You are on page 1of 43


The Specialist Committee on

Unconventional Propulsors
Final Report and
Recommendations to the 22nd ITTC

1 MEMBERSHIP AND MEETINGS • Memorial University of Newfoundland,

Canada, September 1997.
The 21st ITTC appointed the Specialist Com- • Università degli Studi di Genova, Italy,
mittee on Unconventional Propulsors with the February 1998.
following membership: • Bassin d’Essais des Carènes, France, Sep-
tember 1998.
• Dr. Neil Bose, (Canada), Chair; Memorial
University of Newfoundland, St. John’s.
• Dr. Michael Billet (USA), Secretary; Ap- 2 TASK SET FROM THE 21ST ITTC
plied Research Laboratory - Penn State,
State College, PA. Develop guidelines for carrying out propul-
• Dr. Poul Andersen (Denmark); Technical sion tests and extrapolating the results to full
University of Denmark, Lyngby. scale for propellers with ducts, partial ducts,
• Dr. Mehmet Atlar (UK); University of pre- and post-swirl devices, tip plates and z-
Newcastle upon Tyne, Newcastle upon drives.
• Mr. Christian Dugué (France); Bassin
d’Essais des Carènes, Val de Reuil. 3 INTRODUCTION
• Dr.Ing. Marco Ferrando (Italy); Università
degli Studi di Genova, Genova. In order to address extrapolation methodol-
• Dr. Wenhao Qian (China); Marine Design ogy for unconventional propulsors, the Com-
and Research Institute of China, Shanghai. mittee first reviewed known extrapolation
• Dr. Yide Shen (Belgium); University of methods currently in use for ships fitted with
Liège, Liège. different types of these devices. A summary of
these reviews is given in section 4 of this report.
At the first meeting of the Committee, Dr. Section 5 then describes the main extrapolation
Michael Billet was elected Secretary of the methods in further detail.
Committee. Meetings were held as follows:
Many of the extrapolation methods presen-
• University of Liège, Belgium, February tly in use for powering performance prediction
1997. of ships fitted with unconventional propulsors
are based on modifications of the ITTC 1978
method. The ITTC 1978 formulation was de- for powering prediction of ships fitted with un-
veloped primarily for single-screw open- conventional propulsors. Section 7 gives the
propeller ships and established an extrapolation Committee’s Conclusions and Recommenda-
methodology using interaction parameters such tions.
as wake and thrust deduction fraction as well
as Reynolds number based friction corrections.
Organisations have not only modified the ITTC 4. MODEL SCALE AND FULL SCALE
1978 approach for open propellers based on RESULTS OF VARIOUS EXTRAPOLA-
experience, but more importantly in the present TION METHODS FOR UNCONVEN-
context, developed new methodology for each TIONAL PROPULSORS
particular type of unconventional propulsor
treated by using these methods. However, on 4.1 Propeller Boss Cap Fins (PBCF)
the horizon are developing unsteady RANS
codes, new testing procedures for towing tanks, Recent activities concerning effective boss
application of large high-Reynolds number cap designs to improve the efficiency and
water tunnels, etc., that can more rigorously cavitation performance of marine screw pro-
address these issues. Also, there is still a great pellers have focused on the device known as
need for reliable full scale unconventional pro- “Propeller Boss Cap Fins” (PBCF). This was
pulsor performance data to validate any meth- proposed by a group of Japanese inventors,
odology. Ogura et al. (1988), as a means of increasing
the efficiency of a ship screw propeller. Almost
Devices added to the blade tips, such as all the research and development work on
bands, tip fins, bulbs, etc., and those added to PBCF have been reported through a series of
the propeller hub, such as boss cap fins, form a technical articles published by the technical
class of unconventional propulsor that can departments of the Mitsui OSK Lines Ltd.,
utilise relatively reliably the ITTC 1978 ex- West Japan Fluid Engineering Laboratory Co.
trapolation methodology. However, the effect Ltd and the Mikado Propeller Co Ltd. who are
of Reynolds number on this methodology the joint patent holders of the device.
needs to be resolved as devices that locally
modify the flow field to control cavitation or Ouchi (1988) and Ouchi et al. (1988 and
improve efficiency are often tested at relatively 1989) presented early research and develop-
low Reynolds number in the towing tank. The ment work on PBCF involving detailed model
more extreme cases where the propulsor can be tests, flow visualisations and the first full-scale
considered part of the ship hull, such as inte- measurements on the 44,979 GT PCC “Mer-
grated ducted propulsors, form a group of un- cury Ace”. These investigations reported a gain
conventional propulsors where the ITTC 1978 of 3-7% for the propeller efficiency through the
methodology is clearly not adequate. In some “reverse POT” (Propeller Open Test) in a cir-
cases, unconventional propulsors involving culating water channel, 2 to 2.3 % gain in the
pre- and post-swirl vanes, ducts, propeller pods, propulsive efficiency based on self-propulsion
etc., have been treated for extrapolation pur- tests and finally a gain of 4% for the power
poses by using a modified ITTC 1978 method- output on the actual vessel. Further investiga-
ology. In other cases new methods have been tions were also reported by Ouchi (1989),
developed for extrapolation and it is expected Ouchi & Tamashima (1989) and Ouchi et al
that the most reliable methods to be developed (1989,1990 and 1992) including 3-D LDV
in the future will make use of new methods. measurements in the model propeller slip-
stream, full-scale measurements with 11 ves-
Section 6 discusses the rationale behind the sels and cavitation and noise investigations.
Committee’s proposals for the future develop- The comparative analysis of the sea trial results
ment of more general extrapolation methods of 11 ships and their models indicated consid-

erable scale effect between the model and actu- crease. It was anticipated that the total gain
al measurements such that the efficiency gain would be somewhat reduced due to the fric-
at full scale could be two or three times that at tional drag of the fins. Indeed thrust and torque
the model scale. This was mainly attributed to measurements with the same model showed
laminar flow around the boss cap and possible 5% gain in efficiency with the PBCF confirm-
laminar flow separation at the back of the fins. ing the earlier predictions independently. It was
Finally a suitable scale factor, which could be also claimed that it was not apparent from the
determined from a large number of tests on respective studies that the PBCF type device
models and actual ships in the same manner as had to contain the same number of vanes as the
the roughness allowance in the resistance of number of the propeller blades to achieve the
ships, was recommended to compensate for the max efficiency gain as claimed in the PBCF
scale effect. patent.

Ouchi (1988) and Ouchi et al. (1990) also As far as extrapolation methods are con-
reported on various other aspects of PBCF as- cerned, no particular procedure dedicated to
sociated with the extrapolation of the powering PBCF has been reported in the above refer-
performance. They indicated that the presence ences. However, it appears that the patent
of a rudder significantly reduces the strength of holding companies have a well-established
the hub vortex and hence the gain in propeller procedure to quantify the power saving and
efficiency due to PBCF can be reduced by 10- hence the efficiency by the use of PBCF. This
30%. Self-propulsion tests to investigate the procedure is based upon the reverse POT set-
propulsive efficiency by PBCF indicated that up in their circulating water channel and the
the efficiency gain obtained from the self- experience gained over a considerable number
propulsion tests were similar to the gain ob- of full-scale measurements. Moreover, private
tained from the “reverse POT” tests with the communication with the West Japan Fluid En-
rudder. The effectiveness of the PBCF was re- gineering Laboratory Co. Ltd. by this commit-
ported to be hardly affected by the hull wake tee has confirmed that the standard extrapola-
suggesting that most of the efficiency gain was tion technique for the performance prediction is
due to the propeller flow itself, but it was im- based upon the ITTC 1978 procedure, as brie-
portant to take into account the effect of the fly reviewed in Section 5.6.
rudder. The thrust identity analysis of the self-
propulsion tests, in which the PBCF was con- Atlar and Patience (1998) gave a detailed
sidered as an appendage to the propeller, de- review of activities on various boss cap designs
monstrated that the effect of the PBCF pri- as well as on PBCF. Amongst them is a recent
marily appears as an increase in the relative– innovative concept, which is known as the Hub
rotative efficiency due to the reduction in the Vortex Vane (HVV), jointly developed by the
propeller torque due to the PBCF. Potsdam Model Basin, SVA and SCHOTTEL.
The HVV is a small vane propeller fixed to the
In order to shed more light on the function tip of a cone shaped boss cap within a limiting
of PBCF, Gearhart & McBride (1989) carried radial boundary where the tangential velocities
out a detailed flow analysis and power meas- due to the hub vortex are greater than those of
urements with PBCF. The experimental analy- the propeller. In this range the small vane pro-
sis of the efficiency gain from PBCF utilised peller diverts the high tangential velocities in
the results of LDV measurements in the ARL the direction of the jet, thereby generating ad-
Pennstate water tunnel with a model propeller ditional thrust. It is claimed that this mecha-
behind a hull including its rudder. This con- nism is different from the mechanism for
cluded that 2 % out of a total 6% efficiency PBCF where the fins are located usually
gain was due to thrust increase while the re- beyond this limiting radius and the fins them-
maining 4% was associated with torque de-
selves do not generate thrust. Unlike PBCF, the The special geometries are in general only lim-
number of the vanes of HVV is greater than ited modifications of the propeller. For this rea-
that of the propeller blade. In their product son the model testing and scaling of results to
briefs, Potsdam Model Basin(SVA) (1995) and full scale can in principle be done in the same
SCHOTTEL (1995), and in a detailed technical way as for a conventional propeller. In this
report by Schulze (1995), the results of LDV section a review of model testing methods and
measurements and cavitation details with HVV scaling of the results from such propellers is
models are presented. These publications report given. As far as end-plate and similar propel-
remarkable reductions in the hub vortex cavi- lers are concerned the literature often deals
tation by the HVV as well as a successful ap- with the propulsors separately, i.e. if and how
plication of HVV on the full scale claiming an they improve their efficiency, whereas scaling
increase of 3% in the propeller efficiency. and power prediction are not treated. Here,
only the latest references or those related to
scaling and prediction are given.
Atlar and Patience (1998) also reported on
a series of experimental investigations with
Sparenberg & de Vries (1987) by using
various boss caps, which were done in the Em-
linear optimisation theory designed and tested a
erson Cavitation Tunnel at Newcastle Univer-
3-bladed end-plate propeller. Open water tests
sity, over the last decade. These involved com-
were carried out in two different series for two
parative efficiency and cavitation performance
different rates of revolutions. Correspondingly,
measurements with different boss caps through
the Reynolds numbers (with respect to chord
the reverse open water tests in the presence of
length and relative inflow velocity at the 0.85R
no rudder and later on with a rudder for some
section) varied. There were clearly differences
caps, Atlar et al (1998). The cylindrical boss
caps tested were fitted with: fins similar to in the measured open-water curves (KQ and η,
PBCF; fins with end plates; fins as extension to but only small differences in KT) reflecting the
the propeller blades as Trailing Edge Flaps influence of Reynolds number. Over a range of
(TEF); an accelerating duct; a decelerating advance ratios, including the design value, the
duct and Boss Slots (or holes) (BoS). The latter efficiency was higher for the high Reynolds
are designed to increase the pressure in the hub number, as expected. The tests also included
vortex core. The comparison of the perfor- measurements with the end-plate propeller
mances for these caps displayed comparable working behind the model of a tanker, making
efficiency savings relative to a standard cone it possible to present the propeller efficiency in
shape type except for the cap fitted with the the behind condition.
decelerating duct which displayed a large
amount of efficiency loss. The BoS type cap Model testing of an optimally designed
displayed a superior suppression of the hub propeller with two-sided shifted end plates on
vortex cavitation, but also at the cost of a con- the blades was described by de Jong et al.
siderable efficiency loss. No particular mention (1992). The tests consisted of open water tests
has been made of extrapolation issues in these and cavitation tests. The former were carried
investigations. out at a rate of rotation sufficiently high to
avoid laminar-flow effects on the propeller
blades which was checked by further experi-
ments. In the cavitation tests uniform inflow
4.2 Tip Fins was considered. The interest was focused on
the end plate propeller and on comparisons
End-plate and other such propellers that with the same propeller without end plates and
have modified blade tips, have various names with a reference propeller. For that reason scale
like tip-fin, Kappel, winglet, etc., propellers. effects were not considered in greater detail
than as outlined above, nor were interaction

between propeller and ship or aspects of power procedure to be applied to test results with CLT
prediction dealt with. propellers request certain unexpected correc-
tions only known by (the designer) and hence
For the special end-plate propeller known not available for model basins. This makes it
as the CLT propeller, Hollstein et al. (1997) advisable to rely for the predictions of full-
outline the design of such a propeller for a bulk scale performance of the CLT propeller on the
carrier and compare this with data for a sister basis of direct calculations”. In their paper they
ship fitted with a conventional propeller. Ac- show calculated open-water characteristics of
cording to them no model tests were carried out, their CLT propeller as well as the characteris-
although resistance, open-water and self- tics of the conventional propeller. They also
propulsion test results for a conventional pro- give trial trip results for the ships with conven-
peller are available for the designer of the CLT tional and CLT propellers.
propeller. This is because “...the interpolation

Table 1 Propulsion coefficients, power and rate of revolutions obtained by model experiments
and three different extrapolations and by theoretical predictions (Andersen 1996, table 3).

