Alexis Kwasinski, Ph.D.
R. K. Mellon Faculty Fellow in Energy
University of Pittsburgh, Swanson School of Engineering
Overview
 Introduction
 Energy Security and Resilience
 Critical Infrastructure Performance in Past Disasters
 Fundamental Notions and Definitions
 Planning and Design of Resilient Power Systems
 Applications
 Conclusions
 The content of these slides has been or is being supported by NSF and DTRA
 Special acknowledgment: Dr. V. Krishnamurthy
 Additional acknowledgments: Dr. M. Kim, Dr. T. Song and Dr. Andres Kwasinski (RIT)
© A. Kwasinski, 2015
2
Introduction
 What is energy security?
© A. Kwasinski, 2015
3
Introduction
Various definitions of energy security
• Users view: Availability and reduced cost definitions. Wealth
preservation.
• IEA:
• “Energy security is the uninterrupted availability of energy
sources at an affordable price.
• Long term energy security mainly deals with timely
investments to supply energy in line with economic
developments and environmental needs.
• Short-term energy security focuses on the ability of the
energy system to react promptly to sudden changes in the
supply-demand balance.”
© A. Kwasinski, 2015
4
Introduction
Various definitions of energy security
• Users view: Availability and reduced cost definitions. Wealth preservation
• EU:
• About the same to that of the IEA.
• “Achieving energy security requires
• to reduce risks to energy systems, both internal and external,
and
• to build resilience in order to manage the risks that remain.”
• From IEA general energy resilience is the “ability to cope
with…disruptions.”
• Energy security considerations must also be balanced against
competitiveness and environmental concerns – notably those
related to climate change.”
© A. Kwasinski, 2015
5
Introduction
Various definitions of energy security
• Growing users view: Dynamic global view
• China:
• Traditional view (from E. Downs, 2004): “State-centric, supply-side
biased, overwhelmingly focused on oil and with a tendency to equate
security with self-sufficiency “ based on coal and regulating domestic
demand.
• Evolved view from Brookings Institute (as China grew and lost self
sufficiency based on coal):
• Energy security is dynamic, uncertain, and influenced by multiple risks.
• A broader global economic security issue as global energy prices influence
China’s economy and vice-versa.
• “a solution to China’s domestic energy shortage cannot rely just on an
energy usage policy in the narrow sense, as China has traditionally
employed. Rather, energy security for China will require the integration of
energy policy with macroeconomic policy – such as fiscal and monetary
policies – and foreign policy, as well as international cooperation.”
© A. Kwasinski, 2015
6
Introduction
Various definitions of energy security
• U.S. view: A blurred vision
• U.S.A. (view #1):
• EPA/ORNL: Improved energy security is related to a
reduction in both financial and strategic risks of a disruption
in supply or spikes in energy costs.
• In practical terms such risk can be reduced (and, thus, energy
security is improved) by reducing dependence on a given
single source of energy by using diversifying energy sources.
© A. Kwasinski, 2015
7
Introduction
Various definitions of energy security
• U.S. view: A blurred vision
• U.S.A. (View #2 – White House):
• Focus on (energy )infrastructure in Presidential Policy
Directive 21: “The terms "secure" and "security" refer to
reducing the risk to critical infrastructure by physical means
or defense cyber measures to intrusions, attacks, or the
effects of natural or manmade disasters.”
• Improved energy security through:
• Increased energy utilization efficiency
• Source diversification (reduced use of foreign oil, increased
domestic oil and gas production, increased use of renewable
energy, improving use of coal).
© A. Kwasinski, 2015
8
Introduction
Various definitions of energy security
• U.S. view: A blurred vision
• U.S.A. (View #3 U.S. Congress):
• Congressional Budget Office (CBO): “Energy security is the
ability of households and businesses to accommodate
disruptions of supply in energy markets.
• The United States is more secure with regard to a particular
energy source if a disruption in the supply of that source
creates only limited additional costs for consumers.”
© A. Kwasinski, 2015
9
Introduction
Various definitions of energy security
• U.S. view: A blurred vision
• U.S.A. (View #4, State legislatures):
• National conference of State legislatures (NCSL): “Energy security refers to a
resilient energy system. This resilient system would be capable of
withstanding threats through a combination of active, direct security
measures—such as surveillance and guards—and passive or more indirect
measures-such as redundancy, duplication of critical equipment, diversity in
fuel, other sources of energy, and reliance on less vulnerable infrastructure.
• The Kansas Energy Security Act defines security as “ … measures that protect
against criminal acts intended to intimidate or coerce the civilian population,
influence government policy by intimidation or coercion or to affect the
operation of government by disruption of public services, mass destruction,
assassination or kidnapping.”
• Energy security focuses on critical infrastructure; a term that is receiving
increasing attention. The Homeland Security Act of 2002 and the USA Patriot
Act define critical infrastructure as “systems and assets ... so vital to the United
States that the incapacity or destruction of such systems and assets would have
a debilitating impact on security, national economic security, national public
health or safety, or any combination of those matters”
© A. Kwasinski, 2015
10
Introduction
Various definitions of energy security
• U.S. view: A blurred vision
• U.S.A. (View #5, DOD):
• Department of Defense (DOD): “Energy security means
having assured access to reliable supplies of energy and the
ability to protect and deliver sufficient energy to meet
operational needs.”
• The Department of Defense describes climate change as a
“threat amplifier;” … “destabilization driven by climate
change to increase the mission burden of the U.S. Military
(Military Advisory Board, 2007).”
© A. Kwasinski, 2015
11
Introduction
Various definitions of energy security
• NATO view: A diverse vision
• NATO
• In NATO there is no agreement on the definition of energy
security as it differs based on each country's needs.
• “We must make energy diversification a strategic
transatlantic priority and reduce Europe’s dependency on
Russian energy.” from NATO Secretary General Anders Fogh
Rasmussen in March 2014,
• “Energy infrastructure is a critical part of global energy
security, and is subject to a number of vulnerabilities.”
© A. Kwasinski, 2015
12
Introduction
Various definitions of energy security
• Unconventional producers view: Still focusing on domestic use.
• Russia:
• From Decision of the Government of Russian Federation No.
1234 from August 28, 2003 energy security refers to a “state of
protection of the country, its citizens, society, state, economy
from the treats to the secure fuel and energy supply” (note
that the term “secure” is used to define energy “security”).
• Additionally, “the full and secure provision of energy
resources to the population and the economy on affordable
prices that at the same time stimulate energy saving, the
minimization of risks and the elimination of threats to the
energy supplies of the country”.
© A. Kwasinski, 2015
13
Introduction
Various definitions of energy security
• Producers view: Driven by wealth generation / domestic economic
motors.
• Saudi Arabia:
• Demand-oriented security
• Focus on
• Protecting oil production
• Preserving demand (foreign markets)
• Revenue
• Preference for reduction in foreign domestic production
even through lowering own production prices
© A. Kwasinski, 2015
14
Introduction
Various definitions of energy security
• World Economic Forum:
• Energy security is an umbrella term that
covers many concerns linking energy,
economic growth and political power .
• The energy security perspective varies
depending upon one’s position in the
value chain.
• Consumers and energy-intensive
industries desire reasonably-priced
energy on demand and worry about
disruptions.
• Major oil producing countries consider
security of revenue and of demand
integral parts of any energy security
discussion.
© A. Kwasinski, 2015
15
Introduction
Various definitions of energy security
• Some general definitions (from academia)….
• “Energy insecurity can be defined as the loss of welfare that
may occur as a result of a change in the price or availability of
energy” (Bohi, Toman, and Walls, 1996).
• “Energy security is the continuity of energy supplies relative
to demand.” (C. Winzer, 2011)
© A. Kwasinski, 2015
16
Introduction
Energy Security Definition Highlights
• Dependencies and influences:
• External influences include climate change.
• There are opposing goals in terms of energy security between
energy producing nations and energy users nations.
• Systems and services necessary for societies subsistence and
growth depend on each other.
• Dependency is a linkage or connection between two systems,
through which the operation of one of these systems
influences the operation of the other system.
• Such influences may exist across nations’ borders,
© A. Kwasinski, 2015
17
Introduction
Energy Security Definition Highlights
• Dependencies and influences
• In terms of energy security, dependencies can be established
between
• two infrastructure systems
• two social systems
• a social system and an infrastructure system
© A. Kwasinski, 2015
18
Introduction
Energy Security Definition Highlights
• Dependencies and influences
• Financial and banking systems are critically important in
•
•
•
•
influencing energy security among nations.
The existence of global energy markets imply that energy prices
in a given country are influenced by other countries energy
policies and goals.
Energy security through self sufficiency is an utopian goal
because of the world economic ties and the dependencies
among energy infrastructure systems and financial and banking
systems.
Reduced sovereignty and opposing energy security goals among
producers and users is a source of conflict.
Alliances among nations to enhance security is key.
© A. Kwasinski, 2015
19
Introduction
Energy Security Definition Highlights
• Risk and Resilience
• Risk is considered in this context as an exposure to
disruptions.
• Resilience considered in this context as ability to recover
from disruptions
• From IEA Model for Short Term Energy Security (MOSES):
© A. Kwasinski, 2015
20
Resilience
 Energy Security and Resilience
© A. Kwasinski, 2015
21
Resilience
Energy Security Definition Highlights
• Broader Resilience Definition from U.S. PPD-21
• PPD-21 defines resilience as the ability to prepare for and
adapt to changing conditions and withstand and recover
rapidly from disruptions. Resilience includes the ability to
withstand and recover from deliberate attacks, accidents, or
naturally occurring threats or incidents.
© A. Kwasinski, 2015
22
Resilience
Energy Security Definition Highlights
