Reliability Assessment of Distribution System With Distributed Generation

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224376036

Reliability assessment of distribution system with distributed generation

Conference Paper · January 2009


DOI: 10.1109/PECON.2008.4762728 · Source: IEEE Xplore

CITATIONS READS
30 1,575

2 authors, including:

Mahmoud Fotuhi-Firuzabad
Sharif University of Technology
422 PUBLICATIONS   14,151 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Providing flexibility at the distribution network level under regulatory policies View project

Upcoming challenges of future electric power systems (sustainability and resiliency) View project

All content following this page was uploaded by Mahmoud Fotuhi-Firuzabad on 13 October 2015.

The user has requested enhancement of the downloaded file.


2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Reliability Assessment of Distribution System


With Distributed Generation
Pedram Jahangiri* and Mahmud Fotuhi–Firuzabad (Senior Member, IEEE) **
*
Email: pedram_jahangiri@ee.sharif.edu
** Email: fotuhi@sharif.edu

Department of Electrical Engineering


Center of Excellence in Power System Control and Management
Sharif University of Technology
Azadi Ave, Tehran, Iran

Abstract— Reliability assessment is of primary importance include photovoltaic, wind turbines, fuel cells, small and
in designing and planning distribution systems that operate micro-sized turbine packages, internal combustion engine
in an economical manner with minimal interruption of generators, and reciprocating engine generators [4].
customer loads. Distributed generation (DG) is expected to In this paper, DG units are in standby mode and start
play an increasing role in emerging power systems because when some load points have interrupted from the main
they use different type of resources and technologies to serve supply. Its use as backup power could reduce the customer
energy to power systems. DG is expected to improve the interruption cost and improve reliability of the system.
system reliability as its backup generation. Since DG units
are subject to failures as all other generation units, the Since DG units are subject to failures as all other
random behavior of these units must be taken into account generation units, the random behavior of these units must
in the analysis. Existence of DG units in a distribution be taken into account in the analysis. This uncertainty
system will effect on restoration time of load points. In this associated with the availability of DG units will affect on
paper, an algorithm for assessing restoration time of load restoration time of load points. This impact can be
points for a radial distribution system with back up DG assessed using the concept of conditional probability.
units is presented. The Impacts of DG units, their location, In this paper, an algorithm based on analytical method
number of DG units in each location and their availability, for assessing restoration time of load points of a
on distribution system reliability are investigated using a distribution system with DG units as backup generation is
real distribution network in Tehran utility/Iran. presented.
A number of system indices such as SAIFI, SAIDI,
Keywords ü Reliability Assessment; Distribution Systems; CAIDI and AENS are computed for a real radial
Distributed Generation; Backup Generation; Restoration distribution system .A sensitivity analysis is performed to
Time examine the impact of DG units, their location, and their
number, on reliability indices.
I. INTRODUCTION
The Reliability concepts can be applied to virtually any II. RELIABILITY ASSESSMENT OF DISTRIBUTION
engineering systems. In its broadest sense, reliability is a SYSTEM
measure of performance. This measure can be used to help Reliability assessment has become important for utility
systems meet performance criteria, to help quantify planners in recent years.
comparisons between various options, and to help make
Improved service reliability might be motivated by
economic decisions. The ultimate goal of reliability
government regulation or by market competition, but
analysis is to help answer to questions like “is the system
providing superior service at an attractive price is in the
reliable enough?” “Which scheme will fail less?” and
interest of both utilities and customers.
“where can the next dollar be best spent to improve the
system? ’’[1] A. Definitions
The goal of a power system is to supply electricity to its Zone: A circuit downstream of a protective device
customers in an economical and reliable manner. It is (breaker, recloser or fuse). For instance Zone Z1, Z2 & Z3
important to plan and maintain reliable power systems in Fig.1.
because cost of interruptions and power outages can have Sub-Zone: A circuit within a Zone, and downstream of a
severe economic impact on the utility and its customers. sectionalizing device (switch or sectionalizer). For
At present, the deregulated electric power utilities are instance Zone Z1 contains 4 sub-zones:Z1.0, Z1.1, Z1.2,
being restructured and operated as distinct generation, and Z1.3 in Fig.1.
transmission and distribution companies and the
responsibility of maintaining reliability of the overall Restoration Time: the time required to restore service to
power system is shared by all involved companies instead a load after the zone protective device opens to clear a
of by a single electric utility [2]. permanent fault. The restoration time for loads at the fault
location is the time required to repair/replace the faulted
Distributed generation (DG) is normally defined as component. However, appropriate switching action might
small generation units (<10 MW) installed in distribution restore other loads faster by first isolating the fault and
systems [3]. The applications of DG include combined then re-connecting the loads to the original (or to an
heat and power, standby power, peak shaving, grid alternate) source. The restoration time for loads not
support, and stand alone power. The DG technologies

