Ocean Engineering 171 (2019) 259–272

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Prediction of hydrodynamic performance of pump propeller considering the T

effect of tip vortex
Chao Wang, Kaiqiang Weng, Chunyu Guo∗, Lang Gu∗∗
College of Shipbuilding Engineering, Harbin Engineering University, No.145 Nantong Street, Nan Gang District, Harbin, Heilongjiang Province, 150001, PR China

ARTICLE INFO ABSTRACT

Keywords: There is a strong gap flow in the pump jet propulsor, in order to predict the hydrodynamic performance of the
Panel method pump jet propulsor, the steady-state hydrodynamic disturbance model was established by the panel method. The
Pump jet propulsor pump jet propulsor was divided into two systems, rotor-hub and stator-duct, and inter-system interference was
Performance prediction considered by induced velocity. Aiming at the geometric shape of stator, a method of leaf meshes based on
Tip vortex model
circular cone was proposed. Based on the actual situation of the internal flow of the pump, a tip leakage vortex
model suitable for the flat-topped blade was proposed. Comparing the results of pressure, circulation, and
propulsion performance, it was found that when the influence of tip vortex is considered, the overall annular
distribution of rotor blade and tip load are closer to the actual situation, and the performance curve is in good
agreement with the viscous flow, which shows that this method can effectively predict the steady hydrodynamic
performance of the pump jet propulsor.

1. Introduction the same composition as the stator ducted propeller, the geometry of
the various components is quite different. For example, in the case of
Pump jet propulsor in the form of structure belongs to the stator torpedo or submarine pump jet propulsor, due to the restrictions of the
with a combination of duct-style propulsor; the main difference with flow channel boundary, the upper and lower faces of the fixed rotor
the duct propeller is the introduction of pump type blade wheel. may not be parallel to the axis in the axial projection view. Therefore,
Depending on the design application, there are some differences among the conventional method of drawing a geometric shape based on a
the types of ducts used in the pump propeller. Torpedo or submarine concentric cylindrical surface will not be able to be used for such a
pump jet propulsors typically use a decelerating duct, while the inclined rotating blade. In addition, due to the viscous effect of the boundary
surface of the tail of the carrier provides an angle of attack for the axial layer of the wall when the flat top rotary vane is working in the pipe,
section of the duct. This duct design ensures that the pump in the water there are separate vortices, induced vortices and leaking vortices dis-
flow both inhibits and delays the occurrence of cavitation. Pump jet tributed along the chord at the tip of the blade (You et al., 2007), one of
propulsor hydrodynamic research mainly includes experimental re- the leaking vortices as the leading movement, and even the formation
search (Suryanarayana et al., 2010a,b; B. Satyanarayana et al., 2010) of tip vortex cavitation. Zhang et al. (Zhang et al., 2015) summarized
and CFD (Computational Fluid Dynamics) simulation (Pan et al., 2013; motion trail and change rule of the tip vortex of the axial flow pump
Wang and Zhou, 2004; Fu et al., 2016; Shi Yao et al., 2014; Lu et al., through experimental observations and CFD numerical simulation. The
2016a,b; Lu et al., 2017; Guo et al., 2018; Feder et al., 2018). Many compact layout of the pump jet propulsor allows the impact of the tip
numerical studies based on potential flow method have focused on the leaking vortex to be limited not only at the tip of the rotor but also at
ducted propeller with stator. The earliest potential flow methods were the overall system performance.
the lifting surface vortex method (Hughes and Kinnas, 1991a,b) and the In this paper, the panel method based on potential flow theory is
surface panel method (Kawakita, 1992) adopted by Hughes et al. and used to establish the numerical model of the pump jet propulsor. A
Kawakita et al., respectively, and most of the follow-up research work is method of mesh based on a circular truncated cone and a tip leakage
based on these methods (Wang and Yang, 1999; Wang and Liu, 2007; vortex model suitable for flat-topped blades is given, and the effects of
Liu and Wang, 2006; Su, 1999). Although the pump jet propulsor has tip vortex model are analyzed by the calculation results of the pressure

∗
Corresponding author.
∗∗
Corresponding author.
E-mail address: [email protected] (C. Wang).

https://doi.org/10.1016/j.oceaneng.2018.10.039
Received 13 March 2018; Received in revised form 9 July 2018; Accepted 23 October 2018
0029-8018/ © 2018 Elsevier Ltd. All rights reserved.
C. Wang et al. Ocean Engineering 171 (2019) 259–272

incompressible flow of V0 , the perturbation potential at any one point

can be represented by the following integral according to the Green's
formula (Su, 1999):
1 (Q ) 1
4 E (P ) = [ (Q) ( ) ] dS
Sb+ Sw + S
nQ RPQ nQ RPQ (1)
where, S is the boundary of the fluid, which is composed of the object
surface Sb , the wake vortex surface Sw , and the outer boundary surface.
Q is the arbitrary control point on the boundary surface S . RPQ is the
straight line distance between two points P and Q . (Q )/ nQ is the
normal derivative of the point velocity potential—satisfying the non-
penetrating surface condition. E is the Green formula parameter—the
value of E is 0 when the point P is within S , 0.5 when P is on S , and 1
when P is outside S .
On each part of the boundary surface S , the following boundary
conditions should also be satisfied:
0 When S (2)

