numerical and experimental evaluation of waterjet
TRANSCRIPT
11th International Conference on Fast Sea Transportation FAST 2011, Honolulu, Hawaii, USA, September 2011
NUMERICAL AND EXPERIMENTAL EVALUATION OF WATERJET
PROPELLED DELFT CATAMARANS
Manivannan Kandasamy1, Svetlozar Georgiev
2, Evgeni Milanov
2, Frederick Stern
1
1 IIHR-Hydroscience & Engineering, the University of Iowa
2 BSHC- Bulgarian Ship Hydrodynamics Centre
ABSTRACT
The accurate prediction of waterjet propulsion using CFD is
of interest in the standpoint of performance analyses of
existing waterjet designs as well as design optimization of
new waterjet propulsion systems for high-speed marine
vehicles. Currently, the design and analysis of waterjets
follow the ITTC '05 recommended procedures and
guidelines which was validated by a rigorous experimental
campaign through standardized testing. The current study
focuses on validation of detailed duct flow simulations on
catamarans using the Delft catamaran as the model. The
validation work is conducted as a pre-requisite for
subsequent URANS based optimization. The Delft
catamaran model was build at BSHC and a customized
waterjet was designed for the model based on pre-existing
stock waterjet designs. Data from the model testing using
the ITTC '05 procedures include net jet thrust, thrust-
deduction, water-jet volume flow-rate, sinkage, trim, and jet
velocity surveys at nozzle exit. Simulations were performed
over a speed range of 0.4<Fr<0.75 using URANS and an
actuator disk body-force model. The computed net jet thrust,
thrust deduction, sinkage and trim compare well with
experiments indicating that the present approach is an
efficient tool to predict the performance of waterjet
propelled JHSS and paves way for future optimization work.
KEY WORDS
Water-jet, Self-propulsion, URANS, Validation, Catamaran
1.0 INTRODUCTION
Nowadays, there is a growing interest in waterjet propulsion
because it has a lot of benefits over conventional screw
propeller such as shallow draft design, smooth engine load,
less vibration, lower water borne noise, no appendage drag,
better efficiency at high speeds and good maneuverability.
These advantages have increased the demand of waterjet
propulsion systems for a variety of marine vehicles
including high-speed naval sealift. Since waterjet propulsion
systems are relatively new, the powering performance
analysis of waterjet appended hulls using tow tank model
testing has been a recent, ongoing area of research.
The ITTC Waterjet Performance Prediction Specialist
Committee (Van Terwisga, 2005) developed a model testing
procedure for waterjet propulsion. Rigorous experimental
testing was conducted by nine separate ITTC member
organizations using a 5.5 m scale model of the U.S. navy's
research vessel Athena at Fr=0.6. The scatter in resistance,
i.e., the difference between minimum and maximum
measured resistance was 9% of the mean. Dynamic trim
had an overall scatter of 10.6%, which was reduced to 2.9%
if the outliers were removed. Heave had an overall scatter of
117%, reduced to 55% with the elimination of outliers.
Looking at the quantities the dynamic trim was measurable
to within 0.3 degrees and heave to within 16 mm. The
scatter in inlet wake fraction was 7.6%. Of all the methods
used to determine flow rate directly, the most accurate and
repeatable was the use of a high density laser doppler survey
at the inlet opening or internal to the waterjet system. The
scatter in flow rate for equal impeller speed appeared was
0.8%. However, there was a 3.5% maximum deviation from
the mean in the estimation of model waterjet speed for the
self-propulsion point. Consequently, the scatter in the
estimated model thrust at self propulsion at Fr=0.6 was
18%. The committee concluded that the mechanics of the
experimental procedures applied for the submitted data sets
were generally sound, but the complicated nature of the
measurements and data reduction resulted in the substantial
facility bias.
Recent innovations in CFD and high performance
computing have enabled faster and cost effective approach
for predicting waterjet propulsive characteristics. This has
enabled detailed analysis of the flow through the waterjet
ducts, which would require prohibitively expensive Laser
Doppler Velocimetry (LDV) measurements if the whole
flow field has to be measured. Such detailed flow analysis is
required for a deeper understanding of the flow physics
giving insights into further improvement of the performance
characteristics of the waterjet. However, the CFD has to be
thoroughly validated before relying on it for performance
analysis, design, and optimization.
