Documente Academic
Documente Profesional
Documente Cultură
Energy
journal homepage: www.elsevier.com/locate/energy
a r t i c l e i n f o a b s t r a c t
Article history: A computational hydrodynamics method was formulated and implemented for horizontal axis tidal
Received 27 November 2009 turbines. This paper presents a comparative analysis between screw propellers and horizontal axis
Received in revised form turbines, in terms of geometry and motion parameters, inow velocity analysis and the implementation
8 March 2010
methodologies. Comparison and analysis are given for a marine propeller model and a horizontal axis
Accepted 9 March 2010
turbine model that have experimental measurements available in literature. Analysis and comparison are
Available online 22 April 2010
presented in terms of thrust coefcients, shaft torque/power coefcients, blade surface pressure distri-
butions, and downstream velocity proles. The effect of number of blades from 2 to 5, of a tidal turbine
Keywords:
Panel method
on hydrodynamic efciency is also obtained and presented. The key implementation techniques and
Propeller methodologies are provided in detail for the propeller based panel method tool to migrate as a prediction
Wind turbine tool for tidal turbine. While the method has been proven to be accurate and robust for many propellers
Tidal turbine hydrodynamics tested in the past, this numerical tool could be validated further for turbines. To further rene and
validate the panel method for various turbines, it requires substantial additional experimental
measurements. These measurements include downstream velocity prole by using LDV and/or SPIV,
which are essential for numerical wake vortices descritization.
Crown Copyright 2010 Published by Elsevier Ltd. All rights reserved.
1. Introduction Hoogedoorn et al. [4]. To the authors knowledge, these few panel
methods for turbine simulation and prediction include a 2D panel
Computational methods have been widely used for wind or tidal method by Drela et al. mentioned in [1], a 3D time-domain panel
turbine research and development. A comprehensive review of method for wind turbine by Hampsey [5], a rudderepropeller
these methods and their merits and limitations, for example, was interaction panel code by Turnock [6] and a design based simula-
given by Nicholls-Lee et al. [1]. Among these methods, the panel tion work by Greco et al. [7]. Formulation and implementation of
methods, in the most advanced and complicated method group, these panel methods for both wind turbines and tidal turbines are
have both high computing efciency and prediction accuracy as an basically the same though different uid properties such as
engineering tool for turbine simulation and design optimization. viscosity/density and hence Reynolds number, require different
While probably the blade element methods (BEM) are the most scheme and numerical treatment. The primary difference for panel
widely used as preliminary simulation tools for wind and tidal methods as tools for wind turbine and for tidal turbine is that tidal
turbines, much fewer panel methods could be found for turbine in turbines are often exposed to cavitation that would cut off the
literature. One of the many examples of BEM methods is Kishinami negative pressure spike and reduce the energy extraction efciency
et al. [2] that used a combination of momentum, energy and blade substantially even if there does not exist stall or separation. For
element theory and produced some results which agree relatively wind turbine under high speed inow conditions, compressibility
well with available experimental measurements. Another example might be a problem, at least in terms of turbine efciency correc-
of the BEM method is the one by Lanzafame and Messina [3] that tion. A bare panel method developed from scratch also needs many
produced some hydrodynamic characteristics which agree closely other essential numerical components for different application
with the model test data. Other combined methods also have been cases. Establishing these numerical components for a newly
used to simulate exible turbine blades, for example, by developed bare panel method require substantial effort and
implementation development, as described later in this section. In
this paper we provide an efcient turn-key solution that a well-
* Tel.: 1 709 772 4575; fax: 1 709 772 2462. established propeller panel method can be quickly turned into
E-mail address: pengfei.liu@nrc-cnrc.gc.ca a turbine simulation code.
