Research on Security Assessment Index System for Operating Reserve in Large Interconnected Power Grid

doi:10.4236/epe.2013.54B151

Paper Menu >>

Journal Menu >>

Energy and Power Engineering, 2013, 5, 785-791

doi:10.4236/epe.2013.54B151 Published Online July 2013 (http://www.scirp.org/journal/epe)

Research on Security Assessment Index System for

Operating Reserve in Large Interc onnected Power Grid

Mingsong Liu1, Huadong Sun1, Jian He1, Hengxu Zhang2, Jun Yi1, Jian Zhang1

1China Electric Power Researc h I n s t itute, Beij i n g, China

2Shandong University, Jinan, China

Email: [email protected]

Received April, 2013

ABSTRACT

Optimization and placement of spinning reserve is an important issue in power system planning and operation. System-

atic way for security assessment of operating reserve needs to study. A security assessment index system for operating

reserve in large interconnected power grids is presented in th is paper. Firstly, classification and determination methods

of operating reserve at home and abroad are investigated, and operating reserve is divided into transient state operating

reserve and quasi-steady state operating reserve from the view of security assessment. Secondly, assessment indexes

and optimization methods for transient state operating reserve are studied. Thirdly, optimization model, deterministic

and probabilistic optimization methods for quasi-steady state operating reserve are explored. Finally, some principles

for determination of operating reserve are suggested, and a security assessment index system is put forward. The pro-

posed index system, considering both transient and quasi-steady state, both deterministic and probabilistic methods,

provides a systematic way to assessment and arrangement of operating reserve.

Keywords: Security Assessment Index System; Operating Reserve; Spinning Reserve; Large Interconnected Power

Grid

1. Introduction

Several major blackouts occurred in the world recent

years, causing great economic losses and severe social

influence [1-5]. Insufficient reserve capacity and im-

proper placement of operating reserve are among several

main reasons. Optimization and placement of spinning

reserve is an important issue in power system planning

and operation [6-10]. In China, power load grows rapidly

with the development of economy, while power supply is

far behind demand. Spinning reserve is often insufficient

in period of peak load in many provinces, which is a

great threat to security and stability of power system op-

eration. How to arrange spinning reserve reasonably and

effectively in a large interconnected power grid becomes

a technical problem urgent to be solved.

The arrangement of spinning reserve in China is

mainly according to a certain proportion of overall gen-

eration load [11], which is relatively rough. As regional

interconnected power grids, intermittent renewable en-

ergy and power electronic devices increase, system

characteristics and operation modes of power grids be-

come more and more complicated. Higher requirements

have been put forward to spinning reserve.

Much work has been done on optimization and place-

ment of operating reserve, which is very helpful to power

system planning and operation. However, systematic way

for security assessment of operating reserve needs to

study. A security assessment index system for operating

reserve in large interconnected power grids is presented

in this paper. Firstly, classification and determination

methods of operating reserve at home and abroad are

investigated, and operating reserve is divided into tran-

sient state operating reserve and quasi-steady state oper-

ating reserve from the view of security assessment. Sec-

ondly, assessment indexes for transient state operating

reserve are studied, and optimization methods are intro-

duced. Thirdly, optimization model, deterministic and

probabilistic optimization methods for quasi-steady state

operating reserve are explored. Finally, some principles

for determination of operating reserve are suggested , and

a security assessment index system is put forward. The

proposed index system, considering transient and quasi-

steady state, deterministic and probabilistic methods,

provide a systematic way to assessment and arrangement

of operating reserve.

2. Classification and Determination Methods

of Operating Reserve

2.1. Classification and Determination of

Operating Reserve in China

M. S. LIU ET AL.

786

 Classification of operating reserve

In China, according to its function, operating reserve is

divided into load reserve, contingency reserve and main-

tenance reserve [11]. Load reserve is used to balance

instantaneous load fluctuation and load forecasting errors,

which should be spinning reserve. Contingency reserve,

activated in a certain period after disturbance, is used to

compensate large active power imbalance caused by con-

tingencies, part of which should be put into operation

automatically when system frequency decreases. Main-

tenance reserve meets the requirements of maintaining all

generato rs per iodicall y.

Spinning reserve usually consists of the following

generators:

1) Generators that are already put into operation,

without reaching their operating limits, including thermal

power generators, hy dro power gen erators, et c .

