- Email: [email protected]
Static and dynamic resilience of transport infrastructure and demand: the case of the Athens metro
Available Available online online at at www.sciencedirect.com www.sciencedirect.com
ScienceDirect
Available online at www.sciencedirect.com Transportation Transportation Research Research Procedia Procedia 00 00 (2016) (2016) 000–000 000–000
ScienceDirect
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
Transportation Research Procedia 24C (2017) 459–466 www.elsevier.com/locate/procedia
3rd Conference on Sustainable Urban Mobility, 3rd CSUM 2016, 26 – 27 May 2016, Volos, Greece
Static and dynamic resilience of transport infrastructure and demand: the case of the Athens metro a*, Efthymia Apostolopouloubb Alexandros Deloukasa* a, a, b b
Attiko Attiko Metro, Metro, 191-193 191-193 Messogion Messogion Ave., Ave., 115 115 25 25 Athens, Athens, Greece Greece
Abstract Abstract The resilience engineering engineering of of the the urban urban transport transport system system (UTS) (UTS) using using the the example example of of the the Athens Athens metro. metro. The paper paper conceptualises conceptualises the the resilience Resilience Resilience management management guidelines guidelines for for the the UTS UTS have have been been developed, developed, whereas whereas Athens Athens metro metro pilots pilots validate validate the the former. former. The The pilots pilots are are concerned concerned with with the the metro metro attack attack risk risk and and its its consequences. consequences. The The risk risk scenarios scenarios consider consider the the static static and and the the dynamic dynamic resilience resilience of of the the UTS UTS system system with with respect respect to to metro metro threats. threats. The The identified identified actions actions have have measurable measurable impacts impacts improving improving the the UTS UTS resilience resilience in in Athens. Athens. © 2016 The Authors. by Elsevier B.V. © Authors. Published Published © 2016 2017 The The Authors. Published by by Elsevier Elsevier B.V. B.V. Peer-review under responsibility of the organizing committee of the 3rd CSUM 2016. Peer-review under responsibility of the organizing Peer-review under responsibility of the organizing committee committee of of the the 3rd 3rd CSUM CSUM 2016. 2016. Keywords:resilience; guideline; Keywords:resilience; guideline; metro metro risk; risk; vulnerability; vulnerability; disruptive disruptive attack attack Corresponding author. author. E-mail E-mail address. ** Corresponding address. [email protected] [email protected]
1. Introduction The EU-funded project ‘RESilience management guidelines and Operationalization appLied to Urban Transport Environment (RESOLUTE)’ addresses critical components of urban transport systems (UTS) in view of safety and security needs. The H2020 project aims the development of European resilience management guidelines (ERMG) adapted to UTS. The ERMG validation will be executed i.a. through specific Athens metro pilots. Resilience is an emergent property of the UTS function and is not reduced exclusively in risk analysis terms (PARK et al., 2013). Resilience is a broader term embracing the ability of UTS (or other CIs) to perform four functions with respect to disruptions: a. Planning and preparation, b. Absorption, c. Recovery, d. Learning and adaptation. We enrich the risk concept, aiming not minimization of an objective disruption probability but mitigation of uncertain disruption (risk) consequences due to unforeseen events. RESOLUTE follows follows the the resilience resilience management management approach approach developed developed by by Erik Erik Hollnagel. Hollnagel. This This modular modular approach approach RESOLUTE aims a description of the functions (activities) of the system under investigation (e.g. UTS) as well as the couplings aims a description of the functions (activities) of the system under investigation (e.g. UTS) as well as the couplings (dependencies) rather than than outcome-based outcome-based performance performance (dependencies) between between the the functions. functions. The The approach approach promotes promotes processprocess- rather metrics. The study conceptualises the resilience management of the urban transport system using the example of the the metrics. The study conceptualises the resilience management of the urban transport system using the example of Athens metro. metro. The The development development of of resilience resilience guidelines guidelines for for the the UTS UTS has has been been finalised, finalised, whereas whereas their their validation validation Athens through the the Athens Athens pilots pilots is is aa subsequent subsequent cycle. cycle. through Resilience is is defined defined as as the the ability ability of of the the urban urban transport transport system system to to withstand withstand aa major major disruption disruption within within acceptable acceptable Resilience degradation parameters parameters and and to to recover recover within within an an acceptable acceptable time. time. There There is is no no abstract abstract [UTS] [UTS] resilience resilience per per se se but but only only degradation with respect to a known or unknown threat. The paper acknowledges a multi-layer resilience of the UTS, considering with respect to a known or unknown threat. The paper acknowledges a multi-layer resilience of the UTS, considering
2352-1465 © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the organizing committee of the 3rd CSUM 2016. 10.1016/j.trpro.2017.05.082
460
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
an interplay of the physical, service and human layer. The Athens pilots focus on disruptions in the service layer (static resilience) and the human layer (dynamic resilience). 2. Development of Resilience Guidelines for UTS 2.1 FRAM Concept The Functional Resonance Analysis Model (FRAM) concept, developed by E. Hollnagel (HOLLNAGEL, 2012), has been applied to simulate complex socio-technical systems such as the UTS. The FRAM concept recognises that in order to study efficiently a complex socio-technical system, it is more important to understand in detail the system dynamics and the variability of performance than to know only what has gone wrong, studying previous system failures. The FRAM concept is based in four principles. These are: (1) the principle of equivalence of failures and successes, which focuses on the fact that it is equally or even more important to study the operation of complex systems and their components that function properly than to study only the parts of the system; (2) the principle of approximate adjustments, which is based on the variability of day-to-day performance, both human and organisational; (3) the principle of emergence, which denotes that in many cases in complex systems it is not possible to explain what happens as a result of known processes (applying the decomposition and causality principles), but effects are rather emergent than resultant and the underlying system is only partly intractable; (4) the principle of functional resonance, which reflects the fact that the dynamic dependencies among the functions of a system must be accounted for as they develop in a specific situation. In complex socio-technical systems, these dependencies can be far more complex than simple cause-effect relations. 2.2 Resilience Analysis Grid The Resilience Analysis Grid (RAG), also developed by E. Hollnagel, is a way to assess a system’s potential for resilience. The resilience of a complex socio-technical system such as the UTS can be described by its abilities to: (1) anticipate potential threats or opportunities for changes; (2) respond to regular or irregular disruptions by adjusting functioning to existing conditions; (3) monitor both the system and the environment for what could become a threat in the immediate time frame; (4) learn from experiences of both successes and failures. In an effort to assess the potential for resilience of the UTS, these four capabilities should be considered and from that basis a RAG can be developed, i.e. four sets of questions where the answers can be used to construct the system’s resilience profile. Some examples of questions for each of the four system abilities follow: (Anticipating) How long is the organisation willing to look ahead?. (Responding) For which events is there a response ready? How soon can a response be given?, (Monitoring) What are the key performance indicators? How, when and why are they revised?, (Learning) Is learning based both on successes and failures? These sets of questions are answered with the support of group discussions with subject matter experts. Depending on the answers given, ratings for all four resilience abilities of the UTS (deficient, unacceptable, acceptable, satisfactory or excellent) will give the resilience profile of the UTS according to the RAG. 2.3 Generic Guidelines (functions) The workflow to produce resilience guidelines has been organized in steps. First, data have been collected about how the various Critical Infrastructures (CIs), such as the UTS, work in general (daily operation), through stakeholder interviews (at workshops and through face to face interviews). Then, the high-level functions present in all types of CIs have been identified. In total, 26 high-level functions were identified and described. Six main types of high-level functions were defined, shown in Figure 1. It is noted that a function in FRAM terms constitutes an activity.
