Inteer J Nav Architt Oc Engng (20010) 2:1~13

http://dx.doi.org/10.2478/IJNAOE-2013-0014

Comparative stu
udy on thee performaance of Pood type waaterjet
byy experimeent and coomputation n
Moon-Chan
M Kim
m, Warn-Gyu P
Park, Ho-Hwann Chun and Unn-Hwa Jung

Deppartment of Naval Architecturre and Ocean E

Engineering, P
Pusan Nationall University, Puusan, KOREA

AB BSTRACT: A comparative study between n a computationn and an experriment has beeen conducted tto predict the pperformance
off a Pod type wa aterjet for an amphibious
a heeled vehicle. The Pod type waterjet has been
wh b chosen onn the basis of tthe required
sppecific speed off more than 25 500. As the Pod d type waterjett is an extremee type of axial flow
f type waterjet, theoreticaal as well as
exxperimental wo orks about Pod d type waterjetss are very rare.. The main purrpose of the preesent study is tto validate andd compare to
the experimenta al results of thee Pod type wa aterjet with thee developed CF FD in-house codec based on the RANS equuations. The
deeveloped code has been vallidated by com mparing with the experimenntal results off the well-know wn turbine prroblem. The
vaalidation also extended
e aterjet where thhe pressures allong the duct surface and alsso velocities at nozzle area
to thee flush type wa
haave been comp pared with expeerimental resullts. The Pod tyype waterjet haas been design ned and the perrformance of thet designed
waaterjet system including ducct, impeller an nd stator wass analyzed by the previouslyy mentioned iin-house CFD Code. The
prressure distribu
utions and limiiting streamlinees on the bladee surfaces weree computed to confirm
c the peerformance of the
t designed
waaterjets. In adddition, the torq
que and momen ntum were com mputed to find tthe entire efficciency and thesse were comparred with the
model test resultts. Measuremeents were takenn of the flow raate at the nozzlle exit, static prressure at the vvarious section ns along the
duuct and also thet nozzle, revvolution of thee impeller, torrque, thrust annd towing forc rces at variouss advance speeeds for the
prrediction of peerformance as well as for coomparison withh the computattions. Based on these measuurements, the performance p
waas analyzed according
a to thhe ITTC96 staandard analysiis method. Thee full-scale eff ffective and thhe delivered po ower of the
whheeled vehicle were estimateed for the preddiction of the seervice speed. T This paper emp phasizes the coonfirmation off the ITTC96
annalysis methodd and the develo oped analysis code
c for the deesign and analyysis of the Podd type waterjet ssystem.

EY WORDS: Pod
KE P type waterrjet; Amphibious wheel vehiccle; 1996 ITTC D computation;; RANS equatiions.
C Method; CFD

NTRODUCTIION
IN coupledd lifting surfface and RAN NS (Reynoldss Averaged
Navier--Stokes) equattion for waterjet design and analysis. A
A waterjet propulsion systems has many
m advantagges comparrison between the computatioons and experiimental data
coompared to thhe convention nal screw proopeller used for has beeen conducted by Taylor annd Kimball (1999) and a
ammphibious m
military vehhicles becauuse of goood reasonaable correlation was obtaineed. Through the t research
m
maneuverability y at low speeeds, good opeerating ability at carried out for the waterjet desiign method, it i has been
shhallow waters,, and a high thrust at low w speeds to aid a concludded that the R RANS code iss sufficient to predict the
m
maneuverability y and exit from m the water, ettc. Consequenttly, performmance of a waaterjet system with reasonab ble accuracy
thhe demand foor waterjet propulsion sysstems is rapiddly for waterjet sysstem design.
and hennce it can be uused as a tool fo
inncreasing in appplication to varrious marine veehicles. Chun ete al. 2002, Chuun et al. 2003 and Kim et al. 2007 have
Although thhere has been a great deal off research carriied studiedd a flush type wwaterjet systemm for tracked vehicles. In
ouut on waterjet propulsion systems, studies on the Pod tyype ded to a Pod
the present paper, thee study of wateerjets is extend
waterjet especiaally for an am mphibious wheeeled vehicle aare type waaterjet system for a wheeledd vehicle by using a CFD
veery rare. Allison (1993) and d Allison & Jian
J (1998) haave code baased on the RA ANS equationss developed by y Park et al.
reeviewed waterrjet design methodsm which include the t (1996, 2005) as welll as by applyying experimeental works
appplication of tthe lifting surrface theory fo or pump desiggn. using tthe ITTC96 m momentum fluxx method. Th he code has
Taaylor and Kerw win (1998) haave proposed a procedure of a been vvalidated by ccomparing itss computation ns with the
d correlation
existingg model test reesults, showingg a fairly good
betweenn them. The vvalidated codee has been app plied to the
presentt design and aanalysis which were comparred with the
Correspondingg author: Moon
C n-Chan Kim,
experimmental results.
e
e-mail: kmcproop@pusan.ac.kkr
Copyright 2010 Society of Naval Architects of Korea. Production and hosting by ELSEVIER B.V. This is an open access article under the CC BY - NC 3.0 license
(http:/ /creativecommons.org/licenses/by- nc/3.0/).
2 Inter J Nav Archit Oc
O Engng (20110) 2:1~13