DMI extrapola- ITTC-78 DMI-ITTC Design*

V = 9.03 m/s Extrapolation extrapolation
Tip-fin Conv.
Tip-fin Conv. Tip-fin Conv. Tip-fin Conv.
0.30 0.27 0.27 0.26 0.30 0.27 0.26 0.26
0.16 0.16 0.16 0.16 0.16 0.16 - -
ηR 1.00 1.02 1.00 1.02 1.00 1.02 1.00 1.00
ηH 1.21 1.16 1.16 1.14 1.21 1.16 - -
η0 0.63 0.62 0.68 0.66 0.67 0.65 - -
ηD 0.75 0.73 0.79 0.76 0.81 0.77 - -

η D tip fin
η D conventional 1.037 1 1.038 1 1.047 1 - -

PD [kW] 6720 6980 6410 6650 6270 6570 6850* 7260*

n [1/s] 1.95 1.99 1.98 2.01 1.95 1.99 1.97 2.01
CTh 1.58 1.49 1.46 1.42 1.58 1.49 1.55* 1.55*

* Design includes an increase in resistance (allowance) of 9 per cent

For the tip-fin or Kappel propeller with in- basis of model test results. One method was
tegrated fins in the tip region (bent blade tips) due to the model basin (Danish Maritime Insti-
Andersen (1996) carried out a comparative tute) where a correlation allowance and a cor-
study with the conventional propeller actually rection to the wake fraction coefficient was ap-
fitted on the ship as a reference. The model plied. The second method followed the ITTC
tests consisted of traditional tests, i.e. in addi- 1978 with the exception that a more detailed
tion to resistance tests for the ship model, scaling of the influence of friction on the open-
open-water, self-propulsion and cavitation tests water propeller characteristics was carried out.
for both conventional and tip-fin propeller This procedure was based on the flat plate fric-
models. Three different methods were used for tional resistance coefficient. The third method
making the full-scale power prediction on the combined the two others in applying the proce-
dure of the model basin (correlation allowance speed, vanished when the air cavity approached
and correction to the wake) and the correction its ultimate form, i.e. for Froude numbers
to the propeller open-water characteristics. The greater than 3. Hadler and Hecker (1968) indi-
three methods gave different required power, rectly concurred with this hypothesis, but they
demonstrating the need for rational procedures calculated the Froude number using the total
for such estimations (see table 1.). inflow velocity and the immersion of the shaft
as the length parameter. Finally, Olofsson
(1996) acknowledged an influence of FnD, (cal-
culated with the diameter as the length pa-
4.3 Surface Piercing Propellers
rameter), but he set to 4 the limiting value
beyond which the influence disappears. Sum-
marizing, it is widely recognized that the
Surface piercing propellers (SPP) can be
Froude number does affect the behavior of
included into the unconventional propulsor fa-
SPPs, and all authors have suggested the exis-
mily because of their particular mode of opera-
tence of a threshold value that limits this influ-
tion. However, system geometry and compo-
ence. The same general agreement has not been
nents for surface piercing propulsion is quite a
reached on the minimum value that must be
conventional concept for it includes only the
attained to avoid scaling problems. This is due
classical propeller-shaft-rudder layout.
to the different kind of Froude numbers that
have been used. Further research is required to
Current extrapolation methods are focused
identify the minimum value of Fn which must
on two items:
be achieved during open water tests. Never-
theless, provided that open water tests are per-
1) extrapolation to full scale of the model
formed at Fn beyond the threshold value, the
open water performance;
Froude number identity can be avoided during
2) extrapolation to full scale of the propeller
the tests without affecting the full scale per-
hull interaction (wake fraction, thrust de-
formance of the propeller.
duction and relative rotative efficiency).
According to Shiba (1953), surface tension
Comments on item 1). This extrapolation
plays its role when the propeller is about to be
procedure should take into account the influ-
fully ventilated. Complete ventilation is a
ence of two additional parameters, namely the
rather sudden phenomenon that can be corre-
Froude number, Fn, and the Weber number.
lated to a certain value of J called the critical
The influence of the Froude number is relevant
advance coefficient JCR. The critical advance
as the propeller acts at the interface between air
coefficient can roughly be located in the mid-
and water much like hulls. The Weber number
dle of the transition region and the sudden drop
is a ratio between inertial and surface tension
of KT and KQ identify its position. Indeed Shiba
forces. Its influence can be easily foreseen for
found a correlation between Wn and JCR for a
an SPP, which continually pierces the water
single propeller. Later work on this matter
(Ferrando and Scamardella 1996) acknowl-
edged the influence of the Weber number on
The first comments on the role of Fn are
the location of JCR, and the existence of a
due to Shiba (1953). In his investigation he
threshold value of Wn beyond which its influ-
pointed out that gravity affects the shape of the
ence disappears. Unfortunately there is not suf-
air cavity through the Bernoulli equation that
ficient evidence to suggest the existence of a
can be enforced at the boundary between water
unique threshold value. Further research on this
and the atmosphere vented cavity. Accordingly,
subject is therefore strongly needed. Anyway,
the influence of Fn based on the diameter of the
provided that open water tests are performed at
propeller as the length parameter and nD as the
Wn beyond the threshold value, the Weber

number identity can be avoided during the tests big expenditures like extensive tank testing or
without affecting the full scale performance of high standard sea trials. There is not a single
the propeller. known case study in the open literature
providing data for extrapolation purposes.
As far as is known, provided that open wa-
ter tests are performed in agreement with the In some favourable circumstances EHP is
above requirements, the performance of SPPs determined by means of towing tests, while for
can be scaled in the same fashion as conven- the great majority of applications an effective
tional propellers, i.e. by applying a Reynolds horse power estimate is performed using Savit-
number correction only. sky’s (1964) equations or something similar. In
both these instances there is not enough data
Comments on item 2). The propeller is usu- for a true extrapolation of the propeller-hull
ally located quite far from the hull (i.e. 1 to 2 interaction. The absence of sea trial results pre-
diameters away) and the projection in the verti- vents the development of a reliable full scale
cal plane of the immersed hull surface near the extrapolation.
propeller is small. As a consequence, the value
of the thrust deduction factor for SPPs is negli-
gible or zero. Furthermore, from a theoretical 4.4 Oscillating Propulsors
point of view the phenomenon of augmented
resistance for this kind of propulsion can be Oscillating propulsors are still at the resear-
treated with the same physical model used for ch and development stage. Only very few ac-
conventional propellers because no additional tual applications have been built and tested at
thrust affecting devices are involved. Hence the full scale and all of these have been for small
same scaling procedure adopted for conven- craft. Examples are the small boat driven by an
tional propulsion can apply also to SPPs. oscillating propulsor and outboard motor des-
igns built by Isshiki et al. (1987), the experi-
Surface piercing propulsion is mainly em- ments done by a group led by M. Triantafyllou
ployed on planing or semi-displacement craft, at MIT, and various oscillating propulsor des-
exploiting the ventilation of the transom. As igns used on human powered craft (e.g.
SPPs operate far behind the hull the contribu- Bennett 1996), including examples on human
tion of the potential wake is small. On the other powered submersibles (Skidmore et al. 1989).
hand, the viscous wake cannot be neglected, As a result, extrapolation methods for power-
because part of the propeller lies in the ing prediction for oscillating propulsors are
boundary layer of the hull. Considering that rare.
surface piercing propulsion does not include
any additional wake modifying device and that Other propulsors that operate with a similar
the physics of the wake production process is propulsive mechanism to the oscillating pro-
the same as in the case of conventional propul- pulsor are trochoidal propellers (a type of cy-
sion there are no obstacles to using the same cloidal propeller), the most recent example of
scaling procedure as that used in the case of which is the Whale Tail Wheel (Anon. 1998)
conventional propeller arrangements. being fitted to the inland waterways vessel
As regards relative rotative efficiency there “Ludwina”. Such propulsors have been studied
is no particular reason that prevents the appli- at model scale (e.g. Manen 1973, Bose and Lai
cation of traditional scaling methods. 1989, Riijarvi et al. 1994). A special type of
oscillating propulsor, that might be described
Setting aside the preceding considerations, as a hybrid between a conventional and an os-
almost all applications of surface propulsion cillating propulsor, is a propeller with cyclic
are in the pleasure craft trade. Generally, the pitch control (e.g. Gabriel and Atlar, 1998).
budget for the boat design does not allow for
Estimates of the full scale performance of fied ITTC 1978 method have been used to pre-
ships fitted with oscillating propulsors have dict full-scale performance from ship model
been made by Lai (1990) and Yamaguchi and test results. These predictions have been com-
Bose (1994). These have been based on con- pared with the results from sea trials or from
ventional approaches to powering prediction statistical sailing data.
using estimates of the hull efficiency to ac-
count for propulsor/hull interaction. Yamagu-
chi and Bose (1994) assumed neligible Table 2 Pre-swirl energy-saving devices.
hull/propulsor interaction leading to a hull effi-
ciency of one. The rationale for this estimate Name of pre-swirl device Energy- Energy-saving
was that as an oscillating propulsor design saving rate %
mechanism model test/trial
normally has relatively light loading, they are
Reaction fin PRI 4-8 / 4-9
usually large devices with relatively large Novel integrated duct IPI 4-5 / 4-6
swept area and large span relative to the ship Inflow compensate nozzle IPI, AFS 6-11 / 8
beam and/or draft. In view of this much of the Simplified compensate IPI, AFS 4-9 / 4-9
propulsor is not in close proximity to the hull nozzle
and the hull/propulsor interaction was assumed Wake-adapted duct IPI, AFS 4-9 / 4-9
Hydrodynamic fins IPI 4-6 / 3-9
to be small. For the ship under consideration, Fore-propeller hydrody- IPI 4-7 / 4-7
this was a conservative estimate as the hull ef- namic fin sector
ficiency was above one for the comparable Thrust shaft brackets PRI 4.5-12 / 5-8
conventional propeller system. Lai (1990), on Sheathed shaft bracket PRI 5-8 / 5-8
the other hand, assumed similar values of the Hydrodynamic partition IPI, AFS 2-3.7 / 2-3.7
wake and thrust deduction fraction as the con- Aperture fin PRI 2-4 / 2-4
ventional propeller system for the oscillating Stern-appended fin IPI, AFS 3/3
propulsor proposal when considering the pow- Flettner rotor at stern post IPI, AFS 8/
ering performance of four ships. This was Fore-propeller vane-wheel IPI, AFS 3-4 /
done in the absence of more detailed data as
the hull efficiencies for the four vessels varied
from 0.92 to 1.12.
Since 1984 MARIC has developed more
4.5 Pre and Post-Swirl Devices than ten ESDs such as the composite device of
simplified compensate nozzle (SCN) and Costa
Since the 1970s the development and appli- propulsion bulb (CPB), thrust shaft brackets
cation of hydrodynamic energy-saving devices (TSB), fore-propeller hydrodynamic fin sector
(ESD) have been demonstrated all over the (FPHFS) etc. (Qian et al., 1992a, Zhou et al.,
world. Tables 2-4 show energy savings from 1990, Qian et al., 1992b). Eight different types
pre- and post-swirl devices. The propeller in- have been put into use on more than 200 ships
flow compensate nozzles developed by over the range from 500 to 150,000 dwt with
Schneekluth have been installed on more than an annual 50,000 ton fuel-savings. In all cases
600 ships (Schneekluth, 1986). Powering per- a modified prediction method based on the
formance prediction for full-scale ships with ITTC 1978 method has been used in the pre-
pre- and post-swirl devices have been made diction of the full-scale performance from the
based on model test results. For most of these ship model test results. Practice indicates that
the ITTC 1978 method and the two dimension- the energy-saving rate (ESR) or the predicted
al Froude method have been used to predict the speed of the ships installed with ESDs correlate
full-scale performance for ships having a large well with the full scale data.
range of block coefficients. At MARIC several
different prediction methods based on a modi-

described in section 5.4. Included here are some

Table 3 Post-swirl energy-saving devices. observations from a set of tests done over a
range of Reynolds numbers which include, in
Name of post-swirl de- Energy- Energy-saving particular, some trends at higher than normally
vices saving rate % tested values of Reynolds number.
mechanism model test/trial
Fixed guide vane after RRE 4-6 / 5
propeller Extensive propulsion related tests were
Vane-wheel RRE 5-15 / 6-12 recently done on a tugboat model at Bassin
Rudder-appended thrust RRE 4-5 / 4-8 d'Essais des Carènes. This tug-boat was a twin-
fins screw vessel with two ducted propulsors. A
Reaction rudder (asym. RRE 2-4 / 2-4 rudder was placed closely behind each shroud
COSTA propulsion bulb DVP 2-4 / 2-4 (about 1/3 of the chord of the rudder and less
Propeller cap fins DEP 2-5 / 2-5 than 1/2 the length of the shrouds behind) so that
Eddy-eliminating compo- DEP 5.1 / 8.7 the interaction between rudder and propulsor had
site propeller the potential to be strong. The slope of the aft
Integrated duct-vane- RRE 5-10 / end of the boat was high, so that the propulsors
Rudder-appended Flettner PAT 10 / were working in inclined flow conditions. The
rotor distance between the shroud and the hull was
small which created the possibility of detached
Table 4 Pre and post-swirl energy-saving de- flow in some conditions and strong interaction
vices. between the hull and propulsors. Due to
restricted space, the propulsors were highly
Name of pre and post- Energy- Energy-saving
swirl devices saving rate % The model was about 4.5 m long, equipped
mechanism model test/trial with a propeller of about 20 cm diameter. The
Hydrodynamic fins & IPI, RRE / 9-18 speeds tested, based on Froude number scaling,
guide vane-wheel
COSTA propulsion bulb DVP, RRE 4-14 / 4-7.4
were about 1.5 m/s for the transit speed and
& rudder-appended about 0.5 m/s for high towing conditions. Open
thrust fins water, resistance, self-propulsion and varying
Duct-thrust fins after pro- DVP, RRE 4 / 4-5 load tests at all advance ratios were performed,
peller including measurements of the axial thrust of
COSTA propulsion bulb IPI, AFS, 6-12 / 4-12
& simplified compen- DVP
one shroud. The boat was a twin screw vessel,
sate nozzle which implies that the prediction and
Assembly propulsion de- IPI, RRE 5-15 / 5-15 extrapolation methods based on open water,
vices resistance and self propulsion tests should not be
Composite energy-saving IPI, RRE 8-14 / 8-14 affected by the problems of high wake as would
be expected on single-screw ships.
Additional tests were done to investigate
PRI – give a pre-rotation to the propeller inflow
IPI – improve propeller inflow some problems related to prediction and
AFS – alleviate flow separation extrapolation for ducted propellers and these
RRE – recover rotational energy from downstream have relevance also for some other types of
DVP – decrease viscous loss after propeller cap unconventional propulsors. These were:
DEP – decrease eddy after propeller cap - towing conditions at low speed (from 0.2
PAT – produce additional thrust
to 0.4 m/s) to detect any scale effects due to
low Reynolds number, especially on the
4.6 Ducted Propulsors
Some methods for the powering performance
- self propulsion tests without rudders
of ships fitted with ducted propellers are
(transit and towing), to separate effects from
interaction of the propulsor with the hull and The open water test of the whole propulsor
the propulsor with the rudder; shows a Reynolds number effect on KT and KQ
- resistance tests of the hull plus the two that affects the propeller characteristics in the
shrouds, without rudders and without speed range up to 4.5 m/s. The extrapolation of
propellers; the characteristics of the propeller at the Froude
- resistance tests of the shroud alone speed (about 0.5 and 1.5 m/s for towing and
without the propeller ("open water test"). transit) would not give the characteristics of the
propeller at the higher speed tested.
The drag of the shroud in the "open water"
test (figure 1) clearly shows a separation The plot of KT of the propeller (KT obtained
phenomenon at low speed and a transition speed. with the shroud but not including the thrust of
A speed of at least 2 m/s is necessary to get the shroud) versus KQ (figure 2) for the open
stable values away from this transition. This water tests and the two conditions of load
problem seems to be avoided in the behind varying tests (with and without rudders) shows
condition where the values are different. that over the whole range of advance ratio (from
bollard pull to almost no thrust) the three curves
are almost identical. This indicates that the
relative rotative efficiency is very close to 1 with
a precision of less than 1%. The dispersion
duct drag/velocity^2

20 between the curves is due to precision

uncertainty in the measurements, the importance
of which becomes relatively smaller for points
10 close to bollard.
0 .5