1-4244-2405-4/08/$20.00 ©2008 IEEE 1551


2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

actually located at the fault is smaller than that required to removed from operation, would cause an interruption at a
make repairs (r) or to maneuver the switching devices (t). load point. Since the incidence of two or more equipment
failing at the same time is minimal, it is assumed that
B. Example single equipment failures or “first-order” cut-sets cause
Fig. 1 shows a radial distribution system with 3 zones. interruptions.
1) Fault in zone Z2 Using historical data, we assign failure rates and repair
Zone Z2 fuse operates causing interruption to customers time to each equipment.
in zone Z2. Restoration time for Zone 2 is equal to repair For each sub-zone, all equipment failures in the system
time (r). that may cause an interruption (cut-sets) are evaluated and
2) Fault in zone Z1.1 the resulting outage duration is computed.
Main breaker operates causing interruption to ALL The combined effect of all cut-sets for that sub-zone is
feeder customers. The fault is isolated by opening zone reported as load indices of interruption frequency and
Z1.1 switch and zone Z1.2 switch. Since there is no outage duration in that sub-zone.
alternate source, zones downstream of the faulted zone can
only be restored after removing and repairing the faulted λ z = frequency of interruption for sub-zone ‘z’
component. Restoration time for zone 1.1 and all rz = outage duration for sub-zone ‘z’
downstream zones is equal to repair time (r).Restoration
time for upstream zone Z1.0 is the minimum of r and t. cs = cut-set of sub-zone ‘z’
3) Fault in zone Z1.0 rc = outage duration in sub-zone ‘z’ due to cut-set ‘ cs ’
Main breaker operates causing interruption to ALL
feeder customers. However none of the zones can be λcs = interruption frequency in sub-zone ‘z’ due to cut-
restored, until the failed component is repaired. set ‘ cs ’
Restoration time for ALL zones is equal to repair time (r).
λ z = ¦ λcs (failure/yr) (1)
cs

¦ (λ ∗ r cs cs )
rz = cs
(hour) (2)
¦λ cs
cs

E. System Indices Definition

λi = Frequency of interruption for load point i.


Figure 1. A radial distribution system
N i = Number of customer in load point i.
C. Reliability Indices Computation U i = Average annual outage time at load point i.
Customer oriented indices are the interruption
frequency and the outage duration experienced at a load Lai = Average load of load point i
point. These indices are referred to as load-point indices
and are computed for each “sub-zone” in the feeder. All
customers’ loads within a sub-zone experience the “same” • SAIFI (System Average Interruption Frequency
interruption frequency and outage duration. Index): The average number of sustained outages, per
Since interruption frequency and outage duration are year, per customer over a defined area.
known at every load point on the feeder, system indices
such as SAIFI, SAIDI, CAIDI and AENS may then be
computed (see Fig.2). ¦ (λ ∗ N ) i i
SAIFI = i
(int/cust.yr) (3)
¦N i
i

• SAIDI (System Average Interruption Duration


Index): The average length of a sustained customer outage
experienced by a customer.

