Analyzing Routes and Response Times

This is a second preview chapter of a new book in the Primer series from Bradshaw Consulting Services to be titled “Closest Vehicle Dispatch: A Primer for Fire” to be released in time for the FDIC 2017 at the end of April.

Whether you are held to the standards of NFPA 1710, which addresses predominately career fire department responses in the US, or NFPA 1720, which deals specifically with volunteer departments, the challenge of meeting these response time standards is increasingly difficult for many reasons. Higher demands on limited resources and increasing performance expectations from the public are just a couple of those forces opposing response efficiency. Another elementary factor that critically impacts our response times is the route we choose in order to arrive at an incident. In most cases, there is not always a single route that is consistently the best choice at all times of the day or week. These differences can also include seasonal variations or be complicated by special events which may be planned or unplanned (Demiryurek, 2010). The subjectivity of route selection is further complicated by dynamic characteristics such as traffic or weather in addition to the extent of the mental map we develop of a service area or what that map may be lacking in adjoining or mutual aid areas (Spencer, 2011).

Most of the considerations that we process as we consider a potential path of travel in an emergency vehicle are often made subconsciously through personal experience and knowledge. While there is no legitimate argument against knowing your service territory well, the question becomes do we have sufficient awareness to consistently make the best route choices?

According to U.S. Fire Administration statistics for 2005, responding to alarms accounted for 17 percent of firefighter on-duty fatalities (Response, 2007). Deaths in road vehicle crashes are often the second most frequent cause of on-duty firefighter fatalities. In 2014, this percentage dropped to only 10 percent with a total of just 7 fatalities. Although the change is positive, it is too early to consider this to be a trend since it is only the second lowest number of crash deaths over the past 30 years (Fahy, 2015). While these accidents are not all due to their route choice, it can be argued that there are times where crews were clearly in the wrong place at the wrong time. Furthermore, the shortest path is not always the quickest route, and the fastest one may not have the simplest directions either (Duckham, 2003). Given the technology and data available today, there is little doubt that we can make strong decisions provided that we understand how we make these choices and what information may improve them.

In selecting a route for any particular apparatus, we may consider the physical or geographic characteristics of the roadway that determine the maximum speed of travel based on the maneuverability and size of our apparatus. Similarly, we must consider the likelihood of traffic congestion and also the safety of our crews as well as the public. As we increasingly rely on algorithms for making driving decisions, it is important to appreciate the mechanics of how the technology components function together. The Global Positioning System (GPS) is often credited with providing guidance to vehicle operators, but this is not exactly true. The satellite constellation that makes up the American-operated GPS (and similarly the European GLONASS) simply sends accurate time signals by radio waves to our portable receivers who detect the length of time each signal has traveled through space and then triangulates a position based on the calculated distance from those man-made stars (Hurn, 1989). The accuracy of the position that your GPS unit determines is based on the quality of those signals received and the precision of the local clock used to compare the time encoded in the signals. These satellites have no concept of transportation networks or traffic congestion on earth. It is Geographic Information Systems (GIS) that model the street networks and also track the vehicles using them. Unlike the limited number of GPS-like constellations in space that help us derive our position, there are a multitude of GIS-based computer services that offer routing recommendations. Some of these services, like the consumer-based routing applications available on your smartphone, are located on “cloud servers” (although they are quite terrestrial) while others may be hosted privately on local government networks and available only to “trusted client” applications on your Mobile Data Terminal (MDT).MARVLISiOSinFD

Each of these GIS services has unique embedded algorithms for recommending directions or to estimate arrival times (Keenan, 1998). As users of these systems, we become subject to the specific assumptions inherent within their design leaving them far from being equivalent to one another (Psaraftis, 1995). For instance, network models must account for the elevation differences of overpasses in relation to the roadway below in order to prevent suggesting that a vehicle take a turn off of the side of a bridge. The cost of that ill-fated maneuver would be insurmountable, but other legitimate turns have minor costs associated with them because the apparatus must slow down to navigate the curve safely. A traffic light, or oncoming vehicles, can add further to that turn delay. Accounting for these delays requires logic in the GIS routing algorithm as well as valid time estimates coded into the street network data at each intersection.

The most basic feature of any transportation network model, however, is the cost of movement along a road segment in either direction which is known as its “impedance.” Many systems will assume the speed limit over the distance (impedance_time=speed/distance) between intersections to derive a similar “drive time” in both directions. Real world conditions (including traffic, terrain, and weather) will prove that speed limit-based assumption to be overly simplified and can lead to poor routing decisions because of unrealistic impedance values in the model (Elalouf, 2012). Crews will quickly recognize these failures and the lack of trust that these errors engender can compromise the entire routing program. Realistic impedances should be variable based on the time of day or day of the week in addition to the direction of travel.

