TRIBHUVAN UNIVERSITY

INSTITUTE OF ENGINEERING

PASHCHIMANCHAL CAMPUS

THESIS NO : PAS075MSDGE014

"Probabilistic load flow analysis using point estimation method: A case study of
Tandi distribution feeder"

by

Sameep Poudel

A THESIS

SUBMITTED TO THE DEPARTMENT OF ELECTRICAL ENGINEERING

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF SCIENCE IN ELECTRICAL ENGINEERING IN
DISTRIBUTED GENERATION

DEPARTMENT OF ELECTRICAL ENGINEERING

POKHARA, NEPAL

Bhadra, 2078
Probabilistic load flow analysis using point estimation method: A case study of
Tandi distribution feeder"

by

Sameep Poudel

Roll No: PAS075MSDGE014

Thesis Supervisor

Associate Professor Bhrigu Raj Bhattarai

Institute of Engineering, Pashchimanchal Campus

A thesis submitted to the Department of Electrical Engineering in partial fulfillment

of the requirements for the degree of

Master of Science in Electrical Engineering in Distributed Generation

Department of Electrical Engineering

Institute of Engineering, Pashchimanchal Campus

Tribhuvan University

Pokhara, Nepal

Bhadra, 2078

ii
 COPYRIGHT ©

The author has agreed that the Library, Department of Electrical Engineering,
Pashchimanchal Campus, Institute of Engineering may make this report freely available
for inspection. Moreover, the author has agreed that permission for extensive coping of
this thesis report for scholarly purpose may be granted by the supervisor who supervised
the thesis work recorded herein or, in their absence, by the Head of the Department
wherein the thesis report was done. It is understood that the recognition will be given
to the author of this report and to the Department of Electrical Engineering, Institute of
Engineering, Pashchimanchal Campus in any use of the materials of this thesis report.
Copyright or publication or other use of this report for the financial gain without
approval of the Department of Electrical Engineering, Institute of Engineering,
Pashchimanchal Campus and author’s permission is prohibited. Request for the
permission to copy or to make any other use of the material in this report in whole or
in part should be addressed to:

Head

Department of Electrical Engineering

Pashchimanchal Campus, Institute of Engineering

Pokhara, Nepal

iii
CERTIFICATE OF APPROVAL

iv
 ACKNOWLEDGEMENT

On the very outset of this thesis report, I would like to extend my sincere obligation
towards the Department of Electrical Engineering, Pashchimanchal Campus for
providing me the opportunity to pursue my Master’s Degree in Electrical Engineering
in Distributed Generation.

I feel it as my greatest privilege to work under the esteemed guidance of my Supervisor

Associate Professor Bhrigu Raj Bhattarai and M.Sc. program coordinator Assistant
Professor Menaka Karki, Department of Electrical Engineering, Institute of
Engineering, Tribhuvan University. I would like to express my sincere gratitude to
HOD of Electrical Department of Paschimanchal campus, Assistant Professor Ram
Prasad Pandey for his guidance and encouragement. I would also like to express my
gratitute to Assistant Professor Shahabuddin Khan ,Er. Samundra Gurung, Er. Prabhat
Pankaj Er. Nitesh Yadav and Er. Shiva Paudel for their continuous supports and
invaluable suggestions during my graduate program and support during each step when
I feel difficult to move forward to pursue and present this thesis. Finally, I would like
to express my indebtedness to my parents, friends, Faculty member who directly and
indirectly help me for their continuous encouragement, love, wishes and support to
complete this thesis successfully. I have no valuable word to express my thanks, but
my heart is still full of favor received from everyone.

v
 ABSTRACT

With the increase penetration of distributed generation(DG) at various location, the

uncertainty factor in the distribution system has increases rapidly. These uncertainty
factor play a virtal role in maintaining the system reliability. The use of deterministic
load flow(DLF) is unable to show the uncertainty operation of the grid since
deterministic load flow only consider the fixed input variable and change in input
random variable that can be obtained from solar is not consider by the deterministic
load flow so Probabilistic load flow(PLF) analysis is used, to reflect the uncertainty of
DG and load. There are various types of PLF and the most common method is Monte-
carlo simulation(MCS) method. Monte- carlo simulation reflect the uncertainty of the
input variable by the repeated Deterministic load flow of each input random variable.
Deterministic load flow is done using the square root method of load flow. since each
input random variable is individually computed so on increasing the input random
variable increases the accuracy and also the computation time. Hence to achieve the
accuracy large number of computation time is required and if the system is big with
large number of buses then the use of Monte- carlo simulation is inefficient.

To overcome the drawback of Monte-carlo simulation method, point estimation

method(PEM) is proposed here. The Result obtained from the two probabilistic method,
Point-estimation method and Monte- carlo simulation is compared in IEEE-33 Bus
Radial Distribution system. The accuracy of Monte carlo simulation is also achieved
by the point estimation method with very less computational time compared to MCS.
Before implementing Probabilistic load flow using PEM in Integrated Nepalese Power
System(INPS) of Tandi Feeder, it is first tested in IEEE-33 Bus radial distribution
system using the standard data. Bus number 18 and 33 is selected for the DG penetration
of IEEE-33 Bus system. The selection of Bus is done based on least pu voltage and end
of the lateral branch. The effect on pu voltage, power loss before and after DG
penetration is analysed. All the work is done using the MATLAB 2020a software.
Active Power produced by the DG(solar) and Active and reactive power of the load is
taken as uncertainty factor for the study.

After successful completion of using point estimation method on IEEE-33 bus system,
Tandi distribution system as a real feeder with 122 buses is taken for the analysis using
Point Estimation method. TOD load data of Tandi feeder is collected from Tandi DCS

vi
and parsa substation. Hourly variation of solar irradiance from 9am to 5pm of each day
measured by pyranometer installed at Gairapatan, Pokhara based on beta distribution is
taken as real data for PV penetration of tandi feeder. Gram charlier expansion with
chebyshev hermite polynomial equation is consider to calculate the Probability Density
function(PDF) and Cumulative distribution function(CDF) of the random variable.
CDF of pu voltage and active power loss before and after DG penetration is compare
and analysed. Solar as a DG is connected at the different Buses and Probability of over
voltage and under voltage at different bus before and after DG penetration is analysed.
DG location is selected based on the least pu voltage and end of the lateral branch. The
result show the effect of DG penetration on Voltage and power loss of the system.

vii
 TABLE OF CONTENTS

COPYRIGHT © ....................................................................................................... iii

CERTIFICATE OF APPROVAL ............................................................................. iv

ACKNOWLEDGEMENT ......................................................................................... v

ABSTRACT ............................................................................................................. vi

LIST OF TABLES .................................................................................................... x

LIST OF FIGURES .................................................................................................. xi

LIST OF ABBREVIATIONS ................................................................................. xiii

CHAPTER ONE: INTRODUCTION ........................................................................ 1

1.1 Background ................................................................................................. 1

1.2 Problem statement ....................................................................................... 3

1.3 Objective ..................................................................................................... 4

1.4 Scope and limitation .................................................................................... 4

1.5 Report organization ..................................................................................... 5

CHAPTER TWO: LITERATURE REVIEW ............................................................. 6

2.1 Distribution System ..................................................................................... 6

2.2 Radial Distribution System .......................................................................... 6

2.3 Distributed generation ................................................................................. 7

2.4 NEA interconnection Guideline for PV........................................................ 8

2.5 Deterministic load flow ............................................................................. 10

2.6 Monte carlo simulation .............................................................................. 13

2.7 Probabilistic load flow with Point Estimation method ................................ 13

CHAPTER THREE: METHODOLOGY ................................................................ 15

3.1 The probabilistic model of Distribution system .......................................... 15

3.1.1 The probabilistic model of Photovoltaic DG ....................................... 15

viii
 3.1.2 The probabilistic model of load .......................................................... 16

3.2 Point estimation method ............................................................................ 16

3.2.1 Two point estimate (2PEM) ................................................................ 17

3.2.2 Three point estimate (3PEM) .............................................................. 17

3.2.3 Standard central moment .................................................................... 18

3.3 Cumulant Method ...................................................................................... 19

3.4 Gram-Charlier Expansion .......................................................................... 20

CHAPTER FOUR: RESULTS AND DISCUSSIONS ............................................ 23

4.1 Result of IEEE-33 Bus System .................................................................. 23

4.1.1 Point Estimation Method .................................................................... 25

4.1.2 Comparison Between MCS and PEM ................................................. 28

4.1.3 Point Estimation Method with DG ...................................................... 29

4.2 Result of tandi feeder using PEM.............................................................. 31

4.2.1 General layout of Tandi feeder ........................................................... 31

4.2.2 Normally Distributed Data.................................................................. 33

4.2.3 Point Estimation Method .................................................................... 34

4.2.4 Point Estimation Method with DG penetration.................................... 39

CHAPTER FIVE: CONCLUSION AND RECOMMENDATION ......................... 44

5.1 Conclusion ................................................................................................ 44

5.2 Recommendation ....................................................................................... 44

REFERENCES ........................................................................................................ 45

APPENDIX- 1 IEEE-33 BUS SYSTEM .................................................................. 49

APPENDIX- 2 TANDI FEEDER ........................................................................... 50

APPENDIX- 3 CENTRAL MOMENT .................................................................... 58

ix
 LIST OF TABLES

Table 1. 1 Comparison of computation time [11]. ...................................................... 3

Table 4.1 Specification of laptop……………………………………………………..23

Table 4.2 Parameter of IEEE-33 bus system ............................................................ 23

Table 4. 3 Power flow of IEEE-33 bus system. ........................................................ 25

Table 4. 4 Comparison Table between MCS and PEM. ............................................ 28

Table 4. 5 Comparative Table after Penetration using PEM. .................................... 30

Table 4. 6 Key Parameter of Tandi feeder. ............................................................... 32

Table 4. 7 Parameter of Load data............................................................................ 33

Table 4. 8 Minimum and Maximum voltage of Tandi feeder. ................................... 34

Table 4. 9 Power flow detail of Tandi Feeder. .......................................................... 37

Table 4. 10 Beta Distribution Parameter of PV. ....................................................... 40

Table 4. 11 Comparison of Parameter after DG penetration. .................................... 42

x
 LIST OF FIGURES

Figure 2. 1 Radial Distribution system [12]…………………………………………...7

Figure 2. 2 DG as PV connected on distributed network [15]. ................................... 8

Figure 2. 3 Wiring diagram for grid connected solar PV system [19]. ........................ 9

Figure 2. 4 A branch of Radial Distribution Feeder. ................................................. 10

Figure 2. 5 flow chart of Deterministic Load flow.................................................... 12

Figure 3. 1 Probabilistic load Flow using Point Estimation method…………………21

Figure 4. 1 Single line diagram of IEEE-33 Bus system……………………………..24

Figure 4. 2 Pu voltage plot of IEEE-33 Bus system .................................................. 24

Figure 4. 3 Active power loss of IEEE-33 Bus system. ............................................ 25

Figure 4. 4 PDF curve of Bus 18 and 33 .................................................................. 26

Figure 4. 5 CDF curve of Pu voltage of bus 18 and 33. ............................................ 26

Figure 4. 6 Minimum and maximim voltage of IEEE-33 Bus system ....................... 27

Figure 4. 7 Comparison of CDF plot at Bus 18 with 25% Penetration ...................... 29

Figure 4. 8 Comparison of Pu voltage after 25% DG Penetration. ............................ 29

Figure 4. 9 Comparison of Active Power Loss after Penetration. .............................. 30

Figure 4. 10 Single Line diagram of Tandi Feeder. .................................................. 31

Figure 4. 11 Daily Load curve of Tandi feeder ......................................................... 32

Figure 4. 12 Histogram plot of TOD data at Bus 15 ................................................. 33

Figure 4. 13 Voltage magnitude plot of Tandi feeder. .............................................. 34

Figure 4. 14 Minimum and maximum voltage plot of Tandi feeder. ......................... 36

Figure 4. 15 Under voltage probability of tandi feeder ............................................. 36

Figure 4. 16 CDF plot of Pu voltage of Tandi feeder. ............................................... 37

Figure 4. 17 CDF plot of Active Power Loss of Tandi feeder ................................... 38

Figure 4. 18 CDF plot of Reactive Power loss of Tandi feeder. ................................ 38

xi
Figure 4. 19 Intermittency of Solar Power Generation ............................................. 39

Figure 4. 20 Comparison of Pu voltage plot with 25% DG at different Bus ............. 40

Figure 4. 21 Comparison of CDF plot of Pu voltage at Bus 120 ............................... 40

Figure 4. 22 Comparison of CDF plot of pu voltage at Bus 71 ................................. 41

Figure 4. 23 Comparison of CDF plot active power loss at Branch 4. ....................... 42

xii
 LIST OF ABBREVIATIONS

CDF Cumulative Density Function.

