Professional Documents
Culture Documents
by
Gene Preston
g.preston@ieee.org
December 6, 2002
Seminar Purpose
Show the methodology for calculating the
reliability indices using graphics and examples
Define terms used in reliability studies such as
LOLE, LOLP, EUE, FOR, PFOR, pdf, etc.
Provide information to stakeholders concerning
input data and interpretation of study results
for single area and multi-area studies
Generation Adequacy Study Objectives
f(x)
total area = 1
random variables
f(x) F(x)
.5
1- area = value x
expected value or mean value
F(x) = 1 - Pr[generation MW x]
= Pr[generation MW > x]
Combination of two generator pdfs using
convolution
x1 gen 1 Pr x2 gen 2 Pr
.125 .25 .125 .1653
.375 .25 .375 .2123
.625 .25 .625 .2725
.875 .25 .875 .3499
Take all combinations of Prs and MWs
X Pr
0.25 0.0413250
0.50 0.0944000
0.75 0.1625250
1.00 0.2500000
1.25 0.2086750
1.50 0.1556000
1.75 0.0874750
Representation of pdfs with discrete states
(one generator with states: 0, Derated, Pmax)
.8
f(x)
.1 PFOR
.1 FOR
0 x - MW Derated Pmax
Generator failure as an exponential function of time
probability of failure =
t=0 1/
mean time to failure
up Pr[unit is up] = P1
3
3 PFOR = per unit derated time
(partial forced outage rate)
0 MW P3 FOR = per unit down time
Use of the Markov process to represent three
two-state generators
U U U
9 1 4 1 2.33 1
D D D
P2 P3 P5
UUD 2.33
UDU 2.33
DUU
4 9 9 4
1 1 1 1
P4 P6 1 P7
UDD DUD DDU
9 4 2.33
1
1 P8 1
DDD
Markov representation of three generators
3 -2.3 -4 0 -9 0 0 0 P1 0
-1 4.3 0 -4 0 -9 0 0 P2 0
-1 0 6 -2.3 0 0 -9 0 P3 0
0 -1 -1 7.3 0 0 0 -9 P4 0
-1 0 0 0 11 -2.3 -4 0 P5 0
0 -1 0 0 -1 12.3 0 -4 P6 0
0 0 -1 0 -1 0 14 -2.3 P7 0
1 1 1 1 1 1 1 1 P8 1
Markov representation of three generators
P1 .504 45 MW
P2 .216 25 MW
P3 .126 30 MW
P4 .054 10 MW
P5 .056 35 MW
P6 .024 15 MW
P7 .014 20 MW
P8 .006 0 MW
Use of a binary tree for the three generators to
perform the convolution process
Individual States Cumulative
Gen 1 Gen 2 Gen 3 MW Pr MW Pr Pr
20 .7 -- 45 .504 45 .504 .504
15 .8 0 .3 -- 25 .216 35 .056 .560
10 .9 0 .2 20 .7 -- 30 .126 30 .126 .686
0 .3 -- 10 .054 sort 25 .216 .902
20 .7 -- 35 .056 20 .014 .916
0 .1 15 .8 0 .3 -- 15 .024 15 .024 .940
0 .2 20 .7 -- 20 .014 10 .054 .994
0 .3 -- 0 .006 0 .006 1.000
Graph of the 3 generator cumulative distribution
for the probability that generation MW is > x)
.994
1.0 .940
.916 .902
.9
load not served
.8
.686
.7
.6 .560
Pr .504
.5
.4 generation
.3
.2
.1
.000
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Unsuitability of the binary tree and Markov
methods for large systems
400 120
2 10 states (combinations)
1.0
.9
.8
.7
.6 load not served
Pr
.5
.4
.3
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Cumulative distribution for generator 1
generator 1 = {10 MW, FOR=.1}
1.0 .900
.9
.8
.7
.6 load not served
Pr
.5
.4
.3
.2
.1 .000
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Adding generator 2 scales and shifts the initial
distribution for Pr=.8 (up) and Pr=.2 (down)
generator 2 = {15 MW, FOR=.2}
1.0
.9 1.00 x .8 =.80
.8 .900 x .8 =.72
.7
.6
Pr
.5
.4
.3
.900 x .2 =.18
