OPTIMAL SITING AND SIZING OF DISTRIBUTED GENERATION (DG) USING GENETIC ALGORITHM (GA) K. M. C. Marcelino, REE, MSEE, MBA 5/F Technical Services Building (TSB), Ortigas Ave., Ortigas Centre, Brgy. Ugong, Pasig City 1065 [email protected]ABSTRACT This paper presents an approach in searching for the optimal allocation (size) and location of Distributed Generation (DG) for active power loss reduction and voltage profile improvement of the system using Genetic Algorithm (GA) as a search method. GA will be applied in an existing 115 kV/34.5 kV distribution system of Manila Electric Company (MERALCO). DG was also introduced as an alternative to which will help in lessen the costs of expansion of the distribution system (CAPEX projects), ease feeder and delivery point power transformer loading, reduce technical losses, improve power system security, and introduce renewable energy as a source of power for distribution networks, etc. GA has proven in previous studies that GA can be applied in any type of situation and this study presents that DG has significant effect in the loss and voltage improvement of the distribution system. KEYWORDS: Distributed Generation (DG), Genetic Algorithm (GA), Active Power Loss, CIP2 Substation INTRODUCTION The development of Distributed Generation (DG) in the country is considered an essential part of the power system. DG allows electric supply coming from the large power systems or the location of the load center to be isolated from the rest with economic viability of building power lines. As we all know, DG can source its power from fuel cells, diesel, and renewable energy---such as wind turbines, hydro turbines and photovoltaic (PV) modules, among others. The distributed power generation or DG can be located anywhere as long as there is fuel available and there is a load that consumes. Using DG has many benefits such as low technical losses and voltage support, among others. DGs are small-scale power generation technologies used to provide an alternative to or an improvement of traditional power systems. These technologies includes a wide range of energy sources such as wind, solar, biomass, and storage. Using DG lessens a country’s greenhouse gases that can affect the environment. DG will be able to reduce the project costs or capital expenditures (CAPEX) of the DU where in modifications can be deferred at a later date. Distribution Utilities (DU) are currently experiencing low electric power supply due to the depletion of electric power supply generation and transmission constraints, such as low water levels for large hydro power generation and high fuel costs for operating oil-based generators that resulted in lower available generation capacity. DUs or ECs operates and maintains a large and sometimes complex distribution network or system or grid, which is a part or portion of a large power system. It is in the distribution system where loads or consumers are usually found. The distribution system is commonly operated at a voltage level of 13.2/13.8 kV for Electric Cooperatives (EC) and/or DU located in the provinces while 34.5 kV for Metro Manila. Commonly, distribution systems are radial networks, meaning the system is are not interconnected. The Distribution Development Plan (DDP) made by planning engineers for a DU or an EC provides for an orderly and timely modification to the distribution network to cope with the system. The costs of building projects includes large amount of money that will be used to provide customers more reliable, flexible and efficient power systems. DG systems are introduced in this paper to find an alternative which will help in lessen the costs of expansion, ease feeder loading, reduce technical losses, and introduce the use of renewable energy as a source of power for distribution networks, etc. However, this study is focused on reducing system technical losses. Technical losses are the main part or the major part of systems loss component. Reducing technical losses means having more power available, increasing the energy supply of the DU or having more efficient distribution system. This means that by reducing technical losses with the same amount of energy supply, the DU can serve more customers or consumers. Genetic Algorithm (GA) is an optimization method that is biologically inspired. It is a stochastic global search method that mimics the metaphor of natural evolution in order to solve problems and to model evolutionary systems. GA uses biologically inspired techniques such as genetic inheritance, natural selection, mutation, and sexual reproduction (recombination or crossover). Naturally, all living organisms or individuals in the world compete for resources such as food, water and shelter and most importantly for attracting mates. Those who are
15
Embed
OPTIMAL SITING AND SIZING OF DISTRIBUTED GENERATION (DG) USING GENETIC ALGORITHM · PDF file · 2016-11-15optimal siting and sizing of distributed generation (dg) using genetic...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
OPTIMAL SITING AND SIZING OF DISTRIBUTED GENERATION
(DG) USING GENETIC ALGORITHM (GA)
K. M. C. Marcelino, REE, MSEE, MBA
5/F Technical Services Building (TSB), Ortigas Ave., Ortigas Centre, Brgy. Ugong, Pasig City 1065
This paper presents an approach in searching for the optimal allocation (size) and location of Distributed
Generation (DG) for active power loss reduction and voltage profile improvement of the system using Genetic
Algorithm (GA) as a search method. GA will be applied in an existing 115 kV/34.5 kV distribution system of Manila Electric Company (MERALCO). DG was also introduced as an alternative to which will help in lessen
the costs of expansion of the distribution system (CAPEX projects), ease feeder and delivery point power
transformer loading, reduce technical losses, improve power system security, and introduce renewable energy as
a source of power for distribution networks, etc. GA has proven in previous studies that GA can be applied in any
type of situation and this study presents that DG has significant effect in the loss and voltage improvement of the
distribution system.
