You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325466651

WELLS AND IMPULSE TURBINES IN AN OWC WAVE POWER PLANT: A


COMPARISON

Conference Paper · September 2005

CITATIONS READS

7 381

3 authors, including:

Mattia Scuotto Antonio Falcao

6 PUBLICATIONS   192 CITATIONS   
Technical University of Lisbon
146 PUBLICATIONS   4,956 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

OPERA (EU H2020 project) View project

Development of oscillating-water-column wave energy converters and new types of air turbines View project

All content following this page was uploaded by Mattia Scuotto on 31 May 2018.

The user has requested enhancement of the downloaded file.


WELLS AND IMPULSE TURBINES
IN AN OWC WAVE POWER PLANT: A COMPARISON

Mattia Scuotto1, António F. de O. Falcão2


1
Delft University of Technology, Delft, The Netherlands
2
Instituto Superior Técnico, Lisbon, Portugal

rotor blades and guide vanes (if adopted)


Abstract
symmetrical with respect to a cross-sectional plane
The aim is to compare the performances of a given perpendicular to the shaft axis. The turbomachine
Oscillating Water Column (in given wave climate) drives a variable-speed generator rotor. The turbine
when equipped with two different types of air probably represents the most critical element in the
turbines: a Wells turbine and an impulse turbine. energy conversion process; it is essential to optimise
Comparisons between these two types of turbine its performance, subject to cost-related constraints.
have been published by some authors. These works Although the Wells turbine has been chosen to equip
were mainly focused on aerodynamic performances: most of the full-sized devices built so far, other types
if a comparison is to be meaningful it should address of self-rectifying air turbines have been developed
the overall performance of the plant the turbine for wave energy conversion purposes. The Wells
equips, since it is known that the turbine turbine and, more recently, the impulse turbine are,
aerodynamics affects the hydrodynamic process of by far, the most studied and investigated (although
wave energy absorption at the beginning of the not the only ones), theoretically as well as
energy conversion chain, and the quality of the experimentally. Dimensionless aerodynamic
produced electrical power at the end. The results performance curves for the Wells turbine and for the
obtained in numerical simulations put in evidence impulse turbine were taken from published testing
some inherent limits of both machines when adopted data (in both cases with 0.6m-rotor-diameter models
as power take-off system in OWC devices and may and with analogous test rigs). The numerical
address further investigations. simulations were performed in the time domain for
Introduction the non-linear impulse turbine and with a stochastic
model for the approximately linear Wells turbine.
The Oscillating Water Column (OWC) device is a
The spring-like effect due to air compressibility in
first generation wave power plant, of which some
the chamber was taken into account in a linearized
full-sized shoreline prototypes have been built as
form. The computations involved a set of rotor
research facilities. The energy conversion chain from
diameters for each turbine type. In every case, the
wave power to electricity is divided into three stages:
rotational speed was optimised for each sea state
wave power → pneumatic power → mechanical (subject to maximum rotor blade tip speed
power → electrical power. Assuming this scheme, constraints). We wish to remark that in present work
the present work can be regarded as a comparison only the aerodynamic losses, implicit in the
between two energy chains that differ from each efficiency curves of the turbines, were taken into
other in the turbine element. Each conversion step account. In both cases, we may identify at least four
requires the proper sub-device. The action of the other major sources of losses involved in the process:
incident waves causes oscillations of the water hydrodynamic viscous loss in the chamber, air-flow
column, within the submerged part of the inner hysteresis loop, bearing friction and the electrical
structure, which in turn induces variations of losses in the generator rotor. These losses, not
pressure in the air chamber of the device. In an accounted for in the calculations presented here, are
OWC, the mechanical power take-off system consists estimated to reduce the power output by 20-30%.
of a self-rectifying axial-flow air turbine, located in a
duct connecting the air chamber with outside Simulation environment
atmosphere. A self-rectifying turbine has the The Pico (Azores, Portugal) OWC plant was used in
capability to operate without a system of self- the simulations: its hydrodynamic coefficients
rectifying valves, accepting instead the air-flow from (excitation volume-flow Γ(ω), radiation conductance
both sides while rotating always in the same B(ω) and radiation susceptance C(ω)) were known
direction. Hence, these turbines do usually have both from published results computed with a boundary
element code (AQUADYN) (Brito-Melo, 2001). The through the turbine and p the pressure oscillation in
shoreline plant has a concrete structure (of 12×12 m the chamber).
internal square cross-section at free surface level)
that spans a natural gully whose depth is about 8 m.
Irregular waves were considered in the simulations.
Two different wave climates were used in the
comparisons. The wave climates were represented by
a set of 44 sea states, each sea state being
characterized by its Pierson-Moskowitz spectral
power distribution, with the significant wave height
HS and the energy period Te , and its frequency of
occurrence φ. The local wave climate (a
representation of which is given in Figure 1) results
from wave measurements at the plant’s location in Figure 2 – Dimensionless plot of Wells turbine
conjunction with longer-term results from power output versus pressure-head: Π(Ψ) (solid
forecast/hindcast numerical modelling for the same line), Π(σΨ) without valve control (dashed line) and
ocean area. Average wave power level is about 15 with valve control (chain line). (Falcão et al., 1998)
kW/m. In order to simulate and compare
performances in a more energetic environment, a
second wave climate, characterised by twice the
power, was created based on the original one by
simply multiplying, for each sea state, the value of
the significant wave height HS by 2.