• Broader Resilience Definition from U.S. PPD-21
• Four components of resilience:
• Ability to prepare for changing conditions
• Ability to adapt to changing conditions
• Ability to withstand disruptions
• Ability to recover rapidly from disruptions.
• Various temporal scales:
• Short term
• Medium term
• Long term
© A. Kwasinski, 2015
23
Resilience
Critical (energy) infrastructures
• They include:
• Infrastructures used by producers to harvest energy and process
it for transmission and distribution. Examples:
•
•
•
Oil production and refining
Natural gas production
Mining (for coal, uranium, etc. )
• Infrastructure used to deliver energy to users. Examples:
• Electric power grids
• Roads
• Waterways (including ports)
• Infrastructures used to convert energy for service operations .
Examples:
•
•
Power plants of information and communication technologies (ICT)
facilities
Power plants of steel and aluminum industry
© A. Kwasinski, 2015
24
Resilience
Critical infrastructures
• Critical infrastructures (CIs) is an integrated concept of a system
that includes:
• Physical components and its interconnections,
• Cybernetic operations and management platforms, data storage.
• Human resources, processes , contracts, etc. used to build, operate
and maintain such infrastructure.
Physical Domain
Environment
Processes and
human resources
Physical
resources and
components
Critical
Infrastructure
Human Domain
Information and
data storage,
processing and
transmission
Cybernetic Domain
© A. Kwasinski, 2015
25
Resilience
Critical infrastructure and Lifeline Systems.
Dependencies
• A lifeline system is a CI that is necessary for the operation of
another dependent infrastructure.
• Dependencies create vulnerabilities.
• Dependencies can be established through any of the 3 CI
dimensions
Environment
Processes and
human resources
Physical Domain
Physical Domain
Physical
resources and
components
Physical
resources and
components
Critical
Infrastructure
Human Domain
Information and
data storage,
processing and
transmission
Processes and
human resources
Cybernetic Domain
© A. Kwasinski, 2015
Critical
Infrastructure
Human Domain
Information and
data storage,
processing and
transmission
Cybernetic Domain
26
Critical Infrastructure
Performance in Past Disasters
© A. Kwasinski, 2015
27
C.I. Past Performance Examples
Oil and Natural Gas Production
• Natural disasters affect oil and natural gas production, processing and
distribution both directly and indirectly.
• It takes several days in order to restart a refinery after a power outage.
2011 Earthquake and
Tsunami in Japan
Indirect effect
Hurricane Ike
Direct effect
© Alexis Kwasinski, 2015
28
C.I. Past Performance Examples
Oil Distribution
• Gasoline (or diesel) distribution may be a critical issue in any type of
disaster affecting operation of transportation systems and other
infrastructures, such as ICT systems.
Superstorm Sandy
Superstorm Sandy
© Alexis Kwasinski, 2015
29
C.I. Past Performance Examples
Natural gas distribution
• Performance differs depending the type of disruptive events. E.g.
earthquakes affect natural gas distribution significantly ,whereas storms or
hurricanes have a more limited and localized effect on natural gas
distribution.
Hurricane Gustav
© Alexis Kwasinski, 2015
30
C.I. Past Performance Examples
Transportation Networks (Roads)
• Inhomogeneous performance depending the type of disruptive event.
• E.g. earthquakes
2010 Maule
Earthquake and
Tsunami in Chile
© Alexis Kwasinski, 2015
31
C.I. Past Performance Examples
Transportation Networks (Roads)
• Inhomogeneous performance depending the type of disruptive event.
• E.g. hurricanes
Superstorm Sandy
Superstorm Sandy
© Alexis Kwasinski, 2015
32
C.I. Past Performance Examples
Transportation Networks (Roads)
• Inhomogeneous performance depending the type of disruptive event.
• E.g. hurricanes
Hurricane Isaac
© Alexis Kwasinski, 2015
33
C.I. Past Performance Examples
Transportation Networks (Roads)
• Inhomogeneous performance depending the type of disruptive event.
• E.g. hurricanes
Hurricane Isaac
PLAQUEMINES
PARISH BORDER
© Alexis Kwasinski, 2015
34
C.I. Past Performance Examples
Transportation Networks (Roads)
• Lack of electric power lead to performance degradation because control
components (traffic lights) may be out of service or gas stations may not
operate.
2011 Earthquake and
Tsunami in Japan
Superstorm Sandy
© Alexis Kwasinski, 2015
35
C.I. Past Performance Examples
Waterways
• Very important infrastructure affecting other infrastructures and social
services (e.g. economy).
Hurricane Isaac
Superstorm Sandy
© Alexis Kwasinski, 2015
36
C.I. Past Performance Examples
Electric Power Grids
• Some relevant recent hurricanes: Katrina, Gustav, Ike, Irene (2011),
Isaac and Sandy (2012).
• Power outages extended over large areas and lasted from several days to
weeks.
• Extensive damage was mainly
observed in part of the areas
affected by the storm surge.
• Power outages originated primarily
in damage received by the
distribution portion of power grids.
(transmission recovered faster and
few power generation plants were
Hurricane Gustav
damaged)
© Alexis Kwasinski, 2015
37
C.I. Past Performance Examples
Electric Power Grids
• Of all discussed relevant recent hurricanes (Katrina, Dolly (2008), Gustav, Ike,
Irene (2011), Isaac and Sandy (2012)) only Katrina was a major hurricane when
making landfall.
• Katrina was a cat. 3 at landfall but only cat. 1 in New Orleans.
• Gustav (cat. 2) caused more outages in Louisiana than Katrina (cat. 3).
About 1,200K for Gustav vs. about 900K for Katrina.
• Ike’s outages extended from Texas to the Ohio River Valley.
• Irene was mostly a tropical storm, yet it caused about 6M power outages.
Sandy was a moderate cat. 1 hurricane (actually, an extratropical storm)
and caused almost 8.2M power outages.
© Alexis Kwasinski, 2015
38
C.I. Past Performance Examples
Electric Power Grids
• Of all discussed relevant recent hurricanes (Katrina, Dolly (2008),
Gustav, Ike, Irene (2011), Isaac and Sandy (2012)) only Katrina was a
major hurricane when making landfall.
• Ike was a cat. 2 storm, yet…..
© Alexis Kwasinski, 2015
39
C.I. Past Performance Examples
Electric Power Grids
• Some relevant recent earthquakes: Chile (2010), Christchurch (2/2011), and
Japan (3/2011).
• Power outages extended over large areas.
• Noticeable short and long-term drop in power demand
• Except in Japan, power generation issues were relatively minor.
• Issues in NZ with soil liquefaction affecting buried cables. Good
performance of the HVDC tie between the north and south islands.
• In all of these events strong shaking
damaged some substation components.
© Alexis Kwasinski, 2013
40
C.I. Past Performance Examples
Electric Power Grids (Superstorm Sandy)
• Relatively little damage to the power grid but outages were severe
• Areas with underground infrastructure: lower outage incidence but longer restoration
times.
• A short circuit in the substation of a power plant in Manhattan caused a significant
portion of the power outages in that island.
98% restoration
time
Outage Incidence
41
C.I. Past Performance Examples
Electric Power Grids (Superstorm Sandy)
Outage incidence
• Statistically, outages in some areas were much longer than usual.
• Outage incidence depends on local hurricane intensity and
infrastructure characteristics.
• Restoration time also depends on
logistics and restoration crews
management approaches
• Blue dots: data from the
hurricanes of the 2004 – 2008
Atlantic seasons
• Red lines: Regression curves for
the blue dots
Restoration time
• Red dots: data from Sandy
Local hurricane intensity
© Alexis Kwasinski, 2015
42
C.I. Past Performance Examples
Electric Power Grids (Superstorm Sandy)
• Often, damage to power grids is less severe than for residences.
• Storm surge damaged some substations in coastal areas
© Alexis Kwasinski, 2015
43
C.I. Past Performance Examples
Electric Power Grids (Superstorm Sandy)
• No observed damage to wind turbines. Only observed cases of damage in PV
systems:
44
© Alexis Kwasinski, 2015
C.I. Past Performance Examples
2011 Japan’s Earthquake & Tsunami
• Shaking damage was little. Tsunami damage was extensive on the coast.
• Power outages were extensive both on the coast and inland. Power issues and
restoration of all services were affected by Fukushima #1 nuclear power plant
incident. Coal fired and gas power plants were also damaged by the tsunami
and other nuclear power plants went offline.
• There were significant transportation issues specially during the first month
due to limited availability of gasoline, damaged roads in coastal areas and more
traffic (e.g. the army deployed more than 100,000 troops in the area).
© A. Kwasinski, 2015
45
C.I. Past Performance Examples
Power Grids (Japan’s EQ and Tsunami)
© Alexis Kwasinski, 2013
46
C.I. Past Performance Examples
Electric Power Grids (Japan’s EQ and Tsunami)
• Natural disasters are not single events. They are complex events with 4
distinct phases: pre-disaster (long-term aftermath), during the disaster,
immediate aftermath and intermediate aftermath.
Onagawa nuclear power plant: Offline since
the earthquake. Currently, almost all
nuclear power plants in Japan are offline
• As a result of the Fukushima #1 Nuclear Power plant event electric power
utilization in Japan has been affected, particularly during the
summer when rotating outages are likely. Public opinion has
created pressure to discontinue the use of nuclear power in
Europe, Japan and the US.
• In all these countries and regions wind power is seen as
an important alternative to nuclear power.
© Alexis Kwasinski, 2015
C.I. Past Performance Examples
Electric Power Grids (Chile’s EQ and Tsunami)
• Mismatches between earthquake intensity and power grid damage
Earthquake intensity (MMI)
© Alexis Kwasinski, 2015
Power grid damage
48
C.I. Past Performance Examples
Electric Power Grids (World’s Conflicts)
• Targeted in conventional conflicts (e.g. ww2 dambusters, U.S. attacks
on Iraq in 1991). Focus on power plants and substations (central
locations that cause significant disruptions when taken out of service).
• Also targeted in asymmetric conflicts (e.g. Peru’s Shining Path, El
Salvador’s FMNL, Iraq, Pakistan, Yemen, Germany’s Baader-Meinhof,
etc). Focus on transmission line towers (big impact targets difficult to
protect).
UK Ministry of
Defense
© Alexis Kwasinski, 2015
Forbes and alainonline.net
C.I. Past Performance Examples
Electric Power Grids
• Due to their predominately centralized control and power generation
architectures, power grids are very fragile systems in which little
damage may lead to extensive outages.
Power outage incidence
after Ike
Percentage of power grid damage after
Ike
© Alexis Kwasinski, 2015
50
C.I. Past Performance Examples
Electric Power Grids Weaknesses
• Predominant centralized architecture and control as seen by users.
• Passive transmission and distribution.
• Very extensive network (long paths and many components).
• Need for continuous balance of generation and demand.
• Difficulties in integrating meaningful levels of electric energy storage.
• Stability and power quality issues when
integrating significant levels of renewable
energy sources.
