I recently received some data from WMATA, in the form of the ridership for various half-hour portions of a typical May workday. Due to security concerns, the WMATA contact asked me to report the data only as “peaking factors”, which I define as dividing each bin by the maximum bin (see figure at right).
Based on the peak ridership, I decided to look into how many trains per hour WMATA runs to carry the peak load, and compare the change in ridership from the peak to the relative number of trains.
See this figure for the number of trains per hour. Because WMATA does not publish train schedules for peak periods, the data is not perfect, because it was collected by incrementing the departure time in the Ride Guide by 1 minute at a time. The train departure times are apparently not stored internally in any particular order, because though the ride guide reports three different options for travel to your destination, they are not necessarily the next three trains.
Additionally, I was interested in where people were entering the system on the red line. Last year, I obtained data for entries and exits during the morning peak hours, daytime, evening peak and late night, by station and month, for FY 2006 and some of FY 2007. This figure is a summary of red line entries by station for the AM and PM peaks.
Looking at the number of trains per half hour graph, it looks like WMATA runs at least 75% of peak trains between the hours of 7 and 9 in the morning peak, and between 3:30 and 6:30 during the evening peak. These times are also when WMATA ridership is above 60% of peak ridership.
This figure shows passenger entry peak factor divided by train frequency peak factor. This is the best metric I could come up with for estimating how crowded a typical train should “feel”. Since the ridership data is based on entries, I use the train peak factor at Grosvenor-Strathmore (20 minutes away from Farragut North) for the time periods before 12 noon, and the data at Farragut North for the afternoon data.
My hypothesis when I started looking into this was that WMATA was probably reducing train service too early in the evening, and should increase the frequency of trains because ridership would justify the additional service. At least for the gross metrics I could determine based on the data I had available for the red line, I have to reject that hypothesis. My metric shows that for the majority of the day during a normal weekday, my “crowding factor” remains close to 1.0 and averages about 0.8, which is desirable because you wouldn’t want your trains to spend most of the day during non-rush hour at the same standing load as during rush hour. Not only is that less comfortable, but the riders during the day time are more likely to be able to choose some other form of transportation because the roads are not clogged. If the metric had risen significantly above 1.0 during some off-peak period (indicating that WMATA had decreased the number of trains significantly more than ridership) that would indicate more crowded trains during that period. Fortunately for the riders, it looks like WMATA is doing a good job balancing ridership with available system capacity.
Thanks to a source at WMATA for the data.