DG Distributed Generation.

DLF Deterministic load flow.

IEEE Institute of Electrical and Electronic Engineering.

INPS Integrated Nepalese Power System.

MATLAB Matrix Laboratory.

MCS Monte- carlo simulation.

PDF Probability Density Function.

PEM Point Estimation Method.

PLF Probabilistic load flow.

PPF Probabilistic Power flow.

PV Photo-Voltaic.

2PEM Two point Estimation Method.

3PEM Three point Estimation Method.

RAM Random Access Memory.

TOD Time of Day.

xiii
 CHAPTER ONE
INTRODUCTION

1.1 Background

Rapid development of renewable energy resources play a vital role to control carbon
emission. The increasing use of energy creates a challenges for energy saving as a result
use of renewable energy resources has increases in the power system rapidly. with the
development of new technology and innovation, the use of solar energy has increases,
but due to the fluctuating nature of solar output operation and penetration problem
arises [1]. The stochastic nature of solar output is generally due to maximum generation
at the day time and no generation of power during the night. Solar output also vary
seasonally, more at spring than the winter.The climatic condition and meteorological
feature of different location also causes uncertainty generation of the solar in the
distribution system [2].

The integration of Distributed generation(DG) in the power system creates the

randomness in the power output, hence the use of deterministic load flow cannot trace
the uncertainty of power generated by the DG. The use of Probabilistic load flow(PLF)
for such uncertainty of power generation consider all the randomness of the power
generated by the DG and output of Bus voltage and line current is obtained for each
value of photo-voltaic(PV) generation in terms of undervoltage and overvoltage. PLF
is also applied in the field such as renewable energy generation,risk assessment,
stochastic in programming [3]. Hence use of PLF for further research and development
is of great importance [4].

There are normally three method for PLF calculation as analytical method, simulation
method and approximation method. Simulation method mainly include Monte-carlo
simulation(MCS) [5] this changes the each random variable into the sequence of
deterministic load flow problem. MCS can be consider as the most accurate PLF
method among various other method so it is commonly used for analysing the accuracy
of various other probabilistic method [3]. The accuracy of MCS is achieve by the use
of large number of random sampling and iteration calculation which makes this method
computationally inefficient and time consuming hence for the large system with many
buses MCS is not used [6]. so to overcome this drawback, PEM can be used.

1
Analytical method include the cumulant method and convolution method.Analytical
method less computational burden with references to simulation method as it consider
the number of problem into the mathmatical assumption to obtain the accurate
information of output random variable [7].

Point estimation is the method of using sample data to calculate a single value (Point
estimate as it identifies a point in some parameter space). Point estimation uses the
value from the sample of unknown parameter to get the best fit of the output data. Each
sample data from the large number of population is use by the point estimation for the
output parameter. random sample of the population is used to approximate the value
using the mathmatical equation and assumtion to get the best fit of the output data.

There are two types of estimation as point estimation and internal estimation, point
estimation produces a single value whereas internal estimation produces a range of
value [1].

Point estimation method have good balance of accuracy and computational burden. It
need few deterministic power flow runs (it required less number of deterministic
iteration for same number of input variables). Mathmatical equation and derivation are
simple and it require less information of input random variable as compared to other
method [8-9].

Another most popular method to solve the probabilistic power flow compare to MCS
method is PEMs. Mean, variance, skewness and kurtosis are some statistical moment
used as probabilistic function for PEM. Along these 4 central moment, other 2 central
moment in terms of polynomial function can also be used in PEM for accuracy of the
method. PEM is superior than MCS method in the sense that PEM require less data,
less computational burden and it takes less computational time [7]. The accuracy of
PEM can also be improved by increasing the number of estimation point 'm' so it can
be said that 3PEM is more accuracy than 2PEM and less number of estimation point
lead to inaccuracy. It should be noted than on increasing the estimation point the higher
order moment is required but higher order moment greater than 4th order has no physical
meaning [4]. The use of higher order estimation point may arises the following
problem;

1) It is difficult to obtain the higher order central moment.

2
 2) standard location and weight for m> 3 cannot be determined.
3) Higher order central moment may give negative or non real value of location
and weight.

So for improve the accuracy only large number of estimation point shouldnot be used
and on considering the less estimation point arises large inaccuracy. Hence considering
the optimum condition, 3PEM is used in my thesis work.

1.2 Problem statement

The availability of generation at various location and due to high demand of power
system distributed generation are connected to the network. The power output of
distributed generation like PV highly varies as the source of energy is also vary. On
integrating such network to the grid proper analysis must be done for proper operation
of the system. on using deterministic load flow various uncertainties are not consider.

For each value of bus loads, generation patterns and network topologies, large number
of computation effort are required so the use of conventional load-flow computation
seems impractical.

For such stochastic generation of the power probabilistic load flow analysis is accurate
and more effective as it consider all the uncertainties like uncertainty in load demand,
generation outage.

Large number of sample data and iteration are use by the monte carlo simulation for
each of the random output data which make the Monte-carlo simulation time consuming
and computationally inefficient for best result. For the large system on using Monte
carlo simulation require computer having high processor, greater capacity of RAM and
more memory space with large computational time. Such drawback of MCS can be
overcome by the PEM. MCS method is seems less accurate for radial distribution
system than PEM.

Table 1. 1 Comparison of computation time [11].

Parameter MCS PEM

Time(sec) 384.56 8.564

3
1.3 Objective

Main Objectives

• To Perform probabilistic load flow of Tandi Distribution feeder using point

estimation method.

Specific Objectives

• To use IEEE-33 bus system for probabilistic load flow using PEM.
• To compare the result of Point estimation method and Monte-carlo simulation
method in IEEE-33 Bus system.

1.4 Scope and limitation

The various scope of this thesis include;

• Probabilistic Load flow of IEEE-33 bus and Tandi distribution system having
122 buses using Point estimation method with PV penetration is done in
MATLAB environment.
• The Point Estimation method with deterministic load flow is done using Square
root method.
• Maximum load is consider for deterministic load flow of tandi feeder.
• Hourly variation of solar irradiation and load is modeled by beta and normal
distribution respectively.
• The various cases for simulation are no PV generation with normally distributed
load, 25% PV generated of total load at different bus to generate PDF and CDF
of voltage and power loss.

The various limitation are;

• Irradiance data of 4 month measured by pyranometer at pokhara is used for PV

penetration.
• TOD load data of 22 Buses are taken as a reference for 122 buses of Tandi
feeder.
• The power storage system(battery) is not connected to our system, so
intermittancy problem may arises.
• The effect on protection system on network is not studied.

4
1.5 Report organization

This thesis consist of following chapters:

• Chapter Two describe the literature review of distribution system, load flow
analysis and probabilistic load flow of power system with PV interconnection
is also discussed here. Beside this the use of point estimation method over the
monte carlo simulation method is also discussed in this chapter.
• Chapter Three describe the methodology of the work, to meet the require
objective. It include the probabilistic model of PV and load, flow chart, point
estimation method and detail about the Gram-charlier expansion.
• Chapter Four Provide the result obtained from the MATLAB environment. It
include the result of load flow and point estimation method with PV connection
on IEEE-33 Bus Radial Distribution system. The result of Point estimation
method and monte-carlo simulation on IEEE- 33 Bus system is compare and
analyse here. The result of Tandi Distribution feeder of 122 buses with PV
interconnection using point estimation method is also shown here.
• Chapter Five conclude the result and present the further work to do.

5
 CHAPTER TWO
LITERATURE REVIEW

Literature review show the need for additional research. It helps to gain and clearify the
knowledge on the particular subject. It is the initial stage for all research activities as
unsolved problem can be easily found and further research can be continue. some
research activities also help to solve the controversey in a particular field. Research
makes clearification of any matter with justification as well as theoritical or practical
proof.

2.1 Distribution System

After the electricity is generated at the power station, it is then transmitted through the
transmission system followed by the distribution system to the end user. The electricity
is transmitted at high voltage at the transmission system which is then step down and
transmitted through the distribution system.

A drop in voltage levels results when demand for electricity exceeds the capacity of the
distribution system. Medium voltage power is transmitted to the distribution
transformers near the customer location through primary distribution lines. Distribution
transformers again lower the voltage by step down transformer to the utilization voltage
which is then passes through the secondary distribution line to the customer [10].

There are three types of distribution system as;

a. Radial distribution system

b. Loop distribution system
c. Network distribution system

Radial distribution system is more applicable as loop and network system are more
expensive because more switches and more conductors is necessary to construct the
loop and network system.

2.2 Radial Distribution System

In Radial distribution system power is received by as single incoming substation at the

utility supply voltage. The power is then split and converted to individual end user.
Distributor are feed from only one end in radial distribution system from the substation.

6
The power from the main branch is split to sub branch and again split out from the sub
branches in radial distribution system [10]. The cost of radial distribution system is
low compare to other network so it is less reliable although radial distribution network
is prefer when substatin is located at the center of the load and due to its low cost,
simple construction and easy to expand. Radial distribution system is mostly used in
densely populated areas [11].

Figure 2. 1 Radial Distribution system [12].

2.3 Distributed generation

Distributed generation is also called as decentralized generation or on-site generation.

Distributed generation mainly refer as generating electricity on-site or near the
consumer premises. It avoid the transmission of electricity over the electric grid or long
distance transmission. Distributed Generation can be PV, wind, biomass, hydropower
etc.

DG is penetrated in the distribution system according to the requirement of generation

of real and reactive power. The various types of DG categories as their generation
capacity are [13];

1) Only generate active power eg solar.

2) Only generate the reactive power eg Capacitor bank, synchronous condensers.
3) Generate both active and reactive power eg Wind turbine.
4) Maintain Bus voltage. Here reactive power will generated or consumed to
maintain the Bus voltage.

7
Solar is one of the mostly used Distributed Generation in power system. The penetration
of Distributed generation in the distribution system helps to improve the Voltage profile
of the system, decreases the transmission losses and also impove the efficiency. DG
penetration can replace the power transformer, distribution station, can break the power
monopoly, can supply electricity in remote areas, reduced transmission loss, reduced
price volatility, reduced transmission and distribution congestion [14].

Figure 2. 2 DG as PV connected on distributed network [15].

The high penetration of PV in the distributed network is limited by the fluctuating

nature of solar output, protection co-ordination of the system. similarly tripping of the
PV unit during the disturbance also limit the PV penetration on the distributed networks
[16].

PV array is the combination of PV module and are connected in series and in parallel.
The process of obtaining electricity directly from light energy is solar or PV generation
system. Solar mainly depend on photovoltaic effect so it is also called photovoltaic cell.
Electricity is produced from the semico nductor solar cell when light strikes on it and
potential different product across it terminal. The potential different is normally
maintained at 0.5 volt and voltage is independent to intensity of incident light whereas
current depend on intensity and area of the solar panel [17].

2.4 NEA interconnection Guideline for PV

There are many organization and researching center that has certain standards around
the world for PV integration such including the US, Germany, Spain, etc. In the context
of Nepal, according to National Energy Crisis Control and Electricity Development
Decade, 2072 S.N 52’B’, 500 watt or above Solar PV can be connected to the national
grid through Net-Metering system. Net Metering should be bidirectional meter capable
of reading both import and export of electricity. For Net-Metering, Customers must

8
submit an application to the respective Distribution Centers along with Single Line
Diagram [18]. NEA Standards of quality of power of solar PV to be supplied by
consumer to Grid should be of following:

• Frequency – 50 Hz

• Voltage – 230/400V /11KV ±5%

• Voltage waveform must be sinusoidal, Total Harmonic Distortion ≤ 3%

• For Three phase system voltage unbalance tolerance level ±1%

• The power generated by the customers solar PV should have power factor between
0.85 lag to 0.95 lead.

• The fault clearance time for earth fault, overcurrent and frequency fluctuation should
be of standard described by IEC/IEEE

Capacity of solar PV to be connected according to following voltage levels:

• 230V up-to 5 kW

• 400V 5 kW -40 kW

• 11kV above 40 kW

Figure 2. 3 Wiring diagram for grid connected solar PV system [19].

9
2.5 Deterministic load flow

The point estimation method is applied on IEEE 33 bus radial distribution system. The
load flow of such a system is carried out Using Square root method [20].

