Effect of Energy Storage on a Combined Wind and Wave Energy Farm Elisabetta Tedeschi, Eider Robles, Maider Santos, Olivier Duperray, Fernando Salcedo Energy and Environment Unit Tecnalia Research & Innovation Derio, Spain [email protected] Abstract—The combined installation of wind and wave energy converters is being proposed more and more often as a solution to exploit the potential complementarity between these two natural resources and to share the costs of their offshore infrastructures. This paper uses real meteorological data to evaluate the difference in the power performance of a single wind turbine and a combined wind/wave energy farm. The proposed case study considers a wind turbine coupled to wave arrays including a different number of point absorbers and the importance of an optimized tuning of the Wave Energy Converters to maximize the output power is shown. As a further step it is assumed that the farm is equipped with an energy storage device. The impact of its power and energy rating and of its efficiency on the achievable smoothing of the output power is analyzed.

I.

INTRODUCTION

The diversification of the energy portfolio seems to be the winning strategy to foster renewable energy penetration into traditional power systems. In the last decades also wave energy has been explored as a contributor to cover the world energy demand, in addition to well-known and established technologies as photovoltaic and wind onshore. Although promising, the wave energy sector is still immature and needs major research efforts to reach its full potential. One of its problems, common to other renewable energy resources, is its intermittency and variability, which strongly affects the power quality at the connection point with the local power system [1]. An interesting opportunity for wave energy development is the combination with wind energy in colocalized or fully integrated installations, which are being analyzed in different research projects [2-3]. The potential advantages of combined wind/wave farms is, on one hand, the exploitation of the possible complementarities between the resources to reduce their variability: in fact wave energy is more constant than wind energy and Wave Energy Converters (WEC) can still provide power when the wind has stopped blowing. On the other hand, combined solutions would substantially reduce installation and infrastructure costs, as well as the costs for operation and maintenance of the farm. Up to now, however, very few contributions [4-5]

This work has been realized under the SEA2GRID project and it has received the financial support of the European Union under a Marie Curie Intra-European Fellowship for Career Development (FP7-PEOPLE-2010IEF n. 272571)

have dealt with the complementarities of the wind/wave resource and with the expected performances of corresponding combined energy farms. Other studies [6] have been mainly focused on the required sizing of the wave array in order to reach a balance between wind and wave power extraction. A crucial issue to ease the grid interconnection of renewable energy farms is the presence of energy storage devices that can mitigate the grid impact of the resource intermittency. Many studies have been performed to evaluate the effect of energy storage in wind energy applications [7-8] and few contributions [9-10] deal with specific energy storage solutions for WECs, however such analyses have not yet been extended to combined installations. In this paper the attention is focused on the effects of macrometeorological fluctuations, i.e. large-scale weather patterns acting on a time scale of several hours, on the power output of a wind/wave energy farm. Starting from real meteorological data, the expected power performance of a 1.5 MW wind turbine (WT) is compared to that of a system composed of the same WT combined with an array including an increasing number of WECs for an additional maximum power of 1.5 MW. Following, the introduction of an energy storage element in the farm is also considered and the benefits in the mitigation of the output power variability are discussed, with reference to different ratings and efficiencies of the storage device itself. II.

SYSTEM MODELING

The system under investigation is schematically represented in Fig.1. It is a combined offshore wind/wave energy farm composed of a 1.5 MW wind turbine and of an array of WECs deployed at the same location. In the second part of the paper the wind/wave farm is assumed to be also equipped with an onshore energy storage system. A. Input metheorological data In order to evaluate the potential of a combined wind/wave energy farm, wind and wave data measured at the

TABLE I. YEARLY AVERAGE VALUES OF THE RELEVANT METEOROLOGICAL DATA AT SELECTED LOCATIONS

Buoy location and number

v10

HS

Tz

Site 1:Atlantic (South) - 41049

6.3 m/s

2.15 m

6.21 s

Site 2:SWHilo, Hawaii - 51002

8.1 m/s

2.28 m

6.40 s

Site 3: Half Moon Bay, CA - 46012

5.9 m/s

2.41 m

7.66 s

In the case of wave energy, the power per unit of crest width, Praw_wave, for irregular sea waves and under the assumption of deep water, is evaluated by [14]: (2)

_

where ρwa is the water density, assumed to be 1025 kg/m3, and g is the gravity acceleration.

Figure 1. Schematic, not to scale, representation of the combined wind/wave energy farm equipped with a storage device.

same time and at the same location are required. A time resolution t=1 hour is considered sufficient to evaluate the effect of macro-meteorological fluctuations [4, 7]. The oneyear data measured in 2010 by the buoys owned and maintained by National Data Buoy Center [11] at three different locations of both the Atlantic and the Pacific Ocean are considered: site 1 corresponds to the buoy deployed in the North Atlantic offshore the Dominican Republic (Station 41049, 27°30'0" N 63°0'0" W); site 2 corresponds to the buoy deployed South Southwest of Hilo, in the Hawaii Islands (Station 51002, 17°5'39" N 157°48'27" W), and site 3 corresponds to the buoy deployed at Half Moon Bay in California (Station 46012, 37°21'45" N 122°52'52" W). The data from available buoys have been selected in order to have a minimum number of missing measurements (mostly single points). Such few missing measurements have been reconstructed by linear interpolation. Among available data the relevant quantities for the following analysis are: wind average speed (at 10 m) v10, significant wave height Hs and average period of zero up-crossing Tz, which are summarized in Table I for the selected locations. From the wind speed measured at 10 m, v10, the corresponding wind speed, v, at the WT height is derived by applying the wind profile power law [12]. From Tz, the corresponding wave energy period, Te, that is used in the WEC model can be obtained as explained in [13], provided that the sea state can be represented by a Bretschneider spectrum, as discussed below. The first step of the analysis implies evaluating the available raw power associated to the wind and wave resource at the selected sites. In the case of wind the available resource is calculated from the wind speed as power per unit area, Praw_wind, according to: _

(1)

where ρwi is the air density, assumed to be 1.225 kg/m3.

B. Model of the wind turbine Wind turbines are a well-established technology and their performance can be described by a power curve to represent the dependency of the WT power output (Pwind) on the wind speed (v), as specified by IEC61400-12-1 [15]. Such power curve can be analytically expressed by the following equations [16]:

Pwind (t ) = 0 Pwind (t ) = Pn ⎛⎜ v ⎞⎟ ⎝ vn ⎠ Pwind (t ) = Pn Pwind (t ) = 0

for v < v min 3

for v min ≤ v < v n

(3)

for v n ≤ v ≤ v max for v > v max

In (3) vmin, vn and vmax are the minimum, rated and maximum wind speed, respectively and Pn is the nominal power of the wind turbine. According to [17], it is here assumed that vmin = 4 m/s, vn = 11.6 m/s, vmax = 25 m/s and Pn= 1.5 MW, corresponding to the power curve of Fig.2. C. Model of the wave energy converter Calculating the power extracted by the WECs is less straightforward than in the case of the WT. This is due to the higher concept diversification and limited maturity of the wave energy sector. Two different approaches can be considered to calculate WEC extracted power and both of them require information on the significant wave height (Hs) and energy period (Te) measured at the considered location. The most conventional approach, which is the one recommended in the IEC62600100 [18], is highly device specific and relates the transfer characteristics of the considered WEC to the statistical description of the sea state: from the “power operator” of the WEC and the probability of occurrence of each pair Hs-Te, an average power matrix for the specific device can be obtained. In this paper, however, a different method is used, according to what presented in [19]: in this case the model of the WEC is based on the assumption that it can be approximately represented by a damped linear oscillator. Thus, the fraction of available power (2) that the WEC actually extracts is

