An evacuation order is a sentence, and everything that has to happen after it is said is a transportation operation with a fixed road network, a finite fleet, a fuel supply sized for ordinary Tuesdays, and a population that does not all own cars. This piece covers where clearance time comes from and how a decision maker is supposed to use it, what contraflow buys and what it costs, how to find the households that cannot drive themselves out, why paratransit vans and school buses are the actual evacuation fleet, fuel and breakdowns along the route, pets and service animals, and why the failures usually turn out to be logistics rather than public stubbornness.

Clearance time: the number the decision maker needs first

Clearance time is the elapsed time between the moment an evacuation order takes effect and the moment the last evacuating vehicle leaves the area at risk. It is made up of mobilization time, which is how long households take to secure a house, find the dog, fuel a car and get on the road, plus travel time on the network, plus queuing delay at whatever bottleneck controls the route. In the hurricane states the figure comes out of the transportation analysis inside a Hurricane Evacuation Study, a product state emergency management agencies develop with the U.S. Army Corps of Engineers, FEMA and the National Hurricane Center, and the same studies also supply the surge modeling and the behavioral assumptions. Inland, for dam failure or a wildland urban interface zone or a rail corridor, nobody hands you the number and someone in your own jurisdiction has to build it.

Clearance time is not a single number, because it is the output of assumptions that vary. The analysis has to assume a participation rate by zone, a vehicle usage rate for how many vehicles each household takes, tourist occupancy in the coastal counties, the time of day and the day of the week, and a response curve describing what percentage of evacuees leave in each hour after the order. Change the participation rate from sixty percent to eighty percent in one zone and the clearance time moves by hours, which is why the studies publish a matrix of scenarios rather than one figure. Anyone who quotes your county a clearance time without saying which scenario produced it has given you a number you cannot use.

The operational use of the figure is subtraction. You start from when the hazard makes the roads unusable, which for surge is well before landfall and for a wildfire may be a matter of hours, and you work backward through clearance time, notification time and the time it takes your own decision process to produce an order with a signature on it. What is left is your decision deadline, and if that deadline has already passed when the forecast arrives, the honest options are ordering earlier on less certain information, phasing the order by zone, or recommending in-place or vertical sheltering for some areas. That last recommendation belongs to an elected official, in writing, with the clearance time attached, and it is a much easier conversation in March than at two in the morning.

Lane capacity and spreading the demand in time

A lane of road moves a limited number of vehicles per hour and no amount of urgency changes it. Recent editions of the Highway Capacity Manual put base capacity for a freeway segment under ideal conditions in the range of roughly 2,200 to 2,400 passenger cars per hour per lane, and observed sustained throughput during real evacuations tends to come in lower because of heavy vehicles, trailers, merge turbulence, weather and driver unfamiliarity. Check the current edition for the values rather than relying on this paragraph, because the point for planning is the order of magnitude and the location of the constraint. The controlling bottleneck is usually a lane drop, a bridge, a signalized intersection at the end of a rural arterial or an interchange where two evacuation flows merge, and the capacity of that one feature sets the clearance time for everything behind it.

Since you cannot add capacity during the event, the only remaining variable is demand over time. Phased or staged evacuation orders release the most vulnerable zone first, then the next zone some hours later, so that the same total number of vehicles passes through the same bottleneck without all arriving in the first four hours. Phasing works only if the messaging is precise about who is being told to leave now and who is being told to prepare and wait, and it fails when broadcast coverage flattens the distinction into a general warning for the whole coast. There is no enforcement mechanism behind a phase, so the plan depends on people in the later zones believing that their turn is coming and that the road will still be open, which is a credibility question built long before the event.

Shadow evacuation is the demand nobody ordered. In every large coastal event, households outside the designated risk area leave anyway, and they consume the same lanes as the people who were told to go. Ahead of Hurricane Rita in September 2005 the Houston-Galveston region saw an enormous voluntary departure, with commonly cited estimates of the total ranging from about 2.5 million to 3.7 million people; these are estimates rather than counts and the sources differ. Published reviews at the time, including an analysis by the Houston Chronicle, concluded that more deaths were associated with the evacuation itself than with the storm’s direct effects, and the various totals were arrived at by different methods, so I would not treat any single figure as settled. Twenty-three nursing home residents died in a bus fire near Wilmer, Texas on September 23, 2005 during that evacuation, and the National Transportation Safety Board investigated and published a highway accident report, which is the place to read what the Board found about the mechanism.

Contraflow: what reversing a highway actually costs

Contraflow uses the inbound lanes of a divided highway to carry outbound traffic, and in principle it roughly doubles the outbound capacity of the facility. Both South Carolina and Georgia developed contraflow plans for their primary coastal evacuation routes after Hurricane Floyd in September 1999, when interstate traffic out of the Charleston and Savannah areas backed up for many hours and the reversal did not happen in time to help. Louisiana revised its plan after the congestion during Hurricane Ivan in September 2004 and used the revised version ahead of Hurricane Katrina in August 2005. The activation authority, the crossover locations, the termination point and the staffing plan belong to the state DOT and the state police in each state, so get the current plan from your DOT district office rather than working from what you remember.

