Vous êtes sur la page 1sur 8

Renewable Energy 47 (2012) 95e102

Contents lists available at SciVerse ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Realizing the potential of tidal currents and the efciency of turbine farms in a channel
Ross Vennell
Ocean Physics Group, Department of Marine Science, University of Otago, 310 Castle St, Dunedin 9054, New Zealand

a r t i c l e i n f o
Article history: Received 2 December 2010 Accepted 29 March 2012 Available online 11 May 2012 Keywords: Tidal stream power Tidal turbine Farm efciency Power from tidal channels

a b s t r a c t
Tidal turbines in strong ows have the potential to produce signicant power. However, not all of this potential can be realized when gaps between turbines are required to allow navigation along a channel. A review of recent works is used to estimate the scale of farm required to realize a signicant fraction of a channels potential. These works provide the rst physically coherent approach to estimating the maximum power output from a given number of turbines in a channel. The fraction of the potential realizable from a number of turbines, a farms uid dynamic efciency, is constrained by how much of the channels cross-section the turbines are permitted to occupy and an environmentally acceptable ow speed reduction. Farm efciency increases as optimally tuned turbines are added to its cross-section, while output per turbine increases in tidal straits and decreases in shallow channels. Adding rows of optimally tuned turbines also increases farm efciency, but with a diminishing return on additional rows. The diminishing return and ow reduction are strongly inuenced by how much of the cross-section can be occupied and the dynamical balance of the undisturbed channel. Estimates for two example channels show that realizing much of the MW potential of shallow channels may well be possible with existing turbines. However unless high blockage ratios are possible, it will be more difcult to realize the proportionately larger potential of tidal straits until larger turbines with a lower optimum operating velocity are developed. 2012 Elsevier Ltd. All rights reserved.

1. Introduction Underwater turbines in strong tidal currents can contribute to the need for renewable energy sources. To make a signicant contribution many turbines must be grouped into large tidal farms to exploit the energy of high ows through narrow tidal channels. This leads to a critical question, how does power production scale as a turbine farm expands from a single turbine into a large energy farm? While adding turbines to a farm increases its generation capacity, power extraction also enhances drag on the ow along the channel. This enhanced drag slows ows along the channel, reducing power production by all turbines. Flow reduction becomes more signicant as a farm grows and ultimately limits a channels potential to produce power ([1], hereafter GC05). So it is the combined effects of ow reduction and installed generation capacity which determines how much power an expanding farm produces. Part of addressing the question of how power production scales is estimating the maximum power a particular channel can
E-mail address: ross.vennell@otago.ac.nz. 0960-1481/$ e see front matter 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2012.03.036

generate, i.e. the channels potential (see review [2]). Most recent estimates of potential have used the approach developed by GC05 (e.g. [3]). To realize that potential requires the turbines to occupy the entire cross-section of the channel, have perfect electromechanical efciency, that there are no losses to mixing behind the turbines and that there is no drag on any structure supporting the turbines. Realistically none of these are achievable, however GC05s potential provides a useful upper bound for production. The need for navigation along a channel will often restrict turbines to only part of the channels cross-section, Fig. 1. Thus just as important as estimating a channels potential is to determine how much of a channels potential can be realized from a given number of turbines when some ow can bypass the turbines through gaps left to permit navigation, Fig. 2. In addition, ow reduction may also have environmental impacts, such as reduced tidal exchange and sediment transport (e.g. [4]). Thus another constraint on turbine numbers is likely to be an acceptable ow speed reduction. This requires answering the question, how does power production scale with farm size within constraints on how much of the cross-section can be lled with turbines and how much ows within the channel can be reduced?

96

R. Vennell / Renewable Energy 47 (2012) 95e102

Fig. 1. Schematic of a turbine farm in a narrow constricted channel connecting two large water bodies. Differing tidal regimes in the two large water bodies drive oscillating tidal ow through the channel. The example farm has 3 rows of turbines. The arrows around each turbine indicate the stronger ows passing around the turbines and the weaker ows passing through the turbines.

