Second International Symposium on Marine Propulsors

smp’11, Hamburg, Germany, June 2011

Numerical Design and Experimental Verification of a RIM-Driven

Thruster*

Alexey Yu. Yakovlev, Marat A. Sokolov, Nikolay V. Marinich 1

1
Krylov Shipbuilding Research Institute (KSRI), St.Petersburg, Russia

ABSTRACT The issues of design, calculation and experimental

This paper presents the results of a design and investigation of thrusters are addressed in reference with
experimental verification of RIM-driven thruster blading Lebedev et al (1969). The studies on waterjet propulsors
system. The propeller was designed by means of direct (Papir 1970, Kulikov & Khramkin 1965) are also closely
blade pitch optimization method. Blade camber was set to related to this subject.
zero for symmetric port and starboard thruster The theory presented in these papers can be successfully
performance. The strength analysis of blading system applied to the design of thrusters with RIM-driven
under design was carried out and cavitation characteristics propeller because specific features of its hydrodynamic
were estimated. For verification of the design, a thruster configuration have no principal effect on the main
model of tunnel type was manufactured. This model was calculation procedures. Recent publications (Kinnas et al
tested in the deep-water towing tank of the Krylov 2009) confirm this statement, since the modern
Shipbuilding Research Institute. computation techniques are integrated into the traditional
Keywords pattern of calculations.
RIM-driven propeller, thruster, design, experiment 2 STATEMENT OF PROBLEM
1 INTRODUCTION The thruster performance is essentially dependent on the
Thrusters with RIM-driven propellers have been known type on the tunnel shape and, in particular, on the
for a long time. They were used, e.g., for the USSR ships arrangement of propeller within the tunnel. In our case,
Valeriy Chkalov and Rodina, built in the middle of 20th the simplest and a widely-used tunnel design is a straight
century in Germany (Lebedev 1969). However, only in pipe smoothly joined to the ship hull on both sides. In a
recent years have such thrusters come into wide use. general case, the thruster can be assumed to have a
Primarily, this has to do with the modern electric motor propeller or a pair of counter-rotating propellers including
technologies. It is now possible to build reliable RIM- CP propeller option. Additionally, the thruster can be
driven thrusters featuring a number of hydrodynamic and fitted with flow guide vane systems. Propellers are
operating advantages over conventional types of thrusters. traditionally fitted to vertical shafts. In the case under
consideration, a single propeller is placed in a tunnel. The
Advantages offered by the use of a RIM-driven propeller pod is eliminated due to RIM-driven design because the
in thruster unit include: propeller is rotated using the RIM drive, which is also
 Reduced pressure head losses and more uniform used to fix propeller blades. Two propeller design options
wake in propeller disk due to elimination of struts and pod are considered, viz., hub-type propeller design and
in thruster channel. hubless propeller design. Since the thruster is to have
 Reduction in thruster-induced noise and vibration equally efficient port and starboard performance, it was
due to flow equalization in front of the propeller and decided to design a propeller with symmetrical blade
elimination of tip vortices in wake of propeller blades. shape. Usually, the main operating mode for thruster is the
bollard pull mode. In the case under consideration, the
 No risk of cable entrapment for propeller shaft in design advance ratio is assumed equal to J=0.13, which is
case of hubless propeller design. practically very close to the bollard-pull condition. The
 No loss of energy due to gap between propeller formulated requirements determined the type of thruster
blade and thruster channel wall. under study.

