Third International Symposium on Marine Propulsors

smp’13, Launceston, Tasmania, Australia, May 2013

CFD Analysis of Propeller Performance in Oblique Flow

Giulio Dubbioso 1, Roberto Muscari1, Andrea Di Mascio 2

1
CNR-INSEAN (Italian Ship M odel Basin), Rome, Italy
2
CNR-IAC (Istituto per le applicazioni del Calcolo M auro Picone), Rome, Italy

ABSTRACT forces and moments are generated (or at least, amplified) in

The present work is aimed to assess the capability of a the propeller plane which may affect profoundly the
numerical code based on the solution of the Reynolds dynamic response of the vehicle. Viviani et al (2007)
averaged Navier—Stokes Equations for the study of showed that, during turning circle manoeuvres, propeller
propeller functioning in off design conditions; this aspect power/torque demand increase up to 100% and 50% of the
is becoming of central interest in naval hydrodynamics value in the approach phase (straight path) for the external
research because of its crucial implications on design (relatively to the centre of the turn) and internal propellers,
aspects and performance analysis of the vessel during its respectively. Atsvanapranee et al (2010) evidenced that in-
operational life. A marine propeller working in oblique flow plane loads exerted by the propeller blades during drift
conditions is numerically simulated by the unsteady motion can affect profoundly the dynamic response
Reynolds averaged Navier-Stokes equations (uRaNSe) (course stability and turning qualities) of the vessel, by
and a dynamically overlapping grid approach. The test means of circular motion tests on the DDG51 destroyer
case considered is the CNR-INSEAN E779A propeller model. In this regard, moreover, numerical computations
model. Two different loading conditions have been performed on a twin screw single rudder tanker like vessel
considered at different incidence angles (10° to 30°) in (Durante et al 2010, Broglia et al 2013), stressed that usual
order to analyze the propeller performance during idealized actuator disk model (broadly adopted for simulation of self
off-design conditions, similar to those experienced during propulsion tests) do not account for a lateral force. An
a tight manoeuvre. The main focus is on hydrodynamic hybrid model consisting in the actuator disk for treating
loads (forces and moments) that act on a single blade, on the axial component of the flow and the Ribner lateral force
the hub and on the complete propeller; peculiar model (Ribner 1945) for the modeling of oblique effects
characteristics of pressure distribution on the blade will be has been proposed with promising results. Moreover,
presented as well. Verification of the numerical similar computations, carried out on the same vessel
computations have been asses sed by grid convergence equipped with a different configuration, further stressed
analysis. the need for a propeller lateral force model for estimating
Keywords
complex phenomena related to rudder-propeller interaction
Propeller off-design functioning; Dynamic Overset Grids; (Broglia et al 2011).
Oblique Flow; Blade loads. On the basis of the above considerations, investigation of
the propeller functioning in off-design condition is of
1 INTRODUCTION paramount importance in order to improve the propulsion
Propeller performance prediction has been traditionally system sizing and design, as well as to provide an
centred on ship’s design requirements in straight ahead accurate estimation of the vessel dynamic response.
sailing, focusing mainly, in case of standard propulsion Traditional numerical approaches like B.E.M (Boundary
configurations (single or multi-screw), on thrust and Element Method) or VLM (Vortex Lattice Method)
torque hydro-loads. Strictly speaking, a screw propeller is accurately predict the hydrodynamic characteristics of the
designed in order to satisfy a prescribed requirement propeller in typical design conditions, namely in open
(maximization of the overall propulsive efficiency and/or water or in a wake field typical of a ship advancing straight
minimization of the cavitation phenomena) by a suitable on its path or, in general, when the cross flow components
adaptation of the incidence of the blade sections to the are negligible with respect to the axial one. Accuracy of
ship wake experienced at the required speed. However, these approaches for typical propeller off-design
during its operational life, a vessel may experience operations may be spoiled by massive unsteady
different off-design scenarios, like maneuvering and crash separation and/or cavitation phenomena which are
stop, which change completely the working regime of the intrinsically not accounted for in these formulations. On
propeller. In these cases, in fact, propeller loads can the other hand, Computational Fluid Dynamics (CFD)
increase abruptly with respect to those experienced during approaches based on the direct solution of the Navier-
the straight ahead motion, because of complex flow field Stokes equations can be an attractive alternative because
phenomena and interactions between hull, propeller and of their ability to treat viscous effects and separation
rudder; moreover, in these circumstances, additional phenomena.