accessible to a wider audience and then use the results to discuss the number of turbines required to realize a signicant fraction of the potential of two example channels. This work begins by reviewing the physics of tidal channel potential and turbine efciency contained in the works by outlining four key concepts in Section 2. Understanding the physics of farm efciency hinges on understanding how the farms gross drag coefcient affects the ow along the channel. Section 3 looks at the efciency of a single row of turbines. Section 4 examines the effects of adding more rows. Section 5 looks in detail at how many turbines are required to achieve a signicant fraction of the example channels potentials when constrained by cross-sectional occupancy and ow reduction. 2. Background physics The physics underlying the latter sections are reviewed in the next four sub-sections, each of which presents an essential concept. The underlying aim of this section is to make the connection between the farms drag coefcient and the power which is available for electricity production clear. The farms drag coefcient is the link between the number of turbines in the farm and the power available for electricity production. The essential idea is that the farms drag coefcient increases as the turbines ll more of the cross-section, i.e. the blockage ratio increases, or as more rows of turbines are added to the farm. The drag coefcient also changes as the ow through the turbines is adjusted by tuning the pitch of their blades. Channel and farm specic tuning is critical to maxmising the farms output, V10. The underlying models are those of GC05 and GC07 adapted by V10 and V11. In GC05s model for a turbine farm in a short narrow channel, Fig. 1, the farm is modelled as a drag on the ow. The model has oscillating tidal ow driven along the channel by a water level difference between the ends of the channel. This difference, or headloss, is due to the differing tidal regimes in the two large water bodies which are connected by the channel. The water bodies are assumed to be so large that any water owing through the channel does not affect water levels within them. Thus water levels at the ends of the channel are unaffected by a power extraction within it. This is the simplest useful channel geometry. One extension, not included here, has a lagoon at one end of the channel and a large ocean at the other [9]. The nite reservoir of the lagoon means that tides within it depend on the volume of water which ows through the channel. As a result a lagoon can increase or decrease a tidal channels potential depending on whether the amplitude of the head between the ends of the channel is less than or greater than the tidal amplitude in the ocean [10]. In addition any large deep ocean will likely have a shallow continental shelf between it and the entrance to the channel. Frictional dissipation and resonances over the shelf may also inuence the amplitudes of the tides at the entrance to the channel driven by tides in the deep ocean [11]. The GC05 model is given in terms of volume transport, here it is presented in terms of velocity. In short uniform cross-section channels the cross-sectional average velocity does not vary signicantly along the channel [12,13]. Thus the tidal velocity everywhere along a short channel with a rectangular cross-section depends only on time and can be expressed as u u0 sinut fu , where u0 is the amplitude of the velocity and fu its phase. GC05s momentum balance for a uniform cross-section channel can be written in the form

An older method for estimating a channels potential was to use the Kinetic Energy Flux through the undisturbed channel and an arbitrary loading factor of 10e15% to estimate power output from the channel [2]. However the KE ux through a channel without turbines is unrelated to a channels potential as it does not account for the ow reduction due to power extraction [5]. Also, the KE ux does not provides a means to estimate the power available from a given number of turbines which may occupy only part of the cross-section, nor a means to estimate the ow reduction. Both these aspects are critical in assessing the tidal current resource at a particular site, the farms impacts and the economics of developing it. Building on GC05 and GC07 the recent [6,7] (hereafter V10 and V11) works provide the rst physically coherent approach to maximizing the power output from a given number of turbines. The upper bound for the power available for electricity production given by these models can be used to answer the question about how power production scales within constraints, as well as how best to arrange and congure the turbines. In this work a review of GC05 [8], (hereafter GC07), V10 and V11, is used to discuss the question and indicate the scale of farm required to realize a signicant fraction of a channels potential. How much of a channels potential can be realized, i.e. farm efciency, is strongly linked to turbine efciency. Not only the turbines electro-mechanical energy conversion efciency, but also the turbines uid dynamic efciency. This work focuses on uid dynamic efciency, thus addressing the power available for electricity production from a farm with a given number of turbines. Turbines extract energy from the ow passing through the area spanned by their blades. To maximize turbine efciency all turbines must have the strength of the ow passing through them adjusted or tuned, typically achieved by varying blade pitch. To maximize farm output the ow speed through the turbines must be tuned for a particular channel and fraction of the channels cross-section taken up by the turbines, as well as tuned in relation to each other (V10 and V11). This makes maximizing the uid dynamic efciency of a farm complicated. The four fundamental works, GC05, GC07, V10 and V11, are necessarily mathematical and complex. This work aims to rstly review the essential results from these works to make them

  vu g z0 C C u u sinut D F vt L h L
Fig. 2. Schematic of ow through a row of identical turbines and the ow through and around a single turbine within the row. Relative sizes of the velocities are u4  u  u1  u3 . After V11.

(1)

In Eq. (1) the rst term represents the inertia of the ow, the second term the sinusoidal pressure gradient, or head, which forces

R. Vennell / Renewable Energy 47 (2012) 95e102

97

ow along the channel and the last term the drags on the ow. A is the cross-sectional area of the channel, L its length, h its depth and z0 is the amplitude of the oscillating headloss between the two large water bodies connected by the channel and u the angular frequency of the tide. The extension of the above to a variable crosssection channel is straightforward, but not done here in order to keep the presentation simpler, V11. In Eq. (1) the drag on the ow is made up of two components, the rst is due to background bottom friction, which has a drag coefcient of CD based on the channels horizontal area. The second component of drag is due to power extraction by the farm, associated with the farms gross drag coefcient CF based on crosssectional area. The forced due to power extraction from a uniform cross-section channel can be expressed as

Table 1 Tidal power potential and power available for two high ow hypothetical example channels without a constriction. More details on the examples are given in V10. Turbine numbers and power available per turbine are based on the cross-sectional area occupied by the turbine, AT, of 400 m2. A cross-sectional occupancy of 0.1 was chosen to represent a low turbine cross-sectional occupancy, while 0:5 was chosen to represent what may be an unrealistically high occupancy. Shallow channel Tidal strait Length, L Depth, h Width, W Undisturbed velocity amplitude, u0UD Head loss amplitude, z0 peak Farm drag coefcient for peak power lost, CF Velocity amplitude at peak, upeak 0 Flow reduction at peak, upeak =u0UD 0 Tidal current potential, P lost Turbines to ll cross-section, N0 A=AT max Potential per turbine, P lost =N0 Power available from rst turbine installed Farm with 0:1 Power available Farm efciency Number of rows Number of optimally tuned turbines Average power available per turbine Reduced velocity/Relative ow Optimal tunings r1/r3 Farm with 0:5 Power available Farm efciency Number of rows Number of optimally tuned turbines Average power available per turbine Reduced velocity/Relative ow Optimal tunings r1/r3
max