• Difficulties in integrating new loads
• Aging infrastructure
© Alexis Kwasinski, 2015
51
C.I. Past Performance Examples
Electric Power Grid Vulnerabilities
• Sub-transmission and distribution portions of the grid lack redundancy
• E.g., Only one damaged pole among many undamaged causing most of the
island to loose power.
Grand Isle, about 1 week after Hurricane Isaac
Entergy Louisiana
•I.e., an attack to a single pole miles away from a
military outpost may interrupt its power supply.
© Alexis Kwasinski, 2015
52
C.I. Past Performance Examples
Electric Power Grid Vulnerabilities
• Sub-transmission and distribution portions of the grid lack redundancy
© Alexis Kwasinski, 2015
53
C.I. Past Performance Examples
Electric Power Grid Damage Distribution
• Severe damage is often limited to relatively small areas
• During disasters damage distribution is inhomogeneous (e.g. Ike).
© Alexis Kwasinski, 2015
54
C.I. Past Performance Examples
Electric Power Grid – Underground
• Underground infrastructure
• is very costly (e.g. 5x more than overhead)
• is not effective for earthquakes
• with hurricanes, it yields lower failure
probability but it has longer repair times.
© Alexis Kwasinski, 2013
55
C.I. Past Performance Examples
Electric Power Grid – Human component
• Mitigation vs. logistics: a matter of cost and probability
• Effective management of logistics is key to reduce outage time.
• Restoration crews
• Flooded
substation
• About a week
after Isaac
© Alexis Kwasinski, 2015
• Flooded roads
(looking south)
56
C.I. Past Performance Examples
Electric Power Grid – Cybernetic component
• Despite claims on the contrary, past loss of service in public
communication networks have not led to electric power grids
outages (e.g. Buldyrev et. al. 2010 article in Nature).
• Currently, most electric power grids have dedicated
communication networks or use dedicated links.
• Deployment of smart grid technologies may lead to increase
use of public communication networks by electric utilities.
© Alexis Kwasinski, 2015
57
C.I. Past Performance Examples
ICT Networks (Hurricane Katrina)
• Power outages were a significant cause of communications failures
• 2.5 Million PSTN lines lost service.
• Storm surge destroyed 9 central offices and flooded 6 other COs. 5 of the 9
destroyed COs were restored with digital loop carrier (DLC) systems.
• 18 central offices lost service due to engine fuel starvation.
© A. Kwasinski, 2015
58
C.I. Past Performance Examples
ICT Networks (Hurricane Katrina)
• Most of the cell sites and existing digital loop carrier (DLC) systems failed due
to power-related issues. Only a small percentage were damaged (e.g. water
immersion or collapsed tower).
• Inconsistent building practices for cell sites. In a same site some base stations
above flood plane and the others below the flood plane.
• Damaged base stations restored with COWs or COLTs.
• Power restored to most undamaged base stations and DLCs with portable
gensets. Some cell sites had multiple deployed gensets.
© A. Kwasinski, 2015
59
C.I. Past Performance Examples
ICT Networks (Hurricane Gustav)
• Lessons from Katrina allowed to reduce communication outages.
• Power outage was more extensive than that caused by Katrina. Yet,
communication outages were small.
• No CO was damaged because the storm surge was not as strong as Katrina’s.
• Damage assessment identified a CO with genset issues.
• PSTN outages were reduced because many DLCs had been located on
platforms and equipped with permanent gensets since Katrina.
© A. Kwasinski, 2015
60
C.I. Past Performance Examples
ICT Networks (Hurricane Ike)
• Cat. 2 hurricane but the storm surge is comparable with a cat. 4 storm.
• 340,000 Public Switch Telephony Network (PSTN) outages.
• 12 COs lost service. One of those destroyed by the storm surge. One other may
have been damaged by storm surge waters but the remaining lost service due to
power issues.
• Service restored to the damaged CO with a switch on wheels.
© A. Kwasinski, 2015
C.I. Past Performance Examples
ICT Networks (Hurricane Ike)
• Power issues were the most important cause of outages in distributed network
elements and were a significant cause of communications outages.
• Only 3% of the more than 1,000 DLCs that lost service were destroyed.
• Few cell sites were damaged.
• COWs and COLTs were used to restore service or to improve network
coverage.
© A. Kwasinski, 2015
62
C.I. Past Performance Examples
ICT Networks (2010 Chile’s Earthquake & Tsunami)
• Shaking was not particularly intense but, still, power outages lasted in
important areas more than 2 weeks. Power issues were an important cause of
communication systems outages.
• 3 COs were affected by the tsunami. One CO lost service due to high
temperatures when the air conditioner stop working after the genset failed.
© A. Kwasinski, 2015
63
C.I. Past Performance Examples
ICT Networks (2010 Chile’s Earthquake & Tsunami)
•Almost all cell sites and most small remote switches lacked permanent
gensets.
• Shaking damaged batteries, antennas and other base stations equipment.
© A. Kwasinski, 2015
64
C.I. Past Performance Examples
ICT Networks (Feb. 2011 Christchurch, NZ Earthquake)
• Extensive soil liquefaction led to many buried power lines failures.
• Extensive use of micro and nano-cells imply many sites where gensets were
needed. Genset deployment needed to be prioritized.
• Only a few cell sites were destroyed. They were restored with COWs
• Cordoned-out areas in city downtown affected services restoration.
© A. Kwasinski, 2015
C.I. Past Performance Examples
ICT Networks (2011 Japan’s Earthquake & Tsunami)
• PSTN outages peaked at 1.5 Million 2 days after the earthquake.
• 26 COs were destroyed by the tsunami. Some were restored with DLCs or shelters with
switching equipment.
• COs were well constructed. In some towns the CO is one of the few buildings still
standing. Watertight doors reduced damages.
• Power issues affected many COs both on the coast and inland. Many COs require
portable generators to keep operation. Deployment of these generators and refueling
was complicated by road conditions and limited gas
© A. Kwasinski, 2015
C.I. Past Performance Examples
ICT Networks (2011 Japan’s Earthquake & Tsunami)
• Cells out of service peaked 6,720 on March 12th.
• Many cell sites in coastal areas were destroyed by the tsunami. Service was
restored with COWs or by increasing coverage of neighboring undamaged cells.
Also, small microcells linked with satellites were used.
• Power issues affected most of the cell sites that lost service. Few cell sites had
permanent gensets.
• The microgrid in Sendai did not lose service in its dc circuit.
© A. Kwasinski, 2015
67
C.I. Past Performance Examples
ICT Networks (2012 Hurricane Isaac)
• Lessons from Katrina and Gustav allowed to reduce communication outages
• Communication outages limited to distributed networks elements (cell sites, CATV
and PSTN outside plant electronic equipment). These outages lasted in some cases up to
a week.
• Most of the service disruption happened due to power issues in flooded areas south of
New Orleans outside the new levee protected area.
• Several sites had damaged propane tanks at ground level
5 days after the storm
(genset was deployed)
© Alexis Kwasinski, 2015
4 days after the storm (no
genset)
68
C.I. Past Performance Examples
ICT Networks (Superstorm Sandy)
• Flooding affected service in a few central offices.
• Most central offices did not present power issues
140 West St, NYC, 11/3/2012
© Alexis Kwasinski, 2015
69
C.I. Past Performance Examples
ICT Networks (Superstorm Sandy)
• Wireless networks lost coverage in coastal areas
and cities due to power issues and difficult
access to rooftop cell sites.
• Some PSTN vaults flooded
• Extensive use of COLTs and COWs
• Limited damage to outside plant cables.
© Alexis Kwasinski, 2015
70
C.I. Past Performance Examples
ICT Networks (Superstorm Sandy)
• Performance of data centers was an important aspect of the impact of
Sandy on ICT systems.
• Damage to generators pumps due to flooding was reported in two
large data centers.
• Fuel distribution issues also affected operations in some data centers
© Alexis Kwasinski, 2015
71
C.I. Past Performance Examples
Electric Power Grid – Conventional solutions
• They have limited effectiveness because they do not address inherent
vulnerabilities of power grids
• Solutions for reducing damage to power grids:
• Infrastructure hardening:
•Tree trimming programs
• Reinforced poles
• Underground infrastructure
• Solutions to accelerate restoration times once outages occur:
• Mobile transformers
• Portable diesel generators
(“emergency microgrids”)
© Alexis Kwasinski, 2015
72
C.I. Past Performance Examples
Electric Power Grid – Conventional solutions
• “Emergency microgrids” using diesel
generators connected to the power
distribution grid has been used in the
past but it presents issues (e.g. safety,
reliability, power quality….)
© Alexis Kwasinski, 2015
73
C.I. Past Performance Examples
Electric Power Grid – Smart Grid Technologies
• Community energy storage
• Advanced distribution automation
• Integrated communications
• Smart meters
• Phasor measurement units
• Residential photovoltaic (PV) systems
• Advanced loads including electric vehicles
• These solutions are limited because they do not address inherent
problems in power grids.
• Loads are becoming a valuable asset
• Smart meters and other related smart grid technologies facilitate
detecting an outage but they provide very limited improvement to avoid
outages or to mitigate their effects.
© Alexis Kwasinski, 2015
74
C.I. Past Performance Examples
Electric Power Grid – PV Inverters
• Lower 9th Ward after Hurricane Isaac:
The sun was shinning but
no grid = no power
(even with PV arrays
because of IEEE 1547).
Entergy New Orleans
© Alexis Kwasinski, 2015
C.I. Past Performance Examples
Electric Power Grid – Customer-based solutions
• Standby power systems (i.e. systems with local energy storage often in
batteries and/or diesel).
• Issue: high failure to start probability
for standby gensets
• Low availability of gensets for long
operating times
• Microgrids
• They are predominately a power
system (as oppose to energy
systems).
• Reduced energy storage may lead
to lower availability
© Alexis Kwasinski, 2015
Cell site with a standby diesel genset after
Hurricane Ike
Fundamental Notions
 Reliability, Availability and Resilience
© A. Kwasinski, 2015
77
Reliability
• Reliability applies to components. Once they fail, they cannot be
repaired.
• Reliability, R, is defined as the probability that an entity will operate
without a failure for a stated period of time under specified conditions.
• Unreliability is the complement to 1 of reliability (F = 1 – R)
F(t) = Pr{a given item fails in [0,t]}
• F(t) is a cumulative distribution function of a random variable t with a
probability density function f(t).
• Both F(t) and f(t) can be calculated based on a hazards function h(t)
defined considering that h(t)dt indicates the probability that an item fails
between t and t + dt (“event A”) given that it has not failed until t (“event
B”). From Bayes theorem
Pr{B | A}Pr{A} Pr{ A}
h(t )dt  Pr{ A | B} 