¦ (U ∗ N ) i i
SAIDI = i
(hr/cust.yr) (4)
Figure 2. load & system indices
¦N i
i

D. Load Indices Computation


The “Minimal Cut-Set” approach is used for computing
load indices. A “Cut-Set” is a set of equipment which, if

1552
2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

• CAIDI (Customer Average Interruption Duration e) Downstream sub-zone with unavailable DG units
Index): For those customers who experienced sustained thus the restoration time is equal to repair time of the
outages, the average length of the outage. section.
For assessing restoration time of downstream sub-zone
after isolation of the fault the uncertainty associated with
SAIDI
CAIDI = (hr / int) (5) the availability of DG units must be considered. Therefore,
SAIFI a capacity outage probability table (COPT) for all DG
• AENS (Average Energy Not Supplied): Total energy units should be formed in each location. It is a simple
not supplied (due to interruptions) divided by the total array of capacity levels and the associated probabilities of
number of customers. existence. If all DG units in each location are identical, the
capacity outage probability table can be easily obtained
using binomial distribution. If not, the COPT can be
¦ ( La ∗U ) i i
created by recursive technique in which units are added
sequentially to create the final model.
AENS = i
(kwhr/cust.yr) (6)
¦N i
i
For each level of capacity, operator should decide which
switches must be opened to produce an island and restore
healthy sub-zones.
III. MODELING AND EVALUATION TECHNIQUE The decision is based on the fact that for each level,
The following algorithm is for assessing restoration available capacity should be greater than all load points in
time of sub-zones when distribution system contains DG the island.
units. Now the restoration time of each sub-zone can be
First, all sub-zones in the network should be assessed for each contingency (short circuit fault) by
determined. weighting the restoration time with the probability of
existence of that level and sum these weighted terms.
Once a fault occurs in one of the sections of a sub-zone,
it causes main breaker of the substation to be The method is illustrated using the following example.
automatically opened. A. Example
The following steps must be performed: The network shown in Fig. 3 has 6 sections (A, B, C,
D, E, F), 5 spot loads (capacity of each spot load is equal
1) Find location of the fault. to C KVA), 3 DG units at the end of the feeder (capacity
of each unit is equal to C KVA), 5switches and 5 sub-
zones.
2) Isolate the faulted sub-zone:

I) First of all, open the switch that is relating to the


faulted sub-zone (upstream of the fault).

II) Open switches corresponding to adjacent sub-zones


of the faulted sub-zone (downstream of the fault).

3) Close the main breaker.

Figure 3. A radial distribution network


4) Develop several islands by opening some switches
of sub-zones and then utilize available DG units for For each sub-zone, all the contingencies should be
restoring these islands. considered.
The restoration time of these islands is equal to For instance, when a fault occurs in section B .the main
switching time plus time for starting DG units if they are breaker is opened. Once the location of the fault is
available, else the restoration time is equal to the repair determined, the fault should be isolated and then switches
time of the faulted section. The restoration time for some S1&S2 will be opened and thereupon the main breaker is
sub-zones that are down stream of the faulted sub-zone closed. Now, a COPT for DG units should be developed. It
and are not in the island with DG units is equal to repair is shown in Table I. Availability and unavailability of each
time of the faulted section. unit are assumed to be A and U=1-A respectively.
Consequently, after isolating the fault five groups of
sub-zones can be categorized as sorted below:
a) Faulted sub-zone, its restoration time is equal to the
repair time of the section.
b) Upstream sub-zone that can be restored through the
main supply.
c) Downstream sub-zone without DG that can not be
restored until the repair time of the section.
d) Downstream sub-zone with available DG units that
can be restored by starting DG units.

1553
2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

TABLE I. COPT ONCE A FAULT OCCURS IN SECTION B Table II shows the data associated with sections and load
Switches Restored points. Average failure rate of sections is assumed to be
Capacity should sub- 1.49(failure/yr.km).
Probability Time
In be zones
Opened by DG TABLE II. TEST SYSTEM DATA
3× C A 3 - Tst 3,4,5 Load
point Average Number of
2×C 2
3A U S3 Tst + SWT 4,5 Section
Length(km) Number load at end customer at
number
at node(KVA) end node
1× C 3AU 2 S3,S4 Tst + SWT 4 End node
1 0.42 1 90 137
0×C U3 - -(DG off) - 2 0.25 - - -
3 0.294 2 100.1 126
4 0.411 3 18.7 20
For illustration purpose the restoration time for each
5 0.190 - - -
sub-zone due to contingency B (fault in section B) is 6 0.190 4 90 284
calculated using conditional probability method. Sub- 7 0.340 5 269.5 210
zone1 will be restored through the main supply. 8 0.110 - - -
Tres _ subzone1 = FLT + SWT 9
10
0.124
0.030
6
7
50.6
87.6
57
135
Tres _ subzone 2 = FLT + SWT + Trp 11
12
0.124
0.200
8
9
90
85
172
170
13 0.140 10 200 190
T res _ subzone3= FLT + SWT + A3Tst + (1 − A3 )Trp 14 0.104 11 51.04 56
15 0.167 12 26.4 38
T res _ subzone 4= FLT + SWT + A3Tst 16 0.189 - - -
17 0.033 16 45.1 49
+ (3 A2U + 3 AU 2 )(Tst + SWT ) + U 3Trp 18 0.138 13 16.5 10
19 0.481 - - -
T res _ subzone5= FLT + SWT + A3Tst 20
21
0. 200
0.290
14
15
374
90
280
204

