ECpE Department

Distribution System Reliability

Dr. Zhaoyu Wang

1113 Coover Hall, Ames, IA
[email protected]
Distribution System Reliability

Acknowledgement:
The slides are developed based in part on Electric Power and Energy
Distribution Systems, Models, Methods and Applications, Subrahmanyan S.
Venkata, Anil Pahwa, IEEE Press & Wiley, 2022

ECpE Department
1. Motivation
• The main purpose of distribution systems is to provide electricity to
customers to conduct the activities that require electricity. Assessment of
the distribution system to determine the extent to which electricity is
made available to the customers without interruptions provides a
measure of system reliability.

• Reliability is defined as the ability of the system to provide electricity

without interruptions, and resiliency is defined as the ability of the system
to recover from extreme or unplanned events.

• The subject of system resiliency and associated measures are still evolving.

• The focus of the slide is on reliability, although resiliency has an impact on

it.
3

ECpE
Department
1. Motivation
• The main motivation for reliability assessment is to analyze and improve system
performance:
• enhances customer satisfaction and satisfies regulatory requirements
• provides information for maintenance scheduling, the basis for new or expanded system
planning
• determines performance‐based rate making
• Various operation, maintenance, and design strategies can be used to enhance reliability. The
reliability assessment therefore begins at the design and planning stages to build a new
substation, to upgrade existing facilities, to add new feeders, and to identify poorer
performing (weak) sections of the system. Operationally, the assessment will be needed to
reconfigure the system for reduction in the customers affected, to add tie points to other
feeders, or to develop postfault switching plan.
• Performance-based rate making in de or reregulated environments:
• aids in forecasting utility's revenue
• assists in quantifying the “quality” of power delivered to customers in evaluating the
regulatory requirements specified by Public Utility Commissions (PUCs)
• also useful in benchmarking the system performance in comparison to others.
4

ECpE
Department
1. Motivation
• Distribution System Performance & Reliability:
• Overall performance: measured in terms of system losses and voltage profile
• Reliability: measured with respect to probability of experiencing outages and
to customer satisfaction.
• Outages: caused by failure of equipment due to bad quality, aging, human error, or
extreme weather events. As all these causes have uncertainties associated with
them, we rely extensively on probabilistic analysis to quantify system reliability.
• Customer satisfaction is measured in terms of the number of momentary and
sustained interruptions, the duration of outages, the number of customers
affected, and the number of customer complaints.
• Reliability can be improved by hardening or upgrading the entire system: not cost
effective.
• A targeted approach to selectively harden the system will result in optimal results.
Similarly, maintenance techniques can be enhanced to obtain optimal results.

ECpE
Department
2. Basic Definitions
• Various standardized indices are used for measuring reliability and associated
computations. Definitions are given in this section to aid the readers in
understanding the factors that affect the calculation of indices. Many of these
definitions were taken directly from The Authoritative Dictionary of IEEE Standards
Terms, 7th Edition and/or IEEE Standard 1366‐2012.
a) Connected Load: Connected transformer kVA, peak load, or metered demand on
the circuit or portion of circuit that is interrupted. When reporting, the report
should state whether it is based on an annual peak or on a reporting period peak.
b) Customer: A metered electrical service point for which an active bill account is
established at a specific location (e.g. premises).
c) Customer Count: The number of customers either served or interrupted
depending on the usage.
d) Forced Outage: The state of a component when it is not available to perform its
intended function due to an unplanned event directly associated with that
component. 6

ECpE
Department
2. Basic Definitions
e) Interrupting Device: A device whose purpose is to interrupt the flow of power,
usually in response to a fault. Restoration of service or disconnection of loads can
be accomplished by manual, automatic, or motor‐operated methods. Examples
include transmission circuit breakers, feeder circuit breakers, line reclosers, line
fuses, sectionalizers, and motor‐operated switches.
f) Interruption: The loss of service to one or more customers connected to the
distribution portion of the system. It is the result of one or more component
outages, depending on the system configuration. Note that the outage of a
component does not necessarily result in interruption.
g) Interruption Duration: The time from the initiation of an interruption to a
customer until service has been restored to that customer. The process of
restoration may require restoring service to small sections of the system until
service has been restored to all customers. Each of these individual steps should
be tracked, collecting the start time, end time, and the number of customers
interrupted for each step.
7

ECpE
Department
2. Basic Definitions
h) Interruptions Caused by Events Outside of the Distribution System: Outages that
occur on generation, transmission, substations, or customer facilities that result in
the interruption of service to one or more customers. While this is generally a
small portion of the number of interruption events, these interruptions can affect
many customers and may last for an exceedingly long duration.

i) Lockout: The final operation of a recloser or circuit breaker to isolate a persistent

fault or the state where all automatic reclosing has stopped. The current ‐carrying
contacts of the overcurrent protecting device are locked open under these
conditions.

j) Loss of Service: A complete loss of voltage on at least one normally energized

conductor to one or more customers. This does not include any of the power
quality issues such as sags, swells, impulses, and harmonics.

k) Major Event: An event that exceeds reasonable design and or operational limits of
the electric power system. It includes at least one major event day (MED).

