Ocean Engineering 126 (2016) 488–500

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Underwater radiated noise prediction for a submarine propeller in different

flow conditions
crossmark
⁎
M. Cansın Özdena, , Ahmet Y. Gürkana, Yasemin Arıkan Özdenb, Talat G. Canyurta,
Emin Korkuta
a
Istanbul Technical University, Faculty of Naval Architecture and Ocean Engineering, 34469 Maslak, Istanbul, Turkey
b
Faculty of Naval Architecture and Maritime, Yıldız Technical University, 34349 Beşiktaş, Istanbul, Turkey

A R T I C L E I N F O A BS T RAC T

Keywords: International Maritime Organisation (IMO) and other bodies have been trying to set-up guidelines and
Propeller noise regulations to reduce/limit noise levels at sea which influence marine life in particular marine mammals and
URANS certain fish types. Ships with low noise characteristics will be a must in the near future for almost all ship types.
Non-cavitating However for special ships, such as naval surface vessels, fishing vessels, submarines, etc. this has already been
DARPA Suboff submarine
an issue because of their operational requirements. The propeller is one of the main sources of underwater noise
INSEAN E1619 submarine propeller
generated by ships. Therefore, it is important to predict and control the underwater noise characteristics of
Ffowcs-Williams and Hawkings (FWH) model
propellers. In this respect, the main objective of this study is to calculate propeller radiated noise numerically.
Propeller noise has been investigated numerically for the INSEAN E1619 submarine propeller in open water,
behind a generic DARPA suboff submarine and within imposed wake cases at non-cavitating conditions due to
their deeply submerged operations. Flow around the propeller is solved with a commercial CFD software using
Unsteady Reynolds Averaged Navier–Stokes (URANS), while hydro-acoustic analysis is performed using a
model based on Ffowcs-Williams and Hawkings equation. This paper reports the results of the study. The paper
also includes the details of the bodies and discusses further improvement of the methodology applied.

1. Introduction Hydro-acoustic predictions of a propeller based on computational

Ship underwater radiated noise has recently been under investiga-
tion by international bodies, such as International Maritime
Organisation (IMO), classification societies, conservation groups and
governments due to its adverse effects on marine environment.
Increasing shipping activities also increase ambient noise levels in
oceans around the world and has a potential impact on marine wildlife.
Ships with low noise characteristics have always been an important
issue for some ship types; naval vessels, fishing vessels, research vessels
and also submarines due to their specific mission and operational
requirements. The propeller is one of the most dominant sources of
noise on ships. The propeller noise is the only traceable signal by
sonars particularly for naval surface ships and submarines since all
other sources can be eliminated by the use of appropriate insulation
methods. Nevertheless, it determines the detectability, operability and
even survivability of the ship. Therefore, noise predictions for propel-
lers in cavitating and non-cavitating states have always been an
important issue in naval architecture.
fluid dynamics (CFD) methods should be investigated in two steps, first
solving the flow around the propeller, followed by the acoustic
computations.
Based on the above approach, Kawamura et al. (2004) compara-
tively analysed different turbulence models for the prediction of open
water performance for a conventional propeller. Later Li (2006)
estimated open water characteristics of a highly skewed model
propeller employing k–ω turbulence model and validated the study
with experimental data. Gao et al. (2012) simulated numerically the
unsteady viscous flow around an Autonomous Underwater Vehicle
(AUV) with propellers by using the Reynolds-averaged Navier–Stokes
(RANS) equations, shear-stress transport (SST) k–ω model and
Pressure with Splitting of Operators (PISO) algorithm based on sliding
mesh. The hydrodynamic characteristics of an AUV with propellers
such as, resistance, pressure and velocity were reflected well the real
ambient flow field of the AUV. Then, the semi-implicit method for
pressure-linked equations (SIMPLE) algorithm was used to compute
the steady viscous flow field. The computational results agreed well

⁎
Corresponding author.
E-mail addresses: ozden@itu.edu.tr (M.C. Özden), gurkanah@yahoo.com (A.Y. Gürkan), yarikan@yildiz.edu.tr (Y.A. Özden), gokcer.canyurt@gmail.com (T.G. Canyurt),
korkutem@itu.edu.tr (E. Korkut).

http://dx.doi.org/10.1016/j.oceaneng.2016.06.012
Received 3 May 2015; Received in revised form 9 June 2016; Accepted 9 June 2016
Available online 28 September 2016
0029-8018/ © 2016 Published by Elsevier Ltd.
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

