Florida State University Libraries
Electronic Theses, Treatises and Dissertations
The Graduate School
2014
Optimization of Battery Energy Storage
Systems for PV Grid Integration Based on
Sizing Strategy
Ye Yang
Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected]
FLORIDA STATE UNIVERSITY
COLLEGE OF ENGINEERING
OPTIMIZATION OF BATTERY ENERGY STORAGE SYSTEMS FOR PV GRID
INTEGRATION BASED ON SIZING STRATEGY
By
YE YANG
A Dissertation submitted to the
Department of Electrical and Computer Engineering
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
Degree Awarded:
Fall Semester, 2014
Ye Yang defended this dissertation on October 24, 2014.
The members of the supervisory committee were:
Hui Li
Professor Directing Dissertation
Chiang Shih
University Representative
Simon Y. Foo
Committee Member
Petru Andrei
Committee Member
Jim P. Zheng
Committee Member
The Graduate School has verified and approved the above-named committee members, and
certifies that the dissertation has been approved in accordance with university requirements.
ii
ACKNOWLEDGMENTS
First I wish to express my sincere gratitude to my academic supervisor Dr. Hui Li. Her
inspiration and guidance made this dissertation possible. Her generous support and professional
advice enabled me to understand and develop the research. She also taught me the rigorous
attitude towards study and effective research skills, which will be useful for my future work and
life.
I also want to give thanks to Dr. Simon Foo for his encouragement, guidance and support
from the initial to the final level. Furthermore, I am grateful to Dr. Petru Andrei, Dr. Jim P.
Zheng and Dr. Chiang Shih for their great advice and support.
Besides, I would like to thank CAPS staff and members of the research group. In
particular I would like to thank Dr. Yan Zhou, Dr. Xiaohu Liu, Dr. Liming Liu and Dr. Jin Shi
for their guidance and friendship.
Last but not least, I want to give my best appreciation to my parents and family for
always being there for me.
iii
TABLE OF CONTENTS
LIST OF TABLES .......................................................................................................................... v
LIST OF FIGURES ....................................................................................................................... vi
ABSTRACT ................................................................................................................................ viii
1. INTRODUCTION .................................................................................................................... 1
1.1 Research Background ..................................................................................................... 1
1.2 The Need for Battery Energy Storage Systems in PV Systems ..................................... 4
1.3 Research Objectives and Dissertation Proposal ............................................................. 5
2. STATE-OF-THE-ART BESS DESIGN FOR PV APPLICATIONS ....................................... 8
2.1 BESS Design Methods Based on System Requirements ............................................... 9
2.2 BESS Design Methods Based on Economic Analysis ................................................. 11
3. SIZING STRATEGY OF DISPERSED BESS FOR DISTRIBUTED PV SYSTEMS GRID
INTEGRATION ........................................................................................................................... 13
3.1 Sizing Strategy for Dispersed BESS ............................................................................ 14
3.2 System Modeling and Energy Management Strategy .................................................. 16
3.3 Battery Model with Aging Effect Estimation ............................................................... 19
3.4 Case Studies and Simulation Results ........................................................................... 27
3.5 Summary ...................................................................................................................... 43
4. INTEGRATED SIZE AND EMS DESIGN OF BESS FOR UTILITY-SCALE PV PLANT
GRID INTEGRATION ................................................................................................................. 44
4.1 System Description....................................................................................................... 44
4.2 Integrated Size and EMS Design Method of BESS ..................................................... 48
4.3 Case Studies and Optimized Size and EMS Design ..................................................... 52
4.4 Summary ...................................................................................................................... 60
5. CONCLUSIONS AND FUTURE WORK ............................................................................. 61
5.1 Conclusions .................................................................................................................. 61
5.2 Future Work ................................................................................................................. 61
REFERENCES ............................................................................................................................. 63
BIOGRAPHICAL SKETCH ....................................................................................................... 63
iv
LIST OF TABLES
3.1.
Feeder information in the GE distributed power system ..................................................17
3.2
Key parameters related to cost evaluation in (3.3)-(3.7) ...................................................39
4.1
Key parameters related to revenue change maximization in (4.1)-(4.5)............................60
v
LIST OF FIGURES
1.1
Cumulative worldwide capacity for grid-connected PV installations for 2007, 2010 and
2012 (ROW, rest of world) ..............................................................................................................2
1.2
BESS functions for PV applications ...................................................................................5
2.1
State-of-the-art BESS designs for PV applications ............................................................9
3.1
The proposed sizing strategy diagram for dispersed BESS ..............................................14
3.2
The modified GE distribution power system model .........................................................16
3.3
Energy management system .............................................................................................18
3.4
The EMS control flowchart: overvoltage regulation .........................................................20
3.5
The EMS control flowchart: undervoltage regulation and peak load shave ......................21
3.6
LiFePO4 battery equivalent circuit model ........................................................................22
3.7
Cyclation based irreversible capacity loss .........................................................................24
3.8
Experimental measurements of the capacity loss per day .................................................24
3.9
Simulation results without BESS .......................................................................................29
3.10
Simulation results with BESS ............................................................................................29
3.11
BESS lifetime on Bus 205 .................................................................................................30
3.12
Annual stored energy on Bus 205 ......................................................................................31
3.13
Operation time with and without BESS .............................................................................33
3.14
Peaking power generation on Bus 205...............................................................................34
3.15
Annual cost-benefit analysis under λ=0.7: quantitative annual cost ..................................35
3.16
Annual cost-benefit analysis under λ=0.7: normalized annual cost ..................................35
3.17
Annual cost-benefit analysis under λ=0.6: quantitative annual cost ..................................37
vi
3.18
Annual cost-benefit analysis under λ=0.6: normalized annual cost .................................37
3.19
Quantitative annual cost on all buses under λ=0.5 ............................................................38
3.20
Lifetime of LiFePO4 and lead-acid batteries ....................................................................40
3.21
Annual cost-benefit analysis under λ=0.5: quantitative annual cost ..................................41
3.22
Annual cost-benefit analysis under λ=0.5: normalized annual cost ..................................42
4.1
GRU distribution system....................................................................................................45
4.2
PV/BESS plant EMS diagram ...........................................................................................47
4.3
Flowchart for EMS decision making .................................................................................49
4.4
Proposed integrated size and EMS design method of BESS ............................................49
4.5
The power profiles of PV, battery, and grid for eight days ...............................................53
4.6
The annual generation of PV/BESS plant according to EB and d ......................................54
4.7
Battery lifetime according to EB and d...............................................................................55
4.8
The power fluctuation rate according to EB and d..............................................................56
4.9
The average annual peaking power generation according to EB and d ..............................57
4.10
The revenue change of LiFePO4 battery ...........................................................................58
4.11
The revenue change of lead-acid battery ...........................................................................59
vii
ABSTRACT
The market for solar energy has been expanding rapidly worldwide. However, due to the
weather conditions, photovoltaic (PV) systems generally have considerable power variations,
which include voltage fluctuations and frequency variations. As a result, the connected power
systems may experience adverse effects from the fluctuating power generated by the PV system.
The intermittent power generation of a solar farm can perturb the supply and demand balance of
the whole power system. For stability, a power network requires a spinning reserve, which
increases with the growth of PV installations and inevitably degrades the efficiency of power
generation. Therefore, mitigating the adverse effects on the grid from an intermittent PV source
is expected to be essential for increasing the penetration level of PV systems. Recently, battery
energy storage system (BESS) has been seen as a promising solution to help PV integration, due
to the flexible real power control of the batteries. Unfortunately, this technique has not been
applied extensively due to the high cost of batteries. If chosen, battery storage needs to be
designed methodically, which is critical for the owners of PV.
Firstly in this dissertation, an original sizing strategy is proposed for a dispersed BESS in
distribution feeders with distributed PV systems. The main functions of the dispersed BESS are
overvoltage reduction and peak-load shaving. The benefits and cost analysis of the installed
dispersed BESS are conducted. Under high penetration level of PV systems, to assess the effect
of the dispersed BESS on overvoltage reduction, the proposed cost–benefit analysis uses the
work stress of voltage regulation devices as a reference. The factors of load shifting, peaking
power generation, as well as dispersed BESS costs and an estimation of lifetime are considered
in the annual cost calculation. In particular, lithium iron phosphate (LiFePO4) batteries and lead-
viii
acid batteries have been selected to demonstrate the proposed method on the modified GE
distribution feeders. The economic analysis of these two types of battery can determine the lower
cost battery type and the cost-effective size design for the dispersed BESS on different locations
in the distribution system under high PV penetration level.
Secondly, this dissertation proposed a method to optimize the design of a centralized
BESS capacity and the energy management system (EMS) based on a utility revenue analysis for
a large-scale PV plant application. The battery storage, which is controlled by the EMS, aims to
enhance the integration into the grid of a large PV plant by shaping the fluctuating PV plant
output into a relatively constant power and supporting the peak load. LiFePO4 batteries and leadacid batteries are used to demonstrate the proposed method in a utility model. The lifetime and
the systematic performance of these two types of battery are compared. Furthermore, the change
in utility revenue caused by the installed battery storage can be calculated and maximized based
on the proposed method to determine the optimal design of BESS capacity and EMS for a large
PV power plant application.
These two proposed methods can offer insights into the performance and economic
analysis of a BESS in PV applications for project designers and business stakeholders. With the
help of the developed methods, BESS designs can be optimized for any PV application with the
necessary changes according to practical application. Finally, the scope of future work is
discussed.
ix
CHAPTER ONE
INTRODUCTION
1.1 Research Background
1.1.1 Photovoltaic Market Growth
The number of PV installations has grown remarkably since 2000 and the amount of PV
energy has gradually become another major source of power generation for the world. At the end
of 2007, the world’s cumulative installed PV capacity was approaching 10 GW (Fig. 1.1). In
2010, the value had increased more than three times to 40.7 GW. Two years later, more than 100
GW of PV had been installed around the world [1]. At least 110 TWh of electricity is generated
by PV annually. This amount of energy is sufficient for the annual energy consumption of almost
10 million households in the United States [2]. In 2012, Europe led the installed cumulative PV
capacity with more than 70 GW. This represents about 70% of the world’s cumulative PV
capacity. Next in the ranking is China (8.3 GW) followed by the United States (7.8 GW) [1].
1.1.2 Distributed PV Applications and Large-Scale PV Applications
The PV market can be split into two major segments. The first segment includes
distributed PV systems for residential, commercial and industrial applications with rooftopmounted systems. The other includes large-scale PV systems for utility applications with groundmounted systems [3].
Generally, the power rating of residential PV systems is 1–20 kW and the peak power of
commercial or industrial PV systems normally varies from 20 kW to 1 MW. China’s Bureau of
Energy announced approval of distributed PV generation application demonstration parks. More
1
than 1.8 GW of distributed PV pipeline has been approved [4]. Installing PV systems by the enduser also helps reduce transmission and distribution (T&D) losses. About 5–10% of transmission
losses could be saved by placing power at the end-user load. Further, strategically locating PV
Cumulative PV Capacity (GW)
systems or using storage with PV systems can also help to defer transmission upgrades [5].