Bulten (2006) performed a detailed investigation both
experimentally and numerically on a waterjet test setup
where the waterjet inlet was mounted on top of a cavitation
tunnel. The mass flow rate in the tunnel was adjusted to get
the desired inlet velocity ratio (IVR) values. This was
modeled in the CFD using a prescribed velocity profile at
the inlet of the cavitation tunnel and a constant pressure
boundary condition at the outflow plane. The waterjet stator
and rotor geometry was also modeled. Validation
demonstrated that the standard two equation turbulence
model in combination with wall functions was able to
predict the non uniformities in the duct flow field with
© 2011 American Society of Naval Engineers 217
acceptable accuracy. The results showed that the main inlet
flow characteristics such as cavitation inception at cutwater
where the flow to the duct separates from the main flow,
velocity distribution at the impeller plane, flow separation at
the inlet, the shape of the inlet stream tube are related to the
IVR. The author recommends a dedicated inlet design for
each ship since variations in design ship speed and power
density of the installations cause the design IVR to vary.
The analysis of waterjet for the use of amphibian vehicle
was performed by Jang et al., (2004) to provide detail
understanding of complicated three-dimensional viscous
flow phenomena including interactions of intake duct, rotor,
stator, and contracted discharge nozzle. RANS flow solver
with moving, non-orthogonal multi-block grid system was
used. The CFD results were compared with experimental
fluid dynamics (EFD) and the complex viscous flow feature
of the waterjet, such as the secondary flow inside of the
intake duct, the recovery of axial flow by the action of the
stator, and tip vortex were predicted. The performance
prediction of waterjet for the use of similar vehicle by
diameter sizes and weights were investigated both
numerically and experimentally by Kim et al., (2009).
An extensive study was undertaken to analyze the effect of
integrating RANS calculations into experimental waterjet
powering prediction by Delaney et al., (2009). The
experimental data for the validation was provided by Jessup
et al., (2007), who conducted model tests for the joint high
speed sealift (JHSS) powered with four waterjet systems
and testing incorporated all of the approaches explored by
the ITTC. These included LDV surveys, bollard tests, single
total head probes, and direct measure using weight scales.
Two different JHSS models were considered; each model
houses either axial-flow or mixed-flow waterjet. The hull,
waterjet inlets, and shafts were modeled in the simulation.
Multi-element unstructured grids and boundary layer prism
elements were generated around waterjet geometry. The free
surface was treated as a symmetry plane, and the ship was
modeled at sinkage and trim prescribed by the propelled
experiment. RANS simulation used experimentally
determined volumetric flow rates through the pump as a
condition for the thrust provided by the actuator disk model.
The full scale simulations (Fn=0.35, Reynolds number
Rn=5.3×109) were also performed in order to investigate the
scaling effects by comparing boundary layers. The RANS
delivered pump power predictions showed good agreement
with EFD within one percent at model scale, and within two
percent at full scale.
Hino et al., (2009) performed RANS analysis of a free
surface flow around waterjet propelled high-speed ship
(Fn=1.0, Rn=1.0×106). Free surface location was predicted
using single-phase level set approach. An actuator disk
model in which duct geometry is modeled in a
computational grid was used to simulate the self-propelled
condition. The nozzle shape was not modeled, and dynamic
motions were not predicted. The flow fields of waterjet
propelled simulations, such as free surface elevations,
pressure distributions in the duct center planes, and limiting
streamlines on a ship were compared with the towed
simulations; however, the detailed V&V results were not
given. Takai et al., (2010) investigated the capability of
URANS with an actuator disk model for the simulation of
the JHSS appended with axial waterjets, including waterjet-
hull interactions. The computational setup differs from
Delaney et al., (2009) in that the waterjet-hull interactions
and waterjet-wake interactions are also predicted with free
surface and dynamic motions. The effects of waterjet-hull
interaction are highly non-linear as they include the effect of
the dynamic trim on boundary layer ingestion and shape of
inflow stream tube, together with the effect of the waterjet
induced vertical forces on the dynamic motion. The
waterjet-wake interactions do not significantly affect the
propulsion characteristics, but are of interest in the study of
wake signatures. Self propulsion simulations are carried out
at model scale with full scale thrust identity similarity. The
simulations are carried out over a range of ship speed at
different IVR ratios for the waterjet. On the finest grid with
13 million points, the jet volume flow rate was under
predicted by 6% and was attributed to interpolation errors
caused by extensive use of overset grids within the duct;
each duct had five overset grids. It was recommended that
the number of overset grids be restricted within the duct. An
accurate flow rate measurement is very important since the
estimation of power is dependent upon velocity cubed and
thrust by the velocity squared.