with the experimental data, showing that the numerical method has a far-field acoustics.
good accuracy in the prediction of hydrodynamic performance of a
propeller (Gao et al., 2012). A detailed literature review on the 2.1. Numerical methods and flow solver
prediction of open water performance of propellers can be found in
26th ITTC (2011). For the numerical calculations ANSYS 15 Fluent was used to satisfy
Empirical, semi-empirical and Bernoulli-based methods have been the following governing equation for continuity (Alin et al., 2010);
investigated by many researchers (Testa, 2008). However, generation ∂ρ ∂
of a method by aero-acousticians Ffowcs-Williams and Hawkings + (ρvi )=0
∂t ∂xi (1)
(FWH) for calculation of noise of an arbitrary body moving in a fluid
can be considered as a mile-stone in acoustic predictions (Ffowcs where xi and vi are the tensor form of axial coordinates and velocities,
Williams and Hawkings, 1969). Di Francescoantonio (1997) was the respectively. Then the momentum equation becomes;
one who combined the idea of Ffowcs-Williams and Hawkings' porous
∂(ρvi ) ∂(ρui uj ) ∂p ∂ ⎡⎢ ⎛ ∂ui ∂uj 2 ∂u ⎞ ⎤
formulation with the work of Farassat and Myers (1988) on Kirchhoff + =− + μ⎜ + − δij l ⎟ ⎥
formulation. With the development of computing technology and ∂t ∂xj ∂xi ∂xj ⎢⎣ ⎝ ∂xj ∂xi 3 ∂xl ⎠ ⎥⎦
numerical practice, this method became available also for hydro- ∂
+ (−ρui′ u′)
j
acoustic predictions for marine propellers. Seol et al. (2002) investi- ∂xj (2)
gated the non-cavitating propeller noise employing Boundary Element
Methods (BEM) for the calculation of flow around a propeller in time- where δij is Kronecker Delta and −ρui′ u̅ ′j are the unknown Reynolds
stresses.
domain and used FW–H method to predict the far-field acoustics.
Salvatore and Ianniello (2003) published the preliminary results for For the turbulence modelling, SST k − ω turbulence model is
employed due to its good performance on wall bounded boundary
cavitating propeller noise predictions. In their study, a hydrodynamic
model for transient sheet cavitation on propellers in non-uniform layer flows (Li, 2006).
inviscid flow was coupled with a hydroacoustic model based on the FLUENT employs the cell-centred finite volume method. RANS
Ffowcs-Williams and Hawkings equation. They split the noise signa- formulation is used with absolute velocity selection. The transient
ture into thickness and loading term contributions. They demonstrated solution is performed with a second order implicit pressure based
that noise predictions by the FW–H equation were in agreement with solver. Velocity and pressure are coupled via the SIMPLE algorithm.
those obtained by using the Bernoulli equation model (Salvatore and Green Gauss Node Based is used for gradient and Pressure Staggering
Ianniello, 2003). Seol et al. (2005) extended their work to cavitating Option (PRESTO) for pressure discretization. Quadratic Upstream
noise state. They predicted cavity extent by the sheet cavity volume Interpolation for Convective Kinematics (QUICK) scheme was em-
model and used the sheet cavity volume data and time dependent ployed for Momentum, Turbulent Kinetic Energy and Specific
pressure as the input for the FW–H equation to predict far-field Dissipation Rate calculations.
acoustics (Seol et al., 2005). Barbarino and Casalino (2012) studied
and validated noise predictions for a NACA-0012 airfoil. Then they 2.2. Noise predictions
applied the same method to compute the broadband noise spectrum of
an aircraft. 2.2.1. Ffowcs-Williams and Hawkings method
Ffowcs-Williams and Hawkings (FWH) formulation adopts the
In line with the literature, a study has been carried out to
investigate the prediction of propeller noise for submarines and most general form of Lighthill's acoustic analogy, and is capable of
predicting sound generated by equivalent acoustic sources such as
underwater vehicles. The main objective of the study is to obtain
accurate propeller noise prediction and use this information for monopoles, dipoles and quadrupoles. A time domain integral formula-
tion adopted wherein time histories of sound pressure, or acoustic
controlling noise on submarines when designing the submarine and
propeller. This paper reveals the results of the study, following the signals, at prescribed receiver locations are directly computed by
evaluating few surface integrals. Time-accurate solutions of the flow
previous research project (Ozden et al., 2012 & Korkut et al., 2013).
Firstly a validation study has been carried out for noise characteristics field variables, such as pressure, velocity components, and density on
source (emission) surfaces, are required to evaluate the surface
of a fisheries research vessel propeller measured previously (Atlar et al.,
2001). Then, the predictions were performed for open water hydro- integrals. Time accurate solutions can be obtained from unsteady
Reynolds-averaged Navier–Stokes (URANS) equations, large eddy
dynamic characteristics of E1619 generic submarine propeller and
compared with the experimental results published by Di Felice et al., simulations (LES), or detached eddy simulations (DES).
Ffowcs-Williams and Hawkings (FWH) equation is an inhomoge-
(2009), as well as noise characteristics of the same propeller in open
water condition. The calculations of resistance and wake properties of neous wave equation derived from the continuity and Navier–Stokes
equations (Ffowcs Williams and Hawkings, 1969 & Brentner and
DARPA Suboff generic submarine model were carried out and com-
pared with the experimental results given by Liu and Huang (1998); Farassat, 1998).
Crook (1990); Bull (1996) and Chase (2012), respectively. Finally the 1 ∂ 2p′ ∂2 ∂ ∂
−∇2 p′= {Tij H ( f )}− {[Pij nj +ρui (un −vn )] δ ( f )}+
noise predictions of E1619 were conducted in the simulated (imposed) a 02 ∂t 2 ∂xi ∂xj ∂t ∂t
wake and behind DARPA Suboff conditions.
{[ρ0 vn+ρ (un −vn )] δ ( f )} (3)
The method used for the noise prediction is given in Methodology
section. The validation of the method and the geometries of the bodies where
are also presented. The results of the study and some conclusions
drawn from the study are also given. ui : flow velocity in xi direction.
un: flow velocity normal to the surface ( f = 0 ).
2. Methodology vi : surface velocity component in xi direction.
vn : surface velocity component normal to the surface.
The flow around a propeller is solved using a RANS solver with a δ ( f ): Dirac delta function.
SST k–ω turbulence model. Then, the transient solution is performed H ( f ): Heaviside function.
with a second order implicit pressure based solver. The velocity and
pressure are coupled via a SIMPLE algorithm. Time dependent p′, is the far-field sound pressure ( p′=p − p0 ). f = 0 is a mathematical
pressure data is used as the input for the FWH equation to predict

