The URL can be used to link to this page

Home My WebLink AboutLoss evaluation of HVAC and HVDC transmission solutions for large offshore wind farmsElectric Power Systems Research 76 (2006) 916–927 Loss evaluation of HVAC and HVDC transmission solutions for large offshore wind farms N. Barberis Negra a , J. Todorovic b , T. Ackermann c,∗ a Politecnico of Turin, Department of Electrical Engineering, Italy b ELEKTROPRENOS, Banja Luka, Bosnia and Herzegovina c Royal Institute of Technology, Teknikringen 33, 10044 Stockholm, Sweden Received 12 May 2005; received in revised form 29 August 2005; accepted 9 November 2005 Available online 2 May 2006 Abstract This paper presents a comparison of transmission losses for different technical transmission solutions for large offshore wind farms. Three technical solutions are analyzed, i.e. HVAC, HVDC Line Commutated Converter (LCC) and HVDC Voltage Source Converter (VSC). The losses for each technology are calculated for wind farms with different ratings and various distances to shore. In addition, solutions with combinations of two and the three different transmission technologies are analyzed and compared. Based on this comparison, further analysis regarding the economical feasibility can be performed in order to determine the most economic solutions for the transmission system of an offshore wind farm. © 2006 Elsevier B.V. All rights reserved. Keywords:HVAC; HVDC Line Commutated Converter; HVDC Voltage Source Converter; Offshore wind farm 1. Introduction Today’s installed offshore wind farms have relatively small installed rated capacity (max. 160MW) and are placed within a short distance (max. 20km) from the shore [1]. Overall economics of offshore wind farms tend to increase with wind farm size, hence future projects will be sig- niﬁcantly larger and most likely further away from shore. On the one hand, offshore locations have better wind con- ditions than onshore ones: this means higher energy out- put. On the other hand, longer transmission distances lead to higher investment costs as well as higher energy losses [3]. All currently (mid 2005) existing offshore wind farms are connected to shore by HVAC cables and only three of them have offshore substations [1]. For large wind farms, with hun- dreds of MW of rated capacity, and long distances to shore, offshore substations would be necessary for stepping up the ∗Corresponding author. Tel.: +46 7066 39457. E-mail addresses:nicola.barberis@libero.it (N.B. Negra), todorovicjovan@hotmail.com (J. Todorovic),Thomas.Ackermann@ieee.org (T. Ackermann). voltagelevel(HVAC)and/orforconvertingthepowertoHVDC [3]. The connection of large offshore wind farms (100MW) imposeachallengingtask.Thepropertransmissionsolution,i.e. HVAC or HVDC, can have an important inﬂuence on the over- all project feasibility. Small differences in transmission losses between two solutions could cause large differences in energy output over a project time of 20 years. In this paper, system transmission losses for three differ- ent transmission systems, i.e. HVAC, HVDC Line Commutated Converter (LCC) and HVDC Voltage Source Converter (VSC) are compared.Table 1 shows a comparison of the three trans- mission technologies considering current technology and main components. The comparison in this paper considers 500 and 1000MW wind farms located in an area with an average wind speed of 9m/s and varying distance to shore (up to 200km). The trans- missionsystemlossesarecalculatedaslossesoftheannualwind farm production (in percent). It is assumed that the wind farm has an availability of 100%. Further analyses with different size of the wind farm (400 up to 1000 MW), different average wind speed (8–11m/s) and differentdistancesfromshore(upto300km)areperformedand presented in refs.[5,6]. 0378-7796/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2005.11.004 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 917 Table 1 Comparison HVAC-HVDC transmission system [1–6,23] HVAC HVDC LCC HVDC VSC Maximum available capacity per system 800MW at 400kV Up to 600MW (submarine transmission) Up to 350MW installed 380MW at 220kV 500MW announced 220MW at 132kV (1080MW proposed) All up to 100km Voltage level 132kV installed Up to ±500kV Up to ±150kV 220 and 400kV under development (±300kV proposed) Offshore installed projects Many small installation (Table 1.5 in ref.