Вы находитесь на странице: 1из 9

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO.

2, MARCH 2014 1129

Power System Reliability Impact of Energy Storage


Integration With Intelligent Operation Strategy
Yixing Xu, Student Member, IEEE, and Chanan Singh, Fellow, IEEE

Abstract—Electric power industry is experiencing a movement Energy price for period


from the existing conventional electric grid to a more reliable,
efficient and secure smart grid. In order to achieve these goals, Forecasted energy price for future period
components such as energy storage will be included, and poten-
tially in large scale. Many feasible applications of energy storage
in power systems have been investigated. The major benefits of Utilized renewable energy in period
energy storage include electric energy time-shift, frequency regu-
lation and transmission congestion relief. In this paper, we focus Forecasted renewable energy utilization
on the reliability improvement of the bulk power system brought in period
by the utilization of energy storage in the local distribution sys-
Available renewable energy in period
tems integrated with renewable energy generation. An intelligent
operation strategy for energy storage which improves reliability Forecasted available renewable energy
considering the renewable energy integration is presented. The
in period
smart grid communication and control network is utilized to
implement the proposed energy storage operation. A bulk power Energy purchased in power market for
system reliability evaluation framework is proposed to study period
the reliability impact brought by the energy storage integration
and operation. A detailed case study and sensitivity analysis is Energy planned to be purchased in future
performed to demonstrate the effectiveness of the presented oper- period
ation strategy and evaluation framework, and to provide valuable
insights on the power system reliability impact derived from the State of charge level at the end of period
energy storage integration.
Index Terms—Bulk power system, electric energy storage, model Energy storage minimum state of charge
predictive control, reliability, renewable energy resources.
level
Energy storage maximum state of charge
NOMENCLATURE level
Energy storage charging efficiency
Power charged to energy storage in Energy storage discharging efficiency
period
Energy storage maximum charging I. INTRODUCTION
power limit
Power discharged from energy storage
in period I N the environment of smart grid, new infrastructure and
technologies are put into place for better information flow
and communication between all stakeholders in the system, in-
Energy storage maximum discharging cluding generation entities, system operators, and distributions
power limit system load aggregators [1]. By taking advantage of the com-
Load in current period munication network, useful local and global data and informa-
tion could be sent and received from each entity in a timely
Forecasted load in future period fashion. With the rich and timely data and information, players
in the power system can make better decisions to improve reli-
ability and economy for themselves and for the system. In this
Manuscript received January 08, 2013; revised June 19, 2013 and August 09, paper, energy storage devices are operated with an intelligent
2013; accepted August 11, 2013. Date of publication August 21, 2013; date of strategy which fully utilizes the data and information received
current version February 14, 2014. The work reported here was supported in
to make charging or discharging decision in order to improve
part by a PSERC Project. Paper no. TSG-00015-2013.
The authors are with the Department of Electrical and Computer Engineering, the economy of the local distribution system and improve the
Texas A&M University, College Station, TX 77843 USA (e-mail: yixing@neo. bulk power system reliability. Considering the increasing pen-
tamu.edu; singh@ece.tamu.edu).
etration of renewable energy generation [2], the distributed re-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. newable energy resources, such as wind and solar generation,
Digital Object Identifier 10.1109/TSG.2013.2278482 integrated in distribution systems are also considered.

1949-3053 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1130 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

