Review Article

Proc IMechE Part M:

J Engineering for the Maritime Environment
2021, Vol. 235(1) 15–29
A review on the hydrodynamic Ó IMechE 2020
Article reuse guidelines:
characteristics of autonomous sagepub.com/journals-permissions
DOI: 10.1177/1475090220936896

underwater vehicles journals.sagepub.com/home/pim

Jyoti Prakash Panda , Arindam Mitra and Hari V Warrior

Abstract
Autonomous underwater vehicles play an essential role in geophysical data collection, deep water mining, seafloor map-
ping, ocean exploration, and in many other related activities starting from military to scientific applications. A detailed
understanding of hydrodynamic characteristics will lead to better design, better control, and optimal path planning of
autonomous underwater vehicles in the deepest corner of oceans. This article will provide a detailed review of the
hydrodynamic characteristics of autonomous underwater vehicles, starting from different experimental techniques used
in the analysis of hydrodynamic parameters, methods used for fixing the autonomous underwater vehicles in towing
tank, instruments used for measurement of the hydrodynamic parameters. Furthermore, numerical methods employed
in performing computational analysis, hydrodynamics-based shape optimization, studies on drag reduction, and finally a
detailed list of turbulence models used in the computational fluid dynamics–related numerical simulations. The
hydrodynamics-based optimal shape of the autonomous underwater vehicles, the best technique to predict the hydrody-
namic parameters, and the best turbulence model for the computational fluid dynamics–based prediction of hydrody-
namic parameters will be recommended. At last, the hydrodynamic characteristics of different bio-inspired autonomous
underwater vehicles are discussed.

Keywords
autonomous underwater vehicles, hydrodynamic characteristics, experiments, computational fluid dynamics techniques,
turbulence models, drag reduction, bio-inspired autonomous underwater vehicles

Date received: 6 January 2020; accepted: 17 May 2020

Introduction by lithium-ion batteries5,6 since those have higher

energy density and can work for a more extended time.
Autonomous underwater vehicles operate in the ocean AUV can also acquire its power from fuel cells.7 Since
and can navigate into the hazardous environments in AUVs have to perform the uncrewed mission in differ-
the deepest ocean and are pre-programmed so that they ent regions of the ocean for several hours, the power
do not require any operator while working in the dee- requirements should be minimized to reduce the use of
pest corner of the ocean.1,2 They have their propulsion batteries. The significant portion of power loss is due to
system3 for navigation and a computer for decision- the resistance given by the water to the AUV. The
making. Autonomous underwater vehicles (AUVs) detailed understanding of the hydrodynamic character-
have sensors, through which they can collect several
istics will lead to optimized shape, better control, and
data, including temperature, depth level, chemical com-
navigation system of AUV.
position, and so on. They are usually deployed from
research vessels. AUV can take images of the seafloor
using very high-resolution cameras, which are useful in Department of Ocean Engineering and Naval Architecture, Indian
seafloor mapping for tidal energy plants and construc- Institute of Technology Kharagpur (IIT Kharagpur), Kharagpur, India
tion of underwater marine structures.4 AUV has versa-
tile applications in defense, industry, and policy sectors. Corresponding author:
Arindam Mitra, Department of Ocean Engineering and Naval
The oil and gas industries use AUV for geo-hazard Architecture, Indian Institute of Technology Kharagpur (IIT Kharagpur),
assessment activities. AUV can detect crack and dam- Kharagpur 721302, India.
ages in underwater structures. AUV is mostly powered Email: [email protected]
16 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

This article will primarily focus on the discussion of the stagnation pressure. The pressure coefficient cannot
hydrodynamic characteristics of AUV, starting from be more than one, because maximum pressure obtained
the definition of hydrodynamic parameters of AUV, is for the case of stagnation pressure only. It can be zero
experimental, and numerical techniques involved in the or negative, depending on the flow becomes cavitating
estimation of hydrodynamic parameters. A detailed list or gaseous. The pressure coefficient can be more than
of turbulence models used in computational fluid one exclusively for compressible flow. But, in an under-
dynamics (CFD)-based numerical simulations will be water vehicle, the flow is considered as incompressible.
provided, and hydrodynamics of bio-inspired AUV will It is an essential parameter in determining the hydrody-
be discussed. namic characteristic of AUV.
Most of the experimentalists working on AUV
hydrodynamics have mainly studied the first two para-
Hydrodynamic parameters of AUVs
meters (drag and lift coefficients) by varying the angle
The important hydrodynamic parameters of the AUV of attack and submergence depth.
are drag, lift, skin friction, and pressure coefficients.
Those can be defined as the following.
The drag coefficient is given by Experimental methods
Fd The experiments provide detailed insight into the accu-
Cd = ð1Þ rate flow dynamics past underwater vehicles. However,
0:5rU2 A
performing hydrodynamic experiments for every case
where Fd is the drag force acting on the AUV body, A of realistic flows, such as AUV operating at a depth of
is the frontal projected area. This force is working in 4000 m from the ocean surface or the AUV operating
the opposite direction of motion. This area should be near a moving submarine is practically impossible, since
reduced without compromising the other constraints, cost associated with such experiments is very high. It is
so that, drag Fd gets minimized. The lift coefficient can unsafe for a scientist to perform experiments in hazar-
be defined as dous environments. There are very limited experimental
Fl results available in literature in which the hydrody-
Cl = ð2Þ namic characteristics of AUV are studied experimen-
0:5rU2 A
tally or numerically in towing tank, wind tunnel, or
where Fl is the lift force acting on the AUV body, A is water tank. The towing tank can be defined as a long
the projected area toward the upward direction. A water channel over which a carriage is present. The car-
should be maximized to increase the lift coefficient. riage is moving over the tank in which the model is
Due to this reason, streamlined body shape is preferred fixed. The towing tank is shown in Figure 1(a) in which
for AUV body. an AUV is mounted. The different motions of AUV
The skin friction coefficient (Cf ) can be calculated as are shown in Figure 1(b). The earth fixed coordinate
tw system and AUV body-fixed coordinate systems are
Cf = ð3Þ also shown in the figure.
0:5rv2
In this section, three most standard experimental
where t w is the skin shear stress on the wetted surface works will be reviewed. The experimental techniques
of the AUV body, v is the free stream velocity, and the they have used, the method of production of AUV
rest of the symbols have their usual meaning. tw occurs structure, the materials used for the production of
due to the presence of viscosity in the fluid. It is the AUV, the hydrodynamic parameter measuring tech-
opposing force exerted by the fluid on a moving object. niques, and finally, the findings of those articles will be
Cf is usually expressed in terms of Reynolds number, presented. Very first experiments on AUV flow
which in turn, is the ratio of inertia force to viscous dynamics were performed in a wind tunnel.10 The
force. The smooth, streamlined body will provide less hydrodynamic forces on a low drag AUV were mea-
skin friction coefficient than rough surfaces. But some- sured. The AUV was manufactured in three sections,
times, in turbulent flow, uneven surfaces also provide namely the midbody, the tail section, and the fins. The
less Cf . It has been discussed more elaborately in Brzek midsection was manufactured using standard fiberglass
et al.8 techniques. The tail was produced in a turning machine
The pressure coefficient is defined as and finally finished in a computer numerical control
p  p‘ (CNC) milling machine. The fins are symmetrical aero-
Cp = ð4Þ foils with a NACA 66-006 section. The Reynolds num-
0:5rU2‘
bers in those experiments range from 1 to 5 million.
where p is the pressure at the point for which pressure The velocity of air was fixed at 44.6 m/s in all the
coefficient is calculated. p‘ and U‘ represents the pres- experiments in the wind tunnel. They have performed
sure and velocity at the freestream. Cp is different at experiments for both with and without fins cases. The
every point in the flow field. Cp is zero means the pres- drag of the AUV was found to be increased with an
sure is same as the freestream pressure. Unit Cp signifies increase in Reynolds number. For the case of AUV
Panda et al. 17

conducted at a towing tank of subsea R&D Center at

the Isfahan University of Technology. They have used
a standard method to fix the AUV hull in the towing
tank. They mainly have used two struts to fix the AUV
over the carriage. A picture of the detailed arrangement
is shown in Figure 1(a). It was observed that the drag
coefficient is more near the free surface because of the
wave-making resistance. However, with an increase in
depth, a relative decrease in drag was noticed.
Allston et al.12 performed experiments for prediction
of the wake behind a large AUV hydrofoil. They used
particle image velocimetry (PIV)12,13 technique for the
flow visualization past the AUV. They have compared
the visuals of PIV results with the contours of numeri-
cal results obtained through CFD14,15 simulations.
Javadi et al.16 have performed experimental studies on
the effect of bow profiles on the resistance of an AUV.
They considered two different bow shapes, tango and
the standard shape in free surface motion tests. In the
towing tank, the trolley was capable of travel at a speed
range of 0.05–6 m/s. They have used an approximate
formula for the calculation of resistance. From the
experiments, it was observed that the residual resistance
of the standard bow is more than the tango bow.
Manshadi et al.17 have studied the effect of the addition
of vortex generators on the hydrodynamic behavior of
an underwater vehicle. The results show that the cross-
Figure 1. Experimental setup and different motions of an AUV: flow separation reduces in the presence of vortex gen-
(a) towing tank reproduced from Mansoorzadeh and Javanmard9 erators, and subsequently, the drag decreases.
and (b) motions of AUV.