298
In the present work the performance of a rotating propeller time–derivative in the governing equations is
set at incidence with respect to the inflow is investigated approximated by a second order accurate, three–point
by means of an approach based on the numerical solution backward finite difference formula (Di Mascio et al 2004).
of the unsteady Reynolds averaged Navier—Stokes Finally, the rotation of the propeller is simulated by a
equations (uRaNSe). The selected propeller is the four dynamic overlapping grids algorithm (Muscari & Di
bladed INSEAN E779A model, for which high quality data Mascio 2005, Muscari et al 2006).
for CFD validation exist Salvatore et al 2004). The
3 TEST CASE
hydrodynamic solver is the in-house code χnavis, which is
The propeller selected for this study is the E779A INSEAN
a finite volume uRaNSe solver. The rotation of the
model, a modified Wageningen propeller characterised by
propeller with respect to the fixed background is handled
a constant pitch distribution and very low skew. View of
by a dynamic overset grids technique.
the propeller is shown in Figure 2 whereas the essential
2 MATHEMATICAL MODEL geometric characteristics are summarized in Table 1.
The numerical method is based on the integration of the
URANSE, the turbulent viscosity being calculated by
means of a proper turbulence model. The INSEAN code
χnavis implements several turbulence models, ranging
from algebraic ones (Baldwin & Lomax 1978, Smagorinsky
1963) to two–equations differential models (Lam &
Bremhorst 1981, Chang et al 1995). For the present results,
the one–equation model by Spalart & Allmaras (1994) was
used, supported be previous results on a rotating
propeller and rudder-propeller interaction (Muscari et al
(2011)). Figure 2: Propeller E779A overview
Proper boundary conditions are enforced at the physical
boundaries of the numerical domain (sketched in Figure 1). Diameter 0.227m
In particular, on the solid walls the velocity of the fluid is Number of Blades 4
set equal to the local velocity of the body; at the inflow, P/D 1.1
the velocity is set to the undisturbed flow value, whereas Table 1: Propeller E779 characteristics
at the outflow the pressure is set equal to zero; finally, on The propeller has been tested for two different advance
the free surface, whose location is determined by a coefficients J=U∞/nD (namely, J=0.88 and J=0.60),
kinematic condition, the continuity of stresses across the covering a relatively broad set of incidence angles (10°,
surface is enforced (dynamic condition). 20° and 30°). Numerical computations have been carried
out with the medium grid for each case in the test matrix, in
order to investigate and characterize the global behaviour
of the propeller. Furthermore, computations on the finest
grid have also been performed for the two conditions that
may be more representative of a real ship maneuvering
scenario, namely J=0.88 at 10° and J=0.6 at 30°. The
former case is typically met during the firsts instants of the
Figure 1: Topology of the numerical domain. Left: complete manoeuvre (the propeller loading is close to the one
domain; Right: propeller near field experienced during the approach phase, and the low
values of lateral and rotational speed result in low drift
The numerical algorithm is based on a finite volume
angle in correspondence of the propeller location); after
technique with pressure and velocity co–located at cell
the first phase, hull speed reduction and coupled s way-
center. Viscous terms are integrated by a standard second
yaw motion cause the propeller to operate at low advance
order centered scheme, whereas for the convective and
coefficient and at high incidence with respect to the mean
pressure terms three different schemes can be chosen: a
flow, that is at a regime represented by the second fine
second order ENO (Essentially Non Oscillatory) scheme, a
grid simulation. Figure 2 shows the inertial frame of
third order upwind scheme and a fourth order centered
reference adopted for the numerical simulations: the x axis,
scheme (Di Mascio et al 2008). The third order upwind
pointing downstream, coincides with the propeller axis of
scheme was used for the computations shown in the
rotation; z axis is directed upward and the y axis follows
following, as it represents the best trade–off between
the right hand rule. In the numerical simulations, the
small numerical dissipation and good robustness. Of
rotational speed of the propeller has been kept fixed to a
course, because of the treatment of the viscous terms, the
value of n=25rps; the different values of the advance
scheme remains formally second order in space.
coefficients J are obtained by changing the resultant
The simulation of the free–surface is performed by a
inflow velocity U∞=(u ∞2+v∞2) 0.5; the oblique flow
single–phase level set algorithm (Di Mascio et al 2007). An
conditions have been achieved by varying the resultant
ENO technique (similar to the one used for the bulk flow)
angle of attack in the horizontal plane x-y, namely setting
is used to solve the level set equations. The physical