KEYWORDS: Distributed Generation (DG), Genetic Algorithm (GA), Active Power Loss, CIP2 Substation
INTRODUCTION
The development of Distributed Generation (DG) in the country is considered an essential part of the power
system. DG allows electric supply coming from the large power systems or the location of the load center to be
isolated from the rest with economic viability of building power lines. As we all know, DG can source its power
from fuel cells, diesel, and renewable energy---such as wind turbines, hydro turbines and photovoltaic (PV)
modules, among others. The distributed power generation or DG can be located anywhere as long as there is fuel
available and there is a load that consumes. Using DG has many benefits such as low technical losses and voltage
support, among others. DGs are small-scale power generation technologies used to provide an alternative to or an
improvement of traditional power systems. These technologies includes a wide range of energy sources such as
wind, solar, biomass, and storage. Using DG lessens a country’s greenhouse gases that can affect the environment.
DG will be able to reduce the project costs or capital expenditures (CAPEX) of the DU where in modifications
can be deferred at a later date.
Distribution Utilities (DU) are currently experiencing low electric power supply due to the depletion of
electric power supply generation and transmission constraints, such as low water levels for large hydro power
generation and high fuel costs for operating oil-based generators that resulted in lower available generation
capacity. DUs or ECs operates and maintains a large and sometimes complex distribution network or system or
grid, which is a part or portion of a large power system. It is in the distribution system where loads or consumers
are usually found. The distribution system is commonly operated at a voltage level of 13.2/13.8 kV for Electric
Cooperatives (EC) and/or DU located in the provinces while 34.5 kV for Metro Manila. Commonly, distribution
systems are radial networks, meaning the system is are not interconnected. The Distribution Development Plan
(DDP) made by planning engineers for a DU or an EC provides for an orderly and timely modification to the
distribution network to cope with the system. The costs of building projects includes large amount of money that will be used to provide customers more reliable, flexible and efficient power systems. DG systems are introduced
in this paper to find an alternative which will help in lessen the costs of expansion, ease feeder loading, reduce
technical losses, and introduce the use of renewable energy as a source of power for distribution networks, etc.
However, this study is focused on reducing system technical losses. Technical losses are the main part or the
major part of systems loss component. Reducing technical losses means having more power available, increasing
the energy supply of the DU or having more efficient distribution system. This means that by reducing technical
losses with the same amount of energy supply, the DU can serve more customers or consumers.
Genetic Algorithm (GA) is an optimization method that is biologically inspired. It is a stochastic global
search method that mimics the metaphor of natural evolution in order to solve problems and to model evolutionary
systems. GA uses biologically inspired techniques such as genetic inheritance, natural selection, mutation, and
sexual reproduction (recombination or crossover). Naturally, all living organisms or individuals in the world compete for resources such as food, water and shelter and most importantly for attracting mates. Those who are
able to survive and attract a mate will have relatively larger numbers of offspring and those who poorly perform
will produce fewer offspring or even no offspring at all. This means that the genes from the “fittest” will spread
from generation to generation. These are based on Charles Darwin’s “survival of the fittest” in the book of The
Origin of Species.
This study is all about finding optimal allocation and location of DG and to prove whether DG affects the voltage profile and technical losses of the system.
This study provides information that will help engineers in implementing DG as an alternative source of
power with the benefits of reducing costs and threat of vulnerability to terrorism and improving power system
security. Furthermore, this study will give other researchers ideas on how to implement DGs in distribution
systems that will benefit both customers and the utility using the proposed algorithm. The proposed algorithm can
be an aid to other researchers in which it can also be applied to other applications, such as applications in optimal
location and size of capacitors in the distribution system and others alike.
This study will focus on the optimal location and allocation (size) of the DG technology considering the
impacts on the 115 kV sub-transmission system of Manila Electric Company (MERALCO) in Laguna using GA.
GA will be responsible in searching for the best location and size of the DG. This paper will use the existing MERALCO 115 kV sub-transmission system as a test system for simulating the proposed algorithm and will be
programmed using MATLAB.
METHODOLOGY
The operating condition of a distribution systems drastically changes upon the connection of the DG
compared to a system without DG. Planning the installation of DGs should therefore be considered such as the
maximum number of DGs that can be connected into the system and at what capacities, where these DG units
should be installed, what technology suits best, and what connection type should be used, among others.
The problem of DG allocation and location should be approached with caution. If DG units are connected
at non-optimal locations, the system losses may tend to increase while the total system loss considering the point
of view of the grid or system operator (SO) may decrease. Interconnecting a non-optimal DG size at a non-optimal
or optimal location, the system losses may also increase. Previous studies have also shown that connecting DG at
non-optimal locations increases the system’s technical loss and therefore increases the cost of electricity (Kotb,
Shebl, Khazendar, & Husseiny, 2010). In MERALCO’s case, the interconnection of Alternergy Wind One Corp.’s
(AWOC) into MERALCO’s 115 kV system, MERALCO’s system loss increased while the whole or Luzon Grid
system loss decreased.
Of all benefits and objectives of implementing DG, the idea of implementing DG for loss reduction needs
special attention.
Problem Formulation
Solving the placement and allocation of DG units requires the Fitness Function (FF) or an Objective
Function wherein this fitness function will be optimized where the presence of some constraints are needed in
order to define the optimal location and size of the DG. The FF is selected to reduce the system power losses and
increase the voltage stability of the system.
Different scenarios are used in this study to determine the effectiveness of the proposed algorithm (i.e. peak
and off-peak conditions and DG sizes). These scenarios will all be applied with the proposed GA algorithm. GA
starts by automatically proposing different DG sizes and locations within the defined DG size and location limit
then internally executes the load flow program, using Power System Analysis Toolbox (PSAT), which is properly linked with the GA toolbox or package until the minimum solution is obtained for the suggested location and size.