0 0 .4
0 1 2 3 4 5
Kt propeller

velocity m/s
0 .3

Figure 1. Shroud drag. The circles denote the

drag from the “open water test”; the squares 0 .2
are the values found from the resistance of
the hull tested with and without the shrouds
0 .1
fitted. 0 .0 5 0 .0 6 0 .0 7 0 .0 8 0 .0 9 0 .1
torque coefficient Kq
The tests of resistance with the shrouds and
without propellers showed that the shroud drag Figure 2. Thrust coefficient of the propeller
was about 25% of the total. Comparison of these versus torque coefficient. The open squares are
values with those obtained from the open water open water values; the solid squares are the
test on the shrouds, showed that the interaction behind condition with the rudder; and the solid
term was between 1 and 1.5% (the total drag of circles are the behind condition without the
the hull plus shrouds was 1% to 1.5 % higher rudder.
than the value obtained from the addition of the
naked hull resistance and open water drag of the Nevertheless, looking at the detail of the
shrouds). This interaction term is very small curves in figure 2 for high loads, there is an
despite the geometry. important difference between the curves.
Although the torque coefficient plots against
advance coefficient are not shown here, the

torque coefficient in the bollard condition for the only is plotted in figure 3) for open water and
open water and behind condition (KQ = 0.0915) behind condition with and without the rudder
was the same with a precision of less than 0.2 % shows again an effect of the rudder.
for both propellers. However, KQ becomes
0.0945 in the bollard condition with rudders (an
increase of 3%). Also, the KT of the propeller 0.4

thrust coefficient of duct

shows little change from 0.422 in open water to
0.414 in the behind condition without rudder (a 0.3
decrease of 2%), while there is an increase to
0.438 (plus 4%) with the rudder. Clearly, the
rudder has a "postswirl" effect on the propeller 0.1
rotor, increasing the thrust by 6% and the torque
by 3% between the two behind conditions. As a 0
result of this postswirl effect, thrust is increased
more than torque leading to an increase in
0.2 0.25 0.3 0.35 0.4 0.45
efficiency. These arguments together with figure
2 show that the effect of the rudder on the thrust coefficient of the propeller
relative rotative efficiency is far more important
than the difference between operating in open
water and the behind condition. Figure 3. Thrust coefficient of the duct versus
thrust coefficient of the propeller. The open
The plots against advance coefficient (see squares are open water values; the solid
figure 4 for thrust coefficient against advance squares are the behind condition with the
coefficient) show also that a KT identity for the rudder; and the solid circles are the behind
behind condition with rudder, based on the rotor condition without the rudder.
alone is meaningless for low advance ratio: KT
changes from 0.438 in the bollard pull condition Direct comparison
to 0.426 for JS = 0.3 while KT in open water has a the
maximum value of 0.422. A KT identity without open water and behind condition with rudder
rudder is also problematic as a value of KT = gives a confusing picture which at first sight
0.414 is obtained in open water for J0 = 0.25 (i.e. seems to show that there is no effect on the
less than the open water value and leading to a shroud at transit speeds as a result of operating in
negative wake). A KQ identity could be done in the hull wake and in an inclined flow, but that
this latter case, as the bollard KQ is the same in the effect at towing speeds is large. This is a
the open condition and the condition without false conclusion since comparing the open water
rudder, but the slope of the KQ versus JS is too results with the behind condition curve without
small to be precise at least for small J. When a the rudder shows a shift in the thrust of the
plot was drawn (not presented here) showing KT shroud. However, the thrust values at bollard for
of the propeller versus KQ, for the open water these two conditions are within 1%. Everything
condition and behind condition only, but with appears as if the shroud is "seeing" a flow having
additional points from propulsion tests at smaller less speed than the propeller rotor itself, that is to
Reynolds number (simulating towing at full say as if the wake fraction "seen" by the shroud
scale speeds of 2 to 4 knots), there was a big is a little bit larger than that "seen" by the
discrepancy between some of these last points propeller rotor. Is this due to a boundary layer
due to scale effects and/or large errors in effect of the hull on the pressure distribution of
measurement of very small forces. the shroud? Direct comparison of the open water
results with the behind condition with rudder
The two plots of KT of the duct versus KQ and gives then a confusing picture simply due to the
KT of duct versus KT of the propeller (the latter fact that for the same flow around and in the
shroud, the rudder gives a shift on thrust and
torque of the propeller rotor (postswirl effect) 4.7 Partial Ducts
which is an effect as large as the effect of the
hull on the shroud. But the postswirl effect Wake equalizing ducts were developed by
doesn't change the flow around the shroud as the Schneekluth (1986). They consist of two nozzle-
bollard thrust of the shroud indicates. shaped half ring ducts which are installed on
both sides of the stern ahead of the propeller.
These latter points have further implications Their diameters are about the same as the radius
on the identification of wake fraction from the of the propeller and their chord is smaller than
thrust identity. From the plots of thrust the diameter. Sometimes, only one duct is fitted
coefficient against advance coefficient (one of to the stern on one side of the propeller.
which is shown in figure 4) the following can be
seen. Comparing open water and behind The wake behind single-screw ships is non-
condition with rudder gives a wake fraction of homogeneous (i.e. there are very small velocities
about 10% for J around 0.3-0.4 (towing at the top of the propeller disc) and it is assumed
condition). However, in the case without rudder, that improving the homogeneity of the wake will
wake fraction is below 3%. The value of 1-w is improve the propulsion efficiency (essentially
surprisingly more than 1 in this region of the open water propeller efficiency). Optimising
advance ratio. There seems to be a difference of the angle of the partial duct to the stern under
regime between these low J values and the load conditions (accelerating the flow in the top
region of transit. part of the disc and slowing it in the lower part)
is said to improve the homogeneity of the wake.
Power savings of 5 to 10 % have been
thrust coefficient of propeller

0.4 reported (Schneekluth 1986), despite the fact that

this appendage creates an additional viscous drag.
On the other hand, this partial duct can create a
0.3 thrust and reduce eventual separation.

0.25 The extrapolation of powering performance

of wake equalising ducts fitted on large single-
screw ships are a typical example of the
0.15 difficulty of scaling data from model tests of
0 0.2 0.4 0.6 0.8 1 1.2 short appendages. The reason is that these
advance coefficient J devices are usually fitted to very large ships,
which means a very low towing speed in the
tank to comply with Froude number scaling of
Figure 4. Thrust coefficient of propeller versus the full scale ship speed. For a given ship speed,
advance coefficient. The open squares are Froude number scaling imposes a reduction of
open water values; the solid squares are the the model speed for a smaller scale model;
behind condition with the rudder; and the whereas the maximum model size remains
solid circles are the behind condition without basically the same for each facility.
the rudder. Reynolds effects can act in the opposite way:
In the region of transit speeds, the effect of
the rudders is about 0.04-0.05 on the advance - separation which occurs on the model may
coefficient of the rotor. The KT identity based not occur, or may occur over a smaller region,
upon the total KT may work because of a at full scale, hence a partial duct effective at
cancelling effect between the rudder and the reducing this separation on the model may be
shroud. less effective at full scale;

underestimated even for a conventional

- the boundary layer is relatively far smaller propeller.
at full scale than on the model, hence the size
and inclination of the duct optimised for the - The propeller open water efficiency
model may not be optimal on the full scale; actually decreased slightly with the duct for
this model, in contradiction to the principle
- friction of the partial duct is exaggerated on of operation of wake-equalising ducts.
the model due to the very low Reynolds
number on such small appendages (values of - The wake fraction and thrust deduction
Reynolds number on the appendages may be fraction changed slowly, but did not reach a
about 50,000 only). constant value over the range of speed tested.
In these tests the ducts reduced suction and
Friesch and Johannsen (1994) did tests in a increased wake so then the increase in
large cavitation tunnel (HYKAT) on one model propulsive efficiency came from an increase
of the size normally used in model towing tank in hull efficiency.
tests at speeds from 3.4 to 5.5 m/s in order to
investigate these Reynolds number effects. As In conclusion, partial ducts may result in
published results over a large Reynolds number energy saving at full scale, but this was not, and
range are rare, some of the implications of this probably cannot be proven by model tests at
work are described; however the Committee Froude speed by use of the present testing
cautions that published records from many more procedure. These tests done at higher Reynolds
tests of this type are needed for different types of number in a large cavitation tunnel showed an
unconventional propulsor in order to reliably erratic behaviour of the energy saving with test
identify true trends. The equivalent Froude speed speed.
for this model was about 1.2 m/s and
comparisons were made with towing tank tests No conclusion could be drawn from the
done at this speed. Their results are as follows: Froude scale speed tests as they were affected by
unrealistic levels of separation and other
- The wake equalising duct did not reduce the Reynolds number effects. Although not
drag of the ship (without the propeller). necessarily true for all partial ducts, the savings
indicated in this set of model tests were
- There was a reduction of power required for apparently due to increased hull efficiency; the
propulsion, but the behaviour of it was very open-water efficiency seemed to decrease a bit
Reynolds number dependant: from 2.5 % at and the wake was not more homogenous. Tests
Froude speed; to 9.6% at 3.5 m/s; to only 6% where separation is occurring on the hull without
at 5.5 m/s. the propeller and where this is reduced/
suppressed with the propeller, can lead to an
- The propeller characteristics were affected underestimation of suction (thrust deduction)
very little by the presence of the ducts (slight and an overestimation of wake fraction (i.e. an
increase of thrust and torque coefficients), overestimation of hull efficiency) when analysed
but, alarmingly for the testing community, using ITTC 1978 methods.
they became stable for speeds higher than 3.5 More generally, it is difficult to reliably
m/s only (at which speed the torque predict full scale performance of this type of
coefficient was reduced by 20% compared device from model tests at equivalent Froude
with the value obtained at Froude speed for a speed in the towing tank and even tests at higher
comparable thrust coefficient). This shows speeds in the cavitation tunnel can give uncertain
that open water tests can be used, but that trends. To address these issues, and to bridge the
Reynolds number effects on the Reynolds number range, further reliable
characteristics of the propeller may be correlations are needed between: tests done in
large cavitation tunnels at even higher speeds; done by Rains et al. (1981). They presented a
towing tank tests done for some equivalent semi-empirical approach to estimate the total
speeds at the lower end of the cavitation tunnel drag of a podded drive and comparative power
range; and full scale trials results. estimations. The investigation also included a
model test programme with a DD-963 class
Tests performed on a large tanker in the destroyer model fitted with 3 different drive
GTH at the Bassin d’Essais des Carènes at full systems: a conventional twin drive; a non-
speed (11 m/s) with partial ducts optimised in azimuthing twin tractor; and a pusher type
model tests at Froude speed showed a podded propulsor. The tractor type drive has its
completely different effect of the duct on the propeller ahead of the pod while this is re-
wake in the two situations. This implies that versed in the pusher type. The model tests in-
optimisation of the inclination of the duct should volved resistance tests with and without the
be done from the results of high Reynolds propulsion appendages, open water tests of the
number tests. Propulsive tests showed no energy propeller in isolation, which was the same for
saving for the high Reynolds number tests. all three propulsion systems, and self-
Specifically: propulsion tests. In the latter tests, only torque
for each propeller was measured and the pro-
- towing tank tests showed energy savings of pulsive coefficients analysed appear to be
1-2% depending on the propeller tested and based on the torque identity method. Although
speed simulated; the performance comparison of the three pro-
pulsion systems was presented in terms of the
- there was a difference in nature of the wake appendage drag, effective power and delivered
(without the partial duct) between tests at the power on the full scale, no specific reference
Froude speed and those at higher Reynolds was made as to how they were extrapolated to
number, in particular the occurrence of a full-scale.
region of high vorticity in the region where
the duct was to be fitted; Minsaas (1988) investigated a tractor type
non-azimuthing Z-drive, with a steering flap at
- by subtracting the wakes, the effect of the the tail of the drive strut, as a promising alter-
partial duct appeared to be completely native to water jet propulsion for speeds up to
different from the two tests at differing 55 knots. Neglecting the cavitation and pro-
speeds, the smoothing effect being cancelled peller induced drag, semi-empirical formulae
at higher Reynolds number; were given to calculate the viscous drag of the
unit due to the body, the strut within the slip-
- pressure fluctuations were improved with stream and strut outside the slipstream. In these
the duct fitted and reproduced in the higher formulae emphasis was placed on the consider-
Reynolds number tests similar signals to the able scale effect on the resistance estimation as
full scale. well as on its validity for only smooth surfaces.
Reynolds number effects in tests performed Later Halstensen and Leivdal (1990) described
at speeds of about 1 m/s may have more effect the development and full scale application of
on the predictions than not using the Froude this type of podded drive, called “SpeedZ”, to
scaling identity, especially for ship testing at Fjellstrand catamarans “Sleipner” and “Draup-
very small Froude numbers. ner” and to Westamarin SES catamarans “Su-
per Swede” and “Super Dane”. The study re-
ported on the open water tests done with sub
4.8 Z - Drives (Podded Propulsors) cavitating and partially cavitating propellers
fitted to the propulsion unit in the cavitation
An early investigation into the power pre- tunnel at MARINTEK, as well as towing tank
diction of vessels fitted with podded drives was tests with the naked hull. Although no details

of the power performance prediction were dis- screws.

cussed, the complexity of the resistance pre-
diction at full scale due to the effect of the pro- The final group of tests, which were per-
peller was reported. From this study it appears formed at MARIN, involved open water tests,
that the full-scale estimations were made based hull resistance tests with the same hull and self-
on the naked hull resistance tests from the propulsion tests. However, the propeller used
towing tank and the efficiency of the total pro- was a new design developed by the Krylov
pulsion unit obtained from the cavitation tunnel. Ship Research Institute which was customised
The latter included a resistance correction for to the Azipod and hull. Two sets of open water
the full scale. tests were done with this propeller, with and
without the Azipod housing. The calculated net
A recent study on the power performance efficiency of the Azipod in the open water tests,
aspects of azimuthing podded drives was pre- which was subject to low Reynolds number
sented by Kurimo et al. (1998) on the devel- effects, was found to be slightly lower than that
opment of two 14 MW Azipod units for the of the propeller alone at the corresponding ad-
cruise ship “Elation”. The study reported on an vance coefficient. The hull resistance tests were
extensive model test programme done in a done without the Azipod units. Following the
wind tunnel, towing tank and cavitation tunnel self-propulsion tests a value of 5-7% of power
at three institutions. The experiments done in saving was predicted on the full-scale when
the VTT towing tank involved open water tests compared with the original vessel fitted with a
with a stock propeller, hull resistance tests and conventional propulsion system. Although no
self-propulsion tests. The purpose of the open details were given, it was indicated that the
water tests was threefold: to determine the in- prediction was carried out according to MA-
teraction coefficients between the propeller and RIN’s “Standard” method for thrusters.
the unit; to compare the performance of two
pusher and one tractor type unit; and to provide In this study, the use of a special dyna-
a basis for the analyses of the propulsive coef- mometer for the open water tests highlighted
ficients. The open water tests were done using the importance of the complex interaction be-
a special dynamometer that could separately tween the Azipod housing and the propeller for
measure the thrust of the propeller and resis- a tractor type unit when the propeller thrust and
tance of the Azipod housing. Based on these the resistance of the pod housing were meas-
tests and separate wind tunnel experiments, ured separately. Therefore it was argued that
which involved resistance measurements of the adopted scale effect correction methods
various components of the propeller housing, a used in the extrapolation of the thruster propul-
tractor type Azipod was optimised. The resis- sion test results often concentrate on the scal-
tance tests were performed by using an earlier ing of the measured “resistance” of the housing.
hull model which had appendages for conven- This may be acceptable for the pusher type
tional twin screw drive. However these ap- Azipod, but not for the tractor type due to the
pendages were removed and the optimised complex interaction between the propeller and
Azipod housing was fitted. These tests were pod housing which results in an increased local
performed with and without the housing such pressure field behind the working propeller.
that the resistance of the housing could be ob- Therefore, emphasis was placed on the gap
tained. In the self-propulsion tests, the total between the Azipod housing and the propeller
thrust of the each Azipod unit was measured. It since low gap sizes would increase the resis-
was not possible to measure the propeller thrust tance of the housing and the measured thrust of
and the resistance of the pod separately due to the propeller, if they were separately measured,
the small scale of the model. The measured de- although the net thrust of the azipod unit would
livered power indicated a saving of 3-4% over be the same. Based on this argument, it was
the same hull driven by conventional twin claimed that the difference between the total
thrust of the unit and the thrust of the propeller represents one of the earliest applications of
in the tractor type of unit should not be related inviscid blade design theory and, as a result,
to Reynolds number dependent viscous effects introduces the problem of scaling. An ex-
alone. Also, more detailed methods for the ex- trapolation method developed by the ITTC
trapolation of test results from tractor type de- community has been widely used for commer-
vices are recommended to be developed. cial ships having an open screw propeller to
predict full-scale powering performance. The
Mewis (1998) has discussed recent devel- ITTC 1978 methodology uses resistance tests
opments in large podded drives with a specific of the ship model, open-water tests of the
emphasis on the range of their efficiency im- model open propeller, and model-scale self-
provements and difficulties in power perfor- propulsion tests, to determine interaction coef-
mance prediction. This relates mainly to the ficients, which are then scaled to predict full-
tractor types, a factor which was also pointed scale performance. However, this methodolo-
out by Kurimo (1998). A part of the difficulties gy and the scaling of its interaction coefficients
has been associated with the different configu- are based on particular commercial ship data-
rations used in open water tests and the treat- bases. The direct application of this methodol-
ment of the thrust force measured during these ogy to complex propulsors such as ducted pro-
tests. In order to demonstrate these difficulties, pellers or pump jets, water jets and various
open water tests of a tractor type Azipod drive novel forms of open screw propellers continues
have been carried out in three different con- to be an unresolved issue.
figurations as reported by the Propulsion
Committee of this Conference. Any propulsor configuration and its per-
formance are highly dependent upon the hy-
Kurimo (1998) presented the results of sea drodynamic characteristics of the ship
trials with the cruise ship “Elation” which were hull/marine vehicle. Energy caused distortions
carried out in the Gulf of Finland in December are present in the ingested flow due to the skin
1997. The trial results involved the presenta- friction drag of the ship hull/marine vehicle
tion of speed measurements, cavitation obser- and upstream appendages. It is in this envi-
vations, pressure pulse measurements and ma- ronment that wake-adapted complex propulsors
noeuvring tests. The trials indicated that at the excel. Complex propulsors provide more des-
full power of 2 x 14MW, the vessel achieved ign options with which to meet the many addi-
approximately 0.55 knots greater speed than tional performance requirements in these com-
the mean value of the corresponding speeds of plex wakes. These new requirements include,
the six previous sister ships, which have con- but are not limited to: 1. minimization of un-
ventional twin-screw diesel electric drives. De- steady forces, 2. reduction in radiated noise, 3.
spite this, the scatter between the predictions improved efficiency for heavily loaded pro-
from the different model basins, which used pellers, 4. reduction of cavitation and elimina-
different extrapolation methods, was large and tion of specific cavitation types, 5. improved
designers were cautioned to treat test results off-design performance, and 6. enhanced ship-
with great care to avoid optimistic expectations. maneuvering characteristics.

