October 10, 2017

All-Door Boarding Pilot on the Silver Line

The MBTA and BostonBRT piloted all-door boarding on the Silver Line 4 and 5. Here are the results!

The MBTA partnered with BostonBRT to pilot all-door boarding on Silver Line routes 4 and 5 (Washington Street) from May 24th to June 6th, 2017. The goal was to measure how service improved when passengers could board at each door instead of lining up at the front door. Before the pilot began, the MBTA and BostonBRT identified several different metrics on which the SL 4/5 was expected to improve during the pilot, including reliability, reduced dwell time, and customer satisfaction.

For the MBTA a key improvement would be a reduction in the variability of the time it takes the bus to make a trip, whether that is due to dwell times or traffic. Since the MBTA schedules buses to take a certain amount of time, large variance in the time it takes a bus to complete a trip can either cause unreliable service or extra time at terminals to account for the occasional longer trip. Since we are interested in variability, in addition to the median we often look at the difference at the 90th percentile.

Data Sources

Working with BostonBRT, the MBTA developed a short intercept survey to ask passengers alighting the Silver Line during the second week of the pilot. The survey asked passengers about their general perception of the Silver Line, their satisfaction with their most recent trip (during the pilot) and how it compared to their general Silver Line experience, as well as the importance of several elements of bus service. Approximately 900 people answered the survey.

The MBTA used its Automated Passenger Counters (APC) to collect ridership, location, and time data during the pilot. APC are installed at the doors of MBTA buses and essentially count people getting on and off. The system also records the time when the bus opens or closes its doors, allowing the MBTA to calculate the dwell time (the time between when the doors are opened and closed at one stop) or running time between stops (the time between closing the doors at one stop and opening them at the next stop).

One limitation of using the APCs to collect data is that they can't accurately measure the impacts at the terminal stations. At terminals buses may arrive early from the previous trip, allowing operators to leave the door open longer than just the time to board the passengers, so we don't know how much of the door open time is necessary. On the Silver Line 4 and 5 the terminals (Dudley, Temple Place, and South Station) are responsible for 38 percent of the boardings.

The MBTA also used its Automatic Vehicle Location (AVL) system to collect data on the amount of time it took to run each trip. This system also evaluates each trip against the Service Delivery Policy standards to determine whether a trip is considered “on time” or not.

In order to measure the pilot's impact on SL4/5 dwell time and reliability during the pilot, we compared APC and AVL measurements during the pilot to those during normal operations. When making such comparisons, we ideally want to have as similar conditions as possible to the pilot, so that most changes between the pilot and baseline can be explained by the experimental condition of the pilot. However, road conditions are constantly changing, making such “apples to apples” comparisons difficult: temporary construction projects and background traffic change from one year to the next, ridership patterns might change from one month to the next as schools let out for the summer, and the SL4/5 fleet itself changed as new buses came into service in 2017. To try to account for these different factors, we identified three baseline periods: May 2-24, 2016; May 25-June 7, 2016; and May 1-23, 2017. We then compared these to May 24-June 6, 2017 during the pilot period, looking only at non-holiday weekdays, when ridership is typically the highest.

APC Results

We analyzed the data in many different ways, and in general, the results showed modest benefits at the median, and very large benefits at the 90th percentile. One of the reasons for this is that buses make many stops (we collected roughly 24,000 stop records during the pilot, and 64,000 stop records during the three baseline periods), but about 65 percent of those stops involved passengers getting off the bus and two or fewer passengers boarding. These are not the situations when all door boarding is expected to improve bus service. Indeed, dwell time savings during the pilot increased with the number of passengers boarding at a given stop. As the figure below demonstrates, dwell times at stops with 30 or more boardings were cut in half during the pilot, saving roughly one minute each. Such reductions would tend to have a big impact on improving reliability, because they are more effective at shortening our longest dwell times.

Chart showing median stop dwell time by number of boarding at each stop.

In looking at reliability via our AVL data, we focused on on-time departure from the route start points, because this would help us approximate dwell times at the terminals, where APC data are not as useful. In theory, if passengers are able to board more quickly at the terminals, then more trips should depart on time. As the figure below shows, on-time departure increased in both directions during both the AM and PM peak, when ridership is greatest.

Chart showing on-time departure percentage at terminals during the pilot and baseline periods.

Survey results

In general, SL 4/5 riders are happy with their existing service on a typical day (with 80% of respondents to the survey reporting being at least somewhat satisfied). Satisfaction with the trip during the pilot period was even higher, with 90% of respondents being at least somewhat satisfied. Generally in surveys we see higher satisfaction with a particular trip than with MBTA service overall, but the gap for the Silver Line survey is bigger than we normally see.

When directly asked if their trip during the pilot was faster, slower, or about the same as usual, 65% of respondents reported their trip being at least somewhat faster than usual (with an additional 30% reporting no difference). Most importantly, only 5% of respondents reported that their trip was slower than usual, indicating that the improvements in variability reduced the likelihood of very slow trips (see the chart below for full breakdown of responses).

Additionally, 70% of respondents reported being at least somewhat more likely to use the Silver Line in the future as a result of their reported trip.

Chart showing survey responses answering whether their trip was faster, slower or a similar time as normal.

When asked to rate the importance of aspects of bus service that can be improved by introducing BRT features (including bus service frequency, boarding speed, crowding, and travel speed due to traffic congestion), respondents identified service frequency as most important (6.4 on a scale of 1-7, between important and extremely important), followed by fast boarding, unrestricted travel speed, and uncrowded conditions on the bus (see the chart below).

Chart showing survey results ranking the importance of various characteristics of bus service

In order to provide riders with best quality of bus service, incorporating two aspects of BRT into the high-frequency high-ridership routes will provide the most direct benefit in areas of most concern for the riders: dedicated bus lanes to improve service frequency, and all-door boarding to reduce crowding and boarding time.

Although unrestricted travel speed is not as important to the passengers as frequency of service, because the MBTA has a limited fleet of bus vehicles, by freeing up buses from congested conditions, the MBTA can offer more frequent service on the route without additional vehicles and drivers.

When riders can board at all doors instead of just the front door, boarding time is reduced (especially on Silver Line buses, which have multiple rear doors), and crowding is alleviated because riders can select the less-crowded areas of the bus for boarding (instead of fighting through the crowds at the one front door).

Conclusion

The pilot showed promising results for improving bus service with all-door boarding. All-door boarding can reduce variability by shortening dwell time at busy stops. This can improve on-time performance and shorten trips for passengers. The MBTA is working to implement this change with the procurement of a new fare collection system. In addition, the MBTA is working with municipal partners and BostonBRT to pilot other aspects of improved bus service.