This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstra?on under grant agreement no 608166. The contents of this presenta?on are the author's views. The European Union is not liable for any use that may be made of the informa?on contained therein. RAIN Risk Analysis of Infrastructure Networks in Response to Extreme Weather RAIN Workshop Climate Change & Weather Modelling 9 th November 2015 Dr. Alan O’Connor Dept. of Civil Engineering, Trinity College Dublin [email protected]www.rainproject.eu
Transcript
This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstra?on under grant agreement no 608166. The contents of this presenta?on are the author's views. The European Union is not liable for any use that may be made of the informa?on contained therein.
RAIN -‐ Risk Analysis of Infrastructure Networks in
Response to Extreme Weather RAIN Workshop Climate Change & Weather Modelling 9th November 2015
Dr. Alan O’Connor Dept. of Civil Engineering, Trinity College Dublin [email protected] www.rain-‐project.eu
Introduc?on • Project RAIN – Risk Analysis of INfrastructure Networks in Response to
Extreme Weather • FP7 Theme 10 -‐ Security Ac?vity -‐ 10.2 Security of Infrastructures and
U?li?es SEC-‐2013.2.1-‐2 -‐ Impact of Extreme Weather on Cri?cal Infrastructure'. 'Ac;vi;es will concentrate on targets of an incident or disaster of transna*onal importance…, significant sites of poli;cal or symbolic value and u;li;es being those for energy (including oil, electricity, gas), water, transport (including air, sea, land), communica;on (including broadcas;ng), financial, administra;ve, public health, etc. A series of capabili;es are required to cope with this mission area, many of which primarily relate to the phases "protect" but also "prepare”.
The ambi;on is both to avoid an incident and to mi*gate its poten;al consequences. emphasis will be on issues such as: analysing, modelling and assessing vulnerabili*es of physical infrastructure and its opera;ons; securing exis*ng and future public and private cri*cal networked infrastructures, systems and services with respect to their physical, logical and func;onal side; control and alert systems to allow for quick response in case of an incident; protec*on against cascading effects of an incident, defining and designing criteria to build new secure infrastructures and u*li*es.'
To develop a systema(c risk management framework that explicitly considers the impacts of extreme weather events on cri?cal infrastructure and develops a series of mi(ga(on tools to enhance the security of the pan-‐European infrastructure network. To do this we must: – quan(fy the complex interac(ons between weather events and land
based infrastructure systems (i.e. transport, telecoms, energy etc.), – develop an opera?onal analysis framework that considers the impact of
individual hazards and the coupled interdependencies of cri?cal infrastructure through robust risk and uncertainty modelling,
3. Literature and knowledge base for weather impacts on the transport system
34
block roads by felling trees and affecting vehicle behaviour, increasing the probability of accidents. Winds approaching tornado force may also directly damage vehicles, bridges and equipment, as does hail in extreme cases. Changes in temperature affect materials: frozen materials become brittle, and temperature-based expansion and shrink-ing can loosen stones in cuttings, causing them to fall across and block roads.
Weather phenomena and their impacts and consequences form a complex net where many different phenomena, alone or together, can have the same impact and result in a whole range of consequences. In Figure 6, for example, wind causes the sea level to rise. There are also many different phenomena that may lead to flooding also, all with different probabilities. The consequences of flooding are many – some more severe than others, some more likely to happen than others, and some rapid and some having effects over a longer period of time.
TRA
NSP
OR
T I
NFR
AST
RU
CTU
RE/
VEH
ICLE
LO
CA
L N
ATU
RE
W
EATH
ER
Figure 5. Conceptual map of weather phenomena impacts on road infrastructure and transport (source: Makkonen, Törnqvist and Kuusela-Lahtinen, VTT).
