See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/272021083

Improving Propeller Efficiency Through Tip

Loading

CONFERENCE PAPER · NOVEMBER 2014

CITATION READS

1 212

8 AUTHORS, INCLUDING:

Michael Brown Antonio Sánchez-Caja

George Washington University VTT Technical Research Centre of Finland
1 PUBLICATION 1 CITATION 28 PUBLICATIONS 45 CITATIONS

SEE PROFILE SEE PROFILE

Phillip Duerr Ilkka Saisto

Naval Surface Warfare Center - Carderock Div… Aker Arctic Technology
2 PUBLICATIONS 1 CITATION 14 PUBLICATIONS 18 CITATIONS

SEE PROFILE SEE PROFILE

Available from: Antonio Sánchez-Caja

Retrieved on: 14 March 2016
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

Improving Propeller Efficiency Through Tip Loading

Michael Brown1, Antonio Sánchez-Caja2, Juan G. Adalid3, Scott Black1,
Mariano Pérez Sobrino3, Phillip Duerr4, Seth Schroeder1, Ilkka Saisto2
(1 Naval Surface Warfare Center, Carderock Division, U.S.A., 2 VTT
Manufacturing Technology, Finland, 3 Sistemar, Spain, 4Florida Atlantic
University, U.S.A)

improved efficiency (4-8%), reduced cavitation, and

ABSTRACT reduced hull vibration (Adalid and Gennaro, 2011).
Designs of TLPs require careful evaluation and
This paper examines the potential performance gains
unconventional scaling approaches to meet their
of tip loaded propellers (TLP) in general and presents
performance goals at full scale.
specific findings for a sample case. Potential flow
simulations, Reynolds-Averaged Navier-Stokes This paper focuses on the assessment of
(RANS) viscous flow simulations and experimental three distinct propellers, all of which are designed for
methods were used to assess tip loaded propellers the same requirements. A parametric study process
designed by the authors from the Naval Surface developed by NSWCCD is described that provides an
Warfare Center, Carderock Division (NSWCCD) and initial look at the effects on propeller efficiency from
the Sistemar company, for the same performance pressure side and suction side tip features. This
requirements using different design approaches. A parametric design space exploration method is how
reference design of a conventional open propeller NSWCCD developed the propeller particulars for the
was also developed and assessed using potential flow TLP analyzed during this study.
and viscous RANS methods. The differences in
performance between model and full scale were Afterwards, the requirements for the
studied using a viscous flow RANS solver. The comparative design project between the Sistemar
sources of the scaling effects are identified and company and the USN’s propulsor design branch at
differences between the scaling of conventional and NSWCCD are described. The requirements were
TLPs are examined. Both TLPs designed by selected to emulate a notional merchant ship
NSWCCD and Sistemar are predicted by the RANS propeller and included structural, powering and
flow solver to be more efficient than the conventional cavitation requirements.
propeller at full scale. Relative to the conventional
propeller, the NSWCCD-designed TLP is predicted Two of the designs (the Sistemar CLT and
to have higher efficiency at the design point as well the NSWCCD TLP) were manufactured as 250mm
as increased cavitation inception speed. (1/26th scale) models and tested for open water and
cavitation performance at the Canal de Experiencias
Hidrodinámicas de El Pardo (CEHIPAR) in Madrid,
INTRODUCTION Spain. The computational results are compared with
Improving propulsive efficiency is a perpetual goal the experimental data and the correlation is
for naval architects and engineers. Contracted discussed.
Loaded Tip (CLT®) (Adalid and Gennaro, 2011) and
Kappel propellers (Andersen et al., 2005) have been Two different RANS solvers were used to
demonstrated to improve propeller efficiency by analyze the propellers. The simulation results for
shifting the spanwise circulation distribution towards open water performance in terms of thrust and torque,
the tip of the propeller blade while using geometric blade surface pressures and streamlines, and
features to prevent losses over the tip. In principle, Reynolds number scaling corrections are compared.
this concept is similar to how airplane winglets
improve lift-to-drag ratio. The CLT concepts,
developed in Spain over the last several decades,
have been applied to more than 280 ships with

1
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

DESIGN SPACE EXPLORATION The design points analyzed included five

values of propeller thrust loading coefficient (CT),
In this section, an initial assessment and design four advance coefficients (JA), five circulation
process developed at NSWCCD for TLPs is distributions (G), and 80 hydrorake distributions,
described. This includes modifying in-house tools where:
used for geometry definition and potential flow based
analysis for use with the unique geometry associated 𝑇
𝐶𝑇 =
with the tip loaded concept. The goal of the TLP 1
𝜌(𝑉𝐴 )2 𝜋𝑅2
design was to increase propeller efficiency without 2
sacrificing other performance metrics such as
cavitation, pressure pulse, thrust breakdown, 𝑉𝐴
𝐽𝐴 =
unsteady forces and structural strength. 𝑛𝐷