/ nQ = V0 nQ On Sb (3)

p+ p =0
On Sw
Fig. 1. Front stator pump jet propulsor model. ( )+ ( ) =0
nQ1 nQ1 (4)

distribution, circular distribution, and the performance indexes, etc. where, Q1 is the point on the wake vortex surface Sw ; super-
scripts + and - represent the upper and lower surfaces of the wake
vortex respectively.
2. Numerical model
Referring to the boundary conditions of Eqs. (2)–(4), the integral Eq.
(1) on the boundary surface S can be expanded to:
The theoretical method of establishing the numerical model in this
paper was based on velocity potential low-order surface panel method, 2 (P ) = (Q )
1
( )dS + (Q1) (
1
)dS+
in which the discrete elements on the surface of the object are hyper- Sb
nQ RPQ
Sw
nQ1 RPQ1

bolic elements, and the singularities arranged on the elements are the 1
( V0 nQ )( R )dS
mixed distribution of source and dipole. The numerical model of the Sb
PQ
(5)
pump jet propulsor was established by solving the basic equations of a
lifting body, and the tip leakage vortex model was added to the original where, n is the unit normal vector of the corresponding object surface
model. In order to verify the reliability of this method, this paper and points to the flow field. is the velocity potential jump through
considered the front stator and rear stator pump models, as shown in the wake vortex surface and can be written as:
Fig. 1 and Fig. 2. = +
(6)
The value of is obtained using the Kutta condition, which re-
2.1. The basic equation
quires that the pressure difference between the upper and lower surface
of the lift body tail is zero, that is
Assuming that the lift body operates in a irrotational, non-viscous,
+
( p)TE = pTE pTE = 0 (7)
Eq. (5) is the basic equation for solving the flow around a lift body.
Combining Eq. (5) with the Kutta condition Eq. (7), a unique numerical
solution of velocity potential on the surface can be obtained.

2.2. Numerical model of the pump jet propulsor

The components of the pump jet propulsor are divided into two
independently existing lift bodies in the same coordinate system, rotor-
hub system and stator-duct system, as shown in Fig. 3 and Fig. 4. The
two systems respectively constitute the basic equations, and the inter-
action between systems takes into account the perturbation of the flow
field around each system, while the interference effect is expressed in
the form of induced speed. The equations of velocity and induced ve-
locity of each system are solved in turn. By way of induced velocity loop
iteration, the numerical solutions of both systems reach the con-
vergence criterion, so as to obtain the overall calculation results of the
pump jet propulsor. The following is a description of the velocity po-
tential and the induced velocity for the stator-to-hub system and the
stator-to-duct system of the rear stator pump jet propulsor model and
the front stator pump jet propulsor model.
According to Eq. (5), the equations for solving the velocity potential
Fig. 2. Rear stator pump jet propulsor model. of stator-duct system and rotor-hub system at the boundary are

260
C. Wang et al. Ocean Engineering 171 (2019) 259–272

(a) Rotor-Hub System

(a) Rotor-Hub System

(b) Stator-duct System

Fig. 3. Rear stator pump schematics. (a) Rotor-hub system. (b) Stator-duct
system.

expressed as:
(b) Stator-duct System
Fig. 4. Front stator pump schematics. (a) Rotor-hub system. (b) Stator-duct
2 k (P ) = k (Q ) 1
( R )dS + k (Q
1) n ( R
1
)dS+ system.
nQ PQ Q1 PQ1
Ss + Sd Sws + Swd

k 1
(( V0 + Vsd , rh ) nQ )( R )dS 4 k
Vrh k (Q ) 1
, sd (P ) = P n ( R )dS
(8)
PQ
Ss + Sd Q PQ
Ss + Sd
k (Q ) 1
+ 1 P n (R )dS
Q1 PQ1
k (P ) k (Q ) 1 k (Q 1 Sws + Swd
2 = ( )dS
nQ RPQ
+ 1) n ( R )dS+
Q1 PQ1
Sr + Sh Swr k 1 1
+ (( V0 + Vsd, rh ) nQ ) P (R )dS
(11)
PQ
k 1 Ss + Sd
(( V0 + r + Vrh , sd ) nQ )( R )dS
(9)
PQ
Sr + Sh
0
When solving for the first time, Vrh, sd is 0 if the rotor and hub are
where Ss , Sd , Sr , and Sh , respectively, are the stator surface, duct surface, 0
first calculated; otherwise V sd, rh is zero. Eqs. (8)–(11) are the integral
rotor surface, and hub surface; Sws , Swd , and Swr are the vortices of the equations of velocity potential and induced velocity of each system.
corresponding components; is the rotational angular velocity of the According to the discrete integral equations of the grid elements, an
propeller; k represents the kth iterative operation; Vsd, rh represents the equation system consisting of dipoles and source sinks is formed.
induced velocity generated by the rotor-hub system at the stator and Combining the Kutta condition of rotor-hub system and stator-duct
duct surfaces as part of the relative inflow velocity of the stator-duct system, the velocity of panels was obtained by iterative method. Fi-
system, and Vrh, sd is in the opposite direction. By taking the gradient of nally, the pressure distribution on object surface was obtained ac-
the integral Eq. (1) outside the boundary surface, the expressions for cording to the Bernoulli equation. Pump jet propulsor performance
the induced velocities Vsd, rh and Vrh, sd are obtained as follows: indicators can be expressed as:

Tr Qr Ts
k 1 1 KTr = ; K Qr = ; KTs = ;
4 Vsd k (Q ) k (Q ) n2Dr4 n2Dr5 n2Dr4
, rh (P ) = P n ( R )dS
Q PQ
+ 1 P n (R
Q1 PQ1
)dS
Sr + Sh Swr Qs Td KT all J
KQs = ; KTd = ; =
n2Dr5 n2Dr4 K Qr 2
k 1 1
+ (( V0 + r + Vrh , sd ) nQ ) P ( R )dS
PQ
Sr + Sh where KT and K Q are the thrust coefficient and torque coefficient, re-
(10) spectively; the subscripts r, s, and d correspond to the rotor, stator, and
duct, respectively; the total thrust coefficient is KT all = KTr + KTs + KTd ;

261
C. Wang et al. Ocean Engineering 171 (2019) 259–272

Fig. 5. Side projection profiles.

Fig. 9. Geometric model of the duct propeller with pre-swirl stator blades.

