Some cities or communities develop an evacuation plan for use in case of an emergency. There are a variety of the explanation why metropolis officers may enact their plan, a main one being a pure catastrophe, corresponding to a twister, flood, or wildfire. An evacuation plan might help the neighborhood extra successfully reply to an emergency, and so might assist save lives. Nonetheless, it may be tough for a metropolis to guage such a plan as a result of it’s not sensible to have a whole city or metropolis rehearse a full blown evacuation. For instance, Mill Valley, a metropolis in northern California, created a wildfire evacuation plan however lacked an estimate for a way lengthy the evacuation would take.
At present we describe a case examine wherein we teamed up with town of Mill Valley to check and enhance their evacuation plan. We define our method in our paper, “Mill Valley Evacuation Examine”. We began by utilizing a site visitors simulator to mannequin a citywide evacuation. The analysis objective was to offer town with detailed estimates for a way lengthy it might take to evacuate town, and, by learning the egress sample, to search out modifications to make the plan more practical. Whereas our prior work on this topic offered an estimate for the evacuation time and confirmed how the time could possibly be decreased if sure highway adjustments have been applied, it seems the suggestions in that paper — corresponding to altering the variety of outgoing lanes on an arterial — weren’t possible. The present spherical of analysis improves upon the preliminary examine by extra precisely modeling the quantity and beginning areas of automobiles, by utilizing a extra real looking map, and by working carefully with metropolis officers to make sure that beneficial adjustments to the plan are deemed viable.
Geography and methodology
Mill Valley is in Marin County, California, north of San Francisco. Most of the residences are positioned on the steep hillsides of a number of valleys surrounded by dense redwood forests.
![]() |
![]() |
Aerial views of Mill Valley, courtesy of the Metropolis of Mill Valley. |
Lots of these residences are in areas which have just one exit route, towards the city heart. From there the very best evacuation route is towards Freeway 101, which is within the flat a part of town and is the almost certainly space to be removed from potential wildfires. Some neighborhoods produce other routes that lead away from each town and Freeway 101, however these routes cross by way of hilly forested areas, which could possibly be harmful or impassable throughout a wildfire. So, the evacuation plan directs all automobiles west of Freeway 101 to go east, to the freeway (see map under). The neighborhoods east of Freeway 101 should not included within the simulation as a result of they’re away from areas with a excessive fireplace hazard score, and are near the freeway.
Mill Valley has about 11,400 households west of Freeway 101. Most Mill Valley households have two automobiles. Evacuation occasions scale with the variety of automobiles, so it’s within the widespread curiosity to attenuate the variety of automobiles used throughout an evacuation. To that finish, Mill Valley has a public consciousness marketing campaign aimed toward having every family evacuate in a single car. Whereas nobody is aware of what number of automobiles can be used throughout an evacuation, it’s protected to imagine it’s on common between one and two per family. The essential evacuation downside, then, is the best way to effectively get between 11 and 23 thousand automobiles from the varied residences onto one of many three units of Freeway 101 on-ramps.
![]() |
The simulated a part of Mill Valley west of Freeway 101 is contained in the blue border. Freeway 101 is proven in inexperienced. The purple squares point out the three units of Freeway 101 on-ramps. The pink space has the best fireplace hazard score. |
The present work makes use of the identical normal methodology because the earlier analysis, specifically, working the open supply SUMO agent-based site visitors simulator on a map of Mill Valley. The site visitors simulator fashions site visitors by simulating every car individually. The detailed behaviors of automobiles are dictated by a car-following mannequin. Every car is given a degree and time at which to begin and an preliminary route. The routes of most automobiles are up to date all through the simulation, relying on circumstances. To contemplate potential adjustments in driver conduct underneath the excessive stress circumstances of an evacuation, the consequences of the “aggressiveness” of every automobile can also be investigated, however in our case the impacts are minimal. Some simplifying assumptions are that automobiles originate at residential addresses and the roads and highways are initially empty. These assumptions correspond roughly to circumstances that could possibly be encountered if an evacuation occurs in the course of the evening. The principle inputs within the simulation are the highway community, the family areas, the common variety of automobiles per family, and a departure temporal distribution. Now we have to make assumptions concerning the departure distribution. After discussing with town officers, we selected a distribution such that almost all automobiles depart inside an hour.
4 bottlenecks
Mill Valley has three units of Freeway 101 on-ramps: northern, center, and southern. All of the automobiles should use one among these units of on-ramps to succeed in their vacation spot (both the northernmost or southernmost phase of Freeway 101 included in our map). Provided that we’re solely involved with nearly all of Mill Valley that lies west of the freeway, there are two lanes that method the northern on-ramps, and one lane that approaches every of the center and southern on-ramps. Since each car has to cross over one among these 4 lanes to succeed in the freeway, they’re the bottlenecks. Given the geography and present infrastructure, including extra lanes is infeasible. The intention of this analysis, then, is to attempt to modify site visitors patterns to maximise the speed of site visitors on every of the 4 lanes.
