Академический Документы
Профессиональный Документы
Культура Документы
Generation reliability
Andrés Ramos
http://www.iit.comillas.edu/aramos/
Andres.Ramos@comillas.edu
Topic objectives
• To understand
– Why reliability must be studied
– Where does reliability appear
– How to measure or asses reliability
– What is a probabilistic production cost model by analytical
simulation
– How unit output is computed
– How to compute some generation reliability measures
• To see a true model
Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 5
What is reliability?
• The traditional objective of electric system planning is to
supply electricity demand at minimum cost and with an
acceptable long term reliability
– Supply continuity or existence of voltage supply
– Wave quality (value, shape, frequency, …)
• Reliability of electricity supply:
– Adequacy (long term)
• The ability of the electric system to supply the aggregate electrical demand
and energy requirements of the customers at all times, taking into account
scheduled and reasonably expected unscheduled outages of system elements
[ENTSO-E]
– Security (short term)
• The ability of the electric system to withstand sudden disturbances such as
electric short circuits or unanticipated loss of system elements [ENTSO-E]
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 6
Why nobody speaks about reliability?
• Historical way
– Statistics about failures and load shedding from the past
– Studies easy to be done
• Predictive way
– Assess the expected values of failures by a reliability model
– Complex studies
– Data are extracted from historical data
• Introduction
Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 9
Reliability assessment
• Reliability index
– Parameter that measures a particular aspect of reliability of an
electric energy system
• We are going to talk about generation
• The more relevant aspects to be assessed by reliability
indexes are:
– Number or frequency of failures
– Duration of failures
– Incidence of failures
• There is no index able to assess the total reliability
including all the aspects
Minimize:
∑ Operation costs + Investment costs
Subject to:
• Load supply
• Reliability index < maximum reliability index
Minimize:
∑ Operation costs + Investment costs
+ Costs associated to Not served energy
Subject to:
• Load supply
Minimize:
Objective function #1: ∑ Operation costs + Investment costs
Objective function #2: Reliability index
Subject to:
• Load supply
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 12
Contents
• Introduction
• Reliability assessment
Reliability indexes classification
Deterministic
Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 13
Reliability indexes classification (i)
• Deterministic
– Reflect the average behavior of the supply continuity
– Do not consider stochasticity in the operation
– Very frequently used:
• Intuitive
• Simple to determine
• Require few data to compute them
– Easy comparison among systems
• Probabilistic
– Do consider stochasticity in the operation
• Unit failures
• Load variation
• Hydro inflows
– Offer more information and of better quality than the deterministic
indexes
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 14
Contents
• Introduction
• Reliability assessment
• Reliability indexes classification
Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 15
Deterministic reliability indexes (i)
Source: REE
Source: ENTSO-e
RM ( MW )
LU ( pu ) =
Largest unit power
LU>1 → we can loss the largest unit
LU<1 → if we loss the largest unit → not served energy
• Dry year:
– Used in hydro dominated systems
– Criterion more than index
• In the driest year (or drier year series) demand has to be satisfied (deterministically)
• Deterministic indexes
RM(GW)=(10+3-11.5)=1.5 GW → excess of 1.5 GW
RM(pu)=(10+3-11.5)/11.5=0.13 → excess of 13 % of maximum load
LU(pu)=(10+3-11.5)/1=1.5 → excess of 1.5 largest unit
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 22
Probabilistic reliability indexes:
LOLP and LOLE (i)
• Loss Of Load Probability (LOLP):
– According to the name, should be defined as probability of being unable to satisfy
all the power with the available generation
– However, usually is measured in number of hours or days in a year with
insufficient generation. For example, 0.1 day/year
• To avoid this double and ambiguous definition, Billinton defines the
Loss Of Load Expectation (LOLE):
– Number of days (hours) in a year that we expect not to satisfy the demand with
the available generation
LOLE
LOLP =
365 days o 8760 hours
LOLE 1 1
LOLPN = LOLP365 = > LOLP8760 =
N 365 8760
EENS
LOEP =
Total load
5000 MW
5000 MW
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 33
Generation capacity adequacy assessment techniques
1. State Table
– Exact
– Impossible to use for large-scale systems