Waterjet sysstems are classsified into an axial

a flow typee, a 1982) was implemented. Althouggh some comp mputationally
mixed flow type and a radial flow type acccording to thheir
m inexpennsive models have shown reasonable performance p
loading characteeristics. Axial flow type waterjets are ussed (Ockfenn and Matveevv, 2009), it iss believed thatt employing
whhen a high flow w rate is requiired that is equuivalent to a hiigh more sophisticated turbulence model m produuces highly
sppecific speed. Axial flow waterjets are normally sm mall accuratte solutions. Numerical
N critterion of convvergence is
coompared to othher two types of o waterjets. In the present caase, used ass less than 10-33 for all terms of u, v, w, veelocities and
thhe specific speeed is extremely y high as muchh as 2500, whiich pressurre. The multi-bblock grid sysstem is used because
b it is
is adapted to thhe Pod type waterjet.
w A Pood type waterjjet, very difficult to generrate whole sysstem by one bloock. Sliding
thherefore, normaally has high momentum
m effiiciency comparred block scheme
s is used in the interfacce of impeller block
b where
to a flush type waterjet, whille the pump efficiency
e is loow. the bilinnear interpolattion method is applied (Park et al. 1993a,
Thhis is somewhaat similar to thee conventionall ducted propelller 1993b).
altthough the speeed of revolution is normallyy higher than thhat Thee developed coode was firstlyy validated byy comparing
off the ducted proopeller. the expperimental dataa of well-know wn 3-dimensioonal turbine
Although waterjet
w units have
h been widdely used for the t which has been executed by Drinng (1982). Thhe Reynolds
prropulsion systeem of marine vehicles, a unnique method for numberr is 5.9 105 annd the on-com ming velocity iss 23m/s. The
thhe model test anda also for th he performance analysis of the t rotationnal velocity which is 78% of o oncoming velocity
v was
waterjet propulssion system has h not been fixed f yet due to only coomputed in thee present validaation. The statoor and rotor
hiighly complexx hydrodynam mic interactionns in the duuct- has twoo blocks of H-type grid (85 29 29) whicch is shown
immpeller-stator-hhull system. The T traditionaal concept off a in Fig. 1.
thhrust deductionn method (IT TTC 1987) haas a number of
prroblems for thee analysis of waterjet
w systemms; occasionallyy a
neegative value iss found due to the trim difference between the t
seelf-propulsion and resistaance tests. Many differeent
appproaches for the t model testt and performaance analysis are a
stiill being used by different organizations
o w
worldwide. Sinnce
thhe waterjet com mmittee of 21 1st ITTC (19966) suggested the t
m
momentum flux approach (the ITTC96 standdard method), thhis
m
method seems to
t be widely adopted
a worldw wide for waterrjet
peerformance anaalysis. The direect thrust meassurement methhod
is an alternativee to the momeentum flux meethod. The dirrect
m
measurement off the forces annd moments im mparted upon the t
huull by the waaterjet system may provide insight into the t
innteraction, but it is very diffficult to direcctly measure the t
thhrust practicallyy. The so callled ITTC96 method m has beeen
(a) Side
S view of thee turbine.
appplied to the annalysis of the present experim mental results.
In the preseent study, the application off a computationnal
annd an experimeental method to o a Pod type waterjet
w is mainnly
ussed for the puurpose of comp parisons and validation.
v In the
t
neear future, an im mproved desig gn is also expeccted to be carriied
ouut based on thee present researrch of the Pod type waterjet for
thhe amphibious wheel
w vehicle.

W
WATERJET D
DESIGN AND
D ANALYSIS
S

Numerical meth
hod and validation

The three dimensional incompressible Navier-Stokkes

eqquations are soolved by so caalled Iterativee Time Marchiing
M
Method (Park et al. 1993a, 1993b) for thee analysis of the t
prresent Pod tyype waterjet system. In thhis method, the t
coontinuity equattion is solved d by Maker annd Cell methoods
(b) Coomputational grrid system.
(VViecelli, 1969)) and the mom mentum equatiion is solved by
timme marching scheme.
s The sppecial derivativves of convectiive
Fig. 1 Geometry
G and the
t grid system
m of rotor-statoor turbine.
fluux terms are differentiated
d with
w QUICK scheme (Leonaard,
19979). The spatiial derivatives of viscous term
ms and continuuity
Thee sliding blocck scheme is applied to thhe interface
eqquation are diffferenced by central differenncing. To captuure
betweenn the rotor andd stator parts beecause of rotatiing of rotor.
thhe turbulent floows, low Reynolds number k- k model (Chien,
Innter J Nav Archhit Oc Engng (2010)
(2 2:1~13 3