Fig. 10. The whole computational domain and rotation domain.

is the efficiency of the pump jet propulsor; the speed factor is

J = V0/ nDr ; Dr is the maximum diameter of the rotor; n is the rotor
Fig. 6. Truncated blade section diagram three-dimensional map. speed.

2.3. Boundary elements of the division

The meshing method based on the coaxial cylindrical surface is

more common in past research. It could be applied to the stator-to-rotor
model of the front stator pump jet propulsor model in this paper, but it
would not be suitable for the rear stator pump jet propulsor model with
tilted hub in this paper. As can be seen from the side projection profile
in Fig. 5, there is a certain angle between the mid-hub and the inner
wall of the duct, respectively, with the x-axis. The upper and lower
faces of the stator and rotor are also tilted. The cylindrical surface
cannot penetrate the leading edge and the trailing edge of the blade.
Fig. 7. Rear stator pump jet propulsor gap leakage vortex model. Therefore, this paper uses the coaxial circular truncated cone surface to
divide the blade sections of the stator and the rotor, as shown in Fig. 6.
First, the projection of the leading edge and the trailing edge in the
radial direction of the division, the radial positions rl , rt , and the axial
positions xl , x t , of the leading and trailing edge on the jth radial profile
are obtained. Then, the conversion of cone angle and vertex position
x 0 is used to determine the shape and location of the cutting circular
truncated cone. Combined with the jth blade section chord length c , the
leading edge to the generatrix chord distance cl , trim value x r , skew
angle , and other geometric parameters, using the sine and cosine
theorem of arbitrary triangle, are used to determine the distance dOP
between the arbitrary point P of the blade section and the vertex O of
the cone on the plane of the circular truncated cone (the fan-shaped
plane OC1 C2 shown in Fig. 6), and the point P in the circular truncated
cone to yz plane arc distance d PB1 . The blade is segmented by cosine in
the chord direction and extension direction, respectively, so the co-
Fig. 8. Front stator pump jet propulsor gap leakage vortex model. ordinates of the points on the cross section can be obtained from dOP
and d PB1 , and the expression is

262
C. Wang et al. Ocean Engineering 171 (2019) 259–272

Table 1
The results of calculation and experiment.
J = 0.8 Test value 2 million 3.5 million 6.11 million

Kt-all 0.247 0.262 0.255 0.254

Kq 0.486 0.510 0.501 0.493
ƞ 0.642 0.654 0.648 0.655

1.0 EXP CFD-EXP

CFD-Kt-s&d
CFD-Kt-all
0.8
(a) stator and rotor CFD-10Kq

KT 10KQ effiency
0.6

0.4

0.2

0.0
0.2 0.4 0.6 0.8 1.0 1.2
J
(b) duct
Fig. 12. The comparison of open-water performance results on whole condi-
tions.

(c) cross section of rotor

Fig. 13. Sectional view.

2.4. Tip leakage vortex model

Based on previous CFD simulations (You et al., 2007) and experi-

ments (Zhang et al., 2015), it is known that the vortex line of the
leaking vortex is usually emitted from the tip of the blade, after the gap
area of the wall constraints, fully developed in the dorsal side. The pitch
angle of the vortex line decreases with the increase of axial distance
until a point somewhere downstream of the blade trailing edge, which
(d) rotor tip domain close to the pitch angle of the vortex line of the wake vortex in the end.
Fig. 11. The mesh generation. Based on the above conclusions, a geometrical model of rotor tip
leaking vortex was established for the pump jet propulsor in this paper,
as shown in Fig. 7 and Fig. 8.
x = dOP cos x0
The vortex line is emitted from the connection point between the
r = dOP sin
top surface of the rotor and the blade face. The initial pitch angle of
= d PB1 / r (12) each vortex line is the hydrodynamic pitch angle 0 of the starting point
of the vortex line. As the axial distance increases, the pitch angle
decreases continuously until a point somewhere downstream equal to

263
C. Wang et al. Ocean Engineering 171 (2019) 259–272

0.30 Tip vortex free model

Tip vortex model

Ktr
0.25

0.20

0.15
(a) the stator(b) duct 0.6 0.7 0.8
J
0.9 1.0 1.1

(a) Rotor thrust coefficient curve

0.10
Tipless vortex model
0.08 Tip vortex model

0.06

Kts,Ktd
0.04

0.02

(c) the rotor(d) the hub 0.00

Fig. 14. Potential flow calculation model meshing.