With the recent rapid development of energy storage tech- Instead, a load aggregator is an entity who participates in the
nologies, and their flexibility during operation, interest in in- wholesale power market, and purchases the energy for its indi-
tegrating energy storage into power systems to improve sys- vidual customers and delivers the power to them.
tems reliability and economy has been growing. A reliability Normally, load at a given time is determined by customers’
cost/worth evaluation method that can incorporate the impact demand and is inelastic to price except considering demand re-
of wind energy and energy storage utilization in electric power sponse. In some recently implemented demand response pro-
systems is presented in [3]. Authors in [4] evaluates system reli- grams, load could be controlled (curtailed) during some periods
ability considering wind and hydro power coordination, where of time by system operator or customers could adjust their de-
hydro facilities with energy storage capability are utilized to al- mand according to real-time electricity price. Considering the
leviate the impact of wind power fluctuations and also improve relative small scale of such demand program and the scope of
the system adequacy. A methodology for the operation of a hy- this work, in this paper, such demand response programs are not
brid plant with wind power and hydrogen storage to maximize considered. The load is assumed to be determined by customers
economic benefits in a market environment is presented in [5]. and inelastic.
Researchers in [6]–[8] investigated the operation strategies of In a market environment, a load aggregator’s objective is to
energy storage on the bulk power system level and evaluated serve its customers reliably while minimizing cost. With the in-
the reliability impact. tegration of energy storage devices, a load aggregator has more
Previous reported work has been mainly on the operations of flexibility in managing energy transaction and delivery.
energy storage and renewable energy at the bulk power system The distribution network topology with energy storage and
level and evaluate the corresponding reliability impact. How- distributed renewable energy is simplified as in Fig. 1. All the
ever, more and more energy storage devices are being integrated loads, energy storage devices and distributed renewable energy
into the distribution system and used to support the distribu- resources within this distribution system are modeled as lumped
tion system operations. For example, in smart grid demonstra- components for simplicity. However, more detailed modeling,
tion project funded by United States Department of Energy, the such as distributed modeling along the feeders, could also be
community energy storage is deployed at DTE Energy to benefit used. The load in each period , , is inelastic. The en-
consumers, businesses, and utilities [9]. When energy storage ergy storage can be charged with the power purchased in power
is operated to benefit the distribution system, it can also be market and delivered from external grid. The energy storage can
coordinated by the system operator to have an impact on the also be discharged to support the load . Charging power
bulk power system. The smart grid communication and con- and discharging power of energy storage are con-
trol network makes the operation strategies possible which re- trolled by the load aggregator. Charging and discharging ef-
quires real-time information and control signal communication ficiencies are considered. More detailed energy storage model
between multiple entities. This paper focuses on the reliability could be used according to the specific needs. Imported power
impact on the bulk power system considering the energy storage is the power purchased in power market and delivered
operation strategies within the distribution system and coordi- through external grid in each period . If power is allowed
nation by the bulk power system operator. to be sold back to the external grid, could be negative,
In this paper, the energy storage operation strategy for which means power flows from the distribution system to the
distribution system load aggregator is presented to improve external grid. This operation could occur when there is exces-
the economy of distribution system by minimizing energy sive power generated from the distributed generation integrated
purchasing cost in power market. Energy storage coordination in the distribution system. Distributed renewable energy gener-
strategy for system operator is proposed to improve bulk ation provides power to the distribution system. During
power system reliability taking into consideration the local each period , the energy balance is formed with the load, dis-
energy storage operations. In order to accurately quantify the tributed renewable energy generation, energy storage charging/
reliability impact brought by energy storage, a reliability eval- discharging power and the imported power. Load aggregator of
uation framework based on sequential Monte Carlo simulation a distribution system participates in the wholesale power mar-
approach is presented. The reliability impact of energy storage kets to purchase electric energy to serve its customers in the dis-
is then studied and insights are gained. tribution system. Meanwhile, load aggregator is also assumed
This paper is organized as follows: Section II describes to operate the renewable energy generation and energy storage
the energy storage operation strategy in distribution systems. integrated in its served distribution system. Renewable energy
Section III presents the energy storage coordination strategy generation can be controlled by curtailing available renewable
when there is a loss of load event. Section IV describes the energy output. Energy storage devices are operated by deter-
proposed reliability evaluation framework taking into consid- mining the charging/discharging operations. The objective of
eration the energy storage operation and coordination. Case the load aggregator is to serve its customers with reliable power
studies and insights are presented in Section V. Conclusion is supply while minimizing the electric energy purchasing cost in
presented in Section VI. power market.
The following assumptions are used in the study of the
II. ENERGY STORAGE OPERATION IN A DISTRIBUTION SYSTEM system. Only the active power is considered. The power loss is
In many electricity markets, small individual customers lo- not considered. Voltage levels are assumed to be properly regu-
cated in a distribution system normally do not directly partic- lated. These assumptions are normally acceptable in reliability
ipate in the electricity market to purchase energy they need. analysis.
XU AND SINGH: POWER SYSTEM RELIABILITY IMPACT OF ENERGY STORAGE INTEGRATION 1131

a price taker. Thus, the energy purchasing cost for periods


starting from period is

(1)

The energy storage unit is modeled by its operation limits


which include energy storage maximum and minimum state of
charge level, charging/discharging power limit (short for power
limit in this paper), and charging/discharging efficiency. The
Fig. 1. Simplified distribution system integrated with energy storage and dis- state of charge at the end of each period is determined by the pre-
tributed renewable energy generation.
vious period state of charge level and the charging/discharging
operation during this period, it is expressed as