Figure 3 – Efficiency of the Wells turbine versus


pressure-head: η(Ψ) (solid line), η(σΨ) without valve
control (dashed line) and with valve control (chain
line). (Falcão et al., 1998)

If the Reynolds number and the Mach number effects


are assumed negligible, then the dimensional analysis
applied to incompressible flow turbomachinery
Figure 1 – Local wave climate, represented by 44 sea allows us to write Π=fP(Ψ) and Φ=fQ(Ψ), the
states (HS , Te). The area of each circle is functions depending only on turbine shape and not
proportional to the contribution of the sea state to the on its size or rotational speed. The dimensionless
total annual wave energy. characteristics, along with some derived curves
which are explained below, are reported in Figure 2
Numerical modelling of the Wells turbine (dimensionless power output Π versus pressure-head
The reference machine equips the Pico plant and has Ψ) and Figure 3 (efficiency η). The maximum Πmax
an 8-bladed rotor of D=2.3 m outer diameter and is reached at Ψcrit=0.067, the pressure-head critical
Di=1.36 m inner diameter, with a row of guide vanes value beyond which aerodynamic stalling occurs at
on each side of the rotor. The blade chord is constant the rotor blades, resulting in severe drop in
(0.375 m) and its cross-section varies continuously efficiency. The Wells turbine is characterised by
from profile NACA 0015 at the hub to NACA 0012 high rotational speed. Since the blade tip-speed has
at the tip. Five different diameter sizes, from D=2.0 to be constrained (limiting the rotational speed to
m to D=3.8 m, were used in simulations. It is about 1500 rpm in the case of D=2.3 m) in order to
convenient to introduce dimensionless variables to avoid shock waves at blades tip, the stalling occurs
characterise the aerodynamic performance, namely even with moderately energetic sea conditions
the non-dimensional pressure head Ψ=p/(ρaN2D2), because of the high flow-rate caused by the pressure
the non-dimensional flow-rate Φ=µ/(ρaND3) and the difference. This must be avoided by providing, for
non-dimensional power output Π=ηΦΨ. (Here, N is example, the plant with a relief valve. The result is a
the rotational speed in radians per unit time, ρa the less severe downfall of the overall efficiency as sea
constant value of the air density, µ the mass flow rate power level increases (another solution, though
mechanically more complicated, is represented by the
improved version of the Wells turbine with variable pressure oscillation. On the other hand, the available
pitch-blades). pneumatic power to the turbine is
The straight line Φ=K×Ψ=0.6803×Ψ was found to µ
Pavai (t ) = p (t ) (7)
be a good approximation to the curve of the ρa
dimensionless flow-rate Φ=Φ(Ψ) and for this reason and the ratio of Pt(t) to Pavai(t) yields the expression
the machine is said to be linear, that is the ratio of for the turbine efficiency, that is the efficiency of the
mass flow-rate to pressure fluctuation is constant for central step in the energy conversion process
a Wells turbine of given geometry at given rotational Pt (t ) Π
speed, µ/p=KD/N. A proportionality relationship (at η (t ) = = . (8)
Pavai (t ) ΨΦ
constant rotational speed) between turbine flow rate The average available power is
Φ and pressure head Ψ is a basic assumption ∞
pµ ⎛ p2 ⎞

