Database and Model for Dynamic scenario assessment V2

Transcription

Database and Model for Dynamic
scenario assessment V2
ADAI: Miguel Almeida, Luís Mário Ribeiro,
Domingos Viegas
AMRA: Alexander Garcia-Aristizabal, Giulio
Zuccaro, Maria Polese, Stefano Nardone, Marco
Marcolini
AEE: Marianne Grisel, Christophe Coulet
FMI: Karoliina Pilli-Sihvola
31.8.2014
31.08.2014 | i
The research leading to these results has received funding from the European Community's
Seventh Framework Programme FP7/2007-2013 under grant agreement no. 284552 "CRISMA“
Deliverable No.
Subproject No.
SP4
Workpackage No.
42
Authors
Status (F = Final; D = Draft)
File Name
Dissemination level
D42.3
Subproject Title
Models for MultiSectorial
Consequences
Work package Title
Cascade Effects on
Crisis-Dependent
Space-Time Scales
ADAI: Miguel Almeida, Luís Mário Ribeiro,
Domingos Viegas
AMRA: Alexander Garcia-Aristizabal, Giulio
Zuccaro, Maria Polese, Stefano Nardone, Marco
Marcolini
AEE: Marianne Grisel, Christophe Coulet
FMI: Karoliina Pilli-Sihvola
F
CRISMA_D423_public
PU
(PU = Public; RE = Restricted; CO = Confidential)
Contact
[email protected]
[email protected]
Project
Keywords
Deliverable leader
Contractual Delivery
date to the EC
Actual Delivery date to
the EC
www.crismaproject.eu
Name:
Miguel Almeida, Domingos Viegas
Partner:
ADAI
Contact:
[email protected]; [email protected]
31.08.2014
31.08.2014
http://www.crismaproject.eu
31.08.2014 | ii
Disclaimer
The content of the publication herein is the sole responsibility of the publishers and it
does not necessarily represent the views expressed by the European Commission or its
services.
While the information contained in the documents is believed to be accurate, the
authors(s) or any other participant in the CRISMA consortium make no warranty of any
kind with regard to this material including, but not limited to the implied warranties of
merchantability and fitness for a particular purpose.
Neither the CRISMA Consortium nor any of its members, their officers, employees or
agents shall be responsible or liable in negligence or otherwise howsoever in respect of
any inaccuracy or omission herein.
Without derogating from the generality of the foregoing neither the CRISMA Consortium
nor any of its members, their officers, employees or agents shall be liable for any direct
or indirect or consequential loss or damage caused by or arising from any information
advice or inaccuracy or omission herein.
31.07.2014 | iii
Table of Contents
TABLE OF CONTENTS ................................................................................................................ III
LIST OF FIGURES ......................................................................................................................... V
LIST OF TABLES ......................................................................................................................... VII
GLOSSARY OF TERMS ............................................................................................................. VIII
ACRONYMS .................................................................................................................................. IX
EXECUTIVE SUMMARY ................................................................................................................ X
1.
INTRODUCTION ................................................................................................................... 1
1.1. Multi-hazards analysis .................................................................................................. 1
1.2. Cascade Event Chains Database ................................................................................. 1
1.3. Integration of cascade effects into the general framework of CRISMA .......................... 2
2.
SCENARIOS FOR QUANTITATIVE ANALYSIS ................................................................... 5
2.1. Triggering hazard: Extreme Weather Condition (Pilot A) ............................................... 5
2.2. Triggering hazard: Coastal Submersion (Pilot B) .......................................................... 7
2.3. Triggering hazard: Earthquake (Pilot D) ........................................................................ 8
3.
QUANTITATIVE ANALYSIS AND RESULTS ..................................................................... 12
3.1. Triggering hazard: Coastal Submersion (Pilot B) ........................................................ 12
3.1.1. Flood
Flood .................................................................................................. 12
3.1.2. Flood
Damage to Electricity Network
Release of Chemical Substance ..... 14
3.2. Triggering hazard: Earthquake (Pilot D) ...................................................................... 16
4.
3.2.1. Earthquake
Earthquake .............................................................................. 17
3.2.2. Earthquake
Damage to Electricity Network (Cable failure)
Forest Fire...... 22
CASCADE EFFECT MODEL IMPLEMENTATION .............................................................. 35
4.1. Logic of the sequences ............................................................................................... 35
4.2. Cascade effect model code......................................................................................... 36
5.
FINAL REMARKS ............................................................................................................... 38
6.
LIST OF REFERENCES ...................................................................................................... 39
APPENDIX (1) DESCRIPTION OF CHAIN BLOCKS IDENTIFIED FOR POSSIBLE CASCADE
EVENT CHAINS .................................................................................................................. 40
APPENDIX (2) MAIN PARAMETERS OF THE LABORATORIAL TESTS FOR THE
DETERMINATION OF IGNITION BY INDUSTRIAL ELECTRIC DISCHARGE.................... 48
31.07.2014 | iv
APPENDIX (3) ALGORITHM FOR THE MODEL IMPLEMENTATION ......................................... 51
APPENDIX (4) – EXAMPLE OF THE LOGIC SEQUENCE FOR THE PILOT D USE CASES ...... 54
31.07.2014 | v
List of Figures
Figure 1: Business logic of the CRISMA framework. ....................................................................... 2
Figure 2: Integration of the cascade effect model in the business logic of the CRISMA
framework. ...................................................................................................................................... 3
Figure 3: Diagram of cascade event chains identified for the occurrence of extreme weather
conditions........................................................................................................................................ 6
Figure 4: Diagram of cascade event chains identified for the occurrence of flood. .......................... 8
Figure 5: Diagram of the identified cascade event chains for Earthquake (from D42.2)................... 9
Figure 6: Photos of electric poles. ................................................................................................. 10
Figure 7: Images related to forest fires triggered by electric cable failure. a) Oliveira de Frades
(Portugal) in 28/02/2004; b) Schematic view of top of pole 39; red circle indicates where the
conductor failed (Plaintiff, 2013); c) Photograph of fire ignition area (Plaintiff, 2013). .................... 10
Figure 8: Cascade events chain for pilot B. ................................................................................... 14
Figure 9: Maximum water level in the area of Port-Neuf in La Rochelle during Xynthia. ................ 14
Figure 10: Transition matrix developed for pilot B. ........................................................................ 15
Figure 11: Workflow for the application of cascade effect model in Pilot D of CRISMA. ................ 17
Figure 12: Logical flow of information and data for the assessment of the earthquakeearthquake scenario. .................................................................................................................... 18
Figure 13: Building inventory in L’Aquila test case (Pilot D). The computation domain is divided
in a regular grid, and for each grid element the total number of buildings and the proportion of
different building classes are represented. .................................................................................... 18
Figure 14: Simulated seismic sequence located at the NW of L’Aquila city, Italy. The main shock
is a Mw 5.6 event. ......................................................................................................................... 19
Figure 15: Shake map of the triggering event, representing the spatial distribution of the Peak
Ground Acceleration (PGA) in %g values...................................................................................... 20
Figure 16: Probability of having collapsed buildings in the target area after the occurrence of the
main seismic event. ...................................................................................................................... 20
Figure 17: PGA values for different exceedance probability thresholds: (a) 1%; (b) 0.1%, and (c)
0.01%; and (d) resume of the result hazard curves. ...................................................................... 21
Figure 18: Probability of having collapsed buildings in the target for the earthquake-earthquake
scenario. The map represents the probability of collapsed buildings considering the potential
triggered seismicity according with the characteristics of the seismic sequence triggered by the
main seismic event. ...................................................................................................................... 22
Figure 19: Flow of information during the running of cascade effects for the selected scenario.
“Application 1” is referred to the transition EQ-DEN and “Application 2” is referred to the
transition DEN-FF. ........................................................................................................................ 23
Figure 20: Geographic distribution of the electric poles. ................................................................ 24
Figure 21: Map of the triggering earthquake intensity distribution with location of the electric
power network (distribution lines – smaller poles). ........................................................................ 24
31.07.2014 | vi
Figure 22: Fuel map for the region of L’Aquila............................................................................... 25
Figure 23: Sketch of a T&D system for an EPN (TL = Transmission Lines, D = Distribution lines,
TD [HV MV] = Transformation (from high to medium voltage) and Distribution station, TD
[MV LV] = Transformation (from medium to low voltage) and Distribution station, L = Load)
(adapted from D5.2 Deliverable of the EU project SYNER-G). ...................................................... 26
Figure 24: An example of tubular steel poles classification (from a commercial producer). ........... 26
Figure 25: Definition of top displacement XMAX. ............................................................................. 27
Figure 26 – Normalized pseudo-acceleration spectra.................................................................... 27
Figure 27: Representation of the electric poles and the mass on the top....................................... 28
Figure 28: IDA for the pole type 12B14. ........................................................................................ 28
Figure 29: Fragility curves of electric poles to a transversal force T1. ........................................... 29
Figure 30: Sequence of images of a laboratorial test to determine the energy required for
ignition by industrial electric discharge. ......................................................................................... 30
Figure 31: a) Plot of the Cumulative Distribution Function (CDF) of Log-normal, Gamma,
Normal, Weibull, and Exponential competing models, and the empirical CDF of the observed
data; b) CDF of the Log-normal (with parameters as those presented in Table 3), selected as
the best model describing the observations. ................................................................................. 31
Figure 32: Map for probability of cable failure................................................................................ 32
Figure 33: Map for probability of fire ignition due to an electric cable failure. ................................. 33
Figure 34: Map for probability of fire ignition by electric cable failure due to earthquake. .............. 34
Figure 35 – Map of the evolution of the forest fire (a) after the ignition by the electric discharge
from a probable ignition point and other possible outputs from the FireStation software: b) rate
of spread, c) linear intensity. ......................................................................................................... 34
Figure 36: Stages of use of the cascade effect model. .................................................................. 35
Figure 37: General overview of the CE Map transformation mechanism. ...................................... 36
Figure 38: WPS process in detail. ................................................................................................. 37
Figure 39: Diagram of cascade event chains identified for the occurrence of an earthquake......... 40
Figure 40: Diagram of flood cascade event chain.......................................................................... 42
Figure 41: Diagram of forest fire cascade event chain................................................................... 44
Figure 42: Diagram of extreme weather cascade event chain. ...................................................... 45