Let us consider a radial Distribution system with two node as shown in figure 3.2;
V2∠∂2
V1 ∠∂1 P2+ jQ2

R+jX
I2
jQ2 Node 1 Node 2

Figure 2. 4 A branch of Radial Distribution Feeder.

where,

Node 1 = sending end Node

Node 2 = Receiving end Node

P2 +jQ2 = Power at Bus 2

R + jX = Branch Impedance.

V1 , V2 = Voltage magnitude

Then,

P2 +jQ2 = V2 I2* …(2.1)

𝑃2 −𝑗𝑄2
I2 = ...(2.2)
𝑉2∗

Also,
|𝑉1 |∠ ∂1 −|𝑉2 |∠ ∂2
I2 = …(2.3)
𝑅+𝑗 𝑋

Solving Equation (2.2) and (2.3) we get,

𝑃 𝑋− 𝑄 𝑅
∂2 = ∂1 -tan-1(|𝑉 |22+𝑃 𝑅+𝑄
2
) …(2.4)
2 𝑋 2 2

and

10
|V2| =

1
√[{(𝑃2 𝑅 + 𝑄2 𝑋 − 0.5|𝑉1 |2 )2 − (𝑅2 + 𝑋 2 )(𝑃2 + 𝑄2 )}2 − (𝑃2 𝑅 + 𝑄2 𝑋 − 0.5|𝑉1 |2 )]
2 2

…(2.5)

From above equation 2.4 and equation 2.5 gives the angle and the magnitude of the
Voltage respectively at the receiving end Bus.

The above equation for voltage magnitude and phase angle at receiving end node can
be written in generalized form as;

V(m2) =|B(j)-A(j)|1/2

Where,

A(j) = P(m2)*R(j) + Q(m2)*X(j) – 0.5|V(m1))2

B(j) = {A2(j)-[R2(j)+X2(j)]*[P2(m2)+Q2(m2)]}1/2

and
𝑃(𝑚2 )∗𝑋(𝑗)−𝑄(𝑚2 )∗𝑅(𝑗)
∂(m2) = ∂(m1) -tan-1[𝑃(𝑚 2 (𝑚 ]
2)∗𝑅(𝑗)+𝑄(𝑚2 )∗𝑋(𝑗)+𝑉 2)

similarly,

Real and Reactive power loss of respective branch can be calculated as;

𝑅(𝑗)∗[𝑃 2 (𝑚2)+𝑄2 (𝑚2 )]

LP(j)=
|𝑉(𝑚2)|2

𝑋(𝑗)∗[𝑃 2 (𝑚2 )+𝑄2 (𝑚2)]

LQ(j)= |𝑉(𝑚2 )|2

In the above equation,

j= Branch Number.

m1 = sending end node.

m2 =Receiving end node.

P = Active Power.

Q= Reactive Power.

11
 start

Initialize value

Set convergency
criteria

Calculate load at each node

Find voltage and power loss

Update value

Yes
Max error>
Tolerance

No

Display Result

Stop

Figure 2. 5 flow chart of Deterministic Load flow

The deterministic load flow of radial distribution system of IEEE-33 bus system and
real tandi distribution feeder is done using the square root method. The input variable
are line data which contain value of resistance and inductance, and Bus data which
contain the load at each buses. Initially the pu voltage is set one and voltage angle del
is set zero. The load flow is done to calculate the active power, reactive power, voltage
and voltage angle till the value is converge and maximum error reaches below the
tolerance value. finally the result are display and iteration of load flow stop. The detail
of deterministic load flow shown in flow chart is shown in the figure 2.5.
12
2.6 Monte carlo simulation

Monte Carlo simulation is the computational algorithm technique used in random

sampling to obtain the numerical results [21]. Monte Carlo is normally used for three
different cases as;

• Optimization
• Numerical integration
• Probability distribution.

Here Monte Carlo simulation is based on Probability distribution of randomness of

solar output.Solar output is taken as input variable for the Monte-carlo simulation [21].
The varying nature of solar input is obtained by PDF using the beta distribution. Monte
Carlo simulation is done in MATLAB for large number of iterations for accuracy.

2.7 Probabilistic load flow with Point Estimation method

The probabilistic method is mainly based on the uncertainty in the system due to the
stochastic generation of PV. The review of several paper based on point estimation
method has been described below;

Yulong Che[4] estimate the probabilistic moment of Probabilistic power flow using the
improved three point estimation method. IEEE-14 bus system and IEEE-118 buses with
normally distributed input variable are used for two point estimation, three point
estimation and Chebyshev inequality to obtained the result of improved three point
estimation method.

M. Fan [7] analyse the impact of uncertainties of PV generation in transmission system.

To analyse the probabilistic model of PV generation and load, cumulant method is used
here and the obtained result are compared with the monte carlo simulation method.The
result show that the generation dispatch behaviour reduces the uncertainty influence of
PV generation.

Chun Lien Su [8] is based on the probabilistic load flow analysis using point estimation
method. IEEE 6 bus system is used and result are verified using the monte carlo
simulation method. If the stochastic nature of DG output is estimated, all the line
parameter and power flow can be accurately evaluated by using 2PEM through
mathmatical equations. However, more estimation point can be taken for better result.

13
J. Emmanuel Anandraj [11] is based on probabilistic load flow of distribution network
with PV unit. Active power produced by PV and power absorbed by load are taken as
uncertainties. Probabilistic load flow is done in MATLAB Software in 30 bus radial
network using two point estimation and three point estimation scheme and obtained
result are compared using Monte- Carlo simulation method to show that three point
estimation method is better than two point estimation method.

Delgado [22] said that the variation in load demands and power fluctuation of
renewable generators cannot be considered with the deterministic power flow because
it uses specific power values so probabilistic power flow method is used here that reflect
the uncertainty of input variables. This paper extends the point estimation
method(PEM) with dispersed generation and variable power factors of wind and solar.
This paper combine the 3 point estimation scheme with cholesky decomposition to
estimate the cumulative distribution function. This system is tested on the three phase
unbalanced IEEE 123 node test system and result are compared with those obtained
result from the monte carlo simulation method.

B. K. Panigrahi [23] present the Two schemes of PEM (2m and 3m) and their
performance is evaluated by comparing it to the Monte-Carlo based PPF. Point
estimation method is applied in transmission system (30-Bus) and radial distribution
system(33-Bus). The obtained result from both the method show that the point
estimation method has better perfomance in radial networks as compare to monte carlo
simulation method.

It can be summaries from the different paper of literature review that stochastic
behaviour of DG cannot be trace by the deterministic load flow and so the probabilistic
method of load flow seems practical and accurate. The performance of point estimation
method is more accurate for radial distribution system than Monte carlo simulation
method. The output of MCS and PEM is almost same but MCS require more
computational time than PEM. The accuracy of PEM can be maintain by increasing the
number of estimation point. Hence 3 point estimation is more accurate than 2 point
estimation method and require only one more computation time than 2 point estimation
method.

14
 CHAPTER THREE
METHODOLOGY

The methodology show the step by step procedure to achieve the objective of the work.
Literature review show the background information and detail about the point
estimation method. The work is divided into two part;

1) working with IEEE-33 bus system data

2) working with Real feeder of Tandi distribution.

Data of real feeder are collected from the Tandi Distribution center and Parsa substation
which include the line data and Bus data of Tandi Distribution feeder with 122 buses
located at Chitwan district of Nepal.

3.1 The probabilistic model of Distribution system

3.1.1 The probabilistic model of Photovoltaic DG

Due to stochastic nature of solar irradiance(Ib ) the power output obtained from the solar
vary at different time. The variation of solar irradiance and hence the power generated
by the solar can be consider to follow the beta distribution [24]. The equation for power
output of the solar is given by;

PG = η*Ac*Ib

where,

PG =Power generated by the PV system(KW).

Ib = solar irradiance on the surface(KW/m2)

η = efficiency of the PV system

Ac =Array surface area(m2).

For power generation by the PV, the probabilistic density function is given by [25];
Γ (α +β)
fPG(PG ) = Γ (α )∗Γ (β) (Ib ) α-1 *(1-Ib)β-1

for PG ∈ [0,PG(Ib)] ,0≤ Ib≤ 1, α > 0 and β > 0 otherwise 0.

Where;

15
fPG is the probability density function for power generated; α,β denote the shape
parameters of beta distribution with 12.62 and 2.21 respectively[11] and Γ is the gamma
function.

As per the ontario's standard the percentage of penetrated DG unit is limited to less than
30% [26]. According to [27] the best penetration level is shown 21% for IEEE -33 bus
distribution system considering only two factors, power loss and voltage regulation and
other factor such as economic and geographic are not considerations.

3.1.2 The probabilistic model of load

The power ( PL) absorbed by load at any load bus is random and assumed to follow a
normal distribution with certain value of mean and standard deviation [11]. The
probability density function of PL is given by:

−(𝑃𝐿−μ𝑃𝐿 )2
1 2∗(𝜎𝑃𝐿 )2
fPL(PL) = σ * 𝑒 ...(3.1)
𝑃𝐿 ∗√2𝜋

Where;

PL is the load power[kW] ,

fPL is probability density function for load power,

μPL denote the mean value and σPL is the standard deviation.

Equation 3.1, show the variation of load active power. Load reactive power is also
follow normal distribution function and PDF for reactive power can also be obtained
from the equation 3.1 [11].

3.2 Point estimation method

Few point of random variable are consider in calculation the statistical information or
central moment in the point estimation method. Using these point (concentration),
output information is obtained [28].

The various uncertain parameter are the power generated by the solar and normally
distributed load active and reactive power, these uncertain parameter are consider
random variable. For three point estimation method, the concentration of random
variable can be calculated as [28];

ρl,k = µρl + ℇl,k * σρl

16
Where;

ρl,k is the th kth concentration of ρl random variable,

µρl is the mean of random variable,

σρl is the standard deviation of random variable and

ℇl,k is the kth standard location of random variable.

The standard location ℇl,k and weight Wl,k of input random variable is obtained by
solving the below given non-linear equations 3.2 and 3.3 by using the equation derived
by the Miller and Rice [29].
1
∑𝑘𝑘=𝑙 𝑊 l,k = …(3.2)
𝑚

∑𝑘𝑘=𝑙 𝑊 l,k (ℇl,k) j = 𝜆 l,j j= 1,2,…. 2k-1 …(3.3)

Where,

m is the total number of input random variables, i.e. the uncertain generator power and
the active and reactive load powers; 𝜆 l,j denotes the jth standard central moment of the
random variable ρl . The concentration point ρk,l may move away from the mean value
µρ. so in three point estimation method one of the concentration is set to mean to
maintain the accuracy [28].

3.2.1 Two point estimate (2PEM)

For two point estimation method each random variable consider only the two point. The
statistical information can be obtained from the same equation 3.2 and 3.3 at K=2.
standard location and weight are calculated from skewness factor λl,3. Here output
move away from the mean value µρ as variation of input variable changes the vaue of
standard location and concentration Hence uniformity is not obtained. This is overcome
by three point estimation method by considering one concentration as mean value [28].

3.2.2 Three point estimate (3PEM)

In case of 3PEM one concentration points out of three is taken as mean value by
considering one of the standard locations ξl,k as zero. Hence K = 3 is taken for 3PEM.

17
statistical information is obtained from equation 3.2 and 3.3. The standard locations ξl,k
and weight Wl,k of random variable can be obtained by;

λ𝑙,𝑘 3
ξl,k = +(-1)3-k √( λ𝑙,4 − λ 2𝑙,3 for k=1,2
2 4

ξl,3 = 0;

(−1)3−𝑘
wl,k = ξ for k=1,2
𝑙,𝑘 (ξ 1,𝑙 −ξ 2,𝑙 )

1 1
wl,3 = (𝑚 − )
λ𝑙,4 − λ2𝑙,3

Three point estimation method is more accurate than two point estimation method as
input random variable 'm' do not depend on standard location [11].

3.2.3 Standard central moment

The standard central moment is denoted by 𝜆 l,j . It is useful in comparing the shape of
different probability distribution. The moment of the probability distribution about the
mean is the central moment. central moment show the variation of data from the first
raw moment [11]. The various central moment are;

𝜆 l,1 = Mean of random variable;

𝜆 l,2 =Standard Deviation of random variable;

𝜆 l,3 = Skewness of random variable.

𝜆 l,4 = Kurtosis of random variable.

where,

Skewness measure the symmetry of the distribution and study the shape of the curve.
3∗(Mean−Median)
Skewness = standard deviation

Kurtosis show the differences between the tail of the distribution with the Normal
Distribution.