Numerical techniques
with fins, a higher magnitude of drag was observed at
all Reynolds numbers. With the increase in incidence The most common methods of numerical simulations
angles, an increase in drag of AUV was observed. An in marine hydrodynamics are based on potential flow
increase in the lift coefficient was also noticed with an theory, such as panel methods18 and the strip method.19
increase in incidence angle. The magnitude of the lift Both of these methods have been extensively used in
coefficient was more when the experiments were per- surface ships.
formed by making the fins on. Jagadeesh et al.11 per- In strip theory, the ship length is divided into a finite
formed experiments to measure the hydrodynamic number of transverse two-dimensional (2D) slices that
force coefficients of the AUV in towing tank-based are rigidly connected. It is applicable only in slender
experiments. The operating speed of the AUV was var- bodies that is length is much larger than beam or
ied in the range of 0.4–1.4 m/s. The pitch angle was draught. It uses a linear approach, and it works on the
varied from 0° to 15°. A multi-component strain gauge frequency domain. Several strip theories are discussed
type balance was used to measure the hydrodynamic in Beck et al.20 It is the fastest and most straightforward
forces and moments acting on the AUV. The drag, lift, approach to determine the motion in a ship. Milgram21
and pitching coefficient variations were calculated from has addressed the application of strip theory in under-
the measured results. The variation of those parameters water vehicle REMUS (Remote Environmental
was studied against different angles of attack and sub- Monitoring UnitS), including the effect of finite depth
mergence depths. The volumetric Reynolds number and lifting forces on fin surfaces. The disadvantage of
was in the range of 0.1–0.4 million. An increase in drag, this method is that it cannot be used in large-amplitude
lift, and pitching moment coefficients was observed wave motion or nonlinear phenomena. In the panel
with an increase in the angle of attack. A reduction of method, the integral equation is solved from Laplace
drag and lift coefficients was observed with an increase equation and Green’s identity. The different faces on
in Reynolds numbers. However, a reverse trend was the surface ship and the underwater vehicle is divided in
observed for the pitching moment coefficient. to several number of quadrilateral panels. The govern-
Mansoorzadeh and Javanmard9 have studied the ing equations are then solved on each panel to find the
effects of free surface on the drag and lift coefficients of flow parameters. Singularity in the panel method may
the AUV experimentally. The experiments were be a source, a doublets, or both. Singularity strength
18 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

may be of two types—constant (low order) and varying hydrodynamic parameters of an AUV. Salari and
(high order). High-order singularity strength will take Rava28 performed numerical simulations to calculate
more computational time because of more unknown. A the hydrodynamic coefficients of the AUV operating
low-order singularity panel method of an underwater near the free surface. They have considered the k  v
vehicle is discussed in Sahin et al.22 In the panel shear stress transport (SST) model for their numerical
method, viscosity of water is not considered. Those simulations. The experimental results of Jagadeesh
methods are useful for hydrodynamic analysis of prob- et al.11 were utilized for the validation of the numerical
lems involving negligible viscous effects. But, when vis- predictions. The numerical results of drag and lift coef-
cous effects come into existence, CFD techniques are ficients variations against the Froude number were pre-
used. In CFD, the governing mathematical equations sented. The Froude number can be defined as
are solved in the fluid domain for each control volume. Fr = U=(gL)0:5 . The drag coefficient was observed to
First, the fluid domain is discretized by the finite vol- be dependent on the immersion depth and speed. A
ume method, and then, different flow parameters are decrease in the drag coefficient was noticed with an
solved in each control volume.15 Flow domain uses increase in the Froude number. The drag coefficients
either in structured or unstructured meshing depending were observed to be more near the free surface because
upon the complexity of the flow. The details of different of the larger wave-making resistance. When the AUV
meshing methods are discussed in several litera- approaches the free surface, the flow field becomes
tures.14,15,23 Turbulence modeling is an integral part of unsymmetric about its horizontal mid-plan. Sakthivel
CFD because most of the practical flow problems are et al.29 analyzed the hydrodynamic performance of an
turbulent. There are different turbulence models AUV at higher angles of attack up to 20°, using a non-
develop till date to approximate the flow parameters in linear version of k  e turbulence model.30 Chen
a fluid flow. The simple mixing length model is suffi- et al.31 studied the motion stability of an autonomous
cient to capture the attached flow in almost every flow. underwater helicopter with a disk-shaped hull by CFD
But, to capture the flow variables in the separated simulations. The CFD results and Routh stability cri-
region, eddy viscosity models, Reynolds stress models terion31 were applied to check the motion stability of
(RSMs), large eddy simulation (LES), and direct the disk-shaped AUV. Few works on CFD application
numerical simulation (DNS) are developed for finding on AUV are discussed below. Zhang et al.32 have per-
out characteristics of different complex turbulent flow. formed CFD-based numerical simulations on the spir-
But, in comparison to RSM, LES, and DNS, one equa- aling motion of AUV and compared their simulation
tion (Spalart Almaras) and two-equation (k  e, k  v) results against experimental results of a spiraling AUV
eddy viscosity-based models takes less computational in the South China Sea, which was interacting with
time. Although these models are not accurate in captur- strong ocean currents. Isa et al.33 analyzed the hydro-
ing separated flows, they give consistent results with less dynamic coefficients of a hybrid underwater glider
computational time. Moreover, for very highly complex using CFD and strip theory19 techniques. The hybrid
flows, higher models are not feasible due to high com- glider combines the concept of both buoyancy-driven
putational requirements. The description of these tur- glider and traditional AUV. The glider was able to con-
bulence models are discussed by Wilcox24 elaborately. trol the rudder and wings independently. Dantas and
The design and numerical analysis of AUV mainly De Barros34 have studied the effects of control surface
involve CFD simulations,25 since those have complex deflection on the hydrodynamic performance of the
shapes and most of the AUV maneuver at lower AUV. For all the CFD-based numerical simulations,
speeds, for which viscous effects can not be neglected. they used the k  v turbulence model.
Performing experiments in the ocean is a challenging, Tang et al.35 have predicted the hydrodynamic coef-
time-consuming, and unsafe task. However, CFD- ficients of a complex shaped AUV TUNA-SAND
based numerical techniques are comparatively cost- (Terrain-based Underwater Navigable AUV for
effective and straightforward. In this section, a detailed Seafloor And Natural resources Development) by CFD
discussion of different numerical studies on the hydro- simulations which, has a complex block-like structure
dynamic characteristics of the AUV will be provided. (Figure 2) and was developed for seafloor and natural
Tyagi and Sen26 used CFD techniques to calculate resources development. The hydrodynamic coefficients
the transverse hydrodynamic damping force and are useful in the development of the control system of
moment coefficients of the AUV. They have used the the AUV. Song et al.36 performed numerical simula-
two-equation eddy viscosity models for the simulations. tions on flow past a Myring shaped AUV using k  v
The numerical simulations were performed for two dif- SST turbulence model. The resistance of the complex
ferent AUV hull shapes. The numerical predictions are hull was calculated using numerical simulations. De
validated against experimental results available in the Barros et al.37 have performed numerical simulations
literature. The CFD predictions of the coefficients were for the prediction of normal force and moment coeffi-
observed to be nonlinear with the transverse velocity. cients of an AUV at a nonlinear angle of attack and
Those coefficients are useful in the maneuverability sideslip range.
studies of underwater vehicles. Mostafapour et al.27 Reynolds stress models are used to calculate seven
analyzed the effect of Reynolds number on the equations throughout the discretized flow domain. So,
Panda et al. 19

