Professional Documents
Culture Documents
smp2022 Design of The Wageningen F Series
smp2022 Design of The Wageningen F Series
smp2022 Design of The Wageningen F Series
Hull-clearance
Astern torque ratio
The hub diameter and chord length at the hub are both a
polynomial resulting from the analysis of the MARIN
database.
The most commonly applied blade profiles by the industry
are used for the F-series as well. For the camber, the NACA
mean line is selected with a = 0.6 at the hub to delay flow
separation and a = 0.8 at 0.6R towards the tip. For
thickness, the NACA66 DTMB mod distribution is applied
Figure 1: Thrust loading coefficient 𝑪𝑻 and cavitation
perpendicular to the nose-tail line.
number 𝝈𝑵 as function of PD and BAR for the 5-
bladed propellers 3 F-SERIES DESIGN
The tip speed ranges from 30 m/s for bulkers and LNG 3.1 Design Approach
carriers towards 50 m/s for fast displacement ships and Nowadays, a competitive propeller can be designed using
fast container ships. The tip speed is important to specify optimization techniques. Therefore, the F-series design
the relation between the hydrodynamic pressure and the was approached using optimization techniques and
hydrostatic pressure. polynomial fitting of the optimized parameters that
The clearance between the propeller tip and the hull describe the geometry. The optimization and smoothing
surface above the propeller was defined as well to was iterative between the supplied design input, the
evaluate hull excitation. The clearance ranges from 20% description of the design variables and the optimization
to 50% of the diameter for a fast displacement ships and results. This design iteration loop is schematically
bulkers, respectively. presented by Figure 2.
To estimate the global response levels due to propeller
induced hull excitation, DNV (as a participant to this JIP)
proposed to use an empirical transfer function to relate
the excitation force to the vertical vibration in the aft ship.
The first harmonic excitation force on the square virtual
plane of size 1.5 times the propeller diameter is
multiplied by a correlation factor. Hence, vibration is
assumed to be linear with the virtual force. The
correlation factor was determined based on the ship type
and its displacement.
The prescription of the astern torque is important for the
propeller blade thickness in the area around 0.8R,
following from the DNV rules. Despite the ambition to Figure 2: Design iteration loop
allow full torque astern for each propeller in the F-series Performance optimization was performed for a subset of 64
in view of electric drives, an exception was made for the propellers which are well distributed within the F-series
high BAR propellers and high PD with low blade matrix of 150 propellers and represent most of the typical
number. In practice these propellers are generally used ships.
for fast, very high powered ships only.
3.2 Definition of Optimization Problem
2.4 Design Parameters
Multi-objective optimisation techniques provide the
The table below provides a summary of the parameters that
various trade-offs between conflicting objectives and
are used in the design of the F-series. The input and
indicate the influence of design constraints on the Table 1: Objectives and constraints within the
achievable objectives. propeller optimization
To allow large design freedom, the propeller is fully Objectives Constraints
parameterized in the radial pitch, camber and outline (both
Maximize behind Avoid bubble cavitation with at least
chord and skew) distributions. These are defined as Bézier
efficiency 5% margin on cavitation number 𝜎𝑁
curves of which the parameters (control points) form the
optimization variables. Within MARIN and the Minimize hull Comply to the cavity-planform
Cooperative Research Ships (CRS), the multi-objective excitation, criterion for suction-side sheet-
optimisation method PropArt has been developed for calculated as a cavitation such that the sheet
propeller design optimisation studies. virtual force on a cavitation merges smoothly with tip
A genetic optimization algorithm was used to control the flat plate above the vortex cavitation.
computational process and to provide input to the geometry propeller
Comply to the time ratio and aspect
generator. Typically, 96 geometries are generated within
Minimize ratio criteria for pressure-side
each generation. Through subsequent generations, the
cavitation volume cavitation as determined in the EU
genetic algorithms modifies the design parameters to
on all blades EROCAV project as reported by
converge towards the best propellers within the design
Moulijn et al. (2006).
space.
Each propeller geometry is scaled to the correct PD and Table 2: Selection process for the F-series survivor
BAR, after which each geometry is subjected to a series of from each particular optimization
three PROCAL computations:
1. Steady computation of open water characteristics
2. Unsteady computation in the specified wake field,
iterated to the specified thrust coefficient 𝑪𝑻 using the
inflow speed as iteration variable.
3. Cavitation computation for the same condition as the
unsteady fully-wetted computation.
Propeller computations were done with PROCAL
(Bosschers et al, 2008), a boundary element method (BEM)
which computes the inviscid flow around the propeller The entire series of 150 propellers would be analysed or its
operating in the ship’s wake field. Besides efficiency, the performance showing goal and constraint behaviour over
calculations give the pressure distribution and the extent the wider design space.
and dynamics of the sheet cavitation developing on the
In each design iteration (see Figure 2) the best geometries
propeller blades. PROCAL has been developed by MARIN
from the optimizations were selected and smoothed as
within the framework of CRS and has been extensively
function of the main parameters PD, BAR and Z.
validated.
