Optimal Sizing and Placement of BESS in

Distribution Grid with High PV Penetration

Considering BESS Optimal Operation
1st Bach Ta-Duy 2nd Anh Nguyen-Tuan
Dept. Electrical Engineering Dept. Electrical Engineering
School of EEE, School of EEE,
Hanoi Uni. of Science and Technology Hanoi Uni. of Science and Technology
Hanoi, Viet Nam Hanoi, Viet Nam
[email protected] [email protected]

3rd Tuyen Nguyen-Duc* 4th Goro Fujita

Dept. Electrical Engineering Department of Power System
School of EEE, Department of EE,
Hanoi Uni. of Science and Technology Shibaura Institute of Tech. (SIT)
Hanoi, Viet Nam Tokyo, Japan
[email protected] [email protected]

Abstract—Battery Energy Storage Systems (BESSs) are an PV power output increases. Limiting the active power of a PV
effective solution in preventing overvoltage caused by the high system is an effective solution in preventing voltage violations.
penetration rate of renewable energy sources (RES). This pa- While PV energy is green and free energy, curtailment may
per proposes a new strategy to achieve voltage regulation in
distributed power systems in the presence of solar sources and be an unacceptable solution. One of the practical and effective
BESSs. The goal is to find the minimum size of the BESSs and solutions to achieve flexible active power control to solve the
their corresponding position in the network based on the size and overvoltage problem is to deploy an energy storage system
location of the integrated solar generation. The proposed method [2]. Among all possible types of energy storage technologies,
formulates the problem by combining grid sensitivity coefficient battery energy storage systems (BESS) can be considered
analysis and modeling second-order cone programming (SOCP)
in power flow analysis. Using reactive power to adjust voltage widely used and quite developed. The location and size of
regulation is not always the optimal solution in distribution the BESS have a significant impact on the economic benefits
systems in which R/X is large. Therefore, the study only considers aspect when the BESS is integrated into the power grid,
the minimum size and best BESS location achieved by optimizing inappropriate placement and too large size of the BESS can
the amount of solar-curtailed active power through BESS based increase investment costs [3]. Therefore, finding a reasonable
on the grid topology and R/X ratios in the distribution system.
Simulation results on the modified IEEE-33 bus verify the size and location for the BESS along with improving electrical
effectiveness of the proposed method. system performance is the inspiration for this paper.
Index Terms—BESS Sizing, BESS Placement, BESS Otimiza- Many studies have been performed to optimize the size
tion, Second-order Cone Programming (SOCP), Distribution of BESSs in [4], [5]–[9] Most studies focus on maximizing
grid, PV. the economic benefits of BESSs [5], [6], and some aim to
improve the uncertainty due to natural conditions of renewable
I. I NTRODUCTION energy sources such as solar and wind [7], [8]. However, these
Grid-connected photovoltaic (PV) systems are becoming studies have not focused on improving the voltage quality of
increasingly popular as a means of generating energy. With BESSs in the presence of large amounts of renewable en-
the high penetration rate of PV in the distribution grid, excess ergy sources integrated with the distribution system. Research
energy is injected into the grid, creating reverse power flow, in [9], [10] has demonstrated that BESS can help reduce
and causing overvoltage problems. Because of the large R/X the workload on OLTC and SVR in preventing overvoltage
ratio in the distribution network, using reactive power to reg- phenomenon due to high PV penetration rate [11]–[13]. The
ulate voltage is not always effective. The use of conventional optimal size of the BESS has not been considered in these
