You are on page 1of 19

Renewable Energy 30 (2005) 2129–2147

www.elsevier.com/locate/renene

Computational analysis of performance and flow


investigation on wells turbine for wave
energy conversion
T.S. Dhanasekarana,b,*, M. Govardhanb
a
Kyoei Nakasuji Building, Zip: 731-0122, 3-7-18 Nakasuji Asaminami-ku,
Hiroshima-city, Hiroshima, Japan
b
Thermal Turbomachines Laboratory, Department of Mechanical Engineering,
Indian Institute of Technology Madras, Chennai 600 036, India
Received 20 September 2004; accepted 4 February 2005
Available online 7 April 2005

Abstract

Wells turbine is a self-rectifying airflow turbine capable of converting pneumatic power of the
periodically reversing air stream in oscillating water column into mechanical energy. This paper
reports the computational analysis on performance and aerodynamics of Wells turbine with the
NACA 0021 constant chord blades. Studies have been made at various flow coefficients covering the
entire range of flow coefficients over which the turbine is operable. The present computational model
can predict the performance and aerodynamics of the turbine quantitatively and qualitatively. The
model also predicted the flow coefficient at which the turbine stalls, with reasonable accuracy.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: CFD; Wells turbine; Wave energy conversion; Turbomachinery

1. Introduction

With the conventional energy resources likely to get exhausted in a few decades,
the inexhaustible sources of energy have to take their place. Alternate energy from

* Corresponding author. Tel.: C82 831 1190; fax: C82 831 1193.
E-mail address: dhanasekaran@digital-sol.co.j (T.S. Dhanasekaran).

0960-1481/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2005.02.005
2130 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

Nomenclature
ch Blade chord (m)
CA Total pressure coefficient
cm Average axial velocity (m/s)
CT Tangential force coefficient
Ct Torque coefficient
d Diameter (m)
h Hub (m)
l Blade height (m)
I Moment of inertia of the rotor (kg m2)
N rotational speed, power (rpm, W)
Ps Static pressure (N/m2)
Po Total pressure (N/m2)
R Radius ratio (r/rt)
s Blade pitch (m)
t Time (s)
t* Non-dimensional time
T Tangential force (N)
U Peripheral velocity (m/s)
V Volume flow rate (m3/s)
w Relative velocity (m/s)
Z Number of blades
a Incidence angle (deg.)
f Flow coefficient (cm/Ut)
h Efficiency
r Density of air (kg/m3)
s Blockage solidity
t Torque (N m)
u Angular velocity of rotor (1/s)
Subscripts
h Hub
m meridional direction
u Peripheral/tangential direction
t Tip

the ocean is attracting the attention of the researchers in the recent past due to its
perennial availability and minimum health hazards. Of the many possible forms of
ocean energy, wave energy is promising. Wave energy is an alternate form of energy,
which is pollutant free and in near future it is likely to be economically viable.
Countries which are surrounded by sea and possess remotely situated island
communities such as China, Japan, Portugal and India, wave energy conversion is
T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2131

an attractive proposition. The wave energy device is based on the principle of the
oscillating water column (OWC).
The OWC device is a simple rectangular box closed on all sides except for an opening
at its front, just submerged in water, to receive the sea waves. When wave crest enters the
device, water level inside the device rises and compresses the entrapped air. The air rushes
out to the atmosphere through the top opening. When a wave trough enters the device, the
above process is reversed and air rushes into the device from atmosphere. Thus the OWC
device produces periodically reversing (bi-directional) stream of air. If a duct with a
suitable turbine is placed at the top opening, the energy from the oscillating airflow can be
converted into mechanical energy. In the initial stages of development, conventional
turbines were used in conjunction with the OWC device. But these turbines required flow-
rectifying valves and guide vanes. A concept proposed by Prof. A. A. Wells in 1976
eliminated the need for flow-rectifying valves and guide vanes [1]. Hence the name Wells
turbine.
Wells turbine is a self-rectifying axial flow turbine with untwisted rotor blades of
symmetrical aerofoil cross section set radially at 908 stagger. The turbine blading, being
symmetrical with respect to a plane normal to the axis, is not sensitive to the direction of
incoming flow. The tangential force of the rotor works only in one direction although
airflow is oscillating, Fig. 1. Consequently, the turbine rotates in the same direction
without rectifying valves to rectify the oscillating flow and produces power regardless of
which way the air is flowing. There are several reports, which give information on the
performance of the Wells turbine focusing on both the starting and running
characteristics [2–9].
There are few reports on computational analysis of Wells turbine. Watterson and
Raghunathan [10] have studied the effects of solidity on Wells turbine performance,
pressure drop, torque and efficiency. Numerical analysis of performance and aerodynamic
characteristics on Wells turbine with CA9 blade profile has been carried out by Thakker
et al. [11]. The hysteretic characteristics and aerodynamic characteristics of Wells turbine,
which operates in unsteady state like sinusoidal flow condition are investigated by
Setoguchi et al. [12]. To investigate the effect of blade sweep on the performance of the
Wells turbine, the numerical investigation was carried out under steady flow condition
with a fully Navier–Stokes code for two kinds of blades, NACA 0020 and CA9 by
Kim et al. [13].
The objective of the present investigation is to carry out the numerical analysis of
performance and flow investigation on Wells turbine with NACA 0021 blade profile. The
computational studies have been made for various flow coefficients. In order to get the
appropriate inlet velocity profile to turbine, the computational domain has been generated
with inlet hub nose. The computational method employs an unstructured mesh, which
allows inclusion of such features as blade tip clearance and casing treatment.

