IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO.

1, MARCH 2011 181

Distributed Generation Interconnection Planning: A

Wind Power Case Study
Sheng-Yi Su, Chan-Nan Lu, Fellow, IEEE, Rung-Fang Chang, and Guillermo Gutiérrez-Alcaraz, Member, IEEE

Abstract—There is increasing requests for noncontrollable varying mixtures of the commercial, industrial, and residential
distribution generation (DG) interconnections in the medium end-users who impose time-varying load demands. Tie switches
and low voltage networks. Many studies have suggested that between feeders are used to conduct the load transfers to reduce
with proper system planning, DG could provide benefits such
as reliability enhancement, investment deferment, and reduced outage time and system losses. To further enhance the system
losses. However, without network reinforcements, the allowable operation efficiency and reliability, many utilities rely on
interconnection capacity in a network is often restricted due to two-way communications to monitor and control equipment at
fault current level, voltage variation, and power flow constraints. distribution substations and feeders. Through state estimation,
This paper aims to address the issue of optimizing network op- the measurement data from distribution automation (DA) and
eration and use for accommodating DG integrations. A new DG
interconnection planning study framework that includes a coordi- advanced metering infrastructure (AMI) are used in the distri-
nated feeder reconfiguration and voltage control to calculate the bution management system (DMS) to provide more accurate
maximum allowable DG capacity at a given node in the distribu- load estimations of feeder sections for integrated volt/var and
tion network is presented. A binary particle swarm optimization service restoration controls [1].
(BPSO) technique is employed to solve the discrete nonlinear Optimizing grid operation and use is the first priority of the
optimization problem and possible uncertainties associated with
volatile renewable DG resource and loads are incorporated European Union’s smart grid deployment plan [2]. In the fu-
through a stochastic simulation approach. Comprehensive case ture, more innovative monitoring, control, communication, and
studies are conducted to illustrate the applicability of the pro- self-healing technologies will be deployed to enhance grid oper-
posed method. Numerical examples suggest that the method and ation and facilitate customer participation. The connection and
procedure used in the current DG interconnection impact study operation of DG would allow customers to take part in opti-
should be modified in order to optimize the existing grid operation
and usage to facilitate customer participation in system operation mizing the operation and planning of the network.
and planning. Many benefits are achievable by properly integrating DG. A
set of indices, including 1) voltage profile improvement index;
Index Terms—Distributed generation, distribution management
system, distribution system planning, feeder reconfiguration, 2) line loss reduction index; 3) environmental impact reduction
particle swarm optimization, renewable energy, smart grid, index; and 4) DG benefit index, was proposed in [3] for as-
voltage/var controls. sessing the benefits of DG in a quantitative manner. Intentional
islanding operation of DG can also be performed to support net-
work reliability during an emergency. However, DG intercon-
I. INTRODUCTION nections could also cause technical problems in voltage quality,
fault level, and system protection [4]. Their impacts on distri-
ISTRIBUTION networks are divided into subsystems bution network operations would vary depending on the system
D of radial or loop feeders with a number of switches that
are normally closed or opened. Customers are supplied from
strength at the point of common coupling (PCC). Thus, utility
companies have to conduct impact study to ascertain the safety
different substations via routes which are characterized by the operations after DG interconnections.
Conventionally, distribution network volt/var controls in-
volve regulation of voltage and reactive power by on load tap
Manuscript received July 22, 2010; revised October 20, 2010 and December changer (OLTC), line voltage regulators (VR), and shunt ca-
06, 2010; accepted January 01, 2011. Date of current version February 18, 2011. pacitors (SC). Automatic voltage reference setting techniques
This work was supported in part by Taiwan Power Company under Contract are used in voltage regulating relay (VRR) or automatic voltage
TPC-546-4840-9802, and National Science Council of Taiwan under Grants
NSC 98-3114-E-110-004 and NSC98-2221-E-110-076-MY3. Paper no. TSG- relay (AVR). The relay could use local voltage level (LVL)
00097-2010. (Fig. 1) or line drop compensation (LDC) method (Fig. 2) to
S.-Y. Su is with Taiwan Semiconductor Manufacturing Company, Hsinchu
300, Taiwan.
estimate the line voltage drop of a varying load current at a
C.-N. Lu is with the Department of Electrical Engineering, National Sun voltage reference point for voltage control.
Yat-sen University, Kaohsiung 80424, Taiwan (e-mail: cnl@ee.nsysu.edu.tw). DG integrations complicate the validity of LVL and LDC
R.-F. Chang is with Kao Yuan University, Kaohsiung 82151, Taiwan (e-mail:
rfchang@cc.kyu.edu.tw).
control schemes [5]–[7]. A method for designing feasible
G. Gutiérrez-Alcaraz is with National Sun Yat-sen University, Kaohsiung LDC parameters which guarantee the satisfaction of voltage
80424, Taiwan, and also with the Department of Electrical Engineering, Instituto constraints for all possible variations in DG output was pro-
Tecnologico de Morelia, Morelia, Mexico (e-mail: ggutier@itmorelia.edu.mx). posed in [7]. Studies have reported that DG units in power
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. distribution feeders may lead to over-voltages due to excessive
Digital Object Identifier 10.1109/TSG.2011.2105895 power injections. A simple analytical method was used in [8] to
1949-3053/$26.00 © 2011 IEEE
182 IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO. 1, MARCH 2011