ECpE
Department
2. Basic Definitions
l) Major Event Day: A day in which the daily system average interruption duration index (SAIDI)
exceeds a threshold value, TMED. For the purposes of calculating the daily system SAIDI, any
interruption that spans multiple calendar days is accrued to the day on which the interruption
began. Statistically, days having a daily system SAIDI greater than TMED are days on which the
energy delivery system experienced stresses beyond that normally expected (such as severe
weather). Activities that occur on MEDs should be separately analyzed and reported.

m) Momentary Interruption: A single operation of an interrupting device that results in a voltage

zero. For example, two circuit breaker or recloser operations (each operation being an open
followed by a close) that momentarily interrupt service to one or more customers is defined
as two momentary interruptions.

n) Momentary Interruption Event: An interruption of duration limited to the period required to

restore service by an interrupting device. Such switching operations must be completed
within a specified time of five minutes or less. This definition includes all reclosing operations
that occur within five minutes of the first interruption. For example, if a recloser or circuit
breaker operates two, three, or four times and then holds (within five minutes of the first
operation), those momentary interruptions shall be considered one momentary interruption
event.

ECpE
Department
2. Basic Definitions
o) Outage (Electric Power Systems): The state of a component when it is not available to perform its
intended function due to some event directly associated with that component. (Note: An outage
may or may not cause an interruption of service to customers, depending on the system
configuration. This definition derives from transmission and distribution applications and does not
apply to generation outages.)
p) Planned Interruption: A loss of electric power that results when a component is deliberately taken
out of service at a selected time, usually for the purposes of construction, preventative
maintenance, or repair. (Note: This derives from transmission and distribution applications and
does not apply to generation interruptions. The key test to determine if an interruption should be
classified as a planned or unplanned interruption is as follows: if it is possible to defer the
interruption, the interruption is a planned interruption; otherwise, the interruption is an
unplanned interruption.)
q) Planned Outage: The state of a component when it is not available to perform its intended function
due to a planned event directly associated with that component.
r) Reporting Period: The time period from which interruption data is to be included in reliability index
calculations. The beginning and end dates and times should be clearly indicated. All events that
begin within the indicated time period should be included. A consistent reporting period should be
used when comparing the performance of different distribution systems (typically one calendar
year) or when comparing the performance of a single distribution system over an extended period
of time. The reporting period is assumed to be one year unless otherwise stated.
10

ECpE
Department
2. Basic Definitions
s) Step Restoration: A process of restoring interrupted customers downstream from
the interrupting device/component in stages over time.
t) Sustained Interruption: Any interruption not classified as a part of a momentary
event. That is, any interruption that lasts more than five minutes.
u) Total Number of Customers Served: The average number of customers served
during the reporting period. If a different customer total is used, it must be clearly
defined within the report.
v) Unplanned Interruption: An interruption caused by an unplanned outage.

11

ECpE
Department
3. Reliability Indices
• IEEE Standard 1366 (published in 1998) with 12 most significant indices, provides a guide for
utilities to assess the reliability of distribution systems.The indices are categorized into
system-level and customer-level indices.

a) System-level indices
• Frequency of outages (SAIFI, CAIFI, ASIFI)
• Duration of outages (SAIDI, CAIDI, ASIDI)
• Momentary outages (MAIFI, MAIFIE )
b) Customer-level indices
• Frequency (CAIFI, CEMIn, CEMSMIn )
• Duration (CTAIDI)
• The indices can be classified based on sustained or momentary outages. They can also be
classified based on duration and frequency indices. Ultimately, most of these are measures of
availability of the system under study or investigation. Unfortunately, there is no one measure
that can describe the reliability of the distribution system completely because of its complex
nature.
12

ECpE
Department
3.1 Basic Parameters
The following parameters specify the data needed to calculate the indices for the reporting period:
An interruption event
Restoration time for each interruption event
Total number of sustained interruption events
K Number of interruptions experienced by an individual customer
CI Customers interrupted
CMI Customer minutes interrupted
Total number of distinct customers who have experienced a sustained
interruption
Total number of customers who experienced n or more sustained
Interruptions
Total number of customers who have experienced a sustained
interruption of more than S hours
Total number of customers who have experienced more than T hours
of sustained interruptions

13

ECpE
Department
3.1 Basic Parameters
Total number of customers who have experienced n or more combined
sustained and momentary interruption events
Number of momentary interruptions
Number of momentary interruption events
Number of interrupted customers for each sustained interruption event

Number of interrupted customers for each momentary interruption event

Total number of customers served for the area
Connected kVA load interrupted for each interruption event
Total connected kVA load served
Threshold value for the MED identification

14

ECpE
Department
3.2 Sustained Interruption Indices
• Service interruptions lasting more than five minutes are classified as sustained
interruptions.

• Interruptions shorter than five minutes are considered part of a momentary event.

• The demarcation of five minutes is based on the industry's recognition that many
temporary faults can be resolved through reclosing operations within one minute.

• A permanent fault, which causes sustained interruption, needs physical inspections

and rarely can be cleared in less than five minutes.

• The industry has established this demarcation to differentiate between sustained

and momentary interruptions.

15

ECpE
Department
3.2 Sustained Interruption Indices
1) System Average Interruption Frequency Index (SAIFI)
This index gives the average number of times a customer experienced a sustained interruption
over a predefined period.

Σ 𝑖 𝑁 𝑖 CI
¿ =
𝑁𝑇 𝑁𝑇

2) System Average Interruption Duration Index (SAIDI)

This index gives the total average duration of interruption experienced by a customer during a
predefined period. It is typically measured in customer minutes of interruption.
Total Customer Minutes of Interruption
SAIDI ¿
Total Number of Customers Served
¿ ¿
Σ𝑖 𝑟 𝑖 𝑁 𝑖 CMI
¿ =
𝑁𝑇 𝑁𝑇

16

ECpE
Department
3.2 Sustained Interruption Indices
3) Customer Average Interruption Duration Index (CAIDI)
This index gives the average time needed to restore service.

