the point of common coupling is exceeding the required ... required: dynamic grid support â Fault Ride Through ... The input could also come as a 4-20 mA current- ... Low Voltage Fault Ride Through (4.7) ... seconds). Besides the set-point mode the TLX PRO+ can also run in Q(U) mode in order to have a better voltage.
A 12 MEGAWATT POWER PLANT WITH FULLY IMPLEMENTED ANCILLARY SERVICES ACCORDING TO THE GERMAN GRID CODES – THE FIRST RESULTS Søren Bækhøj Kjær {sbk}, Jörg Dannehl {dannehl}, Frederik Mecking {mecking} and Jens Godbersen {jgo} @danfoss.com Danfoss Solar Inverters Jyllandsgade 28, 6400 Sønderborg, DENMARK, +45 7488 1300
ABSTRACT: Since April 2011 all photovoltaic power plants connected in parallel with the German medium voltage network must comply with the TR3, TR4 and TR8 technical guidelines published by the Fördergesellschaft Windenergie - FGW (Federation of German Windpower and other Renewable Energies). This includes requirements to active- and reactive-power, power quality, connection and disconnection of the plant to/from the network, immunity to faults on the network and requirements for certified simulation models. This paper gives a brief description about the implementation of the required functions (ancillary services) and shows some of the first obtained detailed results from a 12 MV photovoltaic power plant located in Busenwurth, northern Germany, equipped with approximate 720 string-inverters. Keywords: Large Grid-connected PV systems, Plant Control, Ancillary Services, Low Voltage Fault Ride Through, Electrical Characteristic
1
INTRODUCTION
As the amount of installed photovoltaic power in the distribution network is increasing, its impact on the network is also increasing. In some cases the voltage at the point of common coupling is exceeding the required 110% on the low voltage network [1] and even the limit of 102% in the medium voltage network [2], when power is being fed back into the network. In other cases, inverters in the network cease to inject power due to a temporary over-frequency, which was the way of handling over-frequency (over-frequency is an indication of over-production of power in the network) before April 2011. A sudden drop in available power cause the frequency to decay below acceptable values, thus customers (load) are being disconnected from the network in order to restore the frequency [3]. The same was the case when the magnitude of the grid voltage got below some certain value. The required behaviour of the inverters was then to cease inject power immediately. This would for sure not help to restore the voltage, and the immunity of the entire system was pretty low. These drawback and others can be solved be incorporating ancillary services into the PV systems (inverters). According to the grid codes for the German medium voltage network [4] the following services are required: dynamic grid support – Fault Ride Through (FRT); active power control – Power Level Adjustment (PLA); and static grid support by reactive power control, etc. The purpose of this paper is to show how these ancillary services are build into a string-inverter and how the more than 700 inverters are performing together in a 12 MW power plant and also to demonstrate the 200 kVA FRT test-facility at Danfoss Solar Inverters. This paper is divided as follows: The approach to the project is given in section 2, followed by detailed results in section 3 and finally a conclusion is given in section 4. 2
METHODS
2.2 Danfoss’ ancillary service solution The control architecture of a modern ancillary service solution includes the possibility to operation in both
open- and closed-loop mode. In open-loop mode (see Fig. 1), the Distribution Network Operator (DNO) can command the reference for reactive power (Q) or power factor (PF) at the Point of Common Coupling (PCC) and the inverters will respond to the set-point. Hence no measured signals of the Q/PF is fed back to them, thus the steady state performance depends on the accuracy of the inverters and the parasitic elements in the network, e.g. inductive current drawn by transformers and capacitive current drawn by cables, etc. If closed loop (see Fig. 2) is utilised, the Q/PF at the PCC is also measured and fed back to the inverters. Present, the Danfoss TLX PRO+ inverters support open-loop control and closed-loop control by third party. A modern system also includes a security scheme in case that the communication between the DNO and the plant is lost, or if communication lines within the plant is broken. In our case, it is possible to set-up “fall back” values for Q/PF, which then is applied if communication within the plant is obstructed.