configurations without considering time-varying effects of

feeder loads and DG power outputs. This approach could lead
to optimistic or conservative results.
A framework that includes a coordinated feeder reconfigura-
tion and volt/var control to calculate the maximum allowable
DG capacity at a given node of the distribution system is pro-
posed in this paper. A BPSO technique is adopted to solve the
discrete nonlinear optimization problem. Possible uncertainties
associated with volatile renewable DG resource and loads are
incorporated by using a stochastic simulation approach. The or-
ganization of the paper is as follows. The problem formulation
and proposed solution procedure are described in Section II.
Section III presents the simulation results of the effect of DG
Fig. 1. Local voltage level control scheme. operations on distribution network voltage controls and possible
DG penetration increase due to coordinated control studied in
the paper. The final section concludes the paper with some re-
marks on DG impact study.

II. PROBLEM FORMULATION AND SOLUTION METHOD

When DG is introduced into the distribution network, de-
pending on the loading and power factor of the feeders, feeder
bus voltages would have different distributions. In the DG inter-
connection impact assessment, the voltage deviations due to DG
injections can be estimated by a distribution load flow (DLF)
program or using analytical formulation. Using the feeder model
shown in Fig. 3, the following simplified equations can be used
to estimate the voltage and power flows on the feeder and check
the security constraints considered in the impact study:
Fig. 2. Line drop compensation scheme.

estimate the feeder voltage profile when placing DG units with

specific active and reactive power generations. DG was mod- (1)
eled in [9] as a PV node and its control was coordinated with where
existing volt/var controls to minimize distribution losses. A
centralized control method to operate the control devices using
a communication system was described in [10]. Such studies
have suggested that active control of DG output and volt/var
regulators is desirable in order to allow for higher levels of
distributed resource interconnections. However, currently, DG
are often not controllable by utility company due to contract
and communication issues. Voltage regulation schemes which disregard DG effects may
Besides volt/var controls, automated feeder switches are result in voltage violations. In this study, according to the avail-
often dispatched by system operators under normal and emer- ability of voltage control devices, two volt/var control schemes
gency situations in order to enhance system efficiency and described in Table I and Fig. 4 are analyzed. Controllers are ad-
reliability. A simple formula was presented in [11] as a feeder justed to maintain feeder voltage within limits by using locally
planning and/or real-time control tool. Approximate power measured data. The voltage limits ( and ) are used to
flow and a loss reduction formula were developed in [12], determine whether a controller adjustment is required. The co-
[13] to aid the search for optimal feeder configuration for loss ordination of voltage controls is based on a proper timing se-
minimization. The number of noncontrollable DG intercon- quence among the controllers to reduce hunting in the voltage
nections in the medium and low voltage network is increasing. controls. It is assumed that SC has a lower adjustment delay
The impact of DG on distribution feeder reconfiguration was time, while VR and OLTC would have relatively higher delay
studied in [14], whereby the objective function was the cost times before moving to the next setting when the controller ad-
summation of electrical energy generated by DG and from justment criterion is satisfied. The control sequence and equip-
substation buses. For DG interconnection, question, such as the ment involved are shown in Table I. The simulation procedure
amount of DG capacity that is allowed at a network location, for voltage control scheme CS-2 is shown in Fig. 4. Control de-
is often raised by DG investors. Currently, network impact vices, SC, VR, and OLTC are adjusted within limits to achieve
assessments are performed by using simple screening formulas bus voltage target values of , and until the
or numerical simulations, and are often based on certain feeder allowed maximum DG integration capability is reached. After
SU et al.: DISTRIBUTED GENERATION INTERCONNECTION PLANNING: A WIND POWER CASE STUDY 183

Fig. 3. A simple feeder model.