2
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
461
Sustained adaptability
Anticipate Monitor
Respond Learn
Boundaries
Core functions
Figure 1. Sustained adaptability
The functions are characterised and categorized (Listed sample of functions, see Note 1). Anticipating functions: This type includes all functions performed when planning for operation of CIs, such as (a) Develop a Strategic Plan, (b) Manage awareness and user behaviour, (c) Train staff. Responding functions: These are functions performed following a system disruption, such as (a) Restore/repair physical infrastructure, (b) Restore/repair operations, (c) Coordinate emergency actions. Monitoring functions: These functions have been defined so that they are flexible enough to increase or relax the data collections according to a potential or ongoing event. Examples of monitoring functions are: (a) Monitor resource availability, (b) Monitor safety and security, (c) Monitor user generated feedback. Learning functions: (a) Collect event information, (b) Provide adaptation and improvement insights. There are also boundary functions (i.e. functions determining the system’s boundaries): (a) Supply Resources, (b) Provide risk warning, (c) Fight the emergency, (d) Regulate domain and operations, as well as core functions, which express objectives central to the UTS, with many interdependencies with other functions, such as (a) Deliver service, (b) Use of the service. In the next step, a function description template has been filled in for each function (Function Description Template, see Note 2). The prevalent function type (Human, Technology or Organisation) has been determined. The interdependencies of a specific function with other functions have also been defined at this step. The six aspects of each function are: input, output, resources, preconditions, control and time. Interdependencies with other functions have been specified for each function by answering questions such as: “What start the function?”, “What does the function act on or change?” (input questions), “What are the function’s main outputs?” (output question), “What are the required resources for the function?” (resource question), “What controls the function?”, e.g. regulations and Key Performance Indicator (KPIs), stakeholders, organisations (control question) and “Are there any time implications for the function?” e.g. deadlines (time question). In a following step, common conditions or sources that affect the performance variability of each function have been specified and functions are linked together (functions’ couplings). Eleven common performance conditions were identified and evaluated for CIs such as the UTS: (1) Availability of resources; (2) Training and experience (competence); (3) Quality of communication; (4) Human-Computer interaction and operational support; (5) Availability of procedures and plans; (6) Conditions of work; (7) Number of goals and conflict resolution; (8) Available time and time pressure; (9) Circadian rhythm and stress; (10) Team collaboration quality; (11) Quality and support of the organisation. Then, a set of questions is assigned to each function for all common conditions applicable to the function. In the
3
462
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
end, a template has been filled in for each Generic Guideline (function). The template contains a guideline abstract, some background facts about the function, the recommended actions and the associated questions that led to the recommended actions. The recommended actions are provided in order to dampen the performance variability for the common conditions (Generic Guideline Template, see Note 3). 2.4 Metro Emergency Preparedness - Emergency Rules and Procedures The Emergency Procedures constitute the corner-stone of the Safety Plan of an organization for addressing an emergency or crisis. Emergency rules and procedures in the Athens metro network include i.a. OCC response to incidents, Fire fighting, Train evacuation in tunnel, Bomb threat, Controlled smoke channelling with ventilators, Station evacuation, Major incidents, Bomb blast – gas attack - Drill exercises There is a Fire Drill Plan, which simulates an incident and the related necessary response actions. According to the exercise, a fire breaks out in a train entering a station, on the platform level. The general purpose of the drill is to address fire incidents in metro stations in the best possible way. At the same time, the drill also serves to evaluate the capabilities of the local Fire Stations and improve the cooperation between Fire Service and the supporting agencies involved. Tangible bodies potentially contributing to resilient UTS in Athens are the General Secretariat for Civil Protection (CP) and its Operational Centre 199 SEKYPS, Region of Attica CP, Police (GADA), EMAK Rescue Team, EKAB first aid and hospitals. - Emergency Communication Plan In case of emergencies, the information and communication requirements become very urgent. An Emergency Communication Plan has to support communication procedures within the metro network including following actions: (1) Preparation of communication plans per eventual incident; (2) Coordinated representation, aiming at providing the absolutely necessary confirmed facts, for informing the public in real time; (3) Introduction of procedures to be notified to the stakeholders involved and be readjusted after drills; (4) Preparation of a common vocabulary to be utilized by the communication section personnel (5) Immediate and valid information, as far as the spokesperson is concerned. 3. Validation of Resilience Guidelines for UTS 3.1 The Athens Transport System and the metro lines 2&3 The Attica region exhibits, according to the 2011 census 3.812.350 permanent inhabitants. During the last decades, the car ownership has increased considerably from 248 cars (MDS, 2000) to 384 cars per 1000 inhabitants (WWF, 2014). From 1996 to 2014 the car owning households have increased from 61% to 82% and the number of cars per household from 0,73 to 1,12. The 2014 ratios pertain to a pooled sample proportionally drawn from Athens and the next four largest Greek metropolitan areas. The Attica General Transportation Plan of the Athens Transit Authority revealed 7.5 mi. daily trips apportioned to 38% car, 34% Public Transport (PT) and 28% other modes. The ongoing deep recession in Greece increased the unemployment rate in Attica to 27,3%, thus cutting the commuter share. The drop of the disposable income by 25% hits also trips for shopping and commercialized leisure. Car mobility level reflects reduced fuel consumption for road traffic in Attica by 23% between 2008 and 2013, due also to fuel price increases. PT supply cuts of vehicle-kms by 12% and real fare increases by 21% during the same period knocked on the PT demand too. Elasticity-based modal estimates of OASA for 2013 (see i.a. Deloukas, 2013), provide approximately 6,5 mi daily trips (including 2,4 mi car trips) of which 30% are commuting ones. Private car flows are, in general terms, more flexible and continuous than urban public transport ones. The former are much less dependent on staff, central control, electric power or telecommunications. The PT network exhibits, in general, a lower density and connectivity than the private car alternative. PT multimodality allows, however, for service substitution. Urban rail, owning an exclusive Right-of-Way (RoW), under normal circumstances is more reliable than car in terms of transport time. 3.2 Metro risk scenarios The threat type of the metro pilots pertains to large-scale malevolent attacks (bomb blast or Chemical, Biological, Radiological, Nuclear, Explosives/CBRNE exposure). The metro lines 2&3 crossing at SYNTAGMA, comprising 36