• Constant Q or PF “Master”
K1-4
PRO+
PRO+
PRO+
PRO+
• Q, PF and P 4-20 mA
Ethernet Q U
• Set-point curves
PF 1
…up to 100 inverters…
P
PF(p) and Q(u)
Figure 1: Open loop control of power-plant. K1-4 is four relays dictating the amount of active power the PV power plant is allowed to generate (signal comes from the DNO). The input could also come as a 4-20 mA currentloop signal or even through the IEC 61850 communication standard for substation automation or the EN 61400-25 series of standard for communication with wind turbines.
Additional 8 MW power plant
Uk = 20 kV Sa = 154 MVA R = 0.19 Ω XL = 2.59 Ω
∑
TR1
“Master” Set- points from DNO
PRO+
PRO+
PRO+
~ 5 km
PRO+
MV cable 3 * 630 RM / 35 R = 47 mΩ/km XL = 105 mΩ/km
Feedback: U, P, Q Setpoints: Q, PF
TR2
Q U PF 1
P
Uk = 0.4 kV Sa = 2 MVA ΔU = 6% R = 0.884 mΩ XL = 5.95 mΩ
Figure 2: Closed loop control of power-plant. R = 0.43 Ω XL = 3.12 Ω
2.2 Busenwurth Photovoltaic Power Plant The PV power plant studied in this paper is located in Busenwurth, Germany, see Fig. 3 and 4. The electrical diagram of the power plant is depicted in Fig. 3. The plant is divided in seven subsections each of them connected to a LV/MV transformer and with 120 pieces of the TLX 15k PRO+ string inverters (nominal apparent power is 15 kVA). These subsections are connected through a MV cable to the main transformer to the HV network some 5 km away.
1.8 MW installed TLX inverter
TR7
Figure 5: Schematic over the 12 MW power plant in Busenwurth, Germany.
2.3 Requirements for PV Power Plants The following requirements are specified in the TR3 technical guidelines (edition 22, 1-7-2011) [7]: Active Power Provision (4.2) Reactive Power Provision (4.3) Power Quality (4.4) Disconnecting the DER from the Grid (4.5) Verification of Cut-in conditions (4.6) Low Voltage Fault Ride Through (4.7)
3
Figure 3: Busenwurth 12 MW PV power plant from the air, courtesy of Möhring Energie GmbH [8].
TR6
MEASURED RESULTS
In this section the performance of ancillary services on plant level are shown for selected aspects. For further results on the requirements of section 2.3 the reader is referred to the TLX unit certificate [9]. All measurements shown in the following are taken at the low voltage side of the transformers in subsections 1 and 6 with Fluke 435 Three Phase Power Quality Analyser and Dranetz BMI Mavowatt 70 Power Xplorer. Due to restrictions in controllability in the field LVRT results have been obtained in the laboratory. 3.1 Active Power Provision
Figure 4: Detailed look on a part of the 12 MW PV power plant equipped with TLX inverters, courtesy of Möhring Energie GmbH [8].
One of the relevant aspects in terms of active power provision of PV plants is the Power Level Adjustment (PLA). In order to have control of the active power flow in the grid the DNO can command to reduce the active power fed to the grid. According to FGW TR3, 4.2.2 the required accuracy is +/- 10 % for PV plants. The figures 6 and 7 show the plant response in case of PLA. It shows that the required accuracy is kept. It takes a transition time of around two minutes until the PLA level is applied in the whole plant.
the maximum set point are chosen to 40%. Figure 9 shows the measured reactive power vs. its reference as well as the impact on the transformer voltage. It can be seen that it possible to vary the voltage within +/- 5 %.
Figure 6: Plant Behaviour on 2011-05-08 during Power Level Adjustment. Figure 9: Test of steady state performance for reactive power. Figure 10 shows the dynamic response to a change of the reactive power set-point. In worst case it takes maximum 5 seconds to settle to another set-point. In normal conditions the settling time will be much faster, because the set-point change from max. under-excited to max. over-excited is very unlikely.