Fig. 4. Simulation procedure of voltage control scheme CS2.

TABLE I
CONTROL SEQUENCES AND EQUIPMENT OF VOLTAGE CONTROLS the DG outputs. In [15], it was shown that by applying the fun-
damental theorem of calculating the probability distribution of
a function of random variable, for a given hour, using the mean
and standard deviation of the wind velocity measured and the
characteristic parameters for wind turbine, the distributions of
active and reactive powers produced by the wind turbine can be
obtained. The proposed procedure considers both normal and
each parameter adjustment a new DLF is conducted. The proce- worst case scenarios by using a probabilistic approach. It is as-
dure for control scheme CS-1 is similar to that of Fig. 4, except sumed that using data available from SCADA, AMI, customer
that the middle logic for line VR is not used. information system (CIS), the uncertainty of hourly bus loads
The time varying effects of bus loads and DG outputs on and DG outputs can be modeled by normal distributions with
volt/var control are included by using a random simulation ap- respective means and standard deviations to indicate the level
proach. Different types of distributions can be chosen to model of variations. For a new DG interconnection, typical generation
184 IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO. 1, MARCH 2011

profile submitted by DG investor in the interconnection appli-

cation can also be used in the impact study. For large varia-
tions of the wind turbine output, the probability density func-
tions of power outputs would have higher standard deviations.
When sampling bus injection data from specified distributions
for load flow studies, it is possible that the randomly selected
data could either be lower than the minimum (0.0) or higher than
the DG capacity. When this happens, they are set to the limits.
This treatment of volatile renewable energy outputs would lead
to a higher probability at the two ends of the power output dis-
tribution than that in the normal distribution.
Fig. 5 shows the proposed stochastic simulation procedure.
For each feeder configuration, bus injection data required for
one hundred load flow simulations are randomly
selected from their respective distributions for each bus and each
hour . The number of the simulated operation con-
ditions (i) can be increased to cover more operation conditions.
Maximum DG capacity at a given node of a feeder configura-
tion is obtained by observing all security constraints in each of
the randomly selected cases (2400 cases). The feeder bus volt-
ages obtained from DLF are used to compare the performance
of voltage control scheme CS-1 and CS-2. At the end of the sim-
ulations, the following items are calculated:
— feeder voltage quality index (VQI);
— average number of voltage control actions during a day; Fig. 5. Distribution load flow simulation procedure.
— maximum DG integration capacity.
VQI is defined as where is the DG capacity interconnected at node and
is the set of all feeder nodes. is the set of control vari-
ables including DG capacity and the status of feeder switches.
VQI
represents system hourly loading conditions. is the set
(2) of feeders receiving power from transformer . is the power
where is the voltage at node i, is the bus nominal flowing to feeder from transformer during loading condi-
voltage, is the preferred service voltage range, and tion. is the rating of transformer , and is the set of
is the number of nodes in the feeder. In our study, and all substation transformers. is the loading of feeder during
are set to 0.95 and 1.05 respectively. VQI has a maximum system loading condition, is the rating of feeder , and
value of 1.0, the higher the better. A voltage violation would re- is the set of all feeders. is the node voltage during
sult in a negative VQI . The average VQI in a service area with system loading condition, and are the voltage se-
N feeder buses can be calculated by curity limits. is the DG output during system loading
condition, and is the maximum DG rating allowed for the
voltage level. is the fault current at substation bus and
VQI VQI (3)
is the maximum interrupting current at bus connected to trans-
former .
Fig. 6 shows the proposed solution procedure. It is good for
To maximize the DG interconnection capacity at a given
single or multiple site interconnection study with or without
node, the problem is formulated as
existing DG in the system. A BPSO technique [16] is used to
search through possible network topology and determine the op-
(4) timal network configuration that has the highest fitness of the
objective function. Before the calculation of the maximum DG
Subject to the radial operation condition of the network and capacity for a specific network configuration, a network trace
security constraints, including procedure is performed to grasp the radial structure and ascer-
tain the availability of control devices in the current topology.
(5) After that DG capacity is increased and the voltage controls are
conducted by either CS-1 or CS-2 scheme until any of the secu-
(6) rity constraints, including bus voltage, fault current, and power
flow constraints, is violated.
(7)
Many previous studies have shown that through coopera-
(8) tion and competition among the population, particle swarm
(9) optimization (PSO) technique can often find good solutions
SU et al.: DISTRIBUTED GENERATION INTERCONNECTION PLANNING: A WIND POWER CASE STUDY 185