*
This project was sponsored by the Grant Council of the Russian Federation President (Grant MD-8150.2010.8).
The purpose of studies was to examine the feasibility of for a wide range of operating conditions. Further, the
developing a successful RIM-driven propeller (hub-type following expressions of characteristics will be used:
or hubless design options) for thruster unit. T
KT 
3 PROPULSION SYSTEM DESIGN n 2 D 4
3.1 Thruster design
Q
Thrusters can be designed with fixed-pitch propellers KQ 
(FPP) or controllable pitch propellers (CPP) or counter- n 2 D 5
rotating propellers. This paper considers only a case of KT J
thruster with single FPP. The main operating mode is   (2)
KQ 2
assumed to be the bollard-pull mode. The design process
includes two phases: estimation of pressure head losses in V
the thruster channel; and determination of propeller J
nD
geometry. The pressure head losses are commonly
estimated using relations of individual component losses VS
JS 
versus a number of geometric and hydrodynamic nD
parameters of thruster. These relations can be obtained by where T – propeller thrust, Q – propeller axial torque,
generalization of extensive experimental data (Idelchik n - number of revolutions, D – propeller diameter,
1992). In the non-dimensional form, these losses can be V - inflow velocity, VS – mean velocity in tunnel,
expressed in terms of resistance coefficients: η - propeller efficiency , KT – propeller thrust coefficient,
p KQ – propeller torque coefficient, J – propeller advance
  I  L  M  P (1) ratio, JS – propeller internal advance ratio.
V2
 S
2 The accuracy of calculations by the method (Vasiliev &
Yakovlev 2001) can be assessed using a case of OD-10
where the following resistance coefficients are introduced:
axial pump (Papir 1970). Fig. 1 compares calculations and
ξI – at the thruster channel entry, ξL – friction over the
experimentally obtained thrust versus advance ratio for
channel length, ξM – grids at the channel entry and exit, ξP
this pump. Since the pump may have different blade pitch
– propeller pod, Δp – pressure difference due to energy
angle settings Φ, Fig. 1 presents the results obtained for a
losses, VS – velocity in thruster channel, ρ - water density.
number of pitch angles. Calculations showed satisfactory
Energy losses at the entry of thruster channel depend on agreement with experimental data for a wide range of
the shapes of water intake edges and inclination angle of pitch angles.
hull sides. Losses due to friction over the channel length
are calculated by a traditional method using Blasius 0.8
1
formula (Idelchik 1992). Losses due to protective grids
2
are represented by a function of grid blockage coefficient
(there were no such losses in our case). Losses due to pod
0.6
depend on the pod length and diameter (for streamlined
pods these losses are limited).
Based on the estimated pressure head losses, the operating
0.4
mode of propeller in tunnel is to be determined. For this
KT