The principles employed and the problems

5 DESCRIPTION OF EXTRAPOLATION encountered in designing either an open screw
METHODS AND TEST PROCEDURES propeller or a complex propulsor are similar.
However, the design methods employed are
different for the two cases. First of all, the
5.1 Momentum Methods design of a wake-adapted complex propulsor
having a duct and/or multi-blade rows treats
The propulsion of a ship or marine vehicle the propulsor as a unit. Therefore, its model-

scale testing procedure and scaling methodolo-

gy will be different than that developed for an A relationship to determine required rotor
open screw propeller. One present-day method thrust is determined by integrating the mo-
of designing complex propulsors includes a mentum in the stream wise direction between a
combination of a vortex lifting-surface method station located far upstream of the propulsor
with a computational fluid dynamics based and a station located after the body and can be
through-flow analysis method. These two
expressed as T = m ( ∆V ) where T is the rotor
methods allow for hull interactions, viscous
effects and blade-to-blade row effects to be in- thrust, m is the mass flow rate and ∆V is the
cluded as well as three-dimensional effects integrated change in stream wise momentum.
such as streamline curvature, radial pressure An important assumption in this relationship is
gradients, and secondary flow (Kerwin et al. that free-stream static pressure exists at each
(1994) and Schott (1996)). station.

Interest in multi-blade row propulsors has An incoming velocity profile near the plane
developed because of increasing efficiency of the propulsor and the predicted hull drag co-
demands for high-speed surface ships and efficient are necessary to start the design proc-
submerged vehicles. The attractiveness of ess. This has been traditionally obtained from
ducted propulsors for these applications is due model scale tests with corrections for variations
to the ability to design a propulsor having a in Reynolds number. In some cases, analytical
lower blade relative velocity and a higher effi- predictions are also being used due to more re-
ciency than can be achieved by an open pro- liable computational fluid dynamic procedures
peller in these applications. The similarities and their validation to specific geometries
which exist between ducted propulsors and (Larsson et al. (1998), Arabshahi et al. (1998),
axial-flow compressors and liquid pumps have Zierke et al. (1997) and Stern et al. (1996).
lead to the application of momentum analysis
methods. Nowhere are different testing meth- It is also necessary to determine the total
odologies more evident between an open pro- drag of the hull, which includes increments due
peller and a complex multi-blade row marine to appendages, etc. The flow field solution re-
propulsor than in reviewing a momentum based quires the calculation of the frictional drag on
design methodology. all the various propulsor components. It must
be remembered that the pressure drag of the
Early efforts in developing momentum afterbody, which is initially included only as a
based design methodology are given by Wisli- bare-body drag coefficient, is substantially
cenus (1960 and 1968), Henderson et al. modified by the addition of the propulsor.
(1967), and Bruce et al. (1974). This method However, this effect is initially estimated from
determines the type, size, and design of any the modified pressure distribution with the pro-
multiblade row propulsor from an optimisation pulsor/hull combination.
of the mass flow through the propulsor to
achieve design goals. An optimised blade-row It must be emphasised that this momentum-
spanwise-circulation distribution is obtained based design procedure is applied where there
from minimising the energy losses in the pro- is incoming vorticity in the flow. The approach
pulsor and in the discharge jet. A detailed uses an inviscid calculation of the flow field
analysis of the flow field at various stations and is coupled with an energy analysis through
through the propulsor follows by solving the the propulsor. Therefore, all the energy losses
momentum, continuity and energy equations. through the propulsor are calculated and the
The upstream boundary conditions are the mass efficiency of the rotor becomes the hydraulic
flow rate, the momentum and the kinetic ener- efficiency, which is the ratio of the energy
gy of the ingested flow. placed in the fluid to the shaft energy. Solu-
tions to the flow field are obtained by using the scale ship drag which must include appendage
Streamline Curvature Method (Treaster, 1969). drag and propulsor/hull interactions. Current
Today, this has been extended to a RANS so- attempts to do this utilise CFD procedures, lar-
lution using computational fluid dynamics ge water tunnel tests, and an extended analysis
modeling. of propulsion data.

This design procedure for complex propul-

sors does not require open water tests for ex- 5.2 Other General Methods
trapolation methodology to predict full-scale
speed and rpm. In fact, testing of these wake- Other extrapolation methods are being de-
adapted complex propulsors in uniform flow veloped using self-propulsion testing, the phi-
significantly reduces propulsor efficiency and losophy of which could be considered for un-
introduces corrections that are not well defined. conventional propulsors. For these methods an
Therefore, only model-scale propulsion test open water test is not necessary, only a load
data is required for extrapolation to full scale. varying self-propulsion test is required.
This model test is conducted over a range of
over/under loading conditions. This is usually A modern MARIN method based on a form
achieved by varying the rotor rpm at constant factor (1+k) concept has been developed for
ship velocity. Resistance test data or calcula- some ship configurations for the extrapolation
tions are only necessary for the design. of propulsion test data to full scale. The form
factor is the ratio of viscous resistance to the
At these model scale conditions, the effi- flat plate drag based on the ITTC 1957 formula.
ciency can be defined as In general, this factor can be determined for
each hull form using low rpm self-propulsion
( RTm − FD )Vm effective power measurement data. The scale effect on resis-
ηD = = tance (FD) is determined from the standard
2π q m nm delivered power
equation including an incremental resistance
coefficient (CA). The measured relationship
and between the thrust coefficient (KT) and the ap-
parent advance ratio (JV) from the propulsion
FD = {(1 + k )(C fm − C fs ) − C A } ρ m S m Vm tests is corrected for both wake and propeller

2 blade friction scale effects to predict power and

From the model scale KT and JV curve, correc- rpm. This is effectively an ITTC 1978 method
tions are made to account for Reynolds number with the wake fraction found from a statistical
differences to full-scale first by skin friction method based on previous test results rather
corrections to the propulsor components than directly through open water tests of the
( ∆K T , ∆K Q ) and secondly, by corrections to actual model components.
the wake ( ∆n ). It is very important to note that
these corrections cannot account for separating An extensive analysis of typical combat-
flows. ant/auxiliary Naval ship data has been conduct-
ed by the Naval Surface Warfare Center,
In summary, the model scale propulsion Carderock Division, to validate model test ex-
test provides a KT versus JV curve. This curve trapolation procedures (Karafiath, 1997). The
must be corrected for Reynolds number effects developed prediction methodology uses a cor-
both on the propulsor and on the hull which relation allowance (CA) to account for differ-
produces a wake scale effect. The weak point ences between ship and model roughness and
of this or any other extrapolation methodology for other variables that influence the powering
remains the determination/prediction of the full prediction. Thus, the propulsion test is con-
ducted at an overload condition with an added

tow force (FD) to overcome the effects of an of the propeller are scaled following ITTC
additional frictional resistance and achieve 1978 procedure while no scale effect correction
equivalent thrust loading. This is accomplished is applied to the thrust/drag produced by the
by varying the propeller rpm at a constant ve- pre-swirl stator.
locity. One basic assumption of this method is
that the efficiency of the model propeller is the Van et al. (1993) state that this method
same as that of the ship propeller; this is ap- provides a better full scale prediction than the
propriate for small wake fractions. This is ITTC 1978 procedure. This statement is sup-
again an ITTC 1978 type method, but dis- ported by the close agreement between model
penses with the form factor, open water tests and full scale predictions presented, that is not
and other corrections. found by using the ITTC 1978 procedure di-
Both of the above methods have limited
application for unconventional propulsors since It is well known that the ITTC 1978 corre-
the previous test data and correlation coeffi- lation procedure fails to correctly scale the per-
cients, needed for the methods to work, do not formance of unconventional propulsion sys-
exist. tems, and this is due to two main causes. The
first one is that whatever device is applied to
the hull, it generally has a longitudinal dimen-
5.3 Performance Prediction Methods For sion which is much less than that of the hull
Ships With Pre-Swirl Stators itself. Given the usual testing speeds, this pro-
duces a Reynolds number corresponding to a
Van, Kim and Lee of KRISO (1993) pro- laminar flow on the device. The second cause
posed two alternative procedures for perfor- is that the model hull has a boundary layer that
mance prediction for ships fitted with pre-swirl differs from the full scale one both in thickness
stators. The authors explicitly state that the two and in velocity distribution. Therefore, the in-
procedures, named A and B respectively, basi- teraction between the hull and the special devi-
cally follow the ITTC 1978 method. In answer ce can seldom be correctly reproduced at model
to a request for clarification Van pointed out scale if it strongly modifies the flow around the
that at their towing tank the form factor method hull.
is not used. Accordingly these procedures are In this particular case of the pre-swirl stator,
not true variations of the ITTC 1978 correla- we can assume that the flow around the stern is
tion methodology as they follow the 2-D ITTC not overly affected. This can be argued fol-
1957 approach. The two procedures are dis- lowing the principle of operation of the pre-
cussed separately. swirl stator that is designed to produce a rota-
tional speed component opposite to that in-
Method A. In this method, the propeller duced by the propeller. For this reason, testing
and the stator are considered as a propulsion the stator together with the open propeller in-
system and are tested together. This assump- sures a higher Reynolds number on the stator
tion implies that in both open water and self- itself due to the propeller-induced velocities.
propulsion tests, the thrusts of propeller and Probably this is the main reason of the better
stator are measured simultaneously and their agreement of Method A performance predic-
sum is used as the thrust of the propulsion sys- tion with the model results.
tem. The hull resistance is scaled according to
the ITTC 1957 method (i.e. a form factor is not Method B. This procedure does not require
used). the joint test of the stator and the propeller be-
cause the stator is tested with and considered
In his correspondence with the Committee, part of the hull. On the other hand, it requires
Van stated that the open water characteristics that a double set of resistance and self-
propulsions test are done with and without the General remarks. Both of the proposed
stator. methods are basically 2-D procedures that can
be regarded as variations of the ITTC 1957 cor-
The scaling process is again the two dimen- relation approach, but in principle these tech-
sional approach of the ITTC 1957 method with niques could be used also with the ITTC 1978
an exception made for the determination of the 3D correlation procedure.
full-scale wake, which is performed by means
of the following formula that closely resembles From a theoretical point of view, the pro-
that suggested by the ITTC 1978 correlation posed methods are acceptable when applied to
procedure: special propulsive devices that do not consider-
ably alter the flow around the hull. In other
wS = (t MO + 0.04 ) words, these methods are suitable for scaling
C FS + C A the performance of propulsive devices whose
+ (wMO − t MO − 0.04 ) + (wMS − wMO ) effect is mainly confined to altering the pro-
C FM peller inflow, without affecting the pressure
field around the hull. Since these methods ad-
where: dress the scaling problem only from the poten-
wS = ship wake tial flow point of view, they are not suited to
wMO = model wake without stator treat such devices that could produce a consid-
wMS = model wake with stator erable variation of the pressure field acting on
tMO = model thrust deduction without stator. the hull. Actually, a hull pressure-modifying
device could alter the characteristics of the
while the standard ITTC 1978 ship wake is: flow around the model hull, e.g. the extent of
laminar separation. In this case, the proposed
wS = (t + 0.04 ) methods will probably produce results as inac-
(1 + k ) C FS + ∆C F curate as those of the unmodified ITTC 1978
+ (wM − t − 0.04) procedure. The lack of sea trials results pre-
(1 + k ) C FM vents a practical evaluation of the capability of
these methods to correlate model test results
The major difference compared with the with the actual performance of the ship. Thus,
ITTC 1978 formulation is the term (wMS – wMO). further effort is required to validate these
Since in the opinion of Van et al. (1993) the methods.
main effect of the stator is the increase of the
angles of attack of the propeller blade sections, From an experimental point of view, if the
the stator action can be considered to be a two proposed techniques are equally reliable, it
mainly potential phenomenon. Thus, the differ- appears that Method B would be preferable.
ence in wakes with and without stator can be Actually, the testing procedure related to
directly transferred to full scale. Method B requires additional resistance and
self-propulsion tests; this is its major drawback,
If the flow on the hull is not overly affected but the procedure is straightforward and does
by the presence of the stator, this assumption not require any special test rig. On the contrary,
looks reasonable and this procedure is accept- an ad hoc test rig is necessary to simultane-
able. The same would not be true in the case of ously measure the thrust of the propeller and of
other devices that accelerate or decelerate the the stator as required by Method A. The time
flow on the hull, like ducts, partial ducts etc. consumption of this technique is comparable
Van et al. (1993) state also that Method B ex- with that of standard tank practice, but the re-
hibits a good agreement with power savings quired special equipment would not be feasible
obtained at model scale. or available in all of the towing tanks.

(1984). In the case of the open water screw test,

5.4 ITTC 1978 Modified Methods for Duct- the equivalent open water velocity in which the
ed Propellers. screw works in the nozzle must be known. The
velocity change is expressed by a ∆J correction
Some ITTC 1978 modified methods have
been proposed for extrapolation of powering UN 1−τ
∆J = = J (1 + 1 + τCT )
performance of ducted propellers. Stierman nD 2τ
(1984) presented three methods: where the
nozzle is considered as an appendage; where where τ is the thrust ratio TP/TT. The purpose of
the nozzle is treated as a part of the propulsion the ∆J correction is to find the correct subdivi-
unit; or where the screw, nozzle and hull are sion of the wake fraction into a potential and a
treated as three interacting objects respectively. viscous part.

