Disaster Creep – Why Disaster and Catastrophe are the Norm - not the Exception

Issue 71: Disaster Creep – Why Disaster and Catastrophe are the Norm - not the Exception

By Michael D. Larrañaga, P.E., Ph.D.

The frequency, intensity, and severity of disasters globally are increasing, and this "disaster creep” has profound effects on modern society.^1,2,3,4 During the year 2011, the U.S. had 14 separate billion-dollar weather disasters followed by 2012 with 11 billion-dollar weather disasters. The economic consequences of these disasters are growing exponentially as a result of increased population densities, and the transition to coastal megacities to foster global trade. As global commerce continues to sustain the world’s economies, the world becomes increasingly networked. As networks grow both in size and flow capacity, they tend to self organize towards efficiency, optimization, and decreasing redundancy. And this has profound effects on the magnitude of disasters.

Recent weather related disasters that support this premise are the 2005 Hurricanes Katrina and Rita ($85 billion), Japan’s 2011 earthquake-Tsunami-nuclear meltdown (>$235 billion), the 2008 Sichuan, China earthquake ($29 billion), and Superstorm Sandy, which is expected to become the costliest disaster in U.S. history. The Chernobyl meltdown ($200 billion), Space Shuttle Columbia ($13 billion), Prestige Oil Spill-Spain ($12 billion), and the BP/Macondo Incident in the Gulf of Mexico ($20 billion-$40 billion) are the costliest non-weather related accidents in world history.^5,6 Other examples include terrorist attacks, earthquakes, electrical blackouts, transportation system collapses, industrial accidents, food shortages, epidemics, and political and social upheaval. The challenges presented by these disasters threaten global supply chain resilience (banking, agriculture, natural resources, energy, etc.). These catastrophes are intensified by the modern evolution of highly connected (networked), optimized, and cost-efficient systems that have eliminated the surge capacity and error tolerance found in complex systems.

Many modern energy, industrial, transportation, health care, telecommunications, and political systems are highly vulnerable to small changes that propagate and develop into major disasters. The systemic tendency to unravel, decay uncontrollably, or move from order to disorder (e.g., disaster) is a characteristic of all natural and human-made systems and is defined by Newton’s Second Law of thermodynamics as "entropy."

Before one can overcome this entropic tendency of all systems to unravel, one must understand the nature of system failure across multiple disciplines. In addition, decay of networked systems is non-Newtonian, meaning that networked disasters decay at a rate much faster or different than what can be predicted. Humans have a tendency to view disasters as outliers, or uncommon events, yet experience shows that catastrophes are the norm and stable systems are the outliers. Human nature forces people to view the world through naïve "disaster” lenses. Perhaps this is the human condition – perpetual optimism. Ted Lewis, author of Bak’s Sand Pile, maintains that life is a series of passages from catastrophe to catastrophe with inconsequential periods of calm in between, thus staking the claim that catastrophe is the new normal.³

Predictable unpredictability is a characteristic of most complex systems. This seems to be contradictory, but it can be seen in natural systems. Natural catastrophes such as epidemics, forest fires, earthquakes, insect foraging, avalanches, population shifts, and extinctions all present a similar curve that can be obtained by plotting the probability of an event occurring (y-axis) against the consequences of the event occurring (x-axis).³ One example is an avalanche in the Swiss Alps. There is a high probability that an avalanche will occur, but the overwhelming majority of these avalanches will be small and inconsequential. At the other extreme, there is a very small chance that a large and catastrophic avalanche will occur. Yet catastrophic avalanches occur every year around the world. The curve obtained by plotting avalanche frequency vs. magnitude exhibits power law behavior and can be used to explore variations in avalanche size through space and time in addition to contributing factors, which can provide valuable information to avalanche managers who strive to prevent catastrophic avalanches.⁷

Another example is forest fires; it's not possible to predict where the next forest fire will ignite due to random lightning strikes, but forest fire data with regards to frequency and magnitude of forest fires follows a power law.

Power laws are defined by the curve 1/x^q (x > 0, q > 0) and have a shape as shown in Figure 1. These patterns are found throughout natural, man-made, and random systems. The exponent, q, is of great importance. The horizontal axis can be anything of interest, including storm damages, casualties of war, or elapsed time between accidents. Hazards with power law exponents below 1.0 are high-risk, while hazards with exponents above 1.0 are low-risk. As such, the exponent has important implication for risk-informed decision-making. Table 1 gives some common examples of power law exponents, the x-axis values, and their associated risk classifications. From this, one can conclude that forest fires present a high risk to society, as do hurricanes, earthquakes, wars, and epidemics.^3,8

Figure 1. Random lightning strikes start forest fires (top) creating the pattern at bottom. The probability – consequence graph (probability of a forest fire vs. percent of forest burned) of a forest fire obeys a power law with an exponent of 0.54 (bottom).

 

A unifying theory amongst all disasters and catastrophes can be found in the power law. By utilizing this tool to manage risk, it is possible to identify high-risk priorities to decrease the potential for a catastrophe.³ Figure 2 represents a power law analysis of the U.S. crude pipeline infrastructure. The long tail in the analysis translates to a tremendous Probable Maximum Loss (PML).