1600 3.5

3.5

Sea spectrum in W/m2 PTF

1400 3

1200

3

2.5

2.5

2

2

2

[W/m ]

[kW]

1000 800 1.5

600

1.5

400

1

1

200

0.5

0.5

0

0

5

10

15

20

25

0 0

30

[m/s]

0.1

0.2

0.3

Figure 2. Power curve for an Acciona wind turbine AW-70/1500, expressing the WT extracted power as a function of the wind speed

represented by a power transfer function (PTF), which is approximated by the Gaussian functional form: 1 ( f − f c )2 PTF ( f , f c ,σ ) = exp( − ) (4) 2 σ2 In (4) f represents the incident wave frequency, fc represents the central frequency the WEC is tuned on, (which depends on the control strategy applied to the WEC), and σ defines the width of the power transfer function. WECs have larger or narrower bandwidth depending on their geometry and size; in the following analysis cylindrical point absorbers are considered, which are typically narrow-banded devices. Each of them is thus assumed to have a width of the PTF: σ = 0.025 Hz and the diameter is d = 20 m. The dashed red curve in Fig.3.a represents an example of PTF for fc = 0.1 Hz. The assumption of linearity of the system allows evaluating the WEC average extracted power by multiplying the PTF for the wave spectral energy density of a representative seastate [20]. This requires modeling the sea energy distribution, S(f), along the incident wave frequencies through a suitable spectrum for the selected location and sea condition. In the following a Bretschneider spectrum [13] is used. The blue solid curve in Fig.3.a represents the energy distribution of a Bretschneider spectrum having Hs = 1.74 m and Te = 9.36 s. Thus, the average power extracted by the WEC can be calculated as: ,

,

(5)

In Fig.3.b the integrand of (5), corresponding to the curves of Fig.3.a is plotted as a function of the frequency. The advantage of the applied PTF approach is that it allows taking into account the effect of different WEC controls on the power performance, which is especially relevant for point absorber type WECs, as shown in the simulation section. D. Model of the energy storage system The energy storage element has been modeled in an extremely simplified way according to the following equations:

0.4

0

0

(a)

[Hz]

0.1

0.2

[Hz]

0.3

0.4

(b)

Figure 3. (a) Bretschneider spectrum for a sea state having Hs = 1.74 m and Te = 9.36 s (blue) and power transfer function of a WEC with σ = 0.025 Hz tuned fc = 0.1 Hz (red). (b) Corresponding power curve for a WEC having a diameter of 20 m in the considered sea state

dEs P (t ) Ps = Preq − Pin =− s , (6) ε ( Ps ) dt In (3) Es is the stored energy, Ps is the power exchanged by the storage system, ε represents the efficiency of the storage process, which is here assumed equal for both charging and discharging stage. Preq represents the desired power profile and Pin the input power profile (due to the wind (3) or wind + wave (5) converted power), when no storage is included. Maximum storage power, Psmax, and energy, Esmax, capacity have been taken into account, while no preferred state of charge for the storage system is considered in this analysis. III.

SIMULATION RESULTS

A. Evaluation of complementarity of wind and wave raw resources In order to evaluate the potential complementarity between the wind and wave resource, their contemporary availability at the same location must be evaluated. This can be done by calculating he cross-correlation, c(τ) of the raw power signals at time τ = 0 s, according to: c (τ ) = 1 n

n −τ

∑

( Praw _ wind (i ) − Praw _ wind _ avg )( Praw _ wave (i + τ ) − Praw _ wave _ avg ) STD raw _ wind STD raw _ wave

i =1

(7) where n is the total number of sample data points along the considered period and STD indicates the (normalized) standard deviation of the corresponding quantity, calculated as exemplified below for the case of raw wind: STD raw _ wind = 1 Praw _ wind _ avg

1 n −1

n

∑(P

raw _ wind

i =1

(i ) − Praw _ wind _ avg ) 2

(8)

1

0.9

0.9

0.9

0.8

0.8

0.8

0.7

0.7

0.7

0.6 0.5 0.4

Correlation c(0)

1

Correlation c(0)

Correlation c(0)

Site 3 – Half Moon Bay, California

Site 2 – SW Hilo, Hawaii

Site 1 – Atlantic (South) 1

0.6 0.5 0.4

0.6 0.5 0.4

0.3

0.3

0.3

0.2

0.2

0.2

0.1

0.1

0

1

2

3

4

0

0.1 1

quarter

2

3

4

0

quarter

1

2

3

4

quarter

Figure 4. Cross-correlation of the wind and wave raw available power in each quarter of 2010 (1:January-March; 2:April-June, 3:July-September, 4:October- December) and average yearly correlation (red) at the three selected locations.

The value of the cross-correlation has been calculated for the three selected locations, both on a yearly basis and for each quarter of the year, to better highlight the seasonal variability of the resources. Corresponding data are reported in Fig.4. For the sake of clarity it is worth recalling that a crosscorrelation equal to 1 means perfect correspondence between the wind and wave resource at the selected location showing a common evolution pattern along the time, while a crosscorrelation equal to 0 means that the two resources are completely unrelated. A cross-correlation of -1 indicates that the resources show a perfect inverse correspondence. A low cross-correlation is generally recommended where the windwave farm is deployed since variability reduction of the combined output power requires that the resources appear in different moments, in order to compensate each other. It can be noted that at site 1 the wind and wave resources show a high yearly average cross-correlation, c(0)= 0.79, while a lower cross-correlation, c(0)= 0.51, is found at site 2. Site 3 shows low cross-correlation, c(0)=0.35. These preliminary results can be explained considering that wind and wave resources show little correlation if there is a prevalence of swell waves that are remotely generated and thus have little correspondence with local wind condition. This is the case of the Half Moon Bay location (site 3), which is characterized by low local wind velocities and consequent swell due to the significant fetch from the North Pacific, mainly during winter season. On the other hand, the Atlantic (South) buoy is located in an area of inconsistent swells with main events occurring during the hurricane season from June to November. An intermediate condition occurs at site 2: Hawaii is potentially a good site for wave energy exploitation [21], however a partial sheltering from significant swells coming from the North due to islands can be noted at the selected site. Based on such observations site 3 is the most suitable for the potential installation of a combined wind and wave energy farm. Thus the following analyses are carried out with reference to site 3. B. Evaluation of complementarity of wind and wave energy converters power production As a following step, the expected power performance of the WT and WEC-array potentially deployed at site 3 has

been evaluated. At first the extracted powers from the WT and single WEC have been evaluated separately. Then a combined farm composed of one WT combined with a single point absorber and with an array of two, three, and four point absorbers has been considered. No smoothing effect due to the interactions of the point absorbers in the array is here taken into account, both because of the small size of the WEC array and because the constructive and destructive interference occurring at different wave frequencies tend to compensate each other when added on a yearly wave spectrum [22-23]. A maximum power limitation of 1500 kW has been assumed both for the WT and the overall WEC array. The power performance of the farm along the entire year has been evaluated in terms of average and maximum power production, and their corresponding ratio. The standard deviation has been selected as an indicator of the output power variability in the considered configurations and it has been calculated adapting (8). From the data reported in Table II it can be noted that the WEC is characterized by a much higher peak to average power ratio with respect to WT. This is typical for point absorbers devices [24] and requires suitable over-ratings of the mechanical and electrical equipment with respect to the average extracted power. The application of a power saturation, whose level is here kept constant for the WEC array irrespective of the number of point absorbers, can help reducing the peak-to-average power ratio and improving their efficiency [25]. TABLE II.