Nozzle as an appendage of the hull. The In this method, the influence of the nozzle
nozzle is primarily considered as a flow regu- on the screw is barely taken into account. A
lator. The resistance test is done with the noz- nozzle limits the radial outflow of a screw and
zle behind the hull and the open water test with therefore, the KT/KQ ratio will not be measured
the screw alone. correctly during the open water test.

The thrust deduction fraction is defined as Nozzle as a part of the propulsion unit.
Here the open water test is done with the screw
T − R H+ N + nozzle system and the resistance test is done
t= P
TP with the naked hull. The thrust deduction frac-
where Tp is the propeller thrust and RH+N is the tion is defined using the naked hull resistance,
resistance of the hull with nozzle. The wake and the total screw + nozzle thrust
fraction is found by propeller thrust identity
V − VA TP + T N − R H
w= t=
where V is the model speed and VA is the ad-
vance velocity of the screw. The total resis- In this case, VA in the defining formula of the
tance of the hull with nozzle is extrapolated by wake fraction is the entrance velocity into the
subtracting the estimated model nozzle resis- screw + nozzle system. The naked hull resis-
tance from the measured total resistance, scal- tance is extrapolated according to the ITTC
ing the resistance of the naked hull according 1978 guidelines. The scaling of the screw
to the ITTC 1978 method, and adding again the thrust and torque from the open water test is
estimated full size nozzle resistance. It is as- performed with the same ∆KTP and ∆KQ correc-
sumed that the formula to estimate the nozzle tions as proposed by the ITTC 1978. The noz-
resistance does not introduce too large errors zle thrust coefficient KTN should also be scaled.
because the nozzle resistance only amounts to a The resistance difference ∆KTN is roughly esti-
few percent of the total resistance. Note that mated using a flat plate friction line and a for-
this is in contrast to the tests described in sec- mula according to Hoerner. The interaction co-
tion 4.6. efficients are extrapolated using the ITTC 1978
method or by considering the effective power
The scaling of the propeller characteristics of the screw + nozzle system in their normal
is performed in exactly the same way as out- position and in a position far behind the hull.
lined in the ITTC 1978 method. The interaction
coefficients are extrapolated either according to The objections against the method are two-
the ITTC 1978 or by the method of Stierman fold:
(1) the nozzle thrust must necessarily be meas-
ured during the self-propulsion test; This method has been used to predict the
(2) the action of the screw + nozzle system is performance of ships fitted with energy saving
different in open and behind conditions. devices such as the simplified compensate noz-
In the behind condition the nozzle produces a zle; COSTA propulsion bulb; thrust shaft
larger thrust due to the contracting inflow at the brackets; fore-propeller hydrodynamic fin sec-
stern. For instance, in an open water condition tor; COSTA propulsion bulb and rudder-
τ = TP/TT = 0.90 may be found, while τ = 0.70 appended thrust fins; composite energy-saving
in the behind condition. Such an effect can be technique; etc.
seen also on a twin screw ship even with no
contracting inflow, perhaps due to boundary The size of MARIC’s towing tank is
layer or inclined flow influences. Due to this 70*5*2.5m. The length of the geosim ship
discrepancy, it is incorrect to determine the models were 3.5 to 4.5m. The diameters of
wake fraction by using the total thrust identity single and twin-propeller models Dm are bigger
axiom. than 0.12m and 0.11m respectively. The fol-
lowing method is used for the energy saving
Screw, nozzle and hull as three interacting devices described in section 4.5 and tested at
objects. To assess the interaction between the MARIC. The method utilises resistance, self-
screw and the nozzle, an open water test is car- propulsion and open water tests. The open wa-
ried out with the ducted screw. The resistance ter test is done with the propeller alone; the re-
test is performed with the nozzle behind the sistance and self-propulsion tests are done with
hull to take into account the interaction be- and without the energy saving devices. This
tween the nozzle and the hull. The thrust de- testing procedure is described in the following
duction and wake fraction are defined as in paragraphs.
method 1, by making use of the hull + nozzle
resistance, the screw thrust measured during Resistance test. The resistance test of the
the self-propulsion test, and the entrance ve- single-screw ship model is usually done
locity VA into the screw disk. The open water without the appendages such as bilge keel etc.
velocity VA in which the screw acts is the sum Measurements are made of the total model re-
of the translation velocity of the screw + nozzle sistance Rm, towing speed Vm and the water
system and the nozzle induced velocity. The temperature tm at the same loaded condition as
screw must be out of the nozzle. The KTP/KQ the propulsion test. To get a reliable value of
ratio remains constant, but the entrance veloc- (1+k) more data are taken in the low speed area
ity is changed. This means a shift of the J-axis than elsewhere. For the power prediction of the
of the open water diagram with a ∆J-correction, full-scale ship with the above mentioned ener-
standing for the dimensionless nozzle induced gy saving devices, the resistance test of the
velocity. The nozzle induced velocity can be ship model is usually done with and without
calculated by momentum methods, and vortex the above mentioned energy saving devices to
ring or sheet methods. compare their influence on the resistance per-
The resistance of the hull and the nozzle is
separately extrapolated, as described in method Open-water test. An open water test is
1. The propeller characteristics are scaled using done for the propeller model used in the pro-
the well-known ∆KT and ∆KQ corrections. pulsion test. In this test the propeller submer-
gence is bigger than its diameter. The advance
speeds vary from 0 to the case of zero thrust
5.5 The Modified Full-Scale Performance while the rotational speed is kept constant
Prediction Method (MARIC Method e.g. during the test. The results of the open-water
Zhao et al., 1988). test are used in the analysis of the self-

propulsion test results. The Rn of the propeller

model in the open-water test should reach the and
critical value 3*105 as far as possible. For the
power prediction of the full-scale ship with the n S = n m / λ1 / 2
above mentioned energy saving devices, the
open-water test is usually carried out without
the energy saving device.
The following equations can be easily de-
Self-propulsion test. When performing the
self-propulsion test, the model is tested at a 2
minimum of 5 different speeds. At each speed PES = RTS VS and RTS = 0.5CTS ρ S VS S S
the propelling forces Zm can be varied by PDm = 2πQm n m λ3.5 ρ S / ρ m
changing the propeller revolution rate to enable
the skin-friction correction FD to vary. The When the ITTC 1978 method is used
thrust produced by the propeller should meet
the following condition: CTS = (1 + k )C FS + CW + ∆C F
Tm (1 − t m ) + Z m = Rm
The ship self-propulsion point is given by the CW = CTm − C Fm (1 + k )
following condition:
Z m = FD = ρ m S mV m {C FM (1 + k ) − C FS (1 + k ) − ∆C F }
2 [
∆C F = 105(K S / LWL )
1/ 3
− 0.64 10 −3
2 K S = 150 × 10 m −6

The self-propulsion points can be calculated

when Vm, nm, Tm, Qm, Zm have been recorded. [
C Tm = Rm / 0.5 ρ mVm S m
For the power prediction of the full-scale ship
with the above mentioned energy saving de- and the friction coefficient is the ITTC 1957
vices, the self-propulsion test of the ship model line.
is usually carried out with and without the en-
ergy saving device in order to compare their When the Froude method is used
influence on the propulsion performance.
CTS = C FS + C R + ∆C F +
The ship model speed Vm is first corrected C R = CTm − C Fm
for blockage (a procedure following the Emer-
son method is used – see the Resistance Com-
mittee report of the 19th ITTC, 1990). The where, ∆C F + is the resistance correction coef-
testing procedure of the ship model with the ficient. It can be adjusted to keep CTS calculated
energy saving device is the same as that from both methods nearly the same. Normally
without the energy saving device. The wetted it can be taken from the following empirical
surface area of the energy saving device is not formulae:
taken into account in the transformation of the
results to the full scale. 10 3 ∆C F + = 0.75 − 0.00352 LWL

The basic transformation. The following or

formulae can be derived from the kinematical
and dynamical similarities
PE = (RTm − FD )Vm λ3.5 ρ S / ρ m
V S = λ1 / 2V m
where, FD is the resistance correction of the rical stern and twin-skeg forms, the ship’s pre-
self-propulsion point. It can be estimated from rotating efficiency η n will lead to a propulsion
the following formulae: efficiency in the form

FD = 0.5 ρ mVm S m [(1 + k )C Fm − (1 + k )C FS − ∆C F ]

η n = n0 / n
FD + = 0.5 ρ mVm S m [C Fm − C FS − ∆C F + ]

where, n is the revolution rate of the propeller

behind the asymmetrical stern and n0 is the cor-
Determination of the propulsion coefficient. responding nominal revolution rate. This can
The coefficients KTm and KQm are obtained from be obtained by taking the average value be-
the self-propulsion test as follows tween the revolution rates of right-hand and
left-hand turning propeller models in the self-
K Tm = Tm / ρ m nm Dm
2 4
) propulsion tests. Or it can be the revolution rate
of this propeller model in the open-water test
K Qm = Qm / ρ m nm Dm
2 5
) with VA as its axial inflow speed when its thrust
reaches the value in the following equation
Based on the thrust identity method J0m, KQ0m
and η 0 m can be read from the propeller model RVS TV A n0 Q0 (1 − t )
ηD = = = η 0η nη Rη H
characteristics whose scale effect correction 2πnQ 2πn0 Q0 n Q (1 − w)
has been made with KTm as the input data. Fol- The prediction of the full-scale performance for
lowing this, the propulsion coefficients can single screw ships. The scale effect of the pro-
therefore be calculated as follows pulsion factors recorded in the self-propulsion
test revised at MARIC by the following for-
η Rm = K Q 0 m / K Qm mulae
1− wm = J 0 m nm Dm / Vm
1− t m = ( Rm − FD ) / Tm wS = t m + (wm − t m )
[C FS (1 + k ) + ∆C F ]
[C Fm (1 + k )]
η Hm = (1− t m ) / (1− wm )
t S = t m − 0.08834 + 0.01262 LWLm
η ′ =η η η
Dm 0 m Hm Rm
η RS = η Rm + 0.08645 − 0.01236 LWLm

where η Dm should be coincident with η Dm (in J0s, KQ0s and η0S can be read from the load coef-
the above equation η Dm = PE / PDm ). ficient of the full-scale propeller KTS/JS2 chart
with the KTS/JS2 value as the input.
Basic transformation for twin-screw ships.
With Tm and Qm values for both propeller mod- K TS / J S =
2 D (1 − t S )(1 − wS )
2 2
els as the input data, the power and the thrust
deduction fraction can be calculated. For de-
termining wm, η 0 m , and η Rm the respective av- The following can then be calculated
erage values for both propeller models should
be used. (1 − wS )VS
ns =
J 0S D
Basic transformation for ships with asym- 3
PDS = 2πρ S D 5 n S K Q 0 S / η RS
metrical stern form or twin-skeg form. For
ships with a big tangential component in the η HS = (1 − t S ) / (1 − wS )
propeller inflow, such as those with asymmet- η DS = PE / PDS or

η DS = η 0 Sη RSη HS Since the presence of a rudder would weaken

the swirl in the propeller slipstream, it is es-
sential to place the rudder behind the propeller
The additional increment of the bilge keel in these tests. By using this configuration, the
and the air resistance of the superstructure over measurements of thrust and torque are taken
the water level can be estimated as 4-6% of PDS.. for the propeller with and without the PBCF.
The rate nS is proportional to (1.04-1.06)1/3. The Based on the measured values of the torque, it
values of CP and CN obtained from this method is possible to estimate the delivered power and
are close to 1, so PD, ns and VS can be used di- hence the ratio of the delivered power with and
rectly to predict the full-scale performance. without the PBCF. The measurement of this
ratio will be the gain (improvement rate) in
For the prediction of the full-scale perfor- propeller efficiency (∆ηpm) due to the PBCF
mance for twin-screw ships with shaft brackets in the model scale.

wS = wm In the POT set up one should bear in mind

that the presence of the boat housing and the
ts = tm rudder behind the propeller will affect the pro-
η RS = η Rm peller advance velocity. Therefore the neces-
sary correction should be made in the advance
coefficient in terms of the wake fraction caused
For twin-skeg ships wm should be corrected by the boat housing and rudder, and compari-
based on experience. The full-scale performan- son of the efficiencies should be made at the
ce prediction method and its procedure are the corrected advance coefficient for the same
same as those for single-screw ships. thrust.

By considering the scale effects mentioned

5.6 Propeller Boss Cap Fins (PBCF) earlier in Section 4.1, the propeller efficiency
gain in the model scale (∆ηpm) is related to the
Although no particular extrapolation improvement rate of the propulsion efficiency
method specific to PBCF has been reported in on the full-scale (∆ηs) as shown, for example,
the open literature, it appears that the PBCF by Ouchi & Tamashima (1989), in figure 5.
patent holding companies have a well estab- This guidance implies that the power saving or
lished procedure to quantify the efficiency gain efficiency gain expected at the full-scale will
by the use of PBCF. This procedure is based be 2 to 3 times greater than the model scale
upon the so-called “reverse POT” (Propeller predictions.
Open Tests) set-up and experience gained over
a considerable number of full-scale measure- In the above outlined procedure the pro-
ments. Furthermore, private communication by peller and PBCF are considered as a unit pro-
this committee with the West Japan Fluid En- pulsor and the effect of the PBCF is included in
gineering Laboratory Co. Ltd has confirmed the propeller open water characteristics. The
that the extrapolation for ship powering per- prediction of the efficiency gain due to PBCF
formance is based upon the ITTC 1978 proce- does not require self-propulsion tests, but relies
dure, Nishimoto (1998). These are briefly re- heavily on the accumulated knowledge and ex-
viewed in the following. perience gained with PBCF on the full-scale.

