Lessons Learned from the August 14, 2003 LOOP Events in North America

1
Lessons Learned from the August 14th, 2003 Loss
of Off-Site Power Events in North America
2
The Outage(s)
3
Lessons LearnedLOOP Frequency
4
Estimating Grid-LOOP Frequency
5
Grid Regions
Millstone
Ginna
PI
IPEC
BVPS
Braidwood
Byron
Fort Calhoun
Surry
Clinton
Calloway
McGuire
Oconee
Browns Ferry
Diablo
Robinson
Vogtle
ANO
SONGS
Farley
St.Lucie
Comanche
TPN
6
Types of Grid-Centered LOOPs
Here, I would like to point out that the LOOP
frequency in the model is a surrogate for a
complete model of the grid, just as LOCA
frequencies are a surrogate for a fracture
mechanics model of the RCS. Grid-LOOPs can be
broken into five categories. Unfortunately, the
more it is parsed, the fewer data points that
exist in each category. The grid-LOOP frequency
should represent the reliability of the grid and
thus characterizing the grid by counting LOOPs in
the NPP switchyard seems to be a poor surrogate.
The community needs to be able to characterize
the likelihood of the grid providing off-site
power from an easily obtainable data set.
7
1997 to 2004 LOOP Events
1G during Spring in WSCC Palo Verde
14Jun2004 4G during Summer in NPCC IPEC, NMP,
Fitzpatrick, Ginna 3G during Summer in ECAR
excludes DB 14Aug2003 EPRI Classed as Ia, Ib,
IIb, as well as 1 ECAR None
8
Grid-LOOP Events At A Site1 By Region-Season
from 1/1/1997 TO 12/31/2004 1 Events that
caused more than one LOOP on the same site are
only counted once.
Grid-LOOP Mid March to Mid June
Grid-LOOP Mid June to Mid September
Grid-LOOP Mid September to Late Dec
Grid-LOOP Late December to Mid March
SpringSummer Fall Winter
9
Multi-Unit Trips
10
Multi-Unit Trips Because of High-Voltage System
Problems
Multi-Unit Sites
11
Expected Summertime Trips
Part of that seasonal dependency (cited in
NUREG-1784) was based on a larger number of
summer time plant trips. The following
compares the published scram frequency in
NUREG-1784 (re Table C-7) combined with data for
2002 and 2003 from the INPO experience web site.
It shows that with one exception, the summer1
plant trip frequency is a relatively
stable fraction of the total number of plant
trips in the course of the year.
12
Energy Information Agency
13
(No Transcript)
14
Grid-LOOP Events At A Site1 By Region-Season
from 1/1/1997 TO 12/31/2004 1 Events that
caused more than one LOOP on the same site are
only counted once.
Grid-LOOP Mid March to Mid June
Grid-LOOP Mid June to Mid September
Grid-LOOP Mid September to Late Dec
Grid-LOOP Late December to Mid March
15
Grid-LOOP Events and Grid Equipment Related
Outages Reported to EIA
16
Grid-Induced Trips
A better characterization of the grid reliability
as seen by the NPPs is the grid-events that led
to a trip of the NPP generator. A list provided
by EPRI was parsed into plant, weather, and grid
by the authors.
17
Grid-LOOP and Grid Induced Trip Events At A
Site1 By Region-Season from 1/1/1997 TO
12/31/2004 1 Events that caused more than one
LOOP on the same site are only counted once.
18
Grid-LOOP and Grid Induced Trip Events At A
Site1 By Region-Season from 1/1/1997 TO
12/31/2004 1 Events that caused more than one
LOOP on the same site are only counted once.
19
Estimating Grid-LOOP Frequency
20
Lessons LearnedLOOP Duration
21
LOOP Duration
22
Practical Durations
To see these values by regional-seasonal factors,
the following was created for WCAP-16316. There
are few data points and charting by region and
season is not practical without a much more
sophisticated statistical method than simply
averaging numbers.
23
LOOP Duration
24
Lessons LearnedConclusions
25

• Data from both the Department of Energy and from
  EPRI suggest that the LOOP Event has a regional
  dependency.
• There is too little data from which to create
  regional-seasonal non-recovery curves. The
  best-estimate grid recovery times used to
  estimate LOOP duration should use the time needed
  to step through the reconnection procedures all
  the way to the first safety-bus.
• Wide-spread transmission failures that tripped an
  operating NPP off-line need to be counted as a
  full LOOP at those susceptible plants (i.e., the
  ones that tripped or would have tripped except
  for an on-going outage). For other plants in the
  same North American Electric Reliability Council
  region (e.g., NPCC, MAIN, ECAR), summer-time
  wide-spread grid events should be a fractional
  LOOP event in the total count of LOOP events
  appropriate in the calculation of a plant
  specific LOOP frequency (a 0.25 factor is
  proposed). Wide-spread grid events in other than
  the summer season seem to be unlikely based on
  the data presented herein. Because of the
  regional nature of grid-disturbances, plants
  outside of the regions where NPPs tripped can
  exclude those grid events from the count of valid
  LOOPs.