120
100
Europe
China
U.S.A
ROW
80
60
40
20
0
2007
2010
2012
Fig. 1.1 Cumulative worldwide capacity for grid-connected PV installations for 2007, 2010 and
2012 (ROW, rest of world)
Examples of some commercial and industrial building PV installations include [5]:
•
Santa Rita Jail, Alameda County, CA, 1.18 MW (PV surface area: 130,680 ft2)
•
Ridgehaven Building, San Diego City, CA, 66 kW (PV surface area: 6,500 ft2)
•
Neutrogena Corporation, Los Angeles, CA, 546 kW (PV surface area: 62,000 ft2)
2
Centralized large-scale PV systems are normally greater than 100 kW in size. In the
United States, 24 utility projects were completed in the first quarter of 2013, ranging in size from
1 to 79 MW [3]. Some of the largest central installations are [5]:
•
3.9 MW in Rancho Seco, CA
•
3.8 MW in Springerville, AZ
•
5 MW project in Prescott, AZ
1.1.3 Adverse Effects of Grid-Tied PV Systems
Grid-tied PV systems produce considerable amounts of green energy and reduce carbon
dioxide emissions significantly. However, the generated power from PV systems varies greatly
due to weather conditions, which leads to unpredictable and intermittent power generation.
Compared with the traditional forms of electricity generation, the rapid increase in PV generation
has introduced new challenges for power systems. The major effects for a traditional power grid
are [6]:
a) Large reverse power flow
b) Interaction with capacitor banks and voltage regulators
c) Voltage variations
d) Real and reactive power fluctuations
e) Potential effects for overcurrent and overvoltage protection systems
f) Increased losses and power factor modification
g) Voltage unbalance (for a single-phase PV)
h) Harmonics (potential increase in total harmonic distortion)
3
In addition, these effects will become more significant with the growing penetration of
PV generation in power systems. These adverse effects have to be mitigated to maintain the
stability and reliability of the power supply with the rapid growth of grid-tied PV systems.
1.2 The Need for Battery Energy Storage Systems in PV Systems
The intermittency of PV technology is due to the output power reduction caused by the
movement of clouds over PV modules. This intermittency causes most of the undesired effects,
such as items (b), (c), (d) and (e) in Section 1.1.3. As a result, the amount of spinning reserve has
to be increased with the growth of PV penetration. Hence, the utility’s overall conventional
generation is not reduced due to the higher requirements for spinning reserve. Intuitively, a
BESS is able to control the power flow flexibly, which could facilitate increased penetration of
PV systems by minimizing their intermittency.
For grid applications, a BESS can provide a variety of services [7], depending on the
power rating, storage capacity and location, as shown in Fig. 1.2. The aforementioned adverse
effects of PV systems can be mitigated by a BESS from generation (large-scale PV applications)
to end-user applications (distributed PV applications).
According to Fig. 1.2, the BESS is capable to mitigate all the adverse effects of grid-tied
PV systems. In this research, the BESS is aiming to solve the overvoltage issue for a distributed
PV application and reduce the power fluctuations for a large-scale PV application respectively.
Of all the kinds of battery, lithium-ion batteries have good power and energy densities and they
have been widely applied for grid applications [7]. In particular, LiFePO4 batteries have
gradually occupied a large part of the market due to their extraordinary safety performance and
long cycle life. In this dissertation, LiFePO4 batteries and traditional lead-acid batteries were
selected to demonstrate the proposed optimal sizing strategies of a dispersed BESS system for a
4
distributed PV application and a centralized BESS for large-scale PV application. The lead-acid
battery will be used as the traditional benchmark.
BESS Location
Large-scale PV
Plant
Application
(Generation)
Residential/
Commercial/
Industrial PV
Application
Duration of Battery Output Energy
Short ( < 5 min)
Medium ( 5 min ~ 1 hr)
Frequency Regulation Services
Long ( 1 hr +)
·
·
·
Improve Power
Quality
Voltage, Harmonics
Grid Support
Peak Shaving
Outage Mitigation
·
(Distribution)
Power Quality
Provide Capacity
Renewable Output
Firming
Energy Shifting
Uninterruptable Power
Supply
(End-User)
·
·
·
Energy
Management
Retail Rate
Optimization
Demand Response
Back-up
Fig. 1.2 BESS functions for PV applications
1.3 Research Objectives and Dissertation Proposal
A reliable BESS size design technique is required to ensure the advantages of the BESS
and prevent high costs. An economic analysis of a BESS for certain PV application is required
for business stakeholders as well. The economic benefits that can be realized from a BESS
depend on the application, the size of the system, the sophistication of the system’s electronic
control equipment, the customer’s rate structure and the operating costs [7].
5
The goal of this research is to develop advanced method to optimize the size design of a
BESS for PV integration applications. To achieve this goal, the dissertation will cover two main
topics.
The first topic is the proposal for a sizing strategy of a dispersed BESS with a high
penetration of PV systems. The main objective of the proposed method is to derive the BESS
annual cost calculation considering the systematic performance of voltage regulation and peakload shaving and size the dispersed BESS based on the annual cost analysis. The following will
be presented:
·
Derive the annual cost calculation of a distributed BESS when it is integrated into the
distribution system with distributed PV systems.
·
Quantitatively analyze the benefits and cost of each battery unit in the dispersed
BESS.
·
Uncover the trade-off between systematic performance and the size design of the
dispersed BESS.
·
Determine the size for a dispersed BESS based on an annual cost analysis.
·
Compare the annual costs of LiFePO4 and lead-acid batteries.
The second topic is to propose the method for optimizing the design of the BESS and the
energy management system (EMS) for enhancing the grid integration of utility-scale PV plants.
The main objective of this method is to compare the change in utility revenue for different
designs of BESS and EMS to support constant power production and peaking power generation.
The following tasks will be included:
·
Investigate the effect of BESS capacity and EMS design on the systematic
performance of the PV plants.
6
·
Quantitatively analyze the systematic benefits of the BESS for different capacities
and EMS designs.
·
Derive the calculation for changes in the utility revenue and derive the optimization
objective function of the BESS for a utility-scale PV application.
·
Quantitatively evaluate the performance of the BESS for reducing power fluctuations
under different capacities and EMS designs.
·
Determine the optimal design of BESS and EMS for LiFePO4 and lead-acid batteries.
7
CHAPTER TWO
STATE-OF-THE-ART BESS DESIGN FOR PV APPLICATIONS
Several researchers have demonstrated that BESS can be used to improve the integration
of PV systems [8–10]. A coordinated control of dispersed BESS with traditional voltage
regulators including the on-load tap changer transformers (OLTC) and step voltage regulators
(SVR) was demonstrated to solve the voltage rise problem caused by the high PV penetration in
the low-voltage distribution network [8]. In [9], the use of battery storage in regulating network
voltage was investigated. Different strategies were employed in controlling the storage unit.
Simulations were coupled with real data, taken from a rural single wire earth return network,
showing that battery storage is capable of boosting the network voltage. Similarly, a BESS was
introduced for mitigating voltage unbalance as well as improving the efficiency of the
distribution network [10]. A control algorithm was developed and implemented in order to study
the ability of the BESS to mitigate network voltage unbalance and reduce losses.
On one hand, in these previous studies, generally an oversized BESS was used to meet
the desired features, which will lead to unnecessary high costs. On the other hand, an undersized
battery may not meet the grid-connection standards. As a result, the right BESS design tool is
important and beneficial for PV owners and power system planners.
Recently, many BESS designs have been presented for residential and commercial PV
systems and large PV plant systems [11–21]. These design methods can be divided into two
major categories as shown in Fig. 2.1: (1) BESS designs based on system requirements and (2)
BESS designs based on economic analysis. The BESS designs based on system requirements
were focused on finding the minimum capacity requirement for certain BESS functions [11–19].
8
Based on studies of BESS performance and economic analyses, several BESS sizing strategies
have been developed for single residential battery storage [20–21].
Maximize the difference between benefit and cost
Peak load shaving ($)
VS
BESS cost ($)
Single BESS Design
for Residential PV
System
Systematic Performance
BESS Designs Based on Economic Analysis
Maximize the difference between benefit
and cost of the battery storage
BESS Design for
Large-scale PV
System
Constant production;
Peaking power generation;
Ramp rate limit;
Calculate the minimum energy requirement
Single BESS Design
for Residential PV
System
Voltage regulation;
Peak load shaving;
Frequency regulation;
BESS Designs Based on System Requirements
Fig. 2.1 State-of-the-art BESS designs for PV applications
2.1 BESS Design Methods Based on System Requirements
For residential PV applications, several BESS designs have been discussed [11–15]. A
small storage capacity at residential PV installations was introduced to prevent overvoltages
[11]. Possibilities for preventing overvoltages using a small battery were investigated. The
relation between battery buffer size and overvoltage reduction was presented for a typical
9
Belgian residential distribution feeder. The influence of the buffer along the feeder was
calculated using load profiles and solar data.
A 50 kWh BESS was used to work with a 22 kW PV system to reduce the intermittency
of the PV energy [12]. In [13], a 2.3 kWh lithium-ion battery storage system was selected to
work with a 5 kW PV system for shifting consumption. In order to size a battery to meet dual
objectives of demand shift at peak tariff times and outage protection for commercial buildings, a
stochastic simulation using Monte Carlo technique was offered in [14]. A procedure for sizing a
BESS for the purposed of shaving the peak demand of a residential distribution feeder was
presented in [15]. The BESS power and energy storage rating are determined from actual load
demand data and desired level of peak reduction using the load following method.
A few researchers have investigated BESS designs for utility-scale PV plant applications
as well. A 1 MWh BESS was integrated into a 1.5 MW PV plant [16]. The BESS was applied to
achieve ramp rate control, frequency droop response, power factor correction, solar time-shifting
and output leveling. These functions of the BESS could optimize the benefits of the plant. In
[17], a 1 MWh BESS was connected to a 1.26 MW PV. This BESS design was able to smooth
the output power and regulate PV plant power fluctuations.
In [18], the BESS capacity was selected as 560 kWh for a 1.2 MW PV plant in Tudela,
Spain. The BESS could help the PV plant achieve constant power production, active power ramp
rate limitation and regulation of frequency and voltage. Reference [19] analyzed the minimum
energy capacity ratings that an energy storage system should accomplish in order to achieve a
defined constant power production in a PV power plant.
In these BESS designs [11-19], the capacity was selected to meet the BESS functions’
requirements. Due to lack of the economic analysis with different BESS capacity designs, the PV
10
owners would not know if they would lose or gain money nor would they know the best size of
design.
2.2 BESS Design Methods Based on Economic Analysis
To obtain a serviceable and economical BESS design, several cost–benefit sizing
methods have been discussed for a single BESS in a residential/industrial PV application [20–
21]. In [20], the cost analysis of a hybrid PV and BESS for demand-side applications was
examined. The forms of input data and output data for three different invest plans were proposed
and discussed. A sensitivity analysis to understand those affecting parameters to the investment
was also investigated. The analysis results can support the users with a trade-off among
economic, operational and environmental factors.