In the reverse POT configuration the pro-

peller is placed downstream of the open water
propeller boat in order to provide a free flow
field for the development of the hub vortex.
9 correction to the propeller characteristics due to
difference in blade drag at model and full scale
is developed for conventional propellers.
Moreover, it relies on data for one representa-
Gain by Ship Trial Test (%)

Ferry tive profile only: at 0.75R. Hence due to differ-

6 ences in geometry, velocity and load distribu-
tions this method of scaling is not believed to

be accurate enough for tip fin propellers. In-
Container Container stead the blade was subdivided into streamwise
Pcc strips and the sectional drag estimated using
Reefer the theoretically calculated velocities and a
simple, flat-plate estimate of the frictional re-
2 sistance which was dependent on mainly the
Container local Reynolds number. To secure a fair com-
1 parison this procedure was applied to both the
tip fin and the conventional comparator pro-
0 1 2 3
pellers. By this procedure the corrections turn-
Gain by Model Test (%) ed out to be bigger for the tip fin propeller, i.e.
∆=η=pm it is more sensitive to scale effects. Unfortu-
nately, no full scale tests have been done, so no
Figure 5. Correlation of Efficiency Gains by confirmation of this scaling procedure can be
PBCF between Model and Actual Ship, made.
Ouchi & Tamashima (1989) 5.8 Extrapolation From an Extended Analy-
sis of Self Propulsion Test Data
An alternative to the above procedure,
which makes use of self-propulsion tests, has Under this category are grouped some
been used by the West Japan Fluid Engineering methods that were developed with the aim of
Laboratory Co Ltd. and is also reported im- improving the powering prediction process by
plicitly by Ouchi (1988). In this procedure the avoiding and replacing some of the assump-
PBCF is considered to be an appendage and tions inherent in their use. The common theme
then the usual open water characteristics of the behind these methods is that the use of the
propeller are predicted from standard series towing resistance is considered meaningless or
data. The effect of the PBCF is included in the misleading when attempting to predict full
self-propulsion factors, which are obtained by scale power.
means of the thrust identity method. The per-
formance of the ship with PBCF is predicted About ten years ago, some researchers
based on the standard ITTC 1978 procedure. (Holtrop 1990), started to make full scale pow-
ering predictions using self propulsion and
open water tests as the sole source of experi-
5.7 ITTC 1978 Modified Method for Tip mental data. This was the first step in the di-
Fins rection of dispensing with towing resistance
data, but this method was still consistent with
The only extrapolation method reported in traditional propulsion analysis since the model
detail in the literature for end plate and similar resistance had to be reconstructed from model
propellers is that reported by Andersen (1996). self propulsion test results.
Since the tip fin propeller is only a slight modi-
fication of a conventional propeller the ITTC Briefly described here are two methods
1978 procedure has been followed with only a which fall into this category, but which have
minor modification. The standard ITTC 1978 not yet been developed to a stage suitable for

general application. These methods, however, The second method described in this sec-
have promise in the development of useful tion, Schmiechen’s (1991) “rational theory”, is
tools for the full scale powering prediction for a method for ship powering analysis based on
vessels equipped with unconventional propul- results from model self-propulsion load vary-
sors. ing tests alone. Stand alone resistance and
propulsor open water tests are avoided.
Iannone (1997a, 1997b), starting from the
ITTC 1978 methodology framework, intro- Two overload tests are done at the same
duced a new propulsive analysis through the steady speed, but different values of the over-
separation of the self propulsion flow into its load. Care must be taken to ensure that the
viscous and potential components. To this end speed is steady to avoid significant accelera-
only open water and British self propulsion tion/inertia forces in the longitudinal momen-
tests are required. In particular, to evaluate the tum equation. The following parameters are
viscous component of the self propulsion flow measured for the two tests: speed, shaft rota-
the innovative concept of self propulsion form tional speed, shaft thrust, torque and the towing
factor KSP is introduced. force. These are designated

In the first version of this methodology the V1 N1 T1 Q1 F1

self propulsion form factor was derived from V2 N2 T2 Q2 F2
model thrust measurements at low speed during
self propulsion tests. This procedure, however,
produced rather uncertain values of KSP, due to The thrust, T, and torque QP, are assumed
the small values of measured thrust and the un- to vary as a quadratic function of the rotational
avoidable presence of laminar flow or separa- speed N such that
T = TO N 2 + TH NV
Iannone’s investigations (1997b) revealed Q p = QP 0 N 2 + QPH NV
that a sort of thrust deduction phenomena oc-
curs during self propulsion tests. Furthermore it
which is equivalent to them being linear func-
was appreciated that the self propulsion form
tions of the ship advance coefficient JH. The
factor at service conditions depends not only
subscripted terms on the RHS of these equa-
on hull trim at speed, but also on speed itself
tions are found from the experimentally meas-
(Iannone 1998). As a consequence the predic-
ured values as follows
tion method was refined (Iannone 1998) by
determining the thrust deduction factor and the
self propulsion form factor by tests at the servi- T1 N 2V2 − T2 N 1V1
T0 = 2 2
ce speed. Moreover a new data reduction N 1 N 2V2 − N 2 N 1V1
method was proposed (Iannone 1998), leading 2
N 1 T2 − N 2 T1

to different relationships among propulsive TH = 2 2

characteristics measured during British self N 1 N 2V2 − N 2 N 1V1
propulsion tests. Q N V − Q2 N 1V1
QP 0 = 21 2 2 2
N 1 N 2V2 − N 2 N 1V1
After some satisfactory applications to 2 2
single and twin screw hulls, Iannone claims N 1 Q2 − N 2 Q1
QPH = 2 2
that the method, though still in a refinement N 1 N 2V2 − N 2 N 1V1
phase, is promising for its suitability to be used
as an extrapolation tool for full scale powering From the steady state form of the longitudinal
predictions for unconventional propulsors. momentum equation
0 = TE + F − R = T (1 − t ) + F − R Jp


and by assuming that the thrust deduction frac-
tion can be modelled as a linear function of the The coefficients of this parabola are found in
ship advance coefficient the following sequence from the coefficients
already calculated:
t = tH J H
the value of tH is found as

T2 + F2 − T1 − F1 K PPH = 2πK QPH

tH =
æ T2V2 T1 − V1 ö æ2ö
1/ 2
çç − K PLPO = K PPO − ç ÷ K TO
3/ 2

è DN 2 DN1 èπ
1/ 2
where D is the propeller diameter. The resis- æ2ö 3 1/ 2 K
K PLP1 = K PPH −ç K TO K TH − T 0
tance is found from èπ 2 2

æ t V ö To obtain KPLP2 use is made of the zero thrust

R = T1 çç1 − H 1 + F1 condition when the equation
è DN 1
K T = K T 0 + K TH J HT = 0
and the advance coefficients of the two steady
states are found as
yields in sequence
V1 V
JH1 = ; JH2 = 2 KTO
DN 1 DN 2 J HT =
The following coefficients are then found KT H
based on Schmiechen’s definitions

KT 0 = ; K TH = H 3
ρD 4
Q Q and a cubic equation for the propeller advance
K QP 0 = P 05 ; K QPH = PH4 coefficient in the zero thrust condition
ρD ρD
K PPH 2 PPLP1 4 ( K PLT − K PLPO )
J PT = − +
Schmiechen then assumes that the power K TH π J PT π J PT

coefficient representing the power losses of the

propeller, the difference between the shaft
power supplied to the propeller and the jet As JPT occurs on both sides of this equation, a
power of the propeller in coefficient form solution here is to assume a value for JPT on the
RHS and calculate a new value for JPT on the
K PL = K PP − K PJ LHS. This procedure can be iterated until the
RHS and LHS values fall within a pre-set limit.
can be represented by a quadratic equation in From this the coefficient KPLP2 can be found
the advance coefficient of the propeller, JP. from

K PLP 2 = 2
( K PLT − K PLPO − K PLP1 J PT ) various parameters in Schmiechen’s “ thrust
deduction theorem”
Using the above basis, the propulsive perfor-
mance of the vessel can be found over a range τ = (1 + CT )1/ 2 − 1
of ship advance coefficients, JH, which can be
assumed. In sequence the thrust and power co- thrust deduction fraction t = tH JH
efficients are found as
é τ t2 ù
K T = K TO + K TH J H ê(1 + τ )t −
(1 − t )

An iteration is then necessary to obtain the energy wake

working value of the propeller advance coeffi-
cient at this ship advance coefficient and this is we χ (1 − w)
done using the loss parabola. = 1−
w w
K PLP 2 J P and the various propulsive efficiencies:
K PL = K PLPO + K PLP1 J P +
K PJ = K PP − K PL propeller efficiency
K PJ 2 K T
JP = − KT J P
KT π K P J ηTP =
In this procedure also, a value of Jp is assumed
jet efficiency
and a new value is calculated; the process is
iterated until the desired accuracy is achieved.
ηTJ = 1
From here the following propulsive coeffi- 1 + (1 + CT ) 2

cients are obtained

K pp pump efficiency
K QP =

K QL = η JP =
2π ηTJ

wake fraction effective thrust coefficient

JP K T (1 − t ) æ TE ö
w = 1− CE = çç =
è ρD V
2 2 2

thrust loading coefficient hull efficiency

8 KT (1 − t )
CP = η ET =
π JP2 (1 − w)
azimuthing unit and the propeller is considered
total propulsive efficiency as a whole and the propulsive coefficients are
related to the interaction between the whole
η EP = η ET ηTP propulsion unit and the hull. In this case the
system’s net thrust must include the drag forces
configuration efficiency on the housing. The resistance test with the hull
is performed without the propulsion unit fitted.
η EP æ PE ö
η EJ = çç = In the second approach, the azimuthing unit
η JP è PJ is considered as an open propeller in isolation
and the propulsive coefficients are related to
energy wake the interaction between the propeller and the
hull appended with the housing. Therefore the
we = w – χ (1-w) propeller thrust and torque will not include the
effect of the drag forces on the housing and the
resistance test must be performed with the
housing unit fitted, but without the propeller in
and normalized pressure level in the wake place.
C p = (1 − w) 2 ( χ 2 + 2 χ ) . In both of the above approaches the prob-
lem arises as to how to scale the drag of the
housing unit due to the low Reynolds number
Although Schmiechen has not done a whole at model scale which results in relatively large
set of evaluations for a series of ships, he has drag forces. In the first approach, a scale effect
presented comparisons between model and full correction can be applied to the model KT val-
scale parameters for detailed tests done on the ues obtained from an open water test with the
Meteor. For this vessel, the following parame- complete unit where the system’s net thrust is
ters were found to be similar for the model and measured. In the second approach, this correc-
full scale: KT, KQL, ηJP, CE, ηEP and ηEJ. tion can be applied to the appended hull resis-
Small variations between model and full scale tance based on the resistance results. The final
occurred in: KQ, w, ηTJ, ηTP , CP, ηET, CE, ηEJ power prediction is expected to be the same for
and t. both cases. However, the total propulsive effi-
ciency will be different due to the difference in
the resistance for the naked and appended hull.
5.9 Z - Drives (Podded propulsors) This is actually a matter of definition of pro-
pulsive efficiency. In some cases a stock pro-
Allied with the recent upsurge in the num- pulsor unit may be used instead of a geometri-
ber of applications of podded drives, there is cally similar model of the actual propulsor.
growing concern with the differences in testing
procedures and extrapolation methods being In a third approach, open water tests of the
used for these propulsors, particularly for trac- podded drive as a whole unit are also done.
tor type units. The extrapolation methods in use However, the thrust and torque measurements
have not been published in the open literature. are taken at the shaft excluding the drag effects
Also there is a lack of full-scale data for these of the housing and are corrected to the full-
systems due their short application history. The scale. The resistance tests with the model hull
following outline approaches to the testing and are done without the podded drive while the
extrapolation for podded propulsors have been resistance of the drive housing and the scale
suggested. effects are estimated numerically.
In the first approach the housing of the In a fourth approach, in addition to resis-

tance and self-propulsion tests, two open water

tests: one with the propeller alone; the other As a result of its work, the Committee feels
with the propeller plus the housing together that where there is weak interaction between
(whole unit), are used for the evaluation of the the unconventional propulsor and the ship hull,
podded drive. In this approach, the podded then methods developed on the core of the
drive is assumed to be an appendage and its ITTC 1978 approach can give levels of accura-
drag is converted to the full-scale without any cy for extrapolation of full scale ship powering
correction for scale effects. The thrust deduc- performance that are of the same magnitude as
tion fraction is calculated from the system’s those obtained for conventional propulsors.
(net) thrust and the total resistance with the Weak interaction occurs with devices such as
pods. The wake fraction is obtained based on a tip fin and tip plated propellers, propeller boss
KQ identity and no correction is made to the full cap fins, and other devices that are only a small
scale. The open water test results for the whole modification to a conventional propeller. Ef-
podded drive are corrected for the difference in fective modifications do need to be made to
Reynold’s number between the open water test account for differences in blade frictional drag
and the self-propulsion test. Further corrections resulting from differences in Reynolds number
are also made to the open water test results for between model and full scale and a method to
the full scale by using ∆KT and ∆KQ values do this for tip fin propellers is referred to in
based on open water test results with the pro- section 5.7. However, where there is strong in-
peller alone. teraction between the device and the hull, then
methods based on the ITTC 1978 approach are
not adequate. Strong interaction occurs with all
6 DISCUSSION OF EXTRAPOLATION types of ducted and partially ducted propellers,
METHODOLOGIES pre- and post-swirl devices, z-drives, etc. The
reason for this is that these latter unconven-
In section 4, this report has given an over- tional propulsors have large, complex and
view of methods of extrapolation used in the strongly modifying effects on the flow around
past for different types of unconventional pro- the hull. In addition, the exact physical mecha-
pulsor. Problems arising during testing and ex- nisms by which some of these devices interact
trapolation have been highlighted in some with the hull is not always clear. As a result,
cases. In section 5, these methods of extrapola- the testing and analysis leading to extrapolation
tion, together with other candidate extrapola- does not lend itself to being broken down into
tion methodologies, have been discussed in the pieces designated by the ITTC 1978 proce-
further detail. dure (i.e. separate resistance, open water pro-
pulsor, and self-propulsion tests and their
To date, many extrapolations for uncon- analysis). Most of these interactions between
ventional propulsors have used methods heav- propulsor and hull are strongly Reynolds num-
ily based on the ITTC 1978 methodology. ber dependant.
However, the modifications to the ITTC 1978
method proposed and used for many of these Benchmark work is necessary to investigate
unconventional propulsors are different: i.e. these flow phenomena and to identify trends in
there are as many or more modifications to the powering performance as the Reynolds number
ITTC 1978 method as there are types of un- is increased. Examples of some of the types of
conventional propulsor. This situation is less test that are necessary are described in sections
than ideal as there is always a question as to 4.6 and 4.7 for models fitted with ducts and
whether any power saving predicted is a fun- partial ducts. To do this work it is necessary to
damental characteristic of the device or a func- use the largest facilities available and to avoid
tion of the testing, analysis and extrapolation strict adherence to Froude scale speeds. New
method in use. test techniques need to be developed in some
cases. A heavy reliance needs to be made on the flow velocities would be made at these lo-
self-propulsion tests with geometrically scaled cations and these would be used to identify the
models of the unconventional propulsor and thrust of the device. (Note that this is analo-
hull arrangements. Use can and should also be gous to the method of analysis proposed for
made of RANS-type CFD calculations to in- waterjets by the Waterjets Group of the 21st
vestigate trends in certain detailed flow be- ITTC.) The thrust deduction effect could be
haviour as the Reynolds number of a particular found from an integration of local pressures
set up is increased. over the afterbody. CFD methods would be
used to supplement velocity values from actual
To scale or extrapolate powering perform- test results and to investigate trends at higher
ance, it is necessary to know the trends in per- Reynolds numbers. In contrast, the reason for
formance as Reynolds number is increased and looking at an extended analysis of self-
this can be found from these specially designed propulsion test data is with the plan that pow-
and perhaps expensive tests. Full scale trials ering performance information can be extracted
results from ships fitted with unconventional from these tests in a more macro manner (than
propulsors are needed for the same cases that by using the integrations referred to above), but
have been studied in depth at model scale. On- by avoiding as far as possible the less realistic
ce this base of knowledge is established, more aspects of the assumptions present in conven-
routine work might be possible following a tional methods of extrapolation. The Commit-
more simple test and extrapolation procedure. tee recommends that work is continued on the
development of new general methods for ex-
In carrying out propulsion tests for uncon- trapolation of the powering performance of
ventional propulsors, the Committee recom- ships fitted with unconventional propulsors
mends the use of extensive load varying tests along these lines.
done at high values of Reynolds number. In
working towards a guideline for extrapolation, The published performance of unconven-
the Committee recommends that the data from tional propulsor systems is sometimes clouded
these tests be used in combination with an for a number of reasons. Often the advantage of
analysis using one or more of: a momentum fitting an unconventional propulsor design is
analysis, as described in outline in section 5.1; not actual energy saving, but the suppression of
the results from RANS-type CFD calculations; cavitation or the reduction of unsteady pressure
and/or an extended analysis of self-propulsion pulses and vibration. There may be no efficien-
tests, some perhaps rather simplistic examples cy gain in some situations. In addition, uncon-
of which are described in section 5.8. The aim ventional propulsors can be fitted to ships
is to avoid assumptions made in the conven- where the initial design of the conventional
tional ITTC 1978 type analysis which are propeller is not optimum. This latter leads to an
known to be less than realistic. An example of apparently large propulsive efficiency gain
one of these is the use of a propeller open water with the device, whereas a proportion of that
test (often done at one Reynolds number) with gain is the result of more focussed design effort
results from the propeller in the behind condi- being placed on the propulsion design in gen-
tion (often done at a different Reynolds num- eral.
ber) to identify a wake fraction. A second ex-
ample is the empirical methods used for the
scaling of that wake fraction from model to full 7 DRAFT CONCLUSIONS AND REC-

