Investigation on Solar PV and Battery System. Penetration in Singapore Distribution Power. Networks. S. X. Chen and H. B. Gooi. Nanyang Technological ...
Investigation on Solar PV and Battery System Penetration in Singapore Distribution Power Networks S. X. Chen and H. B. Gooi
Tham Tzen Woo and Thi Ha Yushein
Nanyang Technological University Singapore
Panasonic R&D Center Singapore
more slowly than load changes, so peaking generation is throttled back to stabilize the power flow into and out of the grid. In addition, when the load on the utility grid reaches new peak levels, system operators must start activating every available generating source, and even minor throttling back of generation may cause the grid voltage to collapse. The designs of the grid-connected PV power system focus on converting as much irradiant power as possible into useful active power i.e., current flowing into the grid which is in phase with the utility-defined voltage. The PV power system can help to meet the typical loads supplied by the electric power utility infrastructure [15] [12] [13] [14]. However, as the installed capacity of this technology grows, at some point this assumption will no longer hold true. In some small areas of the electric power distribution system, e.g., some rural feeders, solar electric power generation has already approached or exceeded the local daytime load. Electric utilities have begun to modify their physical infrastructure, e.g., bigger wire size and voltage control settings, to adapt to this new power flow pattern. If this trend continues, PV power systems will be required to provide more grid support services and to participate, to a greater extent, in utility dispatch and operations processes [16] [17] [18]. This paper will focus on research of grid-connected PV and battery systems in Singapore. PV systems will help make full use of the solar power in Singapore. Battery storage systems will overcome the intermittency of PV power output by charging or discharging themselves. Solar PV and battery systems will be very close to the load in Singapore. It will help to reduce the power transmission loss. If lots of those PV and battery systems are combined together, they can be considered as virtual power plant (VPP), which will be able to participant in the power system frequency market. The PV and battery system is installed in the distribution power system. Fig. 1 shows a common power distribution system in Singapore. The incoming voltage is 66kV and is lowered to 22kV by a 66kV/22kV transformer. After that two 22kV/0.4kV transformers will step down the voltage to 0.4kV further at Bus 4 and Bus 5, which are connected to several buildings. The power flow and voltage at each bus will
Abstract—This paper discusses the penetration of the solar photovoltaic (PV) and battery system in Singapore distribution power networks and its impact on grid. The distribution power network with the installation of the solar PV and battery system will be introduced. The effect of the voltage and power flow will be studied based on the comparison between the results obtained before and after the installation of the solar PV and battery system. The frequency will also be examined by performing the dynamic simulation of the solar PV and battery system and Load Frequency Control (LFC).
I. I NTRODUCTION The global installed capacity of grid-connected photovoltaic (PV) systems has grown dramatically over the last five years [1]. The total PV installed capacity in the Singapore Power (SP) power grid is still less than 1% of the peak electricity load. However, a 5% to 10% level may be attained in less than a decade from now [2] [3]. Such penetration levels are significantly higher than the currently assumed limits under which net energy metering is allowed [4] [5]. Reaching those levels would likely require significant changes to traditional inverter technologies and regulations in order for high penetration of PV to maintain reliable and economical grid operation [6]. The integration of a large number of embedded PV generators will have far reaching consequences on the distribution networks as well as on the national transmission and generation system [7]. If the PV generators are built on the roof tops and at sides of buildings, they will be electrically close to loads. However, these PV generating units may be liable to common mode failures that might cause a sudden or rapid disconnection of a large proportion of the PV operating capacity. Considering the recent grid codes in Germany and Spain [8], PV generators should provide dynamic grid support e.g. voltage stability during voltage drops. This is often referred to as Fault-Ride-Through (FRT) and frequency control capability [9]. In the interconnected electric power grid, generated electric power must be consumed within milliseconds of being generated [10]. Excess power can be accumulated with energy storage systems such as pumped hydro and battery systems [11], but conventional energy storage systems respond much
978-1-4673-4584-2/12/$31.00 ©2012 IEEE
Kenichi Watanabe Energy Solution Center Panasonic Corporation
48
IPEC 2012
one week of Singapore is shown in Fig. 2. In the PV system,
be studied both before and after the installation of the PV and battery system is installed. Besides, the Load Frequency Control (LFC) will also be studied by examining the frequency dynamics of the PV and battery system. This paper will discuss the impact of the PV and battery systems in Singapore distribution power networks based on the simulation results.