∑ ❑𝑟 𝑖 𝑁 𝑙 CMI
𝑖
¿ =
CI CI
Also,
SAIDI
CAIDI =
SAIFI
4) Customer Total Average Interruption Duration Index (CTAIDI)
This index gives the total average time that customers who experienced at least an interruption
were without power. This index is similar to CAIDI except that customers with multiple
interruptions are counted only once.
Total Customer Minutes of Interruption
CTAIDI ¿
Total Number of Distinct Customers Interrupted
¿ ¿
Σ 𝑖 𝑟 𝑖 𝑁 𝑖 CMI
¿ = 17
CN CN

ECpE
Department
 3.2 Sustained Interruption Indices
5) Customer Average Interruption Frequency Index (CAIFI)
This index gives the average frequency of sustained interruptions for those customers experiencing
sustained interruptions. The customer is counted once regardless of the number of times
interrupted for this calculation.

Σ 𝑖 𝑁 𝑖 𝐶𝐼
¿ =
CN CN

6) Average Service Availability Index (ASAI)

This index gives the percentage of time that a customer has received power during the defined
reporting period.
Customer Hours of Service Availability
ASAI ¿
Customer Hours of Service Demand
¿ ¿
If the reporting period is one year, the number of hours are 8760 for normal years and 8784 for a leap
year. 18

ECpE
Department
3.2 Sustained Interruption Indices
7) Customers Experiencing Multiple Interruptions (CEMIn)
This index gives the fraction of individual customers experiencing more than n sustained
interruptions.
Total Number of Customers that Experienced 𝑛
or More Sustained Interruptions
CEMI 𝑛=
Total Number of Customers Served
CN ( 𝑘 ≥ 𝑛 )
¿
𝑁𝑇

8) Customers Experiencing Long Interruption Durations (CELID)

This index gives the fraction of individual customers who experience interruptions with durations
longer than or equal to a given time. The time is either the duration of a single interruption (S) or
the total amount of time (T) that a customer has been interrupted during the reporting period. For
the single interruption duration, we get
Total Number of Customers that Experienced
an Interruption of 𝑆 or More Hours
CELID 𝑆 ¿
Total Number of Customers Served
¿ CN ¿
( 𝑡 ≥ 𝑆 )
¿ 19
𝑁 𝑇

ECpE
Department
3.2 Sustained Interruption Indices
And for the total interruption duration, we get;

CN ( 𝑡 ≥ 𝑇 )
¿
𝑁𝑇

20

ECpE
Department
 3.3 Load Based Indices
• These Indices use load interrupted instead of customers affected.They are useful for
measuring system performance in areas that serve relatively few customers and have
large concentrations of load, such as industrial and commercial customers.

1) Average System Interruption Frequency Index (A SIFI)

This index gives the average number of times the system experienced sustained interruptions
over a predefined period.
Total Connected kVA of Load Interrupted ∑ ❑ 𝐿𝑖
ASIFI ¿
Total Connected kVA Served ¿ 𝑖
𝐿𝑇
¿ ¿
2) Average System Interruption Duration Index (ASIDI)
The index gives the average duration of system load interruption based on connected load.
Total kVA duration of Load Interrupted ∑ ❑𝑟 𝑖 𝐿𝑖
A SIDI = ¿ 𝑖
Total Connected kVA Served 𝐿𝑇

21

ECpE
Department
 3.4 Momentary Interruption Indices
• These indices are based on the momentary interruptions experienced by the customers.

1) Momentary Average Interruption Frequency Index (MAIFI)

This index gives the average frequency of momentary interruptions experienced by customers
over a duration.
Total Number of Customer Momentary Interuptions ∑ ❑ ℑ𝑖 𝑁 𝑚𝑖
MAIFI = ¿ 𝑖
Total Number of Customers Served 𝑁𝑇

2) The Momentary Average Interruption Event Frequency Index (MAIFI E)

This index gives the average frequency of momentary interruption events. It does not include the
events immediately preceding a sustained interruption.
Total Number of Customer Momentary Interuptions Events
MAIFI 𝐸 ¿
Total Number of Customers Served
¿ ¿
∑ ❑ ℑ𝐸 𝑁 𝑚𝑖
𝑖
¿
𝑁𝑇

22

ECpE
Department
3.4 Momentary Interruption Indices
3) Customers Experiencing Multiple Sustained Interruption and Momentary Interruption
Events Index (CEMSMIn)
This index is the ratio of individual customers experiencing n or more of both sustained
interruptions and momentary interruption events to the total customers served. It is useful in
identifying customer issues that are hidden in averages.
Total Number of Customers Experiening 𝑛
or more Interruptions
CEMSMI 𝑛 ¿
Total Number of Customers Served
¿ ¿
CNT (𝑘 ≥ 𝑛)
¿
𝑁𝑇

23

ECpE
Department
3.5 Sustained Interruption Example
• Consider a feeder serving 1000 customers as shown in
the Figure with several single‐phase laterals connected
to it through fuses.
• Protecting scheme:
• The recloser opens whenever there is a fault
downstream.
• The recloser closes after a short delay and opens
again if the fault is not cleared.
• It will reclose multiple times based on the
selected settings for number of reclosing
operations and will lock out after that if the fault
is still there.
• The fuse opens if the fault is downstream of the
Figure. An example distribution feeder.
fuse, or the recloser opens if the fault is on the
main feeder.
• The breaker opens if there is a fault anywhere
downstream or it but upstream of the recloser. It
follows an operation procedure similar to
recloser.
• Opening of the breaker causes an interruption to
all the customers served from this feeder. 24

ECpE
Department
 3.5 Sustained Interruption Example

Table. Log of interruptions in the system

• A part of the outages logged in the system is provided in the table.