2 km 20 m 0.5 km 2.7 m s1 0.17 m 0.57 1.5 m s1 0.57 9 MW 25 0.36 MW 0.99 MW 4.6 MW 0.52 3 8 0.62 MW 2.2 m s1/0.82 0.73/0.5 7.2 MW 0.8 1 13 0.58 MW 1.8 m s1/0.68 0.84/0.73

50 km 100 m 10 km 1.9 m s1 1.4 m 8.5 1.2 m s1 0.61 6 GW 2500 2.3 MW 0.34 MW 2.7 GW 0.48 41 10,000 0.27 MW 1.6 m s1/0.87 0.70/0.46 3.0 GW 0.54 4 5000 0.61 MW 1.5 m s1/0.82 0.67/0.54

F rCF Ajuju

(2)

where r is the uid density. For multi-row farms the gross drag coefcient can be expressed as CF NR CR , where NR is the number of rows of turbines and CR is the drag coefcient of a single row. Understanding how power production from tidal farms scales with farm size hinges on understanding how the farms gross drag coefcient, CF, is connected to the number of turbines in the farm, how they are arranged in rows and their conguration as outlined in Sections 2.3 and 2.4. The power lost by the ow to the farm with multiple rows is simply the total turbine force, Eq. (2), times the velocity through the rows of turbines, Fu, which averaged over a tidal cycle is

P lost

4 rC Au3 3p F 0

(3)

To aid discussion two hypothetical examples of high ow channels were created, Table 1. The rst is a shallow 20 m deep entrance to a large embayment. This example has half the length, width and depth, but with similar velocities to that used in [14]s model to look at the effect of modest energy extractions on channel ows. The second is a 100 m deep tidal strait connecting two water bodies with tides in anti-phase, a smaller and higher ow version of Cook Strait, NZ which has cross-sectionally averaged peak ows closer to 1.1 m s1 [10,15]. These examples are chosen to be dynamically representative of large and small tidal channels. The dynamical balance of a channel without turbines is indicated by the size of the rescaled bottom friction drag coefcient l0 gz0 CD =hLu2 , V10. For the tidal strait example l0 0:35, thus the ows inertia is around three times more important than bottom friction in Eq. (1). For the shallow channel l0 4:8, thus drag dominates the shallow channels dynamics, even before any turbines are installed. Thus the shallow channel is in a quasi-steady state dynamical balance with the forcing due to the headloss between the ends. To estimate the number of turbines required in the farms discussed in this work the only turbine specic information required is the blade area. For this work a nominal blade area of 400 m2 will be used to estimate turbine numbers. This is similar to the blade area of the largest operating tidal turbine which has a pair of 16 m diameter blades [16]. In discussing the power available to a farm no allowance is made for energy losses due to drag on the turbines support structure [17], or that a minimum ow speed may be required to generate power, or that power output may be capped at high ows to limit structural loads. These turbine specic characteristics and less than perfect electro-mechanical efciency will result in electricity production being less than the power available gures discussed here. The largest operating turbine generates 1.2 MW at a free stream steady velocity of 2.25 m s1, with an electro-mechanical efciency for converting power available into

electricity near 90% [16]. If similar high efciency turbines could be installed in the example channels then power outputs are likely be near the power available gures discussed here. The power available for electricity production is determined by the farms uid dynamic efciency. 2.1. Increasing a farms drag coefcient by installing turbines in a narrow channel reduces ow speeds along the entire channel, GC05 Fluid forces on a turbine rotate its blades which generates power. Viewed from the ows perspective, the forces it exerts on the turbine are a drag force on the uid. The turbine drag force is given by Eq. (2), which depends on the farms gross drag coefcient CF. Increasing CF by adding rows or increasing the drag coefcient of a row, CR, increases the total drag within the channel leading to a reduction in ow speeds along the entire short narrow channel, which is inherent in the dynamical Eq. (1). The ow reduction is seen in the approximate analytic solution to this non-linear equation (V10 equation (A3)) and in Fig. 3a). 2.2. The potential of a tidal channel, an upper bound for power production, GC05 The power lost by the ow due to power extraction, Eq. (3), is plotted in Fig. 3b) for the two examples as the upper thick solid and dashed lines, which almost overlap. Conceptually the power lost to the farm is proportional to the product CF u3 0 , Eq. (3). Increasing a small farms gross drag coefcient by adding turbines creates little ow speed reduction, Fig. 3a), thus the power given by the product initially increases as a farm expands due to growing CF, Fig. 3b). A very large farm leads to signicant ow reduction, as a result

