EE5712 Power System Reliability

:: Evaluation Methods
Panida Jirutitijaroen
Outage model
Component modeling
System modeling

REVIEW
 Generating Unit Model
• Generators are modeled as
three state devices.
Up
– State 1: Up state with full
capacity.
– State 3: Down state with zero 1
capacity.
– State 2: Derated state with
partial capacity. Derated Down
• A two state model is a
special case of this model 2 3
where there is no transition
to state 2.
• Unit unavailability is known
as Forced Outage Rate
(FOR)
 Transmission Line Model
Normal Weather Adverse Weather
• Assume to be either in the
up state or failed state.
Up Up • Failure and repair rates are
further assumed to be
1/S 1/N dependent on weather.
U U – λ = Failure rate in the normal
weather.
µ µ’
– µ = Failure rate inthe
λ λ’ normalweather.
– λ’ = Repair rate in
1/S 1/N theadverseweather.
D D – µ’ = Repair rate in the adverse
weather.
Down Down – N, S = Mean durations of
normal and adverse weather.
 AC Power Flow
n
• Denote f(x) = 0 as, Pi   Vi Vk ReYik cos ik  ImYik sin  ik   0
k 1
n
Qi   Vi Vk ReYik sin  ik  ImYik cos ik   0
• The constraints are as follows k 1
P ≤ Pmax
Qmin ≤ Q ≤ Qmax
Fij = gij(x)
F ≤ Ff
-F ≤ Fr
Where
x = (|V|, θ, P, Q)
gij(x) = function describe circuit flow in the network
F ij = Circuit flow between node i and j
 DC Load Flow
• Also called linearized power flow model
Bθ + G = D
• The constraints are as follows
G ≤ Gmax
Fij = (θj - θi)bij
F ≤ Ff
-F ≤ Fr
Where
G = Input capacity from a generator (MW)
D = Demand (MW)
B = Susceptance matrix
bij = Susceptance between node i and j
 = Node voltage angle vector
F ij = Circuit flow between node i and j (MW)
 Linear Network Flow Model
• Also called transportation flow model
AF+G=D
• The constraints are as follows
G ≤ Gmax
F ≤ Ff
-F ≤ Fr
where
A = Node-branch incidence matrix
F = Circuit flow vector
Ff = Maximum forward flow vector
Fb = Maximum backward flow vector
G = Generation vector
Gmax = Maximum generation vector
D = Demand vector
Adequate for multi-area reliability analysis, not accurate enough for
composite system reliability
Simple assessment steps
Power system reliability indexes
Single-area reliability analysis
Composite system reliability analysis
Multi-area power systems reliability analysis
Distribution system reliability analysis
Reliability worth

OUTLINE
Component modeling
System modeling
State selection
Evaluation
Reliability calculation

SIMPLE RELIABILITY ASSESSMENT

STEPS
 Component Modeling
• Analyze failure effects of each component
• Usually presented by Markov model
• Identify capacity of each state associated with
its probability and frequency
• Steady state probability and frequency
calculated from Markov equations or
frequency balance approach
 System Modeling
• Interaction among • Power flow analysis
components
– Series/parallel network

2
3
1 5 8
4
6 7
 Reliability Assessment Steps

Unit & System Models

State Selection

Load Curtailment Evaluation

No Yes

Success Failure Reliability Indices

State State Calculation
 Three Simple Steps
• State selection
– Enumeration
• Evaluation
– By inspection
– From power flow analysis
• Reliability calculation
– Loss of load probability is simple summation over loss of load
states
– Frequency of failure is the multiplication of the probability and
the transition rate from loss of load state to success state
– Expected loss load is the expected value of amount of load loss
and is found from the weighted average of the loss load
Deterministic indexes
Probabilistic indexes
Customer-oriented indexes
Indexes by assessment levels

