See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/283291049

Evaluation of the Propeller Hull Vortex Using a RANS Solver

Conference Paper · June 2011

CITATIONS READS

0 58

5 authors, including:

Jussi Martio Tuomas Sipilä

VTT Technical Research Centre of Finland ABB Marine and Ports
14 PUBLICATIONS   39 CITATIONS    20 PUBLICATIONS   101 CITATIONS   

SEE PROFILE SEE PROFILE

Antonio Sánchez-Caja Ilkka Saisto

VTT Technical Research Centre of Finland Aker Arctic Technology
36 PUBLICATIONS   251 CITATIONS    14 PUBLICATIONS   91 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PropNoise View project

Cavitating propeller flows View project

All content following this page was uploaded by Antonio Sánchez-Caja on 24 October 2017.

The user has requested enhancement of the downloaded file.

 Second International Symposium on Marine Propulsors
smp’ 11, Hamburg, Germany, June 2011

Evaluation of the Propeller Hull Vortex using a RANS solver

Jussi Martio1, Tuomas Sipilä1, Antonio Sanchez-Caja1, Ilkka Saisto1,Timo Siikonen2

1
VTT Technical Research Centre of Finland
2
Aalto University

ABSTRACT regions of the flow field, and the so-called pirouette

The appearance of the propeller hull vortex (PHV) has effect. The pirouette effect emerges as the propeller
been investigated using the FINFLO RANS solver. This works in a loaded condition and the tip clearance is
specific phenomenon occurs when solid surfaces placed small. This may result in situations where the blade’s
in the neighbourhood of the propeller prevent the natural suction side gets starved of water. The pirouette effect
flow contraction resulting from the local flow was considered to be the dominant source for the
acceleration. As the propeller blades near the solid inception of the PHV.
surfaces may not obtain enough mass flow from the inlet A more systematic investigation was carried out by (Sato
– i.e. the propeller is starved of water - the flow may et al, 1986). The procedure included experiments with
reverse locally so that water is sucked from the outlet. six model propellers. A flat plate representing the hull
The small tip clearance combined with the low advance surface was located above the propeller. The tip
number as well as the geometry of the solid surfaces may clearance was altered by moving the propeller vertically.
affect the inception of the PHV. Air bubbles were injected into the flow field in order to
Time-dependent computations were carried out at seven visualize the streamlines on the plate.
advance numbers. During the analysis of the results
According the observations, the flow patterns were
several quantities were examined to detect the inception
classified into six specific categories, as described in Fig.
of the PHV. Streamlines were found to be the best
1. First, the air bubbles are moving uniformly to the
technique to trace the vortices from an extremely
downstream for the plate located far enough away from
complex flow field. The flow patterns on the flat plate
the propeller (Non-reverse flow). Once the plate
were compared to published experimental observations
approaches the blade tips, the flow starts occasionally to
using the limiting streamlines along the plate. Flow
reverse. Still, no vortex is generated under this pattern
patterns similar to those discovered in the experiments
(No-vortex reverse).
were obtained also by RANS simulations, so the RANS
method is able to predict the occurrence of the PHV at
least on a qualitative level.
Keywords
Propeller-Hull Vortex, URANS
1 INTRODUCTION
Propeller-hull vortex (PHV) cavitation is a specific type
of cavitation related to a stationary vortex between the
propeller tip and the hull. The favourable conditions for
the generation of the PHV are typically low advance
numbers J, small clearance between the blade tip and the
hull surface, and hull geometry. Flat plate geometry is
presumed to produce more favourable conditions for the
PHV than curved geometry (Carlton, 2007).
Traditionally, PHV cavitation has been studied by means
of experimental modelling and full-scale observations.
The PHV phenomenon was first reported by (Huse,
1971). Four hypotheses were introduced as theoretical
Figure 1. Bubble trajectories on the plate as tip clearance
groundings for PHV cavitation. These hypotheses
ratio c/D is varied (Sato et al, 1986)
included the starting vortex, the vortices created by the
shear flow in the wake field, the vortices created in other