voltage regulation devices to prevent voltage violations due to studies, leading to oversized BESS usage, resulting in extra
PV reverse power flow such as on-load tap-changing (OLTC) costs. Approaches [14], [15] only considered the optimal
transformers and step voltage regulators (SVR), will increase size for BESS integrated with PV at the same location but
the working time and reduce the life of the devices in case did not consider the coordination and working of multiple
BESSs to solve voltage problems while PVs penetrate the node. The grid is depicted in Fig.1, which is a four-node test
distribution grid. In addition, the optimal location of BESS in medium-voltage distribution network (MVDN) [22]. Loads
the distribution system affecting the voltage at the nodes is a ld1 , ld2 ... are described as constant PQ models. Line models
research topic that needs to be solved. In the papers [16]–[18] L01 , L12 ... are modeled using the RL-direct sequence equiv-
the position of BESS in voltage regulation was considered, alent circuit, and the admitted shunt component is ignored.
but the relationship between position and voltage was not Node 0 represents the Slack Bus, and the adjusted voltage
given. The paper [19] uses sensitivity analysis between BESS amplitude is constant V0 . Thus, the node 0 voltage amplitude
location and grid voltage to optimize the amount of power is the electrical amplitude reference voltage for the whole grid,
reduction of PVs. The modified impedance matrix is applied and we can calculate the voltage of other nodes through node
to describe the relationship of BESS location and grid voltage 0. Thus, by consecutive calculation between nodes, the voltage
but only evaluates the relationship of one BESS location module at node 3 is obtained.
with voltage [20]. Therefore, BESS sizing strategies need to
consider the location and operational coordination of BESSs V3 = V2 − V23 = (V0 − V01 ) − V12 − V13
(1)
to address the challenges caused by PV systems integrated into V3 = V0 − (V01 − V12 − V13 )
the grid. In this paper, a strategy to simultaneously optimize
BESS location and size is proposed to solve the overvoltage Where Vij is the voltage difference between node i and node
and undervoltage problems in a distribution system with high j, Vi is the voltage amplitude of node i. Thus, the voltage
PV penetration. Based on the forecasted capacity of solar difference between node 0 and node 3 is the sum of the voltage
power and the grid’s structure, the BESS’s size and location deviations on the branches going from node 0 to node 3.
will be obtained through a second-order cone programming Normally, the voltage offset across the branches is approx-
(SOCP) optimization mathematical model. In this method, the imated by the formula:
relationship between the position and voltage of the nodes Rij × Psj + Xij × Qsj
will be analyzed through the sensitivity analysis method [22]. Vij ≃ (2)
Vn
Then, this relationship is combined with the Second-order
where Psj is the sum of the active power of the loads at the
Conic Programming (SOCP) model in power flow analysis
node behind node j and the branch active power loss of the
to identify the optimal location and size of BESS.
branches after node j, and Qsj is the sum of the reactive power
The main contributions of this work are as follows:
of the loads at the node behind the node j and branch reactive
• Use second-order cone programming in considering the
power loss of branches after node j. For Ps1 and Qs1 , can
optimal size of BESSs. write:
• Optimizing the BESS size considers two factors: the
location of BESS and the operating schedule of BESS. Ps1 = P1 + P2 + P3 + P4 + PR12 + PR23 + PR24
(3)
• Coordinate the operation of multiple BESSs in improving Qs1 = Q1 + Q2 + Q3 + Q4 + QX12 + QX23 + QX24
voltage and reducing losses in distribution grids with a
Where P1 ,P2 ,P3 ,P4 , and Q1 ,Q2 ,Q3 ,Q4 are the active
high proportion of renewable energy.
