A recent review of the city's library master plan showed that at least one additional library branch is needed to meet the needs of a growing population. The library board estimates that the cost for constructing a new branch will be at least $3.5 million and could go as high as $5 million. The board and city officials realize that, unfortunately, funding for a new branch will not be available in the near future. So the board decides to review the library's programs to see if there are ways to better meet the needs of the community until an additional branch can be built. They ask the city's IT department to compile demographic information about residents living near the existing library branches so they can tailor the services provided by each branch accordingly.
The additional services provided at the existing branches as a result of the review have been well received and library usage is up slightly, but mainly among existing library patrons. The board believes that there are many more residents who could benefit from the resources the library provides, especially among underserved populations in the city, such as seniors, low-income residents, non-English speakers, and children. Many of these residents are unable to travel easily to a branch. The board members—who hail from the community—would like to be more proactive in bringing library resources to these underserved populations.
Books on wheels
The board members are well aware of the county's bookmobile program, and they decide to look into creating a bookmobile program for the city. The first hurdle, as always, is money. Buying a bookmobile, it turns out, would cost $150,000 to $200,000. Even a used one, which would likely need upgrades and repairs, would cost upwards of $50,000. And then there are the operating costs. There is no additional money in the library budget for a bookmobile, and it's unlikely any will be forthcoming from the city council.
With a little research, the board members discover they can apply for grants to provide services to underserved communities and can use the money to help fund the bookmobile. They know, though, that the competition for grant money is fierce, and the amount of funds available is limited in any case. And at this point they're not really sure how many residents would actually use the bookmobile.
After a series of discussions, the members decide their best bet for bringing a bookmobile to the city would be to partner with the county library system. Their plan is to contract one of the county's two bookmobiles one day a week, on days the bookmobile is otherwise idle. The city would cover the operating cost of the bookmobile on those days and would also offer to buy additional materials geared to the underserved population, such as Spanish-language and large-print books.
The members believe this would be much less expensive than the city buying and operating its own bookmobile. It will also allow the city to operate a bookmobile on a trial basis to gauge the response from the community. If the program is successful, the city might decide to buy a bookmobile in the future.
Before presenting their idea to the city council, and ultimately to the county supervisors and the county library administration, the board members want to be ready with a proposal. So they ask the city's IT department to explore the feasibility of creating bookmobile routes that will serve the target populations. The results of the analysis will also be included in the grant proposals the board will eventually submit.
Specifically, the members would like to know where the underserved populations live within the city, where the best bookmobile stops would be to serve the most people in these communities, and whether there are viable routes that can visit the stops in the time allotted on a single day.
Communities in need of books
Over in IT, the same analyst who did the earlier spatial analysis for the library board volunteers to take on this new task. For that analysis she used ArcGIS Online. Since this current project will ultimately involve creating detailed routes and schedules, she will use ArcGIS Pro. ArcGIS Pro includes all of the tools she'll need for the demographic analysis, site selection, and routing.
The board's main goal is to improve access to library materials and services for residents in lower-income areas, particularly if those residents are members of an underserved group—specifically, kids, seniors, or Spanish-speaking adults. The members will want to see if there are particular neighborhoods where these populations tend to live and ensure the bookmobile stops are located in or near these neighborhoods.
The analyst decides to create a map showing income distribution across the city, to provide an overview, and then to create maps of where people in the underserved populations live. Since the analysis is at the city scale, she decides to use census block boundaries as the basis for the maps and the subsequent analysis. Census blocks will provide a finer level of resolution than block groups or census tracts.
The first step is to add the latest demographic information to the census block layer. After perusing the myriad variables available via the Enrich Layer tool, the analyst finds the variables that most closely match the data she wants to map. From the Key US Facts collection she chooses Median Household Income. From the At Risk data collection she chooses Households with Income Below Poverty Level, Children (Age <14), and Seniors (Age 65+). From the Language data collection she chooses Pop 18-64: Spk Sp/Eng NW (adults who speak Spanish and speak some English, but not well).