EFFECT OF THE NUMBER OF WECS IN THE ARRAY ON THE FARM POWER OUTPUT (FIXED CONTROL WITH FC= 0.01 HZ)

WT WEC WT + 1 WEC WT + 2 WEC WT + 3 WEC WT + 4 WEC

Pavg [kW] 591.1 156.4 747.5 898.0 1028.4 11139.8

Pmax [kW] 1500 1496 2969.6 3000 3000 3000

Pmax/ Pavg 2.54 9.39 3.97 3.34 2.92 2.63

STD

0.961 1.107 0.857 0.812 0.771 0.736

180 1

[kW]

160 140

0.98

120 100

0.96

60 0.05

0.1

0.

[Hz] 0.3

STD in p.u.

80 0.94 0.92 0.9 0.88

0.2

0.86

0.1 0.84

0 0.05

0.1

0.

[Hz]

Figure 5. (a) Yearly average power extraction of a single WEC as a function of the control tuning frequency, fc, for a constant control (blue) and for a control hourly tuned on the peak frequency (red). (b) Extracted fraction of the available raw wave power as a function of the control tuning frequency, fc, for a constant control (blue) and for a control hourly tuned on the peak frequency (red).

The applied peak power limitation acting both in the WT and in the WEC array explains the increase of average power extraction for the same peak power in case of arrays of two, three and four point absorbers. It can be clearly noted that the coupling of a WT and a single WEC consistently reduces the power output variability leading to an STD reduction of 10 % compared to the case of WT only and to more than 20 % compared to the WEC alone. Such variability reduction is even increased up to more than 23 and 33 %, respectively, in the case of a 4WEC array, at the expense of a higher equipment investment. To better evaluate if point absorbers can well complement a WT, it is also relevant to analyze the potential impact of their control on the overall power extraction. At first a single WEC having fixed response characteristics, meaning that its control is kept constant along the year, was considered. Different options for the tuning of the WEC have been analyzed. This is obtained by changing the central frequency fc of the PTF in the range [0.05-0.15] Hz, assuming its response to be optimized for different reference sea states (Fig.5). In the case of fixed control a maximum WEC yearly efficiency of 23% compared to the available raw power can be obtained at site 3 for fc=0.09 Hz. However it is shown how such efficiency is more than halved if control is not properly set. It is also worth noting that an hourly tuning of the WEC control based on the present sea state increases the overall WEC efficiency up to 25 % (red line in Fig.5.b), corresponding to an average yearly extracted power of 170 kW from the considered WEC (red line in Fig.5.a).

0.82

0

5

10

15

20

Figure 6. Normalized standard deviation of the power output as a function of the time constant, T, of the moving average filter representing an ideal storage. Blue line: WT only. Red line: single WEC. Black line: combined WT + 4 WECs array farm

C. Effect of the energy storage In the second part of the paper a preliminary evaluation of the energy storage effect in mitigating the output power variability is considered. It is here assumed that the desired power output of the farm corresponds to the one resulting from the application of a moving-average filter acting on the power profile produced by the farm when no energy storage is included, similarly to the approaches followed in [7, 9, 26]. The normalized reduction of the output power STD as a function of the time constant of the moving average filter, T, is reported in Fig.6 for the WT only (in blue with squares), a single WEC (in red with circles) and the combined WT+ 4WEC array (in black with triangles). The trend can be explained considering that wave energy is naturally more constant that wind energy, thus the filtering effect acting on the same time scale produces a higher variability reduction in the WT than in the WEC. It can be seen that a further reduction of the STD of 5.4 % and 9.9 % is obtained for the combined wind/wave farm if a power output corresponding to the one filtered out with a 12- and 24-hour time constant, respectively, is achieved. Such reduction of the power fluctuation is however obtained assuming an ideal energy storage device, having sufficient power and energy capability to ensure a perfect tracking of the desired power profile. The last part of the analysis verifies how an energy storage device of specified power capacity, energy capacity and efficiency actually performs when the desired power reference is the one corresponding to T=12 hours. Several different storage power capacities in the range 500-3000 kW have been considered. This takes into account that for wind energy applications, storage power rating between 25 % and 100 % of the plant rated power are suitable for services requiring a response time of several hours (as production leveling) [8]. The corresponding energy capacity range has been selected in this case to ensure a time response of 24 hours for the smallest power capacity.

0.95

0.95

0.85

0.8

0.8

0.75

0.75

[p.u.]

0.9

0.85

[p.u.]

0.9

0.7 0.65

Psmax = Psmax = Psmax = Psmax = Psmax =

0.6 0.55 0.5 0.45

0.7 0.65

2000

3000

4000

5000

6000

7000

8000

500 kW 750 kW 1000 kW 1500 kW 3000 kW

Psmax = Psmax = Psmax = Psmax = Psmax =

0.6 0.55 0.5 0.45

9000 10000 11000

2000

3000

4000

5000

Esmax [kWh]

6000

7000

8000

500 kW 750 kW 1000 kW 1500 kW 3000 kW

9000 10000 11000

Esmax [kWh]

Figure 7. Fraction of the year when the desired power represented by the input profile output from the WT+4 WECs array averaged over T=12 h can be perfectly tracked, depending on the storage power and energy capacity, for unity storage efficiency

Figure 8. Fraction of the year when the desired power represented by the input profile output from the WT+4 WECs array averaged over T=12 h can be perfectly tracked, depending on the storage power and energy capacity, for storage efficiency of 0.8

In Fig.7 the effect of the different storage ratings on the number of hours in the year when the desired power profile can be perfectly tracked, divided by the total number of hours in the year is shown, for ideal storage efficiency. In Fig.8, the same evaluation has been performed for a storage efficiency of 0.8. In order to take into account the potential effect of the initial state of charge of the energy device, data reported in Figs.7-8 are calculated as the average of the results of the analyses carried out for an initial energy storage level equal to 0, Esmax/2 and Esmax. It can be clearly seen that for the considered application and energy storage capacities, power ratings higher than 1.5 MW would bring no additional advantages. It can be also noted how the reduced efficiency worsens the tracking of the desired profile up to 5% for the highest energy capacities.

power output is shown. Figs 9.c and 9.d show the storage power and storage energy, respectively. It can be seen that due to the actual energy and power constraints of the storage system, a perfect tracking of the desired power profile cannot be obtained, thus worsening the quality of the power output. In this case the variability of the power output from the wind/wave farm is STD = 0.684, meaning that the storage system further reduces of 7% the output variability compared to the case of the wind/wave farm only. The same storage system, coupled with the WT, only reduces the output power variation from STD = 0.9607 to 0.8514, which shows the usefulness of the combined installation to smooth the output power production. It is also worth noting that in the case of WT only, such 11 % of variability reduction is in very good agreement with the results of [7], foreseeing that a 10 % reduction in the yearly fluctuation can be achieved with 2-3 MWh storage capacity per MW of wind power. If an efficiency of 80 % is considered for the energy storage, the variability reduction in the case of WT only decreases of 6.85 %. It has been verified, however, that such variation of the output power can be reduced up to 4 more times with the same non-ideal storage system if a 4 WECs array is deployed in the farm. This result is due to both the complementarity of the wind and wave resource and to a well-tuned control strategy of the WECs, but it is achieved at the expense of an increased installed power capacity in the combined farm.