At the next stage the reverse flow becomes stable and a 2 NUMERICAL METHOD
vortex can be detected just above the propeller. This flow The flow simulation in FINFLO is based on the solution
pattern is classified as the ‘Aft vortex flow’ situation. of the RANS equations by the pseudo-compressibility
For right-handed propellers this specific vortex is method. FINFLO solves the RANS equations by a finite
rotating in a clockwise direction. As the c/D-ratio is volume method. The solution is extended to the wall and
decreased to a magnitude of 0.11, a ‘Double vortex flow’ is based on approximately factorized time integration
-flow field can be detected. Two separate vortices are with local time stepping. The code uses either Roe's flux
originated from the plate so that finally they are located difference splitting or Van Leer's flux vector splitting for
side by side as shown in Fig. 1. A counter-clockwise compressible flows and an upwind-based scheme for
vortex located on the port side of the clockwise rotating incompressible flows. In the latter case, the pressure is
vortex is generated so that the flow image is no longer centre differenced and a damping term is added via a
symmetrical with respect to the centreline. The convective velocity. A multigrid method is used for the
instantaneous flow pattern actually looks like a vertical acceleration of convergence. Solutions on coarse grid
section of a mushroom. levels are used as the starting point for the calculation in
The ‘Fore vortex flow’situation emerges as the counter- order to accelerate convergence. For the time-accurate
clockwise vortex in this case moves close to the computations the interface between the rotating propeller
clockwise vortex. Finally, the vortices unite once the grid blocks and the surrounding stationary blocks is
counter clockwise vortex absorbs the clockwise vortex. treated using a sliding mesh technique, where the grid
lines across the block interface are discontinuous. The
The last flow pattern is achieved at c/D=0.013. This solution for a block is interpolated using the solution in
specific phenomenon was called the ‘Splatter flow’. A the neighbouring block. A mass-conserving interpolation
strong jet flow is formed downstream from a point just is made between the connecting blocks at every time step.
above the propeller tip, as the blade is very close to the A detailed description of the numerical method including
flat plate. A counter-clockwise vortex is generated on the discretization of the governing equations, solution
starboard side of the jet flow. This particular vortex was algorithm, etc. can be found in (Sanchez-Caja et al.,
specified as the ‘splatter vortex’. 1999 and 2000). Chien’s k-epsilon turbulence model was
In Fig. 2. the different flow patterns are shown as a used in the calculation.
function of tip clearance ratio c/D and load factor CT Unsteady flows can be approximated using mixing plane
8 ⋅T 8 ⋅ KT and quasi-steady methods. In the mixing plane approach
CT = = (1) the flow quantities for both the rotating and non-rotating
πρD 2U 2 πJ 2
blocks are circumferentially averaged on both sides of the
where J denotes the advance number, T is the thrust, ρ is common interface and then transferred to the ghost cells
the density, D is the propeller diameter, U is uniform as boundary values. In the quasi-steady approach the
velocity and KT is the thrust coefficient. The suction rotating and non-rotating blocks are connected without
limit and the vortex limit are drawn to separate ‘Non- any averaging process. The simplified approaches may
vortex flow’, ‘No-vortex reverse flow’and ‘Aft vortex give sensible results when the interaction between the
flow’from each other as explained. rotating and stationary domains is weak. For axial flows,
even though the velocities everywhere are expressed in
inertial coordinates, the flow disturbances are convected
using different reference frames on the blocks at each
side of the interface. The transition of the flow at the
interface should be smooth without interferences of solid
boundaries located nearby. In this investigation a time-
accurate approach was considered more appropriate to
avoid interference of the flat plate.
FINFLO is also extended to model two-phase flows
together with the cavitation model. The cavitation model
is based on either Merkle’s or Kuntz’s mass transfer
model (Sipilä et al, 2009). The cavitation model is not
applied in the present study.