numerical model predictions. From their numerical

results, it was observed that the submarine interaction
forces are minimum when the center of gravity of the
AUV is around the amidship of the submarine.
Oceanic flows are highly turbulent. Mitra et al.40
have studied the effects of free-stream turbulence on the
hydrodynamic characteristics of AUV. A series of
experiments and CFD simulations were conducted to
study the effects of free-stream turbulence on the evolu-
tion of the drag, skin friction, and pressure coefficients
of the AUV. It was observed that free stream turbulence
reduces the drag of the AUV. They also have provided
numerical results of hydrodynamic coefficients evalua-
tion for different angles of attack. Such outcomes on
the AUV hydrodynamics will be useful for the design
modification of AUV operating in complex flow fields.
Du et al.41 had studied the hydrodynamics of AUV
approaching sea-bottom. Sea beds have complex con-
figurations and result in complex flow fields. They have
performed CFD simulations by varying the depth of
operation of the AUV. The drag of the AUV observed
Figure 2. The complex shaped AUV TUNA-SAND.35
to be increased when AUV approaches the sea-bed.
Wu et al.42 had studied the hydrodynamic character-
this gives more accurate results than one-equation and istics of an AUV approaching a cone-shaped dock. The
two-equation turbulence models. But, it takes more interaction of the cone-shaped dock with the ocean cur-
computational time and challenging to apply for very rents results in complex flow fields, which has signifi-
complex flows. So, for most of the engineering prob- cant effects on the AUV hydrodynamics. Based on
lems, two-equation models are preferred. ANSYS FLUENT,23 numerical simulations were per-
formed based on the moving mesh methodology to
analyze the movement of the AUV. They noticed that,
The hydrodynamics of AUV operating in when AUV approaches the duck with a uniform speed,
complex flow fields the drag is less when the AUV approaches an unclosed
duck in comparison to a closed duck. When AUV faces
AUV often operates in complex flow fields like near
cross currents, it must change its heading using the rud-
moving submarines, near ships, or sea-beds with com-
der to avoid the sizable lateral drift. Wu et al.43 per-
plex topography.38 AUV works near the surface ship,
formed physics-based numerical simulations on AUV
another AUV or a submersible will experience hydrody- docking, using CFD-based multi-block hybrid dynamic
namic interaction due to the presence of different vehi- grid method. The method employs multi-block mesh
cles. The relative motion between AUV with another topology, moving mesh, and dynamic layer method to
nearby marine vehicle will affect the operation and grid the domain and move the rigid body. The velocity
objective of AUV. Furthermore, AUV operates in the of the AUV increases up to a certain extent in case of
deep ocean come across several underwater structures the constant revolution per minute (RPM) docking,
or topographical undulations. Due to this reason, the and the resistance of the AUV increases when it passes
data collection operation of AUV through different through the neck point of the duck.
sensors gets hampered. The hindrance from other Rattanasiri et al.44 performed numerical simulations
objects in the operating field of AUV should be mini- on a pair of self-propelled AUV operating in tandem.
mized to get accurate data from AUV. In this section, Their principal purpose was to find the optimal dis-
we will discuss available literature related to the hydro- tance between two vehicles for safer operations in mul-
dynamics of AUV in complex flow fields. tiple vehicles’ missions. It was noticed that the flow
Leong et al.39 have studied the hydrodynamic inter- field of one AUV has a significant effect on the flow
actions of a submarine on an autonomous underwater field of the other AUV, results in drag enhancement. A
vehicle. The hydrodynamic interactions of submarines propeller race deduction was proposed to accommo-
can result in an uncontrolled motion of AUV. The pri- date the increase in vehicle drag due to the propulsors
mary purpose of their research was to find the regions of other vehicles.
across the submarine where an AUV can operate safely
with minimal hydrodynamic interaction effects. The
Hydrodynamics-based shape optimization
CFD simulations were performed at different positions
of AUV. The experimental results of the small-diameter The AUV should be designed in such a manner that it
ratio of the two vehicles were used to validate the must enhance energy utilization and produce minimum
20 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

drag while operating in the ocean. CFD can be consid-

ered as the best tool for shape optimization of under-
water vehicles operating in complex hydrodynamic
environments. The traditional AUV has an elliptical
frontal shape, followed by a cylindrical midsection and
a conical back portion. The form of an AUV should be
optimized in such a way it provides low drag, high pro-
peller efficiency, low power consumption, and it can
operate for a long duration.
Joung et al.45 have done an optimization of the noz-
zle angle at different velocities using ANSYS for mini-
mum drag. They have used the design of the experiment
and the direct search method of ANSYS Design
explorer in their investigation. Joung et al.46,47 have
done a design optimization of a ducted propeller AUV.
Three different design optimization parameters chosen Figure 3. The different shapes of frontal, back, and mid
by them are the location of sail, separation of sail and sections of the AUV considered in Gao et al.49
transducer, and the angles of attack of the nozzle. Won
et al.48 have used an external and internal shape optimi-
zation techniques for a duct type AUV. They used the propeller was fixed behind the hull. They observed that
Myring equation and Venturi effect in their shape opti- for a high angle of attack, more thrust was generated
mization. The Taguchi method have been utilized by by the propeller, but more torque is required for the
them to minimize the no. of CFD simulations. Gao rotation of the propeller. The unsteady flow across the
et al.49 have performed CFD simulations to optimize propeller at higher angles of attack leads to the vibra-
the hull shape of AUV for maximum energy utilization. tion of the AUV hull.
They varied the frontal and back portion of the AUV Chen et al.53 performed numerical simulations on
in different combinations from a collection of shapes, the AUV hydrodynamics fitted with a foldable propel-
as shown in Figure 3. Although they didn’t recommend ler. The propellers need to be fully closed when the
a single optimized configuration for AUV, they sug- AUV is working in buoyancy-driven mode and
gested for reducing the length of the cylindrical section becomes fully open in other cruising modes. The fold-
for the drag reduction. able propeller can perform such activities and are use-
The choice of turbulence model and difficulty in ful for hybrid-driven underwater vehicles. They mainly
meshing complex structure is a critical criterion in have calculated the hydrodynamic coefficient of the
determining the optimum shape of an AUV through underwater vehicle during the folding and unfolding of
CFD. It may lead to a significant problem in optimiz- the propeller.
ing the drag of an AUV. Appropriate choice of turbu- De Barros and Dantas54 conducted CFD simula-
lence model is vital in this respect. tions on the maneuverability of AUV in the presence of
a ducted propeller. They validated the numerical model
predictions against the experimental results obtained
Hydrodynamics of AUV with appendages
from towing tank experiments. They observed that with
Most of the researchers have performed numerical and the increase in angles of attack, the normal force and
experimental studies on the hydrodynamic characteris- pitching moment coefficient increases both for the cases
tics of AUV without considering the accessories such as of bare hull and hull with ducted propellers. It was also
fins and propellers. In this section, the AUV hydrody- noticed that the duct increases the normal force coeffi-
namics will be reviewed in the presence of propeller and cient and decreases the pitching moment coefficient.
fins. Mainly the effect of the addition of those acces- Du et al.55 performed simulations of an AUV with the
sories on the hydrodynamic performance of AUV will deflectable ducted propeller. The combination of AUV
be addressed. Liu et al.50 studied hydrodynamics of an with deflectable ducted propeller results in the vector
AUV with different layouts of wings. Zhang et al.51 propulsion system. The propulsion system helps the
performed CFD simulations of an underwater vehicle. AUV to perform various motions. In their seminal
Mainly the resistance performance and movement sta- work, they proposed a CFD model to estimate the per-
bility at deep-sea were simulated using Reynolds aver- formance of the delectable propeller. They performed
aged Navier–Stokes (RANS) simulations. They have simulations longitudinal and lateral 2D motions and
performed simulations with four screw propellers. The three-dimensional (3D) space motions.
resistance performance of the underwater vehicle with a Javaid et al.56 have studied the effects of wing form
propeller for both horizontal and vertical navigation is on the hydrodynamic characteristics and dynamic sta-
analyzed. Hayati et al.52 studied the hydrodynamic per- bility of an underwater vehicle. They mainly have con-
formance of an AUV fixed with propellers. The sidered two different wing types, in their numerical
Panda et al. 21

