You are on page 1of 9

On the Extractable Power from a Tidal Channel

Patrick F. Cummins1

Abstract: The drive to develop marine renewable energy has made resource assessments a matter of pressing need and ongoing research. In
the case of tidal in-stream power generation, a central objective of such assessments is to estimate the maximum power that may be extracted
from tidal motions based on observable properties of the flow in the natural undisturbed state. To this end, analytical models have been
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

developed for simple geometries, such as a channel connecting the open ocean to an inner basin. A key assumption of these models is that
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

the along-channel volume flux is nondivergent. This requires the channel length to be small in comparison to the tidal wavelength and also
that μ ≪ 1, where μ is the ratio of the surface area of the channel to that of the basin. A practical consequence of assuming nondivergent flow is
that the extractable power is then independent of position along the channel. In the present study, a linear one-dimensional model is developed
to study the power potential of a uniform tidal channel that is spanned by a turbine fence and that links a bay to the open ocean. The assumption
of nondivergent flow is relaxed, permitting an exploration of the sensitivity of the extractable power to nondimensional parameters measuring
the relative channel length and area. Depending on parameter values, the results show that assuming nondivergent flow may lead to an under-
estimation of the available power. For long waves, the maximum extractable power is shown to scale as ð1 þ μÞ, thus providing a simple
correction to account for the effects of finite channel area. It is shown that the power potential may vary appreciably with the position along the
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

channel. However, for long waves the optimal location of the turbine fence is always at the mouth where the magnitude of the along-channel
flow is greatest. DOI: 10.1061/(ASCE)WW.1943-5460.0000102. © 2012 American Society of Civil Engineers.
CE Database subject headings: Renewable energy; Tidal power; Resource management; Tidal currents.
Author keywords: Marine renewable energy; Tidal electric power generation; Resource assessment; Tidal currents.

Introduction A series of recent theoretical studies have considered the esti-


mation of the extractable power as a problem in optimization. This
The ocean is receiving increasing attention as a potential source of is based on recognition that the available power increases as tur-
renewable energy. While this idea has long held appeal, the har- bines are introduced, but that the addition of too many turbines will
nessing of ocean energy has been limited to date. In particular, tend to block the flow and lead to reduced power. Garrett and
a small number of tidal barrages exist around the world, and these Cummins (2004) examined the power potential of a uniform fric-
have operated reliably for decades (Hammons 1993). However, as tionless channel connected to a small bay with the influence of the
barrages may have detrimental ecological impacts, recent efforts to turbines represented as a linear or quadratic drag on the tidal flow.
exploit the potential of the tides have tended to focus on the devel- Garrett and Cummins (2005) considered the maximum extractable
opment of in-stream energy-conversion devices that extract power power from a tidal channel of variable cross section that connects
from tidal currents (e.g., Bedard et al. 2010). There have been trial two large basins. For quadratic drag, this is given by
deployments of these submarine turbines at a number of coastal
locations. Such efforts may be expected to continue as tidal power
advances beyond its present precommercial stage of development. PGC05 ¼ γρgaQmax ð1Þ
The development of tidal in-stream technology has been accom-
panied by the appearance of a number of resource assessments that where ρ is the density of seawater, g is the acceleration due to grav-
attempt to place upper limits on the extractable power at different ity, a is the amplitude of the tidally varying elevation difference
sites over the coastal ocean (e.g., Cornett 2006; ABP Marine Envi- between the ends of the channel, and Qmax is the peak volume flux
ronmental Research Ltd. 2008). With few exceptions, these assess- in the undisturbed state. The dimensionless coefficient γ varies
ments have been based on calculations of the time-averaged flux of within a small range, 0.20–0.24, depending on the extent to which
kinetic energy. It is now understood that this approach is generally friction dominates the along-channel momentum balance in the
lacking theoretical support and may lead to anomalous results, natural state. Power generation thus depends on the product of
such as an extreme sensitivity of the extractable power to the the head across the channel, ρga, and the volume flux. An addi-
location where a turbine fence is placed along a channel of variable tional result of the analysis is that available power is independent
cross-section. of the position of the turbine fence along a channel.
1
Blanchfield et al. (2008a) extended these results to the case of a
Research Scientist, Institute of Ocean Sciences, Fisheries and Oceans tidal channel connecting a bay to the ocean. In the case of quadratic
Canada, Sidney, BC, Canada V8L 4B2. E-mail: Patrick.cummins@ drag, they find an expression similar to Eq. (1) for the extractable
dfo-mpo.gc.ca
power, but with a now interpreted as the amplitude of the tidal
Note. This manuscript was submitted on January 5, 2011; approved on
May 2, 2011; published online on May 5, 2011. Discussion period open elevation just outside the channel, in the open ocean. This extended
until June 1, 2012; separate discussions must be submitted for individual model has been applied to estimate the maximum available power
papers. This paper is part of the Journal of Waterway, Port, Coastal, and from Masset Sound, British Columbia (Blanchfield et al. 2008b).
Ocean Engineering, Vol. 138, No. 1, January 1, 2012. ©ASCE, ISSN Karsten et al. (2008) applied a similar analytical model to estimate
0733-950X/2012/1-63–71/$25.00. the potential of Minas Passage, Nova Scotia, and obtained good