4
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
463
kms, 37 stations, 66 trains and carrying 210 mi. pax p.a., constitute the growing backbone of the Athens urban transport system. There is no overlap of line segments. In terms of connectivity and directness, the Athens metro is well integrated (DERRIBLE and KENNEDY, 2010). Stations and vehicles are crowded places. Critical attack scenarios refer to partial or full closure of affected lines in the post-attack phase. Failure propagation among system components is investigated. The metro attack risk (MAR) is conceptualised in the equation (1) as the potential that the threat will successfully exploit vulnerabilities (weaknesses) within the metro system and therefore result in harmful consequences for the public and the operator (KHOUDOUR et al., 2011). MAR = Threat X Vulnerability X Consequences
(1)
Threat of an attack is external force acting on the metro system; it is a rare, highly uncertain event and its evaluation is a matter of intelligence. Vulnerability is the probability that damage occurs given a threat; protecting factors potentially mitigating metro weaknesses are Closed Circuit Television (CCTV) monitoring, network sensors, station design, security procedures or topology. Vulnerability is property of the metro system and its identification & prioritization is a matter of science & engineering. Consequences reflect the severity of damage given a successful attack and an occurrence of damage; factors mitigating consequences are measures of preparedness such as trained staff, frequent drills with first responders, reserve/retracted vehicles, TETRAcom back-up or public address systems as well as a fast emergency response. The latter factors aim a quick recovery of the system in the post-attack phase. Pertaining EU-common risk assessment methodologies have been developed (CAVENNE and ULISSE, 2007). Specific weaknesses and gaps refer i.a. to the scale and the topology of the metro network (BERCHE et al., 2009), whereas external vulnerabilities to the road corridor connectivity with urban sectors. Berche et al. using network graph theory to examine 14 metropolitan PT networks revealed that i.a. the Paris PT system is much more resilient against attacks than the London one. In Athens, the urban rail system is characterised by a simple network configuration with low redundancy and few alternative routes solely for certain inner-city paths in conjunction with line 1. The radial structure does not exhibit significant circular connectivity enabling rerouting out of the inner ring of Athens. A closure of metro connections to Western suburbs could induce severe cross-modal delays in the road links passing the Kifissos barrier. On the other side, resilience by design was a focal issue for the Athens metro since its inception. ‘Fail-safe’ train operations, ample stations with good lighting without dead corners, appropriate cladding materials, laminated glazing and no bins constitute the rule. New relevant developments such as intelligent CCTV monitoring or vehicle robustness are closely followed (O’NEILL et al., 2014). However, targeted attacks to certain links and nodes of the metro (impacting loss of network connectivity), in conjunction with cross-sectoral loss of functionality of the telecommunication system and, even more, of the power system, are classified as highly critical. Screening of risk scenarios is based on metro vulnerabilities and on attack consequences in terms of travellers and traveller flows affected. The next sections elaborate on the static and dynamic resilience of the Athens metro (Figure 2).
5
464
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
Figure 2 Static and dynamic resilience (source: A. Deloukas, Tools and operational data available [Attiko], www.resolute-eu.org/index.php/2015-07-16-15-29-23/resolute-1st-workshop)
3.3 Metro static resilience The static resilience of the UTS demand is related to the UTS robustness – i.e. maintaining system function after the shock without immediate metro infrastructure restoration. The corresponding guideline and function describes how to absorb/respond to the attack. The staffs recall system and reserve capacities (e.g. spare trains) have to be mobilised. Interim traction or station power shortages may be dampened by stationary supercapacitors. The efficiency of modal substitute strategies (bus bridging, interdiction of Single Occupancy Vehicle (SOV) use, bus lane extension or lifting road tolls in conjunction with Variable Message Sign (VMS) re-direction updates) is assessed in advance with the use of a multi-modal transportation model. The retraction of spare buses from proximal depots (out of 7 depots dispersed over Attica) and scheduled bus lines is a critical strategy component. Service Level Agreements (SLAs) for temporary bus operations have to be then activated. Metro-bus-metro and metro-longer bus leg bridging are options considered. Critical drivers are the geographical extent and duration of the metro connectivity loss. Effective replacement management mitigates the transport capacity degradation of the UTS during the disruption. Transportation model runs estimate with static assignments the interplay of disrupted UTS supply and demand. KPI metric for demand recovery is a static resilience indicator measuring the transport demand covered by residual metro and alternate carriage capacity as percentage of the transport demand reduction due to the attack. 3.4 Metro dynamic resilience The dynamic resilience of metro demand aims to re-establish the initial level of demand as briefly as possible after a disastrous attack. This presupposes fast repair of the infrastructure (physical layer) and restoration of the transport operations (service layer). The corresponding guidelines and functions describe how to absorb/respond to and recover after the attack. Line segments, stations, vehicles and other components to be restored as a priority are to be defined. As a tool of preparedness, a pre-event Recovery Plan would include a registry of capacitated contractors and fast-track contracting options, awarding early delivery (NCHRP, 2013). The phased recovery of metro infrastructure breaks down to short-term response measured in days (clearance of rescue ways, damage-reducing evacuation, emergency service, mass care as well as debris removal), mid-term recovery measured in weeks (repair of urgently needed infrastructure, assessment of repairing or replacing further infra components) and long-term recovery measured in months and years (reconstructing metro infrastructure components). Typical is an exponential recovery time path.