299
u ∞=U∞cosβ and v∞=U∞sinβ. Unless otherwise specified, all
quantities have been cast in non-dimensional form by
using as reference values the propeller radius
Lref=0.1135m, the velocity of the blade tip
(Uref=nπD≈17.85m/s) and the density of fluid
(ρ=1000kg/m3). Therefore the non-dimensional period of
revolution is T=2π. The Reynolds number is the same of
the physical experiment carried out in pure axial flow
conditions, i.e. Rn=UrefLref/ν=1.78∙10^6.
4 MESH DETAILS
The computational mesh adopted in the numerical
simulations consists of 75 overlapping blocks for a total of
about 11.3 million volumes. A multigrid technique is
exploited in order to achieve a faster convergence of the
inner (pseudo--time) iteration, so four levels of
computational mesh are used. Each level is obtained from
the finer one by removing every other vertex along each
spatial direction. Details of the cells distribution for the
finest discretization is listed in Table 2.
M EDIU FINE Figure 3: Overview of meshing strategy
M It has to be observed that, in order to achieve a mesh
4 BLADES 0.295M 2.36M 20.9% distribution aligned with the mean flow of the propeller
HUB 0.085M 0.64M 5.6% slipstream in the horizontal plane x-y, the near wake mesh
NEAR WAKE (TIP) 0.388M 3.11M 27.5% has been set at a different (lower) angle with respect to the
NEAR WAKE 0.266M 2.13M 18.8% orientation of the inflow speed U∞ and the background.
(OTHER) Finally, the time step has been set to Δt=0.01745,
BUFFER 0.236M 1.89M 16.7% corresponding to a rotation of one degree for the propeller
BACKGROUND 0.148M 1.18M 10.4% blade, in the case of the finer grid.
TOTAL 1.415M 11.3M 100%
Table 2: Mesh details (finer grid)
The physical domain has been divided into three different
regions: propeller and near wake field, buffer region and
background. Each blade, both fore and aft parts of the hub
have been represented by an "O-grid" topology, whereas
the rest of the mesh is made up by concentric toroidal
blocks that cover the whole computational domain. For the
sake of clarity on the mesh strategy followed, two different
slices in the planes x=0 and z=0 are represented in Figure
3. In particular, cells are clustered up to 5 radii in the
downstream direction, where it can be expected that the
peculiar features of the flow be resolved accurately (tip
and hub vortices). Grid distribution is such that the
thickness of the first cell on the wall is always below 1 in
terms of wall units (y + =O(1)), the boundary layer being
described by about 32 cells. Apart from the blocks fitted
around the geometry, most of the cells are placed in the
toroidal blocks that are used to track the propeller wake;
moreover, the buffer toroidal block is used to improve grid
overlapping quality between the near wake and the
background. For the sake of completeness, the mesh on
the blade surface and on a plane cutting the blade at
r/R=0.7 are represented in