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Summing the two curves gives the combined
generators 1 and 2 cumulative distribution
.80+.18=.98
1.0
.9 .80
.8 .72
.7
.6 load not served
Pr
.5
.4 generation
.3
.2
.1 .000
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Adding generator 3 scales and shifts the distribution
for Pr=.7 (up) and Pr=.3 (down)
generator 3 = {20 MW, FOR=.3}
1.0
.9
.8 1.00x.7=.700 .98x.7=.686
.7 .80x.7=.56
.6 .72x.7=.504
Pr
.5
.4
.98x.3=.294 .80x.3=.24
.3 .72x.3=.216
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Summing the two curves gives the combined
generators 1, 2, and 3 cumulative distribution
(same curve as the one using a binary tree)
.294+.700=.994
1.0
.9
load not served
.8
.686
.7
.6 .560
Pr .504
.5
.4 generation
.3
.2
.1
.000
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Binary tree graph of the 3 generator cumulative
distribution for the probability that generation is
available at x MW
.994
1.0 .940
.916 .902
.9
load not served
.8
.686
.7
.6 .560
Pr .504
.5
.4 generation
.3
.2
.1
.000
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Cumulative distribution Pr[generation is up]
represented in discrete 1 MW steps
1.0
.9
load not served
.8
.7
.6
Pr
.5
.4 generation
.3
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Flip the function over and backwards and the distribution
represents Pr[generation out of svc]
This is the COPT or Capacity Outage Probability Table
1.0
.9
.8
.7
.6
Pr .496
.5 .440
.4 .314
.3
.2 load not served .098
.084 .060
.1 .006 .000
0
45 40 35 30 25 20 15 10 5 0
x (MW)
Representation of Pr[gen out of service] as
piecewise linear increments
1.0
.9
.8
.7
.6
Pr
.5
.4
.3
.2 load not served
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Representation of Pr[gen out of service] as
piecewise quadratic increments
1.0
.9
.8
.7
.6
Pr
.5
for any 3 points,
.4 interpolate between
.3 the left two points
.2 load not served
.1
0
0 5 10 15 20 25 30 35 40 45 50
x (MW)
Relative per unit error introduced by numerous
interpolations of piecewise linear (PL) and
piecewise quadratic (PQ) distributions
0.06
0.05
0.04
PL
per unit error
0.03
x=30%
PQ
x=30%
0.02
PQ
0.01
x=20%
0
0 20 40 60 80 100 120
h grid spacing in MW for the ERCOT 286 generator p roblem
LOLE loss of load expectation
EUE expected unserved energy
EUE = MWH not served (for each hour in the study)
LOLE for one day
1.0 .994
= 1 - Pr[gen up]
.9 .940 .916 .902
= Pr[load loss]
.8 = 1-.56 = .44 d/y
.7
.686
.6
Pr .560
.5 .504
.4 generation
.3
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50 MW
x MW (daily or hourly)
LOLP loss of load probability
.994
1.0 .940
.916 .902
.9
.8
.686
.7
.6 .560
.504
Pr
.5
.4 generation
.3
.2
.1
0
0 5 10 15 20 25 30 35 40 45 50 MW
x MW (daily or hourly)
Generation scheduled maintenance
with
maint
planning
MW reserve
demand
insufficient reserve
1 5 10 15 20 25 30 35 40 45 50 52
with maint
planning
MW reserve
demand
1 5 10 15 20 25 30 35 40 45 50 52
90000
total generation
80000
60000
Megawatts
30000
20000
10000
0
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
Week
ERCOT Automatic Maintenance with 1 wk delay in pk demands
90000