Pr{B}
Pr{B}
© Alexis Kwasinski, 2015
78
Reliability
h(t )dt  Pr{ A | B} 
Pr{B | A}Pr{A} Pr{ A}

Pr{B}
Pr{B}
• Since
• Pr{B|A} = 1
• Pr{A} = f(t),
• Pr{B} = 1 - F(t).
• Then
h(t )dt 
f (t )
1  F (t )
• and
t
h ( ) d

0
F (t )  1  e

© Alexis Kwasinski, 2015
79
Reliability
• The hazards function may take various forms and is a combination of
various factors. Typical forms for electronic components (solid lines) and
mechanical components (doted lines) with the three most characteristics
components (early mortality, random and wear out) are
© Alexis Kwasinski, 2015
80
Reliability
• Considering electronic components during the useful life period, the
hazards function is constant and equals the so called constant failure rate
λ. So,
R(t)
F(t) = 1 – e- λt
f(t) = λe- λt
R(t) = e- λt
t
• And,
E [ f (t )] 


0
tf (t )dt 
1

• The inverse of λ is called the Mean Time to Failure. I.e.,it is the expected
operating time to (first) failure
© Alexis Kwasinski, 2015
81
Reliability
• The failure rate of a circuit is in most cases the sum of the failure rate of its
components.
• General form for calculating failure rate (from MIL-Handbook 217):
adj  base Q T  E O
Production
quality
Thermal stress Electrical
stress
Other factors (power
and operational
environment factors)
• Aluminum electrolytic capacitors tend to be a source of reliability concern
for PV inverters. Although their base failure rate is low (about 0.50 FIT),
the adjusted failure rate is among the highest (about 50 FIT). Compare it
with a MOSFET adjusted failure rate of about 20 FIT.
• NOTE: FIT is failures per 109 hours.
© Alexis Kwasinski, 2015
82
Availability
• Availability applies to systems (which can operate with failed
components) or repairable entities.
• Definitions depending application:
• Availability, A, is the probability that an entity works on demand. This
definition is adequate for standby systems.
• Availability, A(t) is the probability that an entity is working at a specific time
t. This definition is adequate for continuously operating systems.
• Availability, A, is the expected portion of the time that an entity performs its
required function. This definition is adequate for repairable systems.
• Consider the following Markov process representing a repairable entity:
© Alexis Kwasinski, 2015
83
Availability
λ is the failure rate and μ is the repair rate. The probability for a repairable
item to transition from the working state to the failed state is given by λdt
and the probability of staying at the working state is (1-λ)dt. An analogous
description applies to the failed state with respect to the repair rate.
• The probability of finding the entity at the failed state at t = t +dt is
identified by Prf(t + dt) then this probability equals the probability that
the item was working at time t and experiences a failure during the
interval dt or that the item was already in the failed state at time t and it is
not repaired during the immediately following interval dt. In
mathematical terms,
Prf(t + dt) = Prw(t)λdt + Prf(t)(1-µ)dt
© Alexis Kwasinski, 2015
84
Availability
• Hence,
Pr f (t  dt )  Pr f (t )
dt
 Prw (t )  Pr f (t ) 
• Which leads to the differential equation
d Pr f (t )
 (   ) Pr f (t )  
dt
• With solution (considering that at t = 0 it was at the working state)
Pr f (t ) 
Prw (t ) 