Propeller blades with winglet features such Γ

𝐺=
as TLPs do not lend themselves to traditional ITTC 2𝜋𝑅𝑉𝐴
geometry definitions. For modeling single-sided
winglets, a new geometric parameter called Additionally, three blade area ratios (EAR) were
hydrorake is defined at NSWCCD, which is a rake- analyzed to investigate the frictional losses associated
like displacement perpendicular to the local propeller with the blade area. At each point in the parametric
pitch. Hydrorake is defined as: study, a design was completed for comparison with
the baseline.
𝑇𝑜𝑡𝑎𝑙 𝑅𝑎𝑘𝑒 1 𝑟
ℎ𝑦𝑑𝑟𝑜𝑟𝑎𝑘𝑒 = � − 𝑠𝑘𝑒𝑤 ∙ ∙ The family of circulation distributions is
𝐷 2 𝑅
shown in Figure 1. They represent a series with
∙ tan(𝜙)� ∙ 𝑠𝑖𝑛(90 − 𝜙), progressively increased loading at the tip, with the
G5 distribution having a localized increase at the tip.
where R and D are propeller radius and diameter Examples of representative blade shapes for positive
respectively, skew is projected skew in radians, r is and negative hydrorake are shown in Figure 2 and
local radius, and ϕ is local propeller pitch angle. Figure 3, respectively.
Sistemar’s CLT propeller is an example of positive
hydrorake, while the Kappel propeller is an example
of negative hydrorake. The traditional airfoil profiles
defined by thickness and camber distributions were
changed from their traditional constant radius
application to be oriented in a plane perpendicular to
the mid-chord reference line. Spanwise distributions
are defined along the normalized arc length of the
reference line as span instead of radius.

NSWCCD’s lifting-surface design, lifting-

surface analysis and panel-method analysis tools, that
are descendants of codes developed by Kerwin et al.
(2006), Hsin et al. (1991) and Kinnas et al. (1997),
were modified to utilize these geometric models and
used to explore the design space of TLPs.

A parametric study was performed to

explore the efficiency benefits of the tip loaded Figure 1: Circulation distributions considered during
concept across the design space. A nominal parametric study.
merchant ship propeller was selected as the baseline
design for comparison. This propeller is a typical 5-
bladed fixed pitch design. It was redesigned for open The predicted efficiency was highly
water before any modifications were attempted. All sensitive to the parameters that were varied during
simulations were completed using uniform inflow, the parametric study. More specifically, certain
i.e. open water conditions. hydrorake distributions in combination with certain
circulation distributions increased the efficiency
while some combinations did not show any

2
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

geometry is still continuous. The total and the

inviscid efficiencies are shown to compare the
frictional effects as the hydrorake is increased at the
tip. For some of the hydrorake distributions, the
efficiency gain is reduced due to the friction loss
acting on an additional wetted surface area at the tip.

Frictional effects are predicted to be a

function of the advance coefficient. Figure 5 shows a
set of advance coefficients for a fixed circulation and
hydrorake distribution. It is shown in the figure that
at a very low advance ratio of JS = 0.40 the tip feature
degrades the efficiency of the propeller. At lower
advance ratios, losses are dominated by frictional
effects, and the additional wetted surface area of the
tip plate increases frictional losses.
Figure 2: Example of blade shape with positive hydrorake.

Figure 3: Example of blade shape with negative hydrorake.

improvement. The parametric study indicates that the

benefits from these tip features occur when the rake
and circulation distributions are properly matched.

An example of the effects of hydrorake

distributions on efficiency for a single advance
coefficient and circulation distribution is illustrated in
Figure 4. Certain magnitudes of hydrorake at the tip
present on the left portion of the figure do not appear
on the right-hand portion. This is because some
extreme cases failed to converge during the iterative Figure 4: Effects of hydrorake distributions and bend radius
blade-design process, thus producing unrealistic on open water efficiency for a single advance coefficient
results that are not shown. The three sections of and circulation distribution.
Figure 4 (upper, middle and bottom) demonstrate the
effect of varying the radius of the bend between the To identify possible candidates for retrofit
blade and the tip. The upper portion illustrates results and to determine criteria for the applicability of
of geometries with a large radius (gradual) bend, TLPs, the operating design space of propeller thrust
while the lowest portion contains geometries with a loading coefficient (CT) and advance coefficient (JA)
small radius (sharp) bend. It should be noted that was explored. The design space was mapped with
even though the bend is sharp for some cases, the efficiency gains relative to the baseline design at that

3
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

particular point. More specifically, a baseline design DESIGN REQUIREMENTS

was completed for each point of CT and JA and then a
tip modified design was completed for the same NSWCCD and Sistemar collaborated to develop a set
operating point and compared to that baseline. of common requirements for which the TLP and CLT
Efficiency gain as a function of CT and JA is propellers would be designed. These requirements
illustrated in Figure 6. Figure 6 shows that efficiency were based on those of a nominal single-screw
gains are possible in a wide range of the design merchant vessel. Since the propeller experiments
space. Specifically, the tip loaded concept provides were to be conducted in a recirculating water tunnel
the greatest possible increase in efficiency for (with no hull or wake screens upstream of the
operating conditions of high CT and JA. This means propeller) the propellers were designed for uniform
that a more heavily loaded propeller with a low RPM inflow. To simulate the performance of the propeller
has the potential for the greatest efficiency behind a ship hull, additional requirements were put
improvement by adapting the TLP concept. in place for advance coefficients of +/- 20% of the
design advance coefficient.