After running the Enrich Layer tool, the necessary attributes are now in the census block layer. First the analyst creates a map of median household income. It is clear that incomes are generally much lower in the western third of the city than in the eastern two-thirds. The board will want to make sure there are bookmobile stops in the western portion of the city.
The analyst then maps the underserved populations, looking for hot spots—essentially, adjacent blocks that all have a relatively high number of people in a particular category.
The two clusters of households with income below poverty level occur in the western part of the city where incomes are lower overall. The same is true for Spanish-speaking adults. For kids and seniors, there are hot spots in the low-income area, but also in higher-income parts of the city.
After some heated discussions, the board members decide to focus on locating the bookmobile stops to best reach seniors and Spanish-speaking adults. Since many of these residents are concentrated in the lower-income part of the city, locating stops to serve these populations will likely also serve kids living in low-income households in this area. In the end the board members agree that, while ideally it would be good to extend bookmobile services to all the kids in the city, the primary goal is to reach low-income residents.
Creating guidelines
The first topic of discussion is to decide how many stops the bookmobile can realistically make, given that it would be visiting the city only one day per week. Library staff have been researching bookmobile routes and schedules from across the country. Most schedules limit the number of stops per day to no more than three or four, with stops lasting an hour or an hour and a half. That's also true of the county's current bookmobile schedule. With preparation time before the route; travel time between stops; breaks; and additional downtime between stops to reshelve books, process paperwork, and clean up the bookmobile, three or four stops makes for a full day.
The board members are adamant that there need to be four stops at a minimum, but are concerned that even four stops will not reach enough residents to make the bookmobile project worthwhile. The library staff point out that, in fact, most routes run every other week rather than weekly. If the city takes this approach, they will be able to have eight stops in total on the schedule—four one week and four the next. The board members agree that this is a good approach and will greatly increase the number of residents potentially using the bookmobile. They decide that there will be two routes with four stops each, running on alternate Mondays, when the county bookmobile is currently idle.
Now the discussion turns to the specific types of locations to consider for stops. Library staff report that they've found that nursing homes and senior day care centers are common bookmobile stops, as are apartment buildings, mobile home parks, schools, and shopping centers.
For the stops geared toward Spanish-speaking adults, the board members decide to focus on locations that, while located near these populations, will be easily accessible to all nearby residents—specifically, shopping centers and malls. Since seniors are less mobile and less likely to travel very far to visit a bookmobile, the members decide to use nursing homes, assisted living facilities, and adult day care centers as potential stops. These stops will serve at least the residents of the facility and hopefully attract residents from any nearby facilities or retirement communities. The stops geared to seniors will be scheduled during the morning and early afternoon, while the stops at malls and shopping centers will be scheduled later in the day.
The GIS analyst raises the issue of whether it's acceptable to have a bookmobile stop near an existing branch. The library staff report that on the bookmobile routes they researched, about half of stops on a given route are between three-quarters of a mile and two miles from a library, with the rest farther than two miles. They also note that most patrons using the branches live within about a mile of a branch. Since many seniors are unlikely to travel to a branch library more than a few blocks away, the board members decide to include any location that serves a large number of seniors, even if it is near an existing branch. To ensure that at least some stops serve parts of the city not near a branch, they decide the stops geared to Spanish-speaking adults should be at least a mile from a branch. Part of the goal of the bookmobile is to increase visibility for, and access to, library services, so even people living just over a mile from a branch might not have made the effort to go there, but will possibly use the bookmobile if they see it in their neighborhood. That includes, of course, any resident of the city, not only those in the underserved communities.
Finding the best stops
The GIS analyst now has enough information to find the best locations for the eight bookmobile stops.
Location-Allocation analysis allows her to identify the combination of four stops that will serve the most seniors and another set of four stops that will serve the most Spanish-speaking adults. Later, the stops will be grouped into two routes so that there are two stops for seniors and two stops for Spanish-speaking adults in each route.
The analyst starts with finding the optimal stops for seniors. The city's GIS database includes most of the layers from the countywide association of governments database. The IT department has already extracted the portion of each layer within the city's boundary. The GIS analyst sees that the Business Sites layer includes nursing homes and assisted living facilities, which are classified as Rest Homes in the data. Another layer includes adult day care facilities. As it turns out, there are only two of these in the city. She'll merge the Rest Homes and Adult Day Care centers to create a layer of potential stops.