• Note that the outage on 12 May has a total duration of 3 minutes and 11 seconds, which is a momentary
interruption according to the cutoff time of 5 minutes.
• Since this outage interrupted all the customers, it must have been due to operation of the breaker or could
be due to an event in the transmission systems.
• Note that 400 customers experienced at least one sustained interruption, and 150 customers (connected
to lateral with F2, F4, and F6 faults) experienced four sustained interruptions and total interruption of
longer than three hours.
25

ECpE
Department
3.5 Sustained Interruption Example
• Various indices can be computed based on the data in this table.

26

ECpE
Department
3.5 Sustained Interruption Example
• Note that 150 customers experienced four sustained interruptions with a total duration of
200 minutes and 5 seconds, which gives CN(k ≥ 4) and CN(t ≥ 3) equal to 150.
• Also, the longest interruption of 121 minutes and 30 seconds due to F1 affected 400
customers, which gives CN(s ≥ 2) equal to 400.
• If we count the total interruption including momentary and sustained, 250 customers
experienced three or more interruptions, which gives CNT (k ≥ 3). We can use this information to
compute the additional indices.

27

ECpE
Department
3.6 Momentary Interruption
Example
• The table shows a sample of
momentary interruptions
recorded for the system.
• The recorded data shows a
total of 6600 momentary
interruptions (IMi) and 3600
momentary events (IM)E
• The momentary interruption
indices are computed as:
Table. Log of momentary interruptions in the system

28

ECpE
Department
4. Major Event Day Classification
• Distribution systems are designed for handling outages that happen under normal
operation.
• Certain unforeseen events, mainly due to extreme weather, can push the system to the
limit by causing numerous outages. Such events skew the reliability performance of the
system and thus are excluded from the calculations of reliability indices.
• A Major Event Day (MED) is defined as a day in which the daily system SAIDI exceeds
threshold value, TMED.
• Using SAIDI as the index because:
• It has led to consistent results regardless of the utility size.
• It is a good indicator of operational and design stress on the system.
• For calculating the daily system SAIDI, any interruption that spans multiple days is
accrued to the day on which the interruption begins.

• TMED value is calculated at the end of each reporting period (typically 1 year) for use
during the next reporting period.
29

ECpE
Department
4. Major Event Day Classification
• The process called “Beta Method” is used to identify MEDs
• The major purpose of Beta Method is to allow major events to be studied separately from
daily operation and in the process to better reveal trends in daily operation that would be
hidden by the large statistical effect of major events.
• Specific steps of the Beta Method are:
1. Gather daily SAIDI values for five sequential years until the end of the reporting period. If
less than five years of data are available, use all the available historical data, as less than
five years may not yield accurate results.
2. Exclude days with zero SAIDI from the dataset.
3. Compute the natural logarithm (ln) of SAIDI for all the days in the dataset.
4. Calculate the average (α) of the logarithms of the data set.
5. Calculate the standard deviation (β) of the logarithms of the data set.
6. Compute the value of k using the following equation:

or,
30

ECpE
Department
4. Major Event Day Classification
7. The Table provides probabilities and the expected number of
MEDs for different values of k in the Beta Method. The value
of k determines the threshold for identifying MEDs based on Table. Probability of exceeding as a function
the daily system SAIDI. However, there is no analytical of k
method for selecting the allowed number of MEDs per year.

8. To address this, a recommended value of k, namely k = 2.5,

has been established based on consensus among members
of the IEEE Power and Energy Society's Distribution Reliability
Working Group

9. Any day with daily SAIDI greater than the threshold value
TMED that occurs during the subsequent reporting period is
classified as a MED.

31

ECpE
Department
5. Causes of Outages
• Each piece of equipment in a distribution system has a certain probability of failing.
• When first installed, a piece of equipment can fail due to poor manufacturing,
damage during shipping, or improper installation.
• Healthy equipment can fail due to high currents, extreme voltages, animals, and
severe weather.
• Sometimes equipment may fail due to chronological age, chemical decomposition,
contamination, and mechanical wear.
• Failures in distribution systems can be categorized into three groups: intrinsic
factors, external factors, and human factors.
• Intrinsic factors: age of equipment, manufacturing defects in equipment, and the
size of conductors.
• External factors: trees, birds/animals, wind, lightning, and icing.
• Human factors: vehicular accidents, accidents by utility or contractor work crew,
and vandalism.

32

ECpE
Department
5. Causes of Outages
• Since distribution system overhead lines are highly exposed to the atmosphere,
external factors are the major causes of damages or failures. Intrinsic factors, on
their own, generally do not endanger the reliability of such lines. Their effect can be
seen only when they are combined with some other factors.

• For example, a very old small conductor would not break down or burn down by
itself, but if lightning strikes lines with such conductors, probability of its break
down or burn down is much higher than that of a new conductor under the same
situation.

33

ECpE
Department
5.1. Trees
Trees are among the major factors that affect the reliability of an overhead distribution line. Trees can
cause failure of such lines in the following ways:
• Overhead conductors can be damaged when struck by a falling branch of tree.
• Wind can blow a tree branch into overhead conductors, resulting in two wires contacting each
other.
• A growing branch of a nearby tree can push two phase conductors together resulting in a two‐
phase fault.
• During regular tree trimming, a tree branch can be accidentally dropped on the overhead line.
• Ice accumulation on tree branches can cause limbs to break off and fall on the conductors.
Tree Trimming for Overhead Line Reliability:
• Tree trimming means periodic pruning of vegetation near power lines, which is the best
possible solution to avoid overhead line failures caused by trees.
• Trimming is done every 2-6 years in most distribution systems.
• Selective trimming focuses on trees causing more customer interruptions, which can reduce
operating and maintenance costs.
• Some utilities trim only main feeder trunks, not lateral branches.
• Tree trimming should always be performed by a trained crew to ensure safety and direct
regrowth away from the conductor location. 34

ECpE
Department
5.2. Lightning
• Lightning is a transient, high-current electric discharge caused by the breakdown of
air due to large electric fields.
• Lightning can occur within a cloud, from a cloud to the surrounding air, between
adjacent clouds, and from a cloud to the ground.
• Intracloud, cloud-to-air, cloud-to-cloud, and cloud-to-ground lightning are the
different types of lightning discharges.
• Cloud-to-ground lightning is of particular concern due to its threat to power
systems.
• Lightning can affect power systems in two ways:
• Direct strokes: Lightning directly strikes the power system. Although the incidents of
direct strokes are very few, they are very dangerous for the system.