489
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

surface used to facilitate the application of the generalised function Table 1

theory and the free-space Green function to the unbounded space Boundary conditions.
exterior flow problem ( f > 0 ) to reach the solution. This surface ( f = 0)
Propeller hub No slip wall
represents a source (emission) surface and can be used as a surface Propeller blades No slip wall
overlapping with the body (impermeable) or as a permeable surface far Inlet Velocity inlet
from the body. ni is the unit normal vector indicating the exterior Outlet Pressure outlet
region ( f = 0). a 0 is the sound velocity in the far field and Tij is the Outer domain surface Symmetry
Rotating source domain Interface
Lighthill tensor defined as below (Ffowcs Williams and Hawkings,
1969 & Brentner and Farassat, 1998);

Tij =ρui uj +Pij −a 02 (ρ −ρ0 ) δij tion of that variable (Ffowcs Williams and Hawkings, 1969 & Brentner
(4)
and Farassat, 1998).
Pij is the compressive stress tensor. For a Stokesian fluid it is defined as
follows;
⎡ ∂u ∂uj 2 ∂u ⎤ 2.2.2. Set-up for noise calculations
Pij =pδij −μ ⎢ i + + k
δij ⎥ In order to solve the above equations numerically, the moving
⎣ ∂xj ∂xi 3 ∂xk ⎦ (5) frame of reference approach was used and a cylindrical computational
where the free-stream quantities are denoted by the subscript 0. domain was generated with two regions namely; inner and outer
The solution of Eq. (4) is obtained by the use of the free-space domains for all cases. The outer domain simulated the surrounding
Green function (δ (g)/4πr ). The complete solution includes the surface fluid and the inner volume was used to simulate the rotation of the
and volume integrals. While the surface integrals include the effects of propeller where the time step size was selected as the time required for
monopole, dipole and partially quadrupole effects and the volume a 0.1° of reference frame rotation of the propeller.
integrals include only the quadrupole sources except the regions of the In order to check the contribution of hydrodynamic volume
source surface. In cases where the flow is in low subsonic region, the sources, an initial study has been carried out previously for a navy
value of the volume integral diminishes and the source surface encloses propeller and the validation case (FRV propeller) (Ozden et al., 2012 &
the source region. Thus the volume integrals are not included in the Korkut et al., 2013). The propeller blades and hub were initially used as
calculations and the equations below are obtained; the noise sources only. Then, a cylindrical moving domain around the
propeller was used as the noise source (porous surface), based on the
p′ (x ⃗ , t )=pT′ (x ⃗ , t )+pL′ (x ⃗ , t ) (6) moving reference frame approach where the side wall of inner domain
where; was employed as the source surface in this study. It was found that the
use of this porous surface made it possible to include nonlinear terms
⎡ ⎛ ̇ ⎞⎤ and showed better noise prediction values compared the use of the
⎢ ρ0 ⎜Un + Un⎟̇ ⎥
⎢ ⎝ ⎠⎥ blades and hub only. A similar approach was also used by other
4πpT′ (→
x , t) = ∫ dS
f =0 ⎢ r (1 − Mr )2 ⎥ researchers (Ianniello and De Bernardis, 2015).
⎢ ⎥
⎣ ⎦ During the computations, the unsteady RANS solver of the ANSYS
Fluent software was used. The propeller was placed inside a cylindrical
⎡ ρ Un {rMṙ + a 0 (Mr − M2 )} ⎤
+ ∫f =0 ⎢⎣ 0
⎥ dS Chimera block with 1.5D diameter and 2.5D length. The outer wall of
r 2 (1 − Mr )3 ⎦ (7) this block was selected as the acoustic surface in order to capture the
nonlinear terms. The boundary conditions employed are given in
1 ⎡ ̇ ⎤ ⎡ ⎤
4πpL′ (→
x , t) =
a0
∫f =0 ⎢⎣ r (1 −Lr Mr )2 ⎥⎦ dS + ∫f =0 ⎢⎣ r 2L(1r −− LMMr )2 ⎥⎦ dS Table 1.
Receivers were located 1 m away from the propeller centre, as
1 ⎡ ̇ 2 ⎤ shown in Fig. 1, in order to avoid further 1 m correction. A common
+
a0
∫f =0 ⎢⎣ Lr {rMrr 2+(1a−0 (MMrr )−3 M )} ⎥⎦ dS (8) practice in the analysis and presentation of the noise levels is to reduce
the measured or computed values of Sound Pressure Levels (SPL) in
where; each 1/3 Octave band to an equivalent 1 Hz bandwidth by means of the
ρ correction formula recommended by ITTC (2008) as follows:
Ui=vi + (ui −vi )
ρ0 (9)
SPL1 = SPL m − 10 log Δf (12)
L i=Pij n^j +ρui (un −vn ) (10) where SPL1 is the reduced sound pressure level to 1 Hz bandwidth in
When the integration surface is overlapping a closed wall (impene- dB; re 1 μPa (standard reference pressure for water), SPLm is the
trable wall) the expressions pT′ (x ⃗,t ) and pL′ (x ⃗,t ) from Eqs. (6)–(8) are measured or computed sound pressure level at each centre frequency in
referred as the thickness noise term and the loading noise term (Ffowcs dB; re 1 μPa and Δf is the bandwidth for each one-third octave band
Williams and Hawkings, 1969 & Brentner and Farassat, 1998). The filter in Hz. The ITTC also required that the sound pressure levels be
thickness noise term expresses the noise generated by the displacement corrected to a standard measuring distance of 1 m using the following
of the flow and the loading noise term expresses the noise generated by relationship:
the thrust due to the rotation of the blade (Testa, 2008). The terms in SPL = SPL1 + 20 log(r ) (13)
brackets in Eqs. (7) and (8) reveal that the kernels of the integrals are
solved for the retarded time steps (τ ) expressed as in Eq. (11) where t is where SPL is the equivalent 1 Hz at 1 m distance sound pressure level
time and r is the observer distance; (in dB; re 1 μPa) and r is the vertical reference distance for which the
r noise level is measured or receiver locations from the propeller as
τ=t−
a0 (11) r=1 m in the noise calculations. Therefore, 1 m correction was not
applied in this case. Hanning filter was used and results were
The expressions in the equations denoted by a subscript are the graphically presented where; the logarithmic-scaled x-axis represents
^
elements of vectors and unit vectors. For example r ⃗ and n⃗ denote the the centre frequencies in Hz, while the linear-scaled y-axis represents
unit vectors in radiation and wall-normal directions in L r =L ⃗∙r^⃗ =L i ri and the sound pressure levels in dB re 1 μ, 1 Hz, 1 m, for all cases presented
Un=U⃗ ∙n ⃗=Ui ni . The dot over a variable denotes source-time differentia- in this paper.