-0.02
0.6 0.7 0.8 0.9 1.0
J
the geometrical pitch angle g of the rotor tip section, then extends
along the downstream with a fixed pitch angle g . In order to simplify
the method of dividing the tip vortex elements, we assumed that the (b) Stator and duct thrust coefficient curves
pitch angles of the vortex lines tend to be the same at the same axial
Fig. 15. Pump stator, rotor, and duct thrust coefficient curves.
position T , and let T be the point of the axial coordinate xT , as shown in
Fig. 7. The trend of the radial coordinate of the vortex line is consistent
with the inner wall of the duct, and can be obtained by interpolation. As shown in Fig. 9, the geometric model contains 9 stator blades and
The geometric coordinates of the ith unit on the vortex line can be 4 rotor blades, and the rotor diameter D is 0.254 m. The detail in-
expressed as: formation of this calculation model can be obtained in the reference
(Hughes and Kinnas, 1991a,b).
xi = xi 1 + x
ri = f (x i)
3.1.1. Grid partition of the computing domain
i = i 1 + (x i x i 1)/ ri /tan( i ) (13) The computational flow field was divided into two cylindrical
computational domains that simulated the infinity boundary and the
+( 0 )(i 1)/ numT 0 xi xT
=
0 g inner domain that simulates the rotor's rotation. The diameter of outer
i
g xi xT (14) domain is 15D selected to eliminate wall effect on the rotating area. The
cylindrical surface was set as a symmetrical plane, and the speed inlet
Where numT is the number of cells from the start point to the T point. and the pressure outlet were both 10D away from the leading edge of
The connecting facets of the rotor blade face and the top face, as the duct. The inner domain contained only the rotor, and the outer
well as the connecting facets of the blade face and the blade back at the boundary was the inner wall of the duct. The upstream boundary took
trailing edge, satisfy the Kutta condition. the duct entrance, and the downstream boundary was located between
the stators and the rotors. The inner and outer domains delivered data
3. Method validation through the interaction surfaces, as shown in Fig. 10.
The computational domain and the near wall were discretized using
Due to the special application of the pump jet propulsor, there is a Cartesian and prismatic grids respectively. Considering the influence of
lack of published test data. Therefore, the more mature CFD simulation the boundary layer, the thickness of the first layer of the stator and
technology was used instead of the experimental method to verify the rotor surface was set at 0.05 mm, with the corresponding Y+ value of
pump jet proplusor performance prediction method proposed in this 60 or so. In order to control the Y+ value and the number of grids, at
paper and analyze the calculation results. least 3 layers of prismatic grids were arranged, and the ratio of thick-
ness growth was 1.2 through inner layer to the outer layer. The edge
3.1. CFD method feasibility analysis line of each part, the top surface of the rotor, and the tip clearance area
were set densely, as shown in Fig. 11. RNG k-ε turbulence model was
In order to verify the applicability of the CFD method for numerical selected to close the calculation equation, and the Multiple Reference
simulation of a propeller with a stator, the front stator duct propeller Frame (MRF) model was used to forecast the steady performance. The
with test data was used to verify the accuracy of the CFD method. diffusion phase used the central difference format, the convection phase

264
C. Wang et al. Ocean Engineering 171 (2019) 259–272

0.6 propeller, so it can be used to calculate the open water performance of

Experimental value
Tip vortex model pump jet propulsor and follow-up potential flow method validation.
0.5
Tip vortex free model
3.2. Tip vortex model feasibility analysis
0.4
KT-all

3.2.1. Front stator model

0.3 This study used the front stator pump jet propulsion model for
calculation and analysis. This model is same as the calculation model in
0.2 the CFD feasibility analysis. The cross-sectional view of the model is
shown in Fig. 13.
0.1 The model used in this paper is a pre-stator pump jet propulsor
consisting of nine stator blades and four rotor blades, each with a length
0.0 of 2 inches. A NACA63–018section shape was used for the stators. A
0.2 0.4 0.6 0.8 1.0 1.2 complete description of the geometry for both the duct and the pro-
J peller are given in Appendix A. The duct was manufactured from a
transparent plastic in order that the propeller could be observed during
(a) Thrust coefficient
the experiment. By saturating the water in the tunnel with air bubbles,
0.8 the streamlines along the duct on both the inside and the outside could
Experimental value
be observed. The force on the individual propeller and the combined
0.7 Tip vortex free model
force of the propeller, stator and duct are measured by using a re-
Tip vortex model
0.6 movable thrust bearing in the sleeve supporting the duct and the stator.
Force measurements were performed to obtain open water curves cor-
Kq-all

0.5 responding to different pitch angles on the stators. And the velocity
field behind the propulsor was also measured at two axial locations aft
0.4
of the duct using a LDV system.
The maximum rotor diameter Dr is 0.254 m, and the minimum gap
0.3
between the rotor tip and the duct is 3 mm. The mesh of the potential
0.2 flow method is shown in Fig. 14. The number of grids for a single stator
blade is 36 × 12 , and the number of grids for a single rotor blade is
0.1 36 × 8.
0.2 0.4 0.6 0.8 1.0 1.2 1.4
J 3.2.2. Open water performance of front pump jet propulsor
In order to analyze the influence of tip vortex model on the front
(b) Torque coefficient stator pump jet propulsor, the thrust coefficients of the stator, rotor,
Fig. 16. The whole thrust coefficient and torque coefficient of pump jet pro- and duct were calculated by the potential flow theory method. Fig. 15
pulsor. shows the thrust coefficient curves.
It can be seen from the calculation results that the trend of the
change of the thrust coefficient curve of the stator, rotor, and duct with
used the first-order windward format, and the calculation method of
or without the tip vortex model remained constant. This shows that
flow field used the SIMPLEC algorithm.
either using or not using the tip vortex model has little effect on the
stator, rotor, and duct calculation of the front stator pump jet propulsor.
3.1.2. Grid independence analysis Under the same advance speed coefficient, the proportion of the thrust
Following the work of Roy (Roy et al., 2013) on uncertainty re- coefficient of the rotor to the total thrust coefficient was larger,
search of Cartesian grid in viscous flow calculation, multiple sets of reaching more than 80%, indicating that the thrust of the pump jet
grids with various sizes have been established. The mesh size refine- propulsor was generated mainly by the rotor. As the advance speed
ment rate r and the total number of grids satisfy the following re- coefficient increased, the thrust coefficient of the rotor, stator, and duct
lationship: decreased, however, the thrust of the stator and the duct decreased
Nfine more, and the proportion of the thrust produced by the rotor increased
r=( )1/ d , gradually in the total thrust.
Ncoarse
In this paper, the thrust coefficient and torque coefficient of the
Where N is the total number of grids and d is the space dimension of the pump jet propulsor were compared with the experimental results, as
problem. The mesh refinement rates in all three directions of each set of shown in Fig. 16. At the same time, the thrust coefficient of the stator-
grids in the entire computing domain are consistent. In this paper, we duct system and the efficiency curve of the pump jet propulsor were
set r with the value 1.2.Three different meshes were obtained with also compared with the experimental values, as shown in Fig. 17. The
quantity of 2million,3.5million, and 6.11million, respectively. From the advance speed coefficients were 0.6, 0.8, and 1.0 respectively.
results shown in Table 1, as the number of grids increased, the errors of The KT all map for the entire pump model thrust coefficient was
the thrust coefficient and torque coefficient between the calculated KT all = Kts + Ktd + Ktr , and the K Q all map for the entire pump model
value and the test value are decreased. When the number of grids is torque coefficient was K Q all = K qs + K qr . From the calculated total
more than 3.5 million, the calculation results have high accuracy, the thrust coefficient and torque coefficient results, it can be seen that the
error of Kt-all and Kq was less than 4%. thrust and torque coefficients calculated by the two models were similar
Fig. 12 demonstrates the use of 6.11 million grids for forecasting the to the experimental coefficients, and the values of thrust and torque
full condition hydrodynamic performance results. The calculated re- were also in good agreement with the results of viscous fluid. This in-
sults of CFD are consistent with the experimental values within the dicates that the calculation method used in this paper is reliable, and
design conditions, and deviations existed only if the advance speed the potential flow theory can predict the open water characteristics of
coefficient is high. These results show that the CFD method has higher the pump jet propulsor more accurately. As can be seen from the overall
accuracy and reliability in the numerical simulation of the stator-like efficiency curve, the results of the model with or without tip vortex

