� Abstract— High penetration levels of photovoltaic systems on low voltage electricity nets tend to lead to overvoltage problems. Currently, in the residential sector of the Belgian Flanders region, the only way to reduce those problems is by means of completely halting the inverter. This paper investigates the power quality improvements of a fixed method of gradual active power curtailment in order to minimize overvoltage problems, as well as a basic economic impact on the end user. The method is simulated on 3 models of existing low voltage feeders located in the northern region of Belgium. Results show that overvoltage problems are nullified and unbalance is improved. Index Terms— APC, curtailment, droop, power quality, PV I. INTRODUCTION N the past decades the European Union has focused on replacing traditional power sources by low-carbon generation and renewables. Feed-in tariffs and other consumer profit driven incentives lead to an irrefutable increase of Photovoltaic (PV) installations, amongst other technologies, in the Low Voltage (LV) electricity net [1]. That already leads to overvoltage problems [2]. As the Belgian action plan for renewable energy (NREAP) expects that the allotment of renewable energy will keep rising (from 2.6 GW in 2011) to 11.4 GW by 2020 [3], distribution system operators (DSOs) are facing a more complicated control problem, keeping the LV net within ranges of the European EN50160 standard [4]. That standard states that the 10-minute mean rms value of the supply voltage should be ���� ± 10% for 95% of the time. Furthermore, the 10-minute mean rms value of the Voltage Unbalance Factor (VUF) should be below 2% for 95% of the time. Both rms values are measured on a weekly base. To solve the overvoltage problems, methods as proposed in [5], [6], [7] and [8] could be used, but they require large costs for the DSO, while the overvoltage problems rarely occur on the specific nets. Active Power Curtailment (APC) on the other hand does not require those costs for the DSO. It also does not cost consumers much, although the expected yearly profit from PV yield for some consumers is bound to go down [9]. Furthermore, Europe’s PV market share leader Germany already defines a voltage stabilizing droop method to be implemented in generators in VDE-AR-N 4105 [10], which indicates the necessity of such methods. The Local Intelligent Networks and Energy Active Regions (LINEAR) project of the Flanders region in Belgium investigates the optimal use of renewable energy resources [11]. Through 3 different field trails, a practical smart grid methodology is developed. In [12] 4 characteristic feeder topologies are deducted from a group of 28 representative feeders. II. SCENARIOS A. Grid Topology All APC simulations are performed on three 230/400V, 50Hz reference feeders, based on a real city feeder and a real semi- urban feeder as described in [2], and another real feeder with pseudonym TestFeeder1. All grids are TT earthed. As discussed in [2], the scenarios causing overvoltages where the transformer is set on a tap of 230V are: 50% and 100% PV penetration on 1 phase in both city and semi-urban feeders, and 100% of PV penetration on all phases in the city feeder. However, in reality, the tap is not always at 230V; therefore the simulations of the city and semi-urban feeder include a sensitivity analysis of the tap. The PV penetration levels are calculated by: � = �� Where �: ���� �� ��� ����� �: ������ �� ℎ����� �� ℎ � ��� ���� ��� �: �� �� ������ �� ℎ����� �� ℎ� ������ Reducing Overvoltage Problems with Active Power Curtailment –Simulation Results K. Lemkens, F. Geth, P. Vingerhoets and G. Deconinck KU Leuven Department of Electrical Engineering – ESAT-ELECTA Kasteelpark Arenberg 10, B-3001 Leuven, Belgium Email:
[email protected] I 1 978-1-4799-2984-9/13/$31.00 ©2013 IEEE 2013 4th IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), October 6-9, Copenhagen The different simulations are carried out with 3 goals in mind: 1. Study voltage problems at rising degrees of PV penetration. This is simulated on the city and the semi-urban feeder; 2. Examine the effect of APC on an existing problematic feeder. This is simulated on TestFeeder1; 3. Consider the difference in PV profit between an uncurtailed inverter and a curtailed one for the same scenario. This is also simulated on TestFeeder1. The semi-urban feeder topology is shown in Fig. 1, and Fig. 2 depicts that of the city feeder. The multiple house icons drawn at nodes 3, 27, 30, 33 and 38 indicate apartment buildings at those locations. The number of house icons at one location coincides with the number of households in that specific apartment building. The TestFeeder1 topology is shown in Fig. 3. Fig. 1: Semi-urban feeder topology Fig. 2: City feeder topology Fig. 3: TestFeeder1feeder topology B. Generation & Load The bar graph of Fig. 4 represents the sizes of the PV installations when every house has PV installed. The different scenarios contain: - Scenario A: 50% of the households on phase 1 have a PV installation; - Scenario B: All households on phase 1 have a PV installation; - Scenario C: All households on all phases have a PV installation. Fig. 4: PV location and size on semi-urban feeder The PV setup of the city feeder is graphically represented in Fig. 5. Notice that a specific household always has the same PV size. E.g. household 16 in the city feeder has a PV installation of 4kWpeak in every scenario and simulation. Fig. 5: PV location and size on city feeder The PV setup of TestFeeder1 is shown in Fig. 6. 