[6]) Not yet installed Only test project (oil platform in Norway) Black start capability Yes No Yes Technical capability for network support No, SVC are required to supply reactive power No, capacitor banks or Statcom are required to supply reactive power to the valves Yes, reactive power can be generated or absorbed by the VSC devices Offshore station in operation Yes No On an oil platform Decoupling of connected networks No Yes Yes Cable model Resistances, capacitance and induction Resistance Resistance Requirements for ancillary service Not necessary Yes for low wind speeds conditions Yes for low wind speeds conditions Space requirements offshore substation Smallest size Biggest size Medium size Installation costs Small for station (only transformer) high cost for cable High cost for station (transformer, ﬁlters, capacitors banks, thyristor valves etc.) Low costs for cable Station 30–40% more expensive than LCC solution (IGBT more expensive than thyristor valves) cable more expensive than LCC 2. Aggregated wind farm model In order to evaluate transmission losses for a wind farm, it is necessarytoexactlydeﬁnetheoutputofthewindfarmovertime. Thus an aggregated model based on Holttinen and Norgaard [7] has been considered. With this model it is possible to deﬁne an output power curve for the aggregated wind turbines, i.e. wind farm. Input data for the model are: - wind farm size of 500 or 1000MW; - standard 5MW wind turbine; - wind speed in the area with average wind speed of 9m/s rep- resented by Rayleigh distribution; - dimension D of the wind farm equal to 25 for 500MW and 50km for 1000MW (front side in respect to the direction of the wind)[3]; - turbulence intensity I equal to 10%. In order to obtain the aggregated model, it is necessary to generate a normalised distribution that represents the probabil- ity distribution function for the wind speed for individual wind turbines in the area at a given time. To obtain this distribution, it is necessary to get a normalised standard deviation as a function of D and I andthenmultiplythisvaluefortheaveragewindspeed of the area: from the obtained standard deviation it is possible to plot the wind speed distribution. Byapplyingthewindspeeddistributiontothepowercurveof a single wind turbine, it is possible to obtain a smoothed multi- turbine power curve, that is representative for the aggregated power output of the wind farm (Fig. 1). The jth elements of the discrete multi-turbine power curve,PT,j , can be calculated as PT,j = M j Ps,j Ps,j (1) where M is equal to 11,Ps,j the jth element of the single-turbine power curve and Ps,j is the probability of a spatial wind speed Fig. 1. Comparison between a single wind turbine power curve and the wind farm power curve for a 1000MW wind farm and an average wind speed in the area of 9m/s,D =50km. 918 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 Fig. 2. Duration curve of a 1000MW wind farm with different average wind speeds in the area. distribution.Inordertogetagoodresultthesumshouldbedone as a minimum for a wind speed range from −5 to +5m/s around the jth element in the power curve: for this reason M is chosen equal to 11. In Fig. 1, a comparison between a single wind turbine power curveandawindfarmpowercurveisgiven.Uptoaround13m/s, thewindfarmgeneratesmorepowerthanthesinglewindturbine due to the variation of wind around the chosen value of ±5m/s. For instance, considering an average wind speed of 10m/s, it is assumed that in the area of the wind farm, the wind speed varies between 5 and 15m/s. Hence, some turbines operate at lower wind speed and others at higher, but since the power obtained from the wind is proportional to the third power of the wind speed, turbines operating at higher speed give a more relevant contribution to the total power generated by the wind farm. Inordertorepresenttheaggregatedpowercurveforthewhole wind farm, it is necessary to multiple the obtained power curve by the swept area and the number of installed wind turbines. In Fig. 2, the duration curve for a 1000MW wind farm with different average wind speeds (8, 10 and 11m/s) is given. From Fig. 2, it is possible to see that with increasing average wind speed, the rated wind farm capacity can be generated for longertimeperiod.At11m/sthewindfarmwilloperateatrated capacity for almost 40% of the time, which is almost twice as long as for a wind farm would operate at rated capacity at a location with 9m/s average wind speed. 