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 63
agreement with the results of simulations with a numerical tidal
model.
In each of these analytical models, a key assumption is made
that the volume flux along the channel is independent of position
and is a function only of time. Formally, this requires the channel
length to be small compared to the tidal wavelength and also that
the area of the channel be small compared to that of the basin.
While it may be expected that the long-wave approximation will
be well satisfied in most practical circumstances, there are likely
to be cases where one or both of the assumptions for flow
nondivergence are violated. For example, in the case of Masset
Sound (Blanchfield et al. 2008b) neither assumption holds particu-
larly well because the channel area is about 20% of the basin area
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

and the nondimensional wavelength parameter (defined below) is Fig. 1. Schematic of the model domain, which consists of a basin of
not small compared to unity ðk0 L ≈ 0:4Þ.
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

surface area, AB , connected to the ocean via a channel of length, L,


In the following, a simple linear model is constructed to exam- uniform width, W, and surface area, AC ¼ LW; at x ¼ xT , the channel
ine the extractable power in the case of a uniform channel that con- is spanned by a turbine fence that extends over small distance, Δx,
nects a bay to the open ocean. In contrast to previous analytical along the channel
studies, the assumption of nondivergent flow is not made. As a con-
sequence, nondimensional parameters measuring the channel area
and length appear explicitly as independent parameters. In addition,
r ¼ r C þ rT ðx; xT Þ
the extractable power may depend on the position of the turbine 8 9
fence along the channel. The sensitivity of the maximum extract- > 0; L ≤ x ≤ xT  Δx=2 >
< =
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

able power to the governing nondimensional parameters of this ¼ r C þ r 0T ; xT  Δx=2 < x < xT þ Δx=2 ð4Þ
problem is explored, and results are compared with existing theory. >
: >
;
Although the direct applicability of the model is limited because of 0 xT þ Δx=2 ≤ x ≤ 0
its idealized nature, the results nevertheless have practical implica-
tions with regard to the estimation of available power and the use of where r0T is the magnitude of the turbine drag coefficient. The
existing analytical models. neglect of momentum advection in Eq. (3) eliminates exit separa-
tion effects which may balance the head in short channels [Garrett
and Cummins (2005)]. While quadratic drag would be more real-
Linear Model istic, use is made of a linear drag law in Eq. (3) to allow a simple
analytical approach to the problem.
A linear, one-dimensional model is considered for the domain illus-
Following Blanchfield et al. (2008a), a periodic tidal forcing,
trated in Fig. 1. This consists of a uniform rectangular channel of
ζ ¼ a cos ωt, is imposed at the open boundary at x ¼ 0. Here, as
length L that connects the open ocean to an inner basin of surface
in previous studies, it is assumed for simplicity that there is no back
area AB . The channel surface area is AC ¼ WL, while the cross-
effect so that the amplitude of the tidal forcing is unaffected by the
sectional area is ACR ¼ WH with H and W the constant depth
introduction of turbines in the channel. As mentioned in Garrett and
and width, respectively. As depicted in Fig. 1, the channel is
Cummins (2005), this effect is likely to be small if the exterior re-
spanned by a fence of turbines centered at x ¼ xT and extending
gion is large and deep. In other circumstances, the back effect may
over a small interval Δx along the channel. The objective is to cal-
be substantial and lead to an increase in the available power
culate the maximum power that may be extracted from the tidal
(e.g., Garrett and Greenberg 1977). At the basin/channel junction,
motions by the turbine fence. It is recognized that in any practical
a linearized relation is written for conservation of volume for the
setting only a fraction of the energy dissipation in the fence will do
basin
productive work to drive turbines. Inevitably, there will be losses
due to internal turbine inefficiencies and losses to wake mixing. For
maximum power extraction it is assumed that all of the flow is dζ B
AB ¼ ACR u at x ¼ L ð5Þ
through the fence; the efficiency of partial fences was considered dt
in Garrett and Cummins (2007).
Assuming small amplitude motions, the linearized continuity where ζ B ðtÞ is the area-averaged sea-level displacement of
equation for the channel is the basin.
Nondimensional variables are denoted by primes and defined
∂ζ ∂u
¼ H ð2Þ as follows: t ¼ ω1 t0 , x ¼ Lx0 , ðζ; ζ B Þ ¼ aðζ 0 ; ζ 0B Þ, and u ¼
∂t ∂x pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffi
a g=H ðk 0 LÞ1 u0 , with k0 L ¼ ωL= gH the nondimensional
where ζ is the sea-level displacement and u is the along-channel wave-number parameter. This leads to a nondimensionalization
current. The linearized along-channel momentum equation is of the system that is consistent with Garrett and Cummins
(2005) and Blanchfield et al. (2008a). Assuming harmonic time
∂u ∂ζ dependence for the sea-level displacement and along-channel cur-
¼ g  ru ð3Þ 0
∂t ∂x rent, ðζ 0B ; ζ 0 ; u0 Þ ¼ ℜf½~ζ B ; ~ζðx0 Þ; ~uðx0 Þeit g, the continuity Eq. (2)
becomes
The coefficient of the linear drag law in Eq. (3) is assumed to
consist of a uniform background value, r C , and a localized drag, r T ,
1 d u~
associated with the turbine fence. Accordingly, the drag coefficient i ~ζ ¼  ð6Þ
in Eq. (3) has the form of a boxcar function and is written as k20 L2 dx0