The increase cost of adding a specified size of battery energy storage system to a
residential PV system was investigated [21]. To derive the most cost effective battery size, a cost
function based on a proposed physical based battery lifetime model was developed. In addition,
the utility rating was included in the cost calculation-- time of use rating. Real load and PV
power profiles were applied to calculate a lifelike economic factor of the residential PV and
battery system.
However, the previous studies did not consider factors related to BESS sizing on the
supply side of the distribution system, such as the reduced workload on voltage regulation
devices, as well as the reduced peaking power generation costs, which limit the value of the size
planning and evaluation for the BESS. Furthermore, for residential and commercial PV
applications, the BESS installations would be dispersed to work with the distributed PV system.
The aforementioned sizing methods mainly focus on a single BESS with one PV system, which
means that they are not suitable to be applied on the dispersed BESS design. Moreover, the
11
BESS lifetime varies greatly for different usages, depending on the PV-bus feeder location, PV
penetration level, local weather and battery type. As a result, the previous methods need to be
improved because they lack other operational information of the power system with distributed
PV systems and dispersed BESS.
For BESS design in a utility-scale PV application, the design guidelines based on
economic analysis are required and beneficial for the utility companies. In addition, the
PV/BESS plant EMS design also has great impacts on the system performance, which could vary
the BESS capacity selection as well [22-23]. In [22], an optimal power management mechanism
was presented for grid connected PV systems with storage. The objective was to help intensive
penetration of PV production into the grid by proposing peak shaving service at the lowest cost.
Similarly, a price-based EMS was presented for roof-top PV installations with battery and local
load [23]. The power output of the PV/battery systems is scheduled based on the time-varying
price signals that correspond to the demand in the electricity networks. The effects of the EMS
on the battery is significantly, and the BESS design method should take EMS as a factor into
consideration. Consequently, BESS design methods need to be improved for large-scale PV
application as well.
12
CHAPTER THREE
SIZING STRATEGY OF DISPERSED BESS FOR DISTRIBUTED
PV SYSTEMS GRID INTEGRATION
In distributed networks, voltage fluctuation and overvoltage are considered to be the
primary issues of grid-connected distributed PV systems [6]. Traditional voltage control devices,
such as on-load tap changer (OLTC) transformers and step voltage regulators (SVR), are usually
used to regulate voltage within the normal voltage range. However, this voltage regulation
technology will suffer from increased work stress, resulting in reduced system lifetime under
high penetration distributed PV applications. Therefore, the main function of the dispersed BESS
for a distributed PV application is voltage regulation; another major feature of BESS is peak load
shaving.
To obtain an economical and serviceable BESS, we developed a comprehensive method
for cost-benefit analysis and offer design guidelines for both dispersed LiFePO4 and lead-acid
BESS. First, a battery was integrated into each PV bus in the General Electric (GE) distribution
power system model. The real annual load power profile, PV power profile, and temperature data
were adopted. Second, two physical models of a LiFePO4 battery and a lead-acid battery were
developed with an aging effect, so the lifetime information under different operational conditions
could be derived. In addition, operation times of an on-load tap changer transformer (OLTC) and
a step voltage regulator (SVR) in the system were considered as the criteria to evaluate the
overvoltage reduction. Several factors that have an impact on the BESS benefits, including load
shifting and peaking power generation, were recorded. The cost and benefits with respect to
battery size were then analyzed on each bus of the developed distribution power system under
13
different PV penetration levels. Finally, the selection of BESS type could be decided and the
choices of BESS sizes for each bus determined.
3.1 Sizing Strategy for Dispersed BESS
The proposed dispersed BESS design method is shown in Fig. 3.1. It is applied to
investigate battery lifetime, OLTC and SVR work stress, peaking power generation, and peak
load shifting in a distribution power system with BESS and PV units. It consists of two
subsystems. Subsystem 1 includes a distribution power system model, a physical battery model,
and a voltage regulation and peak load shaving oriented energy management system (EMS);
subsystem 2 contains a cost calculation block and a cost-benefit BESS sizing block.
Input Data
PV
Penetration
Annual Load
Profile
Subsystem 1
Annual PV
Annual
Profile
Temperature Profile
Distribution Feeders System Model
- Dispersed BESS & PV
PBESSi
Battery Size
Iterations
Battery
Type
Busi Voltage
SOC
EMS
Physical
Charge -Bus Voltage Control
Battery Model /Discharge
-SOC -SOH
-Peak Load Shaving
EMS Control
Parameters
-Bus Voltage
Reference
-Load Peak
Indicator
-DOD
-Max Charge/
Discharge
Power
Battery Lifetime
Subsystem 2
Cost-benefit Calculation
·
·
·
·
BESS Levelized Cost
Shaved Peaking Power Generation
Load Shifting
OLTC&SVR Work Stress
Cost-benefit
Comparison
Cost-benefit
BESS Size
Fig. 3.1 The proposed sizing strategy diagram for dispersed BESS
14
In subsystem 1, the battery cost-benefit analysis process starts with a three-step cycle for
a given battery size and PV penetration level. First, the distribution power system model reads
current input data, including annual load power profile, PV power profile, and temperature
profile; records the SVR/OLTC operations; delivers the OLTC and SVR work stress, peaking
power generation, and peak load shifting information to subsystem 2; and, finally, generates
dynamic buses’ voltages for the EMS. The EMS is the control center for voltage regulation and
battery protection. Second, the EMS compares the buses’ voltages with the bus voltage
references, which are the upper and lower voltage limits. EMS is able to control BESS to
regulate the bus voltage within normal range by sending charge/discharge requests to the battery.
The OLTC and SVR will take actions if there is still overvoltage or undervoltage on the buses
after BESS operation. Depth of discharge (DOD) and maximum charge/discharge power will be
evaluated to protect the battery. The DOD reveals the battery’s operation range in state of charge
(SOC) [23]. In this project, the operation range of SOC is set from 15 percent to 85 percent. The
output of the EMS block is the charge or discharge command to each physical battery that is
modeled in this paper. In addition, the charge/discharge power is dynamically adjusted to protect
the battery from overcharging/overdischarging, which could degrade its lifetime. Finally, the
physical battery model updates the SOC and state of health (SOH) after obtaining the new
charge/discharge power command. The above three steps will be repeated and new input data
read again until the battery’s SOH reaches the end-of-life condition of battery, which means the
battery is scrapped and the desired battery lifetime evaluation is finished.
After obtaining each battery unit’s lifetime, OLTC and SVR work stress, peaking power
generation, and peak load shifting information from subsystem 1, the economical annual cost
calculation is performed, taking into consideration BESS unit annual cost, the benefit from
15
SVR/OLTC work stress relief, load shifting, and shaved peaking power generation in the
subsystem 2. Using this approach, the cost of each bus can be calculated under every possible
battery size with different PV penetration levels. As a result, a desired cost-benefit BESS size
can be derived.
3.2 System Modeling and Energy Management Strategy
3.2.1 Distribution System Description
Fig. 3.2 The modified GE distribution power system model
A distribution power system model developed by GE [24] is selected and modified in this
paper to investigate voltage rise, BESS usage, and system performance. Fig. 3.2 shows that the
distributed PV units, BESS, and loads are integrated into the GE model. The model includes
some fundamental distribution system components: OLTC as the substation transformer, SVR,
and switched capacitors. This study focuses on Feeder 2. This feeder’s length is about six miles
and the total peak load is 11 MVA. Seven loads represent a mixture of both residential loads and
16
commercial peak loads ranging from 0.3 MW to 5 MW with 0.92 power factor. Hybrid PV and
BESS units are connected on 201 to 207 buses. The primary feeder base voltage is 12.5 kV. The
secondary feeder base voltage is 240 V for residential loads. The service capacity is rated at 1.5
per unit (p.u.) related to the served load. The impedance of transformers is 2.5 percent and the
X/R ratio is 1.5. An average length of 200 feet was selected for the distance from the transformer
to the load. The secondary feeder impedance was calculated based on the conductor with 200 A
of thermal capacity. The detailed feeder information is shown in Table 3.1.
Table 3.1. Feeder information in the GE distributed power system
Feeder 2 Line Impedance
Length (mile)
Conductor
1.3
Z1
0.65
Z1
0.65
Z1
0.97
Z1
1.1
Z2
1.1
Z2
Feeder Conductor Selection
0.648 Ohm/mile
ACSR 556.6 18/1
From
100
201
203
204
205
206
To
201
202
204
205
206
207
Z1
Z2
Type
Transformer
0.768 Ohm/mile
Feeder Voltage Equipment
From
To
5
100
SVR
202
Switched capacitor
205
203
X/R
2
2
2
2
2
2
ACSR 266.8 18/2
Rating
20MVA
10MVA
2.5MVar
3.2.2 EMS
The EMS is the control hub in the distribution power system, with voltage regulation and
peak load shaving functions, as shown in Fig. 3.3. When overvoltage occurs on a certain bus and
17
PV is generating power, the battery will be charged with the excess PV power until the bus
voltage drops to a normal range. Based on the energy stored from this situation, the battery will
discharge to support the bus voltage during peak load time or when voltage sag occurs. However,
if the power reaches the battery power limit or SOC is beyond the limits, the closest voltage
Real Power (kw) Grid Voltage Grid Voltage
control device will change the tap position to regulate the voltage.
0
Upper
Voltage Limit
Without Battery
Voltage Control
With Battery
Voltage Control
Load Shift
PV
Load
Discharging
2
4
6
8
Charging
PV to Grid
Power Limit
Battery
10
12
14
Time (h)
Grid
16
18
Peak
Load
20
22
24
Fig. 3.3 Energy management system
For this study, the Bus 205 is selected as an example to describe the power control
decision-making flowchart in Fig. 3.4 and Fig. 3.5. The same strategy can be applied to other
buses. The initial voltage of Bus 205 is obtained based on the load and PV power data. The bus
voltage is then compared with the voltage upper and lower limits. If the bus voltage is controlled
18
within the normal range, EMS collects next load and PV power data. Otherwise, one of the
following two conditional voltage control strategies is activated:
(a) Overvoltage regulation flowchart: On the one hand, when overvoltage occurs on Bus 205 and
the PV is generating power, battery will be charged with the excess PV power until the bus
voltage drops to normal range. On the other hand, when overvoltage occurs but PV is not
generating power or the battery charge flag shows that battery has reached its power limit,
the closest voltage control device will step down the tap position to reduce the voltage.
(b) Undervoltage regulation and peak load shave flowchart: On the one hand, during peak load
time or when voltage sag takes place, the battery will discharge to support the bus voltage.
On the other hand, when the battery discharge flag shows that the battery has reached its
limit and voltage sag still exists, the closest voltage control device will step up the tap
position to boost the voltage.
