towing vehicle. The reason for this is that the towfish follows the towing vehicle
and the towing cable; therefore the dynamics of the towing vehicle are transferred to the towfish with a time lag. Figs. 12 and 13
show that the interaction force is well reflected in system dynamics, and in the steady state, only the surge-directional interact-
tion force remains. Based on the results of Case 1, we conclude that the dynamics modeling of the semi-submersible AUV
system describes the behavior of the physical system well.

Unauthenticated
Download Date | 9/24/15 11:42 PM
420 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

Fig. 8 Cable alignment: Case 1.

Fig. 9 Linear velocity profile: Case 1. Fig. 10 Angular velocity profile: Case 1.

Fig. 11 Euler angle: Case 1.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 421

Fig. 12 Interaction force of towing vehicle: Case 1. Fig. 13 Interaction force of towfish: Case 1.

Case 2: Turning behavior

This simulation is conducted to investigate the behavior of the system while executing a U turn. In the simulation, the initial
motion is forward running for 55 seconds to allow the system to enter a steady state, and during the rest of the simulation time, a
turning command with a 20-degree rudder deflection is applied. Once the yaw angle of the towing vehicle reached 170 degrees,
the rudder is set back to a zero angle of deflection to make the towing vehicle execute a U-shaped turning motion. Figs. 14-20
show the results of the simulation. As Fig. 14 shows, the turning radii of the towing vehicle and the towfish are significantly
different. The reason for this is that the turning motion of the towing vehicle is transferred to the towfish with a time lag, and the
motion of the towing cable and the towfish have nonlinearity. The difference in the turning radius of the towing vehicle and that
of the towfish is similar to that reported in Grosenbaugh (2007). Also the depth of the towfish increases during the turning
motion and decreases after the turning motion. A comparison of the results obtained by Grosenbaugh (2007) and those obtained
in this study suggests that the shape of the trajectory of the towfish is different from that of the tow vehicle because of the form
of the towfish and its dynamics.

Fig. 14 Overview of trajectory of towing vehicle and towfish: Case 2.

Unauthenticated
Download Date | 9/24/15 11:42 PM
422 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

In Grosenbaugh (2007), the towfish is a simple cylinder in shape, and in this study, a nonlinear equation of motion is
applied to the towfish dynamics. Fig. 16 shows that the pitch angular velocity oscillated around approximately 80 seconds,
and the same phenomenon is observed in Fig. 15 for the heave velocity. The behavior of the towfish provides a clue to the
explanation of this phenomenon. Figs. 15-17 show that the velocity and Euler angle of the towfish exhibit different patterns
from Case 1. A sudden increase (or decrease) in the velocity and Euler angle indicates that the towfish starts to turn after
approximately 80 seconds. This suggests that the effect of the turning motion of the towfish is transferred to the towing
vehicle, and the phenomena described above occur. In addition, we notice this effect in the interaction force between the
vehicles and the cable, as shown in Figs. 18 and 19. The time lag to enter the steady state that is observed in the first simula-
tion case is almost same as the time lag in the second simulation case. Another observation that we can make is that the overall
motion of the towfish during a turning motion is very unstable. We may also need a proper controller to make the motion of
the towfish stable.

Fig. 15 Linear velocity profile: Case 2. Fig. 16 Angular velocity profile: Case 2.

Fig. 17 Euler angle: Case 2. Fig. 18 Interaction force of towing vehicle: Case 2.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 423

Fig. 19 Interaction force of towfish: Case 2.

Fig. 20 Depth variation of towfish during turning motion.

CONCLUSIONS AND FUTURE WORK

This paper presents approach to modeling the dynamics of a semi-submergible AUV system with a towfish towed by an
underwater cable. The towing vehicle and towfish are modeled using a 6-DOF rigid-body equation of motion, and the marine
cable is modeled using ANCF to express the deformation of cable. In modeling the dynamics, hydrodynamic forces are applied
to the system. To verify the completeness of the modeling approach, two numerical simulations are conducted. The first
simulation case is a straight running situation, and the second is a U-turn motion. Based on the results, we can draw the
following conclusions concerning the proposed modeling procedure for a semi-submersible AUV system.

1) The cable in the system can be modeled well using ANCF. This modeling approach for the cable takes into consideration the
deformation of the cable and the interaction force between vehicles.
2) Hydrodynamic forces, drag forces, and forces due to added mass are applied to the cable element, and in the steady state,
those external forces and the internal forces are in equilibrium.

Unauthenticated
Download Date | 9/24/15 11:42 PM
424 Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425

3) In the first simulation case, the towfish motion is sufficiently stable, which suggests that the towfish can perform its mission
well when moving without turning.
4) There is a time lag in the transfer of the effect of the motion of the towing vehicle to the towfish. The time lags are almost the
same in the two simulations conducted.
5) The results of the turning simulation indicate that the towfish motion becomes less stable during turning, which suggests that
it may need to be operated with a proper controller.
6) The results of the simulations for the modeled system compare well with the results of other studies; the overall motion of the
modeled system agreed well in a physical sense with that observed in other studies.