The time domain behavior of a combined WT+4 WEC array farm equipped with a storage device having an energy capacity of 3 MWh, a power capacity of 1.5 MW and unity efficiency is shown in Fig.9. Input power and desired power profile are plotted in Fig.9.a. In Fig.9.b the actual smoothed [kW]

3000 2000 1000

[kW]

0 1000 3000

1050

1100

1150

(a)

1200

1050

1100

1150

(b)

1200

1000

1050

1100

1150

(c)

1200

3000 2000 1000 0 1000

1050

1100

1150

(d)

1200

2000

[kW]

0 1000

[kW]

IV.

1000

1000 0 -1000

[hours]

Figure 9. Input power (blue) from the combined farm and desired power (red) after storage action (T=12 h) (a), actual power output (b), storage power (c) and energy (d) when Esmax = 3 MWh, Psmax = 1.5 MW for unity efficiency

CONCLUSIONS

The goal of this paper is to provide a preliminary evaluation of the reduction in the variability of the power output from a combined wind/wave farm compared to the case of a wind turbine only. The output power performance resulting from the connection of energy storage systems of different ratings is also analyzed. Based on real wind and wave measurements, the analysis considers the impact of macro-meteorological fluctuations affecting the power generation of a combined farm on a time scale of several hours. At first it has been shown how wind and wave resources can show a very good complementarity,

which is however highly site-dependent. Thus, specific preliminary studies are required to identify suitable sites for the deployment of a combined wind-wave energy farms. In the considered test case, coupling a 1.5 MW wind turbine to a 4 WEC array of comparable peak power, a reduction of more than 20 % of the output power variability (and the elimination of hours at zero output power) is obtained compared to the case of the WT only. Following, the importance of point absorber control to improve the WEC array performance is shown and an adaptive tuning based on the actual sea conditions is encouraged. The study of the effect of storage on the combined farm shows that a storage system with energy capacity of 3MWh could further reduce the power output variability of 7 %, when tracking a power reference smoothed out along 12 hours. It is however important to underline that this is a preliminary analysis and that the selection of a specific storage technology for this application is out of the scope of this paper. More accurate evaluations will be carried out for this purpose by introducing detailed models of the storage system. It is also expected that improved performance could be obtained if a smart control strategy, including adapting the reference to the resource availability, is applied to the same energy storage system.

[9]

[10]

[11] [12]

[13] [14] [15]

[16]

[17] [18]

ACKNOWLEDGMENT The authors thank Pierpaolo Ricci from Tecnalia for the valuable discussion about WECs power extraction capability; the authors also thank Iker Marino for the valuable discussion about energy storage issues. REFERENCES [1]

[2] [3] [4]

[5]

[6]

[7] [8]

M. Santos, F. Salcedo, D. Ben Haim, J. L. Mendia, P. Ricci, J.L. Villate, J. Khan, D. Leon, S. Arabi, A. Moshref, G. Bhuyan, A. Blavette, D. O’Sullivan, R. Alcorn; (2011). Integrating Wave and Tidal Current Power: Case Studies through Modelling and Simulation, (Available: www.iea-oceans.org). EU FP7 MARINA Platform project (http://www.marinaplatform.info/, last accessed on: 19 January 2012). EU FP7 ORECCA project (http://www.orecca.eu last accessed on: 19 January 2012). F. Fusco, G. Nolan, J. V. Ringwood, “Variability reduction through optimal combination of wind/wave resources – An Irish case study”, Energy 35 (2010), pp. 314-325. L. Cradden, O. Duperray, H. Mouslim, D. Ingram, “ Joint exploitation of wave and offshore wind power” Proceedings of European Wave and Tidal Energy Conference (EWTEC11), Southampton, UK, 19-23 September 2011, pp.1-7. A. Babarit, J. Hals, “On the maximum and actual capture width ratio of wave energy converters” Proc. of the 9th European Wave and Tidal Energy Conference (EWTEC11), Southampton, UK, 19-23 September 2011, pp.1-7. J. V. Paatero, P. D. Lund, “Effect of Energy Storage Variations in Wind Power”, Wind Energy, 2005, N.8, pp.421-44.1. M. Swierczynski, R. Teodorescu, C. N. Rasmussen, P. Rodriguez, H. Vikelgaard, “Overview of the Energy Storage Systems for Wind

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

Power Integration Enhancement”, , IEEE International Symposium on Industrial Electronics, Bari (IT), 4-7 July 2010, pp.3749 – 3756. N. Daratha, H. Polinder, M. de Sousa Prado "A First-Order Energy Storage Requirements Estimation for an Archimedes Wave Swing Park", Proc. of the Int. Conf. on Sustainable Energy Technologies 2008 (ICSET08), 24-27 Nov. 2008, Singapore, pp.1161-11-65. D. Ravi Kiran, A.Palani, S. Muthukumar, V. Jayashankar,"Steady Grid Power From Wave Energy", IEEE Transactions on Energy Conversion, Vol 22, NO.22 june 2007, pp. 539-540. National Data Buoy Center website. (http://www.ndbc.noaa.gov , last accessed on: 19 January 2012). Peterson, E.W. and J.P. Hennessey, “On the use of power laws for estimates of wind power potential, J. Appl. Meteorology, Vol. 17, pp. 390-394. W. H. Michel, "Sea Spectra Revisited", Marine Technology, Vol. 36 N.4 winter 1999, pp.211-227. IEC 61400-12-1, “Wind turbines - Part 12-1: Power performance measurements of electricity producing wind turbines”, 2005-12-16. J. Falnes “Ocean Waves and Oscillating Systems; Linear Interactions Including Wave Energy Extraction” Cambridge University Press 2002. B. Fox et al. “Wind power integration – connection and system operational aspects”. The institution of Engineering Technology, 2007. Acciona Windpower AW-1500 (http://www.accionaenergia.com/media/315810/AW1500_brochure.pdf ). PD IEC/TS 62600-100, “Marine energy – Wave, tidal and other water current converters - Part 100: Power performance assessment of electricity producing wave energy converters” (standard in development). G. H. Smith, V. Venugopal, J. Fasham, “Wave Spectral Bandwidth as a Measure of Available Wave Power”, Proceedings of the 25th International Conference on Offshore Mechanics and Artic Engineering (OMAE06), Hamburg, Germany, 4-9 June 2006, pp. 1-9. A. Falcao, “Stochastic modeling in wave power-equipment optimization: maximum energy production versus maximum profit.”, Ocean Engineering, Vol.31, is.11-12, August 2004, pp. 1407-1421. International Energy Agency. (http://www.eia.gov/state/state-energy-profiles-analysis.cfm?sid=HI (Renewables and Waste in World in 2009). A. Babarit, “Impact of long separating distances on the energy production of two interacting wave energy converters” Ocean Engineering, vol 37, pp. 718-729, 2010. P.Ricci, J.-B. Saulnier, A.F.deO. Falcão, "Point-absorber arrays: a configuration study off the coast of Portugal" di Proceedings of the European Wave and Tidal Energy Conference (EWTEC 2007), Porto, Portugal. D. L. O’Sullivan, A. Lewis, “Generator requirement and Functionality for Ocean Energy Converters” Proc. Of the XIX Conf. on Electrical Machines, ICEM 2010, Roma, pp. 1-7. E. Tedeschi, M. Carraro, M. Molinas, P. Mattavelli, "Effect of Control Strategies and Power Take-Off Efficiency on the Power Capture From Sea Waves", IEEE Transactions on Energy Conversion, Volume: 26 , Issue: 4 Publication Year: 2011 , Page(s): 1088 – 1098. J. Aubry, P. Bydlowsky, B. Mundon, H, Ben Ahmed, B. Borgarino, “Energy Storage System Sizing for Smoothing Power Generation of Direct Wave Energy Converters”Proc. Of the 3rd Int. Conference on Ocean Energy (ICOE10), 6-8 October 2010, Bilbao, pp.1-7.

