Power plant control in large-scale photovoltaic plants: Design, implementation and validation in a 9.4 MW photovoltaic plant

Power Plant Control in Large Scale PV Plants. Design, implementation and validation in a 9.4 MW PV plant Eduard Bullich-Massagu´ e1, Ricard Ferrer-San-Jos´ e1, M` onica Arag¨ u´ es-Pe˜ nalba1, Luis Serrano-Salamanca2, Carlos Pacheco-Navas2, Oriol Gomis-Bellmunt1 1CITCEA-UPC, Electrical Engineering Department, Technical University of Catalonia, Diagonal 647 Planta 2, Barcelona, Spain 2GreenPowerMonitor, Avda. Josep Tarradellas 123-127, 08029 Barcelona, Spain Keywords: PV plant, control, modelling, simulation, grid code Abstract The paper proposes an algorithm for active and reactive power management in large PV power plants. The algorithm is designed in order to fulfil the requirements of the most demanding grid codes and combines the utilisation of the PV inverters, fixed switched capacitors and STATCOMs. The control algorithm is simulated as required by the grid codes and validated on a real 9.4 MW photovoltaic power plant. 1 Introduction With the electric energy demand increasing and the rising awareness around sustainable growth (e.g. the well-known 20/20/20 objective [1]), renewable energies have experienced a rapid growth in the last few years [2,3]. In the electricity sector, wind power and photovoltaic (PV) power are the technologies with the highest growth in Europe [4]. Currently, the amount of energy generated from PV or wind power has a great importance in the energy mix. With the increase of renewable penetration, the grid support provided by these sources is fundamental. As a result, new grid codes are appearing or being updated, forcing wind and PV power plants to provide grid support [5 – 10]. The most demanding grid codes are normally those of island areas or weak power systems. Power management applied to PV plants has encountered many technical challenges. For instance, the integration of storage systems to deal with the variability of the renewable sources and the appropriate coordination with the power plant control, which has been addressed in [11 – 19]. The authors from [11] propose a control method for a battery energy storage system to be integrated in renewable plants so that the intermittent resource can be dispatched on an hourly basis. In [12], a power plant control for a PV plant is proposed to accomplish grid code requirements, comparing the operation when the PV plant includes storage support and when it does not. Focusing on the ramp rate control, a model to simulate effective dispatch of energy storage units so as to ensure this requirement is shown in [13]. A different approach for PV inverter ramp rate control, also using an integrated energy storage device, is suggested in [14]. It is proposed as a more accurate solution than the traditional moving average method, for allowing to limit the ramp-rate within a desired level. The utilization of PV solar farm inverters as STATCOMs for improving power transfer limits is addressed in [20]. The Low Voltage Ride Through requirement is examined in [21], proposing a control strategy to improve voltage profiles in steady state and when facing load variations at grid buses [16]. The authors from [18] propose a control coordination for capacitor banks and an on-load tap changer in a wind power plant to accomplish the grid code requirements. This proposal is based on the knowledge of the capacitor’s state by the central controller, thus bidirectional communications are required. On the other hand, in [19] an algorithm for the coordinated control of automated devices and photovoltaic generators is presented, based on an optimization approach for minimizing circuit losses and motion of utility controls while solving voltage rise problems. However, the analysis does not take into account the controls needed and their dynamics. The before cited studies analyse specific devices and/or strategies that can enhance the grid integration of PV plants by affecting the power management. A global approach on the active and reactive power controls needed to fulfil the grid codes requirements and their interaction is addressed in this paper, which extends the basic concepts presented in [17]. The control proposed does not need to know the power production state of each converter and the communication system needed is unidirectional. Furthermore, the experience in the PV plant commissioning process is shown and real tests results are presented to validate the algorithms proposed. 