A. Modes of Operation (2)

With the integrated energy storage and distributed renewable All energy storage operation variables are within their oper-
energy generation, the load aggregator of the distribution system ation limits.
has more flexibility during operation. It can utilize the flexible Many methods have been developed for load forecasting [10],
control of the energy storage to adjust the imported power from [11], renewable energy forecasting [12], [13], and energy price
the power market and manage the variable and intermittent re- forecasting [14]–[16]. Load aggregator could utilize these tech-
newable energy generation. Considering the component failures niques to obtain short term forecasts of load, available renew-
and congestions in the power systems, the distribution system able energy and energy price to determine the operation. As
with renewable energy generation and energy storage are oper- forecasting methods are not the focus of this paper, the fore-
ated in four modes: Grid Connected Mode, Coordination Mode, casted information is assumed to be given.
Islanding Mode, and Failure Mode. The basic procedure of MPC is that a finite-horizon optimiza-
Operation Strategy in Grid Connected Mode: In grid con- tion problem determining the series of optimal control opera-
nected mode, there is no fault within the distribution system; tions is solved before each control step, but only the first control
external grid could deliver the required power to the distribu- operation is implemented [17]. After implementing the first con-
tion system. In this mode, power from external grid, renewable trol step, the system updates the actual state of the system and
energy generation and energy storage can all be utilized to serve the future states using a predictive model. Then the control rou-
the load. The objective of the load aggregator is to minimize its tine is repeated to determine the next step’s operation. Applying
energy purchasing cost in power market while meeting the de- the above MPC approach, energy purchasing cost minimization
mand. The management of energy supplies is crucial in deter- problem at period can be implemented as follows:
mining the energy purchasing cost. 1) Obtain the actual load, available renewable energy and
In the environment of smart grid, new infrastructure and tech- price in the current period .
nologies are put into place for better information flow and com- 2) Select a receding optimization horizon of periods (e.g.
munication between all stakeholders in the system. By taking 24 hours). Use load, renewable energy and price forecast
advantage of the communication network, useful data and infor- models to obtain the most updated load, renewable energy
mation could be sent and received from each entity in a timely and price forecasts for the next periods, from period
fashion. With the rich and timely data and information, load ag- to .
gregators can make better decisions to achieve their goal. En- 3) Solve the optimization problem, formulated as follows.
ergy storage devices are operated with an intelligent strategy Objective: Minimizing energy purchasing cost from period
which fully utilizes the data and information received to make to
operation decisions in order to improve the economy of the
local distribution system. A Model Predictive Control (MPC)- (3)
based operation strategy is proposed to minimize the energy
purchasing cost by optimally coordinating the energy purchase
from the power market, energy storage charging/discharging op- The first part is the energy purchasing cost of the
eration, and utilization of renewable energy generation. Before current period . The second part is the
further describing of the proposed operation strategy, power predicted total energy purchasing cost of the following periods
market modeling and energy storage modeling are introduced from to . and are the decision variables
first. to be solved.
In this paper, the power market is simplified as a real time Constraints:
power market model. During each market period (e.g. 5 min- i. Energy storage operation constraints
utes), the amount of energy needed is determined by load ag-
(4)
gregator. Then it submits its offer to get that needed amount
of energy in the market. The energy price is determined by the (5)
market clearing mechanism. Load aggregator is assumed to be (6)
1132 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

where .
ii. Available renewable energy constraints

(7)
(8)

where . The utilized renewable en-


ergy is equal to or less than the available renewable en-
ergy. Extra energy is curtailed according to the operation
constraints. For example, by controlling the pitch angle
of blades of wind turbine, the wind power output can be
curtailed [18].
iii. Power balance constraints