1200
Solar radiation (W/m2)
1000
66kV/22kV 1
800
600
400
200
0
2
3
Mon
Fig. 2.
22kV/0.4kV
22kV/0.4kV
4
9 8
Thu
Fri
Sat
Sun
The solar radiation in one week of Singapore
ps = ηSI(1 − 0.005(to − 25))
11
(1)
where, η is the conversion efficiency (%) of the solar cell array; S is the array area (m2 ); I is the solar radiation (kW/m2 ); and to is the outside air temperature (◦ C). The lithium-ion battery storage system is used to compensate the variability of solar power [21], [22]. The excess solar energy can be stored in the battery system. It can support the grid during the peak load hours [23], [24]. The PV and battery systems have the capability to participate in the frequency market if lots of those systems are aggregated together via VPP concepts.
12 13
7
14
6 Fig. 1.
Wed
the maximum power point tracker will be used to obtain the maximum power. The PV power output (ps ) is presented by (1) [19], [20].
5
10
Tue
A simple distribution network in Singapore
In section II, the PV and battery system is introduced. The equivalent circuit of PV and its output power formulation are presented in section II. The system analytical model is described in section III. The distribution line model is discussed in section III.A. The power system frequency model is shown in section III.B. The LFC function is examined in this section. The case study and result analysis are detailed in section IV. The voltage and energy efficiency study are discussed and three scenarios are compared in section IV.A. The results of the system frequency study with VPP are shown in section IV.B. The conclusion is drawn in section V.
III. S YSTEM A NALYTICAL M ODEL A. Distribution Line Model A distribution line model is used to calculate the power flow and power loss for the distribution power system in Fig. 1. Fixed loads are modeled as constant real and reactive power consumptions at each bus, Pd and Qd , as specified in the bus matrix. The output of the PV and battery system can be considered as the real and reactive power injections. The shunt admittance (Ysh ) of any constant impedance shunt element at a bus are specified by Gsh andBsh as follows:
II. PV AND BATTERY S YSTEM The fundamental building block of solar photovoltaic power is the solar cell or photovoltaic cell [12], [13]. A solar cell is a self-contained electricity-producing device constructed of semi-conducting materials. Light strikes on the semi-conducting material in the solar cell creating direct current (DC) [14]. In the calculation of the power output of a PV module, we assume that a maximum power point tracker will be used. Manufacturers of PV modules supply information on the voltage and current of the maximum power point at reference temperature and reference irradiance. The output current I can be expressed as a function of the output voltage V from the equivalent circuit of the PV module. The solar radiations in
Ysh = Gsh + jBsh
(2)
Each distribution line is modeled as a standard π model. The model has a series resistance R and a reactance X in series with an ideal transformer of a tap ratio τ and a phase shifter with phase shifter angle θshif t . The line has a total line charging capacitance Bc . The model is shown in Fig. 3. Branch voltages and currents from the f rom end to the to end of the branch are related by the branch admittance matrix Ybr as follows: Ifl Vfl = Y (3) br Itl Vtl
49
where Ybr =
(Ys τ12 + j B2c ) −Ys −jθ1shif t τe −Ys jθ1shif t Ys + j B2c
Power systems exhibit a highly non-linear and time-varying nature. However, for the purpose of frequency control synthesis and analysis in the presence of load disturbances, a simple low-order linearized model is used. In comparison with voltage and rotor angle dynamics, the dynamics which affect the frequency response are relatively slow. They are in the range of seconds to minutes. The overall generator-load dynamic relationship between the incremental mismatch power and the frequency deviation can be expressed as