Figure 43: Diagram of cascade event chain for release of chemical substance............................. 47
31.07.2014 | vii
List of Tables
Table 1: Damage Probability Matrix for dike segment vulnerability. ............................................... 13
Table 2: Transition matrix for pilot B .............................................................................................. 16
Table 3: Candidate distributions, PDF, estimated (MLE) model parameters and uncertainties,
and Akaike Information Criteria (AIC). According to the AIC information, the model that best
describes the data is the Lognormal.............................................................................................. 31
Table 4: Description of chain blocks identified for possible cascade event chains after an
earthquake. ................................................................................................................................... 41
Table 5: Description of chain blocks identified for possible cascade event chains after a flood. ... 43
Table 6: Description of chain blocks identified for possible cascade event chains after a forest
fire................................................................................................................................................. 44
Table 7: Description of chain blocks identified for possible cascade event chains for a case of
extreme weather conditions. ......................................................................................................... 46
Table 8: Description of blocks of cascade event chain identified for possible cascade event
chains after a release of chemical substances. ............................................................................. 47
31.07.2014 | viii
Glossary of terms
Term
Domino effect
Cascade effect
Serial domino
(cascade) effect
Parallel domino
(cascade) effect
World state
Multi-hazard
Multi-risk
Damage
Adverse event
Risk source
Event
Definition
"a cascade of events in which the consequences of a previous
accident are increased by following one(s), as well spatially as
temporally, leading to a major accident“ (Delvossalle, 1996)
“the situation for which an adverse event triggers one or more
sequential events (synergetic event)” (Marzocchi et al. 2009)
“Happening as a consequent link of the only accident chain caused
by the preceding event” (Reniers, 2004)
“Happening as one of several simultaneous consequent links of
accident chains caused by the preceding event” (Reniers, 2004)
A particular status of the world, defined in the space of parameters
describing the situation in a crisis management simulation that
represents a snapshot (situation) along the crisis evolvement. The
change of world state, that may be triggered by simulation or
manipulation activities by the CRISMA user, corresponds to a
change of (part of) its data contents.
To determine the probability of occurrence of different hazards either
occurring at the same time or shortly following each other, because
they are dependent from one another or because they are caused
by the same triggering event or hazard, or merely threatening the
same elements at risk without chronological coincidence.
To determine the whole risk from several hazards, taking into
account possible hazards and vulnerability interactions (a multi-risk
approach entails a multi-hazard and multi-vulnerability perspective).
Definition 1 (context of socio-economic vulnerability, related with
concept of impact): the amount of destruction or losses, either in
health, financial, environmental functional and/or other terms as
a consequence of an occurred hazard (Marzocchi et al. 2009,
2012)
Definition 2 (context of structural damages): physical harm that
impairs the value, usefulness, or normal function of something
(Oxford Dictionaries,
http://oxforddictionaries.com/definition/english/damage)
Anything produced by a risk source in a certain area that can
generate phenomena with potentially adverse consequences. The
adverse event can be due to a risk source located inside or outside
the site where the event takes place (Marzocchi et al. 2009).
Element which alone or in combination has the intrinsic potential to
give rise to risk
(ISO/Guide 73:2009(en) Risk management — Vocabulary
https://www.iso.org/obp/ui/#iso:std:iso:guide:73:ed-1:v1:en)
Occurrence or change of a particular set of circumstances. An event
can be one or more occurrences, and can have several causes. An
event can sometimes be referred to as an “incident” or “accident”.
(ISO/Guide 73:2009(en) Risk management — Vocabulary
https://www.iso.org/obp/ui/#iso:std:iso:guide:73:ed-1:v1:en)
31.07.2014 | ix
ACRONYMS
Term
CE
CF
DEN
E
EQ
FF
IDA
k
OOI
P
TPD
PGA
SDOF
T
t
XMAX
WS
Definition
Cascade effect(s)
Cable failure
Damage to electricity network
Energy required to have ignition
Earthquake
Forest Fire
Incremental Dynamic Analysis
Poles stiffness
Object of interest
Power of electric discharge
Transition Probability Data
Pick ground acceleration
Single degree of freedom
Elastic period of poles
Time to ignition
Top displacement of electricity poles
World State
31.07.2014 | x
Executive Summary
This deliverable presents a concept model to assess the eventual occurrence of cascade
effects that was produced under the CRISMA Project. This deliverable is a consequence
of other deliverables produced in the CRISMA Project, namely D42.1 (Garcia-Aristizabal et
al., 2013) and D42.2 (Almeida et al., 2013). In this deliverable, besides an integrative
compiling of the results previously achieved, a description of the final concept model is
carried out. Moreover, an application of the concept model in a scenario where a forest fire
was triggered by an earthquake is detailed. The developed cascade effect model is
available in the CRISMA catalogue (https://crisma-cat.ait.ac.at/).
First chapter of D42.1 introduces the theoretical concepts of cascade effects and
describes the concept model dynamic scenario assessment due to cascade events and its
inclusion in the general CRISMA tool.
Cascade effects will have two applications in the pilots of CRISMA project, namely in Pilot
B, dealing with coastal submersion hazards, and Pilot D, relating an earthquake event to a
forest fire. Additionally an application with no further achievements was also planned for
Pilot A concerning to extreme weather conditions event. The scenarios for quantitative
analysis are described in Chapter 2 of this deliverable. In Chapter 3, the quantitative
analysis and the results achieved by application of the cascade effects conceptual model
are detailed.
The final conclusions and statements on the database and model for dynamic scenario
assessment integrating cascade events in a multi-risk assessment scheme are described
in Chapter 5.
31.08.2014 | 1
1. Introduction
A hazard crisis situation may be due to the occurrence of a single hazard event with large
impacts or due to several hazard events that occur simultaneously. Hazard events
occurring at the same time may have independent causes or may result from a sequence
of triggering hazard events. The outcome of a situation for which an adverse event triggers
one or more sequential events (synergetic event) is called “cascading effects” (Marzocchi
et al., 2009, 2012).
The perception and understanding of the potential occurrence of cascading effects is of
great relevance for planning and response activities since a surprising situation in a hazard
crisis scenario may endanger people and goods, and may nullify a strategy that was
developed accounting for a scenario in which the triggering event was a single occurrence.
1.1. Multi-hazards analysis
The development of a model to manage with cascade effects has several challenges as
several hazards are approached in a single application. A detailed description on existing
multi hazards and multi risks assessment methods was carried out in D42.1.
In the scope of the CRISMA Project, “Multi-hazard analysis” is seen as the determination
of the probability of occurrence of different hazards either occurring at the same time or
shortly following each other, because they are dependent or because they are caused by
the same triggering event or hazard, or merely threatening the same elements at risk
without chronological coincidence. By this definition, multi-hazard can be applied in
different perspectives: (1) multi-hazard seen as the assessment of different independent
hazards that threaten a common area or common exposed elements; (2) multi-hazard
seen as the assessment of triggering, domino, or cascade effects and (3) multi-hazard
seen as the assessment of possible hazard interactions (at vulnerability level) (GarciaAristizabal and Marzocchi, 2012).
In the development of a cascade effect model, the perspective of multi-hazard seen as the
assessment of triggering, domino, or cascade effects is followed. To assess the likelihood
of cascade effect occurrence, a transition between two related hazard events must be
considered. The main goal of the cascade effect model is to provide information about the
occurrence probability of a series of events. Therefore, two main aspects must be
considered: 1) the possible cascade event chains resulting from a triggering hazard event,
and 2) the transition probabilities from the triggering hazard event to the triggered events.
1.2. Cascade Event Chains Database
Based on a methodology proposed by Garcia-Aristizabal et al. (2013), a database with
many identified possible cascade event chains was created for the hazards managed in
the CRISMA project, namely: flood, earthquake, forest fire, release of chemical substances
and extreme weather conditions. This methodology and the preselected scenarios are
described in detail in D42.2 (Almeida et al., 2013).
The event chain scenarios identified are presented in APPENDIX (1)
Description of chain blocks identified for possible cascade event
chains
31.08.2014 | 2
. The sequence of events is often cyclic as a certain event type may occur in the same
chain more than once. This repetition may be direct (e.g. earthquake earthquake) or
indirect (e.g. ex: flood damage to structural protection flood). When a potential
repetition is verified, the symbol … was used to indicate the cyclic chain. In order to
simplify the event chain scenarios diagram, shortenings were created.
The possibility to have certain scenarios strongly depends on the specific case under
analysis. For example, the possibility of having an explosion triggered by damages to an
industrial facility depends on the type of industrial facility damaged in the area of interest.
Under this perspective, the proposed diagrams must be adapted to the characteristics of
the hazard event and to the area of interest.
The event chains database is of great importance as it allows the user to choose the chain
of interest and the cascade event analysis that shall be performed.
1.3. Integration of cascade effects into the general framework of
CRISMA
Figure 1 presents a scheme of the CRISMA framework. There is an initial world state that
is changed by the specific hazard simulation models originating a new world state. Each
model requires different inputs related to the hazard (e.g., earthquake intensity) or related
to the exposed elements (e.g., fuel cover distribution). The CRISMA tool also allows the
user to play with other models associated to the mitigation options, the resources
management and/or the choice of output. Finally, each simulated world state resulting in a
set of impacts which are traduced as indicators, criteria and costs that support the user
decision making.
Figure 1: Business logic of the CRISMA framework.
The integration of the cascade effect model in the CRISMA framework relates two
consecutive world states combining two different hazard events. Figure 2 shows a diagram
that intends to explain the integration of the cascade effect tool in the CRISMA framework.
Facing a given world state (WS0), the user may access the probabilities of occurrence of a
hypothetical hazard event triggered by the hazard event of WS0. The cascade effect model
is so appealed and the user is invited to choose the cascade event chain of interest to
31.08.2014 | 3
perform calculations. This event chain may have one or more transitions between hazard
events. Besides the specific information of WS0 that is required as input for the cascade
effect model, the transition probability data between the two hazard events is also
necessary.
Figure 2: Integration of the cascade effect model in the business logic of the CRISMA framework.
The cascade effect model provides information about the probability of occurrence of a
cascade of hazard events originated by the triggering hazard event. This information is
usually spatial as different probabilities occur for different locations. Since the information
about the probability spatial distribution of the eventual occurrence of a triggered event is
available, the user may simulate a new scenario to assess the impacts of a cascade effect
occurrence simulating and creating a new world state.
The interaction between the CRISMA framework and the cascading effect model (CEM) is
guaranteed by the “Cascade Events Configuration and Interaction View” building block,
hereafter shortened to “cascade effect view (CEV), which is available on the CRISMA
Catalogue (https://crisma-cat.ait.ac.at/bb/Cascade-Events-Configuration-and-InteractionView). The Cascade Effects View is a user interaction building that allows a user to
configure and run a Cascade Effects Scenario. The user can select a triggering event (for
example, an earthquake) and provide may either specify the characterization of the event
(Simulation Control Parameter) and thus initiate a new Simulation Model Run for this
particular event, or select (if available) the output of a past event or an event already