The simulation process will be repeatedly executed until all the solutions and limits are met. The GA algorithm
will end when at least one of the conditions is met (Nallagownden, Thin, Guan, & Mahmud, 2006)
Generations – the algorithm stops when it reaches the specified maximum number of generations.
Time – the algorithm runs for the specified amount of time in seconds.
Fitness Limit – the algorithm stops when the fittest value in the current generation is less than or equal
to the specified fitness limit.
Stall Generations – the algorithm stops when there is no improvement in the fitness function value for a
certain number of generations.
Stall Time Limit – the algorithm stops when there is no improvement in the fitness function for a certain
period of time in seconds.
The suggested algorithm is programed under MATLAB software using both Power System Analysis Toolbox (PSAT) and Genetic Algorithm Toolbox (GA) and will be applied in an existing MERALCO 115 kV-34.5 kV
distribution system in Laguna.
Objective Function
The objective function or the fitness function is the function that needs to be minimized under various
constraints. The main objective of the proposed algorithm is to determine the best locations and sizes for new DG
resources by minimizing active power losses and improving the voltage profile of the system. According to
Pisisca, Bulac, and Eremia (2009) the objective or fitness function in order to minimize transmission loss is shown
as follows:
Minimizing the active power losses:
kji
ji PP,
1 )(
n
j
jiji
ji
jiij
jiji
ji
jiijn
i
QPPQVV
RQQPP
VV
R
11
sincos
where n is the number of buses, ijR is the resistance of the line between buses “i” and “j”, iP and iQ are net
real and reactive power injections in bus “i” and iiV , are the voltage magnitude and angle at bus “i”.
The active power losses are obtained from a load flow program. Further, the main function is subject to
operation constraints, such as the maximum and minimum power the DG can produce, the location of the DG,
etc. These limits are shown below:
DiiDGi PPP ,
n
k
kiikkiikki BGUU1
)sin()cos(
n
k
kiikkiikki BGUU1
)sin()cos(
maxii SS
maxmin iinomi UUU
maxDGDG PP
In addition to the capacity limits of the DG, the following constraints shall be considered in this study to
derive an optimal condition with consideration of system reliability. According to Abou El Ela, Zein El-Din and
Spea in 2006, these constraints or limits that must be followed are as follows:
Reliability Constraint
lossD
N
i
N
j
DGinj PPPPg DG
j
1 1
where injP is the injected power coming from the conventional power generation or delivery point substation at
bus i, jDGP is the power being generated by the DG at bus j, GN is the total number of delivery point buses,
and DGN is the total number of DG buses.
Upper and lower voltage limits
VVV ni
Figure No. 1 Genetic Algorithm Method
In order to assess the performance of the proposed algorithm, this paper used MATLAB as the computing
software to solve for the best location and allocation of the DG. GA Toolbox was used to determine the best
location and allocation of DG within the 115 kV sub-transmission of the MERALCO in Laguna. The power flow
analysis was implemented using PSAT (Milano, 2005). It is an open-source MATLAB Toolbox developed by Dr. Federico Milano; it has the capability to perform power flow analysis and other power system studies like optimal
power flow.
System Description
Manila Electric Company (MERALCO) was established in 1903 as Manila Electric Railroad and Light
Company to provide electric light and power and an electric street railway system to Manila and its suburbs
(MERALCO, 2016). It is a private distribution utility (DU) which holds a congressional franchise under Republic
Act or RA No. 9209 effective June 28, 2003, which grants MERALCO a 25-year franchise valid through June 28,
2028 to construct, operate, and maintain the electric distribution systems in the municipalities and cities of
Bulacan, Cavite, Metro Manila, and Rizal and certain cities, municipalities and barangays in the provinces of
Batangas, Laguna, Pampanga, and Quezon. MERALCO’s head office is located at Lopez Building, Ortigas Avenue, Pasig City.
As of Dec. 31, 2015, MERALCO has a total of ten (10) network sector offices, thrity (30) office branches
and seven (7) auxilliary offices, fourteen (14) extension offices, one (1) payment office and one (1) customer
assistance office aside from the head office in Ortigas Centre. Also, the company’s network facilities consists of
110 substations with a total capacity of 16,642 MVA (delivery point substations and distribution substations), 855
circuits with a total of 21,825 circuit-kilometers (sub-transmission and distribution) and 172,579 distribution
transformers wherein MERALCO’s distribution system in Laguna consists of three (3) delivery point substations,
thirteen (13) distribution susbtations, seven (7) customer-owned substations (6-load and 1-power plant) in looped
configuration. Figure No. 2 shows MERALCO’s franchise are and distribution facilities.