1 Figure 1 shows the sequence followed to manage the PV plant project development in different countries as Romania, South Africa or the U.S.A. Grid code requirements have implications in PV plant design and control. Most of the plants to be controlled have already been constructed, so the focus is to design a control and, if needed, to redesign the PV plant adding, for instance, FACTS devices. After addressing the control algorithms, transmission system operators (TSO) require simulation models of the PV plants including their control. So, the corresponding models are made in PSS/E ® and DIgSILENT Power Factory ® software, as it is indicated in most grid codes [6 – 8]. After performing some simulations and validating the grid code compliance, the implementation is permitted and own tests are made before the TSO performs the validation tests to consider the PV plant able to be operative. Due to different PV plants studied, a general control model is designed using the typical ancillary devices such as FACTS and capacitor banks. Grid code requirements to ensure grid stability PV plant + Control design TSO claims simulation model (usually PSS/E or Digsilent Power Factory) Stability studies PPC implementation + own tests Test certification Operational PV plant No acceptance Acceptance No acceptance No acceptance Acceptance Section III Section IV Performed by TSO Section II PV plant layout Fig. 1: PV plant control design and implementation process 2 Power Plant Control Design 2.1 PV Plant Description Although there is no clear categorization on PV plants size according to the installed capacity, the ones considered in this study could be classified as large scale PV plants for presenting an installed capacity of 9.4 MW, which is in the range from several MW to GW, considered large scale [22]. As shown in Figure 2, each PV inverter is associated to a PV string and connected to a three winding transformer. This transformer elevates the voltage from low to medium voltage and is connected to an internal PV collection grid (usually in tree or ring configuration). Ancillary devices as FACTS or capacitor banks are connected to the internal PV grid or to a collector bus close to the MV/HV transformer. The MV/HV transformer connects the PV collection grid to the high voltage transmission system. Despite having local controls, it is necessary to coordinate PV inverters together to achieve the desired setpoints at the point of common coupling (PCC). Hence, a power plant control (PPC) must act as a master to drive all PV plant devices. In this way, the PPC will read the measurements from the PCC and will send orders (active and reactive power setpoints) to all inverters or FACTS, as well as connection/disconnection orders to capacitor banks if they are present in the PV plant. Then, the inverters will perform their own controls to follow their master (PPC) orders. Only in the case of fault ride through (FRT), inverters and FACTS will omit the PPC orders. This is due to the fact that grid codes require a rapid response during fault events where a communication delay would result in the PV plant being non-compliant to the FRT requirement. Fig. 2: Typical large scale PV plant layout including the proposed power plant control schemes 2.2 Control Requirements Grid code requirements [5 – 10] can be summarized in i) voltage regulation actions, ii) frequency regulation actions, iii) FRT actions and iv) ramp rate restrictions: i) Voltage regulation