EUROCON 2005

Serbia & Montenegro, Belgrade, November 22-24, 2005

Distributed Generation Impacts on Electric Distribution Systems Reliability: Sensitivity Analysis

Hamid Falaghi, Mahmood-Reza Haghifam, Member IEEE

Abstract - Recent developments in the electric power and power supply industries have raised a great deal of interest in distributed power generation (DG). DG will have impacts on distribution networks. In this paper, modelling of DGs in reliability assessment and computing of related indices such as SAIFI, SAIDI, CAIDI is introduced. Using model development and case studies it is discussed and shown that location and numbers of DGs effect in distribution reliability indices.

II. EFFECT OF DG ON SYSTEM RELIABILITY

Keywords - Distribution Systems, Distributed Generation, Reliability Modeling

I. INTRODUCTION

increasing role in future electric power systems. Studies have predicted that DG will be a significant percentage of all new generation going online[1, 2]. DG is related with the use of small generating units installed in distribution feeders, or customer load level [1]. The application of DG include combined heat and power, standby power, peak shaving, grid support, and stand alone power. Impact of DG on distribution systems can be positive or negative depending on the system operating condition[3]. The use of DG to improve system reliability is one of its important applications. Thus, one of the essential studies about this context is evaluation of effect of DG in reliability indices. In [4] a reliability model for determining the DG equivalence to a distribution facility for use in distribution systems planning studies in the new competitive environment is presented. In [5] a six-state model for DG, is proposed for evaluation its effect on distribution reliability. Modelling of DGs as negative loads in reliability calculation is proposed in refs [6-7]. This paper addresses the important area of how DG affect system reliability. It outlines the approach developed and perform a number of sensitivity analysis of system reliability to show the impact of this new generation on reliability indices such as SAIFI, SAIDI, CAIDI, ENS.

DISTRIBUTED Generation is expected to play an

The basic objective of the paper is to present a sensitivity analysis to present distributed generation effects on the distribution system reliability. This paper considers a simple illustrative distribution system as shown in Figure 1. The feeder is operated as radial feeders but it can be supported by a DG through a normally open switch. The average load at each load point is taken 400 kW. For example following a fault occurrence on first section, with no DG on the feeder, service of all load points must be interrupted during repair process but with DG as shown in Fig. 1, load points loads "E", "F", and "G" can be restored via the DG. Existence of DGs can reduce the outage duration and consequently increase the system reliability.
Main Supply
A

E
oad Point

DG Unit

Bmaker

N.C Switch

150 kVA

Fig. 1. A Typical distribution system with one DG

This simple example shows that DG can have a positive impact on distribution system reliability in the case of local utility supply interruptions. Distribution automation system in a distribution system including DGs, load transfer to other feeders via switches operation can be performed in fault condition in order to keep customer supply.
III. MODELLING AND EVALUATION TECHNIQUE

H. Falaghi is with the Department of Electrical Engineering, Tarbiat Modarres University, Tehran, Iran (phone: 98-21-8011001; fax: 98-21 -8005040; e-mail: falaghi@modares.ac.ir). M.-R. Haghifam is with the Department of Electrical Engineering, Tarbiat Modarres University, Tehran, Iran (phone: 98-21-8011001; fax: 98-21 -8005040; e-mail: haghifamgmodares.ac.ir).

A. Reliability Assessment Reliability evaluation of distribution systems has received considerable attention and there are a number of publications dealing with modelling and evaluation techniques. However a continuing need to extend and develop the techniques still remains as new design and operation approaches are introduced into the system [8]. There are several available methods for evaluating the reliability of distribution systems. Most convectional techniques developed for the reliability evaluation of radial distribution systems are generally based on minimal cut set approach or failure mode and effect analysis

The techniques used in this paper are based on FMEA, but they are extended so that the effect of DGs can be quantitatively assessed. The first step evaluates the reliability indices at each load point by identifying the
1465

(FMEA) [9].