1∗∑𝑛
𝑘=𝑙 (𝑋𝑘−𝑋𝑚𝑒𝑎𝑛)^4
Kurtosis= 𝑛∗σ

where,

18
n = No. of random variable.

σ= Standard deviation.

On considering Tandi distribution feeder, Load is assumed normal distribution and PV

as beta distribution so Active and reactive power are consider as random variable. The
different central moment of tandi feeder is obtained from the TOD load data is shown
in Appendix-3.

3.3 Cumulant Method

Cumulant method is an efficient method to generate the PDF of random variable. This
method has less computational burden than the convolution method. The probability
distribution having same moment have identical cumulants so moment determine the
cumulant, similarly cumulant also determine the moments. Different order cumulant
can be obtained using the cumulant generating function [22] as shown below;

1st cumulant = mean

2nd cumulant = variance

3rd cumulant = 3rd central moment

The six order cumulant obtained from the six raw moment are;

k1 = µ1

k2 = µ2 - µ12

k3 = µ3 -3µ2µ1 + 2µ13

k4 = µ4 – 4µ3µ1 -3µ22 +12µ2µ12 – 6µ14

k5 = µ5 -5µ4µ1 -10µ3µ2 +20µ3µ12 + 30µ22µ1 -60µ2µ13 + 24µ15

k6 = µ6 – 6µ5µ1 -15µ4µ2 + 30µ4µ12 – 10µ32 + 120µ3µ2µ1 – 120µ3µ13 + 30µ23

-270 µ22µ12 + 360 µ2µ14 – 120µ16

Four order cumulant is normally used, but for accuracy six order cumulant is used here
in my work. The result of the cumulant is used to obtained the PDF and CDF curve
using the Gram- charlier expansion.

19
3.4 Gram-Charlier Expansion

The density of Gram−Charlier distribution is the polynomial times the normal density.
Gram-charlier expansion find the probability distribution of random variable in term of
its cumulant. If x is mean value and µ is standard deviation σ of random variable, the
standardized variable x̅ has the form x̅ = (x−µ)/ σ . According to Gram-Charlier
expansion, the cumulative density function F (x̅) and probability density functions f(x̅)
can be written as [11];
𝐶1 𝐶2 𝐶3
F(x̅) = 𝜙(x̅) + 1! 𝜙′ (x̅) + 2! 𝜙′′(x̅) + 3! 𝜙′′′(x̅)

𝐶1 𝐶2 𝐶3
f(x̅) =𝜑(x̅) + 1! 𝜑′(x̅) + 2! 𝜑′′(x̅) + 3! 𝜑′′′(x̅)

Where,

𝜙(x̅) and φ(x̅) represent the CDF and PDF of the standard normal distribution with µ
= 0 and σ = 1, respectively, cv constant coefficients with the following form.

co = 1

c1 = c2 = 0

c3 = -β3/σ3

c4 = β4/σ4 -3

c5 = -β5/σ5 + 10 β3/σ3 …

To solve this Gram-charlier expansion equation for PDF and CDF, chebyshev hermite
polynomial equation are required [22], which are as;

He0(x) = 1

He1(x) = x

He2(x) =x2 -1

He3(x) = x3 -3x

He4(x) = x4 -6x2+3

He5(x) = x5 -10x3+15x

He6(x) = x6 -15x4 +45x2 -15

20
 Read input
variable

k =1

l =1

compute,
Standard central moment (λ𝑙,j )
Standard location (ξl,k )
weight (Wl,k )

Determine the location (ρl,k)

if l <= m

Deterministic load flow

Compute the statistical moment of output

if k <= 3

Plot CDF of output variable using

Gram-Charlier expansion

Figure 3. 1 Probabilistic load Flow using Point Estimation method.

21
The probabilistic load flow with PV connected radial distribution network using point
estimation method is computed in MATLAB software with following steps;

Step 1: Read the input data.

step 2: Set concentration or point K =1 to 3( K =3 for 3 point estimation method)

step 3: Set L =1 to m ( m= Random variable)

step 4: for all random variable calculate standard central moment,standard location and
weight for the point.

Step 5: calculate the points for the random variable as per the estimation scheme (three
point estimation)

Step 6: Carry out load flow for all 'k' point.

Step 7: Calculate the moments for all the output.

Step 8: For all output random variable calculate the cumulant and plot the CDF curve
using Gram-Charlier expression.

Effect of PV integration on distribution system is viewed by comparing no PV

connection scheme with PV connected scheme at different buses.

The obtained result from the point estimation method is compared with the result from
the monte carlo simulation method in IEEE-33 bus system. Active and reactive power,
minimum and maximum voltage, overvoltage and Undervoltage is taken as the
parameter for comparison. The execution time with point estimation method and monte
carlo simulation method is also compared.

22
 CHAPTER FOUR
RESULT AND DISCUSSION

In this thesis, a probabilistic load flow algorithm is developed and implemented in

Integrated Nepalese Power System (INPS) network. Before implementing in INPS
system of Tandi Distribution feeder the developed algorithm is checked in the standard
Radial Distribution system of IEEE-33 bus system. All the working is done under
MATLAB R2020a script environment. The various specification of laptop used for the
computation is shown in table 4.1.

Table 4.1 Specification of laptop

Processor 2.4 GHz, core i5-6200U

RAM 8GB

Harddrive 512 HDD, 240 sata SSD

Operating System window 10 pro 64 bit

4.1 Result of IEEE-33 Bus System

The various parameter for load flow analysis of IEEE-33 Bus system is taken as shown
in table 4.2.

Table 4.2 Parameter of IEEE-33 Bus system

S. N Parameter Data

1. Number of Buses 33

2. Number of Branches 32

3. System voltage (KV) 12.66

4. Base MVA 100

5. Total Active Power (KW) 3715

6. Total Reactive Power (KVAR) 2300

23
The single line diagram of IEEE- 33 Bus system is shown in figure 4.1;

Figure 4. 1 Single line diagram of IEEE-33 Bus system

The PU voltage plot of IEEE-33 bus system is shown in Figure 4.2.

Figure 4. 2 Pu voltage plot of IEEE-33 Bus system

It is seen from Figure 4.2 that minimum per unit voltage is 0.9038 and 0.9164 at Bus
18 and 33 of IEEE-33 Bus system.The active power loss at various buses is shown in
figure 4.3.

From figure 4.3 the maximum active power loss is 52.08 Kw at branch number 2.
Active power loss also goes on decreasing as we move away from the source.

24
 Figure 4. 3 Active power loss of IEEE-33 Bus system.

Table 4. 3 Power flow of IEEE-33 bus system.

Active Reactive
S.N Power(KW) Power(Kvar)

Total Load 3715 2300

Total Loss 211 143

Total Power supply 3926 2443

4.1.1 Point Estimation Method

The various result obtained in terms of probability density function(PDF) and

cumulative distribution function (CDF) for Bus number 18 and 33 using PEM is shown
in figure 4.4 and 4.5. The functions for few buses are obtained for analysis purpose
similarly PDF and CDF function can be obtained for other buses too.

From figure 4.4, it can be observed that after obtaining the PDF curve of the required
parameter we can say that what will be the variation range of that parameter for the
considered variation in input parameters. Voltage magnitude of bus-18 can be seen to
be varied approximately in the range of 0.899 to 0.91 p.u. while that of bus-33 in the
range of 0.91 to 0.924 p.u. in considerable amount.

25
 Figure 4. 4 PDF curve of Bus 18 and 33

Mean value of voltage magnitude obtained for bus-18 and bus-33 from calculation are
0.9037 p.u. and 0.9164 p.u. respectively. From normally distributed pdf curve of the
both bus, it can be observed that maximum density value coincides to the corresponding
mean value. In similar way pdf curves for other buses can be obtained and can get
information as per the requirement. Hence, showing that large amount of information
can be obtained from probabilistic load flow in compare with deterministic load flow.
If all the random variable are equally distributed for all pu voltage then flatten type PDF
curve is obtained.

Figure 4. 5 CDF curve of Pu voltage of bus 18 and 33.

26
Figure 4.5 shows the cumulative distribution function (CDF) curve of voltage
magnitude of bus-18 and bus-33 respectively of IEEE-33 bus system obtained from
results of Point Estimation Method technique in MATLAB environment. Here the
curve line is obtained after fitting the obtained output data to normal distribution
function and hence the curve is S shaped as expected. In the curve it can be observed
that maximum value of distribution function is 1 p.u. or 100%. This curves gives us the
information that what is the probability of occurrence of a particular value and less than
that. For bus-18 the probability of occurrence of value 0.905 p.u. and less than that is
0.86 or 86%. The probability of occurring voltage magnitude between 0.902 p.u. and
0.905 p.u. for bus-18 will be 0.76(0.86-0.10) as per normal curve. Similar information
can be obtained from curve of bus-33 and similar curve can be obtained for other buses
and hence the required information. Hence, from the CDF curve we can say that with
the considered range of variation in input parameter what will be the probability of
maintaining the voltage within the required range of the respective buses and hence
other corrective measures can be followed to avoid the voltage collapse, thus improving
the reliability of supply.

1.02

0.98

0.96
Pu voltage

0.94

0.92

0.9

0.88

0.86

0.84
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
Bus Number

Minimum voltage Maximum voltage

Figure 4. 6 Minimum and Maximim voltage of IEEE-33 Bus system

The minimum voltage and maximum voltage that can be obtained from IEEE-33 Bus
system with Normally distributed load having zero skewness and 3 kurtosis is shown

27
in the figure 4.6. The minimum and maximum voltage are 0.8994 and 0.9083 at Bus 18
of IEEE-33 Bus system.

4.1.2 Comparison Between MCS and PEM

Probabilistic load flow of IEEE-33 bus system is carried out by Monte Carlo simulation
technique for 500 iteration and in each iteration 40 samples are taken.

It is seen that the output voltage from Both the method coincide each other and have
the error less than 0.001%. The Active power Loss obtained from both the method are
almost similar with less than 3% error. The comparison of MCS and PEM is shown in
table 4.4.

Table 4. 4 Comparison Table between MCS and PEM.

Parameter MCS PEM

Total Active Power loss (Kw) 211.14 211.503

Total Reactive Power loss (Kvar) 143.135 143.392

Mean Voltage Magnitude 0.9453 0.94531

Mean minimum Voltage 0.9407 0.9407

Mean maximum Voltage 0.9465 0.9465

Minimum Voltage 0.90375 0.90376

Execution Time(IEEE-33 BUS) 42.36 sec 2.956 sec

Hence it can be concluded from the comparison result of Point Estimation method and
Monte- carlo simulation method that the accuracy of Monte- carlo simulation method
is also maintain by Point Estimation method and Point Estimation Method is Prefer
here for further work because,

• Execution time of PEM is around 20 times less than MCS method.

• PEM takes less memory and space than MCS method.
• Only 2m+1 times iteration in PEM than 1000 of iteration in MCS method.
• For large and vague system MCS need powerful processor with extra RAM and
memory, but not by PEM.

28
4.1.3 Point Estimation Method with DG

Injecting 25% of solar at bus number 18 and 33 of IEEE-33 bus system. Bus Number
18 and 33 is selected for solar injection as it has the least pu voltage as compare to other
node. The generation from the solar is given by probability density function using beta
distribution. The value of Alpha and beta is 12.62 and 2.21 having mean value
851kw/m2 and standard deviation of 8.95%[11]. The random variable for the PEM are
Solar active power and load active and reactive power which is based on beta and
normal distribution respectively.

Figure 4. 7 Comparison of CDF plot at Bus 18 with 25% Penetration

1.02
1
0.98
Pu Voltage

0.96
0.94
0.92
0.9
0.88
0.86
0.84
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
Bus Number

Vm 0% DG Vm 25% DG

Figure 4. 8 Comparison of Pu voltage after 25% DG Penetration.

29
From the CDF curve shown in figure 4.7 the Probability of obtaining under voltage at
Bus 18 get improved from 0.8994 to 0.9267 and over voltage get improved from 0.9083
to 0.9521 after 25% of DG penetrated at Bus 18 and 33. Hence voltage get improved,
which is more clearly shown by the figure 4.8.

60
Active Power Loss(KW)

50

40

30

20

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Branches

Branch Active Power Loss(0%) Branch Active Power Loss(25%)

Figure 4. 9 Comparison of Active Power Loss after Penetration.