simulations, those are of rectangular and tapered types. Turbulence models for hydrodynamic
From the analysis of numerical results, it was observed analysis of AUV
that the rectangular wings have more drag in compari-
son to the tapered wings. The drag force of the rectan- Turbulence models24,64,65 play major role in CFD anal-
gular wing was 12% more than the drag force of ysis. The main turbulence modeling techniques in CFD
tapered wings. However, they observed that the AUV are eddy viscosity models, Reynolds stress models,
with rectangular wings are comparatively more stable large-eddy simulations and direct numerical simula-
dynamically since the wetted area of the rectangular tions. Must of researchers working on AUV hydrody-
wing is more than that of the tapered wing, which gen- namics mainly used the two equation-based turbulence
erates a larger lift force. models (k  e and k  v). These models are quite sim-
ple. The computational cost is comparatively less in
comparison to other sophisticated models. However,
Methods of drag reduction of AUV these models may not accurately capture the flow phy-
sics, since those rely on so many assumptions and sim-
The most basic method of reduction is by streamlining plifications. The main assumption for the failure of
the shape of the AUV. Meng et al.57 studied hydrody- such a model is the turbulent viscosity hypothesis, in
namic characteristics of an AUV by adding porous which the turbulent stress field was empirically corre-
materials over its surface. They used two different lated with the strain field of the flow. However, the
AUV models to test the effect of the addition of porous recent emphasis is shifted toward Reynolds stress mod-
material on their hydrodynamic performance. The two els and large eddy simulations. Few of the researchers
different models are REMUS100 and SUB OFF sub- have started working on former models for the hydro-
marine models. By CFD methodologies, they have dynamic analysis of AUV. The Reynolds stress models
incorporated a porous viscosity coefficient over model do not have any empirical relations and directly solve
surfaces. For REMUS100 model, up-to porous viscos- the equations for the Reynolds stress components in
ity coefficient 830 kg=m3 s, the drag of the AUV the flow field to compute the detailed flow field. In the
decreases. However, after a porous viscosity coefficient present scenario, with an increase in computational
830 kg=m3 s, the drag of the REMUS AUV increases. A facilities, researchers can use those models for predic-
similar trend was observed for the SUBOFF submarine tion 3D flow past AUV. Mitra et al.40 have used such
model. The critical porous viscosity coefficient was models to analyze the hydrodynamic parameters of
observed to be 500 kg=m3 s. AUV in the presence of free-stream turbulence and
Saghafi and Lavimi58 have tried to modify the shape over sea-beds with complex topography. Very limited
of AUV for drag reduction. They used the realizable literature is available on the analysis of hydrodynamic
k  e model for the studies on the shape modifications. parameters of AUV using large-eddy simulations.66
They have recommended the best configuration for a LES is very accurate in comparison to the former two
minimal drag of the AUV. So, shape optimization can models. However, one cannot use such a sophisticated
be considered as an alternative method for the drag model, for the flow analysis with complex physics and
reduction of AUV. with a larger fluid domain since the cost associated
Wu et al.59 performed numerical simulations of the with such simulations is very high. However, with LES,
micro-bubble flow around an underwater vehicle the accurate flow pattern in the wake region of AUV
model. The induced micro-bubbles decreases the den- can be obtained.
sity and turbulent viscosity of the fluid adjacent to the In this section, a detailed description of the turbu-
AUV hull, which minimizes the production and dissipa- lence models used will be provided. In addition to the
tion of turbulence kinetic energy along the AUV hull. Reynolds stress model, two more eddy viscosity-based
By utilizing the methodology mentioned above, they models were used to compare the experimental results
have reduced the drag of the underwater vehicle by of hydrodynamic parameters along with AUV of differ-
50%. Anbarsooz60 studied the drag reduction of AUV ent shapes. Those models are k  e, k  v SST. The
with hydrophobic surfaces. They used micron-sized main goal of this work is to show that Reynolds stress
spherical particles over the hull for drag reduction. models has the highest potential to capture the turbu-
Shereena et al.61 have utilized a novel method for lent flow field along the AUV of different shapes in
the drag reduction of the Afterbody1, an AUV model contrast to the eddy viscosity based.
designed by Huang et al.62 They have injected air jet in
the boundary layer of the AUV for the decrease in Standard k   model
drag. The mixture model available in ANSYS63 was
used in conjunction with the k  v SST model for the The k  e model67,68 is the simplest and most widely
simulation of multiphase flow. They have considered used model in industrial and AUV design applications.
two different smooth and tapered shapes of the AUV The cost of computation of k  e model is compara-
stern profiles for their analysis. They achieved a drag tively less than Reynolds stress models, since there are
reduction of 44.8% with the injection of air-jet. only two transport equations to be solved one for the
22 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

turbulence kinetic energy and other for the turbulence the modeled equations for the transport equations of
dissipation rate, which allows to independently deter- the Reynolds stress components.75–77 The Reynolds
mine the turbulence velocity
p ffiffiffi and length scale. The velo- stress transport equation has four terms in its right side,
2
city is calculated
p ffiffiffiffiffi as k and the length scale is those are production, transport, dissipation, and the
2
calculated as k3 =e. The turbulent viscosity was calcu- pressure strain correlation term. The pressure strain
lated using the following formula correlation term mainly accounts for the complex flow
features resulting from the stream line curvature and
k2 flow separation. The Reynolds stress model are much
mt = rCm ð5Þ
e more reliable and accurate than its two-equation coun-
The modeled transport equations used in their for- ter parts. The Reynolds stress transport equation can
mulation can be written as be written as74
   ∂ui uj ∂Tijk
∂ ∂ mt ∂k ∂ t ui uj + U k = Pij   hij + fij ,
∂t rk + (rui k) = m+ + Gk  re, ∂xk ∂xk
∂xi ∂xj sk ∂xj
   where
∂ ∂ m ∂e
∂t re + (rui e) = m+ t ∂Ui ∂Uj
∂xi ∂xj sk ∂xj Pij =  uk uj  ui uk ,
∂xk ∂xk
e e2 ∂ui uj p p
+ C1e Gk  C2 r ð6Þ Tkij = ui uj uk  n + djk ui + dik uj , ð8Þ
k k ∂xk r r
In the first equation, Gk represents the generation of ∂ui ∂uj
turbulence kinetic energy because of mean velocity gra- hij =  2n
∂xk ∂xk
dients and can be written as a function of Reynolds  
p ∂ui ∂uj
stress as follows fij = +
r ∂xj ∂xi
Gk =  rui uj ∂xi uj ð7Þ
Pij is turbulence production, Tijk is diffusive transport,
The model coefficients were calculated by comparing hij is dissipation rate tensor and fij is pressure strain
with experimental datasets of decaying grid turbulence correlation. The Poisson equation for pressure fluctua-
and shear flows in boundary layers, mixing layers, and tions can be written as
jets. A more detailed description of the model is avail-
able in Launder and Spalding.68 1 2 ∂Uj ∂ui ∂2 (ui uj  ui uj )
r (p) =  2  ð9Þ
r ∂xi ∂xj ∂xi ∂xj
SST k  v model The fluctuating pressure term can be decomposed
24 into two terms: slow and rapid pressure term
The k  v model utilizes the transport equations for
p = pS + pR . The terms for the slow and rapid fluctua-
turbulence kinetic energy and the specific dissipation rate.
tions satisfy the following equations
The SST k  v model of Menter69 and Rattanasiri
et al.70 employs the k  v and k  e model in the near 1 2 S ∂2
wall region and far field, respectively. The k  e model r (p ) =  (ui uj  ui uj ) ð10Þ
r ∂xi ∂xj
is converted into the k  v model with following refine-
ments to the standard k  v model:24 1 2 R ∂Uj ∂ui
r (p ) =  2 ð11Þ
r ∂xi ∂xj
1. A blending function is multiplied to both standard The slow pressure term corresponds to the
k  v model and the transformed k  e model. turbulence–turbulence interactions and the rapid term
2. A damped cross diffusion derivative term is incor- accommodates the interaction of mean and fluctuating
porated in the v equation. strain.78–81
3. A modified version of turbulent viscosity is formu- The rapid pressure strain correlation term can be
lated to consider for the transport of turbulent modeled in terms of a fourth rank tensor82
shear stress.
4. The model constants are calibrated against experi- ∂Ul
fR
ij = 4k (Mkjil + Mikjl ) ð12Þ
mental results of flat plate boundary layer, adverse ∂xk
pressure gradient flows, backward facing step flow
and NACA 4412 airfoil flows. where
ð
1 1 ∂2 Rij (r)
Mijpq = dr ð13Þ
Reynolds stress model 8pk r ∂rp ∂rp
Reynolds stress models71–74 do not rely upon any ad where Rij (r) = hui (x)uj (x + r)i.
hoc definition of turbulent stresses in terms of strain The complete pressure strain correlation has the
field, rather the stress field is directly computed from from
Panda et al. 23