1-4244-0049-X/05/$20.00 (C2005 IEEE

failure events leading to failure of the load point. This is achieved by discovering the closed and normally open paths from each load point to the feeding points. Based on these and on further information regarding the system configuration and component reliability data, the contribution of each failure event is calculated and then combined to give the load point reliability indices. The second step is to evaluate system indices from the load point indices [10]. The effect of DGs on the reliability can be assessed using the following algorithm: (a) Consider each load point of the system analyzed. (b) Consider each failure event of the load points in turn. (c) For each failure event determine how service can be restored as follows: In the The service to the load points in the upstream can be restored by switching actions through a main supply. The supply to the downstream area can be restored by switching actions through an alternative supply or DG. The restoration times of load points in each area will not be the same due to restoration sequence. Load points (or nodes) in a distribution system can be divided into five classes according to restoration times. Class I: the load points that are not affected by a fault. Class II: the load points in repaired area are in this class. Outage duration is equals the time to isolate the fault and to repair the faulted component. Class III: Outage duration for load points equal the time for isolating the fault and to restoration of service through a main supply or alternative tie point. Class IV: Outage duration for load points equal the time for isolation of the fault and service restoration through DGs. Class V: the load points that can not restore through tie points or DGs. With respect to capacity limitations, outage durations equal the time to isolate and to repair failed equipments. (d) Deduce load point indices by considering all the events leading to failure of the load point and their associated restoration procedure. (e) Assess the overall system indices by appropriately combining reliability indices of the load points.
B. Effect ofDG failures Unfortunately the DGs are subject to possible failures as all other technical systems. In the event of DG failure, the restoration process by DG become impossible. The implication in this concept is one of conditional probability associated with the outage time (r) is equal to the fault isolation and DG connecting time (TFL+TDG) if the DG work successfully, or egual to the fault isolation and repair time if (TFL+TRP) the DG fail to start. The average of this values can be evaluated using the concept of expectation: r = (TFL + TDG Distributed Generation Success) + (TFL + TP Distributed Generation Unsuccess)

IV. IMPLEMENTATION

Based on the model and algorithm presented in this paper, a computer software package has been developed by Distribution Networks Laboratory in Tarbiat Modarres University, Tehran, Iran. Our product takes a distribution network as an input, with all its connectivity and available data. The required data is saved in the attribute tables of the feeders. For each part of the algorithm, a script was written for perform the analysis and computation required. The entire algorithm was therefore converted to a combination of scripts. The application run by calling a "main" script that acts as the "Supervisory" script. This script calls the needed scripts for a specific stage.
V. APPLICATION STUDIES The test system is shown in Figure 2. This system is a real distribution network in Tehran utility/Iran. The network has 16 load points and the number of customers and average loads connected at each load point are shown in table 1. The average fault location time and repair time for all sections are taken 0.66 and 3 hour respectively. Length of sections are given in table 1. Average failure rate of feeder is taken 1.49 (fail/yr.km). In this study the average startup time of DGs is assumed 0.33 hour.
Lp2

Lp16
Loc5 17 4
5

14

3 2 15 6 6

20

1
Lp13
8 11

21 Lp5

Li<4

Lpl

Lp3

Lp12
DG
7

HV/MV Substation h Point (Distribution Substation) O Normally Closed Switch

lLoad

12

13

14

<ll. (j)Lp4
Loc1

L8Lp9 L10 Lp5

Lpl11
9

DG Distributed Generation Location

i
Lp6

10

--s
Lp7

Section Number
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

TABLE 1: TEST SYSTEM DATA. Loadpoint Length Average Load Number atCutmra

Fig. 2. Test System

[km]

+(TFL where UDG is the probability of DG failure.

= (TFL

+TRP) X UDG

+TDG) X (1 UDG)

0.42 0.25 0.294 0.411 0.190 0.190 0.340 0.110 0.124 0.030 0.124 0.200 0.140 0.104 0.167 0.189 0.033 0.138 0.481 0.200 0.290

end node 1
2 3 4 5

at end node [kVA]

90 100.1 18.7 90 269.5 50.6 87.6 90 85 200 51.04 26.4 45.1 16.5 374 90

Number of Customer at end node 137 126 20 284 210

57

6 7 8 9 10 11 12

135 172 170 190 56 38

49 10 280 204

16 13
14 15

1466

Three cases are studied for investigation of number and location of the DGs on reliability indices. Case A- For assessing of DGs locations on the reliability indices, 5 candidate locations are assumed as shown in Figure 2 from LocI to Loc5. In this case, with connecting of one 652 kVA DG location in each candidate point, effect of DG location in reliability indices are assessed using developed software. The results for this case are shown in Figures 3 - 6. As shown in Figure 3, DG location can not affected in SAIFI, but other indices such as CAIDI, SAIDI and ENS are strongly sensitive to the location of DG.
10
8

Case B- In this case effect of number of connected DGs in reliability indices is evaluated. In this case results for feeder without DG to installation of DGs in all of candidate locations are reported. Capacities of DGs are 650 kVA. The results of this case are shown in Figures 710. Results show that the rate of interruption is constant, but with increasing of DGs numbers, distribution reliability indices (SAIDI, CAIDI and ENS) can be decreased. However, decreasing rates are not constant. Rates of reliability indices improvement approach to zero with installation of sufficient number of DGs network. It is shown that for each feeder with a given configuration and failure rates and repair times we can find optimum number DGs with respect to economical aspects.
10 8
0