Figure No. 2 MERALCO Franchise Map and Distribution Facilities
RESULTS
In order to assess the proposed algorithm in solving the DG allocation and location problem, the method is applied to an actual system of MERALCO with 24 bus, 115 kV with 13.8 kV radial distribution system connected
wherein three (3) cases were shown. The initial system consists of 24 loads with a total peak load of 900 MW and
an off-peak load of 246 MW with a total active power loss of 5.45 MW and 0.75 MW, respectively, and total
reactive power loss of 130.24 MVAR (peak) and 20.97 MVAR (off-peak) using the actual loading of substations
obtained from SCADA. The slack bus is set at Biñan, Sta. Rosa and Calauan delivery point substations. Lastly,
MVA base is set at 100 MVA and maximum DG output is set at 100 MW for Case A, 200 MW for Case B and
293 MW for Case C (343 MVA x 0.95 p.f. x 90%).
Prior to the load flow analysis of the system, the number of populations and generations must be initially
set to determine whether these populations and generations can have an effect in finding the optimal size and
location. Tests were conducted to verify whether the statements are true or false. As described and in the example
by R. L. Haupt and S. E. Haupt in 2004, an initial of eight (8) population size and recommended generations of twenty (20) were used. Based on these settings, finding the optimal location and size is difficult due to the low
population size which the GA searches for the best population available and creates a new set of generation or
children. Increasing the population size to fourteen (14) and generation still set at twenty (20), finding the best
result increases as compared to having an initial population size of only eight (8). Further, increasing the
population size to twenty (20), as suggested and described by R. L. Haupt and S. E. Haupt in 2004, and generations
unchanged, the probability of finding the best result drastically increases because every iteration or every
simulation of the proposed algorithm shows only the result of the fittest. Further increasing the population size to
thirty (30) and generation still unchanged, the results of finding the fittest child suddenly decreases just like in
having an initial population of eight (8). Increasing or decreasing the population size does not mean finding the
best result or the optimal result. On the other hand, when trying to decrease or increase the number of generations
from the recommended generations of twenty (20) generations, the result is the same as increasing or decreasing the number of populations. Meaning, finding the best solution or optimal result decreases when both population
and generation were decreased. Finding the initial population size is very important due to it having an effect on
the evaluation of the cost function. Having a small initial population means it improves initial performance while
having a large initial population size improves long term performance. As described by R. L. Haupt and S. E.
Haupt in 2004, the best initial population size ranges from twenty (20) to thirty (30). Further, population size is a
parameter setting that will have an effect to the evaluation of the cost function or objective function and it also
indicates the performance of GA. The entire algorithm is based on randomness. GA is an algorithm of random
search; repeatability of results is a concern. To address this issue, ten (10) independent runs/tests/iterations were made to verify the high probability of the results.
Selection function is what chooses parents that will be needed in the creation of the next generation.
Roulette-wheel selection was used because the chromosomes with the bigger fitness will likely be selected many
times to produce the next set of generation. Meaning roulette-wheel based selection will produce fitter results
compared to other selection methods such as ordinal selection wherein the chromosomes are chosen randomly.
In this study the following parameters were set and the rest are default by the GA:
Generations – 20
Initial Population – 20
Selection Function – Roulette Wheel
Base Case – No DG
This case discusses the existing condition of the system during peak and off-peak condition. The chosen
distribution system is located at Laguna where large demand customers are located such as various manufacturing
companies such as Toyota Motors, Inc. Samsung Electronics, and other alike. Also, it is in Laguna where various
industrial parks are located making Laguna as the industrial hub of MERALCO’s franchise area. Further, these
customers requires better system voltages compared to other customers like commercial and residential. As these
customers utilizes power quality or voltage sensitive equipment. If the system voltage is poor, these industrial
customers will opt to find other distribution utilities that can provide better system voltages in order for them to
manufacture better. Having poor system voltage will mean bad product or waste product output for the manufactures as they will not be able to sell their products. MERALCO’s distribution system in Laguna has a
peak demand of 900 MW and off-peak demand of 246 MW.
It can be seen in Table No. 1 that the exisitng system loss during peak is 0.0545 p.u. or 5.45 MW and 0.75
MW during of peak condition. Also shown in Figure No. 3 the system voltages at 115 kV bus during peak and
off-peak condition without DG. The figure shows that during peak condition, the system voltages sags due to the
simultaneous operations of the industrial, commercial and residential customers while there is an increase of
system voltages during off-peak condition due to limited operations of the various consumers whithin the area.
The figure also indicates that there is a room to further improve the system voltage of the distribution utility
especially the bus voltages between Sta. Rosa and Calauan delivery point substations even though the system
voltage is still within the prescribed 0.95 p.u. to 1.05 p.u. limit of the Philippine Grid Code (PGC).
Table No. 1 System Loss without DG
Case
Peak Off-peak
Optimal System Loss Optimal System Loss
Location (Substation)
Output (p.u.)
Active (p.u.)
%
Improvement Location
Output (p.u.)
Active (p.u.)
%
Improvement
Base - - 0.0545 - - - 0.0075 -
Figure No. 3 System Voltage without DG
Case A – DG output set to 100 MW
As discussed above, the setting for the initial population for this study is set to twenty (20) with twenty
(20) generations. During peak load system, the optimal location to interconnect the DG produced two (2) locations (Calamba and CIP2 substations) as these locations yielded the highest improvement in active system’s loss or an
improvement of 9.37%. However, when the system load is decreased to off-peak loading condition, the optimal
location yielded only one (1) location which is the CIP2 substation with an 18.89% system loss improvement. It
can be seen in Table No. 2, the optimal locations to interconnect the DG under peak and off-peak system loading
condition. As previously mentioned, the optimal locations to interconnect the DG are at Calamba and CIP2
substations during peak condition due to high similar results indicating that these substations yielded the highest
system loss improvement, one can accept that these results are accurate. However, during off-peak condition, only
one (1) substation location was determined. The equivalent optimal output powers of DG ranges from 0.9984 to
0.9999 p.u. or from 99.84 MW to 99.99 MW during peak condition while from 0.9784 to 0.9994 or from 97.84
MW to 99.94 MW during off-peak condition, which is almost equal to the maximum output power limit of the
DG. The table also shows how the system’s total active power losses decreased from the initial loss of 5.45 MW
down to 4.94 MW, or a 9.37% improvement during peak load system while an 18.89% improvement or from 0.75 MW to 0.61 MW. The chosen locations are considered the load centers of the distribution system in Laguna as
various industrial estates and customers are located within the area covered by these substations and is in between
the Sta. Rosa and Calauan delivery point substations where there is still an opportunity for improvement as
discussed in the Base Case.
