You are on page 1of 7

Indian Journal of Geo Marine Sciences

Vol. 47 (01), January 2018, pp. 89-95

Impact of stern design on hydrodynamic drag of AUV's hull


Aymen Mohamed*1, Hedi Kchaou2, Med Salah Abid3 & Zied Driss4

Laboratory of Electro-Mechanic Systems (LASEM), National School of Engineers of Sfax (ENIS), University of Sfax, B.P.
1173, Road Soukra km 3.5, 3038 Sfax, TUNISIA
*[E-mail: aymenenis@gmail.com]

Received 09 November 2015 ; revised 30 October 2016

Identification of hydrodynamic parameters of the bare hull model is a paramount step in AUV design. CFD investigation using
ANSYS Fluent basing on K-ω SST turbulence model and a different mesh density is established for the velocity ranging from
0.4 m/s to 1.4 m/s. The numerical calculation of the drag coefficient with y+= 2 are revealed in good agreement with
experimental data from towing tank tests of Jagadeesh et al1 despite the low Reynolds regime. Thus, the comparison between
different afterbody models shows the significant effect of the stern design on the hydrodynamic drag of the vehicle and the onset
of flow separation around the stern part.

[Key words: AUV, CFD, hydrodynamic, Turbulence, stern, flow separation.]

Introduction to study the flow around MAYA AUV over


Recently, with the advancing higher angle of attack. They confirm that this last
development of computing performance and behaves well with flow separation and
numerical codes in prediction of fluid flow and reattachment in 3D complex turbulent flows;
pressure fields, computer based simulations using Dantas et al5 used CFD to study the influence of
Computational Fluid Dynamics (CFD) are control surfaces in maneuverability of an AUV.
susceptible to replicate conditions which are They were concluded that the occurrence of the
delicates through the experimentation. Many control surface stall depends on a linear
authors were benefited from this procedure to relationship between the control surface
collect much information about their experimental deflection and the angle of attack.
models. Jagadeesh et al1 used the low Reynolds The accuracy of CFD predictions is
turbulence models to estimate drag, lift and highly dependent on the quality and density of
moment coefficients for various velocities and meshing, settings of the TCM, thus required a
angles of attack. The numerical results show a validation through Experimental Fluid Dynamics
good agreement with measurements from towing (EFD) to ensure the reliability of the CFD model.
tank. Juong et al2 used CFD to optimize the By combining both computational and
design of an AUV hull. They obtained a experimental work, a validated simulation model
reasonable value of the nozzle angle; drag force could be obtained for the evaluation of the
and pressure and velocity fields. Malik et al3 used hydrodynamic characteristics of an AUV and
CFD to calculate the hydrodynamic features for a would be a cheaper, faster and viable approach
submersible AUV model in transient flow regime. compared to purely experimental work.
They were concluded that the CFD method is well In the current investigation, the total drag
capable and economical way to evaluate the coefficient of Afterbody 1 is predicted using
hydrodynamic derivatives of submersible numerical study for different operating speeds
platforms such as submarines, torpedoes and ranging from 0.4 m/s (Re = 105000) to 1.4 m/s
autonomous underwater vehicles. Sakthivel et al4 (Re = 367000) at the depth of submergence
used the standard model and non linear models d=4D.Also, the design of the stern is modified to
90 INDIAN J. MAR. SCI., VOL. 47, NO. 01, JANUARY 2018

characterize the addiction of this modification to


the stability of steering.
The simulation of the hydrodynamic
characteristics is basically performed with
Reynolds-averaged Navier Stokes (RANS)
equations. As the k-ω SST is substantially more
accurate than k-ɛ in the near wall layers, we
established a detailed study about the demeanor
of flow in exchange for inflation layers
dimensions. To validate the drag coefficient, the Fig.3−Modifications of Afterbody design
calculated drag coefficients are compared with
experimental results of Jagadeesh et al1. In this study, an incompressible, steady
and isotherm, Reynolds averaged Navier-Stokes
(RANS) model is applied to solve the (hull of
Materials and Methods AUV) problem on a translating reference frame.
The struts fixed to the hull displayed in The basic idea behind this reference frame is the
figure 1 for pushing the system has a significant assumption that it is the flow field which
effect in changes of flow stream behind the translates, and not the hull, thus means that an
model. We are emphasized the crucial necessity unsteady flow field translates into a steady flow
for the Suitable designs to get an appropriate with respect to the hull. This approach simplifies
experimental results. The main physical the problem in terms of boundary conditions, post
parameters defining the flow field are the processing results, and computational cost.
Reynolds number R e = ρU∇1/3 /μ and the depth
distance d= 4D; where ρ is density of water (1000 For a problem of flow simulation, the
kg/m3), U is the inlet velocity, and ∇ is volume of main control equations6 are:
the body (0.018 m3), and 𝜇 is the viscosity of
water (0.001 kg/m s). Equation of continuity
∇U = 0 (1)