3.3 Battery Model with Aging Effect Estimation
3.3.1 Equivalent Model of the Battery Model
The physical battery model is critical to the investigation of the battery lifetime, including
SOC and SOH. Two physical battery models were developed and applied in this study. One is
based on the LiFePO4 battery from A123 systems (APR18650) [25]. The other is based on the
traditional lead-acid battery from Sigmas Battery (SP6-1.2) [26]. The lifetime estimation for
these two types of batteries has been integrated into the respective battery models as well. The
battery equivalent circuit model is shown in Fig. 3.6. The detailed parameters may change with
temperature and have been described in [27]. In order to estimate the remaining lifetime of the
battery (SOH), two major aging effects are considered. One is the cyclation-based capacity loss
( �
� �
), which represents the aging effect during normal discharge operation; the other is
19
calendrical aging-based capacity loss ( �
�
� �
), which represents the aging effect during
nonoperation time. These two aging effects and the lifetime of the battery can be estimated as
shown in (3.1).
Fig. 3.4 The EMS control flowchart: overvoltage regulation
20
Fig. 3.5 The EMS control flowchart: undervoltage regulation and peak load shave
{
�
�
�
�=
� �
� �
− �
=
= ∙
� �
� �
� �
− �
,
�
�
The total available coulomb of the unused battery ( � �
battery datasheet [25], [26]. The used coulomb ( � �
discharge power and time. Consequently,
calendrical capacity loss per day (
�
� �
� �
�
.
) can be estimated based on
) can be calculated based on the
could be determined. Moreover, the
can be also obtained based on the experimental results
21
in [17], which varies with different temperatures (T) and states of charge (SOC). Based on the
nonoperation time ( ), the �
�
� �
can be calculated.
The initial value of SOH for an unused battery is 1. With the increase of battery utilization
and nonoperation time, the battery’s remaining lifetime, SOH, decreases. The battery will be
scrapped when SOH reaches 0, at which time the battery lifetime can be obtained.
Fig. 3.6 LiFePO4 battery equivalent circuit model
3.3.2 Aging Effect Estimation Methods
The cyclation-based capacity loss is derived by counting the amount of transferred
coulombs. The battery cell (APR18650) is capable of 2,000 cycles at 100 percent DOD [25].
22
Accordingly, the �
� �
can be calculated from the battery capacity change in the available
cycles. Fig. 3.7 shows the battery capacity in terms of the available battery cycles to illustrate the
estimated available volume of coulombs before the battery is scrapped. In order to simplify
analysis, the linear function for capacity reduction rate is selected in this paper to calculate the
volume of coulombs during battery discharge. Once the discharged coulombs are equal to the
total area highlighted in Fig. 3.7, the battery is scrapped. The challenge case is when the battery
is always working at 100 percent DOD. When the battery life ends, the battery cycles will reach
to the limit value of 2,000 cycles. In real-world applications, 100 percent DOD is not practical
and the capacity reduction rate may be non-linear; therefore, battery cycles may be greater than
2,000 cycles when the battery capacity drops to 80 percent.
In addition, variation of environment temperature (T) also has an effect on the
�
� �
calculation. The terminal voltage of the battery changes with T [27]. In this case, the
current and the amount of used coulombs under the same amount of power change with T.
Therefore, the temperature influence is also considered in the evaluation of
�
b).
� �
�
� �
.
can be estimated by dividing the used coulombs by the total available coulombs (1-
The �
�
� �
is the effect of capacity loss occurring at the nonoperating state of the
battery. The three main factors for the calendrical aging effect are T, current SOC, and the
nonoperating time of the battery. The estimation of this aging effect is shown in Fig. 3.8, which
is based on the measurement results from [28]. In these functions, the capacity loss/day can be
determined by the given SOC and T. Finally, the �
capacity loss/day with nonoperating time.
23
�
� �
is estimated by the product of
Fig. 3.7 Cyclation-based irreversible capacity loss
Fig. 3.8 Experimental measurements of the capacity loss per day
24
3.3.3 Cost-benefit Analysis
Fig. 3.1 shows that the BESS lifetime, OLTC and SVR work stress, peaking power
generation, and peak load shifting information can be obtained from subsystem 1. Therefore, the
cost of each BESS and the whole feeder can be calculated based on the BESS lifetime. The cost
evaluation determines the most cost-beneficial BESS size. In order to design a battery size more
accurately, the cost calculation will consider multiple factors, including the battery’s annual cost,
annual benefit from reduced OLTC/SVR work stress, annual benefit from shaved peaking power
generation, and annual benefit from load shifting.
In view of the fact that BESS lifetime varies a lot under different scenarios, the lifetime
levelized annual cost analysis method [29] is adopted in this paper. In general, the levelized cost
over its lifetime CL is given by:
=
�∗ +�
+� −
∗
.
where Q is the current cost; n is its lifetime (years) and d is the discount rate [20]. For
simplification, the levelized factor over n years is denoted as �
=
∗ +
�
�
+
�
�−
.
Therefore, the levelized annual cost of BESS on bus i can be expressed as:
�
�
=
�
+
∗�
�
/� ∗
+
∗�
(3.3)
�
where the BESS on bus i is defined as BESSi. The power conversion system (PCS) cost is
proportional to the battery’s power rating (
�
).
the storage is proportional to each battery’s capacity (�
is the unit cost of the PCS. The cost of
�
).
is the unit cost of the battery.
� is the system efficiency [30]. The installation cost is proportional to the sum of PCS and
storage cost [31], and can be represented by introducing an installation cost factor
is the levelized factors related to the BESSi’s lifetime.
25
�
�
.�
�
To generate the same amount of peaking power, gas combustion turbine plant cost is
reduced due to the unleashed power from BESSi during peak load time. The annual benefit,
depending on the peaking power generation from BESSi, can be derived by:
where
��
��
=
��
∗�
.
is the annual average peaking power generation from BESSi and �
is the unit-
levelized annual cost of the gas combustion turbine [32].
The annual benefit for bus i from load shifting can be calculated by:
�
where
�
=
�
∗
����
−
.
����
is the annual stored energy on BESSi during overvoltage regulation, energy used to
shave peak load during on-peak time.
and
�
load and off-peak load, respectively.
�
are the electricity rate at peak
The annual benefit of the annual saved OLTC and SVR operation is reflected in reduced
operation and maintenance (O&M) costs. Since all seven BESS units contribute to the work load
mitigation of OLTC and SVR, this benefit from BESSi can be calculated as:
&
where
&
�
=
&
∗
&
� �
∗
&
∗
∑�=
7
is the capital cost for OLCT and SVR combined.
�
&
�
�
�
.
is the O&M cost
factor [33]. According to [34], the OLTC transformer requires one major maintenance after
every 150,000 operation times (
time when BESS is applied.
assign the benefit to each BESSi.
).
&
�
�
� �
is the annual saved operation
is the annual operation time of BESSi and is used to
Based on (3.3) to (3.6), the annual cost of installing BESSi to regulate voltage and shave
peak load can be obtained by:
26
�
=�
�
−
��
−
�
−
&
.
�
The BESSi size that can minimize the annual cost is defined as the cost-benefit size.
Therefore, the cost-benefit BESS size on bus i can be obtained by minimizing
�
. (3.7) can
be applied to obtain the annual cost and the cost-benefit size for every BESSi in the distribution
network.
3.4 Case Studies and Simulation Results
The proposed method of cost-effective BESS size evaluation has been applied on the
distribution system model and simulated using Matlab simulation. Four cases were selected to
evaluate the proposed method. Case A compares system performance with and without BESS in
one day. Case B shows the BESS lifetime and usage. Case C highlights OLTC/SVR work stress
and peaking power generation with BESS. Case D presents the BESS cost analysis results under
different scenarios.
The simulation results were derived based on the strategy of Fig. 3.1, where the local
temperature profile is drawn from [35]. The load data is obtained from [36] and the peak load is
rated to the seven buses’ peak load on Feeder 2, according to [24]. The PV power profile is
provided by [37]. The local PV penetration level λ is defined as:
�=
∗
%
.
where PPV is annual PV peak power, and PLPL is the annual local peak load. The data from [28] is
then linearly rated to the desired value on each bus.
The BESSi size in the simulation study can be expressed as:
where
�
is the capacity of BESSi.
ℎ
�
=
ℎ
∗
.
is defined as the number of desired operation hours of
BESSi when it is providing local peak load (
).
27
3.4.1 Case Study A: System Performance With and Without BESS in One Day
In order to investigate the effect of BESS integration with the proposed EMS on system
operation, the system performance with BESS and without BESS in one particular day under λ =
0.5 are first shown in Fig. 3.9 and Fig. 3.10. The BESS size applied for this case study is 2-hour
peak load and the PV, load, and BESS power profiles are based on Bus 205.
The same power profiles in 24 hours of PV and load are shown in (a) of both Fig. 3.9 and
3.9. The BESS action for voltage regulation is illustrated in Fig. 3.10 (a) with the stored energy
in BESS from 8:00 a.m. to 12:00 a.m., which contributes to eliminating the bus overvoltage
caused by a large amount of PV output power. During the peak load time, BESS provides energy
to load in order to shave the peak load. Fig. 3.9 (b) and Fig. 3.10 (b) show the primary and
service voltage profile. The voltage limits of primary voltage and the service voltage are adopted
as 0.97 p.u.~1.05 p.u. and 0.94 p.u.~1.04 p.u., respectively [30], [38]. In Fig. 3.9 (b), the service
voltage is higher than primary voltage due to the reverse power flow. With the BESS operation,
these two voltages are very close, as shown in Fig. 3.10 (b). The OLTC and SVR tap positions in
one day are shown in (c).
Fig. 3.9 (c) shows that the tap position has to be changed frequently in order to regulate
the bus voltage within normal range if BESS is not applied. But the tap position can be kept
constant with the help from BESS in Fig. 3.10 (c). Therefore the work stress of OLTC/SVR can
be relieved with BESS by reducing the tap change operations.
28
Fig. 3.9 Simulation results without BESS
Fig. 3.10 Simulation results with BESS
29
3.4.2 Case Study B: BESS Lifetime and Peak Load Shifting
In this paper, Bus 205 is selected as an example to investigate the BESS lifetime and usage. A
similar analysis can be applied to other buses. Fig. 3.11 shows the lifetime of BESS205 under
different BESS sizes and PV penetration levels for the applications of voltage regulation and
peak load shaving. The BESS lifetime can be estimated based on (3.1). The PV penetration level
and BESS size are defined by (3.8) and (3.9), respectively. It can be observed that under low PV
penetration, BESS205 lifetime increases with the size, but the increasing rate slows after 2-hour
peak load BESS size. On the other hand, under high PV penetration the battery will be used more
frequently and, therefore, the lifetime increases with the size growth in a near-linear fashion.
Moreover, for a chosen size of BESS205, its lifetime decreases significantly with the increase of
PV penetration λ.
Fig. 3.11 BESS lifetime on Bus 205
30
As described in (3.5), the benefit from load shifting depends on the stored energy in
BESS205. Fig.3.12 presents the annual energy stored by BESS205 under different PV
penetration levels and BESS sizes. Under higher PV penetration levels, BESS is required to
Fig. 3.12 Annual stored energy on Bus 205
absorb more energy to solve the overvoltage issue; therefore, the annual stored energy increases
rapidly if the BESS size is expanded. However with lower penetration levels, there may be less
reverse PV power, resulting in bus overvoltage. Therefore, less energy needs to be stored in
BESS for voltage regulation. As a result, a certain BESS size is enough to meet the requirements
of stored energy: expanded BESS size will not contribute to the saved energy. The overall trends
of BESSi lifetime and annual stored energy of other buses are similar to the results of BESS205.