Figure 4: Mesh details on the propeller blade

5 NUMERICAL RESULTS
Numerical results are mainly focused on forces and
moments generated by the propeller; in order to explain

300
the essential features of the phenomenon, pressure In Figure 6 side force and in-plane moments are reported,
distribution on the blade and representative propeller in an alternative fashion, as percentages of longitudinal
sections during a complete cycle will be presented. thrust and torque.
Unfortunately, due to the lack of experimental data in
oblique flow conditions, validation of the numerical
results has not been performed. It has to be remarked,
however, that excellent agreement with experimental
results has been obtained by the same numerical
methodology for analyzing a similar propeller in pure axial
flow condition (Muscari & Di Mascio 2009) and behind a
fully appended hull (Muscari et al 2011), both in terms of
global loads and flow field features in the downstream
wake.
In the following description, propeller forces (pedix T) and
moments (pedix Q) have been non-dimensionalized as
follows:
Ti
K Ti 
N 2 D 4
(1)
Qi
K Qi 
N D2 5

where i indicate the axis of the inertial frame of reference.

5.1 Characterization in Oblique Flow
Propeller E779A performance in oblique flow has been
assessed on the medium grid and results are listed in
Table 3 and Table 4 in terms of absolute values of forces
and moments; the same data, for the sake of clarity, are
Figure 5: Propeller loads
plotted in Figure 5. The behavior of the propeller is
qualitatively similar for the two loading conditions: both 5.2 Blade Loads
thrust and torque increase with drift angle, and moreover, In order to provide a reliable support for synthesizing the
this being the key aspect of present investigation, relevant numerical results that will be presented, the functioning of
forces and moments are generated in the propeller plane a propeller running at incidence with respect to the flow is
due to the asymmetric inflow conditions. In particular, the described by analyzing the flow past a generic blade
propulsive components, namely thrust (KTx) and torque section during a complete revolution. To this aim,
(KQx), increase up to 65% and 21%, respectively, for the velocities in the inertial frame of reference have been
higher advance coefficient (J=0.88), and up to 40% and projected on a frame moving with the blade; as it can be
20% for the lower one (J=0.60). evidenced in Figure 7, the blade section inflow consists of
axial and circumferential components, defined as:
 KTx K Ty K Tz KQx KQy K Qz uaxial  U  cos 
(2)
0° 0.133 ≈0 ≈0 0.298 ≈0 ≈0 vtan  v  v  r  U  sin  sin 
10° 0.144 0.029 0.005 0.308 0.098 0.072 where θ is the azimuthal position, Ω is the propeller rate of
20° 0.173 0.059 0.012 0.350 0.201 0.151 revolution, and r is the radial position of the blade section.
30° 0.220 0.086 0.018 0.419 0.314 0.241 In particular, the axial component and circumferential one,
Table 3: Propeller loads; J=0.88 related to the propeller rotational velocity (i.e. Ωr), are
b KTx K Ty K Tz KQx KQy KQz constant (neglecting self induction velocity for the former
0° 0.269 ≈0 ≈0 0.054 ≈0 ≈0 one), whereas the component related to the transv erse
10° 0.281 0.02 0.003 0.515 0.083 0.072 mean flow (i.e. -U  sinβsinθ) changes during a cycle,
20° 0.299 0.04 0.007 0.541 0.167 0.149 leading to a variable (geometric) angle of attack:
30° 0.327 0.06 0.011 0.583 0.251 0.229 v
 geom    arctan tan (3)
Table 4: Propeller loads; J=0.60 u axial
Focusing on the in-plane loads, the most relevant where Ө is the airfoil geometric pitch angle.
component of the resultant force lies in the horizontal Starting from the position at θ=0°, the circumferential
plane and is oriented in the same direction of the incoming component follows a sinusoidal variation, with the
flow, whereas the vertical one is much smaller. Moreover, minimum at θ =90°, and the maximum at θ =270°. In order
moments in the horizontal and in the vertical plane to emphasize the role of the transverse component in the
(yawing and pitching moment, respectively) consist of a variation of the angle of attack, the blade section
relevant percentage of the propeller torque. hydrodynamics has been sketched in figure Figure 7 in
correspondence of two generic blade angles (θ 1 and θ 2)

301
and compared to the case of pure axial flow condition
(pedix β=0°).