265
C. Wang et al. Ocean Engineering 171 (2019) 259–272

0.20 Experiment value

Tipless vortex model
0.15 Tip vortex model

Kts+Ktd
0.10

0.05

0.00

-0.05
0.2 0.4 0.6 J 0.8 1.0 1.2

(a) Thrust coefficient curve for the stator and duct system

0.8

0.6

0.4

Experimental value
Tip vortex model
0.2
Tip vortex free model

0.0
0.2 0.4 0.6 0.8 1.0 1.2
J

(b) Efficiency curve for the stator and duct system

Fig. 17. Thrust coefficient curve and efficiency curve for the stator and duct system.

model were similar. With the increase of advance speed coefficient, the Fig. 19. The number of grids for both rotor and stator blade is 32 × 10 ,
open water efficiency of pump jet propulsor first increased and then and the number of grids for hub and duct is 56 × 33 and 60 × 99, re-
decreased, which was consistent with the trend of variation of experi- spectively.
mental values of front stator pump jet propulsor. When the advance
speed coefficient reached 0.962, the efficiency of the pump jet pro- 4.2. Pressure and circulation distribution
pulsion reached its maximum value.
Select the results of the model with or without tip vortex at J = 1.94
4. Calculation results and analysis of rear stator pump jet and J = 3.24conditions, the rotor blade pressure distribution at r/
propulsor R = 0.98 and r/R = 0.88, and the pressure distribution at the cross
section of the duct at = 20° are obtained. The results were compared
The CFD method and the potential flow method were both used to with the CFD simulation results based on viscous flow, as shown in
predict the performance of Rear stator pump jet propulsor in this paper. Fig. 20 and Fig. 21. In the figures, s represents the chordwise distance
from the center of the element to the leading edge.
4.1. Rear stator model As can be seen from the rotor pressure distribution in Fig. 20, under
the same advance speed coefficient conditions, the influence of the tip
The model contains 11 rotor blades and 9 stator blades, rotor vortex model on the rotor decreased rapidly with the decrease of the
maximum diameter Dr is 0.238 m, spinning speed n is 2000 rpm, and radius of the blade profile, which has the greatest effect on the blade
the minimum clearance between the rotor tip and the duct is 3 mm. The tip. It can be seen from the figure that when the interference of tip
total number of grids calculated by the CFD method is 9.44 million and vortex was added, the change trend of tip pressure was more in line
the number of rotating domain grids is 4.17 million. The surface mesh is with the trend of viscous flow method. The effect of the tip vortex
shown in Fig. 18 and the mesh of the potential flow method is shown in model is more obvious at low advance speed. For example, in the case

266
C. Wang et al. Ocean Engineering 171 (2019) 259–272

Fig. 18. The grid of CFD calculation model.

where the advance speed coefficient is equal to 1.94, the leaf profile at
the radius of 0.98 is selected, and the pressure distribution of the chord
direction is analyzed. The calculation result of the non-tip vortex model
decreases at the pressure coefficient at the trailing edge, which is in-
consistent with the CFD calculation result, and the calculation results of
the tip vortex model at the trailing edge are more consistent with the
CFD calculation results. This indicates that the tip vortex model is
added to make the calculation result develop in a benign direction.
However, both tip vortex treatments obtained smaller tip loads than
CFD results. This is because, in actual viscous fluids, fluid passing
through the rotor gap area is affected by the boundary layer of the rotor
tip face and the inner wall of the duct, so the tip velocity difference
between the suction surface and the pressure surface is larger, but in the
potential flow, the influence of the gap flow is not considered. In ad-
dition, through the comparison of different advance speed coefficients,
we can find that the smaller the flow rate, the more obvious the in-
fluence of the tip vortex model. This is because the flow rate decreased, Fig. 19. The mesh distribution of potential flow calculation model.
so the velocity potential jump of the connecting surface between rotor
blade and top surface was more severe, and the dipole strength of tip decrease and the pressure on the inner wall surface of the duct begins to
vortex element was increased. Combining the equivalent relationship rise again. It can be seen from the calculation results with or without tip
between dipole and vortex, the vortex strength of tip vortex was also vortex model that the introduction of the tip vortex model changes the
increased. In a physical meaning, when the local speed decreases, the calculation results of the gap between the duct and the rotor tip, while
Reynolds number also decreases and the influence of fluid viscosity the calculation results of the tip free vortex model have not changed. It
turns into remarkable gradually. is indicated that the tip vortex model has an impact on the calculation
Fig. 21 shows the pressure distribution of the duct. As seen in the results of the pump jet propulsor. However, since the influence of
figure, the tip vortex model only influenced the inner surface of the duct viscosity is not considered, the calculation result of the tip vortex model
above the rotor tip and had little effect on the overall pressure dis- is different from the change trend of the CFD calculation result.
tribution of the duct. Additionally, when the flow rate was large, the Fig. 22 is the rotor circular distribution. When the radial distribu-
potential flow results were basically consistent with the CFD results. tion of the circular is known in the model without tip vortex, the
However, when the advance speed coefficient was low, due to the non- amount of the circular at the blade tip tends to be zero, which does not
viscous assumption of the fluid, for the duct pressure distribution in the meet the actual situation of the flat top blade. The introduction of the
gap region, the results of potential flow method and viscous flow tip vortex model corrects the radial distribution of the circular, and the
method were quite different. For example, in the case of an advance increase of the circular will increase the rotor thrust and torque. From
speed coefficient of 1.94, according to the normal flow phenomenon, the chordwise distribution of the circular, it can be seen that the rotor
when fluid flows between the top end of the rotor and the inner wall leading edge load was very small, and when the chordwise distance
surface of the duct, the flow rate of the fluid increases and the pressure increased, the load gradually increased. At the same time, the tip vortex
decreases; and when the fluid flows through the gap region between the intensity also gradually increased, and when the advance speed was
rotor and the inner wall surface of the duct, the flow rate begins to low, the growth trend is faster. This also verifies that the smaller the

