smart international symposium for next generation infrastructure:identifying extreme risks in...

24
www.isngi.org Identifying Extreme Risks in Critical Infrastructure Interdependencies Presented by: A/Prof Kang Tai, Nanyang Technological University

Upload: smart-infrastructure-facility

Post on 06-May-2015

281 views

Category:

Education


0 download

DESCRIPTION

A presentation conducted by A/Prof Kang Tai Nanyang Technological University. Presented on Thursday the 3rd of October 2013. Critical infrastructures like our power generation facilities and water supply form highly interconnected networks that are mutually dependent and any failure can cascade through the network, resulting in devastating impact on health, safety and the economy. These catastrophic events/disruptions can be triggered by environmental accidents, geological/weather phenomena, disease pandemics, etc. The disruptions can be caused/exacerbated by their being unexpected, but they may actually be expected if relevant data have been accounted for. To help account for and thereby anticipate such disruptions, one way is to identify potential unforeseen interdependencies among infrastructure components that can lead to extreme disruptions upon some failure in the network. This paper shows how a simulation model for cascading failures and a risk analysis/optimization approach can be applied to search for unforeseen interdependencies and failure points that give rise to the highest risk in a network.

TRANSCRIPT

Page 1: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

Monday, 30th September 2013: Business & policy Dialogue

Tuesday 1 October to Thursday, 3rd October: Academic and Policy Dialogue

www.isngi.org

ENDORSING PARTNERS

The following are confirmed contributors to the business and policy dialogue in Sydney:

• Rick Sawers (National Australia Bank)

• Nick Greiner (Chairman (Infrastructure NSW)

www.isngi.org

Identifying Extreme Risks in Critical Infrastructure

Interdependencies

Presented by: A/Prof Kang Tai, Nanyang Technological University

Page 2: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

Identifying Extreme Risks in Critical Infrastructure Interdependencies

K. Tai School of Mechanical and Aerospace Engineering, NTU

A. Kizhakkedath School of Mechanical and Aerospace Engineering, NTU

J. Lin School of Mechanical and Aerospace Engineering, NTU

R.L.K. Tiong Institute of Catastrophe Risk Management & School of Civil and Environmental Engineering, NTU

M.S. Sim Information Division, DSO National Laboratories, Singapore

International Symposium for Next Generation Infrastructure SMART Infrastructure Facility, University of Wollongong Wollongong, Australia, 1 – 4 October 2013

Page 3: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

2

Critical Infrastructure

• Critical infrastructure refers to the assets, systems and networks comprising identifiable industries, institutions and distribution capabilities that provide a reliable flow of goods and services essential to the functioning of the economy, the government at various levels, and society as a whole (Clinton 1996). Clinton, W.J. (1996) “Executive order 13010 - Critical infrastructure protection”, Federal Register, Vol.61, No.138, pp.37347-37350

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 4: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

3

Critical Infrastructure for a Modern Society/Economy

Agriculture

and Food

Banking and Finance

Communication and Information

Technology

Drinking Water and Treatment

Plants

Military Installations and Defence

Transportation Systems

Health Care and Civil Defence

Energy

Commercial and Industry

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 5: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

4

Critical Infrastructures form Interconnected Networks with Complex Interdependencies

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 6: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

5

Classification of Infrastructure Interdependencies

• Physical

– A physical or engineering reliance between infrastructures, e.g. material flow from one infrastructure to another

• Information / Cyber

– An informational or control requirement between infrastructures, e.g. a reliance on information transfer between infrastructures

• Geospatial / Geographic

– A relationship that exists entirely because of the proximity of infrastructures, e.g. a local environmental event affects components across multiple infrastructures due to physical proximity

• Policy / Procedural – An interdependency that exists due to policy or procedure that relates a state or event

change in one infrastructure sector to a subsequent effect on another sector, e.g. government’s emergency mandatory orders on a particular area due to the influence of an event

• Societal / Logical – An interdependency that an infrastructure event may have on societal factors, e.g. public

opinion, public confidence, fear, and culture issues

(Rinaldi et al. 2001, Pederson et al. 2006)

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 7: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

6

Complex Interdependencies Lead to Infrastructure Disruptions with Widespread Consequences

9/11 terrorist attacks

2011 Tohoku

earthquake/tsunami

2011 floods in Thailand

2008 global

financial crisis

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 8: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

7

Unforeseen Interdependencies

Banking and Financial

Infrastructures

Transportation Infrastructures

Military Infrastructures Global Impact

9/11 terrorist attacks

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 9: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

8

Black Swans

• The idea that such high impact but highly unexpected events could actually have been expected if the relevant available data had been accounted for was put forth by Taleb in his book “The Black Swan”.