and Ys =
τe
1 R+jX .
As shown in Fig. 3, Ifl is the current from the f rom side of branch l. Similarly Itl is the current from the to side of branch l. Vfl is the voltage at the f rom side of branch l. Likewise Vtl is the voltage at the to side of branch l. Vf From Ending:
If
Vt
Phase Shifter
It
R+jX
ith bus
Tap Transformer
BC /2
Fig. 3.
dΔf (t) + DΔf (t) (4) dt where Δf is the frequency deviation; ΔPT h is the thermal plant power change; ΔPSB is Solar PV and battery system power change; ΔPL is the load change; M is the inertia constant and D is the load damping coefficient. Equation (4) together with the dynamics of the thermal power plant as well as PV and battery system can be represented in the simulation block diagram. The LFC helps to adjust load points of governors of the generation units and control their outputs. The actual frequency is measured at the load dispatch center and is sent back to the LFC function. The LFC model is shown in Fig. 5. ΔPT h (t) + ΔPSB (t) − ΔPL (t) = M
To Ending: j th bus
BC /2
A simple line model in a power system
B. Power System Frequency Model As mentioned before, the individual PV and battery systems at different locations will be aggregated together as a VPP so that their combined capacity is large enough for them to participate in the wholesale regulation market. The control of frequency and power generation is commonly referred to as LFC which is a major function of the AGC system. The purpose of AGC is to maintain system frequency very close to a specified nominal value and to dispatch the generation of individual units. LFC regulates the power flow between different areas at the desired MW interchange values while holding the frequency constant. In an isolated power system, regulation of interchange power is not a control issue, and the LFC task is limited to maintaining the system frequency to the specified nominal value [25]. Only one area is considered in this report and AGC will be discussed in the context of the PV and battery system. Figure 4 shows the basic power system frequency model and it will be studied in this paper.
Solar PV and battery system LFC Thermal Power Plant
Fig. 4.
Fig. 5.
LFC Model
IV. C ASE S TUDY AND R ESULT A NALYSIS A. Voltage and Energy Efficiency Study The distribution network is shown in Fig. 1. There are three voltage levels in this system, namely 66kV for Bus 1, 22kV for Buses 2 and 3, and 0.4kV for other buses. Bus 1 is connected to the upstream network. Buses 6-14 are load buses, which are connected to different buildings. The load information is shown in Table I.
+ +
TABLE I L OAD I NFORMATION IN D ISTRIBUTION S YSTEM
Power System Frequency Model
The study system may have other generation sources in the power system frequency model. To simplify the problem and for the purpose of the PV and battery system research, only the thermal power plant and the PV and battery system are incorporated in the power system frequency model in Fig. 4. In addition, only one area is considered and power interchange among areas is not considered for simplicity.
Bus #
P MW
Q MVAr
Bus #
P MW
Q MVAr
Bus #
P MW
Q MVAr
6 7 8
0.23 0.2 0.2
0.07 0.06 0.03
9 10 11
0.13 0.3 0.2
0.06 0.07 0..05
12 13 14
0.2 0.17 0.08
0.05 0.05 0.03
Figure 6 shows the simulation system in PowerWorld. A generator is used to simulate the upstream network. Three scenarios are simulated in PowerWorld. They are (1) a power flow study without any PV and battery system, (2) a power flow study with the PV and battery system at bus 8, and (3) a power
50
flow study with the PV and battery system at all load buses. The capacity of each PV and battery system installed in the distribution network is about 45kWp and 60kWh respectively.
1.005 Without PV and battery system PV and battery system at bus 8 PV and battery system at each bus
1 0.995
Voltage (pu)
0.99 0.985 0.98 0.975 0.97 0.965
2
4
6
8
10
12
14
Bus
Fig. 7.