6
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
465
Aimed is the reduction of downtime and degraded modus operandi. The critical KPI for the restoration is the recovery time back to 90% operability of the metro network (Zorn and Shamseldin, 2015). The restoration of metro functionalities encompasses time-critical activities planned as standard responses in the Pre-event Recovery Plan and those improvised after the unexpected incident. The re-establishment of the initial level of demand captures the human dimension, including travellers, their perception of the system, their information processing and their actions (cognitive and affective layer). TV & radio addresses the general public in the post-attack phase. Specific and credible information is requested, as well as consistent expert testimony, rapid investigations and confidence-building media coverage. Software apps for mass alerts and bulk SMS are prospective RESOLUTE products instructing people on the move. The reduction of the perceived public risk influences positively peoples’ cognitive & affective judgment of risks and effectuates panic prevention. Aimed is an enhanced confidence to the transport system as a whole. Communication schemes exercised through diverse dissemination channels and social media - when sufficient - reach a higher situational awareness of travelers and the wider population. Travelers may be diverted in the post-event phase to alternative modes and routes. In catastrophic events, panic is to be eroded and fearful people to be switched to concerned or simply worried population. Public response and travel behavior becomes more resilient then. However, there is always a time lag between risk communication and its impact. Strategies of reducing the perceived risk by the adaptively minded public are crucial. The previous experience of less (N.Y. 9/11, London 7/2005: almost 6% traffic reduction) and more resilient behaviour (Madrid 3/2004, Tokyo sarin gas 3/1995: no ridership reduction) are instructive in this respect (WINTERFELDT and PRAGER, 2010). Learning from past experience is an essential step to adapt resilient strategies for the Athens case. The risk threshold of the Athenian public with respect to a potential strike is assessed through a WTA-risk valuation. The experiment comprises a panel of 550 face-to-face structured interviews and provides (stated only) evidence on the (a) level of metro demand alteration due to the dread risk, (b) segmentation of demand response (discretionary vs. compulsory travel), (c) temporal rate of demand alteration (short- vs. mid-term impact). Transportation model runs with extended generalised cost functions incorporate a cost component of perceived risk at a time. Scenario comparisons estimate the recovery rate of metro demand, the latter being a performance process indicator. 4. Conclusion The UTS guidelines and the Athens pilots advance a resilience management enabling the metro organisation to (a) anticipate potential threats, (b) respond to threats in an effective manner, (c) monitor recovery from threat realization with appropriate metrics, (d) understand and learn from past threat realizations adapting its human, technical and organisational capital. In the Athens pilots we are not concerned about the probability of the occurrence of attack scenarios (threat) but about counteracting their consequences. The last step of the FRAM method, as exposed in the ERMG Guidelines, is the identification of effective countermeasures to potential threats. Diverse countermeasures may lead to the same end, i.e. more resilient UTS with respect to an attack. The Athens pilots recommend actions in order to mitigate the UTS performance variability (negative outcome) in case of a large-scale metro attack. The recommended actions improve the static and dynamic resilience of the UTS in this respect. They comply with the four system capacities (RAG dimensions) of anticipating, responding, monitoring and learning and they validate operationally the ERMG guidelines for the UTS function. Novel performance indicators have been developed for both resilience regimes. Notes 1, 2, 3: List and templates available from corresponding author upon request. Aknowledgements “This work has been supported by the RESOLUTE project (www.RESOLUTE-eu.org) and has been funded within the European Commission’s H2020Programme under contract number 653460. This paper expresses the opinions of the authors and not necessarily those of the European Commission. The European Commission is not liable for any use that may be made of the information contained in this paper.”
7
466
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
References Berche B., C. von Ferber, T. and Yu. Holovatch, 2009. Resilience of public transport networks against attacks, The European Physics Journal B 71 (1), 125-137 Cavenne F. and Ulisse A. 2007. Development of a EU-common risk assessment methodology. EURAM Deliverable, November 2007 Deloukas A., 2013. Deep Recession Impact on the Metropolitan UPT System Ridership. Hellenic Institute of Transportation Engineers, Bulletin 184 Derrible S., C. Kennedy, 2010. Characterizing metro networks:state, form and structure. Transportation, 37, 275-297 Hollnagel, E., 2009. RAG-Resilience Analysis Grid, Technical Document, Industrial Safety Chair, Copenhagen Hollnagel E., 2012. FRAM, the Functional Resonance Analysis Method: Modelling Complex Socio-Technical Systems, Ashgate Khoudour I., E. El-Koursi, S. Velastin, N Buch, L. Thiebot, F. Fontaine, 2011. An approach for protecting transport infrastructure. Proc. IMechE Part F: J. Rail and Rapid Transit, Vol. 225, 383-93 MDS (Metro Development Study), 2000. Existing Transport Conditions in Attica. Technical Report 091, Athens NCHRP, 2013. A Pre-event Recovery Planning Guide for Transportation. NCHRP Report 753, Washington D.C. O’ Neill C., M. Markez, N. Sartori, R. Seddon, U. Battista, 2014. Blast resilient metro vehicles. Transport Research Arena Pr oceedings 2014, Paris Park J., T. Seager, P. Rao, M. Convertino, I. Linkov, 2013. Integrating risk and resilience approaches to catastrophe management in engineering systems. Risk Analysis, Vol. 33, 356-367 RESOLUTE Project, 2015. Deliverable D D3.7.ERMG Adaptation to UTS, www.resolute-eu.org/deliverables Winterfeldt D. and F. Prager, 2010. Estimating behavioural changes for transportation modes after terrorist attacks in London, Madrid and Tokyo. METRANS Project Report 08-10, University of Southern California, L.A. WWF Greece, 2014. Travel Modes and Urban Cycling, in: ‘Better Life’ Programme. Public Issue survey, Athens Zorn C., A. Shamseldin, 2015. Post-disaster infrastructure restoration. International Journal of Disaster Risk Reduction 13, 158-166
8
ScienceDirect
Available online at www.sciencedirect.com Transportation Transportation Research Research Procedia Procedia 00 00 (2016) (2016) 000–000 000–000
ScienceDirect
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
Transportation Research Procedia 24C (2017) 459–466 www.elsevier.com/locate/procedia
3rd Conference on Sustainable Urban Mobility, 3rd CSUM 2016, 26 – 27 May 2016, Volos, Greece
Static and dynamic resilience of transport infrastructure and demand: the case of the Athens metro a*, Efthymia Apostolopouloubb Alexandros Deloukasa* a, a, b b
Attiko Attiko Metro, Metro, 191-193 191-193 Messogion Messogion Ave., Ave., 115 115 25 25 Athens, Athens, Greece Greece