64 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
The momentum Eq. (3) reduces to chosen to match the sea-level amplitude ratio, maxðζ B Þ=a, and
the phase lag of the basin relative to the channel mouth in the un-
d ~ζ disturbed state. The analytical solution given in the appendix can be
i ~u ¼   δ~u ð7Þ
dx0 used for this purpose. The dependence of the extractable power on
the parameters of Eq. (13) is examined below.
where δ ¼ ðr=ωÞ ¼ δ C þ δ T ðx0 ; x0T Þ is the nondimensional friction
and x0T ¼ xT =L the nondimensional position of the turbine fence. Long-Wave Theory
Eqs. (6) and (7) may be combined to form a second-order
boundary-value problem for the complex sea-level displacement A simple closed-form expression for the extractable power can be
 obtained under the assumption that the volume flux is independent
~
d 1 d ζ of position along the channel and a function only of time. This is
ð1  iδÞ þ ðk0 LÞ2 ~ζ ¼ 0 ð8Þ
dx0 dx0 referred to in the following as long-wave theory. However, the
assumption of flow nondivergence requires not only that k 0 L ≪ 1
but also that the relative channel area be small, μ ≪ 1. These con-
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

The imposed tidal forcing at the channel mouth leads to the


boundary condition ditions, which were noted in passing by Blanchfield et al. (2008a),
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

are explicated in the appendix.


~ζ ¼ 1 at x0 ¼ 0 ð9Þ Assuming nondivergent flow, Eq. (7) may be integrated from
x0 ¼ 1 to x0 ¼ 0 to yield
To obtain a boundary condition at the basin/channel junction, it
is assumed that the basin responds as a uniform reservoir so that ½i þ ðΔx0 δ0T þ δ C Þ~u ¼ ð1  ~ζ B Þ ð14Þ
area-averaged surface displacement in the basin can be taken as the
elevation at the junction with the channel, ~ζ B ¼ ~ζðx0 ¼ 1Þ. This Here ~ζ B ¼ ~ζðx0 ¼ 1Þ, and use has been made of Eq. (9). From
requires the tidal wavelength in the basin to be much greater than Eq. (5), we have ~ζ B ¼ iβ~u with β ¼ μ=ðk 0 LÞ2 . Substituting and
the scale of the basin. Combining the nondimensional version of solving for ~u gives
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

Eqs. (5) with Eq. (7) and making use of the volumetric relation,
ðλ þ δ C Þ  iðβ  1Þ
ACR ¼ HAC =L, yields the condition ~u ¼ ð15Þ
ðβ  1Þ2 þ ðλ þ δC Þ2
d ~ζ ðk 0 LÞ2
þ ð1  iδ C Þ~ζ ¼ 0 at x0 ¼ 1 ð10Þ The average nondimensional power dissipation is
dx0 μ ð1=2Þðλ þ δ C Þj~uj2 , of which a fraction, λ=ðλ þ δ C Þ, is expended
with μ ¼ AC =AB . in the turbine fence. The average dissipation in the fence is then
The average rate
R of energy dissipation in the turbine fence is 1 λ
given by Pavg ¼ uFdx, where F ¼ ρACR r T u is the drag force P0LW ¼ ð16Þ
2 ðβ  1Þ þ ðλ þ δ C Þ2
2
per unit length exerted by the flow on the fence and the overbar
denotes the time average over a tidal cycle. Substituting Eq. (4) The peak power occurs for λ ¼ λP, where
and integrating yields
λP ¼ ½ðβ  1Þ2 þ δ 2C 1=2 ð17Þ
Pavg ¼ ρACR Δxr 0T u2T ð11Þ
and is given by
where uT ¼ uðxT ; tÞ and it has been assumed that the interval Δx is maxðP0LW Þ ¼ 0:25ðλP þ δC Þ1 ð18Þ
R x þΔx=2
sufficiently small ðΔx=L ≪ 1Þ that xTTΔx=2 u2 dx ≈ Δxu2T . The
If channel friction is unimportant ðδ C =jβ  1j ≪ 1Þ, then
channel cross section has been taken as constant in Eq. (11), con-
maxðP0LW Þ ¼ 0:25jβ  1j1 . On the other hand, if channel friction
sistent with the assumption of small amplitude motions, a=H ≪ 1,
is dominant (δ C =jβ  1j ≫ 1) as in, for example, the case of
required for linearization of the continuity Eq. (2). In nondimen-
Helmholtz resonance (β ¼ 1), then λP ¼ δC and maxðP0LW Þ ¼
sional form, the power dissipation is
0:125δ 1
C .
 
