ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

Distribution Network Reinforcement Planning for

High Penetration Level of Distributed Generation
Chun-Lien Su *1, Hsiang-Ming Chuang *2
*
Department of Marine Engineering, National Kaohsiung Marine University
Kaohsiung, TAIWAN
1
cls@[Link]
2
adward721104@[Link]

Abstract— High penetration levels of distributed generation investment and the operating costs of DG units, costs paid to
(DG) significantly affect the operations of distribution networks purchase energy from the grid, and network reinforcement
that they are connected. Different network reinforcement costs, was proposed in [3] to determine the optimal schemes
alternatives that provide different levels of improvements to fit of sizing, placement, and timing of investments on DG units
with special requirements of individual utility are available.
and/or network reinforcements over the planning period. The
From a cost-benefit point of view, an optimal investment
strategy is necessary for network reinforcements in a cost- Monte Carlo simulation based probabilistic solution method
effective manner. This paper aims presenting a methodology for and a fuzzy satisfying method from [4] and [5] were used to
optimal planning of network reinforcements to accommodate deal with the uncertainties involved in the input parameters,
increased connection of DG to distribution networks. An respectively. [6] proposed an optimization framework to
objective function, which contains three objectives including address and compare the benefits of incorporating flexible DC
investment cost, customer interruption cost, and network losses links with conventional network reinforcement measures into
conversion cost, is minimized subject to system operation radial distribution networks, which are subject to continuous
constraints. A multi-stage approach based on genetic algorithms load growth and higher levels of DG penetration. [7]
(GAs) is then used to derive long-term investment planning and
presented a long-term dynamic multi-objective model, which
network configurations in the planning period. Test results of a
practical distribution feeder system connected to wind power simultaneously optimizes the objectives of the benefits of
generations is selected for computer simulation in order to distribution network operators (DNO) and distributed
ensure and demonstrate performance of the proposed method. generation owners (DGOs), and determines the optimal
schemes of sizing, placement, and, in particular, the dynamics
Keywords: Distributed generation, Network reinforcement, of investments in DG units and network reinforcements over
Investment planning, Service reliability. the planning period, when planning a distribution network. A
methodology was presented in [8] to quantify the deferment
produced by DG, which considers both demand growth- and
I. INTRODUCTION
system security-related investment at the planning stage;
With the vigorous development of renewable energy wherein, reinforcements required by system security
applications, category and type diversification, increased standards are considered.
capacity, and quantity of a distributed generation (DG) system, Currently, there are many network reinforcement
power companies face the shocks caused by the integration of technologies that can provide different levels of improvement
various types of DG systems into a power system. In response to suit the specific needs of different power companies. The
to this change, consideration is given to many distribution main solutions to the problem of DG integration operations
network reinforcement technologies or integration strategies can be classified as into grid voltage control, grid power flow,
for investment, while pending new integration technology. fault current, and other methods of management [9]. The
Thus, a proactive network management [1] has been proposed, above different network reinforcement programs are mutually
expecting to turn to higher order network operation exchangeable. After reviewing relevant network designs and
management by relying on the previous network design operational practices, they can be successively added into an
concept. However, since the problem of integrating a large expansion plan. Many studies suggest that they should be
number of distributed power generation systems has not been simultaneously considered in the investment planning of
considered in the initial design of the existing distribution distribution network reinforcement programs in order to
systems, a DG system integrated into a power system can provide better analysis results.
result in safety concerns regarding system operations, or The purpose of this paper is to build a set of analysis model
affect power supply quality, which requires network for optimal investment strategy for reinforcement programs
reinforcement programs [2]. for the integration of a distribution system and DG, which
There are some network reinforcement studies on uses the multi-stage optimization method for investment
distribution networks with DG penetration [3-8]. A dynamic planning of reinforcement programs for the management of
model for multi-objective DG planning, which considers the integrating a large number of DG units in a distribution
minimization of emissions and total cost, including network. With the overall costs including the investment cost