Table No. 2 Result of iterations for Case A (peak and off-peak conditions)
Figure No. 4 shows the comparative system voltage profile during peak condition between the Base Case
and the two (2) identified optimal substations to interconnect the DG while Figure No. 5 shows voltage profile
during the off-peak condition and the DG connected at CIP2 substation as this is only the identified optimal
location to interconnect the DG. It can be seen in both figures that the voltage profile of the system improved and
is now nearer the unity value or the 1.0 p.u. voltage.
Figure No. 4 System Voltage during Peak Condition
Figure No. 5 System Voltage during Off-peak Condition
Case B – DG maximum output set to 200 MW
In this case, the maximum output of the DG was increased from 100 MW to 200 MW to determine whether
this setting can have an effect in the performance of the GA, the DG optimal location and size, system’s active
power loss and system’s voltage profile. Based on the results found in Table No. 3, increasing the DG power
output improved the system’s loss by more than double during peak condition, from 9.37% improvement in Case
A to 19.46% improvement (5.45 MW to 4.41 MW) and by almost 70% during off-peak condition or from 18.89%
in Case A to 32.19% (0.75 MW to 0.51 MW). During peak condition, the same two (2) substation locations that were identified in Case A were also identified in this case during peak condition. However, it was identified that
Calamba substation is more appropriate to interconnect the DG as this occurred four (4) out of the ten (10)
iterations while CIP2 substation only occurred once. The size of the DG ranges from 1.9965 to 2.0000 p.u. or
from 199.65 MW to 200 MW.
0.9600
0.9700
0.9800
0.9900
1.0000
1.0100
Bus Voltages at 115 kV Bus (p.u.) during Peak Condition
Without DG With 100 MW DG at Calamba With 100 MW DG at CIP2
0.9800
0.9900
1.0000
1.0100
1.0200
1.0300
Bus Voltages at 115 kV Bus (p.u.) during Off-peak Condition
Without DG With 100 MW DG
During off-peak condition, the same substation in Case A was also identified as optimal location to
interconnect the 200 MW DG which is CIP2 substation. However, it is more evident in this case that CIP2 is the
optimal location as seven (7) out of the ten (10) simulation results indicates the CIP2 is the appropriate location
to interconnect the DG. The DG’s power output ranges from 1.9778 to 1.9996 p.u. to from 197.78 MW to 199.96
MW.
Table No. 3 Result of Iterations for Case B (peak and off-peak conditions)
Shown in Figure No. 6 the effect of increasing the DG’s maximum output power during peak condition to
the system’s voltage profile among Base Case, Case A and Case B. Figure No. 7 shows the system’s voltage
profile during off-peak condition with CIP2 as the only identified optimal location to interconnect the DG for
Cases A and B. Both the figure shows that increasing the maximum output power of the DG has a little effect in
the system voltage profile improvement compared with Case A or with DG’s maximum output set at 100 MW
because the lines are now almost or overlapping with each other indicating that there is a very little improvement.
It can be noted that increasing the DG’s output does not have an effect the performance of the GA as the GA still obtained the same results got from Case A during peak and off-peak condition of the system.
Figure No. 6 System Voltage during Peak Condition
0.9600
0.9700
0.9800
0.9900
1.0000
1.0100
Bus Voltages at 115 kV Bus (p.u.) during Peak Condition
Without DG With 100 MW DG at Calamba With 100 MW DG at CIP2
With 200 MW DG at Calamba With 200 MW DG at CIP2
Figure No. 7 System Voltage during Off-peak Condition
Case C – DG maximum output set to 293 MW
In this case, the maximum power output was further increased to 293 MW, again to determine the effect
in the performance of the GA, DG optimal location and size, system’s active power loss and system’s voltage
profile. The maximum output of 293 MW DG was derived using the maximum capacity of the wire or 115 kV
line, which is 2-795 MCM ACSR with a capacity of up to 343 MVA, multiplied by the assumed 95% power factor
and the 90% critical loading criteria of the line or the wire (343 MVA x 90% p.f. x 90% critical loading criteria).
It can be seen in Table No. 4 that during the system peak condition, system loss improved 27.53% or by
8.07% basis points from Case B and 18.16% basis points from Case A. Further, the system loss improvement were lower due to the incremental percentage increase compared from Cases A and B where the incremental
percentage increases were 9.37% and 10.09%, respectively. This means that there is a certain point wherein the
system loss improvement will reach its peak and will now start to improve at a slower pace. This can also be the
cause of the load of the system, where the load of the system is now insufficient to cause a higher system loss
improvement which means that the DG’s maximum effect in the system can also be determined how large or how
big the load or the system is. Further, it is safe to say that the DG’s size is relatively proportional to the system
loss improvement. Meaning, as the DG’s output increases the system loss improvement also increases. Further in
this Case, only Calamba substation was identified as a suitable location to interconnect the DG because of having
the highest system loss improvement compared with the system without the DG or the Base Case and the DG’s
output ranges from 292.60 MW to 292.77 MW.