0360-5442/$ e see front matter Crown Copyright 2010 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2010.03.013
2844 P. Liu / Energy 35 (2010) 2843e2851
Panel methods, are called the boundary element methods, or interacting with an ice body at variable proximity (transient
boundary integral methods (BEM in short as well). Lifting surface hydrodynamic response of a propeller moving towards an ice
and panel methods have been widely used in research and devel- sheet) [21].
opment of aircraft wings, hydrofoils and both aerial and marine In the past decades, many panel methods have been developed
propellers. Zero thickness propeller blade simulated and computed and these panel methods along with their associated numerical
by lifting surface theory in the computational uid dynamics (CFD) schemes and techniques for propellers are well established. These
eld has a history of over 60 years. The use of the surface panel existing panel methods with a minor or moderate revision, could be
method for a simple body surface mesh can be traced back to the quickly used for tidal and wind turbine prediction. The aim of this
early 1960s. Hess and Vararezo [8] probably made the rst panel work is to present the physics and numerical similarities and
method computation for propellers. To deal with complete aircraft difference between propeller and turbine panel methods and then
geometry, panel method codes, PMARC (Panel Method Ames provide detailed implementation techniques and treatments.
Research Center) developed by Ashby et al. [9] and VSAERO by
Maskew [10] are the early examples of panel methods for aircraft 2. Method
wings and propellers. On the other hand, panel methods have also
been used for marine propeller research development and early The current panel method is a multiple-body interaction panel
examples among those are publications by Greeley and Kerwin [11] method. The fundamentals of panel method have been presented in
and Hoshino [12], just to name a few. A time-domain unsteady detail in some textbooks, including the ones by Moran [22] for 2D
panel method code OSFBEM (oscillating foil boundary element foil sections and by Katz and Plotkin [23] for unsteady 3D body and
method) was developed by the author for oscillating propulsors of wings. A detailed formulation and implementation for a low-order,
both chordwise and spanwise exibility to simulate marine time-domain panel method, were given by the author [13].
animals propulsion [13]. To respond the need in simulation of
uid-structure interactive hydrodynamics to predict ice blockage 2.1. Flow physics similarity and differences between propeller and
effects between ice sheet and ice-class propellers, a panel method tidal turbine
code, PROPELLA [14] was developed in 1996, based on OSFBEM.
Since then, continued efforts were made to maintain and enhance Being rotary wings, ow around both propeller and turbine
the capability of the code. The capability of unsteady oblique ow blade sections has both similarities and differences. Flow condi-
and inow wake were presented in early 1998 [15]. Automatic body tions around a wing section determine the hydrodynamic charac-
surface generation for a propeller of arbitrary number of blades, teristics of the wing section. These characteristics are mainly
nozzle, rudder, ice blockage etc., was presented in 2001, along with determined by the effective angle of attack of the blade section. The
velocity prole downstream prediction and wake vortices roll-up similarity and differences of the ow around a blade section
enhancement [16]. Cavitation predictive capability via an empirical between propeller and turbine are shown in the velocity diagrams
formulation was established for PROPELLA and presented in 2001 in Figs. 1 and 2, respectively.
[17]. In the meantime, a pre- and postprocessor were developed for In Fig. 1, blade geometric pitch angle of attack is represented as
the code by using OpenGL and Visual C of Microsoft Foundation aP, and the angle of zero lift of the blade section is a0 due to the
Class, as a 3D unsteady data visualization tool [18] to view the camber of the blade section. Therefore the effective geometric
geometry motion and colour blended results. A novel and robust angle of pitch is aP a0. The inow velocity angle of attack, also
numerical Kutta condition using Broydens iteration was developed called hydrodynamic angle of attack in the literature, is aV tan1
and presented in 2002 [19]. This panel method was used to predict (Va/2prn), where Va is propeller shaft forward velocity, also called
the hydrodynamic characteristics of the wing-in-ground effect for inow velocity in some literature, and r is the blade section local
a dual-foil counter-phase propulsor [20]. Since 2003, this propeller radius and n is the shaft rotation speed in revolutions per second.
panel method has been redeveloped with a multiple-body inter- However, even in open water, as a propeller is rotating, the inow
action formulation to deal with a propeller with pod and strut, velocity to the propeller plane is not unidirectional because of the
called a podded propeller unit and recently for a podded propulsor induced tangential velocity Vt and radial velocity Vr. In fact, these
P. Liu / Energy 35 (2010) 2843e2851 2845
now interchanged, i.e., the suction side and pressure side of the
blade section are swapped for a turbine. The effective angle of
attack now becomes ae aV aP a0 when the induced veloci-
Fig. 1. Propeller blade section velocity schematic diagram.