Figure 8: Blade load (hub frame of reference); J=0.88

In this regard, it has to be observed that the thrust
component is mainly affected by the pressure field
developed on the blades, whereas the side force is mainly
related to viscous effects confined in the boundary layer;
as a consequence, the different time response of the
pressure variation over the blade with respect to the
boundary layer development due to the varying inflow
experienced by the rotating blade, would originate this lag
(Leishmann & Beddoes 1989).
Figure 6: In-plane loads (force and moments) ratio with
respect to thrust and torque
In the former case, the angle of attack is reduced with
respect to a pure axial flow condition because v Ω and v β
are oppositely directed, causing a reduction of the
resultant transverse component; on the other hand, for
θ=θ2 the two terms have the same sign and, consequently,
the angle of attack is greater than the case β=0°.
Figure 9: Blade load (hub frame of reference); J=0.60
The vertical component KTz is phase-shifted with respect
to the other components by a quarter of cycle (i.e. π/2); it
is worth note that, despite the fact that vertical velocity
component of the inflow is zero, the resultant average
value of KTz (smaller than the KTy component) is not, as a
result of the non homogeneous distribution of the self-
induction velocity over the propeller disk (Coleman et al
1945). In fact, the circulation convected with the trailed
and shed vortex system is different for each blade as a
consequence of the variable inflow experienced during a
revolution (see equation (3) and Figure 7) and, therefore,
the induced velocity field is distributed in a non symmetric
fashion. Moreover, because of the inclination of the
propeller slipstream with respect to the propeller rotational
Figure 7: Propeller blade section hydrodynamics axis, a negative induced velocity field (i.e. up-wash) is
As a direct consequence of the non uniform inflow, the generated over the propeller disk which further amplify the
loads generated by the blade are not constant during the non-uniform propeller loading distribution. The effect of
revolution, this being the principal reason for the presence the propeller slipstream deflection on the blade loads is
of forces and moments in the propeller plane. schematically described in Figure 10, where the
This behavior can be evidenced in Figure 8 and Figure 9, hydrodynamics of a generic blade section at the same
representing blade forces and moments in the hub frame of positions close to θ=0° and θ=180° is described.
reference (i.e., the inertial one) for J=0.88 (10°) and It is evident that the deflection of the propeller wake
J=0.60 (30°). For both loading conditions, the wave form causes negative induced velocities u i in the outer portion
of the loads mirrors the harmonic variation of the inflow. of the first and fourth quadrant of the propeller disk; on
The thrust KTx and the lateral force KTy developed during the contrary, the self-induction is more concentrated on
the blade passage on the upper half of the disk the opposite side, where the vortex tube is deflected. As a
(0°<θ<180°) are lower with respect to those developed in consequence, in the former position, the up-wash further
the lower half; moreover, close to θ=270°, their maximum amplifies the effects of the transverse component v β, i.e.,
values show a slight phase shift. the angle of incidence is further increased. The opposite
trend is schematized near θ=180°, where the angle of

302
attack is reduced due to the increase of the axial velocity following discussion is centred on the same conditions
component. analyzed with the finest grid resolution (J=0.88 and
J=0.60 at incidence of 10° and 30°, respectively).