During off-peak system, the system loss improved to 34.85% compared to Base Case or the system loss
only improved by a mere 2.66% basis points compared with Case B and 15.96% basis points compared with Case
A which is lower than the increase compared with Case B from Case A. On one hand, the optimal location to
interconnect the DG is still the same which is CIP2 and is consistent with the simulation results in the previous
cases during off-peak condition as this resulted in the highest system loss improvement. Also, during off-peak
condition, only CIP2 susbtation was identified because the the substation occurred eight (8) out of the ten (10)
iteration or simulation results. Finally, the DG’s output ranges from 282.95 MW to 286.21 MW. It can be noted
that the DG’s maximum output range is almost not equal to the set output of 293 MW, which means that beyond
the DG output range, the system loss improvement will be lesser than the obtained 34.85% system loss
improvement.
0.9800
0.9900
1.0000
1.0100
1.0200
1.0300
Bus Voltages at 115 kV Bus (p.u.) during Off-peak Condition
Without DG With 100 MW DG With 200 MW DG
Table No. 4 Result of Iterations for Case C (peak and off-peak conditions)
Figure No. 8 shows the system voltages among the cases. The figure again shows that there is a very little
room for improvement in terms of the system voltage even as the DG’s output is increased as discussed earlier in
Case B. However, as seen in Figure No. 9 during off-peak condition, Case C has the best system voltage compared with the other Cases as this has the best system voltage.
It can generally be noted that increasing the DG’s output does not have an effect the performance of the
GA as the GA still obtained the same results (optimal location and size of the DG) got from all the case with DG
Bus Voltages at 115 kV Bus (p.u.) during Peak Condition
Without DG With 100MW DG (Calamba) With 100MW DG (CIP2)
With 200MW DG (Calamba) With 200MW DG (CIP2) With 293MW DG (Calamba)
Figure No. 9 System Voltage during Off-peak Condition
Summary of all the Cases
It can be seen in the tables below the optimal locations, DG’s output power range and system loss
improvement during peak and off-peak condition of all the cases.
Table No. 5 Summary of Results of System Loss Improvement during Peak Condition
Case
Optimal System Loss
Location (Substation)
Output (p.u.) Active (p.u.)
%
Improvement
Base - - 0.0545 -
A Calamba 0.9988 to 0.9999
0.0494 9.37% CIP2 0.9984 to 0.9995
B Calamba 1.9965 to 2.0000
0.0439 19.46% CIP2 1.9972
C Calamba 2.9260 to 2.9277 0.0395 27.53%
Table No. 6 Summary of Results of System Loss Improvement during Off-peak Condition
Case
Optimal System Loss
Location (Substation)
Output (p.u.) Active (p.u.)
%
Improvement
Base - - 0.0075 -
A CIP2 0.9784 to 0.9994 0.0061 18.89% B CIP2 1.9778 to 1.9996 0.0051 32.19%
C CIP2 2.4671 to 2.8621 0.0049 34.85%
CONCLUSIONS
Genetic algorithm (GA) can be used as a better tool in finding the optimal solution(s) compared to manually
searching for the optimal location(s) to interconnect and the size(s) of the DG(s) in the distribution system.
Moreover, GA is faster, easier and more accurate to use compared to other search methods. Adding DG in to the
system reduces both active and reactive power losses, improves power system security by sourcing power within
the distribution system instead of sourcing from the bulk power generators thereby reducing the threat of
vulnerability to terrorism and it improves the system’s voltage profile as seen in the various tables and figures
presented in all the cases, respectively. Further, the amount of effect of the DG to the system greatly depends to the optimal location and size of the DG installed in the system and the load of the system itself or the distribution
system. It can also be seen in all the results and cases that the output limit of the DG is somewhat directly
proportional to the system’s active power losses improvement during peak and off-peak condition. However, there
0.9900
1.0000
1.0100
Bus Voltages at 115 kV Bus (p.u.) during Off-peak Condition
Without DG With 100MW DG With 200MW DG With 293MW DG
will come a point that the improvement will be lesser compared with the previous size or output of the DG as
DG’s, as previously mentioned, effect on the system also depends on the load of the system itself. On one hand,
increasing the size or power output of the DG did not have an effect on the performance of the GA as shown in
Case B and C.
Based on the various cases shown above, the optimal substation location to interconnect the DG must be at CIP2 substation during peak and off-peak conditions. This is because most of the large industrial customers are
located near the CIP2 substation; it produces the least system active power loss or has the best active power loss
improvement and resulted in best system voltage. Further, CIP2 substation is the only substation location that has
been identified during off-peak condition while it is one (1) of the two (2) substations indicated in Cases A and
B. Lastly, CIP2 substation is almost equidistant between two (2) delivery point substations, Sta. Rosa and Calauan
DPs.
Another key point in this study is the optimal size of the DG. Based on the various cases, the optimal size
of the DG is 100 MW as this resulted in better system voltage and there is only very little or minimal improvement
in system voltages when the DG’s size is increased. Furthermore, it did not violate any system condition such as
having a power backflow at the 230kV side which will need coordination with the grid owner/operator if ever
backflow is allowed and various re-setting of protection and control relays at the 230kV side.