(a) at 1/4
4H (b) at 2/4H (mid-span) H
(c) at 3/4H

Fiig. 2 Time averraged pressure coefficient (H Rn=5.9 105 baased on impeller chord at 0.7R).
H: rotor span) (R

The bilinearr interpolation method is used in the slidiing This prressure coefficiient is differenntly defined as in equation
bllock scheme. In Fig. 2, the time avveraged pressuure (1) from
m other compuutations in ordeer to compare the Drings
cooefficients of computed valu ues with turbuulent as well as results. Other pressuree coefficients are
a defined as follows;
f
with laminar flowf and expeerimental onees are comparred 1
acccording to thhe span posittion of 1/4(arround the huub), C P = ( p p ) / u 2 (2)
1//2(mid span) and
a 3/4(around d the tip). Presssure coefficiennts 2
arre defined as foollows; Thee good corresppondence is fouund between computation
c
and exxperiment exccept trailing edge
e and thee difference
CP = ( p pT .E. ) / ( pmax pT .E. )
'
(
(1) betweenn laminar andd turbulent moodel is not soo large. The
pressurre at trailing edge region shows some separations
whhere pT.E . is thee pressure at trrailing edge annd pmax is the peeak which invokes
i some difference betw
ween the compputation and
prressure in bladee surface. the expperiment.

(a) 1/6 pitch (b)) 2/6 pitch (c) 3/6 pitch

(dd) 4/6 pitch (ee) 5/6 pitch ( 6/6 pitch

(f)

Fiig. 3 Pressure contour

c distribu
ution around rootor/stator at mid-span
m (Rn=5.9 105 based on impeller chhord at 0.7R).
4 Inter J Nav Archit Oc
O Engng (20110) 2:1~13

Fig. 3 showws the equivaleent pressure diiagrams at eveery
1//6 intervals of rotor pitch where
w continuouus pressure linnes
arre found at thee interface of rotor and stattor region. Theese
reesults show thaat the present treatment forr the interface of
bllocks is reasonaable.
Thee limited streammlines in the symmetry
s planne of intake
duct arre shown in Fig. 6 where some separaations occur
around lip part. The separation
s at liip region comm monly more
occurs than at lamp region due too a severe currvature. The
pressurre distribution along the duuct surface (raamp) of the
computted results concur very weell with the experimental
results, with the excepption of positioon 7, as shown in Fig. 7.

Fiig. 4 Flush typee waterjet arran

ngement.

Another vallidation was conducted byy comparing the t

coomputations with
w the experim mental data foor the case of the
t
fluush type waterj
rjet (Park et al.. 2005) as show
wn in Fig. 4. The
T
whhole domain of o the waterjet system is treatted in the preseent
coomputation as shown
s in Fig. 4,
4 which compprises an intakee, a Fig. 6 Limiting
L stream
mlines around intake
i region.
duuct, a rotor andd a nozzle. The Reynolds nuumber is 4.3105,
baased on the tottal velocity annd chord lengthh at r=0.7R. The
T
nuumber of grid points of intak ke duct is 6033060. The grrids
off each block of o the rotor and
a stator havee 632535 and a
5552135, resppectively. The number of grrid points in tip
cleearance is 69888. The wholee grid system iss shown in Fig.. 5.

Fiig. 5 Grid systeem for flush typ

pe waterjet sysstem.
Fig. 7 Comparison
C forr pressure alonng the duct surfface (ramp).
Innter J Nav Archhit Oc Engng (2010)
(2 2:1~13 5

The compliccated flows might

m be occuurred in front of Finally the momentuum and torquee are also com mpared and
immpeller, which was very diffficult to predict accurately with
w given in
i Table 1, alsso showing a good verificattion for the
thhe pressure meaasurement of averaging
a onlyy four surroundded CFD code
c used. Thhese validationn results show
w why the
pooints. The streeamlines in noozzle and wakke region is also presentt CFD tool haas been develloped and appplied to the
shhown in Fig. 8. The flow beh hind the statorr is straighteninng, presentt Pod type wateerjet system.
whhich shows thee good functioon of the preseent stator. The jet
floow velocity at the nozzle exiit is comparedd in Fig. 9, wheere
ann excellent agreeement can be seen. Table 1 Comparison of momentum
m, torque and thhrust for the
flush tyype waterjet.
M
Momentum [N]

Inlet Noozzle

Coomputation 10.1 9
90.2

Exxperiment 9.8 8
88.9

Torrque Thrust
per unit 10 K Q per KT
Nm]
[N N]
unit[N

Compuutation 1.24 1.200 81.0 0.588

Experriment 1.20 1.177 N/A N/A

Fiig. 8 Streamlines in nozzle an
nd wake regionn.

(a) Vertical (bb) Horizontal

Fiig. 9 Comparison of axial vellocity at nozzlee exit in bollardd condition witth 4000 rpm.