Subsequently, the input was further adjusted towards better
The results of the PROCAL computations are evaluated convergence and smoother polynomial-based description
and converted into objectives and constraints. Three of the F-series geometry. High curvature in the polynomial
objectives are defined for the design of the F-series, as surfaces was avoided for the purposes of consistency and
given in Table 1. Further, in addition to the objectives, smoothness, which was regarded as more important than
three constraints are used to avoid cavitation erosion risk the recording of every detail from the individual
in the prescribed design condition. optimizations.
3.3 Optimization results 3.4 Geometry and Smoothing
For a normal optimization the Pareto front shows the trade- The following summarizes the iterative path towards the
offs of all solutions. An example is given in Figure 11 and geometry description of the F-series:
Figure 12 at the end of this paper. The radial distributions
The blade thickness of the F-series follows the DNV
are shown as well as the Pareto Fronts (converging from
rules. The final polynomial description defines the radial
blue to red) and the open water diagrams. Individual
distribution. For simplicity reasons, it was chosen to
G73P35 was chosen as a winner from this optimization.
describe the thickness with a simple cubic Bezier curve,
For the design and smoothing towards the geometry of the such that both the thickness at 0.6R and 0.8R fulfil the
F-series, the winner from each individual optimization was rules. The thickness at the tip is equal (0.0035D) for all
hand-selected, for which Table 2 provides a scheme. The propellers.
Pareto fronts for the subset of 64 combinations of the main
A fit was made on the input parameters of the blade
parameters Z, PD and BAR. From all fronts a winner was
outline function from the selected propellers from the
selected and polynomial regression determined the relation
optimizations. It was found that the chord distribution
of all geometric parameters as a function of these main
was captured by a 4th order Bezier with control points at
parameters, yielding a series description of the geometry.
the hub, 0.8R and the tip. The majority of the propellers
is relatively well captured by this simple and smooth
description of the chord length.
For the skew a similar procedure was followed. A
parabolic Bezier curve with quadratic polynomial input
captures the majority of the propellers sufficiently
accurate and provides consistency within the F-series. In
general, the wider blades (higher BAR or lower Z) have
higher skew.
The rake of the F-series balances the skew-induced rake
and keeps the blade as tight as possible above the hub so
that the propeller can lay down on the pressure side for Figure 4: Camber polynomials at root and 0.7R
storage. The rake was not optimized for hydrodynamic
4 COMPUTED PERFORMANCE
performance. No special tip-rake was applied.
The performance of the 150 F-series propellers was
For the pitch, initial optimizations allowed much freedom
computed by PROCAL.
with a 5th order Bezier curve. Nonetheless, it was found
that a simple parabolic Bezier curve, always forced 4.1 Efficiency
towards maximum pitch at 0.7R, would describe the The behind efficiency of the F-series as calculated by
optimization results sufficiently accurate. Only the pitch PROCAL will be higher than the B-series as shown by
reduction at hub and tip are required as design Figure 5 (left).
parameters. Bilinear polynomials were found to capture
the trends from the optimizations well. An impression is
given in Figure 3. The low BAR propellers have more
pitch reduction, compared to higher BAR propellers,
both at the hub and at the tip. According to the
optimizations, a pitch distribution like the F-series
provides an optimum balance between efficiency, hull
excitation and the stability and non-erosiveness of the
sheet cavitation.
Non-erosive Growing Complies to Erosion Pressure- Figure 8: Cavitation outline for two other F-series
region phase, outline risk side propellers (see table 3 for explanation of colors)
non- criterion, cavitation
erosive non-erosive No clear difference in cavitation behaviour as function of
blade number was seen, when compared for equal area of
one blade, although more super cavitation is observed for
the propellers with more blades.
For reference, Figure 9 shows an example of the cavitation Figure 10 (middle) considers the relation between the
behaviour of the B-series, computed in the same maximum pressure fluctuation and the integrated vertical
conditions. In general, the B-series show less stable sheet force which shows a poor correlation.
cavitation. The cavitation of the F-series at similar The empirical approach as proposed by DNV was used to
condition is much closer to contemporary designs. relate the first harmonic virtual force to vertical vibrations
Due to the focus on high efficiency, the cavitation volume levels. Figure 10 (right) presents the predicted vibration
of the F-series is not always less compared to the B-series, levels. Levels up to 4 and 10 mm/s are classified as average
but the behaviour is significantly more favourable. For for ships with aft and forward accommodation
higher BAR, the differences between the behaviour of the respectively.
B-series and F-series become smaller. The approach by DNV only uses the first harmonic
contribution to the excitation force due to the substantial
uncertainties in the measurement or computational
methods for the higher harmonics. Therefore, the higher
harmonics are correlated within the empiricism of the
approach. Nevertheless, during the optimization, the
choice was made to minimize the total virtual force, instead
of just the first harmonic force in order to avoid high levels
for the higher harmonics.
Although both the computations with PROCAL and the
DNV approach are engineering tools, they were considered
capable to steer the characteristics of the F-series
propellers. This analysis indicates that the F-Series are
Figure 9: B-series propellers computed in similar designed with a focus on high efficiency, while staying at
conditions as the F-series the average comfort levels.