4 2 0

6
4 2

3 Location

3
Number

Fig. 3. Variation of SAIFI in case A

24 22
on zu
N,

Fig. 7. Variation of SAIFI in case B

25
w

j 18 - =
11

16

a0

7-~
1 2 3 Location 4 5

2015 10
5
O
--*---------*-

>,
"

12 10

c,

3
Number

Fig. 4. Variation of SAIDI in case A

43.5

Fig. 8. Variation of SAIDI in case B

4.2

3 2.52-

2
=
C

3.5

2.8

2.1

1.40.70
0

1.5

3 Location

3 Number

Fig. 5. Variation of CAIDI in case A

45000 40000 35000 30000 25000 20000 0
1 2 3
Location

Fig. 9. Variation of CAIDI in case B

50000 40000 -

30000
2000010000

Number

Fig. 6. Variation of ENS in case A

Fig.

10.

Variation of ENS in case B

1467

Case C- In order to study effect of DG failure, this case assume that there are three DGs with capacity of 650 kVA in at Loc2, Loc3, and Loc4. Reliability indices are calculated for probability of DGs failure as 0, 0.01, 0.05 and 0.1, respectively. The results for this case are shown in Figures 11 - 14. it can be seen clearly that the rate of interruption is constant, but with increasing of DGs failure probability, distribution reliability indices (SAIDI, CAIDI and ENS) can be increased.
10

8
4

sufficient number of these generation units has been added. Developing a complete economical modelling and using optimization methods such as genetic algorithm for problem solving in large scale distribution systems are under study and will be reported in further works.
0 0.01 0.05 0.1

VI. CONCLUSION In this paper the impacts of DGs on reliability of distribution systems are investigated. It has described the model and techniques needed to evaluate reliability of distribution systems with DGs. The proposed model has been applied to a real Iranian distribution network. The analyses show that the reliability indices are highly sensitive to locations of DGs. The reliability indices can be enhanced monotonically as number of DGs increased. It also show that the benefits obtained from DGs installation tend to asymptotes to limiting values when

0 UDG

[1]

Fig. 11. Variation of SAIFI in case C

12. .6
-

[2] [3] [4]

12. .3
1

2-

11. 7
11. .4
-

[5]
0 0.01
UDG

11. .1
0.05
0.1

[6]
[7]

Fig. 12. Variation of SAIDI in case C

1.9
1.85 1.8 1.75
1.7
_

[8]

[9] [10]

REFERENCES T. Ackermann, G. Andersson, and L. Soder, "Distributed generation: a definition", Electric Power System Research, Vol. 57, 2001, pp. 195-204. IEA, "Distributed Generation in Liberalized Electricity Markets", Paris, p. 128. H. Lee Willis and Walter G. Scott, Distributed Power Generation: Planning and Evaluation, Marcel Dekker, Inc., 2000. A. A. Chowdhury, Sudhir Kumar Agarwal, and Don 0. Koval, "Reliability Modeling of Distributed Generation in Conventional Distribution Systems Planning and Analysis", IEEE Transaction on Industry Applications, Vol. 39, No. 5, 2003, pp. 1493-1498 In-Su Bae et al., "Optimal Operating Strategy for Distributed Generation Considering Hourly Reliability Worth", IEEE Transaction on Power Systems, 2004 R. E. Brown, "Electric Power Distribution Reliability", Marcel Dekker, Inc, New York, Basel, 2002. R. E. Brown and L. A. A. Freeman, "Analyzing the reliability impact of distributed generation," in Power Eng. Soc. Summer Meeting, vol. 2, 2001, pp. 1013-1018. Weixing Li, Peng Wang, Zhimin Li, and Yingchun Liu "Reliability Evaluation of Complex Radial Distribution Systems Considering Restoration Sequence and Network Constraints", IEEE Transaction on Power Delivery, Vol. 19, No. 2, 2004, pp. 753-758. R. Billinton, R, N. Allan, "Reliability Evaluation of Power Systems",2nd edition, Plenum Press, New York, 1996. IEEE Standard 1366-1998, IEEE trial-use guide for electric power distribution reliability Index, 1998.

1.65 0 0.01 0.05

0.1

UDG

Fig. 13. Variation of CAIDI in case C

21500 21000 20500 20000 19500 19000 18500 0 0.01 0.05
0.1
-

-A

UDG

Fig. 14. Variation of ENS in case C

1468