(9)
(10)
Fig. 2. Load aggregator communication and control network structure.
where . Load, available renew-
able energy, and price in current period are the actual
values and known. While load, available renewable en- The structure of the communication and control network sup-
ergy, and price in future periods are predicted using fore- porting the operation strategy is presented in Fig. 2.
cast models, thus are given parameters for the optimiza- As shown in Fig. 2, load aggregator utilizes the smart grid
tion problem. The solution of this optimization problem communication and control network to receive real-time up-
gives an optimal operation schedule for energy storage dated information and execute the operations. The received in-
charging/discharging operation, energy purchase and re- formation includes energy price forecast, energy storage state
newable energy utilization from period to . of charge, load forecast, and renewable energy forecast. With
the forecasts information, load aggregator can use the MPC-
1) Implement the first period’s operation of the solved opera- based operation strategy to determine the operations which in-
tion schedule, which is the current period . clude transactions in power market, energy storage charging/
2) Update the energy storage state of charge level, move to discharging operation and renewable energy curtailment.
the next period, and repeat the algorithm from step 1. Operation Strategy in Coordination Mode: When the distri-
The short-term (e.g. next 2–3 hours) forecast is relatively bution system is operated in grid connected mode, failures could
more accurate than the longer term (e.g. 23–24 hours) fore- occur in the bulk power system. These failures could cause loss
cast. The load, renewable energy and price forecasts are updated of load events in the bulk power system. The bulk power system
according to the newest information after each operation step. operator could utilize the energy storage resources in the distri-
Then the most updated forecasts could be effectively integrated bution systems to avoid such loss of load events. This opera-
into the operation decision making process to minimize the en- tion mode requires coordination between the local distribution
ergy purchasing cost. Better operation for current period can be system load aggregator and the bulk power system operator. De-
determined by taking into consideration the future load, renew- tails of this energy storage coordination operation strategy are
able energy and price. presented in Section III.
The solution for energy storage operation schedule is optimal Operation Strategy in Islanding Mode: In islanding mode,
with respect to the given forecast. The accuracy of the fore- the external grid cannot deliver power to the distribution system
cast will affect the optimality of the operation schedule, because due to failures in the external grid. The objective of the load ag-
of the differences between the forecast and the actual values, gregator is to avoid or minimize load curtailment. The available
such as wind power output. In the case study of this paper, we renewable energy is first utilized to serve the load. If it is not
have assumed the forecast to be perfect but if information on enough to cover the load, energy storage is discharged to sup-
characteristics of forecast uncertainty were available, it could port the load. If there is extra renewable energy, it is stored in
be incorporated in the determination of the schedule. The pro- energy storage for future usage. The extra energy which cannot
posed MPC-based operation strategy could significantly mit- be stored in energy storage due to operation constraints is then
igate the negative impact caused by the forecast errors. This dumped.
strategy constantly updates the forecasts, and utilizes the most Failure Mode: In this mode, there is a failure inside the dis-
accurate short-term forecasts to determine the current operation. tribution system. Power cannot be supplied either from the ex-
The detailed study regarding the impact brought by the forecast ternal grid or from the power sources within the distribution
errors could be found in our previous work [19]. system until the failure is cleared.
XU AND SINGH: POWER SYSTEM RELIABILITY IMPACT OF ENERGY STORAGE INTEGRATION 1133

B. Energy Storage Operation Impact on the Distribution and prevent the system going to failure state. If there is still
System Economic Performance not enough energy to cover the entire load, the system is in
failure state. However, even when the system is in failure
The operation of energy storage could improve both the reli-
state, the energy storage still provides its stored energy to
ability level and the economic performance of the distribution
minimize the amount of loss of load for the studied period.
systems. With the MPC-based operation strategy, the distribu-
6) Send the operation instructions to the energy storage and
tion system could obtain considerable economic benefits by op-
execute the operations for this period.
timally controlling the energy storage charging and discharging
7) Load aggregator updates the energy storage SOC.
operation. The detailed analysis regarding the economic impact
8) Go to step 2 until the system is restored to normal state.
by the energy storage operation could be found in our previous
The real-time information communication helps coordinate
work dedicated to this topic [20]. Due to the page limitation and
the energy storage operation between the local distribution
the focus of this paper, the following sections will focus on the
system load aggregator and the bulk power system operator.
reliability impact on the bulk power system through the energy
This coordinated operation eventually benefits both the local
storage coordination.
distribution system load aggregators through energy cost
saving and the system operator with higher bulk power system
III. ENERGY STORAGE COORDINATION reliability.
The energy storage integrated in the distribution system could
not only improve the local distribution system economy but also IV. RELIABILITY EVALUATION FRAMEWORK
contribute to the bulk power system reliability improvement. In order to quantitatively evaluate and study the reliability im-
During the energy storage operation in grid connected mode, pact on the system brought by the energy storage integration, a
there is a certain amount of electric energy stored most of the reliability evaluation framework is proposed. Within this frame-
time. The real-time updated energy storage state of charge in- work, the detailed energy storage operation strategies are simu-
formation is sent to the load aggregator. When a communica- lated to truly reflect its impact on the power system. The relia-
tion channel between a distribution system load aggregator and bility impact on the bulk power system caused by the local en-
the bulk power system operation is established, this real-time ergy storage integration in the distribution system could be ac-
energy storage SOC information can be also sent to the system curately evaluated. This information is valuable for the system
operator. As the system operator has the information of the bulk operator when studying the energy storage integration impact,
power system, it can coordinate the power supplies and use the planning for further energy storage integration or expansion,
energy storage integrated in the distribution system to support evaluating the reliability and economic benefits and conducting
part of the bulk power system’s load when there is a loss of load cost/worthy analysis.
event occurring somewhere in the system. The proposed reliability assessment framework is based on
An unexpected outage of several generation units or simulta- Sequential Monte Carlo Simulation. In the reliability assess-
neous failures of several major transmission lines could cause ment, a power system is considered to be operating in either
insufficient generation for the system or transmission conges- success state or failure state. A system is considered operating in
tion. In either case, loss of load events could occur in the system. success state when it has enough generation capacity to serve the
In this situation, the system operator could analyze the energy load. When generation capacity is not sufficient to meet the load
storage state of charge information which has already been col- demand and loss of load occurs, the system is in failure state.
lected and determine whether and how to discharge the energy The probabilities and durations associated with the system re-
storage in order to avoid the loss of load or reduce the un- siding in success and failure states and energy not served during
served energy, which subsequently leads to higher bulk power failure states are the reliability indices.
system reliability. The coordination procedures during loss of During operation, energy storage could serve as generation
load events are as follows: providing power to the load and could be charged as a load. Cur-
1) Identify the loss of load event. rent period energy storage state of charge level is determined by
2) Obtain the updated available energy storage SOC in the the previous operations. Because of these unique chronological
system. Only the energy storage operated in the grid characteristics of energy storage, its impact on system reliability
connected mode could be utilized by the system operator. is best captured using sequential Monte Carlo method, in which
During loss of load event, the operations required by the its specific operation strategies are integrated. The assessment
system operator have higher priority over the operations flowchart is shown in Fig. 3. Details of the assessment frame-
required by the load aggregator. During other periods, load work are presented as follows.
aggregator fully controls their energy storage operation. The bulk power system and the distribution system integrated
3) Assess the available energy and power from energy storage with energy storage and renewable energy generation are simu-
for a period (e.g. 5 minutes or one hour). Due to the limited lated in parallel. The state of each bulk power system component
stored energy, energy storage could only provide certain is simulated first. With the information of the system topology
power output for a limited period of time. and the simulated distribution system components state, the op-
4) Treat the energy storage as extra power source with the eration mode of the distribution systems integrated with the en-
assessed power from step 3. ergy storage and renewable energy generation is determined.
5) Run power flow analysis to determine the optimal dispatch Then the energy storage operation strategies and the distribu-
with energy storage in order to avoid the loss of load event tion and transmission coordination strategies are implemented
1134 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

