Citation

Lessan J, Kim AM. Safety Sci. 2022; 156: e105894.

Copyright

(Copyright © 2022, Elsevier Publishing)

DOI

10.1016/j.ssci.2022.105894

PMID

unavailable

Abstract

In this study, we explore how short- and no-notice evacuation instructions may be designed by a risk-averse emergency management body, explicitly considering varying levels of evacuee compliance and responses to these instructions in the design. The population compliance rate, or its distribution, are unlikely to be known; even a small error in estimating either could lead to drastically worse outcomes. Our solution, which is based on the idea of hedging, determines optimal evacuation instructions in a range of possible scenarios. As part of the proposed solution algorithm, we utilize conditional-value-at-risk (CVaR) and stochastic optimization paradigms to demonstrate how and when compliance uncertainty is a critical input for improved evacuation instructions. Our findings show that in situations where compliance rates are low or highly uncertain, relying solely on highly risk-averse strategies is unlikely to deliver efficient outcomes in comparison to a naïve or no-information approach (where instructions are designed assuming 100% compliance). Although accounting for compliance uncertainty in designing evacuation instructions can be highly beneficial, the added benefits are realized beyond certain population compliance rate levels and uncertainty ranges. This work provides insights into how emergency managers can best overcome non-compliance to instruction, the benefits which are realized together with efforts to promote compliance (community education initiatives and improved communication mechanisms).

Language: en

Keywords

Compliance uncertainty; Conditional-value-at-risk; Evacuation instruction; Evacuee compliance; Optimization; Wildland fire evacuation