Evacuation Behavior: Challenging

You'd think the shadow evacuees would be balanced out by the cry wolf effect - those people who refuse to evacuate. But that would be way too easy.

The one plan most emergency managers dread writing is the evacuation plan. This is because regardless of how comprehensive and reasonable it is, or how detailed and rational, - there is a high rate of non-compliance from the public.

It doesn’t even matter what kind of disaster we are planning for.

Case in point:  I happen to live with a nuclear power engineer-type who has learned quite a bit about emergency management in the past 10 years.  He points out articles in his world that bridge into mine.  The most recent was from the Atomic Insights blog titled “Increasing Evacuation Zones actually increase risk in case of reactor accident.”

Emergency response for a nuclear power incident has traditionally included evacuation of the surrounding population beyond a arbitrary radius (general 10 miles).  The concept was that the radiation dose would decrease radically from the incident site, and at 10 miles would decrease to an acceptable level. This was all based on an NRC report from 1980.

After 9/11, the NRC initiated a reassessment of the effects of a severe reactor accident, complete with one of those inexplicable acronyms:  “The “State-of-the-Art Reactor Consequence Analyses” (SOARCA) project was completed and released in June 2012.

It is a very detailed and comprehensive and somewhat incomprehensible report (to a layman). The abstract notes that because of previous and ongoing research, “analyses of severe accidents at nuclear power reactors are more detailed, integrated and realistic than at any time in the past.”

Buried down on page 84 of that report, under the subtitle “Sensitivity Analyses on the Size of the Evacuation Zone” is this statement:

Analysis of Figure 15 and 16 also shows that expanding the evacuation size from a 10 mile radius to a 20 mile radius results in increased LCF risk for people in the 0-10 mile area. SOARCA analyses show that an evacuation beyond the area closest to the plant will delay those most at risk, i.e., closest to the plant. The increased risk to the population within 10 miles of the plant is due to slower evacuation speeds because of additional traffic congestion and delays that result from evacuation of a larger population.

Basically, the larger the evacuation zone, the more exposure for those people closest to the event because of the traffic inherent in any evacuation, especially one that is likely to be as fraught with both real and imaginary consequences as an incident at a nuclear power plant.

This isn’t news to emergency managers. Evacuation planning is complicated and made more so by politicians who demand unreasonable planning parameters and a public who don’t pay attention. Emergency Managers also know that regardless of the evacuation limits, disasters come with shadow evacuations – that group of people who are not required to evacuate, but try it anyway, causing even more traffic delay.

You’d think the ‘shadow evacuees’ would be balanced out by the ‘cry wolf’ effect – those people who refuse to evacuate – and then cause their own suite of response problems. But that would be way too easy.

There have been some studies about evacuation behavior, including this one looking at hurricanes in South Carolina.

There weren't many surprises:  fewer people are willing to evacuate for a Category 1 or 2 hurricane, and a third are uncertain whether they were located in a storm surge or Flood zone. It also suggested that shadow evacuations could significantly increase the evacuating population and potentially put more people in harms’ way.

The solution to this problem is difficult.  It falls somewhere between trying to modify human behavior and modifying emergency plans to account for human behavior.  I’m not sure which would be easier.