• Under islanding mode, the distribution systems are dis-


connected from the external grid. The load in the distri-
bution systems is supported through the integrated dis-
tributed generation and energy storage. In order for the
power system to be in success state, there must be suffi-
cient power supply integrated in the disconnected distribu-
tion systems to cover their load, and the power flow anal-
ysis must confirm that the load in the external grid is also
covered. In summary, both the disconnected distribution
systems and the external grid must be in success state to
make the bulk power system in success state. Otherwise,
the bulk power system is in failure state.
• Under Coordination Mode, the bulk power system requires
extra energy from the distribution systems integrated with
energy storage devices in order to avoid the loss of load
event. If there is sufficient energy from the energy storage
devices to cover the potential loss of load in the external
grid, the system is in success state. If there is not enough
energy from energy storage to support the bulk power
system due to the limited stored energy or transmission
congestions, the bulk power system is in failure state.
• Under failure mode, there is a failure inside the distribution
system. Power from either the external grid or the energy
storage cannot cover the distribution systems load due to
either the limited energy or the distribution systems net-
work disconnection. For example, in the radial distribution
Fig. 3. Proposed reliability assessment framework flowchart. systems without distributed generation and energy storage,
the failure or a power line can cause loss of load in the
downstream parts of the distribution system.
during the simulation process. According to the determined op- Once the bulk power system state is determined, it is recorded
eration strategy, the energy storage SOC and the imported power for this period and later used to calculate LOLP and EENS.
from the external grid for this distribution system are obtained. With this reliability evaluation framework, the reliability im-
The SOC could be used by the system operator if there is a loss pact by the energy storage could be quantified. The reliability
of load event in the bulk power system. From the system op- impact brought by the energy storage could be attributed to two
erator’s point of view, the imported power for this distribution main factors. The first one is that in the loss of load event, en-
system is the load at this bus. With the load at each bus known ergy storage could provide extra energy to the system to avoid
and the knowledge of the bulk power system components state the event or reduce the unserved energy. The other one is that
and topology, power flow analysis is performed to determine the because of the correlation between the energy price and load
bulk power system state. The system is in success state if all the demand, the presented operation strategy controls the energy
loads are met; the system is in failure state if loss of load event storage in a fashion that energy storage is charged during non-
occurs. Within the analysis, both the generation and transmis- peak and low energy price periods while discharged during peak
sion lines availability are considered. Transmission line conges- and high energy price periods. This characteristic indicates that
tions could also be included in the power flow analysis. If the the distribution system’s load is adjusted to a lower level during
system is in failure state, energy storage coordination strategy peak period when loss of load events are more likely to occur.
is applied to avoid or minimize the loss of load. During the sim- Thus the total system’s peak load is mitigated. The probability of
ulation period, the system state and energy storage SOC are up- inadequate generation during the peak load periods is reduced.
dated and recorded for further simulation. The bulk power system reliability is improved.
From the distribution system point of view, the system state
determination for reliability analysis is as follows:
V. CASE STUDIES
• Under grid connected mode, the distribution systems with
energy storage devices are connected to the external grid. A detailed case study on IEEE Reliability Test System (24
The bulk power system power flow analysis is conducted buses) [21] is performed to demonstrate the effectiveness of the
to determine whether the entire load is covered by the gen- proposed power system reliability evaluation framework and
eration and energy storage. If the total power supply is not the benefits of the presented intelligent energy storage opera-
sufficient for the entire load, or the transmission conges- tion strategy. The sensitivity analysis provides more insights on
tions cause loss of load in any part of the system, the system how the system reliability is affected through the integration of
is identified as in failure state. Otherwise, it is in success energy storage considering the existence of integrated renew-
state. able energy resources.
XU AND SINGH: POWER SYSTEM RELIABILITY IMPACT OF ENERGY STORAGE INTEGRATION 1135