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1170

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

of reinforcement equipment, customer interruption costs, line loss, with unit investment costs of power generation
loss corresponding to the investment cost of power generation and transmission equipment;
and transmission equipment as the planning objectives. After CCIC is the customer interruption cost;
considering all grid operation safety constraints, it can PGi and QGi are the active and reactive powers
determine the optimal strategy for investment, as well as the
optimal network configuration for reinforcement programs for production of DG at node i, respectively;
management of voltage control, power flow, and fault current PLi and QLi are active and reactive injection powers
problems, as caused by the integration of DG units. It includes of node i, respectively;
reinforcement equipment installation time, location, and pij and q ij are the active and reactive powers of line
capacity, thus, DG units can be effectively integrated to
construct the future high-quality distribution network i-j;
configuration. i is the set of the lines connected to node i;
I Li is the current on line i;
II. PROBLEM FORMULATION
Traditional distribution system configuration expansion I Lmax is the maximum allowed current of line i;
i
mainly addresses the distribution network expansion planning
of the future load growth of the customers. In response to an S si is the electric load of transformer S connected to
increasing number of customers and growth in feeder load of feeder i;
each year, utilities expand their distribution systems and plan S srat is the rating of transformer S;
the construction and expansion of distribution equipment and
Sout is the set of the feeders powered by transformer S;
lines to reduce line loss and enhance supply quality and
reliability [10]. Such a planning problem is generally solved If is the feeder maximum fault current;
using the heuristics optimization method subject to future rat
ICB is the rated short-circuit breaking capacity of
power demands and system operation constraints. With
specific planning objectives, such as maximum power supply feeder CB;
reliability, line loss minimization, minimum investment cost, Vi is node i voltage;
and maximum supply quality, it determines the combination
Vimin and Vimax are the minimum and maximum
of control variables, including line switch and feeder
reconfiguration, line upgrading, construction of a feeder limits of the voltage of node i, respectively.
and/or a transformer substation. This paper aims to build a set
of optimal investment strategy for distribution network When the DG units are connected to an existing
reinforcements against high penetration levels of DG; distribution system, it will affect the power flow, fault current,
therefore, the mathematical equation for the reinforcements of and voltage control of the distribution feeder. Incorporating a
each regional distribution network with DG are illustrated as large number of DG units requires corresponding network
follows: reinforcements for management of the system with DG. As
mentioned above, regarding the three technological
min f ( x)=min CLossConv +CCIC (1) management perspectives, including power flow management,
voltage control, and fault current management, this paper
proposed a number of possible short-term reinforcement
Subject to measures to reduce the impact of the increased amount of
PGi  PLi   pij  0 for all nodes DGs on system operations. In general, the selection of the
ji
(2)
reinforcements should consider the effectiveness and
QGi  QLi   qij  0 for all nodes feasibility of managing the grid power flow, fault current, and
ji
(3)
voltage control under the conditions of normal operations of
Vimin  Vi  Vimax for all load points (4) DGs in the grid and without affecting customer electricity
usage. Table I shows the network reinforcement measures
I Li  I Lmax for all line sections (5) considered in this paper, which are the decision variables to
i
rat be determined.
 S si  Ss (6) In (1), customer interruption cost is calculated on the basis
iSout
rat of cost of annual expected energy not supplied (CEENS) in
I f  I CB (7) order to consider the interruption costs of various types of
Radial network topology (8) customers. In the analysis, it is assumed that DG can support
island operations under the loss of the main power supply
where CLossConv is the investment cost of line loss from utilities, which can improve power supply reliability and
conversion to power generation and transmission reduce customer interruption costs caused by accidents or
equipment, which is equal to the multiplication of line scheduled maintenance blackouts. Therefore, the calculation
of customer interruption costs may consider the main power
supply and/or the DG of an alternative power supply during