The current work extends Takai et al. (2010) to waterjet
propelled catamarans. The Delft catamaran was selected as
the candidate geometry and was fitted with a custom
designed waterjet based on existing stock waterjets by
BSHC. Data from the model testing using the ITTC '05
procedures include net jet thrust, thrust-deduction, water-jet
volume flow-rate, sinkage, trim, and jet velocity surveys at
nozzle exit. The model was shipped to INSEAN after testing
to quantify facility bias, and testing is currently being
performed in INSEAN. This CFD validation study was done
as a prerequisite for subsequent design optimization of both
the water-jet inlet and the hull form. In future, the optimized
hull and waterjet will be built and model tested at INSEAN.
Kandasamy et al., (2010) derived a simplified integral
force/moment waterjet model for ship powering predictions
that includes the effect of waterjet induced sinkage and trim
on the powering performance without requiring detailed
simulations for the waterjet system (nozzle, pump, ducting
system, and inlet). The CFD waterjet model was also
validated for the Delft catamaran since it will be used in the
preliminary bare hull optimization, before resorting to the
waterjet inlet shape optimization.
The remainder of the paper will be structured as follows:
section 2 presents an overview of the ITTC recommended
procedure for model testing and the CFD waterjet model;
section 3 presents the experimental method used at BSHC;
section 4 presents the computational method; section 5
presents the results from the experiments and the
computations, followed by the conclusions in section 6.
218 © 2011 American Society of Naval Engineers
2.0 ITTC & CFD CONTROL VOLUME METHODS
Fig. 1. ITTC control volume
The ITTC waterjet model combines model testing with
control volume/integral analysis for ship powering
predictions. The selected control volume (Fig. 1) is defined
by a streamtube consisting of the nozzle, pump, ducting
system, inlet, and upstream imaginary surface BC in the
flow through which it is assumed no mass transport occurs
by definition with one outlet A6 and inlet A1A. Vertical
reaction force, weight of the working fluid, and pressure and
shear forces acting on BC are neglected. The inlet A1A is
selected to avoid major flow distortions by the intake
geometry and as practical choice is usually one impeller
diameter in front of the ramp tangency point.
The first step is the resistance test and wake-field
measurements where a resistance test is carried out for a
bare hull model with closed intakes that is free to sink and
trim. The total bare hull resistance RTBH is obtained and is
used later to estimate the thrust deduction factor, t. During
the resistance test the boundary layer velocity profile
u1AX(y,z) is measured at Station 1A. This profile will be used
later to calculate the following items: the intake area at
station 1A, the average velocity at station 1A, , and the
momentum correction factor at Station 1A, cm1, which is
calculated at any station N using Eq. (1-2)
21
NMN
ANN
uc dA
A V
(1)
1
N
NA
N
V V n dAA
(2)
Next is the calibration and propulsion test. The purpose of
the calibration test is to establish a relation between a
measurement signal at the jet (often a differential pressure
transducer or Kiel probe is used) and the jet-thrust (TJx)
which is measured through the Bollard pull test. The flow
rate (Q) is then calibrated through the momentum flux
approach, since direct measurement of flow-rate is prone to
higher uncertainties. Assuming negligible inlet axial
velocity, TJx is equal to the momentum flux at the jet nozzle
providing Eq. (3) for estimating Q.
6
6 cos
Jx
m
T AQ
c (3)
The momentum correction factor at the jet exit cm6 is
obtained from detailed jet velocity profiles using LDV. α is
the jet angle relative to the horizontal at the nozzle (station
6). This calibration is assumed to hold good even with non-
zero forward velocity, so that the Kiel probe measurements
taken during self-propulsion tests could be used to estimate
Q.