2) Hydro power generators that can be put into opera-

tion quickly, usually within several minutes.

 Determination of operating reserve

In China, operating reserve is arranged according to a

certain proportion of maximum generation load [11].

Load reserve is 2%-5% of maximum generation load.

Lower proportion is suitable for large systems, while

higher proportion is for small systems. Contingency re-

serve is about 10% of maximum generation load, which

should be greater than the largest generator of a system.

Maintenance reserve is about 8%-10% of maximum gen-

eration load.

2.2. Classification and Determination of

Operating Reserve in America

 Classification of operating reserve

In America, according to its function, operating re-

serve is divided into regulating reserve, contingency re-

serve, additional reserve for interruptible imports, and

additional reserve for on-demand obligations [12].

Regulating reserve is similar to load reserve in China,

and the same is contingency reserve. Additional reserve

for interruptible imports and on-demand obligations

should be made effective within ten minutes, which meet

interruptible imports and on-demand obligations to other

entities or balancing authorities.

Operating reserve consists of spinning reserve and

nonspinning reserve. Spinning reserve means unloaded

generation which is synchronized and ready to serve ad-

ditional demand. Nonspinning reserve means that oper-

ating reserve not connected to the system but capable of

serving demand within a specified time, or interruptible

load that can be removed from the system in a specified

time.

 Determination of operating reserve

In America, operating reserve is arranged in a deter-

ministic way [13]. Regulating reserve is to provide suffi-

cient regulating margin to meet NERC's control per-

formance criteria, while contingency reserve is to meet

NERC disturbance control standard, considering the most

severe single contingency and a certain proportion of

hydro, thermal generation load. The combined unit ramp

rate of each balancing authority's on-line, unloaded gen-

erating capacity must be capable of responding within

ten minutes.

2.3. Classification and Determination of

Operating Reserve in Europe

 Classification of operating reserve

In Europe, according to its function, operating reserve

is divided into primary control reserve, secondary control

reserve, and tertiary control reserve, as shown in Figure

1 [13]. Primary control reserve is activated within sec-

onds, while secondary control reserve is activated within

minutes. Tertiary control reserve implies changes in gen-

eration or load on a contractual, market or regulatory

basis, activated for a period of time, e.g. 15 minutes.

 Determination of operating reserve

In Europe, operating reserve is determined as follows

[13]. Primary control reserve, proportionally distributed

among each control area, is agreed to be 3000 MW,

which is the maximum instantaneous power deviation for

the UCTE synchronous area, based on operational char-

acteristics concerning system reliability and size of load s

and generation units. Secondary control reserve must be

available to cover expected demand and generation fluc-

tuations. A total tertiary control reserve must be available

to cover the largest expected loss of power in the control

area.

Figure 1. Principle frequency deviation and subsequent activation of reserves.

M. S. LIU ET AL. 787

2.4. Classification of Operating Reserve

According to Response Rate

There are some other kinds of classification of operating

reserve. In [14], reserve is divided into 10-minute spin-

ning reserve, 10-minute nonspinning reserve, 30-minute

reserve, 60-minute reserve and cold reserve. In [15], re-

serve is divided into instantaneous reserve, fast reserve,

slow reserve and cold reserve. It is obvious that response

rate is the fundamental characteristics of operating re-

serve. Therefore in power market, it is prone to divide

operating reserve into instantaneous reserve, 10-minute

spinning reserve, 10-minute nonspinning reserve, 30-

minute reserve, 60-minute reserve and cold reserve, as

shown in Table 1.

2.5. Classification of Operating Reserve from the

View of Security Assessment

Instantaneous reserve takes effect during transient state

process immediately after a contingency, which affects

system frequency deviation much. While other kinds of

operating reserve take effect during quasi-steady state

process, usually at least several minutes after a distur-

bance, which mai nl y affect power flow regu l a t ion.

From the view of security assessment for power sys-

tem operation, operating reserve can be divided into

transient state operating reserve and quasi-steady state

operating reserve. Transient state operating reserve con-

sists of instantaneous reserve, while quasi-steady state

operating reserve consists of other kinds of operating

reserve.

3. Transient State Operating Reserve

3.1. System Frequency Response after Loss of

Generation

System frequency response after loss of generation is

introduced in [16]. When a large generator trips, the dis-

turbance spreads to each generator immediately. Due to

dead zone and delay block, generator governors do not

Table 1. Classification of reserve according to response

rate.