The comparison of Active Power Loss after 25% DG penetrated on IEEE-33 Bus
system is shown in figure 4.9. It can be seen that Active Power Loss in each branches
get reduces after PV Penetration.

Table 4. 5 Comparative Table after Penetration using PEM.

NO 25%
Particular
PENETRATION PENETRATION

Mean voltage 0.9453 0.9610

Minimum voltage 0.9037 0.9397

Mean Under Voltage 0.9407 0.9536

Mean Over Voltage 0.9465 0.9643

Total Active power loss 211.138 KW 128.653 KW

Total Reactive Power loss 143.135 kVAr 86.158 kVAr

30
It is seen from the Table 4.5 that after PV interconnection mean voltage goes on
increasing from 0.9453 to 0.961. Minimum voltage of the bus increases from 0.9037 to
0.9397 and active and reactive power loss decreasing as PV is Penetrated.

4.2 Result of tandi feeder using PEM

4.2.1 General layout of Tandi feeder

After successful completion of using Point Estimation Method with IEEE-33 bus
system, Tandi Distribution Feeder with 122 buses as a Radial distribution system is
taken for Analysis. The single line diagram of Tandi Distribution Feeder with Parsa
substation as a source at Bus 1 is shown in figure 4.10;

Figure 4. 10 Single Line diagram of Tandi Feeder.

Table 4.6 shows the parameter of 122 bus Tandi feeder which is one of the feeder of
NEA, Ratnanagar Tandi Distribution center among 4 feeders. The feeder is supply from
the Parsa substation and has feeder length of around 115 km. The data obtained from
the Tandi feeder and corresponding Line data and Bus data is shown in Appendix 2.

31
 Table 4. 6 Key Parameter of Tandi feeder.

S. N Parameter Data

1. Number of Buses 122

2. Number of Branches 121

3. System voltage (KV) 11

4. Base MVA 100

5. Total Active Power (KW) 4674.19

6. Total Reactive Power (KVAR) 2773.49

The maximum KVA is 5439.52 KVA with 0.86 average power factor obtained from
the TOD load data of Tandi feeder. Hence 59.4% Transformer loading is obtained. The
Load data is obtained by substituting the value of resistance and reactance of Dog,
Rabbit and Weasel conductor.

The mean of the load parameter of tandi feeder is taken with reference to 59.4%
transformer loading and 0.86 pf and standard deviation is taken with references to
standard deviation obtained from 22 TOD load data.

The daily load curve of Tandi Feeder is shown in the figure 4.11.

3000

2500

2000
Load (KW)

1500

1000

500

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Times of day (Hours)

Figure 4. 11 Daily Load curve of Tandi feeder

32
Load curves gives the variation of load with respect to time in the Power station. The
load on the power station varies time to time and it never remain constant, The figure
above is the plotted of variation of load for the whole day of 24 hours. The tandi feeder
is the domestic dominated with maximum load occuring during the morning 8am and
afternoon 6 to 7 pm as shown in the figure 4.11.

4.2.2 Normally Distributed Data

The TOD data of the Tandi feeder is checked for the best fit of the distribution. The
result obtained from the TOD data at Bus number 15 are;

Table 4. 7 Parameter of Load data.

S.N Mean Median Skewness Kurtosis

Value 3.82 3.93 -0.27 -0.79

250

200
Frequency

150

100

50

0
1 2 3 4 5 6 7 8
KVA interval

Figure 4. 12 Histogram plot of TOD data at Bus 15

From the Table 4.7 it is seen that the Mean and Median are almost equal to each other
and the value of skewness and kurtosis nearly equal to zero and lies in the interval [-
2,2]. The negative of skewness show that the data are right skewed.

The figure 4.12 also show the histogram plot which are almost symmetrical. All the
TOD load data are checked and can be concluded that the taken data are Normally
distributed.

33
 Figure 4. 13 Voltage magnitude plot of Tandi feeder.

The Pu voltage of tandi distribution feeder obtained from the deterministic load flow is
shown in the figure 4.13. The pu voltage goes on decreasing and is 0.817 at Bus 122
which is the minimum voltage at end bus of the feeder.

4.2.3 Point Estimation Method

The result of Voltage magnitude in terms of Minimum voltage and Maximum voltage
obtained using Point Estimation Method of Tandi feeder is as shown in Table 4.8;

Table 4. 8 Minimum and Maximum voltage of Tandi feeder.

Bus Under Over Bus Under Over

No. voltage voltage No. voltage voltage
1 1 1 62 0.8574 0.8765
2 0.9819 0.9843 63 0.8566 0.8759
3 0.9707 0.9746 64 0.8546 0.8744
4 0.9617 0.9668 65 0.8538 0.8739
5 0.9506 0.9572 66 0.8527 0.8732
6 0.9497 0.9564 67 0.8521 0.8727
7 0.9484 0.9554 68 0.854 0.874
8 0.9478 0.9549 69 0.8531 0.8733
9 0.9476 0.9549 70 0.8525 0.8728
10 0.9474 0.9545 71 0.852 0.8726
11 0.9473 0.9544 72 0.8617 0.8802
12 0.9401 0.948 73 0.8599 0.8786
13 0.9338 0.9426 74 0.8565 0.8758
14 0.9328 0.9418 75 0.8547 0.8743
15 0.9322 0.9413 76 0.8505 0.8708
34
16 0.9257 0.9356 77 0.8493 0.8697
17 0.9156 0.9269 78 0.8489 0.8693
18 0.9076 0.9201 79 0.8487 0.8691
19 0.9074 0.92 80 0.843 0.8648
20 0.9065 0.9195 81 0.8428 0.8645
21 0.906 0.9193 82 0.8395 0.8619
22 0.9056 0.9193 83 0.839 0.8615
23 0.8982 0.9118 84 0.8389 0.8615
24 0.8926 0.9069 85 0.8368 0.8598
25 0.8913 0.9057 86 0.8366 0.8596
26 0.8904 0.9048 87 0.8364 0.8595
27 0.8892 0.9038 88 0.8311 0.8553
28 0.8892 0.9037 89 0.8308 0.855
29 0.889 0.9036 90 0.8305 0.8548
30 0.8884 0.9031 91 0.83 0.8545
31 0.8881 0.9028 92 0.8296 0.8543
32 0.888 0.9027 93 0.8294 0.8541
33 0.8879 0.9026 94 0.8293 0.8541
34 0.8843 0.8998 95 0.8309 0.8552
35 0.8837 0.8992 96 0.8283 0.8531
36 0.8832 0.8989 97 0.828 0.8528
37 0.8829 0.8986 98 0.8248 0.8504
38 0.8826 0.8984 99 0.8224 0.8485
39 0.8765 0.893 100 0.8199 0.8467
40 0.8735 0.8903 101 0.8176 0.8449
41 0.8675 0.8851 102 0.815 0.8431
42 0.8645 0.8825 103 0.8122 0.8413
43 0.8633 0.8814 104 0.8089 0.8391
44 0.863 0.8811 105 0.808 0.8384
45 0.8628 0.8809 106 0.8078 0.8382
46 0.8613 0.8795 107 0.8076 0.8381
47 0.8605 0.8788 108 0.8075 0.838
48 0.8595 0.8779 109 0.8079 0.8384
49 0.8592 0.8776 110 0.8079 0.8383
50 0.8587 0.8771 111 0.8072 0.8378
51 0.8581 0.8765 112 0.807 0.8376
52 0.8574 0.8759 113 0.8069 0.8376
53 0.8571 0.8757 114 0.8079 0.8384
54 0.8578 0.8763 115 0.8066 0.8377
55 0.8577 0.8761 116 0.8058 0.8372
56 0.8575 0.8761 117 0.8049 0.8368
57 0.8578 0.8762 118 0.8042 0.8364
58 0.8576 0.876 119 0.8036 0.8361

35
 59 0.8575 0.876 120 0.8027 0.8357
60 0.86 0.8785 121 0.8022 0.8355
61 0.8588 0.8776 122 0.802 0.8347

Figure 4. 14 Minimum and Maximum voltage plot of Tandi feeder.

The minimum and maximum voltage obtained from each bus of the tandi feeder is
shown in Table 4.8 and Figure 4.14 and it is seen the pu voltage goes on decreases as
we move toward the end buses and variation of pu voltage at end buses is 0.802 to
0.8347pu.
120

100

80
Probability

60

40

20

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Bus Number

Figure 4. 15 Under voltage probability of Tandi feeder

The under voltage probability from bus 1 to bus 20 of tandi feeder is shown in the figure
4.15. There is no under voltage probability from bus 1 to bus 6 and above bus 15 to bus
122 there is 100% probability of getting under voltage before DG penetration which is

36
more clearly shown by the figure 4.15. There is no chance of over voltage probability
in tandi feeder.

The detail of Load and Loss of Tandi feeder is shown in the Table 4.9;

Table 4. 9 Power flow detail of Tandi Feeder.

Active Reactive
Parameter
Power(KW) Power(KVAr)

Total Load 4674.19 2773.49

Total Loss 544.904 633.131

Total Power supply 5219.094 3406.621

Tandi feeder has the total load of 4674.19kw and 2773.49 kVAr. On considering the
given value of conductor of tandi feeder as shown in Appendix 2 the total line losses
are 544.904kw and 633.131Kvar, Hence 5219.09 kw and 3406.621kvar is supplies from
the substation under the normal condition when DG is not connected.

Figure 4. 16 CDF plot of Pu voltage of Tandi feeder.

Figure 4.16 show the CDF curve of voltage magnitude of Bus 40, 80 and 122 of Real
Tandi Distribution feeder. The probability of obtaining voltage 0.82 pu and less than
0.82 pu at Bus 122 is around 90% and there is no chance of obtaining voltage greater
than 0.834 at Bus 122 before DG penetration. The Pu voltage at Bus 122 is 0.802 pu at
worst conditon and 0.834 pu at best condition. similar result can be obtained for Bus
40, 80 from figure 4.16 and for all other buses can also be calculated.
37
 Figure 4. 17 CDF plot of Active Power Loss of Tandi feeder

The Probability of Active Power Loss at branch 2 ranges from 40 to 48 kw as shown in

figure 4.17. The obtained Active Power loss from the load flow is 44.22 Kw which lies
in between the CDF curve of Active power loss shown in figure 4.17. From the CDF
curve there is no chance of loss less than 40 Kw and more than 48 kw at Branch 2 of
Tandi Feeder. similarly the probability of Active power loss at Branch 3 and 4, and
other Branches of Tandi feeder can be obtained.

Figure 4. 18 CDF plot of Reactive Power loss of Tandi feeder.

The probability of Reactive power loss at Branch 2 is 47 to 56 kvar. The obtained

Reactive power loss from the load flow is 51.537 kvar which lies in between the cdf
curve of reactive power loss of Branch 2. From the CDF curve of figure 4.18, there is

38
no chance of loss less than 47 kvar and more than 56 kvar at Branch 2 of tandi feeder.
similarly reactive power loss at branch 3 and 4 and, and other branches of tandi feeder
can be obtained.

4.2.4 Point Estimation Method with DG penetration

It is seen that the pu voltage of the Bus goes on decreasing as we move away from the
source. The Bus 122 which is the end Bus of the Tandi Distribution Feeder has the least
pu voltage i.e 0.817. To improve the Pu voltage of the system and also to decrease the
losses 25% of the solar as per ontario's standard [26] and [27] is penetrated at different
buses of the real system considering only two factors, power loss and voltage regulation
and other factor such as economic and geographic are not considerations. The
probabilistic analysis of tandi distribution feeder is done for three different scheme of
DG connection as , 25% of DG connected at Bus 122, 25% of DG connected at Bus
122,94 and 25% of DG connected at Bus 122,94,71. The Bus is selected based on
Minimum per unit voltage and end of the lateral branch.

The intermittency of solar power generation of average of 8 days irradiance data

obtained from pyranometer of Gairapatan Pokhara is as shown in figure 4.19.

800
PV output irradiance (W/m2)

700
600
500
400
300
200
100
0
9:13
9:32
9:51
10:1
7:00
7:19
7:38
7:57
8:16
8:35
8:54

10:2
10:4
11:0
11:2
11:4
12:0
12:2
12:4
1:01
1:20
1:39
1:58
2:17
2:36
2:55
3:14
3:33
3:52
4:11
4:30
4:49
5:08
5:27
5:46

Time of Day

Figure 4. 19 Intermittency of Solar Power Generation

The solar irradiance for each hour of the day is modeled by the Beta probability density
function based on collected data of solar irradiance from 9am to 5pm of 4 month, taken
as a reference for penetration of PV. The mean value of irradiance obtained is 0.451
KW/m2 and 61.03% standard deviation with beta distribution parameter as shown in
table 4.10.
39
 Table 4. 10 Beta Distribution Parameter of PV.