Equations of motion (N-S Equation)


dU
ρ dt = ρg − ∇p + μ∇2 U (2)

Where U is the velocity vector, 𝜌 is the mass


density of water, p is the pressure, g is the
acceleration of gravity, and 𝜇 is the fluid dynamic
viscosity coefficient.

Fig.1−Experimental setup in the towing tank Turbulence closure model


The k-ω SST turbulence model is a two-
equation eddy-viscosity model )3 and 4
The myring6 design of experimental (improved by Menter7 to combine the robust and
model was described in figure 2 above. accurate formulation of the k-ω model in the near-
wall region with the free-stream independence of
the k-ɛ model in the far field:
∂ ∂
ρk + ρkui
∂t ∂xi
∂ ∂k
= Γk + Gk − Yk
Fig.2−Dimensions of experimental model )Afterbody1(
∂xj ∂xj
+ Sk 3
The modification in CAD model is ∂ ∂
ρω + ρωui
primarily on the rear part of the models. Two ∂t ∂xi
specified different designs are studied in this ∂ ∂ω
paper as shown in figure 3. = Γω + Gω − Yω + Dω
∂xj ∂xj
+ Sω 4
MOHAMED et al.: IMPACT OF STERN DESIGN ON HYDRODYNAMIC DRAG OF AUV'S HULL 91

Where, Gk represents the generation of turbulence Figure 5 shows the grid of the hull model
kinetic energy due to mean velocity gradients, Gω within the area based on the tetrahedral elements
represents the generation of ω, Γk and Γω constructed the viscous sub-surface to the surface
represent the effective diffusivity of k and ω, and tetrahedral one in outer sub layer.
respectively, Yk and Yω represent the dissipation
of k and ω due to turbulence, Dω represents the
cross-diffusion term, Sk and Sω are user-defined
source terms.
K-ω SST model can be used as a Low-Re
turbulence model without any extra damping
functions as k-ω formulation within boundary
layer makes the model usable all the way down to
the wall through the viscous sub-layer.
In Ansys Fluent, the study is carried out Fig.5−Grid surrounding Afterbody1
with velocity inlet for various velocities from 0.4
m/s to 1.4 m/s and outflow condition for outlet8. The accuracy of computational results is
The AUV surface with no slip wall and the other largely affected by meshing density. So, through
surfaces in computational domain with free slip the grid independence test, we try to get a proper
wall as shown in Figure 4. number of the grid which is consistent with
experimental facilities and CPU memory. Thus,
the thickness of the first layer within boundary
layers has an important effect in the calculated
results. For a specified y+, the first layer thickness
∆y 10 can be obtained using:
Fig.4−Computational domain and boundary conditions

Mesh generation ∆y = L∆y + 80 Re −13/14 (6)


The grid is generated with both structured The boundary layer thickness δL can be
11

and unstructured meshes with considering only estimated as:


the half of the bodies due to the flow symmetry.
As proven by Menter7, k-ω SST is a fully δL/L = 0.382/Re −13/14 7
turbulent model which effectively utilized for a
high Reynolds number flow. Therefore, meshing The grid index ratio is calculated as the
near the wall of model should be studied more ratio between the grid indexes for each case to the
accurately than far field due to the significant most refined case, normally the first. The increase
damping pulsation which reduces the Reynolds was defined as equal to 2 , as recommended by
number. So, to benefit of a good detection of flow Eca et al12. The property of the investigated grid
separation and shearing boundary layers features densities for the velocity of 0.4 m/s was described
in our TCM, we required a maximum thickness in table 1.
for the volume adjacent to the surface. Thus, y+ is
non dimensional wall distance which depends on Table 1− Mesh properties of the hull (v=0.4 m/s)
the choice of TCM and characterizes the local
Reynolds number. The first node near the wall y+ First cell Total number of
should be located in the viscous sub-layer8 with thickness ( h i ) elements
y+ closed as possible to 1, can be estimated with
the following relation: Case 1 0.5 0.136 3118792
ρ𝑦 𝑢∗
𝑦+ = 5 Case2 0.7 0.204 2980127
μ
Where μ is the local dynamic viscosity of the Case 3 1 0.272 2839733
fluid, ρ is the density of the fluid, y+ is non
dimensional mesh volume distance from the body Case 4 √2 0.383 2762985
τw
wall surface, u∗ = is the friction velocity, and Case 5 2 0.544 2691288
μ
τw is the shear stress on the body surface. For all
solution residues, we adapted a convergence
criterion in order of 10-4.
92 INDIAN J. MAR. SCI., VOL. 47, NO. 01, JANUARY 2018