Design and anaalysis - Initiially assumed pump

p efficienccy, P = 0.95

Preliminary design has been

b conductedd by panel coode - Ratiio of nozzle annd inlet area, Aj = 0.95
baased on the som
me basic assum
mptions as folloows: - Relaative rotative efficiency,
e r = 0.98
- Wake factor, (1-w) = 0.97 - Nozzzle efficiency, n = 0.99 at deesign speed off

- Inlet recoveryy factor, i = 0.95 5.388knots

6 Inter J Nav Archit Oc Engng (2010) 2:1~13

1 1

Vj =
2
{
Vw + Vw 2 + 4T / ( A j ) } 2
(3)
1

NS = N Q 2 / H 4
3

1 3

where Vw is the inlet velocity and T is the thrust of impeller = 2200 ( 0.448 60 ) 2 / 6.68 4
and Aj is the nozzle area. (8)
= 2748 rpm, m3 / min, m
1 1

Vj =
2
{
2.69 + 2.69 2 + ( 4 418 9.81) / (1000 0.038 ) 2 }

= 11.8m / s Table 2 Preliminarily predicted parameters for the design of
Pod type waterjet.
Q j = Aj V j = 0.038 11.8 = 0.448m3 / s (4)
Vs EHP RT DHP
OPC t
[knots] [kW] [kN] [kW]

Actual jet efficiency 4.33 9.08 2.52 0.22 0.9 45.86

4.89 14.06 3.24 0.22 0.9 71.01

5.38 19.42 4.10 0.22 0.9 98.08

5.94 26.32 5.24 0.22 0.9 132.93

t : transmission efficiency, RT : total resistance
where = Vw / V j = 0.228 , = 1 n = 0.01 , = 0
from Fig. 10 in Allison (1993) and h j = 0 (same height at For amphibious military vehicles there are a number of
special and unique waterjet requirements, some of which
inlet and nozzle).
need to be taken into account for the present design. These
are:
- High thrust at low speeds to aid maneuverability and exit
1 2 0.228 (1 0.228) from waters
j = = 0.395
0.93 1 + 0.01 0.2282 - Resistance to cavitation when power is applied at low
speeds, typically 1.67m/s to 2.78m/s for swimming vehicles
Therefore Overall Performance Coefficient is roughly which corresponds to 6km/h ~ 10km/h.
predicted as follows; - Since the vehicle runs over muddy ground, the tip of the
impeller is wide and thick enough to resist wearing
The duct-impeller-stator is then designed by taking the above
OPC = t r p j (1 t )
items into account.
= 0.97 0.98 0.60 0.395 0.98 (6) The impeller diameter is designed as 330mm for the
power absorption of NCR 73.5kW (100PS) per a unit.
= 0.22 According to the requirement of the resistance to cavitation,
the area ratio of the impeller blade is designed to be large. In
This initially predicted OPC is applied to other speed as particular, the pitch at the tip is decreased to avoid cavitation
shown in Table 2. The effective horse power (EHP) is also since the flow speed on the tip region of the blade is the
predicted from similar vehicle data. If the pump efficiency fastest. According to the same concept, the chord length at
can reach to around 65%, the achieved speed would be higher the tip region is designed to be larger than those at the other
than the design speed of 5.38knots. radii.
The correct design angle of the stator leading edge is
The required water head can be also calculated based on essential to achieve good performance of the pump since this
the above assumptions. plays an important role in straightening rotating flows. The
number of the stator blade also plays a key role in
straightening rotating flows, resulting in good pump
H = V j 2 / 2 g n j Vw 2 / 2 g + h j efficiency. The number of the stator blade was chosen as 11,
which is a somewhat larger number than the flush type due to
= 11.82 / ( 2 9.8 ) 0.99 0.95 2.69 2 / ( 2 9.87 ) a short stator length which is due to the short duct length and
= 6.68m (7) also usual in a Pod type waterjet. The thickness of the tip and
the leading/trailing edges of the impeller and stator were
designed to be thick enough to resist wearing. The principal
The specific speed is also calculated as follows; particulars of the designed pump part are shown in Table 3.
Innter J Nav Archhit Oc Engng (2010)
(2 2:1~13 7

Taable 3 Princippal particulars of the waterj

rjet model (scale As shown in Fig.. 10, the desiggn is iterativelyy conducted
raatio, =3.5). by the analysis code.. The panel coode has been usedu for the
Impelleer diameter [mm] 94.3 prelimiinary design beecause it is nott time-consumiing and also
convennient in initial determination
d o design param
of meters such
Number of blades, Imp
peller 5 as pitchh, camber and so
s on, compareed to the RAN NS code. The
Statorr diameter [mm
m] 94.3 mesh generation
g for panel
p code is shown in Fig. 11 and the
computted pressure distribution is shhown in Fig. 122.
Number of blades, Staator 11
Nozzle diameter [mm
m] 30.3
P/D off Impeller at 0.7
7R 0.764
P/D off Impeller (Meaan) 0.76