2 Fig. 6: PV location and size on TestFeeder1 PV as well as load profiles were derived from [13]. Both profiles only consider active power. Therefor the simulations disregard reactive power. Whilst the load profiles are fifteen minute-based profiles, the PV profiles are minute-based. Furthermore, the PV profiles are scaled to match the kWpeak sizes of the respective household. C. Load Flow Analysis The simulations were run with a static load flow analysis tool using the backward-forward sweep method, implemented in MATLAB [14]. The simulation time span is 1 year, with a time step of 1 minute. III. APC METHOD Curtailment, or droop, is a method to force evaluation parameters within their set boundaries, in this case keep the voltage below 1.1 per unit. Reducing voltages at a given location can be achieved by lowering the power injected at that location (depicted in Fig. 7). By only acting on the active power, the method used in these simulations, is an APC method. Furthermore, because the droop set point (the minimum voltage where the algorithm starts regulating) does not change throughout time and is the same at all locations, the method could be called a ‘fixed droop’ scheme. Due to its fixed character, the method requires no communication and, as discussed by [9], is easy to implement on inverters. The output power of the inverter ��� is given by: � ≤ "#��$: ��� = %&&' "#��$ < < '(: ��� = %&&' − �( − '() ≥ '(: ��� = 0 With %&&' the maximum power point tracked power, "#��$ the droop set point, � the droop coefficient, the actual voltage and '( the absolute maximum voltage according to [4]. "#��$, and '( are in per unit with the tap voltage as reference. The droop coefficient � is calculated as follows: � = '( − '( − "#��$ ∗ %&&' Droop sloops at different droop set points (1.05�� to 1.09��) be depicted in Fig. 7. Fig. 7: Droop slopes at different droop set points IV. RESULTS A. Penetration levels The city and semi-urban feeder are used to determine the impact of rising degrees of PV penetration levels. The statistical results of the voltages on phase 1 and the unbalance (VUF), for the base and drooped case, are shown in Fig. 8. The black box defines percentiles 25 to 75, whereas the whiskers define the outer 5 percentiles. Fig. 8: Statistical results of semi-urban feeder (100% PV on phase 1) After applying APC, with "#��$ = 1.05��, all overvoltages are nullified as shown in the 3rd column of Fig. 8. Note that, even though only phase 1 is controlled, not only the voltages on that phase have returned within the constraints of [4], also the unbalance between the 3 phases again complies with the specification. A droop set point at 1.05�� clearly resolves the unbalance and overvoltage problems in the semi-urban feeder, but nevertheless comes with a cost as shown in Fig. 9. It is apparent that PV installations at the far end of the feeder need to curtail a lot more than those closer to the transformer. 3 Fig. 9: Base case yield vs. droop case yield A sensitivity analysis on the tap set point can be combined with different droop set points for a single feeder to a figure as shown in Fig. 10 and Fig. 11. The rising character of each individual graph clearly indicates that a higher tap on the feeding transformer causes the droop mechanism to curtail more energy on a yearly basis. Another conclusion is that a lower droop set point leads to higher losses. Fig. 10: Tap analysis on Semi-urban feeder with different scenarios and droop set points Note that in the Semi-urban feeder, scenario C is not shown, because no overvoltages arise in that scenario. Furthermore, in scenario C of the City feeder, the energy losses rise about 3 times faster than in scenario B. This coincides with the amount of PV in scenario C, which also about 3 times higher than that in scenario B. Fig. 11: Tap analysis on City feeder with different scenarios and droop set points B. Feeder with problems Overvoltages already occur on TestFeeder1 as indicated by Fig. 12. It shows the voltage statistics of phase 1 and 2, as well as the voltage unbalance for 1 year, before and after a voltage droop with 1.08 per unit is applied. As with the example of the semi-urban feeder, applying curtailment on TestFeeder1 not only nullifies all overvoltages, but solves the unbalance incompliance to [4]. Fig. 12: Voltage statistics comparison on TestFeeder1 during 1 year C. Economic consequences Fig. 13 shows the difference in profit for both a standard inverter, and one with curtailment acting on a droop set point of 1.08 per unit when the price per kWh is 0.1€. The simulation accounted for the switching behavior of a PV inverter, according to [4]. It therefore would switch off when the 10-minute mean rms value is above 1.1 per unit. At most households, the profit goes down when applying curtailment. Nevertheless, household 18 prospers because it has no major PV injectors on its phase. Contrary to household 18, household 25 does have a major PV injector on its phase, namely household 16. However, in between household 16 and 25 none of the PV installations are connected to that phase, causing the voltage to drop faster with curtailment than without, thus less need for curtailment at household 25 then in the standard behavior. 