3 CALCULATIONS
3.1 Propeller and grid specifications
Figure 2. Flow patterns on the plate as a function of c/D The INSEAN E779a propeller was selected as the
and CT (Sato et al, 1986) reference case for the study. The main characteristics of
the model propeller are given in Table 1, while the
surface grid of the propeller is shown in Fig. 3.
 Examples of convergence histories of chosen quantities
are presented in Fig. 5. The quasi-steady computation
was stopped after approximately 20,000 iterations.

Figure 3. The schematic view of the INSEAN E779A

propeller
Figure 4. Schematic view of the asymmetrical grid.
Table 1: Basic data of the INSEAN E779A model
propeller
Propeller diameter D 227.27 mm
Number of blades 4
Rotation Right-handed
Pitch ratio P/D 1.1

The flat plate was located at a given distance from the

propeller tip. The grid configuration is shown in Fig. 4.
The main parameters of the computation are presented in Figure 5. Example of convergence histories of the
Table 2. computations for the J=0.099. From left to
right: L2 norms for the momentum residuals
Table 2. Main parameters used in the study (time- and L2 norms for k/e residuals
accurate computation)
Number of elements ~1.1 million 4 FLOW ANALYSIS
Tip clearance c/D 0.157 4.1 Analyzing process
The most difficult part of the analysis was the
Advance numbers time 0.050, 0.099, 0.249, visualization of single vortices, which was a consequence
accurate 0.326, 0.433, 0.532, 0.606 of the complexity of the global flow field. Such
load factor CT (exp.) 2.0, 2.9, 5.0, 10..0, 18.6, complexity grows as the advance number of the propeller
132.0,. 531.7 decreases. In other words, an individual vortex can be
propeller rotational rate (rps) 11.8 quite difficult to isolate from the global flow field. The
vortices of interest are not stable, but strongly time-
time step [s] 0.0001177 dependent. The size of the vortex compared to the global
flow field can also be considered minimal.
The propeller-hull vortices were traced first by the
pressure and velocity fields. Both absolute and relative
3.2 Time accurate computations
velocities were used in this process. Furthermore, the
The computations were started with the quasi-steady
vorticity isosurfaces were also displayed in order to detect
approach. Once sufficient convergence was reached in
possible PHV formation. All these quantities were found
the quasi-steady solution the computation was continued
to be inconvenient for visualization purposes, mainly
in a time-accurate manner.
because the suitable range for the specific quantity was
found to be too sensitive to be selected properly.
A more convenient technique to display the PHV in Fig. 6c. Clearly, a distinguished vortex is located just
formation was to detect the limiting streamlines on the between the propeller and the plate. Once the loading is
plate in a similar manner as sketched in Fig. 1. The increased so that CT reaches a value of 10.0, the ‘Double
origin of the PHV can be explored with the streamlines vortex flow’condition appears. This is demonstrated also
as well as its general shape. in Fig. 6d. Furthermore, the same pattern is even clearer
The most detailed information about the structure of the with CT=18.6. Finally, an instantaneous ‘Fore vortex
flow pattern was obtained by searching the location of flow’ situation emerges as the load factor is set to a
PHV with the limiting streamlines and using the magnitude of over 130. On average the evaluated flow
instantaneous location of the vortex as the origin of the patterns follow very well the experimental observations.
streamlines in the 3D-flow field. This technique also The suction and vortex limits defined by (Sato et al,
reveals whether or not the PHV is attached to the 1986) are produced reasonably well by the RANS
propeller blades. method.
The limiting streamlines produced for four rotation
4.1 Results
angles are presented in Fig. 7 at advance number
A summary of the findings is shown in Fig. 6. In
J=0.433. It may be concluded that the origin of the vortex
principal the flow should follow the patterns illustrated
remains quite steadily on the flat plate, but the form of
in Fig. 2 with c/D=0.157 for each CT. According to Fig.
the vortex changes as the blades encounter the vortex.
3., the flow should remain as a ‘Non-reverse flow’at a
CT value of about ~3. This is illustrated in Figs. 6a and
6b, since the flow field remains uniform with CT=2.0 but
can be categorized as a ‘No-vortex reverse flow’with a
load factor magnitude of 2.9. The flow starts to swing to
the port side of the propeller; still a distinct vortex cannot
be isolated from the flow field.