00 0 0.002 1100
X (n ) = µˆ =
∑ i =1
xi
• Sample mean n
n
S 2 (n ) = σˆ2 =
∑ i =1
(x i
− X (n ))2
• Sample variance n −1
n
• Coefficient of variation S 2 (n )
cv = n
X (n )
t = 2.581 t = 1.962
1000−1,1% 1000−1,5%
2 2
• Case p=0.9
– Mean: 0.81
– Variance of the r.v.: 0.1539
– Coefficient of variation of the r.v.: 0.484322
• Case p=0.95
– Mean: 0.9025
– Variance of the r.v.: 0.08799375
– Coefficient of variation of the r.v.: 0.328684
• 2000 samples
p=0.9 p=0.95
media 0.802000 media 0.905500
varianza 0.158875 varianza 0.085613
var media 0.000079 var media 0.000043
coef var 0.496997 coef var 0.323132
• 20000 samples
p=0.9 p=0.95
media 0.809200 media 0.904950
varianza 0.154403 varianza 0.086020
var media 0.000008 var media 0.000004
coef var 0.485593 coef var 0.324097
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 42
Statistical basic concepts
f X ( x) FX ( x )
1
0
D Dx D D x
fY ( y ) FY ( y )
1
0
0 y 0 y
P P
35,000
30,000
25,000
Sábado
Demanda [MW]
20,000 Domingo
Lunes
Martes
Miércoles
15,000 Jueves
Viernes
10,000
5,000
-
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hora
35000
30000
25000
20000
Demanda [MW]
15000
10000
3966989 MWh
5000
0
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
Horas
35000
30467 MW 30000
25000
20000
Demanda [MW]
15766 MW 15000
3966989 MWh
10000
5000
0
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
Hora
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 48
Demand probability density function
– Probability of the demand is equal to a certain value f X ( x)
– Obtained from load forecasts for this week
Probability density
0.045
function approximated
in a “deterministic” way
0.040
0.035
0.030
Probabilidad
0.025
0.020
0.015
0.010
0.005
0.000
00
00
00
00
00
00
00
00
00
0
0
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
10
20
30
40
50
60
70
80
90
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Demanda [MW]
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 49
Demand cumulative distribution function
– Probability of the demand is lower or equal to a certain value FX ( x )
– Obtained accumulating the probability density function f X ( x)
1.000
0.900
0.800
0.700
0.600
Probabilidad
0.500
0.400
0.300
0.200
0.100
0.000
00
00
00
00
00
00
00
00
00
0
0
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
10
20
30
40
50
60
70
80
90
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Demanda [MW]
1.000
Probability of the
0.900 demand is >= D =
15766 MW = 1
0.800
0.700
0.600
Probabilidad
0.500
23615 MW
0.400
0.300
0.200
Probability of the
0.100 demand is >= D =
30467 MW = 0
0.000
0
0
00
00
00
00
00
00
00
00
00
0
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
10
20
30
40
50
60
70
80
90
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Demanda [MW]
1.00
0.90
p = 1− q 0.80
0.70
0.60
Probabilidad [p.u.]
0.50
0.40
0.30
0.20
q 0.10
0.00
0 350
Potencia [MW]
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 55
Unit dispatch
30000
25000
20000
Demanda [MW]
15000
10000
Energy of unit 3
5000
Energy of unit 2
Energy of unit 1
0
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
Horas
Power of unit 1
0.9
0.7
Power of unit 3
0.6
Probabilidad [p.u.]
0.5
0.4
0.3
0.2
0.1
0
0 2500 5000 7500 10000 12500 15000 17500 20000 22500 25000 27500 30000 32500 35000
Demanda [MW]
0
Demand
EENS1
Expected energy output of the unit =
0
EENS before unit dispatch –
Demand
EENS after unit dispatch
Unit fails, with probability q
Probability
1
Energy not served if the
unit fails
EENS2
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 62
Thermal unit convolution (i)
EENS if the unit i does NOT fail
Probability
1 Gi ( x − Pi )
EENS after unit i dispatch
EENS1 × pi
Probability
1 Equivalent Load-
0 Duration Curve
Demand ELDC
EENS2 ×qi Gi +1 ( x ) = pi Gi ( x − Pi ) + qi Gi ( x )
0
Demand
Ei = T ∫ [Gi ( x ) − Gi +1 ( x )] dx
D
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 66
Reliability measures (i)
LOLE= days
year
Original System
Original System +
1 unit with PMAX
PMAX
Maximum yearly
load
D1 D2
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 69
Unit contribution to system reliability: ELCC and FCE
(iii)
• Firm Capacity Equivalent (FCE)
– Alike ELCC, it is computed considering only the demand of a given hour
– Firm capacity is defined as the ideal unit always available
• No outages, no maintenance
– Given a generation system and a load, FCE is the power capacity with
failure probability equal to 0 such as if substituted by the unit gives
exactly the same system reliability measure
LOLP
Original System
Original System
- PMAX
1 unit with PMAX Same reliability
LOLP B C measure
FCE
Decrement in firm capacity
=
Load increment
Load
D Study -∆D D Study
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 71
Capacity credit of wind generation
• Contribution of wind power to system
adequacy.