Lω 1 Eq. (16) is a generalization of previous results with linear tur-
P0avg ¼ Pavg ¼ λj~uT j2 ð12Þ bine drag. In particular, for negligible channel friction ðδ C ¼ 0Þ,
ρACR a2 g2 2
Eq. (16) reduces to the solution given by Eq. (34) of Blanchfield
where λ ¼ Δx0 δ 0T is the turbine drag parameter with δ 0T ¼ r 0T =ω and et al. (2008a). Garrett and Cummins (2005) considered a large
Δx0 ¼ Δx=L. Use has been made in Eq. (12) of the identity, u02 basin for which β ¼ 0; setting β ¼ δC ¼ 0 in Eq. (16) results in
T ¼
the nondimensional version of their Eq. (2.5). The linear results
uðx0T Þ.
ð1=2Þj~uT j2 with ~uT ¼ ~
of Garrett and Cummins (2004), given in their Eqs. (14) and (16),
The power that is extracted by the turbine fence from the tidal
are recovered by setting δ C ¼ 0 and assuming β ≫ 1 so that
motions depends on parameters associated with the drag of the
ðβ  1Þ ≈ β in Eqs. (16) and (17). This approximation amounts
fence, λ, and its position, x0T , along the channel. In addition, the
to neglecting the acceleration term in Eq. (3).
extracted power depends on the following three environmental
It is evident that the assumption of nondivergent flow leads to
parameters
important simplifications of the system. In particular, the extract-
AC ωL rC able power given by Eq. (16) is independent of the position of the
μ¼ ; k0 L ¼ pffiffiffiffiffiffiffi ; and δC ¼ ð13Þ
AB gH ω turbine fence along the channel. In addition, the number of non-
dimensional parameters is reduced from the three given in
that measure the relative surface area and length of the channel and Eq. (13) to two parameters, δC and β, with the latter a composite
the strength of background frictional damping. In practical situa- parameter formed from μ and k0 L. Fig. 2 presents the maximum
tions, values for these parameters may be estimated from tidal extractable power with weak and strong channel friction as a func-
sea-level data following the approach of Blanchfield et al. (2008a). tion of μ and k 0 L. The plotted range of these parameters in Fig. 2
The channel and basin areas determine μ, while k 0 L and δC are extends beyond the formal range of validity of long-wave theory.

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 65
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

Fig. 2. Nondimensional average extractable power based on long-wave theory, maxðP0LW Þ, for (a) weak ðδC ¼ 0:1Þ; (b) strong ðδ C ¼ 2Þ background
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

pffiffiffi
channel friction; the dashed curves are the loci of the Helmholtz resonance condition, k0 L ¼ μ
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

Fig. 3. (a) Variation of the nondimensional average extractable power with the turbine drag parameter in the case of a turbine fence located next to the
basin ðx0T ¼ 0:99Þ or at the mouth of the channel ðx0T ¼ 0:01Þ; environmental parameters are set to ðμ; k0 L; δ C Þ ¼ ð0:4; 0:05; 0:1Þ; (b) variation of
the maximum extractable power with position of the turbine fence along the channel; (c) along-channel variation in the magnitude of the current,
j~uðx0 Þj, in the undisturbed state; the dotted horizontal line represents the current magnitude in long-wave theory

The contours in this figure correspond to curves of constant β. For a similar factor, the ratio may be evaluated from Eq. (15) and is given
given value of μ, the power increases with the wave number, reach- by
ing a maximum at Helmholtz resonance, and then decreasing at
larger wave numbers. j~ζ B jλ¼λP
R¼ ¼ ½2ð1 þ δ C =λP Þ1=2 ð19Þ
A measure of the environmental disturbance produced by the j~ζ B jλ¼0
introduction of the turbines is the ratio of the elevation amplitude
in the basin at maximum power extraction to that in the undisturbed In the limit that channel friction is unimportant (δ C =j
state. Since the amplitude of the channel current is reduced by a β  1j ≪ 1), R ¼ 21=2 , and the tidal range inside the basin is

66 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
8 9
reduced to 71% of its original amplitude, independent of the value > A cos kLx0 þ B1 sin kLx0 ; 1 ≤ x0 ≤ x012 >
< 1 =
of β (cf. Garrett and Cummins 2004). Conversely, for a tidal chan- ~ζðx0 Þ ¼ A2 cos k2 Lx0 þ B2 sin k 2 Lx0 ; x012 < x0 < x023 ð20Þ
nel controlled by friction ðλP ≈ δ C Þ, the elevation in the basin is >
: >
;
reduced to half of its natural range. By comparison, Garrett and A3 cos kLx0 þ B3 sin kLx0 ; 0 0
x23 ≤ x ≤ 0
Cummins (2005) found that the transport in a frictionally controlled
channel with quadratic drag was reduced by a similar fraction where kL ¼ k0 Lð1  iδ C Þ1=2 , k 2 L ¼ k0 L½1  iðδC þ δ 0T Þ1=2 ,
(R ¼ 31=2 ≈ 0:58) at maximum power. x12 ¼ xT  Δx =2, and x23 ¼ xT þ Δx0 =2. The boundary condi-
0 0 0 0 0

tion Eq. (9) gives A3 ¼ 1. The five other coefficients in Eq. (20) are
obtained from the condition Eq. (10), and by requiring that ~ζðx0 Þ
Method of Solution and Results and ~uðx0 Þ ¼ ið1  iδÞ1 ðd~ζ=dx0 Þ be continuous at x012 and at
x023 . These conditions can be expressed in matrix form as
The extractable power with a finite nondimensional wave number
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

and channel area is now considered, and results are compared with AC ¼ D ð21Þ
long-wave theory. This requires the solution of Eq. (8), a variable-
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

coefficient ordinary-differential equation. However, with the boxcar


distribution Eq. (4) for the frictional drag, the coefficient in Eq. (8) where C ¼ ½ A1 B1 A2 B2 B3 T , D ¼ ½ 0 0 cos kLx023
is piecewise constant, and the solution may be written in the form 0 K sin kLx023 T , and

