2016 Bienniallnternational Conference on Power and Energy Systems:Towards Sustainable Energy (PESTSE)

Reliability Assessment ofDistribution System with

Renewable Distributed Generation
~ehetaj ~akandar Chandra Shekhar Reddy Atla Prof. Sankar Velamuri
M Tech Student, Dept. of EEE Team Lead Professor in Dept. of EEE
JNTU - Anantapur, India P RDC - Bangalore, India JNTU - Anantapur, India
E-mail: [email protected] E-mail: [email protected] E-mail:[email protected]

Abstract-Now-a-days, in evolving power systems Distributed conventional and renewable DG impacts on the reliability of
Generation (DG) is predictable to perform a vital role to serve imminent distribution systems is done in [2]. The DG
energy to the power systems as they practice various types of penetration impact on the reliability of a radial distribution
resources and advancements. Mainly DG can be related with two network is investigated in [3]. To picture the comprehensive
technologies i.e., conventional and renewable energy technology.
effects of PV systems, a mathematical model is given in [4].
The fluctuations may occur more in the power produced from
renewable resources and it is difficult to forecast. The evaluation For computing the availability, mean time to failure and
of reliability indices can be done using Monte Carlo Simulation failure frequency ofthe network, Markov model is used in [5].
(MCS) or analytical method. In the case of DG, the analytical A simulation based reliability evaluation technique with
technique is very difficult to apply to the problem due to its static detailed battery operation model is developed in [6]. Feeder I
nature of problem formulation. So, MCS method is used to assess of Bus 2 of RBTS is tested for simulation pro gram and system
the renewable models in distribution system reliability. Here the related indices in [7]. Total information needed to execute
DG's role is to boost the level of reliability by supplying the local basic reliability analyses to evaluate the reliability indices
load during interruptions. In this research work, the higher using analytical method are given in [8]-[10]. Modeling by
penetration ofDG will be connected at each customer load points
using comprehensive and Sequential Monte Carlo (SMC)
of the Roy Billinton Test System (RBTS) of bus 6. The foremost
aim of this research work is to enumerate and check reliability simulation, the influence of DG to distribution system
indices for several system structures. The eight state Markov reliability is shown in [11]. The evaluation of reliability
model is implemented in VC++ to estimate the failure rate and indices and system reliability concepts are in [12]-[13].
repair time for solar DGs. This model is used to determine the In this research work, mainly focused on developing a
reliability indices in reliability model. For renewable models reliability evaluation framework for radial distribution feeders
connected at customer end, a computer program is developed in the presence ofSolar DG located with the load. In Section 11,
using MCS technique in VC++ to demonstrate the improvement about the modeling of renewable energy sources is described.
in reliability indices using solar DGs. The tlow chart for Renewable DG is developed using Monte
Carlo simulation and is described in Section III. The developed
Keywords-Solar Distributed Generation (DG), Distribution simulation program is applied on a RBTS of bus 6 on Feeder-l
Network, System Reliability Evaluation, Reliability Indices, Time and validated with published results. In Section IV, the
sequential MeS, Roy Bi/linton Test System (RBTS). description of RBTS test system and validation of system
I. INTRODUCTION reliability indices without DG is given. Section V delivers the
results obtained from simulation program are summarized.
Distribution Generation (DG), a small scale generation
unit is installed in the distribution system which is typically 11. MODELING OF RENEWABLE ENERGY SOURCES
connected at customer load level, on distribution feeders, or at When the DG is not 100% reliable then, the system can be
the substations. As a backup generation, DG helps in modeled with eight state Markov model. The State Transition
improvement of the system reliability. The random actions of Diagram (STD) and the State Transition Matrix (STM) for the
DG generation units should be considered in the analysis as all distribution system model including renewable DG for case I
other generation units. DG mainly contributes reliability to the and case 2 is shown in Fig.l. Markov model proposed in this
customer side rather than on the utility or system side. To paper, consists of three components namely, Solar PV, the
satisfy the customer, one has to provide the best service with units used to store energy (batteries), and the units used to
least amount of power failure. Installation of DG at each power conditioning (inverters). The Mean Time to Failure
customer load point of a bus system will improve system (MTTF) and Mean Time To Repair (MTTR) values for two
reliability indices. During interruptions with the externailoads cases are calculated based on UP and DOWN states of the
to the DG facility, islanding of DG cannot be done in general. STD. The tlow chart for finding MTTF is shown in Fig. 2. The
This creates quality and safety problems to the utility in the failure rate and repair time for Solar DG are evaluated and a
maintenance and restoration processes. computer program is developed using VC++.
For estimation of distribution system reliability, a Stochastic state transition matrix for the Fig. 1 is presented
generalized technique is shown in [1]. Even when the below.
connection may not be simply radial, the assessment of