a) b)

c) d)
Figure 7. From left to right and top to bottom: The
limiting streamlines on the plate with
propeller rotation angles 0 deg, 25 deg, 40
deg and 70 deg. The advance number
J=0.433. 0 deg refers to the blade at the 12
o’clock position.
e) f) The computed KT as a function of the propeller rotation
angle is presented in Fig. 8. The oscillation of the thrust
increases as the advance number is decreased. On the
other hand, the evaluated time-averaged thrust coefficient
remains almost constant in the advance number range
0.050 to 0.099. Also, small phase shifts in oscillations
can be noticed between these advance numbers. At all
Figure 6. From left to right and top to bottom: The advance numbers the force and moment oscillations take
limiting streamlines on the plate with advance place at the blade passing frequency corresponding to 90
numbers 0.606, 0.532, 0.433, 0.326, 0.249, deg rotation.
and 0.050. The corresponding load factors
are 2.0, 2.9, 5.0, 10.0, 18.6 and 531.5. The The oscillation amplitudes of KT reduce significantly
positive x-axis points in the downstream between advance numbers 0.249 and 0.326, as presented
direction in Fig. 8. Still, both situations produce the ‘Double
vortex flow’condition, as shown in Figs. 6d and 6e, on
The ‘Aft vortex flow’situation should emerge once the the flat plate. Fig. 9 shows a more detailed illustration of
load factor reaches an approximate magnitude of 5. The the structure of the PHV. The ‘Double vortex flow’
limiting streamlines for this specific situation are shown formation with two distinct vortices is detected at
J=0.249. The vortices rotating in the clockwise and As the loading on the propeller is decreased, that is, the
counter-clockwise directions are interacting both with the advance number is increased to 0.326, the propeller-hull
blades and the tip vortices, depending on the propeller vortices do not seem to impinge on the blades. This is
rotation angle. Furthermore, in this particular case the the probable explanation also why the fluctuation of the
vortices are also attached to the low-pressure region near thrust coefficient strongly reduces between these advance
the leading edge. numbers.
0.5
0.050
0.099
Finally, the vortex structures of the ‘Aft vortex flow’-
0.249
0.326
condition are studied in Fig. 11. The ‘Vortex limit’was
0.433
0.45
0.532
0.606
defined to separate the ‘Aft vortex flow’and the ‘No-
vortex reverse flow’-situations from each other.
0.4

The traced streamlines at J=0.433 illustrate that also for

KT

0.35
this case the generated vortices emerging from the plate
do not interact with the blades at any position. That is,
0.3
the streamlines denoted with red colour are presumably
affected by the tip vortices, but they do not impinge on
0.25
0 200 400 600 800 1000 1200
the blades for any time step. Furthermore, the ‘Double
Angle [deg]
vortex flow’ -situation is formed at the propeller
0.45
0.249
rotational angle 70 deg. In this particular case the
0.44
0.326
vortices bypass the blades towards the downstream
0.43
direction.

0.42
KT

0.41

0.4

0.39

0.38
0 200 400 600 800 1000 1200
Angle [deg]

Figure 8.Top: The thrust coefficient KT for the time-

accurate computations as a function of rotation
angle for all calculated advance numbers.
Below: The oscillation of KT for cases J=0.249
and J=0.326