2. Experimental set-up and programme

The experimental set-up is shown in Fig. 2. The set-up consists of a 4 kW variable


speed centrifugal blower for delivering the necessary pneumatic power to the Wells
2132 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

Fig. 1. Principles of operation of Wells turbine with guide vanes.

turbine. The blower is capable of developing an optimum static pressure rise of about
2000 N/m2 while delivering a discharge of 1.1 m3/s. The delivery duct of the
centrifugal blower is 2800 mm long and has a diameter of 250 mm. A diverging cone
of length 25 mm connects this duct with the turbine test section. Fig. 2 gives the cross
sectional assembly of the turbine. The turbine test section has an internal diameter of
265 mm and the fabricated rotor had a diameter of 263 mm, leaving a tip clearance of
1 mm. The hub diameter was selected as 163 mm providing a hub to tip ratio of
0.619. The rotor under investigation has NACA 0021 blade profile with a blockage
solidity of 0.6. As the chord is kept constant through out blade length, pitch–chord
ratio and solidity vary from hub to tip. Fig. 1 gives velocity triangles at inlet and
outlet of the turbine.
The apparatus is fully equipped with instrumentation for measuring the essential
parameters like torque, speed and pressure. Output torque measurement was carried out by
T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2133

Fig. 2. Cross sectional assembly of Wells turbine.

means of a dynamometer and the speed by digital non-contact tachometer. Measurement


of pressures was made by means of pneumatic probes. A three hole boundary layer probe
was installed at inlet 33 mm ahead of the rotor to measure the two-dimensional flow
entering into the turbine. A miniaturised five-hole probe was installed 80 mm behind the
turbine to measure the three-dimensional flow behind the rotor. The probes were mounted
on traverse mechanisms fixed to the pipe wall of the test rig. Furness Control micro-
manometer (Model FC012, Furness Controls Ltd, UK) through a 20-channel scanning box
(Model FC 011 Furness Controls Ltd, UK) with a least count of 0.01 mm of water column
were used to read the pressure data. Pressure measurements were carried out at 13 equi-
spaced radial locations. The probe data obtained was reduced using the calibration charts
to evaluate all necessary parameters, including the flow rates. More details on
experimental set-up and measurement can be found in Swaminathan [14]. Raghunathan

Fig. 3. Perspective view of computational domain showing mesh on turbine blades.


2134 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

and Ombaka [15] studied the performance of the turbine under uni-directional and
sinusoidal frequencies of flow reversal and concluded that the change in power output and
efficiency under oscillating flow conditions is within 3 and 5% of the uni-directional flow
conditions. Hence, the present investigations were carried out only in uni-directional flow.
Axial velocity coefficient

0.3
(a)
0.25 Experimental Values

0.2 Computed Values φ = 0.08


0.15
0.1
0.05
0
0.6 0.7 0.8 0.9 1
Radius ratio, R
Axial velocity coefficient

0.3
(b)
0.25 φ = 0.134
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R
Axial velocity coefficient

0.3
(c)
0.25 φ = 0.155
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R
Axial velocity coefficient

0.3
(d)
0.25
0.2

0.15
0.1
φ = 0.206
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R

Fig. 4. Comparison of inlet axial velocity coefficient at various flow coefficients.