Although waterjet
w units have
h been widdely used for the t
prropulsion systeem of the maarine vehicles,, a more uniqque
deesign method thhan that of thee conventional propeller has notn
yeet been developped due to a highly
h complexx hydrodynam mics
innteraction in thhe duct-impelller-stator systeem. Traditionaally
thhe mean streammline method iss used for the design
d of a pum
mp.
Thhis method, however,
h is lim
mited in term ms of taking innto
acccount the effeects of viscosiity and compliicated turbulennce
phhenomena. ManyM differen
nt methods for f design anda
peerformance annalysis are sttill being used by differeent
orrganizations woorldwide. As mentioned
m in thhe introductionn, a
CF FD code based on the RAN NS equation caan be used forr a
m
more accurate design
d and an nalysis, which can be used to Fig. 11 Panel represenntation for Podd waterjet systeem.
annalyze the com mplicated visccous flow. Thhe developed in-
hoouse CFD codee was mainly applied to the present design.

(a) Pressure side.

s

(b) Suction side.

s

Fig. 122 Pressure disttribution on thhe rotor surface computed

Fiig. 10 Design procedure.
p by paneel code.
8 Inter J Nav Archit Oc
O Engng (20110) 2:1~13

This potentiaal code is main

nly used for thee initial designn of sides att the tip regionn is larger thann the other raddii although
im
mpeller pitch which
w is to be absorbed the given power and a the pitcch at tip regionn was designedd to be low, which
w means
gaave the required thrust. the pitcch at the tip reegion is designned to be someewhat larger
than opptimal pitch. With
W the excepption of this phenomenon
p
the oveerall flow behaavior seems to be favorable, as a shown in
Fig. 166. In case of stator, complicated flow phenomena are
indicateed as shown in Fig. 17, which are due d to the
straighttening of the strong
s rotatingg velocity invooked by the
rotor.

Fig. 14 Schematic vieew of the desiggned Pod type waterjet.

w

Fig. 15 Multi-block grid

g system of the
t Pod type waterjet.
w

Fiig. 13 Non--dimensional camber (C) and pitch (P) (

diistribution of thhe designed im
mpeller along thhe radius.
(a) Suction side
The stator isi also accordiingly designedd to adapt to the t
floow passing thhrough the im mpeller. The pitch and cambber
diistribution of thhe finally desiggned impeller are
a shown in Fig.F
133. The overalll configuratio on of the dessigned Pod tyype
waterjet is also shown
s in Fig. 14. The analyssis was iterativeely
coonducted for thhe design of th he present Pod type waterjet by
ussing the validaated code. The generated gridd for the analyysis
off the Pod type waterjet is sh hown in Fig. 7,
7 where the tootal (b) Pressure side
s
m
multi-blocks annd grid num mber are 19 and 550, 3550,
reespectively. Figg. 16~17 show ws the computeed streamline and a Fig. 16 Limiting streaamline and surrface pressure contours on
prressure distribuution on impeeller and statorr blade surfacces. the rotoor blade (Rn=44.3105 basedd on the impelller chord at
Thhe pressure diifference betw ween the suctiion and pressuure 0.7R).
Innter J Nav Archhit Oc Engng (2010)
(2 2:1~13 9

As shown in Tabble 4, the com mputed nozzle momentum

(M7) iss quite similarr to the experrimental resultts while the
inlet moomentum (M1) and especiallly the torque arre relatively
differennt from the expperimental resuults. The differrence of the
inlet momentum
m mayy be due to thee absence of thhe modeling
of the vehicle configguration arounnd the inlet noozzle in the
presentt computation. Regarding thee difference off the torque,
the smaall gap (0.3% ofo diameter) might
m create a difficulty
d not
only inn the experimennt but also in the
t computatioon. In terms
of the experimental results, the tipp clearance waas carefully
maintaiined to be coonstant along the radii based on the
(a) Sucttion side previouus study accordding to Kim et e al. 2007. In the case of
the com mputation, furrther investigaation for the impeller
i tip
region would be expeected because the gap ratio of the flush
type waaterjet for the previous
p validaation case was 0.7% of the
diameteer, the results of
o which weree shown in Tabble 2, where
excellennt agreement ofo torque was found. If the performance
p
is preddicted on the basis
b of the prresent computaation of the
torque, the predicted speed would lower than thaat predicted
by the experiments
e

Table 4 Comparisonn of momenttum and torquue between

computtation and moddel test. Modeel speed : 1.48m/s, RPM :
(b) Pressure side 3400.
R
Relative
Expeeriment Com mputation
nd surface presssure contours on
Fiig. 17 Limitingg streamline an diifference
thhe stator bladee (Rn=6.1105 based on thee stator chord at
M11[N] 5
59 64 8.60%
0.7R).
M77[N] 1
127 126 0.82%
The flow behhind the statorr is not rotatedd as shown in the
t
strream line in Fig.
F 18, which h means that thhe stator playss a Torquue[Nm] 1.64 1
1.91 16.8%
suuccessful role in straightening
g the flow to thhe nozzle. Finaally,
thhe momentum anda torque werre also computed to confirm the t
peerformance of the design sp peed and power and these are a
coompared with the
t model test results.
r MODE
EL TEST
Model and test set-u
up

Thee 1/3.5 scale model

m was consstructed by FRPP, as shown
in Fig.19. The princiipal dimensionns of the wheeeled vehicle
model and
a the designned waterjet syystem are showwn in Tables
1 and 5.
5 The stator and impeller were
w made off aluminum,
which are
a shown in Fig. 20.