267
C. Wang et al. Ocean Engineering 171 (2019) 259–272

1.0
CFD
0.8
Tip vortex free model
0.6 Tip vortex model
0.4

CP
0.2

0.0

-0.2

-0.4

-0.6

-0.8 J=1.94

-1.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
S/C

1.6

1.4 CFD
Tip vortex model
1.2
Tip vortex free model
Fig. 19. (continued) 1.0

CP
0.8
flow rate, the more obvious the effect of the tip vortex model.
0.6
J=3.24
0.4
4.3. Open water performance prediction results
0.2
This study considered five conditions, in which the advance speed
0.0
coefficients were 1.29, 1.78, 2.27, 2.75 and 3.24. The pump jet pro-
pulsor stator, rotor, and overall performance calculations were com- -0.2
0.0 0.2 0.4 0.6 0.8 1.0
pared with the CFD calculations shown in Fig. 23 and Fig. 24. S/C
As we can see from Figs. 23 and 24, the thrust and torque coeffi-
cients of the no tip vortex model were quite different from the CFD (a) r/R = 0.98
results, especially when the flow rate was smaller, the difference is 1.0
more obvious. The introduction of the tip vortex model improved the
0.8 CFD
prediction accuracy of the stator, the rotor and the overall performance. Tip vortex model
0.6
When the advance speed coefficient is low, the tip vortex model im- Tip vortex free model
proves the result more obviously, this also proves when the thrusters 0.4

are under heavy load, rotor tip there is more intense leakage vortex, 0.2
and the tip vortex model can accurately simulate the perturbation of the
CP

0.0
leaking vortex to the surrounding flow field. It can be seen from the -0.2
study, tip vortex model only affects the contact element greatly, both -0.4
the rotor and the stator are connected or intersect with the tip vortex, so
-0.6
obvious changes in hydrodynamic performance, however, the duct
-0.8
elements does not come into contact with the tip vortex model, so the J=1.94
change in duct thrust is small. From the overall efficiency graph, we can -1.0

see that the results of the model with or without the tip vortex were -1.2
0.0 0.2 0.4 0.6 0.8 1.0
similar. The trend of overall efficiency mainly depended on the effi- S/C
ciency curve of the rotor. However, from the rotor results, the tip vortex 1.6
model has less effect on the rotor efficiency. In addition, at high ad-
vance speeds, the calculation of rotor thrust coefficient and efficiency is 1.4 CFD
Tip vortex model
high, this is because the tip vortex and wake vortex are all linear 1.2
Tip vortex free model
structures, simplifies the real flow of leaking vortex.
1.0
CP

0.8
5. Conclusions
0.6 J=3.24
Based on the potential flow theory, this paper presents a method of
0.4
predicting the steady hydrodynamic performance of a pump jet pro-
pulsor based on the panel method. For the first time, a method of mesh 0.2
based on a circular truncated cone and a tip leakage vortex model 0.0
suitable for flat-topped blades is given.
The rotor tip was processed in different ways, and the calculation -0.2
0.0 0.2 0.4 0.6 0.8 1.0
results of the pressure distribution, circular distribution, and the per- S/C
formance indexes were analyzed. The following conclusions were ob-
tained. (b) r/R = 0.88
(caption on next page)
268
C. Wang et al. Ocean Engineering 171 (2019) 259–272

Fig. 20. Pressure distribution of the rotor blade at (a) r/ R= 0.98 and (b). 0.045

r/ R= 0.88. 0.040
J=1.94
0.035

Tip vortex model 0.030

/D/V 0
1.0
Tip vortex free model 0.025
CFD 0.020 J=3.24
0.5
0.015

0.010
CP

Tip vortex free model

0.0
0.005 Tip vortex model
0.000
0.5 0.6 0.7 0.8 0.9 1.0
-0.5 r/R

(a) Radial circular distribution

-1.0 J=1.94
0.035
J=1.94
-1.5 0.030 J=3.24
0.0 0.2 0.4 0.6 0.8 1.0
S/C 0.025

/D/V 0
0.020
1.5
Tip vortex model
0.015
1.0 Tip vortex free model
CFD 0.010

0.5 0.005

0.000
0.0
CP

-0.005
0.0 0.2 0.4 0.6 0.8 1.0
-0.5 S/C

-1.0 (b)Chordwise circular distribution

J=3.24 Fig. 22. Rotor radial circular distribution and chordwise circular distribution.
-1.5

-2.0 improvement on the calculation results.