490
M.C. Özden et al.
Fig. 1. Receiver locations used in calculations.
3. Validation of the methodology
The methodology to calculate noise characteristics of a propeller
have been tested for a fisheries research vessel propeller (Korkut et al.,
2013). Atlar et al. (2001) presented the results of cavitation tunnel tests
carried out with model propeller of a Sumitomo Heavy Industries
Fisheries Research Vessel (FRV) and those of noise measurements with
its full-scale propeller to validate the low-noise performance of this
propeller. The tests involved the simulation of a target wake using a
wake screen and the determination of the noise behind the simulated
wake.
A 4-bladed fisheries research vessel propeller (Fig. 2) was investi-
gated where cavitation tunnel hydrodynamic performance and hydro-
acoustics tests were conducted in Emerson Cavitation Tunnel in the
University of Newcastle upon Tyne, UK. During the tests, the target
wake velocity ratios (Va/Vs) were simulated using different size of
meshes attached to a square frame and the wake velocities were
measured using a pitot static tube comb and “scanivalve” assembly.
Main particulars of the model propeller are given in Table 2.
A cylindrical computational domain (Fig. 3) with 3.500.000 ele-
ments has been generated with 1.5D diameter and 3D length for the
inner domain and 6D diameter and 9D length for the outer domain
491
Ocean Engineering 126 (2016) 488–500
Fig. 2. Fisheries research vessel propeller model.
where the propeller was placed at a distance of 3D from the inlet and
6D from the outlet. Non-dimensional wall distance value of y+≈ 50 was
obtained. Since a hull body was not available, the wake characteristics
in the simulated wake were measured instead. It would be a good
solution to use the measured wake for CFD calculations. In order to
take into account the influence of the wake characteristics, velocity
profile of the measured wake was imposed to the velocity inlet of the
calculation domain to simulate the non-uniformity. Fig. 4 shows the
measured wake and its adaptation to the velocity inlet.
It would be more appropriate to use the test results of propeller
noise in non-cavitating state in order to validate the CFD calculations
performed. The closest test result was the case given in Fig. 5. Both
cavitation tunnel tests and CFD calculations were conducted for a
0.3 m diameter propeller with the rotational speed of 978 rpm at an
advanced speed of 3 m/s and the density of the water in the tunnel was
1002 kg/m3.
The calculation was performed for the J value 0.61. Unsteady flow
solver was employed for the prediction of performances of the propeller
behind the wake condition. Results of KT, KQ and η0 values are given in
Table 3 comparatively with the experimental ones.
KT values were predicted with a better accuracy than KQ values. The
difference in KQ values was reflected on η0 values, as expected.
Acoustic predictions followed the solution of the flow around the
propeller. It should be noted that a hydrophone was placed on top of
the cavitation tunnel to measure noise values. Then, the calculated
noise characteristics at the corresponding receiver (2) were used for
comparison. Concerning noise prediction there is a good agreement
between the predicted results with the experiments for the frequency
range of 200–5000 Hz, as shown in Fig. 6. For the lower and higher
ranges of the frequency, RANS over predicts the flow noise.
4. Geometry of bodies
4.1. DARPA Suboff
DARPA Suboff AFF8 is a generic submarine model geometry with a
length of 4.36 m comprising of 1.02 m forebody, 2.23 m midbody and
1.11 m aftbody. It has a cylindrical cross-section with a maximum
diameter of 0.508 m. The AFF8 has a sail which is located at the top
dead centre of the hull starting at x=0.92 m from the bow and ending at
x=1.29 m. It has a cross shaped rudder where rudders and hydroplanes
are located at x=4 m from the bow. The hull and appendage arrange-
ment of DARPA Suboff AFF8 is shown in Fig. 7 and the main
particulars are given in Table 4 (Groves et al, 1998). Views of the
calculation domain and DARPA Suboff are shown in Figs. 8 and 9,
respectively.
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Table 2
Main particulars of model propeller.