TABLE I TABLE II
BASE CASES RELIABILITY INDICES LOLP WITH 105% LOAD SCALE

In the case study, the most heavily loaded bus, bus 18, is in-
tegrated with energy storage and distributed wind turbine gen-
erations (WTG). The energy storage devices are operated with
the presented strategy and are assumed to be perfectly reliable. TABLE III
Wind power output data from [2] is utilized. The failures of the EENS WITH 105% LOAD SCALE
generation and transmission lines are modeled according to the
IEEE Reliability Test System Reliability Data. The transmis-
sion line capacities are considered. The case study horizon is
set as one year, and the one year horizon is simulated for 100
times. The average annual indices are calculated for evalua-
tion. The system operator utilizes the stored energy from energy
storage to avoid loss of load event or minimize unserved energy.
DC Optimal Power Flow is performed using MatPower [22]
to determine the bulk power system state and dispatch energy
storage discharge. The power system operational constraints,
such as generation ramp rate limit could be integrated into the
power flow analysis. The IEEE Reliability Test System load
profile is used. The original annual peak load of the system is
2850 MW. Considering the original system is relatively reliable
and the annually increasing load is making the system less re-
liable, the studied bulk power system’s load is scaled so that 8.0% . With the integration
the annual peak load is 2992.5 MW (1.05 2850 MW) and of both energy storage and renewable energy generation, the
3243 MW (1.15 2850 MW) respectively in the case studies. system reliability improvement is even more significant.
The studied energy storage capacities range from 0 MWh to 80 In Table II and Table III, it can be noticed that when the energy
MWh, and the energy storage power ratings range from 0 MW storage has power rating from 20 MW to 60 MW, the expansion
to 80 MW. Base cases without energy storage or wind turbine of WTG from 2 MW to 50 MW does not improve LOLP. That
generation are first studied. Commonly used reliability indices, is because in those loss of load events, the added WTG is too
Loss Of Load Probability (LOLP) and Expected Energy Not small to cover the generation inadequacy. However, the added
Served (EENS) are used as the reliability indicators for the bulk WTG reduces EENS from 2.609 GWh to 2.495 GWh. It is still
power system. The results of the base cases are shown in Table I. improving bulk system reliability and reducing customer inter-
ruption cost.
Different sizes of energy storage and wind turbines capacities Table IV and Table V presents the results for the system with
are matched and integrated in the distribution system to form a annual peak load at 3243 MW (1.15 2850 MW). With a much
case. The reliability results for bulk power system with peak larger load demand, the system has the LOLP improved from
load at 2992.5 MW (1.05 2850 MW) are shown in Table II 1.0073% to 0.7784%, and the EENS improved from 14.39 GWh
and Table III. to 13.19 GWh. Compared with the results for the 105% load
The case study results in Table I, Table II and Table III scale system, the same expansion of energy storage and WTG
demonstrate the reliability improvement benefits for the bulk has a more effective impact on the reliability improvement for
power system. The reliability improvement through the energy this less reliable system.
storage integration and proposed operations is very effective. As indicated in the results of case studies, the integration of
For example, in Table I, for the system with peak load at energy storage within the distribution system could contribute
2992.5 MW (1.05 2850 MW), the integration of an en- to the bulk system reliability improvement, especially when the
ergy storage device alone, which has a 20 MW power rating system is heavily loaded and becomes less reliable. However,
(merely 0.67% of the 2992.5 MW peak load), and 40 MWh as the energy storage is the resource owned and operated by the
capacity, could improve the bulk power system’s EENS for local distribution system load aggregator, system operator who
1136 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

