ECEN 615
Methods of Electric Power
Systems Analysis
Lecture 12: August 14 2003 Blackout
Prof. Tom Overbye
Dept. of Electrical and Computer Engineering
Texas A&M University
[email protected] Announcements
• Homework 3 should be done before the first exam
but need not be turned in
• Start reading Chapter 7 (the term reliability is now
often used instead of security)
• First exam is in class on Thursday Oct 1
• Distance learning students do not need to take the exam
during the class period
• Closed book, notes. One 8.5 by 11 inch notesheet and
calculators allowed
• Last’s years exam is available in Canvas with answers
1
Blackouts
• Blackouts are costly, with
some estimates of costs above
$100 billion per year.
• But blackouts are not created
equal. Some are unavoidable due
to large scale system damage
(hurricanes, tornados and ice
storms). Most are local,
distribution issues.
2
Some Electric Grid Risks
Image Source: Enhancing the Resilience of the Nation’s Electricity System, US National Academies Press, 2017
3
The Real Cause of Most Blackouts!
But mostly only
the small ones in
the distribution
system
Photo source: http://save-the-squirrels.com
4
High-Impact, Low-Frequency Events
• In order to enhance electric grid resiliency we need to
consider the almost unthinkable events
• These include what the
North American Electric
Reliability Corporation
(NERC) calls High-Impact,
Low-Frequency Events
(HILFs); others call them Image Source: NERC, 2012
black sky days
– Large-scale, potentially long duration blackouts
– HILFs identified by NERC were 1) a coordinated cyber,
physical or blended attacks, 2) pandemics, 3) geomagnetic
disturbances (GMDs), and 4) HEMPs 5
Avoidable Transmission Level
Blackouts
• Many major blackouts can be prevented.
• Time frames of the blackouts, minutes to hours,
allow for human intervention
– Tokyo 1987 (20 minutes), WECC 1996 (six minutes),
Eastern Interconnect 2003 (about an hour), Italy 2003 (25
minutes)
• And of course many are prevented, and hence do
not make the news. For example, near voltage
collapse in Delmarva Peninsula, 1999.
6
Going Back in Time
• The August 14th 2003 blackout is rapidly moving from
being a “recent event” into history; yet it still has much
to teach us.
• This talk is about the past and the future: what can we
learn from the past to help us prepare for the future
– But not so much about what are the immediate lessons from
the Blackout since many recommendations have already been
put into practice.
• The blackout final report is very readable and available
by googling “August 14 2003 Blackout Report”
7
In contrasting
numbers, the
August 14 2003
Blackout hit about
50 million people,
while Hurricane
Irene (2011) caused
power outages
affecting perhaps
seven or
eight million.
8
August 14, 2003 Hoax Image
This image was
widely circulated
immediately after
the blackout, even
appearing for a
time on a DOE
website. It was
quickly shown
to be a hoax.
What might
immediately give it
away?
9
Actual Before and After Images
10
My Favorite August 14, 2003 Cartoon
11
Causes of the Blackout
• Blackout Final Report listed four causes
– FirstEnergy (FE) did not understand inadequacies of their
system, particularly with respect to voltage instability.
– Inadequate situational awareness by FE
– FE failed to adequately manage their tree growth
– Failure of the grid reliability organizations (primarily
MISO) to provide effective diagnostic support
• Human/cyber interactions played a key role
12
We’ve Come Quite a Ways Since 2003
• Report included 46 recommendations, many of which
have dramatically changed the operation of the
interconnected power grid
– Thirteen were focused on physical and cyber security
• Focus of talk is what can 8/14/03 teach us to help
with the grid in 2020
• Need to keep in mind economic impact of 8/14/03
was above $5 billion; yearly impact of blackouts
could be above $100 billion
13
First Energy Control Center, Recent
(2013)
Image Source: www.wksu.org/news/story/365
14
My Involvement in Blackout
Investigation
• I spend a lot of time talking to reporters on 8/14 to
8/16, before I knew what happened
• Tasked by DOE to do onsite visit to FE on 8/19 to
8/21 with Doug Wiegmann; did similar visit to MISO
the next week.
