Preparation of Papers in a Two-Column Format for the 21st Annual Conference of the IEEE Industrial Electronics Society

M. Ahmed, M. EL-Shimy, and M. A. Badr. Sizing of reactive power compensators for onshore and offshore grid connected wind farms. Industry Academia Collaboration (IAC) Conference, 2015, Energy and sustainable development Track, Apr. 6 8, 2015, Cairo, Egypt. http://www.iacconf.com/ Sizing of reactive power compensators for onshore and offshore grid connected wind farmsM. Ahmed1, M. EL-Shimy*1, and M. A. Badr2

1Electrical Power and Machines Department, Ain Shams University, Abbassia, Cairo, 11517, Egypt

2Faculty of Engineering,Future University in Egypt (FUE), New Cairo, Egypt

Corresponding author: shimymb@yahoo.com; 002 01005639589

Abstract - The analysis of reactive demands of high voltage transmission links connecting wind farms to the central grid is considered in this paper. Both High Voltage AC (HVAC) and High Voltage DC (HVDC) options as well as overhead and cable transmission alternatives are considered in the analysis. Therefore, the presented work is applicable for both onshore and offshore installations considering various transmission technologies. The sizing of the required reactive power compensators for the transmission system is the main objective of the manuscript. The target is to keep acceptable operating voltage limits through appropriate amounts of reactive power injection or absorption at the wind farms interface bus. The considered wind farm is made up with DFIGs. Based on the reactive power capability limits of the DFIGs comprising the wind farm, the minimum rating, and type of external reactive power compensating devices are determined for various transmission options. In addition, the reactive power loading on the compensators at various operating conditions of the wind farm is determined.

Index Terms - Wind power; DFIG; transmission alternatives; loading capability limits; reactive power compensators.

I. IntroductionFrom interconnection system point of view, wind farms are connected to the grid by either overhead or cable transmission lines. Overhead transmission is usually used in onshore installations while the cable transmission is usually used in offshore installations. Most current transmission systems are HVAC while HVDC option is its counterpart. The choice between HVAC and HVDC alternatives is usually based on techno-economic selection criteria. The HVDC alternative is technologically available in either voltage-source converter (VSC-HVDC) or line-commutated converter (LCC-HVDC). Fig. 1 illustrates a guideline for the techno-economic selection of various transmission alternatives.

Fig.1: Choice of transmission technology based on overall system economics [1,4]Generally, as clear from Fig. 1, the HVAC alternative is more techno-economically feasible in comparison with the HVDC option except when the transmitted power is very high (> 400 MW) and/or the transmission distance is very long (> 250 km) [1 - 4]. Other salient technical characteristics and operational constraints of various transmission alternatives can be found in [1 - 7]. Although as shown in Fig. 1 that the LCC-HVDC shows the highest power carrying capability over large distances, its black start is not technically possible while this capability is available with both HVAC and VSC-HVDC. From the grid support point of view, only the VSC-HVDC can absorb or generate reactive power for supporting the grid. This capability is facilitated by the VSC devices. Reactive power compensators are usually needed for grid support in either HVAC and LCC-HVDC installations. With both HVDC alternatives, decoupling of the connected network is available due to the power conversion processes provided by the power converter. The HVAC option does not provide this decoupling capability. From substation space and cost requirements, the HVAC option shows the smallest values in comparison with other alternatives. This is of special importance in offshore applications where the space requirements and associated costs of power substations is one of the main concerns.For both onshore and offshore installations, the wind turbines comprising a wind farm are placed in rows perpendicular to the prevailing wind direction [8]. The major interactions among the turbines result in the energy loss caused by the wind turbine wakes. A wind farm design for the maximum wind energy capture must minimize the wake effects among turbines [9]. The wake effect can be explained considering Fig. 2 [9]. When a uniform incoming wind encounters a wind turbine, a linearly expanding wake behind the turbine occurs. A portion of the free stream wind speed will be reduced from its original speed vup to vdown as shown in Fig. 2(a). As shown in Fig. 2(b), for the given wind direction and placement of turbines, turbine 2 is in the wake of turbine 1. Turbine 3 is in the wake of turbine 1 and turbine 4. Turbine 5 is not affected by the wake of any other turbine. The energy crop maximization and interaction (i.e. wake and turbulence) minimization can be achieved by proper spacing between turbines [9-14]. The spacing definitions are illustrated in Fig. 2(c). It is found that the optimum spacing values are S1 10R and S2 17R where R is the turbine rotor radius shown in Fig. 2(a). With this spacing, the aerodynamic interactions can be neglected and the energy capture is maximized. In this paper, proper spacing between turbines is assumed. Therefore, the wake effect and other aerodynamic interactions are neglected.