Fig. 6. Flow diagram of the coordinated feeder switching and voltage control for increasing DG integration capacity.

for complicated problems. This optimization approach updates interval [0.0, 1.0]. The following logistic function transforma-
the population of individuals by applying operators according tion is used in this study to update the movement [16]
to the fitness information obtained from the environment so
that the individuals of the population can be expected to move
towards better solution areas.
The particle of the BPSO method used in this paper is (10)
represented as , where represents
where is a sigmoid limiting transformation and
switch status, and D denotes total number of switches of the
is a quasi-random number selected from a uniform distribution
system. The best previous position (the position with the best fit-
in [0.0, 1.0]. The update scheme is as follows:
ness value) of the particle is recorded and represented as
. Meanwhile, the rate of position change (ve- for
locity) for particle is expressed as . In
for
a binary space, a particle can move to nearer and farther corners
of the hypercube by flipping various numbers of bits; velocity
of the particle can be described by the number of bits changed (11)
per iteration, or by the Hamming distance between the particle (12)
at time and .
The moving velocity is defined in terms of the changes in the else
probabilities that a bit will be a particular state. Thus a particle
end
move in a state space restricted to 0 and 1 on each dimension,
where each represents the probability of bit taking the end
value 1, since it is a probability and must be constrained to the end
186 IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO. 1, MARCH 2011

TABLE II
SPECIFICATIONS OF VOLTAGE CONTROL DEVICES IN FIG. 7

Fig. 7. Three feeder test system.

TABLE III
AVERAGE VQI DURING PEAK AND OFF-PEAK PERIODS WITH DG INTEGRATION
AND CS-2 CONTROL SCHEME

Fig. 8. Customer type load profiles.

Fig. 10. Daily average VQI of the system with DG integration.

Fig. 9. DG (wind turbine) power output patterns.

where denotes the best previous position of the particle,

represents the best among all the particles and m is the pop-
ulation size. D is the dimension size, , are learning factors,
and is the inertia weight. We set , to 2.0 and w between
0.2 1.2. The number of particles used in the tests is 50.

III. NUMERICAL RESULTS AND DISCUSSIONS

Several systems were tested and simulation results from three Fig. 11. Daily average VQI under different DG output patterns and capacities.
test systems are presented in this paper. In the followings, effects
of DG interconnection on feeder voltage controls are first pre- — The rating of the main transformer (69 kV/22.8 kV) is 25
sented, followed by test results on a possible increase of inter- MVA with an impedance of 9%.
connection capacity when a coordinated feeder switching and/or — Feeders are 3C500XP2 cables with impedance of
volt/var control are conducted. . There are 17 switches in the feeder and each
Fig. 7 shows a 19 bus test system. The system data used are feeder section has a length of 1.6 km.
as follows: — Typical load profiles of different load types are shown in
— The short circuit capacity at bus N1 is 800 MVA. Fig. 8. Hourly real and reactive power demands at each of
SU et al.: DISTRIBUTED GENERATION INTERCONNECTION PLANNING: A WIND POWER CASE STUDY 187

Fig. 12. Five bus example.