After calibration, the propulsion test is carried out to
determine relation between speed, flow-rate and thrust at
self propulsion point. The calibrated Kiel probe
measurements provide Q at the self-propulsion point. Q is
then used to determine the size of the inlet capture area from
the inlet wake-field measurements by applying conservation
of mass. Once the capture area is determined cm1 can be
calculated. All the relevant variables to predict the waterjet
net thrust (RX) from Eq. (4) are now known.
2
6 0
6
cosX M Min
QR c Qc U
A
(4)
Note that the ITTC model ignores vertical forces and
pitching moments. These are not relevant for the purposes
of the original ITTC model, since sinkage and trim are
measured during the thrust evaluation procedure. For a CFD
simulation, however, it is desirable to estimate the resulting
waterjet-induces forces and moments affecting sinkage and
trim, since the final attitude of the ship affects the resistance
and wave-generation characteristics of the ship. Hence, a
modified control volume is selected as shown by the shaded
area in Fig. 2.
Fig. 2. CFD control volume
The choice of the proposed control volume which is for the
waterjet is motivated by the possibility of experimentally
measuring the pressure and velocities at the waterjet inlet,
which allows for the computation of the inlet flow angle ,
momentum flux correction factor CMin, pressure force
in inP A , and pressure distribution at stern which provides
the waterjet/hull interaction stern force ΔFS. The model can
then be used during early design optimization stages to
includes the effect of waterjet induced sinkage and trim on
the powering performance without requiring detailed
simulations for the waterjet system (nozzle, pump, ducting
system, and inlet).
© 2011 American Society of Naval Engineers 219
3.0 EXPERIMENTAL METHODS
Table 1 provides the main particulars of the Delft catamaran
model.
Table 1. Main particulars for the Delft catamaran model
Main Particulars Symbol Model
Length overall, m LOA 3.8220
Length between perpendiculars,
m
LPP 3.6274
Length on waterline, m LWL 3.6274
Breadth moulded, single hull, m B 0.2904
Clearance b/n hull CPs, m - 0.8470
Draft at FP, m TF 0.1815
Draft at AP, m TA 0.1815
Displacement volume, m3 Δ 0.0770
Prismatic coefficient CP 0.6160
Block coefficient CB 0.4027
Longitudinal C.B. LCB -0.0970
Wetted surface area (bare hull),
m2
S 1.4220
A waterjet propulsion unit, consisting of ducting system,
inlet opening, axial flow pump and nozzle, was incorporated
in each of the catamaran hulls. Fig. 3 shows the stock
waterjet pump arrangement used for testing.
Fig. 3. Waterjet propulsion unit used for testing
The model testing was carried out in BSHC deep water
towing tank. The towing tank main dimensions is 200 m
length x 16 m breadth x 6.5 m depth. Two wave absorbing
beaches are arranged, one front and one side dampers. The
towing carriage has a maximum speed of 6 m/s. The
carriage is used to guide the ship model and to hold the
measuring and supplementary equipment. A propeller
dynamometer was installed on the starboard hull for
impeller generated thrust and torque (delivered power)
measurements. Three forms of pressure reading
instrumentation were used for the hull model testing: i)
static pressure taps located around the inlet opening of
starboard hull; ii) one static pitot tube was used to measure
the hull boundary layer velocity profiles and one static pitot
tube for measuring the reference dynamic pressure at the
starboard hull nozzle exit; iii) a 5-hole probe was used to
determine flow-field velocity distribution at the nozzle exit
plane of starboard hull at bollard conditions, necessary to
calculate the nozzle momentum correction factor and
bollard flowrate.
During the resistance test the boundary layer velocities were
measured along a normal to hull surface, one inlet diameter
upstream of inlet centerline, with the inlets closed. The runs
were made at 7 different model speeds, corresponding to
Froude numbers 0.4, 0.45, 0.5, 0.55, 0.6, 0.65 and 0.7. The
measurements were performed by means of a static pitot
tube. It was assumed that the velocity profiles thus obtained
do not change in traverse direction (they vary only in z –
direction) and so they fully describe the flow pattern at
Station 1A. The obtained velocity profiles were later used to
calculate the intake area and the inlet momentum correction
factor Cm1 . There is some controversy on whether the
velocity survey should be made with closed intake (nominal
wakefield) or with open intake and active waterjet (total
wakefield). Terwisga (2005) states that ideally one would
like to measure the effective wake ingested by the intake.