Reserve Response Time Synchronous

instantaneous reserve several seconds Yes

10-minu te spinning reserve <10min Yes

10-minut e nonspinning reserve <10min No

30-minute reserve (10min, 30min) No

60-minute reserve (30min, 60min) No

cold reserve >60min No

take effect at once. System frequency decreases sharply.

Several seconds later, governors start regulating, and

active power of each generator is adjusted according to

its moment of inertia and system frequency deviation.

Ten or more seconds later, primary frequency control is

over, but system frequency does not recover to normal

value. Several minutes later, AGC takes effect, restoring

system frequency to normal value.

3.2. Assessment Index for Transient State

Operating Reserve

Dynamic characteristics of system frequency and assess-

ment index for system frequency and spinning reserve

have been studied [17-23]. There are mainly three kinds

of index shown as follows.

 Transient frequency deviation index

Overall spinning reserve capacity affects system

frequency response after loss of active power. On the

other hand, transient frequency deviation reflects overall

spinning reserve capacity. If overall spinning reserve

capacity is small, transient frequency deviation becomes

large. If overall spinning reserve capacity is large,

transient fr equency deviati on becom e s small.

There are several indexes for transient frequency

deviation shown as follows.

1) Maximum transient frequency deviation. This index,

intuitive and easy, is widely adopted in power system

analysis.

2) Transient frequency deviation index using two-

element table [18]. A set of two-element tables, made up

of transient frequency deviation and its duration, is used

to describe transien t frequ ency dev iatio n accep tab ility for

individual bus or gener ator.

3) Transient frequency deviation index considering

cumulative effect [19]. The index, defined by integration

of frequency deviation and its duration, is used to

quantitatively assess transient frequency deviation

security by taking into account the frequency deviation

cumulative effect．

 Transient frequency time-space distribution index

After a disturbance causing large active power

imbalance, transient frequency presents time-space

distribution features. Since network structure, load level,

generation capacity, spinning reserve capacity, dynamic

model parameters are different among areas, transient

frequency deviation may be different at different

locations.

Some transient frequency time-space distribution

indexes are used in [21], shown as follows.

1) Mean changing rate of frequency. Shortly after a

disturbance, the frequency response curve is usually

linear. The index reflects type and severity of the

disturbance and information of network and generators.

M. S. LIU ET AL.

788

2) Maximum transient frequency deviation at different

locations. The index is used to assess influence degree on

different locations by the disturbance.

3) Transient frequency response delay at different

locations. This index reflects frequency spreading

process.

 Response rate of generators in primary frequency

control

After a large disturbance of active power, different

generators may response in different rates. Generators of

different types, years and capacity have different re-

sponse rate. Generally speaking, thermal generators of

new type, with large capacity response fast. The faster

the generators with reserve capacity response, the less the

maximum transient frequency deviation is, and the

shorter the duration to maximum deviation is. Response

rates of generators in primary frequency control are re-

lated to difference coefficients and dead blocks of their

governors [17, 24].

3.3. Optimization of Transient State Operating

Reserve

In [22], an optimization method for spinning reserve,

considering transient frequency security, is proposed. The

algorithm is separated into two steps. Critical spinning

reserve capacity is calculated based on sensitivity

analysis. Then critical spinning reserve capacity is put

into optimization model as a constraint, and th e model is

solved using priority list method.

In [23], an optimization meth od in which the transient

frequency deviation of power system is taken into

account is proposed. The maximum active power of each

generation unit that is allowed to undertake during

primary frequency regulation is calculated. Then both

spinning reserve cost and moment of inertia of each

generation are comprehensively considered to give the

optimization configuration scheme of spinning reserve.

4. Quasi-Steady State Operating Reserve

4.1. Optimization Model of Quasi-steady State

Operating Reserve

When quasi-steady state operating reserve takes effect, a

system is usually in a quasi-steady state. Transient state

process can be ignored, and it becomes a problem of

power flow. Optimization of quasi-steady state operating

reserve is a problem of optimal power flow with several

constraints, as shown in (1).



min

XB

(1)

The objective function





X represents network

loss, generation cost, reserve procurement cost, social

cost, or some other kinds of cost. The objective is to

minimize give n ki n d of cost.

The equality constraint represents power

flow equations.



=0GX

The inequality constraint

XB represents a set of

constraints, shown as follows.