RECOMMENDATIONS
Objectives in this study have been properly addressed by proving that GA is a better search method
compared to other known methodologies, it minimized the system’s total active losses, found the optimal location
to interconnect and size of the DG and voltage profile improvement of the system. However, further studies must
be conducted on how to minimize both the system’s total active and reactive power losses. Reactive power losses
have also great effect in the system’s capability to transfer power from point to point because reactive power
losses limit the transfer of real power in the system. In this study, the DG must both inject real and reactive power
to the system to reduce both the active and reactive power losses in the system.
Another study must be added is the economic analysis of what type of DG will be best suited for the system.
In the objective function, both the economic and technical analysis must be carefully taken into consideration
because this will determine whether the study will be both beneficial to the people and the investors. This will
easily determine the type, size and location of DG.
Identically with the other studies, the maximum number of DGs must also be studied as this have an effect
in the planning of the whole system. Also, the number of DGs in the system also affect the system’s reduction in
losses and to determine the relationship between the numbers of DGs in the system the voltage profile of the
system because the system may get worse if the number of DGs in the system is too much or the system cannot
afford additional power source.
Lastly, the seasons of the year must also be taken into consideration because every season there are various
characteristics of the system due to the nature of the loads on how they react with the season and the system. For
example, the summer season in the Philippines registers the peak load of the system or the highest load of the
system for the whole year. Moreover, it is the time of the season where the surroundings are very hot and dry.
Taking this into consideration will make the DG suit better. With this known characteristics of the system, adding
the type of DG and knowing the optimal output of both the real and reactive power will make the study better.
There are lots of studies that can be taken into consideration like the impacts of the DG to the system (i.e. finding
the correct size of protective relays in the system, harmonics, etc.).
REFERENCES
Abou El Ela, A. A., Zein El-Din, A., & Spea, S. R. (2006). Multi-objective Genetic Algorithm to Improve the
Power System Operation using Distributed Generation. IEEE Transactions on Power Systems Conference, ISBN
978-1-4244-5111-1, pp 612-619.
Aslanzadeh, S., & Jamali, S. (2007). Study of Distributed Generation Type and Islanding Impact on the Operation
of Radial Distribution Systems. IEEE Transactions on Conferences on Universities and Power Engineering, ISBN
978-1-905593-36-1, pp. 1216-1221.
Beasley, D., Bull, D. R., & Martin, R. R. (1993). An Overview of Genetic Algorithms: Part 1 Fundamentals.
University Computing, Vol. 15 (2), pp. 58-69.
Bharathi Dasan, S. G., Ramalakshmi, S. S., & Kumudini devi, R. P. (2009). Optimal Siting and Sizing of Hybrid
Distributed Generation using EP. IEEE Transactions on Power Systems, ISBN 978-1-4244-4330-7, pp. 1-6.
Brown, R., Gupta, S., Christie, R., & Venkata, S. (1996). A Genetic Algorithm for Reliable Distribution System
Design. IEEE Transactions on Conferences on Intelligent Systems Applications to Power Systems, pp. 29-33.
Chakraborty, D., Sharma, C., Das, B., Abhishek, K., & Malakar, T. (2009). Distribution System Load Flow
Solution Using Genetic Algorithm. IEEE Transactions on Conferences on Power Systems, ISBN 978-1-4244-
4330-7, pp. 1-6.
Chen, P., Chen, Z., Bak-Jensen, B., Villafafila, R., & Sorensen, S. (2007). Study of Power Fluctuation from
Dispersed Generations and Loads and its Impact on a Distribution Network through a Probabilistic Approach.
IEEE Transactions on Conferences on Electrical Power Quality and Utilization, ISBN 978-84-690-9441-9, pp.
1-5.
Conti, S., Raiti, S., Tina, G., & Vagliasindi, U. (2001). Study on the Impact of PV Generation on Voltage Profiles
in LV Distribution Networks. IEEE Transactions on Conferences on Power Tech Proceedings, Vol. 4, pp. 1-6.
Daly, P. A., & Morrison, J. (2001). Understanding the Potential Benefits of Distributed Generation on Power
Systems Delivery. IEEE Transactions on Rural Electric Power, ISBN 0-7803-7012-0, pp. A2/1-A213.
Driesen, J., & Belmans, R. (2006). Distributed Generation: Challenges and Possible Solutions. IEEE Transactions
on Conferences on Power Engineering Society, ISBN 1-4244-0493-2, pp. 1-8.
Eto, J., Koomey, J., Lehman, B., & Martin, N. (2001). Scoping Study on Trends in the Economic Value of Electricity Reliability to the US Economy. LBNL-47911, pp. 1-148.
Goldberg, D. E. (1989). Genetic Algorithms in Search Optimization and Machine Learning. Boston,
Massachusettes: Addison-Wesley Longman Publishing Co., Inc.
Goldberg, D. E. (1999). The Race, the Hurdle, and the Sweet Spot: Lessons from Genetic Algorithms for the
Automation of Design Innovation and Creativity. In P. Bentley, Evolutionary Design by Computers (pp. pp. 105-
118). San Mateo, California: Morgan Kaufmann.
Grainger, J. J., & Stevenson, J. W. (1994). Power System Analysis. Princeton, New Jersey: Mc-Graw Hill, Inc.