T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2135

The experimental studies were made at three different speeds namely; 3500, 4000
and 4500 rpm, respectively. For each speed, the flow survey covered the entire range
of flow coefficients over which the turbine is operable. However, results obtained at
4500 rpm are presented in this paper to validate the computational results. The output

0.5
Axial velocity coefficient

0.45 (a)
0.4 Experimental Values φ = 0.08
0.35
0.3
0.25 Computed Values
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R

0.5
Axial velocity coefficient

0.45 (b)
0.4
0.35 φ = 0.134
0.3
0.25
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R

0.5
Axial velocity coefficient

0.45 (c)
0.4 φ = 0.155
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R

0.5
Axial velocity coefficient

0.45 (d) φ = 0.206


0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Radius ratio, R

Fig. 5. Comparison of outlet axial velocity coefficient at various flow coefficients.


2136 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

power, input power and the efficiency of the turbine are calculated in the following
manner.

"Ð r #
t
P c 2prdr
Input power to the turbine; NI Z V Ð rt o1 1m
rh
(1)
rh c1m 2prdr

The frictional losses in bearing vary with the speed and were estimated by
performing a decelerating test on the turbine. The turbine without load was
accelerated to a maximum speed of about 7000 rpm and then the input power was cut
off by switching off the blower motor and blocking blower entry. The deceleration of
the turbine was found out by recording the turbine speed at regular intervals. The
moment of inertia of the rotating element of the turbine ‘I’, was calculated from the
geometry of the rotor, shaft and brake drum. The mechanical loss at every recorded
speed was computed as

du
Nb Z Iu (2)
dt

Net power from the turbine N0ZtuCNbwhere the torque t was found from the
dynamometer

 
N
Efficiency of the turbine; h Z 0 (3)
NI

100

80

60
η
40
Experimental Values
Computational Values
20

0
0.00 0.10 0.20 0.30
φ

Fig. 6. Variation of efficiency with flow coefficient.


T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2137

Fig. 7. Static pressure distribution on pressure side of turbine blade at various flow coefficients.

3. Numerical method and calculation

A commercial code, FLUENT 6 was used for the numerical analysis. The steady,
incompressible and three-dimensional Navier–Stokes equations were discretized by

Fig. 8. Static pressure distribution on suction side of turbine blade at various flow coefficients.
2138 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

the finite volume method in conservation form. The rotating reference of the turbine was
adopted. The k–3 model was used for solving turbulent flow. In order to get the appropriate
velocity profile at inlet as measured by experiment, the computational domain has been
created with hub nose and with all blades. Fig. 3 shows the computation domain with
unstructured grids on hub and blade surfaces. The total number of grid points is
approximately 620,000. Velocity and pressure are given as boundary conditions at inlet
and outlet, respectively. The rotational speed of the turbine was kept constant and inlet
velocity was increased to obtain various flow coefficients.

Fig. 9. Static pressure distribution on hub surface at various flow coefficients.


T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2139

4. Results and discussions

4.1. Validation of computed results

Figs. 4 and 5 show the comparison between experimental and computed values of axial
velocity at upstream and downstream of the turbine, respectively. The figures show that
the computed values match very well with the measured values for all the flow coefficients.
However, the computer values under predict the measured values at hub and tip regions.

Fig. 10. Static pressure distribution on turbine blade tip at various flow coefficients.
2140 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

The computed and measured efficiency of turbine is shown in Fig. 6. The comparison
between the two is satisfactory. Also, it can be seen that the stalling point of the turbine
observed at around fZ0.2 by the experiment is well predicted by the computation with
reasonable accuracy.

4.2. Three-dimensional turbine aerodynamics

Static pressure contours on pressure and suction side of the turbine blade have been
plotted and shown in Figs. 7 and 8, respectively. From Fig. 7(a)–(d), the variations in
the static pressure distribution on pressure side of the blade can be seen at operating
(fZ0.134, 0.155 and 0.206) and stall condition (fZ0.228). It can be clearly seen from

Fig. 11. Velocity contours at various flow coefficients: RZ0.3.