1) Constraints of generator operation limits.

2) Constraints of generator reserve capacity.

3) Constraints of generator ramp rates.

4) Constraints of line curren t limits.

5) Constrains of section transfer capab ility.

6) Constrains of overall reserve capacity.

7) Constrains of area reserve capacity limits.

8) Constrains of system stability.

These inequality constraints can be used as security

assessment indexes for quasi-state operating reserve.

4.2. Deterministic Optimization Methods

In traditional power industry, generation, transmission

and distribution belong to one power company. Reserve

capacity is provided and used by the same company.

Therefore economic affairs are easy to be coordinated

within the compan y. Power supply reliability and system

operation security are the most important issues. Spinning

reserve capacity is usually determined according to a

certain proportion of maximum generation load, or the

largest generator of a system. These methods are easy

and widely used. However, they are not economically

optimal.

Some optimization methods are studied to minimize

reserve capacity, reserve procurement cost, etc. In [25],

optimal reserve model of interconnected regional power

systems is built up. The adaptive immune genetic algorithm

is used to calculate the optimal reserve capacity, taking

into account offset frequency, load fluctuation and security

reliability constraints of the regional inter- connected

power system. In [26], two approaches are adopted, the

independent modeling of multi-step optimization and

unified modeling of joint optimization for daily spinning

reserve and generation scheduling. Different energy-

saving generation dispatching models are established.

4.3. Probabilistic Optimization Methods

Power supply reliability is a most important issue for

both traditional power industry and power markets.

Security and adequacy are two aspects of reliability.

Probabilistic indexes, such as loss of load probability and

value of lost load, are often used to evaluate system reli-

ability.

Much has been done on probabilistic optimization

methods of spinning reserve. In [27], a stochastic variable

called spinning reserve's gain or loss is defined, to

M. S. LIU ET AL. 789

analyze the risks of spinning reserve scheme from the

prospective of power generating system. A utility

expectation decision-making model is proposed, reflect-

ing both the objective risk of a spinning reserve scheme

and the decision-maker's attitude to risks. In [28], a

method and standards of shareable operational reserve

reliability evaluation for multi-area power networks are

proposed. Such random factors as the constraints of tie-

lines, the status of on-line generations, loss of network

elements, load fluctuations are cons idered. The reliability

of each area and the whole power system are evaluated

separately. In [29], an extended state-space partitioning

based method to assess operating reserve of power grid

connected with large-scale wind farm is proposed.

Multi-state model of wind turbine generator is adopted,

and the total state space, divided into two subspaces, can

be chosen by state enumeration method and Monte Carlo

simulation respectively. In [30], a mathematical model

for deploying distributed generation as reserve is built u p.

Loss of load prob ability is used to evaluate the reliab ility

of power supply, and different optimal deployment

scheme for distributed generators are given under

different reliability targets.

5. Security Assessment Index System for

Operating Reserve

5.1. Some Principles for Operating Reserve

Requirements of operating reserve capacity are closely

related to system scale, network structure, characteristics

of load and generation, etc. It is better to make system-

atic assessment for a power grid before arranging its re-

serve capacity, instead of using a certain proportion roughly.

When a DC line with large active power is connected

to a power grid, severe impacts on the grid may be

caused if an outage occurs in one pole of the DC line.

Spinning reserve capacity needs improving, according to

frequency response characteristics after the outage.

When large amount of wind power is connected to a

power grid, wind power, with characteristics of fluctua-

tion and intermittence, requires even higher spinning

reserve capacity. Spinning reserve capacity needs im-

proving, and the spinning reserve for wind power should

be close to wind power.

Transient frequency deviation is close related to active

power disturbance. For a power grid, the active power

disturbance should be maximum active power imbalance

considering N-1 contingencies. However, if control

measures are designed for a contingency, it should be

analyzed separately.

5.2. Recommended Security Assessment Index

System

A recommended security assessment index system for

operating reserve in large interconnected power grid is

shown in Table 2.

For transient state operating reserve, deterministic as-

sessment methods are mainly adopted. Transient fre-

quency deviation indexes, such as maximum transient

frequency deviation, index using two-element table and

index considering cumulative effect, are used to assess

overall spinning reserve capacity. Transient frequency

time-space distribution indexes, such as mean changing

rate of frequency, maximum transient frequency devia-

tion and transient frequency response delay at different

locations, are used to assess spinning reserve distribution

characteristics. Response rate of generators in primary

frequency control is used to assess spinning reserve re-

sponse.