interactions between weather events and land based transportation systems (source: Makkonen, Törnqvist and Kuusela-Lahtinen, VTT)
To develop a systema(c risk management framework that explicitly considers the impacts of extreme weather events on cri?cal infrastructure and develops a series of mi(ga(on tools to enhance the security of the pan-‐European infrastructure network. To do this we must: – quan(fy the complex interac(ons between weather events and land
based infrastructure systems (i.e. transport, telecoms, energy etc.), – develop an opera?onal analysis framework that considers the impact of
individual hazards and the coupled interdependencies of cri?cal infrastructure through robust risk and uncertainty modelling,
– considering cascading hazards, cascading effects and ?me dependent vulnerability
– Develop Technical and Logis(c solu(ons to minimise the impact of these extreme events, include novel early warning systems, decision support tools and engineering solu?ons to ensure rapid reinstatement of the infrastructure network
Hurricane and Severe Storm damage (2012) -‐ Hurricane Sandy -‐ largest Atlan?c storm on record devastated por?ons of the Caribbean, Mid-‐Atlan?c and North Eastern United States – 193 people were killed along the path of the storm in
seven countries – damage >$20 billion(USD) losses (including business
interrup?on) > $50 billion – Jamaica: winds lee 70% of residents without electricity,
blew roofs off buildings, killed one, and caused about $55.23 million in damage.
5
In recent years, the complex interdependencies of the European infrastructure networks have been highlighted through multiple failures during extreme weather events. These failures have been the driver for this project concept and the aim of RAIN is to better predict the sensitivity of European in-frastructure to such widespread disruption so that mitigation measures can reduce the impact of possi-ble future events. Examples of such failures include:
New York Hurricane and Severe Storm damage (2012): The largest Atlantic storm on record, hur-ricane Sandy, devastated portions of the Caribbean, Mid-Atlantic and North Eastern United States in late October 2012. At the time of preparing the RAIN project, Sandy was estimated to have caused damage of at least $20 billion USD and further estimates of losses (including business interruption) surpassed $50 billion. At least 193 people were killed along the path of the storm in seven countries. In Jamaica, winds left 70% of residents without electricity, blew roofs off buildings, killed one, and caused about $55.23 million in damage. In Haiti, Sandy's outer bands brought flooding that killed at least 54, caused food shortages, and left about 200,000 homeless. In the Dominican Republic, two died. In Puerto Rico, one man was swept away by a swollen river. In Cuba, there was extensive coastal flooding and wind damage inland, destroying some 15,000 homes, killing 11, and causing $2 billion in damage. In the United States, Hurricane Sandy affected at least 24 states, from Florida to Maine and west to Michigan and Wisconsin, with particularly severe damage in New Jersey and New York. Its storm surge hit New York City on October 29, flooding streets, tunnels and subway lines, Figure 1.1(a), and cutting power in and around the city. New York governor Andrew Cuomo called in the National Guard members to help the state. Fi-nancial services were severely affected, with the New York Stock Exchange remaining closed for trading for two days, the first weather closure of the exchange since 1985. The East River over-flowed its banks, flooding large sections of Lower Manhattan. Battery Park had a water surge of ap-proximately 4.0m. Seven subway tunnels under the East River were flooded, which the Metropolitan Transportation Authority stated early on October 30 “that the destruction caused by the storm was the worst disaster in the 108-year history of the New York City subway system”. Gas shortages throughout the region led to an effort by the U.S. federal government to bring in gasoline and set up mobile truck distribution at which people could receive up to 10 gallons of gas, free of charge. This caused queues of up to 20 blocks long and was quickly suspended. The full extent of the infrastructure damage may not be quantified for some time yet but it is clear from preliminary reports that the hurricane had a devastating impact on a wide vari-ety of infrastructure networks.
Flood damage: In September 2012, twelve people died and hundreds had to be evacuated as a result of the flash floods in Andalucía and Murcia in the south of Spain. This single event caused by torren-tial rain resulted in damage to infrastructure networks and building alike, including the collapse of two bridges on two motorways. The cascading effects and interdependencies of these infrastructure failures are yet to be determined.