31
3.4.3 Case Study C: OLTC/SVR Work Stress and Peaking Power Generation
As shown in Fig. 3.9, the BESS is helpful for reducing the work stress on OLTC/SVR
and achieve peaking power generation. The integrated BESS on all seven buses cooperatively
regulate the buses’ voltages and reduce the OLTC/SVR operation times. Fig.3.13 shows the
annual operation times with and without BESS action under different PV penetration levels and
BESS sizes. The top line shows the annual operation times of OLTC and SVR without BESS
action. The operation times increase significantly with the growth of PV penetration level. The
number under λ=0.7 is almost double the one under λ=0.3. The lines below show the annual
operation times with different selected BESS sizes. The difference between these lines and the
top one represents the saved operation times. Under higher λ and larger BESS size, the difference
is more apparent. This means the saved operation time is greater.
This result is used to evaluate the benefit of mitigating the work stress of OLTC/SVR in
(3.6). It also can be seen from Fig. 3.13 that operation times under small BESS sizes increase
quickly with λ. But the number is still much smaller than the one without BESS action. In
addition, it is obvious that a large size BESS helps to reduce operation times rapidly under higher
λ. However, a certain BESS size, even a small size, is enough to achieve the desired operation
times under lower λ. Consequently, the optimized BESS size should be decided based on λ and
desired operation times.
32
OLTC+SVR Operation Times
4000
3500
Without BESS
With 0.5 h BESS
With 1.0 h BESS
With 2.0 h BESS
With 3.0 h BESS
With 4.0 h BESS
With 5.0 h BESS
Without
BESS Action
3000
2500
Baseline
OT=1946
2000
1500
30
40
50
60
70
PV Penetration (%)
Fig. 3.13 Operation time with and without BESS
The peaking power generation due to BESS205 integration dominates the benefit derived
in (3.4). Fig. 3.14 shows the annual average peaking power generated by different BESS5 sizes
under different values of λ. Since the power supplied by BESS during peak load time is
determined by the energy stored in BESS205, the trend of the shaved peaking power is similar to
the annual saved energy in Fig. 3.12. It is clearly shown that the peaking power generation
increases with the PV penetration level and BESS size under higher values of λ. But the peaking
power generation will be constant under lower values of λ when the BESS size reaches a certain
value. Accordingly, BESS size can be selected based on the λ and peaking power generation
requirement.
33
Fig. 3.14 Peaking power generation on Bus 205
3.4.4 Case Study D: BESSi Cost-benefit Analysis Results
In order to illustrate the BESSi cost evaluation, the BESS annual costs on all 7 buses of
Feeder 2 are analyzed and compared according to BESS sizes. The BESSi size will change from
half-hour peak load power provision to 5-hour peak load power support. The annual costs are
also explored under different λ as shown in Fig. 3.15–3.19. Table 3.2 shows some key
parameters applied in this analysis to derive the results of Fig. 3.13–3.19 from (3.7). These
parameters can be changed based on the user’s applications.
34
Fig. 3.15 Annual cost-benefit analysis under λ=0.7: quantitative annual cost
Fig. 3.16 Annual cost-benefit analysis under λ=0.7: normalized annual cost
35
Fig. 3.15 shows the quantitative annual costs of BESSi under λ=0.7. Due to different
peak load on these buses, the BESS actual capacity is different, especially for Bus 207, which
has the largest peak load [16]. It can be observed that the annual costs of all seven BESS are
positive with all the possible BESS sizes. But, the lowest annual cost can be determined from
this result. In order to clearly reveal the relationship between size of BESSi and its annual cost,
the quantitative annual cost results are normalized in the range from -1 to 1 based on the peakpeak annual cost of BESSi. Fig. 3.16 shows the normalized cost results. It illustrates much more
clearly the variation trend of BESSi annual cost with growth of size. Now, the lowest cost can be
found and the cost-benefit BESSi size can be obtained. In the demonstrated system, the 3-hour
peak load BESS size can yield the minimum cost for BESS205; 2.5-hour peak load for
BESS206; 2-hour peak load for BESS204; and 1.5-hour peak load for BESS207. The cost keeps
increasing with the BESS size for others. The normalized results enable us to find the most costeffective BESS size of all the buses, but the real value of the annual cost should be obtained from
Fig. 3.15.
The BESSi quantitative annual costs under λ=0.6 are presented in Fig. 3.17. Similarly,
the normalized annual cost results are shown in Fig. 3.18. For BESS205, the cost-benefit size is
around 2-hour peak load. A 1.5-hour peak load is the cost-benefit BESS size for BESS204 and
BESS206. The cost keeps increasing with the BESS size for others. It can be seen from Fig.
3.15–3.19 that the cost-benefit size of the same BESS decreases with the PV penetration level.
Fig. 3.19 shows the annual cost of BESSi changes with BESS size under lower PV
penetration level λ=0.5. The BESSi quantitative annual costs always increase with the BESS
size. Therefore, it is difficult to find the cost-benefit size for BESS when the PV penetration level
is 50 percent or lower.
36
Fig. 3.17 Annual cost-benefit analysis under λ=0.6: quantitative annual cost
Fig. 3.18 Annual cost-benefit analysis under λ=0.6: normalized annual cost
37
Fig. 3.19 Quantitative annual cost on all buses under λ=0.5
3.4.5 Case Study E: LiFePO4 and Lead-acid Comparison
In this case study, BESS 205 is selected as an example to investigate the BESS lifetime
and stored energy. Fig. 3.20 shows the BESS205 lifetime with different sizes and PV penetration
levels for LiFePO4 and lead-acid batteries. It is clear that LiFePO4’s longevity is greater than the
lead acid battery’s, but the lifetime trends of these two batteries are similar. Under low PV
penetration, BESS205’s lifetime increases with the size, but the increasing rate slows after 2hour peak load BESS size. Under high PV penetration, batteries are used more frequently;
therefore, the lifetime increases with the size growth in a near-linear fashion. Moreover, for a
38
chosen size of BESS205, its lifetime decreases significantly with the increase of PV penetration
λ.
Table 3.2 Key parameters related to cost evaluation in (3.3)-(3.7)
Parameters
Discount rate
Unit PCS cost
Installation cost factor
LiFePO4 battery cost
Unit-levelized annual
cost of gas
combustion turbine
Electricity rate
at peak load
Electricity rate
at off-peak load
Symbol
d
Value
0.03~0.04
400$/kW
5%
300$/kWh
Remarks
NREL [39]
SANDIA [30]
EPRI [31]
Balqon Corporation[40]
120$/kW-y
PJM Interconnection [32]
EPonpeak
21.715c/kWh
Tallahassee Utility[41]
EPoffpeak
7.536c/kWh
Tallahassee Utility[41]
�
�
CW
�
Under 50 percent PV penetration level, the BESS annual costs on all seven buses of
Feeder 2 are compared between LiFePO4 and lead-acid batteries, according to BESS sizes. The
quantitative
�
of two batteries on all the buses are shown in Fig. 3.21. Although the cost of
te lead-acid battery is lower, the
�
of the lead-acid battery is greater than for the LiFePO4
battery, due to the lifetime performance. In order to clearly reveal the relationship between the
size of BESSi and
�
, the quantitative
to 1 based on the peak-peak
�
�
results are normalized in the range from -1
, as shown in Fig. 3.22.
39
25
LiFePO4
Lifetime (years)
20
15
Lead Acid
10
5
0
70
4
3
2
30
BESS Size
(hour peak load)
Fig. 3.20 Lifetime of LiFePO4 and lead-acid batteries
60
50
5
1
40
PV Penetration (%)
It is much clearer to illustrate the variation trend of BESSi annual cost with increased
size. Then, the cost-benefit size for BESSi can be obtained. To meet the requirement of
overvoltage reduction, the BESS size should be greater than a 1-hour peak load for both
LiFePO4 and lead-acid batteries. The cost-benefit battery sizes for lead-acid type are: 2.5-hour
peak load for BESS204; 4-hour peak load for BESS205; 3-hour peak load for BESS206; and 2hour peak load for BESS207.
40
Lead Acid Annual Cost (k$)
700
Bus 201
Bus 202
Bus 203
Bus 204
Bus 205
Bus 206
Bus 207
600
500
Lead-acid
Battery
400
300
200
100
0
0.5
1
LiFePO4 Annual Cost (k$)
700
1.5
2
2.5
3
3.5
4
Battery Size (Hour Peak Load)
Bus 201
Bus 202
Bus 203
Bus 204
Bus 205
Bus 206
Bus 207
600
500
4.5
5
4.5
5
LiFePO4
Battery
400
300
200
100
0
0.5
1
1.5
2
2.5
3
3.5
4
Battery Size (Hour Peak Load)
Fig. 3.21 Annual cost-benefit analysis under λ=0.5: quantitative annual cost
41
Lead-acid Normalized Annual Cost (k$)
1
0.6
0.4
Lead-acid
Battery
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0.5
LiFePO4 Normalized Annual Cost (k$)
Bus 201
Bus 202
Bus 203
Bus 204
Bus 205
Bus 206
Bus 207
0.8
1
1.5
1
2
2.5
3
3.5
4
Battery Size (Hour Peak Load)
Bus 201
Bus 202
Bus 203
Bus 204
Bus 205
Bus 206
Bus 207
0.8
0.6
0.4
4.5
5
4.5
5
LiFePO4
Battery
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0.5
1
1.5
2
2.5
3
3.5
4
Battery Size (Hour Peak Load)
Fig. 3.22 Annual cost-benefit analysis under λ=0.5: normalized annual cost
42
The cost keeps increasing with the BESS size for other buses and for all LiFePO4
batteries. Therefore, 1-hour peak load is the cost-benefit size for them. The normalized results
facilitate finding the most cost-effective BESS size of all the buses, but the real value of the
annual cost should be obtained from Fig. 3.21.
This proposed method can be applied to other distribution systems that include
distributed PV and dispersed BESS units. If there is no OLTC/SVR, the benefit in (3.6) can be
ignored. In addition, the system parameters of Table 3.1 and some key parameters related to the
cost evaluation of Table 3.2 need to be changed in accordance with different applications. If the
applied battery’s life cycle is available, the battery aging effect could also be estimated using
(3.1). EMS can also be modified flexibly to provide other functions of BESS besides voltage
regulation in the applied distribution system.
3.5 Summary
This chapter has proposed a new method for performing a cost-benefit analysis for
dispersed BESS when it is applied on distributed feeders to improve PV integration. The costbenefit of two types of batteries has been analyzed and compared under certain PV penetration
levels. The tradeoff between cost and benefits of the dispersed BESS can be evaluated
quantitatively for every battery unit. Therefore, the lower-cost battery type can be obtained, and
the most cost-effective battery size can be determined for each one of the dispersed BESS, based
on the results. A modified GE system model was selected for use in this study, but the analysis
method can be applied to any distribution feeders with dispersed BESS and PV units with proper
modifications.