That is the flow field including the suction effects on the
flow about the hull, without any effect of the intake flow
itself. This so called “effective wake” is however difficult to
measure (intake flow is always present) and depends on the
working point of the intake. For that reason, it is suggested
to measure the boundary layer velocity profile with closed
intake openings, which contradicts the previous tentative
procedure described in the ITTC Quality Manual(4.9-03-03-
05.2).
Bollard pulling forces were measured with both pumps
operating within a range of impeller revolutions. The
obtained data were used later for flowrate calibration. The
reference nozzle dynamic pressure (measurement signal)
necessary for flowrate calibration was measured with a
static pitot tube located at about 70% of nozzle exit radius,
i.e. about the middle of the nozzle flow area, at 12 o‟clock
position. The main objective of these tests was to define the
model self-propulsion parameters and to gain data necessary
for self-propulsion flowrate estimation. No skin friction
correction force was applied during the tests, thus all results
refer to model scale conditions. For all the tests, the model
was free to heave and pitch, but was restrained in yaw and
roll. The tests were performed with outward turning
impellers. 3D velocity survey at nozzle exit (Station 6) was
performed using 5-hole pressure probe. The obtained local
axial velocity distribution was used to evaluate the nozzle
momentum correction factor Cm6 at that Station, used for
flowrate calculation.
4.0 COMPUTATIONAL METHODS
The Unsteady Reynolds-Averaged Navier Stokes (URANS)/
Detached Eddy Simulation (DES) flow solver, CFDSHIP-
IOWA has been developed at IIHR –Hydroscience &
Engineering– over the past 15 years for ship hydrodynamics
applications (Carrica et al., 2007). For the present work,
220 © 2011 American Society of Naval Engineers
URANS with the blended k-ε/k-ω turbulent model is
selected as a flow solver. The free surface location is
predicted by a single phase level set method. A second order
upwind scheme is used to discretize the convective terms of
momentum equations for URANS. A pressure-implicit split-
operator (PISO) algorithm is used to enforce mass
conservation on the collocated grids. The pressure Poisson
equation is solved using the PETSc toolkit (Belay et al.,
2002). All the other systems are solved using an alternating
direction implicit (ADI) method. For a high performance
parallel computing, a MPI-based domain decomposition
approach is used, where each decomposed block is mapped
to one processor. The code SUGGAR (Noack, 2005) runs as
a separate process from the flow solver to compute
interpolation coefficients for the overset grids and
communicates with a motion controller (6DOF) within
CFDSHIP at every timestep. The software USURP (Boger
and Dreyer, 2006) is used to compute area and forces on the
surface overlapped regions. In addition, a simplified body
force model is used for waterjet propelled simulation to
prescribe axisymmetric body force with axial and tangential
components (Paterson et al., 2003). The propeller model
requires thrust, torque, and advance coefficients as input and
provides the torque and thrust forces. These forces appear as
a body force term in the momentum equations for the fluid
inside the propeller disk. The location of the propeller is
defined in the static condition of the ship and moves
according to the ship motions.
Based on recommendations from Takai et al. 2010, care was
taken to limit the number of overset grids in the duct and
reduce interpolation errors for volume flow rate. The duct is
discretized using a single structured grid which overlaps
with the hull grid at the inlet ant the nozzle exit (Fig. 4).
Fig. 4. Overset grid system for Delft catamaran with
symmetry plane.
A symmetry boundary condition was used and a body-fitted
“O” type grids are generated around the port side ship hull
geometry. A rectangular background grid is used with
clustered grid near the free surface to resolve the wave field.
A cylindrical refinement block was generated immediately
following the nozzle exit to better resolve the exiting jet. In
the present work, the shaft and the downstream rotor are not
included in order to avoid the complexity of the grid design
since the present work is prerequisite for optimization work.
For self propelled simulations, a total of 6.4 million grid
points is split into 64 blocks with an average of 100K grid
points/block by the MPI based domain decomposition. The
simulation domain is extended to [-0.5, 2.5], [-0.7, 0.7], [0,
1.3] in streamwise, vertical and spanwise directions,
respectively. The boundary conditions are detailed in Table
2.