Abstract Abstract The resilience engineering engineering of of the the urban urban transport transport system system (UTS) (UTS) using using the the example example of of the the Athens Athens metro. metro. The paper paper conceptualises conceptualises the the resilience Resilience Resilience management management guidelines guidelines for for the the UTS UTS have have been been developed, developed, whereas whereas Athens Athens metro metro pilots pilots validate validate the the former. former. The The pilots pilots are are concerned concerned with with the the metro metro attack attack risk risk and and its its consequences. consequences. The The risk risk scenarios scenarios consider consider the the static static and and the the dynamic dynamic resilience resilience of of the the UTS UTS system system with with respect respect to to metro metro threats. threats. The The identified identified actions actions have have measurable measurable impacts impacts improving improving the the UTS UTS resilience resilience in in Athens. Athens. © 2016 The Authors. by Elsevier B.V. © Authors. Published Published © 2016 2017 The The Authors. Published by by Elsevier Elsevier B.V. B.V. Peer-review under responsibility of the organizing committee of the 3rd CSUM 2016. Peer-review under responsibility of the organizing Peer-review under responsibility of the organizing committee committee of of the the 3rd 3rd CSUM CSUM 2016. 2016. Keywords:resilience; guideline; Keywords:resilience; guideline; metro metro risk; risk; vulnerability; vulnerability; disruptive disruptive attack attack Corresponding author. author. E-mail E-mail address. ** Corresponding address. [email protected] [email protected]
1. Introduction The EU-funded project ‘RESilience management guidelines and Operationalization appLied to Urban Transport Environment (RESOLUTE)’ addresses critical components of urban transport systems (UTS) in view of safety and security needs. The H2020 project aims the development of European resilience management guidelines (ERMG) adapted to UTS. The ERMG validation will be executed i.a. through specific Athens metro pilots. Resilience is an emergent property of the UTS function and is not reduced exclusively in risk analysis terms (PARK et al., 2013). Resilience is a broader term embracing the ability of UTS (or other CIs) to perform four functions with respect to disruptions: a. Planning and preparation, b. Absorption, c. Recovery, d. Learning and adaptation. We enrich the risk concept, aiming not minimization of an objective disruption probability but mitigation of uncertain disruption (risk) consequences due to unforeseen events. RESOLUTE follows follows the the resilience resilience management management approach approach developed developed by by Erik Erik Hollnagel. Hollnagel. This This modular modular approach approach RESOLUTE aims a description of the functions (activities) of the system under investigation (e.g. UTS) as well as the couplings aims a description of the functions (activities) of the system under investigation (e.g. UTS) as well as the couplings (dependencies) rather than than outcome-based outcome-based performance performance (dependencies) between between the the functions. functions. The The approach approach promotes promotes processprocess- rather metrics. The study conceptualises the resilience management of the urban transport system using the example of the the metrics. The study conceptualises the resilience management of the urban transport system using the example of Athens metro. metro. The The development development of of resilience resilience guidelines guidelines for for the the UTS UTS has has been been finalised, finalised, whereas whereas their their validation validation Athens through the the Athens Athens pilots pilots is is aa subsequent subsequent cycle. cycle. through Resilience is is defined defined as as the the ability ability of of the the urban urban transport transport system system to to withstand withstand aa major major disruption disruption within within acceptable acceptable Resilience degradation parameters parameters and and to to recover recover within within an an acceptable acceptable time. time. There There is is no no abstract abstract [UTS] [UTS] resilience resilience per per se se but but only only degradation with respect to a known or unknown threat. The paper acknowledges a multi-layer resilience of the UTS, considering with respect to a known or unknown threat. The paper acknowledges a multi-layer resilience of the UTS, considering
2352-1465 © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the organizing committee of the 3rd CSUM 2016. 10.1016/j.trpro.2017.05.082
460
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
an interplay of the physical, service and human layer. The Athens pilots focus on disruptions in the service layer (static resilience) and the human layer (dynamic resilience). 2. Development of Resilience Guidelines for UTS 2.1 FRAM Concept The Functional Resonance Analysis Model (FRAM) concept, developed by E. Hollnagel (HOLLNAGEL, 2012), has been applied to simulate complex socio-technical systems such as the UTS. The FRAM concept recognises that in order to study efficiently a complex socio-technical system, it is more important to understand in detail the system dynamics and the variability of performance than to know only what has gone wrong, studying previous system failures. The FRAM concept is based in four principles. These are: (1) the principle of equivalence of failures and successes, which focuses on the fact that it is equally or even more important to study the operation of complex systems and their components that function properly than to study only the parts of the system; (2) the principle of approximate adjustments, which is based on the variability of day-to-day performance, both human and organisational; (3) the principle of emergence, which denotes that in many cases in complex systems it is not possible to explain what happens as a result of known processes (applying the decomposition and causality principles), but effects are rather emergent than resultant and the underlying system is only partly intractable; (4) the principle of functional resonance, which reflects the fact that the dynamic dependencies among the functions of a system must be accounted for as they develop in a specific situation. In complex socio-technical systems, these dependencies can be far more complex than simple cause-effect relations. 2.2 Resilience Analysis Grid The Resilience Analysis Grid (RAG), also developed by E. Hollnagel, is a way to assess a system’s potential for resilience. The resilience of a complex socio-technical system such as the UTS can be described by its abilities to: (1) anticipate potential threats or opportunities for changes; (2) respond to regular or irregular disruptions by adjusting functioning to existing conditions; (3) monitor both the system and the environment for what could become a threat in the immediate time frame; (4) learn from experiences of both successes and failures. In an effort to assess the potential for resilience of the UTS, these four capabilities should be considered and from that basis a RAG can be developed, i.e. four sets of questions where the answers can be used to construct the system’s resilience profile. Some examples of questions for each of the four system abilities follow: (Anticipating) How long is the organisation willing to look ahead?. (Responding) For which events is there a response ready? How soon can a response be given?, (Monitoring) What are the key performance indicators? How, when and why are they revised?, (Learning) Is learning based both on successes and failures? These sets of questions are answered with the support of group discussions with subject matter experts. Depending on the answers given, ratings for all four resilience abilities of the UTS (deficient, unacceptable, acceptable, satisfactory or excellent) will give the resilience profile of the UTS according to the RAG. 2.3 Generic Guidelines (functions) The workflow to produce resilience guidelines has been organized in steps. First, data have been collected about how the various Critical Infrastructures (CIs), such as the UTS, work in general (daily operation), through stakeholder interviews (at workshops and through face to face interviews). Then, the high-level functions present in all types of CIs have been identified. In total, 26 high-level functions were identified and described. Six main types of high-level functions were defined, shown in Figure 1. It is noted that a function in FRAM terms constitutes an activity.