3. HVAC transmission system The production of large amounts of reactive power can be considered the main limiting factor of HVAC cable utilization in transmission systems for long distances. A comparison of the transmission capacity of different cables operated at certain voltagelevels(132,220and400kV)anddifferentcompensation solutions (only onshore or at both ends) is presented in Fig. 3. Cable limits, i.e. maximal permissible current, voltage swing of receiving end between no-load and full load (<10%) and phase Fig. 3. Transmission capacity of different HVDC transmission cables for three voltage levels, 132, 220 and 400kV. variation (<30 ◦), should not be exceeded, see also Brakelmann [8]. For the here considered cables, the maximal current is the only limit that is reached, the other two are not critical con- straints. The critical distance is achieved when half of the reactive current produced by the cable reaches nominal current at the end of one cable. For the here considered cables, the critical distances is: -Lmax,132kV =370km; -Lmax,220kV =281km; -Lmax,400kV =202km. 3.1. Components of the HVAC transmission solution The voltage level within an offshore wind farm grid is typ- ically in the range of 30–36kV, hence for large offshore wind farms and/or long distances to shore, a substation is necessary to step up the voltage for the transmission to shore. An HVAC transmission system used for connection of large offshore wind farms to the onshore grid contains: - HVAC submarine transmission cable(s); - offshore transformer(s); - compensation units, thyristor controlled reactors (TCR), both onshore and offshore; - onshore transformer(s), depending on a grid voltage. Thesecomponentsprovidethetransmissionfromanoffshore collection point of the wind turbines’ power (offshore substa- tion) to a grid connection point placed onshore. 3.2. Loss calculations 3.2.1. Models and assumptions Cable loss calculations are performed based on Brakelmann [9]. Loss calculations take into account the current distribution along cable line and temperature dependence. N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 919 Table 2 Cables’ parameters and main characteristics Cable 132kV 220kV 400kV Resistance (/m) 48 × 10−6 48 × 10−6 45.5 × 10−6 Inductance (mH/km) 0.34 0.37 0.39 Capacitance (mF/km) 0.23 × 10−3 0.18 × 10−3 0.18 × 10−3 Nominal current (A) 1055 1055 1323 Cable section (mm 2 ) 1000 1000 1200 Max operating temperature ( ◦C) 90 90 90 For 132 and 220kV voltage transmission levels, three core XLPE insulated submarine cables are used while for 400kV levelthreesinglecoreXLPEsubmarinecablesareconsideredin trefoil formation. Cable characteristics are tabulated in Table 2. According to Brakelmann in ref.[9], the cable losses per unit length can be calculated as followed P =(Pmax −PD )I IN 2 υθ +PD (2) where Pmax is the nominal total cable loss,PD the dielectric loss, per core,I the load current,IN the nominal current,υ θ the temperature correction coefﬁcient that is calculated as: υθ =cα cα +αT θ max 1 −I IN 2 (3) where αT isthetemperaturecoefﬁcientoftheconductorresistiv- ity(1/K),cα theconstant,i.e.cα =1−αT (20 ◦C −θamb ),θ max the maximal temperature rise, i.e. 90 −15=75◦C, the ambient temperature is supposed to be θ amb =15◦C. As the cable current along the cable is not constant for a speciﬁc wind farm output, but depends on the position along a cable,i.e.I=f(x),thefollowingintegralmustbeusedtocalculate the cable losses: Pl0 =Pmax l0 I 2 N l0 x=0 I 2 (x)υθ (x)dx +PD (4) Solving integral (4)for length l0 , the cable losses per unit length are obtained. Multiplying the integral with actual cable length l0 , the cable losses in W are achieved. This method pro- vides an accurate calculation of the cables losses [9]. In order to calculate transformer losses, equivalent param- eters like Rfe , representing iron losses and Rcu , representing copper losses, are deﬁned. These data can be obtained from nominal transformer loss data. As TCRs are used as compensa- tion units, it is assumed that they have the same no load losses as an equivalent transformer with the same VA rating and half of load losses of an equivalent transformer with the same VA rating [10]. 3.2.2. Results System losses for average wind speed of 9m/s, for three transmission voltage levels (132, 220 and 400kV) and for two wind farm conﬁgurations of 500 and 1000MW are presented in Tables 3 and 4, respectively. Transmission system losses l%have been calculated as l%= N i Plost,i pi ha N i Pgen,i pi ha (5) where Plost,i represents the transmission losses at wind speed i, Pgen,i the power generated by the wind farm at wind speed i, Table 3 Transmission losses of a 500MW wind farm, with 9m/s of average wind speed in the area in % of annual wind farm production Table 4 Transmission losses of a 1000MW wind farm, with 9m/s of average wind speed in the area in % of annual wind farm production 920 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 Fig.4. Participationofeachtransmissioncomponentintotaltransmissionlosses for 500MW wind farm, 9m/s of average wind speed, at 100km transmission distance, three three-core 132kV submarine cables [5]. N the number of wind speed class considered for the model,pi the probability to have a certain wind speed I, which is obtained by the Rayleigh distribution,h the number of hours in a year and a is the availability