purpose the waterjet theory is applied (Rusetskiy &

Mavlyudov 2009) to find the propeller design advance
ratio, which is determined based on the internal velocity, 0.2 Ф=10.0
Ф=-7.5
as well as the required thrust coefficient of propeller.
Then, in accordance with the prescribed procedures, the Ф=-5.0 Ф=0.0
propeller is to be designed for the given mode using 0.0
available propeller diagrams for the thruster. However, it 0.2 0.4 0.6 0.8 1.0 1.2
is impossible to use the above-said diagrams for RIM- J
driven propellers and a special-purpose method is to be
used for calculation of propeller in tunnel. Figure 1: Comparison of calculated and experimentally
obtained characteristics of the OD-10 pump at different
3.2 Calculation of propeller in tunnel angles of blade pitch change Ф (deg).
In this study the propeller in tunnel is calculated using the 1 – experiment, 2 – analytical method (Vasiliev & Yakovlev
method (Vasiliev & Yakovlev 2001) based on calculation 2001).
of flow around 2D foil lattices. In spite of using a
simplified mathematical model, this method calculates
with sufficient accuracy the thrust coefficient and 3.3 Propeller design
efficiency as well as the pressure distribution over blade The method presented above provides solution to the
direct problem, i.e., estimation of propeller characteristics
in tunnel. For propeller design it is required to solve a so- involves a danger of blades intersection. For avoiding
called inverse problem, i.e., to determine the propeller such situations, the minimum allowable pitch Hmin is to be
blade shape for given characteristics. From the prescribed. This requirement is the second constraint in
mathematical point of view, the inverse problem can be (3).
quite simply formalized and reduced to multiple solution For direct numerical solution of problem (3) it is reduced
of the direct problem. This approach is called “direct to the problem of mathematic programming with finite
optimization of propeller” in order to differentiate it from number of unknowns. For this purpose, the sought
more traditional design methods. function H(r) is represented as a linear combination of
Strictly speaking, it is required to solve an optimization basic functions:
problem including the objective function and several N
constraints: H r    Ak  H k r  (4)
k 1
1   H   k  max CP H   min
 r , The basic functions H k r  are given a priori, and only
 K T H , J S 0   K T 0 their coefficients Ak are sought. Thus, the initial problem
 (3)
(3) in infinite-dimensional space is reduced to the problem
 H r   H min r 
 of seeking N values of Ak, which ensure the minimum of
f 0 objective function within given constraints. Such a
where H  H r  - function describing blade pitch problem can be solved using standard methods of non-
distribution over radius, CP – blade surface pressure linear programming. If these are applied, functions η, CP
coefficient, k – balance coefficient for two-parameter and KT in problem (3) are calculated at each step using
optimization procedure, f – camber of cylinder blade prediction of propeller performance in tunnel (Vasiliev &
sections, JS0 and KT0 – design internal advance ratio and Yakovlev 2001). The above-described approach is a
thrust coefficient. simplification of axial pump design method (Yakovlev
2008) for the case of symmetric blade section outline.
The choice of objective function depends on specific
practical requirements regarding blading system. The 3.4 Results of thruster design
most common case is the requirement to obtain the In the thruster design process, it was assumed that the
maximum efficiency of blading system η; however, other propeller diameter was D=0.2 m and the number of
optimization criteria are also possible. The requirement propeller revolutions was n=15 rps. Under these
may be to delay, e.g., propeller blade cavitation as much conditions, it was required to obtain the thrust coefficient
as possible, and in this case, the cavitation number σ is to KT of about 0.337 at advance ratio J=0.13, i.e., near
be minimized. The optimum point may also be specified bollard-pull mode.
in terms of blade strength, levels of force and pressure Based on the shape of the water channel of the thruster
fluctuations. It is also possible to have mixed optimization under design, the resistance coefficient was estimated to
criteria when the blading system should be a trade-off be ξ=0.38. In this case, the losses due to friction and flow
between a number of parameters. In this paper, two- past channel entry were taken into account. Considering
parameter optimization is considered, viz., traditional that the channel has a constant radius, the required thrust
requirement of maximum propeller efficiency and, was to be achieved at internal advance ratio JS0=0.83.
simultaneously, minimization of pressure reduction on
The propeller was designed using the method described
blade surface. The unknown blade geometry parameter is
above. Thus, the optimum pitch distribution over radius
distribution of pitch H over radius r. There is no
was found with other blade geometry parameters given.
cylindrical blade section camber in this problem f=0.
The blade width distribution was chosen in accordance
For optimization of propeller efficiency combined with with the propeller case presented by Lebedev et al (1969).
good cavitation performance, one should set k≠0 (within The blade thickness was initially chosen based on beam
the first or even second decimal places). Under this theory estimations, and then it was corrected using
approach the optimization is primarily focused on the propeller strength estimates presented below. The blade
efficiency, while among propeller geometry options with section outline is symmetric, lens-shaped and similar to all
close efficiency values the preference is given to the case cylinder blade sections.
with the smallest pressure reduction. If we set k=0, then
Design process convergence to optimum solution is
the propeller is designed for the optimum efficiency but
illustrated in Fig.2. In this case, the variations of
without consideration of optimum pressure reduction
efficiency and CP were considered for coefficient values
parameter.
k=0 and k≠0 in (3), i.e., in the process of efficiency-based
The first constraint in (3) is the condition which governs optimization without pressure reduction considerations
the propeller operating mode. In our case, the thrust and in the process of optimization for efficiency as well as
coefficient KT0 is assumed for operating advance ratio JS0. pressure reduction minimization. The solutions
In solving problem (3), one may face with the situations obtained are compared. It is seen that at k=0 the propeller
when the blade geometry considered for the current pitch efficiency is smoothly increasing, but the pressure on
blade is further reduced. In comparison with the two-
parameter optimization, the optimization based on the studies, decisions are taken on improvement of propeller
propeller efficiency only results in a 0.4% efficiency design. In this case, the design was performed for research
increment, but the penalty is a 30% extra pressure purposes and propeller adjustments were not undertaken.
reduction on blade. In choosing the final design, it was
4.1 Propeller blade strength study
found advisable to ensure better cavitation performance.
Evaluation of propeller blade strength is an integral part
Therefore, further consideration will be given to the
of propeller design process. At present, the blade strength
geometry derived at k≠0.
analysis is performed using FEM-based techniques. This
type of approach to the propeller strength analysis
1
(Volkov & Postnov 1990) has proved to be efficient for a
1,6
2 number of full-scale structures. The calculation procedure
-CP 3 (Volkov & Postnov 1990) intended for conventional type
1,4 4
of propellers was modified to cover the case of RIM-
1,2 driven propeller. However, these modifications have not
altered the principles of this method.
For illustration of a typical stress distribution pattern for
0,790
propeller blades, Fig.4 gives distributions of equivalent
S
0,785 stresses in blades fixed in hub only, in rim only and both
in hub and rim. The first case is a conventional propeller;
0,780 two other cases refer to RIM-driven hubless propeller and
0 2 4 6
M
hub-type propeller. The thickness distribution over the
blade radius for this case is assumed to feature smooth
Figure 2: Convergence of propeller efficiency ηS and blade increase in thickness around fixing location.
pressure reduction in the optimization process.
1 – CP at efficiency-based optimization (k=0), 2 – CP at
optimization for efficiency and CP (k≠0), 3 – η at efficiency-
based optimization (k=0), 2 – η at optimization for efficiency
and CP (k≠0), M – number of iteration step. ηS – efficiency
of propeller in pipe.
In the process of these studies, two propeller options were
designed: propeller with hub (hub diameter d=0.3*D); and
hubless propeller. The geometry of propellers is similar.
The hubless propeller blades are different in that the
thickness is reduced to internal radii and the blades are
elongated.
Fig.3 shows photos of hub-type propeller and hubless
propeller manufactured based on the design under
consideration.