POWER SYSTEM RELIABILITY

INDEXES
 Deterministic Indexes
• Operating reserve margin
– Excess generation capacity in case of emergency
• Percentage reserve
– Amount of reserve capacity as a percentage of the
total peak load
• Reserve margin as the largest unit online
– Amount of reserve equals to the capacity of the
largest unit online
 Probabilistic Indexes
• Probability index
– Loss of load probability (LOLP), no unit
– Hourly loss of load expectation (HLOLE), multiply LOLP
by number of hours in a year
– Daily loss of load expectation (DLOLE), multiply LOLP
by number of days in a year
• Frequency index
– How often the system fails in a year (failure/year)
• Expected value index
– Energy Not Supplied (EENS) or Expected Unserved
Energy (EUE)
 Customer-Oriented Indexes
Index Description
SAIFI System average interruption Average number of interruptions per
frequency index customer served per year
SAIDI System average interruption Average interruption duration per customer
duration index served per year
CAIFI Customer average Average number of interruptions per
interruption frequency index customer affected per year
CAIDI Customer average Average interruption duration per customer
interruption duration index affected per year
ASAI Average service availability Ratio of the total number of customer hours
(ASUI) (unavailability) index that service was available (unavailable) during
a year to the total customer hours demanded
AENS/ Expected energy not Average energy not supplied per customer
EENS supplied served per year
Singapore Indexes
 Indexes by Assessment Levels
• Single-area reliability analysis
– Reserve margin
• Multi-area power systems reliability analysis
– Loss of load probability/expectation
– Expected unserved energy
• Composite system reliability analysis
– Loss of load probability/expectation
– Expected unserved energy
– Operating reserve, spinning reserve, non-spinning
reserve
• Distribution system reliability analysis
– Customer-oriented indexes
Objective
Generation system model
Unit addition algorithm
Capacity outage states rounding
Load modeling
Clustering algorithm
Generation reserve model
Continuous distribution approximation

SINGLE AREA RELIABILITY ANALYSIS

 Single Area Reliability Analysis
• Single bus analysis
• Generators and loads
are within the same bus G1 G2 G3

• Each generators has

their own performance Load
indexes
• No transmission
capacity limits
 Objective
• Single bus analysis: Generators and loads are
connected to the same bus.
• Interest to find out the ability of existing
generation to serve load
– Generation system model
– Load model
– Generation reserve model
 Generation System Model
• Based on failure and repair rate of each
generator, calculate three vectors C, P, and F

(C,P,F) Description
C Capacity outage levels
P Probability of capacity outage greater than
or equal to C (cumulative probability)
F Frequency of capacity outage greater than
or equal to C (cumulative frequency)
 Generation Model Example
• A system with two
λ generators, each with
1U2U 1U2D
200 MW 100 MW 100 MW
0 MW 100 MW
μ • Calculate C, P and F
μ λ μ λ
– C: capacity outage table
λ
1D2U 1D2D – P: cumulative probability
100 MW 0 MW
100 MW 200 MW – F: cumulative frequency
μ

Status
Capacity In
Capacity Out
 Unit Addition Algorithm
• Given capacity outage table with cumulative
probability and frequency
• Add one more unit in
• Find the updated capacity outage table,
cumulative probability and frequency
• Use conditional probability concept!
 Unit Addition Example
1U
λ
2D
• A system with two
50 MW μ 0 MW generators, each with
0 MW 50 MW
100 MW
1U2U
λ’
1U2D
• Add new capacity of 50
200 MW 100 MW MW in
0 MW 100 MW
μ’ • Calculate C, P and F of a
μ’ λ’ μ’ λ’ new system
λ’
1D2U 1D2D
– C: capacity outage table
100 MW
100 MW
0 MW
200 MW
– P: cumulative probability
μ’ – F: cumulative frequency
Status
Capacity In
Capacity Out
 Original Capacity Outage Table
i Cap out Cap in P : Cumulative F: Cumulative
C probability frequency
(Cap out ≥ C ) (Cap out ≥ C )
1 0 200 P1 F1
2 100 100 P2 F2
3 200 0 P3 F3

X up, Pup X down, Pdown

Cap out = 0 MW Cap out = 50 MW Prob(Cap out ≥ 250)
0 50 = P3 × Pdown
100 150 Freq(Cap out ≥ 250)
200 250 = F3 × Pdown + P3 × Pdown × μ