TABLE IV A reliability evaluation framework is proposed to quantify the


LOLP WITH 115% LOAD SCALE energy storage reliability impact on the bulk power system.
Considering the potential large scale deployment of energy
storage devices and renewable energy generation, the quantified
study on the power system reliability impact is crucial for the
system operator to determine the value of the energy storage
integration. This work provides valuable insights on how the
energy storage could effectively and positively impact the bulk
system reliability. These insights could be beneficial for further
studies on the planning of energy storage and renewable energy
integration and expansion, cost/worthy analysis, energy storage
revenues streams evaluation, and economic feasibility analysis,
etc. They could also serve to help develop more appropriate
business models for energy storage deployment.

REFERENCES
[1] United States Department of Energy, 2010 Smart Grid System Report
TABLE V Washington, DC, February 2012, p. 20585.
EENS WITH 115% LOAD SCALE [2] North American Electric Reliability Corporation, Accommodating
High Levels of Variable Generation Apr. 2009 [Online]. Available:
http://www.nerc.com/files/IVGTF_Report_041609.pdf
[3] B. Bagen and R. Billinton, “Reliability cost/worth associated with wind
energy and energy storage utilization in electric power systems,” in
Proc. PMAPS, 2008, pp. 1–7.
[4] R. Karki, H. Po, and R. Billinton, “Reliability evaluation considering
wind and hydro power coordination,” IEEE Trans. Power Syst., vol.
25, no. 2, pp. 685–693, 2010.
[5] M. Korpas and A. T. Holen, “Operation planning of hydrogen storage
connected to wind power operating in a power market,” IEEE Trans.
Energy Conversion, vol. 21, no. 3, pp. 742–749, 2006.
[6] P. Hu, R. Karki, and R. Billinton, “Reliability evaluation of gener-
ating systems containing wind power and energy storage,” Generation,
Transmission & Distribution, IET, vol. 3, no. 8, pp. 783–791, Aug.
2009.
[7] Z. Y. Gao, P. Wang, L. Bertling, and J. H. Wang, “Sizing of energy
storage for power systems with wind farms based on reliability cost
and wroth analysis,” in Proc. Power Energy Society General Meeting,
2011 IEEE, Jul. 24–29, 2011, pp. 1–7.
[8] Z. Y. Gao, P. Wang, and J. Wang, “Impacts of energy storage on reli-
ability of power systems with WTGs,” in Proc. PMAPS, June 14–17,
benefits from it should compensate the load aggregator for the 2010, pp. 65–70.
reliability improvement benefits it received. The quantified re- [9] H. Asgeirsson, “Status of the community energy storage deployment
liability improvement through the proposed evaluation frame- at DTE energy,” in Proc. Power and Energy Society General Meeting,
2011 IEEE, July 24–29, 2011.
work provides the basis for determining the proper compensa- [10] T. Senjyu, P. Mandal, K. Uezato, and T. Funabashi, “Next day load
tion. If the reliability improvement value needs to be converted curve forecasting using hybrid correction method,” IEEE Trans. Power
to economic terms in order to determine its value, a cost/ben- Syst., vol. 20, no. 1, pp. 102–109, 2005.
[11] J. W. Taylor and R. Buizza, “Neural network load forecasting with
efit analysis based on the customer interruption cost method weather ensemble predictions,” IEEE Trans. Power Syst., vol. 17, no.
could be utilized. Our previous work [23] proposed an economic 3, pp. 626–632, 2002.
evaluation method for reliability improvement based on the cus- [12] R. J. Bessa, V. Miranda, and J. Gama, “Entropy and correntropy
against minimum square error in offline and online three-day ahead
tomer interruption cost method. wind power forecasting,” IEEE Trans. Power Syst., vol. 24, no. 4, pp.
1657–1666, 2009.
[13] G. Sideratos and N. D. Hatziargyriou, “An advanced statistical method
VI. CONCLUSION
for wind power forecasting,” IEEE Trans. Power Syst., vol. 22, no. 1,
The smart grid communication and control infrastructure pp. 258–265, 2007.
[14] N. Amjady and F. Keynia, “Day-ahead price forecasting of electricity
could be utilized to better manage energy storage within the markets by mutual information technique and cascaded neuro-evo-
distribution system. This paper has presented an operation lutionary algorithm,” IEEE Trans. Power Syst., vol. 24, no. 1, pp.
strategy for a distribution load aggregator to utilize its energy 306–318, 2009.
[15] P. Mandal, T. Senjyu, N. Urasaki, T. Funabashi, and A. K. Srivastava,
storage to better manage its energy procurement and distributed “A novel approach to forecast electricity price for PJM using neural
renewable energy generation. The communication and control network and similar days method,” IEEE Trans. Power Syst., vol. 22,
structure for implementing this operation strategy is also pre- no. 4, pp. 2058–2065, 2007.
[16] C. Jie, D. Shi-Jie, and H. Xiaoming, “Electricity price curve modeling
sented. The communication network also allowed the energy and forecasting by manifold learning,” IEEE Trans. Power Syst., vol.
storage coordination during loss of load event in the bulk 23, no. 3, pp. 877–888, 2008.
power system. With the proposed energy storage coordination, [17] L. Xie and M. D. Ilic, “Model predictive economic/environmental
dispatch of power systems with intermittent resources,” in Proc. Proc.
energy storage integrated in the distribution system could also IEEE Power & Energy Soc. 2009 General Meeting, Calgary, AB,
contribute to the bulk power system reliability improvement. Canada, 2009.
XU AND SINGH: POWER SYSTEM RELIABILITY IMPACT OF ENERGY STORAGE INTEGRATION 1137