simulated. When the triggering event has been selected and characterized, the CEV
shows the possible paths of event chains that are available. The user may select one of
the paths and the Cascade Effects View will highlight eventual secondary events triggered
after the previous one. In this way, the user may select a specific chain of events that are
interested to assess or to interrupt a chain if he decides to stop the analysis in an
intermediate point. For each of the events, the user may either characterize the event by
providing the respective simulation control parameters or select the output of a past event.
In parallel, the user may access to the available transition probability data (TPD) using this
interactive building block. After selecting the event chain, the list of the available related
transition probability data are shown for each transition. The user shall select the data to
be used for each transition. The interaction view allows the creation of a new TPD and the
editing of the existing TPD.
31.08.2014 | 4
The cascade effect model will use several inputs provided by the cascade effects view in
order to produce the probability map(s) that will be shown by the same interactive building
block. The CEM will not interact with the user as this will be carried out by the CEV. CEM
is a black box that will generate cascade effects probability maps that will be show to the
user by the CEV.
31.08.2014 | 5
2. Scenarios for quantitative analysis
In D42.2, possible adverse event chains were identified and described for five different
triggered hazards, namely: earthquakes, floods, forest fires, extreme weather conditions
and release of chemical substances. Only three of these hazards were planned to use the
cascade effect conceptual model in CRISMA Pilots. The scenarios will be detailed in the
following paragraphs. In all, the triggering event shall be seen as an already occurred
event (pre-determined occurrence) and the consequent events shall be seen as episodes
which the end-user intends to evaluate the probability of happening and the possible
impacts associated.
2.1. Triggering hazard: Extreme Weather Condition (Pilot A)
Extreme weather conditions can cause various cascade events chains. In Pilot A, the
original purpose was to assess the probability of a crisis scenario taking place in northern
Finland due to extreme winter weather. The scenario is initiated when a low pressure
system forms in southern Scandinavia in mid-December. The system moves towards
Finland, bringing lots of snow (30 cm/day over land) which, together with freezing drizzle
causes very poor road conditions. Crown snow-load starts to accumulate on roads, trees
and power lines. A second low pressure system with snow storms and winds gusting up to
20 m/s on land arrives one day after the first one. Two days later, a third low pressure
system hits with extremely heavy winds (gusts on land 30–40 m/s), causing major
problems on road, electricity and communication networks. Finally, the low pressure centre
moves slightly southeast and cold air starts to flow from the northwest. Temperatures fall
widely below -10°C. During the following 2-3 weeks, the cold weather spreads into the
area and daytime temperature falls widely down to -20–45°C, causing the need for
evacuation. The return period for this kind of event has not been estimated, but based on
expert judgement it is once in several hundred years.
The main focus of Pilot A is the event chain from the initial weather event to the damage to
the electricity power lines, which, combined with the cold spell, causes health problems to
vulnerable communities, such as the elderly, causing a need for evacuation. Two chains of
events cause damage to the power lines: the falling snow and heavy winds damage either
the power lines directly or the damage is caused by leaning or falling trees (see Figure 3).
The initial weather event is pre-determined; therefore the probability of the event is
assumed to be one. However, the extent of the damage is not known and it depends on
several factors, such as wind direction, forest cover, soil etc.
31.08.2014 | 6
Figure 3: Diagram of cascade event chains identified for the occurrence of extreme weather
conditions.
As this kind of an event has never happened before in northern Finland, no historical data
could be used to assess the transition probability data for the cascading event. Therefore,
a method called expert elicitation was used to obtain subjective probabilities of the event
chain. The study was conducted at the Pilot A demo seminar in Kemi, Finland on the 7-9th
of April, 2014. The experts used in the elicitation were emergency service actors and
representatives of the relevant municipalities and electricity distribution companies, who
participated in the event.
The result of the elicitation was that the probability of the original scenario of pilot A is
practically zero due to three reasons: 1) over 70% of the electricity distribution cables in
Kemi area are underground cables; therefore, storms cannot damage them; 2) for
overhead cables, trees so high up north are too short and weak to cause any major
damage; and 3) the damage caused by snow and other perils has been historically
repaired in a short period of time (the majority in 30 minutes, all in less than two days).
The outcome of the expert elicitation resulted in a modification of the scenario description
in Pilot A. The new description takes a long power blackout as given and assumes direct
impacts of occurring because of that. Therefore, no cascade event scenario can be
described or assessed. Due to the abovementioned reasons, it was decided to not use
31.08.2014 | 7
cascade effect model in Pilot A. The main decision was the exclusion of these event
chains from the general diagram. Therefore, hereinafter, this use case for cascade effect
will be not developed.
2.2. Triggering hazard: Coastal Submersion (Pilot B)
In these particular applications the triggering flood is referred to a costal submersion based
on the Xynthia coastal floods occurred in France in 2010. Two of several event chains
(Figure 4) having a flood as the triggering event are planned to test in the Pilot B of
CRISMA Project.
The first event chain is: Flood Damage to Structural Protection Flood. The violence of
the tides may damage or destroy the dams and dikes which are protecting the land area.
This scenario of destruction exposes the terrestrial elements to water, even for lower water
levels that would be not so relevant if the protection barriers were intact.
The second event chain is: Flood Damage to Industrial Facility Release of chemical
substance. The flooding waters may damage the electricity network, leading to a power cut
and causing a dysfunction of a wastewater treatment plant. The wastewater that cannot be
treated nor stored is so discharged in the ocean affecting its environmental quality and
possible triggering to eutrophication problems.
31.08.2014 | 8
Figure 4: Diagram of cascade event chains identified for the occurrence of flood.
2.3. Triggering hazard: Earthquake (Pilot D)
Figure 5 lists the several identified event chains that can be initially triggered by an
earthquake. In Pilot D, two applications of cascade effects will be demonstrated: (1)
earthquake earthquake; and (2) earthquake
damage to electricity network
forest
fire.
31.08.2014 | 9
Figure 5: Diagram of the identified cascade event chains for Earthquake (from D42.2).
In the first scenario, the case of a seismic sequence triggered by a main shock is
analysed. This case is based on the fact that the occurrence of a medium-to-big
earthquake has the capacity to produce a perturbation of the stress field around the source
of the main shock, stimulating the occurrence of triggered seismicity and, therefore,
increasing the seismic hazard in the short-term. Accounting for the characteristics of the
triggered seismicity, the objective of this scenario is to assess the effects of the generated
seismic sequence in the short-term seismic hazard, and to calculate the possible damages
associated with the triggered seismicity. : this scenario both the updated building inventory
(after the occurrence of the main shock) and time-dependent vulnerabilities are
considered.
The second scenario links an earthquake to a forest fire. In this scenario, a Mw 5.6
earthquake is the triggering event and the end-user plays with CE (cascade effect) model
to evaluate the possibility of having a triggered forest fire, initiated by a cable failure in the
31.08.2014 | 10
electric network. In this scenario, the earthquake provokes damages to the electricity
network specifically in the cables joints/couplings device (Figure 6a) near to the pole
causing a rupture in the electric cable. As this electric cable is energized, it ionizes the air
to the ground and consequently an electric discharge happens. In Figure 6b shows an
example of air ionization caused by the proximity of a tree branch and Figure 6c shows a
fire ignition caused by an electric discharged. Both examples follow the same principle of
the application managed in Pilot D however in the CRISMA application the area reached
by the electric arc is covered by surface fine forest fuels possible triggering an ignition
which may develops to a forest fire.
(a)
(b)
(c)
Figure 6: Photos of electric poles.
There are several forest fire events triggered by an electric cable failure. Figure 7a shows
an image of a cable damaged which was sufficient to create an electric arc to the ground
originating a large forest fire. The most important forest fire event triggered by an electric
failure occurred in Victoria (Australia) in February, 8th 2009 (the Kilmore Fire). This fire
event was originated by an electric cable failure of the pole 39 (Figure 7b) driving to the fall
of the cable in a wild land area (Figure 7c). The safety system interrupts the energy supply
in 0.25s and three attempts to turn on the power for a total time of 4.48s were registered.
Among several impacts, this fire results in 119 people dead. The court decision
condemned the responsible to pay about 500 million Australian dollars as compensation
for the losses.
Cable damaged
over the ignition
point
(a)
(b)
(c)
Figure 7: Images related to forest fires triggered by electric cable failure. a) Oliveira de Frades
(Portugal) in 28/02/2004; b) Schematic view of top of pole 39; red circle indicates where the
conductor failed (Plaintiff, 2013); c) Photograph of fire ignition area (Plaintiff, 2013).
A scenario of a forest fire following an earthquake assumes a great interest in the
operational point of view. On one side the earthquake event requires the deployment of all
31.08.2014 | 11
the available civil protection means, including firefighters. If a forest fire stars, some of
these units must be allocated to fight the fire as soon as possible to avoid the enlargement
of the fire and the appearance of a new hazard event requiring a major concern. On the
other side, if the fire spreads to threat a village, people cannot stay inside houses because
of the earthquake impacts and of eventual replicas, but cannot stay outside if the village
become immersed in smoke. In this scenario, the evacuation may be the only option which
must be planned in advance because the evacuation routes may also be disrupted by the
earthquake and by the fire. Therefore, a preventive evaluation of this chain of events is of
great interest and shall be object of reflection. The cascade effect tool would help in this
reflection and would support the decision making.
Running of cascade effects model in this scenario will provide the spatial distribution of the
probability of having a fire ignition in a fuel bed caused by an electric cable failure triggered
by an earthquake. With this information, the end user may use a forest fire behaviour
prediction model (as for example FireStation, which has been included in the CRISMA
Catalogue) in order to assess the possible impacts. This information may support planning
in a short or long term.
31.08.2014 | 12
3. Quantitative analysis and results
The scenarios for application of the tool for assessing cascade effects were previously
described. Once the event chains of interest are selected, a number of input data is
required as for example information related to the world state and information about the
probabilities for the occurrence of transition between events.
Considering a specific scenario, after the occurrence of the triggering event the initial world
state changes as a consequence of the impacts resulting from that hazard event. However
some elements may be time dependent vulnerable and therefore the word state may be
continuously changing after the triggering event. Time dependent vulnerability is presented
in in deliverables D43.1 and D43.2, where much information on this subject is available.
Nevertheless to access the cascade effects tool, the world state in the moment of the
plausible transition must be defined.
The transition probability data (TPD) are essential to determine the likelihood of the
transition. TPD can be previously uploaded to the system in the platform database by the
administrator or may be provided or changed by the user during the utilization of the
cascade effect model. The inputs must be available as geo referenced data or text data
allowing calculations with geo reference data.
3.1. Triggering hazard: Coastal Submersion (Pilot B)
3.1.1. Flood
Flood
3.1.1.1. General perspective
In case of coastal submersion, several phenomena can damage dikes such as external
erosion, internal erosion and overflow.
Three damage levels are considered for the exposition of dikes towards storm surges or
floods. First, the dike resists to the hazard and thus there is “no failure”. Second, the dike