Figure 4: Blade stress distributions for the cases when blades

are fixed in hub (a), in rim (b) and both in hub and rim (c).
Level lines correspond to fractions of maximum stresses in
each case.
As indicated by the obtained stress distributions, the area
of maximum stresses is expected to shift to the areas
around the blade joints. However, it should be noted that
if the thickness of a rim-fixed blade is reduced, the
a b maximum stress location may shift to the blade middle
Figure 3: Models of designed propellers: hub-type design (а) radii.
and hubless design (b). If the blade is fixed at both sides, the maximum stresses
are reached near the hub and rim, but on different sides of
the blade. In this case, the maximum stress location is
4 ADDITIONAL STUDIES ON DESIGNED
found to be shifted to the leading edge.
PROPELLERS
Usually, propeller design is not limited to the estimations The peak stresses in case of traditional hub-type design
described above. After design calculations, an in-depth are 1.5 times higher than in the case of rim-fixed blades.
study of propeller performance is undertaken; first Reduction of stress levels is achieved not only because of
analytically and then experimentally. Based on these wider blade tip sections, but also due to specific load
distributions over propeller blade. Rim-fixed blades were measured. The forces generated on the pipe were not
feature shorter arms of forces applied to tip sections and measured in these tests. In case of the hub-type design, the
longer arms of forces applied to root sections. Since the propeller hub was connected to the dynamometer and the
loads on tip sections are usually much higher than the propeller thrust and torque was transmitted to the
loads on root sections, the bending moment in the case of dynamometer via hub (see Fig.6). For testing the hubless
rim-fixed blades prove to be less than in the case of hub- propeller, a special-purpose strut was made to connect the
fixed blades. rim of hubless propeller to the dynamometer shaft and
Stresses are reduced dramatically when blades are fixed transmit the thrust and torque (see Fig.7). Fig.8 shows a
both to hub and rim. In this case, the stress values are general view of the tested model at the test rig. The model
about 1/4 of the stress values observed for the rim-fixed represents a hubless propeller.
blades.
Thus, the fixing arrangement in RIM-driven propellers
makes it possible to reduce the propeller weight and
increase propeller efficiency by using blades of reduced
thickness.
4.2 Analysis of propeller cavitation
performance
Fig.5 provides examples of propeller blade cavitation
predictions. Calculations were done for the cavitation
number σn=3.0:
p  p
n  2 (5)
 n2 D2
Figure 6: Propeller hub/dynamometer connection.
where pν – saturated vapor pressure.
It can be seen that the propeller designed with due account
of blade pressure distribution (at k≠0 in Eq. (3)), features
a relatively narrow cavitation area extending along the
leading edge. While the propeller designed only for
optimum efficiency (at k=0 in Eq. (3)) features a wide
cavitation area at tip radii, which may result in an earlier
fall-off of propeller curves.