Table 2. Boundary conditions
Description p U V W
In.
Resist.
Self-
propelled
Exit
Bottom, sides
Top
Symmetry
No slip (ship
hull)
For barehull resistance computations, the ship is initially
static on calm water. The ship is then allowed to pitch and
heave under a constant inlet fluid velocity until a steady
state is reached. Ship-fixed coordinate system is used, which
means that there is no surge motion allowed for the ship and
the background grid. For self-propulsion simulation, an
actuator disk model is used to prescribe axisymmetric body
force with axial and tangential components. The simulations
mimic the experiment; the ship accelerates until the
resistance equals the prescribed thrust and added tow force
and converges to the self propulsion point. 2-5 nonlinear
iterations are required for convergence of the flow field
equations within each time step. Convergence of the
pressure equation is reached when the residual imbalance of
the Poisson equation drops six orders of magnitude. All
other variables are assumed convergence when the residuals
drop to 10-3
.
5.0 RESULTS
For all the tests, the model was free to heave and pitch, but
was restrained in yaw and roll. The model resistance,
sinkage and trim were investigated in a Froude number
range 0.055 - 0.700, corresponding to towing speed range
of 0.298 m/s - 4.175m/s. The measured barehull resistance,
sinkage and trim are compared with the CFD calculations in
Fig. 5. The sinkage was measured with two string pots for
displacement measurements, one located on the leading
post (1708 mm forward of midship) and a second located on
the tow post (97 mm aft of midship). The measured data
were used to calculate the trim angle. The sinkage is
measured at the tow post location.
© 2011 American Society of Naval Engineers 221
(a)
(b)
(c)
Fig. 5. EFD vs CFD comparisons: (a) barehull resistance
and net jet thrust; (b) Sinkage, and (c) trim
Table 2. Model self-propulsion points
V
m/s
Fn
-
n
rps
Q
Nm
Rx
N
PdREF,
kgf/m2
2.3856 0.400 24.578 1.265 37.97 813.7
2.6838 0.450 29.885 1.888 57.86 1208.2
2.9820 0.500 33.112 2.310 71.62 1511.3
3.2802 0.550 34.808 2.567 80.44 1706.8
3.5785 0.600 36.574 2.833 87.68 1863.8
3.8767 0.650 38.433 3.132 95.96 2069.3
4.1749 0.700 40.010 3.359 101.72 2238.7
Table 2 shows the impeller rps, thrust and torque at model
self-propulsion point. The flow rates Q during self
propulsion were calculated using the reference pressure
Pdref, at Station 6 which were taken at the same static pitot
tube location as at bollard condition and the self propulsion
flow rates were using the bollard flowrate calibration
relation. The measurements were taken at the same Static
Pitot tube location, as at bollard condition. Fig 6 compares
the computed volume flow rates with the measured values.
CFD underpredicts the volume flow rate by ~2% over the
speed range.
Fig. 6. Comparison of volume flow rates
The flow rate was used to determine the size of the inlet
capture area by applying conservation of mass. For all tested
craft speeds, the inlet capture area was assumed to be
rectangular and its size was determined from BH inlet
velocity-field measurements. Fig. 7 shows the CFD inlet
inlet capture area got by seeding the flow through the duct
with stream lines which is elliptical similar to previous CFD
simulations (Takai et al., 2010). Van Terwsaga (2005)
concluded that the inlet capture area for Athena was also
elliptical, but the shape does not have significant effect on
the ingested momentum and energy flux. However, it was
noted that there is at least one reference (Roberts and
Walker, 1998) claiming that the choice of a rectangular
capture area may lead to an under-prediction of gross thrust
by ~10% for a typical high speed ferry.
Fig. 7. Inlet capture area obtained from CFD for Fr=0.53
222 © 2011 American Society of Naval Engineers
The net jet thrust Rx was determined using Eq. (4) and
shown in figure 5a. Overall, the net jet thrust Rx is ~10%
higher than the barehull resistance. Simulations using the
CFD waterjet model, which include the effects of just the
waterjet induced sinkage and trim on the barehull indicate
an increase of ~5% in barehull resistance. The model scale
thrust deduction fraction, t, accounting for all the interaction
effects of the propulsion system on bare hull system is
shown in Fig. 8.