actions: the PV power plant is required to help maintaining the grid voltage level. A minimum reactive power capability of the PV power plant is established. Additional ancillary equipment, as FACTS devices, can help to reach the capability limits. Depending on the TSO needs, the actions required in voltage regulation can be chosen from: •Reactive power setpoint: the TSO sends a reactive power setpoint that must be exchanged at the PCC. • Voltage regulation by droop curve: the TSO specifies a droop curve which consists of predefining the reactive power depending on the voltage level at the PCC. •Power factor setpoint: the TSO sends a power factor setpoint to be established at the PCC. ii) Frequency regulation actions: the frequency support is required to maintain the grid frequency between specified ranges around its nominal value. The frequency support may require, depending on the country, some kind of energy storage system [8]. The basic requirements in this field may be summarized as: •Active power curtailment: the TSO sends an active power setpoint to be injected at the PCC. • Frequency regulation by droop curve: The TSO specifies a curve which predefines an increase or decrease of the active power delivered at PCC as a function of the measured frequency. In absence of power reserve provided by energy storage system or auxiliary generation system (e.g. diesel generator), the increase of power generation when the plant is operating at its maximum power point (MPP) cannot be done. So, agreements with TSO are performed and in these conditions (MPP operation and absence of reserve) underfrequency droop curve is not applied. Note that the frequency droop function is also required to be applied during curtailment events. In this case, underfrequency support can be performed. iii) FRT actions: the fault support specifies requirements under abnormal conditions at PCC. The main specifications in this field are a dynamic reactive power injection requirement under fault conditions, and the time that the power plant must remain connected to the grid depending on the voltage and frequency levels reached during the fault. iv) Ramp rate restrictions: The active power variation may be restricted to a ramp rate when transitions (like curtailment setpoint) occur if the plant does not include energy storage systems [6, 7]. When a power plant is provided with energy storage systems as required in [8], it is possible to limit the power output variation at any time. Ramp rates also may be applied to reactive power output [7]. 2.3 Power Plant Control Solution Taking into account that PV inverters have the capability to perform their own local controls following active and reactive power setpoints, the PPC will generate these setpoints in order to achieve the desired value at PCC. PV inverters including their local control are already built. So, only the PPC, which drives the voltage and frequency support actions listed above is described here. The FRT requirement is fulfilled by the local controls. The active power control scheme is shown in Figure 2. The control is divided in the reference computation block, the controller and the dispatch system. The reference computation block calculates the active power setpoint that must be achieved at PCC. Despite the TSO may send a curtailment setpoint, PT SO , a frequency droop is applied continuously so that it modifies the desired setpoint at PPC, P∗ pre−ramp . Furthermore, there is a ramp rate limitation provided by the grid code. So, P∗ pre−ramp is limited by a ramp rate controller which computes the desired active power at the PCC, P∗. If there is not a curtailment event, PT SO is set to the nominal PV plant power, Pplant . The frequency droop curve is set in the most generic shape which corresponds to that described in [6] and shown in Figure 3(a), where Pavailable is the maximum available active power, PT SO is the TSO curtailment setpoint, Pmin is the active power that the PV plant has to deliver when a maximum