Bus voltage comparison for three scenarios
in the distribution network has increased with the installation of the PV and battery system. The more the PV and battery systems are installed on the load bus, the higher the voltage will increase. Bus 1 is connected with the upstream network. Hence, it has the maximum voltage of 1.0 pu. Fig. 6.
TABLE III C OMPARISON R ESULTS OF T HREE S CENARIOS
Distribution network simulation in PowerWorld
In the second scenario, the output of the PV and battery system at bus 8 is set at 39kW for active power injection and 19kVAr for reactive power injection. For the third scenario, the outputs of the PV and battery systems are shown in Table II.
P kW
Q kVAr
Bus #
P kW
Q kVAr
Bus #
P kW
Q kVAr
6 7 8
39 39 39
19 19 19
9 10 11
39 38 39
19 19 19
12 13 14
39 39 40
19 20 20
Minimum voltage (pu)
Maximum voltage (pu)
Average voltage (pu)
Transmission loss (kW)
Efficiency enhanced (%)
One Two Three
0.96799 0.96831 0.97352
1.0 1.0 1.0
0.98767 0.98804 0.99083
27 26 18
N/A +3.7% +33%
Compared with the base case in scenario one, the power loss is reduced by 1kW in scenario two. Its energy efficiency enhancement is 3.7%, which can be obtained by using 1kW divided by 27kW. The energy efficiency for scenario three is increased by 33% compared with that of scenario one. There are two reasons for the improvement of energy efficiency. One is because the PV and battery system is installed at the load bus and the active power can be consumed at the location it is generated. It avoids the power loss via the transmission lines. The other one is that the PV and battery system can supply reactive power at the load bus. It will regulate the load bus voltage and thus will help to reduce the power loss of the transmission lines. The voltage dynamic analysis for the PV and battery system bus in scenario two is simulated in Matlab Simulink as shown in Figure 8. Two buses are simulated in Figure 8. One is the Bus 4, which is the substation bus and the other is Bus 8 which is connected to the PV and battery system. The PV and battery system is under the control of the AGC signal. The frequency dynamic response will be discussed in the section IV.B. The maximum voltage fluctuation at Bus 4 is from 0.0781% to 0.1895%. Referring to the standard of IEC61000-3-3, this voltage fluctuation is less than 3% and it is acceptable [26].
TABLE II O UTPUTS OF PV AND BATTERY S YSTEMS IN S CENARIO T HREE Bus #
Scenario #
The voltage values of each bus during these three scenarios are shown in Figure 7. By comparing the results of scenario one and scenario two, one can tell that the voltage is enhanced at the bus which was installed with the PV and battery system. The voltage at other buses without the PV and battery system is nearly the same. As shown in Figure 7, the voltage of bus 8 is increased by 0.00354 pu from 0.98083 pu to 0.98437 pu. The distribution loss inside the distribution network is 27kW without the PV and battery system. After the PV and battery system is installed at bus 8, the power flow simulation shows that the power loss changes to 26kW, a reduction of 1kW. In scenario three, nine PV and battery systems were installed at nine different load buses. From the curves in Fig. 7, one can tell that the voltage has been improved by comparing voltages of scenario one from bus 4 to bus 14 with those of scenario two. The total transmission loss in scenario three is 18kW. Table III shows the comparison results of three scenarios. The minimum voltage and average voltage of all the buses
B. System Frequency Study with VPP The frequency response is studied by aggregating all the PV and battery systems installed in the distribution network as a single generating source and injecting its active power output into the power system frequency model. The block
51
Fig. 9.