Parameter Alpha Beta

solar 1.025 1.248

Figure 4. 20 Comparison of Pu voltage plot with 25% DG at different Bus

The figure 4.20 show the Pu voltage plot with three different scheme of DG connection.
It is seen that Pu voltage is improved with each scheme of DG connection. The
maximum improvement in Pu voltage is seen at scheme 1 when 25% of DG is
connected at the end bus.

Figure 4. 21 Comparison of CDF plot of Pu voltage at Bus 120

40
The comparison of CDF plot of Pu voltage at Bus 120 for 25% DG penetrated at
different Bus is shown in figure 4.21. From the figure 4.21 it is seen than Probability of
pu voltage at Bus 120 is maximum when 25% of DG is connected at the end Bus.

Figure 4. 22 Comparison of CDF plot of pu voltage at Bus 71

From figure 4.21 and 4.22 of pu voltage, it can be observed that although the minimum
voltage probability remain almost same but maximum voltage Probability is increases
for each scheme of DG connection. The maximum increases in voltage of Bus 120 is
seen at scheme 1 when 25% of DG is connected at bus 122 and maximum increases in
voltage of Bus 71 is seen at scheme 3 when 25% of DG is connected at Bus 122,94
and 71. Hence it is seen that pu voltage is improved at each scheme of DG connection.
The minimum voltage probability remain almost same although the DG is connected
because of the intermittance character of PV.

Similarly, for all other Buses of tandi feeder CDF can be plotted and probability of
minimum voltage and maximum voltage at each buses can be determined. The CDF
plot of active power loss at Branch 4 is shown in figure 4.23, It is seen that Probability
of active power loss at branch 4 is decreases with each scheme of DG connection.

Under the normal condition when DG is not connected, the mean active power loss at
branch 4 is 43 KW which can be reduced maximum and reaches to 25KW under the
best condition of DG connection and over active power loss at branch 4 is around 46

41
kw for each scheme under worst condition of DG connection which is as similar to no
DG connection scheme as shown in figure 4.23.

Figure 4. 23 Comparison of CDF plot active power loss at Branch 4.

Hence for all other branches of tandi feeder the probability of maximum and minimum
power loss can be obtained for each scheme of DG connection.

Table 4. 11 Comparison of Parameter after DG penetration.

25% DG connected at

Bus Bus
Parameter No DG Bus 122 122,94 122,94,71

Total Active Power loss(KW) 541.24 422.206 419.05 423.68

Total Reactive Power loss(Kvar) 628.91 490.309 486.63 491.86

Mean minimum voltage 0.8618 0.868 0.863 0.866

Mean maximum voltage 0.8815 0.9163 0.9106 0.9064

Minimum Voltage 0.8182 0.8513 0.8469 0.8417

Mean Under voltage 46.20% 42.90% 43% 43.40%

42
Table 4.11 show the comparison of various parameter for each scheme after 25% DG
connected at different Bus. The Minimum voltage of 0.8182 at Bus 122 is improved to,
0.8513 when 25% of DG connected at Bus 122, 0.8469 when 25% of DG connected at
Bus 94,122, 0.8417 when 25% of DG connected at Bus 71,94,122. Hence with each
scheme of DG connection the pu voltage get improved and maximum improvement in
minimum voltage occur when 25% of DG is connected at the end bus. The active and
reactive power loss is decreases and mean minimum voltage and maximum voltage
probability is seen increases for each scheme of DG connection.

Bus 1 to Bus 20 is taken to calculate the mean under voltage probability because from
bus 20 to end bus 122 there is 100% probability of under voltage and under voltage
probability is not change although DG is connected at each scheme. 46.2% under
voltage probability at normal condition is reduced to 42.9% at scheme 1 of DG
connection which is more clearly seen by the table 4.11. Hence maximum reduction in
under voltage is achieved when DG is connected at end bus. There is no chance of over
voltage at each scheme although the DG is connected.

43
 CHAPTER FIVE
CONCLUSION AND RECOMMENDATION

5.1 Conclusion

In this paper, Point Estimation method is used for Probabilistic load flow analysis of
PV This paper show the comparative result of PEM and MCS method. 42.36 sec
computational time of MCS is reduced to 2.95 sec by PEM and accuracy of MCS is
also maintain by the PEM. Hence for the large system it is efficient and fast to use PEM.
Hourly variation of solar irradiance is consider using beta distribution. Bus number
122,94 and 71 is chosen for PV interconnection. The minimum voltage of 0.8182 is
improve to 0.8513 at scheme 1 when 25% of DG is connected at Bus 122, which is the
best case considering the improvement in per unit voltage. similarly total active power
loss of 541.24 KW is reduced to 419.05 KW and reactive power loss of 628.91 KVar
is reduced to 486.63 KVar when 25% of DG is connected at Bus 122 and 94, which is
the best cases considering the minimizing the Power loss as it has maximum loss
reduction as compare to other cases. The mean minimum voltage and maximum voltage
probability is also improve for each scheme of DG connection.

PV connection in distribution system reduces the branch power loss, reduces the
undervoltage probability but increases the overvoltage probability. The use of PEM in
the real system help to calculate the undervoltage and overvoltage at each buses of the
system which is helpful in maintaining the voltage limit, calculate the maximum and
minimum power loss at each branches and power flow through each branches. Hence
probability analysis is used to view the best cases and worst cases at each buses, which
is used to analyse the uncertainty and maintain the reliability of the system.

5.2 Recommendation

The various opportunities for extending the scope of these thesis are;

➢ Protection co-ordination can also be consider for better analysis of the

distribution system.
➢ Some other uncertainties parameter such as wind can be consider for evaluating
probabilistic model of the system with optimum cost analysis.
➢ Battery Energy storage system can be used for balancing the intermittancy of
DG output.
44
 REFERENCES

[1] Point Estimators - Definition, Properties, and Estimation Methods. (2021).

Retrieved 12 March 2021, from https://corporatefinanceinstitute.com /resources/
knowledge /other/point-estimators/

[2] Wu, C., Wen, F., Lou, Y., & Xin, F. (2015). Probabilistic load flow analysis of
photovoltaic generation system with plug-in electric vehicles. International Journal of
Electrical Power & Energy Systems, 64, 1221–1228.

[3] Prusty B Rajanarayan, Jena D. An over-limit risk assessment of PV integrated power

system using probabilistic load flow based on multi-time instant uncertainty modeling.
Renew Energy 2018;116:367–83

[4] Che, Y., Wang, X., Lv, X., & Hu, Y. (2020). Probabilistic load flow using improved
three point estimate method. International Journal of Electrical Power & Energy
Systems, 117, 105618.

[5] Liu Jun S. Monte Carlo strategies in scientific computing. Springer; 2001.

[6] Kroese DP, Taimre T, Botev ZI. Handbook of Monte Carlo methods. John Wiley &
Sons; 2013.

[7] M. Fan, V. Vittal, G. T. Heydt and R. Ayyanar, "Probabilistic Power Flow Analysis
With Generation Dispatch Including Photovoltaic Resources," in IEEE Transactions
on Power Systems, vol. 28, no. 2, pp. 1797-1805, May 2013.

[8] Mensah, L., Yamoah, J. and Adaramola, M., 2019. Performance evaluation of a
utility-scale grid-tied solar photovoltaic (PV) installation in Ghana. Energy for
Sustainable Development, 48, pp.82-87.

[9] B. K. Panigrahi, S. K. Sahu, R. Nandi and S. Nayak, "Probablistic load flow of a

distributed generation connected power system by two point estimate method," 2017
International Conference on Circuit ,Power and Computing Technologies (ICCPCT),
Kollam, 2017, pp. 1-5.

[10] A. B. Krishna, N. Gupta, K. R. Niazi, and A. Swarnkar, “Probabilistic power flow

in radial distribution systems using point estimate methods,” in 2017 4th International
Conference on Advanced Computing and Communication Systems (ICACCS), 2017,
pp. 1–6.

45
[11] J. Emmanuel Anandraj, "Point estimate method of Load Flow for distribution
network with photovoltaic generators," 2013 International Conference on Energy
Efficient Technologies for Sustainability, Nagercoil, India, 2013, pp. 24-29.

[12]www.theelectricalportal.com/2017/06/primary-distribution-system-secondary
html

[13] S. Zadeh, O. Nejad, S.hasani, A.A.Gharaveisi and GH.Shahgholian, Optimal DG

Placement for Power Loss Reduction and Improvement Voltage Profile using Smart
Methods, 3 ed., vol. 1, International Journal of Smart Electrical Engineering, 2012,
pp. 141-147.

[14] N. K. Roy, and H. R. Pota, “Impact of High PV Penetration into Distribution

Networks under Contingencies” Smart Energy Grid Engineering (SEGE), 2013 IEEE
International Conference on 28-30 Aug. 2013, Oshawa, ON, Canada.

[15] Xiaotian Xu, Xiao Lv, Xinyuan Zhang, Xiaotian Xu," Research on Distributed
Photovoltaic Grid-connected Voltage Cooperative Control Strategy Considering Local
Load," American Journal of Electrical and Electronic Engineering. 2020.

[16] S. Gurung, S. Naetiladdanon and A. Sangswang, "Impact of Large Photovoltaic

Penetration on small signal stability," in 2017 IEEE Region 10 Conference (TENCON).,
pp. 646-650.

[17] Liu, S., Liu, P. X., & Wang, X. (2016). Stochastic Small-Signal Stability Analysis
of Grid-Connected Photovoltaic Systems. IEEE Transactions on Industrial Electronics,
63(2), 1027–1038.

[18] Nepal Electricity Authority, Course of 2074 related to energy obtained from solar
PV, Nepal Electricity Authority, 2076.

[19]www.researchgate.net/publication/318654444_Design_Estimation_of_Rooftop_G
rid-tied_Solar_Photovoltaic_System

[20] Hong, L., & Goudie, D. (1977). The Square Root Method in Power Flow

Computation and Contingency Analysis. IFAC Proceedings Volumes, 10(1), 364-368.

[21] A. M. Leite da Silva and A. M. de Castro, "Risk Assessment in Probabilistic Load

Flow via Monte Carlo Simulation and Cross-Entropy Method," in IEEE Transactions
on Power Systems, vol. 34, no. 2, pp. 1193-1202, March 2019.

46
[22] Delgado, C., & Domínguez-Navarro, J. A. (2014). Point estimate method for
probabilistic load flow of an unbalanced power distribution system with correlated
wind and solar sources. International Journal of Electrical Power & Energy Systems,
61, 267–278.

[23] B. K. Panigrahi, S. K. Sahu, R. Nandi and S. Nayak, "Probablistic load flow of a

distributed generation connected power system by two point estimate method," 2017
International Conference on Circuit ,Power and Computing Technologies (ICCPCT),
Kollam, 2017, pp. 1-5.

[24] Conti. S, Raiti. S, “Probabilistic Load Flow for Distribution Networks with
Photovoltaic Generators. Part 1 – Theoretical concepts and models,” International
Conference on Clean Electrical Power, Capri, Italy, pp. 132-136 May 21st-23rd 2007.

[25] K. Dheeraj Kumar, “Optimal planning of distributed generation system,” Ph.D.

thesis, Department of Electrical Engineering, IIT, Roorkee, 2007.

[26] Ontario Power Authority.” [Online]. Available: http://www. powerauthority.on.ca/

[27] S. H. Dolatabadi, M. Ghorbanian, P. Siano and N. D. Hatziargyriou, "An Enhanced

IEEE 33 Bus Benchmark Test System for Distribution System Studies," in IEEE
Transactions on Power Systems, vol. 36, no. 3, pp. 2565-2572, May 2021.

[28] J. M. Morales and J. Perez-Ruiz, "Point Estimate Schemes to Solve the

Probabilistic Power Flow," in IEEE Transactions on Power Systems, vol. 22, no. 4, pp.
1594-1601, Nov. 2007.

[29] A. C. Miller and T. R. Rice, “Discrete approximation of probability distributions,”

Management Science, vol. 29, no. 3, pp. 352-362, March 1983.

[30] R. S. Al Abri, E. F. El-Saadany and Y. M. Atwa, "Optimal Placement and Sizing

Method to Improve the Voltage Stability Margin in a Distribution System Using
Distributed Generation," in IEEE Transactions on Power Systems, vol. 28, no. 1, pp.
326-334, Feb. 2013.