overfrequency deviation, fmax , occurs (for frequencies over fmax it is permitted to disconnect), f4 establishes when the overfrequency droop finishes and Pmin must be delivered, fn is the nominal frequency (the TSO can modify it sightly according to its necessities), f2 and f3 determine a deadband zone where the frequency droop is not applied, f1 establishes when the underfrequency droop finishes and fmin is the maximum underfrequency deviation where the PV plant must remain connected. The definition of the frequency droop curve is done according to [6], where TSOs specify the dead band, fmin , Pmin , fmax , Droop 1 and Droop 2. As mentioned before, agreements with TSOs are made to implement the frequency droop curve in absence of power reserves. In the case of the PV plant operating at the MPP (no curtailment required), P=Pavailable ⩽PT SO =Pplant . Under this condition, PT SO is greater than Pavailable and the TSOs have agreed to implement the curve depicted in Figure 3(b). In this situation, once the frequency exceeds the threshold, f3 , PT SO is fixed at the current active power value and the over frequency droop operation is performed. During curtailment events, PT SO < Pavailable and the curve of Figure 3(a) is implemented. Once P∗ is obtained, the controller computes the aggregated power, Ptot , that must be generated by all PV inverters. The controller is based on a typical PI controller which ensures the error between P∗ and the measured power at PCC, P ,tobe0ina steady state. The dispatch system is applied using p.u. signals as in [23]. However, the present approach does not need any information of the available power. The dispatch system takes the Ptot and distributes it among all PV inverters. It is dispatched in a per unit system so that there is only 1 signal to be sent despite different PV inverter power ratings. In this way, Ptot is divided by the nominal PV plant power, Pplant , to obtain α that is sent to all inverters. Each inverter ireceives the α signal and computes its local active power setpoint according to the expression (1). P∗ inv,i =α·Pnom,i (1) Where Pnom,i and P∗ inv,i are the nominal active power and the local active power setpoint of the inverter irespectively. The reactive power control is performed similarly to the active power control. Figure 2 depicts its corresponding scheme. In addition to PV inverters, FACTS devices or capacitor banks are commonly found in a PV plant. So, the control is designed for a generic PV plant which can contain all these elements. To do so, a priority criteria has been established. First, capacitor banks are managed to deliver the major part of reactive power (only when capacitive power is required). These banks deliver discrete blocks of reactive power so, the fine regulation is performed by FACTS and PV inverters. FACTS have priority over PV inverters as they are installed for this particular application. However, when a FACTS device reaches a specified level of reactive power (not necessarily its nominal power) the remaining amount of reactive power is delivered by both (FACTS and PV inverters). Contrary to the frequency regulation actions, the voltage regulation actions do not require simultaneous operations as for example reactive power setpoint plus voltage droop. So, a mode selector is implemented to determine the way to calculate the desired reactive power setpoint, Q∗ pre−ramp . If the TSO sends a reactive power setpoint, QT SO , then Q∗ pre−ramp =QT SO . When power factor setpoint is set, the corresponding desired reactive power is calculated as (2) . When a voltage droop mode is set, the Q∗ pre−ramp is calculated according to a curve depicted in Figure 3(c). In this case, due to the whole plant operation, it is needed to filter the voltage measurement, V , to obtain the droop input, V0 . This filtering is to avoid multiple connections/disconnections of the capacitor banks (a connection of a capacitor bank provokes a voltage increase and so, a decrease of Q0 and the