1 e


 (    )t


1
   e(    )t

© Alexis Kwasinski, 2015

85
Availability
• When plotted:
• If we denote the inverse of λ as the Mean Up Time (MUT), TU, when the
system is operating “normally” and the inverse of μ as the Mean Down
Time (MDT or off-line time), TD, then as t tends to infinity
A  Prw (t  ) 
• That is,
Availability =
TU
TU



MTBF TU  TD   
Expected time operating “normally”
Total time (“normal” operation + off-line time)
© Alexis Kwasinski, 2015
86
Availability
• Notes:
• Unavailability is defined as
Ua 
MDT


MTBF   
• Mean time between failures (MTBF) is the sum of TD and TU
UP
DOWN
• Ways of improving availability
• Modularity
• Redundancy (parallel operation of same components)
• Diversity (use of different components for the same function
• Distributed functions
© Alexis Kwasinski, 2015
87
Availability
• About the common claim of data center operators of having “diverse power feeds:” Two
power paths imply redundancy, not diversity because the grid is one.
© A. Kwasinski, 2015
88
Availability
• Now consider a two-components system (A and B). The Markov process
is now
T
• So,
 dP 
T

P
A


 dt 
• Where,
A
B
0
  (  A  B )





(



)
0

A
A
B
B

A


B
0
(  B   A )
A


0



(



)

B
A
A
B 

PT  PrS1 (t ) PrS2 (t ) PrS3 (t ) PrS4 (t )
© Alexis Kwasinski, 2015

89
Availability
• The expected time that the system remains in each of the states is given
by
Ti 
1

 aii
1
NS
a
ij
j 1
j i
• The probability density function of being at state Si is
fTi (Ti   )  aii eaii
• the frequency of finding the system in state Si is
i  aii PrSi (t  )
© Alexis Kwasinski, 2015
90
Availability
• Hence, for the two-components system (A and B).
© Alexis Kwasinski, 2015
91
Availability
• If in a system all components need to be operating in order to have the
system operating normally, then they are said to be connected in series. This
“series” connection is from a reliability perspective. Electrically they could
be connected in parallel or series or any other way. The availability of a
system with series connected components is the product of the components
availability.
AS   ai
• If in a system with several components, only one of them need to be
operating for the system to operate, then they are said to be connected in
parallel from a reliability perspective. The system unavailability equals the
product of components unavailability, where the unavailability, U, is the
complement to 1 of the availability (U = 1 – A).
U P   ui
© Alexis Kwasinski, 2015
92
Availability
• For a series two-components system
(both A and B need to operate
for the system to operate).
System availability
Working state
© Alexis Kwasinski, 2015
Failed states
93
Availability
Failed
state
• For a parallel two-components system
(either A or B need to operate
for the system to operate).
System unavailability
Working states
© Alexis Kwasinski, 2015
94
Availability
• The most common redundant configuration is called n + 1 redundancy in
which n elements of a system are needed for the system to operate, so one
additional component is provided in case one of those n necessary
elements fails.
• n +1 redundant configuration.
But more modules is not always
better:
A
a = 0.97
A  (n 1)anu  an1
• Availability decreases when n
increases to a point where A < a
© Alexis Kwasinski, 2015
95
Availability
• For more complex systems, availability can be calculated using minimal cut
sets
• A minimal cut set is a group of components such that if all fail the system
also fails but if any one of them is repaired then the system is no longer in a
failed state. The states associated with the minimal cut sets are called minimal
cut states.
• Much simpler than Markov approaches.
© A. Kwasinski, 2015
96
Availability
•Unavailability with minimal cut sets:
 MC

US  P  K j 
 j 1 
• Calculation:
M c i 1
Mc
 P( K )   P( K
i 1
i
i 2 j 1
i
Mc
Mc
i 1
i 1
K j )  U S  1   [1  P( Ki )]   P( K i )
• Approximation with highly available components:
cj
MC
MC
j 1
j 1 l 1
U S   P( K j )   ul , j
© Alexis Kwasinski, 2015
u


97
Availability
• Typical availabilities standby power plants
Ac mains: 99.9 %
 Power plant: 99.99 %
(without batteries)
- 48 V
Genset: 99.4 % (includes TS)
(failure to start = 2.41 %)
Each rectifier: 99.96 %
n+1 redundant configuration is
used for improved availability
© Alexis Kwasinski, 2015
98
Availability
• Availability Calculation standby power plants before batteries
• Binary representation of Markov states:
• 1st digit: rectifiers (RS) with n+1
redundancy
• 2nd digit: ac mains (MP)
• 3rd digit: genset (GS) (failure to start
probability given by ρGS
• Availability of power plant without batteries:
APP
where RS
  GS  GS  MP  MP
 1 
 MP (  MP  GS )

n (n  1)

(n  1)R   R
2
R
 RS 

 ATS ARS

2r2 rn n 1Cn 1
n 1
C
i
i 0
n 1
ri rn 1i
© A. Kwasinski, 2015
k
n
k!
Cn    
 k  (n  k )! n !
99
Availability
• Availability Calculation standby power plants before batteries
• System availability equation: P(t )  AT P(t )
0
(1  GS )MP
GS MP
RS
0
0
0
 (MP  RS )



GS
( GS  MP  RS )
0
MP
0
RS
0
0


 MP
0
(GS   MP  RS )
GS
0
0
RS
0


0
 MP
GS
( GS   MP  RS )
0
0
0
RS


A

 RS
0
0
0
(MP   RS )
0
(1  GS )MP
GS MP


0
 RS
0
0
GS
( GS  MP   RS )
0
MP


0
0
 RS
0
 MP
0
(GS   MP   RS )
GS



0
0
0
 RS
0
 MP
GS
( GS   MP   RS ) 

• Failure probability (in time):
PPPf (t ) 
P
S i F
Si
(t )  1 
P
Si W
Si
(t )
• The probability density function fPPf(t) associated with the probability of
leaving the set of failed states after being in this set from t = 0 and entering
the set of working states at time t + dt is
f PPf (t )  a e a t
where aF = 3μRS + μMP + μGS
© A. Kwasinski, 2015
100
Availability
• Availability Calculation standby power plants before batteries
• Notice that aF = 3μRS + μMP + μGS is the sum of the transition rates from failed
states (called minimal cut states) to immediately adjacent working states.
© A. Kwasinski, 2015
101
Availability
• Availability Calculation standby power plants before batteries
• The probability of discharging the batteries is, then
PBD (t  TBAT )  1  
 TBAT
 0
f PPf ( )d  e  a
TBAT
• System unavailability or outage probability:
PO  e aF TBAT lim PPPf (t )  e  aF TBAT U a
t 
• Two cases are exemplified:
• Case A: With a permanent
genset.
• Case B: Without genset
© A. Kwasinski, 2015
102
Availability
• Importance of local energy storage
Needed availability
> 0.99999
• Telecom power plants are needed in order
to overcome grid’s low availability.
• Battery energy storage is essential in order to
reach telecom-grade availability levels. Still,
Power availability for air conditioners is below
the minimum required in telecom applications
Grid availability = 0.999
A = 4-nines
A = 3-nines
A > 5-nines
A = 2.5-nines
A = 4-nines
Typical availability in normal conditions
© Alexis Kwasinski, 2015
103
Availability
• In general, when batteries are considered the unavailability is
U w / B  U w / oB e