Pgxx = Cumulative probability that a capacity out is greater than or equal to xx or capacity in is less than or equal to xx
Frgxx = Cumulative frequency that a capacity out is greater than or equal to xx or capacity in is less than or equal to xx
 Frequency of Failure
X up X down
Success
A+ B+
Frf,1 = Fr{system fails | X up} Frf,2 = Fr{system fails | X down}
× Pr {X up} × Pr {X down}

Fail B-
A-

• Frequency of failure comes from the following

– Frequency of encountering failure states when the component X is up
– Frequency of encountering failure states when the component X is
down
– Frequency of encountering failure states when component X change
status (if any)
 X up X down
Success
A+ B+
Frf,1 = Fr{system fails | X up} Frf,2 = Fr{system fails | X down}
× Pr {X up} × Pr {X down}

Fail B-
A-

Frf,2 = ( Pr{system success| X up} – Pr{system success| X down} ) × Pr(X up)

× (transition rate X up to X down)

Frf,2 = ( Pr{system fails| X down} – Pr{system fails| X up} ) × Pr(X down)

× (transition rate X down to X up )
Modified Capacity Outage Table

i Cap Cap in P : Cumulative F: Cumulative frequency

out C probability (Cap out ≥ C )
(Cap out ≥ C )
1 0 250 P1 × Pup + P1 × Pdown
2 50 200 P2 × Pup + P1 × Pdown
3 100 150 P2 × Pup + P2 × Pdown
4 150 100 P3 × Pup + P2 × Pdown
5 200 50 P3 × Pup + P3 × Pdown
6 250 0 P3 × Pdown F3 × Pdown + P3 × Pdown × μ
 Capacity Outage States Rounding
• Reduce number of capacity outage states
• Construct capacity outage states as multiples
of specific increment
• Rounding off the capacity states that are not
multiples of specific increment
 Rounding Off a Two-State Unit
• Before rounding • After rounding
λ12
56 MW,
0 μ21 α λ12
P 60 MW,
0 μ21 αP
• Increment of 10 MW μ21
• 2 capacity states, 50 and 60 (1-α) λ12
50 MW,
MW (1-α)P
• α : ratio of the difference
between the integral Freq(60 MW) = αPμ21
capacity outage to actual Freq(50 MW) = (1-α)Pμ21
capacity and the increment
– α = (60-56)/10 = 0.4
 Load Modeling
• Develop from hourly load data
• Discretized load model
• Characterized by
– L: Load level
– P: Probability of load greater than or equal to L
– F: Frequency of load greater than or equal to L
 Load Model Example

Load = 4600 MW

P(4600) = 11.5/24

F(4600) = 2/24

P(Li) = (#of hours load is ≥ Li) / (#of hours in interval)

F(Li) = (#of transitions from load < Li to load ≥ Li) / (#of hours in interval)
This may result in numerous load states!! We can group them using clustering
algorithm!!
 Clustering Algorithm
• Basic K-means clustering
algorithm Initial Mean clusters
• Group load based on
proximity Compute new Calculate Euclidean distances
of original load states to
mean clusters
• Less load states mean clusters

• Let Yes
Regroup
– Li = load state i clusters?
No
– N = number of cluster
Final mean cluster
• Euclidean distance is
– √(mean cluster – load state)²
 Generation Reserve Model
• Develop model for capacity margin at each bus
• Margin = Generation – Load

(M,P,F) Description
M Margin capacity
P Probability of margin less than or equal to M
F Frequency of margin less than or equal to M

• Use conditional probability concept!!