In the period from 30th July to 8th August 2010, Finland was hit with severe storms, particularly 'downbursts' following an unusually period of high temperatures. Falling trees cut off roads, destroyed buildings and caused devastation to property. The water and electricity networks were cut off in wide
Figure 1.1(a): Flooding in New York Subway following Hurricane Sandy (30th October 2012)
In recent years, the complex interdependencies of the European/Interna?onal infrastructure networks have been highlighted through mul?ple failures during extreme weather events. These failures have been the driver for this project, e.g.
• Hai?: Sandy's outer bands brought flooding that killed at least 54, caused food shortages, and lee about 200,000 homeless,
• Cuba: extensive coastal flooding and wind damage inland, destroying some 15,000 homes, killing 11, and causing $2 billion in damage,
• United States: affected at least 24 states, from Florida to Maine and west to Michigan and Wisconsin, with par?cularly severe damage in New Jersey and New York.
– Its storm surge hit New York City on October 29, flooding streets, tunnels and subway lines and cuLng power in and around the city.
– New York Stock Exchange remaining closed for trading for two days. – 7 subway tunnels under the East River were flooded, which the Metropolitan
Transporta?on Authority stated early on October 30 “that the destruc;on caused by the storm was the worst disaster in the 108-‐year history of the New York City subway system”.
Flash Floods Central Europe, 2002 • deaths of approximately 150 people and an es?mated
€150 Billion worth of damage, • In Germany and the Czech Republic, the worst affected
areas – e l e c t r i c i t y f a i l u r e s , d i s c o n n e c t e d
telecommunica(on links, damage to approximately 250 roads and 256 bridge structures,
– disrup?on to the Gas service due to damaged pipelines and contamina(on of clean water with flood water.
– restora?on of important services to full capacity took approximately 1 month for electricity, 2 months for Gas and 3 months for telephone communica?ons.
Flash Floods Finland August 2010 • Finland was hit with severe storms, par?cularly
'downbursts' following an unusually period of high temperatures. Falling trees cut off roads, destroyed buildings and caused devasta?on to property,
• The water and electricity networks were cut off in wide areas in Central and South-‐East Finland,
• Forest damage represented some 8.1 million m3 and 240, 000 hectares.
• 35,000 kilometres of the electricity network was destroyed or damaged. 9,000 distribu?on substa?ons were lee without electricity, leaving 480,000 households in the dark. Repair work to the damaged networks amounted to nearly 200,000 hours (over 120 man-‐years).
Underpass was flooded by rainwater in Oulu on Sunday.Photo Str / Lehtikuva
Mon, 01 Jul, 2013 12:00:24 AMFluctuating weather to continue this week
Downpour in Oulu disrupts powersupply, vehicular movementsFTimes-STT Report, July 1
The Finnish meteorological department has predicted the unusual weather is likely tocontinue in the first week of July along with scorching heat, rainfalls and thunderstorm insome parts of the Nordic country. Finland has been experiencing a fluctuating weather patter ever since the summer beganthis year. Normal activities in the northern city of Oulu have been disrupted as heavy rain causedpower cuts, water logging-triggered traffic tailback in many part of the city, residents andpolice said. The downpour flooded most of the underpasses and roads when number of vehicles wentinoperative due to the water in the city, they said adding that the ground floor of manyhomes were submerged by the rain eater. Power supply to some 1,600 household remained snapped for about two hours Sunday.
Like 0
Find us on Facebook
Finland Times
Like
635 people like Finland Times.