I.

INTRODUCTION

The diversification of the energy portfolio seems to be the winning strategy to foster renewable energy penetration into traditional power systems. In the last decades also wave energy has been explored as a contributor to cover the world energy demand, in addition to well-known and established technologies as photovoltaic and wind onshore. Although promising, the wave energy sector is still immature and needs major research efforts to reach its full potential. One of its problems, common to other renewable energy resources, is its intermittency and variability, which strongly affects the power quality at the connection point with the local power system [1]. An interesting opportunity for wave energy development is the combination with wind energy in colocalized or fully integrated installations, which are being analyzed in different research projects [2-3]. The potential advantages of combined wind/wave farms is, on one hand, the exploitation of the possible complementarities between the resources to reduce their variability: in fact wave energy is more constant than wind energy and Wave Energy Converters (WEC) can still provide power when the wind has stopped blowing. On the other hand, combined solutions would substantially reduce installation and infrastructure costs, as well as the costs for operation and maintenance of the farm. Up to now, however, very few contributions [4-5]

This work has been realized under the SEA2GRID project and it has received the financial support of the European Union under a Marie Curie Intra-European Fellowship for Career Development (FP7-PEOPLE-2010IEF n. 272571)

have dealt with the complementarities of the wind/wave resource and with the expected performances of corresponding combined energy farms. Other studies [6] have been mainly focused on the required sizing of the wave array in order to reach a balance between wind and wave power extraction. A crucial issue to ease the grid interconnection of renewable energy farms is the presence of energy storage devices that can mitigate the grid impact of the resource intermittency. Many studies have been performed to evaluate the effect of energy storage in wind energy applications [7-8] and few contributions [9-10] deal with specific energy storage solutions for WECs, however such analyses have not yet been extended to combined installations. In this paper the attention is focused on the effects of macrometeorological fluctuations, i.e. large-scale weather patterns acting on a time scale of several hours, on the power output of a wind/wave energy farm. Starting from real meteorological data, the expected power performance of a 1.5 MW wind turbine (WT) is compared to that of a system composed of the same WT combined with an array including an increasing number of WECs for an additional maximum power of 1.5 MW. Following, the introduction of an energy storage element in the farm is also considered and the benefits in the mitigation of the output power variability are discussed, with reference to different ratings and efficiencies of the storage device itself. II.

SYSTEM MODELING

The system under investigation is schematically represented in Fig.1. It is a combined offshore wind/wave energy farm composed of a 1.5 MW wind turbine and of an array of WECs deployed at the same location. In the second part of the paper the wind/wave farm is assumed to be also equipped with an onshore energy storage system. A. Input metheorological data In order to evaluate the potential of a combined wind/wave energy farm, wind and wave data measured at the

TABLE I. YEARLY AVERAGE VALUES OF THE RELEVANT METEOROLOGICAL DATA AT SELECTED LOCATIONS

Buoy location and number

v10

HS

Tz

Site 1:Atlantic (South) - 41049

6.3 m/s

2.15 m

6.21 s

Site 2:SWHilo, Hawaii - 51002

8.1 m/s

2.28 m

6.40 s

Site 3: Half Moon Bay, CA - 46012

5.9 m/s

2.41 m

7.66 s

In the case of wave energy, the power per unit of crest width, Praw_wave, for irregular sea waves and under the assumption of deep water, is evaluated by [14]: (2)

_

where ρwa is the water density, assumed to be 1025 kg/m3, and g is the gravity acceleration.

Figure 1. Schematic, not to scale, representation of the combined wind/wave energy farm equipped with a storage device.

same time and at the same location are required. A time resolution t=1 hour is considered sufficient to evaluate the effect of macro-meteorological fluctuations [4, 7]. The oneyear data measured in 2010 by the buoys owned and maintained by National Data Buoy Center [11] at three different locations of both the Atlantic and the Pacific Ocean are considered: site 1 corresponds to the buoy deployed in the North Atlantic offshore the Dominican Republic (Station 41049, 27°30'0" N 63°0'0" W); site 2 corresponds to the buoy deployed South Southwest of Hilo, in the Hawaii Islands (Station 51002, 17°5'39" N 157°48'27" W), and site 3 corresponds to the buoy deployed at Half Moon Bay in California (Station 46012, 37°21'45" N 122°52'52" W). The data from available buoys have been selected in order to have a minimum number of missing measurements (mostly single points). Such few missing measurements have been reconstructed by linear interpolation. Among available data the relevant quantities for the following analysis are: wind average speed (at 10 m) v10, significant wave height Hs and average period of zero up-crossing Tz, which are summarized in Table I for the selected locations. From the wind speed measured at 10 m, v10, the corresponding wind speed, v, at the WT height is derived by applying the wind profile power law [12]. From Tz, the corresponding wave energy period, Te, that is used in the WEC model can be obtained as explained in [13], provided that the sea state can be represented by a Bretschneider spectrum, as discussed below. The first step of the analysis implies evaluating the available raw power associated to the wind and wave resource at the selected sites. In the case of wind the available resource is calculated from the wind speed as power per unit area, Praw_wind, according to: _

(1)

where ρwi is the air density, assumed to be 1.225 kg/m3.

B. Model of the wind turbine Wind turbines are a well-established technology and their performance can be described by a power curve to represent the dependency of the WT power output (Pwind) on the wind speed (v), as specified by IEC61400-12-1 [15]. Such power curve can be analytically expressed by the following equations [16]:

Pwind (t ) = 0 Pwind (t ) = Pn ⎛⎜ v ⎞⎟ ⎝ vn ⎠ Pwind (t ) = Pn Pwind (t ) = 0

for v < v min 3

for v min ≤ v < v n

(3)

for v n ≤ v ≤ v max for v > v max

In (3) vmin, vn and vmax are the minimum, rated and maximum wind speed, respectively and Pn is the nominal power of the wind turbine. According to [17], it is here assumed that vmin = 4 m/s, vn = 11.6 m/s, vmax = 25 m/s and Pn= 1.5 MW, corresponding to the power curve of Fig.2. C. Model of the wave energy converter Calculating the power extracted by the WECs is less straightforward than in the case of the WT. This is due to the higher concept diversification and limited maturity of the wave energy sector. Two different approaches can be considered to calculate WEC extracted power and both of them require information on the significant wave height (Hs) and energy period (Te) measured at the considered location. The most conventional approach, which is the one recommended in the IEC62600100 [18], is highly device specific and relates the transfer characteristics of the considered WEC to the statistical description of the sea state: from the “power operator” of the WEC and the probability of occurrence of each pair Hs-Te, an average power matrix for the specific device can be obtained. In this paper, however, a different method is used, according to what presented in [19]: in this case the model of the WEC is based on the assumption that it can be approximately represented by a damped linear oscillator. Thus, the fraction of available power (2) that the WEC actually extracts is

1600 3.5

3.5

Sea spectrum in W/m2 PTF

1400 3

1200

3

2.5

2.5

2

2

2

[W/m ]

[kW]

1000 800 1.5

600

1.5

400

1

1

200

0.5

0.5

0

0

5

10

15

20

25

0 0

30

[m/s]