of the wind park. In our calculations the availability is deﬁned as 100%. Gray cells in Table 3 represent the best transmission solu- tions with the lowest losses, while number of cables indicate the number of cables required for the particular solution. In the 132kVcolumn,numberofcablespresentsthenumberofcables required for the 200km. Within the loss calculations, a new cable is added to the wind farm whenever the transmission system requires more capacity (depending on the wind speed). The same approach applies for Table 4. Fig. 4 shows the share each transmission component con- tributes to the total transmission losses for a 500MW wind farm with a transmission distance of 100km using a 132kV cable. It can be seen that cable losses represent by far the highest share of the total transmission losses. Thus, in order to decrease the totaltransmissionlosses,specialattentionshouldbegiventothe cable selection. From Tables 3 and 4, it can be seen that only 220 and 400kV solutions are considered. These two submarine XLPE cable designs are, however, still under development [11]. Especially the400kVXLPEsubmarinecableiscurrentlyonlyavailablefor short lengths as appropriate joints and splices for longer lengths are not available yet. Considering distances longer than 200km, 132kV solutions prevail [5], as at such long distances, 220 and 400kV cables generate large amounts of reactive power. 4. HVDC System with Line Commutated Converter Line Commutated Converter (LCC) based transmission sys- temshavebeensuccessfullyinstalledinmanybulkpowertrans- missionsystemsoverlongdistancesbothonlandandsubmarine allaroundtheworld,seerefs.[16,17].Adrawbackofthistrans- mission solution is the required reactive power to the thyristor valves in the converter and may be the generation of harmonics in the circuit [16]. 4.1. Components of the transmission system Main components of the transmission system based on LCC devices are: - AC and DC ﬁlters; - converter transformer; - converter based on thyristor valves; - smoothing reactor; - capacitor banks or STATCOM; - DC cable and return path; - auxiliary power set; - protection and control devices (i.e.: cooling devices, surge arrester). All these components are considered in the following loss calculations, except the STATCOM: for the inﬂuence of the STATCOM on total losses we would like to refer to ref.[15]. 4.2. Loss calculations 4.2.1. Models and assumptions In order to calculate transmission losses for different wind farmsizes,datafromexistingHVDCLCCinstallationsarecon- sidered, see also refs.[16,17]. Converter stations have been built in sizes of 250, 440, 500 and 600MW. Losses have not been calculated in this case, but they are obtained from Siemens: they vary typically with a lin- ear trend between 0.11% (no load) and 0.7% (rated power) of the rated power [16]. Both monopolar and bipolar solutions are considered depending on the number of converter stations and on the size of the wind farm. CablemodelsarebasedonBrakelmann’stheory [8]andmod- els take into account variations of temperature in the cable in order to obtain more realistic results. In Table 5, an overview of the selected cable solutions are presented: all the conﬁgurations are based on mass impregnated cables with conductors made of copper. Lost power Pcab in the cable is calculated with formulae: Pcab =PL max I 2 I 2 N vθ (6) Table 5 Cables data for the model compiled and calculated from [8,17–19] Rated power (MW) 250 440 500 600 Voltage level (kV) 250 350 400 450 Nominal current (kA) 1 1.257 1.25 1.333 Cable section (mm 2 ) 1000 1400 1200 1600 Resistance (/km) at 20 ◦C 0.018 0.013 0.02 0.011 Max operating temperature ( ◦C) 55 55 55 55 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 921 Table 6 Transmission losses for different converter station layouts with 9m/s of average wind speed in the area in % of annual wind farm production CS, converter station. PL max =R0 cm I 2 N lcable (7) cm =1 +α20 (θ L max +θU −20) (8) cα =1 −α20 (20 −θU ) (9) vθ =cα cα +α20 θ L max 1 −I IN 2 (10) where R0 is the DC resistance of the conductor at 20 ◦C per unit length [18,19],α20 the constant mass temperature coefﬁcient at 20 ◦C [18,19],PL max the lost power in the cable at its max- imum operating temperature,θ Lmax =55◦C is the maximum operatingtemperatureoftheinsulator,θ U =15◦Cistheambient temperature,IN the nominal current of the cable,I the current ﬂowing into the cable and lcable is the length of the cable used for the transmission. When more than one converter station is used for the trans- mission, the total power is divided between the different con- verter station depending on the conﬁguration that gives the lowest total losses. Converter stations are shut down when low power is gener- ated in the wind farm: in these conditions, only the losses of protection and control devices are considered and these devices are supplied by the auxiliary power set. 4.2.2. Results Three different layouts are considered for a 500MW wind farm and four for a 1000MW wind farm: these conﬁgurations are listed in Table 6 with the corresponding system losses. Transmission system losses l%have been calculated with (5) and data from Table 5. The gray marked cells in Table 6, represent the conﬁguration with the lowest losses. Fig. 5. Loss participation to the overall system losses from data in Table 5 (CS, converter station). 