• Indirect strokes: Most of the lightning strikes are of this type. They do not strike the
power system directly, instead they strike some nearby objects such as a tall building or a
tree. In this case, a traveling voltage wave is induced, which is less severe than the direct
strokes. 35

ECpE
Department
5.2. Lightning
• Lightning can cause severe damage to overhead lines, which cannot be fully
avoided but can be reduced by careful application of shield wires and surge
arresters.
• Surge arresters should be inspected for any manufacturing defects, and also, very
old arresters should be replaced by new ones to prevent any damage caused by
lightning.
• The level of damage caused by lightning depends on some other factors also. For
example, lightning could be more destructive for a very small and very old
conductor compared to a big and new conductor.

36

ECpE
Department
5.3. Wind
• The probability of equipment failure increases rapidly with increasing wind speed because the
pressure exerted on trees and poles is proportional to the square of the wind speed.

• Wind can lead to supply interruptions in the following ways:

 Wind can cause a tree branch to touch two phase conductors together, resulting in a
fault.
 Wind induces several types of conductor motions – swinging, galloping, and aeolian
vibrations. If the swing amplitude is high, phase conductors could touch each other.
Conductor galloping is the phenomenon when conductor starts moving up and down
harmonically due to wind. Aeolian vibrations are generated by the air turbulence on the
downwind side of the conductor. All these conductor motions are not good from the
reliability point of view.
 To mitigate interruptions caused by wind:
 In highly windy areas, maintain a larger spacing between conductors.
 Use twisted pair conductors to prevent the mentioned conductor motions.

37

ECpE
Department
5.4. Icing
• Ice storms occur when supercooled rain freezes on contact with tree branches and
overhead conductors and forms a layer of ice, which can cause outages in multiple
ways:
 Heavy accumulation of ice on tree branches can cause them to break off and
fall on the conductors.
 Ice places heavy physical load on conductors and support structures.
 Combination of ice and wind can result in sagging of conductor. The possibility
of sagging is more when the conductor is of very small size. When ice breaks
off, it can cause the conductor to jump into the conductor located above it.
 To prevent failures caused by ice storms:
 Overhead conductors and supporting structures should have high strength to
withstand the physical load imposed by ice.
 Areas prone to ice storms should consider using larger conductors to improve
resilience against ice-related issues. 38

ECpE
Department
5.5. Animals/Birds
• Animals and birds can cause harm to an overhead distribution line in several ways. Following
are a few possibilities that may result in customer interruptions:
• Squirrels cause faults by bridging grounded equipment with the phase conductor.
• Raptors and roosting birds cause faults by bridging conductors with their wings.
• Woodpeckers cause damage to utility poles by pecking holes in them.
• Large animals, such as cattle, horse, and bear, can also do physical damage to utility
poles, making the system more prone to future outages.
• To avoid failures caused by various animals and birds, several remedies can be implemented:
• Install plastic animal guards on bushings and insulators to prevent squirrels from
simultaneously touching the tank and phase conductors.
• Use anti‐ roosting devices on attractive perches to prevent birds from roosting.
• Use steel or concrete poles instead of wooden poles to avoid problems caused by
woodpeckers.
• Use barricades near wooden poles to reduce the problems caused by large animals.

39

ECpE
Department
5.6. Vehicular Traffic
• Vehicular accidents can lead to damage in distribution systems:
• Collisions between fast-moving vehicles and distribution poles can result in
pole damage, conductor sagging or swinging, and equipment damage.
• Pad‐mounted equipment are also vulnerable to vehicular accidents.
• Methods to prevent damage caused by vehicular accidents:
• Use concrete barriers and concrete poles to reduce the frequency of
automobile collisions.
• Concrete barriers should be employed to protect pad-mounted equipment
from vehicular accidents.

40

ECpE
Department
5.7. Age of Components
• Each component of a distribution feeder has its own probability of failure.
• It is often assumed that the performance of distribution system components
deteriorates when they reach approximately 30 years of age.
• However, the age of a component alone does not create reliability issues. Its effects
become apparent when combined with other factors such as conductor size, wind
velocity, and lightning intensity in the area.
• Components that are around 30 years old and contribute to reliability problems
should be replaced with new ones on feeders experiencing such issues.

41

ECpE
Department
5.8. Conductor Size
• Conductor size, when considered alone, does not cause any reliability problem.
• Similar to “age,” its effects can also be seen when combined with some other
factors.
• For example, in high wind velocity areas, the swinging amplitude of small
conductors will be high compared to big conductors.