(a)

(b)

(c)Fig. 2: Wake effect [9]. (a) Wind turbine wake model. (b) Turbines affected by the wake of the other turbines. (c) Tower spacingA general layout of a grid-connected wind farm is shown in Fig. 3. The grid-interconnection system can be divided into two parts. The first part is the infield connections (or grid) while the second part is the bulk power transmission link [8, 12, 13, 15]. The infield connections are usually cables in either onshore or offshore installations. These connections, gather the produced energy and bring it to a central collection point. The collection point is then tied to the main grid via the bulk power transmission link.

Fig. 3: General layout of grid connected wind farms

The produced energy at the generator terminals is usually of a low voltage (e.g. 690 V). This low voltage is then increased to the level of the infield connection voltage (up to 66 kV [4, 7]). This is provided by turbine transformers that are installed directly in or close to the basement of each wind turbine. At the collection point, the voltage is increased to the grid voltage level by the collection point transformers. In the Zafarana farm in Egypt as an example, the infield connections are provided at an AC voltage of 22 kV level while the collecting substation increases this voltage to an AC voltage of 220 kV which is the voltage of the AC central grid [16]. Previous researches [1, 2] show that the impact of infield connections and the collecting transformers may ne neglected. This assumption is based on the fact that the infield cables usually produce reactive power for all operating conditions due to their high capacitance while the transformers consumes reactive power for all operating conditions. Therefore, in this study, the negligence of the impact infield connections and the collecting point transformers is reasonable for the simplification of the reactive power analysis and compensator sizing. This paper presents a detailed analysis of the reactive power demanded by various transmission systems connecting wind farms to the grid. Both HVAC and HVDC options as well as overhead and cable transmission alternatives are considered in the analysis. The minimum size and type of the required reactive power compensators at the collection point are determined. This is based on the difference between the reactive power capability of the wind farm and the reactive power needed at the collecting point for acceptable voltage level and grid support. The considered wind farm is made up with DFIGs. The reactive power capability of the DFIG and the DFIG-based wind farms is determined according to the recent modeling advances [17]. Standard models of the transmission links are used for the power flow analysis. The PSAT 2.1.9 [18] is used as a simulation environment. In addition, the reactive power loading on the compensators at various operating conditions of the wind farm is determined.

II. The study system and modelingA hypothetical 150 MW wind farm is considered. The layout of the wind farm takes the form shown in Fig. 3. This wind farm is composed of 100 DFIGs. The rating of each DFIG 1.5 MW. It is assumed that the spacing between turbines is large enough for the negligence of the wake effect, turbulence, and other aerodynamic interaction. Therefore, the same value of the wind speed is assumed to affect all the turbines. The reactive power production from the infield cable grid is assumed to approximately compensate the reactive power absorbed by the infield and the collection point transformers. The considered hypothetical wind farm is used in the analysis of onshore and offshore installations. For the onshore installation, the wind farm is connected to the grid via 230 kV overhead HVAC transmission link while the offshore installation is assumed to be connected to the grid via a submarine HVAC transmission link. For both installations the impact of LCC-HVDC cable as used as a transmission link is investigated. A set of transmission lengths as well as the range of the active power production from the wind farms is simulated. These simulations in conjunction with the capability limits of the wind farm are used for estimating the minimum size and operational loading of the required compensators. The modeling and simulation of the capability limits of the DFIGs are based on [17] while the models for the HVAC and HVDC options are based on [18 20]. The parameters of the DFIG are available at [1, 2, 17] while the parameters of various HVAC transmission options (overhead and cables) are available at [21]. The parameters of the LCC-HVDC line is obtained from [22].