+ (3 A 2U )(Tst + SWT ) + (3 AU 2 + U 3 )Trp


In this network there are five locations for possible
installation of DG units that is shown in Figure 4 (these
Trp : repair time points are defined by the utility considering various
T st = start time of DG unit factors)
The purpose is to determine the best location and
Tres = restoration time . optimum number of DG units in order to maintain the
predefined reliability level. The capacity of each DG unit
FLT : fault location time is assumed to be 650 KVA.
SWT : switching time A computer program has been developed to perform
necessary computations. This program first computes
IV. STUDY RESULTS failure rate( λ ),restoration time(r), and annual outage
time(u) for each load point then calculate system indices
The application of the proposed technique to a multi- such as SAIFI, SAIDI, CAIDI and AENS as described
load point distribution system is clarified using the before. Three different case studies are considered. For
distribution system shown in Fig. 4 each case it is assumed that 0, 1, 2, 3, or 4 DG units can be
The test system is a real distribution network in Tehran established in each location and then system indices are
utility/Iran. The network has 16 load points and 2138 calculated.
customers. Because DG units are assumed as standby units they
only affect on outage time and doesn’t affect on
interruption frequency, hence SAIFI will be constant in
each case and is equal to 6.5932 (int/cust.yr).
A. Case 1
The effect of location of DG units on system indices are
investigated in this case. The results are shown in Fig. 5-6.
These Figures show the system reliability indices (SAIDI,
CAIDI, and AENS) of the test system for the comparative
studies. In these figures, each column pattern is associated
with a special mode (No DG, 1 DG unit, 2 DG units, 3 DG
units, and 4 DG units).
Figure 4. Test System Results show that indices such as CAIDI, SAIDI and
AENS are strongly sensitive to the location of DG units
The average fault location time and repair time for all consequently the location of DG units in a distribution
sections are assumed to be respectively 0.66 and 3 hours.

1554
2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

system is very important and the best location for the studies and to consider the impacts on system indexes of
certain number of DG units can be chosen from the results. DG units’ availabilities. This is illustrated in this case by
varying the availabilities of DG units on location 2 and
B. Case 2 determining their effects on system indexes. The results
The effect of number of DG units in each location on for SAIDI, CAIDI, and AENS for this location are shown
system indices is investigated in this case. The results are in Fig. 8-10. The results clearly demonstrate the reliability
shown in Fig. 5-7. The optimum number of DG units can enhancement of the system indexes (SAIDI, CAIDI, and
be found for each location by considering financial factors AENS) when the availabilities associated with DG units
and suitable level of indices. are increased.

NO DG 1DG 2DG 3DG 4DG 1 DG 2 DG 3 DG 4 DG


25 25

20 20
SAIDI(hr/cust.yr)

SAIDI(hr/cust.yr)
15 15

10 10

5 5

0 0
1 2 3 4 5 1 0.9 0.8 0.6 0.4 0.2
Location Availability Of DG Units

Figure 5. case1&2.SAIDI Figure 8. case3.SAIDI

NO DG 1 DG 2 DG 3 DG 4 DG 1 DG 2 DG 3 DG 4 DG
4
4
3.5 3.5

3 3
CAIDI(hr/int)
CAIDI(hr/int)