The propellers were required to be five-

bladed, fixed pitch and right handed. Also, the
designs were required to include 30 degrees of
projected skew. The diameter was required to be 6.50
meters at full scale with a hub diameter of 1.46
meters. The combination of required thrust loading
coefficient and advance coefficient was expected to
result in designs with pitch-to-diameter ratios larger
than typical values.

The propellers were required to produce a

full-scale thrust of 1,383 kN (310,900 lbf) with an
advance velocity of 8.00 m/s (26.25 ft/s) and a shaft
speed of 80 revolutions per minute. This results in a
design propeller CT and JA of 1.27 and 0.923,
respectively. Blade stress was required not to exceed
62 MPa (9,000 psi) at the design condition. Design
Figure 5: Effects of advance ratio on hydrorake efficiency goals included maximizing open water efficiency at
impact. the design point, maximizing suction side cavitation
inception speed and minimizing the amount of
suction side cavitation present for ahead conditions.
The designs were required to achieve an inception
speed greater than 4.0 m/s for all blade surface forms
of cavitation, and no pressure or suction side
cavitation at the propeller design speed and RPM.
There was to be no pressure side cavitation allowed
throughout the range of +/- 20% of the design
advance coefficient at the 8.0 m/s design advance
speed. There were no requirements for tip or hub
vortex inception speed. The designs were required
not to experience thrust breakdown at the propeller
design point, thrust breakdown being defined as the
ratio of cavitating thrust relative to non-cavitating
thrust being less than 0.95. Cavitating thrust was to
be measured at design advance coefficient and design
cavitation number. Non-cavitating thrust was to be
measured at design advance coefficient, but at a
sufficiently large cavitation number such that there is
no cavitation on the propeller.
Figure 6: Estimated increase in efficiency by employing a
TLP over a conventional propeller.

4
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

PROPELLER DESCRIPTIONS (see Figure 9). This design was developed by Bailar
Marine Consulting using NSWCCD design
The NSWCCD tip loaded propeller (TLP) model as processes, including advanced blade sections (Bailar
fabricated by CEHIPAR using bronze is shown in et al. 1991). This propeller was not manufactured, but
Figure 7. It was designed for a spanwise circulation the open water and cavitation performance were
distribution with a maximum at 0.80R. The TLP analyzed using potential flow tools and the viscous
includes a blended, pressure side winglet (positive flow RANS solver, NavyFOAM (Schroeder et al.,
hydrorake) designed using NSWCCD blade shaping 2009).
methods modified for use on TLPs. The blade
sections include a NACA 66 thickness distribution
and a custom chordwise loading distribution with
more trailing edge loading than the more traditional
NACA a=0.8.

Figure 9: Baseline conventional propeller.

Figure 7: NSWCCD tip loaded propeller (TLP). Figure 10 shows spanwise chord and
thickness distributions while Figure 11 shows
The Sistemar contracted-loaded tip propeller spanwise pitch and camber distributions for all three
(CLT) is shown in Figure 8. The CLT was designed propellers. The NSWCCD-designed propellers
using Sistemar’s in-house methods and includes a include a large amount of camber along the entire
pressure side end-plate (positive hydrorake). span, while the CLT propeller has less camber but a
higher pitch especially near the tip. In Figures 10 and
11, “span” is defined as the ratio of the local radial
distance from the hub over the radial distance
between the hub and the tip.

VISCOUS FLOW SOLVERS

The CLT and TLP propellers were analyzed at
model- and full-scale Reynolds numbers (RN) using
two viscous flow solvers; NavyFOAM and FINFLO.
The baseline propeller was analyzed only with
NavyFOAM. For this study, the “model-scale” RN is
approximately 4.32x105 (to match the model tests)
and “full-scale” RN is approximately 30x106. Where
RN is:
Figure 8: Sistemar contracted-loaded tip (CLT) propeller.
𝑛 ∗ 𝐷 ∗ 𝑐ℎ𝑜𝑟𝑑0.7𝑅 𝑟 2 0.5
𝑅𝑁 = �𝐽𝐴 2 + � ∗ 𝜋� �
A conventional propeller was also designed 𝜈 𝑅
as a baseline using the same design requirements as
the TLP and CLT for the purposes of comparison

5
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

description of the NavyFOAM RANS solver and

validation for rotating propulsors can be found in
Schroeder et al. (2009). The CEHIPAR cavitation
tunnel hub geometry was included in the grid, and
consists of a bullet-shaped nosecone upstream and a
faired taper down to a smaller diameter driveshaft
downstream.

During the NavyFOAM analysis of the three

propellers, a grid sensitivity study was performed for
each Reynolds number and at multiple advance
coefficients. Through this study, model-scale grids of
approximately 2.0 million cells and full-scale grids of
approximately 2.6 million cells were selected. An
example of the blade surface discretization for the
full-scale Reynolds number is shown in Figure 12 for
the TLP.
Figure 10: Spanwise chord and thickness distributions.