43
CHAPTER FOUR
INTEGRATED SIZE AND EMS DESIGN OF BESS FOR
UTILITY-SCALE PV PLANT GRID INTEGRATION
Based on the sizing method for dispersed BESS, an improved approach to designing
BESS capacity and EMS for the grid-tied PV plant application is presented. The 1-minute
interval PV plant output with load data from Gainesville Regional Utilities (GRU) and the actual
temperature data are applied to achieve more realistic results. Both the LiFePO4 battery and
lead-acid battery are selected and applied, due to the good performance of the former and the low
cost of the latter. In order to compare their systematic performance, LiFePO4 and lead-acid
physical battery models have been developed with the lifetime estimation feature. EMS is
developed to maintain the constant power generation and support peak load. The power
production value in EMS and battery capacity would be varied to study the utility benefits,
including improved production, peaking power generation, and reduced line losses. As a result,
the revenue change can be obtained under all possible circumstances. Finally, the optimized
design of battery capacity and EMS can be determined by maximizing the revenue change for
the LiFePO4 battery and lead-acid battery, and the suitable battery type for the PV plant
applications can be identified accordingly.
4.1 System Description
This section presents the GRU utility system model, physical battery model and EMS.
They were developed and applied in this research to demonstrate the proposed method to
optimize battery and EMS design for large-scale PV plant application.
44
4.1.1 GRU Utility System Model
The GRU distribution system model was developed and applied in this study. This
system is an overhead 12 kV circuit for approximately 4.5 miles. The total system load of GRU
is 11 MVA. The PV penetration level of the GRU system is close to 24 percent with the gridconnected 2.61 MW PV plant. As shown in Fig. 4.1, the GRU distribution system is reduced to a
20-node model. Four 0.9 MVar capacitor banks and a voltage regulator are installed. The per unit
(p.u.) impedances of the transmission lines are labeled in Fig. 4.1 as well. A 2.06 MW PV plant
is connected to Bus 13. Two smaller PV systems are connected to Bus 12 and Bus 16; they are
0.25 MW and 0.35 MW, respectively. The focus of this study is the 2.06 MW PV plant on Bus
13; the battery storage is integrated to the same bus to help the PV plant achieve constant
generation and support peak load.
8
9
10
12
11
13
7
OLTC
Substation
1
2
3
4
5
6
0.00021
0.00063
0.00021
0.00014
0.00049
0.00049
0.00028
0.00063
0.00069
0.00028
0.00049
0.00098
Battery
Storage
0.00118
0.00027
0.00035
0.00014
0.00021
0.00007
0.00035
0.00076
0.00076
0.00076
0.00056
0.00028
0.00090
14
0.00354
OLTC: On-Load Tap Changer
PV: Photovoltaic
SC: Switched Capacitor
Vbase = 12kV
Sbase = 1MVA
Zbase = 144ohm
15
16
PV
PV
(0.25MW)
(2.06MW)
18
17
19
0.00021
0.00028
0.00028
0.00014
0.00021
0.00021
0.00083
0.00083
0.00083
0.00042
0.00063
0.00063
PV
(0.25MW)
Fig. 4.1 GRU distribution system
45
R
X
0.9 MVar SC
4.1.2 EMS
The EMS controls the power flow between the battery storage, the PV plant, the load, and
the grid. The constant power production and the peaking power generation are the two main
functions considered in this study. The EMS was developed and modified based on [18], as
shown in Fig. 4.2, with different d, where d is defined as the ratio of the desired plant power
generation reference Pref to PV peak power PPPV. Based on the weather forecast,
could be
predicted one day ahead. In Fig. 4.2 (a), with d=60 percent, the battery could store some energy
while supporting Pref for six hours (9:30 a.m. – 3:30 p.m.). For the GRU utility whose customers
are mostly residents, the peak load normally happens between 7:00 – 11:00 p.m.. During this
peak load period, the battery generates constant power to the grid until its SOC reaches 50
percent. The peaking power generation is decided by the energy stored in the daytime and the
duration of peak load time. When d increases to 80 percent in Fig. 4.2 (b), the discharged and
charged energy are equal, which means that the battery SOC remains around 50 percent in the
daytime utilization. Consequently, no extra energy is stored in the battery to shave the peak load
at night. But the energy production in the daytime will be greater with higher d and the
production could reach the maximum value when d is 100 percent. In this case (Fig. 4.2 (c)), the
SOC of the battery could be less than 50 percent after supporting Pref for six hours. The battery
needs to be recharged to 50 percent SOC to ensure the next-day usage. The battery will be
charged constantly from the grid during the light load period (2:00 – 5:00 a.m.). The charging
power depends on the battery daytime usage and the duration of light load time. In general, for a
certain battery capacity, higher d could increase the daytime production of the plant, but the
peaking power generation of battery would be limited and the battery lifetime would be
shortened due to frequent discharge according to (3.1). Since d selection in EMS could affect the
46
performance and lifetime of the battery, the optimized design of d in EMS is required and
necessary.
Power
PV + Battery Storage System
PV
Pref = d∙Pppv
Charging
Support Peak Load
PPG
Discharging
0
2
4
6
8
10
12
14
16
18
20
22
24
20
22
24
20
22
24
Time (h)
(a) d = 60%
PV + Battery Storage System
Power
Pref = d∙Pppv
PV
Charging
Discharging
0
2
4
6
8
10
12
14
16
18
Time (h)
(b) d = 80%
Power
Pref = d∙Pppv
0
PV + Battery Storage System
PV
Charging
Light Load
Charging
2
4
Discharging
6
8
10
12
14
16
18
(c) d = 100%
Fig. 4.2 PV/BESS plant EMS diagram
47
Time (h)
The flowchart for EMS decision making has been designed based on the battery functions
as shown in Fig. 4.3. The constant power production, peaking power generation, and light load
charging functions are realized according to this flowchart. In addition, the required power of the
battery storage for different function could be planned as well. In this study, the SOC Min and
SOCMax are selected as 15 percent and 85 percent respectively to protect the battery from
overuse. EB is the capacity of battery storage. Pref depends on d and PPPV, as shown in Fig. 4.2.
PGrid is the sum of the power of PV, load, and battery storage, which means PGrid is the plant
power generation or consumption. PGrid will be compared with Pref . And the power of battery
PB is used to compensate the power mismatch when SOC is within the acceptable range and
SOH is greater than 0. PGrid varies according to � and it is limited by the EB selection as well.
Moreover, battery peaking power generation (PPG) is also dependent on EB and d. This process
will repeat until SOH reaches 0, which means the battery is scrapped. Then the utility revenuerelated battery systematic information can be obtained, including battery lifetime, plant
production, peaking power generation, and power line losses under certain EB and � design.
4.2 Integrated Size and EMS Design Method of BESS
The proposed method to optimize the battery and EMS design in large-scale PV plant
application is presented in Fig. 4.4. EB and d are the variables used in the proposed method.
Based on the system description in Chapter Two, EMS controls the PV and battery storage to
achieve the constant generation and peaking power generation until the battery is scrapped. Then
the utility revenue change can be calculated under all possible designs of EB and d, and the
optimized design of battery storage and EMS can be determined by maximizing the revenue
change.
48
Start
PV Plant Output & Local
Load Data Collection
PB
PPV + PLoad
+
PGRID
|PGRID – Pref|>
P_DeadBand/2
No
Peak Load = 1
Light Load = 1 No
No
Yes
Yes
Yes
PGRID > Pref
SOC > 50%
& SOH > 0
No
SOC > SOCMin
& SOH > 0
Yes
SOC < SOCMax
& SOH > 0
No
Yes
Charge
PB = PGRID - Pref
SOC < 50%
& SOH > 0
Yes
Yes
No
Pbat = 0
No
Yes
No
Constant Discharge
PPG = PB = (SOC-50%)EB/TP
Constant Charge
PB = (SOC-50%)EB/TL
Peaking Power Generation
Light Load Charging
Discharge
PB = PGRID - Pref
Constant Power Production
Fig. 4.3 Flowchart for EMS decision making
Utility System Model
Systematic
Performance
EB
Physical Battery Model
- LiFePO4 - Lead-acid
SOC
& SOH
Load &
PV Plant
Charge /
Discharge
Annual Peaking Power
Generation Benefit
Annual Reduced Power
Line Losses
Annual Generation
Benefit
Battery Levelized
Annual Cost
Load & PV
Power Outputs
EMS
d
Revenue Change
Calculation
- Constant Power Production
- Peaking Power Generation
Revenue Change
Maximization
Optimized
Design of
EB and d
Fig. 4.4 Proposed integrated size and EMS design method of BESS
In the revenue change calculation block (Fig. 4.4), the benefits of the installed battery can
be obtained through the improved power production and the peaking power generation.
49
Moreover, since the PV/battery plant is close to the loads, the line losses of the GRU network
would be reduced as well. Therefore, the utility revenue change calculation can be derived and
the proposed battery and EMS design optimization could be formulated in (4.1)–(4.5). The
objective of (4.1) is to maximize the utility revenue change while enhancing the PV plant grid
integration and satisfying some prevalent constraints (4.2)–(4.5). In (4.1),
�,
�,
and
are
the annual benefits of the improved power production, peaking power generation, and reduced
line losses, respectively. The battery storage levelized cost (
is incorporated in (4.1) as a
penalty. Traditionally, the output of a large-scale PV plant is curtailed to 50 percent of its
maximum power generation in order to reduce the power fluctuation [42]. Accordingly, the
profit of the curtailed PV generation (
��
is applied as the revenue baseline to evaluate the
utility revenue change. The EB and �, which can maximize utility revenue change, will be
considered as the optimized battery capacity and EMS design.
Objective Function:
max
�,
�
+
�
�
+
=∑
�
+
��
�
�=
�
��
∙
��{
∑�=
=
=∑
=(
−
�
�=
)
�=
50
��
� � �{
, �}
, �}
∙
+
=∑
−
�
�
�
{
%
∙�
�
∙
∙
� � �}
+
+
�� ��� .
�
�
−
.
�
is the annual benefit of the plant generation.
is the price of electricity. The time
�
index in minutes is represented by �; n is equal to 525,600, the number of minutes in a year.
� � �
is the plant power generation to the grid.
�
represents the benefit of battery peaking
power generation. For generating the same amount of peaking power, gas combustion turbine
plant cost is reduced, due to the unleashed power from battery storage during peak load time.
��
is the battery power generation during peak load that can be determined by EB and �; m is
the number of the minutes for peak load time in a year, and �
the gas combustion turbine [32].
is the levelized annual cost of
is the benefit of the reduced line loss, where
is the reduced line losses that are related to
� � �.
�
{
� � �}
is the annual levelized cost of battery that
with its lifetime � . The battery power conversion system
depends on the capacity rating
(PCS) cost is proportional to the battery power rating (
power rating (2.06MW) in this study.
proportional to the battery capacity (
), which is equal to the PV plant’s
is the unit cost of the PCS. The cost of the storage is
is the unit cost of the battery. � is system efficiency.
).
The installation cost is proportional to the sum of PCS and storage cost. As a result, the
installation cost factor
�
�
is introduced.
is the interest rate.