1
underlying the whole analysis based on the stochastic P avai = exp⎜ − ⎟dp (9)
methods developed by Falcão (Falcão, 2002), to 2π σ p ρa ⎜ 2σ p ⎟
−∞ ⎝ ⎠
which we remand for further details. whose dimensionless form is
It was assumed that the local wave climate may be ∞
⎛ Ψ2 ⎞

1
represented by a set of sea states, each being a Π avai = f Q (Ψ ) exp⎜ − ⎟dΨ (10)
stationary stochastic ergodic process. Hence, the 2π σ Ψ ⎜ 2σ 2 ⎟
−∞ ⎝ Ψ ⎠
probability density function of sea surface elevation ς Then, under the assumption of the linear relationship
at a fixed observation point is supposed to be
Φ=KΨ, it could be proved that the expression (10)
Gaussian with variance

for the average available power may be easily turned

∫ Sζ (ω )dω
into
σ ζ2 = (1)
Π avai = K ⋅ σ Ψ
2 (11)
−∞
where ω is radian frequency and Sς(ω) is a two sided Thus, the average efficiency of the turbine is
spectral density (per unit radian frequency Π Π
η= = (12)
bandwidth). Π avai Kσ Ψ
2

The assumptions made, together with the adoption of The curves represented by equations (6), (10) and
linear water wave theory and the modelling of the (12) turned out to be more useful when dealing with
spring-like effect of air in the chamber by a real random waves.
linearised isentropic relationship between air The method presented above is computationally
pressure and density, lead to a fully linear system much less time consuming if compared to the
whose pressure oscillation p(t) has a Gaussian traditional time domain analysis, but can be applied
distribution with variance given by only to linear systems. However, if the aerodynamic

stalling occurs (as the pressure-head becomes larger

2
σ 2p = 2 Sζ (ω ) Γ(ω )Λ (ω ) dω (2) than ψcrit ), the system ceases to behave linearly
0 (experimental results on the Wells turbine show that
−1
⎛ ⎛ KD ⎞ ⎛ ωV ⎞⎞ the flow-rate/pressure-head linear approximation is
Λ = ⎜ ⎜⎜ + B(ω ) ⎟⎟ + i⎜⎜ 0 + C (ω ) ⎟⎟ ⎟ (3) no longer valid). In the numerical simulations the
⎜ ρ N p γ ⎟
⎝⎝ a ⎠ ⎝ a ⎠⎠ presence of an ideal relief valve has been simply
where V0 is the undisturbed air chamber volume, γ is simulated by replacing the actual value of turbine
specific heat ratio for the air and pa is the instantaneous power output Pt with the maximum
atmospheric pressure. From the expression of the power the turbine can produce Pt,MAX=ρaN3maxD5Πcrit
instantaneous turbine power output when Ψ >Ψcrit . The curves of the average values of
P(t )t = ρ a N 3 D 5 f P (Ψ (t ) ) (4) the power output and the efficiency are modified as
the expression for the average value of the turbine shown in Figure 2 and 3. A properly designed set of
power output is derived, according to the assumed real valves can approximate the behaviour of an ideal
pressure distribution properties valve. Implementation of such devices leads to
∞ ⎛ p2 ⎞ reliability and mainteinance-related problems.
exp⎜ − ⎟ f (Ψ ( p ) )dp
1
Pt =
2π σ p ∫ ⎜ 2σ 2 ⎟ P
(5) The present procedure is aimed to maximize the plant
−∞ ⎝ p⎠ annual energy production. This aim is achieved, in
or, in dimensionless form, each individual sea state, by assigning the turbine a
∞ constant rotational speed N, chosen to match sea
⎛ Ψ2 ⎞
⎟ f P (Ψ )dΨ