Figure 10: Effect of propeller wake deflection

It is evident that the deflection of the propeller wake
causes negative induced velocities u i in the outer portion
of the first and fourth quadrant of the propeller disk; on
the contrary, the self-induction is more concentrated on
the opposite side, where the vortex tube is deflected. As a
consequence, in the former position, the up-wash further
amplifies the effects of the transverse component v β, i.e., Figure 12: Pressure field; J=0.88
the angle of incidence is further increased. The opposite In Figure 12 and Figure 13 the pressure generated on the
trend is schematized near θ=180°, where the angle of propeller blades, at prescribed circumferential locations,
attack is reduced due to the increase of the axial velocity are visualized for the two advance coefficients in terms of
component. CP =p/0.5ρU2ref. It has to be pointed out that blade
In Figure 11 blade loads components in a reference frame positions are represented in the same frame of reference
fixed to the blade are reported. It is worthy of notice that considered for the hub loads (Figure 8 and Figure 9). In
the "spindle" torque, namely the torque oriented along the both cases, the pressure field has a behavior consistent
blade span direction, denotes a wavy variation with an with the trend of the angle of attack defined in equation
higher harmonic content than the forces. (3). At θ=90°, the blade is in the lowest loaded condition
and this is confirmed by the relatively small intensity of
the pressure on its surface; on the contrary, at θ=270° the
pressure intensity is highest on relevant portion of the
pressure and suction sides of the blade. Considering the
two horizontal positions, θ=0° and θ=180°, slight
asymmetries in the pressure field are evident at J=0.88 on
the pressure side. On the other hand, with the increasing
of propeller loading, the asymmetry of pressure
Figure 11: Blade loads (rotating frame of reference) distribution with respect the z axis is more evident; in
particular, at θ=0°, the pressure intensity on the front and
Moreover, the two loading conditions have different
back side of the blade is higher, and moreover, extended
characteristics: at the lowest J, the moment experiences
on a wider portion, with respect the one experienced at
two different sign inversions for two short intervals,
θ=180°. This general overview of the pressure
probably because at this conditions the centre of pressure
distribution further explains the nature of the in -plane
of the blade is located at an higher distance from the
loads (in the hub frame of reference) developed by a
leading edge with respect to the J=0.88 case.
propeller working in oblique flow: the different pressure
Pressure field description in next section will provide a
distributions along the vertical axis generate the positive
further support and a finer overview of the different
side force and pitching moment; on the other hand, the
hydrodynamic behavior of the blades at both advance
pressure distribution along the horizontal axis generates
coefficients.
the vertical force and the yawing moment.. Moreover, it
5.3 Pressure field should be stressed that the former effect is mainly due to
The pressure field developed on the blade and on the lateral component of the inflow, whereas the latter one
representative blade sections during a revolution is is a consequences of the three--dimensional self-induction
described in order to gain a better insight into the nature effects, as described in the previous section.
of the loads generated in oblique flow conditions. The