∂vk used LES methodologies to capture the turbulence

fij = eAij (b) + kMijkl (b) ð14Þ
∂xl boundary layers over the body of the underwater vehi-
cle. They observed that the wake of the underwater
The most standard form of slow pressure strain cor-
vehicle is mostly affected by the shear layer from the
relation can be written as
trailing edge of the fins and the turbulent boundary
 
1 layer growing along the stern. Posa and Balaras91 have
fSij = b1 bij + b2 bik bkj  IIb dij ð15Þ performed LES-based simulations on the effects of
3
Reynolds number on the flow around an appended
Established slow pressure strain correlation models underwater vehicle. At the highest Reynolds number,
including the models of Panda et al.,73 Rotta,83 Sarkar they observed a weaker peak of turbulence kinetic
and Speziale,84 and Warrior et al.85 use this general energy in the outer layer over the stern.
expression.
The linear form of the expression for the rapid pres-
sure strain correlation can be written as Contrasting turbulence models prediction
  of hydrodynamic parameters of AUV
fRij 2
= C2 Sij + C3 bik Sjk + bjk Sik  bmn Smn dij Here, we will consider the k  e, k  v, and a Reynolds
k 3
stress model to analyze the hydrodynamic parameters
+ C4 (bik Wjk + bjk Wik ) ð16Þ
of three different AUVs and will contrast those model
Here bij = (ui uj =2k)  (dij =3). The rest of the terms in predictions against experimental results available in the
the formulation for the rapid pressure strain correlation literature and finally the best turbulence model in terms
has their usual meaning. Rapid pressure strain correla- of cost and accuracy will be recommended.
tion models like the models of Panda and Warrior,72 Although few of the researchers have performed
Mishra and Girimaji,81 and Speziale et al.86 use this numerical simulations with large-eddy simulations, the
general expression. For this work, the Reynolds stress technique requires high-performance computing, and
model of Gibson and Launder87 is used. The detailed the computational cost of such simulations very high,
description of the modeling parameters and the model which may not be useful in practical applications.
constants are available in Gibson and Launder.87 However, with LES, the flow field can be visualized
more accurately. Here, the focus will be given on the
contrast of RANS model prediction of hydrodynamic
Large eddy simulation parameters of AUV against experimental results avail-
Large eddy simulation methodology involves solving of able in the literature.
a filtered Navier–Stokes equations88 Three axisymmetric underwater bodies were consid-
ered for our simulations, as shown in Figure 4. Those
∂i u~i were named as AUV111 and AUV2,10 respectively. The
= 0,
∂i xi length and diameter of AUV1 are 1.4 and 0.14 m,
ð17Þ
∂i u~i ∂i u~i u~j ∂P~ ∂t ij 1 ∂2 u~i respectively. More information on the geometric
+ =  + + fi characteristics of AUV1 is available in Jagadeesh
∂t ∂xj ∂xi ∂xj Re ∂xj xj
et al.11 The AUV2 has three sections fore-body, tail,
In the above equation, the filtered velocity and pres- and fins. The fin sections are not shown in Figure 4.
sure are denoted as u~ and p~, respectively. Re is the The AUV2 has a maximum diameter of 0.666 m and
Reynolds number and fi is forcing term. The x and t length 3 m.10
denote the spatial and time coordinates. A subgrid-scale The AUV1 was tested in a towing tank. AUV1 and
(SGS) tensor (t ij = uf
iuj  u
~i u~j ) was introduced to differ- AUV2 were tested on water and wind tunnel,
entiate scales of turbulence.88 Although different SGS
eddy viscosity models are available in the literature, for
use in LES studies, the most standard one is the wall-
adapting local eddy-viscosity (WALE) model,89 since it
has a lower computational cost.
Da Silva Costa et al.66 performed LES-based numer-
ical studies to analyze the influence of stern shape on
the wake formation pat the AUV. They have consid-
ered a fishtail shaped AUV for their simulations. Da
Silva Costa et al.90 studied the hydrodynamics of flat
fish-shaped AUV using LES. Posa and Balaras88 have
studied the hydrodynamics of an underwater vehicle
model using LES. In their simulations, they have con-
sidered all the appendages. They performed numerical Figure 4. Diagram of the layout of the autonomous
simulations for two different Reynolds numbers. They underwater vehicles (a) AUV1 and (b) AUV 2, not to scale.
24 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

stress model prediction of volumetric drag matches well

with the experimental results and comparatively better
than the model predictions of k  e, k  v SST models.
The experimental data with and without fins on
AUV2 were collected from Huggins and Packwood10
at different Reynolds numbers and angles of attack.
The experiments of Huggins and Packwood10 were
conducted in a low-speed wind tunnel. The maximum
achievable speed of the wind tunnel was 100 m/s.
Figure 6 shows the variation of the drag coefficient
with Reynolds number for the fin off condition. From
both the figures, it is observed that Reynolds stress
model predictions are better than the prediction of
eddy viscosity models.

Bio-inspired AUV
A human being has an extraordinary brain, can see
Figure 5. The effect of Reynolds number on coefficient of drag. with his both eyes in one direction (single vision), the
Dashed line k  e model, dashed dot line k  v shear stress transport turtles have rigid bodies, and the fishes can swim effec-
(SST), and solid line represents Reynolds stress model predictions. tively without any propeller. Different fishes have dif-
ferent swimming methods and have different modes of
movement of fins.92 The box-fish propels by oscillating
a tail fin without affecting the body. For propulsion,
pitching, and maneuverability, different fishes employ
different strategies.93 Few of the fishes swim using the
pectoral fins alone. The pectoral fins help the fish to
maintain their stability. Fishes can control their buoy-
ancy in water by controlling their volume; by this, they
can travel up and down. Similarly, turtles have many
unique features in their body design and their modes of
propulsion. They have five longitudinal ridges on their
carapace. These ridges are misaligned slightly to the
streamline around the body for generating vortices in
the stream-wise direction, which enables it to delay or
suppress flow separation on the framework, which
enhances the hydrodynamics performance at different
swimming modes.94 Thrust in the turtles is generated
by movements of forelimbs and hindlimbs.95
AUV must have minimum drag so that the powering
system of the AUV can work more efficiently and can
cruise and collect data from deeper oceans for a longer
Figure 6. The effect of Reynolds number on coefficient of drag period of time. Although several researchers have pro-
for different Reynolds numbers. vided different streamlined shapes for better operation
Dashed line k  e model, dashed dot line k  v shear stress transport of AUV, the recent emphasis of the research commu-
(SST), and solid line represents Reynolds stress model predictions. nity has been shifted toward bio-inspired AUV96 and
bio-inspired propulsions.97 The researchers are making
AUV shapes like different fishes.98–101 Few of the
respectively. For AUV1, the experimental results of researchers also made their AUV design like turtles102–
volumetric drag coefficients from Jagadeesh et al.11 104
to achieve better hydrodynamic performance. AUV
were considered at different Reynolds numbers ranging can have a shape like dolphins. Numerical studies on
from 1:053105 to 13:673105 . For simulations the inlet the flow past a modeled dolphin are presented in
velocity was taken as 0.4, 0.6, 0.8, 1, 1.2, and 1.4 m/s Pavlov and Rashad.105
respectively. The experimental datasets at different Honaryar and Ghiasi106 have made the AUV design
angles of attack were also included for comparison. like a catfish, as shown in Figure 7. The left and right
Three sets of simulations for each turbulence model figures show top, side, and bottom views of the actual
were performed at a particular speed and angle of fish and the model AUV, respectively. The hydrody-
attack. From Figure 5, it is observed that Reynolds namic damping derivatives were calculated using
Panda et al. 25