Facebook social plugin
Downpour in Oulu disrupts power supply, vehicular movements
Risk Analysis of Infrastructure Networks must therefore: 1. quan(fy the complex interac(ons of the exis?ng infrastructure systems
and their interrelated damage poten?al in the event of specific extreme weather events,
The degree of a node is the number of edges the node is con-nected to, it has further insights on the connectivity properties ofthe network. For the power network most of the nodes have fairlylow degree, 67% of the nodes degree is less than or equal to 2, nev-ertheless, there are some high degree node. In addition, it has high-er average node degree with the gas pipeline system. Characteristicpath length is the average shortest path length of pairs of nodes dij,which is defined as hdi ¼ 2
NðN#1ÞP
i;j2V ;i–jdij, the characteristic pathlength of the power and gas networks are 8.247 and 4.313, namely,much differences in initial efficiency between the power grid andthe gas pipeline systems. Another metric is the average clusteringcoefficient, the clustering property of node i is defined as ci ¼ 2jEi j
kiðki#1Þand the clustering coefficient of the network is defined asC ¼ 1
N
PNi¼1ci, where ki is the degree of the node i, and Ei is the set
of the edges of the subgraph constructed by the neighboring nodesof node i. |Ei| represents the number of components in the set of Ei.It is applied to describe the connectivity of the network at a locallevel. As shown from the table, it is as small as 0.047 for the powernetwork, nevertheless, it is bigger for the gas pipeline system, indi-cating a better connection of its neighboring nodes. Suppose thatpower is routed through the shortest path, the betweenness of anode is a substitution for how much power it is transmitting. Thebetweenness centrality of node l is defined as bl ¼
Pi;j2V ;i–j
nijðlÞnij
,where nij is the number of the shortest paths between node i andnode j, nij(l) is the number of shortest paths between node i andnode j passing the node l. Average values 797 give the average fre-quency of shortest path passing a node.
2.3. Functional characteristic of the infrastructure systems
Then based on operation mechanism, functional characteristicto each of the infrastructure systems should be further considered.For the power system, inspired by the model proposed by Wangand Chen (2008) and Wang and Rong (2011), a load model whichcontains both node and edge informations is adopted here. It is as-sumed that the initial load of edge eij is defined as Wij = (Di % Dj)h,where Di and Dj are the degrees of nodes i and j, h is an adjustableparameter which controls the strength of the initial load of theedge. When edge eij is damaged, the load of the broken edge willbe redistributed to the neighboring edges connecting to node iand node j (see Fig. 3). The additional load received by edge eim
is defined as follows
DFim ¼ Fij %Wim
Pa2Ci
Wia þP
b2CjWbj
! " ð1Þ
where Ci and Cj are the sets of neighboring nodes of i and j. It is as-sumed that the capacity Cij of edge eij is proportional to its initialload. Namely, Cij = (1 + b)Wij, where the constant b is the toleranceparameter. Meanwhile the load of a node is defined as the between-ness of the node (Motter et al., 2002). The node capacity is the max-imum load that the node can handle. It is also assumed that thecapacity Cj of node j is proportional to its initial load Lj. Cj = (1 + a)Lj
where the constant a is the tolerance parameter.
The gas pipeline system includes gathering system, transporta-tion system, distribution system and pipelines linking them. Thispaper assumes that transport occurs along the shortest paths inthe gas pipeline system. A generalized betweenness centralitymodel proposed by Carvalho et al. (2009) is modified and utilizedhere for modeling gas pipeline system. Consider the gas pipelinenetwork GG = (VG, EG) with nodes set VG and edges set EG. Let TK,L
be the flow from source subgraph (VK, EK) to sink subgraph (VL, EL).Generalized betweenness centrality of eij e EG is defined as follows
Gij ¼X
s 2 Vk
t 2 VL
TK;L
jVK jjVLjrs;tðeijÞ
rs;tð2Þ
where eij e EG, rs,t is the number of shortest paths from node s tonode t and rs;tðeijÞ is the number of these paths passing through linkeij. Denote bg
j be the decision variable designating the flow of node j.Mathematical formulation is given as follows
bgj ¼
X
i
Gij #X
m
Gjm ð3Þ
2.4. Modeling of interdependency
With view to the concept of interdependency, we know infra-structure does not exist in isolation, especially with the develop-ment of scientific technology and social economy, manyinfrastructure systems life become more and more interconnectedand interdependent. This paper considers infrastructure interde-pendency as a bidirectional relationship, co-located as well as mu-tual interdependency are considered here. Physical components ofthe power and gas pipeline systems are situated within the samegeographical region. Activities of one of the systems are dependentupon the activities of the other infrastructure system. For example,gas transmission nodes depend on power supply to ensure theirnormal operation while some gas-based generators are driven bygas for fuels. Define P and G as the power and gas pipeline systems.Let i e I(G, P) denotes the node of the gas pipeline system which re-ceives the power input, j e O(P, G) denotes the power node whichprovides supply to the gas pipeline system. A simple mathematicalrepresentation of interdependency d1(i, j) for the load of node i independence of load of node j is given as bellows
bgi ¼
bgi bp
j –0
0 bpj ¼ 0
(ð4Þ
where bgi is the load of node i in G while bp
j is the load of node j in Pwhich links to i. Let s e I(P, G) denotes node in the power systemwhich depends on the gas pipeline system, t e O(G, P) denotes nodes
Table 1General properties of the power and gas pipeline systems.