2
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
461
Sustained adaptability
Anticipate Monitor
Respond Learn
Boundaries
Core functions
Figure 1. Sustained adaptability
The functions are characterised and categorized (Listed sample of functions, see Note 1). Anticipating functions: This type includes all functions performed when planning for operation of CIs, such as (a) Develop a Strategic Plan, (b) Manage awareness and user behaviour, (c) Train staff. Responding functions: These are functions performed following a system disruption, such as (a) Restore/repair physical infrastructure, (b) Restore/repair operations, (c) Coordinate emergency actions. Monitoring functions: These functions have been defined so that they are flexible enough to increase or relax the data collections according to a potential or ongoing event. Examples of monitoring functions are: (a) Monitor resource availability, (b) Monitor safety and security, (c) Monitor user generated feedback. Learning functions: (a) Collect event information, (b) Provide adaptation and improvement insights. There are also boundary functions (i.e. functions determining the system’s boundaries): (a) Supply Resources, (b) Provide risk warning, (c) Fight the emergency, (d) Regulate domain and operations, as well as core functions, which express objectives central to the UTS, with many interdependencies with other functions, such as (a) Deliver service, (b) Use of the service. In the next step, a function description template has been filled in for each function (Function Description Template, see Note 2). The prevalent function type (Human, Technology or Organisation) has been determined. The interdependencies of a specific function with other functions have also been defined at this step. The six aspects of each function are: input, output, resources, preconditions, control and time. Interdependencies with other functions have been specified for each function by answering questions such as: “What start the function?”, “What does the function act on or change?” (input questions), “What are the function’s main outputs?” (output question), “What are the required resources for the function?” (resource question), “What controls the function?”, e.g. regulations and Key Performance Indicator (KPIs), stakeholders, organisations (control question) and “Are there any time implications for the function?” e.g. deadlines (time question). In a following step, common conditions or sources that affect the performance variability of each function have been specified and functions are linked together (functions’ couplings). Eleven common performance conditions were identified and evaluated for CIs such as the UTS: (1) Availability of resources; (2) Training and experience (competence); (3) Quality of communication; (4) Human-Computer interaction and operational support; (5) Availability of procedures and plans; (6) Conditions of work; (7) Number of goals and conflict resolution; (8) Available time and time pressure; (9) Circadian rhythm and stress; (10) Team collaboration quality; (11) Quality and support of the organisation. Then, a set of questions is assigned to each function for all common conditions applicable to the function. In the
3
462
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
end, a template has been filled in for each Generic Guideline (function). The template contains a guideline abstract, some background facts about the function, the recommended actions and the associated questions that led to the recommended actions. The recommended actions are provided in order to dampen the performance variability for the common conditions (Generic Guideline Template, see Note 3). 2.4 Metro Emergency Preparedness - Emergency Rules and Procedures The Emergency Procedures constitute the corner-stone of the Safety Plan of an organization for addressing an emergency or crisis. Emergency rules and procedures in the Athens metro network include i.a. OCC response to incidents, Fire fighting, Train evacuation in tunnel, Bomb threat, Controlled smoke channelling with ventilators, Station evacuation, Major incidents, Bomb blast – gas attack - Drill exercises There is a Fire Drill Plan, which simulates an incident and the related necessary response actions. According to the exercise, a fire breaks out in a train entering a station, on the platform level. The general purpose of the drill is to address fire incidents in metro stations in the best possible way. At the same time, the drill also serves to evaluate the capabilities of the local Fire Stations and improve the cooperation between Fire Service and the supporting agencies involved. Tangible bodies potentially contributing to resilient UTS in Athens are the General Secretariat for Civil Protection (CP) and its Operational Centre 199 SEKYPS, Region of Attica CP, Police (GADA), EMAK Rescue Team, EKAB first aid and hospitals. - Emergency Communication Plan In case of emergencies, the information and communication requirements become very urgent. An Emergency Communication Plan has to support communication procedures within the metro network including following actions: (1) Preparation of communication plans per eventual incident; (2) Coordinated representation, aiming at providing the absolutely necessary confirmed facts, for informing the public in real time; (3) Introduction of procedures to be notified to the stakeholders involved and be readjusted after drills; (4) Preparation of a common vocabulary to be utilized by the communication section personnel (5) Immediate and valid information, as far as the spokesperson is concerned. 3. Validation of Resilience Guidelines for UTS 3.1 The Athens Transport System and the metro lines 2&3 The Attica region exhibits, according to the 2011 census 3.812.350 permanent inhabitants. During the last decades, the car ownership has increased considerably from 248 cars (MDS, 2000) to 384 cars per 1000 inhabitants (WWF, 2014). From 1996 to 2014 the car owning households have increased from 61% to 82% and the number of cars per household from 0,73 to 1,12. The 2014 ratios pertain to a pooled sample proportionally drawn from Athens and the next four largest Greek metropolitan areas. The Attica General Transportation Plan of the Athens Transit Authority revealed 7.5 mi. daily trips apportioned to 38% car, 34% Public Transport (PT) and 28% other modes. The ongoing deep recession in Greece increased the unemployment rate in Attica to 27,3%, thus cutting the commuter share. The drop of the disposable income by 25% hits also trips for shopping and commercialized leisure. Car mobility level reflects reduced fuel consumption for road traffic in Attica by 23% between 2008 and 2013, due also to fuel price increases. PT supply cuts of vehicle-kms by 12% and real fare increases by 21% during the same period knocked on the PT demand too. Elasticity-based modal estimates of OASA for 2013 (see i.a. Deloukas, 2013), provide approximately 6,5 mi daily trips (including 2,4 mi car trips) of which 30% are commuting ones. Private car flows are, in general terms, more flexible and continuous than urban public transport ones. The former are much less dependent on staff, central control, electric power or telecommunications. The PT network exhibits, in general, a lower density and connectivity than the private car alternative. PT multimodality allows, however, for service substitution. Urban rail, owning an exclusive Right-of-Way (RoW), under normal circumstances is more reliable than car in terms of transport time. 3.2 Metro risk scenarios The threat type of the metro pilots pertains to large-scale malevolent attacks (bomb blast or Chemical, Biological, Radiological, Nuclear, Explosives/CBRNE exposure). The metro lines 2&3 crossing at SYNTAGMA, comprising 36