Before merging the layers, she checks to see which, if any, fields are common to both layers. She notices that the Business Sites layer does not have a Name field, only an owner name, which is not necessarily the name of the facility. It also includes a street address stored in several separate fields. The Adult Day Care layer does contain a Name field, as well as a single address field (ADDR).
From previous experience, the analyst knows that the Location-Allocation tool, Solve Location-Allocation, automatically carries forward a field called NAME if it exists in the input facilities layer. If the field is blank or doesn't exist in the input layer, the tool assigns a generic name:Location 1, Location 2, and so on. She decides it's better to have something more meaningful—even an address—that can be used to identify the stop. And, in fact, she's pretty sure using the address as an identifier will be more helpful for people trying to find the bookmobile than a facility name would be.
After merging the layers, the output layer contains the Name field (carried over from the Adult Day Care layer). The analyst then assigns the street address to the Name field.
A conundrum—the Rest Homes layer contains an owner name field that is not unique and is not the name of the actual facility. The street address is contained in three separate fields. The Adult Day Care layer does contain a facility name (in the NAME field), and also includes a street address (in a single field). The goal is to have a unique identifier for each facility stored in the Name field. The identifier will be carried through the Location-Allocation analysis.
By merging the layers and calculating the NAME field equal to the address field (or the concatenated address fields in the case of the Rest Homes layer) a unique and meaningful identifier is created for each potential bookmobile stop.
The merged layer contains twelve potential stops: ten rest homes (nursing home or assisted living facilities) and two adult day care centers.
There are 12 senior facilities representing potential bookmobile stops. The facilities at 111 and 183 3rd Avenue are adjacent and appear as a single point on the map. Most of the facilities are located in the lower-income western portion of the city and are near senior population hot spots.
Now the analyst can proceed to run the Location-Allocation analysis to find the four optimal stops. Location-Allocation essentially assigns census blocks to the closest stop, totals the number of nearby seniors for each stop, and then finds the combination of four stops that results in the highest number of seniors near a stop. While the hot spot map showed concentrations of seniors, Location-Allocation uses the population of seniors in all census blocks within a specified distance from a potential stop. The tool requires the demand locations (census blocks, in this example) to be point features and the population associated with each block to be stored in a field named Weight. The analyst converts the census blocks to points, adds the Weight field, and calculates the values to be equal to those in the original field containing the senior population—Seniors (Age 65+).
The analyst uses the Maximize Attendance problem type, which assumes fewer seniors will visit as the distance from the bookmobile stop increases. She also assumes, for the purposes of the Location-Allocation analysis, that seniors who have to walk, or even drive, more than a mile to a stop will not travel to that stop. The analyst knows that, in reality, some seniors may not be willing to travel a mile to a stop, while others will travel more than a mile. Her goal here is to enter some reasonable assumptions to help determine which combination of four stops will potentially serve the most seniors relative to all of the possible combinations.
The Location-Allocation analysis identifies the four stops that serve the greatest number of seniors. The allocation lines on the map indicate the chosen stops and the census blocks that are assigned to each stop. The chosen stops are indicated in the table. While two of the stops will be visited one week and two the next, together the four stops will provide access to the bookmobile for the greatest number of seniors. The analyst sees that several of the stops are near an existing library branch, but she decides it is not a problem based on the library board's discussion about distances between libraries and bookmobile stops.
Next the analyst turns her attention to selecting the set of stops that will serve Spanish-speaking adults. She can reuse the layer of census block points she created. To start, she calculates the existing Weight field equal to the population of Spanish-speaking adults in each block. This value will be used in the Location-Allocation analysis to ensure the four selected stops serve the greatest number of Spanish-speaking adults overall.
With the census block layer now ready for the analysis, the analyst sets about creating a layer of potential bookmobile stops for this population. The database includes a layer called Places, which contains landmarks such as government buildings, hospitals, schools, and churches. It also includes malls and retail centers, which the library board decided would be the best locations for the stops serving Spanish-speaking adults. She selects these features and creates a new layer of potential stops. There are 30 in total.