The objective function is subject to several constraints that are derived in (4.2)-(4.5).
Constraints:
� � �
�
� ℎ�
=
≤
�
�
�
51
+
�
{
≤
�
≤
{
, �} +
, �} ≤
�
≤
�� .
� �
�
�
ℎ�
.
.
.
.
The first constraint (4.2) maintains the power balance among the PV, the local load, the
battery and the grid. Constraint (4.3) ensures that the charge/discharge power of the battery
remains within the upper and lower bounds specified by the battery manufacturer. The third
constraint (4.4) guarantees that the SOC is kept in the acceptable range. The annual power
fluctuation rate (FR) is limited in (4.5).
is the percentage of the power fluctuation violation
time in one year. Additionally, the power variation limit should be less than 40 kW/min, as is
defined in [42]. The FR of the original GRU PV plant is 20 percent. When the output of the PV
plant is curtailed to 50 percent of its maximum power output [42], the
percent (
to
��
.
��
. By deploying the battery storage, the
�
is reduced to 15
is desired to be comparable
4.3 Case Studies and Optimized Size and EMS Design
The proposed battery and EMS optimized design method has been applied on the GRU
PV plant and simulated using Matlab. In this study, the range of
is from 0.5 MWh to 15 MWh
and � varies from 50 to 100 percent. Four cases are selected to examine the proposed method.
Case A shows the PV plant power production with battery storage for eight days and the annual
plant net generation with battery. Case B compares the lifetime of the LiFePO4 battery and the
lead-acid battery according to EB and d. Peaking power generation and FR are illustrated in Case
C. Case D highlights the analysis results of utility revenue change and the optimized design.
4.3.1 Case Study A: PV Plant Power Production with Battery Storage
To investigate the effect of battery integration with the developed EMS on the PV plant
generation, power profiles of PV, battery, and grid are first presented for eight consecutive days
in Fig. 4.5. The selected EB is 5.75 MWh and � is 80 percent.
52
The battery is controlled by EMS to compensate for the difference between the PV power
and the power production reference. The power profiles of battery and PV for eight days are
shown in Fig. 4.5(a). The plant power generation to grid is constant for six hours during the day
time except for the seventh day, as shown in Fig. 4.5 (b). Furthermore, the battery can support
the peak load at night using the stored energy in five days. However, for the rest of three days,
the battery needs to be recharged during the light load time to ensure next-day usage. For this
certain EB and d, the battery storage can successfully achieve constant production and peaking
power generation in most of these eight days. Note that, on day seven a larger EB or lower d is
needed to achieve the desired functions.
Power Profile (MW)
2
Ppv
Pbat
1
0
-1
0
1440
2880
4320
5760
7200
8640
10080
11520
8640
10080
11520
(a) Ppv and Pbat for eight days
1.5
Power Profile (MW)
Ppv
1
Pgrid
0.5
0
-0.5
0
1440
2880
4320
5760
7200
Time (min)
(b) Pgrid and Ppv for eight days
Fig. 4.5 The power profiles of PV, battery, and grid for eight days
53
The annual generation of PV/battery plant according to different EB and d is shown in
Fig. 4.6. The annual energy generation with the LiFePO4 battery is greater than with the leadacid battery under the same EB and d. This is because the LiFePO4 battery has higher
charge/discharge power limits. The curtailed PV plant generation is 2152 MWh, which is labeled
in Fig. 4.6 as well. It can be observed that larger EB or higher d leads to greater generation by the
plant. However, the growth of the annual generation with EB becomes slow after 6 MWh. As a
result, a large capacity battery is not necessary for generation purposes under this EMS.
2450
Annual Generation (MWh)
2400
2350
2300
2250
2200
Curtailed PV Annual
Generation
2152 (MWh)
2150
2100
d = 60%
d = 80%
d = 100%
d = 60%
d = 80%
d = 100%
2050
2000
1950
2
4
6
8
10
Battery Capacity (MWh)
12
LiFePO4
Battery
Lead-acid
Battery
14
Fig. 4.6 The annual generation of PV/BESS plant according to EB and d
54
4.3.2 Case Study B: Lifetime of the LiFePO4 Battery and the Lead-acid Battery
The battery lifetime of the LiFePO4 battery and the lead-acid battery with different EB
and varied d are shown in Fig. 4.7. The battery lifetime is estimated based on (3.1). Under certain
EMS, the lifetime of the LiFePO4 battery can be five times greater than that of the lead-acid
battery, and the lifetime increases with the capacity. In addition, for a battery with small
capacity, the lifetime with larger d is longer. This is because EMS prevents the battery from
being used, due to the limits of the battery power and SOC. For batteries with large capacity, the
lifetime with smaller d is longer, since less power is required. For EB within 4 MWh to 11 MWh,
the battery lifetime is longer when d=80 percent. From the battery lifetime point of view, d=80
percent is a better choice for most of EB.
d = 60%
d = 80%
d = 100%
d = 60%
d = 80%
d = 100%
Lifetime (years)
20
LiFePO4
Battery
Lead-acid
Battery
15
10
5
0
2
4
6
8
10
Battery Capacity (MWh)
12
Fig. 4.7 Battery lifetime according to EB and d
55
14
4.3.3 Case Study C: Power Fluctuation Rate and Peaking Power Generation
The power FR of the PV/battery plant is varied according to EB and d. The power FR
reduction performances of the LiFePO4 battery and the lead-acid battery are very similar and
have the same trend. Therefore, the power FR of LiFePO4 battery is selected and presented in
Fig. 4.8. The power FR of the original PV is 20 percent; it drops to 15 percent with the curtailed
method. When the battery is applied with PV to achieve constant production, the FR decreases
with an increase in the size of the battery and the rate of FR decrease slows dramatically after 10
MWh. Moreover, the expected power FR reduction should be comparable to the curtailed
35
d = 50%
d = 60%
d = 70%
d = 80%
d = 90%
d = 100%
Power Fluctuation Rate (%)
30
25
Original PV
Power FR
20
Curtailed PV
Power FR
15
10
5
0
0
2
4
6
8
10
Battery Capacity (MWh)
12
Fig. 4.8 The power fluctuation rate according to EB and d
56
14 15
method. Consequently, a 4MWh EB could ensure the performance of the power FR reduction
when d=100 percent. Notably, for this constant power generation EMS, a small size battery
could aggravate the power fluctuation violation.
The average annual peaking power generation according to EB and d is shown in Fig. 4.9.
According to the EMS, d determines the desired PV/battery plant output (Pref). Higher d requires
the battery to release more energy during the day. Therefore, the battery can only store limited
energy from the daytime for the peak load support at night. Under this circumstance, the peaking
power generation of the battery is lower with higher d. The peaking power generation of the
lead-acid battery is lower than for the LiFePO4 battery due to the different charge/discharge
Average Annual Peaking Power Generation (kW)
power limits. For the peaking power generation purpose, lower d with larger EB is preferred.
600
LiFePO4
Lead-acid
500
d = 50%
400
300
d = 60%
200
d = 90%
100
4
6
8
10
12
Battery Capacity (MWh)
14
Fig. 4.9 The average annual peaking power generation according to EB and d
57
4.3.4 Case Study D: Analysis Results of Revenue Change and the Optimized Design
To determine the optimized design of battery storage and EMS, the revenue change is
analyzed and compared according to EB and d for the LiFePO4 battery and the lead-acid battery,
as shown in Fig. 4.10 and Fig. 4.11, respectively. Table 4.1 shows the key parameters applied in
(4.1) – (4.5). These parameters can be changed based on the user’s applications and the projected
values in the future.
Among all the conditions, the revenue change reaches the maximized value when a 5
MWh LiFePO4 battery system is applied with d=100 percent as shown in Fig. 4.10. Therefore, 5
MWh capacity with d=100 percent can be considered to be the optimized design for the LiFePO4
Fig. 4.10 The revenue change of LiFePO4 battery
58
battery storage for this PV plant application. For lower d, a larger battery would be required to
reach their greatest revenue change.
The lead-acid battery lifespan is considerably shorter than the lifetime of the LiFePO4
battery, as shown in Fig. 4.7. Therefore, the levelized cost of the lead-acid battery is significantly
higher than that of LiFePO4 battery. Consequently, the revenue changes with lead-acid battery
under all values of EB and d are always negative (Fig. 4.11), which means that the GRU utility
cannot gain any profit if the lead-acid battery is applied within the PV plant. Moreover, this
analysis demonstrates that d=80 percent is the most effective EMS design for most of EB for a
lead-acid battery.
Fig. 4.11 The revenue change of lead-acid battery
59
4.4 Summary
This chapter has proposed an effective method for determining the optimized design of
battery capacity and EMS when battery storage is applied to improve PV plant grid integration.
This method has been demonstrated on a real utility model for a LiFePO4 battery and a lead-acid
battery. The utility revenue change can be determined for all possible battery capacity and EMS
designs to select the optimized combination of these factors. The proposed method can also be
applied to other utility power systems with large-scale PV units. In addition, when battery
storage is used to achieve other functions, including spinning reserve and frequency regulation
[45], this method can also be applied to optimize the battery storage and EMS design with
necessary modifications.
Table 4.1. Key parameters related to revenue change maximization in (4.1)-(4.5)
Parameters
Symbol
Value
Remarks
Discount rate
Unit PCS cost
Installation cost factor
d
0.03~0.04
400$/kW
5%
NREL [39]
SANDIA [30]
EPRI [31]
Balqon
Corporation[40]
PJM
Interconnection
[32]
Ultralife [43]
New Solar
Investment [44]
LiFePO4 battery cost
Unit-levelized annual cost of
gas combustion turbine
Lead-acid battery unit cost
Electricity price
�
�
CW
300$/kWh
�
120$/kW-y
CW
100$/kWh
EP
0.60$/kWh
60
CHAPTER FIVE
CONCLUSIONS AND FUTURE WORK
5.1 Conclusions
In this dissertation, the proposed BESS design strategy can be applied to size dispersed
battery storage systems for distributed PV systems. Using the proposed strategy, planners can
determine the most cost-effective size of every battery unit in the distribution system according
to the results of the annual cost analysis. It can also be applied to other power systems with
different type of batteries.
For large-scale PV plant applications, the proposed design method can reveal the tradeoff between the design of battery capacity/EMS and the utility revenue. The proposed design
method can determine the optimized design of battery size and the parameters of the EMS
according to the change in utility revenue. The proposed method can also be applied on other
utility-owned PV systems with different type of batteries/EMS.
With the proposed BESS design methods, the PV owners or system planners will have a
sophisticated and comprehensive design tool to optimize BESS size and EMS for their practical
applications.
5.2 Future Work
Based on the research presented above, future work may focus on developing the method
to design hybrid BESS and ultra-capacitor systems. To reduce the capacity requirements of
BESS and prolong the battery’s operation life, an ultra-capacitor can be integrated into BESS
due to its high power density and fast response. This hybrid battery and ultra-capacitor system is
potentially a promising solution for both residential/commercial PV systems and utility-scale PV
61
plant applications. The relationship between the cost of an extra ultra-capacitor and its benefit to
the battery storage system should be investigated.