98
0 undisturbed

R. Vennell / Renewable Energy 47 (2012) 95e102

a
Power/ P
max

1 0.8 0.6 0.4 0.2

Flow reduction, U /U

0.8

0.6

0.4 0

0.5 1 CF / CFpeak

1.5

0 0

Decreasing tuning parameter, r3 >

0.5 1 CF / CFpeak

1.5

Fig. 3. Effects of increasing farms drag coefcient for two example channels in Table 1. Solid lines are for shallow channel and dashed lines for tidal strait. Horizontal axis is the farm drag coefcient relative to the drag coefcient required to realize a channels potential, i.e. at the peak in the power lost curve. a) Reduction in ow relative to ow in undisturbed channel. b) Thick lines are the power lost to the turbines and thin lines are the power available from 5 rows of turbines in the shallow channel and 40 rows in the tidal strait when turbines occupy 20% of the channels cross-section. Solid dots show peak in power available curves at optimal tuning, i.e. the upper bound for how much power is available from these turbines.

adding turbines decreases the power lost to the turbines i.e. CF u3 0 decreases as the benet of growing the farm to increase CF is outweighed by the reduction in u0 due to the additional drag. Between these two extremes there is a maximum power which can be lost by the ow to the turbines (GC05) as seen in the power lost curves max in Fig. 3b). This maximum power lost to the turbines, P lost is a channels potential, which is the upper bound for how much power a channel can generate. The optimal gross farm drag coefcient CF peak required to realize the potential can be estimated from the analytic solution to GC05s model using (V10-A4) or (V11max 2.10). To realize a channels potential, P lost , turbines must occupy the entire cross-section. Were it possible around 25 turbines would ll the cross-section of the shallow channel and 2500 the crosssection of the tidal strait, Table 1. Many turbines sweep a circular area, thus cannot ll a rectangular channels cross-section unless contained within a square shroud. Thus the number of turbines required to ll the cross-section, N0, may not always be of practical use. However N0 is a useful as a reference value for the size of a farm in later discussions on turbine numbers where they only ll part of the cross-section or are spread amongst rows. GC05 found a remarkably simple approximate expression for estimating the potential of a channel based on the transport amplitude through the undisturbed channel, U0UD Au0 , and the amplitude of the headloss, z0 , which is given by

reduction in the wake behind the turbines, i.e. they found CR ; r3 . Though the functional relationship is complex and given in equations (GC07-2.23, V10-2.6) and (GC07-2.9, V10-2.7), for the purpose of this work it is only essential to understand that the drag coefcient of a row only varies due to the changes in the blockage ratio or changes in the ow reduction behind the turbines. CR ; r3 increases as either is increased or the ow in the wake, r3, is decreased. The fraction of the cross-section occupied by the turbines, i.e. the blockage ratio, is simply

AT A

where M is the number of turbines in the row, AT is area blocked by the rotating by the blades of a single turbine. The average power lost by the ow is given by Eq. (3). This is not the same as the power available to the turbine for electricity production due to mixing losses behind the turbines, GC05 and [19]. Using their results the power available is the smaller work done by the ow through the turbine, Fu1, whose average is

P avail

4 rAr1 NR CR u3 0 3p

(5)

P lost 0:22rg z0 U0UD

max

(4)

[10] gives a method to estimate both the potential and the ow speed reduction which only requires the transport along the undisturbed channel, which is easily measured using a vessel mounted ADCP, e.g. [18]. 2.3. Gaps between turbines allows ow to bypass turbines reducing the power available for power production, GC07 When there are wide gaps between turbines within a row to allow for navigation some ow bypasses the turbines altogether and does not contribute to power production. The mixing of the retarded ow passing through the turbines with the faster ow passing around the turbines also dissipates some of the ows energy as heat, [19] and GC07. The ows near a row of turbines are shown schematically in Fig. 2. GC07 extended classic Lanchester-Betz actuator disc theory for an isolated turbine [20,21] to a row of turbines in a narrow channel, Fig. 2. They found that the rows drag coefcient, CR, depends only on the blockage ratio , the fraction of the channels cross-sectional area occupied by the turbines, and the ratio r3 u3 =u, which quanties the ow

where r1 u1 =u is the ratio of the velocity though the turbines to the velocity upstream of the row. Comparing Eqs. (3) and (5) shows that due to mixing losses only the fraction r1 of the power lost by the ow to the turbines is available for electricity production, where 0  r1  1. The power available Eq. (5) represents an upper bound for electricity which could be produced from a farm in a channel. 2.4. Turbines must be adjusted or tuned for a particular channel and in relation to each other to maximize the power available, V10 & V11 Maximizing the power lost by the ow due to the farm, Eq. (3), is not the same as maximizing the power available for generation as they differ by a factor of r1, Eq. (5). Fig. 3b) demonstrates this, where the power lost curve peaks at CF peak, while the lower power available curves peak at a smaller drag coefcient CF opt indicated by the dots. To maximize the power available the ow through the turbines must be adjusted or tuned to give the optimal farm drag coefcient CF opt. Typically tuning is done by adjusting the pitch of the turbines blades. Here this tuning is done mathematically by adjusting the value of the ow reduction, r3, which affects both CR and r1, GC07. GC07s extension of Lanchester-Betz theory to turbines in a channel assumed the ow upstream of the row, u, in Fig. 2 was