A true momentum analysis would identify General technical conclusions

streamtubes passing from ahead of to behind
the propulsor unit. Detailed measurements of

• The exact physical mechanism by which propulsors when applied to ships fitted with
some unconventional propulsors interact tip fin and tip plated propellers, propeller
with the hull is not clear. boss cap fins, surface piercing propellers
and other devices which are modifications
• Extrapolation of full scale powering for of a conventional propeller.
ships fitted with unconventional propulsors
may be developed from extensive load Recommendations to the Conference
varying tests and in addition one or more of
momentum analysis, CFD computations, or • For powering prediction it is recommended
an extended analysis of self propulsion test that ship models fitted with unconventional
data. propulsors, such as propellers with ducts,
partial ducts, pre- and post- swirl devices,
• Use of extrapolation methods for uncon- z-drives, etc., should be tested as a unit and
ventional propulsors based on modifica- not broken down into component tests of
tions to the ITTC 1978 method show simi- the hull, propulsor and rotor and stator
lar levels of variation of the powering pre- components.
diction found between methods as the level • Extrapolation methods of full scale power-
of the power saving expected with the de- ing for unconventional propulsors should
vice. In other words, the accuracy of these be done using self-propulsion load varying
extrapolation methods is in most cases of tests of the geometrically similar ship
the same order of magnitude as the level of model and geometrically similar propulsor.
power saving of the device under analysis.
• It is recommended that for self-propulsion
• There is a shortage of accurate data from tests with unconventional propulsors the ef-
full scale trials supporting extrapolation fect of the rudder on the propulsion system
predictions made for unconventional pro- be considered due to the influence of the
pulsors. downstream swirl on the rudder.

• Extrapolations cannot be reliably made of • To reduce/ eliminate the scaling of flow

self-propulsion test data if flow separation separation effects during self-propulsion
which occurs on the unconventional pro- tests these tests should be done at higher
pulsor and/or the ship hull is not scaled cor- levels of Reynolds number than can be
rectly. achieved by rigid adherence to Froude
• It is recognised that most methods of ex- number scaling.
trapolation currently in use are modifica-
tions to a greater or lesser degree of the Recommendations for future work
ITTC 1978 method. It is recommended that
the ITTC 1978 method is only used with • It is recommended that work is continued
caution as a guideline for extrapolation of on the development of new general
model test results to full scale ship power- methods for the extrapolation of model test
ing prediction for unconventional propul- results for unconventional propulsors. The
sors. extrapolation methods should be validated
against extensive model tests and full scale
• Extrapolation methods for unconventional trials.
propulsors based on modifications of the
ITTC 1978 method are expected to give • Extensive tests for unconventional propul-
powering predictions to the same level of sors over a wide range of Reynolds num-
accuracy as that achieved for conventional bers have shown that some usual assump-
tions of trends in performance are false. peller with high aspect ratio blades, Marine
There is a need for detailed tests over a Technology, Vol. 26, No. 3, 192-201.
wide range of Reynolds numbers to be do-
ne on ship models fitted with each different Bruce, E. P., Gearhart, W.S., Ross, J.R., and
type of unconventional propulsor. Where Treaster, A.L., 1974, The Design of
possible such test results should be com- Pumpjets for Hydrodynamic Propulsion,
pared with CFD computations and full Proceedings of Fluid Mechanics, Acoustics,
scale trials. and Design of Turbo-Machinery, NASA,
Washington, D.C.

Ferrando, M. & Scamardella, A. 1996 Surface
Piercing Propellers: Testing Methodologies,
Andersen, P., 1996, A Comparative Study of
Result Analysis and Comments on Open
Conventional and Tip-Fin Propeller Per-
Water Characteristics. In Proceedings Small
formance. In Proc. 21st Symp. on Naval
Craft Marine Engineering Resistance &
Hydrodynamics. Trondheim, Norway.
Propulsion Symposium, pp.5-1 – 5-27. Yp-
Anon., 1998, Whale tail hits the road, Marin
silanti: University of Michigan
News, No. 65.
Friesch, J. & Johannsen, C. 1994 Propulsion
Arabshahi, A., Beddhu, M., Briley, W., Chen, Optimization Tests at High Reynolds
J., Gaither, A., Gaither, K., Janus, J., Jiang, Numbers, SNAME Transactions, Vol. 102,
M., Marcum, D., McGinley, J., Pankajak- pp.1 – 21
shan, R., Remotigue, M., Sheng, C., Gabriel, R. and Atlar, M., 1998, Calculation of
Sreenivas, K., Taylor, L., Whitfield, D., the performance of a ship propeller with
1998, A Perspective on Naval Hydrody- cyclic blade pitch control, International
namics Flow Simulations, Twenty-Second Shipbuilding Progress, 45 (443), 201-223.
Symposium on Naval Hydrodynamics,
Washington, D.C., August. Gearhart, W. S., McBride, M. W., 1989,
“Performance Assessment of Propeller
Boss Cap Fin Type Device”, 22nd ATTC, St
Atlar, M., Hannah, J., Takinaci, A.C., and John’s., Newfoundland.
Korkut, E., 1998, “Effect of various boss
caps on the efficiency and cavitation per- Hadler J.B. & Hecker R. 1968 Performance of
formance of a propeller.”, Intl Symposium partially submerged propellers. In Pro-
Honoring Tarik Sabuncu On the Occasion ceedings, Seventh Symposium on Naval
of His 75th Birthday, Istanbul Technical Hydrodynamics (ed. R.D.Cooper & S.W..
University. Doroff), pp. 1449-1493. Arlington: Office
of Naval Research – Department of the
Atlar, M., Patience, G., 1998, “An Investiga- Navy.
tion into Effective Boss Cap Designs to
Eliminate Hub Vortex Cavitation”, Halstensen, O. and Leivdal, PA., 1990. The
PRADS’98, The Hague. development of the speedZ propulsion
system. The 7th high speed surface craft
Bennett, J., 1996, Of fiddles and fins, Wooden conf., London.
Boat, 132, September/October, 40-43.
Henderson, R.E., McMahon, J.F. and Wisli-
Bose, N. and Lai, P.S.K., 1989, The experi- cenus, G.F., 1967, A Method to Design
mental performance of a trochoidal pro-

Pumpjets, Underwater Missile Propulsion,

Compass Publications. Kerwin, J. E., Keenan, D.P., Black, S.D. and
Diggs, J.G., 1994, A Coupled Vis-
Hollstein, H.J., Gomez, G.P. & Gonzalez- cous/Potential Flow Design Method for
Adalid, J. 1997 Bulk carrier speed trial- Wake-Adopted Multi-Stage, Ducted Pro-
swith a Sistemar CLT propeller. The Naval pulsors using Generalized Geometry,
Architect, February 1997, pp. 33-6. Transactions of the Society of Naval Ar-
chitects and Marine Engineers, New York,
Holtrop J., 1990, Are Resistance Tests Indis- pp. 2-28.
pensable?, Proceedings 20th ITTC, Madrid,
Vol. 2, pp. 454-459. Kurimo, R. et al.,1998. Azipod propulsion for
passenger cruisers: Details of the hydrody-
Iannone L., 1997a, Power Performance Analy- namic development and experience on the
sis and Full-Scale Predictions without propeller design for “Fantasy”-class cruise
Towing Tests, Proceedings of Nav & liners. NAV & HMSV Intl. Conference,
HSMV 1997, Sorrento, Italy, 18 – 21 Sorrento.
Iannone L., 1997b, Self-Propulsion Flow Com- Kurimo, R., 1998. Sea trials experience of the
ponents Analysis for Power Performance first passenger cruiser with podded propul-
Predictions, Proceedings Eight Congress of sors. PRADS’98, The Hague.
the International Maritime Association of
Mediterranean (IMAM), Istanbul, Turkey. Lai, P.S.K. 1990 Oscillating foil propulsion,
Ph.D. thesis, University of Glasgow.
Iannone L., 1998, Fattore di Forma e Risucchio
nelle Condizioni di Servizio dall’Analisi Larsson, L., Regnstrom, B., Broberg, L., Li, D-
del Flusso di Autopropulsione, INSEAN A, and Janson, C-E, 1998, Failures, Fanta-
(Rome model Basin) Report No. 1996-61, sies, and Feats in the Theoretical/Numerical
Rome, Italy. Prediction of Ship Performance, Twenty-
Second Symposium on Naval Hydrody-
Isshiki, N., Hashimoto, K. and Morokawa, H. namics, Washington, D.C. August.
1987 Studies on fin ship propelled by os-
cillating tail fin, Proceedings International Manen, J.D. van, 1973, Non-conventional pro-
Marine Engineering Conference, Chinese pulsion devices, International Shipbuilding
Society of Naval Architecture and Marine Progress, Vol. 20, No. 226, 173-193.
Engineering, Shanghai.
Mewis, F., 1998. Podded drives im vormarsch
de Jong, K., Sparenberg, J.A. Falcão de Cam- hydrodynamische aspekte. Schiff & Hafen,
pos, J.A.C. & van Gent, W. 1992 Model 11.
Testing of an Optimally Designed Propeller
with Two-Sided Shifted End Plates on the
Blades. In Proc. 19th Symp. on Naval Hy- Minsaas K.J. 1998. Propulsion systems for
drodynamics, pp. 461-75. Washington DC: high speed marine craft. MARINTEK A/S
National Academy Press. Report, May.

Karafiath, G, 1997, U. S. Navy Ship-Model Nishimoto, H., 1998, Private Communications

Powering Correlation and Propeller RPM with West Japan Fluid Engineering Labo-
Predictions, Propellers/Shafting ’97, So- ratory Co., Ltd.
ciety of Naval Architects and Marine Engi-
neers. Ogura, M., Koizuka, H., Takeshita, H., Kohno
Y., Ouchi, K., Shiotsu, T., 1988, “A Screw Propeller Efficiency -”, Naval Architecture
Propeller Boss Cap with Fins”, UK Patent and Ocean Engineering, Vol. 28, Papers
Application , GB 2194295 A. Edited by Society of Naval Architects of
Japan. Published by Ship and Ocean Foun-
Olofsson, N. 1996 Force and Flow dation, Japan 1992.
Characteristics of a Partially Submerged
Propeller. Doctoral Thesis. Goteborg: Ouchi, K., Tamashima, M., Arai, K., 1992
Chalmers University of Technology “Propeller Noise Reduction caused by
Department of Naval Architecture and PBCF”, PRADS’92, Newcastle upon Tyne.
Ocean Engineering.
Potsdam Model Basin (SVA), Propeller Design,
Ouchi, K. 1988, “Research and Development 1995, “TVV and HVV Propellers - an In-
of PBCF (Propeller Boss Cap Fins) to En- novative Concept developed by SVA in
hance Propeller Effciency”, The Motor Cooperation with SCHOTTEL”, Postdam
Ship 10th. International Marine Propulsion Model Basin (SVA) Product Brief.
Conference, London
Qian, W.H. et al. 1992a The exploratory de-
Ouchi, K., Ogura, M., Kono, Y., Orito, H., velopment of simplified compensate nozzle
Shiotsu, T., Tamashima, M., Koizuka, H., and its composite device. Proceedings of
1988, “ A Research and Development of the 2nd International Symposium on Pro-
PBCF (Propeller Boss Cap Fins) - Im- peller and Cavitation, Hangzhou, China
provement of Flow from Propeller Boss -”,
Journal of Society of Naval Architects of Qian, W.H. et al. 1992b A new type of ship
Japan, Vol. 163. energy-saving device—fore-propeller hy-
drodynamic fin sector, Journal of Kansai
Ouchi, K., 1989, “Research and Development Society of Naval Architects of Japan, No.
of PBCF (Propeller Boss Cap Fins) – Im- 218.
provement of Flow from Propeller Boss –“,
The Proceedings of the International Sym- Rains et al. 1981. Hydrodynamics of podded
posium on Ship Resistance and Powering ship propulsion. Journal of Hydronautics,
Performance’89 (ISRP), Shanghai. Vol. 15, Nos 1-4.
Ouchi, K., Tamashima, M., 1989, “Research Riijarvi, T., Li, J., Veitch, B., Bose., N., 1994,
and Development of PBCF – New and Experimental performance and comparison
Practical Device to Enhance Propeller Effi- of performance prediction methods for a
ciency -”, PRADS’89. trochoidal propeller model, International
Shipbuilding Progress, Vol. 41, No. 426,
Ouchi, K., Tamashima, M., Kawasaki, T., Koi- 113-136.
zuka, H., 1989, “A Research and Develop-
ment of PBCF (Propeller Boss Cap Fins) – Savitsky D. 1964 Hydrodynamic Design of
2nd Report: Study on Propeller Slipstream Planing Hulls, Marine Technology, Vol. 1
and Actual Ship Performance-”, Journal of No. 1, Oct. 1964
Society of Naval Architects of Japan, Vol
165. Schmiechen, M., 1991, Proceedings of the 2nd
International Worshop on the Rational
Ouchi, K., Tamashima, M., Kawasaki, T., Koi- Theory of Ship Hull-Propeller Interaction
zuka, H., 1990, “ Research and Develop- and its Applications, VWS, The Berlin Ship
ment of PBCF (Propeller Boss Cap Fins) – Model Basin, HEFT 56, June 13-14.
Novel Energy-Saving Device to Enhance

Schneekluth, H. 1986 Wake equalizing duct, Van, S-H, Kim, M-C. and Lee, J-T, 1993
The Naval Architect, April, 1986 / A pam- “Some remarks on the powering perfor-
phlet from Schneekluth Hydrodynamik, mance prediction method for a ship
Entwicklungs- und Vertriebsges. MBH equipped with a pre-swirl stator-propeller
system”, Korea Research Institute of Ship
Schott, C.G., 1996, Design and Analysis of and Ocean Engineering, Daejeon, Korea,
Axial Flow Turbo-Machinery Blades in ITTC 1993, Vol. 2.
Steady Incompressible Flow by a Combi-
nation of Momentum and Singularity Wislicenus, G.F., 1968, Pumping Machinery
Methods, Master of Science Thesis, The for Marine Propulsion, ASME Symposium
Pennsylvania State University. on Pumping Machinery for Marine Propul-
SCHOTTEL, 1995, “TVV and HVV Propeller,
A Joint Development of SCHOTTEL and
SVA”, SCHOTTEL-Werft Product Brief. Wislicenus, G.F., 1960, Hydrodynamic and
Propulsion of Submerged Bodies, J. Am.
Schulze, R., 1995, “SVA- Nabenkappenflossen Rocket Soc., Vol. 30, No. 12, pp. 1140 –
fur Schiffspropeller”, SVA (Postdam Model 1148.
Basin) Report No: 2218.
Yamaguchi, H. and Bose, N. 1994 Oscillating
Shiba H. 1953 Air-Drawing of Marine Propel- foils for marine propulsion, Proceedings of
lers. Transportation Technical Research In- the Fourth International Offshore and Polar
stitute, Report No. 9. Tokyo: the Unyu- Engineering Conference, Osaka.
Gijutsu Kenkyujo.