Fiig. 18 Profiles of pressure disstribution and streamline

s
(Rn=4.3105 baased on the imp peller chord at 0.7R).
Fig. 19 Model profilee of the amphibbious wheeled vehicle.
v
100 Inter J Nav Archit Oc Engng (20010) 2:1~13

Taable 5 Princippal dimensionss of the modeel vehicle (scale Thee model speedd of the impeeller (or shaft)) revolution
raatio, =3.5). should be determinedd to adjust to a given
g towing force
f (FD) at
Length between
b perpen
ndiculars [m] 1.891 each sppeed. The maxximum model revolution peer minute is
initiallyy predicted to be about 40000 in consideraation of the
Length at designed waater line [m] 1.693 kinemaatic relationshipp with the full--scale speed off revolution.
B
Breadth mouldeed [m] 0.771 For thee transmissionn of power froom a motor too a waterjet
system,, the high tennsile belt systeem was used since, even
Draught[m] 0.341 though a spiral bevel gear is used, a bevel gear syystem is not
reliablee in the case of
o a very high speed revolutiion of more
than 30000rpm. To reeach such a higgh speed of reevolution in
model tests, two 4kW W servomotors were used wiith a torque
meter. The
T torque waas measured byy a 2Nm dynam mometer and
the speeed of revoluution was alsoo measured by b a digital
tachommeter of Ono Sookki Co. It waas found that thhe measured
speed ofo revolution has a good corrrelation with thhe input rpm
within less than 0.005% (in casee of 3000rpm input, the
measurred value was 3001rpm). Thhe pressure seensor of the
Validynne Company P305D
P model inncluding an am mplifier was
used for f the meassurement of pressure. Thhe pressure
differennce between thhe areas aheadd of and behinnd the pump
was useed for the estimmation of the pump
p efficienccy and flow
Fiig. 20 Manuffactured rotorr and stator model made of rate. The
T photo off the set-up profile for thhe pressure
aluuminum. measurrement is show wn in Fig. 23.

The duct waas also made of

o aluminum annd the stator was
w
buuilt within thhis housing, as
a one body. The assemblled
waterjet system is shown in Fig
g. 21~22.

Fig. 23 Set-up profilee for pressure measurement.

m

Self-prropulsion test

Thee self-propulsioon tests were conducted in the t range of

Fiig. 21 Assemblled waterjet system with belt..
Fiig. 22 Experim
mental set-up prrofile with belt driven.
the speeed revolution of an impelleer that can cover the full
scale power at the deesign speed (55.38knots). Thee tests were
conduccted at model speeds of 0 (bollard condiition), 1.19,
1.34 annd 1.48m/s, which
w are equuivalent to thhe full-scale
speeds of 0, 4.33, 4.87 and 5.38knots k , respectivvely. Fig.24
shows a photo of the model runninng in a self-proopulsion test
at threee speeds of 4.33,
4 4.87 andd 5.38knots. It I was very
difficullt to reach a sppeed of revoluution higher thaan 4000rpm
becausee of a belt prooblem. Conseqquently, the maaximum test
speed was
w 5.38knots while the finall speed at the NCR N power
(73.5kWW per one unitt) might be hig gher than the design
d speed
of 5.38knots
k .
In the
t bollard conndition, the grross thrust estiimated from
the mom mentum changge between the inlet and outleet should be
the sam
me or similar (within
( the expperimental erroor range) as
the nett thrust (say, negative resistance) measuured by the
dynamoometer. As shoown in Table 6, two results are similar,
showinng the reliabilitty of the expeeriments conduucted in the
presentt study. It is estimated thaat the overall uncertainty
includinng bias and random
r errorss is around 2%
of the
measurred value at dessign speed.
Innter J Nav Archhit Oc Engng (2010)
(2 2:1~13 11

Table 6 Results of bollard test with model waterjets.