• Or, probability convolution technique in Lecture 3.
 Reserve Model Example
Load Cumulative
Probability
Cumulative
Frequency
• Five generation capacity
(Load ≥ L) states with exact
0 (1) Pl0 Frl0 probability
50 Pl50 Frl50
100 Pl100 Frl100
• Three load states with
exact probability
Generation Cumulative Cumulative
Capacity In Probability Frequency • Find the reserve model
(Cap in ≤ C) for this system
200 (1) Pg200 Frg200
(homework)
150 Pg150 Frg150
100 Pg100 Frg100
50 Pg50 Frg50
0 Pg0 Frg0
 Original Capacity Outage Table

Cap in\Load 100 50 0 (Pl100 = 1)

200 (Pg200 = 1) 100 150 200
150 50 100 150
100 0 50 100
50 -50 0 50
0 -100 -50 0

Prob(Margin ≤ -50) = (Pg50 – Pg0)× Pl100 + Pg0 × Pl50

Freq(Margin ≤ -50) = Freq(gen) + Freq(load)
= Frg50 × Pl100 + Frg0 × (Pl50 – Pl100) + Frl100 × (Pg50 – Pg0) + Frl50 × Pg0
Pgxx = Cumulative probability that a capacity out is greater than or equal to xx or capacity in is less than or equal to xx
Plxx = Cumulative probability that load is greater than or equal to xx
Frgxx = Cumulative frequency that a capacity out is greater than or equal to xx or capacity in is less than or equal to xx
Plxx = Cumulative frequency that load is greater than or equal to xx
 Continuous Distribution Approximation

• Moment/Cumulant matching
• Probability distribution by Gram-Charlier
Series
• Calculate the moments/cumulants from raw
data of capacity outage table
• Refer to
http://www.ee.tamu.edu/People/bios/singh/c
oursenotes/part4.pdf
Problem statement
Objective
Component modeling
Reliability evaluation techniques

COMPOSITE SYSTEM RELIABILITY

 Composite System Reliability
• Interconnected single-
area
• Concern with generation
and transmission
capability adequacy
• High-voltage transmission
lines
• May include high-voltage
transformers, circuit
http://www.powerworld.com/images/7-bus%20Oneline.jpg
breakers
Problem Statement

How will the system

survive (be able to
serve the load) under
contingencies?
 Contingencies?
• Random failures in the system, generators/
transmission lines can fail
• Disturbances, faults in the network
• Randomness in demand, load changes
• Our ‘state space’ IS contingencies!
• How to incorporate them in the analysis?
– Case-by-case, enumerate all contingencies
– If we have n components in the system, we will have
2ⁿ contingencies.
– Need to rank and prioritize contingencies
 Objective
• Deterministic
– Maintain adequate service under most likely
outages and accept some degradation under low
probability outages
• Probabilistic
– System reliability is below an acceptable
threshold, for example, loss of load expectation is
less than 1 day in 10 years
• Two analysis cases, before and after remedial
actions.
 Component Modeling
• Generator
• Transmission line
• Transformer
• Circuit breaker
• Load
 Generating Unit Model
• Generators are modeled as
three state devices.
Up
– State 1: Up state with full
capacity.
– State 3: Down state with zero 1
capacity.
– State 2: Derated state with
partial capacity. Derated Down
• A two state model is a
special case of this model 2 3
where there is no transition
to state 2.
• Unit unavailability is known
as Forced Outage Rate
(FOR)
 Transmission Line Model
Normal Weather Adverse Weather
• Assume to be either in the
up state or failed state.
Up Up • Failure and repair rates are
further assumed to be
1/S 1/N dependent on weather.
U U – λ = Failure rate in the normal
weather.
µ µ’
– µ = Failure rate inthe
λ λ’ normalweather.
– λ’ = Repair rate in
1/S 1/N theadverseweather.
D D – µ’ = Repair rate in the adverse
weather.
Down Down – N, S = Mean durations of
normal and adverse weather.
 Transformer and Bus Model
• Like the transmission
lines, transformer and
bus are treated as two Up Down

stage devices. µ λ
U D
• Failure and repair rates
are assumed to be
independent of the
weather.
 Circuit Breaker Model
• Fault models classified by three modes of failure
• Ground faults
– Two-stage model as transformer or bus
• Failure to open
– Healthy components may be isolated due to the operation of
secondary zone protection.
– May caused by latent faults in the breaker or the associated
protection system
– Characterized by a probability p, a probability that it may not
respond
• Undesired tripping
– Characterized as a failure rate; its effect will be an open line.
 Load Model Example