the 17 load buses are assumed and standard deviations of that when DG capacity is high, those with more uniform output
hourly load variations are assigned by using certain per- (Pattern 3) would have lower impact on feeder voltage quality.
centages of the mean value. With a good coordination of the controllers, adding more reg-
— The requested DG interconnection site is at bus 13. ulation facility should either increase the VQI or keep it un-
changed. However, if controllers are not properly coordinated,
A. Effects of DG Interconnection on Feeder Voltage Controls simulation results show that with the VR or OLTC adjustment,
the setting and condition of SC might be inappropriate or even
For testing DG impact on feeder voltage, three DG power in violation. This could lead to a hunting in the controller ad-
output patterns obtained from certain wind speed characteristics justment and deteriorate voltage quality.
are assumed and shown in Fig. 9. Generation output Pattern 1 Test results also show that under unbalanced feeder loading
matches roughly with the system load profile shown in Fig. 8. conditions, due to DG output variations, adjustment of OLTC
Pattern 2 has a higher output during early morning hours when at the substation might result in over and/or under voltage at
the load is low, and Pattern 3 has a relatively steady output as other feeders if a specific branch current measurement is used
compared to the other two. To model the possible variations for voltage control. Under this condition, if VR is available, it
of the output, at each hour, the values shown in Fig. 9 are the can be used to enhance voltage controls. If not, then according to
means of normal distributions that have corresponding standard the feeder section load profiles, adjustment of the feeder loading
deviations. by feeder reconfiguration would provide an opportunity to re-
Table II shows the voltage control devices and their specifica- solve the voltage quality problem and allow for additional DG
tions, including the control mode, adjustment range, and target interconnection capacity.
value of the regulated bus. Considering the time varying load
and DG output, Table III shows the VQI before and after DG in- B. Effects of Coordinated Feeder Reconfiguration and Voltage
terconnection for the system shown in Fig. 7 when the volt/var Controls on DG Interconnection Capacity
control scheme CS-2 is used. It can be seen that during peak load In this subsection, comparisons of increased DG interconnec-
period, the distribution feeder voltage quality could be main- tion capacities due to voltage control, feeder reconfiguration,
tained or improved with a lower number of controller adjust- and a coordinated control, respectively are given.
ments thanks to DG operations (Pattern 1). However, Fig. 10 Five Bus Test System: A simple example shown in Fig. 12
shows that, with an excessive DG interconnection, the average is first used to explain the benefit of the proposed scheme. It
feeder daily VQI are lower indicating voltage quality deteriora- is assumed that all feeder section impedances are all equal to
tion. , and the bus loads and shunt capacitance are
In the simulations, changes of VR setting are recorded to in- , ,
vestigate the effect of DG output patterns on VR operations in , , and
scheme CS-2. For the cases tested, the controller setting for VR respectively. Power factor of the DG is assumed equal
would need to be adjusted during day times before DG inter- to unity and OLTC uses local voltage measurement to maintain
connection. A DG at bus 13 with output Pattern 1 could help proper secondary voltage.
regulate the feeder voltage during the peak load period and re- Table IV shows the numerical results of test cases using dif-
sult in a lower number of VR switching. With a roughly reverse ferent control schemes. It can be seen that without considering
output profile (Pattern 2), the tap of VR would need to be ad- volt/var and feeder reconfiguration controls (case 2) the allow-
justed during early morning hours if DG penetration is high. able DG capacity at bus 3 is 0.438 pu. The maximum DG ca-
For a more stable output (Pattern 3), scheme CS-2 would have a pacity would increase if the voltage control is included. Under
lower number of controller operations. Fig. 11 shows the daily optimal feeder configuration with proper voltage controls (case
average VQI of the system with different DG capacities and 5) the allowable DG penetration could reach 0.724 pu, which is
output patterns using control scheme CS-2. Test results indicate a 65% increase from case 2.
188 IEEE TRANSACTIONS ON SMART GRID, VOL. 2, NO. 1, MARCH 2011

TABLE IV TABLE VI
AVERAGE VQI DURING PEAK AND OFF-PEAK PERIODS RESULTS OF TEN PSO SOLUTIONS
WITH DG INTEGRATION AND CS-2 CONTROL SCHEME

Case 1: Base case before DG interconnection at bus 3.

Case 2: Base case without voltage control and feeder reconfiguration.
Case 3: Case with voltage control and without feeder reconfiguration.
Case 4: Case without voltage control and with feeder reconfiguration.
Case 5: Case with both voltage control and feeder reconfiguration.

TABLE V
RESULTS OF FEEDER RECONFIGURATION FOR MAXIMIZING DG PENETRATION

Fig. 14. 33 bus feeder [12].