System frequency simulation in Matlab
(a) Frequency change (Hz) 0.2
Bus 8 Switchboard
Service Cable of Length (L) Bus 4 Substation
0 charging
discharging
PV and Battery System
−0.2
Pn
0
50
100
150
200 250 300 350 Time (s) (b) PV and battery system output change (MW)
400
450
500
0
50
100
150
400
450
500
0
50
100
150
200 250 300 Time (s) (d) Load fluctuations (MW)
350
400
450
500
0
50
100
150
200
350
400
450
500
20 10 0
Fig. 8. Bus voltage dynamic simulation for distribution network with PV and battery system
−10
200 250 300 350 Time (s) (c) Thermal plant generation change (MW)
200 0
diagram shown in Fig. 9 is used to simulate the frequency response of the VPP. The total load of the power system is 4000MW and the load fluctuation is under 200MW which is 5% of the total load. 4000MW is smaller than the average of the demand in Singapore. The total capacity for the thermal power plant participated in the frequency market is 200MW. The capacity of the PV and battery system participated in the frequency market is 20MW which is 10% of the thermal power generation. The results of the system frequency study with VPP are shown in Fig. 10. The load fluctuation is shown in Fig. 10(d). The active power output of thermal plant and PV and battery system is shown in Fig. 10(c) and Fig. 10(b) respectively. They are controlled by the LFC. The system frequency response is shown in Fig. 10(a).
−200
200 0 −200
Fig. 10.
250 Time (s)
300
Results of system frequency study with VPP
adjust the system frequency with the help of the thermal power plant. The frequency change is always regulated and controlled to be less than 0.2 Hz using the PV and battery system. V. C ONCLUSION This paper introduced a sizable grid connected PV and battery system in Singapore. The PV and battery system is installed in the distribution network. Fig. 1 is studied in this
As shown in the Fig. 10, the output of the PV and battery system can vary according to the command of LFC and can
52
[14] W. Xiao, M. Lind, W. Dunford, and A. Capel, “Real-time identification of optimal operating points in photovoltaic power systems,” IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 1017 –1026, Jun. 2006. [15] H. B. Gooi, P. L. So, E. K. Chan, E. Toh, and H. Gan, “Strait ahead: Toward a sustainable, economic, and secure electricity supply in singapore,” IEEE Power Energy Mag., vol. 10, no. 4, pp. 65–74, Jul. 2012. [16] R. Perez, R. Seals, P. Ineichen, R. Stewart, and D. Menicucci, “A new simplified version of the perez diffuse irradiance model for tilted surfaces,” Solar Energy, vol. 39, no. 3, pp. 221–231, 1987. [17] P. Kirawanich and R. O’Connell, “Potential harmonic impact of a residential utility-interactive photovoltaic system,” in 9th Harmonics & Quality of Power Conf., vol. 3, 2000, pp. 983–987. [18] S. Chen and H. Gooi, “Scheduling of energy storage in a grid-connected pv/battery system via simplorer,” in TENCON IEEE Region 10 Conf., Nov. 2009, pp. 1–5. [19] C. Tao, D. Shanxu, and C. Changsong, “Forecasting power output for grid-connected photovoltaic power system without using solar radiation measurement,” in Power Electron. for Distrib. Generation Syst. (PEDG), 2010 2nd IEEE International Symposium on, Jun. 2010, pp. 773–777. [20] A. Yona, T. Senjyu, and T. Funabashi, “Application of recurrent neural network to short-term-ahead generating power forecasting for photovoltaic system,” in Power & Energy Soc. General Meeting, Jun. 2007, pp. 1–6. [21] B. Borowy and Z. Salameh, “Optimum photovoltaic array size for a hybrid wind/pv system,” IEEE Trans. Energy Convers., vol. 9, no. 3, pp. 482–488, Sept. 1994. [22] S. X. Chen, K. J. Tseng, and S. S. Choi, “Modeling of lithium-ion battery for energy storage system simulation,” in Power and Energy Engineer. Conf., Asia-Pacific, Mar. 2009, pp. 1–4. [23] C. Venu, Y. Riffonneau, S. Bacha, and Y. Baghzouz, “Battery storage system sizing in distribution feeders with distributed photovoltaic systems,” in 2009 IEEE Bucharest, PowerTech, Jul. 2009, pp. 1–5. [24] (2010, Jul.) Battery and electric vehicle report. [Online]. Available: http://bit.ly/fGaZPB [25] M. Arita, A. Yokoyama, and Y. Tada, “Evaluation of battery system for frequency control in interconnected power system with a large penetration of wind power generation,” in Power System Technology Conference, Oct. 2006, pp. 1–7. [26] IEC, “Iec61000-3-3 standard,” International Electrotechnical Commission, Tech. Rep., 2001.