The analyst notices that some of the potential stops are very close together or even basically at the same location. After taking a closer look, she sees that in some cases a large retailer within a shopping center is included as a separate feature in the Places layer (as a Retail Center), and both the Retail Center and the Mall show up on the map. In other cases, a shopping center is listed twice—once as a Mall and once as a Retail Center. The analyst knows that Location-Allocation will not choose two stops that are very close, since they would both serve the same population. Instead, it will try to maximize the population served by choosing stops that are farther apart. So, in this case, having essentially duplicate stops will not impact the analysis.
The library board decided that, unlike for seniors, bookmobile stops for this population should be more than a mile from a branch library. So the analyst next creates 1-mile areas around the library branches (assuming travel over streets, not straight-line distance) and uses this layer to eliminate any of the potential stops within a mile of a branch. That leaves 16 potential stops.
Since the assumption is that Spanish-speaking adults might in fact visit a branch if it is nearby, rather than using the bookmobile, the analyst decides to apply the Maximize Market Share problem type for this Location-Allocation analysis. This will allow the analyst to include the library branches as competitors in the analysis. The assumption is that the closer someone lives to a branch, the less likely they are to travel to the bookmobile stop. Of course, that may not always be the case, but the end result for the analysis will be to favor those locations that are farther from a branch while still serving the greatest number of Spanish-speaking adults.
For the analysis, the branch libraries—which will be treated as competitors—have to be included in the layer of potential stops, so the analyst merges the layer of 16 potential stops with the library layer to create the input layer of potential stops.
As with seniors, the analyst assumes that the maximum distance people will travel to a bookmobile is one mile.
Maximize Market Share includes the option to specify how much more likely people are to visit one of the candidate facilities (in this case, bookmobile stops) versus one of the competing facilities (the branch libraries). The Location-Allocation analysis considers both the distance people live from a branch or a stop, as well as the relative attractiveness of each, as it determines which stops will serve the most people.
At this point the analyst doesn't really know how much more likely people are to visit a bookmobile versus a branch, so she assumes they are equally likely to visit one versus the other (assuming the distance to each is the same). In the future, once the bookmobile is up and running, the city can survey patrons and get actual data that can be incorporated when the routes are reviewed and revised.
The analyst runs the Location-Allocation analysis, and identifies the four stops that will serve the greatest number of Spanish-speaking adults.
The Location-Allocation tool finds the combination of four stops that will serve the greatest number of Spanish-speaking adults within a mile. The allocation lines on the map indicate which census blocks the stops and the competing library branches draw population from. In the table, the FacilityType field indicates the chosen locations.
Creating the routes and schedules
With the eight optimal stops identified, the analyst can now create the two routes. The first task is to combine the two sets of stops into a single layer.
To make it easier to track and display the two types of stops (retail centers and senior facilities), the analyst adds a Type field to the combined layer and assigns the appropriate value to each stop.
The analyst will use the Solve Vehicle Routing Problem tool to group the eight stops into the two most efficient routes (that is, shortest overall travel distance), with four stops on each route. Solve Vehicle Routing Problem creates routes that are the most efficient (shortest time or distance) within specified parameters such as time windows, break times, and route start and end times and locations. The parameters are specified in a set of feature layers and stand-alone tables.
The analyst specifies that each route will have two stops for seniors, which will be 90 minutes long and will occur between 9:00 a.m. and 3:00 p.m., and two stops at shopping centers serving Spanish-speaking adults (and other nearby residents), which will be 60 minutes long and will occur between 3:00 p.m. and 8:00 p.m.
In addition, the library board wants to offer residents the option of requesting books from any of the branch libraries and picking them up at the bookmobile. To save time, any requested books will be sent ahead of time to the Civic Center Branch, where the bookmobile will pick them up. Similarly, any books being returned will be dropped off at the same branch at the end of the route. So the analyst adds the Civic Center Branch to the stops layer—this location will be both the start depot and the end depot.