Results and Discussion most adequate value of non dimensional wall


Meshing investigation is basically distance. Thus, the results of drag coefficient in
performed in terms of the thickness of the first this case is revealed in good agreement with the
inflation layer as k-ω SST is a full turbulent experimental results of Jagadeesh1 in despite of a
model which is characterized by a special marginal errors over the different level of
independency to boundary layers features. As velocity.
displayed in figure 6, y+= 2 is exhibited as the
5.2
y+ = 0.5 y+ = 0.7 y+ = 1
5
4.8 y+ = 1.41 y+ = 2 EXP
4.6
Cd ) 10¯² (

4.4
4.2
4
3.8
3.6
3.4
0.4 0.6 0.8 1 1.2 1.4
V ) m/s (
Fig.6−Relation between Cd and y+

Figure 7 illustrates the variation of drag increased. Thus, Afterbody2 presented the most
coefficient in terms of the increase in Reynolds adequate model with minimum drag coefficient
number for a fully submerged AUV model (H= notably observed after Re = 250000 when the
4d), as well the comparison between the present flow is more disturbed. Thus, the level of
numerical results with chosen designs and the turbulence of flow is a critical aspect for
referred experimental data of Jagadeesh et al1. developing evaluation of AUV conception.
The decrease of the drag coefficient of different
after body designs were revealed very clear as Re
5.1
4.9 Afterbody 3 (CFD) Afterbody 2( CFD)
Afterbody 1( CFD) Afterbody 1 (EXP)
4.7
4.5
Cd ) 10¯² (

4.3
4.1
3.9
3.7
3.5
104462 140986 177510 214035 250559 287083 323607 360132
Re
Fig.7−Relation between Cd and Re

For the different after body designs, constant. The onset of separation produces a small
pressure starts high (stagnation point) and drops reduction in pressure; then the low velocity of the
rapidly as the flow accelerates past the bow. flow current behind the stern part causes the
Then, it increases slightly to reach the level of significant increase in pressure coefficient as
pressure of the free-stream (zero pressure shown in figure 8.
coefficients) as the cross-sectional area remains
MOHAMED et al.: IMPACT OF STERN DESIGN ON HYDRODYNAMIC DRAG OF AUV'S HULL 93

Afterbody2 has the most significant specific design which is characterized by a


pressure coefficient at the tip of stern portion with regarding uniform cross section within this small
another stagnation point which presents portion of the stern. For this, pressure maintains a
confluence of flow streams. Also, we relative stability in this region before continuing
distinguished a little curvature in pressure its dropping in harmony with decreasing of cross
coefficient profile of Afterbody3 displayed with section.
green color in figure 8 for a distance between 1.36
m and 1.38 m from the bow tip. It is caused by a

Fig.8−Pressure coefficient vs position

The behavior of flow streams behind the disturbance which generated a sharp shearing of
stern part characterized with varied velocities inflation layers appears near the hull's wall due to
generated a turbulent mixing of flow currents in the developed cavitations. Whereas, Afterbody2
this region as shown in figure 9 with levels is shown more adaptable to the current of flow
between different designs. Also, the reducing of with a short wake and a weak rate of eddies
the velocity in this location increases the unlike to Afterbody3 which shows an acceptable
differential pressure thus producing the rate of wake in respect to the others conception.
phenomenon of cavitations. The design which promotes laminar flow
Afterbody1 seems to be the noisiest is the best as the level of skin friction is mainly
design with high level of eddy viscosity in depends to the behavior of the flow. As the cross
addition to longer boundary layers separation. section is increased gradually from the nose to
Thus, Afterbody3 is characterized with an early generate an adequate pressure gradient over the
separation of the boundary layers with the forward part of the hull, the flow was laminar.
observation of high velocity currents near the Otherwise, Afterbody3 highlighted the flexibility
hull's wall. However, Afterbody2 still is stable of its back form with boundary layers separation
with the minimum rate of disturbance. through a low skin friction decreased smoothly
Figure 10 shows the features of velocity compared to the other designs as shown in figure
vectors behind the stern as the main portion to 11. As the high shearing stress within boundary
study in this paper. Boundary layers separation at layers is resulted from the level of disturbance of
the trailing edge introduced low velocity flow, flow currents, Afterbody3 represents the best
caused by the surface with no-slip condition, efficient concept against flow disturbance.
thereby forming the wake zone. Afterbody1 has
the largest thickness of wake with a full level of
94 INDIAN J. MAR. SCI., VOL. 47, NO. 01, JANUARY 2018