III. Simulation and resultsA flowchart showing the main actions in the mathematical simulation of the hypothetical system is shown in Fig. 4. The simulation is performed with the aid of PSAT PSAT 2.1.9 [18].The capability chart of the wind farms is determined based on the model presented in [17] and the results are shown in Fig. 5 where various operational power factor values are considered. It is clear from Fig. 5 that the DFIG wind farm is capable of absorbing reactive power for all active power production values (i.e. wind speed values). The reactive power absorption limit is independent on the active power production except at high values. The wind farm capability of producing reactive power is highly dependent on the active power. In addition, at a power production of 120 MW, the wind farm will only be able to absorb reactive power. The wind farm loading capability region as affected by the PF setting of the grid-side converter (GSC) of the DFIG is also shown in Fig. 5.It is depicted from the figure that the both the lower limits of the capability region and the intersection point between the upper and lower limits are highly sensitive to the PF settings. Changes in the PF cause regular changes in the lower limits while the upper limit is irregularly changed. The sensitivity of the upper limit varies with the total power of the DFIG; the sensitivity increases with the increase in the total output power. Therefore, the inductive reactive power capability of the DFIG is highly sensitive to the changes of GSC PF setting while the capacitive reactive power capability is less sensitive. Considering unity PF as a reference, lagging PFs cause downward shift (increase) in the inductive reactive power capability while leading PFs causes upward (decrease) in this capability. As shown in figure, the impact of the GSC PF on the capacitive reactive power capability is minor and irregular; at a power output of about 90 MW, the impact of the PF settings is reversed. The results shown in Fig. 5 are important for assessing the capability of the DFIG-based wind farm for bus voltage control or reactive power support. If, for example, the reactive power required from the wind farm for keeping the interface bus (or collecting point shown in Fig. 3) voltage with specific limits is within the wind farm capability limits, then no external compensators are needed. Otherwise, reactive

Fig. 4: System simulation flowchart

Fig. 5: capability chart of the considered wind farm

power compensators are required to be installed on the interface bus. This is for supplying the mismatch between the demanded reactive power and the available reactive power from the DFIGs comprising the wind farm. According to the sign of the reactive power mismatch, the proper compensator type will be determined while the maximum mismatch determines the minimum size of the required compensators. If for all power production values, inductive reactive power is to be absorbed by the compensator, then thyristor controlled reactors (TCR) may be the suitable compensator type. On the other hand, if capacitive reactive power is required to be injected for all values of the active power production, then thyristor switched capacitors (TSC) are among the proper compensator types. If both capaciative and inductive reactive power is required during the production of active power values, then compensators such as SVCs or STATCOMs are among the proper alternatives. The sizing and type of proper reactive power compensators as well as their loading during various active power production from the wind farm will be demonstrated by several numerical examples.With the HVAC overhead line option, the total required reactive power at the interface bus is shown in Fig. 6(a) while Fig. 6(b) shows the results of Fig. 6(a) combined with the capability chart of the wind farm.

(a)

(b)

(c)Fig. 6: HVAC overhead lines. (a) Active and reactive power characteristics for various lengths; (b) Line characteristics integrated with wind farm capability limis. (c) Loading and sizing of the required capacitive compensator (TSC) for unity PF operation of the GSCsBased on Fig. 6(b), it is clear that the wind farm is having limited reactive power capability for compensating the requirements of the lines. Table I shows the maximum active power that can be transmitted without the need of reactive power compensators. The results shown in the table are derived from the results shown in Fig. 6(b).

Table IMaximum MW transmitted (or wind farm production) without the need of compensatorsLine length (km)PF of the GSCs of the DFIGs

Unity PF0.95 lag0.9 lag0.95 lead0.9 lead

Maximum MW transmitted (MW)