Figure 7: Propeller rim/dynamometer connection.

The advance ratio was varied during the hydrodynamic
tests (in the full range of advance ratio values) by
а b
changing the towing speed V. The propellers were tested
Figure 5: Predicted cavitation areas for propeller blades. in the range of 0 ≤ J ≤ 1.2, corresponding to the speed
а – propeller optimized for k≠0, range of 0 to 3.6 m/s. For avoiding the wave-making
b – propeller optimized for k=0. effects in open water conditions, the propeller and duct
were immersed to the depth equal to two propeller
diameters as measured from the propeller axis to water
5 EXPERIMENTAL SETUP surface.
The purpose of the tests was to analyze and compare the
In open-water tests, it is required to measure the thrust and
main characteristics of hub propeller and hubless
torque generated on propeller blades only, since the size
propeller used in RIM-driven thrusters as well as to check
and shape of the hub and rim depend on the specific
the method of design applied for propellers of this type.
thruster design and will be different for different ships.
Two rim propellers with were manufactured (see Fig. 3) When these components were manufactured for tests, their
with a pipe section to model the channel where the dimensions were also chosen with a view to structural
thruster propeller is operating. The tests were performed considerations. In tests, the hub and rim are treated as
in the deep-water towing tank of the Krylov Shipbuilding integral part of hub-type propeller. For a proper account
Research Institute. Propeller revolutions, thrust and torque of forces and moments arising on these structures, a
dummy hub and a dummy rim were manufactured to
model the shape of the real hub and rim. Prior to propeller
tests, the force and moments on propellers were measured 0,4
experimentally and then deducted from the force and
moments generated on the propeller outfitted with hub and
0,3
rim. Thus, the force and torque arising only on blades
were determined. A similar method was used for hubless
propeller tests, but in that case the strut and dummy rim

KT
0,2 1
(instead of dummy hub and dummy rim) were examined. 2
For this test method, it is essential to take into account the 3
0,1 4
force and moment on strut, because the strut is used only
in tests and it does not exist in the real thruster.
4 0,0
0,0 0,2 0,4 0,6 0,8 1,0 1,2

8 J

Figure 9: Analytical and experimental relations of propeller

1
2 thrust coefficient versus advance ratio.
1 – experimental data for hub propeller, 2 – experimental
data for hubless propeller, 3 – calculations for hub
propeller, 4 – calculations for hubless propeller.