total generation
80000
60000
Megawatts
30000
20000
10000
0
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
Week
Effect of 1 week delay in sched. maint. on LOLE
1.00E+00
1.00E-01
1.00E-02
delay in deferring maintenance
1.00E-03
LOLE - days/year
1.00E-04
1.00E-05
1.00E-06
1.00E-07
1.00E-08
1.00E-09
1.00E-10
1
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
Week
ERCOT generation (MW)
forced out of service and derated
1.00E+00
9.00E-01
actual more generators
7.00E-01
6.00E-01
5.00E-01
~5000 MW ~9000 MW
4.00E-01
~6% ~11%
3.00E-01
12.5%
2.00E-01
1.00E-01
0.00E+00
84316 82316 80316 78316 76316 74316 72316 70316
0 2000 4000 6000 8000 10000 12000
x MW out of service with 84316 MW installed
DC tie considerations
probability of a DC tie failure is nearly 0
probability of generation supply being available
in the other region is expected to be nearly 1
transmission constraints in the other region
may reduce the probability of DC tie capability
to less than 1
DC tie capacity can be included or excluded
from the LOLP calculations (affects the LOLE)
DC tie capacity may or may not be used to
serve firm load in ERCOT (affects the reserve)
DC tie considerations
DC tie in LOLP calculations
Yes No
LOLP:100% DC MW LOLP: X MW DC
LOLP:100% DC MW LOLP: 0 MW DC
LOSS OF LOAD EXPECTATION = 0.019290 DAYS/YR USING DAILY PEAK LOADS AND NO LOAD UNCERTAINTY
LOSS OF LOAD EXPECTATION = 0.001565 DAYS/YR USING HOURLY LOADS AND NO LOAD UNCERTAINTY
Transmission model considerations
Simplified transmission network (NARP)
requires the development of an equivalent
computationally fast
questionable circuit flow results
Full transmission network (PLF)
eliminates the need to develop an equivalent
computationally fast if PDFs are used
results are in agreement with AC load flow
Transmission PDFs from AC load flows
PDF = power distribution factor
= MW ckt flow / MW power transfer
difference in two shift factors
flow
load buses
gen
Harmful Generators for Austrop-Sandow Ckts
PDFs for ckt: 3429 SANDOW 345 - 7040 AUSTRO34 345 1
221 LOSTPN 2 -0.1734601 load flow bus 7008
222 LOSTPN 3 -0.1734599 load flow bus 7009
220 LOSTPN 1 -0.1734593 load flow bus 7007
257 BEP GT2 -0.1689710 load flow bus 7808
256 BEP GT1 -0.1689707 load flow bus 7807
258 BEP ST1 -0.1689678 load flow bus 7809
223 FPP G1 -0.1501281 load flow bus 7010
224 FPP G2 -0.1501194 load flow bus 7011
225 FPP G3 -0.1498940 load flow bus 7012
229 HAYSN 4 -0.1423499 load flow bus 7017
228 HAYSN 3 -0.1423486 load flow bus 7016
226 HAYSN 1 -0.1423371 load flow bus 7014
227 HAYSN 2 -0.1423371 load flow bus 7015
247 MCQUEE06 -0.1342581 load flow bus 7605
322 SANDH G1 -0.1341542 load flow bus 9016
323 SANDH G2 -0.1341542 load flow bus 9017
325 SANDH G4 -0.1341535 load flow bus 9019
324 SANDH G3 -0.1341533 load flow bus 9018
246 CANYHY06 -0.1338861 load flow bus 7487
313 DECKR G2 -0.1316094 load flow bus 9001
312 DECKR G1 -0.1315037 load flow bus 9000
More Generator PDFs for Austrop-Sandow
252 GUALUP 3 -0.1303046 load flow bus 7802
250 GUALUP 1 -0.1303038 load flow bus 7800
251 GUALUP 2 -0.1303038 load flow bus 7801
248 SCHUMA13 -0.1302463 load flow bus 7609