corresponding capacitor disconnection). With this filter and an hysteresis applied to capacitor bank dispatcher, the multiple connections/disconnections are avoided. When there are not capacitor banks, the time constant of the filter is set to 0. Q∗ pre−ramp =P·sin (ϕ)T SO cos (ϕ)T SO (2) Where Pis the measured active power at PCC and cos (ϕ)T SO is the power factor setpoint. Once Q∗ pre−ramp is obtained, it can be limited (or not, depending on the grid code) by a ramp rate limiter obtaining the desired reactive power at PCC, Q∗ . At this point, if Q∗ is capacitive, capacitor banks (if they are available) generate the major part of Q∗ . This is performed by taking the setpoint and calculating the number of capacitors to be connected in the capacitor bank dispatcher. The connection orders of capacitor banks are set according to the following criterion (3) and (4) and are represented in Figure 4. Connection/disconnection orders for the i-th capacitor bank: SETCAPi=Q∗>(i−0.4) ·QCAP (3) RESETCAPi=Q∗<(i−0.6) ·QCAP (4) Where QCAP is the reactive power supplied by a capacitor bank at nominal voltage. Then, the finer control is performed first by FACTS and with PV inverters afterwards. A factor K∈[0,1] determines the amount of reactive power that is supplied only by FACTS devices. In a first stage, QF ACT S1 is calculated according to (5) with a maximum absolute value of K·QF ACT S , where QF ACT S is the nominal reactive power of the FACTS device. QF ACT S1=Q∗−N·QCAP (5) Where Nis the number of capacitor banks connected. Then, the controller computes the rest of the reactive power that FACTS plus PV inverters have to supply, Qtot . It is performed by a PI controller as shown in Figure 2, and the corresponding p.u. value β is calculated by dividing Qtot by Qplant , where Qplant is the nominal reactive power of the PV plant. At this point, as the available reactive power remaining in the FACTS device is (1 −K)·QF ACT S , the additional part of FACTS contribution is calculated as β·(1 −K)·QF ACT S . The total reactive power setpoint to the FACTS device in a per unit system is calculated as (6). γ=QF ACT S1+QF ACT S ·β·(1 −K) QF ACT S (6) Each PV inverter ireceives the βsignal and computes its local reactive power setpoint according to the expression (7). Q∗ inv,i =β·Qnom,i (7) Active Power Frequency Pavailable PTSO Pmin fmin f1f2f3 fnf4fmax Dead band Droop 1 Droop 2 (a) Generic frequency droop curve Active Power Frequency PTSO=Pavailable Pmin fmin f1f2f3 fnf4fmax Dead band (b) Frequency droop curve in absence of curtailment event Q’ V’ Dead band Capacitive Inductive Qmax Qmin VTSO (c) Voltage droop curve Fig. 3: Droop curves for frequency and voltage regulation Q* QCAP 2·QCAP 3·QCAP (i-1)·QCAP 0.6·QCAP 1.6·QCAP 2.6·QCAP (i-0.4)·QCAP i·QCAP ((i-1)-0.4)·QCAP 0.4·QCAP 1.4·QCAP 2.4·QCAP (i-0.6)·QCAP ((i-1)-0.4)·QCAP QCAP 2·QCAP 3·QCAP (i-1)·QCAP i·QCAP Q delivered by Capacitor banks disc3 con3 Disci-1 Disci coni-1 coni con2 con1 disc2 disc1 Fig. 4: Capacitor bank connection/disconnection criterion Where Qnom,i and Q∗ inv,i are the nominal reactive power and the local reactive power setpoint of the inverter irespectively. The FACTS device receives the γsignal and computes its local setpoint according to (8). Q∗ F ACT S =γ·QF ACT S (8) Where QF ACT S and Q∗ F ACT S are the nominal reactive power and the local reactive power setpoint of the FACTS device. Remark: for under frequency events, the PV plant will tend to operate at its MPP, while for over frequency operation, the plant will not run at its full capacity and PV power will be wasted. Despite this drawback, the system stability must be ensured. When large amount of PV or any intermittent power generation is (and will be) connected to the grid, this stability will not only depend on the conventional generation response but also on the operation of these renewable power plants. A waste of PV power will be a requirement to maintain the grid frequency between the limits and hence, the power quality during these events, except if there are energy storage systems capable of shifting the power generation. The energy storage systems are still expensive and most of them under development, demonstration or early commercialized [24]. 