Figure 9: Top: the distribution of CP on the suction side Figure 10: Top: the distribution of CP on the suction side
and streamlines with propeller rotation angle 0 and streamlines with propeller rotation angle 0
deg. Below: the same for rotation angle 25 deg. deg. Below: the same for rotation angle 25
The advance number is J=0.249. deg. The advance number is J=0.326
 the RANS method was considered to be satisfactory. The
time-dependent phenomena were also reproduced
reasonably well. It could be stated that the time-accurate
viscous flow solver was able to predict the occurrence of
the propeller-hull vortex on a qualitative level reasonably
well. The detailed analysis of the flow structure also
revealed that before PHV inception a vorticial structure
can be seen in the vicinity of the plate. In this initial
situation the vortex is not yet impinging on the blade.
Since the fundamental mechanisms causing the PHV can
be modelled at least with some accuracy using the RANS
method, the use of cavitation models would be the next
step to improve the simulations at full scale. Also, the
combined method using the vorticity, the helicity, or the
mass flow together with the streamlines could be tested
in the future to detect the ‘pirouette effect’ and to
Figure 11: From left to right and top to bottom – the measure the strength of the PHV.
streamlines and the distribution of CP on the
suction side with propeller rotation angles 0 ACKNOWLEDGMENTS
deg (both top figures), 40 deg and 70 deg. This work has been was conducted in VIRKOOT-project
The advance number is J=0.433 sponsored by TEKES, the Finnish Funding Agency for
Technology and Innovation. The authors are indebted to
5 CONCLUSIONS
Dr. Jaakko V. Pylkkänen, who proposed PHV inception
The fundamental mechanisms causing the propeller-hull as a topic for CFD research before his retirement.
vortex were modelled using the time-accurate RANS
solver. Both steady-state and time-accurate approaches
were utilized in the investigations. REFERENCES
The parameters affecting the strength of the PHV are Carlton J., Marine Propellers and Propulsion, ISBN: 978-
blade loading and the tip clearance. The present 07506-8150-6, 2nd edition 2007
computations were carried out using seven separate
advance numbers and keeping the tip clearance constant. Huse, Propeller-Hull Vortex Cavitations Norw. Ship
Model Exp. Tank Publ. No. 106. May 1971.
The basic phenomena related to the so-called pirouette
effect were detected at least with qualitative accuracy. As Sato R. et al., Observation of Flow on a Horizontal Flat
the tip clearance between the propeller blade tip and the Plate above a Working Propeller and Physics of
wall is decreased, the blade’s suction side does not obtain Propeller-Hull Vortex Cavitation, Proceed. Internat.
enough quantity of water from the inlet side. That is, Symposium on Propeller and Cavitation, Wuxi,
water is also sucked from the downstream, causing the China, 1986
pirouette effect and the finally PHV inception. Sánchez-Caja, A., Rautaheimo, P., Salminen, E., and
The detection and measurement of the pirouette effect’s Siikonen, T. (1999). ‘Computation of the
strength were found to be major issues. Several flow Incompressible Viscous Flow around a Tractor
quantities were tested, and finally the limiting Thruster Using a Sliding Mesh Technique’.
streamlines on the flat plate were found to be the most Proceedings of the 7th International Conference in
appropriate method of tracing the vortices. Once the Numerical Ship Hydrodynamics, Nantes (France).
location of the vortex or vortices was determined on the Sánchez-Caja, A., Rautaheimo, P. and Siikonen, T.
plate, a set of streamlines departing from this region (2000). ’Simulation of Incompressible Viscous Flow
could be created. This technique allowed the structure of Around a Ducted Propeller Using a RANS Equation
vortex mechanism to be studied in a more detailed Solver’, 23rd Symposium on Naval Hydrodynamics,
manner. Unfortunately, this technique does not provide Val de Reuil (France).
any specific method to evaluate the strength of the Sipilä T. et al, Cavitating Propeller Flows Predicted by
vortex. RANS Solver with Structured Grid and Small
The flow patterns divided into six categories by (Sato et Reynolds Number Turbulence Model Approach,
al, 1986) were simulated and the agreement between the Proceedings of 7th International Symposium on
experimental observations and the results produced by Cavitation Nr. Paper 45, 1-11, 2009

View publication stats