starts to fail that means a “breach” is formed. Within CRISMA, in the scope of the
simulation of dike breaches (with TELEMAC-2D), we will consider that the breach formed
within the segment will be set at 10% of the segment’s length that means at a maximum of
25 meters. Third, it is the “total failure”, the dike totally collapsed.
When the dyke is damaged, it loses its capacity to protect population, buildings, networks,
etc. behind it. If a second coastal submersion occurred before the reparation of the dykes
the flood extension can be much more important than in the first case.
3.1.1.2. Input Data
Two input data are required:
1. The localisation and the initial status of the dike
2. The water level above the dike during the first meteorological event
31.08.2014 | 13
3.1.1.3. Transition Probabilities
According to the physical status of the dike, the consequences of an overflow above the
dike will be different. It is possible to express these different responses to the hazard
intensity (water level overflowing above the dike segment) in a table (see Table 1), where
the probabilities of attaining the different damage levels are synthesized.
The result for each segment of dike is a probability to resist, breach or fail. The end user
will be able to see on a map the segments of dike with a colour:
Green if the maximum of chance is to resist
Orange if the maximum of chance is to breach
Red if the maximum of chance is to totally fail
Table 1: Damage Probability Matrix for dike segment vulnerability.
Water level
above the
dike
Status
Good
Medium
Poor
<20 cm
99,9% No failure
0,1% Breach
0% Total Failure
99% No failure
1% Breach
0% Total Failure
10% No failure
80% Breach
10% Total Failure
20 to 50 cm
99% No failure
1% Breach
0% Total Failure
10% No failure
80% Breach
10% Total Failure
5% No failure
15% Breach
80% Total Failure
>50 cm
98% No failure
2% Breach
0% Total Failure
5% No failure
15% Breach
80% Total Failure
0,1% No failure
4,9% Breach
95% Total Failure
>1 m
10% No failure
80% Breach
10% Total Failure
0,1% No failure
4,9% Breach
95% Total Failure
0,0% No failure
0,1% Breach
99,9% Total Failure
On the pilot B application, the user have to decide the behaviour of each dike segment
according the classification and of course its own expertise. If this expertise is not
available, it’s possible to automatically fix the behaviour of the dike segment according to
the vulnerability classification.
If after an event (or a simulation of the event), the dike segment vulnerability is classified:
with an “orange colour” meaning that it is likely to breach or if the user decide
this segment to breach, then, it will be reclassified in the poor status.
with a “red colour” meaning that it is likely to totally fail or if the user decide this
segment to fail, then, it will be reclassified in the “collapsed” status
After the original event, we are able to modify the classification of the dykes in the World
State and a new simulation can be done for another flood.
3.1.1.4. Output results
As previously mentioned, each dike segment will be classified according to the
vulnerability and the probability to fail. If the user assumes the failure of one dike segment,
a new World State (concerning the dykes) and a new flood extension will be created by the
specific software provided by the general CRISMA tool.
31.08.2014 | 14
3.1.2. Flood
Damage to Electricity Network
Release of Chemical Substance
The first event is a coastal submersion. The impacts of the submersion include damages
to the electricity network. These damages trigger a power cut and immediately the total
dysfunction of a wastewater treatment plant. The wastewater then cannot be treated and is
directly discharged in the ocean which pollutes the environment (ocean and submerged
areas) (Figure 8).
Figure 8: Cascade events chain for pilot B.
The aim is to describe the cascade chain event with a transition matrix of probabilities.
Within the pilot B, we use the historical event of Xynthia storm surge that occurred in
February 2010 and led to coastal submersions. During Xynthia, in the municipality of La
Rochelle, several areas were submerged. More particularly, the area of Port-Neuf where
the wastewater treatment plant is located was flooded (Figure 9). Due to the water, the
electricity network was damaged in this area. This triggered the dysfunction for several
weeks of the wastewater treatment plant of Port-Neuf. The wastewater directly flowed in
the natural environment (release of chemical substance).
Figure 9: Maximum water level in the area of Port-Neuf in La Rochelle during Xynthia.
The power cut leading to the plant’s dysfunction has two possible sources. First, the
submersion reaches the nearest electrical converter leading to a power cut. Second, the
submersion reaches the latest pumping station where the effluents are regulated following
31.08.2014 | 15
a short circuit due to water entering in contact with electrical components. In the two
cases, a bypass of the effluents would be immediately created and the natural
environment would be polluted by the discharge of wastewater. The daily volume of
wastewater discharged as well as the concentration of polluting elements can be
calculated considering the plant capacity.
Moreover, it can be noted that the plant requests power to function. In other words, a
power cut leads irremediably to a total dysfunction of the plant and a bypass of the
wastewater. Then, we can state that the probability that a power cut trigger to pollution by
wastewater bypass is 100%.
3.1.2.2. Input Data
There are 3 input data:
1. The localisation of the wastewater plant
2. The characteristic of the original flood on the localisation of the wastewater plant
3. The characteristic of the daily volume of wastewater discharged. The discharge
of the wastewater as well as the concentration of the pollutant can be calculated
with the capacity of the wastewater treatment plant. The Port-Neuf plant
capacity is 170 000 population equivalents. Then, the pollution is characterised
by:
Wastewater discharge: 25 500 m3/day
Concentration in biochemical oxygen demand (BOD5): 2.35x10-3 mg/L
Concentration in chemical oxygen demand (COD): 5.29x10-3 mg/L
Concentration in phosphorus: 1.57x10-4 mg/L
Concentration in Total Kjeldahl Nitrogen (TKN): 5.88x10-4 mg/L
The transition matrix is developed to express the probability that a coastal submersion
triggers pollution by wastewater in the ocean. The intensity of coastal submersion is
expressed with the water level above ground level. The intensity of pollution is expressed
by the wastewater discharge and the concentration of the released pollutant substances
within the wastewater. These are unique values depending on the plant capacity. Indeed,
the wastewater release in the natural environment due to the power cut is the amount of
waste water arriving in the plant. Then, we will assess the probability to have pollution from
wastewater for different coastal submersion intensity. Actually, the matrix is a vector
(Figure 10).
Figure 10: Transition matrix developed for pilot B.
31.08.2014 | 16
The probability P(Poll|Sub) within the transition matrix to assess is the combination of two
probabilities:
The probability P(Power cut|Sub) that the coastal submersion leads to a power
cut in the plant area.
The probability P(Poll|Power cut) that the power cut lead to the total dysfunction
of the plant and the waste water discharge in the ocean.
We stated above that the probability P(Poll|Power cut) that the power cut lead to the total
dysfunction of the plant is 100%. Then by combining the two probabilities, the probability
P(Poll|Sub) that the coastal submersion triggers the pollution is the probability P(Power
cut|Sub) that the coastal submersion triggers a power cut. (Equations 1, 2 and 3)
P(Poll|Sub) = P(Power cut|Sub) X P(Poll|Power cut)
[1]
= P(Power cut|Sub) X 1
[2]
= P(Power cut|Sub)
[3]
It can be noted that in case of storm surge, the power is not put back to service before 48
hours. In the case of pilot B, we only model 48 hours and thus the power will not be
restored at the end of the simulation.
In the municipality of La Rochelle, the electrical converters are not in the flooded areas.
Therefore, only the power cut by the submersion of the pumping station will be considered.
We suppose that the electrical components of the latest pumping station before the plant is
at 20 cm above the ground level. Then, as soon as the water level reaches these 20 cm,
the probability to have a short circuit is 100%. If the water level is lower than 20 cm, the
power cut is possible due to humidity or due to the salinity. Considering different classes of
submersion intensity (water level above ground level), it is possible to assess the
probabilities to have a power cut. (Table 2)
Table 2: Transition matrix for pilot B
Water level
0–5 cm
5–10 cm
10–15 cm
15–20 cm
>20 cm
Probability to have a
power cut
0.01
0.1
0.15
0.2
1
As it was noticed before, the probability that the submersion triggers a pollution by
wastewater spill is the same that the probability that the submersion triggers a power cut.
Results come from a new simulation of the coastal event with pollutant transport. It hasn’t
be done yet, in reason of the too long computation duration of TELEMAC model. Output
results will be shown in Deliverable D53.1 – Demonstrator of Pilot B – to be delivered in
month 38th.
3.2. Triggering hazard: Earthquake (Pilot D)
As it was previously mentioned, the cascade effect application having an earthquake as
the triggering hazard deals with two different event chains: (1) earthquake
earthquake;
31.08.2014 | 17
and (2) earthquake
damage to electricity network
forest fire (fire ignition). Figure 11
shows the workflow for the use of cascade effect in Pilot D of CRISMA having an
earthquake as the initial triggering event. This diagram shows the several inputs, the
functions and the world states required for the two scenarios used for demonstration.
Figure 11: Workflow for the application of cascade effect model in Pilot D of CRISMA.
3.2.1. Earthquake
Earthquake
As can be seen in Figure 11, the earthquake-earthquake scenario consists on the
assessment of the potential impacts of the triggered seismicity that characteristically
happens after a seismic event of certain magnitude. Forecasting the future behaviour of
seismic sequences is not an easy task, and currently is a subject of intense research on
applied seismology. Assuming that a given model can be used to forecast in the shortterm the likely seismicity rates (e.g. in terms of number of events/day) and its expected
spatial distribution, then short-term seismic hazard assessment can be performed and the
expected damages caused by the triggered seismicity can be continuously updated. A
logical representation of the data requirements and interactions is represented in Figure
12.
31.08.2014 | 18
Figure 12: Logical flow of information and data for the assessment of the earthquake-earthquake
scenario.
3.2.1.2. Input Data
Different kinds of input data should be available to assess this scenario. First, an initial
building inventory in the target area needs to be created. In this example, the initial
building inventory is represented in Figure 13. The target area is divided in a regular grid,
and the building inventory is represented, for each grid element, as the number of
buildings within each cell and the proportion of different building classes. Each building
class corresponds with a specific set of fragility functions.
Figure 13: Building inventory in L’Aquila test case (Pilot D). The computation domain is divided in a
regular grid, and for each grid element the total number of buildings and the proportion of different
building classes are represented.
Other input data necessary to build this example is (1) an earthquake acting as the
triggering event, and a seismic sequence occurring as a consequence of the initial
triggering event. In this example, we simulate the occurrence of a main shock – aftershock
sequence occurring in a zone located at the NW of L’Aquila city, as shown in Figure 14.
31.08.2014 | 19
The characteristics of the seismic sequence are simulated assuming seismicity rates, sizefrequency distribution and spatial distribution of the events of similar past sequences in
this region. In a near real-time application, these parameters can be fixed assessing using
the occurring seismic sequence.
Figure 14: Simulated seismic sequence located at the NW of L’Aquila city, Italy. The main shock is a
Mw 5.6 event.
The shake map of the triggering event is plotted in Figure 15 and represents the spatial
distribution of the Peak Ground Acceleration (PGA) in %g units. These intensity values,
together with the fragility functions for the different building classes, can be used to
calculate the probability of having building damages (e.g. collapse) in the target area. The
results of the damage probabilities for the main shock are represented in Figure 16.
31.08.2014 | 20
Figure 15: Shake map of the triggering event, representing the spatial distribution of the Peak
Ground Acceleration (PGA) in %g values.
Figure 16: Probability of having collapsed buildings in the target area after the occurrence of the
main seismic event.
Using the triggered seismicity, short-term seismic hazard assessment can be performed
using the data (i.e. spatial location, magnitude, shake maps) of the forecasted seismicity.
Figure 17 shows examples of the spatial distribution of PGA values for different
exceedance probability thresholds: 1% (Figure 17a), 0.1% (Figure 17b), and 0.01%
(Figure 17c). The resulting transition probabilities for this scenario are the exceedance
probabilities associated with different PGA values. At each grid element of the calculation
domain, these transition probabilities can be represented as a hazard curve. A summary