303
 coarse 0.139 0.033 0.003 0.328 0.100 0.065
medium 0.144 0.029 0.005 0.307 0.098 0.072
fine 0.142 0.029 0.007 0.295 0.097 0.073
Table 5: Convergence results; J=0.88
GRID K Tx KTy KTz 10K Qx 10K Q 10KQz
y
coarse 0.031 0.065 0.005 0.579 0.253 0.208
medium 0.327 0.06 0.011 0.583 0.251 0.23
fine 0.325 0.057 0.012 0.565 0.251 0.229
Table 6: Convergence results; J=0.60
load medium fine E Extr. U N%
K Tx 0.144 0.141 2.2E-3 0.139 1.58
K Ty 0.029 0.029 0.0 0.029 0
K Tz 0.005 0.007 -1.2E-3 0.008 -14.63
10KQx 0.307 0.295 1.27E-2 0.282 4.5
10KQ 0.098 0.097 2.0E-4 0.097 0.21
y
10KQz 0.072 0.073 -1.5E-3 0.075 -2
Figure 13: Pressure field: J=0.60
Table 7: Grid convergence analysis; J=0.88
5.4 Verification
As a final remark, it can be said that the propeller loads
For the two test conditions analyzed, computations have
computed via the medium grid are very close to the finer
been performed considering three different mesh
grid results, and, for the purpose of performance
resolutions, in order to get a rigorous evaluation of the
prediction, can be considered reliable.
numerical uncertainty. In Table 5 and Table 6 propeller
load medium fine E Extr. U N%
loads are summarized for three grid levels. When following
the procedure suggested in (Roache 1997), grid K Tx 0.327 0.325 1.6E-3 0.323 0.49
convergence results with three grid level were poor, K Ty 0.06 0.057 2.5E-3 0.055 4.55
oscillatory convergence being observed in some cases. K Tz 0.011 0.012 -8E-4 0.013 -6.2
This is probably to be ascribed to the inability of the 10KQx 0.583 0.565 1.85E-2 0.546 3.39
coarse grid to resolve the main flow features. As a 10KQ 0.251 0.251 1E-4 0.251 0.04
consequence, only results on the medium and fine grids y
have been taken into account to estimate grid 10KQz 0.229 0.228 1E-3 0.227 0.44
convergence. In particular, the solution variation has been Table 8: Grid convergence analysis; J=0.60
computed as:
6 CONCLUSION
The INSEAN E779A propeller in oblique flow conditions
E   f 2  f1  /(1  r ) p
(4)
has been investigated by means of a dynamic overlapping
grid technique; in particular, a relative wide range of angle
where f1 and f2 are the solution on the fine and medium of incidence (up to 30°) has been considered for two
grid, r the refinement factor (r=2, as stated before) and p is propeller loading conditions, namely at moderate (J=0.88)
the "formal" order of accuracy (p=2, for the present and at medium-high (J=0.60) loading. The focus has been
solver). Then, the uncertainty is evaluated by: mainly centred on the behavior of the propeller in terms of
U N  FS E (5) global loads. As a result of the non symmetric inflow
where FS is a safety factor assumed equal to 3, according condition, relevant in--plane loads, which are usually not
to Roache. of concern in pure axial flow condition or in hull--behind
Outcome of the grid convergence is summarized in Table 7 condition in straight ahead motion, are generated by the
and Table 8: the uncertainty is lower than 5% for the propeller. In particular, in a common off-design scenario of
propeller loads at both conditions, with the exception for ship operation (i.e., maneuvering), these components may
the vertical component KTz; it has to be pointed out, increase up to 20% of the thrust and 40% of the torque, for
however, that this component is much smaller than the the side force and in plane moments (pitching and
others, and consequently, the numerical solver and the yawing), respectively. The description of the pressure
grid strategy adopted can be regarded as successful in developing on the blade during a complete revolution
capturing the essential peculiar phenomena. further supported the comprehension of force generation.
Unfortunately, due to the lack of experiments in oblique
GRID KTx KTy KTz 10K Qx 10KQ 10K Qz flow conditions, a validation of the numerical
y computations has not been carried out; in this regard,

304
however, it has to be remarked that the very accurate Di Mascio, A., Broglia, R. & Muscari, R. (2008). ‘Numerical
results obtained in open water tests (axisymmetric flow) simulations of viscous flow around a naval combatant
and the low uncertainty evaluated by means of two grid in regular head waves’. Proceedings of 6th Osaka
levels in oblique test, can be considered a valuable Colloquium on Seakeping and Stability of Ships ,
support for the accuracy of present computations. Osaka, Japan.
ACKNOWLEDGMENTS D. Durante, R. Broglia, R. Muscari, A. Di Mascio, (2010),
This work was partially supported by the Italian Ministry ‘Numerical Simulations of a turning circle manoeuvre
of Education, University and Research through the for a fully appended hull’, Proceedings of 28th
research project RITMARE. Numerical computations Symposium on Naval Hydrodynamics , Pasadena,
presented here have been performed on the parallel California.
machines of CASPUR Supercomputing Center (Rome); J. G. Leishmann, T. S. Beddoes, (1989) ‘A semi-empirical
their support is gratefully acknowledged. model for dynamic stall’, Journal of American
REFERENCES Helicopter Society 34 (3) pp.3-17.

