Mitigation of Extreme Event Risk: E lectric Power Outage and Infrastructure Failure Interactions

1. Mitigation of Extreme Event Risk:Electric Power Outage and Infrastructure Failure Interactions Presented by Timothy L. McDaniels University of British Columbia USC Center for Risk and Economic Analysis of Terrorism Events (CREATE) Symposium on the Economic Costs and Consequences of Terrorism 20-21 August 2004

2. Acknowledgements Co-authors Stephanie Chang, University of British Columbia Dorothy Reed, University of Washington Krista Peterson, University of British Columbia Contributor Joey Mikawoz Funded by a grant from the National Science Foundation

3. Outline • Introduction • Structure and components of overall project • Infrastructure interdependencies • Conceptual framework for IFI measures • Examples of IFI measures for major outages • 2003 Eastern NA blackout • 1998 Quebec Ice Storm • 1993 Inauguration Day Storm in PNW

4. 1. Introduction • Terrorism: an extreme kind of “extreme event” • “New kind of trouble”: no “all clear”, deliberate • NSF definition of extreme event: • Low p, high consequence, systems interactions, nonlinear response, outside of standard “coping” range • New metrics, analytical methods needed

5. Introduction (2) • Major electrical outages are the outcome of many kinds of “extreme events” • These outages create spillovers into other infrastructure systems • We call these “infrastructure failure interdependencies” (IFIs) • Managing IFIs helps mitigate the effects of big outages or other extreme events

6. Introduction (3) • Conference theme: analyzing economic costs of terrorism attacks • Helpful for guiding priorities for prevention and mitigation investments • Our project: guidance for mitigating effects of big power outages in terms of effects on other infrastructure systems (IFIs); helpful also for terrorism

7. 2. Project Components • Three broad components (three years): 1. Develop metrics and methods for characterizing IFIs (this paper) 2. Develop influence diagrams and scenarios as simple models of IFI patterns 3. IFI mitigation options; ranking by experts

8. Kobe earthquake interdependencies

9. Influence Diagram of Electric Power Reliability (after Keeney et al., 1995)

10. Approaches Guiding Mitigation Ranking • Large literature on risk ranking generally • Our approach: not E(V), but rather: conditional E(V), given a major outage (Haimes, 1998, partitioned multiobjective risk method), so no probabilities of outages or extreme events needed • Multiple objectives matter as basis for ranking mitigation alternatives • Domain experts to create and rank alternatives

11. 3. Concepts for IFI Analysis • Rinaldi, et al, IEEE, 2001 provides important framework for “critical infrastructure interdependency” studies • Definition: “bidirectional relationship between two infrastructures through which the state of each infrastructure influences or is correlated to the state of the other.” • Often called “lifeline interactions”

12. 4. Conceptual Framework for IFIs • We are interested in finding out which interdependencies (and failures) are most important to society • We propose a new framework that builds on previous work just discussed • Provides a typology of IFIs for systematically gathering, structuring and analyzing data about actual outage events • We focus on electrical outages; could be used for any IFI

13. Conceptual Framework (2) • We want to address two questions: • What are the channels by which electric power outages cause failures in other infrastructure systems? • Which of these channels are the most important? • Three components: aspects of the outage (initial failure), aspects of the interactions, and aspects of the consequences

14. Conceptual Framework (3)

15. Conceptual Framework (4)

16. Conceptual Framework (5)

17. Comparing our categories to others

18. Applications of IFI Framework • Used published newspaper accounts and reports to create a database of IFIs for a given major outage event • Used the categories of the framework to structure the database • Developed a “severity index” for categorizing consequences • Developed graphic approaches to summarize findings

19. Disruption Scale

20. 2003 Northeast Blackout Largest blackout in North American history Photo by Reuters/Chip East