Skidmore, J.E., Lueschen, J.D., Renzo, J.A. Zhao, H.H. et al. 1988 Experimental research
and Landrum, C. 1989 Design of a human on the scale effect of ship geosim, Pro-
powered wet submersible for competition, ceedings of Ship Performance Symposium
Oceans ‘89, Seattle. of China Shipbuilding Engineering Society.
Sparenberg, J.A. & de Vries, J. 1987. An Op-
timum Screw Propeller with Endplates. In- Zhou, Z. M. et al. 1990 Energy-saving shaft
ternational Shipbuilding Progress, vol. 34, bracket suitable for multi-screw vessels,
no. 395, pp. 124-33. China Shipbuilding, No. 2, (in Chinese).

Stern, F., Paterson, E. G., Tahara, Y., 1996. Zierke, W.C. (editor) and 18 authors, 1997, A
CFDSHIP-IOWA: CFD Method for Sur- Physics-Based Means of Computing the
face Ship Boundary Layers and Wakes and Flow Around a Maneuvering Underwater
Wave Fields, IIHR Report #381. Vehicle, Technical Report No. TR 97-002,
Applied Research Laboratory, The Penn-
Stierman, E.J. 1984 “Extrapolation methods for sylvania State University.
ships with a ducted propeller”, International
Shipbuilding Progress, Vol.31, No. 356.

Treaster, A. L., 1969, “Computerized Applica- .

tion of the Streamline Curvature Method to
the Indirect, Axisymmetric Turbomachine
Problem,” ARL TM 514.2491-16, The
Pennsylvania State University.





  on pressure fluctuation on the oscillating foil

surface has been published. Such data would
  be invaluable for calibration for CFD work.
       There is also a need for some research and
    development work on control mechanisms for
full scale oscillating propulsors, but this is
by Pengfei Liu, Institute for Marine Dynamics outside of the scope of the ITTC.

This report demonstrates an excellent job

by the Committee on such a difficult and Building on the success of the workshop
comprehensive subject in such a short time. It held by the Propulsion Committee to compare
is a pity that this Specialist Committee will not RANS and panel methods to predict propeller
continue. Continuation of the work by this performance I suggest a workshop on CFD
Committee is important, because the methods modeling of propellers operating in ducts.
of predicting, both numerically and Many issues could be addressed in such a
experimentally, the performance capabilities of workshop. Both experimental and CFD
oscillating propulsors, tip fins on screw predictions of unsteady pressure distribution on
propellers and others are still not yet fully the inner surface of a duct are difficult to find
developed. There is a lot of remaining work to in the literature, although some authors have
be done. presented the steady pressure distribution
inside the duct without experimental
A bigger database covering more aspects verification. Benchmark data including
of the performance of energy saving devices, measurements of pressure fluctuation on the
and alternatives or modifications to the duct’s surface would be very useful for
conventional propulsors would be valuable. calibration and verification of panel methods
Experiment data for calibration and validation and RANS codes. A workshop would focus
of computer codes for unconventional efforts in this field.
propulsors is very scarce. For example, it is
difficult to find experimental data on tip fin or    
end plate propellers. Such data need to include
not only the propeller shaft thrust and torque by M.X. van Rijsbergen (MARIN)
coefficients but the blade pressure distribution
as well. As stated in the report, the In the discussion on extrapolation of
development of oscillating foil propulsion oscillating propulsors, the report of the
systems is in its infancy. Little experiment data committee concentrates on propulsor-hull

interaction. MARIN would however like to ask The figure above shows that the torque
attention for scale effects on the oscillating coefficient decreases with 30 to 60% for
propulsor itself. Reynolds numbers in the range from 2·10 to
7·10, dependent on the blade angle motion and
MARIN has been thoroughly involved in the condition. The torque coefficient of the
the development of the first prototype Whale wheel is dependent on both the lift and drag
Tail Wheel (see e.g. van Manen et al.,1996). coefficients of the blades.
This is a cycloidal propulsor which can operate
in a trochoidal blade motion. From open water
tests at scale factor 2 and several rotation rates, The observed decrease in torque coefficient
it appeared that mainly the torque coefficient is of the Whale Tail Wheel occurs in the same
considerably sensitive to the Reynolds number region of Reynolds numbers where a
(see the figure below). significant decrease in drag coefficient of
streamlined foils has been reported (see e.g.
The Reynolds number is defined as: Hoerner, 1965).

VR c
Re =
ν Trochoidal propellers can operate at the
condition where at some point in the blade path,
where V is the average resultant velocity the entrance velocity to the blade equals zero or
relative to the blade, c is the chord length and ν practically zero, causing extremely low
is the kinematic viscosity of water. The torque Reynolds numbers. Special precautions during
coefficient is defined as: testing and for extrapolation of results may
have to be taken. Little is known on this issue.
KQ =
ρn D4L 2
Due to the sometimes large angles of attack of
the flow on the blade, in combination with low
where Q is the torque delivered by the two Reynolds numbers, leading edge separation
main shafts (corrected for friction), ρ is the may occur, which is also likely to be prone to
density of water, n is the rotation rate, D is the scale effects.
diameter of the wheel and L is the span of the
blades. The torque coefficient is normalised References:
relative to the value at the lowest Reynolds
number for each condition and blade angle van Manen, J.D. and van Terwisga, T., “A new
motion (designated ‘bam A’ or ‘bam B’). way of simulating Whale Tail Propulsion”, 21
Symposium on Naval Hydrodynamics,
Reynolds effect on torque coefficient
Whale Tail Wheel Trondheim, June 1996
bollard pull 'bam A'

100% design condition 'bam A'

Hoerner, S.F., “Fluid-dynamic drag”, Hoerner
design condition 'bam B'
fluid dynamics, 1965


40% by Hassan Ghassemi, Institute for Marine


Of the many different unconventional
2.0E+05 3.0E+05 4.0E+05 5.0E+05 6.0E+05 7.0E+05 propulsors considered by the Committee,
podded propulsors are of increasing practical

interest at the present time. Due to their better

maneuverability, regular inflow to the propeller, (a) The fourth general technical conclusion
reduction of cavitation and vibration, is: “There is a shortage of accurate data
compactness, different types of podded from full scale trails supporting
propulsion systems (pulling, pushing and extrapolation predictions made for
shuttle) are becoming popular in current unconventional propulsors”. My
construction using electrical propulsion in question is: Was any such data
place of conventional propulsion systems. available to the Committee; did the
committee validate extrapolation
Some recent computational research on a procedures with full-scale data?
pulling type of podded propulsion system at the
Institute for Marine Dynamics (IMD)  raised (b) The fourth recommendation to the
issues concerning the best approach to take in conference is: “To reduce the scaling
making reliable predictions. For example, the of flow separation effects, self-
drag of the strut/pod may be affected by the propulsion tests should be done at
downstream propeller wake and its modeling. higher levels of Reynolds number than
Thus it is important to find a reasonable vortex can be achieved by adherence to Froude
wake model to calculate the drag of strut/pod in number scaling.” My comment is: In
the propeller slipstream using a panel method. this context turbulence stimulators
By carrying out experiments, Halstansen and should account for this. I would like to
Leivdal showed that strut/pod drag is have recommendations for the use of
compensated by recovery of rotational energy turbulence stimulators included in the
in the slipstream. What method does the committee report.
Committee recommend for the strut/pod drag
of the pulling type of podded propulsion in the         
vortex wake generated by the propeller?
by Friedrick Mewis, HSVA

The hydrodynamics of different types of First I would like to give my thanks to the
podded propulsion systems (pushing, pulling committee for it’s really good work. In chapter
and shuttle) may change the overall 5.9 (page 240) are described very well the
hydrodynamic performance of the vessel. Can possible methods for using open water test
the Committee comment on these differences results for podded drives.
and how they may influence the approach to
making performance predictions? Some factors are mentioned which affect
the measured values of propeller thrust,
• Ghassemi, H. and Allievi A. 1999: Fluid propeller torque and unit thrust (system thrust)
Analysis and Hydrodynamic Performance and additionally four different approaches to
of Conventional and Podded Propulsion extrapolation methods are described.
Systems, Oceanic Engineering International,
Vol. 3 No. 2. Unfortunately, the estimation of the
propeller thrust in the case of pulling pods is
   very difficult and the correct measurement is
impossible. The reason for that fact is the gap
by Jaakko V. Pylkkanen (VTT) between the propeller hub and the pod housing.
The pressure in this gap influences to a
First of all I would like to express my substantial level the measured propeller thrust.
appreciation to the Committee for its This substantial inner force in the gap is a
competent report. result of the high pressure behind the working

propeller, the pressure change due to the propulsor are becoming more common and
stagnation point of the pod shaft coupled with experiments are being conducted in water
the large gap area. The width of the gap, which tunnels to measure the flowfield.
can vary in different model tests, and in
different towing tanks, influences the measured  !  $"%"  &   
thrust too.
Having been involved to some extent
The error can be up to 10% dependent on [Riijarvi et al. 1994] with testing and
the propeller loading, the pod-geometry, the computational work on trochoidal propellers,
width of the gap and other reasons. The we agree that Reynolds number scale effects
measured unit thrust is not affected by this are very important in the extrapolation of open
problem because the force in the gap is an inner water and self propulsion data on these
force of this unit. My proposal is to use the propellers. As the discussor explains, large
measured unit-thrust as the basis for the angles of attack do occur on the blades of a
estimation of the so called “small figures” like trochoidal propeller, both during each
thrust deduction fraction, effective wake revolution and at low advance ratios of
fraction, propulsor efficiency… in pod operation. Work has shown, in fact, a sort of
propulsion only. critical advance ratio below which stall
effectively reduces both the thrust and torque
coefficients of the propeller.

     A multiple stream tube theory [Bose,
1987] using experimental lift/drag coefficients
 !  " # at different Reynolds numbers for the foil
sections in a quasi steady manner can be used
First of all, we would like to thank Dr. Liu to predict the performance of a trochoidal
for his very supportive comments on the propeller in a way that models this behaviour
Committee’s efforts and his discussion. The and this critical advance coefficient, at least in
most difficult problem for the Committee was a qualitative way. A more accurate method
to identify reliable model/full-scale data for any would require experimental characteristics for
unconventional propulsor. In almost every case the foil sections during dynamic stall from
these databases are incomplete for the oscillating foil tests for the appropriate
calibration and validation of computer codes; conditions and over a range of Reynolds
however, it is very important to note that only numbers.
model-scale data exists for most
unconventional propulsors. We could not References:
identify any experimental database that
includes details of the local steady/unsteady Riijarvi, T., Li, J., Veitch, B.J., Bose, N., 1994,
pressure and velocity fields associated with the Experimental performance and comparison of
unconventional propulsor. In most cases the performance prediction methods for a
unconventional propulsor was designed to trochoidal propeller model, International
improve powering thus only nominal powering Shipbuilding Progress, Vol. 41, No. 426, pp.
measurements were made. At this time the 113-136.
Committee could not identify a specific
reference that presents unsteady duct pressure Bose, N., 1987, Rotary foil propellers, Papers
measurement data. of the Ship Research Institute, Tokyo, Japan,
Vol. 24, No. 5, pp. 45-67.
We agree that a workshop on CFD
modeling of propellers operating in ducts is
needed. These types of unconventional

 !  '" (
  $ As far as the performance prediction of
different types of pods is concerned: whether it
The committee would like to thank Dr. is a pusher or a tractor type, both
Ghassemi for his questions. configurations are unconventional, the latter
being less conventional and hence more
Although the methods for predicting the complex to handle.
drag of the pod housing is not the main concern
of this committee, Dr. Ghassemi is asking for However, if the performance prediction is
the committee’s opinion on this issue. He is based on model test procedures, it would not
particularly interested in the pulling (or tractor) make much difference to look for fundamental
type in which case the pod housing will be in differences in the approaches depending on the
the wake of complex propeller flow. type of the propulsor. This is because the
current procedures, which are not in the public
As reported in Section 4.8 of this domain although they are outlined in Section
committee’s report, studies due to Rains et al. 5.9 based on the best of our knowledge, utilize
(1981) and Minsaas (1988) include some standard tests of the resistance open water and
information on the drag characteristics of self-propulsion. In any case, these tests will be
pusher and tractor type propulsors. In the performed for both types in a similar fashion,
method for the tractor type, Minsaas neglects although the scale effect correction for the pod
the effect of cavitation and propeller induced drag (due to low Rn) will be more complex for
drag, and provides a semi-empirical formulae the tractor type.
for the viscous drag of the housing components
depending upon whether the component is in or However, our recommendation to the
outside the propeller’s slipstream. ITTC as outlined in Section 7, is that the
On the other hand, amongst the reported extrapolation of podded propulsors should be
computational methods, one can mention done based on a self-propulsion test which
studies e.g. Dumez (1997) (reported in Le includes the hull, pod and propeller as unit and
magazine du Bassin d’Essais des Carenes) and not by breaking the test up into components
Korpus et al (1998) (reported in PRADS `98). such as open water, resistance etc.
The former of these studies is based on the
panel technique while the later utilized a References:
RANS technique. There is also a recent useful
work reported by Vartdal et al (1999) Dumez, F-X, 1997. PODS, some encouraging
(published in the CFD `99 workshop, Norway) conclusions, Le magazine du Bassin d’Essais
comparing these two procedures on a tractor des Carenes, No. 7, January.
type pod drive.
Korpus, R. et al., 1998. Hydrodynamic design
In the light of this information, the of integrated propulsor/stern concepts by
committee feels that panel methods are Reynolds – Averaged Navier – Stokes
adequate to model the potential flow effects techniques. PRADS ’98, The Hague
taking into account the influence of the
propeller's vortex wake. However, the major Vartdal, L., et al, 1999. On the use of CFD
contribution due to the viscous effects still Methods in Developing the Azipull Podded
remains as the problem which will require Propulsion System, CFD Conference ’99,
complex boundary layer and even separated Ulsteinrik, Norway
flow models. For these, the committee
recommends use of RANS techniques, as Minsaas, K.J., 1988. Propulsion Systems for
successfully applied by Korpus et al (1998), High Speed Marine Craft, MARINTEK A/S
including the effect of the propeller. Report, May issue.

 !  )" !** 

The following briefly summaries the

verbal reply given at the conference.

(a) Where full scale data was available,

estimates of energy efficiency gains
were made using this data and the
results presented in the report reflect
this. An example of this is included in
the efficiencies quoted in tables 2.4.

(b) The effectiveness of turbulence

stimulators in controlling separation is
extremely uncertain and cannot take the
place of tests done at increased levels of
the Reynolds number.

 !  +" $ ,

The following briefly summarises the

verbal reply given at the conference.

We thank the discusser for his comments

and we agree that the gap between propeller
and pod in tractor units has a great influence on
the measured thrust values of the propeller.
Again we recommend self propulsion tests to
be used on their own as the basis of
extrapolation of full scale power. The
intermediate values in this process, such as
thrust deduction fraction, wake, etc., may not
have the same meaning, or values, as in the
powering prediction for conventional
propulsion systems.