and reactive power of load ld1 , ld2 , ld3 , ld4 , respectively.
The rest of the paper is organized as follows. In Section II,
PR12 ,PR23 ,PR24 and QX12 ,QX23 ,QX24 are the loss on
the principles of the proposed strategy are described. Then
branches L12 , L23 , L24 . These active and reactive losses are
the simulation results are obtained by applying the proposed
negligible compared to the load powers. Applying this assump-
strategy to the IEEE 33-bus grid system, as presented in
tion to (2), (3) leads to:
Section III. Finally, the conclusion is in Section IV.
Ps1 ≃ P1 + P2 + P3 + P4
II. M ETHODOLOGY
Qs1 ≃ Q1 + Q2 + Q3 + Q4
A. Sensitivity Computation
N
P bus N
Pbus (4)
Rij × Pk + Xij × Qk
k≥j k≥j
Vij =
Vn
Thus, by combining (2), (4), we get the voltage amplitude at
the nodes Vi which is a function dependent on the variables
being the active and reactive power of the load.
Vi = f (P1 , P2 , ...Q1 , Q2 , ...) (5)
In this study, we only consider adjusting the active power
Fig. 1. One-line diagram of a three-phase symmetrical, radial distribution (∆Q = 0). Therefore, the voltage change (∆Vi ) when chang-
network.
ing the active power of the load is calculated as follows:
This section briefly outlines the theory of the relationship ∂Vi ∂Vi ∂Vi
between the voltage modulus and the power injected into the ∆Vi = × ∆P1 + × ∆P2 + × ∆P3 + ... (6)
∂P1 ∂P2 ∂P3
Thus, by taking the partial derivative of a linear function, line segments alongside longer ones, particularly when em-
we get a matrix of active power sensitivity Sp . With this ploying single precision arithmetic. To enhance the numerical
matrix, we see the strong or weak change of voltage at node reliability of the method, the authors in [8] strongly advocate
i when there is a change in active power at node j. In the utilizing double-precision arithmetic.
following section, this Sensitivity Factor Sp will be calculated
and applied to the proposed optimal model. C. Application of the proposed method
1) Objective Function: In this section, the combination of
B. Distribution Load Flow the sensitivity coefficient method and the SOCP model to
Considering the single-line equivalent circuit in [21] (all solve the size optimization problem for BESS to improve grid
relevant quantities are per unit), the line model without shunt voltage when the PV system has a high penetration rate. The
connections is sufficient to describe fully the concepts in this objective function is written as the following formula:
paper. The real/reactive power flow from node i to note j are T NX
Bess X
T
X X
Pij = Gij Vi2 − Gij Vi Vj cosθij + Bij Vi Vj sinθij (7) F = (Pij,t + Pji,t ) + (pcj,t + pdj,t )
ij∈ΩL t=1 j t=1
Qij = Bij Vi2 − Bij Vi Vj cosθij − Gij Vi Vj sinθij (8)
| {z } | {z }
f1 f2
√ NX
Bess
where θij = θi − θj . By defining ui = Vi2 / 2, Rij =
Vi Vj cosθij , and Iij = Vi Vj sin θij , (7) and (8) become: + Ebati (14)
j
√
Pij = 2Gij ui − Gij Rij + Bij Iij (9)
| {z }
f3
√
Qij = 2Bij ui − Bij Rij − Gij Iij (10) The objective function has three components: f1 represents
total grid loss and ensures optimal BESS scheduling, while f2
In (9) and (10), Rij and Iij are constrained such that: and f3 determine the minimum BESS size to ensure voltage
2
2ui uj = Rij 2
+ Iij (11) standards. Component f2 with formula (20) ensures either
charging or discharging at any time. Here, Ebati is the size
Equations (9)-(11) can be used to define the radial load flow of the BESSs at node i. pcj,t and pdj,t are positive values,
problem. Let N be the number of nodes in the distribution corresponding to the charge and discharge power of BESS,
system with node 1 being connected to the power substation. respectively.
It is assumed that the magnitude of the voltage at this node is 2) Constraints: Based on the voltage amplitude sensitivity
specified. The power injection constraints at each of the (N-1) coefficient will be written as follows:
remaining nodes (i=2,...,N) are Nbess
X
PV
pcj,t × st,j,i − pdj,t × st,j,i