Figure 11: Spanwise pitch and camber distributions. Figure 12: Blade surface mesh for TLP NavyFOAM full-
scale Reynolds number analyses.
NavyFOAM is an unstructured, finite-
volume based flow solver with RANS and LES FINFLO is also a finite-volume based flow
capabilities. The version used for this project was a solver with RANS capabilities. The implicit solution
single-phase RANS solver in a rotating reference is based on approximately factorized time-integration
frame. One blade passage was gridded and periodic with local time-stepping. In the present
boundary conditions were employed to include the incompressible case the code uses an upwind-biased
effects of the other four blades of the propeller. The approximation for convective fluxes, while either
domain was discretized with a block-structured, Roe's flux-difference splitting or Van Leer's flux-
body-fitted mesh using ANSYS ICEMCFD software. vector splitting is applied for compressible flows. In
Semi-circular trailing edges of 0.2mm radius (at the incompressible case, a Rhie and Chow (1983)
model scale) were enforced on all three propellers. type method has been implemented in the code in the
For both model-scale and full-scale Reynolds simplified form presented by Johansson and
numbers, wall functions were used to simulate the Davidson (1995). The pressure is central-differenced
boundary layer, and a y+ at the wall of 30 was taken and a damping term is added via a convective
as a target. For this study, the H. R. Wilcox k-Omega velocity. A multigrid method is used for the
turbulence model was used. A more complete acceleration of convergence. Solutions on the coarse

6
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

grid levels are used as a starting point for the MODEL HARDWARE AND
calculation in order to accelerate convergence. A
viscous solution is extended to the wall, i.e. no wall EXPERIMENTS
functions were used and the wall y+ was Models of the TLP and the CLT propellers were
approximately 1. Calculations were made using constructed of bronze at 1/26th scale. The diameter of
Chien’s k-epsilon and the SST k-omega turbulence the model-scale propellers was 250mm. The models
models by VTT, however, only the SST k-omega were constructed by the model manufacturing facility
turbulence model results are discussed here. The at the Canal de Experiencias Hidrodinámicas de El
solution of the RANS equations can be obtained by Pardo (CEHIPAR). After fabrication, the two models
either the pseudo-compressibility or pressure were measured using a portable coordinate measuring
correction method. The computations presented in machine in the CEHIPAR manufacturing facility.
this report have been made with the pressure The measurements indicated that the models were
correction method. A more complete description of within the tolerance specified by ITTC
the FINFLO numerical method including recommendations (ITTC procedures 7.5-01-01-01 &
discretization of the governing equations and solution 7.5-01-02-02) and were suitable for hydrodynamic
algorithm can be found in Sánchez-Caja, et al. (1999 testing.
and 2012).
The open water tests were conducted in the
The grids were generated for the FINFLO calm water towing tank at CEHIPAR. The rotational
analyses using an in-house, VTT developed program speed of the models was set at 15 revolutions per
with the help of templates. An O-C topology was second (RPS), and the speed of the carriage was
used around the blades and endplates with the aim of varied in order to obtain thrust and torque values over
reducing the number of cells for a given accuracy. a range of advance coefficients. The Reynolds
This allowed for control of the grid orthogonality number for the open water experiments at the design
over the propeller surfaces. A zero-thickness trailing advance coefficient was approximately 4.32x105.
edge was enforced for both propellers by the gridding
scheme. Figure 13 demonstrates the distribution of The cavitation inception and pressure pulse
cells on the blade surfaces for the CLT propeller. The measurements were conducted in the cavitation
grids around the propeller consisted of 2.4 million tunnel at CEHIPAR. The nominal rotational speed of
cells per blade distributed in 19 blocks for both the the models was again 15 RPS and resulted in similar
full-scale and model-scale calculations. Reynolds numbers as the open water experiments.
Tunnel corrections for measured thrust and torque are
not performed in this case as the model disk area to
tunnel area ratio was 0.061. A thrust identity method
was used to set conditions in the tunnel.

OPEN WATER PERFORMANCE

Measuring the open water thrust and torque was
necessary to establish the conditions for the
cavitation experiments, as well as to check the
accuracy of the RANS analyses. Figure 14 shows
comparisons of the thrust, torque and efficiency of
the CLT and TLP propellers, from both the
experiments and computations. The dashed lines
represent the experimental data for the CLT, while
solid lines represent the experimental data for the
TLP. Symbols indicate the results of the model-scale
RANS computations, with filled symbols showing
the NavyFOAM results and hollow symbols showing
Figure 13: Constant index cuts of the mesh in the direction the FINFLO results. The vertical line represents the
normal to the blade of the CLT propeller. design advance coefficient.

7
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

Solid lines: TLP experiment

Dashed lines: CLT experiment
Solid symbols: NavyFOAM calculations
10KQ Hollow symbols: FINFLO calculations

KT

ETAO

Figure 14: Open water thrust, torque and efficiency of TLP and CLT propellers at model scale.

The nomenclature for thrust, torque and higher-than-design advance coefficients. The
efficiency follows the standard ITTC definition and increased torque in the experiment could be due to
are provided here: the formation of a separated leading edge vortex,
present in the experiments but not fully captured in
𝑇 either NSWCCD’s or VTT’s RANS simulations. The
𝐾𝑇 =
𝜌𝑛2 𝐷4 FINFLO analysis does show a small area of
separation near the leading edge, between r/R values
𝑄 of approximately 0.7 to 0.9, at an advance coefficient
𝐾𝑄 =
𝜌𝑛2 𝐷 5 of 1.108. This separation is indicated by sharp bends
in the surface streamlines in Figure 15. It is possible
𝐽𝐴 ∗ 𝐾𝑇 that the extent of this separated region is under-
𝐸𝑇𝐴𝑜 =
2𝜋 ∗ 𝐾𝑄 predicted in the RANS, causing the slight under-
prediction in the torque at greater than design
advance coefficients.