3 Modelling and Simulation All controls explained above have been modelled in a way that they can be treated as a black box where the user (TSO) can connect the required measurements and the outputs ( α, β, γ and capacitor banks orders) to the required devices. Most of the parameters detailed in section III are configurable: droop curves, PI controller parameters KP , Ki and Kw (antiwindup constant), ramp rate limits, sample times, communication delays, etc. 3.1 PV Plant Modelling Aspects The active and reactive power management algorithm model has been created in FORTRAN language for PSS/E ® software where corresponding simulations are performed in RMS values. The need to use this software comes from the grid operator that includes it into the corresponding grid code. The PV plant model corresponds to the Vanju-Mare PV plant (Figure 5). The PV plant is located in Romania close to the village of Bucara covering a total area of 23.4 ha ( 234 ·103m2 ) [25]. It consists of 15 PV inverters with a total peak power of 9.4 MW [26]. The PV inverters are connected to a 20 kV PV collection grid in ring configuration and then, to a 110 kV transmission grid through a MV/HV transformer. Tables 1-5 summarize the simulation model parameters. The PV inverters are the SMA Sunny Central HE series (SMA500HE and SMA630HE). These inverters are voltage source inverters (VSI) and are classified as high-frequency, pulse-width modulated current-regulated inverters. A STATCOM (GPCOM model) of 2 MVar is added at node 91. When capacitor banks are used, they are connected at bus 100. Node Name Voltage [kV] 1 XFMR.STAT.1 20 2 XFMR.STAT.2 20 3 XFMR.STAT.3 20 4 XFMR.STAT.4 20 5 XFMR.STAT.5 20 6 XFMR.STAT.6 20 7 XFMR.STAT.7 20 8 XFMR.STAT.8 20 11 SMA500HE.01 0.270 12 SMA500HE.02 0.270 21 SMA630HE.03 0.315 22 SMA630HE.04 0.315 31 SMA630HE.05 0.315 32 SMA630HE.06 0.315 41 SMA630HE.07 0.315 42 SMA630HE.08 0.315 51 SMA630HE.09 0.315 61 SMA500HE.10 0.270 62 SMA500HE.11 0.270 71 SMA630HE.12 0.315 72 SMA630HE.13 0.315 81 SMA500HE.15 0.270 82 SMA500HE.16 0.270 91 GPCOM 0.700 100 OUT.PV.PCC 20 101 CEZ.POI 20 102 VANJ.MAR.POI 110 301 BANOVITA 110 501 MV.XFMR.09 20 502 MV.XFMR.10 20 503 HV.XFMR.09 110 504 HV.XFMR.10 110 Table 1: Grid nodes Node Name Voltage [kV] Short circuit power [MVA] Short circuit ratio (X/R) 102 Vanj.MAR.POI 110 1000 10 Table 2: Equivalent grid data Node 1 Node 2 Vp[kV] Vs[kV] Snom [MVA] r [p.u] x [p.u] 503 501 110 20 16 0.0 0.676 504 502 110 20 10 (out of service) 0.0 0.112 100 91 20 0.69 2 0.0 0.06 5 51 20 0.315 0.63 0.0114 0.0589 Table 3: 2 winding transformer data Node 1 Node 2 Node 3 Vp[kV] Vs[kV] Vt[kV] Snom [MVA] r+jx [p.u] 1 (Primary) 0.0053+j0.0592 (p-s) 1 11 12 20 0.27 0.27 0.5 (secondary) 0.0059+j0.0597 (p-t) 0.5 (tertiary) 0.0055+j0.0592 (s-t) 1.26 (Primary) 0.0050+j0.0598 (p-s) 2 21 22 20 0.315 0.315 0.63 (secondary) 0.0042+j0.0599 (p-t) 0.63 (tertiary) 0.0051+j0.0598 (s-t) 1.26 (Primary) 0.0050+j0.0598 (p-s) 3 31 32 20 0.315 0.315 0.63 (secondary) 0.0042+j0.0599 (p-t) 0.63 (tertiary) 0.0051+j0.0598 (s-t) 1.26 (Primary) 0.0050+j0.0598 (p-s) 4 41 42 20 0.315 0.315 0.63 (secondary) 0.0042+j0.0599 (p-t) 0.63 (tertiary) 0.0051+j0.0598 (s-t) 1 (Primary) 0.0053+j0.0592 (p-s) 6 61 62 20 0.27 0.27 0.5 (secondary) 0.0059+j0.0597 (p-t) 0.5 (tertiary) 0.0055+j0.0592 (s-t) 1.26 (Primary) 0.0050+j0.0598 (p-s) 7 71 72 20 0.315 0.315 0.63 (secondary) 0.0042+j0.0599 (p-t) 0.63 (tertiary) 0.0051+j0.0598 (s-t) 1 (Primary) 0.0053+j0.0592 (p-s) 8 81 82 20 0.27 0.27 0.5 (secondary) 0.0059+j0.0597 (p-t) 0.5 (tertiary) 0.0055+j0.0592 (s-t) Table 4: 3 winding transformer data Node 1 Node 2 Line R [Ω] Line X [Ω] Line C [µF] 1 2 0.0244 0.0201 0.0597 2 3 0.0863 0.0107 0.2111 3 4 0.0293 0.0241 0.0716 4 6 0.0494 0.0407 0.1209 5 6 0.0244 0.0201 0.0597 5 7 