922 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 For some conﬁgurations, participation of each component in the system losses of the system is shown in Fig. 5. Converterstationsareresponsibleforthehighestshareofthe overall system losses; participation of the cable increases with cable lengths. It should be noted that the calculation method for ofthecablelossesissigniﬁcantlymoredetailedthanthemethod forthecalculationofthelossesintheconverterstation,however, indiscussionwiththemanufactureroftheequipmentourresults were conﬁrmed. 5. HVDC System with Voltage Source Converter VoltageSourceConverter(VSC)technologyisnewerthanthe previousoneandrelevantprojectshavebeeninstalledonlyfrom 1997 [17].TheadvantageofHVDCVSCsolutionsarebasedon itscapabilitytosupplyandabsorbreactivepowerandtosupport power system stability; on the other hand line to ground faults can be problematic. 5.1. Components of the transmission system Main components of the transmission system based on VSC devices are: - VSC converter station circuit breaker; - system side harmonic ﬁlter; - interface transformer; - converter side harmonic ﬁlter; - VSC unit; - VSC DC capacitor; - DC harmonic ﬁlter; - DC reactor; - DC cable or overhead transmission line; - auxiliary power set. All these components are considered in the following loss calculation, except the auxiliary power set due to lack of infor- mation about its losses. 5.2. Loss calculations 5.2.1. Models and assumptions In order to calculate system losses for different wind farm sizes,andduetolackofdatafrommanufactures,dataaremainly used from installed projects. For instance, system loss data are extracted from installed projects such as the Cross Sound Cable [21]and the Murray Link Project [22]. By calculating the transmission losses of those projects, see ref.[9], it is pos- sible to calculate the losses for the total converter station (350 and 220MW). In general, the idea is to divide the total system losses into three components (two stations+the cable) in order to use the data for further calculations. Considering the system represented in Fig. 6, that is valid for the VSC and LCC HVDC transmission solutions, and assuming the percent losses xS are equal in both converter stations, it’s possible to obtain P1 =(1 −xs )Pin (11) PC =P1 −P2 =RI 2 =R P1 VC 2 (12) Pout =(1 −xs )P2 (13) where VC is the rated voltage of the cable and I is the current ﬂowing in it. Deﬁning then equation R V 2 C (1 −xs )3 P 2 in −(1 −xs )2 Pin +Pout =0 (14) it is possible to calculate the value xS since all the other param- eters are known. In order to consider the temperature dependence of the resis- tance, it is possible to follow the procedure shown by Brakel- mann [9]for an AC cable, taking into account the DC nature of the system (Eqs.(7)–(10)). It is then possible to calculate the resistance of the cable as R =PL max vθ I 2 N (15) Since the data for the cable provide only the input and the output power for the whole transmission system, it is neces- sary to solve the calculations in a loop with (7)–(10),(14)and (15)in order to obtain the value of the current in the cable and thuscalculatetheresistance.Manufacturesareworkingonlarger converter station ratings; however, no detailed data are publicly available for those new converter stations. Hence, the losses for a 500MW converter station are estimated from the losses of a 350MW converter station. Again, for the loss calculations of the cable Brakelmann’s theory [9]is used and the model takes into account variations of temperature in the cable in order to obtain more realistic results. In Table7,thecabledatausedinthefollowingarepresented:all cables are based on PE solution, conductors are made of copper and rated voltage is 150kV. The same approach described in Eqs.