Since the routes will last the better part of a day, the analyst includes a 30-minute lunch break on each route. She specifies that the break should occur sometime between 12:00 p.m. and 2:00 p.m.
The Solve Vehicle Routing Problem tool takes into account the travel time between stops when creating the routes, but the analyst adds an additional 20-minute period before the first stop and between subsequent stops to allow for heavy traffic, maneuvering the bookmobile into position at each stop, paperwork, reshelving books, and cleanup.
With all the parameters for the routes in place, the analyst runs the tool to create the routes. To make the results easier to understand, she creates a map for each route showing the stops on that route.
When sorted by RouteName and Sequence, the output table shows the order of the stops for the two routes. In the StopType field, a value of 0 indicates a stop, 1 a depot, and 2 a break. The FromPrevTravelTime field shows the time required to travel between stops plus the 20 minutes of additional allotted time between stops.
Garnering political support
With the analysis complete, the analyst needs to present the results to the library board, and ultimately to the city council. To do this, she creates a story map showing not only the final stops and routes, but also a series of maps that show how many people from each of the underserved populations are within a mile of a proposed bookmobile stop.
She already has the layer of 1-mile drive areas around the libraries that she created earlier, so she only needs to create the layer of 1-mile drive areas around the combined library branches and proposed bookmobile stops. She then sums each demographic category within each layer.
As it turns out, adding the bookmobile stops nearly doubles the number of people served in each category.
The library board members review the story map showing the results of the analysis and are cautiously optimistic. They decide to go ahead and brief the city council.
The reaction from the council members is positive. However, one member is adamant that a State Veterans Home that is located in the city be included as a stop (as it happens, the Veterans Home is in the council member's district). The library board members point out that there are four stops on each of the two bookmobile routes that are being proposed, which is already pushing the limit of a full workday. Adding a fifth stop to one of the routes will likely entail additional costs, such as overtime pay for staff. Nonetheless, there is strong support among the council members for including the Veterans Home. The city manager, who is also attending the meeting, quickly estimates any additional costs for the extra stop are likely to be less than $10,000 per year. The council members agree to include funding for the bookmobile in their next budget to cover the cost.
After the meeting, the GIS analyst sets about adding the Veterans Home stop to the schedule.
Since most of the veterans at the home are elderly, the analyst will treat it as a senior facility. To ensure that the revised routes are still the most efficient, she'll need to rerun the Vehicle Routing Problem analysis. The Places layer contains the Veterans Home-she selects it, adds it to the layer of stops, and re-runs the analysis. The new routes are created and the Veterans Home is assigned to Route A.
However, as expected, Route A is now much longer—over 10 hours from start to finish. The analyst will need to add a second break to that route late in the day, between 5:30 p.m. and 6:30 p.m.
The analyst runs the Solve Vehicle Routing Problem tool one more time to create the final routes and schedules.
The analyst notices an issue with the schedule: the time window for 1382 Tobias Drive starts before the 3:00 p.m. cutoff, but extends to 4:20 p.m., which is outside the preferred time window for stops at senior facilities. Since the bookmobile arrives and departs before dinner time, she decides this is an acceptable compromise in order to fit in the stop at the Veterans Home.
The library board and city council members sign off on the final stops and routes and form a committee to present the plan to the county officials and begin the grant process.
Rollout of the plan
A few months later, two of the grants come through, and the city signs a one-year contract with the county for use of the bookmobile, with an option to extend the contract at the end of the first year. The county library management's only stipulation is that since the materials are part of the county's collection, city residents will need to have a county library card in order to check out materials. However, residents will be able to obtain a card at the bookmobile and use it immediately.
The GIS analyst creates a story map of the bookmobile stops and schedule for the library's website—her final task for this project, for now. To make the map and schedule more user friendly, she changes the route names. Taking a cue from the colors of the stop markers on the map, Route A becomes the Green Route, and Route B becomes the Blue Route. She also rounds the schedule times to the nearest five-minute interval.
Library staff will monitor attendance and material use at each stop, and a decision will be made at the end of the three months on whether the stops or routes need to be adjusted. If the program is successful, and additional funds are available, the city may eventually buy its own bookmobile and extend library services to even more residents across the city.