R. Vennell / Renewable Energy 47 (2012) 95e102

99

xed and found an optimal value of r3 1=3 (equivalent to r1 2=3) which maximised the power available, the same optimal values as those for an isolated Lanchester-Betz turbine. They also found that at most 2/3 of the power lost to the turbines was available for power production, which occurred if the turbines took up the minimum possible fraction of the cross-section. However in a channel with ows driven by head loss between its ends the ow upstream of the row, u, is not xed but decreases as the farms gross drag coefcient, CF NR CR ; r3 , increases. Thus CF depends on the tuning r3. Consequently changing the tuning changes the strength of the ow along the entire channel via its effect on the gross drag coefcient, V10. V10 went on to show that a consequence of this is that turbines need r3 to be tuned to values between 1/3 and 1 to maximize the power available. Thus tuning a large tidal farm is very different from tuning a single isolated device. A farms opt , depends on a channels geometry and optimal tuning, r3 dynamical balance, as well as the blockage ratio, . V10 found that by optimally tuning turbines it is possible to exceed GC07s maximum of 2/3 of a channels potential which is available for power production. V11 went on to show that not only does a row of turbines need to be tuned for a particular channel it must also be tuned in the presence of other rows to maximize the power available. The need to tune rows in-concert has implications for modellers who must include idealized turbines in their hydrodynamic models to assess the power available from a proposed site, V11, and for the operators of turbine farms as turbines come in and out of service. Tuning in-concert may require many model runs or complex interdependent adjustment of operating turbines to nd the optimal set of turbine tunings. The need to tune turbines in-concert arises because rows of turbines in narrow channels interact with each other via the farms gross drag coefcient, CF, even if they are separated widely enough for the disturbed ow through one row to fully mix before encountering the next row, as CF affects ow along the entire channel. 3. Farm and turbine efciency for a single row 3.1. Farm uid dynamic efciency A measure of a farms efciency is the fraction of GC05s potential which is available for power production, V10, i.e.

FE

P avail P lost
max

(6)

Fig. 4a) shows FE for the two example channels as optimally tuned turbines are added to a single row. The curves for both examples converge on 1 as turbines take up most of the crosssection, making most of the channels potential available for power production. For the shallow channel farm efciency initially grows rapidly as turbines are added, with reduced gains at higher numbers, indicating a diminishing return on additional turbines. In contrast for the tidal strait FE increases more rapidly at the higher occupancies, indicating an increasing return on new turbines added to the row, V11. Though a farm in the shallow channel is more efcient than the tidal strait for a given fraction of the cross-section occupied, in absolute terms the tidal strait has a much larger power available and requires many more turbines to make up this given fraction. The thin lines in Fig. 4a) show the reduction in ow speeds due to adding turbines to the row. Flow in the shallow channel decreases more rapidly as turbines are added, consistent with its higher farm efciency. Fig. 4a) shows the rst of two extreme ways to almost realise a channels potential, i.e. approach 100% farm efciency, is to have the optimally tuned turbines ll the cross-section, i.e. blockage ratio /1. Turbines are normally thought of as extracting energy from the ows Kinetic Energy by reducing ows through the turbines. However, paradoxically, an optimally tuned tidal farm can almost realise a channels potential without reducing the ow through the turbines relative to the ow upstream, i.e. the optimal r3 and r1 /1 as /1, V10. The resolution lies in understanding the source of the farms energy. For an isolated turbine generation is a result of reducing ows through the turbine, where for a Betz turbine the optimal tuning is r3 1=3, corresponding to r1 2=3. As the cross-section is lled with turbines the farms energy source changes gradually from the ows KE to the potential energy of the ow as optimal tunings increase. So that at high blockage ratios the energy source becomes the drop in water level between the upstream and downstream sides of the farm, i.e. the source is the headloss across the farm. It is also worth emphasizing that mixing losses behind the turbines approach zero at high blockage ratios, which must happen if farm efciency is to approach 100%. At high blockage ratios turbine farms approach an extreme form of a hydroelectric dam with low head and high volume ow. For the two examples the head loss is only around 0:1 1:0 m, while the peak volume ows are 27; 000 1; 700; 000 m3 s1 . In contrast a large river dam has high head and low ow, i.e. a 100 m head and 500 m3 s1 . 3.2. Power available per turbine Fig. 4b) gives the power available per turbine as the crosssections are lled with turbines. A single isolated turbine, /0, in the shallow channel makes 0.99 MW available per turbine,

FE is the farms uid dynamic power efciency and is maximised opt at the optimal tuning, r3 , V11. This is the headline efciency, which is the upper bound for the fraction of a channels estimated potential that can be turned into electricity from a given number of turbines.

a
0UD

or U /U

0.8 0.6 0.4 0.2 0 0


Fa

redu

ctio

Power per turbine, MW

b
Flow

2 1.5 1 0.5 0

max

/P

avail

rm

c ffi

ien

cy

0.2 0.4 0.6 0.8 1 Fraction of crosssection occupied,

0.2 0.4 0.6 0.8 1 Fraction of crosssection occupied,

Fig. 4. Effect of the fraction of cross-section occupied by the turbines, or blockage ratio, on efciencies and ow speeds for a single row of optimally tuned turbines in a uniform cross-section channel. Solid curves are for shallow channel and dashed lines for tidal strait examples. a) Thick lines are farm efciency Eq. (6), the fraction of GC05s potential which is available for power production, and thin lines ow relative to undisturbed channel. b) Power available per turbine in MW.