Figure 9. A bio-inspired hull shape (like a turtle) of AUV.102

106
Figure 7. A bio-inspired hull shape (like a starfish) of AUV.
consumption, navigation, sensor, and communication
related to AUV operation have been solved a little bit.
Full autonomy can be solved using the intelligence sys-
tem. Several battery technologies have been developed
for efficient use in AUV. Cost-effective acoustic trans-
ponder systems do navigation. Embedded sensors are
used to overcome decision making in an emergency.
The acoustic communication system is used instead of
laser or RF communication system due to several prob-
lems such as small availability of bandwidth in water,
strong signal attenuation in water, uncertainty in pro-
pagation channel, and so on. Acoustic communication
has a major disadvantage of a low data rate. Still, there
are many unsolved problems related to AUV, such as
dynamic obstacle avoidance, switching of the forma-
tion structure depending on the situation, and so on,
Figure 8. A bio-inspired hull shape (like a boxfish) of AUV.92 which require attention in the future. The use of the
hybrid controller in future AUVs will become a neces-
numerical simulations. They compared the numerical sity. Furthermore, hydrodynamic interaction of AUV
results of the newly developed catfish model with the operation in complex flow fields is required to be taken
DARPA SUB OFF model.107 The hydrodynamic stabi- care of due to proper navigation and communication
lity of the newly proposed bio-inspired AUV was found of data to the observer. Fuel cell or the solar cell can be
to be 99% more than the conventional model. used as a source of energy in AUV for long-range oper-
Li et al.108 performed a hydrodynamic analysis of an ation. The shape of the AUV can also be optimized for
AUV with bio-inspired wings. The wings were capable minimum drag, maximum thrust, and highest propeller
of undergoing traveling wave motion. The effect of the efficiency according to the operating condition and
hydro-wing parameters on the propulsive efficiency of objective. Moreover, the use of data driven algorithm
the underwater vehicle was studied. They used an (machine learning/deep learning–based) for the predic-
immersed boundary method to capture the moving tion of hydrodynamic performance of AUVs will be of
boundary of the hydro wings. Kadiyam and Mohan92 severe importance in future.
designed a bio-inspired propulsion system for the hori-
zontal motion of an AUV. They made the hull shape
like a boxfish as shown in Figure 8. The U-CAT AUV Concluding remarks
presented by Allotta et al.109 shown in Figure 9.
In this article, we provide a thorough review of the
hydrodynamic characteristics of AUV. The study has
Current state and future perspective
been from the experimental techniques used in the anal-
Many researchers build several AUVs till date for ful- ysis of hydrodynamic parameters, the methods used in
filling different objective functions under the deep fixing the AUV in the towing tank, methods used in
ocean. The development of AUV technology, starting the production of AUV hull, and finally, the techniques
from the exploration of potential AUVs, testing with involved in experimental measurements.
the prototype to the development of cooperative Established numerical methods for the analysis were
motion control technology, has been developed for the introduced. The hydrodynamically optimized shape
past five decades. The challenges in autonomy, energy was recommended.
26 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

Finally, the CFD techniques for the flow prediction computational and experimental fluid dynamics meth-
along the AUV are presented. The different commer- ods. J Fluid Struct 2014; 51: 161–171.
cially available CFD solvers are introduced, and the 10. Huggins A and Packwood A. Wind tunnel experiments
turbulence models available in commercial CFD sol- on a fully appended laminar flow submersible for ocea-
vers are presented. Based on numerical analysis and nographic survey. Ocean Eng 1995; 22(2): 207–221.
11. Jagadeesh P, Murali K and Idichandy V. Experimental
contrasting the model predictions with experimental
investigation of hydrodynamic force coefficients over
results of hydrodynamic characteristics, the best turbu-
AUV hull form. Ocean Eng 2009; 36(1): 113–118.
lence model for the flow analysis past AUV was recom- 12. Allston T, Munroe J, Lewis R, et al. Predicting the wake
mended. The hydrodynamics of bio-inspired AUV behind a large AUV hydrofoil. Meth Oceanogr 2014; 10:
were also discussed in detail. 166–177.
13. Adrian L, Adrian RJ and Westerweel J. Particle image
Declaration of conflicting interests velocimetry, vol. 30. Cambridge: Cambridge University
The author(s) declared no potential conflicts of interest Press, 2011.
with respect to the research, authorship, and/or publi- 14. Anderson JD and Wendt J. Computational fluid dynamics,
vol. 206. Berlin: Springer, 1995.
cation of this article.
15. Versteeg HK and Malalasekera W. An introduction to
computational fluid dynamics: the finite volume method.
Funding New York: Pearson Education, 2007.
16. Javadi M, Manshadi MD, Kheradmand S, et al. Experi-
The author(s) received no financial support for the
mental investigation of the effect of bow profiles on resis-
research, authorship, and/or publication of this article.
tance of an underwater vehicle in free surface motion. J
Mar Sci Appl 2015; 14(1): 53–60.
ORCID iD
17. Manshadi MD, Hejranfar K and Farajollahi AH. Effect
Jyoti Prakash Panda https://orcid.org/0000-0003- of vortex generators on hydrodynamic behavior of an
2839-6185 underwater axisymmetric hull at high angles of attack. J
Arindam Mitra https://orcid.org/0000-0002-6216- Visualiz 2017; 20(3): 559–579.
5762 18. Newman JN. Panel methods in marine hydrodynamics.
In: Proceedings of the conference eleventh Australasian
fluid mechanics, Hobart, TAS, Australia, 14–18 December
References 1992. Hobart, TAS, Australia: University of Tasmania.
1. Sahoo A, Dwivedy SK and Robi P. Advancements in the 19. Newman JN. Marine hydrodynamics. Cambridge, MA:
field of autonomous underwater vehicle. Ocean Eng MIT Press, 2018.
2019; 181: 145–160. 20. Beck RF, Cummins WE, Dalzel JF, et al. Motions in
2. Wynn RB, Huvenne VA, Le Bas TP, et al. Autonomous waves. In: Lewis EV (ed.) Principles of naval architecture,
underwater vehicles (AUVs): their past, present and vol. 3. Jersey City, NJ: The Society of Naval Architects
future contributions to the advancement of marine and Marine Engineers, 1989, pp.1–190.
geoscience. Mar Geol 2014; 352: 451–468. 21. Milgram JH. Strip theory for underwater vehicles in
3. Allotta B, Pugi L, Bartolini F, et al. Preliminary design water of finite depth. J Eng Math 2007; 58(1–4): 31–50.
and fast prototyping of an autonomous underwater vehi- 22. Sahin I, Crane JW and Watson KP. Application of a
cle propulsion system. Proc IMechE, Part M: J Engineer- panel method to hydrodynamics of underwater vehicles.
ing for the Maritime Environment 2015; 229(3): 248–272. Ocean Eng 1997; 24(6): 501–512.
4. Shome S, Nandy S, Pal D, et al. Development of modu- 23. ANSYS. ANSYS FLUENT 14.0 user’s guide. Canons-
lar shallow water AUV: issues & trial results. J Inst Eng burg, PA: ANSYS, 2011.
India Ser C 2012; 93(3): 217–228. 24. Wilcox DC. Turbulence modeling for CFD, vol. 2. La
5. Raman N, Briscoe JD and Grivel T. Lithium-ion bat- Canada, CA: DCW Industries, 1998.
teries for autonomous underwater vehicles. In: Proceed- 25. Hu ZQ. Numerical calculation methods for hydrodynamics
ings of the 2002 workshop on autonomous underwater of unmanned marine vehicles and their application. PhD
vehicles, San Antonio, TX, 21 June 2002, pp.45–49. New Dissertation, Shenyang Institute of Automation, Univer-
York: IEEE. sity of Chinese Academy of Sciences, Beijing, China,
6. Bradley AM, Feezor MD, Singh H, et al. Power systems 2013.
for autonomous underwater vehicles. IEEE J Ocean Eng 26. Tyagi A and Sen D. Calculation of transverse hydrody-
2001; 26(4): 526–538. namic coefficients using computational fluid dynamic
7. D’Amore-Domenech R, Raso MA, Villalba-Herreros A, approach. Ocean Eng 2006; 33(5–6): 798–809.
et al. Autonomous underwater vehicles powered by fuel 27. Mostafapour K, Nouri N and Zeinali M. The effects of
cells: design guidelines. Ocean Eng 2018; 153: 387–398. the Reynolds number on the hydrodynamics characteris-
8. Brzek B, Bailon-Cuba J, Leonardi S, et al. Theoretical tics of an AUV. J Appl Fluid Mech 2018; 11(2): 343–352.
evaluation of the Reynolds shear stress and flow para- 28. Salari M and Rava A. Numerical investigation of hydro-
meters in transitionally rough turbulent boundary layers. dynamic flow over an AUV moving in the water-surface
J Turbul 2009; 10(5): 1–28. vicinity considering the laminar-turbulent transition. J
9. Mansoorzadeh S and Javanmard E. An investigation of Mar Sci Appl 2017; 16(3): 298–304.
free surface effects on drag and lift coefficients of an 29. Sakthivel R, Vengadesan S and Bhattacharyya S. Appli-
autonomous underwater vehicle (AUV) using cation of non-linear k-e turbulence model in flow
Panda et al. 27