42

ECpE
Department
6. Outage Recording
• Utilities record daily outages in the service territory.
The recorded data include time, duration, location,
number of customers affected, and the possible cause
of the outage.
• The Table shows the number of outages and their
causes recorded for a period of two years from January
2003 to December 2004 for 66 feeders with a total
length of approximately 1000 km in the distribution
system of a city in Kansas.
• The data show that trees/vegetation and
animals/wildlife caused 53.47% of outages followed by
equipment failure and unknown.
• Environmental factors, which include lightning, extreme
wind, trees, animals, ice storm, and debris, caused a
total of 1290 outages, which is 61% of all the outages.
• A significant number of outages are reported as
unknown or other causes. The cause of the outage is
recorded by the field crew based on inspection and any
circumstantial evidence at the site of the outage.
Sometimes, the crew is not able to determine the
cause, and they declare it as unknown. Table. Causes and number of outages in a service
territory in Kansas in 2003 and 2004.
43

ECpE
Department
6. Outage Recording
Seasonal variation in the number of outages:
• The figure showing outages occurring in
each month during the study period.
• The number of outages is higher during the
summer months (June, July, August, and
September). The main reason for this is
due to thunderstorms and windy
conditions during the summer months.
• The graph also shows a large number of
outages for January 2004. Most of these
outages were due to trees, which fell on
the feeders during icy conditions caused by
winter storms.
Figure. Monthly outages in a service territory in
• Although spring and fall that have quieter Kansas.
weather will lower the probability of
outages, most of the outages during these
seasons are caused by squirrels.
• Nice weather promotes higher animal
activity. Also, the end of winter and the
end of summer coincide with the birth of
new litter of squirrels, which increases the
probability of them causing outages. 44

ECpE
Department
7. Predictive Reliability Assessment
• The data‐based approaches for reliability evaluation are able to provide an
assessment of the system performance during a period in the past.
• These approaches are not capable of predicting the expected reliability in the
future.
• Predictive reliability assessment requires the integration of component failure
models, network topology, and reduction techniques.
• Next Slides will focus on exploring various aspects of predictive reliability
assessment.

45

ECpE
Department
 7. 1 Component Failure Models
• Every Components in distribution systems are designed to operate without failure
throughout their lifetime.
• However, failures can occur due to intrinsic defects or external causes.
• The failure rate of components is typically higher in the early stages of deployment
due to manufacturing defects, shipping damage, or incorrect installation.
• After the break-in period, components are expected to perform well during their
expected life, but the failure rate may increase towards the end of their life due to
aging.
• The failure of components can be modeled using a hazard function or failure rate.
• The bathtub curve is a commonly used representation of component reliability,
showing a high failure rate in the early and late stages, and a constant failure rate
during the useful life.

46

ECpE
Department
7. 1 Component Failure Models
• Hazard rate or failure rate of a component measures the probability of failure at
time t, given that the component has been functioning until that time.
• The lifetime of the component can be treated as a random variable.
• Failures can be represented by a probability density function (PDF) f(t).
• The PDF describes the likelihood of the component failing at different time points.
• The cumulative probability distribution function (CDF) F(t) gives the probability of
the component failing before or at time t.
• An expression for hazard function h(t) is given by:

• Exponential function is commonly used for modeling constant failure rate.

47

ECpE
Department
7. 1 Component Failure Models
• In exponential modeling, the probability of a component failing at time t, given that
it is working at time t, is independent of t.
• The exponential probability distribution function provides expressions for the
probability density function (PDF) f(t) and the cumulative probability distribution
function (CDF) F(t).

• On substitution on equation of h(t), we get,

48

ECpE
Department
7. 2 Network Reduction
• Power distribution systems consist of
components connected in series or
parallel configurations.
• In a series connection, all components
must be functional for electricity
delivery.
• In a parallel connection, at least one Figure. Two components connected in series
functional component is required for
electricity delivery.

• If all components in a parallel connection become unavailable, it will result in a

disruption of service.

• Consider that two components with probability of availability P1 and P2 are

connected in series as shown in the figure.

49

ECpE
Department
 7. 2 Network Reduction
• P1 and P2 can be computed from the
failure rate and mean time to repair
(MTTR).

• So, for a duration of one year, with λ1

as the annual failure rate, and R1 as the
MTTR of component 1, P1 for a year is
8760 − 𝜆1 𝑅 1 Figure. Two components connected in series
𝑃 1=
8760
And,
𝑄1 =1 − 𝑃 1
where is the annual probability of unavailability of component 1. Hence, the
combined probability of availability of both components 1 and 2 is

50

ECpE
Department
7. 2 Network Reduction
• If there are n components in series, we
can generalize the equation to find
probability of all of them being
available, or

Figure. Two components connected in series

which is,

And,
𝑃 parallel =1 −𝑄 parallel
• If there are n components in parallel, the equation can be generalized to
determine the probability of unavailability of all of them, or

51

ECpE
Department
Example
• Consider a network of six
components connected as shown
in Figure (a). Various steps Figure (a). A network of six components
required to reduce this network for
reliability evaluation are shown in
Figure (b).
• The given probabilities of
availability of these components
are P1 = 0.9, P2 = 0.8, P3 = 0.7,
P4 = 0.6, P5 = 0.5, and P6 = 0.4. We
can determine the probability of
series connected components 1–2
and 4–5.

Fig (b). Steps for network reduction 52

ECpE
Department
Example

Figure (a). A network of six components

And,

Also,

Therefore,

Or,

Fig (b). Steps for network reduction 53

ECpE
Department
Example
Now, we find the probability of
continuity of the whole network,
which is Figure (a). A network of six components

Note: Network reduction works well

for simple systems but becomes very
tedious for large systems. Therefore,
it is not suitable for large complex
systems with many components.

Fig (b). Steps for network reduction 54

ECpE
Department
7. 3 Markov Modeling
• Markov modeling is a popular approach for reliability assessment of power
systems.
• Markov modeling is based on the Markov process, which involves defining different
states of the system.
• The states represent different operating states of the system, including fully
operational and failure states.
• Transitions between states occur randomly and are determined by the probabilities
of component failure and repair.
• Failure and repair probabilities in Markov modeling follow exponential
distributions, resulting in constant failure rate (λ) and constant repair rate (μ).
• An important property of the Markov process is its memoryless nature, meaning
that future probabilities only depend on the present state and not on the past.