Fig. 8. Thrust deduction factor
Fig. 9 compares the EFD and CFD flow fields at Fr=0.55
and 0.53, respectively. The diverging wave fronts from the
bow show qualitative agreement. EFD shows considerable
splashing at the jet exit. The splash is qualitatively resolved
by the CFD (Fig. 10), but without any air entrainment
effects. Fig. 10 also shows the pressure distribution inside
the duct with a rapid pressure rise after the actuator region,
which get converted into kinetic energy while passing
through the nozzle.
Fig. 9. Flow field comparison
Fig. 10. Duct pressure distribution and exit jet splash
6.0 CONCLUSIONS
URANS simulations for both barehull and waterjet
propelled Delft catamaran is presented and compared with
model test results. CFDSHIP-IOWA V.4 is employed as a
flow solver, which solves URANS with the blended k-ε/k-ω
turbulent model, single-phase level set method, and
simplified body force model are adopted to simulate the
waterjet propelled ship flow. The present work was
performed to investigate the capability of URANS for the
accurate simulation of waterjet propelled catamarans as a
pre-requisite for CFD based design optimization of both the
hull form and the intake duct shape.
Due to the complexity of the EFD measurement process,
significant facility bias errors were encountered during the
validation of the standardized testing (Van Terwisga, 2005)
by nine separate ITTC member organizations. Hence, the
current Delft model which was built and model tested by
BSHC was shipped to INSEAN to quantify facility bias
errors and testing is currently being done.
Overall, the CFD results match reasonably well with the
experiments. The bare hull resistance was under predicted
by ~2.5%. The flow rate was under-predicted by ~3%.
Previous work on the JHSS (Takai et al., 2010) with the
same code with multiple overset grids in the duct region
showed ~6% underprediction of flow rate due to
interpolation errors. The lower error for the current case can
be attributed to the fact the duct is discretized using a single
grid, and hence there are no interpolation errors. The CFD
under-predicts the net jet thrust by ~5% over the entire
speed range.
It should be noted that certain evaluation procedures differ
between CFD and EFD. The inlet capture area for CFD is
elliptical based on streamtubes entering the duct, whereas a
rectangular capture area was used for the EFD. Also, the
velocity profiles used to determine the capture area in EFD
were measured with the inlets closed as per ITTC
recommendations, but this is not the case with CFD, where
the elliptical capture area obtained with the ducts open was
used. Differences are expected to be negligible as per Van
Terwisga (2005). However, a thorough quantification of the
differences in wake fraction due to the inlet shape changes
and differences in velocity profile with open and closed
inlets using CFD will be beneficial.
© 2011 American Society of Naval Engineers 223
Including the effects of the waterjet induced sinkage and
trim using the CFD waterjet model captures some of the
waterjet system- bare hull system interaction effects, and
indicates an increase in resistance by 4%. This accounts for
some of the interaction effects which contribute to the thrust
deduction, but does not account for the effect of the
boundary layer ingestion on thrust which is one of the most
dominant jet-hull interaction as addressed by Van Terwisga
(1995) and Hoyt et al. (1999).
In addition to the thrust deduction, which accounts for the
effects of waterjet hull interaction, the propulsive efficiency
is an important parameter needed for accurate assessment of
delivered powering requirements. Calculation of the overall
propulsive efficiency requires accurate estimation of the
pump efficiency which necessitates modeling of the shaft,
hub, rotor and stator (Bulten and Van Esch, 2007) which is
out of scope of the present work and needs to be addressed
in future work.
The present work demonstrates the feasibility of using
URANS for performance analysis of hull-integrated waterjet
propelled ship with free surface and dynamic motions and
work paves way for waterjet inlet optimization studies with
main objective of decreasing powering requirements by
increasing the inlet efficiency through modification of
intake duct shape (Kandasamy et al., 2011).