250150144142157161

200152144142159164

150153144142161165

100154144142162167

It is clear from Table 1 and Fig. 6(b) that operating the GSC of the DFIGs at 0.9 leading PF results in maximization of the active power that can be transmitted over the line without the need of compensators. Since, unity power factor operation is recommended by the German Electricity Association (VDEW) [1, 2, 17], then unity power factor operation is used for sizing the required compensator as shown in Fig. 6(c). The figure shows the operational loading of the required capacitive compensator for various power production values considering various line lengths. The minimum size of the compensator is equal to the maximum loading value. It is clear from Fig. 6(c) that the compensator size increases with the increase of the line length. In addition, it is clear that the reactive power loading on the compensator increases with the increase in the active power transmission (i.e. the wind farm active power production). Now considering the XLPE HVAC transmission cable, the total required reactive power at the interface bus is shown in Fig. 7(a) while Fig. 7(b) shows the results of Fig. 7(a) combined with the capability chart of the wind farm.It is clear from the results shown in Fig. 7 indicates that the connection of the wind farm to the grid requires large inductive reactive power compensation for all power transmission (i.e. power production) values and for all lengths of the cable. This is in contrast with the HVAC overhead line (Fig. 6(c) and Table I) where the magnitude of the required compensating capacity is much less and its usage is limited to the situations of high power production. The continuous loading of the compensator in the HVAC cable system is expected to cause faster wear out (i.e. lifetime reduction) of its components in comparison with the compensator in the HVAC overhead case. In addition, the large size of the compensator in the HVAC cable case is expected to cause much higher interconnection cost in comparison with the HVAC overhead line case. It is also known that the costs associated with cable installations are much higher than those associated with overhead installations. Therefore, the proper selection of grid connection system should be carefully based in a detailed techno-economic analysis.

(a)

(b)

(c)Fig. 7: HVAC XLPE cable. (a) Active and reactive power characteristics for various lengths; (b) Line characteristics integrated with wind farm capability limis. (c) Loading and sizing of the required inductive compensator (TCR) for unity PF operation of the GSCs

With the LCC-HVDC line, the results are shown in Fig. 8. The reactive power is only needed for supplying the converters while the conductors of the line do not need reactive power. This is because DC line conductors are neither absorbing nor generating reactive power. Therefore, the reactive power needed for the LCC-HVDC line connecting the wind farm to the grid is expected to be independent on the line length and the installation type (i.e. overhead or cable); however, of course, the reactive power needed by the converters increases with the increase in their capacity of loading as shown in Fig. 8(b). The results show the LCC-HVDC line requires capacitive compensation. Similar to the HVAC overhead lines (Fig 6), the loading on the compensators is not continuous and starts at high active power transmission, however, the compensator in the HVAC overhead line starts to supply reactive power at an active power transfer that is significantly higher in comparison with the LCC-HVDC. In addition, the size of the capacitive compensator needed for the LCC-HVDC is much higher in comparison with the HVAC overhead line. This is should be considered in the overall techno-economical evaluation of various interconnection alternatives.

(a)

(b)Fig. 8: LCC-HVDC. (a) Line characteristics integrated with wind farm capability limis. (c) Loading and sizing of the required capacitive compensator (TSC) for unity PF operation of the GSCs

IV. ConclusionsThis paper presented a detailed analysis of the reactive power characteristics of DFIG-based grid connected wind farms.Various AC and DC transmission alternatives are considered. In addition, the paper presented a methodology for sizing and type determination of the required reactive compensators for the considered transmission options. Various lengths of the transmission lines are also considered. The results show that HVAC overhead line and LCC-HVDC lines connecting the wind farm to the grid needs capacitive compensators while the HVAC cable option requires inductive compensators. The size of the compensators needed for the HVAC overhead and HVAC cable alternatives is highly dependent on the length of the lines while with the LCC-HVDC option the required compensator size is independent of the line length. This is because reactive compensators in LCC-HVDC installations are mainly needed to supply the reactive power needed by the converter while the line conductors do not need reactive power. The operational loadings of the compensators needed for various transmission alternatives are significantly different. This is for the same active power transmission (i.e. wind farm active power production). In both HVAC overhead and LCC-HVDC lines, the compensators supply reactive power only at high active power transmission; however, the loading and size of the capacitive reactive power compensator for LCC-HVDC based system are much higher in comparison with the HVAC overhead basec system. With the HVAC XLPE cable, the required inductive reactive power compensator is needed for all values of active power transmission and for all considered lengths (100 km 250 km) considering the hypothetical 150 MW wind farm. From the points of view of compensator size and compensator loading, the HVAC overhead transmission option offers the minimum values in comparison with the HVAC XLPE cables and LCC-HVDC lines; however, the determination of the optimal alternative needs careful detailed techno-economic feasibility analysis.