62
REFERENCES
[1] European Photovoltaic Industry Association, “Global market outlook for photovoltaics 20132017,” 2013.
[2] U.S. Energy Information Administration, [Online]. Available: http://www.eia.gov/tools/faqs
/faq. cfm?id=97&t=3.
[3] S. Kann, “US Solar Market Insight: 1st Quarter 2013 Executive Summary,” GTM Research
and the Solar Energy Industries Association, 2013.
[4] S. Han, “Commercial PV Rooftop Market in China Set for Strong Growth,” NPD Solarbuzz,
2013. [Online]. Available: http://www.solarbuzz.com/resources/blog/2013/08/commercialpv-rooft op-market-in-china-set-for-strong-growth.
[5] Navigant Consulting Inc., “Relative Merits of Distributed vs. Central Photovoltaic (PV)
Generation,” 2004. [Online]. Available: http://www.energy.ca.gov/reports/2004-09-27_50004-065 .PDF
[6] J. R. Aguero, and J. S. Steve, “Integration challenges of photovoltaic distributed generation
on power distribution systems,” IEEE Power and Energy Society General Meeting, pp.1-6,
July 2011.
[7] D. Rastler, “Electricity energy storage technology options: a white paper primer on
applications, costs and benefits. Electric Power Research Institute,” Electric Power Research
Institute, 2010.
[8] 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,” IEEE Trans. Smart Grid, vol. 3, no. 2, pp. 897-906, Jun. 2012.
[9] M. Zillmann, R. Yan, and T. K. Saha, “Regulation of distribution network voltage using
dispersed battery storage systems: a case study of a rural network,” in Proc. 2011 Power and
Energy Society General Meeting, pp. 1-8, Jul. 2011.
[10] K. H. Chua, Y. S. Lim, P. Taylor, S. Morris, and J. Wong, “Energy storage system for
mitigating voltage unbalance on low-voltage networks with photovoltaic systems,” IEEE
Trans. Power Delivery, vol. 27, no. 4, pp. 1783-1790, Oct. 2012.
[11] J. Cappelle, J. Vanalme, S. Vispoel, T. V. Maerhem, B. Verhelst, C. Debruyne and J.
Desmet, “Introducing small storage capacity at residential PV installations to prevent
overvoltage,” in Proc. Smart Grid Communications, pp. 534-539, Oct. 2011.
63
[12] S. A. Zabalawi, G. Mandic, and A. Nasiri, “Utilizing energy storage with PV for residential
and commercial use,” Industrial Electronics, 2008. IECON 2008. 34th Annual Conference of
IEEE, pp.1045-1050, Nov. 2008.
[13] M. Braun, K. Budenbender, D. Magnor, and A. Jossen, “Photovoltaic self-consumption in
Germany - using lithium-ion storage to increase self-consumed photovoltaic energy,” 24th
European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 2009.
[14] C. W. Tan, T. C. Green, and C. A. hernandez-Aramburo, “A stochastic method for battery
sizing with uninterruptible-power and demand shift capabilities in PV (photovoltaic)
systems,” Science Direct Energy Journal, pp. 5082-5092, Sept. 2010.
[15] C. Venu, Y. Riffonneau, S. Bacha, and Y. Baghzouz, “Battery storage system sizing in
distribution feeders with distributed photovoltaic systems,” in Proc. 2009 IEEE PowerTech
Bucharest, Jul. 2009.
[16] C. A. Hill, M. C. Such, D. Chen, J. Gonzalez, and W. M. Grady, “Battery Energy Storage
for Enabling Integration of Distributed Solar Power Generation,” IEEE Trans. Smart Grid,
vol.3, no.2, pp.850-857, June 2012.
[17] L. Xiangjun, H. Dong, L. Xiaokang, “Battery Energy Storage Station (BESS)-Based
Smoothing Control of Photovoltaic (PV) and Wind Power Generation Fluctuations,” IEEE
Trans. Sustainable Energy, vol.4, no.2, pp.464-473, April 2013.
[18] H. Beltran, E. Bilbao, E. Belenguer, I. Etxeberria-Otadui, and P. Rodriguez, “Evaluation of
Storage Energy Requirements for Constant Production in PV Power Plants,” IEEE Trans.
Industrial Electronics, vol.60, no.3, pp.1225-1234, March 2013.
[19] W. F. Su, C. E. Lin, and S. J. Huang, “Economic analysis for demand-side hybrid
photovoltaic and battery energy storage system,” in Proc. Industry Applications Conference
Thirty-Fourth IAS Annual Meeting, vol. 3. pp. 2051-2057, 1999.
[20] A. Aichhorn, M. Greenleaf, H. Li, and J. Zheng, “A cost effective battery sizing strategy
based on a detailed battery lifetime model and an economic energy management strategy,”
2012 IEEE Power and Energy Society General Meeting, pp. 1-8, July 2012.
[21] Y. Riffonneau, S. Bacha, F. Barruel, and S. Ploix, “Optimal Power Flow Management for
Grid Connected PV Systems with Batteries,” IEEE Trans. Sustainable Energy, vol.2, no.3,
pp.309-320, July 2011.
[22] C. L. Nge, O. M. Midtgard, L. Norum, “PV with battery in smart grid paradigm: Price-based
energy management system,” in 38th IEEE Photovoltaic Specialists Conference (PVSC), pp.
575-579, June 2012.
[23] T. Guena, and P. Leblanc, “How Depth of Discharge Affects the Cycle Life of LithiumMetal-Polymer Batteries,” in 28th Annual International Telecommunications Energy
Conference, pp.1-8, Sept. 2006.
64
[24] E. Liu and J. Bebic, “Distribution system voltage performance analysis for high-penetration
photovoltaic,” GE Global Research, ADC-7-77032-01, Niskayuna, NY, Feb. 2008.
[25] [Online]. Available: http://www.a123systems.com
[26] [Online]. Available: http://sigmastek.com/EN/products.cfm?seriesID=1
[27] M. Greenleaf, “Physical based modeling and simulation of LiFePO4 secondary batteries,”
Dissertation, Dept.ECE., Florida State Univ. Tallahassee, 2010.
[28] A. Farman, “Erstellen eines messtechnisch gestützten Modells zur Berechnung der
kalendarischen Alterung von LiFePO4-Batterien”, Hochschule München, Sept. 2010.
[29] H. L. Willis and W. G. Scott, Distributed Power Generation Planning and Evaluation, New
York: Marcel Dekker, Inc., pp. 141-144, 2000.
[30] S. M. Schoenung, and V. H. William, “Long-vs. Short-Term Energy Storage Technologies
Analysis. A Life-Cycle Cost Study. A Study for the DOE Energy Storage Systems Program,”
Sandia National Laboratories, 2003.
[31] D. Raslter, A. Akhil, D. Gauntlett and E. Cutter, “Energy Storage System Costs 2011
Update Executive Summary,” EPRI, 2012. [Online]. Available: http://www.eosener
gystorage.com/documents/EPRI-Energy-Storage-Webcast-to-Suppliers.pdf.
[32] S. A. Newell, J. M. Hagerty, K. Spees, J. P. Pfeifenberger, Q. Liao, C. D. Ungate, and J.
Wroble, “Cost of New Entry Estimates for Combustion-Turbine and Combined-Cycle Plants
in PJM,” PJM Interconnection, L.L.C., May 2014.
[33] P. Gomatom, and J. Ward, “Feasibility Evaluation of Distributed Energy Generation and
Storage for Cost And Reliability Using the" Worth Factor" Criterion,” in Proc. 2012
Frontiers of Power Conference, Engineering Energy Laboratory, Oklahoma State University,
2002.
[34] [Online].Available:http://www05.abb.com/global/scot/scot252.nsf/veritydisplay/f74113036
3e7c6de85256d6f007aade1/$file/030129_On-load.pdf.
[35] [Online]. Available: http://wunderground.com.
[36] [Online].Available:https://www.sce.com/wps/portal/home/regulatory/loadprofiles/!ut/p/b0/0
4_Sj9CPykssy0xPLMnMz0vMAfGjzOINLdwdPTyDDTwtfAKNDTydnDz9zdxMjA28jfQ
Lsh0VAY010s4!/.
[37] [Online]. Available: http://egauge699.d.egauge.net/index.html.
[38] Willis, H. Lee. “Power distribution planning reference book,” CRC press, 2004.
[39] [Online]. Available: http://www.nrel.gov/analysis/tech_lcoe.html.
[40] [Online]. Available: http://evcargo.com/2012/05/23/121/.
65
[41] [Online].Available:http://www.talgov.com/you/you-customer-helpful-rates-res-elec.aspx.
[42] SunPower Corporation, “Methods of Integrating a High Penetration Photovoltaic Power
Plant into a Micro Grid,” in 35th IEEE PVSC, Honolulu, Hawaii, 2010. [Online]. Available:
http://energy.sandia.gov/wp/wp-content/gallery/uploads/Methods-of-Integrating-a-HighPenetration-Photovoltaic-Power-Plant-into-a-Micro-Grid.pdf.
[43] [Online]. Available: http://www.newsolarinvestment.com/2mw-ground-mounted-solar-proj
ect/.
[44] [Online]. Available: ultralifecorporation.com/download/275/.
[45] D. Ton, C. Hanley, G. Peek, and J. D. Boyes, “Solar energy grid integration systems–energy
storage (SEGIS-ES),” Sandia National Laboratories, Report# SAND2008-4247, 2008.
66
BIOGRAPHICAL SKETCH
Ye Yang received the B.S. degree in information engineering from Tianjin University in
2008, Tianjin, China, and the M.S. degree in electrical engineering from Texas Tech University
in 2010, Lubbock, TX. He started his Ph.D. program in Department of Electrical and Computer
Engineering at the Florida State University from 2011. Meanwhile, he worked as a graduate
research assistant at Center for Advanced Power Systems. His research interests include
integration of renewable energy sources and large scale battery energy storage system
optimization.
He has four journal and conference publications listed below since 2011.
[1]Y. Yang, H. Li, A. Aichhorn, J. Zheng, M. Greenleaf, "Sizing Strategy of Distributed Battery
Storage System with High Penetration of Photovoltaic for Voltage Regulation and Peak Load
Shaving," IEEE Transactions on Smart Grid, vol.5, no.2, pp.982-991, 2014.
[2]Y. Yang, N. Yang, H. Li, "Cost-benefit study of dispersed battery storage to increase
penetration of photovoltaic systems on distribution feeders," IEEE PES General Meeting, 2014,
Washington, D.C, USA, pp.1-5.
[3]Y. Yang, H. Li, "Performance analysis of LiFePO4 battery energy storage for utility-scale PV
system," IEEE Energy Conversion Congress and Exposition (ECCE), 2014, Pittsburgh, PA,
USA, pp.414-419.
[4]R. Mo, Y. Yang, H. Li, "Power hardware-in-the-loop simulation of integrated voltage
regulation and islanding detection for distributed PV systems on GRU model," IEEE Energy
Conversion Congress and Exposition (ECCE), 2014, Pittsburgh, PA, USA, pp.4004-4007.
67