of the resulting hazard curves for the whole domain are represented in Figure 17d.
31.08.2014 | 21
a)
b)
c)
d)
Figure 17: PGA values for different exceedance probability thresholds: (a) 1%; (b) 0.1%, and (c)
0.01%; and (d) resume of the result hazard curves.
Using the updated building inventory (after assessing the direct impact of the triggering
earthquake) and the transition probabilities shown in Figure 17, it is possible to calculate
the expected impact of the likely triggered seismic sequence. Figure 18 shows the
probability of having collapsed buildings in the target area for the earthquake-earthquake
scenario, according with the characteristics of the seismic sequence triggered by the main
seismic event. A direct comparison with the direct impacts expected after the main event
(e.g. between Figure 16 and Figure 18 may be performed in order to assess the expected
effects of the triggered seismicity.
31.08.2014 | 22
Figure 18: Probability of having collapsed buildings in the target for the earthquake-earthquake
scenario. The map represents the probability of collapsed buildings considering the potential
triggered seismicity according with the characteristics of the seismic sequence triggered by the main
seismic event.
3.2.2. Earthquake
Damage to Electricity Network (Cable failure)
Forest Fire
Figure 19 shows the structure of circulation of information in the application of cascade
effects to this scenario. Besides the probability function regarding to the fragility curve of
the electric system and to the probability of having ignition after the electric cable failure,
three categories of inputs must be previously available on the world state, namely: the
location of the electricity network, the intensity distribution of the earthquake and the fuel
cover in the area of interest.
31.08.2014 | 23
Figure 19: Flow of information during the running of cascade effects for the selected scenario.
“Application 1” is referred to the transition EQ-DEN and “Application 2” is referred to the transition
DEN-FF.
The probability of having a cable failure in the electric system (Application 1 in Figure 19)
and the consequent formation of an electric arc results from the intensity distribution of the
earthquake and the fragility curve of the electric devices. The intensity of the earthquake
along the electric network is a part of the world state and information is available on the
shake map. The fragility curve relates the intensity of the earthquake to the potential
damage caused in the electric network along the line.
The probability of having an ignition started by an electric discharge (Application 2 in
Figure 19) results from the fuel classification of the area where the electricity cable failure
is located and the probability model of ignition by electric discharge. The fuel classification
is a part of the world state and is provided by the fuel map of the area. If the electric
discharge occurs in a non-fuel area, as for example a road, the probability of ignition is
null. However, if the electric discharge occurs in a fuel area, as for example a grass land,
the probability of ignition shall be determined by the use of the probability ignition model
that will be detailed later.
If we attend to the whole event chain EQ-DEN-FF, the probability of having a fire ignition
due to an earthquake is given by the product of the probability of having an electric cable
failure and the probability of having a fire ignition triggered by an electric cable failure.
3.2.2.2. Input data from the wold state
Since the cascade effect model is appealed and the user selects the event chain linking
the earthquake to damage to electricity network (cable failure) and to forest fire, some
information regarding to the world state is required. That information is following detailed.
Location of the electric network
The location of the electric network used for this example is not real since it was defined in
the most opportune locations. For convenience of Pilot D, the electric line was designed to
pass close to the Village of Castel del Monte in order to have a forest fire threating this
community. On the other hand, this line was designed to cross the area of interest to have
31.08.2014 | 24
as many probabilities calculations as possible. Apart those two assumptions, the poles are
randomly located only taking precautions to do not put a pole in an unlikely place such as
in the middle of a river. The separation between two successive poles is around 300m.
Figure 20: Geographic distribution of the electric poles.
Map of EQ intensity distribution
The occurrence of a Mw5.6 earthquake in the NE part of the domain has been simulated
as the triggering event for this scenario. The shake map of this event, represented by the
intensity of the ground motion in the area of interest, is shown in Figure 21. Using the
shake map of this event, the peak ground acceleration (PGA, in %g) of the ground motion
is calculated at the base of each of the poles of the electric network considered for this
example.
Figure 21: Map of the triggering earthquake intensity distribution with location of the electric power
network (distribution lines – smaller poles).
31.08.2014 | 25
Fuel map
The fuel map is important to evaluate the probability of ignition as it has the information of
the class of fuel where each electricity pole is located.
To be harmonized with the fire behaviour prediction model FireStation, the classes used in
the fuel map are consistent with those used in FireStation. As can be seen in Figure 22, in
the area of interest there are seven different fuel classes. Grassland is that one with higher
representativeness.
Figure 22: Fuel map for the region of L’Aquila.
The fuel map showed in Figure 22 was developed, in April 2014, by the ArcFUEL Project
Consortium from a cooperation protocol established between CRISMA Project and the
European LIFE+ Project ArcFUEL.
3.2.2.3. Transition probability data for the cable failure triggered by an earthquake
The cables connected to low voltage distribution lines (evidenced in red in Figure 23) are
those considered as the most vulnerable ones to external shocks (e.g. seismic excitation).
Small poles have a high probability to fall down during or after an earthquake than larger
poles. However, during a seismic event, all the poles (low, medium and high voltage)
shake and the stress induced in the electric cables may cause its breakage.
31.08.2014 | 26
Figure 23: Sketch of a T&D system for an EPN (TL = Transmission Lines, D = Distribution lines, TD
[HV MV] = Transformation (from high to medium voltage) and Distribution station, TD [MV LV] =
Transformation (from medium to low voltage) and Distribution station, L = Load) (adapted from D5.2
Deliverable of the EU project SYNER-G).
In particular, the electricity poles used in electricity distribution lines may be subdivided
based on the material (wood, reinforced concrete or tubular steel), diameter and height.
The tubular steel electric poles are very common and for this reason we concentrated on
this typology. In Figure 24 there is an example table classifying the electricity poles of this
category according to the diameter and height.
Figure 24: An example of tubular steel poles classification (from a commercial producer).
In order to assess the probability of the cable failure, a number of pole classes are studied.
In particular, referring to a generic class, the fragility curve represents the probability of
attaining a limit value of displacement at the pole top varying the intensity of the seismic
input. Such limit value is represented as a fraction of a displacement XMAX (three
31.08.2014 | 27
hypotheses are considered: 0.5 XMAX, XMAX and 1.25 XMAX, where XMAX is defined as the
displacement corresponding to the static application of the design force T1 (see Figure 25)
Figure 25: Definition of top displacement XMAX.
The intensity input may be expressed with various measures. Here we adopt the peak
ground acceleration.
In order to determine the probability of attaining xlim, an Incremental Dynamic Analysis
(IDA) is performed (Vamvatsikos and Cornell, 2002). Figure 26 shows the normalized
pseudo-acceleration spectra of the selected records, together with the mean spectrum and
comparison with EC8 based representation.
Figure 26 – Normalized pseudo-acceleration spectra.
In order to perform the IDA, the pole is schematized as a Single Degree Of Freedom
System (SDOF) with a concentrated mass on top (see Figure 27), suitably characterized
by stiffness k and elastic period T; 5% critical damping is assumed.
31.08.2014 | 28
Figure 27: Representation of the electric poles and the mass on the top.
Figure 28 shows an example IDA obtained for the pole type 12B14.
Figure 28: IDA for the pole type 12B14.
Elaborating IDA results with the approach proposed in (Porter et al., 2007) the fragility
curves can be obtained. Figure 29 show the fragility curves derived for pole type 12B14.
31.08.2014 | 29
Fragility 12B14
1
0.5 X
0.9
MAX
X
0.8
MAX
1.25 X
MAX
3
P[X>XL ]
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
5
10
15
20
25
30
35
40
PGA(g)
Figure 29: Fragility curves of electric poles to a transversal force T1.
In order to simplify the use case, all the electricity poles of the area of interest will follow
the fragility curve XMAX.
Transition probability data for fire ignition triggered by cable failure
The transition from the electric cable failure to a fire ignition has two evolutionary steps: 1)
the production of the electric arc from the electricity cable to the fuel bed, and 2) the
ignition of the fuel bed by the electric arc.
It is assumed that the failure of an electric cable imperatively causes an electric arc. Due
to the existing protection systems the electric charge is normally interrupted in case of
failure and after that, for brief instants, the residual electric charge continues flowing. After
the cable breakage, the cable takes some time to land or to reach a sufficient distance to
establish the electric arc. The time to ignition by electric arc with high voltage is very short
(tenths of seconds) and therefore in this sense it is reasonable to consider the production
of an electric arc for a short period as inevitable. On the other hand, the time to ionize the
air path from the cable failure to the fuel bed and the consequent production of the electric
arc is strongly dependent on the distance and on the resistance of this path. This
dependency is a function of many additional factors such as topography, cable height, air
humidity, and pressure. For simplification reasons we will not take those aspects into
consideration in this document, but rather assume that the cable failure will always drive to
an electric arc that reaches the fuel bed.
In a laboratory environment, several tests were carried out in order
to determine the total amount of energy required to ignite a certain
amount of fuel. These tests were performed for straw and pine
needles (Pinus pinaster) for a range of fuel moisture content (FMC)
between 9.2% and 12.2% (APPENDIX (2) Main parameters of the
laboratorial tests for the determination of ignition by industrial
electric discharge.
31.08.2014 | 30
). A bouquet of the fuel material was exposed to an industrial electrical arc of certain power
(P [kVA]) as can be seen in Figure 30. The time (t [s]) elapsed from the beginning of
discharge until the evidence of combustion was measured. The energy (E[kJ]) required for
ignition was determined using Equation 4.
a)
b)
c)
d)
Figure 30: Sequence of images of a laboratorial test to determine the energy required for ignition by
industrial electric discharge.
E
P t
[4]
It was found that the probability of ignition was not significantly different for the several
tests at different conditions of FMC or different types of fuel (straw or pine needles) and
also the fuel load does not seem really relevant for the probability of ignition. This
statement should be considered preliminary since not many tests with fuels other than
straw and pine needles were carried out and the range of fuel moisture content was not
very extensive.
Using the data collected in the laboratory experiments, we estimated the parameter of
different competing probabilistic models in order to find the distribution providing the best
description of the laboratory observations. As possible competing models we used the
Log-normal, Gamma, Normal, Weibull and Exponential distributions. We estimate the
model parameters for each candidate model using a Maximum Likelihood Estimate (MLE)
approach, and use the Akaike Information criteria (AIC, Akaike 1974) for the model
selection. The AIC is a tool based on the concept of entropy, and offers a relative measure
of the information lost when a given model is used to describe some data (a tradeoff
between accuracy and complexity of the model).
Table 3 summarizes the functional form of the PDF, estimated (MLE) parameters (and
uncertainties), and the AIC for all the probabilistic models considered. Using a
Kolmogorov–Smirnov test, we cannot reject the Log-normal, Gamma, and Normal
hypotheses (at significance level of 0.05), which means that, from a statistical point of
view, all these probability models successfully explain the observed data. However,
according with the AIC values, the preferred model (i.e. that with the lowest AIC value) is
the Log-normal.
31.08.2014 | 31
Table 3: Candidate distributions, PDF, estimated (MLE) model parameters and uncertainties, and
Akaike Information Criteria (AIC). According to the AIC information, the model that best describes
the data is the Lognormal.
Model
Log-normal
( , )
Gamma
(a,b)
Normal
( , )
Weibull
(a,b)
Exponential
( )
Probability density
(
= ( | , )=
= ( | , )=
= ( | , )=
( )
)
(
= ( | , )=
= ( | )=
)
Parameters
(MLE)
AIC
=6.334 [6.329,6.339]
=0.246 [0.243,0.250]
– 62803