21. Duration of outage by city

22. Outage Begins IFIs Begin 5 Hours People trapped in elevators rescued 11 Hours Water pumps begin working again 1 Day Trading light on stock exchange Outage Ends Power restored to alarms, appliances, gas and water pumps, and sewage treatment plants 2 days Bank branches open Public health concerns, such as food poisoning from spoiled food, decrease Cellular phones service reliability restored Limited service at airport 1500 traffic signals still not working properly 4 days Canadian National Exposition opens 4 days late Subways and streetcars start running Oil refineries resume operations Fresh food supplies still low in grocery stores, food banks Weeks Blood and vaccine supplies low Many government offices closed for 11 days to conserve energy Major industries operate at half capacity to conserve energy Some nuclear reactors take weeks to get back online Homes flooded with sewage cleaned up IFIs End Timeline of Toronto’s IFIs

23. Elevators Nuclear Power Conservation Computer Systems Restaurants HVAC Sanitation Wastewater Retail Hotels Security Water Oil Manufacturing Mines Building Support Mass Transit Fire Utilities Business Police Roads 911 Rail Ambulance Emergency Services Transportation Air Bus 2003 NE Power Outage ATM Gas Stations Tele- communications Banks Finance Credit Cards Media Stock Exchange Cellular Telephones Health Care Food Supply Internet Storage Government Land Line Telephones Transportation Hospitals Public Health Preparation Offices Cable Production Services Event Affected Infrastructure Systems Affected Infrastructure Subsystems Very disruptive to many people Very disruptive to few people Moderately disruptive to many people Moderately disruptive to few people or minor inconvenience to many or few people

24. 1998 Ice Storm • Affected Northeastern U.S. and Canada • 1.4 million people in Quebec and 230,000 in Ontario without power Photo by CP PHOTO/Jacques Boissinot

25. HVAC Security Plumbing Garage Doors Water Oil Building Support Mass Transit Utilities Business Roads Schools Air Restaurants Daycare Retail Gas Stations Transportation Education Manufacturing 1998 Ice Storm and Outage Hotels Tele- communications Emergency Services Health Care Finance Fire Hospitals Media Shelters Food Supply Government Public Health Cable 911 Services Land Line Telephones Police ATM Banks Storage Production

26. 1993 Inauguration Day Storm January 20 windstorm that affected the Pacific Northwest Photo by Seattle City Light

27. Elevator problems in high-rises Interpoint manufacturing plants down, Boeing 777 plant briefly out, Microsoft out for 12 hours Coordination of repair crews from BC Hydro and western states Raw sewage overflow and dumping Alderwood Mall closes early Building Support Pumps unable to fill storage tanks: extreme water conservation measures imposed Gas stations closed: commuters unable to drive to work with low or no gasoline, gasoline required for small generators Utilities Business Gasoline pumps not working 1993 Storm Power Outage 911 flooded with calls about power restoration Traffic lights dark on surface streets Emergency Services Transportation Reversible lanes on I-5 not working Many restaurants close, others may have sanitary problems due to lack of hot water Public safety threat: Many power lines down and lying in the streets Health Care Food Supply Hospitals report more cases of communicable diseases such as E.Coli, possibly due to lack of hot water for properly washing hands Government Schools close; some students remain overnight 32-year-old man electrocuted when a treetop hit a power line Olympia legislative session curtailed Very disruptive to many people Very disruptive to few people Moderately disruptive to many people Moderately disruptive to few people or minor inconvenience to many or few people Event Affected Infrastructure Systems Description of Event Interdependencies

28. Summary and Conclusions • Study seeks to identify important IFIs and promising mitigation strategies for them • Preliminary analysis shows IFIs vary depending on type and duration of outage, early indications are that impacts on transportation are especially severe • Next step is expanding the database using the conceptual framework

29. Conclusions and Comparison • Our approach contrasts with the work of other researchers here • We are working from the bottom up to identify “low hanging fruit” in terms of opportunities to mitigate the IFIs from electrical outages from any cause • The result would be more resilient response to disasters

30. Thank You