ties are taken into account and a aV aP when the induced
velocities and angle of zero lift are both neglected. It is also obvious
induced velocities are not in the same plane as shown in the gure. that the effective angle of attack must be positive for the blade
The total inow velocity relative to the local blade section in 3D section to be in turbine mode. Now using the propeller code PRO-
! ! !
space is then V a V t V r with the resultant velocity angle of PELLA for turbine prediction is reduced to two key modications
attack aV . Therefore, the effective angle of attack of the blade
when preparing for input, that is, the blade sectional coordinates
section, that determines the hydrodynamic characteristic of a blade interchange and the formulation and implementation of the angle
section, is ae aP a0 aV for which induced velocities have of shed wake vortices. Using the same motion and geometric
been taken into account and is a aP aV without including the variables for propellers, the angle of shed wake vortices for
induced velocities and angle of zero lift. The back side of a propeller a turbine is taken as a at local radius of 0.7 R. In the following non-
blade is often referred to suction side and the face side is referred to dimensional analysis, the wake pitch angle of the turbine was
the pressure side, when operating in the propulsion mode at the obtained as follows:
rst quadrant (positive rotational speed and positive forward
p
speed). It is obvious that for a propeller to produce positive thrust, i. Va np JnD D
a tan1 tan1 tan1 tan1 D
e., to be in propulsion mode, the effective angle of attack must be 2prn 2prn 2prn 2pr
positive. In the current version of the code PROPELLA, the angle of J 1 pD
tan1 r tan
shed wake vortices at the blade trailing edge is taken as a at pR pRr
r 0.7 R, where R is the radius of the propeller, when the angle of
(1)
zero lift is neglected. When shed wake vortices roll-up is taken into
account, the angle of shed wake vortices is modied by the induced where J is the advance coefcient, J Va/nD, of turbine and PD is the
velocity during a wake vortex relaxation procedure. While the pitch local non-dimensional pitch, pD p/D, based on the diameter of the
angle of shed wake vortices has substantial effect on the accuracy of turbine. The same as propeller simulation, turbine is moving in
thrust and torque predictions, it is even more important to the a quiescent uid; the pitch angle is taken for wake vortices
accuracy of the eld velocity prediction downstream of a propeller. formulation as pWD J pD, where PD is the pitch to diameter ratio
In addition, to avail of a multiple-body computational capability, all at 0.7 R. The angle of shed wake vortices for a propeller by using the
bodies in the uid domain move individually in a quiescent uid. similar formulation gives good thrust and torque prediction and
A similar blade section velocity diagram for a turbine is shown induced velocity estimation downstream. However, for turbine
in Fig. 2. With all the same variables as those for a propeller, the mode after extensive numerical test runs, the angle of shed wake
offset coordinates of the surfaces of the propeller blade section are vortices with an additional angle of inow taken as pWD 2J pD
2846 P. Liu / Energy 35 (2010) 2843e2851
gave better results. Hoshino [12] recommended a wake vortex Therefore, for a desired TSR a value of corresponding advance
descritization formulation based on LDV measurements down- coefcient J can be obtained for a propeller code. When the results
stream of a propeller, including wake pitch, wake contraction, Kt and Kq are obtained from a propeller code, they can be inter-
ultimate wake region, and transition. This formulation is valid only polated as Ct and Cpow, respectively, for turbine as a function of non-
for propulsion mode, not turbine mode. For turbine, if this wake dimensional speed TSR.
vortex descritization is used, wake contraction becomes
a substantial ination (800 times). This occurred at a low tip speed 3. Results and discussion
ratio of about TSR 4.0 corresponding to J 0.7855, with a ratio of
J/[p/D] at about 3, in the tidal turbine example (see the results and 3.1. Geometry and motion parameters
discussion section below).
Fig. 3 shows the discretized wake vortices behind a 3-blade Predictions for turbine hydrodynamic characteristics were
P4119 propeller [24] at J 0.8330 and a 20 root pitch tidal turbine obtained for a tidal turbine base model plus two root pitch offsets.
at TSR 7.0 (J 0.4488). In this paper, the tidal turbine refers to the For these models, experimental measurements are available for
tidal turbine model by Bahaj et al. [25]. For presentation purposes, comparison. Table 1 shows, as an example, a list of coordinates for
wake vortices in Fig. 3 are shown only for one blade. a foil section with 18% thickness and a section prole shape of the
NACA 63-8xx [26]. The turbine blades sectional maximum thick-
2.2. Parametric interpolation between propeller and tidal turbine ness varies from 24% at the root section to 15.6% at the tip section.