a=16.68 [16.22,17.14]
b=34.84 [33.87,35.83]
– 62845
=581.0 [578.1,583.8]
=146.3 [144.2,148.3]
– 63319
a=637.1 [633.7,640.6]
b=3.87 [3.81,3.92]
– 63832
=581.0 [569.7,592.6]
– 72815
Figure 31a shows the cumulative probability (CDF) of the candidate distributions and the
empirical CDF of the observed data, while the Figure 31b shows the CDF and related
uncertainties of the Log-normal model.
a)
b)
Figure 31: a) Plot of the Cumulative Distribution Function (CDF) of Log-normal, Gamma, Normal,
Weibull, and Exponential competing models, and the empirical CDF of the observed data; b) CDF of
the Log-normal (with parameters as those presented in Table 3), selected as the best model
describing the observations.
Therefore, the model selected to describe the energy required to start an ignition is a Lognormal with parameters =6.334 [6.329,6.339] and =0.246 [0.243,0.250], where the
values in parenthesis represent uncertainty bounds. The Log-normal model is therefore
used for the determination of the probability of fire ignition given the occurrence of an
electric discharge. As for simplicity we consider that an electric cable failure will always
generate an electric discharge, this model consequently also represents the probability of
a fire ignition given a cable failure.
31.08.2014 | 32
As previously mentioned, there was no significant difference in the results achieved for
pine needles as compared to straw. Therefore, for simplification, we assume that the
model can provide the probability of ignition for all fuel classes other than urban areas,
where the probability of ignition by electric discharge is assumed to be zero. This
assumption seems reasonable as commonly, for prevention reasons, the area below
electric cables is cleaned of heavy fuels.
3.2.2.4.
With all the required inputs available, the cascade effect model is prepared to show the
most interesting outputs to the end user. In this application there are two different use
cases: (1) earthquake – cable failure, and (2) earthquake – cable failure – fire ignition.
Earthquake – Cable failure
Figure 32 shows the distribution of probabilities to have a cable failure in the electric
network after a seismic event with a Mw 5.6. Probability distribution results from the
earthquake intensity distribution (Figure 21) and the fragility curve of the poles (Figure 29).
Figure 32: Map for probability of cable failure.
Obviously, only in the electric network is possible to have cable failure. As it was
previously explained the cable failures which we are dealing with occur close to the electric
poles. Therefore, the information on the cable failure probability is presented around the
poles. In this case, a unique electric poles fragility curve was used (XMAX=1) and therefore
the probability of cable failure depends exclusively on the distance between each pole and
the seismic epicentre.
Earthquake – Cable failure – Fire ignition
Besides the visualization of the map for probability cable failure due to the earthquake, the
user may want to access the probabilities of fire ignition due to cable failure. This
information is provided in the form of a distribution map as showed in Figure 33. The map
for distribution of fire ignition by electric cable failure is obtained from the fuel map (Figure
31.08.2014 | 33
22), the geographic location of the electric network (Figure 20) and the probability function
of fire ignition (Equation 5).
As previously explained, the probability of fire ignition is very dependent on the energy
released by the electric discharge. The value of energy released during the discharge was
determined assuming a trigger time (electric interruption) of 0.1s, a power of 8000kVA
(current of 53A and voltage value of 150kV) and consequently 800kJ of energy for the first
20km of electric line (first 67 poles on the right side of Figure 33). It was also assumed a
withdrawal of 100kJ of energy due to energy consumption and losses for every 20km of
electric line having as consequence a decrease of fire ignition probability along the electric
network.
Figure 33: Map for probability of fire ignition due to an electric cable failure.
If the user intends directly evaluate the fire ignition probability triggered by electric cable
failure due to a seismic event, it is necessary to apply a combined probability. In this use
case, the probability of fire ignition is the product of the probability of cable failure due an
earthquake of a certain seismic intensity distribution and the probability of having a fire
ignition due to a cable failure (Equation 6). Figure 34 shows the probability map of fire
ignition triggered by electric cable failure due to a seismic event.
)
[6]
31.08.2014 | 34
Figure 34: Map for probability of fire ignition by electric cable failure due to earthquake.
Supported by the information on the probability to have a fire ignition around each pole,
the user may simulate an ignition in a certain location according to the most likely spot to
have an ignition, the proximity of an important infrastructure or any other convenience. The
simulation may be performed accessing FireStation fire behaviour simulation tool in order
to simulate the spread of the fire and other important parameters (Figure 35). The
comparison of impacts caused by both earthquake and forest fire in a cascade effect event
may be an important information to support the decision making. The cascade effect tool is
used only to assess the probability of occurrence of a predefined event chain, in the
present case Earthquake
damage to electricity network (cable failure)
fire ignition.
The use of FireStation and the analysis of impacts are out of the use of cascade effect tool
but it is integrated in the general CRISMA tool where the user must return to perform this
last simulation.
(a)
(b)
(c)
Figure 35 – Map of the evolution of the forest fire (a) after the ignition by the electric discharge from a
probable ignition point and other possible outputs from the FireStation software: b) rate of spread, c)
linear intensity.
31.08.2014 | 35
4. Cascade effect model implementation
The cascade effect analysis has two stages (Figure 36): 1) setting of elements to
determine the transition probabilities, and 2) running of the cascade effect model to
determine the scenario probabilities. The settings are provided by the cascade effect view
(CEV), which is out of the scope of this deliverable. The cascade effect model, based on
the input provided by CEV will run to determine the transition probabilities and to produce
the transition probability maps in the area of interest.
Figure 36: Stages of use of the cascade effect model.
4.1. Logic of the sequences
As it was previously mentioned, the cascade effect model needs three types of inputs to
run: 1) event chain selected by the user in the CEV; 2) one transition probability data,
selected or created in the CEV, for each transition of the chain; and 3) geo-referenced
information about the world state.
When the cascade effect model is appealed, the chain of events is selected and the
TPD(s) are designated by the user; the required input world state data is automatically
read by the cascade effect model. Each transition requires different information of the
world state and the CEM will upload it automatically. For the same transition, more than
one TPD may be available for different damage states. For example the transition
“earthquake
Damage to electricity network” may be related to the fall of a pole or to a
cable failure. The user shall select that of interest and to edit it if necessary.
In order to generalize the format of the transition probability inputs, the TPD have a text
format with three columns of data respectively named “triggering hazardintensity
measure“, “triggered hazard intensity measure” and “probability of transition”, which is the
conditional probability of having the intensity measure of the triggered event given the
intensity of the triggering event.
When all the setting data is available and the TPD is chosen for each transition, the user
may run the CE model many times as needed. After the running, the CE model shall
produce a shape file to be ready to be exported to the cascade effect view building block.
In this document, different examples of outputs of the CE model were presented.
31.08.2014 | 36
4.2. Cascade effect model code
The code produced is a standalone software that needs the world state information
provided independently. In this case, this information will be provided by the general
CRISMA platform. The code of the developed software is based on the algorithm
presented in APPENDIX (3.
Figure 37: General overview of the CE Map transformation mechanism.
The CE Map Transformation mechanism (CEMT, Error! Reference source not found.)
consists of a Web Processing Service (WPS) and a Web Map Service (WMS) server. The
server, through the use of WPS has the ability to run processes, which can perform
operations on the input data. The WMS allows it to store and publish the data that results
from the operations performed. The input data is comprised of shape files, event chains
and probability function transformation matrices (PFTM), which are stored in an external
data repository. The output data is returned to the CE tool by the server in the form of a
rendered image representing the transformations performed on the input data.
The PFTMs contain structured data about the probability of an event triggering another
event in the chain, represented in the form of three columns in a multi-row layout. As an
example, the PFTM representing the event of electricity poles collapsing in case of the
occurrence of an earthquake, the first column in the PFTM determines the intensity of the
triggering event in the event chain; the second column specifies the intensity of the
triggered event (in this specific case a Boolean variable indicating the event occurrence),
and finally the third column specifies the probability of occurrence of a determined effect
on the world.
In order to perform a transformation on a set of data, the server must receive a WPS
HTTP GET request from the CE Tool specifying the location and the names of the data to
be used in the transformation process (shape files, event chains and transformation
matrices). The server launches a WPS process that is responsible for performing the
desired transformations on the world state information regarding the chain of events being
31.08.2014 | 37
analysed. It is also responsible for storing the resulting map on the WPS/WMS server,
which returns a rendered image to be displayed to the user.
Figure 38: WPS process in detail.
The WPS process (Error! Reference source not found.) is launched upon the reception
of a WPS HTTP GET request in the WPS/WMS server. It retrieves and opens a shape file
located in the data repository designated in the request. The data contained in the shape
file is extracted and transformed taking into account the data present in the transformation
matrix corresponding to the current event chain. The data that results from the
transformation process is inserted in a new shape file and then rendered to a specified
image format. Finally, the WPS/WMS server returns the resulting image to the client CE
tool for graphical visualization of the results.
31.08.2014 | 38
5. Final Remarks
This deliverable describes the cascade effect model and its implementation in a multi-risk
assessment scheme for the CRISMA platform. Applicative examples from Pilots B and D
are presented. Two different transitions for each of the two pilots were described and the
required input data was presented. The software code of the CE model was produced and
described. This model covers two stages: After the user selects the events chain of
interest, in the first stage an appropriate setting up of the system is performed by loading
the required preliminary information for the correct functioning of the system (namely data
from the world state and the transition probability data). The second part of the CE model
is referred to the running of the system, applying the respective TPD to the geo referenced
data of the world state and producing a spatial distribution of the probabilities of transition
for the area of interest.
Using the cascade effect tool, the user may assess the probability of occurrence of a preselected cascade event chain and plot its spatial distribution in a map. Based on this
information, the user may select a location to simulate the occurrence of the triggering
event and then evaluate the impacts resulting from the hypothetical occurrence of that
event and its likely sequence of triggered events. In this perspective, the information
provided by the cascade effect model is of great relevance in the decision making both for
planning activities and for response actions.
The integration of the CE model in the CRISMA software is the next step of this process.
The link between both pieces of software must take into account the compatibility of the
tools. In one side, the CE model is available to read the world state input data. On the
other side, the geo referenced output data produced must be read by the general CRISMA
software.
31.08.2014 | 39
6. List of References
Akaike H (1974). A new look at the statistical model identification. IEEE Trans Automat Contr AC
19:716–723.
Almeida, M.; Reva, V.; Viegas,D. X.; Garcia-Aristizabal, A.; Polese, M.; Zuccaro, G.; Agnes, C.;
Coulet, C.; Cossalter, A.; Pilli-Sihvola, K; Poussa, L; Molarius, R. (2013). Database and Model for
Dynamic scenario assessment V1. Deliverable D42.2 of the European Integrated Project CRISMA,
FP7-SECURITY- 284552.
Delvossalle, C. (1996). Domino Effects Phenomena: Definition, Overview and Classification. First
European Seminar on Domino Effects. Leuven. Belgium, FederalMinistry of Employment, Safety
Administration, Direction Chemical Risks, Brussels,Belgium, pp. 5-15.