Load = 4600 MW

P(4600) = 11.5/24

F(4600) = 2/24

P(Li) = (#of hours load is ≥ Li) / (#of hours in interval)

F(Li) = (#of transitions from load < Li to load ≥ Li) / (#of hours in interval)
 System Modeling
• DC flow model • Need to know
• AC flow model – System configurations
– Line admittances
1 2
– Net injection at each bus
G1 Load (except the slack bus)
– Thermal limits (MW) of
the line
G2 – Voltage limit of each bus
3 4 – Reactive power limit
(MVAR) of each
generator bus
 DC Load Flow
Minimize curtailment C
• Subject to
Bθ + G + C = D
G ≤ Gmax
C≤D
Fij = (θj - θi)bij
F ≤ Ff
-F ≤ Fr
Where
G = Input capacity from a generator (MW)
C = Load curtailment (MW)
D = Demand (MW)
B = Susceptance matrix
bij = Susceptance between node i and j
 = Node voltage angle vector
F ij = Circuit flow between node i and j (MW)
 Reliability Evaluation Techniques
• Analytical
– Contingency enumeration and selection
• Simulation
– Monte Carlo simulation
– Other variance reduction techniques
• Hybrid
• Simulation will be given in the next lecture!!
 Analytical Techniques
• Contingency enumeration
– Number of contingency increases exponentially
with number of components
• Contingency ranking and selection
– Rank by a performance index
– Rank by a probability
• Truncation (N-m contingency)
– Up to m-order to contingency is considered
• Implicit enumeration
Contingency Enumeration
Base Case

Select a contingency and Evaluate the system

No Loss of
load?
Yes
Take remedial action

No Problem
still exist?
Yes
Calculate reliability indexes

No All
contingency?
Yes
System reliability indexes
Problem formulation
System modeling
Max-flow min-cut algorithm
Reliability evaluation techniques
Example

MULTI-AREA RELIABILITY ANALYSIS

 Multi-Area Reliability Analysis
• Consider thousands of
nodes then simplify the
system to small workable
nodes (areas), each of
them form a control area
• Generator and load from
different nodes within the
same area are grouped
into one.
• Tie-line capability
between areas
 Problem Statement
Represent the transmission
limits, between areas, Not a
physical flow !

Tie-line
Single Area 1 Single Area 2

Single Area 3 Single Area 4

Interest to find out if area generation or tie-line capability are adequate to

serve load
 Objective
• To calculate reliability indices
– Loss of load probability (LOLP)
– Expected unserved energy (EUE)
• Indexes of each area and the entire system
 System Modeling
• Each area is modeled by
– Generation model, capacity outage table
– Load model, discrete load states
• Tie-lines are modeled by
– Capacity states with probabilities
– Transfer capability “Available Transfer capability” that
accounts for inter-transmission capacity within each
area
– Not a physical flow, admittances are not considered
• Use Transportation flow model to describe flows
in the network
 Area Injections
Load Loss Sharing Policy No Load Loss Sharing Policy
• In case of loss of load in • In case of load in area A, the
area A, the other area will other area will not offer
share the load loss help unless its load is
• Separate generation and satisfied.
load model • Combined generation-load
(reserve) model
 Multi-Area System Representation
• Capacity flow network
• Network random variables: area generation capacity, load, tie
line capacity between areas
• Discrete probability distribution functions of all random
variables are constructed.
gi   li  
tij  
…….

…….
s Power system network t
(only tie line capacity)

Types of arcs in this network

Generation arc, connected source node to each node i
Transmission arc, transmission line connect node i to node j
Load arc, connect each node i to sink node
 A Three-Area Example
• Three area with the
following data.
1
Area 1 Area 2 Area 3
(load = 400) (load = 500) (load = 400)
Cap Prob Cap Prob Cap Prob s 2 t
(MW) (MW) (MW)
7 600 0.262144
6 500 0.32768 500 0.393216 500 0.32768
5 400 0.40960 400 0.245760 400 0.40960 3
4 300 0.20480 300 0.081920 300 0.20480
3 200 0.05120 200 0.015360 200 0.05120
2 100 0.00640 100 0.001536 100 0.00640
All transmission lines have
1 0 0.00032 0 0.000064 0 0.00032
maximum capacity of 100 MW.