Number of blades, Z 4
Propeller diameter, D 0.30 m
Pitch ratio at 0.7R, P/D 0.8464
Expanded blade area ratio, AE/A0 0.55
Boss ratio, rh/R 0.276
Rake 0°
Skew 40°
Direction of rotation Right handed

Fig. 5. Cavitation observation on fisheries research vessel propeller.

Table 3
Comparison of CFD results with experiment.

J KT 10KQ η0

Experiment 0.61 0.12 0.17 0.66

CFD 0.61 0.11 0.19 0.59
% Error 1.12 8.93 11.69
Fig. 3. Computational domain for the validation case.

Laser Doppler Velocimetry (LDV) system in the large circulating water

4.2. INSEAN E1619 submarine propeller
channel at INSEAN. Results were presented by Di Felice et al. (2009).
The main particulars of the E1619 submarine propeller are given in
The propeller used for the study was INSEAN E1619 generic
Table 5 and 3-D views are shown in Fig. 10. In Fig. 11 the detailed fine
submarine propeller. The propeller was a seven-bladed highly skewed
mesh used for the study is also presented.
submarine propeller with an unloaded tip blade design and the main
particulars of the E1619 submarine propeller are given in Table 5 (Di
Felice et al. (2009)). This propeller has been analysed in open water, 5. Calculations of INSEAn e1619 propeller in uniform flow
behind the imposed wake and self-propelling DARPA Suboff AFF8
conditions. Open water experiments were performed in the INSEAN 5.1. Performance prediction
towing tank, and wake velocity measurements were carried out by a
Open water performance calculations of E1619 propeller was

Fig. 4. a-Measured wake (top), b-adapted (imposed) wake (bottom) black circle representing measurement domain.

492
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Fig. 6. Comparison of CFD results at receiver 2 with the experiment at J=0.61.

Fig. 7. Geometry of DARPA Suboff AFF8.

Table 4
Main particulars of DARPA Suboff AFF8 (Groves et al, 1998).

Generic submarine type DARPA Suboff AFF8

Description Symbol Magnitude

Length overall Loa 4.356 m

Fig. 9. Cut away view of DARPA Suboff.
Length between perpendiculars Lpp 4.261 m
Maximum hull radius Rmax 0.254 m
Centre of buoyancy (aft of nose) FB 0.4621 Loa Table 5
Volume of displacement ∇ 0.718 Main particulars of INSEAN E1619 submarine propeller (Di Felice et al., 2009).
Wetted surface Swa 6.338
Propeller type INSEAN E1619

Advanced speed 1.68 m/s

RPM 280 rpm
Diameter 0.485 m
Number of blades 7
AE/A0 0.608
Hub/diameter ratio 0.226
Pitch to diameter ratio, P/D at 0.7R 1.15

Reynolds number, based on the section at 0.7R, was about 2.3×105

(Di Felice et al., 2009)).
The results of the fine mesh are presented in Fig. 13, indicating that
RANS calculations captured the thrust, torque and open water effi-
ciency values very well at the given advanced ratios.
Fig. 8. Calculation domain of DARPA Suboff.

carried out at J=0.74 and J=0.85. A cylindrical computation domain
has been generated similar to the validation case. A mesh indepen-
dence study was performed from coarse to fine meshes, using
6.386.638, 8.065.679 and 10.513.205 cells, respectively with non-
dimensional wall distance value of y+≈ 50. The convergence of grid
study can be seen in Fig. 12 with the above grid properties in
comparison with the experimental values for thrust and torque
coefficients at J=0.74. During the experiments, the average blade
5.2. Noise prediction in open water
After the steady computations, described in Section 5.1, which were
carried out for the performance predictions of E1619 generic sub-
marine propeller, calculations were made transiently to predict noise
characteristics. The domain which contains the propeller body was
selected as the source of noise and a number of receiver locations was
defined for the calculation. For the calculations performed for E1619
propeller in open water condition, acoustic predictions were made for a

493
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Fig. 10. 3-D views of INSEAN E1619 propeller.

Fig. 12. Comparison of convergence of CFD values and experiment results from coarse
to fine mesh.

Fig. 11. Detailed fine mesh for INSEAN E1619.

total of 4 receivers located 1 m away from the propeller reference point

(Fig. 1). Fig. 14 shows the noise predictions for different receivers at
1.68 m/s flow speed and 280 rpm of rotational speed where J=0.74.
Since the measured noise data of E1619 model propeller is not
available in open literature a comparison of the computed noise data
with experimental data was not made.

Fig. 13. Comparison of open water characteristics calculation for E1619 with experi-
6. Calculations of DARPa suboff in non-uniform flow mental results by Di Felice et al. (2009).