REFERENCES
Balay, S., Buschelman, K., Gropp, W., Kaushik, D.,
Knepley, M., Curfman, L., Smith, B. and Zhang, H.,
(2002) „PETSc User Manual,‟ ANL-95/11-Revision
2.1.5, Argonne National Laboratory
Boger, D.A. & Dreyer, J.J. (2006) „Prediction of
Hydrodynamic Forces and Moments for Underwater
Vehicles Using Overset Grids‟, AIAA paper 2006-1148,
44th AIAA -Aerospace Sciences Meeting, Reno,
Nevada, 2006
Bulten, N.W.H. (2006). „Numerical Analysis of Waterjet
Propulsion System‟, PhD thesis, Technical University of
Eindhoven, ISBN-10: 90-386-2988-5. Library
Eindhoven University of Technology
Bulten, N.W.H. & Van Esch, B.P.M. (2007) „Fully transient
CFD analyses of waterjet pumps‟, Marine Technology,
44(3), pp. 185-193
Carrica, P. M., Wilson, R. V., and Stern, F. (2007) „An
unsteady single-phase level set method for viscous free
surface flows‟ International Journal for Numerical
Methods in Fluids, Vol. 53, pp. 229-256
Delaney, K., Donnely, M., Elbert, M., and Fry, D. (2009)
„Use of RANS for Waterjet Analysis of a High-Speed
Sealift Concept Vessel‟, 1st International Symposium on
Marine Propulsors, Trondheim, Norway
Hino, T.., Ohashi, K. (2009) „Numerical Simulation of Flow
around a Waterjet Propelled Ship‟ 1st International
Symposium on Marine Propulsors, Trondheim, Norway
Hoyt III, J.G. (Chairman), 1999, “Report of theSpecialist
Committee on Waterjet”, 22ndITTC, Seoul/Shanhai.
Jang, J.H., Park, W.G., Boo, J.S., Chun, H.H., & Kim, M.C.
(2004) „Numerical Simulations of Waterjet with Rotor-
Stator Interaction‟, 10th
international symposium on
transport phenomenon and dynamics of rotating
machinery, Hawaii
Jessup, S., Donnelly, M., Fry, D., Cusanelli, D., & Wilson,
M. (2008) „Performance Analysis of a Four Waterjet
Propulsion System for a Large Sealift Ship‟, 27th
symposium on Naval hydrodynamics, Seoul, Korea
Kandasamy, M., Ooi, S.K., Carrica, P., & Stern, F. (2010)
„Integral force/moment water-jet model for CFD
simulations‟, Journal of Fluids Engineering, Vol. 132,
101103-1
Kandasamy, M., He, W., Tahara, Y., Peri, D., Campana, E.,
Wilson, W., & Stern, F. (2011) 'Optimization of waterjet
propelled high speed ships - JHSS and Delft catamaran',
submitted to FAST 2011, Hawaii
Kim, M-C., Chun, H-H., Kim, H. Y., Park, W. K., Jung, U.
H. (2009) „Comparison of waterjet performance in
tracked vehicles by impeller diameter‟, Ocean
Engineering, Vol. 36, pp. 1438-1445
Noack, R. (2005) „SUGGAR: a General Capability for
Moving Body Overset Grid Assembly‟, AIAA paper
2005-5117, 17th
AIAA Computational Fluid Dynamics
Conference, Toronto, Ontario, Canada
Paterson, E. G., Wilson, R. V., and Stern, F. (2003)
„General-Purpose Parallel Unsteady RANS Ship
Hydrodynamics Code: CFDSHIP-IOWA‟, IIHR
Technical Report, #432, The University of Iowa
Roberts, J.L., and Walker, G.J., 1998,“Boundary layer
ingestion effects in flushwaterjet intakes”, International
conferenceon Waterjet Propulsion II, RINA,
Amsterdam,The Netherlands.
Takai, T., Kandasamy, M., & Stern, F., (2011). 'Verification
and validation study of URANS simulations for axial
waterjet propelled large high speed ship', submitted to
Journal of Marine Science and Technology.
Van Terwisga, T. et al. (2005) „Report of the Specialist
Committee on Validation of Waterjet Test Procedures‟,
Proceedings 24th
Int. Towing Tank Conference; II: 471-
508
ACKNOWLEDGEMENTS
This work is sponsored by the US Office of Naval Research
through research grants N00014-08-0491, under the
administration of Dr. Ki-Han Kim. The simulations were
performed on 4.7GHz IBM Power 6 machine „DaVinci‟ at
the DoD NAVO center.
224 © 2011 American Society of Naval Engineers