For quasi-steady state operating reserve, both determi-

nistic and probabilistic methods are adopted. Indexes

such as available transfer capability and generator ramp

rates are used to assess security of operating reserve.

Indexes such as loss of load probab ility and value of lost

load are used to assess operating risk considering operat-

ing reserve.

6. Conclusions

Optimization and placement of spinning reserve is an

important issue in power system planning and operation.

Classification and determination methods of operating

reserve at home and abroad are investigated at first, and a

new classification of operating reserve from the view of

security assessment is presented. Then assessment in-

dexes and optimization methods for both transient state

and quasi-steady state operating reserve are studied. At

last a security assessment index system for operating

reserve in large interconnected power grids is put for-

ward. The proposed index system takes into account

Table 2. Recommended security assessment index system.

Reserve TypeAssessment

Methods Assessment

Object Assessment

Indexes

Capacity Transient frequency

deviation index

Distribution Transient frequency

time-space di s tribution

index

Transient

state

operating

reserve

Deterministic

Response Response rate of

generators in primary

frequency control

DeterministicSecurity Available transfer

capability, generator

ramp rates, etc.

Quasi-steady

state

operating

reserve Probabilistic Operating

risk loss of load probability,

value of lost load, etc.

M. S. LIU ET AL.

790

transient and quasi-steady state, deterministic and prob-

abilistic methods, and provides a systematic way to as-

sessment and arrangement of operating reserve.

REFERENCES

[1] Y. Xue and S. Xiao, “Comprehensively Defending High

Risk Events with Low Probability,” Automation of

Electric Power Systems, Vol. 35, No. 8, 2011, pp. 1-11.

[2] G. Andersson, P. Donalek, R. Farmer, et al., “Causes of

the 2003 Major Grid Blackouts in North America and

Europe, and Recommended Means to Improve System

Dynamic Performance,” IEEE Transactions on Power

Systems, Vol. 20, No.4, 2005, pp. 1922-1928.

doi:10.1109/TPWRS.2005.857942

[3] Y. Tang, G. Bu and J. Yi, “Analysis and Lessons of the

Blackout in Idian Power Grid on July 30 and 31, 2012,”

Proceedings of the Chinese Society for Electrical Engi-

neering, Vol. 32, No. 25, 2012, pp. 167-174.

[4] W. Lin, Y. Tang, H. Sun, et al., “Blackout in Brazil

Power Grid on February 4, 2011 and Inspirations for

Stable Operation of Power Grid,” Automation of Elec-

tric Power Systems, Vol. 35, No. 9, 2011, pp. 1-5.

[5] Y. Fang, “Lessons from September 8, 2011 Southwest

America Blackout for Prevention and Control of Cas-

cading Outages,” Automation of Electric Power Systems,

Vol. 36, No. 15, 2012, pp. 1-7.

[6] X. Shu, Z. Ming, L. Gengyin, et al., “On Spinning Re-

serve Determination and Power Generation Dispatch

Optimization for Wind Power Integration Systems,” in

Power and Energy Society General Meeting, 2012, pp.

1-7.

[7] M. Q. Wang, H. B. Gooi and S. X. Chen, “Optimising

Probabilistic Spinning Reserve using an Analytical Ex-

pected-energy-not-supplied Formulation,” Generation,

Transmission & Distribution, IET, Vol. 5, No.7, 2011,

pp. 772-780. doi:10.1049/iet-gtd.2010.0805

[8] A. Hoballah and I. Erlich, “Dynamic Stability and Net-

Work Constrained Optimal Spinning Reserve Alloca-

tion,” in Power and Energy Society General Meeting,

2011, pp. 1-5.

[9] C. Wang, Y. Qiao and Z. Lu, “A Method for Determina-

tion of Spinning Reserve in Wind-thermal Power Sys-

tems Considering Wind Power Benefits,” Automation of

Electric Power Systems, Vol. 36, No. 4, 2012, pp. 16-21.

[10] G. Zhao, T. Liu and X. Li, “Optimal Deployment of

Distributed Generation as Backup Generators,” Automa-

tion of Electric Power Systems, Vol. 33, No. 1, 2009,pp.

85-89.