6.1. Flow field calculations
Before computing noise predictions of the DARPA Suboff AFF8
with E1619 propeller, a set of computations was performed for the
resistance characteristics for the DARPA Suboff hull without the
propeller in fully appended configuration. Computations were carried
out in a cylindrical domain with a diameter of 9 m and a region of
4.5 m ahead of the model and 9 m after the model, as shown in Fig. 8.
Two different grids were generated with 7.812.122 and 11.511.137
cells where y+≈50. Resistance predictions with bigger cell arrangement
were compared with the experimental results carried out by Liu and

494
M.C. Özden et al.
Fig. 14. Comparisons of noise predictions in open water for different receiver positions at J=0.74.
Fig. 15. Comparison of resistance predicted by CFD method with experiment (Liu and
Huang, 1998).
Huang (1998). Fig. 15 shows the comparison of the computed
resistance values with the experimental values. The resistance predic-
tions made by RANS were 2.5% and 7.6% less than those measured at
the velocities of 3.05 m/s and 5.14 m/s, respectively. This indicates
that the resistance predicted by RANS computations is in a good
agreement with the experimental ones.
Furthermore, steady RANS calculations were also performed to
obtain the wake characteristics of the fully appended submarine model
without INSEAN E1619 propeller. The computationally predicted wake
at the location of x/L=0.978 was compared with the measured wake of
Crook (1990) and Chase (2012). For the wake measurements Reynolds
number, based on the length, was 1.2×107.
In Fig. 16, the hole at the middle of each figure is due to the body of
the submarine. The overall pattern of both wakes is very similar. The
reduced flow velocity characteristics can be observed at the top due to
the wake of the sail. A “V” shaped higher velocity field on the top is
different than those at the sides and bottom where the horse-shoe
vortices occur which deplete the boundary layer at the centre and send
low momentum flow to the sides. In the experimental wake data, the
low velocity field between lower rudder and bottom results from the
presence of the struts which were used during the wind tunnel tests.
6.2. Noise predictions
Noise predictions were performed for two different conditions; in
the imposed wake and behind the DARPA Suboff wake conditions. In
both conditions the INSEAN E1619 propeller, which was scaled to
0.262 m diameter, was rotating at the self-propulsion point. In the
imposed wake condition, the DARPA Suboff body did not exist but the
computed wake of the body at 2.75 m/s in Fig. 16 was imposed at the
inlet. For the behind body condition the propeller was fitted to the
DARPA Suboff model with the appendages and the noise computations
495
Ocean Engineering 126 (2016) 488–500
were carried out at the self-propulsion point as in the imposed wake
condition.
6.2.1. Propeller in simulated (imposed) wake
Noise calculations for the E1619 propeller were carried out in the
imposed wake condition to investigate applicability of this method in
the case of only propeller and wake data being available, without the
hull body being available. A computational grid has been generated as a
cylindrical domain with a diameter of 9 m and a region of 8.5 m ahead
and 9 m after the propeller geometry, having 9.834.381 cells.
Computational domain for the imposed wake condition is schemati-
cally described in Fig. 17. With this grid, previously computed wake of
the submarine model at 2.75 m/s was imposed to the inlet of the
domain as the DARPA Suboff body itself was absent. In the calcula-
tions, the scaled INSEAN E1619 propeller was rotated at 563.3RPM,
corresponding to the self-propulsion point of J=1.18. URANS itera-
tions were carried out for the acoustics computations. Time step has
been selected as the time required for a 0.1° rotation of the propeller.
Receiver locations were given in Fig. 17.
6.2.2. Propeller behind DARPA Suboff
The DARPA Suboff AFF8 with the E1619 propeller has been studied
by several researchers to investigate characteristics of self-propulsion
point (Chase, 2012, Chase and Carrica, 2013), cycle-to-cycle blade
loading (Liefvendahl and Toerng, 2011) and hull interactions Alin et al.
(2010).
Noise predictions of the E1619 propeller behind the DARPA Suboff
AFF8 were performed using the URANS under self-propelled condi-
tion. The computational domain for the self-propelled condition was
schematically described in Fig. 18. The computations were carried out
in a cylindrical domain with a diameter of 9 m and a region of 4.5 m
ahead and 9 m after the model geometry. The total number of cells was
19.777.345. The flow velocity and propeller rotation rate were the same
as those used in the in the imposed wake condition. Reynolds number
based on the length was 1.2×107 identical to the DARPA Suboff's
experimental wake surveys which were performed in David Taylor
Research Centre Anechoic Flow Facility (Chase, 2012). The time step
has been selected as the time required for a 0.1° rotation of the
propeller and the locations of receiver were also identical to the ones
for the imposed wake condition.
Fig. 19 shows the calculation domain and cut away views. Fig. 20
also shows the perspective view from the stern where contour plots of
the axial velocities are shown on transverse and longitudinal cross-
sections along the hull.
7. Presentation and discussion of results
Fig. 21 shows distributions of axial velocity contours in the wake
trajectories of the INSEAN E1619 propeller behind DARPA Suboff and
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Fig. 16. Measured and computed wake of DARPA Suboff AFF8 configuration.

Fig. 17. Set up for INSEAN E1619 propeller in imposed wake case.