References[1] M. El-Shimy, Reactive power management and control of distant large-scale grid-connected offshore wind power farms, International Journal of Sustainable Energy, vol. 32, no. 5, pp. 449-465, 2013. [2] M. EL-Shimy, Modeling and analysis of reactive power in gridconnected onshore and offshore DFIG-based wind farms, Wind Energy, vol. 17, no. 2, pp. 279295, 2014[3] L. Xu and B. R. Andersen, "Grid Connection of Large Offshore Wind Farms Using HVDC," wind energy, vol. 9, pp. 371382, 2006.[4] T. Ackermann, " Transmission Systems for Offshore Wind Farms," in Wind Power in Power Systems, ed: John Wiley & Sons, 2005, pp. 529-553.[5] T. Ackermann, N. Barberis Negra, J. Todorovic, L. Lazaridis, "Evaluation of Electrical Transmission Concepts for Large Offshore Wind Farms," presented at the In Copenhagen offshore wind conference and Exhibition Copenhagen, 2005.[6] N. Barberis Negra , J. Todorovic , T. Ackermann, "Loss evaluation of HVAC and HVDC transmission solutions for large offshore wind farms," Electric Power Systems Research, vol. 76, pp. 916927, 2006.

[7] F. Renaudin, "Integration and Stability of a Large Offshore Wind Farm with HVDC Transmission in the Norwegian Power System," Master of Science in Electric Power Engineering, Norwegian University of Science and Technology, 2009.[8] P. Bresesti, W. L. Kling, R. L. Hendriks, and R. Vailati., "HVDC Connection of Offshore Wind Farms to the Transmission System," IEEE transactions on energy conversion, VOL. 22, NO. 1, vol. 22, March 2007.[9] J. Schachner, "Power Connections for Offshore Wind Farms," Diploma Thesis, Department of Electrical Engineering, University of Leoben, Austria, 2004.[10] A. Kusiak and Z. Song, "Design of wind farm layout for maximum wind energy capture," Renewable Energy vol. 35, pp. 685-694, 2010.[11] G. Marmidis, S. Lazarou, E. Pyrgioti, "Optimal placement of wind turbines in a wind park using Monte Carlo simulation," Renewable Energy vol. 33, pp. 14551460, 2008.[12] S. Lundberg, "Evaluation of wind farm layouts," Europian Power Electroics and Drives (EPE) vol. 16, 2006.[13] S. Lundberg, "Wind Farm Configuration and Energy Efficiency Studies - Series DC versus AC Layouts," Thesis for the degree of doctor of philosophy, Department of Energy and Environment, Chalmers University of Technology, Sweden, 2006.[14] A. Emami and P. Noghreh, "New approach on optimization in placement of wind turbines within wind farm by genetic algorithms," Renewable Energy vol. 35, pp. 1559-1564, 2010.[15] P. Lazaros Lazaridis, "Economic Comparison of HVAC and HVDC Solutions for Large Offshore Wind Farms under Special Consideration of Reliability," Masters Thesis, Royal Institute of Technology, Stockholm, 2005.[16] New & Renewable Energy Authority (NREA). Available at: http://nrea.gov.eg/english1.html[17] M. Ahmed, M. EL-Shimy, and M. Badr, Advanced modeling and analysis of the loading capability limits of doubly-fed induction generators, Sustainable Energy Technologies and Assessments, vol. 7, pp. 79-90, 2014.[18] Power System Analaysis Toolbox (PSAT). Available at: http://faraday1.ucd.ie/psat.html[19] PSAT manual. Available at: http://licence.ucd.ie/tech/500 [20] F. Milano, Power system modelling and scripting, Springer Science & Business Media, 2010.[21] P. Kundur, Power system stability and control. Edited by Neal J. Balu, and Mark G. Lauby. Vol. 7. New York: McGraw-hill, 1994.[22] N. Barberis, J. Todorovic, and T. Ackermann. "Loss evaluation of HVAC and HVDC transmission solutions for large offshore wind farms." Electric Power Systems Research, Vol. 76, No. 11, 2006, pp. 916-927.