100

R. Vennell / Renewable Energy 47 (2012) 95e102

whereas only 0.34 MW is available in the tidal strait due to much weaker ows. However as the cross-section is lled the power available per turbine for the shallow channel decreases, while for the tidal strait power per turbine increases. So paradoxically at higher occupancies the tidal strait delivers up to six times more per turbine, despite having much weaker ows than the shallow channel. The curves in Fig. 4b) are the result of two competing effects. Firstly, occupying more of the cross-section increases a rows drag coefcient, CR , increases the power available Eq. (5). Secondly, increasing a rows drag coefcient reduces the ow along the channel, U0, which acts to decrease the power available in Eq. (5). The net effect differs for the two examples. For the shallow channel the more rapid reduction in ow speed as the cross-section is lled, Fig. 4a), outweighs the enhanced drag coefcient, leading to reduction in the power per turbine from 0.99 MW to only 0.36 MW at high occupancy. In contrast the more gradual ow reduction in the tidal strait leads to an increasing power available per turbine as the cross-section is lled from 0.34 MW to 2.3 MW. A result is that above 35% occupancy the tidal strait delivers more per turbine than the shallow channel despite the much weaker ows in the strait. The differing performance of the turbines as the cross-section is lled is linked to their differing dynamical balances. At higher occupancies in the shallow channel the power lost to the turbines is similar to that lost to background friction (this can be inferred from (V10-A4) where CF peak is around twice the scaled background bottom friction coefcient CD for near steady state channels, i.e. large l0 channels). This signicant energy loss to bottom friction and turbine drag is associated with a more rapid decrease in ows, a diminishing return on new turbines and the decreasing power available per turbine in Fig. 4. In contrast in the tidal strait bottom friction is almost unimportant and relatively little energy is lost to bottom friction. This gives the strait a proportionately higher potential and an increasing return on additional turbines as they become more efcient when occupying more of the cross-section of the duct formed by the channel. The energetics of channels with turbine farms is discussed in detail in [17]. 4. Efciency of multi-row farms Adding capacity to a farm by adding rows of optimally tuned turbines increases farm efciency as demonstrated in Fig. 5a). For the tidal strait 15 rows makes only 40% of it potential available, so as would be expected, the associated ow reduction is modest. In contrast for the shallow channel 3 rows make 70% of its potential available, with a signicant associated ow reduction. Fig. 5 illustrates a second extreme way to realize most of a channels potential, having a large number of rows. However, for

both channels there is a diminishing return on additional rows. Farm efciency peaks at a very large number of rows, indicating an optimal farm size. However, the diminishing return on new rows is so harsh near this peak that farm size will be restricted to a much smaller of number rows, [17]. V11 showed that for two similar examples the best strategy for growing a farm of optimally tuned turbines is the intuitive one. Fill the rst row up to the maximum permitted by navigational constraints and then add rows up to a maximum that can be economically justied in the light of the diminishing returns inherent in Fig. 5a). 5. Realizing the potential: a trade off The decision on turbine numbers is an economic trade off between the income from the power made available by adding turbines to a farm against their cost and any environmental impacts. Once the maximum permissible cross-sectional occupancy has been achieved then the trade off is in the face of a diminishing return on additional turbines, Fig. 5a). How this trade off plays out depends strongly on the maximum fraction of the channels cross-section which can be occupied in order to maintain navigation along the channel. It also strongly depends on the dynamical balance of the undisturbed channel, be it that of a bottom friction dominated shallow channel or that of an inertia dominated tidal strait, Fig. 5. This diminishing return is inherent in the upper row of plots in Fig. 6, where for both a low cross-sectional occupancy 0:1 and what may be an unrealistically high occupancy, 0:5, power available increases more slowly as rows are added to an optimally tuned farm. While the ultimate choice on number of turbines comes down to an economic and environmental cost benet analysis which is beyond the scope of this work, the plots however do contain essential information to underpin this analysis. For the shallow channel at the high occupancy 1 row of 12 turbines makes 80% of its 9 MW potential available, Fig. 6a). In contrast the same number of turbines spread across 5 rows at the lower occupancy makes only 65% of the potential available for power production. Fig. 6c) shows that the power available per turbine for both occupancies decreases as rows are added. Fig. 6e) illustrates that for both occupancies a farm making 80% of the shallow channels potential available will reduce ows by 30%. If such a signicant ow reduction is not acceptable then, at the lower occupancy, installing around 3 turbines reduces ows by only 10% while making around 25% of its potential available. The curves for the tidal strait in Fig. 6b), d) and f) are similar to those of the shallow channel, but the number of turbines involved is much larger. Almost 80% of the straits 6 GW potential can be realized at the higher occupancy, but this requires an astonishing

a1
0.8 0.6 0.4 Farm Efficiency

b
1

0.8

Flow reduction

0.6 0.2 0 0 5 10 Number of Rows, N 15 0.4 0 5 10 Number of Rows, N 15

Fig. 5. Effect of adding rows to an optimally tuned farm where turbines occupy 20% of the channels cross-section, i.e. a blockage ratio of 0:2. Solid curves are for the shallow channel and dashed lines for the tidal strait. a) Farm efciency Eq. (6). b) Velocity relative to undisturbed velocity.