Fig. 18. Computational domain for DARPA Suboff with propeller.

in the imposed wake condition, respectively. Comparisons of wake propeller hub with a slow velocity field. This can also be observed on
contours predicted by self-propelled and imposed wake conditions are the transverse cross-section views of the wake characteristics, pre-
presented in Fig. 22 with a 0.005x/L interval. Figs. 23–25 show the sented in Fig. 22.
results of the noise predictions in both conditions.
The main difference between the two velocity contours comes from
the use of different arrangements of the hub region for each case. For
the behind body condition the DARPA Suboff body was present while a
streamlined hub geometry was used in the imposed wake case. This
affects the region about one diameter distance downstream of the

496
M.C. Özden et al.
Fig. 19. Calculation domain and cut away views of the bodies.
Self propelled Imposed wakeat x /L
= 1.005at x / L
= 1.010at x / L
= 1.015at x / L
= 1.020
Fig. 23 shows the noise predictions of the E1619 propeller in
behind hull condition for varying receiver positions. Fig. 24 shows
noise predictions of the E1619 propeller in the case of imposed wake
for varying receiver positions. A comparison of the noise predictions in
both cases is presented at receiver 2 in Fig. 25.
As far as the noise predictions are concerned, SPL curves of both
conditions displayed similar magnitudes with slight differences, de-
pending on the receiver location and frequency range. For the behind
Fig. 20. Axial velocity cross-sections around DARPA Suboff with E1619.
497
Ocean Engineering 126 (2016) 488–500
Fig. 21. Cross section plan view of the wake for self-propelled (top) and imposed wake
(bottom) conditions.
body condition receiver 1 was the noisiest of four receivers particularly
in high frequency range. However, the receiver locations in the
imposed wake condition displayed a different trend compared to that
of behind body condition. Nevertheless the comparison of noise levels
at receiver 2 in both conditions shows a reasonable correlation,
indicating that the imposed wake may be used to calculate noise levels
of a propeller when wake characteristics of a ship exists, but the hull
geometry is unavailable.
8. Conclusions
A numerical study has been carried out to predict noise character-
istics of the 7-bladed INSEAN E1619 submarine propeller with a
certain accuracy in different flow conditions, such as open water,
imposed wake and behind DARPA Suboff AFF8 fully appended body by
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Fig. 22. Comparisons of wake contours predicted by self-propelled and imposed wake at different positions.

498
M.C. Özden et al. Ocean Engineering 126 (2016) 488–500

Fig. 23. Underwater noise predictions for varying receiver positions in the case of self-propelled DARPA Suboff by E1619 propeller.

Fig. 24. Underwater noise predictions for varying receiver positions in the case of DARPA Suboff with imposed wake.

Fig. 25. Comparison of underwater noise predictions for self-propelled DARPA Suboff by E1619 propeller with those by imposed wake at receiver 2.

using a Reynolds Averaged Navier–Stokes (RANS) method. The aim of
the study was to try to correlate this information for controlling noise
characteristics on submarines and underwater vehicles when designing
such a vessel and its propeller. The conclusions drawn from the study
are as follows:
• The method used, predicted the open water performance of the
INSEAN E1619 propeller well. Steady computations of resistance
and wake characteristics also showed a good agreement with
experimental measurements. This indicates that RANS method with
a SST k − ω turbulence model is an effective tool in prediction of
performance of a propeller, resistance and wake characteristics of
the DARPA Suboff AFF8 fully appended body.
• During the imposed wake computations, the hub shape in the
absence of submarine body resulted in lower velocity characteristics
compared to those in the self-propelled condition. The use of the
imposed wake did not make a significant difference in the noise
values, compared to the behind body condition while the number of
mesh elements for the imposed wake condition was reduced
significantly, hence the computation time. However the selected
receiver locations were far from the hull. As a future work, receivers
should be located close to the hull and rudders to investigate
scattering effects.
• Noise propagation used in this study was a spherical spreading
model and different receiver locations indicated different values of
noise signal and this requires further investigation.
• Experimental noise data are not available publicly for neither
INSEAN E1619 submarine propeller in open water condition nor
the propeller behind DARPA Suboff submarine body condition.
Future studies may include experimental measurements of noise
characteristics of the propeller in different flow conditions.