Figure 7: Zoom during transition to PLA limit. Data here is sampled with each five seconds. The figure 8 shows the ramp up of active power in the plant after the grid voltage was shut down. The active power ramps up with 116 kW per minute this is around 6.5 % and therefore below the limit of 10 %.
Figure 10: Test of dynamic response for Q-set-point. From 0% to 60% over-excited to 60% under-excited to 60% over-excited and back to 0% (measured every 0.5 seconds).
Ramp up in active power and grid voltage after plant reconnection 2011-07-06. 3.2 Reactive Power Provision The TLX+ inverter is capable of generating up to 60% reactive power, both under- and over-excited, corresponding to a PF between 0.8 and 1.0. By the grid code is no higher reactive current requested than PF = 0.9 thus this is used as maximum reference for testing concerning the power factor. At full power (maximum accessible active power during reactive power injection) a PF of 0.9 relates to a reactive power level slightly above 40% of nominal apparent power. Thus for Q(U)
Besides the set-point mode the TLX PRO+ can also run in Q(U) mode in order to have a better voltage control. Figure 11 shows the implemented characteristic used during the tests. A ‘dead-band’ around the nominal value would also be a possible solution. For example that the reactive power reference is zero for +/- 1 or 2 percent. This would prevent that unnecessary reactive power is injected and therefore reduce the losses. However, the Q(U) curve in Fig. 11 was applied. Figure 12 shows the resulting voltage over one entire day. The voltage is kept within 230V and 235V that is around 2%. In the same way the TLX PRO+ can be operated in PF(P) mode. Hence the power factor is adjusted to the actual active power. Figs. 13 and 14 show the implemented characteristic and the results, respectively.
Q/Smax 40 % overexcited
95% 100% 105%
Voltage/ nominal Voltage
40 % underexcited
Figure 11: Applied Q(U) reference curve during test.
term flicker. Therefore the short term flicker (Pst) over a day is included in the following figures, together with the grid voltages and the active power produced. Figure 15 shows a day with lots of fluctuation in the irradiation due to passing clouds and Figure 16 with mainly constant irradiation. These Pst results are internally calculated by the power analyser each 10 minutes based on the measured voltage. This means it includes both the flicker generated in the PV plant and the influence of the flicker present in the medium and high voltage network. The highest flicker values appear during plant start up in the morning and shut down in the evening. The comparison of the figures shows that the fluctuation in irradiation has no significant influence on the flicker values. This fits to the conclusion presented in [5], that the fluctuation on a bigger power plant level has no influence due to shedding and averaging effect of the passing clouds.
underexcited overexcited
Figure 12: Results of Q(U) open loop control 2011-0707 PF
50 %
80 %
100 % active power/ nominal power
0.95 0.90
Figure 13: Applied PF(P) reference curve during test.
Figure 15: Flicker on a day with lot of changes in irradiation: Grid voltage, active power and short term flicker on 2011-07-10.
Figure 14: Results of PF(P) open loop control 2011-0708 3.3 Power Quality According to the BDEW guideline chapter 2.4.3. [2] the long term flicker value is limited to P LT≤0.46. This is the limiting value for generating units at the point of common coupling. In the FGW TR3 4.4.2. [7] the calculation of a flicker coefficient for different grid impedance angles is required. This calculation is based on the measured short
Figure 16: Flicker on a day with quite constant irradiation: Grid voltage, active power and short term flicker on 2011-07-11.