978-1-4673-6658-8/16/$31.00 ©2016 IEEE

 0 Axy ADe 0 ADS 0 0 0
Ayx 0 0 ADe 0 ADS 0 0
flDC 0 0 Axy 0 0 ADS 0
0 flDG Ayx 0 0 0 0 ADS
(J=
0 0 Axy 0
(I) Reo.d tS- matri.... ( Transition
flDS 0 0 ADG
0 flDS 0 0 Ayx 0 0 ADe lll.atri....)
0 0 flDS 0 flDC 0 0 Axy
0 Ayx 0 Form S - TTlatrix ( Inmcaterl
0 0 0 flDS flDC
matrix)

Solve N - Cl - S) l.
N - Fundaillental Man-ix

2:
n

MTTF1. = ,'\lu
j = l

T
Meon tUne to Failure OVlTTF)

Fig. 2. Flow chart for computing MTTF using Markov models

Analytical method is more complicated to process and not

suitable to evaluate system reliability for large-scale systems
and it is major advantage of MCS technique over analytical
assessment methods. Analytical techniques often need
contingency enumeration of a large number of states before
they can be minimized. MCS technique avoids this difficult by
sampling an illustrative set of system states.
There are two techniques in MCS, one is state sampling
Fig. I. STD for distribution system model including renewable DG (DG is technique (SST) and other is time sequential technique (TST).
not 100% reliable) In this research work, TST is used and to evaluate complex
distribution system, time sequential MCS is used and is showed
A. Case 1 in [3] and [14].
For evaluating the MTTF and MTTR, it is required to fmd
the absorbing states and non-absorbing states. In case 1, if the The safety elements like alternate supplies, fuses and
system can operate with either component 2 or 3 or both in an disconnect switches, and distribution lines/cables are the
operable state, then for this system the up states (non-absorbing primary elements of Distribution system. In Fig.l, the element
states) are SI, S3 and S4 used to find MTTF and the states S2, Ss- in the functioning state represented by up state and down state
S8 are down states (absorbing states) which are used to find the is nonfunctional due to failure.
MTTR [4]. Failure process is a conversion from the up state to down
B. Case 2 state. Due to the absence of element for maintenance or failure
of an element conversion occurs from up state to down state.
In this case, the up states (non-absorbing states) SI-SS, is
used to find MTTF and the states S6-S8 are down states With various probability distributions and random variables
(absorbing states), which are used to fmd the MTTR as shown the Time To Failure (TTF) parameter is generated. By using
in Fig. I. Flow chart for computing MTTF using Markov unifonn random number generator, uniform distribution is
models is shown in Fig. 2. Implementation of failure rate and directly generated and this numbers are transformed into TTF
repair time for the Solar DG is done in VC++ using tlow chart. by following equation

III. MONTE CARLO SIMULATION WITH RENEWABLE DG TTF= - (In U/A) *8760 (2)

The average annual unavailability (U), average outage time

(r) and load point average failure rate (A) are the basic Here, unifonnly distributed random variable is indicated as
reliability indices, which are used to fmd the load point indices U in the interval [0, 1], T= 8760, exponentially distributed.
and these can be calculated using (3) to (5), the system The parameter Time To Repair (TTR) is directly
reliability indices like SAIFI, SAlDI, CAIDI, ASAI, ASUI, considered as repair time. The detailed procedure for MCS
ENS and AENS are detennined using (6) to (12) given in tlow chart has been analyzed in Fig. 3.
Appendix - A.
 TAßLE I. SYSTEM RELTABILTTY INDICES WITHOUT DG

Published By Developed
System Indices Analytical Method ByMCS
[SI Technique
SAIFI 0.33566 0.33649

SAlDI 3.68285 3.63997

Generate random number for each element in the system and convert
it into HF, T1F~ C
- log (U)I A) X 8760 CAIDI 10.97195 10.8173

Find the element 'e' with minimum TlF ASAI 0.99957 0.99958

ASUI 0.00042 0.000415

Deline ' e' as faited element and generate random number and
convert it into repair time (R1) ENS 4.23152 4.17159