(6)–(10)is used for the calculations of the losses in cables. When more than one converter system is considered for the transmission, the total power is divided between the different Fig. 6. System block diagram. N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 923 Table 7 Cable data for the VSC model compiled and calculated from [9,17–19] Rated power (MW) 220 350 500 Nominal current (kA) 0.793 1.2 1.68 Cable section (mm 2 ) 1300 1300 2000 Resistance (/km) at 20 ◦C 0.014 0.014 0.01 Max operating temperature ( ◦C) 70 70 70 VSC systems based on the lowest overall transmission system losses. Converter stations are shut down when power production in the wind farm is lower than transmission system losses: in these conditions, only the losses of protection and control systems are considered and these devices are supplied by the auxiliary power set. However, due to lack of information, these losses are neglected. 5.2.2. Results Three different layouts are considered for a 500MW wind farm and two for a 1000MW wind farm: these conﬁgurations areshownin Table8 withthecorrespondingtransmissionlosses of each system. Transmission system losses l%have been calculated with (5) and data from Table 7. The gray cells in Table 8 represent the conﬁguration with the lowest losses. For some conﬁgurations, contribution of each component to the overall system losses is shown in Fig. 7.It can be seen that the converter stations contribute most to the overall system losses; and that the share of the cable increases with lengths. Table 8 Transmission losses for different converter station layouts with 9m/s of average wind speed in the area in % of annual wind farm production CS, converter station. 6. Comparison of different solutions In this section, a comparison of the three different trans- mission system is presented. From results in Sections 3–5,it can be concluded that AC solution provides the lowest losses for a distance of 50km from shore, while for 100, 150 and 200km from the shore the HVDC LCC solution has lowest transmission losses (Tables 9 and 10). In the tables, ‘Conﬁg.’ stands for the rated power and the voltage level (between break- ers) of the transmission for the HVAC system and the rated power of the converter station for the two HVDC solutions and ‘Nr Cables’ stands for the number of cable required for the transmission. In Fig. 8, the overall system losses are shown for all three transmission systems (HVAC, HVDC LCC and HVDC VSC) for 400 up to 1000MW wind farm at 0–300km from the shore. Fig. 7. Loss participation to the overall system from data in Table 7, VSC system (CS, converter station). 924 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 Table 9 Loss comparison for 500MW wind farm at 9m/s average wind speed in the area CS, converter station. Fig. 8 shows that an HVAC system leads to the lowest transmission system losses for a distance of up to 55–70km (depending on the size of the wind farm, which inﬂuences the conﬁguration of the transmission system). For longer dis- tances, HVDC LCC becomes the solution with lowest losses. Dash–dotted lines in Fig. 8 shows the 1, 1.6 and 2% loss line depending on wind farm sizes and distances. These lines show that in the AC-area, losses do not vary so much and they remain nearly constant for increasing wind farm capacity and almost constant distances to shore. In the LCC-area instead total trans- mission losses vary much more with changing wind farm size and distance: this behaviour is caused by the conﬁguration cho- sen for the transmission of the power with the LCC system. In factforeachwindfarmsizeadifferentcombinationofconverter stations is considered. After having calculated these losses, it could be interesting to consider investment costs in order to obtain a wider overview andamorerealisticcomparisonoftheanalyzedsolutions.Based on the results presented in previous sections, an analysis of the investment costs was presented by Lazaridis and Ackermann in ref.[24]. 6.1. Combination of two transmission systems In some cases it might be beneﬁcial to combine different transmission solutions in order to improve some features of Table 10 Loss comparison for 1000MW wind farm at 9m/s average wind speed in the area CS, converter station. N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 925 Fig. 8. MW-km plane, comparison HVAC-HVDC LCC for different wind farm size (400–1000MW) and different distances to shore (0–300km) for average wind speed of 9m/s. the overall system (system reliability, power stability, etc.). For example, an HVDC VSC transmission system, can be used to improve the stability of the system, since it can control the gen- eration and absorption of reactive power in the system. Conﬁgurations are deﬁned according to the current technol- ogy and data for the components are taken from the previous sections. When a combination is chosen, it is assumed that the system with highest losses is the main transmission component and the lowest one is installed with lower transmitted power in order to decrease the total system losses. When instead system losses of both systems are close, it is assumed that both trans- mission systems transmit the same amount of power. When a HVDC VSC solution is considered, some limitations in the pos- sible combinations must be considered due to the small range of rated power of the available converter station. Results are presented in Tables 11 and 12: in row ‘Conﬁg.’ theratedcapacityoftherelativetransmissionsystemisgiven(in brackets: the voltage level of the HVAC system), in ‘Nr Cables’ the number of cables necessary for each transmission system and ‘at x km’ the corresponding system losses for this distance are given. In the tables, symbol ‘+’ divides the kind of system used for the transmission. From the tables, it can be seen that the combination of two different transmission systems never improves the overall sys- temlossescomparedtoconﬁgurationswithasingletransmission system.However,systemlossesofthesystemwithhighestlosses decrease with the combination with another system. For exam- ple, a HVDC VSC system has losses of 4.05% (Table 9)ifit operates alone at 50km from the shore, but its losses could be Table 11 Comparison of combined transmission solutions losses for a 500MW wind farm at 9m/s average wind speed AC+VSC AC+LCC LCC+VSC Conﬁg. 280MW (400kV)+220MW 150MW (220kV)+350MW 200MW (220kV)+300MW 60MW (220kV)+440MW 300MW+220MW 250MW+350MW No. of cables 1+2 1+2 1+1 1+1 1+2 1+2 At 50km 2.02 3.11 1.54 1.70 2.61 2.86 Conﬁg. 280MW (400kV)+220MW 150MW (220kV)+350MW 370MW (400kV)+130MW 250MW (400kV)+250MW 300MW+220MW 250MW+350MW No. of cables 1+2 1+2 1+1 1+1 1+2 1+2 At 100km 3.21 3.94 2.57 2.55 2.89 3.22 Conﬁg. 280MW (220kV)+220MW 150MW (132kV)+350MW 370MW (220kV)+130MW 250MW (132kV)+250MW 300MW+220MW 250MW+350MW No. of cables 1+2 1+2 2+1 2+1 1+2 1+2 At 200km 6.88 6.98 6.89 6.55 3.46 3.93 Table 12 Comparison of combined transmission solutions losses for a 1000MW wind farm at 9m/s average wind speed AC+VSC AC+LCC LCC+VSC Conﬁg. 200 MW (220kV)+800MW 200MW (400kV)+800MW 300MW (400kV)+700MW 500MW+500MW 250MW+800MW No. of cables 1+4 1+2 1+2 1+4 1+6 At 50km 3.20 1.44 1.31 2.46 3.18 Conﬁg. 500MW (400kV)+500MW 800MW (400kV)+250MW 900MW (400kV)+130MW 500MW+500MW 250MW+800MW No. of cables 2+4 1+1 2+1 1+4 1+6 At 100km 3.02 2.56 2.32 2.70 3.58 Conﬁg. 500MW (220kV)+500MW 800MW (220kV)+250MW 900MW (220kV)+130MW 500MW+500MW 250MW+800MW No. of cables 2+4 3+1 4+1 1+4 1+6 At 200km 6.66 6.68 7.18 3.16 3.93 926 N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 Fig. 9. 1000MW wind farm at different distances to shore, case 1. decrease up to 2% if it is combined with a HVAC transmission system. 6.2. Combination of three transmission systems Large wind farms (up to 1000MW) are supposed to be installed over a wide geographical area. Large offshore wind turbines (5MW) could be placed at distance of 1km from each other. Such a wind farm might require more than one offshore substation with different distance to shore and different grid connection conditions. One example of such a wind farm con- ﬁguration is presented in Fig. 9 (based on a proposed 1000MW offshorewindfarminScotland).Fromtheconsidered1000MW windfarm,powercanbetransmittedbythreedifferenttransmis- sion systems, characterized by different distances to shore and different grid strengths at the onshore connection point. The AC system might be used at short distances with small amount of transmitted power and connected to a weak grid. The HVDC VSC system might be considered the best solution to supportpowersystemstabilityandthereforemightbeconnected to a medium-strong grid with transmission of a larger amount of power. The HVDC LCC solution has the lowest transmission losses and thus it is used for transmission of a large amount of power for a long distance to a strong grid connection point. Three cases of power distributions among the three transmis- sion system options are considered: 1. 80MW by AC, 220MW by HVDC VSC and 700MW by HVDC LCC; 2. 50MW by AC, 350MW by HVDC VSC and 600MW by HVDC LCC; Table 13 Percent of average transmission losses of system in Fig. 2, and participation of each transmission system in total losses Cases Losses (%) AC participation (%) LCC participation (%) VSC participation (%) Case 1 2.60 5.27 72.26 22.47 Case 2 2.71 4.58 56.8 39.19 Case 3 2.31 11.62 63.71 25.37 3. 