Amtrak's poor on-time performance for Surfliner

Every month Amtrak, along with Metrolink and Coaster, provides on time performance data to the LOSSAN board. However, the numbers are not strictly comparable. While Metrolink and Coaster use an on-time definition of “within 5 minutes” for their trains, Amtrak, following the definitions of the Passenger Rail Investment and Improvement Act, defines an on-time arrival as within 10 minutes of the scheduled terminus. The question then arises as to what the performance is when using comparable definitions.

As detailed in the agenda for the March 5th, 2012 LOSSAN meeting, Pacific Surfliner performance between Los Angeles and San Diego had dropped to only 74.6% on time performance in January of 2012 under Amtrak’s definition, despite a schedule change on January 9th that was supposed to resolve passenger train conflicts. By contrast, San Diego’s Coaster achieved a 94.2% on-time performance and Metrolink’s Orange County Line was on-time for 97.7% of all trains.

When we delve into the data and calculate figures based on an on-time definition of within five minutes, rather than ten, Amtrak’s performance figures, as would be expected, drop significantly. Between January 9th, the first day of the new schedule, and March 5th, the last day for which information was available for all trains when I began the data collection, the weekday performance for trains terminating in San Diego was only 59.5%. 

Somewhat surprisingly, since many trains were late due to track work north of Los Angeles, the weekday performance for trains arriving in Los Angeles from San Diego was even worse, dropping to 58.75%. Considering only 500 series (which do not run north of Los Angeles), southbound train performance improved to 66.8%

Certain trains, of course, contributed far more than their fair share. 567, which leaves San Diego at 8:05am and has a scheduled arrival of 10:50am, had an on-time arrival only 8 out of the 41 days studied, a pathetic performance of only 19.5%. During February, that dropped to 14.3% with an average delay of 17 minutes and a typical delay of 12.5 minutes for the single worst performing train in the Surfliner route.

As was previously recounted on this blog, the Pacific Surfliner Express train, 599, actually performed worse than the rest of the line. When adjusting to a 5 minute definition, it’s performance improves to be slightly better than the rest of the pack, at 63.4%; however, it arrived on-time less than half the time in February, with only a 42.9% OTP and a typical delay of 9 minutes (however, half of the delays were 30 minutes or longer). This essentially wipes out the time savings which the express train is supposed to represent and is an extreme business failure. 

Weekend performance is slightly better, as would be expected given the greatly restricted number of train meets. Metrolink’s weekend service consists of only four trains per day, only two of which extend past Laguna Niguel to Oceanside, while Coaster has a somewhat more robust weekend schedule with a total of eight trains operating between Oceanside and San Diego. Nevertheless, Amtrak’s on time performance fails to exceed 70%, with 68.18% for northbound trains and 67.74% for trains terminating in San Diego, although a restriction to 500 series trains improves results to 77.9% for southbound trains.

This poor performance is despite extensive schedule padding which allows trains to make up as much as fifteen minutes. The padding is sufficiently egregious that the scheduled time from Solano Beach to San Diego ranges from 38-48 minutes, longer than the scheduled time for Coaster to make the same journey with two additional stops in Sorrento Valley and Old Town San Diego (scheduled time for Coaster being 37-40 minutes). Train 774, in particular, has some rather large schedule padding. While the statistics may be somewhat inflated due to a number of its trips being cancelled or shortened without arrival information available, it had an extremely high percentage of trips arrive substantially early. 21% of trains arrived late, but an astonishing 47% of trains arrived in excess of five minutes early to San Diego. Removing five minutes of padding from its schedule would result in only one additional of the trains which ran during the timeframe studied being declared late under the 5 minute threshold and none at all under Amtrak’s 10 minute threshold. 580 may also be subject to this as not a single additional train would be late with a five minute reduction in schedule time (however, it had a dreadful January performance with no weekday train running on time until the 31st; post-January it managed an overall 79.5% OTP).

While the Amtrak status archive does not give reasons for the delays, given the limited performance gain for the almost entirely commuter train free weekends, it seems evident that much of the problem is unrelated to train meets or preferential treatment by Metrolink and Coaster dispatchers, but rather to problems which are entirely of Amtrak’s own making, such as unreliable equipment leading to breakdowns.