A model propeller P4119 by David Taylor Model Basin (DTMB)
The geometric parameters between a propeller blade and a tidal along with its results is used for comparison [24,27]. A surface
turbine are the same. For example, the expanded area ratio EAR is panel mesh arrangement for the propeller and the tidal turbine
the ratio of the total blade area to the area of the propeller disk. model [15] are shown in Fig. 4.
Therefore the solidity of a turbine rotor is equivalent to the EAR of To obtain a range of tip speed ratio TSR from 4 to 10, corre-
a propeller. For a tidal turbine calculation using a propeller code sponding input values of J and required shaft revolution speed N in
such as PROPELLA, the only geometry manipulation needed to rpm for a 0.8-m diameter tidal turbine model were obtained. These
prepare for code input, is to interchange the blade upper side with J values were calculated based on a constant inow speed of 1.5 m/s
the lower side, i.e., to swap the suction side and the pressure side. to simulate the actual cavitation tunnel test conditions by Bahaj
However, to extrapolate the results in propeller format corre- et al. [25] and are listed in Table 2.
sponding to a turbine convention, three major variables need to be Table 3 shows the sectional effective angle of attack (AOA) at
correlated. They are advance coefcient J versus tip speed ratio TSR, shaft speed of n 4 rps corresponding to a TSR of 6.7021. The values
propeller thrust coefcient Kt versus turbine thrust coefcient Ct,
and propeller torque coefcient Kq versus power coefcient Cpow.
With the denitions of propeller advance coefcient, J Va/nD, Table 1
Blade section coordinates of NACA 63-818.
propeller thrust coefcient Kt Tthrust/rn2D4, propeller shaft torque
coefcient Kq Qtorque/rn2D5, turbine tip speed ratio, TSR uR/Va, Xup/c % Yup/c % Xlow/c % Ylow/c %
turbine thrust coefcient Ct Tthrust =1=2rVa2 A, and turbine 0.0000 0.0000 0.0000 0.0000
power coefcient Cpow Qtorque uR=1=2rVa3 A, the three parame- 0.1180 0.3830 0.0971 0.4639
0.6380 0.7660 0.1943 0.9278
ters for tidal turbine in terms of propeller, are then
1.4525 1.4020 0.3885 1.9963
2.5572 1.9886 1.2782 3.1285
uR 2pnR pnD p 4.3138
TSR ; (2) 4.0489 2.5317 2.5561
Va Va Va J 5.8763 3.0233 4.2510 5.5421
8.0182 3.4889 6.3561 6.7491
10.4830 3.8984 8.8287 7.9177
Tthrust Kt rn2 D4 8Kt 13.2733 4.2623 11.6263 9.0046
Ct ; (3)
1=2 rVa2 A 1=2 rVa2 pR2 pJ 2 16.3429 4.5514 14.7503 10.0473
19.6417 4.8009 18.2008 10.9646
23.2062 4.9687 21.8870 11.7551
and 26.9727 5.0470 25.8143 12.4000
30.9059 5.0316 29.9560 12.8741
Qtorque uR K rn2 D5 2pnR 16Kq n3 D3 16Kq 34.9885 4.9166 34.2642 13.1506
Cpow q 3 : (4) 39.1988 4.6685 38.6933 13.1997
1=2 rVa3 A 1=2 rVa3 pR2 Va3 J
43.5016 4.2981 43.2089 13.0304
47.8647 3.8230 47.7721 12.6568
52.2564 3.2669 52.3428 12.1006
Performance curves for the 15 pitch turbine (0 offset) 56.6441 2.6568 56.8814 11.3890
60.9941 2.0207 61.3497 10.5519
Ct experiment Ct numerical
65.2710 1.3861 65.7122 9.6201
Cpow experiment Cpow numerical 69.4376 0.7816 69.9365 8.6241
Thrust and power coefficients
Ct experiment Ct numerical
Table 2 0.8
Cpow experiment Cpow numerical
Input turbine motion parameters in terms of J and N.