0.0731 0.0603 0.0179 7 8 0.0540 0.0445 0.1322 1 100 0.0423 0.0348 0.1034 8 100 0.1313 0.1082 0.3213 100 101 0.3605 0.2013 1.771 102 301 0.0000 0.0121 0.0000 Table 5: Line data 0 1000 2000 3000 4000 5000 6000 7000 8000 0 2000 4000 6000 8000 10000 Time [s] Active Power [kW] 5800 5900 6000 2500 3000 3500 Time [s] Active Power [kW] 6600 6800 7000 1500 2000 2500 3000 Time [s] 8200 8300 8400 8500 2200 2400 2600 2800 3000 3200 Time [s] Measured Active Power TSO setpoint (a) Active power response 0 1000 2000 3000 4000 5000 6000 7000 8000 −2000 −1500 −1000 −500 0 500 Time [s] Reactive Power [kVar] Measured Reactive Power TSO setpoint 3700 3800 3900 4000 4100 4200 −1500 −1000 −500 0 Time [s] Reactive Power [MVar] 4500 4600 4700 4800 4900 5000 −2000 −1500 −1000 −500 0 500 Time [s] (b) Reactive power response Fig. 10: Active and reactive power response in Vanju-Mare PV plant A generic PSS/E ® (and DIgSILENT Power Factory ® ) power plant controller model has been created to be used by system operators or other users. It should allow them to perform their own studies. Some simulation results have been presented showing its appropriate behaviour. 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 0.92 0.94 0.96 0.98 1 Time [s] Power Factor Measured PF at PCC PF setpoint 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 −1000 −500 0 500 Time [s] Reactive Power [kVar] Measured Reactive Power Calculated setpoint 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 2000 4000 6000 8000 10000 Time [s] Active Power [kW] Measured Active Power Active Power setpoint (a) Power factor 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 0.92 0.94 0.96 0.98 1 Time [s] Power Factor Measured PF at PCC PF setpoint 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 −1000 −500 0 500 Time [s] Reactive Power [kVar] Measured Reactive Power Calculated setpoint 8000 8050 8100 8150 8200 8250 8300 8350 8400 8450 8500 2000 4000 6000 8000 10000 Time [s] Active Power [kW] Measured Active Power Active Power setpoint (b) Reactive power (c) Active power Fig. 11: Power Factor response in Vanju-Mare PV plant Finally, after monitoring a 9.4 MW Romanian PV plant, real results have been presented. Acknowledgements The authors would like to thank EDPR for providing the PV plant data, which have been utilized in the simulation models as well as for giving permission to publish the results obtained in its real PV plant and Marta Massagu´ e for proofreading this paper. This work has been funded by the Spanish Ministry of Economy and Competitiveness under the projects ENE2012-33043 and ENE2013-47296. This research was co-financed by the European Regional Development Fund (ERDF). The research leading to these results has received support of the Secretaria d ' Universitats i Recerca del Departament d ' Economia i Coneixement de la Generalitat de Catalunya and has been co-funded by the European Social Fund. The research leading to these results has received funding from the European Union Seventh Framework Program FP7-ICT-2013-11 under grant agreement 619610 (Smart Rural Grid). References [1] European Commission. Communication from the commission to the European Council and the European Parliament - An energy policy for Europe. Commission of the European Communities, 2007. [2] Medium-Term Market Report Executive Summary 2014. Market Analysis and Forecasts to 2020. IEA, International Energy Agency. [3] Rabia Ferroukhi et al. REthinking Energy, 2014. IRENA, International Renewable Energy Agency, 2014. [4] Jacopo Moccia Ivan Pineda, Sarah Azau and Justin Wilkes. Wind in power. 2013 European statistics. EWEA, European Wind Energy Association, 2014. [5] Berndt et al. Transmission code 2007 - Network and system rules of the German transmission system operators. Verband der Netzbetreiber - VDN - e.V. beim VDEW, 2007. [6] Grid connection code for renewable power plants (RPPs) connected to the electricity transmission system (TS) or the distribution system (DS) in South Africa. 