Guan, F. H., Zhao, D. M., Zhang, X., Shan, B. T., & Liu, Z. (2009). Research on Distributed Generation Technologies and its Impacts on Power Systems. IEEE Transactions on Conferences on Sustainable Power and
Generation Supply, ISBN 978-1-4244-4934-7, pp. 1-6.
Holland, J. H. (1975). Adaption in Natural and Artificial Systems. Ann Arbor, Michigan: Univeristy of Michigan
Press.
Jie, W. (2009). Control Technologies in Distributed Generation System Based on Renewable Energy. IEEE
Transactions on Conferences on Power Electronics Systems and Applications, ISBN 978-1-4244-3845-7, pp. 1-
14.
Kazemi, A., & Sadeghi, M. (2009). Distributed Generation Allocation for Loss Reduction and Voltage Improvement. IEEE Transactions on Power and Energy Engineering, ISBN 978-1-4244-2486-3, pp. 1-6.
Kumar, Y., Das, B., & Sharma, J. (2006). Genetic Algorithm for Supply Restoration in Distribution system with
Priority Customers. IEEE Transactions on Conferences on Probabilistic Methods Applied to Power Systems,
ISBN 978-91-7178-585-5, pp. 1-6.
Liang, H., Gu, X., & Zhao, D. (2010). Optimization of System Partitioning Schemes for Power System Black-
Start Restoration Based on Genetic Algorithms. IEEE Transactions on Power and Energy Engineering, ISBN
978-1-4244-4812-8, pp. 1-4.
Milano, D. F. (2005). An Open Source Power System Analysis Toolbox. IEEE Transactions on Power Systems,
Vol. 20 (3), pp. 1199-1206.
Mitchell, M. (1996). Introduction to Genetic Algorithms. Boston, Massachusettes: MIT Press.
Nallagownden, P., Thin, L. T., Guan, N. C., & Mahmud, C. M. (2006). Application of Genetic Algorithm for the
Reduction of Reactive Power Losses in Radial Distribution System. IEEE Transactions on Power and Energy
Conference, ISBN 1-4244-0273-5, pp. 76-81.
O'Kane, P. J., Fox, B., & Morrow, D. J. (1999). Impact of Embedded Generation on Emergency Reserve. IEEE
Transactions on Proceedings on Generation, Transmission and Distribution, Vol. 146 (2), pp. 159-163.
Pecas Lopes, J. A. (2002). Integration of Dispersed Generation on Distribution Networks - Impact Studies. IEEE
Transactions on Conferences on Power Engineering Society, Vol. 1, pp. 323-328.
Pisica, I., Bulac, C., & Eremia, M. (2009). Optimal Distributed Generation Location and Sizing using Genetic
Algorithms. IEEE Transactions on Conferences on Intelligent System Applications to Power Systems, ISBN 978-
1-4244-5097-8, pp. 1-6.
Rashidi, M., Rashidi, F., & Monavar, H. (2003). Tuning of Power System Stabilizers via Genetic Algorithm for
Stabilization of Power Systems. IEEE Transactions on Conferences on Systems, Man and Cybernetics, Vol. 5,
pp. 4694-4654.
Ravibabu, P., Venkatesh, K., & Kumar, C. S. (2008). Implementation of Genetic Algorithm for Optimal Network
Reconfiguration in Distribution Systems for Load Balancing. IEEE Transactions on Computational Technologies in Electrical and Electronics Engineering, ISBN 978-1-4244-2133-6, pp. 124-128.
Samaan, N., & Singh, C. (2005). Using Genetic Algorithms for Reliability Calculations of Complex Power
Systems. IEEE Transactions on Conferences on Circuits and Systems, Vol. 5, pp. 4191-4195.
Sastry, K., Goldberg, D., & Kendall, G. (2005). Genetic Algorithms. In E. K. Burke, & K. Graham, Search
Methodologies (pp. 97-125). Springer.
Slootweg, J. G., & Kling, W. L. (2002). Impacts of Distributed Generation on Power System Transient Stability.
IEEE Transactions on Conferences on Power Engineering Society, Vol. 2, pp. 862-876.
Sundhararajan, S., & Pahwa, A. (1994). Optimal Selection of Capacitors for Radial Distribution Systems Using a Genetic Algorithm. IEEE Transactions on Power Systems, Vol. 9 (3), pp. 1499-1507.
Surrender, D., Jain, D. K., & Kumar, A. (2006). Power Quality Issues of Embedded Generation. IEEE
Transactions on Conferences on Power India, ISBN 0-7803-9525-5, pp. 1-4.
Tiwari, P. K., & Sood, Y. R. (2009). Optimal Location of FACTS Devices in Power System Using Genetic
Algorithm. IEEE Trasanctions on Conferences on Nature & Biologically Inspired Computing, ISBN 978-1-4244-
5053-4, pp. 1034-1040.
Virginia Tech. (2007). Benefits of Distributed Generation. Retrieved June 19, 2016, from Distributed Generation:
http://www.dg.history.vt.edu/ch1/benefits.html
Waseem, I., Pipattanasomporn, M., & Rahman, S. (2009). Reliability Benefits of Distributed Generation as a
Backup Source. IEEE Transactions on Power and Energy Society, ISBN 978-1-4244-4241-6, pp. 1-8.
Zoka, Y., Sasaki, H., Kubokawa, J., & Yokoyama, R. (1995). An Optimal Deployment of Fuel Cells in
Distribution Systems by Using Genetic Algorithms. IEEE Transactions on Conferences on Evolutionary