The data shown in Figure 14 demonstrates The baseline propeller, while not tested in
good agreement between the experiments and both open water, was analyzed using the NavyFOAM flow
RANS solvers, with the NavyFOAM analysis slightly solver. The open water thrust, torque and efficiency
closer to the experimental data for the torque of the values calculated using the NavyFOAM solver for
TLP and CLT propellers. the three designs at the model-scale Reynolds number
are shown in Figure 16.
While the correlation between the RANS
computations and the experiment is generally good, At the design advance coefficient, the TLP
both RANS simulations under-predict the torque of has slightly greater efficiency than the baseline,
the TLP propeller around an advance coefficient of which in turn has slightly greater efficiency than the
1.1, resulting in a slight over-prediction of model- CLT. However, each propeller produces a somewhat
scale efficiency compared to the experiment at different thrust at the design advance coefficient, so

8
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

the efficiency of the three propellers is shown REYNOLDS SCALING EFFECTS

compared to the propeller thrust loading coefficient
(CT) in Figure 17. The open water efficiency from the The differences in non-dimensional thrust and torque
FINFLO k-Omega computations is also shown in of model- and full-scale propellers have been
Figure 17 for the TLP and CLT propellers. The figure examined for conventional designs thoroughly in the
illustrates that the FINFLO computations predict the past by the members of the ITTC and for CLT
CLT efficiency to be higher than predicted by propellers by Pérez Sobrino, et al. (2005). In
NavyFOAM, resulting in efficiency parity between predicting full-scale powering performance, it is
the CLT and baseline propeller at model scale. common practice to either invoke the ITTC’78
method for scaling open-water thrust and torque
(ITTC procedures 7.5-02-03-01.4), or to demonstrate
that the propeller open water experiments have
reached a high enough Reynolds number to be
independent of scaling. The latter approach requires
the use of model facilities and hardware of a fairly
large scale, available in only a few locations around
the world.

Figure 15: FINFLO limiting streamlines and pressure

calculation of TLP (pressure side) at JA=1.108.

10KQ

ETAO Figure 17: NavyFOAM (green curves) and FINFLO (blue

curves) open water efficiency for the TLP (filled squares),
CLT (triangles) and baseline (hollow squares, dashed line)
at model-scale Reynolds number.

KT
In Pérez Sobrino, et al. (2005) it is asserted
that the ITTC’78 scaling procedure must be enhanced
to capture the scaling effects of CLT propellers,
because the ITTC’78 procedure uses a similarity
assumption that reduces the spanwise quantities to a
Figure 16: NavyFOAM open water results for the TLP single value of 0.75R. For CLT propellers this
(filled squares), CLT (triangles) and Baseline (hollow assumption is not valid due to the unique tip
squares) at full-scale Reynolds number. geometry and special radial load distribution with
non-zero load at the tip that leads to rather different
Additional numerical simulations using FINFLO with flow characteristics compared to conventional
the k-epsilon model were performed by VTT in order propellers at model scale, so corrections to the
to check the influence of partially laminar flow on the ITTC’78 procedure were proposed.
results. Further discussion on this topic can be found
in Sánchez-Caja, et al. (2014). The present work aims to further this
research by applying the scaling procedure of Pérez
Sobrino to a new CLT design as well as the

9
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

NSWCCD-developed TLP design, and to compare percent. The NavyFOAM-calculated thrust and
the results of the scaling procedure with multiple torque, however, are approximately 5% higher than
RANS assessments. Hsin, et al. (2010) found that the scaled experimental data.
RANS assessments showed an additional scaling
effect for propellers with end-plates.

The full-scale open water thrust and torque

coefficients are obtained by including the corrections 10KQ
described in the Sistemar scaling procedure, as
described in Pérez Sobrino, et al. (2005). Based on
the experience of Sistemar and its internal data base
for CLT propellers, it was decided to adopt the
following correlation coefficients for this project: KT

∆CF = - 0.00169
κ = 0.0094
End plate form factor = 0.5; 1+k= 1.5.
ETAO
The same correlation coefficients were
applied for the TLP and CLT propellers. Figure 18
shows the open water curves from the experiments
corrected by the Sistemar procedure, as well as the
full-scale computational results from FINFLO and
NavyFOAM for the TLP. Figure 19 shows the open Figure 19: Full-scale open water predictions for CLT from
water curves from the experiment corrected by the Sistemar scaling procedure (lines), FINFLO full-scale
Sistemar procedure, as well as the full-scale analysis (hollow points) and NavyFOAM (solid points).
computational results from FINFLO and NavyFOAM
for the CLT.
Figure 20 shows the scaling corrections as a
percent change of KT, KQ and ETAo (at the design
advance coefficient of 0.923) as predicted by the
10KQ various methods for the TLP.

KT

ETAO

Figure 18: Full-scale open water predictions for TLP from

Sistemar scaling procedure (lines), FINFLO full-scale
Figure 20: Reynolds scaling corrections of various methods
analysis (hollow points) and NavyFOAM (solid points).
for the TLP at the design advance coefficient.