• It is estimated by determining the
capacity of conventional plants displaced
by wind power, whilst maintaining the
same degree of system reliability, in
other words an unchanged probability of
failure to meet the reliability criteria for
the system.
• Alternatively, it is estimated by
determining the additional load that the
system can carry when wind power is
added, maintaining the same reliability
level.
• M. Amelin Comparison of Capacity Credit Calculation Methods for Conventional Power Plants and Wind Power IEEE
Transaction on Power Systems, 24(2): 685-691 May 2009 10.1109/TPWRS.2009.2016493
• NERC Methods to Model and Calculate Capacity Contributions of Variable Generation for Resource Adequacy Planning March
2011 http://www.nerc.com/files/ivgtf1-2.pdf
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 72
Contents
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
Discrete convolution
• Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 73
Discrete convolution method
EENS = 53 MW
LOLP = 0.595
Complementary
cumulative
distribution function
Unit 2 Expected
0.70 Output
90.85 MW
0.60
Probability [p.u.]
G6 ( x) G2 ( x )
0.20
G3 ( x )
EENS = 53 MW
0.10
LOEP
0.00 = 0.065
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
Demand [MW]
0.90
RUN-OF-THE-RIVER
HYDRO
0.80
EXPECTED OUTPUT
2500 MW
0.70
0.60
Probabilidad [p.u.]
DEMAND
0.50
23615 MW
STORAGE HYDRO
0.40
EXPECTED
OUTPUT
1830 MW 0.30
0.20
0.00
EENS = 7 MW
00
50
00
50
00
50
00
50
00
50
00
50
0
0
75
50
25
00
75
50
25
00
75
50
25
00
75
50
25
00
75
50
25
00
75
50
25
00
75
50
25
00
15
22
30
37
45
52
60
67
75
82
90
97
10
11
12
12
13
14
15
15
16
17
18
18
19
20
21
21
22
23
24
24
25
26
27
27
28
29
30
Demanda [MW]
LOEP = 0.0003
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 78
Contents
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
Limited energy plants dispatch
• FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 79
Dispatch of limited energy plants
• LEP (limited energy plants)
– Thermal units with maximum fuel available (take-or-pay contracts)
– Hydro units with limited unregulated flows
• If they are dispatched first, because their variable cost is 0, they will
produce more energy than available
• Unregulated flows are supposed to be deterministic (expected value)
• They are dispatched at maximum capacity to minimize total variable costs
(but, which is their maximum capacity?)
• Algorithm
– We try to dispatch the unit at every step until the energy produced by
the LEP unit is lower to its available energy
– Convolution invariance property: the sum of the energy produced by
the thermal unit and the LEP unit is the same regardless the order with
which they are dispatched
– Energy produced by the thermal unit is recalculated
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 80
Contents
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
FLOP model
• Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 81
FLOP model (i) (www.iit.comillas.edu/aramos/FLOP.htm)
• Introduction
• Reliability assessment
• Reliability indexes classification
– Deterministic
– Stochastic
• Probabilistic production cost model
• Statistical concepts
• Unit dispatch
• Unit convolution
• Reliability measures
• Discrete convolution
• Limited energy plants dispatch
• FLOP model
Summary
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 86
Summary of probabilistic production cost model
• Advantages:
– Demand and generation are independent random variables
– Computation of units’ output
– Computation of reliability measures
– Computation speed
• Disadvantages:
– Single loading order (heuristically obtained)
– No minimum load, no startup or shutdown
– No extensions for electricity markets
Demand
• For this electric system 200 400
– Load varies linearly (see figure) from 200 to
Capacity EFOR Variable cost
400 MW
[MW] [p.u.] [€/MWh]
– Period duration 1000 h
200 0.10 25
– Thermal generation of the table dispatched
by increasing variable cost 150 0.15 40
120 0.05 30
100 0.05 50
• This assignment is done individually
100 0.20 65
• Due date: Wednesday, December 9
50 0.15 75
• Discussion of homework: Thursday,
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
December 17
Decision support models in electric power systems – Generation reliability - 88
Generation reliability homework. Assignment A
Generation reliability
Andrés Ramos
http://www.iit.comillas.edu/aramos/
Andres.Ramos@comillas.edu
ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA
Decision support models in electric power systems – Generation reliability - 94