AENS 5.53865 5.4602

Determine the loeation of'e' and find the load points and faUure
duration which are affected due to failed element 'e' . The DG is connected at the load points (LPI-LP6) of
Feeder-l are shown in FigA. The failure rate and repair time
Generate a random number for ' e' and convert this number into new
for Solar DG is evaluated by Markov model analysis. The Bus-
H F. F t+ TIR+ newTlF 6 of the RBTS for Feeder-I (shown in Fig. 4) is helpful to
assess the reliability of the networked distribution system
including DG. All the load points and their customer types and
YES
reliability data can be found in [8, 9]. To assess the efficiency
of a radial distribution system with Solar DG using TST MCS,
a program is developed in VC++.
Calcu\ate the average value of load points faHme rate & ll KV
failure duration for the sampie years

YES

Calculate the system indices for the sampie

years

Fig. 4. Radial Distribution Network with Solar DG

Fig. 3. Flow chart for Monte-Carlo simulation The data is used from [8, 9]. Failure rates of lines and
cables are nearly proportional to their lengths. Then, in this
IV. SYSTEM DESCRIPTION AND VALIDATION network the failure rates are 0.065 f/ km-yr and 0.015 f/ km-yr
The main section includes a distribution line which is a for main feeder sections (LI-L6) and transformer respectively.
combination of line and disconnecting switches is installed in And repair time considered as 5hours for main and lateral,
each end or both ends. And the section which in general 200hours for transformer sections and 1 hour for switching
contains a line, fuse or their combination and trans former is time. The failure rate and repair time for Solar DG is used from
known as lateral section. Some constituents which are not Table III. Some constituents which are not considered are
considered are assumed to be 100% reliable. assumed to be 100% reliable.

Assurne each element as a constant failure rate, switching V. SYSTEM ANALYSIS

and repair times are exponentially distributed and assurne The failure rate and repair time data of the three
switching time, trans former replace time and distribution line components are tabulated in Table 11.
repair time standard deviations as one hour. Failure rates of
lines and cables are nearly proportional to their lengths. The TABLEII. FAlLURE RATES AND REPA1R RATES FOR 3 COMPONENTS
main feeder sections failure rate is 0.065f/km-yr (LPI-LP6).
s_
System Reliability Indices without DG is developed in VC++ No.
AnG /LnG hy AYx Ans /Lns
using MCS and are validated with published results. The 0.02083 3.5706 1.4305
results are presented in Table I; it is observed that the obtained I 2 *10.6 24 *10.4 107
3
results are in-line with published results.
 By using Table 11 as inputs for finding failure rate and ApPENDIX-A
repair time for solar DG is implemented in VC++ and results The equations used to calculate load point and system
obtained are shown in Table 111. indices are given as:
TABLE III. FAILURE RATE AND REPAIR TIME OF SOLAR DG ATTAINED
FROMVC++
(3)

S.No. Parameters Casel Case2 (4)