For both propellers, the scaled experimental

The difference between the Reynolds
efficiency agrees well with the RANS results. The
scaling corrections predicted the NavyFOAM RANS
scaled experimental thrust and torque coefficients
solver and the other methods (including the FINFLO
also agree with the FINFLO calculations within a few
RANS solver) motivated additional investigation.

10
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

The integrated pressure and shear stress for thrust and show velocity in the chord-wise direction normalized by
torque predicted by NavyFOAM were examined inflow velocity.
separately to better understand the differences
between the model-scale and full-scale computations.
If only changes in the viscous shear stresses are
considered, the model-to-full-scale corrections
predicted by NavyFOAM are reduced greatly. These
values are shown in Figure 20, labeled “NavyFOAM
(Cf only)”. Clearly, there must be a Reynolds number
effect on the blade pressure distributions not captured
by the friction-coefficient-based ITTC’78 method,
and possibly captured to a lesser extent by the
FINFLO RANS solver and Sistemar method.

To investigate the Reynolds number effect

on the blade pressure distribution, constant-radius
slices of the volume flow field were extracted from
the NavyFOAM model- and full-scale solutions at
0.7R. These slices reveal a critical flow feature
difference near the trailing edge, shown in Figure 21
and Figure 22.
Figure 22: Constant-radius slice of NavyFOAM flow
The separated flow region near the trailing solution for TLP at full-scale Reynolds number, contours
edge of the model-scale simulation (Figure 21) is show velocity in the chord-wise direction normalized by
larger than that of the full-scale simulation (Figure inflow velocity.
22). It is to be noted that a semi-circular trailing edge
shape was used in both of the calculations considered The scaling corrections from the various
in this comparison. The separated region has the methods discussed above for the CLT propeller are
effect of unloading the trailing edge, and changes the presented in Figure 23.
angle of the streamlines departing the trailing edge by
3 degrees, enough to explain the 7% change in the
pressure integration value for thrust. Furthermore, it
follows that the FINFLO RANS analysis would
produce smaller scaling corrections for this case, due
to a reduced separated area from the zero-thickness
trailing edge geometry simplification.

Figure 23: Reynolds scaling corrections of various methods

for the CLT at the design advance coefficient.

Again, the viscous-only correction from

Figure 21: Constant-radius slice of NavyFOAM flow NavyFOAM is similar to the friction coefficient
solution for TLP at model-scale Reynolds number, contours based ITTC’78 method, though in both cases the

11
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

viscous scaling predicted by the RANS is greater a greater extent than predicted by only a change in
than the ITTC’78 method due to the fully-turbulent the friction coefficient.
boundary layer assumption in the NavyFOAM solver.
In contrast to the TLP, for the CLT propeller the
overall scaling corrections predicted by the two
RANS methods are much closer to each other than
for the TLP. This result opens the question: if the
difference between the NavyFOAM and the FINFLO
results for the TLP was due to a difference in trailing
edge geometry (that is also present in the CLT
analyses), why do the two solvers correlate better for
the CLT? In an attempt to answer this question, the
NavyFOAM CLT solution was interrogated using the
same 0.7R constant-radius slice procedure as the
TLP. The model-scale and full-scale flows are shown
in Figure 24 and Figure 25, respectively.

Figure 25: Constant-radius slice of NavyFOAM flow

solution for CLT at full-scale Reynolds number, contours
show velocity in the chord-wise direction normalized by
inflow velocity.

Figure 24: Constant-radius slice of NavyFOAM flow

solution for CLT at model-scale Reynolds number,
contours show velocity in the chord-wise direction
normalized by inflow velocity.

The CLT propeller has less camber and less

trailing edge loading than the TLP design. Because of
these differences in the blade sections, there is much
less trailing edge separation at model scale predicted
by NavyFOAM on the CLT propeller design.
Figure 26: Surface streamlines and pressures on CLT
propeller at JA = 0.923 at model scale (computed by
However, both of the RANS indicate that NavyFOAM).
there are increased scaling corrections required for
the CLT propeller compared with the other scaling
procedures. This difference is attributed to the The baseline propeller was also analyzed
separated leading edge vortex captured by both using NavyFOAM in order to assess the difference in
solvers (NavyFOAM result shown in Figure 26, scaling caused by the presence of the tip feature.
FINFLO result shown in Figure 27) at model scale. Figure 28 illustrates the predicted scaling effects
At full scale, there is a lower pressure in the vortex between the ITTC’78 method and the NavyFOAM
core that increases the lift and torque on the blade to scaling corrections.

12
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

shown versus the thrust loading coefficient in Figure

29. The NavyFOAM results indicate that at full scale,
the efficiency of the TLP and CLT propellers are 3%
and 0.2% greater than the baseline propeller,
respectively. Similar to the model-scale results, the
FINFLO computations predict a higher efficiency for
the CLT propeller. The FINFLO computations
predict the TLP and the CLT to be 3.7% and 2.0%
more efficient at the design thrust loading coefficient
than the NavyFOAM prediction of the baseline
propeller at full scale, respectively.