The costs are real and they are mostly about access. Reversing the inbound lanes closes the highway to inbound traffic, which means mutual aid apparatus, utility crews, fuel tankers and ambulances either travel on surface roads or travel in the contraflow lanes under escort with a written exception. Every entrance ramp along the reversed segment has to be physically blocked or staffed, because a driver entering a contraflow segment the ordinary way is a head-on collision, and the number of officers and barricades required to do that for eighty or a hundred miles is larger than most agencies assume. The reversed lanes also have no functioning service plazas, no rest areas on the correct side, and no shoulders configured for the direction of travel, so a breakdown in a contraflow lane is harder to clear than a breakdown in a normal lane.

Contraflow also has a beginning and an end that have to be managed as separate problems. It takes hours to set up and it takes hours to take down, the termination point where four lanes fold back into two becomes the new controlling bottleneck, and the decision to reverse has to be made early enough that the setup completes before the surge of departures arrives. Taking it down while people are still leaving is worse than never starting, so the state needs a stated trigger for termination that accounts for the weather closing the roads to the workers who have to remove the barricades.

The households that cannot drive out

The single most useful data pull an emergency manager can do for evacuation planning is households with no vehicle available, which the American Community Survey publishes down to the census tract in its tables on vehicle availability. Zero-vehicle households cluster geographically rather than distributing evenly, they correlate with age, income and renter occupancy, and the map of those clusters is where your pickup points go. Figures commonly cited from the 2000 Census put the share of New Orleans households with no vehicle available at roughly a quarter to 27 percent before Hurricane Katrina, and different analyses of the same census data produce somewhat different shares depending on whether they count households or people, but any figure in that range describes tens of thousands of people who were never going to drive anywhere.

The census count understates the problem, because vehicle ownership is not the same as the ability to use a vehicle on the day. A car that will not pass a hundred-mile trip, a car with no fuel and a household with no money until Friday, a licensed driver who cannot drive at night or cannot drive at all because of vision or cognition, a single vehicle that the working adult took to a job forty miles away, and a household with three people and five animals all produce the same result as having no car. Add to that the residents of nursing homes and assisted living facilities, hospital inpatients, people on home dialysis or home ventilators, group home residents, jail and detention populations, and anyone whose household includes a person who cannot be moved without a lift, and you have a transportation-assisted population that a passenger-car clearance time never accounted for.

These people have to be found before the event, by name and by address where the law permits, because the alternative is finding them by search after the roads close. Paratransit rider manifests are the closest thing most communities have to a working list of residents who cannot drive, home health and oxygen supplier customer lists are another, and some states operate voluntary special needs registries through the health department or emergency management agency with specific rules about who may hold the data and what it may be used for. Verify what your state’s registry actually promises and what your agency is permitted to do with the records, because a registry that residents believe guarantees a ride, staffed by an agency that reads it as a courtesy list, is a failure waiting for a hurricane. New Orleans built a City Assisted Evacuation Plan after 2005 using designated neighborhood pickup points and a downtown processing point, and the city exercised it during Hurricane Gustav in 2008.

Paratransit vans and school buses are the fleet

The vehicles that move a transportation-assisted population are the lift-equipped vans that the transit agency runs for complementary paratransit under the Americans with Disabilities Act, the transit agency’s fixed-route buses, the school district’s bus fleet, the ambulances and wheelchair vans under contract to hospitals and long-term care facilities, and in some places motorcoaches under a state contract. Paratransit vans are the only part of that fleet designed for passengers who cannot climb steps, and there are never many of them, so the sequence matters: the lift-equipped vehicles go to the people who require lifts, and everyone else moves in school buses. School buses bring their own limits, since most of them have no lift, no restroom, no air conditioning in much of the country, and no space for an oxygen concentrator plus a wheelchair plus a suitcase.

The binding constraint is drivers rather than vehicles. School bus drivers are frequently part-time employees of a school district that answers to a superintendent and a board, not to the emergency manager, they hold commercial licenses with passenger and school bus endorsements, they have their own families to evacuate, and their availability on a Saturday night in August is not the same as their availability on a school day. An evacuation that draws on district buses needs a signed agreement covering who requests them, who pays, who insures the vehicles and the passengers, what happens to the drivers and buses at the destination, and how the drivers get home. The Federal Motor Carrier Safety Administration has an emergency relief provision that suspends most hours of service rules for drivers providing direct assistance during a declared emergency, and the scope of that relief is narrower than people assume, so confirm the current regulation and your state’s adoption of it rather than assuming the clock stops.