[18] T. Senjyu, R. Sakamoto, N. Urasaki, T. Funabashi, H. Fujita, and H. Chanan Singh (S’71–M’72–SM’79–F’91) is cur-
Sekine, “Output power leveling of wind turbine Generator for all oper- rently Regents Professor and Irma Runyon Chair
ating regions by pitch angle control,” IEEE Trans. Energy Conversion, Professor in the Department of Electrical and Com-
vol. 21, no. 2, pp. 467–475, Jun. 2006. puter Engineering, Texas A&M University (TAMU),
[19] Y. Xu, L. Xie, and C. Singh, “Optimal scheduling and operation of College Station, TX, USA.
load aggregator with electric energy storage facing price and demand He is also a Guest Professor at Tsinghua Univer-
uncertainties,” in Proc. NAPS, Aug. 4–6, 2011. sity. From 1995 to 1996, he served as the Director
[20] Y. Xu and C. Singh, “Distribution systems reliability and economic of Power Program at the National Science Founda-
improvement with different electric energy storage control strategies,” tion, and from 1997 to 2005, he served as the Head
in Proc. Power and Energy Society General Meeting, Jul. 24–29, 2011. of the Electrical and Computer Engineering Depart-
[21] P. M. Subcommittee, “IEEE reliability test system,” IEEE Trans. ment at TAMU. His research and consulting interests
Power Apparatus Syst., vol. PAS-98, no. 6, pp. 2047–2054, Nov. are in the application of probabilistic methods to power systems. He has au-
1979. thored/co-authored more than 300 technical papers and three books and has
[22] R. D. Zimmerman, C. E. Murillo-Sánchez, and R. J. Thomas, “MAT- contributed to several books. He has consulted with many major corporations
POWER: Steady-state operations, planning and analysis tools for and given short courses nationally and internationally.
power systems research and education,” IEEE Trans. Power Syst., vol. Dr. Singh was the recipient of the 1998 Outstanding Power Engineering
26, no. 1, pp. 12–19, Feb. 2011. Educator Award given by the IEEE Power Engineering Society. For his
[23] Y. Xu and C. Singh, “Adequacy and economy analysis of distribution research contributions, he was awarded a D.Sc. degree by the University of
systems integrated with electric energy storage and renewable energy Saskatchewan, Saskatoon, SK, Canada, in 1997. In 2008, he was recognized
resources,” IEEE Trans. Power Syst., vol. 27, no. 4, pp. 2332–2341, with the Merit Award by the PMAPS International Society. In 2010, he was the
Nov. 2012. inaugural recipient of the IEEE-PES Roy Billinton Power System reliability
Award.
Yixing Xu (S’08) received the Ph.D. degree from
Texas A&M University, College Station, TX, USA,
in 2012, and the B.E. degree from Tsinghua Univer-
sity, Beijing, China, in 2007.
His research interest includes power systems
reliability evaluation, power system optimization,
renewable energy integration, energy storage in-
tegration, smart grid and electricity market. His
industry experience includes an internship (June
2008–August 2008) at Entergy Corporation, and an
internship (June 2012–August 2012) at KEMA Inc.

Вам также может понравиться