One additional aspect of propeller scaling

effect was studied using NavyFOAM: the effect of
the trailing edge (TE) bevel (or, anti-singing edge) on
open water thrust, torque and efficiency. This detail
is too small to manufacture at model scale, and
therefore is not included in propeller open water
Figure 27: Surface streamlines and pressures on CLT
propeller at JA = 0.923 at model scale (computed by
model tests. To assess this effect, the standard US
FINFLO). Navy TE bevel was applied to the TLP and baseline
geometries, while the Sistemar bevel was applied to
the CLT. The beveled geometries were then gridded
and open water coefficients were computed with
NavyFOAM at the full-scale Reynolds number.
Figure 30 shows the effect on the KT, KQ and ETAO
of the TE bevel at full-scale Reynolds number, as a
percentage of the unbeveled value.

Figure 28: Reynolds scaling corrections of various methods

for the baseline propeller at the design advance coefficient.
Figure 29: NavyFOAM (green curves) and FINFLO (blue
curves) open water efficiency for the TLP (filled squares),
The NavyFOAM viscous scaling effect on CLT (triangles) and baseline (hollow squares, dashed line)
efficiency for the baseline is less than that of the CLT at full-scale Reynolds number.
and TLP (1.9% versus 2.6% and 3.1%, respectively)
while the ITTC’78 efficiency scaling effect is within
0.1% for all three propellers. This evidence suggests The results indicate that both styles of TE
that there is an increase in the viscous scaling effects bevel have similar effects: an increase in thrust and
due to the added surface area at the tips of the CLT torque, accompanied by a small drop in efficiency.
and TLP propellers. The Sistemar bevel is predicted to have a milder
effect than the USN bevel. The efficiency of the three
The full-scale Reynolds number propeller beveled propellers at full-scale Reynolds number
efficiency computed by NavyFOAM and FINFLO is

13
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

versus the thrust loading coefficient (CT) is shown in The cavitation inception data demonstrates that the
Figure 31. TLP had no suction side cavitation, but a tip vortex at
the design point. The CLT propeller had much earlier
suction side cavitation inception due to its greater
pitch and higher sectional angle of attack.

VA= 8.0m/s

Figure 30: Effect of USN and Sistemar TE bevels on open Figure 32: TLP cavitation inception curves.
water coefficients.

VA= 8.0m/s

Figure 31: NavyFOAM open water efficiency for the TLP

(filled squares), CLT (triangles) and baseline (hollow Figure 33: CLT cavitation inception curves.
squares) at full-scale Reynolds number with TE bevels.
Images from the test (Figure 34 for the TLP
and Figure 35 for the CLT) show the cavitation
CAVITATION INCEPTION patterns present on the two propellers at the design
The cavitation inception curves measured in the thrust coefficient and cavitation number representing
water tunnel at CEHIPAR are shown in Figure 32 a full-scale advance velocity of 8.0m/s.
and Figure 33 for the TLP and CLT, respectively.
The curves show the cavitation inception number (σ) While the baseline propeller was not built or
as a function of thrust coefficient (KT), where: tested at model scale, RANS comparisons of
cavitation inception were made using NavyFOAM. It
𝑃𝑂 − 𝑃𝑉 must be noted that these analyses used single-phase
𝜎=
0.5𝜌𝑛2 𝐷2 computations only, so while cavitation inception will

14
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

be captured fairly accurately, the shape and extent of

the cavitation post-inception will not. Figure 36 and
Figure 37 show the full-scale flow solutions of the
TLP and baseline propellers at the design point, with
an isosurface of sigma shown to highlight the regions
of the flow where the pressure is below the vapor
pressure.

Figure 36: NavyFOAM flow solution of TLP at full-scale

Reynolds number with an isosurface of sigma
(demonstrating no cavitation predicted).

Figure 34: Cavitation pattern on the TLP at the design point

(JA=0.911, sigma=4.092).

Figure 37: NavyFOAM flow solution of baseline at full-

scale Reynolds number with an isosurface of sigma.

CONCLUSIONS
In this paper, the authors have presented a
comparative design and performance of tip loaded
propellers, TLP and CLT, with a particular focus on
Reynolds number scaling effects on open water
performance. The computational results showed good
agreement with experimental data, and the
differences between the results have been discussed.
From this study the authors have concluded that:
Figure 35: Cavitation pattern on CLT propeller at design
point (JA=0.94, sigma=4.092). - There is definite potential for improving
efficiency and cavitation inception through
the implementation of tip loaded propellers,