[31] Mohammadi, M., Shayegani, A., & Adaminejad, H. (2013). A new approach of
point estimate method for probabilistic load flow. International Journal of Electrical
Power & Energy Systems, 51, 54–60.

47
[32] Ruiz-Rodriguez FJ, Hernández JC, Jurado F. Voltage unbalance assessment in
secondary radial distribution networks with single-phase photovoltaic systems. Int J
Electr Power Energy Syst 2015;64:646–54.

[33] Kroese DP, Taimre T, Botev ZI. Handbook of Monte Carlo methods. John Wiley
& Sons; 2013.

[34] Morales JM, Baringo L, Conejo AJ, et al. Probabilistic power flow with correlated
wind sources. IET Gener Transm Distrib 2010;4(5):641–51.

[35] P. Zhang and S. T. Lee, “Probabilistic load flow computation using the method of
combined Cumulants and Gram-Charlier expansion,” IEEE Trans. Power Syst., vol.
19, no. 1, pp. 676–682, Feb. 2004.

[36] C. L. Su and C. N. Lu, “Two-point estimate method for quantifying transfer

capability uncertainty,” IEEE Trans. Power Syst., vol. 20, no. 2, pp. 573–579, May
2005.

[37] J. Shukla, B. K. Panigrahi,” An Analytical Approach for Optimal Size of

Distributed Generation Unit” IEEE International Conference on Recent Advances and
Innovations in Engineering, May 09-11, 2014, Jaipur, India.

48
 APPENDIX- 1 IEEE-33 BUS SYSTEM

IEEE-33 bus system load data and Bus data is as shown in the Table below.

Bus
From To R X P(KW) Q(KVAR)
No.
1 1 2 0.0922 0.0477 0 0
2 2 3 0.493 0.2511 100 60
3 3 4 0.366 0.1864 90 40
4 4 5 0.3811 0.1941 120 80
5 5 6 0.819 0.707 60 30
6 6 7 0.1872 0.6188 60 20
7 7 8 1.7114 1.2351 200 100
8 8 9 1.03 0.74 200 100
9 9 10 1.04 0.74 60 20
10 10 11 0.1966 0.065 60 20
11 11 12 0.3744 0.1238 45 30
12 12 13 1.468 1.155 60 35
13 13 14 0.5416 0.7129 60 35
14 14 15 0.591 0.526 120 80
15 15 16 0.7463 0.545 60 10
16 16 17 1.289 1.721 60 20
17 17 18 0.732 0.574 60 20
18 2 19 0.164 0.1565 90 40
19 19 20 1.5042 1.3554 90 40
20 20 21 0.4095 0.4784 90 40
21 21 22 0.7089 0.9373 90 40
22 3 23 0.4512 0.3083 90 40
23 23 24 0.898 0.7091 90 50
24 24 25 0.896 0.7011 420 200
25 6 26 0.203 0.1034 420 200
26 26 27 0.2842 0.1447 60 25
27 27 28 1.059 0.9337 60 25

49
 28 28 29 0.8042 0.7006 60 20
29 29 30 0.5075 0.2585 120 70
30 30 31 0.9744 0.963 200 600
31 31 32 0.3105 0.3619 150 70
32 32 33 0.341 0.5302 210 100
33 60 40

APPENDIX- 2 TANDI FEEDER

The detail of conductor Type, Transformer size and corresponding length of Tandi
feeder is as shown in the table below;

Tr. Tr.
Conductor Length Conductor Length
SN RN Size SN RN Size
Type (km) Type (km)
(kVA) (kVA)

1 2 Dog 0.8 50 62 63 Dog 0.4 50

2 3 Dog 0.5 50 63 64 Rabbit 0.65 100

3 4 Dog 0.4 50 64 65 Rabbit 0.5 50

4 5 Dog 0.5 50 65 66 Rabbit 0.75 100

5 6 Rabbit 0.6 100 66 67 Rabbit 0.7 200

6 7 Rabbit 1 50 64 68 Rabbit 0.5 25

7 8 Rabbit 0.5 50 68 69 Rabbit 1 50

8 9 Rabbit 0.5 100 69 70 Rabbit 0.9 100

8 10 Rabbit 0.8 100 70 71 Weasel 1 50

10 11 Rabbit 0.7 50 42 72 Dog 0.3 50

5 12 Dog 0.5 50 72 73 Dog 0.2 100

12 13 Dog 0.3 50 73 74 Dog 0.4 50

13 14 Rabbit 1 150 74 75 Dog 0.2 50

14 15 Rabbit 1.5 100 75 76 Dog 0.5 100

13 16 Dog 0.4 50 76 77 Rabbit 1 100

50
16 17 Dog 0.5 100 77 78 Rabbit 0.5 100

17 18 Dog 0.4 100 78 79 Rabbit 0.6 100

18 19 Rabbit 0.7 50 76 80 Dog 1 100

18 20 Rabbit 1 25 80 81 Weasel 0.9 50

20 21 Rabbit 0.5 150 80 82 Dog 0.5 50

21 22 Rabbit 1 100 82 83 Rabbit 0.7 100

18 23 Dog 0.5 100 83 84 Rabbit 0.5 50

23 24 Dog 0.3 100 82 85 Dog 0.4 200

24 25 Dog 0.7 100 85 86 Rabbit 0.5 50

25 26 Dog 0.6 100 86 87 weasel 0.5 50

26 27 Dog 0.8 200 85 88 Dog 1 50

27 28 Dog 0.1 50 88 89 Dog 0.4 100

28 29 Dog 0.2 200 89 90 Dog 0.4 25

29 30 Rabbit 1 50 90 91 Dog 0.8 100

30 31 Rabbit 0.7 50 91 92 Dog 1 50

31 32 Rabbit 0.5 25 92 93 Dog 1 50

32 33 Rabbit 0.8 25 93 94 Dog 1 25

24 34 Dog 0.5 50 88 95 Rabbit 1 50

34 35 Dog 0.7 50 88 96 Dog 0.6 50

35 36 Dog 0.6 100 96 97 Weasel 0.5 100

36 37 Dog 0.5 50 96 98 Dog 0.8 100

37 38 Dog 0.6 200 98 99 Dog 0.6 50

34 39 Dog 0.5 100 99 100 Dog 0.6 50

39 40 Dog 0.2 50 100 101 Dog 0.6 50

40 41 Dog 0.4 50 101 102 Dog 0.7 100

51
41 42 Dog 0.2 100 102 103 Dog 0.8 50

42 43 Dog 0.2 100 103 104 Dog 1 50

43 44 Weasel 0.7 50 104 105 Dog 0.8 50

44 45 Weasel 0.5 50 105 106 Rabbit 0.5 25

43 46 Dog 0.4 50 106 107 Weasel 0.5 50

46 47 Dog 0.3 150 107 108 Weasel 0.5 25

47 48 Dog 0.4 150 105 109 Weasel 0.4 25

48 49 Weasel 0.6 100 105 110 Dog 0.1 50

48 50 Dog 0.5 50 110 111 Rabbit 0.8 50

50 51 Dog 0.35 100 111 112 Dog 0.7 100

51 52 Weasel 0.6 100 112 113 Dog 0.5 25

52 53 Weasel 0.5 100 104 114 Dog 0.5 100

51 54 Weasel 0.5 50 114 115 Dog 0.7 50

54 55 Weasel 0.5 25 115 116 Dog 0.5 50

55 56 Weasel 0.6 25 116 117 Dog 0.5 50

51 57 Dog 0.5 200 117 118 Dog 0.5 25

57 58 Dog 1 50 118 119 Dog 0.4 25

58 59 Weasel 0.8 25 119 120 Dog 0.6 100

46 60 Dog 0.5 50 120 121 Dog 0.4 25

60 61 Dog 0.5 100 121 122 Dog 2 300

61 62 Dog 0.6 100

52
The Active and Reactive Power flow and Branch Active and Reactive Power Loss of
122 Buses of Tandi Feeder is shown in Table Below;

From P Q Branch P Branch Q

Bus To Bus flow(Kw) flow(Kvar) Loss(Kw) Loss(Kvar)

1 2 5147.225 3323.155 71.633 83.197

2 3 5077.509 3256.515 44.22 51.537

3 4 5017.032 3200.444 34.966 40.951

4 5 4948.17 3134.8 43.352 50.525

5 6 230.223 136.617 0.217 0.135

6 7 178.919 106.169 0.22 0.137

7 8 153.297 90.962 0.081 0.051

8 9 51.084 30.311 0.009 0.006

8 10 76.629 45.469 0.032 0.02

10 11 25.542 15.156 0.003 0.002

5 12 4653.15 2937.394 39.063 45.527

12 13 4604.269 2895.287 23.352 26.966

13 14 127.738 75.795 0.117 0.073

14 15 51.084 30.311 0.028 0.017

13 16 4422.017 2770.469 28.87 33.811

16 17 4360.748 2713.674 35.742 41.657

17 18 4281.788 2650.715 27.886 32.66

18 19 25.542 15.156 0.008 0.005

18 20 140.568 83.411 0.152 0.095

20 21 127.734 75.793 0.063 0.04

21 22 51.084 30.311 0.024 0.015

18 23 4032.941 2485.033 31.504 36.717

23 24 3963.299 2433.292 18.563 21.436

53
24 25 409.366 243.262 0.453 0.527

25 26 357.985 212.605 0.297 0.345

26 27 306.609 181.955 0.292 0.339

27 28 204.425 121.315 0.016 0.019

28 29 178.858 106.13 0.025 0.029

29 30 76.644 45.479 0.046 0.029

30 31 51.088 30.314 0.014 0.009

31 32 25.543 15.157 0.003 0.002

32 33 12.771 7.578 0.001 0.001

24 34 3478.354 2131.172 24.048 28.028

34 35 204.468 121.398 0.115 0.134

35 36 178.851 106.155 0.075 0.087

36 37 127.735 75.807 0.032 0.037

37 38 102.168 60.622 0.025 0.029

34 39 3227.235 1970.015 21 24.474

39 40 3167.956 1930.193 8.197 9.514

40 41 3126.429 1896.315 15.988 18.725

41 42 3092.952 1871.95 7.936 9.211

42 43 1206.819 718.934 1.201 1.394

43 44 51.088 30.314 0.025 0.01

44 45 25.542 15.156 0.004 0.002

43 46 1102.627 655.964 1.994 2.335

46 47 575.765 342.13 0.412 0.476

47 48 498.729 296.183 0.41 0.48

48 49 51.084 30.311 0.022 0.008

54
48 50 370.713 220.065 0.284 0.332

50 51 344.998 204.708 0.174 0.201

51 52 102.186 60.629 0.086 0.034

52 53 51.084 30.311 0.018 0.007

51 54 51.09 30.314 0.021 0.008

54 55 25.543 15.157 0.005 0.002

55 56 12.771 7.578 0.001 0.001

51 57 140.489 83.364 0.041 0.048

57 58 38.315 22.735 0.006 0.007

58 59 12.771 7.578 0.002 0.001

46 60 500.392 297.601 0.516 0.602

60 61 474.385 281.903 0.465 0.542

61 62 422.856 251.075 0.445 0.517

62 63 371.543 220.496 0.229 0.268

63 64 345.359 204.937 0.642 0.402

64 65 179.001 106.218 0.133 0.083

65 66 153.313 90.971 0.146 0.091

66 67 102.168 60.622 0.061 0.038

64 68 115.085 68.29 0.056 0.035

68 69 102.224 60.656 0.09 0.056

69 70 76.637 45.471 0.045 0.028

70 71 25.542 15.156 0.011 0.004

42 72 1829.637 1116.447 4.213 4.865

72 73 1801.364 1098.121 2.732 3.171

73 74 1745.166 1061.821 5.115 5.99

55
74 75 1717.119 1043.757 2.506 2.909

75 76 1685.539 1021.565 6.038 7.037

76 77 153.315 90.972 0.202 0.126

77 78 102.184 60.632 0.046 0.029

78 79 51.084 30.311 0.016 0.01

76 80 1471.565 889.27 9.375 10.888

80 81 25.542 15.156 0.008 0.003

80 82 1390.731 838.904 4.201 4.896

82 83 76.635 45.473 0.044 0.027

83 84 25.542 15.156 0.009 0.006

82 85 1285.631 774.877 2.88 3.372

85 86 51.096 30.317 0.016 0.01

86 87 25.542 15.156 0.012 0.005

85 88 1126.731 677.402 5.62 6.527

88 89 178.902 106.216 0.057 0.067

89 90 127.789 75.871 0.029 0.034

90 91 114.97 68.237 0.048 0.056

91 92 63.866 37.903 0.02 0.023

92 93 38.315 22.737 0.008 0.01

93 94 12.771 7.578 0.002 0.003

88 95 25.542 15.156 0.006 0.004

88 96 894.548 538.324 2.135 2.48

96 97 51.084 30.311 0.019 0.007

96 98 815.518 490.08 2.385 2.77

98 99 762.864 457.945 1.571 1.824

56
99 100 735.855 441.084 1.468 1.705

100 101 708.945 424.339 1.368 1.589

101 102 681.918 407.455 1.485 1.729

102 103 629.379 375.454 1.455 1.69

103 104 602.16 358.351 1.677 1.947

104 105 204.443 121.321 0.158 0.183

105 106 51.098 30.317 0.012 0.008

106 107 38.315 22.735 0.012 0.005

107 108 12.771 7.578 0.002 0.001

105 109 12.771 7.578 0.001 0

105 110 115.013 68.255 0.006 0.007

110 111 89.411 53.061 0.059 0.037

111 112 63.856 37.89 0.014 0.016

112 113 12.771 7.578 0.001 0.001

104 114 371.699 221.319 0.319 0.372

114 115 320.281 190.62 0.333 0.388

115 116 294.538 175.229 0.201 0.235

116 117 268.828 159.877 0.168 0.196

117 118 243.148 144.561 0.137 0.16

118 119 230.279 136.868 0.098 0.115

119 120 217.377 129.137 0.132 0.153

120 121 166.242 98.767 0.051 0.06

121 122 153.252 90.934 0.219 0.255

Loss 544.904 633.131

57
 APPENDIX- 3 CENTRAL MOMENT

The different central moment of Tandi feeder is as shown in table below;