N, number of nodes, E, number of edges, G, number of generators, hki, the averagedegree, C, the average clustering coefficient, hdi , characteristic path length, B, nodesaverage betweenness.
Fig. 3. The schematic diagram of load redistribution after the breakdown of edge.
S. Wang et al. / Safety Science 51 (2013) 328–337 331
1. Introduction
9
Snow storms
Storm winds
Thunderstorms Traffic controlsystems
Power supplysystems
Switches
Electricity shocks
Power failures
Falling trees Line cuts
Stacking snow
Lightning strikes
Frozen switches
Phenomena Impact ConsequencesSystem
failure
Time delays
Accidents
Customer dissatisfaction
Disturbances are operations
Increased maintenance / repair costs
Figure 1. Simplified example of harmful weather impacts and consequences to rail transport.
Figure 2 shows how the causal map can be used to assess risk of consequences of dif-ferent weather phenomena and impacts. The causal relationships contain many uncer-tainties along the chain, and to credibly assess true risk and vulnerability levels such uncertainties must be taken into account.
Figure 2. Linking phenomena, impacts and consequences in the EWENT context.
Risk Analysis of Infrastructure Networks must therefore: 3. develop systems that will accelerate re-‐establishing infrastructure links
post an extreme event. This objec?ve will require developing engineering solu?ons to assess the safety of land based structures, networks and substructures which facilitate rapid replacement of key infrastructure and emergency systems where necessary. Addi?onally we must seek to op(mise the use of the in-‐tact network post any significant event.
Risk Analysis of Infrastructure Networks must therefore accept that extreme events (e.g. “100 year events”) are happening with alarming frequency. Thus preparing for these events is of vital importance. We must arempt to quan?fy and thereby reduce uncertainty and gain a berer understanding of how our cri?cal infrastructures will cope and adapt to weather events to help ensure the security of vital u?li?es. Requires interac?on between several en??es i.e. emergency planners, u?lity operators, first responders, engineers and most importantly the ci?zens living in the area of the extreme event.
entities such as; emergency planners, utility operators, first responders, engineers and most impor-tantly the citizens living in the area of the extreme event. Given the diversity of those involved in such an event – the answers to improving the outcomes of such events cannot be considered in isola-tion by any one discipline. The main underlining ethos of the RAIN project is that the solution is one that MUST be achieved within an interdisciplinary team.
The RAIN consortium brings together experts from transportation, energy, risk assessment, climate prediction, social sciences, engineering and telecommunications with the goal of predicting how ex-treme weather events will impact upon critical European infrastructure networks collectively, Figure 1.1(b). The consortium also draws from a cross-section of research institutes, universities, small and large companies and utility providers, all of whom are engaged in the delivery of improved research methods and standards for critical infrastructure provision.
Figure 1.1(b): RAIN Concept
One of the key components of the RAIN project will be to consider the citizen. The citizen is the most important consideration in an extreme event. The RAIN approach puts the societal impacts of infra-structure failures in extreme weather events at the heart of the approach and develops the risk mitiga-tion strategies to minimise the risks of loss of life and disruption to quality of life.