180MW by AC, 220MW by HVDC VSC and 600MW by HVDC LCC. Depending on the wind farm production, different transmis- sion paths to shore are considered. For example in case 1, if the wind farm produces less than 80MW, only the AC transmis- sion system is used. For the range 80–700MW, only the HVDC LCC solution is operated, whereas between 700 and 780MW, HVDC LCC operates at 700MW and remaining power is trans- mitted by the AC system. For a power production of between 780and1000MW,HVDCLCCandACsystemsoperateatrated capacity and the remaining energy is transmitted by the HVDC VSC system. This conﬁguration is chosen in order to obtain the lowest value for the total losses of the transmission sys- tem. It must be mentioned that in reality the operation mode might be differ from the here proposed solution, particularly in regards to the operation of the HVDC VSC which might be used to support the operation of the onshore grid during certain times. Losses for the wind farm are presented in Table 13. HVDC LCC system causes the highest share of the total transmission system losses, as expected, due to the fact that it transmits the greatest amount of power. The higher the VSC share, the higher the overall losses. Vice versa, if more power is transmit- ted via the HVAC solution, total transmission losses decrease, but this might create problems for the stability of the onshore grid as the onshore grid connection point can be considered weak. 7. Conclusions Interest in large offshore wind farms has increased over the last years. Design and speciﬁcation of the transmission system to shore is critical for the economic feasibility of very large (200MW) offshore wind farms. This paper investigates the total transmission losses of three transmission solutions, i.e. HVAC, HVDC LCC and HVDC VSC. In general, HVAC solution leads to the lowest losses for dis- tances of up to 55–70km from the shore, whereas after this distance, HVDC LCC solution has lower losses and is therefore preferable from the losses point of view. However, many other aspects will inﬂuence the ﬁnal choice of the design, e.g. num- ber of cables required, reliability, life cycle costs, integration intotheonshorepowersystem,etc.Furtherstudiesaretherefore neededtodeterminethepreferreddesignoftransmissionsystem between the offshore wind farm and the shore. N.B. Negra et al. / Electric Power Systems Research 76 (2006) 916–927 927 References [1] Information found at http://www.home.wxs.nl/∼windsh/offshore.html (last visit January 2005). [3] R. Gasch, J. Twele, Wind Power Plants: Fundamentals, Design and Oper- ation, Solar praxis AG, Germany, 2001. [5] J. Todorovic, Losses Evaluation of HVAC Connection of Large Offshore Wind Farms, Master Thesis, Royal Institute of Technology, Stockholm, Sweden, December 2004. [6] N. Barberis Negra, Losses Evaluation of HVDC Solutions for Large Offshore Wind Farms, Master Thesis, Royal Institute of Technology, Stockholm, Sweden, January 2005. [7] H. Holttinen, P. Norgaard, A multi-machine power curve approach, in: Presented at Nordic Wind Power Conference 1–2 March 2004, Chalmers University of Technology, G ¨oteborg, Sweden, 2004. [8] H. Brakelmann, Efﬁciency of HVAC power transmission from offshore- windmills to the grid, in: IEEE Bologna PowerTech Conference, Bologna, Italy, June, 2003, pp. 23–26. [9] H. Brakelmann, Loss determination for long three-phase high-voltage submarine cables, ETEP (2003) 193–198. [10] S. Lundberg, Performance Comparison of Wind Park Conﬁgurations, Ph.D. Thesis, Department of Electric Power Engineering of Chalmers University of Technology, G ¨oteborg, Sweden, 2003. [11] F. Rudolfsen, G.E. Balog, G. Evenset, Energy transmission on long three core/three foil XLPE power cables, in: JICABLE—International Conference on Insulated Power Cables, 2003. [15] P. Cartwright, L. Xu, C. Sasse, Grid integration of large offshore wind farm using hybrid HVDC transmission, in: Presented at Nordic Wind Power Conference, Chalmers University of Technology, Goteborg, Swe- den, 2004. [16] Losses of converter station,www.siemens.com (last visit January 2005). [17] List, of projects found at www.abb.com (last visit January 2005). [18] IEC Conductors on Insulated Cables, IEC Standards 60228, 1978. [19] IEC Electric Cables, IEC Standards 60287, 1994. [21] B.D. Railing, J.J. Miller, P. Steckley, G. Moreau, P. Bard, L. Ronstr ¨om, J. Lindberg, Cross sound cable project—second generation VSC tech- nology for HVDC, in: Paper presented at the Cigr´e conference, Paris, France, August 29–September 03, 2004. [22] I. Mattsson, A. Ericsson, B.D. Railing, J.J. Miller, B. Williams, G. Moreau, C.D. Clarke, Murray link—the longest underground HVDC cable in the world, in: Paper B4-103 presented at the Cigr ´e conference, Paris, France, August 29–September 03, 2004. [23] T. Ackermann (Ed.), Wind Power in Power System, Wiley, UK, 2005. [24] L. Lazaridis, A. Ackermann, Economic comparison of HVAC and HVDC solutions for large offshore windfarms under special consideration of reliability, in: Presented at the Fifth International Workshop on Large-Scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms, Glasgow, Scotland, April, 2005.