Vi,t = Vi,t + (15)
√ X X j
− 2ui Gij + (Gij Rij − Bij Iij )=PLi (12)
j∈k(i) j∈k(i) where Vi,t is the voltage amplitude at node i in the period time
√ X X i. We designate the state of the grid before BESS participates
− 2ui Bij + (Bij Rij + Gij Iij )=QLi (13) in the grid as the initial state. In the initial state the values of
j∈k(i) j∈k(i) the power of load, and capacity of PV are predefined, so we
In (12) and (13), k(i) denotes the set of nodes connected can completely use the AC power flow calculation methods
to node i, and PLi /QLi denote the real/reactive power loads to solve the voltage amplitude of the nodes in this state. The
at node. Equations (12) and (13), when evaluated for i = voltage amplitude of the nodes in the initial state is denoted
PV
2, ..., N , define a linear system 2 × (N − 1) with equations. Vi,t , st,j,i is the sensitivity coefficient between the active
Note that the number of lines in a radial network is (N − 1). power that BESS acts on node j and the voltage at node i for
Because Rij = Rji and Iij = −Iji , the total number of vari- time t, pcj,t and pdj,t are the charging and discharging power
ables is 3 × (N − 1). Therefore, (N − 1) additional equations of the BESS at node j for time t, respectively.
are required to solve for the variables. These equations result • Voltage Amplitude Limits:
from (5) enforced for all lines.
Vmin ≤ Vi,t ≤ Vmax (16)
The set of equations presented in (11)-(13) can be resolved
to derive the solution for radial load flow. Upon computation • Branch Flow Constraints:
of the newly introduced variables, it becomes straightforward √ X X
− 2ui,t Gij + (Gij Rij,t − Bij Iij,t ) =
to infer the original variables. Expósito and Ramos [8] have
j∈k(i) j∈k(i)
introduced a technique for radial load flow, which involves
c d
solving a comparable equation system through the Newton PLi,t + (Pi,t − Pi,t )
√ X X
method. In their discussion, they highlighted that employing − 2ui,t Bij + (Bij Rij,t + Gij Iij,t ) = QLi,t
block arithmetic and symbolic computations can mitigate the j∈k(i) j∈k(i)
ill-conditioning caused by the presence of excessively short (17)
 • Branch Flow Limits: energy, efficiency as well as SoC of BESSs. The minimum and
max
−Pij,t max
≤ Pij,t ≤ Pij,t maximum SoC limits are set to 20% and 80%, respectively,
−Qmax ≤ Qij,t ≤ Qmax (18) and the charging and discharging efficiency of the BESS is set
ij,t ij,t
(Pij,t )2 + (Qij,t )2 ≤ (Sij,t
max 2
) at 96%. The obtained results from the proposed algorithm are
applied to the power load flow to validate their accuracy.
• SOC of BESS Constraints:
∆t
soci,t = soci,t−1 + ηc × ∆t × pci,t − × pdi,t
ηd
(19)
soci,t ≤ 0.8 × Ebati , soci,t ≥ 0.2 × Ebati
soc0i ≤ Ebati , soc0i ≥ 0
• Energy balance of BESS Constraints:
T
X pdi,t
− ηc × pci,t = 0 (20)
t
ηd
Fig. 3. Test System IEEE 33 Bus Distribution System using for optimizing
3) Procedure to Solve the Optimization: model.
There are three-step in the optimization problem shown in
Fig.2:
1. Calculate the initial state and the sensitivity factor. 2500
Total load of grid
2. Solving optimization (14) to seek sizing and placement PV power generation
2000
optimal for BESSs.
3. Resolve optimization model with an objective function
Power [kW] 1500
(14) which removes component f3 with the results of
sizing and placement obtained in step 2. 1000