The RAIN approach will minimise the risk of chaos in extreme weather events by predicting, using the most advanced statistical methods, how both weather patterns are likely to emerge and then how our infrastructures will react under these events. The RAIN approach will show how reducing uncer-tainty and considering the impacts of society can yield significant economic, social and humanitarian benefits.
The RAIN consortium are committed to cooperating with other projects and related actions under the ENV theme to ensure optimal resource allocation and dissemination of results of activities.
B.1.1.4 RAIN S&T objectives The objectives and tasks of the RAIN project are well aligned with the call text and address all of the topics specifically mentioned in the call. This framework will give support to optimising the risk as-sessment process, quantifying weather related hazards and reducing system vulnerability, with a view to developing a more resilient European infrastructure network. A comparison between the call text and the project objectives are shown in Table 1.1(a).
Risk Analysis of Infrastructure Networks must: 1. consider the ci(zen as central. The ci?zen is the most important considera?on in
an extreme event. We must put the societal impacts of infrastructure failures in extreme weather events at the heart of any approach and develop risk mi?ga?on strategies to minimise the risks of loss of life and disrup?on to quality of life.
2. minimise the risk of chaos in extreme weather events by predic?ng, using the most advanced sta?s?cal methods, how weather paXerns are likely to emerge and then how our infrastructures will react under these events.
B.1.3.1 Overall strategy of the work plan The research activities are organised around six technical work packages, each organised around a core topic necessary for the complete description of the problem dynamic. One of these, WP2, focuses on hazard identification, while the second and third work packages, WP3 and WP4 respectively, focus on the elements of critical infrastructure addressed in RAIN. Development of the Risk Analysis framework, identifying measurable risks and benefits, and developing mitigation strategies are cov-ered in WP5, WP6 and WP7. Management and dissemination activities complete the work plan, as indicated in Figure 1.3(a). The diagram also serves to show the interaction and interdependencies be-tween WP’s.
Figure 1.3(a): Work Plan Strategy & Methodological Diagram
B.1.3.2 Timing of the work plans and their components The timings of each task and the milestones are identified in Figure 1.3(b). It is proposed that the RAIN project will commence in October 2013 and will run for 36 months. The majority of the techni-cal work packages will be completed by Month 24 (WP2 to WP5). While the website and dissemina-tion and communication plan will both be completed at an early stage of the project, they will be sub-jected to reviews on a regular basis.
– quan(fy the complex interac(ons between weather events and land
based infrastructure systems (i.e. transport, telecoms, energy etc.), – develop an opera?onal analysis framework that considers the impact of
individual hazards and the coupled interdependencies of cri?cal infrastructure through robust risk and uncertainty modelling,
– considering cascading hazards, cascading effects and ?me dependent vulnerability,
– develop Technical and Logis(c solu(ons to minimise the impact of these extreme events, include novel early warning systems, decision support tools and engineering solu?ons to ensure rapid reinstatement of the infrastructure network.
Risk Analysis of Infrastructure Networks must therefore:
22
B.1.3 S/T methodology and associated work plan
B.1.3.1 Overall strategy of the work plan The research activities are organised around six technical work packages, each organised around a core topic necessary for the complete description of the problem dynamic. One of these, WP2, focuses on hazard identification, while the second and third work packages, WP3 and WP4 respectively, focus on the elements of critical infrastructure addressed in RAIN. Development of the Risk Analysis framework, identifying measurable risks and benefits, and developing mitigation strategies are cov-ered in WP5, WP6 and WP7. Management and dissemination activities complete the work plan, as indicated in Figure 1.3(a). The diagram also serves to show the interaction and interdependencies be-tween WP’s.
Figure 1.3(a): Work Plan Strategy & Methodological Diagram
B.1.3.2 Timing of the work plans and their components The timings of each task and the milestones are identified in Figure 1.3(b). It is proposed that the RAIN project will commence in October 2013 and will run for 36 months. The majority of the techni-cal work packages will be completed by Month 24 (WP2 to WP5). While the website and dissemina-tion and communication plan will both be completed at an early stage of the project, they will be sub-jected to reviews on a regular basis.