0.1

0.2

0.3

Figure 2. Power curve for an Acciona wind turbine AW-70/1500, expressing the WT extracted power as a function of the wind speed

represented by a power transfer function (PTF), which is approximated by the Gaussian functional form: 1 ( f − f c )2 PTF ( f , f c ,σ ) = exp( − ) (4) 2 σ2 In (4) f represents the incident wave frequency, fc represents the central frequency the WEC is tuned on, (which depends on the control strategy applied to the WEC), and σ defines the width of the power transfer function. WECs have larger or narrower bandwidth depending on their geometry and size; in the following analysis cylindrical point absorbers are considered, which are typically narrow-banded devices. Each of them is thus assumed to have a width of the PTF: σ = 0.025 Hz and the diameter is d = 20 m. The dashed red curve in Fig.3.a represents an example of PTF for fc = 0.1 Hz. The assumption of linearity of the system allows evaluating the WEC average extracted power by multiplying the PTF for the wave spectral energy density of a representative seastate [20]. This requires modeling the sea energy distribution, S(f), along the incident wave frequencies through a suitable spectrum for the selected location and sea condition. In the following a Bretschneider spectrum [13] is used. The blue solid curve in Fig.3.a represents the energy distribution of a Bretschneider spectrum having Hs = 1.74 m and Te = 9.36 s. Thus, the average power extracted by the WEC can be calculated as: ,

,

(5)

In Fig.3.b the integrand of (5), corresponding to the curves of Fig.3.a is plotted as a function of the frequency. The advantage of the applied PTF approach is that it allows taking into account the effect of different WEC controls on the power performance, which is especially relevant for point absorber type WECs, as shown in the simulation section. D. Model of the energy storage system The energy storage element has been modeled in an extremely simplified way according to the following equations:

0.4

0

0

(a)

[Hz]

0.1

0.2

[Hz]

0.3

0.4

(b)

Figure 3. (a) Bretschneider spectrum for a sea state having Hs = 1.74 m and Te = 9.36 s (blue) and power transfer function of a WEC with σ = 0.025 Hz tuned fc = 0.1 Hz (red). (b) Corresponding power curve for a WEC having a diameter of 20 m in the considered sea state

dEs P (t ) Ps = Preq − Pin =− s , (6) ε ( Ps ) dt In (3) Es is the stored energy, Ps is the power exchanged by the storage system, ε represents the efficiency of the storage process, which is here assumed equal for both charging and discharging stage. Preq represents the desired power profile and Pin the input power profile (due to the wind (3) or wind + wave (5) converted power), when no storage is included. Maximum storage power, Psmax, and energy, Esmax, capacity have been taken into account, while no preferred state of charge for the storage system is considered in this analysis. III.

SIMULATION RESULTS

A. Evaluation of complementarity of wind and wave raw resources In order to evaluate the potential complementarity between the wind and wave resource, their contemporary availability at the same location must be evaluated. This can be done by calculating he cross-correlation, c(τ) of the raw power signals at time τ = 0 s, according to: c (τ ) = 1 n

n −τ

∑

( Praw _ wind (i ) − Praw _ wind _ avg )( Praw _ wave (i + τ ) − Praw _ wave _ avg ) STD raw _ wind STD raw _ wave

i =1

(7) where n is the total number of sample data points along the considered period and STD indicates the (normalized) standard deviation of the corresponding quantity, calculated as exemplified below for the case of raw wind: STD raw _ wind = 1 Praw _ wind _ avg

1 n −1

n

∑(P

raw _ wind

i =1

(i ) − Praw _ wind _ avg ) 2

(8)

1

0.9

0.9

0.9

0.8

0.8

0.8

0.7

0.7

0.7

0.6 0.5 0.4

Correlation c(0)

1

Correlation c(0)

Correlation c(0)

Site 3 – Half Moon Bay, California

Site 2 – SW Hilo, Hawaii

Site 1 – Atlantic (South) 1

0.6 0.5 0.4

0.6 0.5 0.4

0.3

0.3

0.3

0.2

0.2

0.2

0.1

0.1

0

1

2

3

4

0

0.1 1

quarter

2

3

4

0

quarter

1

2

3

4

quarter

Figure 4. Cross-correlation of the wind and wave raw available power in each quarter of 2010 (1:January-March; 2:April-June, 3:July-September, 4:October- December) and average yearly correlation (red) at the three selected locations.

The value of the cross-correlation has been calculated for the three selected locations, both on a yearly basis and for each quarter of the year, to better highlight the seasonal variability of the resources. Corresponding data are reported in Fig.4. For the sake of clarity it is worth recalling that a crosscorrelation equal to 1 means perfect correspondence between the wind and wave resource at the selected location showing a common evolution pattern along the time, while a crosscorrelation equal to 0 means that the two resources are completely unrelated. A cross-correlation of -1 indicates that the resources show a perfect inverse correspondence. A low cross-correlation is generally recommended where the windwave farm is deployed since variability reduction of the combined output power requires that the resources appear in different moments, in order to compensate each other. It can be noted that at site 1 the wind and wave resources show a high yearly average cross-correlation, c(0)= 0.79, while a lower cross-correlation, c(0)= 0.51, is found at site 2. Site 3 shows low cross-correlation, c(0)=0.35. These preliminary results can be explained considering that wind and wave resources show little correlation if there is a prevalence of swell waves that are remotely generated and thus have little correspondence with local wind condition. This is the case of the Half Moon Bay location (site 3), which is characterized by low local wind velocities and consequent swell due to the significant fetch from the North Pacific, mainly during winter season. On the other hand, the Atlantic (South) buoy is located in an area of inconsistent swells with main events occurring during the hurricane season from June to November. An intermediate condition occurs at site 2: Hawaii is potentially a good site for wave energy exploitation [21], however a partial sheltering from significant swells coming from the North due to islands can be noted at the selected site. Based on such observations site 3 is the most suitable for the potential installation of a combined wind and wave energy farm. Thus the following analyses are carried out with reference to site 3. B. Evaluation of complementarity of wind and wave energy converters power production As a following step, the expected power performance of the WT and WEC-array potentially deployed at site 3 has

been evaluated. At first the extracted powers from the WT and single WEC have been evaluated separately. Then a combined farm composed of one WT combined with a single point absorber and with an array of two, three, and four point absorbers has been considered. No smoothing effect due to the interactions of the point absorbers in the array is here taken into account, both because of the small size of the WEC array and because the constructive and destructive interference occurring at different wave frequencies tend to compensate each other when added on a yearly wave spectrum [22-23]. A maximum power limitation of 1500 kW has been assumed both for the WT and the overall WEC array. The power performance of the farm along the entire year has been evaluated in terms of average and maximum power production, and their corresponding ratio. The standard deviation has been selected as an indicator of the output power variability in the considered configurations and it has been calculated adapting (8). From the data reported in Table II it can be noted that the WEC is characterized by a much higher peak to average power ratio with respect to WT. This is typical for point absorbers devices [24] and requires suitable over-ratings of the mechanical and electrical equipment with respect to the average extracted power. The application of a power saturation, whose level is here kept constant for the WEC array irrespective of the number of point absorbers, can help reducing the peak-to-average power ratio and improving their efficiency [25]. TABLE II.