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1171

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

the outage [11]. As the loads in non-faulted regions are m

energized by the utility and/or DG, it should fit with system TCLi   CLossConv, j (10)
j 1
operational constraints, such as thermal capacity, voltage
change, DG maximum active and reactive powers. This type m
TCCi   CCIC , j (11)
of problem can be ensured through load flow computations. If j 1
the islanding grid is energized only by one DG, the DG can be
n
modeled as a swing bus. When there are a number of DGs, the TCIi   Rk  Ck  (1  DR) y (12)
DG of the maximum installed capacity can be modeled as a k 1
swing bus, while the remaining can be modeled as a PQ or PV
bus according to the reactive power control capability of the
where m is the final year of the planning period;
generator [12]. However, in some cases, to satisfy system
n is the number of types of reinforcement equipment in
operation constraints, it may have to reduce the power
the ith reinforcement program;
generation of the DG unit by the control of inverters or energy
Rk is the total number of the kth reinforcement devices;
storage devices or performing load shedding. The load
Ck is the unit cost of kth reinforcement equipment;
shedding strategy used in this paper considers customer
DR is the discount rate for reinforcement equipment
interruption costs, including customer type, load, and load
investment ;
priority, thus, the loads at the downstream will be shed first
Y is the years for use of kth reinforcement equipment
until all constraints are satisfied. The DG power generation
since its establishment during the total planning period.
reduction strategy reduces the output of the DG power
generation with larger installed capacity until no constraints
The proposed method calculates the TCR of all
are violated.
reinforcement programs, and the reinforcement program with
TABLE I
the minimum TCR is the optimal reinforcement equipment
REGIONAL DISTRIBUTION NETWORK REINFORCEMENT MEASURES investment portfolio. The network configuration integration
Items Reinforcement Measures of reinforcement equipment in different years is the optimal
- Add line voltage drop compensator network configuration, in response to the DG connection and
- Replace line conductor load growth of the year. The change in the network
Voltage Control - Install additional line switch or tied switch
- Distribution networker configuration configuration of each year during the final planning period is
- Construct dedicated line or network the optimal network configuration change. Other candidate
- Replace line conductor reinforcement programs, and their network configuration
Power Flow - Install additional line switch or tied switch changes, will be determined as secondary optimal solutions.
Management - Distribution network reconfiguration
- Construct dedicated line or network
The optimization problem formulated in (1)-(8) is a
- Add fault current limiter combinatorial constrained problem with non-linear and non-
- Upgrade distribution protection switching differential objective functions. Many optimization solution
Fault Current devices approaches, such as mixed integer programming (MIP) or
Management - Install additional line switch or tied switch
- Distribution network reconfiguration
artificial intelligent (AI) search techniques, can be used to
- Construct dedicated line or network solve the problem. However, the use of many models has
been restricted in some fields of analysis where the problem is
large in size and there are a large number of non-linear
III. TOTAL COST ESTIMATION AND SOLUTION TECHNIQUE interactions. In most cases, the use of linear approximations
The reinforcement program total cost considered in this or simplification of the model has been necessary in order to
paper includes line loss conversion costs, customer derive a solution.
interruption total costs, and the total investment cost for the Genetic algorithms (GAs) are an evolutionary optimization
adopted reinforcement measures of the optimal network approach which is an alternative to traditional optimization
configuration for the planned period of time. Regarding the methods. It follows the concept of solution evolution by
ith reinforcement program, its total cost TCRi is represented, stochastically developing generations of solution populations
as follows: using a given fitness statistic. It is particularly applicable to
problems which are large, non-linear and possibly discrete in
nature, features that traditionally add to the degree of
TCRi =TCLi  TCCi  TCIi (9) complexity of solution. Due to the probabilistic development
of the solution, GAs do not guarantee optimality even when it
where TCLi, TCCi, and TCIi represent total line loss may be reached. However, they are likely to be close to the
conversion cost, total customer interruption cost, and global optimum. This paper does not focus on the solution
total reinforcement equipment investment cost for the techniques of optimization; therefore, only a commonly-used
ith optimal network configuration in the planning GA-based algorithm [13] is implemented to determine the
period, respectively. They can be calculated by optimal solution.
The computational procedure of the proposed method is
summarized and shown in Fig. 1.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1172

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

In the analysis, with two DG units as a group, we implement

the load flow and short-circuit default in parallel sequence of
the two units in order to explore grid power flow, grid voltage,
and fault current management of different DG capacities, with
respect to the short-circuit capacity of the connection point.
According to the test results of system impact for different
load demand expansions, feasible regional network
reinforcement programs are developed and their
corresponding costs are calculated.