	Asset/Sector	Consequence	Exponent
Low Risk	S&P500 (1974-1999)	$Volatility	3.1-2.7
	Large Fire in Cities	$Loss	2.1
	Airline Accidents	Deaths	1.6
	Tornadoes	Deaths	1.4
	Terrorism	Deaths	1.4
	Floods	Deaths	1.35
	Forest Fires in China	Land Area	1.25
	East/West Power Grid	Megawatts	1
	Earthquakes	Energy, Area	1
	Asteroids	Energy	1
	Pacific Hurricanes	Energy	1
High Risk	Hurricanes	$Loss	0.98
	Public Switched Telephone	Customer-Minutes	.91
	Forest Fires	Land Area	.54-.66
	Hurricanes	Deaths	.58
	Earthquakes	$Loss	.41
	Earthquakes	Deaths	.41
	Wars	Deaths	.41
	Whooping Cough	Deaths	.26
	Measles	Deaths	.26
	Small Fires in Cities	$Loss	.07

Table 1: Power law exponents of various incidents and disasters.³

Figure 2: Network analysis of the U.S. crude oil pipeline infrastructure. The curve on the bottom left reflects the power law defined by exponent (1.1655) for the national pipeline network, while the curve to the right defines Petroleum Administration Defense District (PADD) 5 (west coast, including Hawaii and Alaska, exponent=0.3943).9 The probability – consequence graph (probability of a random network fault vs. percent of network outage per given fault) of the networks fault obey a power law. The long tail (bottom right) indicates increased probability and consequence (risk) of network down time associated with the West Coast Crude Oil Distribution Network (top). See Figure 3

Figure 3: Probable Maximum Loss of the National Crude Oil Network vs. the West Coast Network (PADD 5). The analysis shows that the West Coast Network maintains a high risk of network downtime due to a single random fault.

Clearly, the west coast analysis presents tremendous PML risk (Figure 3), as almost any random fault in the west coast crude delivery system can result in 100% loss of the network, dwarfing the PML risk of a random fault in the entire national network. PML risk is determined by multiplying the probability values (y-axis) by the percent consequence values (x-axis) in terms of percent of network loss (e.g., risk = probability x consequence) where percent consequence (x-axis) is defined as the percent of the network lost due to a single random fault.9

 

So if one wanted to increase the resiliency of the entire national crude oil network, the west coast would be a good place to start; to become more resilient against catastrophic failure by adding redundancy, hence decreasing optimization while increasing resiliency against catastrophic failure.

Michael D. Larrañaga is with Oklahoma State University

1. Smith, P., Bennett, J., Darken, R., Lewis, T. and Larrañaga, M. "Network-Based Risk Assessment of the U.S. Crude Pipeline Infrastructure,” International Journal of Critical Infrastructures, (in press).
2. Hallegatte, S. and Przyluski, V. "The economics of natural disasters: Concepts and methods,” The World Bank, Washington, DC, 2010.
3. Perrow, C. The Next Catastrophe: Reducing Our Vulnerabilities To Natural, Industrial, and Terrorist Disasters. Princeton University Press, Princeton, NJ: 2011.
4. Lewis, T. Bak’s Sand Pile: Strategies for a Catastrophic World. AGILE Press, Williams, CA: 2011.
5. Bouwer, L., Crompton, R., Faust, E., Höppe, P., & Pielke Jr, R. "Confronting Disaster Losses," Science, 318 (5851), p. 753. November 2, 2007.
6. Larrañaga, M. et al., "Incident Analysis: Macondo Prospect 252-1 Deepwater Horizon Well Control Incident," Oklahoma State University, Stillwater, OK, 2011.
7. Smith, P., Kincannon, H., Lehnert, R., Wang, Q. and Larrañaga, M. "Human Error Analysis of the Macondo Well Blowout”, Process Safety Progress (in press).
8. Naisbitt, W. Forster, R., Birkeland, K. & Harrison, W. "Avalanche Frequency and Magnitude: Using Power-Law Exponents To Investigate Snow Avalanche Size Proportions Through Time And Space," Proceedings Whistler 2008 International Snow Science Workshop, Montana State University, Bozeman, MT, 2008.
9. Bak, P. How Nature Works - the Science of Self-organized Criticality, Copernicus - Springer Verlag New York Inc., New York: 1996.

---

Related Articles:

2nd Quarter 2011 -- After the Alarm Sounds: Historical, Present and Future Perspectives – Jason D. Averill, Erica D. Kuligowski, Ph.D., and Richard D. Peacock
An historical review on egress with information on recent trends in egress modeling. The authors also present five proposed research initiatives that were recently agreed upon as part of a prioritized, consensus-based research agenda. A description of each initiative and its goals is provided. READ MORE

3rd Quarter 2010 -- Wildfire: Past, Present and Future – Ronny J. Coleman and Kate Dargan
The problem of wildfires is increasingly complex and traditional solutions are losing effectiveness. Coleman and Dargan describe how experts are redefining the way they approach wildfires and highlight five approaches for combating wildfires currently being developed or implemented in California: firesafe land use planning, building construction and design, vegetative fuels management, community education, and first responder training. READ MORE

Summer 2008 -- Planning for Disasters – Donald L. Schmidt
Most businesses know they have to be prepared for disasters; however, many are uncertain what preparedness entails. "Disaster plans,” "contingency plans,” and "emergency action plans” are components of what should be an integrated emergency management and business continuity program. This article discusses many of the essential elements of NFPA 1600, often referred to as the "National Preparedness Standard.” The author provides guidance for getting started; risk assessment; resource management; plan development; training, drills, and exercises; and program evaluation and revision. READ MORE

---