15
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

as demonstrated by the higher efficiency and ONR Code 331, under the administration of Dr. Ki-
cavitation inception of the TLP compared Han Kim.
with the baseline propeller. The CLT
propeller matched the baseline propeller for
efficiency at the design thrust coefficient. REFERENCES
However, its efficiency was reduced due to a Adalid, J.G., and Gennaro, G., “Latest experiences
separated leading edge vortex caused by the with Contracted and Loaded Tip (CLT) propellers,”
large sectional angle-of-attack at the design Sustainable Maritime Transportation and
condition. It is expected that a redesign of Exploitation of Sea Resources, 1st ed., Vol. 1, Taylor
the CLT propeller with reduced angle-of- & Francis Group, London, 2011, pp. 47-53.
attack loading would have a higher
efficiency. Andersen, P., Friesch, J., Kappel, J.J., Lundegaard,
- The scale effects have been investigated and L., and Patience, G., “Development of a Marine
demonstrated to come from multiple sources Propeller With Nonplanar Lifting Surfaces,” Marine
(viscous and pressure effects as well as the Technology, Vol. 42, issue 3, 2005, pp. 144-158.
presence of separation). The computational
results show a large sensitivity to trailing Bailar, J.W., Jessup, S.D., and Shen, Y.T.,
edge details (i.e. sharp vs. rounded) in “Improvement of Surface Ship Propeller Cavitation
shaping the flow at model-scale Reynolds Performance Using Advanced Blade Sections,”
numbers. The computational results also Proceedings of the 23rd American Towing Tank
demonstrate the loss of lift due to camber Conference, vol. 1, 1991, pp. 185-193.
and trailing edge loading at model scale. The
TLP and the CLT have been identified as Hsin, C.-Y., Chang, K.-K., Chi, R.-C., and Chen, P.-
having an increased scaling effect due to F., “Design and Analysis of the End-Plate Effect
added wetted surface area in the outer span Propellers,” Proceedings 28th Symposium on Naval
region. Hydrodynamics, 2010.

In the effort to further improve and innovate Hsin, C.-Y., Kerwin J.E., and Kinnas, S.A., "A Panel
in the field of propeller design, the concepts Method for the Analysis of the Flow around Highly
demonstrated here should continue to be studied and Skewed Propellers," Proceedings of the SNAME
developed. The design-by-analysis approach Propellers '91 Symposium, Society of Naval
employed during the detail design of the TLP Architects and Marine Engineers, vol. 1, 1991.
propeller generated a successful design by current
measures, but perhaps a superior design could be Johansson, P., and Davidson, L., “Modified
achieved with better design methods and tools Collocated SIMPLEC Algorithm Applied to
specific to the case of the tip loaded propeller. Buoyancy-Affected Turbulent Flow Using a Multi-
Furthermore, while the current study was limited to Grid Procedure,” Numerical Heat Transfer, Part B,
open water design and performance, it is Vol. 28, pp. 39-57, 1995.
recommended that the performance of tip loaded
propellers in the ship wake (i.e. non-uniform inflow) Kerwin, J.E., Michael, T.J., and Neely, S.K.,
be investigated to assess the effects of more realistic “Improved Algorithm for the Design/Analysis of
flow fields on the efficiency and cavitation Multi-Component Complex Propulsors,” Proceedings
performance. of the Propellers/Shafting Symposium, Society of
Naval Architects and Marine Engineers, vol. 1, 2006.

ACKNOWLEDGEMENTS Kinnas, S.A., Griffin, P.E., and Mueller, A.C.,

"Computational Tools for the Analysis and Design of
High Speed Propulsors," The International CFD
Support for Phillip Duerr was provided by the U.S. Conference, Ulsteinvik, Norway, pp.7/1-1/14, May
Department of Defense SMART program. The 1997.
design, model construction, testing and CFD
performed by Sistemar, CEHIPAR and VTT were Pérez Sobrino, M., Minguito Cardeña, E., Garcia
funded through ONR NICOP Grant N62909-12-1- Gómez, A., Masip Hidalgo, J., Quereda Laviña, R.,
7087 managed jointly by Dr. Ki-Han Kim (ONR) and Pangusion Cidales, L., Pérez Gómez, G., and
Drs. Richard Vogelsong and Woei-Min Lin at ONR Gonzalez-Adalid, J., “Scale Effects in Model Tests
Global. Funding for NSWCCD was provided by with CLT propellers,” Proceedings of the 27th Motor

16
 30th Symposium on Naval Hydrodynamics
Hobart, Tasmania, Australia, 2-7 November 2014

Ship Marine Propulsion Conference, Bilbao, Spain.

January 2005.

Rhie, C.M., and Chow, W.L., “Numerical Study of

the Turbulent Flow Past an Airfoil with Trailing
Edge Separation,” AIAA Journal, Vol. 21, No. 11,
Nov. 1983, pp. 1525-1532.

Sánchez-Caja A., González-Adalid J., Pérez-Sobrino,

M., and Saisto, I., “Study of End-Plate Shape
Variations for Tip Loaded Propellers Using a
RANSE Solver,” Proceedings 29th Symposium on
Naval Hydrodynamics, 2012, pp. 26-31.

Sánchez-Caja A., González-Adalid J., Pérez-Sobrino

M., and Sipilä T. “Scale Effects on Tip Loaded
Propeller Performance Using a RANSE Solver,”
Ocean Engineering, April 2014,
http://dx.doi.org/10.1016/j.oceaneng.2014.04.029i

Sánchez-Caja A., Rautaheimo, P., Salminen, E., and

Siikonen, T., “Computation of the Incompressible
Viscous Flow around a Tractor Thruster Using a
Sliding Mesh Technique,” Proceedings of the 7th
International Conference in Numerical Ship
Hydrodynamics, 1999.

Schroeder, S., Kim, S.-E., and Jasak, H., “Toward

Predicting Performance of an Axial Flow Waterjet
Including the Effects of Cavitation and Thrust
Breakdown,” Proceedings of the First International
Symposium on Marine Propulsors – SMP’09, vol. 1,
2009, pp. 387-394.

17