EFFECT OF THE NUMBER OF WECS IN THE ARRAY ON THE FARM POWER OUTPUT (FIXED CONTROL WITH FC= 0.01 HZ)

WT WEC WT + 1 WEC WT + 2 WEC WT + 3 WEC WT + 4 WEC

Pavg [kW] 591.1 156.4 747.5 898.0 1028.4 11139.8

Pmax [kW] 1500 1496 2969.6 3000 3000 3000

Pmax/ Pavg 2.54 9.39 3.97 3.34 2.92 2.63

STD

0.961 1.107 0.857 0.812 0.771 0.736

180 1

[kW]

160 140

0.98

120 100

0.96

60 0.05

0.1

0.

[Hz] 0.3

STD in p.u.

80 0.94 0.92 0.9 0.88

0.2

0.86

0.1 0.84

0 0.05

0.1

0.

[Hz]

Figure 5. (a) Yearly average power extraction of a single WEC as a function of the control tuning frequency, fc, for a constant control (blue) and for a control hourly tuned on the peak frequency (red). (b) Extracted fraction of the available raw wave power as a function of the control tuning frequency, fc, for a constant control (blue) and for a control hourly tuned on the peak frequency (red).

The applied peak power limitation acting both in the WT and in the WEC array explains the increase of average power extraction for the same peak power in case of arrays of two, three and four point absorbers. It can be clearly noted that the coupling of a WT and a single WEC consistently reduces the power output variability leading to an STD reduction of 10 % compared to the case of WT only and to more than 20 % compared to the WEC alone. Such variability reduction is even increased up to more than 23 and 33 %, respectively, in the case of a 4WEC array, at the expense of a higher equipment investment. To better evaluate if point absorbers can well complement a WT, it is also relevant to analyze the potential impact of their control on the overall power extraction. At first a single WEC having fixed response characteristics, meaning that its control is kept constant along the year, was considered. Different options for the tuning of the WEC have been analyzed. This is obtained by changing the central frequency fc of the PTF in the range [0.05-0.15] Hz, assuming its response to be optimized for different reference sea states (Fig.5). In the case of fixed control a maximum WEC yearly efficiency of 23% compared to the available raw power can be obtained at site 3 for fc=0.09 Hz. However it is shown how such efficiency is more than halved if control is not properly set. It is also worth noting that an hourly tuning of the WEC control based on the present sea state increases the overall WEC efficiency up to 25 % (red line in Fig.5.b), corresponding to an average yearly extracted power of 170 kW from the considered WEC (red line in Fig.5.a).

0.82

0

5

10

15

20

Figure 6. Normalized standard deviation of the power output as a function of the time constant, T, of the moving average filter representing an ideal storage. Blue line: WT only. Red line: single WEC. Black line: combined WT + 4 WECs array farm

C. Effect of the energy storage In the second part of the paper a preliminary evaluation of the energy storage effect in mitigating the output power variability is considered. It is here assumed that the desired power output of the farm corresponds to the one resulting from the application of a moving-average filter acting on the power profile produced by the farm when no energy storage is included, similarly to the approaches followed in [7, 9, 26]. The normalized reduction of the output power STD as a function of the time constant of the moving average filter, T, is reported in Fig.6 for the WT only (in blue with squares), a single WEC (in red with circles) and the combined WT+ 4WEC array (in black with triangles). The trend can be explained considering that wave energy is naturally more constant that wind energy, thus the filtering effect acting on the same time scale produces a higher variability reduction in the WT than in the WEC. It can be seen that a further reduction of the STD of 5.4 % and 9.9 % is obtained for the combined wind/wave farm if a power output corresponding to the one filtered out with a 12- and 24-hour time constant, respectively, is achieved. Such reduction of the power fluctuation is however obtained assuming an ideal energy storage device, having sufficient power and energy capability to ensure a perfect tracking of the desired power profile. The last part of the analysis verifies how an energy storage device of specified power capacity, energy capacity and efficiency actually performs when the desired power reference is the one corresponding to T=12 hours. Several different storage power capacities in the range 500-3000 kW have been considered. This takes into account that for wind energy applications, storage power rating between 25 % and 100 % of the plant rated power are suitable for services requiring a response time of several hours (as production leveling) [8]. The corresponding energy capacity range has been selected in this case to ensure a time response of 24 hours for the smallest power capacity.

0.95

0.95

0.85

0.8

0.8

0.75

0.75

[p.u.]

0.9

0.85

[p.u.]

0.9

0.7 0.65

Psmax = Psmax = Psmax = Psmax = Psmax =

0.6 0.55 0.5 0.45

0.7 0.65

2000

3000

4000

5000

6000

7000

8000

500 kW 750 kW 1000 kW 1500 kW 3000 kW

Psmax = Psmax = Psmax = Psmax = Psmax =

0.6 0.55 0.5 0.45

9000 10000 11000

2000

3000

4000

5000

Esmax [kWh]

6000

7000

8000

500 kW 750 kW 1000 kW 1500 kW 3000 kW

9000 10000 11000

Esmax [kWh]

Figure 7. Fraction of the year when the desired power represented by the input profile output from the WT+4 WECs array averaged over T=12 h can be perfectly tracked, depending on the storage power and energy capacity, for unity storage efficiency

Figure 8. Fraction of the year when the desired power represented by the input profile output from the WT+4 WECs array averaged over T=12 h can be perfectly tracked, depending on the storage power and energy capacity, for storage efficiency of 0.8

In Fig.7 the effect of the different storage ratings on the number of hours in the year when the desired power profile can be perfectly tracked, divided by the total number of hours in the year is shown, for ideal storage efficiency. In Fig.8, the same evaluation has been performed for a storage efficiency of 0.8. In order to take into account the potential effect of the initial state of charge of the energy device, data reported in Figs.7-8 are calculated as the average of the results of the analyses carried out for an initial energy storage level equal to 0, Esmax/2 and Esmax. It can be clearly seen that for the considered application and energy storage capacities, power ratings higher than 1.5 MW would bring no additional advantages. It can be also noted how the reduced efficiency worsens the tracking of the desired profile up to 5% for the highest energy capacities.

power output is shown. Figs 9.c and 9.d show the storage power and storage energy, respectively. It can be seen that due to the actual energy and power constraints of the storage system, a perfect tracking of the desired power profile cannot be obtained, thus worsening the quality of the power output. In this case the variability of the power output from the wind/wave farm is STD = 0.684, meaning that the storage system further reduces of 7% the output variability compared to the case of the wind/wave farm only. The same storage system, coupled with the WT, only reduces the output power variation from STD = 0.9607 to 0.8514, which shows the usefulness of the combined installation to smooth the output power production. It is also worth noting that in the case of WT only, such 11 % of variability reduction is in very good agreement with the results of [7], foreseeing that a 10 % reduction in the yearly fluctuation can be achieved with 2-3 MWh storage capacity per MW of wind power. If an efficiency of 80 % is considered for the energy storage, the variability reduction in the case of WT only decreases of 6.85 %. It has been verified, however, that such variation of the output power can be reduced up to 4 more times with the same non-ideal storage system if a 4 WECs array is deployed in the farm. This result is due to both the complementarity of the wind and wave resource and to a well-tuned control strategy of the WECs, but it is achieved at the expense of an increased installed power capacity in the combined farm.

The time domain behavior of a combined WT+4 WEC array farm equipped with a storage device having an energy capacity of 3 MWh, a power capacity of 1.5 MW and unity efficiency is shown in Fig.9. Input power and desired power profile are plotted in Fig.9.a. In Fig.9.b the actual smoothed [kW]

3000 2000 1000

[kW]

0 1000 3000

1050

1100

1150

(a)

1200

1050

1100

1150

(b)

1200

1000

1050

1100

1150

(c)

1200

3000 2000 1000 0 1000

1050

1100

1150

(d)

1200

2000

[kW]

0 1000

[kW]

IV.