R. Vennell / Renewable Energy 47 (2012) 95e102

101

Shallow Channel 1 0.8


1 5 10 3

Tidal Strait 1 0.8


10 5 50

max

Power/ P

0.6 0.4
1

Power/ Pmax

0.6 0.4
20 1

0.2 0

0.2
5 1

10

10 20 30 Number of Turbines

40

5 10 Thousands of Turbines

15

c
Power Per Turbine, MW

1.2 1 0.8 0.6 0.4 0.2 0 0 10 20 30 Number of Turbines 40

d
Power Per Turbine, MW

1.2 1 0.8 0.6 0.4 0.2 0 0 5 10 Thousands of Turbines

e
max

1
25

f
15 25 15

1 0.8

0.8 0.6 0.4 0.2 0

15000 10000 15000 5000 10000

10

Power/ Pmax

Power/ P

0.6 0.4

2000

5000

0.2 0

2000 1000

0.6

0.7 0.8 0.9 Flow Reduction

0.6

0.7 0.8 0.9 Flow Reduction

Fig. 6. Effect of the number of optimally tuned turbines on power available and ow reduction for the two example channels based on a 400 m2 blade area. Solid curves are for 0:1 and dashed lines for 0:5. a) and b) Farm efciency versus number of turbines, labelled dots give number of rows in the farm. c) and d) Power available per turbine in MW. e) and f) Farm efciency verses ow reduction. Labelled dots give the number of turbines in farm.

16,000 turbines spread over 13 rows! At the lower occupancy a more affordable 5000 turbines would make 30% of the straits potential available while reducing ows by 7%. 6. Discussion The principle aim of this paper was to show how many turbines are required to realize a given fraction of a channels potential in order to improve understanding of how power production scales with farm size. The models in V10 and V11, which build on those of GC05 and GC07, provide the rst physically coherent way of connecting the number of turbines in a farm with the maximum power available for electricity production. The models do not require the use of arbitrary loading or safety factors applied to account for unknown effects. The models are however simple. They use an extended form of classic turbine theory and they only apply to the cross-sectional average velocities, thus do not account for the effects of any variation of ow across the channel or with depth. They also do not allow for any effect of the turbulence shed by a row on the efciency of downstream rows. They do not allow for minimum

turbine operating ows or for loss due to their electro-mechanical efciency or transmission and conversion losses. Thus there is much work to be done to rene and extend the models. However, the models do illuminate signicant aspects of developing large tidal turbine farms. Most of these are a result of power extraction reducing the ows along the entire narrow channel, via the farms gross drag coefcient, CF. They include the necessity to tune turbines for the particular channel and in relation to each other, which will result in signicant computational cost to those modelling turbine farms and will require on going optimization of turbine tuning in large operational farms as turbines come in and out of service. They also include how farms have increasing or diminishing returns on turbines added to the cross-section depending on the importance of bottom friction and how there is always a diminishing return on additional rows, aspects which are fundamental to understanding the economics of farm development. The turbine numbers given by the simple models should be viewed as estimates of the magnitude of the numbers required, rather than precise estimates. These magnitudes are useful in understanding how realistic it is to realize a signicant fraction of

102

R. Vennell / Renewable Energy 47 (2012) 95e102

a channels potential. For the shallow channel around eight 400 m2 turbines, corresponding to three rows with 0:1, makes half of its potential available and 13, corresponding to one row with 0:5, makes 80% available, Table 1. The given values for the power available per turbine are also useful in understanding economic feasibility. For blockage ratios of 0:1 and 0.5 each turbine makes around 0.6 MW available. The detailed economics and environmental costs are beyond the scope of this work, however making an average of 0.6 MW available over the tidal cycle when existing turbines are producing around 1 MW at steady ows around 2 m s1 may be a reasonable return. Thus it appears possible to make a signicant fraction of the shallow channels potential available using 5e10 turbines, with a blade area similar to that of the largest currently operating tidal turbine, spread amongst 1e3 rows which occupy less than 50% of the cross-section. In contrast to make 50% of the tidal straits 6 GW potential available takes 10,000 turbines at 10% blockage. This high number is a result of the weaker ows in the tidal strait which yield only 0.27 MW per turbine. However unlike the tidal shallow channel, the output per turbine rises as more of the straits cross-section is occupied. So that at 50% blockage only 5000 turbines are required to make 50% of its potential available at a much higher 0.61 MW per turbine, similar to that for the shallow channel. Tidal straits have both a much larger potential in absolute terms and in proportion to their size, due to relatively low energy losses to bottom friction. This is seen in their having six times the potential per turbine when turbines ll the cross-section in a single row, Table 1. However tidal straits typically have weaker cross-sectionally averaged ows than shallow tidal channels. Thus at low blockage ratios a very large number of rows are required to make a signicant fraction available. If higher blockage ratios are permitted, then the enhanced efciency of turbines in tidal straits seen in Fig. 4 boosts the output per turbine and reduces the total number of turbines required. [17] uses the channels energy balance to explain why and when tidal straits have an increasing return on turbines added to the crosssection and why shallow channels have a diminishing return. 7. Conclusions Critical questions in developing a farm are what fraction of a channels potential is available for power production from a given number of turbines as a farm scales up from a single turbine into a large farm. The V10 V11 works provide the rst physically coherent way to connect the number of turbines in a farm to both the maximum power available for electricity production and to the degree of ow reduction, which is a consequence of power extraction. Though simple they illuminate several important aspects of developing large farms which have implications for both the economics and environmental impact of farms. Such as the necessity of tuning turbines in large farms for the particular channel, how much of the cross-section they occupy and the number of rows, as well as how to best arrange and congure the turbines. For example to maximize farm efciency the rst row turbines should be lled up to the maximum permitted by navigational constraints, before adding new rows. Also, while farm efciency always increases as optimally tuned turbines are added, once the cross-sectional occupancy limit is attained the power available per turbine decreases as rows are added. Thus increasing power production by increasing the farms installed capacity faces a diminishing return on additional rows. This development strategy may alter when more realistic models are developed which allow for variation of ow across the channel and with depth. However,