Then do the cycle arithmetic, because a bus does not make one trip, it makes a round trip over and over. A bus carrying forty passengers ninety miles at reduced speed, with loading at the pickup point and unloading at the shelter and fuel somewhere in between, may need six or seven hours per cycle. Moving ten thousand people at forty per load is two hundred and fifty loads, which fifty buses complete in five cycles, which is well over thirty hours of continuous operation and therefore more than one driver per bus. Run that calculation with your own numbers and the answer usually shows that the order has to be given many hours before anyone thought, or that the destination has to be closer, or that the fleet is not the size the plan assumed. The site’s case study on the 2001 waterborne evacuation of Lower Manhattan is the example of what an improvised fleet can accomplish and what it costs to have no plan behind it, and it is worth reading alongside this one rather than treated as a model.

Fuel, breakdowns and the roadside

Retail fuel along an evacuation route is sized for ordinary daily traffic, with underground tanks that turn over on a delivery schedule built around normal demand, so a corridor carrying ten or twenty times its usual volume empties its tanks in hours. Resupply then has to move against the flow, which is difficult on a congested route and impossible in a contraflow segment without a specific exception, and the pumps and card readers need electricity, which is why a station on a route with no generator connection is a closed station the moment the power goes. Florida and Louisiana both enacted requirements after their hurricane seasons of the 2000s obliging certain stations, generally those near designated evacuation routes, to be pre-wired for generator connection, so ask your state emergency management agency which stations on your routes are covered and whether anyone verifies it.

Consumption also goes up in ways the trip mileage does not predict, because a vehicle idling in a twelve-hour queue in September heat with the air conditioning running burns fuel without covering distance. That produces the breakdown pattern familiar from any bad evacuation, which is overheating, running out of fuel in a travel lane, and cooling system failures in older vehicles, and each of those events costs the route capacity as described above. The countermeasures are unglamorous and they work: pre-staged wreckers with push bumpers at intervals along the route, fuel trucks positioned to service stranded vehicles on the shoulder rather than only stations, water distribution, and a clearance authority that does not require the vehicle owner’s consent.

People in a stalled evacuation need the things people always need, and the route is where they need them. Plan for restrooms, water, shade, baby formula, medication refills and a medical presence, because a queue carrying a hundred thousand vehicles includes people in insulin-dependent diabetic crisis, people whose portable oxygen has a battery life measured in hours, women in labor and elderly passengers overheating in the back seat. EMS staging along the route is harder than it sounds, since the ambulance has to get to the patient and then get out with them through the same congestion, so the plan needs agreed access points, an inbound path and a receiving hospital that is not itself evacuating.

Pets, service animals and livestock

People refuse to leave animals, and the planning response to that is logistical rather than rhetorical. Congress amended the Stafford Act with the Pet Evacuation and Transportation Standards Act, signed in October 2006, which requires state and local emergency plans to account for the needs of individuals with household pets and service animals and permits federal assistance for rescuing, caring for and sheltering household pets and service animals under a declaration. Look up the current statutory language and FEMA’s implementing policy rather than the summary in this paragraph, because what the law requires of a plan and what a household expects at a shelter door are two different things and the gap between them is where the arguments happen.

Operationally, animals turn into vehicle capacity, crate inventory, veterinary staffing and space. A household evacuating with three dogs takes a vehicle and often a second one, a bus load of forty passengers with pets needs forty carriers that somebody purchased and stored, and a co-located pet shelter needs a separate ventilated space, a supply of food, a waste plan, an intake record that reliably reconnects an animal with its owner, and a licensed veterinarian who has agreed in advance to be the medical authority. Many states run an animal response capability through the department of agriculture, a state animal response team, or the state veterinarian’s office, and county animal control plus the local humane society are usually the people who will actually do the work, so they belong at the planning table rather than on a call list.

Livestock and large animals will not be evacuated at the last minute, and pretending otherwise wastes planning effort. Moving cattle or horses requires trailers, drivers, receiving facilities with pens and water, and lead times measured in days rather than hours, which means the agricultural piece of an evacuation is an early-notice function tied to the forecast and not to the evacuation order. The practical products are a list of trailer resources with the phone numbers of people who tow them, agreements with fairgrounds and equestrian facilities inland, and a written position on what advice the county gives an owner who cannot move the animals, which is generally about identification, moving stock to high ground and opening gates rather than about transport.

What gets called non-compliance is usually a missing ride

After a bad evacuation the public explanation tends to land on the people who stayed, described as stubborn, complacent or unwilling to heed the warning. The behavioral analyses conducted for state hurricane evacuation studies have repeatedly identified other reasons underneath, and the ones that recur are having no vehicle, not having the cash for fuel and a motel room, employers who will not release workers, a household member who cannot travel, animals, worry about losing housing or possessions, and concern about contact with officials over immigration or law enforcement status. Each of those is something a jurisdiction can supply or fail to supply, so describing the outcome as a compliance problem hides the part the government owned. Talk to the researchers your state emergency management agency uses for its behavioral analysis and read the study for your own coast, because the participation rates in your clearance time came from that work.