This means that we need to deliver capacity from source node to sink node.
 Straightforward Approach
• Goal: To compute LOLP, EUE of each area and
entire system
• Enumerate all state spaces
• Evaluate each state if loss of load or not -> How?
• LOLP = add probability associated to all failure
states
• EUE = expected value of loss of load
• This is impractical!!
 State Evaluation
• Flow calculation
• Find maximum flow the network can deliver
from source node to sink node
• By inspection for small system
• Ford-Fulkerson algorithm for large system
 Max-Flow Min-Cut Theorem
For any network, the value of the maximal flow
from source to sink is equal to the capacity of the
minimal cut
• Definitions
– Forward arc = Arc leaving a node with respect to that node
– Backward arc = Arc entering a node with respect to that node
– Path = A sequence of arcs starting from the source node and
ending at the sink node
– Cycle = A path whose beginning and end are the same
– Cut = A partition of set of all nodes in the network into two sets
– Minimal cut = The cut with smallest capacity
 Ford-Fulkerson Algorithm
• Max-flow min-cut algorithm
Flow = 0
• Use a search process to find
augmenting path that a Flow += Search algorithm
positive flow can be sent flow

from source to sink. Yes Feasible

• Compute maximum flow path?
No
that it allows to be sent
Final maximum flow
along this path
• Stop when no augmenting
path can be found
 Search Algorithm
• Search for an augmenting path from source to sink.
– Breadth-first search (BFS)
– Depth-first search (DFS)
• BFS
– Start from source and explore neighboring nodes until sink node
is found
– Explore the unexplored neighbor nodes
• DFS
– Start from source and explore deeper till sink is found
– Otherwise, backtracking when it hit a leaf
• Augmenting path has been found.
• Label nodes: Unexplored, Visited, Explored.
 Example
• Find maximum flow of the following network

1
7 9

s 3 t

9 8
2
 Reliability Evaluation Techniques
• Contingency • Simulation methods
enumeration – Monte Carlo sampling
• Analytical methods • Hybrid
– Network reduction
– Cut-sets
– *** State space
decomposition ***
Interruption indexes
Radial network
Parallel network
Mesh network

DISTRIBUTION SYSTEM RELIABILITY

 Distribution System Reliability
• Network configuration/
topology
• Analysis takes into
account reliability of the
following low-voltage
components,
– Lines/cables
– Transformers
– Circuit breakers http://www.tpub.com/content/construction/14027/css/14027_63.htm

– Isolators
 Problem Statement

Supply A B C
A radial distribution system

L1 L2 L3

L1

A B
Supply
A ring distribution system
C D

L3
L2

What is the reliability of each load point in the network?

 Objective
• Interest to find out the reliability level at load
point, and the system.
Index Description
SAIFI System average interruption Average number of interruptions per customer
frequency index served per year

SAIDI System average interruption Average interruption duration per customer served
duration index per year
CAIFI Customer average Average number of interruptions per customer
interruption frequency index affected per year
CAIDI Customer average Average interruption duration per customer affected
interruption duration index per year
ASAI Average service availability Ratio of the total number of customer hours that
(ASUI) (unavailability) index service was available (unavailable) during a year to
the total customer hours demanded
AENS/EE Expected energy not supplied Average energy not supplied per customer served
NS per year
 Reliability Evaluation Techniques
• Series/parallel network
• Network reduction
• Conditional probability approach
• Cut-set or tie-set approach
Review of a Two-State Component
• λ = 1/MTTF λ

• μ = 1/MTTR UP DOWN
μ
• MCT = MTTR + MTTF
• f = 1/MCT = 1/(MTTR + MTTF) = λμ/(λ+μ)
• Pf = MTTR/(MTTR + MTTF) = f × MTTR
• Most of the time, data is given as λ, failure
rate (failures/per year) and MTTR, average
repair time (hours)
 Frequency of Failure Approximation