55

ECpE
Department
7. 3 Markov Modeling
• if we consider that the system is in state i at time t, we can write an expression for
transition probabilities at time (t + Δt) for a small Δt

and

where is the probability that the state will change to , and is the probability that the
state will remain at time . We further define transition intensities, which are

And,

56

ECpE
Department
7. 3 Markov Modeling
Since the sum of probabilities of being in any state is 1 , we can write

which gives,

Further, we can define a transition intensity matrix

Let P[X(t) = i] be pi(t), which is the unconditional probability of being in state i at time t.
If we define as the time derivative of pi(t), we can write an equation in matrix form for
all the states of the system.

57

ECpE
Department
7. 3 Markov Modeling
Note that and are row vectors,

and

• Differential equation can be solved to find probabilities of being in different states

as a function of time, given the initial probabilities or p(0).
• For state probabilities in the long run, we let t → ∞ and . The above equation
reduces to N ordinary linear equations for a system with N states.
• However, these equations are dependent, and thus, the determinant of A is zero.
Thus, to solve this equation, we remove one of the equations and replace it by

or the probabilities of states must sum to 1 . 58

ECpE
Department
Example
• In a distribution system, we can use the Markov
process by defining the states representing failure
of a component, such as the section of a feeder.
The failures and repairs of components are
represented by exponential probability
distribution functions giving constant failure and
repair rates.

• Consider the system shown in page 23. To

simplify, we consider only the failures of Segment
1 and Segment 2. Thus, as shown in the Figure,
the system will have three states, which are
system fully operational (State 1), Segment 1 Figure. State transition diagram for the
failed (State 2), and Segment 2 failed (State 3). system of Figure 9.1 for outages on
Segment 1 and Segment 2.
• It is assumed that once a segment has failed, no
additional failures will take place in the system
until repairs on the failed segment are completed.
Segment 1 is 1‐mile long and Segment 2 is 2‐miles
long. The failure rates are 0.1 faults per mile per
year, and the MTTR is three hours.

59

ECpE
Department
Example
• From the given data, we can determine the
following values for the Markov model:

• The A matrix for the for the system is

Figure. State transition diagram for the
system of Figure 9.1 for outages on
Segment 1 and Segment 2.
• Now, using Eq. (9.37) for steady state, we get:

60

ECpE
Department
Example
• Delete the first column of the matrix and replace
it by 1s to represent p1 + p2 + p3 = 1:

Or

Figure. State transition diagram for the

system of Figure 9.1 for outages on
• Solving the equation, we have p1 = 0.99989722, Segment 1 and Segment 2.
p2 = 0.0000342599, and p3 = 0.0000685198. As
expected, the system spends most of the time in
fully functional state.

61

ECpE
Department
Example
• In the next step, we compute the total time the
system would be in states of partial failure, which
are 0.0000342599 × 8760 = 0.3 hour in State 2 and
0.0000685198 × 8760 = 0.6 hour in State 3.

• Further, we compute the customer minutes of

interruptions in these two states as follows:

State 2: 0.3 × 1000 × 60 = 18000

State 3: 0.6 × 400 × 60 = 14400

• Thus, we get a total CMI of 18 000 + 14 400 = 32

Figure. State transition diagram for the
400.
system of Figure 9.1 for outages on
• Therefore Segment 1 and Segment 2.

• If we expand the problem to include failures of all

the laterals in addition to the main feeder, the
system will have 11 states. Since distribution
systems are typically much larger than the
example we have considered, implementation of
this method becomes very tedious.

62

ECpE
Department
7.4 Failure Modes and Effects Analysis
(FMEA)
• FMEA is an effective method for reliability analysis of radial distribution
systems.
• The method incorporates system topology, device models, and system
restoration models for system restoration.
• Capabilities include models for temporary and permanent faults,
protection and switching including backup protection, isolation through
protective device operation and sectionalizers, and full restoration
through repair and partial restoration through switching.
• In the FMEA method, failure on every segment of the system is
considered as a Failure Mode.
• Since each failure mode causes interruption of service to a part of the
system, these interruptions are identified as Effects on the system
reliability.

63

ECpE
Department
7.4.1 FMEA Method Assumptions
• Temporary and permanent faults are considered independent and mutually
exclusive in the analysis.
• Each segment of the system is assumed to have a constant failure rate for faults,
which follows an exponential probability distribution.
• The repair rate is also considered constant, implying that the probability of repair
at time t after a failure has occurred follows an exponential probability
distribution.
• The necessary data for the analysis include:
• System topology: The configuration and arrangement of the distribution
system.
• Line segment failure rate or Mean Time Between Failures (MTBF): The
average time between consecutive failures for a line segment.
• Repair rate or Mean Time To Repair (MTTR): The average time required to
repair a failed component or line segment.
64

ECpE
Department
7.4.2 FMEA Procedure
• Identify all failure modes and their system-wide effects.
• Determine the effects in terms of the number of customer outages and the
duration of customer outages.
• Sum the effects of all failure modes to obtain the cumulative number of customer
outages and the duration of customer outages over a specific period, such as a
year.
• Compute system-wide indices such as SAIFI (System Average Interruption
Frequency Index) and SAIDI (System Average Interruption Duration Index).