0.0 0.2 0.4 0.6 0.8 1.0 There are many deficiencies and unresolved problems in this study.
S/C For example, the boundary layer flow is the predominant flow pattern
Fig. 21. Pressure distribution in longitudinal section of duct ( ath°). in the narrow gap area of the pump. If we do not consider the influence
of gap flow on the surrounding flow field, error will occur in the rotor
(1) There is a strong tip vortex in the pump jet propulsor internal flow tip load forecast results. The authors attempted to improve the pressure
field, affect the rotor tip and the flow field around the downstream distribution at the tip using the gap flow model proposed by Mchugh
stator, the tip vortex model can simulate the perturbation effect of (McHugh, 1997).However, when both the gap flow model and the tip
the tip leakage vortex on the surrounding flow field, thereby im- leakage vortex model exist simultaneously, the vertex element of the
proving the forecast accuracy of the rotor and stator thrust and trailing edge has to meet two Kutta conditions at the same time, leading
torque performance indicators. to the calculation processes not being converged properly. Therefore,
(2) Verification of the accuracy of the calculation results shows the the question of how to consider the influence of the gap flow and the
effectiveness of the method proposed in this paper for predicting non-linear processing of the leakage vortex model will be addressed in
pump jet propulsor hydrodynamic performance is demonstrated. follow-up research work in order to further improve the method of
(3) The introduction of the tip vortex model has little effect on the forecasting the hydrodynamic performance of the pump jet propulsor so
hydrodynamic performance of the front stator pump. For the rear as to improve the prediction accuracy.
stator pump jet propulsor, the introduction of the tip vortex model
makes the pressure distribution at the tip of the rotor closer to the
result of viscous flow. At the same time, the overall distribution of Acknowledgment
the blade circular is closer to the actual condition, improving the
prediction accuracy of the stator, the rotor and the overall perfor- The research was financially supported by the National Natural
mance. Science Foundation of China (Grant NO.51679052) and the Defense
(4) At low advance speed coefficients, there is a intense leakage vortex Industrial Technology Development Program (Grant no.
in rotor tip, and the introduction of tip vortex model has obvious JCKY2016604B001).

269
C. Wang et al. Ocean Engineering 171 (2019) 259–272

1.2 1.2
Tip vortex free model Tip vortex free model
CFD CFD
Tip vortex model 1.0
1.0 Tip vortex model
r 0.8
KTr,K Qr, r

KT-all,KQ-all
0.8
0.6
K Tr K T- all
0.6 0.4

KQ r 0.2
0.4 K Q - all
0.0

0.2
1.0 1.5 2.0 2.5 3.0 3.5 -0.2
J 1.0 1.5 2.0
J
2.5 3.0 3.5

(a) Rotor (a) Thrust coefficient and torque coefficient curves

0.70
0.4 KQ s
0.65

0.3
KTS,KQS

0.60

0.2
0.55
K Ts

0.1 Tipless vortex model

0.50 CFD
Tip vortex free model
CFD Tip vortex model
0.0
Tip vortex model 0.45

1.0 1.5 2.0 2.5 3.0 3.5 1.0 1.5 2.0 2.5 3.0 3.5
J J

(b) Stator (b) Efficiency curves

Fig. 24. (a)Thrust coefficient and torque coefficient curves and (b) efficiency
curves.
0.1

0.0

-0.1

-0.2
Kd

-0.3

-0.4
CFD Tip vortex free model
Tip vortex model
-0.5

-0.6
1.0 1.5 2.0 2.5 3.0 3.5
J

(c) Duct
Fig. 23. Performance prediction results of rotor, stator and duct.

Appendix A. A geometry of the propeller and duct

Description of propeller geometry.

Modified mit propeller #76.
Number of Blades = 4.
Airfoil Sections.
Diameter = 10.0 inches.

270
C. Wang et al. Ocean Engineering 171 (2019) 259–272

R/R0 P/D RAKE SKEW C/D F/C T/D

1.0.3483 2.1.2283 3.0.0 4.3.0 5.0.3024 6.0.0348 7.0.0429

8.0. 4644 9.1.2417 10.0.0 11.1.6 12.0.3193 13.0.0309 14.0.0371
15.0. 6805 16.1.2264 17.0.0 18.2.7 19.0.3359 20.0.0267 21.0.0304
22.0.6966 23.1.1881 24.0.0 25.3.7 26.0.3406 27.0 .0268 28.0 .0247
29.0.8127 30.1.1795 31.0.0 32.6.0 33.0.3346 34.0.0220 35.0.0190
36.0.9288 37.1.1788 38.0.0 39.9.0 40.0.30490 41.0.0202 42.0.0140

Offsets for the duct.

Note:Origin located at the duct trailing.
Edge On the propeller axis.
All dimensions are in inches.

Duct inner surface Duct outer surface

X R X R

7.575 5.902 7.575 5.905

7.551 5.816 7.539 5.989
7.466 5.735 7.443 6.064
7.321 5.667 7.289 6.122
7.119 5.606 7.080 6.168
6.864 5.548 6.818 6.204
6.560 5.488 6.508 6.235
6.211 5.426 6.153 6.257
6.824 5.365 5.761 6.270
5.404 5.305 5.337 6.273
4.958 5.253 4.887 6.268
4.492 5.209 4.419 6.253
4.015 5.172 3.942 6.222
3.532 5.142 3.461 6.172
3.053 5.125 2.986 6.106
2.585 5.125 2.523 6.026
2.136 5.125 2.079 5.937
1.712 5.136 1.661 5.843
1.318 5.171 1.277 5.750
0.962 5.213 0.930 5.660
0.651 5.254 0.627 5.577
0.389 5.292 0.373 5.505
0.180 5.324 0.171 5.446
0.029 5.347 0.025 5.403
0.000 5.352 0.000 5.395

References Yokohama National Uni.