I Fai Iure rate 0.032506 0.001955
(5)
2 Repairtime 287.5924 80.3331

Here, the component I failure rate and average repair time

The results obtained from VC++ for the two cases are are Ai and ri, and there pair time, unavailability and average
validated with published results shown in Table IV and Table failure rate at load point p are rp, Up and Ap respectively.
V respectively.
System Average Interruption Frequency Index
TAßLE IV. LOAD POINT UNAVAILABILITY FOR FEEDER I (SAIFI): total number of customer interruptions
total number of customers served
Published by Oeveloped by MCS with OG k
Load
points
Analytical method
Case I Case2
SAIFI = Lii/'iNi (6)
without OG ISJ Li=l Ni
I 3.67 2.55 l.09809
System Average Interruption Duration Index
2 3.69 1.206 1.3543
3 3.72 2.752 0.882844 (SAIDI):sum of all customer interruption durations
total number of customers served
4 3.67 2.792 1.1952
5 3.68 2.187 l.41401
(7)
6 3.68 l.623 0.804488
Customer Average Interruption Duration Index
TABLE V. SYSTEM PERFORMANCE RELIABILITY INDICES
(CAIDI):sum of customer interruption durations _SAlDI
total number of mterrupted customers SAIFI
Published by Developed by MCS with DG
System Analytical (8)
Indices method without Case 1 Case2
DG[SI
SAIFI 0.33566 0.369469 0.338785
Average Service Availability Index
SAlDI 3.6828 2.20675 1.12093 (ASAI):total number ofcustomer hours availableduring a year
CAIDI 10.9719 5.97277 3.30867 total customer hours demanded
ASAI 0.99957 0.999748 0.999872
(9)
ASUI 0.00042 0.000251912 0.000127
ENS 4.23152 2.48934 1.29243
Average service Unavailability Index
AENS 5.53865 3.25829 l.69166
(ASUI): totalnumberofcustomerhoursnotavailable
totalnumbero f customershoursdemanded
From Tables IV and V, it is observed that the ASUI= I-ASAI (10)
Unavailability of each load point with DG (shown in Table IV)
is improved for both the cases. As shown in Table V, it is Energy Not Supplied Index: total energy not delivered by
perceived that the perfonnance of system reliability index is the system (ENS):
amended while SAIFI is not improved. Hence system
reliability indices are enhanced in the presence of Solar DG at ENS = I~=l UiLa(i) (11)
each customer load point. Average Energy Not Supplied (AENS):
VI. CONCLUSION AENS = __
to_t_al_e_n_er..:;g:.::.y_n_o_ts_u-,-p,-p_lie_d_ _
totalnumbero f customersserved
A reliability evaluation method in the presence of Solar DG
to reduce the outage is evaluated in this paper. The modeling of (12)
renewable resources and procedure for evaluation of system
reliability indices is developed by MCS technique and applied
Here, at load point I, the annual outage time is Ui and
on RBTS of bus 6 of Feeder-I. The tlow chart and program is
number of customers is Ni and average load connected is La(i)
developed for implementing time sequential MCS method in
and number of hours in a calendar year are 8760.
VC++ to calculate load point and system indices. The main
contribution of DG to the system reliability is observed in both
the SAlDI and CAIDI indices.
 REFERENCES [8] R.N. Allan, R. Billinton, I. Sjarief, L. Goel, K.S. So, "A reliability test
system for educational purposes- Basic distribution system data and
results", IEEE Trans. on Power Syst., vol. 6, no. 2, pp. 813-820, May
[I] Roy Billinton and Peng Wang, "A Generalized Method for Distribution 1991.
System Reliability Evaluation", in Proc. IEEE WESCANEX, , pp. 349-
354, 1995. [9] Satish Jonnavithula, "Cost! Benefit Assessment of Power System
Reliability", Ph.D. thesis, Dept. of Elect. Eng., University of
[2] Mohammad AI-Muhaini and Gerald T. Heydt, "Evaluating Future Power Saskatchewan, 1997.
Distribution System Reliability Inc1uding Distributed Generation", IEEE
Trans. Power Delivery, vol. 28, no. 4, Oct. 2013. [10] Binendra Shakya, "Repair duration eflects on distribution system
reliability indices and customer outage costs". M.S. thesis, Dept. of
[3] M.Hlatshwayo, S.Chowdhury, S.P.Chowdhury and K. O. Awodele, Elect. and Comp. Eng., University of Saskatchewan Saskatoon,
"Impacts of DG Penetration in the Reliability of Distribution Systems", Saskatchewan, Canada, Feb. 2012.
in Proc. IEEE Int. Conf. on Power Syst. Tech., pp. 1-8,2010.
[11] Fangxing Li and Nura Sabir, "Monte Carlo simulation to evaluate the
[4] Zulu Esau and Dilan Jayaweera, "Reliability assessment in active reliability improvement with DG connected to Distribution Systems", In
distribution networks with detailed effects of PV systems", J. Mod. Proc. 8tl' WSEAS Int. Conf. on Elect. Power Syst., High Voltages,
Power Syst. Clean Energy, pp. 59-68,2014. Electric Machines (Power '08), USA, pp 177-182,2008.
[5] Mohammad Almuhaini, "An Innovative Method for Evaluating Power [12] Roy Billinton, Ronald N. Allan, "Reliability Evaluation of Engineering
Distribution System Reliability," Ph.D thesis, Dept. of Elect. Eng., Systems Concepts and Techniques", Springer publications, Hyderabad,
Arizona State University, Dec. 2012. 2007.
[6] Avinash Banajiger and Visvakumar Aravinthan, "Radial Feeder [13] V. Sankar, "System Reliability Concepts", Himalaya Publishing House,
Reliability Evaiuation in the Presence of Battery Storage using Modified 2015.
Sequential Monte Carlo Simulation", IEEE Conf., Dept. of Elect. Eng.
and Comp. Sci., Wichita State University, Wichita, Kansas, USA, pp. 1- [14] Roy Billinton and Peng Wang, 'Teaching Distribution System
6,2014. Reliability Evaluation Using Monte Carlo Simulation", IEEE Trans.
Power Syst., vol. 14, no. 2, pp. 397- 403, May 1999.
[7] Nisha R. Godha, Surekha R. Deshmukh, Rahul and V. Dagade, L Time
Sequential Monte Carlo Simulation for Evaluation of Reliability Indices
ofPower Distribution System", in Proc. ISCI, pp. 1-9,2012.