4
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
463
kms, 37 stations, 66 trains and carrying 210 mi. pax p.a., constitute the growing backbone of the Athens urban transport system. There is no overlap of line segments. In terms of connectivity and directness, the Athens metro is well integrated (DERRIBLE and KENNEDY, 2010). Stations and vehicles are crowded places. Critical attack scenarios refer to partial or full closure of affected lines in the post-attack phase. Failure propagation among system components is investigated. The metro attack risk (MAR) is conceptualised in the equation (1) as the potential that the threat will successfully exploit vulnerabilities (weaknesses) within the metro system and therefore result in harmful consequences for the public and the operator (KHOUDOUR et al., 2011). MAR = Threat X Vulnerability X Consequences
(1)
Threat of an attack is external force acting on the metro system; it is a rare, highly uncertain event and its evaluation is a matter of intelligence. Vulnerability is the probability that damage occurs given a threat; protecting factors potentially mitigating metro weaknesses are Closed Circuit Television (CCTV) monitoring, network sensors, station design, security procedures or topology. Vulnerability is property of the metro system and its identification & prioritization is a matter of science & engineering. Consequences reflect the severity of damage given a successful attack and an occurrence of damage; factors mitigating consequences are measures of preparedness such as trained staff, frequent drills with first responders, reserve/retracted vehicles, TETRAcom back-up or public address systems as well as a fast emergency response. The latter factors aim a quick recovery of the system in the post-attack phase. Pertaining EU-common risk assessment methodologies have been developed (CAVENNE and ULISSE, 2007). Specific weaknesses and gaps refer i.a. to the scale and the topology of the metro network (BERCHE et al., 2009), whereas external vulnerabilities to the road corridor connectivity with urban sectors. Berche et al. using network graph theory to examine 14 metropolitan PT networks revealed that i.a. the Paris PT system is much more resilient against attacks than the London one. In Athens, the urban rail system is characterised by a simple network configuration with low redundancy and few alternative routes solely for certain inner-city paths in conjunction with line 1. The radial structure does not exhibit significant circular connectivity enabling rerouting out of the inner ring of Athens. A closure of metro connections to Western suburbs could induce severe cross-modal delays in the road links passing the Kifissos barrier. On the other side, resilience by design was a focal issue for the Athens metro since its inception. ‘Fail-safe’ train operations, ample stations with good lighting without dead corners, appropriate cladding materials, laminated glazing and no bins constitute the rule. New relevant developments such as intelligent CCTV monitoring or vehicle robustness are closely followed (O’NEILL et al., 2014). However, targeted attacks to certain links and nodes of the metro (impacting loss of network connectivity), in conjunction with cross-sectoral loss of functionality of the telecommunication system and, even more, of the power system, are classified as highly critical. Screening of risk scenarios is based on metro vulnerabilities and on attack consequences in terms of travellers and traveller flows affected. The next sections elaborate on the static and dynamic resilience of the Athens metro (Figure 2).
5
464
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
Figure 2 Static and dynamic resilience (source: A. Deloukas, Tools and operational data available [Attiko], www.resolute-eu.org/index.php/2015-07-16-15-29-23/resolute-1st-workshop)
3.3 Metro static resilience The static resilience of the UTS demand is related to the UTS robustness – i.e. maintaining system function after the shock without immediate metro infrastructure restoration. The corresponding guideline and function describes how to absorb/respond to the attack. The staffs recall system and reserve capacities (e.g. spare trains) have to be mobilised. Interim traction or station power shortages may be dampened by stationary supercapacitors. The efficiency of modal substitute strategies (bus bridging, interdiction of Single Occupancy Vehicle (SOV) use, bus lane extension or lifting road tolls in conjunction with Variable Message Sign (VMS) re-direction updates) is assessed in advance with the use of a multi-modal transportation model. The retraction of spare buses from proximal depots (out of 7 depots dispersed over Attica) and scheduled bus lines is a critical strategy component. Service Level Agreements (SLAs) for temporary bus operations have to be then activated. Metro-bus-metro and metro-longer bus leg bridging are options considered. Critical drivers are the geographical extent and duration of the metro connectivity loss. Effective replacement management mitigates the transport capacity degradation of the UTS during the disruption. Transportation model runs estimate with static assignments the interplay of disrupted UTS supply and demand. KPI metric for demand recovery is a static resilience indicator measuring the transport demand covered by residual metro and alternate carriage capacity as percentage of the transport demand reduction due to the attack. 3.4 Metro dynamic resilience The dynamic resilience of metro demand aims to re-establish the initial level of demand as briefly as possible after a disastrous attack. This presupposes fast repair of the infrastructure (physical layer) and restoration of the transport operations (service layer). The corresponding guidelines and functions describe how to absorb/respond to and recover after the attack. Line segments, stations, vehicles and other components to be restored as a priority are to be defined. As a tool of preparedness, a pre-event Recovery Plan would include a registry of capacitated contractors and fast-track contracting options, awarding early delivery (NCHRP, 2013). The phased recovery of metro infrastructure breaks down to short-term response measured in days (clearance of rescue ways, damage-reducing evacuation, emergency service, mass care as well as debris removal), mid-term recovery measured in weeks (repair of urgently needed infrastructure, assessment of repairing or replacing further infra components) and long-term recovery measured in months and years (reconstructing metro infrastructure components). Typical is an exponential recovery time path.