• Did return visit in Oct
• Many folks played far
larger roles; I was only
involved extensively early
on
15
Footprints of Reliability Coordinators
in Midwest
16
August 13, 2003
• It is important to realize that immediately before the
blackout few people thought the system was on the
verge of a catastrophe.
• NERC 2003 Summer
Assessment did not
list Ohio as an area
of particular concern
NERC 2003 Summer Assessment is available at http://www.nerc.com/files/summer2003.pdf
17
August 14, 2003: Pre-blackout
(before 14:30 EDT)
• It had mostly been a normal summer day at First
Energy
– Most generation was available though the 883 MW Davis-
Besse Nuclear unit was on a long-term outage
– At 13:31 EDT the Eastlake 5 unit (a 597 MW plant on Lake
Erie) tripped when the operator tried to up is reactive output,
but this was not seen as a severe event
• It had been a busy day at MISO, with their reliability
coordinators dealing with a relatively small outage in
Indiana around noon
– Their state estimator failed at 1215 EDT but no one know this
18
Cinergy Bedford-Columbus 345 kV
Line Tree Contact at 12:08 EDT
19
Trees were Finally “Trimmed” Two
Months Later
20
At 14:27 EDT Star-South Canton 345
kV Line Trips and Recloses
• Star-South Canton is a tie between AEP & FE
• FE missed seeing this event since their alarms had
hung several minutes earlier (14:14)
• Line was back in service so it appeared normal in SCADA
• FE IT folks knew about computer problems
• AEP called FE at 14:32 to check on event; FE says
they saw nothing. A repeat call by AEP to FE at
15:19 also discusses event indicating ground current
was detected.
21
Estimated High Level Voltage Profile
at 15:00 EDT
22
Estimated Flows in Northeast Ohio at
15:00 EDT on August 14th 2003
Chamberlin-
Harding
345 kV Line
trips at 15:05,
an event
that was
missed by
both FE and
MISO
23
Estimated Flows in Northeast Ohio at
15:06 EDT
24
Line Outage Distribution Factors
(LODFs)
• LODFs are used to approximate the change in the flow
on one line caused by the outage of a second line
– typically they are only used to determine the change in the
MW flow
– LODFs are used extensively in real-time operations
– LODFs are state-independent (calculated using dc power
flow approximations) but do dependent on the assumed
network topology
– Below value tells change of real power flow on line for the
assumed outage of line k; ƒk0 is (obviously) pre-contingent
f = d ,k f k
0
25
Flowgates
• The real-time loading of the power grid is accessed
via “flowgates”
• A flowgate “flow” is the real power flow on one or
more transmission element for either base case
conditions or a single contingency
– contingent flows are determined using LODFs
• Flowgates are used as proxies for other types of
limits, such as voltage or stability limits
• Flowgates are calculated using a spreadsheet
26
Flowgate #2265
• Flowgate 2265 monitors the flow on FE’s Star-Juniper
345 kV line for contingent loss of the Hanna-Juniper
345 Line
– normally the LODF for this flowgate is 0.361
– flowgate has a limit of 1080 MW
– at 15:05 EDT the flow as 517 MW on Star-Juniper, 1004
MW on Hanna-Juniper, giving a flowgate value of
520+0.361*1007=884 (82%)
– Chamberlin-Harding 345 opened at 15:05; FE and MISO all
missed seeing this
27
The Bad LODF that Maybe Blacked
Out the Northeast
• At 15:06 EDT (after loss of Chamberlin-Harding
345) #2265 has an incorrect value because its
LODF was not automatically updated.
– Value should be 633+0.463*1174=1176 (109%)
– Value was 633 + 0.361*1174=1057 (98%)
• At 15:32 the flowgate’s contingent line opened,
causing the flowgate to again show the correct
value, about 107%
28
Flows at 15:33 EDT
29
Estimated Northeast Ohio 138 kV
Voltage Contour: 15:33 EDT
30
IT Issues
• MISO RCs had gotten many hundreds of “alarms”
• Contingency analysis results were giving pages of
violations.