2 3
sin kL þ ðkL=μÞ cos kL cos kL  ðkL=μÞ sin kL 0 0 0
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

6 cos kLx 0
sin kLx 0
 cos k 0
2 Lx12  sin k2 Lx012 0 7
6 12 12 7
A¼6
6 0 0 0
cos k 2 Lx23 sin k2 Lx023  sin kLx023 7
7
4 K sin kLx012 K cos kLx012 K 2 sin k2 Lx012 K 2 cos k 2 Lx012 0 5
0 0 K 2 sin k2 Lx023 K 2 cos k 2 Lx023 0
K cos kLx23

with K ¼ kL=ð1  iδ C Þ and K 2 ¼ k 2 L=½1  iðδ C þ δ 0T Þ. A numeri- to the mouth of the channel. Nearly 40% greater power is available
cal routine based on Gaussian elimination is used to solve Eq. (21) for a turbine fence situated at the open boundary. In contrast, the
to obtain the complex amplitude coefficients of Eq. (20). dependence of the power with λ in long-wave theory Eq. (16) is
Given values for the parameters in Eq. (13) and a position x0T for indistinguishable from the curve given in Fig. 3(a) for the fence
the turbine fence, δ 0T is varied and the nondimensional average located next to the basin.
power dissipation in the turbine fence, P0avg , is evaluated from The 40% linear increase in available power approaching the
Eq. (12) with ~uT ¼ iK 2 ðB2 cos k 2 Lx0T  A2 sin k 2 Lx0T Þ. P0avg de- mouth is associated with a similar linear variation in the magnitude
pends on the turbine drag parameter, λ ¼ Δx0 δ 0T , but is independent of the channel current in the undisturbed state [Fig. 3(c)]. As shown
of the choice of Δx0 . The results presented below are with in the appendix, the elevation amplitude is uniform along the chan-
Δx0 ¼ 0:02, which satisfies the condition Δx0 ≪ 1 for the approxi- nel for long waves, while the magnitude of the flow increases by a
mation Eq. (12) to be valid. In test cases, it was found that virtually factor ð1 þ μÞ between the ends of the channel. This increase is
identical results are obtained with Δx0 ¼ 0:01 and Δx0 ¼ 0:05. The required to accommodate the tidal variation in the volume of the
response is considered over a fairly wide parameter range ð0 < channel. As indicated by Eq. (1), the extractable power in long-
k 0 L ≤ 2; 0 < μ ≤ 1Þ for cases with weak ðδ C ¼ 0:1Þ and strong wave theory is proportional to the volume flux in the undisturbed
ðδ C ¼ 2Þ channel friction. By varying the turbine drag and the state. It is understandable then that the available power should in-
fence position, 0:99 ≤ x0T ≤ 0:01, the maximum extractable crease as the flow rate increases approaching the channel mouth.
power, along with the optimal value of x0T , can be determined The amplitude of the current in long-wave theory, obtained from
for a given set of environmental parameters. Eq. (14), is consistent with the result from the model at x0 ¼ 0:99
in Fig. 3(c). Accordingly, long-wave theory predicts the power
Maximum Extractable Power available for a turbine fence at this location. However, the magni-
Examples of the variation in the extracted power, P0avg , with λ are tude of the flow at the basin is smaller than that at the mouth by a
presented in Fig. 3(a) with the parameters in Eq. (13) assigned the factor of ð1 þ μÞ, and the maximum available power is underesti-
values ðμ; k0 L; δ C Þ ¼ ð0:4; 0:05; 0:1Þ. This choice is outside the mated in long-wave theory by a similar factor.
range of validity of long-wave theory because the relative channel The maximum average extractable power, Pmax ðμ; k0 L; δ C Þ, was
area is not small compared to unity. Consequently, the extractable determined by evaluating Eq. (12) with varying turbine drag param-
power depends on the position of the turbine fence as demonstrated eter, λ, and fence location, x0T . Results are presented in Figs. 4(a)
in Fig. 3(a), which shows the response with the fence placed next and 4(b) as contour plots in ðμ; k0 LÞ space for cases with weak and
to the basin ðx0T ¼ 0:99Þ and at the mouth of the channel strong channel friction, respectively. For small to moderate wave
ðx0T ¼ 0:01Þ. As is generally the case, power dissipation in the numbers ðk 0 L ≤ 0:4Þ, the pattern of the contours in these plots
turbine fence increases with increasing turbine drag, reaches a is qualitatively consistent with long-wave theory [Figs. 2(a) and
maximum at some point, and then declines as larger drag chokes 2(b)]. Figs. 4(c) and 4(d) present the ratio of the maximum extract-
the flow. Fig. 3(b) demonstrates that the maximum extractable able power to that given by long-wave theory, Eq. (18), for these
power increases linearly with x0T from the junction with the basin two cases. As expected, the ratio is close to unity for small μ and

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 67
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