TABLE VII
Fig. 13. Optimal configuration for accommodating higher DG integration at MAXIMUM DG INTEGRATION CAPACITY BASED ON
bus 13. THREE FEEDER CONTROL SCHEMES

19 Bus Test System: Using voltage control scheme CS-2,

the feeder structure shown in Fig. 7 could accommodate 11.9
MW without violating considered security constraints. After the
BPSO search, it is found that a higher DG integration is prac-
ticable. Table V shows that with coordinated feeder reconfigu-
ration and voltage controls, an integration of 18.7 MW DG at
bus N13 is feasible. Fig. 13 shows the best configuration for
DG connection at bus 13. Since PSO is one of the stochastic
optimization methods, the performance of the solution would feasible feeder configurations. The execution time required is
hinge on the choice of seeds of random numbers and the search higher if higher particle number is specified in BPSO for solving
strategy adopted in each implementation. In our study, the per- the problem. The number of particles used for obtaining results
formance of BPSO method is evaluated statistically. As shown shown in Table VI is 50.
in Table VI, in most of the tested cases, the solutions obtained by 33 Bus Test System: Fig. 14 shows a 12 kV feeder with
BPSO method require lower execution time, and are the same as total real and reactive loads of 5,058.25 kW and 2,547.32
those obtained from using a brute force approach that tests all kVAr respectively. Without taking the possible gain of voltage
SU et al.: DISTRIBUTED GENERATION INTERCONNECTION PLANNING: A WIND POWER CASE STUDY 189