T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2141

the figure that the high static pressure in the leading edge region increases as the flow
coefficient is increased due to increase in incidence. It is computed that the region of high
pressure occupies 9, 10, 18.3 and 31.25% of chord length (CL) from the leading edge for
the flow coefficients of 0.134, 0.155, 0.206 and 0.228, respectively. It is interesting to note
that the high-pressure region at fZ0.228 exceeds the position of maximum thickness of
airfoil (30.3% of chord) and hence leads to stalling of the turbine. In mid chord region,
static pressure also varies from hub to tip in normal operating range (Fig. 7(a) and (b)). As
far as the static pressure distribution on suction surface of the blade is concerned,

Fig. 12. Velocity contours at various flow coefficients: RZ0.77.


2142 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

Fig. 8(a)–(d), the static pressure near the leading edge reduces as the flow coefficient is
increased. Also, it can be seen that the negative sign of static pressure at trailing edge
changes to positive sign when the flow coefficient increases from 0.206 to 0.228 (Fig. 8(c)
and (d)). This may be due to reduction in size of wake near suction surface. Effect of tip
clearance leakage flow on suction surface static pressure distribution can also be seen from
the above figure. Even though the effect seems negligible in low-flow coefficient (Fig. 8(a)
and (b)), significant effect can be seen at higher flow coefficient near tip of the suction
surface from 40 to 97% blade CL from the leading edge.
In order to find out the stagnation point on the blade profile and effective leakage
area through the tip gap, static pressure distribution on hub and blade tip has been
plotted and shown in Figs. 9 and 10, respectively, for various flow coefficients. From
Fig. 9(a), it can be observed that the stagnation point is very near to leading edge of
the turbine blade at flow coefficient fZ0.08. As the flow coefficient increases, it
moves towards mid-portion of the blade (Fig. 9(b)–(e)). While considering the static
pressure distribution on hub near trailing edge of the suction surface of the blade, it
can be clearly seen that the blade passage flow enters through the trailing edge to
mid-portion of suction surface. This entrainment is considerably more at higher flow
coefficient (near stall region) and hence the size of wake has reduced noticeably as
discussed previously. From the static pressure distribution on tip surface of the blade
for lower flow coefficient (Fig. 10(a)), it seems that the tip clearance leakage flow is
considerably higher in the trailing edge portion. But, as the flow coefficient increases,
leakage flow region advances towards leading edge (Fig. 10(d)) causing large mass
flow of air to leak through the gap.
Velocity contours at various planes of radius ratios (RZ0.3, 0.77) have been plotted
and shown in Figs. 11 and 12 to investigate the flow field at three flow coefficients;

Fig. 13. Location of angular planes denoted with blade chord length (CL).
T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2143

fZ0.134, 0.206 (before stall), 0.228 (after stall). Velocity contours at RZ0.3 indicate
that there is no significant change in the flow field except increase in velocity with the
flow coefficient. But, at higher radius ratio of RZ0.77 (Fig. 12(b) and (c)), swirling
components are comparatively high and wakes behind the blades grow as flow
coefficient increases.
In order to capture the size of wake behind the blade and its interaction with
adjacent blade wake, velocity contours have been plotted at various angular planes
(Fig. 13). Fig. 14(a)–(g) shows the velocity contours at 4, 20, 35, 51, 66, 82, 98

Fig. 14. Velocity contours at various radial planes at flow coefficient fZ0.134.
2144 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

percentage points of CL, respectively at low-flow coefficient of 0.134. From the


Fig. 14(a) and (b), it is seen that there is no sign of wake in these planes near leading
edge. In the region of 35% of CL (Fig. 14(c)), the flow separates from the suction
wall and noticeable size of wake is formed behind the blade and the wake is centered
near the tip region. As the CL increases, it increases in size and moves away from the
suction surface (Fig. 14(d)–(g)). In case of higher flows coefficient of 0.206 (Fig. 15),
the flow separation starts near 20% CL itself (Fig. 15(b)) and it grows as CL

Fig. 15. Velocity contours at various radial planes at flow coefficient fZ0.206.
T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2145

increases (Fig. 15(C)–(g)). It can also be observed that the wake is being pushed
away by the tip clearance flow jet. This gives the reason for advancement in stall as
tip clearance decreases as proved from the experiments by Raghunathan [16]. Also, it
is found from the figure that unlike the case of low-flow coefficient, wake created by
the adjacent blade still appears throughout the CL (Fig. 15(a)–(g)) at tip region and
about five times the blade thickness from suction surface. But, the size of the adjacent
blade wake is well noticeable at 4% CL and it diminishes in size as CL increases.
Similar kind of flow pattern is seen in the flow coefficient of 0.228 also (Fig. 16).