Comparison between experimental data and Pressure coefficient Cp on the pressure side of a blade
computation for P4119 propller for propeller P4119
8.0
St1L St2L St3L St4L
6.0 St5L St6L St7L St8L
10Kt_exp St9L St10L St11L St12L
Thrust and power coefficient
Cpow_exp
4.0
10Kt_PT Cpow_PT
-0.2
Pressure Coefficient Cp
2.0
-0.1
0.0 0.0
0.50 0.90 1.30 1.70 2.10 0.1
-2.0
0.2
-4.0 0.3
0.4
0.075
Advance coefficient
0.175
-8.0
0.275
St4L
0.375
0.475
0.575
0.675
Fig. 6. Comparison of predicted thrust coefcient Ct and power coefcient Cpow with St1L
0.775
0.875
0.975
the measurements [25] for the turbine model of 5 offset angle relative to the base
model.
Chord Location x/c from the LE
tidal turbine measurements on thrust and power coefcients, and Fig. 8. Comparison of predicted thrust coefcient Kt and power coefcient Cpow with
experimental data by LDV or PIV for the velocity prole down- the measurements for the DTMB P4119 propeller [24].
stream of a tidal turbine, makes it difcult for numerical method
validation. Fig. 8 shows a set of thrust and power coefcient for the
DTMB propeller by the current method PROPELLA. (positive rotational speed and positive forward speed). After
It can be seen that prediction by PROPELLA agreed very well a robust numerical Kutta condition was applied, the pressure
with the measurements. As described above, the wake pitch was difference between the suction side and the pressure side at the
taken as the value of the angle of attack at 0.7 R and this wake pitch trailing edge, as can be seen in the gure, is close to zero.
seems a proper one based on the close agreement between the Fig. 10 shows colour blended pressure distribution on the
prediction and the measurements. Further comparison will be suction side of the DTMB P4119 propeller blade (viewing from
given for the induced velocity at a location downstream of upstream). Both the propeller and turbine were modelled as right-
propeller and turbine. handed revolution propeller/rotor viewing from downstream.
Fig. 11 shows the pressure distribution on a blade of the 20 root
pitch tidal turbine at TSR 7. It is noted that, contrary to a propeller,
3.3. Pressure distribution of a propeller versus the a tidal turbine the values of the pressure coefcients on the back side of the tidal
-0.2
0.0
0.2
0.025
0.125
0.225
0.325
Blade Strip
0.425
St10U
0.525
0.625
St7U
0.725
No.
St4U
0.825
0.925
St1U
Fig. 7. Comparison of predicted thrust coefcient Ct and power coefcient Cpow with
the measurements [25] for the turbine model of 10 offset angle relative to the base Fig. 9. (a) Blade suction (upper) and (b) pressure (lower) side sectional pressure
model. distribution of the DTMB P4119 propeller at J 0.8330.
P. Liu / Energy 35 (2010) 2843e2851 2849
0.0
Pressure Coefficient Cp
0.2
0.4
St1U St2U St3U
0.6 St4U St5U St6U
St7U St8U St9U
St10U St11U St12U
0.8 St13U St14U St15U
St16U St17U St18U
1.0
Blade Strip
St16U
1.2
St13U
St10U
0.042
0.208
No.
St7U
0.375
0.542
St4U
0.708
0.875
St1U
Fig. 10. Colour blended pressure distribution on the DTMB P4119 propeller blades.
0.3
Vt/Va_PROPELLA
-1.0 Vr/Va_Exp
0.25 Vr/Va_PROPELLA
0.2
-0.5
0.15
0.0
speed
0.1
St16L
Blade Strip
0.5
St13L
No.
St7L
0
0.458
0.625
St4L
St1L
0.958
-0.05
the LE -0.15
Fig. 11. Unfolded blade section pressure distribution of the 20 root pitch tidal turbine Fig. 13. Velocity prole at a disk plane of 0.16405 times the diameter of the DTMB
at TSR 7. P4119 propeller downstream at J 0.8330, prediction versus experiment [24].