paper. The power flow and voltage at each bus are studied both before and after the installation of the PV and battery system. The VPP concept is used in the system frequency response study by aggregating all the PV and battery systems. During the power flow simulation for the distribution network where the PV and battery system is examined, three scenarios are studied. The results show that the PV and battery system can help to enhance the voltage regulation at the load bus and increase the energy efficiency as well. In the system frequency study, the concept of VPP participating in the frequency market is examined. The results show that VPP can follow the command of LFC and help to adjust the system frequency. The intermittency of the solar power will be considered in the future case studies and more scenarios will be added to examine the voltage and frequency issues due to a large penetration of solar power plants. The PV and battery system will be integrated with the energy management system server and microgrid in lab of clean energy research laboratory (LaCER). The frequency response of the solar PV and battery system will be tested using the AGC function in the microgird. Besides, two-area LFC will be simulated based on the solar PV and battery system for VPP concept. ACKNOWLEDGMENT The authors would like to thank Panasonic and NTU for their generous funding support and Mr. Thomas Foo for his technical support in the equipment usage. R EFERENCES [1] (2007) Natural forcing of the climate system. [Online]. Available: http://bit.ly/vEz2t7 [2] Next-generation electricity: The emerging negawatt and micropower revolutions. [Online]. Available: http://sg.sg/N80b2m [3] EMA, “Handbook for solar pv systems,” Energy Market Authority, Tech. Rep., 2011. [4] W. Vermass. (2007) An introduction to photosynthesis and its applications. [Online]. Available: http://bit.ly/N80mdT [5] R. Somerville, “Historical overview of climate change science,” Intergovernmental Panel on Climate Change, Tech. Rep., 2007. [6] M. B. Ferguson, Renewable energy grid integration. Technical performance and requirements. Hauppauge: Nova Science Publishers, 2011. [7] S. X. Chen, H. B. Gooi, and M. Q. Wang, “Sizing of energy storage for microgrids,” IEEE Trans. Smart Grid, vol. 3, no. 1, pp. 142–151, Mar. 2012. [8] M. Datta, T. Senjyu, A. Yona, T. Funabashi, and C.-H. Kim, “A voltage and frequency control approach by grid-connected mw class pv systems,” in Electrical Machines and Systems Conference, Oct. 2010, pp. 475–480. [9] O. Morton. (2008) Solar energy: A new day dawning? [Online]. Available: http://bit.ly/N80pGI [10] M. R. Chuck Whitaker, Jeff Newmiller and B. Norris, “Distributed photovoltaic systems design and technology requirements,” Sandia National Laboratories, Tech. Rep., 2008. [11] S. X. Chen and H. B. Gooi, “Jump and shift method for multi-objective optimization,” IEEE Trans. Ind. Electron., vol. 58, no. 10, pp. 4538–4548, Oct. 2011. [12] R. Perez, R. Seals, P. Ineichen, R. Stewart, and D. Menicucci, “A new simplified version of the perez diffuse irradiance model for tilted surfaces,” Solar Energy, vol. 39, no. 3, pp. 221–231, 1987. [13] S. Chiang, K. Chang, and C. Yen, “Residential photovoltaic energy storage system,” IEEE Trans. Energy Convers., vol. 45, no. 3, pp. 385 –394, Jun. 1998.
53
