I run service area drive times constantly for my job. I always run the same parameters: toward facilities, 20 and 40 minute cut-offs, day of week (Monday, 8am), standard precision split rings.
The problem I have only occurs in a couple of areas of the country, most notably Phoenix, AZ. In the Phoenix area, when I run the drive times for all our retail locations, I encounter a couple of issues.
1) Our store in the suburb of Peoria generates a very tiny polygon for the 0-20 minute drive time, but that polygon is nowhere close to the store. The 20-40 minute polygon includes all the area immediately surrounding the store, as if it takes 20 minutes just to exit the parking lot! I've included a screenshot showing store #3 with the yellow area being the 20-40 minute drive time, and the red circled area showing the microscopic 0-20 minute drive time.
2) We have 2 stores in the suburb of Chandler, very close to each other (approx. 0.8 miles straight-line distance). The 0-20 minute drive time for store #14 is very small, and there is no 20-40 minute drive time polygon. However, the other store (#10) has a 0-20 minute drive time that completely surrounds the 0-20 drive time of the other store! I've included a screenshot showing the 2 stores and their drive times. The 0-20 drive time for store #14 is in light red, the 0-20 drive time for store #10 is in light yellow, and the 20-40 drive time for store #10 is in darker yellow.
I've also included the attribute table from this process. When I try to run this using high precision, I always get an error that prevents the analysis from running. When I use generalized precision, issue #1 seems to go away but it does not help with issue #2.