500

0
0 2 4 6 8 10 12 14 16 18 20 22
Time [h]

Fig. 4. The PV power generation and total load profile of grid.

1.06
1.05 Bus2
Bus3

1.04 Bus4
Bus5
Bus6

1.03
Bus7
Bus8
Bus9
Bus10
1.02 Bus11
Bus12
Bus13
V [pu]

1.01 Bus14
Bus15
Bus16

1 Bus17
Bus18
Bus19

0.99
Bus20
Bus21
Bus22
Bus23

0.98 Bus24
Bus25
Bus26

0.97 Bus27
Bus28
Bus29

0.96
Bus30
Bus31
Fig. 2. Flowchart of proposed algorithm model. Bus32
Bus33
0.95
0.94
III. S IMULATION AND R ESULTS 0 2 4 6 8 10 12 14 16 18 20 22
Time [h]
In this section, we present the simulation results for the pro-
posed algorithm that provides a better understanding of finding Fig. 5. Network daily voltage profile with PV system at different buses.
the minimum BESS size besides its best place. IEEE 33 Bus
Test System is considered as the case study as shown in Figs.3, This work compares the optimal size of the BESS in two
without loss of generality, the solar power source is located at cases, when the placement is located at node 31 (location
note 31 while the active power is updated every 30 minutes of the PV system), and optimizes placement. The BESS is
through the 24-hour operating period. Meanwhile, the PV designed to regulate the node voltages by ANSI C84.1 with an
power generation and total load profile of the radial test feeder acceptable voltage tolerance range of ±5% during the whole
are shown in Fig.4. For analysis, a plain model is used to PV variation period. Initially, the effect of PV power on the
simulate BESS operation, considering battery charge/discharge voltage profile of the feeder without BESS is evaluated, and
voltage violations are observed at nodes 30,31,32,33 as shown
in Fig.5.

Bus33
Bus32
Bus31
Bus30
Bus29
Bus28
Bus27
Bus26
Bus25
Bus24
Bus23
Bus22
Bus21
Bus20
Bus19
Bus18
Bus17
Bus16
Bus15
Bus14
Bus13
Bus12
Bus11
Bus10
Bus9
Bus8
Bus7
Bus6
Bus5
Bus4
Bus3
Bus2
Bus1
0 0.5 1 1.5 2 2.5 3
dv/dp 10-5
[pu/kw]

Fig. 8. Network daily voltage profile with BESS is fixed with the PV system
Fig. 6. Sensitivity between power at all buses with voltage at bus 31.

500 Fixed Location

Bess-Bus31
Bess-Bus32
Bess-Bus33
400
Power [kW]

300

200

100

8.5 9 9.5 10 10.5 11 11.5 12.5 13 13.5

Time [h]

Fig. 7. The absorbed power in two cases of optimized and fixed placement Fig. 9. Network daily voltage profile with the placement of the BESS
of BESS. considered optimal

Additionally, Fig.8 and Fig.9 revealed that overvoltage 2500 BESS-Bus31

incidents are efficiently mitigated even at nodes 30, 31, 32, BESS-Bus32
BESS-Bus33
Power of PV
and 33 which had the highest violation in both the optimal and 2000
Power at Slack Bus (Without BESS)
Power at Slack Bus (With BESS)

fixed BESS location cases. This was due to BESSs absorbing

the high excess power of PVs (Fig.7), limiting the reverse 1500
power flow over the grid lines during the whole period of high
Power [kw]