Fig. 4. Maximum average extractable power, Pmax ðμ; k0 L; δ C Þ, in the case of (a) weak ðδ C ¼ 0:1Þ; (b) strong ðδC ¼ 2Þ background channel friction;
(c) and (d) give the ratio of the power in (a) and (b) to that given by long-wave theory, maxðP0LW Þ, for weak and strong friction as presented in
Figs. 2(a) and 2(b), respectively

k 0 L. In particular, for the case considered in Garrett and Cummins converge so that the power becomes effectively independent of
(2005) of a channel connecting two large basins ðμ → 0Þ, long- the relative channel area.
wave theory is accurate to within 10% for k0 L as large as 0.38
(0.45) with weak (strong) channel friction. On the other hand, Optimal Fence Location
long-wave theory may underestimate the extractable power by a
In long-wave theory the maximum extractable power is indepen-
significant factor for larger values of μ and k 0 L.
dent of the position of the turbine fence along the channel. Con-
Of particular interest is the behavior for relatively long waves
(k0 L ≤ 0:1). As anticipated from the discussion of Fig. 3, it is evi- versely, for shorter waves or finite channel area, the available power
dent from Figs. 4(c) and 4(d) that the maximum extractable power depends on x0T . The case illustrated in Fig. 3(b), for example, had a
exceeds the predictions of long-wave theory by a factor ð1 þ μÞ, linear variation in the available power with position along the chan-
with very little dependence on the level of background channel nel; maximum power extraction required that the fence be placed
friction. This suggests a simple correction to Eq. (18) to account near the mouth, where the flow rate also is maximal. With weak
for finite channel area such that channel friction, the available power varies nearly linearly along
the channel for k 0 L ≤ 1:5, and the optimal position of the turbine
1þμ fence is found at one of the ends of the channel. Fig. 6(a) gives the
maxðP0LW Þ ¼ ð22Þ
4ðλP þ δC Þ ratio of the power potential between the ends of the channel as a
function of μ for several values of k 0 L with weak channel friction
with λP given by Eq. (17). This correction is equivalent to a redefi- ðδ C ¼ 0:1Þ. Consistent with long-wave theory, the ratio is close to
nition of the basin area to include the area of the channel. Compar- unity for small values of k 0 L and μ. Away from these limits, the
ing with model solutions, Eq. (22) is found to be accurate to within ratio scales linearly with μ and can deviate significantly from unity
2% for k 0 L ≤ 0:1 and 0 ≤ μ ≤ 1. Thus, one of the main limitations such that the fence position has an important influence on the avail-
of long-wave theory is easily removed by this correction. able power.
The response for larger values of the nondimensional wave In the long-wave limit, the optimal position of the fence is al-
number is presented in Fig. 5, which gives the maximum ways at the mouth where, as noted above, the magnitude of the
extractable power as a function of k 0 L for selected values of relative current is also largest. For shorter waves, the optimal position of
channel area, μ, with weak and strong damping. As seen above, a the turbine fence may lie elsewhere along the channel. Fig. 6(b)
pronounced local peak associated Helmholtz resonance is evident gives the optimal position of the turbine fence in ðμ; k0 LÞ space
in the case of weak channel friction [Fig. 5(a)]. However, at larger with weak channel friction. This shows that there is a transition,
values of k0 L all of the plotted curves for both weak and strong as k0 L increases, such that the optimal position of the fence
friction show a linear increase in the extractable power and tend switches to the opposite end of the channel and lies adjacent to

68 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

Fig. 5. Maximum average extractable power, Pmax ðμ; k0 L; δ C Þ, for selected values of the channel area parameter, μ, with (a) weak ðδC ¼ 0:1Þ;
(b) strong ðδC ¼ 2Þ channel friction
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

Fig. 6. (a) Ratio of the extractable power for a turbine fence located next to the channel mouth ðx0T ¼ 0:01Þ to that for a fence adjacent to the basin
ðx0T ¼ 0:99Þ as a function of relative channel area for several values of the wave number parameter; the channel friction parameter is δ C ¼ 0:1;
(b) optimal position of the turbine fence along the channel as a function of the wave number parameter and relative channel area, with δ C ¼ 0:1;
(c) position of the undisturbed peak current amplitude, x0 , such that j~uðx0 Þj ¼ maxðj~ujÞ, with δ C ¼ 0:1; (d) as in (b), but with δC ¼ 2

the basin. For larger values (1:6 < k 0 L < 2), Fig. 6(b) indicates that small values of relative channel area and moderate values of
the optimal fence position lies within the interior of the channel. k0 L, the optimal position shifts to locations near the junction with
For comparison, Fig. 6(c) shows the location along the channel the basin. These shifts again bear no relation to the peak current
where the current magnitude, j~uj, is maximum in the undisturbed magnitude in the undisturbed state, which is located at the mouth
state. It is evident that no general correspondence exists between over the entire parameter range plotted in Fig. 6(d).
the optimal location of the turbine fence and the peak tidal current.
Disturbance of the Tide
The optimal turbine position with strong channel friction
ðδ C ¼ 2Þ is shown in Fig. 6(d). Over most of the plotted parameter Placement of a turbine farm in a tidal channel may lead to wide
range, the optimal turbine position is at the mouth. However, for range of impacts to the natural environment affecting, for example,

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 69
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