UP

DOWN
Tup
Tdown

• If MTTR << MTTF,

• f = 1/MCT = 1/(MTTR + MTTF) ≈ 1/MTTF ≈ λ
• Pf = f × MTTR ≈ λ × MTTR
 Series Network

A B

• Let PsA and PsB be the success probability of

component A and B, system availability is
A = PsA × PsB
• Equivalent Failure rate = Frequency of
failure/Probability of success
λeq = A × (λ1 + λ2) / A = λ1 + λ2
Pf Approximation in Series Network
• A = PsA × PsB
• Pf = U = 1 – A
• PsA = 1 - PfA , PfA = fAMTTRA
• PsB = 1 - PfB , PfB = fBMTTRB
• Pf = 1 – (1 – fAMTTRA)(1 – fBMTTRB)
= fAMTTRA + fBMTTRB – fAfB MTTRAMTTRB
• Assume MTTR << MTTF, f ≈ λ,
Pf ≈ λAMTTRA + λBMTTRB
• In general case, unavailability (hour per year)
Pf ≈ Σ λiMTTRi
MTTReq Approximation in Series Network
• Equivalent repair rate = Frequency of
success/Probability of failure,
μeq = Ps × λeq / Pf.
• Assume that MTTR << MTTF, Ps ≈ 1 and Pf ≈
λAMTTRA + λBMTTRB. Then,
μeq ≈ λeq / λA MTTRA + λB MTTRB.
• In general case, equivalent repair time (hour)
MTTReq ≈ Σλi MTTRi / Σλi
 Parallel/Mesh Network
• Let PfA and PfB be the failure
probability of component A
and B, system unavailability A
is
U = PfA × PfB B
• Equivalent repair rate =
Frequency of MTTReq = 1/μeq
success/Probability of = 1/(μ1 + μ2)
failure
= 1/(1/MTTR1 + 1/MTTR2)
μeq = U × (μ1 + μ2) / U = μ1 + μ2
= MTTR1MTTR2/(MTTR1 + MTTR2)
 Pf Approximation in Parallel Network
• Pf = U = PfA × PfB
• PfA = fAMTTRA (hour/year) and PfB = fBMTTRB
(hour/year)
• Pf = fAfB MTTRAMTTRB (hour²/year²)
• Assume MTTR << MTTF, f ≈ λ,
Pf ≈ λAλBMTTRAMTTRB (hour²/year²)
• In general case, unavailability (hour per year)
Pf ≈ Π λiMTTRi (× 8760ⁿ¯¹)
 λeq Approximation in Parallel Network
• Equivalent failure rate = Frequency of
failure/Probability of success,
λeq = Pf × μeq / Ps.
• Assume that MTTR << MTTF, Ps ≈ 1 and Pf ≈ λAMTTRA
+ λBMTTRB. Then,
λeq ≈ μeq × λA λBMTTRAMTTRB. (hour/year²)
• μeq = (MTTR1 + MTTR2)/MTTR1MTTR2
• In two component case, equivalent failure rate
(failures/year)
λeq ≈ Πλi × ΣMTTRi / 8760
 Summary

Indexes Series Network Parallel Network

Unavailability (hours Σ λiMTTRi Π λiMTTRi / 8760ⁿ¯¹
per year) (n-component case)
Equivalent failure Σλi Πλi × ΣMTTRi / 8760
rate ( per year) (2-component case)
Equivalent mean Σλi MTTRi / Σλi MTTReq = 1/Σ(1/MTTRi )
repair time (hours)
 Distribution Network Example 1
• Single breaker from
Supply A B C supply
L1 L2 L3
• Trip whenever there is
fault on the network
Line λ (failures MTTR • Imply same reliability at
per year) (hours) all load point
A 0.20 6.0
B 0.10 5.0 Load λ (failures Unavailability MDT
point per year) (hours/year) (hours)
C 0.15 8.0
1 0.45 2.9 6.4
2 0.45 2.9 6.4
3 0.45 2.9 6.4
 Distribution Network Example 2
• Three 100% reliable
Supply A B C breakers at load point
L1 L2 L3
• Imply different
reliability at all load
Line λ (failures MTTR
point
per year) (hours)
A 0.20 6.0 Load λ (failures Unavailability MDT
point per year) (hours/year) (hours)
B 0.10 5.0
1 0.20 1.2 6.0
C 0.15 8.0
2 0.30 1.7 5.7
3 0.45 2.9 6.4
 Distribution Network Example 3
• A& B in Parallel
Supply A C
• Then in series with C
B L2
L1