 MCS ,i TBAT  

  

 i mcs

Total unavailability
Base unavailability
(without batteries)
Heavily depends on unavailability of
the electric grid tie

Total availability
 1  Aw / B
Repair rate from a
minimal cut state to an
operational state
(Depends on logistics,
maintenance processes,
etc.)
Batteries (local
energy storage)
autonomy
Local energy storage contributes to
reduce unavailability
Optimal sizing of energy storage depends on
expected grid tie performance and local power
plant availability
© Alexis Kwasinski, 2015
104
Availability
• In general, when batteries are considered the unavailability is
U w / B  U w / oB e



 MCS ,iTBAT 
  
 i mcs


 1  Aw / B
Related with minimal cut
states
© Alexis Kwasinski, 2015
105
Resilience
Relevance of Resilience in Security
• The Presidential Policy Directive 21 identifies “energy and
communications systems as uniquely critical due to the enabling
functions they provide across all critical infrastructure sectors.”
• From EU’s security vision: “Achieving energy security requires to
reduce risks to energy systems, both internal and external, and to
build resilience in order to manage the risks that remain.”
• Resilience and security are implicitly related in other definitions of
energy security.
Resiliency (from PPD21):
• “The ability to prepare for and adapt to changing conditions and
withstand and recover rapidly from disruptions.”
© A. Kwasinski, 2015
106
Resiliency Metrics
Resiliency (from PPD21):
• The ability to prepare for and adapt to changing conditions
and withstand and recover rapidly from disruptions.
• “Withstand” refers to an “up” time
• Rapid recovery refers to a “down” time
• Inclusion of an up and a down time points towards an
equivalence between the concept of base resiliency and that of
availability.
• Preparation and adaptation relates to the influence of processes
through the down time
© Alexis Kwasinski, 2015
107
Resilience Metrics
• Base resiliency:
• N is the total number of customers in a given area, TU,i is the time when
customer i receives electric power during the total measured time T (no
“mean” behavior here).
• TU,i (withstanding characteristic) is mostly related with hardware issues.
• TD,i (recovery speed) is the down time, which is influenced by human
processes and aspects, such as logistical management, as well as hardwarerelated issues.
• Base resiliency is analogous to the average service availability index (ASAI) or,
more generally, to availability. However, resilience does not consider a longterm steady state performance.
© Alexis Kwasinski, 2015
108
Resilience Metrics
• Accounting for dependencies….. Local resilience:
where TS is the energy storage autonomy, μ is the combined repair rates from
the minimal cut states (for a simple grid tie it is just the inverse of the down
time for that grid tie).
• RI is the individual resilience measured as
• The inverse of μ acts as an “inertia” indicating how much resistance there is
locally for a change in local resilience or how much more or less energy storage
is needed for a same local resilience
© Alexis Kwasinski, 2015
109
Resilience Metrics
•Key observations
• Degree of dependence could be measured based on energy storage
requirements.
• Dependencies add vulnerabilities potentially reducing resilience.
• Degree of dependence could be managed through local energy storage.
I.e. local energy storage are buffers regulating (and in an extreme case
decoupling) dependencies.
Environment
Processes and
human resources
Physical Domain
Physical Domain
Physical
resources and
components
Physical
resources and
components
Critical
Infrastructure
Human Domain
Information and
data storage,
processing and
transmission
Cybernetic Domain
Processes and
human resources
Critical
Infrastructure
Human Domain
© Alexis Kwasinski, 2015
Information and
data storage,
processing and
transmission
Cybernetic Domain
110
Resilience Metrics
Resistance vs. resilience
• Let’s define an outage incidence θ as
where no is the number of users experiencing an outage.
The maximum outage incidence θmax is observed when
no equals the peak number of outages No observed during T.
• Then, individual resistance φI is
Resilience vs. restoration speed
• Restoration speed for N users and for 1 user are, respectively
and
© Alexis Kwasinski, 2015
111
Planning and Design of
Resilient Power Systems
© A. Kwasinski, 2015
112
Planning for Improved Resilience
• Does power infrastructure have to be repaired to its pre-natural
disaster condition even when significant demand is lost?
Bolivar Peninsula
after Hurricane
Ike (2008)
Minamisanriku (Japan) after the 2011 tsunami
© Alexis Kwasinski, 2015
113
Planning for Improved Resilience
• Planning difficulties impacting network reconstruction decisions: City
repopulation
The Brookings Institution Metropolitan
Policy Program & Greater New Orleans
Community Data Center
Alexis Kwasinski,
© A.©Kwasinski,
2015 2015
114
Planning for Improved Resilience
• Optimum level of protection (vs. cost):
• Tool: Risk assessment (Risk = Probability*Impact)
• Human perception influences risk calculation
• There is no certainty a given event will happen
Added portion of the seawall
Otsuchi. Was it a “sufficient” level of protection?
Onagawa nuclear power plant
© Alexis Kwasinski, 2015
115
Planning for Improved Resilience
• Costs and decision making
• By quantifying resilience (availability) one can evaluate downtime cost
and compare powering options with different availabilities.
• Objective evaluation tool: risk assessment.
Risk = (Probability of an event to happen) x (Impact of the event)
Event: loss of power as a result of a given disaster
Impact: cost of not having electric power during and after a disaster
• Decision approach to choose a new technology (e.g. microgrid):
Retrofit: Risk of existing technology > Lifetime cost of new technology
New: Lifetime cost existing tech. > Lifetime cost of new tech.
- Note: Lifetime cost includes risk, O&M costs and capital cost.
© Alexis Kwasinski, 2015
116
Solutions for Improved Resiliency
• Microgrids:
• are locally confined and independently controlled electric power grids
in which a power distribution architecture integrates loads and
distributed energy resources—i.e. local distributed generators and
energy storage devices—which allow the microgrid to operate
connected or isolated to a main grid.
• Well designed microgrids
can achieve very high
availability levels and provide
a solution for resilient power
supply when a disaster strikes
• Microgrids have operated
satisfactorily after Irene, Sandy
and the 2011 earthquake in Japan
© Alexis Kwasinski, 2015
117
Solutions for Improved Resiliency
• Key observation that leads to microgrid-based solutions:
• During disasters damage distribution is inhomogeneous
© Alexis Kwasinski, 2015
118
Microgrids Resilience
RMG  AMG  1  (1  RB )e
Total resiliency
(availability)
Base resiliency
(without batteries)
Heavily depends on unavailability
of power sources




( iTES ) 


 i M mcs


Repair rate – related
to the inverse of the
down time
(Depends on
logistics)
May depend on lifeline
performance (if they are not
renewable energy sources)
Local energy
storage (e.g.
batteries)
autonomy
Local energy storage improves
resiliency
• Two types of sources:
• Those which depend on another infrastructure (lifeline).
• Renewable sources
© Alexis Kwasinski, 2015
119
Microgrids
Key characteristics
• Sustainable systems, in the sense that they endure, are resilient systems.
• Higher efficiency, and less volume and area also supports resiliency.
• Distributed generation leads to a de-centralized control architecture.
• Distributed generation adds active elements which support independent
control strategies.
• Micro-grids require diverse
power inputs because each
distributed generation technology has worse availability
than power grids.
© A. Kwasinski, 2015
120
Microgrids
Lifeline-Dependent Power Sources
• There are two types of sources: those that depend on a lifeline and those that
do not depend on lifelines (e.g. renewable energy sources).
• Lifelines are infrastructures used by local power generators to receive energy.
Lifelines are affected differently by various natural disasters.
Transportation and fuel
delivery
Electric grid
© A. Kwasinski, 2015
Natural gas
121
Microgrids
• Lifeline dependency. E.g. Hurricane Isaac
• Electric service
interruption
Port Sulfur,
Oct. 2010
• Communication roads interrupted = Flooded roads made impossible to
deliver fuel for permanent diesel gensets
© Alexis Kwasinski, 2015
122
Microgrids
• Lifeline dependency and restoration logistics. E.g. Hurricane Isaac
• Fuel supply close to the flood
line
• Electric utilities’ trucks being concentrated 1 mile behind
the flood line waiting to enter flooded area
Port Sulfur,
Oct. 2010
123
Microgrids
Lifeline-Dependent Power Sources
• Approaches to limit the negative impact of lifeline dependencies on microgrid
availability:
• Use of diverse power source technologies (e.g. combine natural gas and
diesel, or natural gas and renewable energy sources)
• Local energy storage
© Alexis Kwasinski, 2015
124
Microgrids
Lifeline-Dependent Power Sources
• Continuous fuel supply (e.g. through a pipeline)
f
Continuous
uCOFS 
operation
 f  f
uSB 
MG (Gen  Gen  MG )
 MG (  MG  Gen )
Standby
• Discontinuous fuel supply (with local energy storage… e.g. diesel storage)
PE  1  PE*  P{td  TTC }  1  
td TTC
td  0
f d (td )dtd
E.g. triangular fd(t)
uf 
TM  TTC
PE*
3 TUf  3TTC  TM  TTC
PE
© Alexis Kwasinski, 2015
125
Microgrids
Lifeline-independent Power Sources - Renewables
• Most renewable energy sources do not require lifelines, but…..
• Issues with PV systems: large footprints. Solution:
• Size PV arrays for less than the required load and use it to support
another power source rated at full capacity.
• Renewable energy sources have, typically, variable output. Solutions:
• Local energy storage (e.g. batteries)
• Source diversification (combine wind and PV)
• Aesthetics
2x350 kW
natural gas
generators
50 kW
PV array
© Alexis Kwasinski, 2015
126
Microgrids – Renewables with Batteries
Max
SOC
 k11
k
 11
 p 2