Fig. 7. Contours of R, the ratio of the elevation amplitude in the basin at maximum power to that in the undisturbed state, for (a) weak ðδ C ¼ 0:1Þ;
(b) strong ðδC ¼ 2Þ background channel friction; the dashed-line contours are based on Eq. (19), the ratio in long-wave theory

sediment transport and the benthic ecology. A review of such In the present study, the problem of a tidal channel linking the
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

potential impacts is provided by Boehlert and Gill (2010). Tidal ocean to a bay is reconsidered in a linear one-dimensional model
motions will also be affected, both immediately within the channel that allows for divergence of the flow. For the problem to remain
and basin and possibly in the far field as well (e.g., Polagye et al. analytically tractable, the channel is assumed to be of constant
2009). For the configuration of Fig. 1, power extraction must lead width and depth. As in previous analytical studies, the back effect
to a reduction of the volume transport in the channel, which will of power extraction on the tidal amplitude at the channel mouth is
affect the tidal range in the basin. neglected. The analysis shows that additional nondimensional
As mentioned previously, a simple measure of the disturbance to parameters arise as flow divergence is admitted. In particular,
the physical environment is the ratio, R, of the tidal range in the the extractable power now depends on the relative position of a
basin at peak extraction to the tidal range in the undisturbed natural turbine fence in the channel. In addition, parameters measuring
state. This ratio is presented in Fig. 7 for cases with weak and the relative channel area, μ, and length, k 0 L, appear independently
strong channel friction and compared with Eq. (19). In accordance rather than through a composite parameter.
with theory, the basin tidal range is reduced for long waves by a The results show that the extractable power is in excellent agree-
factor of approximately 21=2, independent of the relative channel ment with long-wave theory, provided that ðμ; k0 LÞ ≪ 1. For larger
area. Also consistent with long-wave theory, the perturbation to the values of these parameters, the available power may vary from the
basin tide at maximum power remains bound such that 0:5 ≤ R ≤ predictions of long-wave theory by an appreciable factor. The case
21=2 over the parameter range shown in Fig. 7. In fact, this bound of long waves ðk 0 L ≤ 0:1Þ is of particular practical interest. In this
on the disturbance of the basin tide is also found to hold for 1 ≤ limit, the results show that maximum power extraction requires that
k 0 L ≤ 2 (not shown). Overall, the results of Fig. 7 demonstrate that the turbine fence be placed at the channel mouth. The power in the
long-wave theory provides a remarkably good description of the long-wave limit is found to scale with channel area as ð1 þ μÞ. This
disturbance to the tidal range of the basin and the flow in the chan- provides a simple correction that removes one of the constraints
nel over a wide parameter range, particularly for strong background imposed by long-wave theory. Since this correction is just a means
friction.
of accounting for tidal variation of the channel volume, it does not
depend on the level of frictional damping in the channel. Moreover,
although this scaling is based on results with linear channel friction,
Summary and Conclusions
it may be expected also to hold for a quadratic drag law. Accord-
The prospect of turbine farms deployed in the ocean has led to the ingly, the results indicate that application of Eq. (1) to estimate the
preparation of resource assessments that attempt to place upper lim- extractable power should be made on the basis of the volume flux at
its on the available power. As the methodology associated with the channel mouth. If it is more practical to estimate the volume
such assessments has been fraught with uncertainty, recent studies flux into the basin, for example, by using tide gauge data from in-
have sought to establish a theoretical foundation for estimating the side the basin, then Eq. (1) should be scaled by ð1 þ μÞ. This leaves
maximum extractable power. A central result from such studies is Eq. (1) unchanged for the case considered in Garrett and Cummins
Eq. (1), which provides an estimate of the power that is available (2005) of two large basins connected by a relatively short channel.
from tidal motions in a channel linking the ocean to a bay or linking For larger values of the wave number parameter, long-wave
two large basins. This result is quite general and allows for variable theory and Eq. (1) are inapplicable. The utility of the present model
channel cross-sectional and quadratic drag. However, a key is also limited; due to its idealized nature it cannot be applied
assumption is made that the flow in the channel is nondivergent. directly to provide an accurate estimate of the available power
This places limits on the dimensions of the channel, which is re- in realistic settings. Nevertheless, the results indicate that the dis-
quired to be relatively short and of small area. An immediate con- turbance to the tide at maximum power is reasonably predicted by
sequence of assuming nondivergent flow is that the extractable long-wave theory, well beyond its formal range of validity. In ad-
power becomes entirely independent of the location of turbine dition, away from the long-wave limit, the present results demon-
fences along the channel. strate that no simple relation exists between the optimal location for

70 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
a turbine fence and the channel currents in the undisturbed state neglects tidal modulation of the channel volume and accounts only
[e.g., Fig. 6]. Thus, unless the waves are long, it may be difficult for variation of the basin volume. This is valid, provided that the
to determine from an inspection of the natural tidal flow where channel area is small compared to that of the basin.
along a channel the greatest power may be obtained. In such sit-
uations, experimentation with a detailed hydrodynamic model
(e.g., Sutherland et al. 2007; Karsten et al. 2008; Walkington
References
and Burrows 2009) may be the most effective means of establishing
the optimal location of a turbine fence and estimating the extract- ABP Marine Environmental Research Ltd. (2008). “Atlas of UK marine
able power. renewable energy resources: Technical report.” Rep. No. R.1432,
Southhampton, UK.
Bedard, R., Jacobson, P. T., Previsic, M., Musial, W., and Varley, R. (2010).
Appendix. Limits for Nondivergent Flow “An overview of ocean renewable energy technologies.” Oceanography,
23(2), 22–31.
No other uses without permission. Copyright (c) 2012. American Society of Civil Engineers. All rights reserved.