Line λ (failures MTTR Load λ (failures Unavailability MDT

per year) (hours) point per year) (hours/year) (hours)
A 0.20 6.0 1 2.5×10¯⁶ 6.81×10¯⁶ 2.7
B 0.10 5.0 2 0.1500025 1.200018225 7.999988
C 0.15 8.0
Interruption cost
Outage duration
Customer damage function

RELIABILITY WORTH
 Cost-Benefit Analysis
• High reliability achieved with high cost
• Is it worthwhile to have high reliability?

http://www.eppo.go.th/power/ERI-study-E/ERI-EOCS-1-E.html
 Reliability Worth
• Assess worth of
Generation
providing reliable
service
MW Load loss
• By evaluating impact
Duration load loss and monetary loss due
to power failure
Load
• Using customer
interruption costs
 Interruption Cost
• Monetary loss from electric supply curtailment
– Energy not supplied/unserved energy
• Class of customers
– Residential load
– Small/Medium/Large Industrial load
• Outage duration
• Evaluation methods
– Cast studies
– Customer survey (favorable choice)
 Outage Duration
• State duration in hours ( D ),

24
D  ml

 g  g  t  t  l


iI
i


iI

i
 
 

i , jI
ij
i , jI
ij
k 1
k

i j i j

– x = Equivalent transition rate of component x from a

capacity of state ω to higher capacity (per day)
– x = Equivalent transition rate of component x from a
capacity of state ω to lower capacity (per day)
 Customer Damage Function
• Customer interruption
cost
– Customer type
– Interruption duration
– Season, weekday,
weekend

c D   e
l 6.48005  0.38489 D  0.02248 D2

L. Lawton, M. Sullivan, K. V. Liere, A. Katz, and J. Eto, "A framework and review of
customer outage costs: Integration and analysis of electric utility outage cost
surveys" (Nov. 1, 2003). Lawrence Berkeley National Laboratory. Paper LBNL-
54365. http://repositories.cdlib.org/lbnl/LBNL-54365
 Summary
• Three simple reliability assessment steps
– State selection
– State evaluation
– Indices calculation
• Levels of power system reliability analysis
– Single-area: Only generation and load
– Multi-area: With Tie line capability between areas
– Composite system: Transmission line is included
– Distribution system: Network connection
 Reliability Assessment Steps

Unit & System Models

State Selection

Load Curtailment Evaluation

No Yes

Success Failure Reliability Indices

State State Calculation
 Next lectures
• Realize that the state space is too large to
enumerate
• Need for simulation methods to evaluate
reliability
• Monte Carlo simulation
– Random sampling
– Sequential sampling
• Other variance reduction techniques
 References
• “Engineering Reliability-New Techniques and Applications”,
B. S. Dhillon and C. Singh, chapter 9,
http://www.ece.tamu.edu/People/bios/singh/enggreliability/
• Electrical Power System Reliability Course notes, C. Singh,
part 3 to 6.
http://www.ee.tamu.edu/People/bios/singh/coursenotes/
• “System Reliability Modeling and Evaluation”, C. Singh,
chapter 4,
http://www.ece.tamu.edu/People/bios/singh/sysreliability/
ch4.pdf
• “Risk Assessment of Power Systems”, W. Li, chapter 4 and 5.
• “Reliability Evaluation of Power Systems”, R. Billinton, and
R. N. Allan, chapter 7, 8, and 13.