Service areas

<< Click to Display Table of Contents >>

Navigation:  Network and Location Analysis > Location and Service Area Problems >

Service areas

The maps illustrating demand assignment in Figure 7‑15D and Figure 7‑16B provide examples of service area definition. Defining service areas or zones (often without regard to specific demand levels) is a widely supported facility within many GIS packages. These areas are typically discrete zones that are closer in distance, time or cost to specific points in the network than to any other specified point (rather like Voronoi regions in the plane — see also, Figure 4‑35).

Note that the specification of service areas in this case is not the same as solving coverage problems. The latter does not assume that the locations or numbers of facility sites are known — these are outcomes of the modeling process. In the pure or set coverage problem the number of facilities must be increased until demand from all eligible customers is met within the given service parameters. This will often result in unacceptably high numbers of facilities, requiring some form of relaxation of the constraints to be of use in real-world situations. An example of such relaxation, known as the maximum coverage problem, is to pre-specify the number of facilities (or vehicles etc.) to service demand, and then seek to maximize the customer coverage within the service constraints. Relaxing the service constraints (e.g. allowing the time to reach a patient with an ambulance not to be an absolute maximum, but an average for example) leads to p-center and p-median type problem specifications.

An example of GIS-based service area functionality is provided in Figure 7‑17A. This shows the locations of three ambulance stations situated in an urbanized region. In Figure 7‑17B all streets closest in network distance to each station are assigned as the primary service area for each station. As noted above, this form of simple network partitioning assumes pre-defined facility locations and takes no account of variations in demand (e.g. expected accident or illness rates) or in supply (e.g. number of vehicles available). The example illustrated was generated using the TransCAD software, but similar facilities are provided in many other GIS packages, such as the and functions in GRASS, the “service areas” facility in ArcGIS Network Analyst, the Voronoi function in SANET (see also, Figure 4‑35) and the Allocation functionality provided in the TNTMips Network Analysis facility. The latter tool, for example, incorporates options to allocate service areas represented by network links based on flows to or from selected locations, with or without specifying from a range of parameters (such as demand levels, capacity constraints and travel impedance).

Figure 7‑17 Service area definition

A. Ambulance locations

B. Service areas (distance bands)



Travel time zones

A related facility to service area definition, also provided in many GIS packages, is that of identifying travel time zones. These are typically generated as a polygon layer, overlaid on the network, indicating bands of travel times or distances. Figure 7‑18 shows a set of bands that have been constructed around a point location on the Salt Lake City street network. These bands have been generated using distances in meters, with the edge of the band being the upper limit of the distance interval. Note that these zones include areas that have no roads, and the procedures for constructing such areas varies between packages, and may or may not allow zone construction procedures to be defined. Manifold implements three different options: (i) a convex hull; (ii) a buffer; and (iii) a zoning system, similar to the buffer operation, but in which off-road speeds (i.e. walking, off-road driving) are separately specified. Computation times can be very lengthy depending on the options selected and the complexity of the local network.

Figure 7‑18 Travel-time or drive-time zones

Drive time zones — Bands: 500,1000,2000,5000 m