the decision on turbine numbers will remain an economic trade off between the power produced and installation, maintenance and environmental costs. How this trade off plays out depends strongly on the maximum fraction of the channels cross-section which can be occupied and the dynamical balance of the undisturbed channel. The examples demonstrate that it may be possible with existing technology to realize much of the MW potential of shallow tidal channels. The GW potential of tidal straits is both larger in absolute terms and also proportionately larger than that of shallow channels, due to relatively low energy losses to bottom friction. At low cross-sectional occupancies the typically lower ows of the strait result in both a low return per turbine and a very large number of turbines being required to realize a signicant fraction of their proportionately higher potential. Thus unless a large fraction of the straits cross-section can be occupied to take advantage of a higher output per turbine, it will be difcult to realize a substantial fraction of the GW potential of tidal straits until larger turbines are developed which are able operate economically in low ows. References
[1] Garrett C, Cummins P. The power potential of tidal currents in channels. Proceedings of the Royal Society A 2005;461:2563e72. [2] Blunden LS, Bahaj AS. Tidal energy resource assessment for tidal stream generators. Proceedings of the Institution of Mechanical Engineers Part A Journal of Power and Energy 2007;221(10):137e46. doi:10.1243/ 09576509JPE332. [3] Blancheld J, Garrett C, Rowe A, Wild P. Tidal stream power resource assessment for Masset Sound, Haida Gwaii. Proceedings of the Institution of Mechanical Engineers Part A Journal of Power and Energy 2008;222(5): 485e92. [4] Neill SP, Litt EJ, Couch SJ, Davies AG. The impact of tidal stream turbines on large-scale sediment dynamics. Renewable Energy 2009;34(12):2803e12. doi:10.1016/j.renene.2009.06.015. [5] Garrett C, Cummins P. Limits to tidal current power. Renewable Energy 2008; 33:2485e90. [6] Garrett C, Cummins P. Tuning turbines in a tidal channel. Journal of Fluid Mechanics 2010;663:253e67. doi:10.1017/S0022112010003502. [7] Vennell R. Tuning tidal turbines in concert to maximise farm efciency. Journal of Fluid Mechanics 2011;671:587e604. [8] Garrett C, Cummins P. The efciency of a turbine in a tidal channel. Journal of Fluid Mechanics 2007;588:243e51. [9] Blancheld J, Garrett C, Wild P, Rowe A. The extractable power from a channel linking a bay to the open ocean. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy 2008;222(3):289e97. [10] Vennell R. Estimating the power potential of tidal currents and the impact of power extraction on ow speeds. Renewable Energy 2011;36:3558e65. doi: 10.1016/j.renene.2011.05.011. [11] Arbic B, Garrett C. A coupled oscillator model of shelf and ocean tides. Continental Shelf Research 2010;30(6):564e74. [12] Vennell R. Oscillating barotropic currents along short channels. Journal of Physical Oceanography 1998;28(8):1561e9. [13] Vennell R. Observations of the phase of tidal currents along a strait. Journal of Physical Oceanography 1998;28(8):1570e7. [14] Byden IG, Grinsted T, Melville GT. Assessing the potential of a simple tidal channel to deliver useful energy. Journal of Applied Ocean Research 2004;26: 198e204. doi:10.1016/j.apor2005.04.001. [15] Vennell R. ADCP measurements of tidal phase and amplitude in Cook Strait, New Zealand. Continental Shelf Research 1994;14:353e64. [16] Douglas C, Harrison G, Chick J. Life cycle assessment of the Seagen marine current turbine. Proceedings of the Institution of Mechanical Engineers Part M Journal of Engineering for the Maritime Environment 2008;222(1):1e12. doi: 10.1243/14750902JEME94. [17] Vennell R. The energetics of large tidal turbine arrays. Renewable Energy, in press. [18] Vennell R. ADCP measurements of momentum balance and dynamic topography in a constricted tidal channel. Journal of Physical Oceanography 2006; 36(2):177e88. [19] Corten G. Heat generation by a wind turbine, Vol. In: 14th IEA symposium on the aerodynamics of wind turbines; 2000. p. 7. ECN report ECN-RX-01-001. [20] Lanchester FW. A contribution to the theory of propulsion and the screw propeller. Transactions of the Institution of Naval Architects 1915;LVII: 98e116. [21] Betz A. Das Maximum der theoretisch mglichen Ausnutzung des Windes durch Windmotoren. Gesamte Turbinenwesen 1920;17:307e9.

Vous aimerez peut-être aussi