1000

1000 0 -1000

[hours]

Figure 9. Input power (blue) from the combined farm and desired power (red) after storage action (T=12 h) (a), actual power output (b), storage power (c) and energy (d) when Esmax = 3 MWh, Psmax = 1.5 MW for unity efficiency

CONCLUSIONS

The goal of this paper is to provide a preliminary evaluation of the reduction in the variability of the power output from a combined wind/wave farm compared to the case of a wind turbine only. The output power performance resulting from the connection of energy storage systems of different ratings is also analyzed. Based on real wind and wave measurements, the analysis considers the impact of macro-meteorological fluctuations affecting the power generation of a combined farm on a time scale of several hours. At first it has been shown how wind and wave resources can show a very good complementarity,

which is however highly site-dependent. Thus, specific preliminary studies are required to identify suitable sites for the deployment of a combined wind-wave energy farms. In the considered test case, coupling a 1.5 MW wind turbine to a 4 WEC array of comparable peak power, a reduction of more than 20 % of the output power variability (and the elimination of hours at zero output power) is obtained compared to the case of the WT only. Following, the importance of point absorber control to improve the WEC array performance is shown and an adaptive tuning based on the actual sea conditions is encouraged. The study of the effect of storage on the combined farm shows that a storage system with energy capacity of 3MWh could further reduce the power output variability of 7 %, when tracking a power reference smoothed out along 12 hours. It is however important to underline that this is a preliminary analysis and that the selection of a specific storage technology for this application is out of the scope of this paper. More accurate evaluations will be carried out for this purpose by introducing detailed models of the storage system. It is also expected that improved performance could be obtained if a smart control strategy, including adapting the reference to the resource availability, is applied to the same energy storage system.

[9]

[10]

[11] [12]

[13] [14] [15]

[16]

[17] [18]

ACKNOWLEDGMENT The authors thank Pierpaolo Ricci from Tecnalia for the valuable discussion about WECs power extraction capability; the authors also thank Iker Marino for the valuable discussion about energy storage issues. REFERENCES [1]

[2] [3] [4]

[5]

[6]

[7] [8]

M. Santos, F. Salcedo, D. Ben Haim, J. L. Mendia, P. Ricci, J.L. Villate, J. Khan, D. Leon, S. Arabi, A. Moshref, G. Bhuyan, A. Blavette, D. O’Sullivan, R. Alcorn; (2011). Integrating Wave and Tidal Current Power: Case Studies through Modelling and Simulation, (Available: www.iea-oceans.org). EU FP7 MARINA Platform project (http://www.marinaplatform.info/, last accessed on: 19 January 2012). EU FP7 ORECCA project (http://www.orecca.eu last accessed on: 19 January 2012). F. Fusco, G. Nolan, J. V. Ringwood, “Variability reduction through optimal combination of wind/wave resources – An Irish case study”, Energy 35 (2010), pp. 314-325. L. Cradden, O. Duperray, H. Mouslim, D. Ingram, “ Joint exploitation of wave and offshore wind power” Proceedings of European Wave and Tidal Energy Conference (EWTEC11), Southampton, UK, 19-23 September 2011, pp.1-7. A. Babarit, J. Hals, “On the maximum and actual capture width ratio of wave energy converters” Proc. of the 9th European Wave and Tidal Energy Conference (EWTEC11), Southampton, UK, 19-23 September 2011, pp.1-7. J. V. Paatero, P. D. Lund, “Effect of Energy Storage Variations in Wind Power”, Wind Energy, 2005, N.8, pp.421-44.1. M. Swierczynski, R. Teodorescu, C. N. Rasmussen, P. Rodriguez, H. Vikelgaard, “Overview of the Energy Storage Systems for Wind

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

Power Integration Enhancement”, , IEEE International Symposium on Industrial Electronics, Bari (IT), 4-7 July 2010, pp.3749 – 3756. N. Daratha, H. Polinder, M. de Sousa Prado "A First-Order Energy Storage Requirements Estimation for an Archimedes Wave Swing Park", Proc. of the Int. Conf. on Sustainable Energy Technologies 2008 (ICSET08), 24-27 Nov. 2008, Singapore, pp.1161-11-65. D. Ravi Kiran, A.Palani, S. Muthukumar, V. Jayashankar,"Steady Grid Power From Wave Energy", IEEE Transactions on Energy Conversion, Vol 22, NO.22 june 2007, pp. 539-540. National Data Buoy Center website. (http://www.ndbc.noaa.gov , last accessed on: 19 January 2012). Peterson, E.W. and J.P. Hennessey, “On the use of power laws for estimates of wind power potential, J. Appl. Meteorology, Vol. 17, pp. 390-394. W. H. Michel, "Sea Spectra Revisited", Marine Technology, Vol. 36 N.4 winter 1999, pp.211-227. IEC 61400-12-1, “Wind turbines - Part 12-1: Power performance measurements of electricity producing wind turbines”, 2005-12-16. J. Falnes “Ocean Waves and Oscillating Systems; Linear Interactions Including Wave Energy Extraction” Cambridge University Press 2002. B. Fox et al. “Wind power integration – connection and system operational aspects”. The institution of Engineering Technology, 2007. Acciona Windpower AW-1500 (http://www.accionaenergia.com/media/315810/AW1500_brochure.pdf ). PD IEC/TS 62600-100, “Marine energy – Wave, tidal and other water current converters - Part 100: Power performance assessment of electricity producing wave energy converters” (standard in development). G. H. Smith, V. Venugopal, J. Fasham, “Wave Spectral Bandwidth as a Measure of Available Wave Power”, Proceedings of the 25th International Conference on Offshore Mechanics and Artic Engineering (OMAE06), Hamburg, Germany, 4-9 June 2006, pp. 1-9. A. Falcao, “Stochastic modeling in wave power-equipment optimization: maximum energy production versus maximum profit.”, Ocean Engineering, Vol.31, is.11-12, August 2004, pp. 1407-1421. International Energy Agency. (http://www.eia.gov/state/state-energy-profiles-analysis.cfm?sid=HI (Renewables and Waste in World in 2009). A. Babarit, “Impact of long separating distances on the energy production of two interacting wave energy converters” Ocean Engineering, vol 37, pp. 718-729, 2010. P.Ricci, J.-B. Saulnier, A.F.deO. Falcão, "Point-absorber arrays: a configuration study off the coast of Portugal" di Proceedings of the European Wave and Tidal Energy Conference (EWTEC 2007), Porto, Portugal. D. L. O’Sullivan, A. Lewis, “Generator requirement and Functionality for Ocean Energy Converters” Proc. Of the XIX Conf. on Electrical Machines, ICEM 2010, Roma, pp. 1-7. E. Tedeschi, M. Carraro, M. Molinas, P. Mattavelli, "Effect of Control Strategies and Power Take-Off Efficiency on the Power Capture From Sea Waves", IEEE Transactions on Energy Conversion, Volume: 26 , Issue: 4 Publication Year: 2011 , Page(s): 1088 – 1098. J. Aubry, P. Bydlowsky, B. Mundon, H, Ben Ahmed, B. Borgarino, “Energy Storage System Sizing for Smoothing Power Generation of Direct Wave Energy Converters”Proc. Of the 3rd Int. Conference on Ocean Energy (ICOE10), 6-8 October 2010, Bilbao, pp.1-7.