Fig. 16. Velocity contours at various radial planes at flow coefficient fZ0.228.
2146 T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147

But in this case the size of wake and wake from adjacent blade is very high
compared to earlier case. In addition, as the lift force decreases due to high incidence
in this flow coefficient, velocity of blade passage flow increases near tip region. This
flow is identified in-between blade wake and wake from the adjacent blade. The flow
is clearly visible from the Fig. 16(c)–(g). The above explanation complements the
reason for large decrease in efficiency of turbine beyond the solidity ratio of 0.5 given
in Raghunathan [16].

5. Conclusions

The present computational model has been validated with experimental results with
reasonable accuracy and found well suitable for further design analysis. Especially,
the value of stall point of the turbine is computed very close to experimentally
measured value. The computation has been carried out at normal operating range and
at stall condition to understand the flow physics and to improve the operating range
and performance of turbine. It is found from the computed results that the wakes
behind the turbine blades merge rigorously in the portion of RZ0.45–1.0, which leads
the turbine to stall. The physical explanation for advancement of stall due to decrease
in tip clearance of the turbine has been made with CFD results.

Acknowledgements

The authors wish to thank National Institute of Ocean Technology (NIOT) for
providing the necessary financial support.

References

[1] Raghunathan S. The Wells turbine. Internal report. Queens University of Belfast; 1980.
[2] Raghunathan S, Tan CP, Wells NAJ, Mc Ilhagger DS. Efficiency, starting torque and prevention of run-
away with Wells self rectifying turbines. Proceedings of second international symposium on wake and tidal
energy. Cambridge, England: BHRA; 1981 p. 207–217.
[3] Raghunathan S, Setoguchi T, Kaneko K. The wells air turbine subjected to inlet flow distortion and high
level of turbulence. Int J Heat Fluid Flow 1987;8:165.
[4] Raghunathan S, Setoguchi T, Kaneko K. Predictions of aerodynamic performance of Wells turbines from
aerofoil data. Trans ASME, J Turbomach 1990;112:792–5.
[5] Raghunathan S, Tan CP. Aerodynamic performance of a wells air turbine. J Energy 1983;7(3):226–30.
[6] Raghunathan S, Tan CP. Effect of blade profile on the performance of the Wells self rectifying air turbine.
Int J Heat Fluid Flow 1985;6:17–22.
[7] Grant RJ, Johnson CG, Strudge DP. Performance of a wells turbine for use in wave energy system. Proc
Third Int Conf Future Energy Concepts IEE 1981;192:117–22.
[8] Setoguchi T, Kaneko K, Inoue M. Determination of optimum geometry of Wells turbine rotor for wave
power generator. (Part I-Performance of isolated airfoils and rotors). Proc Curr Pract New Tech Ocean Engg
1986;2:435–40. New Orleans.
T.S. Dhanasekaran, M. Govardhan / Renewable Energy 30 (2005) 2129–2147 2147

[9] Inoue M, Kaneko K, Setoguchi T. Determination of optimum geometry of Wells turbine rotor for wave
power generator. (Part II-Considerations for practical use). Proc curr pract new tech Ocean Engg 1986;2:
441–6. New Orleans.
[10] Watterson JK, Raghunathan S. Computed effects of solidity on Wells turbine performance. JSME Int J
Series B 1998;41(1):199–205.
[11] Thakker A, Frawley P, Sheik Bajeet E. Numerical analysis of Wells turbine performance using a 3D Navier–
Strokes explicit solver. Proceedings of the 11th ISOPE, Stavanger, Norway, June 17–22; 2001.
[12] Setoguchi T, Kinoue Y, Kim TH, Kaneko K, Inoue M. Hysteretic characteristics of Wells turbine for wave
power conversion. Renewable Energy 2003;28:2113–27.
[13] Kim TH, Setoguchi T, Kaneko K, Raghunathan S. Numerical investigation on the effect of blade sweep on
the performance of wells turbine. Renewable Energy 2002;25:235–48.
[14] Swaminathan G. Performances and flow investigations on Wells turbine. PhD Thesis. IIT Madras; 1990.
[15] Raghunathan S, Ombaka OO. The Wells turbine in an oscillating flow. Proceedings of 19th intersociety
energy conversion engineering conference. San Francisco; 1984.
[16] Raghunathan S. The Wells air turbine for wave energy conversion. Prog Aerospace Sci 1995;13:335–86.

You might also like