1
Π= exp⎜ − (6) conditions in order to maximize the produced energy.
2π σ Ψ ⎜ 2σ 2 ⎟ However, the turbine’s rotational speed is subject to
−∞ ⎝ Ψ ⎠
the constraint ND<2Mmaxca , where ca is sound speed
where σΨ=σp/(ρaN2D2) and Π = Pt /(ρaN3D5). in normal air conditions and Mmax is maximum
Equations (5) and (6) represent the turbine power allowable Mach number (usually 0.5-0.6). If φj is the
output as a function of the root-mean-square of the frequency of occurrence of the j-th sea state, then the
maximum of average annual power output produced The impulse turbine is non-linear, in the sense that
is given by the linear relationship between Φ and Ψ is no longer
44 valid and must be replaced by the experimental curve
∑ Pt ,max ⋅ φ j
ann ( j)
P t , max = (13) Φ=fQ(Ψ). As a consequence, the stochastic method
j =1 cannot be applied and had to be replaced a more
traditional time domain analysis. Thus, the time-
Numerical modelling of the impulse turbine
average of the observed behaviour of the turbine in
The reference machine model (built and tested by given conditions for a reasonably long time interval
Thakker et al., University of Limerick) has a 30- (several minutes) here represents the power out of
bladed rotor of D=0.6 m outer diameter and a hub-to- the plant in those conditions (namely, sea state and
tip ratio of 0.6, with a row of 26 plate (no airfoil rotational speed).
profile) guide-vanes on each side. Rotor blade The same assumptions as before were kept on what
sections at mean blade-height exhibit an elliptical concerns linear wave theory and the linear isentropic
profile at the suction surface and circular profile at relationship for air compressibility in the chamber.
the pressure surface. Chord length is 0.1 m. The Compressibility (Mach number) effects of the air-
geometry was designed following optimisation flow through the turbine were neglected (a single
criteria (for further detail see Setoguchi et al., 2000). curve for the machine efficiency was assumed). The
The published experimental data, taken from equation to be integrated is (remember that
(Thakker et al., 2003) and (Thakker et al., 2004), was p=ρaN2D2Ψ and µ=ρaN2D2Φ )
extended by the authors by linearly extrapolating the t
V0 dp µ(t)
curve of the efficiency down to zero, (based the
highest flow rate experimental points), and beyond Ψ paγ dt
− ∫ gr (t −τ ) p(τ )dτ + ρa − qd (t ) = 0 (14)
−∞
= 1.77 the non-dimensional power output was
the symbols assuming here the same meaning as
assumed to be zero. Six different diameters were
before and ρa is an averaged value of the air density
used in simulations, from D=1.2 m to D=2.7 m.
over a sufficiently long period of time (say, ten or
twenty times the energy period). (Variations in the
instantaneous value of ρa in the turbine flow were
ignored, in spite of the occurrence of relatively large
pressure peaks in the more energetic sea states.) The
radiation and diffraction flow rates are, respectively
t
qr (t ) = ∫ gr (t − τ ) p(τ )dτ (15)
−∞
N
Anωn
qi (t ) = 2b ∑ kn
sin( 2k n a ) cos(ωn t + θ n ) (16)
n =1
where An is the n-th wave component amplitude, kn
the wave number and θn the phase, while N is the
Figure 4 – Impulse turbine, non-dimensional power number of regular waves whose superposition gives
output versus non-dimensional pressure head. the irregular sea surface of given sea state. The
function gr(t) in the convolution integral is a function
that depends on system geometry and is defined as
the inverse Fourier transform of the positive and real-
valued function B(ω)