Active Power Reactive Power

Bus
Mean S.D Skewness Kurtosis Mean S.D Skewness Kurtosis
No.

2 25.542 3.33 0.414 1.7394 15.16 3.12 1.547 2.0013

3 25.542 21 1.9717 7.4322 15.16 17 0.425 7.3478

4 25.542 8.42 0.0508 2.1343 15.16 7.46 0.0465 2.1475

5 25.542 7.52 0.6918 2.1009 15.16 7.45 0.746 2.0142

6 51.084 6.62 0.3636 2.0408 30.31 5.14 0.4621 2.0511

7 25.542 7.95 0.7312 2.698 15.16 7.95 0.7312 2.698

8 25.542 24.9 0.1108 1.3528 15.16 24.9 0.1108 1.3528

9 51.084 20.1 2.8842 10.057 30.31 20.1 2.8842 10.057

10 51.084 5.77 2.287 10.3681 30.31 5.77 2.287 10.3681

11 25.542 17 -1.3942 4.2209 15.16 17 -1.3942 4.2209

12 25.542 32 -0.1649 3.0013 15.16 32 -0.1649 3.0013

13 25.542 5.09 1.0711 3.7296 15.16 5.09 1.0711 3.7296

14 76.626 15.6 -0.7002 3.7536 45.47 15.6 -0.7002 3.7536

15 51.084 5.8 1.4106 5.9964 30.31 5.8 1.4106 5.9964

16 25.542 5.13 0.2294 1.3942 15.16 5.13 0.2294 1.3942

17 51.084 33.3 0.8486 2.661 30.31 33.3 0.8486 2.661

18 51.084 12.5 0.3875 2.5183 30.31 12.5 0.3875 2.5183

19 25.542 52 0.1296 2.9313 15.16 52 0.1296 2.9313

20 12.771 10.8 -1.8668 5.7726 7.578 10.8 -1.8668 5.7726

58
21 76.626 7.84 0.0794 1.8102 45.47 7.84 0.0794 1.8102

22 51.084 24 0.3478 2.0742 30.31 24 0.3478 2.0742

23 51.084 34 -0.4473 1.9103 30.31 34 -0.4473 1.9103

24 51.084 6.5 0.643 2.6692 30.31 6.5 0.643 2.6692

25 51.084 8.12 -0.3059 2.7052 30.31 8.12 -0.3059 2.7052

26 51.084 4.58 0.0438 1.7312 30.31 4.58 0.0438 1.7312

27 102.17 5.54 0.836 2.6825 60.62 5.54 0.836 2.6825

28 25.542 12.3 -0.268 1.6335 15.16 12.3 -0.268 1.6335

29 102.17 6.03 0.8704 2.8889 60.62 6.03 0.8704 2.8889

30 25.542 13.3 1.1457 3.0506 15.16 13.3 1.1457 3.0506

31 25.542 12.8 -0.6744 2.969 15.16 12.8 -0.6744 2.969

32 12.771 22.3 -0.1676 2.3248 7.578 22.3 -0.1676 2.3248

33 12.771 18.2 0.4375 1.3761 7.578 18.2 0.4375 1.3761

34 25.542 17.9 -0.5915 2.3038 15.16 17.9 -0.5915 2.3038

35 25.542 5.52 -0.1408 2.4748 15.16 5.52 -0.1408 2.4748

36 51.084 6.62 0.2594 3.0591 30.31 6.62 0.2594 3.0591

37 25.542 7.95 -0.4498 1.3558 15.16 7.95 -0.4498 1.3558

38 102.17 12.3 0.0922 3.2961 60.62 12.3 0.0922 3.2961

39 51.084 20.1 2.4699 13.3174 30.31 20.1 2.4699 13.3174

40 25.542 11.6 0.8367 2.8926 15.16 11.6 0.8367 2.8926

41 25.542 17 -0.3463 2.8974 15.16 17 -0.3463 2.8974

42 51.084 15 -0.8582 4.1284 30.31 15 -0.8582 4.1284

43 51.084 14 0.414 1.9507 30.31 14 0.414 1.9507

44 25.542 12 0.0326 1.786 15.16 12 0.0326 1.786

59
45 25.542 12.7 -0.3698 4.7706 15.16 12.7 -0.3698 4.7706

46 25.542 12 0.3811 1.87 15.16 12 0.3811 1.87

47 76.626 8 -0.8911 4.7936 45.47 8 -0.8911 4.7936

48 76.626 9 0.548 2.2559 45.47 9 0.548 2.2559

49 51.084 8.46 -1.8421 6.0553 30.31 8.46 -1.8421 6.0553

50 25.542 12 0.5413 4.7126 15.16 12 0.5413 4.7126

51 51.084 5 -1.1223 9.249 30.31 5 -1.1223 9.249

52 51.084 6 -0.2568 2.1112 30.31 6 -0.2568 2.1112

53 51.084 7 -0.3424 3.0278 30.31 7 -0.3424 3.0278

54 25.542 15 3.4818 16.9979 15.16 15 3.4818 16.9979

55 12.771 12 2.696 11.9564 7.578 12 2.696 11.9564

56 12.771 25 1.1457 3.0506 7.578 25 1.1457 3.0506

57 102.17 5.08 -0.6744 2.969 60.62 5.08 -0.6744 2.969

58 25.542 12.2 -0.1676 2.3248 15.16 12.2 -0.1676 2.3248

59 12.771 16.1 0.4375 1.3761 7.578 16.1 0.4375 1.3761

60 25.542 13.4 -0.5915 2.3038 15.16 13.4 -0.5915 2.3038

61 51.084 9.25 -0.1408 2.4748 30.31 9.25 -0.1408 2.4748

62 51.084 10.1 0.2594 3.0591 30.31 10.1 0.2594 3.0591

63 25.542 16.3 -0.4498 1.3558 15.16 16.3 -0.4498 1.3558

64 51.084 8.34 0.0922 3.2961 30.31 8.34 0.0922 3.2961

65 25.542 14.5 2.4699 13.3174 15.16 14.5 2.4699 13.3174

66 51.084 9.08 0.8367 2.8926 30.31 9.08 0.8367 2.8926

67 102.17 10.1 -0.3463 2.8974 60.62 10.1 -0.3463 2.8974

68 12.771 14.6 -0.8582 4.1284 7.578 14.6 -0.8582 4.1284

60
69 25.542 19 0.1456 7.94 15.16 19 0.1456 7.94

70 51.084 8.34 0.147 6.45 30.31 8.34 0.147 6.45

71 25.542 28 -0.287 3.254 15.16 28 -0.287 3.254

72 25.542 13.9 0.151 2.87 15.16 13.9 0.151 2.87

73 51.084 8.81 0.378 2.56 30.31 8.81 0.378 2.56

74 25.542 13.6 0.245 3.25 15.16 13.6 0.245 3.25

75 25.542 23.6 6.847 14.56 15.16 23.6 6.847 14.56

76 51.084 6.36 1.008 4.241 30.31 6.36 1.008 4.241

77 51.084 9.08 1.749 11.24 30.31 9.08 1.749 11.24

78 51.084 5.58 -0.915 2.86 30.31 5.58 -0.915 2.86

79 51.084 13.2 3.14 6.134 30.31 13.2 3.14 6.134

80 51.084 6.68 2.79 5.641 30.31 6.68 2.79 5.641

81 25.542 5.13 -1.09 0.524 15.16 5.13 -1.09 0.524

82 25.542 10.1 7.16 12.244 15.16 10.1 7.16 12.244

83 51.084 12.5 1.057 2.479 30.31 12.5 1.057 2.479

84 25.542 57 3.542 11.179 15.16 57 3.542 11.179

85 102.17 15.5 -2.113 2.533 60.62 15.5 -2.113 2.533

86 25.542 23.5 5.261 5.496 15.16 23.5 5.261 5.496

87 25.542 25 0.522 2.56 15.16 25 0.522 2.56

88 25.542 18 1.295 1.025 15.16 18 1.295 1.025

89 51.084 6.34 3.272 12.287 30.31 6.34 3.272 12.287

90 12.771 20 0.162 2.89 7.578 20 0.162 2.89

91 51.084 9.08 0.256 11.34 30.31 9.08 0.256 11.34

92 25.542 15.9 0.868 4.12 15.16 15.9 0.868 4.12

61
93 25.542 12.3 -0.824 3.59 15.16 12.3 -0.824 3.59

94 12.771 54 7.275 9.147 7.578 54 7.275 9.147

95 25.542 16.4 0.25 7.94 15.16 16.4 0.25 7.94

96 25.542 22 1.25 6.45 15.16 22 1.25 6.45

97 51.084 6.87 -0.287 3.25 30.31 6.87 -0.287 3.25

98 51.084 9.61 0.151 2.58 30.31 9.61 0.151 2.58

99 25.542 26.6 0.378 2.12 15.16 26.6 0.378 2.12

100 25.542 18 0.245 2.256 15.16 18 0.245 2.256

101 25.542 10.2 6.847 14.25 15.16 10.2 6.847 14.25

102 51.084 6.85 1.008 4.241 30.31 6.85 1.008 4.241

103 25.542 17.6 1.749 2.676 15.16 17.6 1.749 2.676

104 25.542 25 -0.915 3.32 15.16 25 -0.915 3.32

105 25.542 41 1.057 2.479 15.16 41 1.057 2.479

106 12.771 25.5 3.542 7.25 7.578 25.5 3.542 7.25

107 25.542 9.79 -2.113 2.533 15.16 9.79 -2.113 2.533

108 12.771 21 5.261 2.497 7.578 21 5.261 2.497

109 12.771 15 0.522 4.12 7.578 15 0.522 4.12

110 25.542 14.3 1.295 1.025 15.16 14.3 1.295 1.025

111 25.542 19.2 3.272 8.25 15.16 19.2 3.272 8.25

112 51.084 12.3 0.162 2.268 30.31 12.3 0.162 2.268

113 12.771 47 0.156 8.25 7.578 47 0.156 8.25

114 51.084 16.4 0.868 3.48 30.31 16.4 0.868 3.48

115 25.542 17.8 -0.824 4.25 15.16 17.8 -0.824 4.25

116 25.542 13.7 7.275 9.147 15.16 13.7 7.275 9.147

62
117 25.542 13 0.25 7.94 15.16 13 0.25 7.94

118 12.771 28 1.256 6.45 7.578 28 1.256 6.45

119 12.771 16 -0.287 3.54 7.578 16 -0.287 3.54

120 51.084 8.34 0.151 5.67 30.31 8.34 0.151 5.67

121 12.771 23 0.378 2.245 7.578 23 0.378 2.245

122 153.25 15 0.245 2.256 90.93 15 0.245 2.256

63