control and feeder reconfiguration into account, the allowable [6] T. Niknam, A. M. Ranjbar, and A. R. Shirani, “Impact of distributed
DG penetration at bus 17 is 1.8 MW. Table VII shows the generation on volt/var control in distribution networks,” in Proc. IEEE
Bologna Power Tech Conf., Bologna, Italy, Jun. 2003, vol. 3, pp. 1–7.
opened switches and the maximum DG capacity. It can be [7] M. Kim, R. Hara, and H. Kita, “Design of the optimal ULTC parameters
seen that with the inclusion of a coordinated volt/var control in distribution system with distributed generations,” IEEE Trans. Power
by SC and VR, the allowable DG capacity could reach 3.4 Syst., vol. 24, no. 1, pp. 297–305, Feb. 2009.
[8] C. Dai and Y. Baghzouz, “On the voltage profile of distribution feeders
MW. On the other hand, if volt/var control is disabled and with distributed generation,” in Proc. IEEE Power Eng. Soc. Gen.
feeder reconfiguration is executed, the allowable DG capacity Meet., Toronto, ON, Canada, Jul. 2003, pp. 1136–1140.
in the studied feeder would become 3.6 MW. Finally, with [9] J.-H. Teng, C.-Y. Chen, C.-F. Chen, and Y.-H. Liu, “Optimal capac-
itor control for unbalanced distribution systems with distributed gener-
coordinated feeder reconfiguration and volt/var control, the ations,” in Proc. IEEE Int. Conf. Sustainable Energy Technol. (ICSET
allowable capacity could reach 5.4 MW for DG interconnection 2008), Singapore, pp. 755–760.
at bus 17. The voltage rise problem due to DG interconnection [10] H. Hatta, S. Uemura, and H. Kobayashi, “Demonstrative study of con-
trol system for distribution system with distributed generation,” pre-
could be relieved by load transfers. sented at the IEEE/PES Power Syst. Conf. Expo., Seattle, WA, 2009.
[11] S. Civanlar, J. J. Grainger, H. Yin, and S. S. H. Lee, “Distribution feeder
IV. CONCLUSION reconfiguration for loss reduction,” IEEE Trans. Power Del., vol. 3, no.
3, pp. 1217–1223, Jul. 1988.
In order to optimize the use of the existing distribution net- [12] M. E. Baran and F. F. Wu, “Network reconfiguration in distribution sys-
work and reduce the cost of interconnection facility and dis- tems for loss reduction and load balancing,” IEEE Trans. Power Del.,
tribution system modification, this paper proposes a new DG vol. 4, no. 2, pp. 1401–1407, Apr. 1989.
[13] C. C. Liu, S. J. Lee, and K. Vu, “Loss minimization for distribution
interconnection planning study framework that includes a co- feeders: Optimality and algorithms,” IEEE Trans. Power Del., vol.
ordinated feeder reconfiguration and voltage control. The pro- PWRD-3, pp. 1281–1289, Apr. 1989.
posed method uses a probabilistic approach to account for the [14] J. Olamaei, T. Niknam, and G. Gharehpetian, “Impact of distributed
generators on distribution feeder reconfiguration,” in Proc. IEEE Lau-
time varying effects of feeder bus loads and DG output varia- sanne Power Tech, Lausanne, Switzerland, 2007, pp. 1747–1751.
tions, and provides an optimal feeder configuration to facilitate [15] N. D. Hatziargyriou, T. S. Karakatsanis, and M. Papadopoulos, “Prob-
higher DG capacity. Simulation results demonstrate the effec- abilistic load flow in distribution systems containing dispersed wind
power generation,” IEEE Trans. Power Syst., vol. 8, no. 1, pp. 159–165,
tiveness of the proposed method. Even though the test scenario Feb. 1993.
is based on wind power generation, the proposed method is also [16] S. A. Yin and C. N. Lu, “Distribution feeder scheduling considering
applicable to other types of DG integrations. It is shown that variable load profile and outage costs,” IEEE Trans. Power Syst., vol.
24, no. 2, pp. 652–660, May 2009.
DG output patterns could have positive or negative impacts on
feeder voltage control, and feeder reconfiguration could relieve Sheng-Yi Su received the M.S. degree from National Sun Yat-sen University,
some of the overvoltage problems due to DG interconnections. Kaohsiung, Taiwan, in 2009.
He is now with Taiwan Semiconductor Manufacturing Company, Hsinchu,
In order to observe operation security under DG operations, the Taiwan.
time varying effects of feeder section loads and output patterns
of existing and new DG are considered. Numerical results have
shown that a coordinated feeder switching and volt/var control
Chan-Nan Lu (M’82–SM’92–F’08) received the Ph.D. degree from Purdue
could permit an existing distribution network to accommodate University, West Lafayette, IN, in 1987.
higher DG capacity and consequently, reduce the network rein- He held positions at General Electric Co. and Harris Corp., and has been with
forcement investment for DG interconnections. National Sun Yat-sen University, Kaohsiung, Taiwan, since 1989. His research
interests are in power system operations and power quality.
REFERENCES
[1] I. Roytelman and S. M. Shahidehpour, “State estimation for electric
power distribution systems in quasi real-time conditions,” IEEE Trans.
Power Del., vol. 8, no. 4, pp. 2009–2015, Oct. 1993. Rung-Fang Chang received the Ph.D. degree from National Sun Yat-Sen Uni-
[2] The SmartGrids Technology Platform, The SmartGrids Strategic De- versity, Kaohsiung, Taiwan, in 2002.
ployment Document Apr. 2010 [Online]. Available: http://www.smart- He is an Associate Professor of the Department of the Electrical Engineering,
grids.eu/documents/SmartGrids_SDD_FINAL_APRIL2010.pdf Kao Yuan University, Kaohsiung, Taiwan. His research interests are optimiza-
[3] P. Chiradeja and R. Ramakumar, “An approach to quantify the tech- tion of power system and load research.
nical benefits of distributed generation,” IEEE Trans. Energy Conv.,
vol. 19, no. 4, pp. 764–773, Dec. 2004.
[4] R. C. Dugan, “Challenges in considering distributed generation in the
analysis and design of distribution systems,” in Proc. IEEE Power En- Guillermo Gutiérrez-Alcaraz (M’99) received the Ph.D degree from Iowa
ergy Soc. Gen. Meet., Pittsburgh, PA, Jul. 2008, pp. 1–6. State University, Ames, in 2009.
[5] H. Y. Li and H. Leite, “Increasing distributed generation using auto- He has been with the Department of Electrical Engineering, Instituto Tecno-
matic voltage reference setting technique,” in Proc. IEEE Power En- logico de Morelia, Morelia Mexico, since 1996. Currently, he is a postdoctoral
ergy Soc. Gen. Meet., Pittsburgh, PA, Jul 2008, pp. 1–7. fellow at National Sun Yat-sen University, Kaohsiung, Taiwan.