The same logic applies to the things that go wrong inside an evacuation that people did join. Buses that arrive without drivers, pickup points with no shade and no water where elderly residents wait four hours, contraflow that starts too late, fuel that ran out at hour six, a receiving shelter that cannot accept a power wheelchair, and a nursing home that discovered its transport contractor had signed the same contract with eleven other facilities are all supply and coordination failures with names and owners attached. They show up in after-action reports as coordination findings, and the fix for each one is a document, a contract term, a driver roster or a piece of equipment rather than a message campaign.

Re-entry is the second half of the transportation problem and it gets a fraction of the planning attention. The returning flow arrives faster than the outbound flow did, it arrives into an area with debris in the roads, no signals, no retail fuel and possibly no water system, and it is politically much harder to hold back than it was to push out. Decide in advance who authorizes re-entry, whether it is phased by zone, whether essential workers and utility crews get credentials that actually work at a checkpoint, and how the contraflow segments get restored to normal operation while traffic is already moving. Write that into the evacuation annex you already have rather than starting a separate plan, because the same fleet, the same fuel and the same roads are involved in both directions.

What to do at your agency

• Have your emergency manager pull the American Community Survey vehicle availability data for every census tract in your jurisdiction this month, map the zero-vehicle households, and give the map to the transit director and the school transportation director with a request for comment on pickup point locations.
• Ask your state DOT district engineer in writing for the current clearance time scenarios that apply to your evacuation zones and for the contraflow plan covering your primary route, then write the controlling numbers and the activation authority into the evacuation annex that already exists.
• Ask the transit agency’s paratransit manager for the current count of lift-equipped vehicles, the number of drivers who could be reached on a weekend, and the number of active riders on the paratransit manifest, and record those three numbers in the plan with the date they were provided.
• Take the school district transportation director to lunch, get a copy of the signed agreement covering emergency use of district buses, and if the agreement does not name who requests the buses, who pays, who insures the passengers and how drivers get home from the destination, put those four items on the agenda of the next meeting you already have with the superintendent’s office.
• Run the bus cycle arithmetic for one real scenario using your own pickup points and your own receiving shelter, and take the resulting number of hours to the elected official who signs evacuation orders so the decision deadline is understood before a season begins.
• Have your animal control supervisor confirm in writing how many pet carriers your agency can put at a pickup point within four hours, who the veterinarian of record is for a co-located pet shelter, and how animal intake records link an animal to its owner’s shelter registration.
• Add one paragraph on re-entry to the existing evacuation annex naming who authorizes return, whether it is phased by zone, and which credential a utility crew or essential worker presents at a checkpoint, and brief that paragraph at the next law enforcement shift commanders’ meeting.

Takeaways

• Clearance time measures mobilization, travel and queuing from the effective time of an order until the last evacuating vehicle leaves the risk area, and it is always the output of stated assumptions about participation, vehicle usage and timing, so a clearance time quoted without its scenario is unusable.
• The decision deadline is the hazard arrival time minus clearance time minus notification and decision time, which frequently means the order has to be given on less forecast certainty than a decision maker would prefer.
• Lane capacity is fixed during the event and the controlling bottleneck sets the clearance time, so the only manageable variable is the distribution of demand over time through phased orders and credible messaging.
• Contraflow roughly doubles outbound capacity while closing the highway to inbound traffic, requiring every ramp along the segment to be blocked or staffed, and its setup and termination each take hours that have to be triggered deliberately.
• Households with no vehicle available cluster geographically, the census count understates the transportation-assisted population once broken cars, non-drivers, medical dependency and institutional residents are added, and those people have to be identified before the event.
• Lift-equipped paratransit vans are the scarce vehicle and drivers are the scarce resource overall, and the round-trip cycle time of a bus, rather than the number of seats, determines how many people can actually be moved before the roads close.
• Retail fuel along a route empties in hours, resupply has to travel against the flow, pumps need electricity, and a disabled vehicle in a travel lane costs more capacity than most plans account for, which makes quick clearance an evacuation function.
• The Pet Evacuation and Transportation Standards Act of 2006 requires plans to account for household pets and service animals, service animals under the Department of Justice ADA regulations stay with their handlers in the general shelter population, and livestock movement is an early-notice function that cannot be done at the last minute.
• When an evacuation goes badly, the findings usually name missing vehicles, missing drivers, missing fuel, late contraflow and receiving sites that were not ready, all of which are supply problems with owners rather than evidence that the public ignored the warning.