Fig. 1 Computational procedure of the proposed method

IV. TEST RESULTS AND DISCUSSION

A practical distribution feeder system shown in Fig. 2 is
selected for computer simulations to ensure and demonstrate
the performance of proposed method. The feeder includes 16 Fig. 2 Test feeder system
buses and its voltage level is 11.4 kV and the peak power of
the feeder is 4078 kW and the off-peak power is 1010 kW. TABLE II
The power lines consist of underground cable 3C500XP2, REINFORCEMENT EQUIPMENT C OSTS
overhead cable 3A477XPW, and 3C100mm2. The three- (a) Line switch
phase short-circuit currents of Buses B1, B4, B5, B8, and B10 Specifications Cost (NTD/each)
are 7.27 kA, 4.79 kA, 3.6 kA, 1.79 kA, and 1.62 kA, feeder terminal unit 181,410
(b) Conductor line
respectively. Regarding customer interruption cost data,
Model Voltage (kV) Cost (NTD/m)
please refer to [14]. The detailed information about peak and
3C500XP2 11.4 812.4
off-peak loads, and cable type, length, and impedance can be
3A477XPW 11.4 207
found in [15]. In the analysis, the annual load growth rate,
3C100mm2 11.4 390
discount rate, and total planning time period are assumed as (c) Feeder CB
2.5 %, 3.65 %, and 10 years, respectively. Moreover, each Specifications Capacity (kA) Cost (NTD/each)
reinforcement program consists of, at most, three different 23kV 600A VCB 12 900,000
types of devices. The line loss conversion unit cost is 21,891 (d) Line voltage drop compensator
NTD(New Taiwan Dollar)/kW. Table II shows the data of the Specifications Cost (NTD/each)
equipment costs of the reinforcement equipment involved in 10MVA 11.4kV 6,000,000
the programs considered in this paper. (e) Fault current limiter
To analyze the impact of different DG connections on the Specifications Cost (NTD/each)
feeder, it is assumed the capacity of each DG unit (wind 11.4kV 1250A fault current 8,930,000
power generation) is 2.3 MW, the rated voltage is 400 V, and (f) Power cable
the second transient reactance, transient reactance, and Conductor Voltage Impedance Capacity Unit Price
line Model (kV) (Ohm/km) (A) (NTD/m)
steady-state resistance are 0.2 p.u., 0.25 p.u., and 1.95 p.u.,
1C500XP2 11.4 0.090+j0.122 479 812.4
respectively. It is assumed that the DG power factor is 0.95.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1173