2012. [7] Technical transmission grid code of the Romanian power system. Romanian Power Grid Company TRANSELECTRICA S.A., 2004. [8] Minimum technical requirements for interconnection of photovoltaic (PV) facilities. Puerto Rico Electric Power Authority. PREPA, 2012. [9] Ben Noone. PV integration on Australian distribution networks. Literature review. The Australian PV Association, 2013. [10] Standard large generator interconnection agreement (LGIA) among interconnection customer and California Independent system operator corporation. [11] S. Teleke, M.E. Baran, S. Bhattacharya, and A.Q. Huang. Rule-based control of battery energy storage for dispatching intermittent renewable sources. IEEE Transactions on Sustainable Energy, 1(3):117–124, Oct 2010. [12] H. Gaztanaga, J. Landaluze, I. Etxeberria-Otadui, A. Padros, I. Berazaluce, and D. Cuesta. Enhanced experimental pv plant grid-integration with a mw lithium-ion energy storage system. In 2013 IEEE Energy Conversion Congress and Exposition (ECCE), pages 1324–1329, Sept 2013. [13] R. van Haaren, M. Morjaria, and V. Fthenakis. Utility scale pv plant variability and energy storage for ramp rate control. In 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC), pages 0973–0979, June 2013. [14] M.J.E. Alam, K.M. Muttaqi, and D. Sutanto. A novel approach for ramp-rate control of solar pv using energy storage to mitigate output fluctuations caused by cloud passing. IEEE Transactions on Energy Conversion, 29(2):507–518, June 2014. [15] Guishi Wang, M. Ciobotaru, and V.G. Agelidis. Power smoothing of large solar pv plant using hybrid energy storage. IEEE Transactions on Sustainable Energy, 5(3):834–842, July 2014. [16] Weidong Xiao, K. Torchyan, M.S. El Moursi, and J.L. Kirtley. Online supervisory voltage control for grid interface of utility-level pv plants. IEEE Transactions on Sustainable Energy, 5(3):843–853, July 2014. [17] E. Bullich-Massagu ´ e, M. Arag ¨ u ´ es-Pe ˜ nalba, R. Ferrer-San-Jos ´ e, L. Serrano-Salamanca, C. Pacheco-Navas, and O. Gomis- Bellmunt. Power plant control experience in large scale pv plant. Modelling, control, simulation and implementation. In 4th International Workshop on Integration of Solar into Power Systems, Berlin, November 2014. [18] Jorge Mart ´ ınez, Pedro Rodriguez, Philip C. Kjær, and Remus Teodorescu. Design and coordination of a capacitor and on-load tap changer system for voltage control in a wind power plant of doubly fed induction generator wind-turbines. Wind Energy, 15(4):507–523, 2012. [19] Jaesung Jung, Ahmet Onen, Reza Arghandeh, and Robert P. Broadwater. Coordinated control of automated devices and photovoltaic generators for voltage rise mitigation in power distribution circuits. Renewable Energy, 66(0):532 – 540, 2014. [20] R.K. Varma, S. Rahman, and T. Vanderheide. New control of pv solar farm as statcom (pv-statcom) for increasing grid power transmission limits during night and day. IEEE Transactions on Power Delivery, PP(99):1–1, 2014. [21] Liu Yaoyuan, Zeng Chengbi, Miao Hong, and Fu Wenwen. Research on a new method to achieve low voltage ride through of pv. In 2014 International Conference on Power System Technology (POWERCON), pages 1064–1070, Oct 2014. [22] IEA PVPS Task 8: Project Proposals on Very Large Scale Photovoltaic Power Generation (VLS-PV) Systems in Deserts, volume 2. May 2006. [23] M. Datta, T. Senjyu, A. Yona, T. Funabashi, and Chul-Hwan Kim. A coordinated control method for leveling pv output power fluctuations of pv-diesel hybrid systems connected to isolated power utility. Energy Conversion, IEEE Transactions on, 24(1):153–162, March 2009. [24] Xing Luo, Jihong Wang, Mark Dooner, and Jonathan Clarke. Overview of current development in electrical energy storage technologies and the application potential in power system operation. Applied Energy, 137:511 – 536, 2015. [25] Ehs annual monitoring report. 2012. [26] http://en.grupotsk.com/proyectos/vanju-mare [accessed in january 2015]. [27] http://www.schrack-seconet.com [accessed in january 2015].