simulation over underwater axisymmetric hull at (computational fluid dynamics) analysis. J Ocean Eng
higher angle of attack. J Nav Archit Mar Eng 2011; Technol 2009; 23(1): 48–53.
8(2): 149–163. 47. Joung TH, Sammut K, He F, et al. Shape optimization
30. Kimura I and Hosoda T. A non-linear k-e model with of an autonomous underwater vehicle with a ducted pro-
realizability for prediction of flows around bluff bodies. peller using computational fluid dynamics analysis. Int J
Int J Numer Meth Fl 2003; 42(8): 813–837. Nav Archit Ocean Eng 2012; 4(1): 44–56.
31. Chen CW, Jiang Y, Huang HC, et al. Computational fluid 48. Won DJ, Kim J and Kim J. Design optimization of duct-
dynamics study of the motion stability of an autonomous type AUVs using CFD analysis. Intell Serv Robot 2015;
underwater helicopter. Ocean Eng 2017; 143: 227–239. 8(4): 233–245.
32. Zhang S, Yu J, Zhang A, et al. Spiraling motion of 49. Gao T, Wang Y, Pang Y, et al. Hull shape optimization
underwater gliders: modeling, analysis, and experimental for autonomous underwater vehicles using CFD. Eng
results. Ocean Eng 2013; 60: 1–13. Appl Comput Fluid Mech 2016; 10(1): 599–607.
33. Isa K, Arshad M and Ishak S. A hybrid-driven underwater 50. Liu F, Wang Y, Niu W, et al. Hydrodynamic perfor-
glider model, hydrodynamics estimation, and an analysis of mance analysis and experiments of a hybrid underwater
the motion control. Ocean Eng 2014; 81: 111–129. glider with different layout of wings. In: Proceedings of
34. Dantas JLD and De Barros E. Numerical analysis of con- the OCEANS 2014, Taipei, Taiwan, 7–10 April 2014,
trol surface effects on AUV manoeuvrability. Appl Ocean pp.1–5. New York: IEEE.
Res 2013; 42: 168–181. 51. Zhang BJ, Lu XD and She WX. Resistance performance
35. Tang S, Ura T, Nakatani T, et al. Estimation of the simulation of remotely operated vehicle in deep sea con-
hydrodynamic coefficients of the complex-shaped auton- sidering propeller rotation. Proc IMechE, Part M: J Engi-
omous underwater vehicle TUNA-SAND. J Mar Sci neering for the Maritime Environment 2019; 234: 585–598.
Technol 2009; 14(3): 373–386. 52. Hayati AN, Hashemi SM and Shams M. A study on the
36. Song FX, Zhang LH, Wu ZL, et al. On resistance calcu- behind-hull performance of marine propellers astern
lation for autonomous underwater vehicles. In: Jiang Z, autonomous underwater vehicles at diverse angles of
Li S, Zeng J, et al. (eds) Advanced materials research, attack. Ocean Eng 2013; 59: 152–163.
vol. 189. Zürich: Trans Tech Publications, 2011, 53. Chen Z, Yu J, Zhang A, et al. Design and analysis of
pp.1745–1748. folding propulsion mechanism for hybrid-driven under-
37. De Barros EA, Dantas JL, Pascoal AM, et al. Investiga- water gliders. Ocean Eng 2016; 119: 125–134.
tion of normal force and moment coefficients for an 54. De Barros E and Dantas JLD. Effect of a propeller duct
AUV at nonlinear angle of attack and sideslip range. on AUV maneuverability. Ocean Eng 2012; 42: 61–70.
IEEE J Ocean Eng 2008; 33(4): 538–549. 55. Du X, Cui H and Zhang Z. Dynamics model and maneu-
38. Mitra A, Panda JP and Warrior HV. Experimental and verability of a novel AUV with a deflectable duct propel-
numerical investigation of the hydrodynamic characteris- ler. Ocean Eng 2018; 163: 191–206.
tics of autonomous underwater vehicles over sea-beds 56. Javaid MY, Ovinis M, Hashim FB, et al. Effect of wing
with complex topography. Ocean Eng 2020; 198: 106978. form on the hydrodynamic characteristics and dynamic
39. Leong ZQ, Ranmuthugala D, Penesis I, et al. Quasi-sta- stability of an underwater glider. Int J Nav Archit Ocean
tic analysis of the hydrodynamic interaction effects on an Eng 2017; 9(4): 382–389.
autonomous underwater vehicle operating in proximity 57. Meng L, Yang L, Su TC, et al. Study on the influence of
to a moving submarine. Ocean Eng 2015; 106: 175–188. porous material on underwater vehicle’s hydrodynamic
40. Mitra A, Panda JP and Warrior HV. The effects of free characteristics. Ocean Eng 2019; 191: 106528.
stream turbulence on the hydrodynamic characteristics of 58. Saghafi M and Lavimi R. Optimal design of nose and tail
an AUV hull form. Ocean Eng 2019; 174: 148–158. of an autonomous underwater vehicle hull to reduce drag
41. Du XX, Wang H, Hao CZ, et al. Analysis of hydrody- force using numerical simulation. Proc IMechE, Part M:
namic characteristics of unmanned underwater vehicle J Engineering for the Maritime Environment 2019; 234:
moving close to the sea bottom. Defen Technol 2014; 76–88.
10(1): 76–81. 59. Wu CS, He SI, Zhu DX, et al. Numerical simulation of
42. Wu L, Li Y, Su S, et al. Hydrodynamic analysis of AUV microbubble flow around an axisymmetric body. J
underwater docking with a cone-shaped dock under Hydrodyn 2006; 18(3): 217–222.
ocean currents. Ocean Eng 2014; 85: 110–126. 60. Anbarsooz M. A numerical study on drag reduction of
43. Wu L, Li Y, Liu K, et al. A physics-based simulation for underwater vehicles using hydrophobic surfaces. Proc
AUV underwater docking using the MHDG method and IMechE, Part M: J Engineering for the Maritime Environ-
a discretized propeller. Ocean Eng 2019; 187: 106081. ment 2019; 233(1): 301–309.
44. Rattanasiri P, Wilson PA and Phillips AB. Numerical 61. Shereena S, Vengadesan S, Idichandy V, et al. CFD study
investigation of a pair of self-propelled AUVs operating of drag reduction in axisymmetric underwater vehicles
in tandem. Ocean Eng 2015; 100: 126–137. using air jets. Eng Appl Comput Fluid Mech 2013; 7(2):
45. Joung T, Sammut K, He F, et al. A study on the design 193–209.
optimization of an AUV by using computational fluid 62. Huang T, Santelli N and Belt G. Stern boundary-layer
dynamic analysis. In: Proceedings of the nineteenth inter- flow on axisymmetric bodies. In: Proceedings of the
national offshore and polar engineering conference, Osaka, twelfth symposium on naval hydrodynamics, Washington,
Japan, 21–26 July 2009. Cupertino, CA: International DC, 5–9 June 1978, pp.125–157.
Society of Offshore and Polar Engineers. 63. ANSYS. ANSYS CFX-solver theory guide (ANSYS CFX
46. Joung TH, Sammut K, He F, et al. A study on the motion Release). Canonsburg, PA: ANSYS, 2009, pp.724–746.
analysis and design optimization of a ducted type AUV 64. Lumley JL. Turbulence modeling. J Appl Mech 1983;
(autonomous underwater vehicle) by using CFD 50(4b): 1097–1103.
28 Proc IMechE Part M: J Engineering for the Maritime Environment 235(1)