65

ECpE
Department
Example

• Again, consider the system

shown in page 23. Additional
data for the system are given
in the Table.
• Segment 1 of the main feeder
is from the breaker to the
recloser, and Segment 2 is
downstream of the recloser.
Table. Data for the distribution system

66

ECpE
Department
Example
• We start the computations by considering failures on each component one by
one and recording their effects. So, for Segment 1, the expected number of faults
is 0.1, which is obtained by multiplying the failure rate per mile by the length in
miles.
• Note that the failure rates are fractional numbers. These numbers are
determined from the historical data for the specific utility. For example, if the
utility has 2000 miles of laterals in its service territory, and 400 faults were
recorded on these laterals, the failure rate comes out to be 0.2 faults per mile per
year. This implies that not every section of the line will see a fault.
• However, for calculations, we use the expected number of faults. A fault on
Segment 1 will interrupt service to all the customers, and they will have an
expected interruption of 3 hours, which gives CMI of 100 × 180 or 18,000 minutes
of interruption.

67

ECpE
Department
Example
• We can account for effects of
faults on each component as
shown in the Table.
• Further, we can compute the
expected SAIFI and SAIDI for
the system as shown below:

and
Table. Computation of CMI

68

ECpE
Department
Example
• Now, consider the option of using the tie switch at the end of the feeder for
restoration of power to the customers connected to Segment 2 whenever
Segment 1 has a fault
• This is done by opening the recloser and closing the tie switch. Consider that this
process takes four minutes. Hence, the 400 customers connected to Segment 2
will experience an interruption of only four minutes whenever a fault takes place
in Segment 1.
• Since this interruption is less than five minutes, which is the cutoff between the
momentary and sustained interruptions, the number of interruptions as well as
the CMI for these customers will be removed from SAIFI and SAIDI calculations.
• Therefore, the total customer interruption reduces to 325, and CMI reduces to
47 400, which gives SAIFI of 0.325 and SAIDI of 47.4 minutes.
• Note that we are able to get information only on expected values of SAIFI and
SAIDI, which are useful for comparison of different system topologies during the
planning stages or for decisions relating to system upgrades.
69

ECpE
Department
7.5 Monte Carlo Simulation
• FMEA provides an assessment of system reliability based on estimated mean
values of reliability indices.
• Detailed assessment requires Monte Carlo simulation, which can be time-
consuming.
• In addition to knowing the mean value of failure rate (or time between failures)
and time to repair (or repair rate), probability distribution of these two
parameters is needed.
• While modeling failures with a fixed average rate and exponential probability
distribution function is a good assumption, considering fixed average repair rate
with exponential probability distribution function is strictly not true.
• Prior research shows that exponential probability distribution function provides
good approximation. Also, the average failure rates and the associated probability
distribution functions can change due to external conditions, such as storms.
• Since analysis with variable failure rates becomes very complex, we consider fixed
failure rates.
70

ECpE
Department
7.5 Monte Carlo Simulation
Simulation process:
• The simulation begins by building failure scenarios for the system. Using a
random generator, we determine the failure rate and the repair rate of each
component from their respective probability distributions.
• Next we find the effects of each component's failure on the numbers of
customers affected and the customer minutes of interruptions. The cumulative
count of customer interruptions and customer minutes of interruptions provides
the values of SAIFI and SAIDI.
• Repeat the process multiple times to get a distribution of SAIFI and SAIDI. The
number of simulations is decided based on the desired results. The number of
simulations is determined based on desired results and convergence to a stable
value. For example, simulations can be continued until the mean of all
simulations converges to a stable value.
• The results provide probability distribution of system indices from which the
average and standard deviation as well as confidence levels can be computed.
71

ECpE
Department
7.5 Monte Carlo Simulation
• The size of the system has significant impacts on the results. Specifically, the
standard deviation of SAIFI and SAIDI increases as the system size decreases. The
spread of SAIDI is usually higher than that of SAIFI.
• Systems with higher failure rates or more faults have less standard deviation.
• As we zoom into the system, the spread of the reliability indices increases.
• The results provide annual performance standards for individual feeders or parts
of the system.
• The results do not accurately forecast the system performance in the future.
• These results are valuable in relative comparison of feeders or systems against
one another for planning improvements.

72

ECpE
Department
8. Regulation of Reliability
• Maintaining adequate level of reliability requires investments in system upgrades. However,
there are trade‐offs between cost and reliability.

• While the reliability of the bulk system, which includes transmission and generation, is
regulated at the federal level in the United States, electric distribution is regulated at the
state level.

• As reported, 35 states out of 50 states and District of Columbia actively regulate distribution
system reliability. Frequencies and durations are commonly used metrics to measure the
overall system performance.

• Typically, frequencies and durations are used to measure the overall system performance.

• Some regulators require utilities to report reliabilities in a smaller region, such as a

geographic area or feeder, and may require identification of circuits with worst reliability. If
a utility serves both urban and rural customers, the regulators may ask for separate reports
for urban and rural regions.

• Some regulators have included reliability performance in their revenue regulations. These
regulations may include penalties only for not meeting the standards or also rewards for
exceeding the performance.
73

ECpE
Department
8. Regulation of Reliability
Other examples include setting quality of service target without any penalty or
rewards and reporting only without targets. While the targets set by regulators and
utilities vary widely. Some examples of targets are listed below:
• SAIDI of the worst performing feeder exceeding system SAIDI by 300%.
• SAIDI of a feeder greater than four times the system SAIDI or in the top 10% for
two consecutive years.
• More than 90% of the interrupted system restored in 36 hours for all events
except extreme events, and more than 90% restored in 60 hours for extreme
events.
• Customers experiencing more than six outages per year for three consecutive
years or outages with total duration of more than 18 hours per year for three
consecutive years.

74

ECpE
Department
Thank You!

75

ECpE
Department