P





0
p1
p 1
0
p 2
p 1






p2 

k NN 
k NN  N  N
Courtesy: Ted Song
127
p1
p2
0
p1
p 1
0
Capacity  ( N  1)T 
 P
Energy difference
between two states
A  1  E
Microgrids – Renewables with Batteries
Solar
Wind
Load
Courtesy: Ted Song
128
Microgrids – Renewables with Batteries
Solar
Dist.
Solar
Rnd.
Wind
Dist.
Wind
Rnd.
Load
Dist.
Load
Rnd.
Courtesy: Ted Song
129
Microgrids – Renewables with Batteries
PB  PPV  L
Courtesy: Ted Song
130
Microgrids – Renewables with Batteries
 0.3041 0.0247 0.0248
 0.3041
0
0.0247

 0.2890 0.0329
0
P










0
0.6959 

0.0329 0.6959  301301
Courtesy: Ted Song
131
Microgrids – Renewables with Batteries
  0.0042 0.0001
π1
π2
0.0036 0.7930 1301
πN
π3
 P
Courtesy: Ted Song
132
Microgrids – Renewables with Batteries
Courtesy: Ted Song
133
Microgrids
Lifeline-Dependent Power Sources
• Renewable energy sources with batteries
PV
PV (75%)
© Alexis Kwasinski, 2015
134
Microgrids
Accounting for power source interfaces
• Many possible configurations. Test cases:
Configuration D
Configuration A
Configuration B
Configuration E
© A. Kwasinski, 2015
Configuration C
Configuration F
Microgrids
Accounting for power source interfaces
• Application of minimal cut sets. Case A:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Application of minimal cut sets. Case B:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Application of minimal cut sets. Case C:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Application of minimal cut sets. Case D:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Case E:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Application of minimal cut sets. Case F:
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Observations:
• Configuration A (center converter) is an order of magnitude worst
than the others (about 5-nines for Configuration A vs. 6-nines for all
other configurations).
• Source diversity is essential in order to achieve ultra-high availability
(Availability equals 0.85 if fuel cells with no redundancy are used, 0.96
if microturbines with no redundancy are used, 0.99 if fuel cells with
redundancy are used, or 0.9994 if microturbines with redundancy are
used).
• Availability drops by about 1 nine without redundancy.
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Other examples:
• Load fed through converters by a microturbine fueled by natural gas (represented in Fig. A)
• Case 2: Same as Case 1 but with two microturbines in parallel so each of them can power the load
alone.
• Case 3: Load fed through converters by an engine generator fueled by diesel delivered by truck and
stored in a local tank.
• Case 4: Same as Case 3 but with two engine generators in parallel so each of them can power the load
alone.
• Case 5: Two power paths to the load; one is as indicated by Case 1 and the other by Case 3. Each path
can power the load alone (represented in Fig. B).
• Case 6: Combined PV and energy storage powering the load through a converter.
• Case 7: Same as Case 6 but combining PV and wind.
• Case 8: Same as Case 5 but with the diesel generator path replaced by the path indicated in Case 6.
• Case 9: Same as Case 5 but with the diesel generator path replaced by the path indicated in Case 7.
Figure B
Figure A
© A. Kwasinski, 2015
Microgrids
Accounting for power source interfaces
• Results:
Note: in cases 6 to 9: subcase “a” considers that there is sufficient energy storage to yield an
availability of 5-nines at the output of the renewable energy source and sub-case “b” considers that
their output availability is 2-nines.
© A. Kwasinski, 2015
Power Distribution Architecture
• Underground (buried) infrastructure.
• Issues:
• It is not effective for earthquakes
• With storms, it has lower failure probability but longer repair times.
• Very costly (e.g. > 5x more than overhead).
• A storm may never end up happening, but “normal” cable failures will
surely happen (and will take longer
to repair than with overhead lines).
Christchurch, NZ
145
Power Distribution Architecture
• Redundant/geographically dispersed power paths
• Issues with conventional protection devices :
• Reliability (particularly with dc)
• Selectivity planning
• Series (high-impedance) fault detection, particularly with power
electronics circuits
Increasing resiliency
Circuit breaker availability model
 (    )
1  ACBC  C CB CB C
C ( C  CB )
146
Power Distribution Architecture
• Power electronic circuits realizing active power distribution
nodes (APDN) (i.e. a set that includes solid state transformers).
• Concept: Place power electronic circuits in key system nodes (could be
integrated as part of a source or load interface).
• APDNs can independently control
power flow in its input and output
ports and may include energy
storage (e.g. batteries).
147
Power Distribution Architecture
• Case studies:
Circuit Breaker
Circuit Breaker
Power Converter
Power Converter
Power Converter +Storage
Power Converter
+Storage
Slide, courtesy of Dr. Myungchin Kim
148
Power Distribution Architecture
• Unavailability comparison:
Ideal C/B
• Effect of Higher Battery Backup Time
: C vs E- Identical Interface with only storage added
• Circuit Breaker & Power Electronics + Storage
 (B vs C) , (B vs D)
 Higher Part Count can be overcome by storage with longer autonomy
Slide, courtesy of Dr. Myungchin Kim
149
Power Distribution Architecture
• Simplified Cost Analysis among different connection options
(Overall Cost)= (Materials Cost) + (System Lifetime Downtime Cost)
where RL,i is the local resilience for case “i,” TL,i is the expected life time for case
“i” (i.e. resilience is assessed over the entire life time) and cDT is the down time
cost per unit time.
• Notes: Overall cost is different from lifetime cost. Lifetime cost also includes
other factors, such as operation and maintenance cost. Since resilience is
evaluated over the entire life time, it is implicitly considering the probabilities
associated with observing a given extreme event.
© Alexis Kwasinski, 2015
150
Power Distribution Architecture
Overall Cost (USD)
Downtime Cost (USD)
Overall Cost (USD)
• Simplified Cost Analysis among different connection options
• Downtime cost acts as a leverage which could offset higher materials cost.
• Hence, overall cost of using active power distribution nodes with embedded
energy storage could be lower than overall cost of a circuit breaker, particularly
if underground cables are considered for the case with circuit breakers and
overhead lines are considered for the case with APDNs (and equal life times).
Materials Cost (USD)
Downtime Cost (USD)
Configuration with
high downtime cost
and low materials
cost
Configuration with
low downtime cost
and high materials
cost
Downtime Cost (USD)
© Alexis Kwasinski, 2015
151
AC vs. DC
• Design needs favoring dc over ac power architectures:
• Inclusion of energy storage
• Power source technology diversification and redundant
source arrangements imply the need for simple paralleling.
• Complex power distribution architectures.
• Increased use of power electronic circuits, even at the
distribution level.
• Preferred use of renewable energy sources
• Other advantages of dc over ac: simpler control, more flexible
power distribution architectures, and potential for higher
efficiency. Also, most modern loads are inherently dc.
© Alexis Kwasinski, 2015
152
Applications
 Sustainable wireless area
 Residential photovoltaic systems
© A. Kwasinski, 2015
153
Application
Wireless communication networks
 Issues when integrating renewable energy in cell sites:
 Power generation footprint >> load footprint


