Express an interest in transit at just about any cocktail party in the Rainier Valley, and you’ll hear how what Sound Transit really needs to do is provide a shuttle to get people to the stations. Inevitably, people are proposing a solution to their specific problem without much awareness of scale or efficiency. Much like park-and-ride spaces, shuttles are probably more effective at allowing people to conceive a way to use light rail than actually providing that access at scale. On the other hand, Metro and ST seem to have worked their way into a contract that projects a pretty good yield from what some might call coverage service.
The Federal Transit Administration is willing to give shuttles a shot, possibly anticipating that autonomous vehicles will eventually transform the economics. Metro and Sound Transit won $350,000 from FTA in a research project combined with LA Metro. This sum, combined with $100,000 each from Metro and ST, would have funded a peak-only shuttle at a couple of Seattle stations and Tukwila International Blvd, according to Project Manager Casey Gifford of Metro.
Enter Seattle, with Transportation Benefit District funds that Metro doesn’t have the capacity to serve with more buses. Its $2.7m contribution dramatically expanded the concept to include four Seattle stations and service over the full span of Link operations. Tukwila, which didn’t top it up, is only available from 6-9am and 3:30-6:30pm.
For the next 12 months or so, riders traveling between a Link Station and the areas shown above can use an app or phone number to summon a minivan operated by Via. They can pay for the ride just like any Metro bus, except for cash: an ORCA that fully transfers to Link, or a Transit GO ticket that doesn’t. Although this payment scheme will shift some more ORCA revenue from ST to Metro, only the small amount of Transit GO tickets (and additional volume) would put more fare revenue into the system as a whole.
The $3.2m budget funds a year of Via vehicles, plus startup and marketing costs. The external vendor provides a few different benefits. The use of non-unionized “contractors” reduces labor costs. Less controversially, like many transportation network companies, Via claims that its data collection and algorithms will rightsize the vehicles operating at any given time, position them to optimize the waiting time for riders, and combine trips.
Metro and ST pay Via per vehicle hour. The current budget is based on some assumptions on how many vans (out of a fleet of 18) Via needs to deploy, for a total of 75,565 hours. If lower-than-projected ridership requires fewer vehicles, there would be money left over. If high demand requires more vehicles than planned, the agencies would need to either come up with more money or accept higher customer wait times. This comes out to about $42/hour including overhead, but the marginal rate is $40/hour, over three times cheaper than a Metro bus.
The contract has a number of Key Performance Indicators that will help both determine the number of active vehicles and whether the pilot should continue beyond the initial year. These include serving 80% of requested trips, a rider rating of 4.5 stars or higher, wait times of 10 minutes or less, and vehicles serving 2.5 riders per hour. More vehicles would reduce times and increase success rates while reducing efficiency, so there are difficult tradeoffs for the algorithm. The 2.5/hr figure means that the cost would be $16 per rider. For comparison, ST Express runs about $7/rider and Sounder runs about $11 for many more passenger miles.
Other press reports mentioned another metric of 1,000 rides per week overall. While this is indeed in the contract, Ms. Gifford characterized that as “an error,” copied over from an example LA Metro contract and not numerically consistent with the other estimates in the program. Indeed, these other metrics suggest over 3,600 rides per week. She emphasized that the agencies would “look holistically at performance,” including “who is being served and what gaps it is filling.”
Although it’s very early going, Ms. Gifford was able to share some promising early results. Over the first two weeks, each vehicle carried 3 passengers per hour (over the target of 2.5) and the average wait time was 7 minutes (vs. 10). If Via holds the 3 riders/hr metric, that would bring cost per rider to about $13. While not competitive with the most productive routes, this is competitive with other routes that might be viewed as “coverage” — providing service to underserved areas. If Via maintains this performance, the case for continuing will be relatively strong.