4 Fig. 13: PV yield - Standard inverter behavior vs. drooped behavior (1.08 per unit) on TestFeeder1 When a curtailment method, such as the one studied in this paper, is introduced on a feeder, careful consideration must be made as to how to compensate for losses. In the case of entire TestFeeder1 it comes down to about 1200€/year. V. CONCLUSIONS Fixed APC is a quick and easy way to improve the voltage quality on problematic feeders. Not only can overvoltage problems be nullified, but the unbalance between the different phases is also automatically improved. Selecting the optimal droop set point to minimize yield losses is dependent on feeder topology and thus needs to be simulated. By doing so, DSOs can find the perfect setting for keeping their grid within specification for any PV penetration level, without reducing the clients profit too much. On existing feeders, as TestFeeder1, overvoltage problems do arise. APC is a cost-friendly alternative to solve those problems as opposed to double-cabling and other types of current solutions. The major drawback of APC is that not all clients are curtailed equally as it depends on their location on the feeder, phase connection and size of their PV installation. For APC to be accepted, compensating the client ought to be considered. VI. ACKNOWLEDGMENT The work is supported via the project Linear organized by the Institute for Science and Technology (IWT). VII. BIBLIOGRAPHY [1] P. Djapic, C. Ramsay, D. Pudjianto, G. Strbac, J. Mutale, N. Jenkins en R. Allan, „Taking an active approach,” Power and Energy Magazine, IEEE, vol. 5, nr. 4, pp. 68- 77, 2007. [2] C. G. De Miguel, „Integrating DER in LV distribution networks: Simulation results concerning power quality issues,” 2012. [Online]. Available: http://www.linear- smartgrid.be/. [3] „Belgium National Renewable Energy Action Plan,” Federal-Regional Energy Consultation Group CONCERE-ENOVER, Nov. 2010. [4] „Voltage Characteristics of Electricity Supplied by Public Electricity Networks,” CENELEC Std. EN 50 160, Jul. 2010. [5] O. Haas, O. Ausburg en P. Palensky, „Communication with and within Distributed Energy Resources,” Industrial Informatics, 2006 IEEE International Conference on, pp. 352-356, 2006. [6] P. C. Ramaswamy en G. Deconinck, „Smart Grid Reconfiguration Using Simple Genetic Algorithm and NSGA-II,” in 3rd IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Berlin, 2012. [7] L. A. Talat, M. R. Salem en H. M. Soliman, „Voltage control by tap-changing transformers for a radial distribution network,” Generation, Transmission and Distribution, IEE Proceedings, vol. 144, nr. 6, pp. 517- 520, 1997. [8] Y. Ueda, K. Kurokawa, T. Itou, K. Kitamura, K. Akanuma, M. Yokota, H. Sugihara en A. Morimoto, „Advanced analysis of grid-connected PV system's performance and effect of batteries,” Elect. Eng. Japan, vol. 164, pp. 247-258, 2008. [9] R. Tonkoski, L. A. C. Lopez en T. H. M. El-Fouly, „Coordinated Active Power Curtailment of Grid Conencted PV Inverters for Overvoltage Prevention,” IEEE Trans. Sustainable Energy, vol. 2, pp. 139-147 Apr. 2011. [10] „Erzeugungsanlagen am Niederspannungsnetz, Technische Mindestanforderungen für Anschluss und Parallelbetrieb von Erzeugungsanlagen am Niederspannungsnetz,” Verband der Elektrotechnik Elektronik Informationstechnik, Aug. 2011. [11] B. Dupont, P. Vingerhoets, P. Tant, K. Vanthournout, W. Cardinaels, T. De Rybel, E. Peeters en R. Belmans, „LINEAR breakthrough project: Large-scale implementation of smart grid technologies in distribution grids,” in Innovative Smart Grid Technologies (ISGT Europe), 2012 3rd IEEE PES International Conference and Exhibition on, 2012. [12] J. Geuns, C. Gonzalez De Miguel, S. Weckx en T. Wijnhoven, „Grid topology & Feeder selection,” January 2012. [Online]. Available: http://www.linear- smartgrid.be/. [13] E. Peeters, C. Develder, J. Das, J. Driesen en R. Belmans, „Linear: Towards a Breaktrough of Smart Grids in Flanders,” in i-SUP conference, Bruges, Belgium, 2008. [14] C. Cheng en D. Shirmohammadi, „A three-phase power flow method for real-time distribution system analysis,” IEEE Trans. Power Syst., pp. 671-679, 1995. 5 /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 200 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 400 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description > >> setdistillerparams > setpagedevice