65. Durbin PA. Some recent developments in turbulence clo- 85. Warrior HV, Mathews S, Maity S, et al. An improved
sure modeling. Ann Rev Fluid Mech 2017; 50: 77–103. model for the return to isotropy of homogeneous turbu-
66. Da Silva Costa G, Ruiz A, Reis M, et al. Numerical anal- lence. J Fluid Eng 2014; 136(3): 034501.
ysis of stability and manoeuvrability of autonomous 86. Speziale CG, Sarkar S and Gatski TB. Modelling the
underwater vehicles (AUV) with fishtail shape. Ocean pressure–strain correlation of turbulence: an invariant
Eng 2017; 144: 320–326. dynamical systems approach. J Fluid Mech 1991; 227:
67. Jones W and Musonge P. Closure of the Reynolds stress 245–272.
and scalar flux equations. Phys Fluids 1988; 31(12): 3589– 87. Gibson M and Launder B. Ground effects on pressure
3604. fluctuations in the atmospheric boundary layer. J Fluid
68. Launder BE and Spalding DB. The numerical computa- Mech 1978; 86(3): 491–511.
tion of turbulent flows. In: Patankar SV, Pollard A and 88. Posa A and Balaras E. A numerical investigation of the
Singhal AK (eds) Numerical prediction of flow, heat trans- wake of an axisymmetric body with appendages. J Fluid
fer, turbulence and combustion. Amsterdam: Elsevier, Mech 2016; 792: 470–498.
1983, pp.96–116. 89. Nicoud F and Ducros F. Subgrid-scale stress modelling
69. Menter FR. Two-equation eddy-viscosity turbulence based on the square of the velocity gradient tensor. Flow
models for engineering applications. AIAA J 1994; 32(8): Turb Combust 1999; 62(3): 183–200.
1598–1605. 90. Da Silva Costa G, Almeida M, de Melo Filho A, et al. 3-
70. Rattanasiri P, Wilson PA and Phillips AB. Numerical D realistic simulations over a flatfish shaped AUV sub-
investigation of a fleet of towed AUVs. Ocean Eng 2014; mitted to ocean currents. Appl Ocean Res 2019; 90:
80: 25–35. 101849.
71. Mishra AA and Girimaji SS. Pressure–strain correlation 91. Posa A and Balaras E. A numerical investigation about
modeling: towards achieving consistency with rapid distor- the effects of Reynolds number on the flow around an
tion theory. Flow Turb Combust 2010; 85(3–4): 593–619. appended axisymmetric body of revolution. J Fluid Mech
72. Panda JP and Warrior HV. A representation theory- 2020; 884: A41.
based model for the rapid pressure strain correlation of 92. Kadiyam J and Mohan S. Conceptual design of a hybrid
turbulence. J Fluid Eng 2018; 140(8): 081101. propulsion underwater robotic vehicle with different pro-
73. Panda JP, Warrior HV, Maity S, et al. An improved pulsion systems for ocean observations. Ocean Eng 2019;
model including length scale anisotropy for the pressure 182: 112–125.
strain correlation of turbulence. J Fluid Eng 2017; 139(4): 93. Lauder GV and Madden PG. Learning from fish: kine-
044503. matics and experimental hydrodynamics for roboticists.
74. Panda JP. A review of pressure strain correlation model- Int J Autom Comput 2006; 3(4): 325–335.
ing for Reynolds stress models. Proc IMechE, Part C: J 94. Bang K, Kim J, Lee SI, et al. Hydrodynamic role of long-
Mechanical Engineering Science 2019; 234: 1528–1544. itudinal dorsal ridges in a leatherback turtle swimming.
75. Panda JP, Mitra A, Joshi A, et al. Experimental and Sci Rep 2016; 6: 34283.
numerical analysis of grid generated turbulence with and 95. Rivera G. Hydrodynamics of freshwater turtles: maneu-
without mean strain. Exp Therm Fluid Sci 2018; 98: 594– verability, stability, and effects of shell shape, 2009,
603. https://tigerprints.clemson.edu/cgi/
76. Mishra AA and Girimaji SS. Intercomponent energy viewcontent.cgi?article=1480&context=all_dissertations
transfer in incompressible homogeneous turbulence: 96. Katzschmann RK, DelPreto J, MacCurdy R, et al.
multi-point physics and amenability to one-point clo- Exploration of underwater life with an acoustically con-
sures. J Fluid Mech 2013; 731: 639–681. trolled soft robotic fish. Sci Robot 2018; 3(16): eaar3449.
77. Mishra AA. The art and science in modeling the pressure- 97. Singh SN, Simha A and Mittal R. Biorobotic AUV man-
velocity interactions. PhD Thesis, Texas A&M University, euvering by pectoral fins: inverse control design based on
College Station, TX, 2014. CFD parameterization. IEEE J Ocean Eng 2004; 29(3):
78. Mishra AA and Girimaji SS. On the realizability of 777–785.
pressure–strain closures. J Fluid Mech 2014; 755: 535– 98. Costa D, Palmieri G, Palpacelli MC, et al. Design of a
560. bio-inspired autonomous underwater robot. J Intell
79. Mishra AA and Girimaji SS. Hydrodynamic stability of Robot Syst 2018; 91(2): 181–192.
three-dimensional homogeneous flow topologies. Phys 99. Sverdrup-Thygeson J, Kelasidi E, Pettersen KY, et al.
Rev E 2015; 92(5): 053001. The underwater swimming manipulator–a bio-inspired
80. Mishra AA and Girimaji SS. Manufactured turbulence AUV. In: Proceedings of the 2016 IEEE/OES autonomous
with Langevin equations, https://arxiv.org/pdf/ underwater vehicles (AUV), Tokyo, Japan, 6–9 Novem-
1611.03834.pdf ber 2016, pp.387–395. New York: IEEE.
81. Mishra AA and Girimaji SS. Toward approximating 100. Lauder GV and Drucker EG. Morphology and experi-
non-local dynamics in single-point pressure–strain corre- mental hydrodynamics of fish fin control surfaces. IEEE
lation closures. J Fluid Mech 2017; 811: 168–188. J Ocean Eng 2004; 29(3): 556–571.
82. Pope S. Turbulent flows. New York: Cambridge Univer- 101. Willy A and Low K. Development and initial experi-
sity Press, 2000. ment of modular undulating fin for untethered biorobo-
83. Rotta J. Statistische theorie nichthomogener turbulenz. tic AUVs. In: Proceedings of the 2005 IEEE international
Zeitsch Phys 1951; 129: 547–572. conference on robotics and biomimetics (ROBIO), Sha-
84. Sarkar S and Speziale CG. A simple nonlinear model for tin, China, 5–9 July 2005, pp.45–50. New York: IEEE.
the return to isotropy in turbulence. Phys Fluid A Fluid 102. Chemori A, Kuusmik K, Salumäe T, et al. Depth con-
Dyn 1990; 2(1): 84–93. trol of the biomimetic U-CAT turtle-like AUV with
experiments in real operating conditions. In: Proceedings
Panda et al. 29

of the 2016 IEEE international conference on robotics view through experiment and numerical analysis. J Bio-
and automation (ICRA), Stockholm, 16–21 May 2016, nic Eng 2018; 15(6): 950–959.
pp.4750–4755. New York: IEEE. 107. Leong Z, Ranmuthugala D, Penesis I, et al. RANS-
103. Konno A, Furuya T, Mizuno A, et al. Development of based CFD prediction of the hydrodynamic coefficients
turtle-like submergence vehicle. In: Proceedings of the of DARPA suboff geometry in straight-line and rotat-
7th international symposium on marine engineering, ing arm manoeuvres. Int J Marit Eng 2015; 157(A1):
http://fluid.mech.kogakuin.ac.jp/;minnie/Research/ A41–A52.
papers/published/Paper140.pdf 108. Li Y, Pan D, Zhao Q, et al. Hydrodynamic performance
104. Zhao W, Hu Y, Wang L, et al. Development of a flipper of an autonomous underwater glider with a pair of
propelled turtle-like underwater robot and its CPG-based bioinspired hydro wings–a numerical investigation.
control algorithm. In: Proceedings of the 2008 47th IEEE Ocean Eng 2018; 163: 51–57.
conference on decision and control, Cancun, Mexico, 9–11 109. Allotta B, Costanzi R, Ridolfi A, et al. The ARROWS
December 2008, pp.5226–5231. New York: IEEE. Project: robotic technologies for underwater archaeol-
105. Pavlov VV and Rashad AM. A non-invasive dolphin ogy. In: IOP conference series: materials science and
telemetry tag: computer design and numerical flow engineering, volume 364, Florence Heri-Tech - The future
simulation. Mar Mammal Sci 2012; 28(1): E16–E27. of heritage science and technologies, Florence, Italy, 16–
106. Honaryar A and Ghiasi M. Design of a bio-inspired hull 18 May 2018. IOP Publishing Ltd.
shape for an AUV from hydrodynamic stability point of