[11] “Guide on Security and Stability for Power System,” DL

755-2001, Beijing: China Electric Power Press, 2001.

[12] NERC, “Operating Reserves,” WECC Standard

BAL-STD-002-0, 2007.

http://www.nerc.com/page.php?cid=2|20

[13] ENTSO-E, “Load-Frequency Control and Performance

[C],” UCTE OH – Policy 1, 2009.

https://www.entsoe.eu/publications/system-operations-r

eports/operation-handbook/

[14] J. Wang, “Study on Reserve Capacity in Power Market,”

Ph.D. Thesis, Xian Jiaotong University, Xian, 2003.

[15] Y. Lai, Y. Xue, D. Wang, et a1., “Risk Deci-

sion-marking for Reserve Capacity Market,” Automation

of Electric Power Systems, Vol. 30, No. 16, 2006, pp.

1-5.

[16] B. Cai, “Power System Frequency,” Beijing: China

Electric Power Press, 1998.

[17] H. Zhou, L. Ni and T. Xu, “Study on Power-Frequency

Dynamic Characteristic of Power Grid,” Power System

Technology, Vol. 33, No. 16, 2009, pp. 58-62.

[18] T. Xu and Y. Xue, “Quantitative Assessments of Tran-

sient Frequency Deviation Acceptability,” Automation

of Electric Power Systems, Vol. 26, No. 19, 2002, pp.

7-10.

[19] H. Zhang, Y. Liu and Y. Xue, “Quantitative Assessment

of Transient Frequency Deviation Security Considering

Cumulative Effect,” Automation of Electric Power Sys-

tems, Vol. 34, No.24, 2010, pp. 24.

[20] H. Zhang, and Y. Liu, “Disturbance Identification Con-

sidering Differences in Frequency Deviation Propaga-

tion Speed,” Automation of Electric Power Systems, Vol.

34, No. 22, 2010, pp. 62-66.

[21] Z. Hou, “Dynamic Characteristic Analysis of Power

System Frequency,” Master Thesis, Shandong Univer-

sity, Jinan, 2012.

[22] M. Liu, J. He, H. Zhang, et al., “Study on Optimization

and Implementation of Spinning Reserve in Large In-

terconnecting power grid,” China Electric Power Re-

search Ins titute, 2011.

[23] M. Liu, H. Sun and J. He, “An Optimization Method for

Spinning Reserve in Primary Frequency Control Con-

sidering Transient Frequency Deviation,” Power System

Technology, Vol. 35, No. 8, 2011, pp. 129-133.

[24] Q. Zhao, L. Wang, Z. Liu, et al., “Study on Dynamic

Frequency Characteristics and Coordinative Un-

der-Frequency Load Shedding Scheme for Nationwide

Interconnected Power Grid of China,” Power System

Technology, Vol. 33, No. 8, 2009, pp. 35-40.

[25] J. Wu, M. Shi, G. Chen, et al., "Immune Genetic Algo-

rithms for Modeling Optimal Reserve Capacity of In-

terconnected Regional Power Systems,” Proceedings of

the CSEE, Vol. 29, No. 1, pp. 14-20, 2009.

[26] G. Xie, L. Zhang and J. Ge, “Optimization of Spinning

Reserve Scheduling of Energy-saving Generation Dis-

patching,” Automation of Electric Power Systems, Vol.

33, No. 13, 2009, pp. 43-46.

[27] X. Qi and M. Ding, “Risk Analysis and Utility Deci-

sion-making of Spinning Reserve Scheme in Power

Generating System,” Automation of Electric Power Sys-

tems, Vol. 32, No. 3, 2008, pp. 9-13.

[28] Y. Jia, Z. Yan, Y. Sun, et al., “Shareable Operational

Reserve Reliability Evaluation for Multi-area Power

Networks,” Automation of Electric Power Systems, Vol.

33, No. 16, 2009, pp. 13-18.

[29] J. He, H. Sun and M. Liu, "Extended State-Space Parti-

M. S. LIU ET AL.

791

tioning Based Operating Reserve Risk Assessment for

Power Grid Connected With Wind Farms,” Vol. 36, No.

3, 2012, pp. 257-263.

[30] G. Zhao, T. Liu and X. Li, “Optimal Deployment of

Distributed Generation as Backup Generators,” Automa-

tion of Electric Power Systems, Vol. 33, No. 1, 2009, pp.

85-89.