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

capacity upgrading investment, which are introduced into the

The impact analysis results have shown that grid voltage, grid during the second and sixth years. As revealed in Table
power flow, fault current, and DG equipment capacity have a III, the cost of Program B is the highest among all the
great relationship with the connection point short-circuit candidate programs, as it adds new line voltage drop
capacity. To conduct cost analysis of the network compensators and upgrades parts of the line in the second,
reinforcement programs of regions with different power fourth, and sixth years. As the cost of adding new power
supplies, as based on the impact test results of the system in supply lines is estimated by line length and unit price, other
Fig. 2, and according to the different power demands of the costs for installing the power supply line, such as road rights
regional distribution grid, it can plan the possible and construction costs, are not considered. Hence, the cost of
combinations of network reinforcement programs, which a line voltage drop compensator is much higher than the cost
correspond to different DG connections, while effectively of the specific line. As a result, the total cost of Program A is
managing power flow, voltage control, and fault current. more than that of Program E by about 1.62 times. However,
Based on the reinforcement equipment in Table II and by although Program E and Program D use different
following the computational procedure of Fig. 1, the analysis reinforcement equipment (the main difference is the use of a
results of the reinforcement program candidate grid specific line and line upgrading), the total cost is nearly the
configuration and network configuration changes can be same. It is noteworthy that, as the DG connection point is
obtained and showed in Fig. 3. In the analysis, it is assumed located at the end of the feeder, there is an impact on the fault
that the DG connection is from the first year to the sixth year, current. Therefore, reinforcement programs do not include the
and is located at the feeder end (Bus 10). Two DG units, with upgrading of protection switches or proactive fault current
a single unit capacity of 2.3 MW, are added each year to constraint devices. According to the test results, with the
conduct reinforcement program strategic configuration addition and configuration of line and interconnection
analysis. As shown in Fig. 3, the network configuration with switches, coupled with network reconfiguration, it can often
new DG connections is labeled by color. The No. 0 network result in maximum investment benefits, in terms of equipment
configuration is the initial planning year (current status) with investment, customer interruption loss, and line loss. Through
any DG connection and load growth. Network configuration the proposed method, as we can easily compare the total cost
changes, as denoted by dotted lines, means the requirements of the reinforcement program, we can determine the optimum
of the reinforcement equipment are due to violations of investment strategy.
operational constraints. On the other hand, solid lines denote
network configuration changes that do not require
reinforcement equipment. As the DG connections are few in
the first year of planning, and there is no network constraint
violation, it does not require the introduction of a network
reinforcement program. Since DG connections have resulted
in violations of the constraints of network voltage in the
second year of planning, the proposed analysis method can
automatically generate two groups of reinforcement programs,
including the addition of a new line voltage drop compensator
in line section 8, and the addition of a new line switch and
interconnection switch in line sections 10 and 17, according to
the violated constraints. According to reference, regarding the
final year of planning, it can generate five groups of
reinforcement programs and network configuration change
processes.
This study used the analysis procedures, as shown in Fig. 1,
to estimate the total cost of the five grid reinforcement
programs. Table III shows the final evaluation results,
including the year of establishment, reinforcement equipment
types, installation location, and ranking according to total cost.
As shown in Table III, the total cost of Program D is 1.145
million NTD, which is the minimum among the five
candidate reinforcement programs. Hence, all network
configurations contained in the program are the optimal
network architectural change procedures of the feeder during
the planning years. More specifically, optimal network
configuration changes require three types of reinforcement
equipment, within the future 10 year planning period, Fig. 3 Analysis results of candidate reinforcement programs and network
configuration change processes
including the line switch, interconnection switch, and line

