In late 2010, the City of Colorado Springs, Pikes Peak Area Council of Governments (PPACG), and PPACG’s travel model consultant team completed a year-long cooperative wildfire evacuation planning process. The process was supported by the simulation of wildfire evacuation scenarios using an adaptation of the PPACG’s VISUM-based four-step travel model. Analysis using the adapted model supported the development of a clear understanding of the challenges that would be faced in evacuating Ute Pass communities and westside Colorado Springs neighborhoods in response to wildfire threat, as well as the development of estimated times-to-evacuate, choke point identification, and comprehensive traffic control plans for twenty two Colorado Springs neighborhoods and the City of Manitou Springs. The model adaptation used an hourly volume/capacity metric to estimate times-to-evacuate, and was built on a worst-case scenario in which the simulated wildfire event was set during the PM peak hour commute (background traffic) with all affected households both part of background commute traffic and taking part in the evacuation of their neighborhoods. It was believed that this worst case scenario with redundant commute and evacuation traffic was implausible and, thus, presented a contingency for planning purposes.

On Tuesday, June 26, 2012, the simulated worst case became reality. In all, 32,000 persons were evacuated in response to the wildfire event, more than half during one and a half hours on the 26th of June. Before the wildfire was fully contained, 18,247 acres had burned and 346 homes and other structures had been destroyed. There were two fatalities and at least six injuries. With insurance claims totaling $352.6M, the event set a new record for the state of Colorado.

That the evacuation was as successful as it was rests on the extensive planning that occurred in advance of the wildfire event and, at least to some extent, on the accuracy of the simulation-supported scenario planning. As an example, for key evacuated neighborhoods, modeled estimated time-to-evacuate of 1.28 to 1.63 hours compare with the experienced time–to-evacuate of roughly 1.5 hours. This paper will explore how the model results were borne out by the reality at a detailed level. Recommendation to improve model performance as well as traffic control plan detail will also be presented from a “lessons learned” perspective.