The conditions that must be satisfied for the flow to be nondiver- Blanchfield, J., Garrett, C., Wild, P., and Rowe, A. (2008a). “The extract-
gent in a tidal channel connecting two large basins were discussed able power from a channel linking a bay to the open ocean.” Proc. Inst.
Downloaded from ascelibrary.org by Sardar Vallabhbhai Natl Inst. on 07/26/12. For personal use only.

by Vennell (1998). Here, a uniform channel connecting a basin of Mech. Eng., Part A, 222(3), 289–297.
finite size to the open ocean is considered, omitting the effects of Blanchfield, J., Garrett, C., Rowe, A., and Wild, P. (2008b). “Tidal stream
advection and rotation that were included in Vennell (1998). In the power resource assessment for Masset Sound, Haida Gwaii.” Proc. Inst.
undisturbed state with no turbine drag, Eq. (8) may be solved sub- Mech. Eng., Part A, 222(5), 485–492.
ject to Eqs. (9) and (10). This yields the following expressions for Boehlert, G. W., and Gill, A. B. (2010). “Environmental and ecological
effects of ocean renewable energy development: A current synthesis.”
the elevation amplitude and channel current (Cummins et al. 2010):
Oceanography, 23(2), 68–81.
1 Cornett, A. (2006). “Inventory of Canada’s marine renewable energy
~ζðx0 Þ ¼ cos kLx0  sin kL þ μ kL cos kL sin kLx0 ð23a Þ resources.” Technical Rep. CHC-TR-041, Canadian Hydraulics Centre,
cos kL  μ1 kL sin kL
J. Waterway, Port, Coastal, Ocean Eng. 2012.138:63-71.

National Research Council, Ottawa, ON, Canada.


  Cummins, P. F., Karsten, R. H., and Arbic, B. K. (2010). “The semi-diurnal
kL sin kL þ μ1 kL cos kL tide in Hudson Strait as a resonant channel oscillation.” Atmos.-Ocean,
~uðx0 Þ ¼ i sin kLx0 þ cos kLx 0
ð1  iδ C Þ cos kL  μ1 kL sin kL 48(3), 163–176.
Garrett, C., and Cummins, P. (2004). “Generating power from tidal
ð23b Þ
currents.” J. Waterw. Port, Coastal, Ocean Eng., 130(3), 114–118.
with kL ¼ k0 Lð1  iδ C Þ1=2 . Assuming finite μ, as k0 L → 0 these Garrett, C., and Cummins, P. (2005). “The power potential of tidal currents
in channels.” Proc. R. Soc. A, 461(2060), 2563–2572.
expressions may be approximated as
Garrett, C., and Cummins, P. (2007). “The efficiency of a turbine in a tidal
1 0 channel.” J. Fluid Mech., 588, 243–251.
~ζðx0 Þ ≈ 1  ðkLÞ2 ð1 þ μ Þx ≈ 1 ð24a Þ Garrett, C., and Greenberg, D. (1977). “Predicting changes in tidal regime:
1
1  μ ðkLÞ2 The open boundary problem.” J. Phys. Oceanogr., 7(2), 171–181.
  Hammons, T. J. (1993). “Tidal power.” Proc. IEEE, 81(3), 419–433.
kL kLð1 þ μ1 Þ
~uðx0 Þ ≈ i kLx0 þ Karsten, R. H., McMillan, J. M., Lickley, M. J., and Haynes, R. D. (2008).
ð1  iδ C Þ 1  μ1 ðkLÞ2 “Assessment of tidal current energy in the Minas Passage, Bay of
ðkLÞ2 Fundy.” Proc. Inst. Mech. Eng., Part A, 222(5), 493–507.
≈ i ½x0 þ ð1 þ μ1 Þ ð24b Þ Polagye, B., Kawase, M., and Malte, P. (2009). “In-stream tidal energy
ð1  iδ C Þ potential of Puget Sound, Washington.” Proc. Inst. Mech. Eng., Part A,
223(5), 571–587.
Thus, in the long-wave limit, the elevation amplitude tends to
Sutherland, G., Foreman, M., and Garrett, C. (2007). “Tidal current energy
become uniform while the current magnitude varies linearly be- assessment for Johnstone Strait, Vancouver Island.” Proc. Inst. Mech.
tween the basin and the mouth. The ratio of the current (and hence Eng., Part A, 221(2), 147–157.
the ratio of the volume flux) at x0 ¼ 0 to that at x0 ¼ 1 is Vennell, R. (1998). “Oscillating barotropic currents along short channels.”
~uð0Þ=~uð1Þ ¼ 1 þ μ. Thus, for the volume flux to be approxi- J. Phys. Oceanogr., 28(8), 1561–1569.
mately constant along the channel, it is necessary that μ ≪ 1. Walkington, I., and Burrows, R. (2009). “Modelling tidal stream power
By assuming that the transport is nondivergent, long-wave theory potential.” Appl. Ocean Res., 31(4), 239–245.

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 71

You might also like