More complex online routing services now offer near real-time traffic updates. While this traffic feedback can be invaluable to most drivers, its practicality to emergency vehicles appears limited in general. If our task was to deliver pizzas, we would be constrained by normal traffic regulations. Knowing where traffic congestion is at any given moment would allow us an opportunity to seek an alternative to bypass a congested intersection. This is a common type of need for drivers and therefore many consumer routing apps seek to address that specific function (Ruilin, 2016). But when our duty is to respond to the accident at that same intersection that is causing the delay for others, these typical consumer routing applications may fail our unique requirement. This objection is especially valid where emergency vehicles are not strictly constrained by the driving patterns of other vehicles on the roadway. In certain situations, it may be allowable for an apparatus to use the road shoulder for travel or even cross a median to use an on-coming traffic lane or to traverse a one-way street in the wrong direction (Harmes, 2007). The only reasonable exceptions to this generality are those dense urban areas where congestion is excessive and these “open” lanes or roadway shoulders simply do not exist to allow apparatus to circumvent that traffic. In a recent trip to New York City, I visited a fire station in downtown Manhattan. They received a call and exited the station with red lights and sirens blaring, but even the air horn was unable to move traffic. The engine sat at the traffic light behind the rest of the cars until the intersection cleared enough to allow drivers to create a path up to the next intersection.

In general, when we look to leverage technology for our unique demands in public safety, a system would ideally be able to learn our peculiar patterns of travel and record typical impedances based on how our own fleet resources travel. Additionally, these impedances will likely be different during certain hours of the day or on specific days of the week and vary even further seasonally based on whether school is in or out of session. These cyclical patterns will have a huge impact on actual drive times and any route recommendations must account for them accordingly. Current consumer routing applications are continually improving their ability to recognize and address the needs of passenger cars or ordinary delivery trucks, but this still does not necessarily translate to better routing of emergent public safety vehicles in most cases.

Finally, the last critical piece of route selection is a review after the call. Comparing the actual route traveled with the recommended path is an important feedback mechanism to both ensure that the system is operating as intended and to build confidence within your crews that encourage them to trust the system. This is not to suggest a blind obedience to technology, but constructing a learning process for everyone in developing tools that function to improve overall performance. No technology is perfect in the real world, just as no person has ultimate knowledge at all times. But cooperatively, we can learn to make improvements in either the computer or human systems as needed to enhance awareness in the other. The most successful implementations of routing assistance create cooperative relationships between responders and the GIS staff responsible for maintaining the data. Failures discovered in any system should not be used to condemn an otherwise useful technology, but seen as opportunities for improvements in either the algorithms behind it or the data that fuels it.

One of the critical outcomes of route selection, aside from arriving safely, is the total time of travel. No matter when the clock starts for measuring your response time, it is the minutes and seconds that the wheels are rolling that often consume the majority of it. The longer that time or distance, the higher the cost. A cost that can be measured both in actual vehicle operating expenses as well as the risks associated with its operation; not to mention the losses adding up on scene prior to your arrival. In general, the shorter the time (and distance) between dispatch and your safe arrival on scene, the better it is for everybody.

 

References:

Demiryurek, U., Banaei-Kashani, F., Shahabi, C. “A case for time-dependent shortest path computation in spatial networks.” GIS ’10 Proceedings of the 18th SIGSPATIAL International Conference on Advances in Geographic Information Systems. ACM, November, 2010; 474-477.

Duckworth, M., Kulik, L. “’Simplest’ Paths: Automated Route Selection for Navigation in Spatial Information Theory.” Foundations of Geographic Information Science. (2003) 169-185. Berlin: Springer-Verlag.

Elalouf, Amir. “Efficient Routing of Emergency Vehicles under Uncertain Urban Traffic Conditions.” Journal of Service Science and Management, (2012) 5, 241-248

Fahy, R. F., LeBlanc, P., Molis, J. Firefighter Fatalities in the United States-2014. NFPA No. FFD10, 2015. National Fire Protection Association, Quincy, MA.

Harmes, J. Guide to IAFC Model Policies and procedures for Emergency Vehicle Safety. 2007. IAFC: Fairfax, VA.

Hurn, Jeff. GPS: A Guide to the Next Utility. (1989) Sunnyvale: Trimble Navigation.

Keenan, Peter B. “Spatial Decision Support Systems for Vehicle Routing”. Decision Support Systems. (1998);22(1):65-71. Elsevier, Salt Lake City.

Psaraftis, H.N. “Dynamic vehicle routing: Status and prospects.” Annals of Operations Research (1995) 61: 143.

“Response-Time Considerations.” Fire Chiefs Online. ISO Properties, 2007. Web. 20 May 2016.

Ruilin, L., Hongzhang, L., Daehan, K. “Balanced traffic routing: Design, implementation, and evaluation.” Ad Hoc Networks. (2016);37(1):14-28. Elsevier, Salt Lake City.

Spencer, Laura. “Why the Shortest Route Isn’t Always the Best One.” Freelance Folder, November 2011. Web. 7 December 2016.

1 Comment

Leave a Reply

Your email address will not be published. Required fields are marked *