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1174

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

[2] P. N. Vovos and J. W. Bialek, “Combinational mechanism for

generation capacity and network reinforcement planning,” IET
Generation, Transmission & Distribution, vol. 1, no. 2, March 2007, pp.
TABLE III
303-311.
ESTIMATION RESULTS OF THE TOTAL COSTS OF C ANDIDATE REINFORCEMENT
[3] A. Soroudi, M. Ehsan, and H. Zareipour, “A practical eco-
PROGRAMS
Total
environmental distribution network planning model including fuel cells
Reinforcement Year of Reinforcement Line to be Cost and nonrenewable distributed energy resources,” Renewable Energy,
Ranking
Program No. Establishment Equipment Type Installed (million vol. 36, no. 1, January 2011, pp. 179-188.
NTD)
[4] A. Soroudi, R. Caire, N. Hadjsaid, and M. Ehsan, “Probabilistic
Add new line voltage
2
drop compensator
8 dynamic multi-objective model for renewable and non-renewable
A 20.15 3
4
Add new power supply
11
distributed generation planning,” IET Generation, Transmission &
line Distribution, vol. 5, no. 11, November 2011, pp. 1173-1182.
Add new line voltage
2
drop compensator
8 [5] A. Soroudi and M. Ehsan, “A distribution network expansion planning
B
4 Line upgrading 6,8,11
21.77 5 model considering distributed generation options and techo-economical
6 Line upgrading 1,2,3,4,9,10 issues,” Energy, vol. 35, no. 8, August 2010, pp. 3364-3374.
2
Add new line voltage
8 [6] E. Romero-Ramos, A. Gómez-Expósito, A. Marano-Marcolini, J. M.
drop compensator
Maza-Ortega, and J. L. Martinez-Ramos, “Assessing the loadability of
4 Line upgrading 6,8,11
C Add new line switch 11 21.12 4 active distribution networks in the presence of DC controllable links,”
Add new IET Generation, Transmission & Distribution, vol. 5, no. 11, November
6 17
interconnection switch 2011, pp. 1105-1113.
Line upgrading 1,2,3,4 [7] A. Soroudi, M, Ehsan, R. Caire, and N. Hadjsaid, “Hybrid immune-
Add new line switch 10
2 Add new
genetic algorithm method for benefit maximisation of distribution
D 17 11.45 1 network operators and distributed generation owners in a deregulated
interconnection switch
6 Line upgrading 1,2,3,4 environment,” IET Generation, Transmission & Distribution, vol. 5, no.
Add new line switch 10 9, September 2011, pp. 961-972.
E 2 Add new 12.41 2
interconnection switch
17 [8] D.T.-C. Wang, L. F. Ochoa, and G. P. Harrison, “DG impact on
investment deferral: network planningand security of supply,” IEEE
Trans. on Power Systems, vol. 25, no. 2, May 2010, pp. 1134-1141.
V. CONCLUSIONS [9] A. Collinson, F. Dai, A. Beddoes and J. Crabtree, “Solutions for the
A methodology for strategic planning of regional connection and operation of distributed generation,” EA Technology Ltd.
distribution network reinforcements for DG penetration has Technical Report, July 2003.
[10] H. Lee Willis, Power Distribution Planning Reference Book, Marcel
been proposed. A practical distribution feeder system has Dekker, Inc., 1997.
been tested to ensure and demonstrate the performance of [11] F. Katiraei, M. R. Iravani, and P. W. Lehn, “Micro-grid autonomous
proposed method. According to the test results, when the DG operation during and subsequent to islanding process,” IEEE Trans. on
connection point is closer to feeder end, the voltage control Power Delivery, vol. 20, vo. 1, 2005, pp. 248-257.
[12] Y. Mao and K. N. Miu, “Switch placement to improve system reliability
and power flow management costs of the reinforcement for radial distribution systems with distributed generation,” IEEE Trans.
programs will be the major portion of investment costs. The on Power Systems, vol. 18, no. 4, November 2003, pp. 1346-1352.
cost of fault current management tends to increase when the [13] D. E. Goldberg, Genetic Algorithms: Search, Optimization and
connection point is closer to the feeder end. Therefore, when Machine Learning. Reading, MA: Addison-Wesley, 1989.
[14] C. L. Su, “Distributed generation and network upgrades placements for
determining network reinforcements, DG connection location distribution system expansion planning,” Advanced Materials Research
should be considered in the analysis. If the location is closer Journal- Materials Science and Information Technology, Vol. 433-440,
to the line end, the grid may be reinforced by adding new January 2012, pp. 1740-1744.
equipment or equipment upgrading. In this situation, the cost [15] C. L. Su, “Cost-effective analysis of network reinforcements for
accommodating increased connection of distributed energy resources to
will be lower than that of a power supply through dedicated distribution networks,” in Proc. 2010 International Conference on
lines. It is noteworthy that, the addition and configuration of Renewable Energy, Yokohama, Japan, 27 June-2 July 2010.
line and interconnection switches, coupled with network
reconfiguration, may result in the maximum investment
benefits, in terms of equipment investment, customer
interruption loss, and line loss. With the proposed method, the
total cost of different reinforcement options can be easily
compared and thus, the optimum investment strategy can be
determined.

ACKNOWLEDGEMENT
The authors gratefully acknowledge financial support for
this work from the National Science Council of Taiwan under
grant NSC 100-2221-E-022-001.

REFERENCES
[1] L. F. Ochoa, C. J. Dent, and G. P. Harrison, “Distribution Network
Capacity Assessment: Variable DG and Active Networks,” IEEE Trans.
on Power Systems, vol. 25, no. 1, February 2010, pp. 87-95.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1175