Garcia-Aristizabal, A. & Marzocchi, W. (main Authors) (2012). Review of existing procedures for
multi-Hazard assessment. Deliverable D3.1 of MATRIX (New methodologies for multi-hazard and
multi-risk assessment methods for Europe) project. Contract No. 265138
Garcia-Aristizabal, A.; Polese, M.; Zuccaro, G.; Almeida, M.; Reva, V.; Viegas,D. X.; Rosqvist, T.,
Porthin, M. (2013). Dynamic scenario concept models. Deliverable D42.1 of the European
Integrated Project CRISMA, FP7-SECURITY- 284552.
ISO/Guide 73:2009(en) Risk management – Vocabulary
https://www.iso.org/obp/ui/#iso:std:iso:guide:73:ed-1:v1:(en)
Marzocchi, W.; Mastellone, M.L.; Ruocco A.Di.; Novelli, P.; Romeo, E.; Gasparini, P. (2009).
Principles of multi-risk assessment. Interaction amongst natural and man-induced risks. Project
Report, FP6 SSA Project: Contract No. 511264
Marzocchi, W., A. Garcia-Aristizabal, P. Gasparini, M. L. Mastellone, and A. Di Ruocco (2012).
Basic principles of multi-risk assessment: a case study in Italy, Nat. Hazards, 62(2), 551-573 DOI:
10.1007/s11069-012-0092-x
Oxford Diccionary at 26/June/2014 – http://www.oxforddictionaries.com/definition/english/damage
Plaintiff (2013). The 2009 Victoria Bushfires Royal Commission Final Report – Opening
Submissions. Prepared by Maurice Blackburn Lawyers. Filled on behalf of The Plaintiff.
http://www.royalcommission.vic.gov.au/Commission-Reports/Final-Report.html
Porter, K.; Kennedy, R.; Bachman, R. (2007). Creating Fragility Functions for Performance-Based
Earthquake Engineering, Earthquake Spectra, Volume 23, No. 2, pages 471–489.
Reniers, G., Dullaert, W., Soudan, K. (2004). A Domino Effect Evaluation Model, University of
Antwerp, Faculty of Applied Economics.
Vamvatsikos, D.; Cornell, A. (2002). Incremental dynamic analysis. Earthquake Engineering
Structure Dynamics. 31 (3), 491-514.
31.08.2014 | 40
APPENDIX (1) Description of chain blocks identified for possible
cascade event chains
Figure 39: Diagram of cascade event chains identified for the occurrence of an earthquake.
31.08.2014 | 41
Table 4: Description of chain blocks identified for possible cascade event chains after an earthquake.
Block name
Dam 1 – Damage to road/rail
transport infrastructure
Dam 2 – Building damage
Dam 3 – Damage to gas (pipeline)
network
Dam 4 – Damage to electricity
network
Dam 5 – Damage to water (river)
network
Dam 6 – Damage to industrial
facilities
Dam 7 – Damage to drinking water
network
Dam 8 – Damage to waste water
network
Dam 12 – Damage to structural
protection (dam/dike)
Landslide
Flood
Disability of transport
infrastructures
Explosion
WUI/ urban fire
Forest fire
Industrial fire
Release of gas/ liquid flammable
substances
Long term power supply
interruption
Release of chemical substances
Water contamination
Soil contamination
Tsunami
Description
The roads and rails infrastructures including bridges can be damaged
or destroyed by the horizontal actions due to earthquake.
Horizontal actions produced by the earthquake can damage or
destroy the buildings
Ground accelerations may cause pipelines failure with consequent
leakage of gas by provoking possible fire ignitions
All the structural elements for the transmission of electrical power
(both overhead and underground power cables ) as well as power
plants and electrical substations located near demand centers can be
affected by the earthquake by causing possible blackouts
River diversions may occur due to movements of the ground
Industrial equipment and systems can suffer structural damage when
hit by earthquakes, so that accidental events as fire, explosion and
dispersion of toxic substances can take place.
Ground accelerations may cause pipelines failure with consequent
leakage of water in the soil
Earthquake may cause significant damage to wastewater network
(also including the pumping stations and the treatment plants) and as
a result causes water and soil pollution as well as difficulties for
residents of affected urban areas
Both natural and artificial structures for the containment of river can
be damaged or destroyed by the earthquake
Due to earthquake, the ground can be overloaded by water. This can
cause a landslide when the slope is important.
Landslides as well as damages to dikes can provoke the overflow of
water with submersion of the neighboring areas
Partial or total collapse of buildings facing the roads/railways may
cause disability to transport infrastructures
Damage to gas network as well as to industrial facility that make use
of inflammable material could initiate an explosion
Damage to gas network could initiate an ignition
Fire caused by explosions or short circuits that could affect the
electricity network may spread to forest areas
Industrial facilities that make use of inflammable material could
initiate an ignition
Damages to the stocks in industrial facilities as well as failure to the
gas pipeline due to earthquake could generate the release of several
kinds of flammable substances
Severe damages to the power grid may cause a long term power
supply interruption (people trapped in the lifts, surgery problems, etc.)
Damages to the stocks in industrial facilities due to earthquake could
generate the release of several kinds of dangerous/pollutant chemical
substances
When a pipe is broken, some intrusions (salinity, bacteria or other)
can be observed in the water (both potable as non-potable)
When a pipe is broken, some intrusions (salinity, bacteria or other)
can be observed in the soil
Sometimes the earthquake cause a fast deformation of the seafloor
than it could happen that the overlying water is displaced vertically
and a tsunamis can be generated
31.08.2014 | 42
Figure 40: Diagram of flood cascade event chain.
31.08.2014 | 43
Table 5: Description of chain blocks identified for possible cascade event chains after a flood.
Block name
Dam 12 - Damage to
structural protection
(dike)
Dam 1 - Damage to
transport (road/rail)
infrastructure
Dam 2 - Building
damage
Dam 3 - Damage to
gas (pipeline) network
Dam 4 - Damage to
electricity network
Dam 5 - Damage to
water (river) network
Dam 6 - Damage to
industrial facility
Dam 7 - Damage to
drinking water network
Dam 8 - Damage to
waste water network
Dam 9 - Damage to
telecommunication
network
Dam 10 - Damage to
irrigation network
Dam 11 - Damage to
Agriculture
Landslide
infrastructure
Explosion
Release of chemical
substance
WUI / urban fire
Industrial fire
Water contamination
Soil contamination
Description
The pressure forces applied on the dikes due to high water level or strength of
waves can make them breach or collapse.
The water level is higher than dikes, therefore, the dikes are overtopped and can
breach.
The roads or rails infrastructures as well as traffic signals can be destroyed due to
high water velocities.
The flood can spread to building areas and they can be flooded or even
destroyed.
Floods could generate scour around pipeline support and by consequence
breaches in the pipe that can origin release of toxic gas plume, dust cloud or
gas/liquid flammable substances.
The vent holes can be flooded.
The production or transformation posts can be flooded. These kinds of failures can
totally or partially stop the service.
The electricity distribution system (power lines) can be destroyed.
High water discharges cause large river bed loads and sedimentations. These
discharges can also destroy the banks.
A flood could also force a closure of the industry and potentially some damages to
the stocks or the machinery.
The pipes can be flooded and intrusions (salinity, bacterium or other) can be
observed in the drinking water network.
Pumps can be destroyed due to the overload of water or just not working due to
electricity failure.
The transportation pipes can be flooded. The pumping stations can be destroyed
by the overload of water or just not working due to electricity failure. These effects
are followed by the spill of waste water outside the network.
The cables which are put in the core of dikes could be cut when the dikes breach.
The high water velocities can damage the irrigation network infrastructure.
Flooding of agricultural land could destroy plantings. In the case of marine flood,
this could generate a salinization of soils.
Due to flood, the ground can be overloaded by water. This can cause a landslide
when the slope is important.
The roads or rails can be flooded and thus impracticable.
Flooding of industrial facility could generate electric shortcuts which could initiate
an explosion.
Damages to gas network or destruction of industrial stocks with inflammable
material could initiate an explosion.
Flooding of industrial facility and damages to the stocks could generate release of
different kinds of chemical substances
Flooding of buildings could generate electric shortcuts which could initiate a fire.
Flooding of industrial facility could generate electric shortcuts which could initiate a
fire.
Intrusions (salinity, bacterium or other) can be observed in the water (both potable
as non-potable)
Intrusions (salinity, bacterium or other) can be observed in the soil
31.08.2014 | 44
Figure 41: Diagram of forest fire cascade event chain.
Table 6: Description of chain blocks identified for possible cascade event chains after a forest fire.
Block name
Smoke cloud
Disability of
transport
infrastructures
WUI/ urban fire
Industrial fire
Dam 2 - Building
damage
Dam 6 - Damage
to industrial
facilities
Forest fire
Description
Large forest fires, WUI/urban fires and Industrial fires normally release smoke
clouds with concentration and mass that can cause other adverse events such as
intoxication or disability of transport infrastructures.
Interruption of transport communication due to smoke cloud that reduce visibility or
due to fallen objects (trees, electricity pillows, etc.), flames and spotting (firebrand
projection).
Forest or industrial fires can spread to WUI zone (the physical space where
vegetation and structures coexist in a fire prone environment) and consequently to
urban zone. On the other hand, a forest fire can pass directly to an urban zone by
spotting (firebrand projection).
Forest or WUI/urban fires can spread to industrial zone and cause fire on industrial
facilities or storage materials.
Combustible building materials may be burned by the effect of the fire. Noncombustible materials may be thermal expanded leading to building damage or
even collapse.
Combustible industrial facility materials may be burned by the effect of the fire. Noncombustible industrial facility materials may be thermal expanded leading to
damage or even collapse.
WUI/urban or industrial fires can pass to forest fire directly by flame spreading or by
spotting.
31.08.2014 | 45
Figure 42: Diagram of extreme weather cascade event chain.
31.08.2014 | 46
Table 7: Description of chain blocks identified for possible cascade event chains for a case of
extreme weather conditions.
Block name
EW1 - Heat waves
EW2 - Cold waves
EW3 - Drought
EW4 - Strong winds
EW5 - Heavy rain
EW6 - Snow storm
EW7 - Lightning strike
Forest fire
Explosion
Collapse / leaning of
trees
Flood
infrastructures
Landslide
Dam 1 - Damage to
transport (road / rail)
infrastructure
Dam 4 - Damage to
electricity network
Dam 7 - Damage to
drinking water network
Dam 9 - Damage to
telecommunication
network
Dam 11 - Damage to
agriculture
Dam 12 - Damage to
structural protection
(dam/dike)
Description
Specific extreme weather conditions. Meteorological drought defined as
precipitation's departure from normal over some period of time (Assessment of the
Regional Impact of Droughts in Europe, p. 3)
Heat waves and droughts cause drying of plants and trees, and sparking forest
fires.
Lightning strikes can start a forest fire by bringing the wood to its flash point.
Under the condition of heat waves, overheating of chemical substances (or
combustible materials) may lead to explosion.
Collapse / leaning of trees due to strong winds or snow accumulated on the tree
crowns. It depends on the tree species, soil, and other characteristics.
Coastal flooding can be caused by strong winds blowing waves onto the land.
Floods caused by heavy rains near rivers, lakes, basins and sea.
Disability of transport infrastructure due to falling trees, which cause blocking of
transport infrastructures.
Disability of transport network due to heavy snowfall accumulating on the streets.
Due to heavy rain, the ground can be overloaded by water. This can cause a
landslide when the slope is important.
Cold waves cause damage to rail transport systems.
Strong winds can bring down power lines by damaging the poles.
Damage to electricity network due to both collapsed and leaning trees which
damage power lines causing power outages.
Damage to electricity network due to snow accumulating on power lines, causing
power outages.
Lightning strikes can damage fuses, transformers and other electricity distribution
systems.
Prolonged cold waves cause freezing of water pipes.
Damage to drinking water network due to power outages in water delivery plants.
Damage to telecommunication network due to power outages in mobile telephone
base stations
Damage to crops due to drought
The pressure forces applied on the dikes due to high water level or strength of
waves can make them breach or collapse.
The water level is higher than dikes, therefore, the dikes are overtopped and can
breach.
31.08.2014 | 47
Figure 43: Diagram of cascade event chain for release of chemical substance.
Table 8: Description of blocks of cascade event chain identified for possible cascade event chains
after a release of chemical substances.
Block name
RSC - Release
of chemical
substance
RSC1 - Release
of toxic gas
plume/dust
cloud
RSC2 - Release
of gas / liquid
flammable
substances
RSC3 - Release
of liquid / solid
substance
Water
contamination
Soil
contamination
Description