Photovoltaic modules footprint = about 200 W/m2
Base station footprint = a few kW/m2
 Variable output of renewable energy sources
 Solutions to these issues for increased resilience:
 Source diversification
 Use of locally stored energy (e.g. in batteries). This is the
role of energy storage in microgrids for increased use of
PV systems in wireless communication networks.
© A. Kwasinski, 2015
154
Application
Sustainable Wireless Area (SWA)
 Concept:
 Sustainable wireless areas (SWAs) are dc microgrids created
by interconnecting a few (e.g. 7) base stations.
 Renewable energy sources are
placed in base stations or nearby
locations where there is sufficient
space.
 Resources (generation and energy
storage) are shared among all base
stations in the SWA.
 traffic and electric energy
management is integrated. I.e., traffic
is regulated (or shaped) based on local
energy resources availability and forecast.
© A. Kwasinski, 2015
155
Application
Sustainable Wireless Area (SWA)
 Potential implementation in urban areas
BS
BS
PV
© A. Kwasinski, 2015
156
Application
Sustainable Wireless Area (SWA)-Control structure
• Hierarchical structure
• Top Level:
• Optimizes SWA operation by coordinating all power sources, energy storage
devices and loads.
• Bottom Level:
• Local autonomous controller in charge of regulating local traffic and power
generation based on top-level commands.
• If the top level controller fails, this controller can maintain local operation but
at a
suboptimal level
(i.e. like in a conventional
system)
157
Application
Sustainable Wireless Area (SWA)-Control structure
• Central controller:
• Function: optimize power generation, energy storage level (state of
charge) and load in an integrated way
• A base station nominal load depends on traffic that, in turn, relates to
an utilization factor ν:
• Traffic shaping can be implemented through a parameter σ
• The central controller adjusts σ in order to optimize power generation
and energy balance within the SWA.
© Alexis Kwasinski, 2015
158
Application
Sustainable Wireless Area (SWA)
• Availability can be improved without
additional energy storage by
modifying the transition probabilities.
• Transition probabilities can be
modified by controlling the load (e.g.
managing traffic) based on batteries
state of charge or based on the
present or expected future condition
of the local power generators
(including PV arrays).
159
Application
Residential PV systems
• Not for IEEE1547 compliant inverters.
• Application is not in these relatively few cases:
© A. Kwasinski, 2015
160
Application
Residential PV systems
• Application is for these far many cases
© A. Kwasinski, 2015
161
Application
Residential PV systems
• And even these cases…..
© A. Kwasinski, 2015
162
Application
Residential PV systems
• General architecture
(it may scale up):
• Critical points:
• PV (grid isolated)
• Other generators
• EV charging
• Communications
163
Application
Residential PV systems
The value of the load: A note about smart grids and smart loads
•Evolving loads: more intelligence and more controllable. More
residential critical loads requiring local power (e.g. phones – wifi calling
and electric vehicles) may increase for residential users the value of
alternative power during disasters.
• Load management objectives:
• Improved efficiency
• Improved resilience
• Load prioritization
• Electric vehicles duality
• A critical load
• A valuable resource
© Alexis Kwasinski, 2015
164
Application
Residential PV systems
• PV modules survival to hurricanes (e.g. Ike)
© A. Kwasinski, 2015
165
Application
Microgrids in 2011 Japan’s earthquake and tsunami
• Natural disasters are not single events. They are complex events with 4
distinct phases: pre-disaster, during the disaster, immediate aftermath and
long-term aftermath (when power generation, transmission and/or
distribution capacity may be lacking).
Onagawa nuclear power plant: Offline since
the earthquake. Currently, almost all
nuclear power plants in Japan are offline
• As a result of Fukushima #1 Nuclear Power plant event electric power
utilization in Japan has been affected, particularly during the summer when
rotating outages are likely.
• Microgrids may also help to limit these and other
effects during the long term aftermath.
© Alexis Kwasinski, 2015
166
Application
Microgrids in 2011 Japan’s earthquake and tsunami
• NTT/NEDO Microgrid in Sendai: This microgrid was designed to provide
different power quality levels to part of an university campus, including a clinic.
Operational on 3/11/11
© Alexis Kwasinski, 2015
167
Application
Microgrids in 2011 Japan’s earthquake and tsunami
NTT/NEDO Microgrid in Sendai
2x350 kW
50 kW
• 1) Earthquake happens. Natural gas generators fail to start.
• 2) Manual disconnection of all operating circuits except the dc one
• 3) and 4) Natural gas generators are brought back into service by
maintenance personnel. A few minutes later, the circuits that were
intentionally disconnected are powered again.
• 5) Power supply from the main grid is restored.
© Alexis Kwasinski, 2015
168
Application
Microgrids in 2011 Japan’s earthquake and tsunami
NTT/NEDO Microgrid in Sendai
• Natural gas infrastructure in Sendai: contrary to most of the city that relied
on natural gas supply from a damaged facility in the port, the microgrid
natural gas was stored inland and was not damaged.
© Alexis Kwasinski, 2013
169
Application
Microgrids in 2011 Japan’s earthquake and tsunami
NTT/NEDO Microgrid in Sendai. Key Lessons
• Local energy storage in batteries were a key asset to keep at least the most
critical circuit operating.
• PV power only played a complementary role.
• Natural gas supply did not fail thanks to an almost exclusive design for the
distribution pipelines for this site.
• Flexible remote operation is very important during extreme events
conditions.
• Connection of generators or their components through ac buses seem to
increase failure to start probability.
• Source diversification is important.
© Alexis Kwasinski, 2015
170
Application
Microgrids in the United States
• Originally, in the U.S. microgrids have implemented with the goal of
reducing peak load, improving efficiency, etc. but not necessarily with the
goal of improving availability (resilience).
• These environmental-related goals still serve to improve resilience of power
grids to droughts, heat waves and other similar disasters.
• These microgrids have performed well during Superstorm Sandy (e.g.
Verizon’s Garden City Central Office).
• More recently, the U.S. DOD has been implementing microgrids for
improved resilience in military bases. In order to reduce vulnerabilities
associated to logistics, designs with reduced dependencies are desired.
© Alexis Kwasinski, 2015
171
Conclusion
 Power grids are fragile systems with an original design that
limits improvements in their resiliency performance.
 Microgrids may provide a better alternative for enhanced
resiliency but only if they are well design:
 Power generation sources diversification.
 Use of energy storage (reduced capacity compared to standby
power plants if diverse sources are used through a
combination of renewable and non-renewable sources).
 Advanced power distribution architectures, likely requiring
the use of APDNs.
 Requirements for enhanced resiliency favors dc over ac
microgrids. Additionally most modern loads are dc.
© A. Kwasinski, 2015
172
Alexis Kwasinski (akwasins@pitt.edu)
Practicum
© A. Kwasinski, 2015
174
Practicum
Task #1
 Agree on a definition of energy security and of electric
energy security.
© A. Kwasinski, 2015
175
Practicum
Task #2 – Case study part
 Consider the following case:
 You need to decide a power supply option for a critical facility
which has a downtime cost of $30K/hour. The expected
lifetime is 10 years. The potential hazard is a cat. 2 hurricane
with a probability of occurrence of 0.31 for those 10 years
(more likely to happen in the year 5.82). In normal conditions
grid availability is 3-nines. The microgrid availability is 5nines regardless of the conditions. Power supply options:


Simple grid tie for $200K
Microgrid for $2.4M
 Decide a power supply option
© A. Kwasinski, 2015
176
Practicum
Task #2 – Case study part
 Calculations:
 Cost option 1:
0.2+(168*0.03+(87600-168)*(1-0.999)*0.03)*0.31+
+(1-0.31)*(0.03*(1-0.999)*87600)=
=0.2+(5.04+2.62)*0.31+2.63*0.69=0.2 +2.37+1.81=
=$4.39M
 Cost option 2:
2.4+(1-0.99999)*(87600)*0.03=$2.42M
© A. Kwasinski, 2015
177
Practicum
Conclusion
 Discussion
 What is the most significant cost component besides




equipment costs?
What is the second most significant component?
How can this component be reduced? Who pays for this fix?
How do these two options compare considering only the
disruptive event.
Are there other potential solutions? Who pays for this
solution?
© A. Kwasinski, 2015
178
Practicum
Conclusion
 Ringo Bonavena a boxer from the 70s once said “Experience
is a comb that life gives you after becoming bald.”
 Likewise
 Knowledge about events is a comb that planners receive after
becoming bald.
© A. Kwasinski, 2015
179