253 GUALUP 4 -0.1302456 load flow bus 7805
254 GUALUP 5 -0.1302456 load flow bus 7805
255 GUALUP 6 -0.1302456 load flow bus 7805
261 RIONO G3 -0.1293995 load flow bus 7812
259 RIONO G1 -0.1293991 load flow bus 7810
260 RIONO G2 -0.1293990 load flow bus 7811
315 DECKR G4 -0.1291926 load flow bus 9003
316 DECKR G5 -0.1291926 load flow bus 9004
317 DECKR G6 -0.1291926 load flow bus 9005
314 DECKR G3 -0.1291925 load flow bus 9002
263 RIONO G5 -0.1291662 load flow bus 7814
262 RIONO G4 -0.1291648 load flow bus 7813
264 RIONO G6 -0.1291633 load flow bus 7815
249 LAKEWD06 -0.1285502 load flow bus 7624
276 LAR #3 -0.1281178 load flow bus 8288
153 SAND 4 G 0.1279722 load flow bus 3432
SOUTH -0.1045032 load flow area 4
NORTH 0.0786292 load flow area 2
WEST 0.0504448 load flow area 1
HOUSTON -0.0243481 load flow area 3
Maximum Flows for Austrop-Sandow Ckts
1
ckt rating ckt rating
Pr [ flow>x ]
overload
0
0 MW
Removing a transmission overload
Pr Pr
circuit
rating
p
Pr
0
F(x)
Load shedding to remove Austrop-
Sandow circuit overloads
LINE-GENERATOR-AREA LOAD SHEDDING REPORT:
BUS# BUS NAME BUS# BUS NAME ID MW MWH GENERATOR > LOAD AREA PDF
3429 SANDOW 345 - 7040 AUSTRO34 345 1 167. 0.114381 LOSTPN 2 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 1 167. 0.043348 LOSTPN 3 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 1 167. 0.016571 LOSTPN 1 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 1 167. 0.006372 BEP GT2 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 1 167. 0.002422 BEP GT1 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 1 166. 0.000880 BEP ST1 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 1 584. 0.000473 FPP G1 NORTH 0.22876
3429 SANDOW 345 - 7040 AUSTRO34 345 1 584. 0.000009 FPP G2 NORTH 0.22875
3429 SANDOW 345 - 7040 AUSTRO34 345 2 167. 0.022171 LOSTPN 2 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 2 167. 0.010692 LOSTPN 3 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 2 167. 0.004809 LOSTPN 1 NORTH 0.25209
3429 SANDOW 345 - 7040 AUSTRO34 345 2 167. 0.002008 BEP GT2 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 2 167. 0.000778 BEP GT1 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 2 166. 0.000277 BEP ST1 NORTH 0.24760
3429 SANDOW 345 - 7040 AUSTRO34 345 2 584. 0.000138 FPP G1 NORTH 0.22876
3429 SANDOW 345 - 7040 AUSTRO34 345 2 584. 0.000002 FPP G2 NORTH 0.22875
1
Pr [gen is in service]
single area
unserved load
additional unserved load due
to a transmission constraint
0
0 x MW load
Overall effect of removing all
transmission overloads on the LOLE
1
Pr [gen is out of service]
single area
unserved load
0
Generation MW 0
Capability increasing load
An Example of ERCOT Increase in LOLE Due to
Transmission Constraints
(CSC Areas, Monitor 345 kV Circuits, No Circuit Outages)
100
10
LOLE - days/year
0.1
generation
0.01 unserved load
0.001
0.0001
transmission
0.00001
unserved load
0.000001
4
6
9
2
5
9
3
8
16
23
33
.3
.8
.4
.7
.4
.2
.9
.8
.8
.9
.2
0.
1.
2.
4.
5.
6.
8.
9.
11
12
14
17
19
21
24
26
28
30
35
LOLE when the peak demand reaches the above % reserve level
See my dissertation on egpreston.com for more details.
The End