• Black Swan events are highly improbable events (outliers), and highly impactful, and can be caused and/or exacerbated by their being unexpected (Taleb 2007).

• However, in spite of being highly unexpected, it is natural that experts (and even casual observers) will retrospectively be able to construct explanations for their occurrences after they have occurred, making them explainable and expected. Taleb, N.N. (2007) The Black Swan: The Impact of the Highly Improbable, Random House

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 10: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

9

Analyzing Vulnerabilities in Critical Infrastructure Networks by Network Modelling/Analysis

interdependency (known)interdependency (unforeseen)

sector 3sector 2sector 1

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 11: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

10

Multiobjective Optimization of Risk

• Risk Analysis Framework

– risk = f (probability, impact)

• Multiobjective Optimization Problem

– searching for maximum probability of occurrence of failure/hazard/threat – searching for maximum impact of disruption (minimum giant component

size)

• Decision Variables – unforeseen interdependencies – failure point(s)

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 12: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

11

Decision Variables in Multiobjective Optimization Problem

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 13: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

12

Multiobjective Optimization by Genetic Algorithms (GA)

yes

no

begin

initialize population ofnetworks

termination/convergence

criteria

stop

selection andrecombination/mutation to populate next generation

compute failure probabilities & compute disruption

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 14: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

13

Anticipating Extreme Risks by Multiobjective Optimization

disruption/impact

prob

abili

ty

Pareto front

Black S

wanBlack Swan

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 15: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

14

Experimental Test Problem

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 16: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

15

Experimental Test Problem – Agent-Based Modelling/Simulation Using NetLogo

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 17: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

16

Experimental Results – Single Objective Opt. (Maximize Impact with Single Node Failure/Attack)

Node failed No. of unforeseen

interdependencies added

Unforeseen interdependencies added Giant component size

28 0 Nil 36

28 1 One of (7→3, 8→3, 11→3, 15→3, 17→3, 28→3, 31→3)

32

28 2 One of (7→3, 8→3, 11→3, 15→3, 17→3, 28→3, 31→3) and one of (3→19, 4→19, 13→19, 15→19, 21→19, 28→19)

30

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 18: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

17

Experimental Results – Single Objective Opt. (Maximize Impact with Single Node Failure/Attack)

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 19: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

18

Experimental Results – Multiobj. Opt. (2 Unforeseen Interdepend. & Single Node Failure/Attack)

Node failed

Probability Giant comp. size

28 0.1 30

9 0.26 33

15 0.36 34

31 0.41 35

1, 5, 12, 43

0.5 36

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 20: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

19

Experimental Results – Multiobj. Opt. (2 Unforeseen Interdepend. & Single Node Failure/Attack)

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 21: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

20

Experimental Results – Multiobj. Opt. (2 Unforeseen Interdepend. & Double Node Failure/Attack)

Nodes failed Probability Giant comp. size

28, 41 0.0080 19

28, 36 0.0130 21

28, 14 0.0200 22

28, 16 0.0320 27

28, 1 0.0500 28

9, 15 0.0936 30

27, 15 0.1152 31

27, 43 0.1600 32

(1,15),(5,15), (12,15),(43,15)

0.1800 33

(1,31),(5,31), (12,31),(43,31)

0.2050 34

(12,43),(1,12), (1,5),(1,43), (5,43),(5,12)

0.2500 35

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 22: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

21

Experimental Results – Multiobj. Opt. (2 Unforeseen Interdepend. & Double Node Failure/Attack)

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 23: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

22

Concluding Remarks

• The experiments show that unforeseen interdependencies can indeed exacerbate the disruption consequences/impact, with the extreme disruptions interpreted as Black Swan events.

• The methodology can serve as a tool for scenario planning, by

helping policymakers to anticipate and thereby focus on the “worst case” scenarios.

• The multiobjective optimization approach also provides a way for policymakers to analyze the “trade-off” between the high-probability/low-impact events and the low-probability/high-impact events.

Tai et al. – Identifying Extreme Risks in Critical Infrastructure Interdependencies (ISNGI 2013)

Page 24: SMART International Symposium for Next Generation Infrastructure:Identifying Extreme Risks in Critical Infrastructure Interdependencies

Contact : Associate Professor K. Tai School of Mechanical and Aerospace Engineering, Nanyang Technological University Phone : 67904444 Email : [email protected] URL : http://www.ntu.edu.sg/home/mktai/