Fig.9−Velocity fields for v = 0.4 m/s

Fig.11−Skin friction

Cavitations behind the AUV have


generated vortices which tend to increase the self-
noise of propulsion system. Afterbody2 represents
the suitable design with the least level of vortex
as shown in figure 12.Thus, the reduction of cross
sectional area along the length of the hull, causing
the flow to decelerate gradually and providing an
Fig.10−Velocity vectors around the sterns acceptable rate of flow currents disturbance.
MOHAMED et al.: IMPACT OF STERN DESIGN ON HYDRODYNAMIC DRAG OF AUV'S HULL 95

Fig.12−Vortices created behind Afterbody designs

In spite of the large amount of vortex behind 2. Juong, T., Sammut, K., He, F. and Lee, S.K. A., Study on
Afterbody3, we can clearly distinguish the area of the Design Optimization of an AUV by Using
Computational Fluid Dynamic Analysis, Proceedings of
relative stability in pressure with absence of the 19th International Offshore and Polar Engineering
disturbance in this location which confirmed the Conference, Osaka, 2009.
observation in the figure of the pressure 3. Malik, S.A., and Guang, P., Transient Numerical
coefficient curves. Simulation for Hydrodynamic Derivates Predictions of
an Axisymmetric Submersible Vehicle. Research
Journal Of Applied Sciences. Engineering Technology,
Conclusions 5(2013) 5003-5011.
In this paper, we are interested on the 4. Sakthivel, R., Vengadesan, S., and Bhattacharyya, S.K.,
prediction of the drag coefficient and the flow Application of non-linear κ – ε turbulence model in flow
behavior of the bare hull AUV considering simulation over underwater axisymmetric hull at higher
angle of attack. Journal of Naval Architecture and
various stern designs, which proves the necessity Marine Engineering, 8(2011), 149-163.
to revise the configuration of the struts in 5. Dantas, J.L.D., and Barros, E.A., Numerical Analysis of
experimental facility. Using k-ω SST TCM, the Control Surface Effects on AUV Maneuverability.
numerical results confirmed by experimental data Applied Ocean Research, 42(2013) 168-181.
6. Myring, D.F., A Theoretical Study of Body Drag in
from towing tank shows that Afterbody2 is the Subcritical Axisymmetric Flow. (Quartely). 1976, 186-
best model with the minimum rate of drag 194.
coefficient, vortices and thickness of wake. 7. White, F.M., Viscous Fluid Flow, )third edition
However, Afterbody3 is characterized by the University of Rhode, Island ( 2006, pp. 395-396.
region of a fairly stability which can be a suitable 8. Menter, F.R., Two-equation eddy-viscosity turbulence
models for engineering applications, AIAA Journal,
location to install the control surfaces in order to 3(2003) 346-351.
have a good maneuverability. 9. Praveen, P.C., and Krishnankutty, P., Study on the effect
of body length on the hydrodynamic performance of an
Acknowledgments axi-symmetric underwater vehicle. Indian J. Mar. Sci.,
8(1994) 1598-1605.
Authors are grateful to the National School of 10. Muhamad, H., Zahurin, S., and Mohd, R. A., CFD
Engineers of Sfax (ENIS( and in particular the simulation of cooperative AUV motion. Indian J. Mar.
Laboratory of Electro-Mechanic Systems Sci., 8(1994) 1598-1605.
(LASEM),Tunisia, for providing the CPU time 11. ANSYS. ANSYS FLUENT theory guide. )Canonsburg,
required for the current numerical analysis. PA: USA( 2009, pp. 114-115.
12. Eca, L., Vaz, G., and Hoekstra, M., A verification and
validation exercise for the flow over a backward facing
References step. Europe conference on computer fluid dynamics.
1. Jagadeesh, P., Murali, K., and Idichandy, V.G., Lisbon, 2010.
Experimental investigation of hydrodynamic force
coefficients over AUV hull form. Ocean. Eng. J,
36)2009(113-118.

You might also like