• SE would fail because of severe system stress
• Inadequate procedures for dealing with SE failure.
• FE control center would get “many phone calls;”
information was not effectively shared.
31
Estimated Flows in Northeast Ohio at
15:46 EDT on August 14th 2003
32
Estimated Northeast Ohio 138 kV
Voltage Contour: 15:46 EDT
33
What Could Have Been Done?
Sammis-Star Flow Sensitivities
DOE/NERC
report said
about
1500 MW
of load
shed would
have been
needed
34
Estimated Flows in Northeast Ohio at
16:05 EDT on August 14th 2003
35
Estimated Northeast Ohio 138 kV
Voltage Contour: 16:05 EDT
36
Path to Cleveland Blocked after Loss
of Sammis-Star 16:05:57
Remaining
Paths
Image Source: August 14 2003 Blackout Final Report
37
345 kV Lines Trip Across Ohio to
West at 16:09
ONTARIO
38
Generation Trips 16:09:08 – 16:10:27
ONTARIO
Image Source: August 14 2003 Blackout Final Report
39
Parts of Ohio/Michigan Served Only
from Ontario after 16:10:37
Image Source: August 14 2003 Blackout Final Report
40
Major Power Reversal: 16:10:38
Image Source: August 14 2003 Blackout Final Report
41
Ontario/Michigan Interface Flows and
Voltage
Image Source: August 14 2003 Blackout Final Report
42
Ties from PJM to New York Open:
16:10:44 (North Ohio Black)
Image Source: August 14 2003 Blackout Final Report
43
System Islands Break Up and
Collapse: 16:10-16:13
Some Local Load
Interrupted
Areas Affected by the Blackout
Service maintained
in some area
Image Source: August 14 2003 Blackout Final Report
44
Are DC LODFs Accurate?
August 14th Crash Test
• Here are some results from August 14th
Time Contingency Element LODF MW (pred) MW (act)
15:05 Chamberlin- Hanna-Juniper 0.362 179 176
Harding 345 345
15:32 Hanna-Juniper 345 Star-Juniper 0.465 545 527
345
15:46 CantonCentral- Sammis-Star 0.164 48 54
Cloverdale 138 345
15:46 same Cloverdale-Star 0.234 68 64
138
16:06 Sammis-Star 345 Star-Juniper numerous 517 676
Star-Urban 138 345
W.Canton-Dale 138
16:06 same Ashtabula- numerous 319 408
Perry 345
45
The Results are Actually Quite Good!
• The initial LODF values were accurate to within
a few percent
• Even after more than a dozen contingencies, with
many voltages well below 0.9 pu, the purely DC
LODF analysis was giving fairly good (with
25%) results
46
What Could Have Occurred on
August 14th?
• With 20/20 Hindsight the blackout probably could
have been prevented. A smarter grid might have
provided the necessary situational awareness, and/or
provided the dynamic load reduction necessary to keep
the system from cascading.
• But key issues are 1) which grid improvement costs are
cost justified, and 2) what are we missing?
47
How Could a Smart Grid Help?
• Under frequency and under voltage relays can provide
quick reduction in the load, but they need to be smart
enough to make the right decision
• Dynamic pricing (LMPs) can help customers make
economic decisions, but they depend upon a variety of
“advanced applications” in order to calculate the
LMPs: state estimation converging to provide the
model for the SCOPF
48
Some Thoughts on Current Needs
• The data used in the models for interconnect wide
studies still have significant problems
• In US we have 100 GW of wind resources, but do not
always have adequate models for transient stability
studies; there are also potential low voltage ride
through issues with solar
• Power grid is rapidly changing which can result in
some operational “surprises”
• High impact, low frequency events are also a concern
• We need people with a deep knowledge of power
systems and (fill in the blank)!
49