3 0,6
5

7
6
0,4
10*KQ

Figure 8: General view of test model setup. 1

1 – inflow velocity, 2 – shaft rotation, 3 – dynamometer shaft, 2
4 – structural component of the test rig, 5 – bracket, 6 – rim, 0,2 3
7 – blades of hubless propeller. 4

In the process of the tests, the dynamic and kinematic

parameters obtained by direct measurements were used to 0,0
0,0 0,2 0,4 0,6 0,8 1,0 1,2
calculate non-dimensional thrust, torque and efficiency
factors. These factors were treated as functions of advance J
ratio J. Figure 10: Analytical and experimental relations of propeller
torque coefficient versus advance ratio.
6 TEST RESULTS AND ANALYSIS
The relations of propeller thrust, torque and efficiency Symbols as per Fig. 9
factors versus advance ratio are compared with estimated
relations. This comparison is performed for hub propeller
and hubless propeller designs (Fig. 9-11). 0,8
It should be noted that the presented results refer to the
propeller blading system and do not include forces on
0,6
pipe and hub or moment induced by rim. Advance ratio J
corresponds to propulsor advance.
In general, the estimates compare fairly well with the 0,4


1
experimental data for moderate loads. Some discrepancies 2
are observed at small advance ratios in particular for the 3
0,2
torque coefficient KQ. For the propeller efficiency there is 4
a good agreement between estimates and experiments with
some overestimate of maximum values. The discrepancies 0,0
between estimations and experimental data can be put 0,0 0,2 0,4 0,6 0,8 1,0 1,2
down to calculation inaccuracy as well as uncertainty in J
assessment of moment on rim. Final conclusions can be Figure 11: Analytical and experimental relations of propeller
obtained after further studies. efficiency versus advance ratio.
Symbols as per Fig. 9
Comparison of data for hub propeller and hubless REFERENCES
propeller indicate that the efficiency curves are similar, Idelchik, I. Е. (ed.) (1992). Handbook on hydraulic
but hubless propeller has higher thrust and torque as resistance. Mashinostroenie.
compared to hub propeller. The latter is due to the fact
that in the hubless case, there is a small reduction of the Kinnas, S. A., Chang S.-H., He, L. & Johannessen, J. T.
flow velocity through the propeller disk and the radial size (2009). „Performance prediction of a cavitating RIM
of blades is larger. driven tunnel thruster‟. First International Symposium
on Marine Propulsors. SMP‟09, Trondheim, Norway.
7 CONCLUSIONS
Kulikov, S. V. & Khramkin, М. F. (1965). Waterjet
In the process of this investigation, a hub propeller and a
propulsors (theory and design). Sudostroenie.
hubless propeller were designed as options for the RIM-
driven thruster. Blade strength and cavitation Lebedev, E. L., Pershitz, R. Y., Rusetskiy, А. А.,
characteristics were estimated for these propellers. Avrashkov, N. S. & Tarasyuk, А. B. (1969). Ship
Experimental relations of thrust and torque of propeller steering units. Sudostroenie.
blading system versus advance ratio were obtained. Papir, А. N. (1970). Waterjet propulsors of small craft.
Special-purpose measurement procedures as well as Sudostroenie.
accessory equipments were developed for these
Rusetskiy, А. А & Mavlyudov, М. А. (2009). Waterjet
experimental studies.
propulsors. Krylov Shipbuilding Research Institute,
Based on the results of the studies the following St.Petersburg.
conclusions can be drawn:
Vasiliev, A.V. & Yakovlev, А. Yu. (2001). „Analytical
 The design method presented here for the RIM- method for estimation of axial pump hydrodynamic
driven thruster, including propeller blading system design, characteristics‟. Summaries of Papers presented at XL
makes it possible to achieve the required thruster Krylov Readings, pp.69-71, St. Petersburg, Russia.
performance.
Volkov, Yu.А. & Postnov, В.А. (1990). „Determination of
 The test rig developed for measuring the forces stressed-strained state of blades with complex
and moment on RIM-driven propeller meets the geometry using finite element method‟. Proceedings of
formulated requirements and proves to be suitable for XV All-Union Conference on shell and plate theory,
experimental investigation of hub and hubless propellers. Izdatelstvo Kazanskogo Universiteta.
 The conducted analytical and experimental Yakovlev, А. Yu. (2008). „Design of blading systems by
studies identified the ways for improvement of design direct optimization‟. Proceedings of the Krylov
methods for RIM-driven thrusters. Shipbuilding Research Institute 35, pp. 111-121.