499
M.C. Özden et al.
Acknowledgements
Authors would like to thank Prof M. Atlar of Newcastle University
and Dr. Sasaki of former member of Sumitomo Heavy Ind. for sharing
the geometry, wake and noise data of Sumitomo Fisheries Research
Vessel. We would also like to thank M. Felli and F. Salvatore of
INSEAN for sharing the 3D geometry of INSEAN E1619 submarine
propeller for use in our academic studies. The authors are grateful to
Assist. Prof. O. Unal and Prof. O. Goren of ITU for useful discussions
and use of the high speed computer at Numerical Hydrodynamics
Laboratory. The authors thank to Prof. M. Taylan of Istanbul Technical
University for his comments on the preparation of the paper. Authors
also wish to acknowledge Mr. B. Can of Wichita State University for his
invaluable help and Mr. E. Demir and Mr. A. Ak from DATUM Ltd for
the preparations of 3D geometries.
References
Alin, N., Chapuis, M., Fureby, C., Liefvendahl, M., Svennberg, U., Troeng, C., 2010. A
numerical study of submarine propeller-hull interaction. In: Proceedings of the 28th
Symposium on Naval Hydrodynamics. Pasadena, California.
Atlar, M., Takinaci, A.C., Korkut, E., Sasaki, N., Aono, T., 2001. Cavitation tunnel tests
for propeller noise of a frv and comparisons with full-scale measurements. In:
Proceeedings of the 4th International Symposium on Cavitation CAV2001. Pasadena,
California, USA.
Barbarino, M., Casalino, D., 2012. Hybrid analytical/numerical prediction of propeller
broadband noise in time domain. Int. J. Aeronaut. 11 (2), 157–175.
Brentner, K.S., Farassat, F., 1998. An analytical comparison of the acoustic analogy and
Kirschhoff formulations for moving surfaces. AIAA J. 36, 8.
Bull, P., 1996. The validation of CFD predictions of nominal wake for the Suboff fully
appended geometry. In: Proceedings of the 21st Symposium on Naval
Hydrodynamics. June. pp.1061–1076.
Chase, N., 2012. Simulations of the DARPA Suboff Submarine Including Self-propulsion
with the E1619 Propeller (M.Sc. Thesis). University of Iowa, USA.
Chase, N., Carrica, P.M., 2013. Submarine propeller computations and application to
self-propulsion of DARPA suboff. Ocean Eng. 60, 68–80.
Crook, B., 1990. Resistance for DARPA Suboff as Represented by Model 5470. David
Taylor Research Center Report, Maryland, USA, Report No: DTRC/SHD-1298-07.
Di Felice, F., Felli, M., Liefvendahl, M., Svennberg, U., 2009. Numerical and
experimental analysis of the wake behavior of a generic submarine propeller. In:
Proceedings of the 1st International Symposium on Marine Propulsors. Trondheim,
Norway.
500
Ocean Engineering 126 (2016) 488–500
Di Francescoantonio, P., 1997. A new boundary integral formulation for the prediction of
sound radiation. J. Sound Vib. 202 (4), 491–509.
Farassat, F., Myers, M.K., 1988. Extension of Kirchhoff's formula to radiation from
moving surfaces. J. Sound Vib. 123 (3), 451–460.
Ffowcs Williams, J.E., Hawkings, D.L., 1969. Sound generation by turbulence and
surfaces in arbitrary motion. Philos. Trans. R. Soc. A264, 321–342.
Gao, F.D., Pan, C.Y., Han, Y.Y., 2012. Numerical computation and analysis of unsteady
viscous flow around autonomous underwater vehicle with propellers based on sliding
mesh. J. Cent. South Univ. Technol. 19 (4), 944–952.
Groves, N., Huang, T., Chang, M., 1998. Geometric Characteristics of DARPA SUBOFF
Models (DTRC Model Nos. 5470 and 5471. David Taylor Research Center Report,
Maryland, USA, Report No. DTRC/SHD-1298-01.
Ianniello, S., De Bernardis, E., 2015. Farassat's formulations in marine propeller
hydroacoustics. Aeroacoustics 14 (1 & 2), 87–103.
ITTC, 2008. Propulsion committee report. In: Proceedings of the 25th ITTC,
International Towing Tank Conference. Fukuako, Japan. Vol. I. pp. 83–141.
ITTC, 2011. Specialist committee on computational fluid dynamics report. In:
Proceedings of the 26th International Towing Tank Conference. Rio de Janeiro,
Brazil. pp. 337–375.
Kawamura, T., Watanabe, T., Takekoshi, Y., Maeda, M., Yamaguchi, H., 2004. Numerical
Simulation of Cavitating Flow around a Propeller. JSNA. pp.195
Korkut, E., Ozden, M.C., Avci, G.A, Goren, O., Takinaci, A.C., Danisman, D.B., 2013.
Numerical calculations of noise characteristics of surface and underwater navy ships
generated by propeller. Turkish Scientific and Research Council (TUBITAK) Project.
Project No: 110M327
Li, D.-Q., 2006. Validation of RANS predictions of open water performance of a highly
skewed propeller with experiments. In: Proceedings of the Conference of Global
Chinese Scholars on Hydrodynamics. Journal of Hydrodynamics. Vol. 18. Issue 3.
Supplement. pp. S520–S528.
Liefvendahl, M., Toerng, C., 2011. Computation of cycle-to-cycle variation in blade load
for a submarine propeller, using LES. In: Proceedings of the 2nd International
Symposium on Marine Propulsors, SMP11. Hamburg, Germany.
Liu, H.-L, Huang, T., 1998. Summary of DARPA SUBOFF Experimental Program Data,
Report No. CRDKNSWC/HD-1298-11, June.
Ozden, M.C., Avci, G.A., Korkut, E., 2012. A numerical study on prediction of noise
characteristics generated by a propeller. In: Proceedings of the 10th International
Conference on Hydrodynamics – ICHD2012. St. Petersburg, Russia.
Salvatore, F., Ianniello, S., 2003. Preliminary results on acoustic modelling of cavitating
propellers. Comput. Mech. 32, 291–300.
Seol, H., Suh, J.C., Lee, S., 2002. Prediction of non-cavitating underwater propeller noise.
J. Sound Vib. 257 (1), 131–156.
Seol, H., Suh, J.C., Lee, S., 2005. Development of hybrid method for the prediction of
underwater propeller noise. J. Sound Vib. 288 (1), 345–360.
Testa, C., 2008. Acoustic Formulations for Aeronautical and Rotorcraft Noise Prediction
Based on the Ffowcs Williams and Hawkings Equation. Universita Degli Studi di
Roma Tre, Italy.