Information Technology Reference
In-Depth Information
Table 1 Historical load data statistics in MW
—
year 2007
Period
Mean
Median
Maximum
Full year
7,729
7,640
13,607
Summer (June
September)
6,609
6,600
9,182
-
Winter (October
March)
8,585
8,539
13,607
-
Winter (December
-
February)
9,290
9,307
13,607
Winter (December
February)
—
weekdays
9,758
9,823
13,607
-
Winter (December
February)
—
week 8 h to 22 h
10,350
10,284
13,607
-
total baseload amounting to about 13,500 MW. The most constraining contingency
is a busbar fault in the Brittany area that trips nearby generation units, which may
lead to a voltage collapse under extreme conditions. The parameters sampled to
generate operating conditions are variable part of total SEO load, SVC unavail-
ability and generator group unavailability in Brittany area. The unavailability of
main production units, consisting of nuclear groups at Civaux, Blayais, St-Laurent,
Flamanville, and Chinon are sampled such that each of these 5 unavailabilities is
represented in 1/6th of the total basecases. The unavailabilities of two SVCs at
Plaine-Haute and Poteau-Rouge are sampled such that 25 % of the cases have both,
25 % do not have both and 50 % have only one of them. The variable part of total
load, a continuous multivariate parameter, is sampled using our proposed ef
cient
sampling method. The power factor of loads is kept constant. All the load samples
are systematically combined with SVC and generator group unavailabilities
respecting their respective sampling laws to form various operating conditions.
Contingency analysis and database generation: For each condition, an optimal
power
ow
constraints in N. Then consequences of busbar fault are studied with a quasi steady
state simulation (QSSS) tool, where the simulation is run for 1,500 s and the con-
tingency is applied at 900 s. Scenarios are characterized as unacceptable if any of
SEO EHV bus voltage falls below 0.8 p.u or the simulation does not converge. Then a
learning dataset is formed using pre-contingency attributes of every scenario (sam-
pled at 890 s of QSSS) that drives voltage stability phenomenon, such as voltages,
active/reactive power
fl
flow is performed, minimizing the production cost under voltage, current,
fl
cations.
Then security rules are produced using decision tree to detect a probable voltage
collapse situation contingent upon the severe event. An independent test set is used to
validate the tree.
The software tools used in the study are:
fl
flows, productions etc., and their respective classi
1. ASSESS
Special platform for statistical and probabilistic analyses of power
networks (Available at:
http://www.rte-france.com/htm/an/activites/assess.jsp
)
2. TROPIC
—
—
Optimal Power Flow tool, embedded with ASSESS, to create initial
base cases
3. ASTRE
—
Simulating slow dynamic phenomena (QSSS), embedded with ASSESS
4. SAS
Statistical analysis and database processing
5. ORANGE, WEKA
—
—
Decision tree tools
Search WWH ::
Custom Search