2850 P. Liu / Energy 35 (2010) 2843e2851
Circumferential averaged axial, tangential and radial velocities of the 20 Power Coefficients C of the 20 root pitch turbine with 2, 3, 4 and 5 blades at the
pow
root pitch turbine model at TSR = 7.0 at x = 0.16405D (downstream) same EAR 0f 0.0669
-0.60
3-blade 2-blade 4-blade 5-blade
Vx/Va-1_PROPELLA -0.65
Normalized Velocities Based on Shaft
Vt/Va_PROPELLA
-0.40 Vr/Va_PROPELLA
-0.55
Power coefficients
Forward speed
-0.20 -0.45
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 -0.35
0.00
-0.25 Tip speed ratio TSR
[9] Ashby DL, Dudley MR, Iguchi SK, Browne L, Katz J. Potential ow theory and [18] Liu P. Design and implementation for 3D unsteady CFD Data visualization
operation guide for the panel code PMARC. NASA TM 102851; 1991. 99 p. using object-oriented MFC with OpenGL. Journal of Japan Society of Compu-
[10] Maskew B. Program VSAERO theory document, NASA Contractor Report 4023; tational Fluid Dynamics 2002;11(3):335e45.
1986. [19] Liu P, Bose N, Colbourne B. A broyden numerical Kutta condition for an
[11]. Greeley DS, Kerwin JE. Numerical methods for propellers design and analysis unsteady panel method. International Shipbuilding Progress 2002;49
in steady ow. SNAME Transactions 1982;90:415e53. (4):263e73.
[12] Hoshino T. Hydrodynamic analysis of propellers in steady ow using a surface [20] Liu. P. Propulsive performance of a twin-rectangular-foil propulsor in
panel method. In: 2nd Report, ow eld around propeller, autumn meeting of a counter-phase oscillation. Journal of Ship Research 2005;49(3):207e14.
the society of naval architects of Japan; Nov. 1989. p. 79e92. [21] Liu P, Akinturk A, He M, Islam M, Veitch B. Hydrodynamic performance
[13] Liu P. A time-domain panel method for oscillating propulsors with both evaluation of an ice class podded propeller under ice interaction, OMAE2008-
chordwise and spanwise exibility, PhD Thesis, Memorial University of 57013. In: Proceedings of 27th international conference on offshore
Newfoundland, Canada; 1996. mechanics and arctic engineering; 2008. 9 p.
[14] Liu P. Software development on propeller geometry input processing and [22] Moran J. Introduction to theoretical and computational aerodynamics. New
panel method predictions of propulsive performance of the R-Class propeller. York: John Wiley and Sons; 1984.
In: Progress report on transport canada ice-class propeller research project [23] Katz J, Plotkin A. Low-speed aerodynamics: from wing theory to panel
(TDC Contract #T8200-6-6507-001/XSD), MMC Engineering & Research, methods. McGraw-Hill; 1991.
Newfoundland, Canada; 1996. [24] Jessup J. An experimental investigation of viscous aspects of propeller blade
[15] Liu P, Bose N. An unsteady panel method for highly skewed propellers in non- ow, PhD Thesis, The Catholic University of America; 1989. 249 p.
uniform inow. In: 22nd ITTC propulsion committee propeller RANS/Panel [25] Bahaj AS, Batten WMJ, McCann G. Power and thrust measurement of marine
method workshop, 5e6 April, Grenoble, France; 1998, p. 343e49. current turbines under various hydrodynamic ow conditions in a cavitation
[16] Liu P, Bose N, Colbourne B. Automated marine propeller geometry generation tunnel and a towing tank. Renewable Energy 2007;32:407e26.
of arbitrary congurations and a wake model for far eld momentum [26] Abbot IH, von Doenhoff AE. Theory of wing sections. Dover Publications, Inc;
prediction. International Shipbuilding Progress 2001;48(4):351e81. 1949. 693 p.
[17] Liu P, Base N, Colbourne B. Incorporation of a critical pressure scheme into [27] Gindroz B, Hoshino T, Pylkkanen JV. Proceedings of 22nd ITTC propulsion
a time-domain panel method for propeller sheet cavitation. In: International committee propeller RANS/Panel method workshop, Grenoble, France;
workshop on ship hydrodynamics, Wuhan, China; 2001. p. 1e9. 1998.