irradiation. Therefore, with optimized BESS placement (the 1000

optimal model selected the location with the highest sensitivity
coefficient (Fig.6) as the BESS location), the size of BESS is 500
significantly smaller than without considering optimal BESS
0
placement in improving grid voltage.
Further analysis of the operation of BESSs with optimal -500
locations is shown in Fig.10 which breaks down that BESSs
not only prevent overvoltage caused by PV but also improve -1000
the total load curve of the grid. Especially at noon, when 0 2 4 6 8 10 12 14 16 18 20 22
solar radiation suddenly drops sharply for many reasons, so Time [h]
the overvoltage phenomenon does not occur. Taking advantage
of this moment, BESS at bus 33 has discharged power to Fig. 10. The optimal charging and discharging schedule of three BESSs is
located at Bus 31,32,33.
recovery SOC to be ready to support the grid in the next hours.
Finally, the proposed model in this paper demonstrated the ef- [11] M. Zillmann, R. Yan and T. K. Saha, ”Regulation of distribution network
fectiveness of optimizing the size of the BESS considering its voltage using dispersed battery storage systems: A case study of a
rural network,” 2011 IEEE Power and Energy Society General Meeting,
operation plan and location. This not only ensures compliance Detroit, MI, USA, 2011, pp. 1-8, doi: 10.1109/PES.2011.6039094.
with technical requirements but also aligns with the economic [12] J. Cappelle et al., ”Introducing small storage capacity at residential PV
imperatives associated with the operation of the distribution installations to prevent overvoltages,” 2011 IEEE International Confer-
ence on Smart Grid Communications (SmartGridComm), Brussels, Bel-
grid amidst the escalating integration of Distributed Energy gium, 2011, pp. 534-539, doi: 10.1109/SmartGridComm.2011.6102380.
Resources (DERs). [13] K. H. Chua, Y. S. Lim, P. Taylor, S. Morris and J. Wong, ”Energy
Storage System for Mitigating Voltage Unbalance on Low-Voltage
IV. C ONCLUSION Networks With Photovoltaic Systems,” in IEEE Transactions on Power
Delivery, vol. 27, no. 4, pp. 1783-1790, Oct. 2012, doi: 10.1109/TP-
This work introduces a new strategy for designing BESS WRD.2012.2195035.
under the high grid penetration of PV. Using the SOCP [14] Y. Yang, H. Li, A. Aichhorn, J. Zheng and M. Greenleaf, ”Sizing
optimize model which allows simultaneous calculation of the Strategy of Distributed Battery Storage System With High Penetration of
Photovoltaic for Voltage Regulation and Peak Load Shaving,” in IEEE
BESS’s optimal layout and scheduling size to ensure the Transactions on Smart Grid, vol. 5, no. 2, pp. 982-991, March 2014,
voltage standard and improve the load curve of the grid. doi: 10.1109/TSG.2013.2282504.
The application of sensitive analysis in the proposed strategy [15] Liao, Jian-Tang, et al. ”BESS-sizing optimization for solar PV system
integration in distribution grid.” IFAC-PapersOnLine 51.28 (2018): 85-
supports optimal operation and clearly shows the relationship 90.
between voltage and active power in the voltage regulation of [16] Zhang, Weiwei & Wang, Shuliang. (2022). Optimal Allocation of BESS
the distribution network. Meanwhile, the results of the location in Distribution Network Based on Improved Equilibrium Optimizer.
Frontiers in Energy Research. 10. 936592. 10.3389/fenrg.2022.936592.
and size of the BESS also depend on the maximum capacity [17] Sirote Khunkitti, Punyawoot Boonluk, Apirat Siritaratiwat, ”Optimal
at each location. In the future, studies will be implemented Location and Sizing of BESS for Performance Improvement of Distri-
to find a BESS size that ensures both economic and technical bution Systems with High DG Penetration”, International Transactions
on Electrical Energy Systems, vol. 2022, Article ID 6361243, 16 pages,
aspects of the power system. 2022. https://doi.org/10.1155/2022/6361243