2
g r (t ) =
π ∫ B(ω ) cos(ωt )dω. (17)
0
In numerical simulations, the diffraction flow rate
was computed directly from the wave power
spectrum, with 225 wave components, and a random
number generation routine was implemented to
ensure random phases. The convolution integral (15)
was truncated to t-30 s (t = present time), limiting the
system memory to half a minute, to save
Fgure 5 – Impulse turbine, efficiency versus non- computational time without significant loss in
dimensional pressure-head. accuracy. For given diameter and rotational speed,
the integration of equations (14)-(16) was performed
numerically using a trapezoidal rule with a time step
of 0.1 s. The integration interval was limited to 9000 hand rising branch of the four curves, it can be seen
points, corresponding to 15 minutes. For a given that, for the same time-averaged power output, the
turbine size and sea state, the rotational speed Nopt impulse turbine is significantly smaller than the
that ensured the maximum average power output was Wells turbine (in terms of rotor diameter). Also
searched within a proper interval N*opt+∆N , whose keeping into account that the plant equipped with a
central value N*opt is the optimal rotational speed Wells turbine must be provided with a set of active
computed applying the stochastic method (in almost relief valves, this would lead to reconsider the
all cases the difference Nopt - N*opt was significant). impulse turbine when maximum profit (rather than
On what concerns the limit of maximum blade tip- maximum energy production) is searched for. For a
speed, a constraint analogous to that of the Wells brief introduction to the subject and some results, see
turbine was adopted. However, NoptD/2 never (Falcão, 2003) and (Scuotto, 2004). Another thing
exceeded this limit, allowing the turbine to be that should be noted in Figure 6 is that the impulse
properly tuned to whichever matched sea conditions. turbine type seems to gain more advantage from
The average annual produced power is then given by increased power availability: the position of the
equation (13). upper curve (double power wave climate) with repect
to the lower one (local wave climates) shows in
Results and further analysis average a higher increase, in percentage terms, of
In order to compare results obtained with the two power output if compared with the two Wells turbine
different approaches, we had to prove that with both type curves. Further investigations were done on the
methods (time domain and frequency domain) the subject by selecting a turbine of each type and
performances of the Wells turbine were equally running simulations (time domain) with very
predicted. Several simulations in time domain with energetic sea conditions. Diameter D=1.8 m was
Wells turbines of various sizes in very different sea chosen for the impulse turbine, D=2.3 m for the
states were done and the results showed differences, Wells turbine, the two machine being comparable
with respect to stochastic methods, in average power from the point of view of the power output in both
output not exceeding about 2-3%. wave climates. Results are shown in Figures 7 and 8.
In Figure 6 the results for average annual power The absence of limiting constraints on rotational
output are presented. Though in both climates the speed would seem to grant the impulse turbine an
competition for maximum produced power is won by almost unlimited power production capability as the
the Wells turbines, it is clear that the size of these available sea power increases, which is not realistic.
machines is considerably larger. Looking at the left-

Figure 6 – Turbine average annual power output versus diameter size, in both considered wave climates.
Impulse turbines performance is represented by left-hand curves, Wells turbines are on the right-hand side.
developed by the generator rotor (Falcão, 2002-b),
derived by integrating the equation that expresses
the dynamic of the system, for example
dN Pt (t )
I = − 0.0592 × N (t ) 2.5375 (18)
dt N (t )
where I is turbine inertia in kg/m2 and the second
term on the right-hand side is the requested
electromagnetic torque as a function of N(t). The
exponent and the number that multiplies the
rotational speed are determined by interpolating the
pairs (Nopt , Pt,max) obtained in several sea states
Figure 7 – 2.3 m Wells turbine (chain line) and 1.8 with a turbine of given size (in equation (18) are
m impulse turbine (solid line) average power output reported the values computed with diameter
in different sea states with fixed wave height (5 m); D=1.8m using the performance at constant N in the
Ej represents power per meter of wave crest. 44 sea states of the local wave climate).
The graph in Figure 9 suggests that compressibility
effects of the air in the case of the impulse turbine
may turn out not to be negligible beyond a certain
wave power level. Possibly this could occur and be
significant in the more energetic wave climate
simulated here.
Conclusions
The impulse turbine was tested with a constant
axial air velocity slightly below 10 m/s and with
rotational speeds ranging from about 100 to about
Figure 7 – 2.3 m Wells turbine (chain line) and 1.8 1250 rpm (Thakker et al., 2003 and 2004). The
m impulse turbine (solid line) average power output Reynolds number was assumed large enough for
in different sea states with fixed energy period (10 Reynolds effects to be negligible. Possibly due to
s); Ej represents power per metre of wave crest. inherent limits of the experimental facilitiy, in the
tests the air-flow Mach number was kept low.
Figure 9 shows the Mach number as a function of Curves for the efficiency and other dimensionless
time in a critical section of the machine, an impulse characteristics may result different at higher values
turbine of diameter D=1.5 m, in three different sea of the Mach number. By means of a more complex
states chosen from local climate. theoretical analysis some significative corrections
The curves were obtained after modifying the are possible. However the prediction of the impulse
program used for time domain analysis: integration turbine aerodynamic performance at high Mach
time step is 0.01 s, air density ρ(t) is a function of numbers (i.e. in highly energetic sea states) is
time as well as the rotational speed N(t) which is unreliable without further turbine testing at
controlled by an algorithm that set the proper appropriate Mach numbers.
instantaneous electromagnetic torque that has to be