The main influence on the flicker values are the ACnetwork conditions. The flicker value increases when there are fast changes in the grid voltage independent from the amount of injected active power. The harmonic currents are measured according to FGW TR 3 chapter 4.4.3. for different power levels in steps of 10%. For the measurement results in the field two different power levels were picked where the irradiation was quite constant: Operation at 85 % active power and 50 % active power. The harmonic spectra based on ten minutes intervals are shown below. The results are based on the same data as the flicker results above on 2011-07-11. The THD values are mainly caused by the 2nd, 9th, 11th and 13th as they appear as the highest values in the spectra. %
1.00
Figure 20: Recorded voltages and currents during a 3phase FRT with 2% residual voltage, P = 145 kW, in total 10 inverters in the network. Channel 1 and 3: grid current, 200 A/div., channel 2 and 4: grid voltage, 250 V/div.
0.75
4
0.50
0.25
0.00 THD
H10
H20
H30
H40
H50
A IHarm
Figure 17: Harmonic current spectrum at an active power level of 85% with a measured current THD of 1.09%. Total RMS: DC Level: Fundamental(H1) RMS: Total Harmonic Distortion THD:
0.00 A 1.16 A 2233.30 A 1.09 %FND (Even: 0.53 %FND, Odd: 0.95 %FND)
%
C re ate d wi th D ran V ie w 6.5 .0
2.5
2.0
1.5
1.0
0.5
0.0 THD
H10
H20
H30
H40
H50
CONCLUSION
This paper has presented the first results from a 12 MW PV power plant, which fulfils the German Grid Codes. Most relevant aspects of ancillary services have been shown. The Danfoss’ ancillary service solutions, both in open loop and in closed loop, have been presented. The main conclusion of this paper is that it is possible to operate large string inverter-based PV power plants with good power quality and control performance. The PV plant consisting of more than 700 inverters complies with the requirements of the grid codes. Further investigations on PV plant with even higher power will follow
A IHarm
Total RMS: DC Level: Fundamental(H1) RMS: Total Harmonic Distortion THD:
0.00 A 5.30 A 1257.12 A 2.49 %FND (Even: 0.79 %FND, Odd: 2.36 %FND)
Figure 18: Harmonic current spectrum at an active power level of 50% with a measured current THD of 2.49%. C re ate d wi th D ran V ie w 6.5 .0
3.4 Low Voltage Fault Ride Through A 200 kW FRT test facility has been designed and constructed in accordance to [6, 7] in order to validate the inverters, also when multiple inverters are connected to a common network, see Fig. 19. The test results from test with 10 inverters in parallel with the network are depicted in Fig. 20. Total power prior to the voltage sag equals 145 kW without reactive power. The plant keeps connected during the entire dip. Moreover reactive current is supplied as required by the grid code. The time requirements are fulfilled as well. Further results with single inverters can be found in [9]. Busbar Public network B1
3 * 690 V
T1 Δ Y
L1
R1
Im
m Ug
m
3 * 690 V
T2 3 * 400 V m B3 Δ Y U
m
TLX
SMPS
m
TLX
SMPS
m
TLX
SMPS
m
R2 B2
3*u 3*i RTU
Figure 19: Schematic for the 200 kVA Fault Ride Trough test facility at Danfoss Solar Inverters.
5 REFERENCES [1] Standard EN 50160, Voltage characteristics in public distribution systems. [2] BDEW, Technical guideline: Generating plants connected to the medium-voltage network, 2008. [3] Letter to FNN in VDE from Bundesnetzagentur, dated 29-9-2010. [4] http://www.concentratingpv.org/darmstadt2009/pdf/papers/24-TroesterGermanGridCodes.pdf [5] Søren Bækhøj Kjær, Flicker and photovoltaic Power Plants, Proc. of 25th European Photovoltaic Solar Energy Conference and Exhibition /5th World Conference on Photovoltaic Energy Conversion, 6-10 September 2010, Valencia, Spain. [6] Standard IEC 61400-21, Wind turbines - Part 21: Measurement and assessment of power quality characteristics of grid connected wind turbines, 2008. [7] http://www.wind-fgw.de/ [8] http://www.moehring-energie.de/ [9] www.danfoss.com/BusinessAreas/Solar+Energy/