2.5 2.5

2 2

1.5 1.5

1 1
0.5 0.5
0 0
1 2 3 4 5 1 0.9 0.8 0.6 0.4 0.2
Location Availability Of DG Units

Figure 6. case1&2.CAIDI Figure 9. case3.CAIDI

No DG 1 DG 2 DG 3 DG 4 DG 1 DG 2 DG 3 DG 4 DG
20 20
18 18
AENS(kwhr/cust.yr)

AENS(kwhr/cust.yr)

16 16
14 14
12 12
10 10
8 8
6 6
4 4
2 2
0 0
1 2 3 4 5 1 0.9 0.8 0.6 0.4 0.2
Location Availability Of DG Units

Figure 7. case1&2.AENS Figure 10. case3.AENS

C. Case3 (sensitivity analyze) V. CONCLUSION


The impacts on system indices of availability of DG This paper proposed an analytical approach to study the
units are illustrated in this case. In the results presented in DG impacts on distribution system reliability indices. The
the previous cases, it was assumed that the availabilities of method assumes DG units as backup generation and
the DG units are 0.95. It is important to perform sensitivity

1555
2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

considers availability and unavailability of DG units. The


method is then applied to a real distribution network in
Iran and system indices such as SAIFI, SAIDI, CAIDI
and AENS has been computed for 3 cases with the
developed computer program. The results show when DG
units are applied as standby units, only affect on outage
duration and don’t affect on interruption frequency so
SAIFI will be constant. In addition, the results show that
indices are too sensitive to location, number and
availability of DG units. Hence, the optimum number of
DG units for the best location in distribution system can
be obtained using the proposed method.
REFERENCES
[1] R.E. Brown and S. Gupta and R. D. Christie and S. S. Venkata and
R. Fletcher “Distribution System Reliability Assessment Using
Hierarchical Markov Modeling” IEEE Transactions on Power
Delivery, Vol. 11, No. 4, October 1996
[2] A. A. Chowdhury and D. O. Koval, “Generation reliability impacts
of Industry-owned distributed generation sources,” in Industry
Applications Conference, vol. 2, Oct. 2003, pp. 1321–1327.
[3] Ackermann,T.; Andersson,G.; Soder, L.; "Distributed Generation:
A Definition”, Electric Power Systems Research, Vol 57, 2001, pp
195– 204.
[4] P. A. Daly and J. Morrison, “Understanding the potential benefits
of distributed Generation on power delivery systems,” in Proc.
Rural Electric Power Conf., 2001, pp. A2/1–A2/13.
[5] R. Billinton, R, N. Allan, "Reliability Evaluation of Power
Systems",2nd edition, Plenum Press, New York, 1996.
[6] R. Billinton and R. Allan, Reliability Evaluation of Engineering
Systems: Concepts and Techniques: Plenum Press, 1994.
[7] IEEE Standard 1366-1998, IEEE trial-use guide for electric power
distribution reliability Index, 1998.

VI. BIOGRAPHIES

Pedram Jahangiri was born in Iran. He


received B.Sc. Degree in Electrical Engineering
from Isfahan University of Technology (IUT),
Isfahan, Iran in 2006. He is currently working
toward the M.Sc. Degree in Electrical
Engineering at Sharif University of Technology
(SUT). His research interests include Power
System Reliability and Automation.

M. Fotuhi-Firuzabad (IEEE Senior Member,


99) was born in Iran. Obtained B.Sc. and
M.Sc. Degrees in Electrical Engineering from
Sharif University of Technology and Tehran
University in 1986 and 1989 respectively and
M.Sc. and Ph.D. Degrees in Electrical
Engineering from the University of
Saskatchewan in 1993 and 1997 respectively.
Dr. Fotuhi-Firuzabad worked as a postdoctoral
fellow in the Department of Electrical Engineering, University of
Saskatchewan from Jan. 1998 to Sept. 2000 and from Sept. 2001 to Sept.
2002 where he conducted research in the area of power system
reliability. He worked as an assistant professor in the same department
from Sept. 2000 to Sept. 2001. Presently he is an associate professor and
Head of the Department of Electrical Engineering, Sharif University of
Technology, Tehran, Iran. He is a member of center of excellence in
power system management and control.

1556

View publication stats

You might also like