Suryanarayana, C., Satyanarayana, B., Ramji, K., et al., 2010a. Experimental evaluation of
pumpjet propulsor for an axisymmetric body in wind tunnel. Int. J. Naval Architect.
Feder, D.F., Dhone, M., Kornev, N., et al., 2018. Comparison of different approaches Ocean Eng. 2 (1), 24–33.
tracking a wing-tip vortex. Ocean Eng. 147, 659–675. Suryanarayana, C., Satyanarayana, B., Ramji, K., et al., 2010b. Cavitation studies on
Fu, Jian, Song, Zhenhai, Wang, Yongsheng, et al., 2016. Numerical prediction of hy- axisymmetric underwater body with pumpjet propulsor in cavitation tunnel. Int. J.
drodynamic noise of pump propeller. Chin. J. Ship Mech. 20 (5), 613–619. Naval Architect. Ocean Eng. 2 (4), 185–194.
Guo, Q., Zhou, L., Wang, Z., et al., 2018. Numerical simulation for the tip leakage vortex Wang, G.Q., Liu, X.L., 2007. A potential based panel method for prediction of steady and
cavitation. Ocean Eng. 151, 71–81. unsteady performances of ducted propeller with stators. J. Ship Mech. 11 (3),
Hughes, M.J., Kinnas, S.A., 1991a. An Analysis Method For A Ducted Propeller With Pre- 333–340.
Swirl Stator Blades. Stators. Wang, C., Yang, C., 1999. Hydrodynamic performance prediction of ducted propeller with
Hughes, M.J., Kinnas, S.A., 1991b. An Analysis Method For A Ducted Propeller With Pre- stators. J. Ship Mech. 3 (3), 1–7.
Swirl Stator Blades. Stators. Wang, Tao, Zhou, Liandai, 2004. Numerical Simulation and Mechanism Study of the
Kawakita, C., 1992. Hydrodynamic Analysis of a Ducted Propeller in Steady Flow Using a Interaction between the Gap Flow and the Main Flow in a Pumped Propeller [C]//
Surface Panel Method. The West-Japan Society of Naval Architects, pp. 11–22. Marine Hydrodynamics Conference.
Liu, X.L., Wang, G.Q., 2006. A potential based panel method for prediction of steady Yao, Shi, Pan, Guang, Wang, Peng, et al., 2014. Numerical analysis of cavitation char-
performance of ducted propeller. J. Ship Mech. 10 (3), 26–35. acteristics of pump propeller. J. Shanghai Jiao Tong Univ. (Sci.) 48 (8), 1059–1064.
Lu, L., Pan, G., Sahoo, P.K., 2016a. CFD prediction and simulation of a pumpjet propulsor. You, D., Wang, M., Moin, P., et al., 2007. Large-eddy simulation analysis of mechanisms
Int. J. Naval Architect.Ocean Eng. 8 (1), 110–116. for viscous losses in a turbomachinery tip-clearance flow. J. Fluid Mech. 586 (586),
Lu, L., Pan, G., Wei, J., et al., 2016b. Numerical simulation of tip clearance impact on a 177–204.
pumpjet propulsor. Int. J. Naval Architect. Ocean Eng. 8 (3), 219–227. Zhang, D., Shi, W., Esch, B.P.M.V., et al., 2015. Numerical and experimental investigation
Lu, L., Gao, Y., Li, Q., et al., 2017. Numerical investigations of tip clearance flow char- of tip leakage vortex trajectory and dynamics in an axial flow pump ☆. Comput.
acteristics of a pump jet propulsor. Int. J. Naval Architect. Ocean Eng. 10 (3), Fluids 112 (1), 61–71.
307–317.
McHugh, Paul G., 1997. Advances in Ducted Propulsor Analysis Using Vortex-lattice
Lifting-surface Techniques. Massachusetts Institute of Technology. List of symbols
Pan, Guang, Hu, Bin, Wang, Peng, et al., 2013. Numerical simulation of steady hydro-
dynamic performance of pump propeller. J. Shanghai Jiao Tong Univ. (Sci.) 47 (6), V0 : Velocity
932–937. S : Boundary of the fluid
Roy, C., Heintzelman, C., Roberts, S., 2013. Estimation of Numerical Error for 3D Inviscid Sb : Object surface
Flows on Cartesian Grids. Sw : Wake vortex surface
Su, Y., 1999. A Study on Design of Marine Propellers by Lifting-body Theory. Ph D Thesis Q : Arbitrary control point on the boundary surface

271
C. Wang et al. Ocean Engineering 171 (2019) 259–272

RPQ : Straight line distance between two points P and Q : Efficiency of the pump jet propulsor
(Q)/ nQ : Normal derivative of the point velocity potential Dr : Maximum diameter of the rotor
E : Green formula parameter J : Advance speed coefficient
Q1 : Point on the wake vortex surface n : Rotor speed
n : Unit normal vector of the corresponding object surface and points to the flow field r l : rt Radial positions
: Velocity potential jump through the wake vortex surface x l : xt Axial positions
Ss : Stator surface : The conversion of circular table conical angle
Sd : Duct surface x 0 : Vertex position
Sr : Rotor surface c : The jth leaf section chord length
Sh : Hub surface cl : The leading edge to the generatrix chord distance
Sws : Vortices of the stator surface xr : Trim value
Swd : Vortices of the duct surface : Side rake angle
Swr : Vortices of the hub surface dOP : Distance between the arbitrary point P of the leaf section and the vertex O of the
: Rotational angular velocity of the propeller cone on the plane of the circular table
k : The kth iterative operation dPB1: Point P in the circular mesa to yz plane arc distance
Vsd, rh : Induced velocity generated by the rotor-hub system at the stator and duct surfaces 0 : The initial pitch angle of each vortex line
Vrh, sd : Induced velocity generated by the rotor-hub system in the opposite direction of the : Pitch angle
stator and duct surfaces g : The geometrical pitch angle of the rotor tip section
KT : Thrust coefficient xT : T be the point of the axial coordinate
K Q : Torque coefficient numT : The number of cells from the start point to the T point
KT all : Total thrust coefficient

272