[18] Ling Ai Wong, Vigna K. Ramachandaramurthy, Sara L. Walker, Phil
R EFERENCES Taylor, Mohammad Javad Sanjari, Optimal placement and sizing of bat-
[1] Y. Yang, H. Li, A. Aichhorn, J. Zheng and M. Greenleaf, ”Sizing tery energy storage system for loss reduction using whale optimization
Strategy of Distributed Battery Storage System With High Penetration of algorithm, Journal of Energy Storage, Volume 26, 2019,100892,ISSN
Photovoltaic for Voltage Regulation and Peak Load Shaving,” in IEEE 2352-152X,https://doi.org/10.1016/j.est.2019.100892.
Transactions on Smart Grid, vol. 5, no. 2, pp. 982-991, March 2014, [19] S. Hashemi and J. Østergaard, ”Efficient Control of Energy Storage
doi: 10.1109/TSG.2013.2282504. for Increasing the PV Hosting Capacity of LV Grids,” in IEEE Trans-
[2] F. Marra, G. Yang, C. Træholt, J. Østergaard and E. Larsen, ”A Decen- actions on Smart Grid, vol. 9, no. 3, pp. 2295-2303, May 2018, doi:
tralized Storage Strategy for Residential Feeders With Photovoltaics,” 10.1109/TSG.2016.2609892.
in IEEE Transactions on Smart Grid, vol. 5, no. 2, pp. 974-981, March [20] H. Nazaripouya, Y. Wang, P. Chu, H. R. Pota and R. Gadh, ”Optimal
2014, doi: 10.1109/TSG.2013.2281175. sizing and placement of battery energy storage in distribution system
[3] Ebrahimian, Homayoun et al. “The price prediction for the energy based on solar size for voltage regulation,” 2015 IEEE Power &
market based on a new method.” Economic Research-Ekonomska Is- Energy Society General Meeting, Denver, CO, USA, 2015, pp. 1-5, doi:
traživanja 31 (2018): 313 - 337. 10.1109/PESGM.2015.7286059.
[4] Y. Yang, H. Li, A. Aichhorn, J. Zheng and M. Greenleaf, ”Sizing [21] R. A. Jabr, ”Radial distribution load flow using conic programming,” in
Strategy of Distributed Battery Storage System With High Penetration of IEEE Transactions on Power Systems, vol. 21, no. 3, pp. 1458-1459,
Photovoltaic for Voltage Regulation and Peak Load Shaving,” in IEEE Aug. 2006, doi: 10.1109/TPWRS.2006.879234.
Transactions on Smart Grid, vol. 5, no. 2, pp. 982-991, March 2014, [22] Brenna, M., Berardinis, E.D., Foiadelli, F., Sapienza, G., & Zaninelli,
doi: 10.1109/TSG.2013.2282504. D. (2010). Voltage Control in Smart Grids: An Approach Based on Sen-
[5] A. Aichhorn, M. Greenleaf, H. Li, and J. Zheng, ”A cost-effective sitivity Theory. Journal of Electromagnetic Analysis and Applications,
battery sizing strategy based on a detailed battery lifetime model and an 2, 467-474.
economic energy management strategy,” 2012 IEEE Power and Energy
Society General Meeting, San Diego, CA, USA, 2012, pp. 1-8, doi:
10.1109/PESGM.2012.6345103.
[6] Y. Ru, J. Kleissl and S. Martinez, ”Storage Size Determination
for Grid-Connected Photovoltaic Systems,” in IEEE Transactions
on Sustainable Energy, vol. 4, no. 1, pp. 68-81, Jan. 2013, doi:
10.1109/TSTE.2012.2199339.
[7] ML. Xu, X. Ruan, C. Mao, B. Zhang and Y. Luo, ”An Improved Optimal
Sizing Method for Wind-Solar-Battery Hybrid Power System,” in IEEE
Transactions on Sustainable Energy, vol. 4, no. 3, pp. 774-785, July
2013, doi: 10.1109/TSTE.2012.2228509.
[8] T. K. A. Brekken, A. Yokochi, A. von Jouanne, Z. Z. Yen, H. M. Hapke,
and D. A. Halamay, ”Optimal Energy Storage Sizing and Control for
Wind Power Applications,” in IEEE Transactions on Sustainable Energy,
vol. 2, no. 1, pp. 69-77, Jan. 2011, doi: 10.1109/TSTE.2010.2066294.
[9] H. Ravindra, M. Omar Faruque, K. Schoder, M. Steurer, P. Mclaren, and
R. Meeker, ”Dynamic interactions between distribution network voltage
regulators for large and distributed PV plants,” PES T&D 2012, Orlando,
FL, USA, 2012, pp. 1-8, doi: 10.1109/TDC.2012.6281599.
[10] X. Liu, A. Aichhorn, L. Liu and H. Li, ”Coordinated Control of
Distributed Energy Storage System With Tap Changer Transformers for
Voltage Rise Mitigation Under High Photovoltaic Penetration,” in IEEE
Transactions on Smart Grid, vol. 3, no. 2, pp. 897-906, June 2012, doi:
10.1109/TSG.2011.2177501.