6
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
465
Aimed is the reduction of downtime and degraded modus operandi. The critical KPI for the restoration is the recovery time back to 90% operability of the metro network (Zorn and Shamseldin, 2015). The restoration of metro functionalities encompasses time-critical activities planned as standard responses in the Pre-event Recovery Plan and those improvised after the unexpected incident. The re-establishment of the initial level of demand captures the human dimension, including travellers, their perception of the system, their information processing and their actions (cognitive and affective layer). TV & radio addresses the general public in the post-attack phase. Specific and credible information is requested, as well as consistent expert testimony, rapid investigations and confidence-building media coverage. Software apps for mass alerts and bulk SMS are prospective RESOLUTE products instructing people on the move. The reduction of the perceived public risk influences positively peoples’ cognitive & affective judgment of risks and effectuates panic prevention. Aimed is an enhanced confidence to the transport system as a whole. Communication schemes exercised through diverse dissemination channels and social media - when sufficient - reach a higher situational awareness of travelers and the wider population. Travelers may be diverted in the post-event phase to alternative modes and routes. In catastrophic events, panic is to be eroded and fearful people to be switched to concerned or simply worried population. Public response and travel behavior becomes more resilient then. However, there is always a time lag between risk communication and its impact. Strategies of reducing the perceived risk by the adaptively minded public are crucial. The previous experience of less (N.Y. 9/11, London 7/2005: almost 6% traffic reduction) and more resilient behaviour (Madrid 3/2004, Tokyo sarin gas 3/1995: no ridership reduction) are instructive in this respect (WINTERFELDT and PRAGER, 2010). Learning from past experience is an essential step to adapt resilient strategies for the Athens case. The risk threshold of the Athenian public with respect to a potential strike is assessed through a WTA-risk valuation. The experiment comprises a panel of 550 face-to-face structured interviews and provides (stated only) evidence on the (a) level of metro demand alteration due to the dread risk, (b) segmentation of demand response (discretionary vs. compulsory travel), (c) temporal rate of demand alteration (short- vs. mid-term impact). Transportation model runs with extended generalised cost functions incorporate a cost component of perceived risk at a time. Scenario comparisons estimate the recovery rate of metro demand, the latter being a performance process indicator. 4. Conclusion The UTS guidelines and the Athens pilots advance a resilience management enabling the metro organisation to (a) anticipate potential threats, (b) respond to threats in an effective manner, (c) monitor recovery from threat realization with appropriate metrics, (d) understand and learn from past threat realizations adapting its human, technical and organisational capital. In the Athens pilots we are not concerned about the probability of the occurrence of attack scenarios (threat) but about counteracting their consequences. The last step of the FRAM method, as exposed in the ERMG Guidelines, is the identification of effective countermeasures to potential threats. Diverse countermeasures may lead to the same end, i.e. more resilient UTS with respect to an attack. The Athens pilots recommend actions in order to mitigate the UTS performance variability (negative outcome) in case of a large-scale metro attack. The recommended actions improve the static and dynamic resilience of the UTS in this respect. They comply with the four system capacities (RAG dimensions) of anticipating, responding, monitoring and learning and they validate operationally the ERMG guidelines for the UTS function. Novel performance indicators have been developed for both resilience regimes. Notes 1, 2, 3: List and templates available from corresponding author upon request. Aknowledgements “This work has been supported by the RESOLUTE project (www.RESOLUTE-eu.org) and has been funded within the European Commission’s H2020Programme under contract number 653460. This paper expresses the opinions of the authors and not necessarily those of the European Commission. The European Commission is not liable for any use that may be made of the information contained in this paper.”
7
466
Alexandros Deloukas et al. / Transportation Research Procedia 24C (2017) 459–466
References Berche B., C. von Ferber, T. and Yu. Holovatch, 2009. Resilience of public transport networks against attacks, The European Physics Journal B 71 (1), 125-137 Cavenne F. and Ulisse A. 2007. Development of a EU-common risk assessment methodology. EURAM Deliverable, November 2007 Deloukas A., 2013. Deep Recession Impact on the Metropolitan UPT System Ridership. Hellenic Institute of Transportation Engineers, Bulletin 184 Derrible S., C. Kennedy, 2010. Characterizing metro networks:state, form and structure. Transportation, 37, 275-297 Hollnagel, E., 2009. RAG-Resilience Analysis Grid, Technical Document, Industrial Safety Chair, Copenhagen Hollnagel E., 2012. FRAM, the Functional Resonance Analysis Method: Modelling Complex Socio-Technical Systems, Ashgate Khoudour I., E. El-Koursi, S. Velastin, N Buch, L. Thiebot, F. Fontaine, 2011. An approach for protecting transport infrastructure. Proc. IMechE Part F: J. Rail and Rapid Transit, Vol. 225, 383-93 MDS (Metro Development Study), 2000. Existing Transport Conditions in Attica. Technical Report 091, Athens NCHRP, 2013. A Pre-event Recovery Planning Guide for Transportation. NCHRP Report 753, Washington D.C. O’ Neill C., M. Markez, N. Sartori, R. Seddon, U. Battista, 2014. Blast resilient metro vehicles. Transport Research Arena Pr oceedings 2014, Paris Park J., T. Seager, P. Rao, M. Convertino, I. Linkov, 2013. Integrating risk and resilience approaches to catastrophe management in engineering systems. Risk Analysis, Vol. 33, 356-367 RESOLUTE Project, 2015. Deliverable D D3.7.ERMG Adaptation to UTS, www.resolute-eu.org/deliverables Winterfeldt D. and F. Prager, 2010. Estimating behavioural changes for transportation modes after terrorist attacks in London, Madrid and Tokyo. METRANS Project Report 08-10, University of Southern California, L.A. WWF Greece, 2014. Travel Modes and Urban Cycling, in: ‘Better Life’ Programme. Public Issue survey, Athens Zorn C., A. Shamseldin, 2015. Post-disaster infrastructure restoration. International Journal of Disaster Risk Reduction 13, 158-166
8