Loss of containment due to any type of event. Chemical substance may be in
gaseous, liquid or solid form.
Release of specific chemical substance. Chemical substance releases in gaseous
form or dust to the air forming gas plume/dust cloud. The dispersion of the
plume/cloud depends on the meteorological circumstances and the physical properties
of the released substance.
Release of specific chemical substance. Flammable chemical substance releases in
gaseous or liquid form. The released chemical ignites can cause fire (jet fire, pool fire)
leading to forest fire, WUI/urban fire or industrial fire. In case of flammable gas release
an explosion can occur.
Release of specific chemical substance. Chemical substance releases in liquid or solid
form. Release causes soil contamination and/or water contamination depending on the
release point and environment. Liquid chemical may evaporate from the chemical pool
on the surface of the soil and form toxic or flammable gas cloud.
Chemical substance may dissolve, mix or react with water causing pollution of water.
Contamination refers to the presence of harmful or toxic chemicals in water.
Chemical substance may soak into the soil. Contamination refers to the presence of
harmful or toxic chemicals in soil. Contaminated soil may cause ground water / water
contamination.
31.08.2014 | 48
APPENDIX (2) Main parameters of the laboratorial tests for the
determination of ignition by industrial electric discharge.
Reference
Fuel moisture
content (%)
Voltage
(kV)
Time to
ignition (s)
Energy to
ignition (kJ)
140311_LAT01
140311_LAT02
140311_LAT03
140311_LAT04
140311_LAT05
140311_LAT06
140311_LAT07
140311_LAT08
140311_LAT09
140311_LAT10
140311_LAT11
140311_LAT12
140311_LAT13
140311_LAT14
140311_LAT15
140311_LAT16
140311_LAT17
140311_LAT18
140311_LAT19
140311_LAT20
140311_LAT21
140311_LAT22
140311_LAT23
140311_LAT24
140311_LAT25
140311_LAT26
140311_LAT27
140311_LAT28
140311_LAT29
140311_LAT30
140312_LAT01
140312_LAT02
140312_LAT03
140312_LAT04
140312_LAT05
140312_LAT06
140312_LAT07
140312_LAT08
140312_LAT09
140312_LAT10
140312_LAT11
140312_LAT12
140312_LAT13
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
78.0
78.0
78.0
78.0
78.0
110.0
95.0
95.0
95.0
90.0
120.0
120.0
130.0
130.0
118.0
140.0
150.0
145.0
125.0
125.0
120.0
120.0
120.0
120.0
120.0
109.0
104.1
109.0
99.1
99.1
94.1
94.1
94.1
94.1
89.2
109.0
109.0
114.0
109.0
109.0
79.3
79.3
84.2
2.00
2.12
2.36
3.00
2.24
2.16
2.08
1.48
2.00
1.60
1.76
1.48
1.68
1.56
1.80
1.64
1.72
2.32
2.64
2.32
1.76
1.92
2.16
1.96
1.48
2.16
1.92
1.96
3.08
2.48
1.92
2.32
1.00
1.12
2.12
2.00
2.08
2.16
1.76
1.68
2.28
2.40
1.28
600
636
708
900
672
648
624
444
600
480
528
444
504
468
540
492
516
696
792
696
528
576
648
588
444
648
576
588
924
744
576
696
300
336
636
600
624
648
528
504
684
720
384
31.08.2014 | 49
Reference
Fuel moisture
content (%)
Voltage
(kV)
Time to
ignition (s)
Energy to
ignition (kJ)
140312_LAT14
140312_LAT15
140312_LAT16
140312_LAT17
140312_LAT18
140312_LAT19
140312_LAT20
140312_LAT21
140312_LAT22
140312_LAT23
140312_LAT24
140312_LAT25
140312_LAT26
140312_LAT27
140312_LAT28
140312_LAT29
140312_LAT30
140312_LAT31
140312_LAT32
140312_LAT33
140312_LAT36
140312_LAT37
140213_LAT01
140213_LAT02b
140213_LAT03b
140213_LAT04c
140213_LAT05b
140213_LAT06b
140213_LAT07c
140213_LAT08b
140213_LAT09d
140213_LAT10b
140213_LAT11
140213_LAT12
140213_LAT14b
140213_LAT15
140213_LAT16b
140213_LAT17b
140213_LAT18b
140213_LAT19b
140213_LAT21
140213_LAT22b
140213_LAT23b
140213_LAT24
140213_LAT25
140213_LAT26b
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
9.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
84.2
84.2
84.2
84.2
84.2
84.2
84.5
84.5
79.5
74.6
74.6
74.6
74.6
99.4
109.3
109.3
109.3
109.3
124.3
119.3
119.3
149.1
98.9
94.0
94.0
84.1
82.9
80.9
77.0
77.0
78.0
104.1
109.0
111.0
123.9
109.0
143.7
143.7
143.7
138.7
153.6
143.7
153.6
158.6
158.6
208.1
2.36
2.28
1.80
1.28
1.72
1.04
1.80
3.44
1.56
1.44
2.68
2.40
1.68
1.52
2.36
2.56
1.72
1.88
1.68
1.60
2.00
2.00
1.08
1.79
0.38
1.29
1.50
0.96
2.88
1.67
2.50
2.04
1.83
1.75
1.67
1.88
1.71
1.46
1.71
0.42
1.29
1.71
1.75
1.79
1.50
0.92
708
684
540
384
516
312
540
1032
468
432
804
720
504
456
708
768
516
564
504
480
600
600
325
663
663
1363
1075
763
1713
938
2338
1138
550
525
1000
563
513
588
638
625
388
1075
1038
538
450
775
31.08.2014 | 50
Reference
Fuel moisture
content (%)
Voltage
(kV)
Time to
ignition (s)
Energy to
ignition (kJ)
140213_LAT27b
140213_LAT28
140214_LAT01b
140214_LAT02
140214_LAT03
140214_LAT04b
140214_LAT04c2
140214_LAT05
140214_LAT06c
140214_LAT07b
140214_LAT08b
140214_LAT09
140214_LAT10b
140214_LAT11b
140214_LAT12
140214_LAT13b
140214_LAT14b
140214_LAT15
140214_LAT16c
140214_LAT17b
140214_LAT18b
140214_LAT19
140214_LAT20
140214_LAT21
140214_LAT22b
140214_LAT23
140214_LAT24
140214_LAT25b
140214_LAT26b
140214_LAT27b
140214_LAT28
140214_LAT29b
140214_LAT30
12.2
12.2
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
10.6
213.1
213.1
207.9
212.9
212.9
212.9
212.9
212.9
159.0
159.0
159.0
159.0
159.0
119.3
119.3
119.3
109.3
119.3
99.4
94.4
94.4
94.4
94.3
74.5
71.5
73.5
72.5
71.5
74.5
72.5
79.4
72.5
79.4
1.46
1.54
1.79
2.13
1.13
1.38
0.46
1.88
1.63
0.63
0.75
1.50
0.88
1.08
2.08
1.00
0.50
2.38
2.33
2.21
1.42
1.96
2.54
2.29
0.83
2.00
1.88
0.46
0.42
0.58
1.92
0.63
1.46
775
463
763
638
338
775
663
563
1313
663
700
450
763
875
625
775
613
713
1863
1275
1063
588
763
688
925
600
563
650
850
613
575
625
438
31.08.2014 | 51
APPENDIX (3) Algorithm for the model implementation
The following pseudo code shows the algorithm handling the I/O information necessary to perform
calculations of cascade effects scenarios.
A.1 Used variables
VARIABLE
DESCRIPTION
EVENT(i)
Name of the ith event in the chain of events of the selected scenario
Flag_KindData(i)
A variable identifying the kind of data representing the i th event. It can take
values among: ‘IM’ (if it represent the intensity measure of an event); or
‘TPD’, if the data in this node is a transition matrix (e.g. conditional
probabilities).
IM_DATA(i)
Spatial distribution of the Intensity measure
TPD(:,:,i)
Transition probability data (conditional probabilities). This variable is a
multi-dimensional matrix according with the IM thresholds of both the
triggered and triggering events. It has 3 columns: triggering event intensity
measure, triggered event intensity measure and probability of transition.
IMTMTriggering(:,i)
Intensity measure of the triggering (IMtriggering) event associated with the
probability values of the TPD(:,:,i)
IMTMTriggered(:,i)
Intensity measure of the triggered (IMtriggered) event associated with the
probability values of the TM(:,:,i)
N
Number of nodes of the selected scenario
IMevent(:,i)
Intensity values of the event in the previous node of the chain. This variable
is used to keep track of the intensities of the triggered events associated
with transition matrix used in TM(:,:,i)
Ppath(i)
This variable is used to store the probability value used at the ith node of the
chain for the calculation of the scenario occurrence probability
Pscenario(i)
Variable used to store the value of the scenario occurrence probability at a
given location in the calculation grid
A.2 Reading the scenario database
1. READ input data from the world state (building inventory, road network, etc).
2. READ database of cascade effects scenarios, and select a path of interest for calculations:
a. Select a triggering event form the list:
i. EVENT(1) = Selected triggering event
b. Show the possible triggered events. User selects the triggered event of interest, and
do it recursively up to the end of the scenario:
i. EVENT(2) = selected triggered event from the list;
ii. EVENT(3) = selected triggered event (by the EVENT(2));
iii. EVENT(i) = selected triggered event (by the EVENT(i-1));
3. Identify the kind of data at each element of the scenario:
a. FOR each element of the selected scenario DO:
31.08.2014 | 52
i. Read the ith element of the scenario
ii. IF EVENT(i) is the “triggering event”, THEN:
1. Flag_KindData(i) = ‘IM’ (intensity measure)
2. IM_DATA(i) = spatial distribution of the Intensity measure of the
triggering event
3. SET: TPD(i) and IMTM*(i) to: “ND” (no data)
iii. IF EVENT(i) is a “transition probability data”, THEN:
1. Flag_KindData = ‘TPD’ (transition probability data)
2. TPD(:,:,i) = Conditional_Probability IMtriggered(k) | IMtriggering(m)
3. IMTMTriggering(:,i) = Intensity measure of the triggering (IMtriggering)
event associated with the probability values of the TPD(i)
4. IMTMTriggered(:,i) = Intensity measure of the triggered (IMtriggered)
event associated with the probability values of the TPD(i)
A.3 Perform calculations
The following calculations are performed at each point of the calculation domain (i.e. they are
referred to a specific grid element of the calculation domain)
1. FOR i ranging from 2 TO N (N=the number of nodes of the scenario; nothe that i=1 will be
always associated with the triggering event whose kind of data will be a IM distribution):
a. When i=2, then the event in the previous node is the triggering event, and therefore
we set:
i. IF i==2, THEN: IMevent(:,i-1) == IM_DATAi-1(location) in the particular grid
element of the calculation domain (location).
b. Flag_KindData(i) is ‘TPD’:
i. Find the probability of the triggered event in node i, given the intensity of the
triggering event in node i-1:
1. j = FIND IMTMTriggering(:,i) == IMevent(:,i-1)
2. Ppath(i-1) = TM(j,:,i)
3. IMevent(:,i) = IMTMTriggered(:,i)
2. Calculate the probability of the selected scenario (for the grid element of the domain under
analysis), as the product of all the elements in the variable Ppath:
a. Pscenario = PRODUCT[Ppath(:)]
b. Calculate expectations (e.g., number of collapsed buildings), according with the
building inventory read from the world state.
OUTPUT: Pscenario, for the specific location of the calculation domain under analysis (e.g. point X,
Y). Running this code for the whole calculation domain will produce an output map with the spatial
31.08.2014 | 53
distribution of the scenario occurrence probability. PXYscenario(X,Y) = Pscenario, calculated in location X,
Y. Another possible output is eventually the expectation values (if calculated).
A.4 Post processing:
Plotting the probability map: PXYscenario(X,Y)
31.08.2014 | 54
APPENDIX (4) – Example of the logic sequence for the Pilot D use
cases
Use case 1: Triggering earthquake 1
Triggered seismicity (EQ
EQ)
0. PREVIOUS INFORMATION
0.1. It is assumed that the database has been created and uploaded with the transition probability
data (EQ/EQ)
0.2. Time dependent vulnerability is available
1. INFO IN THE WORLD STATE 1 – after the EQ1
1.1. Triggering event: EQ1
1.1.1. Map of intensity measure (ex: PGA) distribution (shape file F1)
1.1.2. Updated building inventory (F2)
1.1.3. Updated road network (F3)
1.1.4. Updated population distribution (F4)
1.2. Hazard assessment associated to triggered seismicity
1.2.1. Updated building inventory (F2’)
1.2.2. Updated road network (F3’)
1.2.3. Updated population distribution (F4’)
RUN
1.
2.
3.
4.
5.
6.
7.
8.
9.
Read database
Select triggering event – load F1, F2 and F3
List the possible triggered event (from database)
Choose the secondary event of interest (in this example EQ)
Load the shape file of transition probability data EQ/EQ
The User selects the exceedance probability of interest
Calculate probability of damage and update the building inventory (F2’)
If of interest do the same calculation for road network, population…..
Update the new world state
31.08.2014 | 55
Use case 2: Earthquake
FI)
Damage to electricity network
Fire ignition (EQ
DEN
0. PREVIOUS INFORMATION
0.1. It is assumed that the database has been created and uploaded with the transition probability
data (EQ/EQ)
0.2. Time dependent vulnerability is available
1. INFO IN THE WORLD STATE 1 – after the EQ and before the FI
1.1. Triggering event: EQ
1.1.1. Map of intensity measure (ex: PGA) distribution (shape file F1)
1.1.2. Updated building inventory (F2)
1.2. Damage to electricity network
1.2.1. Map of the poles (shape file F3)
1.2.2. Fragility of the poles (one text file F4)
1.2.3. Properties of the power line(s): voltage + intensity (one text file F5 )
1.3. Fire ignition
1.3.1. Interactive selection of the pole(s) – USER action
1.4. Forest fire
1.4.1. Fuel map (shape file F6)
2. INFO IN WORLD STATE 2 – after the forest fire ignition
2.1. Fire spread map (raster image F9)
2.2. Intensity map (raster image F10)
2.3. Rate of spread map (raster image F11)
2.4. Smoke release map (raster image F12)
RUN
10. Read database
11. Select triggering event – load F1 and F2
12. List the possible triggered event (from database)
13. Choose the secondary event of interest (in this example DEN)
14. Load the map of the poles
15. Calculate the probability of cable failure P(CF|EQ)
16. Use the properties of the power line and the TPD of fire ignition to calculate P(FI|CF). The final
probability of ignition P ig=P(FI|CF)*P(CF|EQ).
17. The values P(CF|EQ), P(FI|CF) can be new attributes of the shape file of the map of poles and
the User may select which of them want to visualize on the map.
18. The User selects the pole(s) of interest for the ignition
19. Return to CRISMA general software, load the inputs for the FireStation runs and execute it
20. Read the outputs of interest
21. Calculate the impacts of EQ+FF
22. Update the new world state

Database and Model for Dynamic scenario assessment V2

Transcription

Similar documents

silicon valley

Melihat bencana dari kacamata anak yang lucu (gambar kacamata

- Included with Park Millennium Cable TV Package

now, even more reasons to buy your ewp from a boise cascade dealer

woodscapes sherwin williams

1100 sr pdf.wps - Katana Owners Club UK

2014 TBL Report - Cascade Engineering

ignition - Curtin Business School

Building on Innovation

PRE-CUT VERSA-LAM STAIR STRINGERS