April 17, 2017

Rethinking the park and ride for the 21st century: Part I

Two major trends in mobility mean that it is time for transportation agencies to rethink the role of the humble park & ride as a tool for promoting more sustainable transportation. By enhancing pickup and dropoff facilities and accommodating self-parking cars, a park & ride can increase both the number of travelers served and the quality of the travel experience. In a future post, I will explore the case for spending public funds subsidizing mobility services for first/last mile service, instead of subsidizing parking through park & rides.

As new mobility services such as transportation network companies (TNCs, also known as ride-sourcing services) continue to gain in popularity, it remains an open question whether they will complement or compete with traditional mass transit services. There are plausible arguments that TNCs could either complement or compete with mass transit. On the one hand, TNCs can provide door to door transportation with excellent comfort, speed, and reliability, at a modest price, allowing the traveler to multitask, and without the cost and hassle of parking. From an individual traveler’s perspective, they deliver most of the benefits of transit, but without the downsides. On the other hand, TNCs are still more expensive than transit, especially for longer trips. TNCs are most likely to complement transit when a short, cheap TNC ride can be linked to a longer bus or train ride.

(Before going any further, you might ask: Does it even matter if TNCs draw ridership away from mass transit? I share Yanbo’s assessment that the answer to this question really depends on when and where. In low-demand areas and/or at off-peak hours, on-demand car services directly from origins to destinations can make more sense both economically and environmentally than infrequent fixed route service using near-empty buses . But on high-demand routes at peak hours, buses (and trains) remain significantly better than cars (even shared cars) in terms of emissions, cost, and capacity (i.e. the number of passengers per hour that can be moved through a given amount of transportation right-of-way). Thus, I am working from the premise that we ought to try and promote greater use of mass transit during peak hours on busy routes, and that the most socially beneficial role for TNCs in this context is to provide a first/last mile solution to make mass transit more attractive.)

It is hardly a new revelation that cars can complement transit when the former are used as a first-mile solution. This is the basic idea behind the park & ride: allow people to drive to a central collection point from which they can use mass transit to get to their ultimate destinations, thus reducing the number of cars that must drive into and park in higher-density areas of the city. The park & ride is a beautifully simple idea, but it also runs up against some simple problems.

One problem with the park & ride concept is that people really don’t like transferring between multiple modes in the course of a trip. Researchers Martin Trépanier, Catherine Morency, and Isabelle Gossmann of Ecole Polytechnique de Montréal found that travelers value the inconvenience of each minute of walking time at a park & ride as equivalent to six minutes of time riding the bus. Researchers in the Netherlands found that travelers valued increases in transfer time several times more heavily than increases in travel time. Naturally, the more people who are using a park & ride, the larger the parking lot must be and so the farther each traveler must walk from their car, on average. Thus, as a park & ride grows, it appears destined to become a victim of its own success, as it becomes less convenient as a transfer point (although this may be offset to some degree by the ability of a larger facility to support more routes or greater frequency).

A second problem is that park and ride lots offer limited capacity. In the Seattle region, for example, 30% of King County Metro’s park & ride lots, comprising more than half of the system’s parking spaces, are routinely at capacity by 9:00 AM on weekdays.

Accommodating TNCs can help with both of these problems. In the near term, a dedicated drop-off area right next to the station entrance or bus loop can be reserved for those being picked up or dropped off (a “kiss & ride”, although I don’t recall the last time I kissed my Uber driver goodbye). This reduces the time spent walking through a (possibly cold, wet, dark) parking lot, making the transfer less burdensome. Second, the cars don’t need to remain parked at the park & ride all day, since the drivers will use them to pick up other passengers. This means that a single space can support pickups and dropoffs for hundreds of travelers each day, instead of just one parked car.

Improved and expanded kiss & ride facilities will also be important as vehicles become more automated. An early application of automation is in self-parking:

A parking lot represents a relatively controlled environment with low speeds, in which a car with self-parking (auto-valet) capability could drop its passenger(s) at the kiss & ride and then park itself. This functionality is likely to be available sooner than the full self-driving capabilities that would enable automated taxis. Moreover, self-parking cars could park themselves in the back of the parking lot, leaving more desirable spots for manually driven cars. Finally, self-parking cars may be able to increase the capacity of parking lots. Researchers from Shanghai Jiaotong University presented work at the 2016 Automated Vehicles Symposium (not available online) about optimizing aisle widths and parking angles for self-driving cars:

Unfortunately, this work has not yet been completed, but as a lower bound, we can consider the width of cars versus that of parking spaces. At  King County Metro’s Eastgate Park & Ride, parking spaces are 2.6 m (8.5 feet) in width.

In comparison, the width of a typical car is 6-6.5 feet, as shown by the examples in the following table:

Make & Model	Width (inches)
Toyota Camry	71.7
Toyota Corolla	69.9
Honda Pilot	78.6
Chevy Malibu	73

Let’s assume that these cars can be parked in parking spaces with an average width of 7 feet, since the car can optimize its path in and out of the space, and doors do not need to be opened and closed while the car is parked. Compared with spaces 8.5 feet wide, this would allow 20% more cars to be parked in the same area, even before reducing aisle widths.

Increasing vehicle automation, particularly Level 4+ automation, will make these changes more urgent. To the extent that automation is adopted by TNCs, it can reduce the cost of TNC trips to the park & ride, making TNCs more attractive for these first-mile trips. This is a major opportunity to increase transit ridership. In their recent working paper, Cost-based Analysis of Autonomous Mobility Services, researchers from ETH Zurich showed how automation can make the costs of taxi-type transportation competitive with those of personal car travel (Read the figure as a paired bar graph. The red and blue bars do not indicate positive vs negative values.):

But at the same time, this also creates risks. Since automation is expected to decrease the cost and and increase the convenience of automobile travel, travelers may be tempted to take automated cars all the way to their final destination. The increased traffic will be less of a disincentive to increased travel than it is today, since travelers will be able to multitask while they let their cars deal with the traffic. Given these trends, it will be all the more essential for park & rides to accommodate automated vehicles – in both dropoff areas for TNCs and parking for privately owned cars – in order to make mass transit travel as competitive as possible.