Part 1. Background
The city of Long Beach, California is a city in southern California with a population of 450,000 people. Though considered a suburb of Los Angeles, it is a city within itself with its own city government, police department, and fire department. Long Beach is a diverse city with great neighborhoods, an oceanfront beach, a vibrant downtown for younger people to live and play, park and recreation areas, and a bicycle-friendly outlook. As all larger cities, it does have problems with crime, homelessness, and graffiti.
The city contains a number of opportunities for start-ups and entrepreneurs. Stationed at California State University Long Beach is the Institute for Innovation and Entrepreneurship. Downtown Long Beach contains the city’s first coworking community in Work Evolution, a WeWork, and a well-funded Long Beach Accelerator. In addition, the city itself has a Smart City Initiative which advances solutions to civic challenges by engaging the community in solving inefficiencies.
One of the latest Smart City Initiatives is an effort to eliminate inefficiencies in responding to service requests by city field representatives. These non-emergency service requests are called in or reported via a website to a ServiceNow platform (Ref. https://www.servicenow.com) for nuisances such as downed trees, dumped trash in alleyways and parks, inoperable traffic lights, and graffiti removal. The service requests are then pushed out to city field representatives for disposition. The Smart City Challenge is seeking solutions to make the city field representatives more efficient in their responding to service requests and transport between requests.
Part 2. Current Process
The current process to responding to service requests is to simply assign requests tracked in ServiceNow with the next available field service representative. This is a first- in, first-out approach where the requests are assigned to city personnel regardless of how far they have to travel to get to the new assignment.
Figure 1 shows the city of Long Beach along with the 2020 city council districts. (The city council districts are shown for reference only and have no effect on service assignments. These are the 2020 district boundaries before the updates in 2022 to reflect the 2020 census results.) Shown as red dots are all of the ServiceNow requests for service for a given time period. Also shown are green dots which represent the location of city field representatives trying to disposition the service requests.
Figure 2 shows the path that each field representative travels in a given time period to disposition the service requests that he/she has been assigned. Again, these paths were generated using a first-in, first-out assignment to field representatives as they become available after dispositioning their previous assignment. With this approach field representatives are spending a lot of time driving between calls (as shown in blue) in an almost chaotic or random fashion. Some field service representative may traverse the four corners of the city with the first-in, first-out method of assigning service requests.
Figure 3 shows the path taken by one field representative as an example. He/she bounces from call to call throughout the city and covers a lot of ground. All of that travel time is unproductive time in terms of responding to service requests. In addition, the time spent driving adds stress to the field representative’s life.
Part 3: Proposed Process
A first attempt at improving the efficiency of field representatives is to assign service request calls to the nearest field service representative. As the field representative completes a service request, his/her next assignment is provided as the nearest open service request. Though this process loses the chronological order of service requests, it certainly improves the efficiency of field representatives in dispositioning service requests.
Figure 4 shows the same set of service requests and initial field representative locations used in the baseline process. Now each assignment to a field representative is the nearest open service request. As seen from Figure 4, all field representatives are staying close to their initial location and shows greatly reduced driving between calls. In fact, the minimum distance algorithm almost yields a map where each field representative has their own area of service even though that was not the objective of the algorithm.
Figure 5 isolates one single field representative on his/her path through the service request assignments. Again the distance traveled is minimized for each assignment.
Part 4. Efficiency Improvements
The above sections show visual comparisons of a baseline first-in, first-out method of distributing open service requests to field representatives and an alternative minimum distance method. The visualization shows that the minimum distance method can greatly reduce the amount of driving (and downtime) for field representatives between service request calls.
The actual changes in mileage for the seven field representatives is shown in Table 1 along with the expected change in costs. The time was calculated assuming a travel speed of 20 mph while the costs were calculated using a rate of $100 per hour. Cost and downtime due to travel were reduced by a factor of four using the minimum distance method. This reduced transit time can both speed up the disposition of open service requests and allow field representatives a chance to take a short break between calls to destress.
Part 5. Commentary
This effort shows that a simple change in distributing service requests to field representatives can have a big savings in time and cost. A factor of four reduction in time and costs was achieved with a relatively simple minimum distance algorithm for distributing service request items to field representatives.
This effort ignored the fact that some service requests, though all are non-emergency requests, may have higher priority than others. It also ignores the fact that not every field service representative can handle every type of service request. It would not be difficult to accommodate these rules into the minimum distance algorithm.