Figure 9 – Impulse turbine, D=1.5 m; Mach number at the outlet edge of the first row of guide-vanes in three
diffrent sea states taken from local climate: HS=1.56m, Te=7.93s (solid line), HS=3.25m, Te=11.5s (dashed line),
HS=4.25m, Te=14.5s (chain line). Annual contribution to energy production can be extimated from Figure 1.
A. Thakker, T.S. Dhanasekaran, 2003. ”Computed
Nomenclature
effects of tip clearance on performance of
ω radian frequency impulse turbine for wave energy conversion”,
HS characteristic wave height Renewable Energy 29 529-547.
Te energy period
φ frequency of occurrence A. Thakker, F. Hourigan, 2004. “Modelling and
p relative pressure scaling of the impulse turbine for wave power
ρ air density applications”, Renewable Energy 29 305-317.
µ mass-flow rate
A. Thakker, Z. Usmani, T.S. Dhanasekaran, 2004.
D turbine diameter
“Effects of turbine damping on performance of
N rotational speed (rad/s)
an impulse turbine for wave energy conversion
Pt power output
under different sea conditions using numerical
Pavai available power
simulation techniques”, Renewable Energy 29
k wave number
2133-2151.
Ψ dimensionless pressure-head
Φ dimensionless flow rate M. Scuotto, 2004. “Wells and impulse turbines in
Π dimensionless power output an OWC wave power plant: a preliminary
σX variance of quantity X comparison”, Master Thesis, Università degli
η efficiency Studi di Padova, Italy.
B, C, Γ hydrodynamic coefficients
X average value of quantity X

Acknowledgements
The present paper is based on the first author’s
Master thesis work (Scuotto, 2004), partly carried
out while he was at the Instituto Superior Técnico
(Lisbon, Portugal) under the supervision of
Professor António Falcão.

Corresponding the author: mattia.scuotto@gmail.com .

References
A. Brito-Melo, 2001. Thèse de Doctorat (PhD
Thesis) – Université de Nantes (Nantes,
France) / Instituto Superior Técnico (Lisbon,
Portugal).
A. F. de O. Falcão, P.A.P. Justino, 1998. “OWC
wave energy devices with air flow control”,
Ocean Engineering 26 1275-1295.
A.F. de O. Falcão, 2002. “Stochastic modelling of
OWC wave power plant performance”,
Applied Ocean Research, 24 59-71.
A.F. de O. Falcão, 2002. “Control of an OWC
wave power plant for maximum energy
production”, Applied Ocean Research 24 73-
82.
A.F. de O. Falcão, 2003. “Maximum energy
production and maximum profit as alternative
criteria for wave power equipment
optimization”, 5th European Wave Energy
Conference, Cork, Ireland, September 2003.
T. Setoguchi, M. Takao, S. Santhakumar, K.
Kaneko, H. Maeda., 2001. “A review of
impulse turbine for wave energy conversion”,
Renewable Energy 23 261-292.

View publication stats

You might also like