Evacuation plan
After we began this analysis, Mill Valley had a preliminary evacuation plan. It included modifying site visitors patterns — disabling site visitors lights and altering site visitors guidelines — on just a few highway segments, in addition to specifying the assets (site visitors officers, signage) essential to implement the adjustments. For example, a two-way highway could also be modified to a one-way highway to double the variety of outgoing lanes. Briefly altering the route of site visitors is named contraflow.
The plot under exhibits the simulated fraction of automobiles which have departed or reached their locations versus time, for 1, 1.5, and a pair of automobiles per family (left to proper). The dashed line on the far left exhibits the fraction which have departed. The strong black strains present the preliminary evacuation plan outcomes and the dotted strains point out the conventional highway community (baseline) outcomes. The preliminary evacuation plan considerably hastens the evacuation.
![]() |
The cumulative fraction of automobiles vs. time in hours. The demand curve is proven within the dashed line on the far left. The strong strains present the preliminary evacuation plan curves for 1, 1.5 and a pair of automobiles per family (left to proper). The dotted strains present the identical for the baseline case. |
We will perceive how efficient the preliminary evacuation plan is by measuring the charges on the bottlenecks. The under plots present the speed of site visitors on every of the 4 lanes resulting in the freeway on-ramps for the case of 1.5 automobiles per family for each the baseline case (the conventional highway guidelines; proven shaded in grey) and the preliminary evacuation plan (proven outlined in black). The common charge per lane varies tremendously within the totally different circumstances. It’s clear that, whereas the evacuation plan results in elevated evacuation charges, there may be room for enchancment. Particularly, the center on-ramps are fairly underutilized.
![]() |
The charges of site visitors on the 4 lanes resulting in Freeway 101 on-ramps for each the baseline case (regular highway guidelines; proven shaded in grey) and the preliminary evacuation plan (proven outlined in black). |
Ultimate evacuation plan
After learning the map and investigating totally different options, we, working along with metropolis officers, discovered a minimal set of recent highway adjustments that considerably decrease the evacuation time in comparison with the preliminary evacuation plan (proven under). We name this the ultimate evacuation plan. It extends the contraflow part of the preliminary plan 1000 ft additional west, to a predominant intersection. Crucially, this enables for one of many (usually) two outgoing lanes to be devoted to routing site visitors to the center on-ramps. It additionally creates two outgoing lanes from that predominant intersection clear by way of to the northern on-ramps, over ¾ of a mile to the east.
![]() |
A map of the primary adjustments within the last evacuation plan. The purple line exhibits that site visitors heading north on Camino Alto will get diverted to the center Freeway 101 on-ramps. The blue line exhibits site visitors within the northern lane of E Blithedale Ave will get routed on the brand new contraflow part. |
The speed per lane plots evaluating the preliminary and last evacuation plans are proven under for 1.5 automobiles per family. The simulation signifies that the ultimate plan will increase the common charge of site visitors on the lane resulting in the center on-ramps from about 4 automobiles per minute to about 18. It additionally will increase the by way of charge of the northern on-ramps by over 60%.
![]() |
The charges of site visitors on the 4 lanes resulting in Freeway 101 on-ramps for each the preliminary case (proven shaded in grey) and the ultimate evacuation plan (proven outlined in black). |
The under plot exhibits the cumulative fraction of automobiles vs. time, evaluating the circumstances of 1, 1.5 and a pair of automobiles per family for the preliminary and last evacuation plans. The speedup is kind of important, on the size of hours. For instance, with 1.5 automobiles per family, it took 5.3 hours to evacuate town utilizing the preliminary evacuation plan, and solely 3.5 hours utilizing the ultimate plan.
![]() |
The cumulative fraction of automobiles vs. time in hours. The demand curve is proven within the dashed line on the far left. The strong strains present the ultimate evacuation plan curves for 1, 1.5 and a pair of automobiles per family (left to proper). The dotted strains present the identical for the preliminary evacuation plan. |
Conclusion
Evacuation plans will be essential in shortly getting many individuals to security in emergency conditions. Whereas some cities have site visitors evacuation plans in place, it may be tough for officers to learn the way properly the plan works or whether or not it may be improved. Google Analysis helped Mill Valley take a look at and consider their evacuation plan by working site visitors simulations. We discovered that, whereas the preliminary plan did pace up the evacuation time, some minor adjustments to the plan considerably expedited evacuation. We labored carefully with town throughout this analysis, and Mill Valley has adopted the ultimate plan. We have been capable of present town with extra simulation particulars, together with outcomes for evacuating town one space at a time. Full particulars will be discovered within the paper.
Detailed suggestions for a specific evacuation plan are essentially particular to the realm underneath examine. So, the particular highway community adjustments we discovered for Mill Valley should not straight relevant for different cities. Nonetheless, we used solely public information (highway community from OpenStreetMap; family info from census information) and an open supply simulator (SUMO), so any metropolis or company might use the methodology utilized in our paper to acquire outcomes for his or her space.
Acknowledgements
We thank former Mayor John McCauley and Metropolis of Mill Valley personnel Tom Welch, Lindsay Haynes, Danielle Staude, Rick Navarro and Alan Piombo for quite a few discussions and suggestions, and Carla Bromberg for program administration.