ACKNOWLEDGMENT

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF)
funded by the Ministry of Education, Science and Technology (NRF-2012R1A1A2008633).

REFERENCES

Berzeri, M. and Shabana, A.A., 2000. Development of simple models for the elastic forces in the absolute nodal co-
ordinate formulation. Journal of Sound and Vibration, 235(4), pp.539-565.
Buckham, B., Nahon, M., Seto, M., Zhao, X. and Lambert, C., 2003. Dynamics and control of a towed underwater vehicle
system, part I: model development. Ocean Engineering, 30(4), pp.453-470.
Choc, Y.I. and Casarella, M.J., 1971. Hydrodynamic resistance of towed cables. Journal of Hydronautics, 5(4), pp.126-131.
Curado, T.F., Pedro, A.A. and Pascoal, A., 2010. Nonlinear adaptive control of an underwater towed vehicle. Ocean
Engineering, 37(13), pp.1193-1220.
Gerstmayr, J. and Shabana, A.A., 2006. Analysis of thin beams and cables using the absolute nodal co-ordinate formulation.
Nonlinear Dynamics, 45(1-2), pp.109-130.
Gerstmayr, J., Sugiyama, H. and Mikkola, A., 2013. Review on the absolute nodal coordinate formulation for large defor-
mation analysis of multi-body systems. Journal of Computational and Nonlinear Dynamics, 8(3), pp.031016.
Grosenbaugh, M.A., 2007. Transient behavior of towed cable systems during ship turning maneuvers. Ocean engineering,
34(11), pp.1532-1542.
Huang, S., 1994. Dynamic analysis of three-dimensional marine cables. Ocean Engineering, 21(6), pp.587-605.
Jun, B.H., Park, J.Y., Lee, F.Y., Lee, P.M., Lee, C.M. and Oh, J.H., 2009. Development of the AUV ‘ISiMI’ and a free
running test in an Ocean Engineering Basin. Ocean engineering, 36(1), pp.2-14.
Kamman, J.W. and Nguyen, T.C., 1990. Modeling towed cable system dynamics (No. NCSC-TM-492-88). Panama City,
FL: Naval Coastal Systems Center.
Kamman, J.W. and Huston, R.L., 2001. Multi-body dynamics modeling of variable length cable systems. Multi-body
System Dynamics, 5(3), pp.211-221.
Kim, K.W., Lee, J.W. and Yoo, W.S., 2012. The motion and deformation rate of a flexible hose connected to a mother ship.
Journal of mechanical science and technology, 26(3), pp.703-710.
Park, H.I., Jung, D.H. and Koterayama, W., 2003. A numerical and experimental study on dynamics of a towed low tension
cable. Applied Ocean Research, 25(5), pp.289-299.
Shabana, A.A. and Yakoub, R.Y., 2001. Three-dimensional absolute nodal coordinate formulation for beam elements:
theory. Journal of Mechanical Design, 123(4), pp.606-613.
Shabana, A.A., Hussien, H.A. and Escalona, J.L., 1998. Application of the absolute nodal coordinate formulation to large
rotation and large deformation problems. Journal of Mechanical Design, 120(2), pp.188-195.
Seo, D.C., 2009. A study on the underwater glider design based on stability analysis and motion behavior. Ph.D. Disserta-
tion, Seoul National University.

Unauthenticated
Download Date | 9/24/15 11:42 PM
Int. J. Nav. Archit. Ocean Eng. (2015) 7:409~425 425

Takehara, S., Terumichi, Y. and Sogabe, K., 2011. Motion of a submerged tether subject to large deformations and dis-
placements. Journal of System Design and Dynamics, 5(2), pp.296-305.
Vaz, M.A. and Patel, M.H., 1995. Transient behaviour of towed marine cables in two dimensions. Applied Ocean Research,
17(3), pp.143-153.
Wu, J. and Chwang, A.T., 2001. Investigation on a two-part underwater manoeuvrable towed system. Ocean Engineering,
28(8), pp.1079-1096.
Yakoub, R.Y. and Shabana, A.A., 1999. Use of Cholesky coordinates and the absolute nodal coordinate formulation in the
computer simulation of flexible multi-body systems. Nonlinear Dynamics, 20(3), pp.267-282.
Yakoub, R.Y. and Shabana, A.A., 2001. Three dimensional absolute nodal coordinate formulation for beam elements: im-
plementation and applications. Journal of Mechanical Design, 123(4), pp.614-621.
Yuan, Z., Jin, L., Chi, W. and Tian, H., 2013. Finite difference method for solving the nonlinear dynamic equation of
underwater towed system. International Journal of Computational Methods, 11(4), pp.1350060

Unauthenticated
Download Date | 9/24/15 11:42 PM