M
Momentum Net thrust
N Error
R
RPM
[N] [N] [%]

2
2200 104.1 96.1 -8.26
2
2600 142.7 134.5 -6.09
3
3000 182.0 180.0 -1.15
3
3400 231.1 230.6 -0.20

Thee self-propulsion model testt should be caarried out in

Fiig. 24 Profile of the model in self-propulssion test at 1.19,
1.34 and 1.48m/s/s.
the samme dynamic condition as in full-scale.
f In orrder to keep
the samme condition, thhe towing forcce (FD) was deetermined to
compennsate the diffference betweeen full- and model-scale
m
frictionnal resistances.. Although thee shape of thee vehicle is
differennt from that off the conventioonal ship whichh is more of
a box than a ship, the ITTC78 extrapolation method is
appliedd in the preseent analysis. To T consider thhe effect of
roughness difference between the fuull- and model--scale, CA is
determiined from the result of the existing
e full-sccale sea-trial
of a tracked vehicle (C Chun et al. 20003).
Thee detail descripption of the ITT
TC96 method is i omitted in
this papper (see the refference; Chun et al. 2003). TheT analysis
data intterpolated at thhe given towingg force is givenn in Table 7
by the ITTC standardd method. The pump efficienncy is lower
than thhe initially asssumed value (0.6) while the overall
efficienncy is similar to
t the predicteed value of 0.222 as shown
in Table.

Taable 7 Self-proopulsion test an

nalysis results (ITTC1996
( meethod) by adjussting towing foorce (FD).
per one
S
Ship Mom
mentum Jet system Effective Flow rate
uunit model Rev. Area
sppeed Fluux[N] Power Pump Power Per unit
speed
IVR
VM VS n A3 E 7- E 1 PPE = E5 - E3 QJ
M1 M7 M 7 M1
[m/s] [
[kts] [rpm]] [m2] [W] [W] [m3/sec]

1.19 4
4.33 2771 0.0041 15.0 79..4 64.4 166.52 166.61 0.0156 3.18

1.34 4
4.87 3141 0.0041 15.5 101.5 86.0 245.09 245.19 0.0177 3.20

1.48 5
5.38 3542 0.0041 11.8 1233.5 111.7 326.95 327.15 0.0196 3.21

Full scale
Model Ship
S Pump Eff. Effecctive Shaft
Rev. QJ / A3
speed sppeed Efficiency O
Overall powwer power Momentum
m Effective Shaft
flux Power power

VM VS n (E5 - E3) PE 2 Qns M 7 M1 PE PD

[m/s] oa
[m/s] [[kts] [rpm]] / PD [W
W] [W] [N] [kW] [kW]

1.19 4
4.33 2771 3.785 0.508 0.233 766.6 328.2 2759 6.6 26.2

1.34 4
4.87 3141 4.287 0.517 0.243 1155.2 474.5 3684 10.0 41.1

1.48 5
5.38 3542 4.749 0.515 0.260 1655.3 635.8 4789 14.7 56.5
122 Inter J Nav Archit Oc Engng (20010) 2:1~13

Thaat is probably due

d to the decrrease of resistannce because
of imprrovement of boow shape whicch makes jet veelocity slow
and inccrease of jet efficiency.
e Thee initially obtaained design
speed ofo 5.38knots is given at the delivered
d poweer of around
58.8kW W (per a unit)), which is soomewhat loweer than the
initiallyy required NCRR power of 73..5kW (100PS) as shown in
Fig. 255. The main reeasons for the decrease of thhe designed
power might
m be due to
t the improveement of the boow shape as
shown in Figs. 26~277 and to the facct that the pressent waterjet
system might be connsidered to be properly desiggned. Based
on the present analyssis the achieveed speed woulld be higher
than thhe design sppeed of 5.38kknots (10km/hh) at NCR
conditioon as seen in Fig.
F 25.

CONC
CLUSIONS

Commparative studdies not only by b computatioons but also

by expperiments havee been perforrmed to proviide a detail
Fiig. 25 Estimateed delivered po
ower vs. speed. understtanding of com mplicated flow phenomena and a a proper
design methodology for f a Pod type waterjet system m. Based on
the pressent study, the following connclusions can be drawn:

y The developed
d codde has been vaalidated by com mparing the
experrimental data of o the well-knnown 3D turbiine problem
wheree excellent agreement of pressure distrribution on
turbinne blade was found
f except at
a the end of trrailing edge.
That is probably duue to some diifficulty at meeasuring the
pressuure in the trrailing edge regionr as welll as some
difficulty in the commputation due to the separatioon. Another
validaation continuedd to the flush axial type watterjet which
has been
b used for
fo amphibiouus tracked veehicle. The
pressuures along the duct intake annd velocities att nozzle exit
Fiig. 26 Improveement of resisstance perform
mance by usingg a were compared witth the experim mental results where
w good
cuurved type vanee. corresspondences weere found. Throough these valiidations, the
develloped code with multi-blocck grid system m and with
slidinng mesh technique is thoughht to be validaated for the
analyysis of the preseent waterjet sysstem.
y The validated
v in-hoouse CFD codde can effectively be used
for thhe design as weell as the analyysis of a Pod tyype waterjet
system m. The presentt designed resuults show that the
t designed
statorr plays importaant role to straaighten the floow behind a
pumpp part althoughh some severe separations arre found on
the sttator surface which
w is inevittably invoked by the high
speedd of revolution of impeller.
y The performance
p o the designedd waterjet waas compared
of
usingg the model test combined with w the ITTC C96 analysis
methood and by thhe CFD compputations. Thee difference
betweeen the compuutations and thhe experimentss was rather
largerr than those of o flush type waterjet casee. The inlet
momeentum differennce is larger thhan that of outtlet (nozzle)
momeentum. The diffference in inleet momentum betweenb the
compputation and thhe experiment would decreasse if a more
accurrate on-comingg flow is sim mulated arounnd the inlet
regionn in computations. More inveestigations are expected to
be coonducted in order
o to determine the reasson for the
largesst difference of
o torque betw ween the compputation and
Fiig. 27 Naked model
m resistance versus speedd. the exxperiment.
Inter J Nav Archit Oc Engng (2010) 2:1~13
y The design speed of 5.38knots can be obtained at the power
of 57.4kW per a unit based on the experimental results
which is lower than the initial NCR design power (73.5kW
per one unit). If the computational result is used for the
prediction of speed performance, which means only
replacement of torque, the required power would be larger
as 66.6kW which is, however, still lower than the NCR
power of 73.5kW.
ACKNOWLEDGEMENT
This research was supported by the ROTEM Co. The
support is gratefully acknowledged.
REFERENCES
Allison, J.L., 1993. Marine waterjet propulsion. Trans.
SNAME., 101, pp. 275-335.
Allison, J.L. and Jiang, C.B., 1998. Modern tools for waterjet
pump design and recent advances in the field.
International Conference on Waterjet Propulsion ,
RINA, Amsterdam, pp. 1-19.
Chien, K.Y., 1982. Prediction of channel and boundary-layer
flows with a low-Reynolds number turbulence model.
AIAA Journal, 20, pp. 33-38.
Chun, H.H. Park, W.K. and Jun, J.G., 2002. Experimental and
CFD analysis for rotor-stator interaction of waterjet pump.
Proc. of 24th Symposium on Naval Hydrodynamics, Japan,
8-13 July 2002.
Chun, H.H. Kim, M.C. Ahn, B.H. and Cha, S.M., 2003. Self-
propulsion test and analysis of an amphibious tracked
vehicle with waterjet. Proceedings of World Maritime
Technology Conference and SNAME Annual Meeting,
Paper No. D6 (D-133), USA.
Dring, R.P. Joslyn, H.D. Hardin, L.W. and Wagner, J.H., 1982.
Turbine rotor-stator interaction. Journal of Engineering
for Power, 104, pp. 729-742.
13
Kim, M.C. and Chun, H.H., 2007. Experimental investigation
into the performance of the axial-flow-type waterjet
according to the variation of impeller tip clearance.
Journal of Ocean Engineering, 34, pp. 275-283.
Leonard, B.P., 1976. A stable and accurate convective
modeling procedure based on quadratic upstream
interpolation. Computer Method in Applied Mechanics
and Engineering, 19, pp. 729-742.
Ockfen, A.E. and Matveev, K.I., 2009. Aerodynamic
characteristics of NACA 4412 airfoil section with flap.
International Journal of Naval Architecture and Ocean
Engineering, 1(1), pp. 1-12.
Park, W.G. and Sankar, L.N., 1993a. An iterative time marching
procedure for unsteady viscous flows. ASME-FED, 20.
Park, W.G., Sankar, L.N., 1993b. Numerical simulation of
viscous flow around a propeller. AIAA Paper No. 93-3056.
Park, W.G. Kim, H.W. Jung, Y.R. and Park, E.D., 1996.
Unsteady incompressible turbulent flow simulation of the
rotor-stator configuration. Proceedings of the International
Symposium on Transport Phenomena and Dynamics of
Rotating Machinery, 2, pp. 257-267.
Park, W.G. Jang, J.H. Chun, H.H. and Kim, M.C., 2006.
Numerical flow and performance analysis of waterjet
propulsion system. Journal of Ocean Engineering, 32, pp.
1740-1760.
Taylor, T.E. and Kerwin, J.E., 1998. Waterjet pump design and
analysis using a coupled lifting surface and RANS
procedure. Proceedings of International Conference on
Waterjet Propulsion, RINA, No. 5, Amsterdam, pp. 1-9.
Taylor, T.E. and Kimball, R.W., 1999. Experimental validation
of a coupled lifting-surface/RANS procedure for waterjet
pump and design analysis. Proceedings of FAST 99,
Seattle, pp. 893-900.
Van Manen, J.D., 1962. Effect of radial load distribution on the
performance of shrouded propellers. Spring Meeting
SNAME.
Viecelli, J.A., 1969. A method for including arbitrary external
boundaries in the MAC incompressible fluid computing
technique. Journal of Computational Physics, 4(4), pp.
543-551.