P. Atsavapranee, R. Miller, C. Day, J. Klamo, D. Fry, (2010), Muscari, R. & Di Mascio, A. (2005). ‘Simulation of the flow
‘Steady-Turning Experiments and RANS simulations around complex hull geometries by an overlapping grid
on a Surface Combatant Hull form (Model#5617)’. approach’. Proceedings 5th Osaka Colloquium on
Proceedings of 28th Symposium on Naval advanced research on ship viscous flow and hull form
Hydrodynamics, Pasadena, California. design by EFD and CFD approaches , Osaka, Japan.

Baldwin, B. & Lomax, H. (1978). ‘Thin-layer approximation Muscari, R., Broglia, R. & Di Mascio, A. (2006). ‘An
and algebraic model for separated turbulent flows’. overlapping grids approach for moving bodies
AIAA Paper 1978-257, 16th Aerospace Sciences problems’. Proceedings of the 16th International
Meeting, Huntsville, Alabama, USA. Offshore and Polar Engineering Conference (ISOPE
2006), San Francisco, California, USA.
R. Broglia, G. Dubbioso, A. Di Mascio, (2011), Prediction
of Manoeuvering Properties for a Tanker Model by R. Muscari, M. Felli, A. Di Mascio, (2011) ‘Analysis of the
Computational Fluid Dynamics’, Proceedings of Flow Past a Fully Appended Hull with Propellers by
Specialist Meeting on Assessment of Stability and Computational and Experimental Fluid Dynamics’,
Control Prediction Methods for NATO Air and Sea Journal of Fluid Engineering 133, pp.1-16.
Vehicles (AVT-189), UK. R. Muscari, A. Di Mascio, (2009) ‘Simulation of the
R. Broglia, G. Dubbioso, D. Durante, Di Mascio A., (2011), viscous ow around a propeller using a dynamic
‘Simulation of Turning Circle by CFD: Analysis of overlapping grid approach’ Proceedings of First
different propeller models and their effect on International Symposium on Marine Propulsors
manoeuvring prediction’, Applied Ocean Research 39 (SMP'09), Trondheim, Norway.
(1) pp.1-10. H. Ribner, (1943), ‘Propeller in yaw’, NACA TECHNICAL
Chang, K. C., Hsieh, W. D. & Chen, C. S. (1995). ‘A REPORT.
Modified Low–Reynolds–Number Turbulence Model P. J. Roache, (1997) ‘Quantification of Uncertainty in
Applicable to Recirculating Flow in Pipe Expansion’. Computational Fluid Dynamics’, Annual Review Fluid
ASME Journal of Fluids Engineering 117, pp. 417–423. Mechanics 29 pp.123-160.
R. Coleman, A. Feingold, C. Stempin, (1945) ‘Evaluation of Smagorinsky, J. (1963). ‘General circulation experiments
the induced velocity fields of an idealized helicopter with the primitive equations. I. The basic experiment’.
rotor’, NACA REPORT TR-29, 1945. Monthly Weather Review 91(3), pp. 99–164.
Di Mascio, A., Broglia, R., Muscari, R. & Dattola, R. (2004). P. R. Spalart, S. R. Allmaras, (1994), ‘A One-Equation
‘Unsteady RANS Simulation of a Manoeuvring Ship Turbulence Model for Aerodynamic Flows’, La
Hull’. Proceedings of the 25th Symposium on Naval Recherche Aerospatiale 1 pp.5-21.
Hydrodynamics, St. John’s, Newfoundland and M. Viviani, C. Podenzana Bonvino, S. Mauro, M. Cerruti,
Labrador, Canada. D. Guadalupi, A. Menna, (2007), ‘Analysis of
A. Di Mascio, R. Broglia, R. Muscari, (2007) ‘On the Asymmetrical Shaft Power Increase During Tight
application of the One{Phase Level Set Method for Manoeuvre’, Proceedings of 9th International
Naval Hydrodynamic Flows’, Computer and Fluids 36 Conference on High Performance Marine Vehicles
(5) pp.868-886. (FAST), Shangai, China.

305